Production Scheduling for the Process Industries [1 ed.] 1032302356, 9781032302355

Citation preview

“Don’t underestimate the revolutionary nature of the concepts described and recommended in this book. This should be required reading for leaders in operations roles. Had more of our manufacturing organizations been built on this structured scheduling methodology, I believe our [Covid] response would’ve been stronger, quicker, and far less painful to our manufacturing teams, sales teams, customers, and consumers.” Dave Rich – VP, Strategic Sourcing & Fulfillment, Litehouse Foods “Effective production scheduling is a critical tool to optimize productto-product transitions and one of the most critical factors to achieve truly effective use of your production resources. Peter King made a significant contribution to understanding and improving production scheduling in his first book. I have personally used his concepts with great benefit. Peter, Mac, and Noel have continued that work in this fine new book that will surely be of great value to process operators.” Raymond Floyd – SVP Suncor Energy (Retired) Current member of Manufacturing Hall of Fame, The Shingo Academy and the Baldrige Award Board of Overseers “By implementing planning wheels we were able to move from fill rates of ~75% to over 99% reliably in a 3 month timeframe. The approach to working with people on the floor captured in this book is key to managing the change needed to stabilize manufacturing. Having a predictable cycle of changeovers is huge to improve performance and improve morale on the factory floor.” David Kaissling – Chief Supply Chain Officer, Shearer’s Snacks; and head of supply chain for several fortune 500 companies. “… a comprehensive resource on the "how" and “why” of production scheduling and how it enables improved manufacturing performance and business success.” Dave Rurak – Executive VP, Integrated Operations and Supply Chain, W. L. Gore & Associates

“In my 35+ years in supply chain, it is rare to come across such an esteemed and knowledgeable group of practitioners in the area of production scheduling. This book is an outstanding reference and step by step guide on how to plan and schedule any repetitive manufacturing operation. … a ‘must-read’.” Paul Baris – VP Planning Strategy, enVista “The concept in this book along with the Phenix planning tools allowed us to move very complex scheduling rules from head knowledge into a cloud-based system. It has improved our speed of scheduling and the consistency of scheduling to our established rules.” Dave Stauffer – Director of Supply Chain, Advanced Food Products “This book is a wonderful overview of the benefits of Product Wheels including all the pressure testing our wheels have had in the most disruptive of environments. An international pandemic, labor compression, and record inflation have really made plant scheduling even more challenging than it has ever been. Product Wheels have been the backbone of which to “grab on to” for these difficult environments.” Mike Evans – Senior VP Operations, Bellisio Foods “Production scheduling has long been a massively neglected part of the equation for maximizing customer service and shop floor performance, while minimizing cost and capital. [This] is an exceptional read on the value, mechanisms and alternatives to optimize shop floor performance. Kudos! to King, Jacob and Peberdy for providing such a comprehensive and unbiased handbook to practitioners and leaders everywhere!” Mike Wittman – Formerly Chief Supply Chain Officer, Pinnacle Foods, now Senior Advisor, Boston Consulting Group

Production Scheduling for the Process Industries This book is aimed at manufacturing and planning managers who struggle to bring a greater degree of stability and more effective use of assets to their operations, not realizing the degree to which production scheduling affects those objectives. It has been reported that 75% of the problems on the manufacturing floor are caused by activities outside the plant floor. Poor production scheduling strategies and systems are often the biggest contributors to the 75%. The book explains in detail that no scheduling strategy, and especially no transition to a different and better scheduling strategy, will succeed without strong commitment and guidance from senior leadership. Leadership must understand their active role in the transition, that people will feel uncomfortable and even threatened by change, and that they will need to be measured by different standards. Effective scheduling requires that following the schedule and production to plan is more important than trying to maximize each day’s throughput. The book explains the advantages of a structured, regularly repeating schedule: how it can increase throughput, right-size inventory based on cycles and variabilities and therefore make it more usable, and improve customer delivery. It will explain the trade-offs between throughput, inventory, and delivery performance, how those trade-offs are actually decided in production scheduling, and how an appropriate scheduling strategy can make the tradeoffs and their ramifications visible. It discusses several popular structured scheduling concepts, their similarities and differences, to allow the readers to decide which might fit best in their environments. In addition, the authors discuss what makes an appropriate scheduling software system, why a package designed for structured scheduling offers capabilities well beyond the Excel workbooks used by many companies, and how it offers much more design capability and ease of use than the finite scheduling modules in SAP or Oracle. Finally, the authors offer a proven roadmap for implementation and critical success factors necessary to achieve the full potential and give examples of operations that have done this well. In addition, a guide for leaders and managers post-implementation is provided to help them fully exploit the advantages of a structured, repeating scheduling strategy.

Production Scheduling for the Process Industries Strategies, Systems, and Culture

Peter L. King, Mac Jacob, and Noel Peberdy

A PRODUC TIVIT Y PRESS BOOK

First published 2023 by Routledge 605 Third Avenue, New York, NY 10158 and by Routledge 4 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN Routledge is an imprint of the Taylor & Francis Group, an informa business © 2023 Peter King, Mac Jacob & Noel Peberdy The right of Peter King, Mac Jacob & Noel Peberdy to be identified as authors of this work has been asserted by them in accordance with sections 77 and 78 of the Copyright, Designs and Patents Act 1988. All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe. ISBN: 978-1-032-30235-5 (hbk) ISBN: 978-1-032-30236-2 (pbk) ISBN: 978-1-003-30406-7 (ebk) DOI: 10.4324/9781003304067 Typeset in Garamond by Deanta Global Publishing Services, Chennai, India

Contents Acknowledgments.....................................................................................xiii About the Authors......................................................................................xv Preface.........................................................................................................xix Section 1 INTRODUCTION 1 Business Imperatives: Why Scheduling Matters............................ 3 The Scheduler’s World Has Been Turned Upside Down............................ 3 The Challenge of Scheduling....................................................................... 4 Scheduling Is Even More Important............................................................. 5 Scheduling Is a Foundation of Manufacturing Performance....................... 6 Why Now?..................................................................................................... 7 2 Characteristics of Process Operations – And Scheduling Challenges..................................................................................... 9 Changeover Difficulty..................................................................................10 Starting Up after a Changeover...................................................................10 Sanitation Cycles..........................................................................................11 Shelf-Life Constraints...................................................................................11 Multi-Step Manufacturing.............................................................................11 Balancing Limited Resources.......................................................................12 Divergence vs Convergence.........................................................................16 Examples of “V” Type Process in Process Plants....................................18 Product Differentiation Points.....................................................................19 Limited Extra Capacity.................................................................................20 Summary......................................................................................................20 3 Overview of Production Strategies.............................................. 23 4 Scheduling Processes and Software............................................ 27 Production Planning................................................................................... 28 Scheduling....................................................................................................31 Supporting Processes...................................................................................31 Scheduling Software....................................................................................33 Goal-Seeking Algorithms.............................................................................34 Repetitive Scheduling...................................................................................35 

v

vi  ◾ Contents

The Scheduling Process...............................................................................35 Software Selection........................................................................................37 5 Example Process.......................................................................... 39 The Process..................................................................................................39 Scheduling Information Flow: Communication between Systems.............41 The Products................................................................................................43 Product Differentiating Characteristics........................................................43 Cultural Challenges......................................................................................45 Section 2  SCHEDULING STRATEGIES 6 Repetitive Scheduling Strategies................................................. 49 Product Wheels............................................................................................50 Product Wheel Design.............................................................................51 Synergy with Lean...................................................................................59 Benefits of Product Wheels.....................................................................60 Repetitive flexible Supply (RfS)...................................................................61 Rhythm Wheels............................................................................................63 Fixed Sequence Variable Volume (FSVV).................................................. 64 Summary..................................................................................................... 64 7 Dealing with Disruption.............................................................. 67 The Nature of Disruption............................................................................67 Ability to Deal with Disruption.................................................................. 68 An Example: The Story of P&G Luvs Diapers............................................70 Section 3  SCHEDULING PROCESSES, SYSTEMS, AND SOFTWARE 8 The Role of Forecasting............................................................... 75 Forecast Value Add......................................................................................76 Bias and Accuracy........................................................................................76 Coefficient of Variation................................................................................78 Timing and Aggregation..............................................................................79 Different Forecast Goals............................................................................. 80 Choice of Demand Forecasting Unit...........................................................81 Product Transitions......................................................................................81 Product Segmentation for Forecasting.........................................................82 Summary......................................................................................................83 9 The Role of Inventory.................................................................. 85 Components of Inventory........................................................................... 86 Managing Inventories...................................................................................87 An Inventory Management Example...........................................................89 Cycle Stock and Safety Stock...................................................................... 90 Calculating Safety Stock...............................................................................91 Variability in Demand..................................................................................92

Contents  ◾  vii

Seasonality................................................................................................97 Variability in Lead Time...............................................................................97 Combined Variability...............................................................................97 Cycle Service Level and Fill Rate................................................................ 98 Safety Stock and Lot Size Impact............................................................... 99 Summary................................................................................................100 10 Typical Scheduling Process Steps.............................................. 103 The Planning and Scheduling Process......................................................103 Exception Management.............................................................................104 Preparing to Plan.......................................................................................105 Creating the Production Plan.....................................................................105 Creating the Detailed Schedule.................................................................106 Communicating the Plan...........................................................................106 The Packing Line Schedule................................................................106 ERP and Shop Floor Systems.............................................................107 The Mixing Schedule.........................................................................107 The Spice and Liquid Prep Rooms....................................................107 Preparing for Tomorrow............................................................................108 The Detailed Scheduling Process..............................................................108 Scheduling the Constraint.......................................................................... 110 Manual Scheduling................................................................................. 110 Just-in-Time Scheduling......................................................................... 110 Repetitive Sequence Scheduling............................................................ 111 KPI-Based Algorithms and Solvers............................................................ 111 Resources.................................................................................................... 112 Evaluating and Adjusting the Schedule..................................................... 113 Releasing Firm or Committed Orders....................................................... 113 11 Multi-Level Scheduling...............................................................115 Product Mix and Moving Bottlenecks....................................................... 116 Types of Scheduling Problems.................................................................. 117 Degrees of Freedom between Levels.................................................... 117 Impact of the Constraint’s Location....................................................... 118 More Than Two Levels.............................................................................. 118 Batch and Lot Size Restrictions................................................................. 118 Distribution Rules....................................................................................... 119 Logical Relationships between Levels.......................................................121 Linking between activities.........................................................................122 The Multi-Level Scheduling Process..........................................................122 Scheduling with Inventory Constraints between Levels...........................123 12 Tanks, Bins, and Flow Paths..................................................... 125 Tank and Bin Scheduling..........................................................................127 Tank Scheduling Example.........................................................................127

viii  ◾ Contents

Specific Flow Paths....................................................................................129 Converging Flows...................................................................................129 Diverging Flows.....................................................................................130 Before and after APS Implementation...................................................130 Simplifying the Complex........................................................................... 131 13 The Role of ERP Systems in Planning and Scheduling............. 133 Assumption of Infinite Capacity................................................................133 Daily Time Resolution................................................................................134 Assumption of Independence...................................................................134 ERP Scheduling Modules...........................................................................135 Repetitive Scheduling in an ERP System...................................................136 Quality Management..................................................................................137 System of Record........................................................................................137 14 Excel as a Finite Scheduling Tool.............................................. 139 Business Continuity...................................................................................140 Critical Features of Scheduling Software.................................................. 141 Issues with Excel........................................................................................ 141 Visibility of Attributes and Sequencing.....................................................143 Time Offsets...............................................................................................143 Lot Sizing and Multi-Level Scheduling......................................................144 Summary....................................................................................................144 15 Software Designed for Production Scheduling..........................145 Supporting Processes................................................................................. 145 Scheduling Requirements..........................................................................146 Repetitive Scheduling Requirements......................................................... 147 Multi-Level Requirements.......................................................................... 147 Software Selection...................................................................................... 147 16 Critical Ingredients, Raw Materials, and Components...............151 Availability Checking................................................................................. 151 Critical Materials......................................................................................... 152 The Firm Zone Strategy............................................................................. 152 Strategy Examples...................................................................................... 153 Summary.................................................................................................... 155 17 Scheduling Software: Security and Privacy................................157 Security....................................................................................................... 158 Privacy........................................................................................................ 159 Section 4  PREREQUISITES TO GOOD SCHEDULING 18 The Role of the Plant Leader......................................................163 Future Proof the Plant................................................................................163 Raw Material Supply Risk......................................................................164

Contents  ◾  ix

Standardizing Packaging Raw Materials................................................164 New Product Development Involvement..............................................164 Transportation Risks..............................................................................165 Labor Risk...............................................................................................165 Simplifying the Product Portfolio..........................................................165 Selective Automation..............................................................................165 Improve Changeovers............................................................................165 Example..................................................................................................165 Dealing with Disruption............................................................................166 Collaboration..........................................................................................166 Physical Triage Meetings........................................................................166 Implementing a Virtual Team in the Plant............................................ 167 What Is Needed of the Plant Leader?....................................................168 Reinforcing Repetitive Patterns of Production..........................................168 Summary....................................................................................................169 19 Scheduling Readiness Criteria....................................................171 Readiness and Sustainability...................................................................... 173 Project Roles............................................................................................... 173 Readiness Examples................................................................................... 176 20 Accessible, Accurate, and Complete Data...................................177 Master Data and Transaction Data............................................................177 Examples of Data Accuracy and Timeliness Problems............................ 178 Data Audits or Checking Practices............................................................ 178 Documenting the Process.......................................................................... 178 Checking Data against a Standard............................................................ 179 Measuring and Tracking Results against a Goal....................................... 179 Analyzing the Root Cause of Gaps...........................................................180 Leadership Visibility...................................................................................180 Planning and Scheduling Data..................................................................180 Summary....................................................................................................181 21 Effective Production and Capacity Planning............................. 183 The Importance of Planning.....................................................................183 Resolving Overloads..................................................................................184 Automated Planning...................................................................................185 Planning Example......................................................................................186 Characteristics of a Good Production Plan...............................................187 Managing Inventory Targets and Constraints...........................................188 Summary....................................................................................................189 22 Workforce Engagement...............................................................191 Selling the Idea.......................................................................................... 191 Designing the New Process....................................................................... 193 Executing the New Process....................................................................... 193

x  ◾ Contents

23 Changeover Reduction – SMED..................................................195 SMED and Its Origins................................................................................195 SMED Concepts..........................................................................................196 Process Industry Changeovers...................................................................197 Automotive Fluids Packaging.....................................................................198 Diaper Manufacturing................................................................................199 SMED Beyond Product Changes...............................................................200 A Non-Manufacturing Example.................................................................201 SMED Applied to Blue Lakes Packaging...................................................201 Summary....................................................................................................202 24 Production Stability................................................................... 203 Total Productive Maintenance...................................................................204 TPM Relevance in Process Industries........................................................205 TPM Saves Money..................................................................................206 Overall Equipment Effectiveness (OEE)....................................................206 Availability..............................................................................................206 Performance...........................................................................................207 Quality....................................................................................................207 Calculation of OEE.............................................................................209 VSM Data Boxes: OEE........................................................................209 Non-Standard OEE Metrics........................................................................210 Summary.................................................................................................... 211 25 Cellular Manufacturing.............................................................. 213 Typical Process Plant Equipment Configurations.....................................213 Cellular Manufacturing Applied to Process Lines..................................... 215 Synthetic Sheet Manufacturing Example.................................................. 217 Virtual Cell Implementation in a Synthetic Rubber Production Facility.. 219 Would Cellular Flow Apply to the Salad Dressing Operation?.................223 Group Technology.....................................................................................223 Summary....................................................................................................225 26 Managing Bottlenecks and Constraints..................................... 227 Poor Scheduling Can Cause Bottlenecks..................................................228 Moving Bottlenecks...................................................................................228 Scheduling Moving Bottlenecks................................................................230 Summary....................................................................................................233 Section 5  GETTING TO SUCCESS 27 Leading Scheduling Improvements to Drive Value: Five Steps for Leaders................................................................................. 237 Laying the Foundations for Effective Scheduling......................................238 Five Steps to Value for Leaders..................................................................238 Step 1: Layout the Improvement Goals and Plan..................................238

Contents  ◾  xi

1. Develop a Tangible Vision............................................................238 2. Communicate to Leaders and Other Stakeholders.......................239 3. Identify Supporters and Cheerleaders..........................................240 4. Develop an Incremental Implementation Plan.............................240 5. Develop a Change Plan.................................................................241 Step 2: Work on the Culture..................................................................241 6. Bring the Voice of the Customer into the Plant...........................241 7. Improve Shop Floor Discipline......................................................242 8. Implement Weekly Reporting and Drive Improvement...............242 9. Freeze the Frozen Horizon!...........................................................243 10. Dealing with Schedule Disruption..............................................243 Step 3: Improve Scheduling...................................................................245 11. Implement Simple Product Wheel Scheduling as a Team..........245 12. Drive Further Improvements.......................................................246 13. Celebrate Successes.....................................................................246 14. Align the Plant to the Wheel Rhythm.........................................246 Step 4: Take Stock..................................................................................247 15. Review Progress...........................................................................247 16. Lessons Learned...........................................................................247 17. Decide on the Full Plant Rollout.................................................247 18. Select Scheduling Software.........................................................247 19. Select an Implementation Consultant..........................................247 20. Get Budget Approval...................................................................248 21. Plan the Full Implementation......................................................248 Step 5: Sustaining the Gains..................................................................248 22. Ownership Is Key........................................................................248 23. Establish Sustainable Practices Early...........................................248 24. Verify That Sustainment Practices Are Working.........................249 25. Formalize Training, Qualification, and Coaching.......................249 26. Track the Key Benefits................................................................250 27. Take Advantage of Vendor Software Improvements..................250 28. Implement a Planning Community of Practice (COP)...............250 28 Where to Begin: A Roadmap to Project Success.........................251 Initial Preparation......................................................................................252 Scheduling System Design.........................................................................254 Strategy Design..........................................................................................255 Final Preparation........................................................................................257 Sustaining...................................................................................................258 Summary....................................................................................................259 29 Critical Success Factors...............................................................261 Scheduling Strategy Critical Success Factors.............................................261 Scheduling System Critical Success Factors...............................................262 Cultural and Behavioral Critical Success Factors......................................262

xii  ◾ Contents

30 Success Stories: Examples of Scheduling Best Practices........... 265 Dean Bordner – Nature’s Bounty..............................................................265 Formerly Senior VP of Operations, Nature’s Bounty (Now The Bountiful Company)...............................................................................265 Mike Evans – Bellisio Foods......................................................................267 VP of Operations...................................................................................267 Dave Rich – Litehouse Foods....................................................................268 Vice President, Strategic Sourcing & Fulfillment...................................268 James Overheul – BG Products.................................................................269 Formerly Operations Director................................................................269 Ryan Scherer – Appvion............................................................................270 Former Organizational Excellence and Capacity Manager...................270 David Kaissling – Shearer’s Snacks...........................................................270 Formerly Chief Supply Chain Officer for Shearer’s Snacks, With 40 Years in CPG as Head of Supply Chain for Fortune 500 Companies.......................................................................................270 Raymond Floyd – Exxon Mobil.................................................................271 SVP Suncor Energy (Retired), Current Member of Manufacturing Hall of Fame, The Shingo Academy, and the Baldrige Award Board of Overseers................................................................................271 Ethylene Co-Polymers – Sabine, TX..........................................................271 Martin Fernandes – Dow Chemical..........................................................273 Director of Supply Chain Innovation....................................................273 Dave Stauffer – Advanced Food Products.................................................273 Director of Supply Chain.......................................................................273 Index.......................................................................................................... 275

Acknowledgments In writing this, we are simply documenting the thoughts, ideas, concepts, and real work done in collaboration with a number of our colleagues: ◾ First and foremost, Alan Nall, who has been collaborating with us for the past seven years on product wheel implementation and on developing the processes and software to enable them. Alan contributed significantly to the concepts, insights, and text contained herein. He is a business consultant with Zinata specializing in manufacturing improvement, program, and project management. Alan previously spent 28 years with Eli Lilly, Procter & Gamble, and DuPont. ◾ John Peberdy, an experienced CTO and software architect, whose work showed us the possibilities that could be achieved with modern software tools, and who was the primary developer of Phenix Planning and Scheduling. John authored Chapter 17 and contributed his thoughts and ideas to other chapters. He has been the architect of a pharmacovigilance application, currently $50+ MM ARR, and has had roles in startups and major software houses. ◾ Bennett Foster, a 40-year veteran of the DuPont Company and product wheel SME, who provided simulation modeling and analytical capabilities, leading to the development of advanced algorithms in Phenix PS. ◾ Sasha Velykoivanenko, with a PhD in Probability and Statistics and 23 years practicing these skills with Procter & Gamble, who developed many of the advanced analytics underlying Phenix PS. ◾ Mia Nall, who recently graduated with a degree in Applied Math and Computer Science from the University of Pittsburgh, has been instrumental in the development of Monte Carlo safety stock simulations and other tools inherent in Phenix PS. ◾ Cheryl Boerjan and Kathy Skinner, supply chain consultants with Zinata, and former P&G supply chain managers, who wrote much of the text in Chapter 20 on accessible, accurate, and complete data and have practiced what they recommend for many years.

xiii

xiv  ◾ Acknowledgments

A number of others contributed their skills and expertise to making Zinata and Phenix PS viable, successful businesses: John Theron, Keri Akmezikyan, Tim Ollivier, Laurence Grisel, Andrea Sikorski, and Chris Thompson. We’ve had the pleasure of collaborating with a number of clients who have provided an environment for us to test, refine, and further develop the concepts described here, including: ◾ Advanced Food Products: Dave Stauffer, Spencer Heisey, and Cleann Pauley ◾ Bellisio Foods: Mike Evans, Cindy Miller, Kristi Wedel, Mariah Benia, and Lisa Caudill ◾ Litehouse Foods: Dave Rich, Charity Hegel, and Dan Munson ◾ The Bountiful Company: Paul Bittinger, Mitch Slade, Dean Bordner, Adelino Rivera, Jaci Souza, Benvinda Santos, Joseph Ricchetti, Ivy Garcia, and Cheryl Alker ◾ BG Products: James Overheul, Matt Peterson, Lisette Walker, and Gregg McCabe ◾ Shearer’s Snacks: David Kaissling, Greg Cook, Eric Krizay, and Vince Amato Finally, our wives, Bonnie, Suzie, and Ellie, who provided encouragement and support throughout this effort. Pete King, Mac Jacob, and Noel Peberdy

About the Authors Peter L. King is the president of Lean Dynamics, LLC, where he has spent the last 15 years applying lean concepts and tools to a diverse group of clients in the chemical, food and beverage, consumer products, and nutraceutical industries. Prior to founding Lean Dynamics, Pete spent his career with the DuPont Company, in a variety of control systems, manufacturing automation, continuous flow manufacturing, and lean manufacturing and lean supply chain assignments. The last 18 years at DuPont were spent applying lean techniques to a wide variety of products, including sheet goods like DuPont™ Tyvek®, Sontara®, and Mylar®; fibers such as nylon, Dacron®, Lycra®, and Kevlar®; automotive paints; performance lubricants; bulk chemicals; adhesives; electronic circuit board substrates; and biological materials used in human surgery. On behalf of DuPont, Pete consulted with key customers in the processed food and carpet industries. Pete retired from DuPont in 2007, leaving a position as Principal Consultant in the Lean Center of Competency. Pete has also served as a senior business consultant with Zinata, Inc. for the past nine years, providing product wheel guidance to clients in the nutraceutical and food industries. Pete received a bachelor’s degree in Electrical Engineering from Virginia Tech, graduating with honors. He is Six Sigma Green Belt certified (DuPont, 2001), Lean Manufacturing certified (University of Michigan, 2002), and an APICS Certified Supply Chain Consultant (CSCP, 2010). He is a member of the Institute of Industrial and Systems Engineers, the Association for Manufacturing Excellence, the Association for Supply Chain Management, and APICS. Pete has authored four books published by Productivity Press: Lean for the Process Industries, first and second editions (2009, 2019), The Product Wheel Handbook (2013), and Value Stream Mapping for the Process Industries (2015). He has authored a dozen magazine articles, and is a frequent presenter at technical society conferences. Pete is an avid runner, having completed a marathon and close to two hundred 5K, 5-mile, and 10K races. He currently resides in Rehoboth Beach, DE, with his wife, Bonnie King.

xv

xvi  ◾ About the Authors

Mac Jacob has implemented four generations of Advanced Planning and Scheduling Software and SAP MRP II at over 100 Procter & Gamble sites worldwide. He started as a project engineer at a manufacturing site and moved through assignments as production line manager, production planning manager, site logistics manager, and North American planning manager for Luvs Diapers. He began to see how the lack of supply chain systems prevented the diaper business from executing its product and manufacturing strategy and led a project to improve P&G’s planning systems. He was the business leader, developed the business processes, and wrote the original training materials for most of P&G’s supply chain processes: Production Execution, Warehouse and Shipping, Distribution Requirements Planning, Site Planning, Category Planning, and Supply Chain Master Data. Mac is a recipient of P&G’s Magnus Award for lifetime contribution to supply chain improvement. He is APICS Certified in Production and Inventory Management (CPIM), with the SCOR-P endorsement. He is certified by Oliver Wight as an MRP II instructor and was a P&G Lead Instructor and Master for Site Planning, DRP, and Supply Chain Master Data. Since retiring from P&G, Mac has worked on several global supply chain management projects as a consultant, and is currently the Head of Product for Phenix Planning and Scheduling. Mac graduated Cum Laude from the University of Michigan with a degree in Naval Architecture and Marine Engineering. He was on the sailing team and captain of the ski team. He earned his MBA from Xavier University. He now lives in Harbor Springs, Michigan, with his wife of 39 years, Suzie, and they have two children. When not designing and implementing production planning software, he is active in sailing, bicycling, skiing, and as an assistant coach of the Harbor Springs High School Ski Team. Noel Peberdy has had a multi-faceted career across the value chain of process manufacturing operations, focusing on people and process, harnessing technology to drive transformative change. His early career centered on solving complex system-level problems using dynamic simulation to optimize mineral processing, metallurgical and food/beverage plants, and operations in deep underground mines. Building a plant in the computer, using mathematical models to simulate the reality of the system, led to his career focus – “connecting the dots” and “bridging silos.” He was also fortunate to have led numerous rescue operations on failed or troubled projects. A newly developed gold mine in which the extraction plant control system didn’t work and put people’s lives at risk. An offshore gas platform on which the safety systems and emergency shutdown systems did not work. A Brewing plant that was performing well below potential. Navigating through these crises – still meeting key dates – demanded out-of-the-box methods, ruthless focus on the ingredients for success – and importantly, an awareness of the interdependencies between the moving parts – People, Process, Technology, and Data aspects – to architect long-term success.

About the Authors  ◾  xvii

At the other end of the spectrum, he has had a central role in conceiving, designing, and bringing several transformative greenfield projects to life. This gave him and his team opportunities to architect plants of the future, fully realizing lean maturity stages 3–4 capabilities and performance. Production scheduling in a process manufacturing operation is a complex systems-level challenge. In his consulting work, Noel realized that scheduling is frequently one of the most significant disconnects in the process manufacturing value chain. Noel has an MSc in Engineering in Distributed Computer Control Systems from the University of Cape Town. He has founded four successful companies in Africa and North America. He is an avid outdoor person. Canoeing and hiking in the wilds of Canada is his passion. He has been married to Ellie Zweegman for 37 years, with whom he has three children. He lives in Southern Ontario, Canada.

Preface

Why We Decided to Write This Book The three of us decided that a book like this was needed because of the complete lack of any comprehensive treatment of the three things necessary for optimal production scheduling of process industry operations: ◾ A strategy that recognizes changeover difficulties and complexity and equipment constraints and limitations rather than relying on Manufacturing Requirements Planning (MRP) concepts that emphasize due dates over operating efficiency, ignoring that both are achievable ◾ Electronic tools that make implementing that strategy straightforward and facilitate effective and efficient weekly scheduling processes ◾ The need for leadership to recognize and practice appropriate behaviors to enable effective scheduling, especially if getting there requires any degree of transformation. And that there are often cultural barriers that must be overcome in the transition. We feel well qualified to discuss these topics because each of us has more than 30 years of process industry experience, with each having focused on one or more of the above topics throughout most of our careers. Our central message is that good scheduling is vitally important to manufacturing success, and in turn to business success; without good scheduling, business profitability will suffer (Figure 0.1). By a good schedule, we mean one that has the agility and flexibility to meet disruption, is accepted and supported by the production organization, can be executed, minimizes total delivered costs, and meets the customer’s quantities and due dates. Note that we are talking about a good schedule,

xix

xx  ◾ Preface CUSTOMER INPUT

PLANNED ORDERS

DEMAND HISTORY MARKETING INPUT

FORECASTING (ERP)

DEMAND MANGMT (ERP)

MRP (ERP)

SCHEDULING ERP? Excel? Custom S/W?

FINITE SCHEDULE

PRODUCTION (MES)

SEASONALITY WHERE THE RUBBER HTS THE ROAD PROMOTIONS

OFTEN TAKEN FOR GRANTED NOT GIVEN THE ATTENTION OR RESOURCES IT DESERVES

Figure 0.1  Scheduling can be the most consequential part of the planning process.

not a perfect one. An effective schedule depends much more on agility than on perfection. Each of the above-mentioned requirements for a good schedule is covered in detail in the following chapters.

How the Book Is Organized The book consists of five sections.

Section 1: Introduction ◾ Chapter 1 – Business Imperatives – Why Scheduling Matters – The connection between good scheduling and business success is explained. The chapter introduces a central theme of the book, that the financial success of any manufacturing business is closely tied to manufacturing productivity, which is tightly linked to how well production scheduling is done. ◾ Chapter 2 – Characteristics of Process Operations – and Scheduling Challenges – Process operations face several challenges that discrete operations experience to a far lesser degree. We explain why scheduling these are more complex than similar tasks in discrete parts assembly manufacturing. ◾ Chapter 3 – Overview of Production Strategies – Repetitive, structured strategies have been well proven in practice to produce more effective schedules than the seemingly random schedules that an MRP-based system will generate. Several variations of this concept will be introduced here, with a detailed explanation of each in Chapter 6. ◾ Chapter 4 – Scheduling Processes and Software – An overview of the planning and scheduling process steps and the software available to support them is presented. ◾ Chapter 5 – Example Process – A plant producing many flavors of salad dressing, packaged in several configurations, will be used to illustrate all of the concepts and practices recommended throughout the book. The product families this plant makes, the challenges its varied product mix introduces to any scheduling process, and how they had been dealing

Preface 

with these challenges before the transition to these practices, will be described in this chapter.

Section 2: Scheduling Strategies ◾ Chapter 6 – Repetitive Scheduling Strategies – This expands on the strategies touched on in Chapter 3. It explains the advantages of a structured, regularly repeating sequence of all products. It defines Product Wheels, one such strategy, in detail, and gives an example of a product wheel on one of the process lines described in Chapter 5. It describes other, quite similar strategies, Rhythm Wheels, Fixed Sequence Variable Volume, and Repetitive Flexible Supply, and explains similarities and differences. It explains how each of these achieves the lean goal of production leveling in a way that is much more relevant to process operations than traditional Heijunka methods. ◾ Chapter 7 – Dealing with Disruption – All manufacturers are facing unprecedented disruptions today, including highly volatile forecasts, material shortages, staffing problems caused by Covid (and likely future pandemics), logistics interruptions, and the effects of extreme weather events. How an agile, repetitive scheduling strategy better positions you to deal with all of these interferences will be covered, as will additional countermeasures that can be taken to further mitigate the problems.

Section 3: Scheduling Processes, Systems, and Software ◾ Chapter 8 – The Role of Forecasting – This chapter focuses on forecasting and how it impacts scheduling, that while forecasts should influence schedule design, weekly schedules should be based on real consumption rather than forecasts. Forecast bias is defined, and why it should be recognized and reduced or eliminated covered. The need to record forecast errors with the appropriate time windows is described. ◾ Chapter 9 – The Role of Inventory – This chapter discusses determining appropriate inventory levels and managing inventories within target bands to meet customer service goals while producing efficiently and minimizing working capital. ◾ Chapter 10 – Typical Scheduling Process Steps – This chapter describes the steps in a typical scheduling process from exception management through communicating the schedule to the shop floor, the choices of how the schedule can be created, and the impact of resource availability.

◾  xxi

xxii  ◾ Preface

◾ Chapter 11 – Multi-Level Scheduling – Multi-level schedules fall into categories based on the location of the process constraint and the degree of freedom between the levels. This chapter discusses the considerations and techniques for scheduling a multi-level manufacturing process to efficiently schedule the constrained level and align the other levels’ schedules to the constraint. ◾ Chapter 12 – Tanks, Bins, and Flow Paths – In some multi-level operations with tanks, bins, and specific flow paths between operations, greater detail is needed to visualize the contents of the tanks and bins at any point in time, and to model or encode the flow path restrictions into the scheduling software. ◾ Chapter 13 – The Role of ERP in Planning and Scheduling – The role of Enterprise Resource Planning (ERP) systems in scheduling, and as a system of record, is described. The functions they perform well, and the functions they struggle with are covered. We explain why they sometimes have difficulty with structured, repetitive scheduling strategies. ◾ Chapter 14 – Excel as a Finite Scheduling Tool – Many companies use an internally developed Microsoft Excel workbook to do their finite production scheduling. This chapter covers the things people find attractive about that approach as well as the difficulties in meeting the critical features that are required for scheduling, and the risks of errors and business continuity. ◾ Chapter 15 – Software Designed for Production Scheduling – Several commercially available software products overcome the problems people have experienced with Excel. They are easier to configure for the scheduling strategies we recommend than most ERP systems and generate far more efficient schedules. The advantages these packages provide are explained, and desirable features are covered. ◾ Chapter 16 – Critical Ingredients, Raw Materials, and Components – It almost goes without saying that raw materials are required to produce a finished product; this chapter discusses checking material availability during scheduling and aligning production and procurement strategies to support material availability, production efficiency, and customer service. ◾ Chapter 17 – Scheduling Software – Security and Privacy – Some of the advanced scheduling systems described reside “in the cloud,” i.e., on remote servers. Some process manufacturers have concerns about their proprietary process data being exposed. This chapter describes the steps that can be taken to provide very effective protection and security.

Section 4: Prerequisites to Good Scheduling ◾ Chapter 18 – Role of the Plant Leader – The plant leader has never had a more challenging role. This chapter describes key steps that plant leaders

Preface 

should take to future-proof the plant in today’s disrupted supply chains and establish virtual cross-functional teams to respond to adverse events in real time. The plant leader plays a critical role in championing these changes, which impact ways of working, collaborative problem-solving under pressure, and rapid decision-making with limited information. Leading and managing the changes at the plant to achieve improved agility and resilience is the challenge – all while maintaining production performance. ◾ Chapter 19 – Scheduling Readiness Criteria – The entire organization must be ready to participate in and accept the changes that any transition will entail for it to be fully successful. The factors that determine the level of readiness are described. ◾ Chapter 20 – Accessible, Accurate, and Complete Data – Accurate and timely data is needed to create a schedule that can be followed on the factory floor, and that won’t need to be frequently changed. The best practices for auditing or checking data are covered, including the types of data to check, and the importance of each data element having an easily accessible single source of truth. ◾ Chapter 21 – Effective Production and Capacity Planning – Without an effective plan and an accurate understanding of manufacturing capacity, you can’t have confidence that the schedule is achievable. Examples of companies that did this well are described. ◾ Chapter 22 – Workforce Engagement – It is critical that the entire workforce is engaged in the transition, that they feel included in decisionmaking, and that they understand that they are contributing to the change rather than having the change imposed on them. A corollary of this is to build value for Standard Work instead of embracing the frequently found firefighter mentality. ◾ Chapter 23 – Changeover Reduction – SMED – Long or costly changeovers constitute one of the most negative influences on production scheduling. They steal plant capacity. Reducing changeover time and cost can provide more latitude in designing a scheduling strategy and thus lead to better schedules. Examples using elements of the Single Minute Exchange of Dies (SMED) methodology will be given. ◾ Chapter 24 – Production Stability – For any schedule to be reliably achieved, operational stability is a must; the manufacturing process must be reliable and predictable. How Total Productive Maintenance (TPM) can drive this will be covered. Overall Equipment Effectiveness (OEE), the most widely used measure of production stability, will be defined. ◾ Chapter 25 – Cellular Manufacturing – Cellular Manufacturing, in cases where the plant footprint warrants it, can greatly simplify scheduling, increase useful capacity, and reduce manufacturing cycle time. If the plant footprint is not appropriate, Group Technology, a subset of Cellular

◾  xxiii

xxiv  ◾ Preface

Manufacturing, is often equally beneficial. Virtual cellular flow and group technology will be explained, with examples of both. ◾ Chapter 26 – Managing Bottlenecks and Constraints – This chapter defines bottlenecks and describes why bottlenecks can be hard to see and hard to manage in a process line, and that the bottleneck can move up or down the line depending on the specific Stock Keeping Unit (SKU) being produced. An example of how opening a bottleneck led to improved schedules will be given.

Section 5: Getting to Success ◾ Chapter 27 – Leading Scheduling Improvements to Drive Success – Five Steps for Leaders – This is a practical guide to leaders implementing scheduling improvements, focused mainly on the human and performance management aspects. The critical topic of focused change management is a central theme, aimed at driving improved plant performance. ◾ Chapter 28 – Where to Begin – A Roadmap to Project Success – A project implementation roadmap is laid out in detail and defines how to proceed through an entire scheduling transformation project. It has several on-ramps which can be taken depending on your current situation and readiness for change. ◾ Chapter 29 – Critical Success Factors – Certain things must be in place or developed early in the implementation of any scheduling strategy and for any scheduling process that implements that strategy to be successful. ◾ Chapter 30 – Success Stories – Examples of Scheduling Best Practices – Several cases where process manufacturers employed the concepts we recommend are described, with the operating and business improvements they enabled.

Who the Book Is Written For The book should interest Operations VPs and managers; Supply Chain VPs and managers; Planning VPs and managers; plant managers; production schedulers; Operational Excellence, Continuous Improvement and Industrial Engineering VPs and SMEs; Lean practitioners; and manufacturing company IT departments. While all of these groups will benefit from the insights throughout the book, some may find specific chapters more relevant to their challenges. For that reason, each chapter has been written to be relatively self-contained. We recommend that people in each of the following groups focus most of their attention on the chapters indicated in the table below, Figure 0.2.

CONTINOUOS IMPROVEMENT

PLANNING & SCHEDULING

ELECTRONIC SYSTEMS IT

CHAPTER TITLE

X

X

X

X

x x

x x

x

x

x

x x

x x

x x

LEADERSHIP

CHAPTER NUMBER

OPERATIONS

Preface 

X x

x x

FRONT MATTER Front Front Front 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

Acknowledgements About the Authors Preface

INTRODUCTION Business Imperaves - Why Scheduling Maers Characteriscs of Process Operaons - and Scheduling Challenges Overview of Producon Strategies Scheduling Processes and So„ware Example Process SCHEDULING STRATEGIES Repeve Scheduling Strategies Dealing with Disrupon SCHEDULING PROCESSES, SYSTEMS, SOFTWARE The Role of Forecasng The Role of Inventory Typical Scheduling Process Steps Mul-Level Scheduling Tanks, Bins, and Flow Paths The Role of ERP Systems in Planning and Scheduling Excel as a Finite Scheduling Tool So„ware Designed for Producon Scheduling Crical Components Scheduling So„ware Security and Privacy PREREQUISITES TO GOOD SCHEDULING Role of the Plant Leader Scheduling Readiness Criteria Accessable, Acurate and Complete Data Effecve Producon and Capacity Planning Workforce Engagement Changeover Improvement - SMED Producon Stability Cellular Manufacturing Managing Bolenecks and Constraints GETTING TO SUCCESS Leading Scheduling Improvements to Drive Value -Five Steps for Leaders Where to Begin – a Roadmap to Project Success Crical Success Factors Success Stories - Examples of Best Pracces

x x

x

x x x x x x x x x x

x x

x x x x x

x x x x x x x x

x x x x x

x x

x x x

x x

x

x x

x

x

x

Figure 0.2  Recommended chapters.

A Final Note When we started writing this book in 2021, COVID-19 was in full swing. Supply chains were so severely disrupted that no one could have imagined that disruption at that level might be here to stay. Yet almost three years since the onset of the pandemic, we hear from so many that their supply chains have never been more disrupted. Certainly, disruption had been a growing phenomenon before the pandemic for many years. Common wisdom now accepts that disruption will be a reality in our future at elevated levels. And even if it does subside, the threat is ever-present: We need to be prepared for future pandemics, extreme weather, politically driven disruptions to supply, and so on. A new model is needed.

◾  xxv

xxvi  ◾ Preface

This book provides pointers to scheduling in the future. It is not possible to state solutions definitively since the situation is so fluid, and it’s not clear what the new reality will be. However, a few things are certain. Historically, big transformations focused investment in sunny day scenarios – how can we organize ourselves in an optimum manner? Today we have to accept that much of our efforts in the future will be dealing with unexpected situations – and that practices and investments must include dealing with volatility. Linear plan–execute–replan processes are a thing of the past. The speed of problem detection and the speed of response replaces accuracy as the mantra of planning. Concurrent planning and scheduling is a new reality. Planning and scheduling can no longer be put into separate buckets and treated independently. Most plants have been designed when stability was the norm. Plant performance has typically been measured on efficiency and cost metrics. Plant design has targeted the same objectives, resulting in long runs where changeovers were of much less concern. This severely limits agility, and consequently resilience. The repetitive scheduling approaches described in this book aim to make the best use of the plant assets while dealing with disruption. It is a key capability for the future.

INTRODUCTION

1

Chapter 1

Business Imperatives: Why Scheduling Matters

The Scheduler’s World Has Been Turned Upside Down As we write this book in late 2022, supply chains are still heavily disrupted by COVID-19-induced staff shortages, supply shortages, and port closures. A worldwide shortage of food starch, triggered partly by a shortage of propylene oxide necessary for starch production, coupled with transportation issues, caused innumerable downstream impacts; for example, see Figure 1.1. The lack of availability of equipment spares has resulted in plants temporarily closing down lines until they can source replacement parts, thus curtailing production. The resulting shortages in one company can lead to a shift of demand to another. In severe cases, when all suppliers experience the same shortage, their customers are forced to move to alternate materials. The compounding effect of these issues on manufacturing companies has severely disrupted their entire operations. This is further compounded by political events in Ukraine and elsewhere, impacting energy costs and causing sudden shifts in the food supply chain. The fiscal policies that have triggered rampant inflation compound the political uncertainties and affect the cost of borrowing and inventory. And the ever-growing impacts of extreme weather events add to the long list of disturbances that complicate the lives of all involved in the supply chain today. The level of volatility is expected to decline but to a higher residual level than before. Also, the threat of another pandemic-like situation hangs over the industry, dictating that companies must continue to build and retain their capabilities to deal with more uncertainty and volatility. The rubber meets the road in all aspects of supply chain execution, but nowhere more so than in the plant. And it is the plant scheduler that is DOI: 10.4324/9781003304067-2

3

4  ◾ Introduction

Figure 1.1  The knock-on impacts of raw material shortages. Source: https://www​.callifd​.com​/uploads​/1​/3​/0​/7​/130789654​/ventura_-​_supply​_disruption​_notification_-​_modified​_food​_starch​_7​.1​.2021​.pdf.

impacted most, faced with impossible choices every day. Armed with Excel in most cases, the complexity of how to deal with a host of disruptions is hard enough. Efficient scheduling is almost impossible without better tools.

The Challenge of Scheduling The life of the scheduler is stressful and frustrating. Every day he or she is expected to solve a complex puzzle: how to deal with multiple, compounding disruptions, to meet the ever-changing demand most efficiently. The weaknesses of higher-level planning systems based on classic Manufacturing Requirements Planning (MRP) algorithms have become clear. Inadequate constraint planning means schedulers are deluged with planned orders that exceed plant capacity. The monthly planning processes have been thrown out the window as largely irrelevant when such rapid change occurs. Despite the schedulers’ best efforts, many plants have found that their effective capacity has dropped significantly by 10%–15% or more. The scheduling puzzle changes so rapidly that expediency supersedes efficiency.

Business Imperatives  ◾  5

What planners and schedulers need most today is a way to defend against this chaos: how to plan and replan to quickly schedule the plant to best meet the changing circumstances in the most efficient way possible. The challenge is to package efficient campaigns that best execute the company’s strategies despite all the craziness that the world is throwing at us – and to scale this across all plants, warehouses, and transportation. Enterprise planning and scheduling have assumed a new level of strategic importance for supply chain leaders.

Scheduling Is Even More Important Many of a business’s strategic objectives are tied to how well manufacturing performs. Manufacturing strategy is the connective tissue that balances customer service performance, effective use of capital assets, and working capital. This is not an easy needle to thread. In plants with a complex Stock Keeping Unit (SKU) portfolio, with numerous lines and equipment types, figuring out the best schedule can be deceptively challenging. This is particularly the case in process manufacturing operations like food, beverage, CPG, batch chemicals, pharma, and biotech. Production scheduling is often the weakest link in the entire customer satisfaction process. It is rarely given the resources, attention, and funding required to succeed. Companies spend millions of dollars on higher-level planning improvements but neglect the last step in the planning chain – scheduling. This is reflected by the fact that Excel is still the most widely used scheduling software. Later chapters will explore what good scheduling looks like. In a nutshell, a good schedule is one that has the agility and flexibility to meet disruption, is accepted and supported by the production organization, can be executed, minimizes total delivered costs, and meets the customer’s quantities and due dates. Furthermore, we mustn’t forget the human side: We’re not dealing with lights-out plants. The best equipment in the world won’t perform if the shop floor teams aren’t motivated and capable of operating it effectively. Team members intuitively understand good schedules. They are closest to the equipment and its foibles. They become frustrated when schedules dictate line changeovers between products that are patently “wrong,” resulting in wasted time, effort, and resources – particularly when it impacts their Friday nights and weekends. Production scheduling determines when materials are consumed and produced. Execution of the schedule determines the flow of materials through the supply chain. Effective scheduling strategies will naturally create a cadence or rhythm driven by the best sequence of making products that the entire supply chain will echo, improving supply predictability.

6  ◾ Introduction

Ad hoc or poor scheduling results in variability that ripples through the supply chain upstream and downstream, reducing supply stability, and making operations more difficult to manage. Time after time, we have seen evidence of the opportunities that lie in good scheduling. Tangible improvements abound, like up to a 30% increase in throughput, which translates to the top line on oversold lines. In both oversold and undersold lines, throughput increases translate to significant improvements in the cost of goods sold and working capital. Good scheduling usually translates into improved morale and quality of life for employees when chaos is replaced with a more stable work environment.

Scheduling Is a Foundation of Manufacturing Performance Ten years ago, two of the authors supported a lean initiative in a well-known nutraceutical company. It was very hard work. The plant was part of an earlier acquisition (described in Chapter 29). It ran poorly. Even though it was a relatively simple packaging operation, Overall Operating Efficiency, OEE, was in the low-thirty percent range. A typical day involved multiple schedule changes due to rush orders, a shortage of the raw materials needed for the day’s schedule, or equipment failures. They could not maintain the necessary inventory levels to insulate the plant from the inevitable external disruptions. Overtime was the norm, often running well into the weekend. Labor turnover was high, and there was never any time to do proactive maintenance, further exacerbating an already chaotic situation. The leadership team was exhausted, frequently being called into the plant multiple times a night. Implementing lean in this situation sounds like a sensible plan. The reality was that it pushed the entire operation to a breaking point. Training sessions kept people in the plant before or after their shifts. It overloaded the leaders even more. It was not working. Pete King, one of the authors of this book, was part of our team. He recommended we implement a structured, repetitive scheduling process, on the basis that it would increase plant throughput significantly. The VP of planning agreed that this was an excellent next step. The lean work was slowed while the scheduling project was done. The new process was in place a few months later, and the results were almost immediate. The shop floor environment stabilized, and throughput improved. Initially, it was difficult to fend off the commercial team, who struggled to understand that they were part of the problem, constantly injecting new demands. As the inventory levels built up, service levels noticeably improved. The sales team realized the benefits of respecting the frozen window, although old habits were hard to break. Maintenance windows, including deep cleans, were included in the schedule to compensate for years of neglect. With OEE rising quickly into the 40s and then 50s, overtime and late-night callouts gradually reduced. The operational

Business Imperatives  ◾  7

excellence team got a new lease on life, and their efforts started to compound the benefits of a more disciplined, structured scheduling process. Within a year, overtime was eliminated, and two lines were no longer in use: Overall throughput increased by about 35%, triggered by the scheduling improvements. Three years later, OEE was in the 80%–90% range. While this was taking place it became a much more enjoyable place to work for all. Associates were able to know what their day would look like; it was predictable and growing and not chaotic anymore. Management time moved from fire-fighting to planning and strategy to unlock new areas of productivity. Our high-performance teams started to take off as we could keep the teams stable and have the time to work with them. The numbers told a great story – Service levels rose significantly. Capital needs for new lines for existing products were reduced to zero. Downtime dropped dramatically. Production costs were reduced by $1.5 Million per year. (Dean Bordner, SVP of Operations)

Why Now? COVID has created an even more challenging time. During this period, a frozen meals manufacturer lost 13 points of OEE, which amounted to 16% of plant output when they needed it most, due to unplanned changeovers driven by labor challenges, late deliveries, and outright unavailability of ingredients. They could not fully capitalize on the demand spike triggered by people being forced to eat at home. Many customers tell us that they don’t have the time or priority to address planning and scheduling now and that some big project is coming. Here is a typical email response when we suggest they sort out scheduling: Had a long conversation with my boss on what direction we are heading as a company, and if she sees any need for the systems you provide. Her thoughts were that the implementation of Enterprise Resource Planning (ERP) software will take three to four years and we don’t have the proper support or depth to try and handle anything else in this time frame. Let’s look at some of the problems with this line of thinking: ◾ The typical plant scheduling improvement project will improve manufacturing efficiency by 5%–30% and can be accomplished in three to four months, with immediate payback.

8  ◾ Introduction

◾ Typically, a third of a planner’s time is freed up, which can be used to support the longer-term ERP rollout. Therefore, the up-front investment in scheduling actually creates time in the scheduler’s busy day to dedicate to the ERP project – repaying that investment before the ERP rollout is complete. ◾ As described above, the entire plant team will experience the benefits of good scheduling and inventory practices so that they can apply the necessary share-of-mind to other improvement work like an ERP project or lean. ◾ The discipline imposed by the planning and scheduling project is a prerequisite for a successful ERP implementation. ◾ All too often, scheduling is an afterthought of a large ERP project: If it is not addressed upfront, it will need to be tackled post-project – but you will have lost four years of a 5%–30% efficiency improvement. That could pay a large portion of an ERP implementation! I think of a simple analogy: If you set out to paint your house, move some walls, and transform its interior, it pays to eliminate the inevitable clutter built up over the years beforehand. Move furniture out of the way. Throw out the old stuff. That’s what good scheduling does – it eliminates the clutter, creates greater visibility, and puts in place a disciplined process that enables you to make the best use of your physical, human, and financial assets, all else being equal. Of course, not all scheduling approaches will yield the same benefits. And there is no silver bullet. This book will help you understand how to access the hidden value that most process manufacturing operations have where an agile, optimal scheduling process has not been established.

Chapter 2

Characteristics of Process Operations – And Scheduling Challenges Manufacturing processes can be categorized into two broad groups: discrete parts assembly manufacturing and process industry manufacturing. Assembly manufacturing generally consists of the manufacture of individual parts and components that operators and machines then weld, bolt, or otherwise fasten together into a finished product. Examples include automobiles, aircraft, motorcycles, cell phones, computers, power tools, television sets, and hair dryers. Process industries are characterized by operations that include chemical reactions, mixing, blending, extrusion, sheet forming, slitting, baking, and annealing. Process companies sell finished products in solid form packaged as rolls, spools, sheets, or tubes; or in powder, pellet, or liquid form in containers ranging from bottles and buckets to tank cars and railcars. Examples include automotive and house paints, processed foods and beverages, paper goods, plastic packaging films, fibers, carpets, glass, and ceramics. Process industry output may be sold as consumer products (food and beverages, cosmetics, pharmaceuticals) or as ingredients or components for other manufacturing processes. A key difference between the two is that the number of different part types converges as material flows through an assembly operation, while the product variety increases as material flows through a process operation. That is, assembly manufacturing starts with a large number of components and ends with a small number of finished product Stock Keeping Units (SKUs), while process operations are the opposite; few raw materials become highly differentiated as material flows through the process, ending with a large number of finished SKUs. That and other differences have very significant implications for how the operations are scheduled. DOI: 10.4324/9781003304067-3

9

10  ◾ Introduction

Changeover Difficulty One of the most significant differences in process operations is that changeovers are generally more time-consuming and often more difficult. In the assembly industries, product changes often involve changing tooling and then adjusting or calibrating the machine with the new tooling. In process operations, changeovers can be more complex and time-consuming and frequently involve material losses. In food production, products often contain allergens such as dairy or peanuts. Where volume warrants it, dedicated lines can be the answer, but in many cases, this is not economically justified. Consequently, plants must do extensive clean-ups after running the allergen-based products, with complex decontamination processes and testing to ensure a contamination-free environment afterward. Organic and kosher requirements also require extensive clean-ups. In pharma and nutraceuticals, in addition to allergens, bottle sizes and cap types can take time to change. And setting up a desiccant machine for a different bottle size can be a tricky operation. Batch chemical process equipment is typically large and designed to be general-purpose, to be run under a wide range of conditions and settings to produce a variety of grades. Thus setups or changeovers tend to take much longer than typical setups in assembly plants, to allow time for vessel cleanout and for the process equipment to reach the new temperature or pressure and stabilize. Plants producing plastic film products generally make a number of different widths, and the mechanical apparatus that stretches the film in the transverse direction takes significant time to adjust for a different width. All of these factors tend to encourage plants to run a large campaign on the current material before switching to the next. The scheduling process can reduce the desire for long campaigns by sequencing products so that the changeovers are as simple as possible.

Starting Up after a Changeover In cases where a number of mechanical adjustments were made during the changeover, such as a vitamin tablet bottling line or a food packaging line, several minor stops can occur soon after restart if the adjustments weren’t made precisely. It can take several attempts at fine-tuning before the line operates smoothly. Thus what are called “minor stops” can collectively steal a lot of capacity. In the synthetic rubber industry, process capability is sometimes poor enough to require significant production time to get properties such as viscosity back within specifications. And it may require significant testing laboratory time and facilities to determine when properties are finally on aim.

Characteristics of Process Operations – And Scheduling Challenges  ◾  11

Both changeover difficulty and restart difficulty can be significant enough to have a strong influence on scheduling, and any strategy should try to minimize these difficulties by sequencing products as similar to the previous product as possible.

Sanitation Cycles Some plants run regular sanitation cycles, such as the third shift every night, or on Saturday and Sunday on alternate weekends. Any scheduling strategy must recognize and accommodate these requirements.

Shelf-Life Constraints In some process operations, especially food production, shelf life is an issue that must be considered when determining production frequencies and campaign sizes. It is typical for products like baby food, salad dressings, tomato ketchup, and fruit juices, to have a shelf life of 12 months or less. In these situations, the retailer may want nine to ten months of that window; they don’t want to put a product on the shelf which is due to expire in a month or two. The supply chain (warehousing, logistics, distribution centers, etc.) may consume a month or more, leaving the plant with only a month or so. Therefore, every product must be made every month at a minimum, where other considerations like changeover time and difficulty might suggest less frequent production. For baked and fried foods, like donuts and potato chips, the on-plant shelf life will typically be much shorter, typically only a week or two. To further limit the time available for manufacture, variable demand increases the risk of products that must be scrapped or donated because the plant has exceeded its portion of the allowable shelf life. These are not generally concerns in assembly operations; the components don’t usually deteriorate over reasonable periods of time. What some of these operations do face is product obsolescence, particularly in very dynamic industries like consumer electronics.

Multi-Step Manufacturing A large number of process plants have two or more steps in series that need to be considered when deciding on a scheduling strategy. If there is a oneto-one relationship between steps, i.e., there is a straight-line flow from one step to the next, there is no added scheduling complexity; the same schedule can manage all steps in series. A line in a donut plant is a good example of this, where dough mixing, baking, cream injection, frosting, and applying

12  ◾ Introduction

sprinkles all happen in one closely coupled line, with no holdup or inventory other than that stored for seconds on an accumulating conveyor. Thus the same schedule controls the entire line. There are, however, three other equipment configurations that require a little more thought: 1) Several pieces of equipment at each step, with crossing flow lines where an asset in step 1 can feed a number of assets in step 2, with very little intermediate inventory to buffer separate schedules at each step. 2) A similar footprint, but with significant in-process inventory to allow a high degree of independent scheduling. 3) A similar footprint, with some degree of in-process inventory, thus somewhere in between cases 1 and 2. Figure 2.1 illustrates case 1 and depicts the manufacture of vitamin tablets. The three major steps are blending active ingredients with excipients, pressing the powder into tablets, and coating. There are several pieces of equipment at each step, and any blender can feed any press, each of which can feed any coater. There are only a few hours of inventory between steps, so the scheduling methodology must carefully coordinate the schedules for each area. Figure 2.2 illustrates the second case, a process making very short polyester fiber filaments used to fill comforters and ski jackets. The key steps are mixing ingredients and cooking them at high temperature and pressure to form a viscous polymer, spinning the polymer into fiber strands, stretching and annealing the fibers, then cutting them into short pieces and baling them in packaging that resembles cotton bales. There are intermediate inventories with enough storage to de-couple the work centers so each piece of equipment can be scheduled separately. Figure 2.3 illustrates the third case, a salad dressing plant. The major steps are mixing and packaging. The dressings can be directly conveyed from the mix tanks to the packing lines or offloaded into portable stainless steel totes when necessary. There are two days of storage capacity available, thus providing some degree of de-coupling of mixing and packaging, so that their schedules can be somewhat independent, but a degree of coordination is required if both operating areas are to be scheduled for minimum changeover loss and maximum throughput.

Balancing Limited Resources Many process plants have resources that are needed in different amounts for different products and often don’t have a sufficient supply that all combinations of products can be made at the same time. A plant making cheese sauces for fast food taco restaurants is limited in the amount of whey available at any time, so they have to schedule so that all lines are not consuming whey at the same time.

SMALL MOBILE BLENDERS (5) WIP B

(BLEND STORAGE IN SUPERSACKS)

Figure 2.1  Major steps in vitamin tablet manufacturing.

BLENDING QUEUE

LARGE STATIONARY BLENDERS (6)

HIGH POWERED PRESSES (3)

HIGH – SPEED PRESSES (7)

SUPER SACKS

WIP C

COATING QUEUE

C5, C6 ACELA 48" COATERS

C4 72" VECTOR COATER

C1, C2, C3 VECTOR COATERS (3)

SUPERSACK LOADING

Characteristics of Process Operations – And Scheduling Challenges  ◾  13

FLAKE SILOS (12)

Figure 2.2  Major steps in staple fiber manufacturing.

RAIL CARS

POLYMERIZATION (2)

FIBER SPINNING (6) FILAMENT TUBS

(4)

DRAW STEAM ANNEAL CUTTER BOXES

(3)

CUT – BALE

BALES

14  ◾ Introduction

REFRIGERATED WIP STORAGE

DRY WIP STORAGE

MIXING FORMULA PREP

LARGE MIX TANKS 600 GAL (4)

SMALL MIX TANKS 300 GAL (4)

Figure 2.3  Major steps in salad dressing production.

• Batching • Weighing

LIQUID PREP KITCHEN

• Blending • Batching • Weighing

SPICE ROOM

CAPACITY = APPROX 2 DAYS

PORTABLE TOTES (120) 300 Gal ea

PACKAGING

RETAIL BOTTLES

PACKET FILLING LINES

JARS & TUBS FILLING LINES

REFRIGERATED FG STORAGE

DRY FG STORGE

SHIPPING

Characteristics of Process Operations – And Scheduling Challenges  ◾  15

16  ◾ Introduction

A plant making frozen dinners has a large number of meals containing noodles and has limited noodle-making capacity on site. Because purchased noodles are more expensive, the schedule must limit the number of lines making noodle-containing meals on any day. The salad dressing plant described in Chapter 5 has a reverse osmosis system feeding filtered water to the dressing mixing vessels but doesn’t have enough capacity to feed all possible recipes that could be made at any time. In a detergent plant, different packaging sizes used different filling nozzles, and there were a fixed number of nozzles for each size. Therefore, the schedule could not require too many nozzles of each size at the same time. The same is often true of labor; different products require different numbers of people to staff the line. Similar to production and labor constraints, there are often changeover equipment constraints. At a potato chip plant, there was a limited number of washout stations necessary for changing flavors. Flavor changeovers had to be staggered based on washout system availability. Specialized tooling or mechanical expertise may also be a constraint. Thus any scheduling process must recognize these limitations and set schedules to level out the need for limited resources.

Divergence vs Convergence One of the more significant differences between assembly plants and process industry plants is that the flow patterns and dynamics are often the opposite of each other. As described in Umble and Srikanth’s Synchronous Manufacturing, the predominant flow characteristic in an assembly plant is the convergence of part types, whereas the flow in process plants is often characterized by product type divergence. Figure 2.4 represents product type convergence in a typical assembly plant, with material flow from bottom to top. At the start, there may be hundreds, thousands, or tens of thousands of individual screws, nuts, bolts, springs, sheets of plastic and sheet metal, and so on. As these parts are processed and assembled into subassemblies, then subsystems, then complete systems, and finally into the finished product, the number of different part types diminishes dramatically. That is, there is a significant convergence of parts into assemblies and then systems into the final product. Considering the manufacture of Toyota Camrys, for example, there may be tens of thousands of part types at the start, while the final product comprises two trim lines, in several colors, for a total end item variety of perhaps two dozen or so. The flow pattern for assembly plants has been called an “A” type process because it resembles the letter A. Process industry operations often follow the opposite flow pattern, sometimes called a “V” type process as shown in Figure 2.5. The process starts with a few raw materials, which may be mixed, reacted, and then cast or

Characteristics of Process Operations – And Scheduling Challenges  ◾  17

Finished Products Part nos. or SKU’s

Material Flow

Systems

Subsystems Assemblies Subassemblies Parts

Figure 2.4  Schematic diagram of an “A” type process.

Material Flow

Finished Products

Raw Materials

Figure 2.5  Schematic diagram of a “V” type process.

18  ◾ Introduction

extruded as fibers, sheets, or pellets, and then further processed to create tremendous final product variety. In the manufacture of nylon yarns for apparel or seat belts, the process starts with a few raw materials: adipic acid, diamine, TiO2, and demineralized water. These are mixed, polymerized to create a highly viscous molten plastic, and then extruded as groups of extremely fine fibers, which can be of different thicknesses. They can then be stretched at different ratios to build a desired tenacity level, annealed to permanently set the properties, dyed to fit the particular end use, and wound on one of a wide variety of rolls or spools. What started as four primary ingredients ends as hundreds or thousands of end item SKUs. The predominant flow pattern is one of divergence. Similarly, in the production of tortilla chips, there may be relatively few basic chip types, which are then coated with different combinations of spices and seasonings, packaged in different size bags, and then packed into different size cartons. So a single type of corn may result in several dozen final packaged SKUs.

Examples of “V” Type Process in Process Plants As a specific example of a “V” type process, consider a sheet goods manufacturing line (Figure 2.6). Six different grades of raw materials, in the form of

PACKAGED ROLLS = 2000 TYPES

CUT ROLLS = 1800 TYPES

SLIT ROLLS = 1000 TYPES BONDED ROLLS 200 TYPES MASTER ROLLS = 50 TYPES

RAW MATERIALS = 6 TYPES Figure 2.6  SKU fan out in the sheet goods example.

Characteristics of Process Operations – And Scheduling Challenges  ◾  19

small polymeric plastic pellets, are stored in input silos. Pellets are pneumatically conveyed to one of four forming machines, melted, and then formed into a wide sheet and rolled up on a “master roll.” There can be 50 different types of master roll, with the differentiation based on pellet type, sheet width, and formed thickness. The master rolls are then unwound on machines that heat the sheet and compress it into the final end-use thickness creating 200 types of bonded rolls. The rolls are then unwound and passed over a rotating knife-cutting apparatus, where 1,000 different types of slit rolls can be created from the 200 bonded rolls. The slit rolls are moved to one of three choppers, which cut the sheet to the desired length. Different chopping lengths generate 1,800 chopped roll varieties. Thus in this process flow, we have significant product differentiation at each major step: at sheet forming (6 → 50), at bonding (50 → 200), at slitting (200 → 1,000), at chopping (1,000 → 1,800), and at finishing (1,800 → 2,000). As you can see in Figure 2.6, we have transformed six types of raw materials into 2,000 final products. This high level of divergence, of product differentiation, is often found in process plants. In the manufacture of pharmaceutical and nutraceutical products, one or more active ingredients are blended with various excipients to form blends with different activity levels or strengths and then pressed into tablets, filled into capsules, or coated with gelatin to form soft gels. Thus a single active can be transformed into several dose types and strengths. These doses are then packaged in bottles of various sizes and colors, with standard caps, flip caps, or childproof caps. Some may contain cotton and/or desiccants, others not. And the same bottle may be labeled differently for different retailers and different countries of sale. So any single active formula can result in perhaps a hundred packaged SKUs.

Product Differentiation Points The presence of a number of product differentiation points, where a single material can be transformed into one of several varieties, is at the core of the difference in “A” type and “V” type processes. Scheduling is critical at each differentiation point; if you make the wrong differentiation decisions, it can profoundly affect your manufacturing system performance. Your incorrect decision will produce a currently unneeded variety, which flows to finished product inventory, filling the warehouse with currently unneeded products. You have also consumed valuable production capacity, which is then unavailable to make needed products. So you end up with excessive finished product inventory, but ironically with poor customer service.

20  ◾ Introduction

A plant making salad dressings can be vulnerable to poor differentiation decisions. If a batch of a particular flavor is packaged in currently unneeded bottle sizes, they will sit in inventory for some time, while needs for that dressing in other bottle sizes may go unmet. Further, if that dressing sits in inventory for some time, it may then exceed shelf-life limits and have to be scrapped. In the fiber industry, in the elongation (“drawing”) of fibers used in industrial ropes and seat belts, scheduling decisions must be made about products made at different temperatures and draw ratios, which affect end properties like tenacity and modulus. Again, you have the opportunity to make products that are inappropriate to the current customer needs. There is complexity in both discrete parts assembly operations and process operations, but it manifests itself at different ends of the operation. Because discrete parts assembly starts with hundreds or thousands of parts, scheduling the arrival and flow of each of those parts must be done very carefully and precisely. MRP-based systems are generally capable of doing this. Process operations have the opposite situation, stemming from the difference between the A and V flow patterns. Managing the arrivals and quantities of raw materials for a process operation is not trivial by any means, but it is inherently simpler than for an assembly operation, because of the relatively few types of raw materials involved. The scheduling complexity begins to appear as material moves through the operation, and step after step involves deciding which of the several or many output varieties should be made, as described above. MRP-based systems can deal with this, but will typically generate what appear to be random sequences of products, with little emphasis on simplifying changeovers and increasing throughput (Figure 2.7).

Limited Extra Capacity Most process plants run a 24 by 5 or a 24 by 7 weekly schedule, so there is less opportunity to make up for poor schedules with overtime.

Summary Of all the differences between assembly manufacturing and process operations, perhaps the one with the most implications for scheduling is the number of product characteristics that may require some equipment change or adjustment on any changeover and the capacity lost during that time. The organic requirements and allergen cleaning found in food and vitamin tablet plants don’t have any corollary in parts assembly. Similarly, the cooling down, heating up, and getting properties on aim found in batch chemicals aren’t generally seen in assembly operations.

Characteristics of Process Operations – And Scheduling Challenges  ◾  21

Manage product differentiation steps

Manage the Bill of Materials Manage Lead Times

NUMBER OF PART TYPES OR MATERIAL TYPES

HUNDREDS OR THOUSANDS

PA RT S

NS

AS

SEM

BLY

SS CE

IO AT ER

OP

O

PR

FEWER THAN 20 RAW MATERIALS

PROCESS FLOW

FINISHED GOODS

Figure 2.7  Parts assembly and process operations – where the complexity lies.

The limited on-plant shelf life and how it influences scheduling is rarely seen in parts assembly. The existence of multi-step operations, with little in-process inventory, requires that the scheduling process coordinates the sequential steps to ensure the required flow can be achieved. Finally, the diverging product mix requires the scheduling strategy and software to make the right differentiation decisions; otherwise, unneeded inventory is created and capacity is lost. How scheduling strategies, processes, and software can manage these complexities will be covered in the next chapters.

Chapter 3

Overview of Production Strategies A good schedule is one that has the agility and flexibility to meet disruption, is accepted and supported by the production organization, can be executed, minimizes total delivered costs, and meets the customer’s quantities and due dates. The production strategy is the foundation of scheduling. It’s impossible to create a good schedule if the production strategy can’t be executed in practice or was not designed to properly balance costs and customer service. Many companies, large and small, have found that a well-designed, structured, repetitive scheduling strategy that focuses attention on throughput and operating efficiency can yield financial benefits over the methodologies typically used in ERP systems. These repetitive strategies don’t sacrifice customer delivery performance nor require excessive inventories; in fact, they right-size the inventory for each SKU to meet whatever fill rate targets the business sets. These strategies set the optimum lot size and frequency of production based on changeover cost, inventory carrying cost, minimum lot sizes, shelf life, and whatever other criteria the business feels are important. They then define a repetitive product sequence that minimizes changeover difficulty and time. The results are higher throughput, which increases revenue in oversold situations and reduces manufacturing cost/unit in all cases. These strategies are particularly applicable to the process industries because of the changeover difficulty and difficulty re-starting after changeovers described in Chapter 2; these difficulties make process operations unsuitable for dispatching methods relying only on start dates or due dates. Almost every planner can tell you that there is a best sequence in which to make their products and preferred lines for each product. In a sense, repetitive scheduling strategies are a way of determining the optimum sequence upfront, so that the effort doesn’t have to be repeated every cycle. DOI: 10.4324/9781003304067-4

23

24  ◾ Introduction

One of the most widely used repetitive scheduling strategies is Product Wheels, described in The Product Wheel Handbook (Productivity Press, 2013), by Peter and Jennifer King. Product Wheel design decides on the best frequency for each SKU to be produced, using calculations based on the classic EPQ (Economic Production Quantity) calculations, sometimes augmented by advanced analytics based on heuristics developed by noted statisticians Silver and Meal (Production and Inventory Management, 1973) and Jackson, Maxwell, and Muckstadt (IIE Transactions 17). Their techniques consider that products with similar characteristics must run together, and be scheduled as families, not as individual products. This sets the basic wheel cycle time and supports decisions on which products should be made every cycle, every second cycle, every fourth cycle, etc. Finally, each cycle is sequenced to minimize changeover time and difficulty. Figure 3.1 shows the two cycles for a vitamin tablet packaging line, where high-volume products like 55402 and 55414 are made every cycle, and lowvolume products like 54932 and 59508 are made on alternate cycles. Product wheel scheduling has been used extensively by large global enterprises like Dow Chemical and DuPont, and smaller companies like Appvion (formerly Appleton Paper) and BG Products. A similar strategy is called Rhythm Wheels as described by Josef Packowski in Lean Supply Chain Planning (Productivity Press, 2013). Pharmaceutical companies like AstraZeneca, Novartis, and Eli Lilly have employed this brand of the strategy. Repetitive Flexible Supply, as highlighted in Ian Glenday’s books, Lean RfS (Productivity Press, 2013) and Breaking Through to Flow (Lean Enterprise Academy Ltd, 2006), is a different concept but built on some of the same foundational concepts and with more similarities than differences. Fixed Sequence Variable Volume, described by Ray Floyd in Liquid Lean (Productivity Press, 2010), is very similar, but with key differences necessitated by the unique challenges of Floyd’s Exxon Mobil plant in Bayport, TX. Line 3 Cycle 1 54935 59132

54932

Line 3 Cycle 2

55675

55336 55327

54933

59596

55675

55327 55325

55325 56808

59108

65675

59058 55326

55414 59596

55402 59108 55402

Figure 3.1  A two-cycle product wheel for a vitamin tablet line.

55414

Overview of Production Strategies  ◾  25

What all of these have in common is a structured sequence, repeated cycle after cycle, that minimizes changeover time and losses and thus reduces cost and increases useful capacity. And it creates a learning curve effect, further increasing efficiency. While a number of companies are using this type of scheduling to great benefit, many are not. Some are aware of the concepts, and are trying to follow them, but with limited success. The most frequent stumbling block is that they try to ease into a structured pattern one product at a time, or one weekly schedule at a time. Without a comprehensive design process that takes all lines and all products into account, it’s difficult to make any structured strategy work for very long. Others start with good intentions but don’t have sufficient management commitment and support to work through the temporary difficulties that any significant transition faces. Any of these versions of a regularly repeating, predictable, well-sequenced scheduling methodology are preferable to MRP-type scheduling strategies that generally meet the due date and inventory targets, but at the sacrifice of operating performance. But to be effective, they must be encoded into the planning and scheduling system to achieve their full benefit. The master data in the ERP or planning and scheduling system must be synchronized with the repetitive schedule’s design so that the system will recommend orders with the appropriate frequencies and quantities to meet efficiency, customer service, and inventory goals. Otherwise, the scheduler will be left with the difficult task of adjusting the timing and quantity of every order to fit with the scheduling strategy and inventory targets. Although these concepts are designed in a very structured way and intended to be followed from cycle to cycle, they don’t have to be rigid and inflexible. If designed and executed properly, they can be dynamic and agile, reacting to current conditions as required to meet business objectives. The design process should include a collaborative discussion on all likely disruptions and the most appropriate countermeasures for each situation. Chapter 7 goes into more detail on dealing with disruption. Chapter 6 gives more detail on the similarities and differences of the various repetitive strategies, the advantages and disadvantages of each, how these strategies operate, how the benefits accrue, and key design steps.

Chapter 4

Scheduling Processes and Software A good schedule has the agility and flexibility to meet disruption, is accepted and supported by the production organization, can be executed, minimizes total delivered costs, and meets the customer’s quantities and due dates. This chapter will introduce the scheduling processes and software necessary to create a good schedule. They will be discussed in depth in later chapters. While scheduling is a short-term activity, a good schedule doesn’t just happen at the last minute: It requires a longer-term set of supporting processes to be successful. It may be helpful to think about the planning and scheduling process as one of continuous improvement as the schedule moves from the long-term horizon to execution on the shop floor. In the long term, many of the inputs to planning and scheduling are unknown and will change by the time they are executed. In the short term, equipment must be scheduled and sequenced, materials must be available, and staffing and other resources must be committed. In between, the task is to understand the customer’s requirements and match them with production capabilities, materials, and resources so that the detailed schedule can be reliably executed, achieves business priorities, and meets the customer’s expectations. Depending on the methodology one subscribes to, the long term could be called Sales & Operation Planning (S&OP), Integrated Planning (IBP), Business Planning, or something else. The search for precision and accuracy in this horizon is pointless since conditions and assumptions will change by the time the plans are executed. Instead, the goal is to look at direction and ranges, with the precision necessary to make business judgments, develop what-if scenarios, and decide which ones will be covered by the authorized production plans.

DOI: 10.4324/9781003304067-5

27

28  ◾ Introduction

Production Planning For this book, we will call the next zone production planning. The goal is to create a production plan with enough detail and precision to secure the materials, line capacity, people, and other resources necessary to execute it. We’ve also heard it called volume planning, capacity planning, rough cut planning, master production scheduling, and Sales & Operational Execution (S&OE). It may be helpful to make a distinction between planning and scheduling. Planning is deciding which products to run, and in what quantities, during each time period. Scheduling is sequencing the products for each time period, in a way that setup costs are minimized, without running out of stock or missing customer due dates. It’s critical for the planner to manage the handoff or transition of the production plan to the detailed scheduling system. At the point of transition, the plan should meet the following requirements: ◾ It should require no more than the available capacity, resources, and components, after taking all losses, such as setups, maintenance, team meetings, process improvement time, or scrap, into account. ◾ It uses realistic or demonstrated production rates for every product in the plan. ◾ All products required to be produced must be included, in the correct quantities, and in the correct time period. ◾ The agreed-upon demand from all sources is included. ◾ Inventory is at target levels at the end of each planning period. ◾ When sequential production levels are involved, the plan meets the lot size criteria at each level. The first four points are more or less self-evident, but the last two are worth a short discussion now. Inventory at target levels ensures that uncertainty is buffered. A plan that shortcuts inventory brings more risk that the schedules to be developed in the next step won’t be executed as planned. Production efficiency is compromised once schedules are disrupted, which brings more risk of further disruption, increased costs, supplier issues, and a vicious circle ensues. When a plan can’t be created that meets inventory targets, it’s time to manage demand or increase capacity. We will discuss this in Chapter 9 on inventory targeting. Lot size criteria at each level may not be obvious to those not used to multi-level planning and scheduling. In the example of the Blue Lakes salad dressing plant that we will present in the next chapter, it can be seen from the value stream map that they create their mix formulas in 300- and 600-gallon tanks. A mixing run will serve many different packaging runs. To minimize cleaning, the total requirements of the orders for jars, tubs, and

Scheduling Processes and Software  ◾  29

bottles should completely use up the mix run. Any mix we don’t use will be scrapped to free up the tank for the next mix, and it would age and go bad anyway if not used quickly. The jar, tub, and bottle runs each have their own lot size criteria. Ideally, they should be in full pallet increments, and at a minimum, full pallet layers to avoid damage and excessive handling. Therefore, each packaging run must meet customer requirements for the period, rounded to multiples of pallets or pallet layers, and the total of all the dependent requirements for each mix must round to 300 or 600 gallons. It’s quite a complex calculation, and typical ERP software can’t make the connection that related packaging items on different production lines should run together at the same time and that the requirements of their components must round to the upstream mix multiples. Sometimes, a distinction has been made between production planning and capacity planning based on older software and techniques, but we believe that this distinction is no longer relevant. The production plan must be capacity, component, and resource feasible in order to be executed. Further, if an unconstrained plan is used to drive component procurement, it will result in over-ordering component materials, or ordering them for the wrong production period. This is a typical issue when using unconstrained ERP/MRP systems to create the production plan, and we will discuss it in more detail in the ERP systems chapter (Chapter 13). There may be other constraints besides production capacity. For example, the component materials must be available, the staffing for the lines must be available, and often other resources are necessary. For example, we’ve seen cases where the conveyors leading out of the production room were unable to handle certain mixes of production. Schedules can often be constrained by the ability to change over between products. For example, there may be a limited number of washout facilities, or there may be a special set-up crew. Too many changeovers that require the washout or crew can’t be scheduled at once. With a good production plan, the scheduling task is relatively straightforward and becomes one of choosing the best timing and sequence for the planned orders to make them feasible to execute on the shop floor and maximize customer service and production efficiency. Without a good plan, no amount of scheduling effort will make the schedule fit. Even the most advanced and expensive advanced planning and scheduling system will be unable to create a schedule that satisfies customer demand and is feasible to execute. We will give an example of this in Chapter 21 when we talk about production planning and capacity management in more detail. Business planning usually works in time buckets of months, quarters, and years. Production planning moves into weekly and daily buckets. In scheduling, the timeline is continuous: Some activities may need resolution to the hour and minute level (Figure 4.1).

30  ◾ Introduction

IBP, S&OP, Business Planning Months, Quarters, Years

Production Planning Days, Weeks, Months

Scheduling Minutes, Hours, Days

Figure 4.1  Planning and scheduling zones and time periods.

The best planning programs will have flexible aggregation that starts at the daily level and can group by weeks, months, quarters, years, and other custom periods as needed. Even when the evaluation is done over longer periods, it’s important that the building blocks of planning start in days. Calculating time offsets in a weekly bucketed period is difficult: A four-day lead time moves the receipt to the following week if the start is on Friday, but the receipt is in the same week if the start is on Monday. If the lowest unit of time is not days, finding a common denominator between periods is difficult. Months don’t end evenly at the end of weeks, and the result is all kinds of odd accommodations such as technical periods or a mix of four- and five-week months. To minimize the effort in planning and scheduling, a principle to follow is that it’s usually easier to solve problems at the planning level than it is at the scheduling level. In detailed scheduling, attempting to add orders to an already full schedule can be overwhelming. Something will no longer fit, but which order or orders should be removed to make room? Ideally, the scheduler should cut products in the best inventory position or those with the lowest margin or lowest business priority. Perhaps several orders could be slightly reduced in size to make room without materially affecting service. Examining all the possibilities in scheduling is difficult, and because it’s difficult, we may accept the first option that works. The detailed scheduling stage is also too late. We may disappoint customers at the last minute when we realize that we can’t meet their requests. At the detailed level, we are more likely to go down the wrong path. By waiting to sort out our priorities during detailed scheduling, we may have already wasted capacity on less important products or customers before we realized that we had a problem. By planning ahead, we will be able to make tradeoffs, produce more efficiently, satisfy more customers, and reduce our costs. If we move the analysis up a level to planning, a well-constructed planning program will show the capacity requirements and inventory positions of each product against availability and inventory targets, allow evaluation of resource and component constraints, and make it easy to adjust the planned quantities

Scheduling Processes and Software  ◾  31

and see the overall impact. Once a feasible plan is developed, scheduling is relatively easy because the required quantities in each period will fit. Planning is a bucketed activity by period, whether days, weeks, or months. It captures the work going on within the bucket, and the situation at the beginning and end, but not the details of what is happening within the bucket. It’s mostly simple math. As one of our old bosses used to say, “add the P (for production) and subtract the S (for shipments).” Most planning problems, even those with component, resource, and inventory constraints, can be solved by goal-seeking optimizations that are relatively simple to construct. The optimizations can be guided by priorities that match business objectives. For example, a typical set of optimization priorities might highly penalize inventories below zero or out of stock, moderately penalize inventories below a low inventory zone, and lightly penalize inventories above maximum. It would be required to respect capacity constraints and lot sizes at each level. Just these simple constraints can produce a viable plan that will be easy to manage in the detailed scheduling process. This allows the planner to concentrate their value-added effort on the upcoming periods, where the plan will be handed off to the detailed scheduling program. They can make sure that its planning parameters are correct and that it incorporates things like the most recent downtime plans.

Scheduling In contrast, scheduling is more complex, because it’s always dealing with the sequence, specific pieces of equipment, and the exact timing. The mathematical solutions to its problems are not so easy. Even a simple single-level schedule is a form of the traveling salesperson problem that has been studied since the late 1800s and is known to be difficult or impossible to solve. It attempts to find the shortest route between a list of cities. Scheduling a production site is even harder; there may be things like alternative production lines, coordination of multiple levels of production, intermediate inventory limits, tanks, flow path restrictions, and resource or component constraints. Reducing the size and scope of the scheduling problem through effective planning allows it to be solved faster and better, whether manually or by automation. Cellular flow or virtual cellular flow, as described in Chapter 25, can also reduce scheduling complexity in cases where the equipment arrangement facilitates it.

Supporting Processes Surrounding the planning and scheduling process are demand management, distribution planning, inventory management, and procurement.

32  ◾ Introduction

Demand management has several roles: to make sure that the demand from all sources is flowing to the planning and scheduling process, to deliver a reliable level of forecast error for each product code, and to manage demand when capacity, components, or resources are insufficient to satisfy the unconstrained demand. Looking beyond the plant into the distribution system, the planners and planning software start by working backward from the customer’s due dates and shipping points and consider the lead times for transportation, staging, and quality management to determine when production needs to be ready. Often, there are expedited transportation lanes that can be used at additional cost, and alternate shipping points that might have stock available when the customer can’t be supplied from their normal shipping point (Figure 4.2). This is often handled by Distribution Requirements Planning (DRP) software in a separate module; however, more sophisticated software can consider the end-to-end process with several advantages: maximizing the use of capacity and inventory, minimizing touches or handoffs between people, and trading off distribution and inventory costs for production efficiency. The purpose of inventory management is to ensure that the released or firmed production schedules can be consistently executed, given the likely level of forecast error and production schedule achievement. We will discuss inventory management in more detail in Chapter 9. Consistently achieving production schedules is important. It allows purchasing to procure the necessary materials in an orderly fashion, without requiring excessive inventory, since only the materials with lead times greater than the committed or firmed horizon will need to be buffered by inventory. The production lines will benefit from the stability of remaining in an efficient changeover sequence. Without reliable schedule achievement, every material must be buffered, since an order could be inserted into the schedule at any time. For everything inserted, something else must be delayed, and delayed orders mean that the components already inbound won’t be used until later.

Figure 4.2  DRP diagram.

Scheduling Processes and Software  ◾  33

There are several possible actions when schedules don’t hold up in practice. The zone of detailed scheduling could be reduced to a shorter period where there is a higher confidence that the schedule can be executed, demand projections could be improved, buffer inventories could be added, or customer service expectations could be changed. This isn’t to say that there aren’t cases where orders should be inserted or delayed in the firmed schedule. We will talk about this in Chapter 7 on dealing with disruption. But changes to the firm schedule should be an orderly process with rules. With a solid scheduling process, a proposal to insert or delay an order can be quickly evaluated for its impact on production efficiency and whether the necessary components can be procured. A value judgment can be made on the merits of accepting the change versus its costs.

Scheduling Software With these processes as a background, let’s briefly review the software to support them. Good scheduling software will help the planner to navigate through all these constraints and considerations in a fast and usable way. The critical things that the software must be able to show are the schedule, the production critical characteristics of each product, the changeover losses between each order, the capacity loading of each production line, the constraints, and the instantaneous inventory of the finished products and their intermediates. Excel is probably the most popular planning and scheduling software in the world. It’s available on almost every PC, and it can be used to model almost any horizon of the supply chain. Unfortunately, this ease of availability and flexibility comes with a cost. A simple example of the difficulty in modeling planning and scheduling in Excel is lead time offset, required for almost every production, distribution, or purchasing decision, for example, produced or received materials that must go through a microbial test that requires several days to complete. Or shipments to a distribution center that leave today and arrive several days later, depending on the mode of transit. Or purchased components with different lead times depending on the source. Think about the difficulty of writing this into the formulas and cells of an Excel workbook to calculate projected inventory. How do you write a lead time of four days into weekly bucketed cells? What if the lead times are not the same for all materials? What if the lead times change due to process improvements, or different vendors? Think of all the cells, formulas, and lookups that must be changed and validated. The authors have witnessed schedulers working all day in Excel to construct schedules that don’t hold up in practice when executed. In contrast, with capable specialized scheduling software, we usually see a 30% reduction

34  ◾ Introduction

in scheduling effort and a 5%–30% increase in production efficiency from schedules that are executable and efficient on the shop floor. While most ERP and MRP systems have a scheduling module, they typically lack the refinement and usability of the best specialized scheduling software. A typical issue is the lack of product attribute visibility when scheduling. Many programs will only display the product’s material code and description, or perhaps a few attributes. This leaves it to the scheduler’s memory to understand what a product is, and how it interacts with the equipment and the products that run before and after it. We typically find that it takes seven or eight product attributes to describe and understand the best sequence and equipment choices for a given product. ERP or high-level planning systems typically cannot provide these details. For example, at Blue Lakes, the scheduler must consider allergens, organic, religious rules, color, container size and shape, particulates, viscosity, labels, neckbands, and caps. When software relies on a scheduler’s memory for product characteristics and changeover implications, business continuity and vacation or absence coverage for the scheduler can be a problem. Few substitutes can fully understand another scheduler’s complex Excel program or adequately remember product characteristics when scheduling in the ERP/MRP system. Schedulers in this environment are typically overstressed. They are reluctant to take time off because they feel like they will be letting the business down, and they are always on call, even when on the beach. One of us has a daughter who was a scheduler for a water bottling plant in Orange County, CA. She and her husband went to LA one Sunday afternoon to see an Angels’ baseball game but had to leave in the fifth inning so she could go to the plant to resolve a scheduling issue. In contrast, specialized scheduling software allows the business to model its production process, product attributes, changeover times, and constraints in a logical way. The best graphical scheduling systems allow an immediate understanding of the schedule at a glance. In one view the scheduler can see capacity utilization, the sequence of product attributes, changeover time and losses, downtime, relationships of upstream and downstream schedules, resource usage, and alerts for component and inventory shortages.

Goal-Seeking Algorithms However, a watch out is that some detailed scheduling programs rely on goal-seeking algorithms or optimizations using KPIs (Key Performance Indicators) that must be carefully tuned to achieve business results. The skills required for testing and tuning the algorithms are higher-level skills, and the business must allocate the right people with sufficient time to maintain the scheduling program as conditions, equipment, and products change. We’ve

Scheduling Processes and Software  ◾  35

seen examples where the business didn’t do this, and consequently, the automation gave poor results, the schedulers simply accepted this as normal, and less-than-optimal schedules were built manually, using drag-and-drop scheduling. Even with proper tuning, it can be difficult to understand or explain how an automated system arrived at a particular schedule. Because of the difficult nature of the scheduling problem, it’s often impossible to find the absolute best schedule, and the scheduler can usually find examples in the automated sequence that can be improved. Therefore, it’s important to apply the law of diminishing returns. While it’s always possible to find a better schedule, how much effort and time is it worth? The goal is to quickly deliver an automated schedule that’s good enough to achieve most of the business results and be executable in practice.

Repetitive Scheduling This is where repetitive scheduling concepts can help. Every scheduler can understand that there are products that should be run together and that there is a best sequence in which to run their products. In a sense, repetitive scheduling is pre-computing an optimum sequence and providing the supporting conditions to reliably achieve it. With properly maintained scheduling software, we’ve seen examples where planners from other sites were able to step in at a moment’s notice, working remotely, to cover an unexpected absence or illness because they were trained in the same software, and could quickly understand the schedule, equipment, and constraints at another site.

The Scheduling Process The scheduling process itself falls into a few basic categories depending on whether it is a single- or a multi-level manufacturing process, and when there are multiple levels, the location of the constraint, and the amount of freedom or inventory between the levels. For a multi-level process, scheduling should start by scheduling the constraining operation first, in order to maximize its capacity. If there is little freedom or inventory between the levels, this may be sufficient, as all the levels must follow along together. However, when there is inventory and freedom between the levels, the scheduling methods diverge depending on the location of the constraint. Upstream is the direction towards component materials, ingredients, and suppliers. Downstream is the direction of finished products and customers. For downstream-constrained multi-level processes, once the downstream

36  ◾ Introduction

constraint is scheduled, a classic MRP approach will serve to translate the requirements upstream and synchronize the upstream production to the downstream constraint. For example, at Blue Lakes, during the times of the year when packaging is the constraint, they would create a best sequence for packaging, and the upstream mixing tanks would follow along to create any mix necessary (Figure 4.3). Upstream-constrained multi-level processes are more difficult to schedule. First, customer demand must be translated to the upstream intermediate processes to determine the quantity and timing required. Then the upstream-constrained process is scheduled for maximum efficiency. Last, the downstream orders are aligned and synchronized to the upstream intermediate orders. The process to align the downstream to follow an efficient upstream sequence is more difficult than the other way around because classic MRP logic works from lower levels to higher levels and not in reverse (Figure 4.4). Downstream Constrained Schedule

Upstream Line 1 Upstream Line 2 Downstream Line Figure 4.3  A downstream-constrained schedule.

Upstream Constrained Schedule Upstream Line Downstream Line 1 Downstream Line 2 Downstream Line 3

Figure 4.4  An upstream-constrained schedule.

Scheduling Processes and Software  ◾  37

Software Selection Finally, to close this chapter, a word about software selection. The start of the selection process should be a vision of how your business and equipment should be scheduled to meet your business goals. The vision should be translated into a series of requirements to meet the goals, then user stories, and examples of the requirements in practice. We’ve found that demonstration of specific user requirements and stories is the only way to assess the ability of scheduling software in practice. Almost any vendor can and will answer yes to a requirement in an RFI or RFP (Request for Information or Request for Proposal). The question is whether they can meet the requirement in a usable and efficient way. How many keystrokes and screens are required? What is the visualization? Only a demonstration of specific user requirements can show the usability. However, keep in mind that no software is likely to meet every requirement, and the software may meet the requirement in a different way than expected. What counts is that the overall new process will be more efficient.

Chapter 5

Example Process The Process To demonstrate the concepts covered in the book, we’ll describe a Blue Lakes salad dressing plant as an example. Blue Lakes employed most of the concepts we will describe at their manufacturing plant in eastern Pennsylvania. They had been facing all of the challenges described below and benefitted significantly from transitioning to a product wheel strategy and by replacing their legacy Excel-based finite scheduling process with a commercially available, professionally supported software product capable of scheduling manufacturing in structured, repetitive patterns. The following describes the scheduling process as it was practiced prior to any changes, with specific improvements described in the following chapters. The process equipment described here was the same before and after; no hardware changes were required to achieve the scheduling benefits. Blue Lakes manufactures salad dressings in a number of flavors and packages them in several formats: bottles of various sizes to be sold in grocery stores, larger tubs to be sold to institutions and restaurants, and small packets sold to delicatessens and grocery store salad bars. In addition to salad dressings, Blue Lakes also produces ketchup and several flavors of barbeque sauce. The process in Pennsylvania begins with the receipt of ingredients and packaging components. They are placed in inventory, and a receipt transaction is performed. Some ingredients must be put in refrigerated or frozen storage. Quality tests are done on the ingredients. The packaging materials are tested to ensure they will run correctly on the lines and conveyors. There is a single Spice Room, where the lower-volume solid ingredients are weighed out for each batch, some are blended, and all needed for that batch are kitted on a pallet. There is a Liquid Prep Kitchen, where some of the ingredients are weighed, blended, and sent to refrigerated Work in Process (WIP) storage.

DOI: 10.4324/9781003304067-6

39

40  ◾ Introduction

The plant has eight mix decks, where the ingredients are mixed, cooked, and cooled. Higher-volume ingredients are piped into the mixing vessels, while lower-volume ingredients like flavorings and spices are manually added. Four of the tanks make a 300-gallon batch, and the other four make 600-gallon batches. The finished batches are pumped from the eight mixing/cooking vessels into portable stainless steel totes. The plant has a total of 120 totes, of about 300 gallons each, so each batch fills either one or two totes. Some totes go to refrigerated storage, while others go directly to the packaging lines. There is also direct piping from the mixing tanks to the tub filling lines and the retail bottle lines that is used on occasion. There are two lines that fill and package the large tubs for restaurant and institutional sales, three retail bottle lines, and seven packet lines. Each line fills and seals the individual containers, applies a label to the bottles or tubs (the packets are enclosed in film with all product information and SKU code printed by the film supplier), puts the individual containers into cartons, seals the cartons, and palletizes them. The carton erection, filling, sealing, and palletizing are all automated. In fact, the entire plant is highly automated. The requirements for operating labor are: ◾ Forklift drivers to receive and store the incoming goods, move materials into and out of storage, transport the dressing totes from the mix decks to storage and to the packaging lines, and to move finished pallets into storage. ◾ Technicians to staff the quality lab and the weigh room. ◾ Operators to load minor ingredients into the mix tanks, monitor temperatures during cooking and cooling, and clean the vessels on flavor changes. ◾ Packing operators to load the empty bottles and tubs into the supply hoppers, change labels and packet film as required, monitor the lines and resolve upsets and jams, and perform the cleaning and mechanical adjustments on changeovers. The plant makes extensive use of bar coding technology to track all material movements, from raw material receipts, through QA testing and pre-mixing, on through the manufacturing operation, and into and out of the finished product warehouse. The various containers used in each step in the operation are barcoded as are all of the rack locations: Thus the location and the content can be verified at each step, so inventory accuracy is very high. The plant operates 24 hours per day, 7 days per week, although some equipment is not needed for all shifts. The prior scheduling process started with the packaging lines, and once those schedules were created, the mix decks had to be scheduled to feed

E xample Process  ◾  41

packaging lines with the proper dressings at the scheduled times. The stainless steel totes provided some degree of time buffering between the two areas, but only a few hours’ worth. Schedulers spent a lot of time managing this coordination and frequently had to modify the packaging schedules to achieve a feasible mixing schedule. Figure 5.1 is a very simplified Value Stream Map of the Blue Lakes PA plant.

Scheduling Information Flow: Communication between Systems All of the information processing needed to create production schedules had been done in a combination of the corporate SAP ERP system and an Excel finite scheduling system. The Excel tool was created and supported by the Blue Lakes IT team. A Forecasting module in SAP created a statistical forecast and allowed adjustments based on forecasts received from customers, inputs from the sales team, and any expected trends from marketing and business leaders. A consensus forecast was created at the finished SKU level, looking forward for 18 months in monthly increments. The upcoming 120 days were broken down into weekly increments. On a weekly basis, the ERP system employed MRP logic to take the consensus forecast, current inventories, currently booked sales, and firm production orders to create new planned orders for each SKU for the next 18 months. There are two schedulers at the Pennsylvania plant, one for bottles and tubs, and the mix decks, and one for packets. Each scheduler followed a weekly process to load their planned orders for the next four weeks into the Excel finite scheduling tool and create schedules for the next four weeks. The intent was for the schedules to be frozen for the next two weeks, but they rarely stayed firm anywhere near that long. Schedule changes had to be made almost daily, to accommodate machine outages, ingredient shortages, packaging component shortages, or more operator call-outs than expected. The weekly process required almost an entire day from each scheduler, and schedule changes consumed about an hour per day each, for the rest of the week. The schedulers loaded the weekly schedules and all updates back into SAP, which used MRP logic to create material and component requirements for Procurement. Procurement replenished components and inexpensive materials on a fixed interval basis, and more expensive materials on a fixed quantity basis. Procurement set safety stock levels but did so on a two weeks of coverage target rather than an analysis based on forecast error and supply variability.

RECEIVING

BULK STORAGE

FROZEN INGREDIENT

THAW (4 – 7 days)

REFRIGERATED INGREDIENT

DRY INGREDIENT STORAGE

PACKAGING MATERIAL STORAGE

PACKAGING COMPONENT STORAGE

INGREDIENT STORAGE

CURRENT INVENTORIES

SCHEDULERS

• Batching • Weighing

LIQUID PREP KITCHEN

• Blending • Batching • Weighing

SPICE ROOM

DAIRY

REWORK

REFRIGERATED WIP STORAGE

EXTERNAL GREEK YOGURT SUPPLIERS

Failures ~ 2%

TESTING

MIXING FORMULA PREP

LARGE MIX TANKS 600 GAL (4)

PORTABLE TOTES (120) 300 Gal ea

STARCH

PRODUCTION SUPERVISOR

SAP

DEMAND PLANNING

CURRENTLY SCHEDULED PRODUCTION

SMALL MIX TANKS 300 GAL (4)

WATER TANK (1000 Gal)

THE SCHEDULE (EXCEL)

PLANNED ORDERS • QUANTITIES • DUE DATES

DRY WIP STORAGE

Dry eggs

Yogurt Greek yogurt Buttermilk Sour cream

RO WATER SYSTEM

EXCEL

PRODUCTION SCHEDULING

Figure 5.1  Blue Lakes value stream map (VSM).

BULK SUPPLIERS

REFRIG, FROZEN SUPPLIERS

DRY INGRED SUPPLIERS

PACKAGING MATERIAL SUPPLIERS

RAW MATERIAL ORDERS PACKAGING COMPONENT ORDERS

RE-ORDER POINT LOGIC IS USED TO TRIGGER ORDERS

SAP

PROCUREMENT

PRODUCTION SCHEDULES

BLUE LAKES SALAD DRESSING CO. TOWANDA, PA SIMPLIFIED FLOW DIAGRAM

PACKAGING

PACKET FILLING LINES (7)

RETAIL BOTTLES (3)

QR HOLD 5 – 7 DAYS

REFRIGERATED FG STORAGE

DRY FG STORGE

INVENTORY STATUS

FORECASTS FROM CUSTOMERS

SAP

FORECASTING

TUB FILLING LINES (2)

INVENTORY STATUS

CONSENSUS FORECAST

SHIPPING

PICK TICKETS

SHIPMENT DATA

SAP

ORDER FULFILLMENT

TRENDS MARKETING INPUT BUSINESS INPUT

DEMAND HISTORY

GROCERY CHAIN FAST FOOD DISTRIBUTION CENTERS

ORDERS

BOTTLES

CANS

LARGE JARS

RETAIL CUSTOMER DISTRIBUTION CENTERS

RESTRAUNTS & INSTITUTIONS DISTRIBUTION CENTERS

PACKETS FOR DELICATESSENS SALAD BARS

EDI ORDERS • SHIP DATES • DELIVERY DATES

CUSTOMER SERVICE REPS

42  ◾ Introduction

E xample Process  ◾  43

The Products The plant produces a total of 540 packaged SKUs, considering all flavors and package sizes. That includes 250 individual flavors. Packets comprise 200 SKUs, tubs 190, and retail bottles 150. Packets are produced in eight sizes, ranging from four to six inches in width and three to six inches in length. Bottle sizes include 11 oz, 12 oz, 13 oz, 16 oz, and 32 oz. Larger containers are ½ gallon, one gallon, four gallons, and six gallons. Figure 5.2 breaks down the number of finished product SKUs by container size.

Product Differentiating Characteristics The Blue Lakes schedulers attempt to sequence the products run on any line to minimize changeover time and difficulty. There are a number of product characteristics that require significant equipment cleaning and sanitization if not sequenced according to the cleaning rules, the most significant being kosher, organic, and allergen content. The lines must be cleaned between any flavor change, but the cleaning before a run of organic product or before a kosher product must be much more thorough, so the goal is to run as many organic products together as possible, and kosher products should be run together to the full extent possible. Each salad dressing flavor can contain one or more allergens. Key among them are eggs, dairy, soy, nuts, fish, and mustard. Flavors with different allergens can be run in sequence without a full allergen clean if the new product

MARKET

CONTAINER TYPE

CONTAINER SIZE

Retail Retail Retail Retail Retail Hospitals, Schools, Restraunts Hospitals, Schools, Restraunts Hospitals, Schools, Restraunts Hospitals, Schools, Restraunts Delicatessens, Salad Bars Delicatessens, Salad Bars Delicatessens, Salad Bars Delicatessens, Salad Bars Delicatessens, Salad Bars Delicatessens, Salad Bars Delicatessens, Salad Bars Delicatessens, Salad Bars

Boles Boles Boles Boles Boles Tubs Tubs Tubs Tubs Packet Packet Packet Packet Packet Packet Packet Packet

11 Oz 12 Oz 13 Oz 16 Oz 32 Oz 1/2 Gal 1 Gal 4 Gal 6 Gal 1/2 Oz 1 Oz 1.5 Oz 2 Oz 2.5 OZ 3.3 Oz 4 Oz 6 Oz

Figure 5.2  Blue Lakes SKUs by container size.

PACKET LENGTH

3" 3" 3" 3" 3" 3" 6" 6"

PACKET WIDTH

4" 4.5" 5" 5.5" 6" 6.5" 5" 6"

No of SKUs 28 54 38 18 12 12 76 48 54 18 20 44 64 22 12 10 10

44  ◾ Introduction

contains all the allergens in the prior product with perhaps an additional one or two. But if the upcoming product lacks an allergen contained in the previous product, a full allergen clean is required. Thus sequencing flavors as much as is practical so that allergens are the same as before or with some additions is a scheduling priority. The allergen situation is made more complex by the fact that some, but not all, products are exported to Canada, and Canada has a slightly different list of recognized allergens (Figure 5.3). The one of most interest here is mustard, an ingredient in several flavors, and considered an allergen in Canada but not in the US. Thus products going to Canada must consider mustard in allergen sequencing, while products sold only in the US can ignore it. Some dressings contain particulates like seeds, which can get lodged in the minute openings in the machinery and then contaminate the next flavor. An extensive cleaning must be done to completely rid the apparatus of any residual particulates, thus another consideration to include in the scheduling logic that determines the optimum production sequences. Beyond kosher and organic requirements, allergens, and particulates, the next most time-consuming changeovers involve package sizes. Bottle size changes take a relatively long time because everything up and down the packaging line must be adjusted. Experience has shown that it is easier to start with the larger sizes, then narrow the fixtures and guides down to run smaller sizes, rather than going from small to large. Because this plant has three bottling lines to run the five bottle sizes, one (of several) goals is to minimize the number of different bottle sizes run on any line. The bottle

ALLERGEN

RECOGNIZED IN U.S.

RECOGNIZED IN CANADA

Eggs

Yes

Yes

Fish

Yes

Yes

Milk

Yes

Yes

Mustard

No

Yes

Peanuts

Yes

Yes

Sesame seeds

No

Yes

Shellfish

Yes

Yes

Soybeans

Yes

Yes

Sulphites

No

Yes

Tree nuts

Yes

Yes

Wheat

Yes

Yes

Figure 5.3  Ingredients considered allergens in the USA and Canada.

E xample Process  ◾  45

scheduler had tried to do this, but on an ad hoc basis rather than any holistic, structured design, and so had difficulty making it work. Salad dressing packets, sold to fast food restaurants and grocery store delicatessen counters, come in different lengths and widths. Adjusting the filling machine after a length change is not difficult as the pre-printed packaging film has an eye mark that is optically sensed to know where to cut the film between packets. But changing the packet width requires very precise machine adjustments; if not done correctly, there will be a number of minor stops after the size change to fine-tune the machine adjustments. Therefore, the scheduling process must try to minimize packet width changes. Salad dressing production requires large quantities of clean water. The water supply to this plant has salt and other contaminants, so a reverse osmosis (RO) system is employed to purify the water. However, the RO system doesn’t have enough capacity to supply all possible product combinations that could be running simultaneously, so the scheduling process must recognize and stay within the limitations. Some products are shelf-stable; they can be stored at room temperature once packaged and sealed. But others must be refrigerated, and there is a limited amount of refrigerated warehouse space available, so the inventory implications of any schedule must be calculated and schedules adjusted as necessary to stay within the available space. This may require more frequent campaigns of the higher-volume refrigerated products so their cycle stock doesn’t contribute to any excess inventory. Thus the scheduling process has a number of priorities to reconcile: organic products, kosher products, allergens, particulate content, flavors, bottle sizes, packet sizes, RO water requirements, and refrigerated storage space. And that is further complicated by the need to coordinate packaging schedules with mixing schedules.

Cultural Challenges Optimal production scheduling, which best balances the trade-offs between capacity utilization and efficiency, inventory and working capital, and customer delivery performance, requires appropriate leadership behavior. Managers must value performance that matches business strategy, and track KPIs that measure and give value to those strategic drivers. Managers must also recognize that some of the changes will not be readily embraced will not be easy, and therefore they must make and demonstrate a full commitment to accomplishing the changes. This includes providing guidance, resources, and any funding needed to make the effort successful. Blue Lakes management recognized early in the transformation that there were some deep-rooted behaviors and beliefs that would be challenging to redirect. For example, leadership had always stressed the importance of

46  ◾ Introduction

maximizing throughput every day, and operating people felt pride in exceeding the daily goals. When a new production record was set, it was a cause for great celebration. Management recognized that repetitive, structured patterns would on some days produce less than average output due to that day’s scheduled product mix, while other days would produce much more than average, and on a weekly or monthly basis, the available throughput would increase significantly. And while they recognized this on a logical basis, it was difficult for them emotionally to reverse years of highly valued behavior. It was difficult to accept that practices that had long been accepted and viewed as a key to their success could have been wrong. Another example was their blind allegiance to OEE metrics. While OEE (Overall Equipment Effectiveness – a widely accepted measure of asset productivity that penalizes for equipment downtime, yield losses, and inability to meet designed rates) can be a very effective way to measure and drive improvement in asset capability, there can sometimes be steps that can be taken that will reduce OEE but will improve business profitability. For example, if an asset sits idle because there is no demand for its product, that’s not considered an OEE detractor. But time spent in product changeovers is, so increasing the number of changeovers will reduce OEE. However, if a manufacturing asset or production line has excess capacity and therefore idle time, it may make sense to make more frequent changeovers of some products to reduce campaign size, cycle stock, and the working capital tied up in inventory. So a very appropriate step to improve profitability would reduce OEE. This disconnect was exacerbated at the Blue Lakes plant by the fact that the OEE metric is owned by the maintenance function, while inventory metrics are owned by production, thus requiring reconciliation of these conflicting goals and adjustments to KPIs by higher levels of leadership. A third example was the lack of value for Standard Work and the need to follow established practices. Production supervisors, and even operators, often deviated from the set plan to deal with some type of unexpected event. While a change in plan was warranted in most of these cases, it was frequently done without input from the schedulers and others with relevant points of view, thus some ramifications of the change weren’t anticipated. Even worse, acceptance of this practice emboldened operators to make changes as they saw fit, even when there was no situation requiring a change. It took frequent coaching across the entire organization, starting at the top, to turn these behaviors around. To summarize, the Blue Lakes salad dressing plant experienced a number of technical and cultural challenges in its transition to a more effective production scheduling strategy and to the supporting software systems. The concepts they employed to meet those challenges are described in detail in the rest of this book.

SCHEDULING STRATEGIES

2

Chapter 6

Repetitive Scheduling Strategies The primary reason that many process companies embrace the structured, repetitive strategies introduced in Chapter 3 is that these methods can unleash hidden capacity by simplifying and shortening changeovers, thus making more time available for production. The analytical determination of appropriate campaign sizes generally leads to less frequent runs of low-volume products, thus reducing the number of changeovers, and the sequencing logic puts changeovers in a pattern that reduces the number of parameters changed on each one. The reduction in the number of changeovers and the shortening of those remaining allows that time to be spent making product. For companies struggling to meet demand, this means more sales, more revenue, and better customer service. For companies staying comfortably within their capacity, this allows fewer shifts or fewer lines to be run, reducing operating cost. The Bountiful Company (formerly Nature’s Bounty) was able to completely de-staff two full lines at one of their packaging plants as a result of applying product wheels. They also eliminated all Saturday work and most overtime, which improved team morale. Another reason for the attraction is that operations will run much more efficiently if production is done at a uniform, level rate. This tends to maximize equipment productivity and labor productivity and smooth out the requirement for raw materials and support facilities. Unfortunately, for most plants, demand is not uniform at all. Variation in demand can be due to several factors: the normal, random variation that will almost always be present; seasonality; discounts and sales promotions; and longer-term trends due to new product acceptance or to sales decline of mature products. Heijunka is the lean technique for smoothing production to meet average customer demand. As described by Jeffrey Liker in The Toyota Way:

DOI: 10.4324/9781003304067-8

49

50  ◾  Scheduling Strategies

Heijunka is the leveling of production by both volume and product mix. It does not build products according to the actual flow of customer orders, which can swing up and down wildly, but takes the total volume of orders in a period and levels them out so the same amount and mix are being made each day. Leveling the mix is important to balance the use of people, equipment, and materials. If different products require different operations and, therefore, have different labor content, the utilization of labor can be smoothed out by producing a small number of each rather than producing in large lots. Similarly, if different products require different machine times and different materials, these will all be smoothed out by producing in reasonably sized lots. Appleton Paper (now Appvion) found that production leveling was one of the greatest benefits of their product wheel implementation. Ryan Scherer, Appleton’s Organizational Excellence and Capacity Manager and the primary architect of their wheel design and implementation, describes it this way: To deal with large production runs and the many inefficiencies and increased costs they created, wheels were employed as a structured way to reduce run length. Implementing wheels on the paper coaters allowed us to level load our large runs to eliminate the peaks and valleys in the schedule, and thus reduce overtime, create better flow, reduce WIP, and allow for a more predictable production schedule. And that was only the beginning. These scheduling concepts provide businesses with the alternatives of running the shortest possible cycles to enhance flexibility, agility, and responsiveness or extending cycles to gain efficiency. Analytics can be applied to find the balance that best meets business goals. In summary, the repetitive strategies described below, Product Wheels, Repetitive flexible Supply, Rhythm Wheels, or Fixed Sequence Variable Volume, will almost always outperform a more traditional scheduling process. They enable the business to make conscious decisions about the trade-off between Cash (throughput), Cost (inventory), and Service (fill rate) rather than let them fall wherever they may from a scheduling process that doesn’t place a priority on those factors. And following the same plan cycle after cycle, changeover after changeover, allows people to get accustomed to the repeating patterns. Practice does make perfect (or at least better and better).

Product Wheels Of the various structured scheduling strategies, Product Wheels are the most widely used. A product wheel is a visual metaphor for a regularly repeating

Repetitive Scheduling Strategies  ◾  51

sequence of the production of all the materials to be made on a specific piece of equipment, within a reaction vessel, or within a process system. Therefore, we will describe them first. Product wheels can be employed in a make-to-stock (MTS), make-to-order (MTO), or finish-to-order (FTO) environment. In fact, MTO and MTS products are often produced on the same wheel, as will be seen in the example later in this chapter. A product wheel can be operated in a push or a pull mode, although pull operation is far better. Pull operation of wheels is more responsive to fluctuations in demand and requires less safety stock for any specific delivery performance target. That is, the wheel will be designed based on average historical demand or forecast demand, but what is produced on any spoke is just enough to replenish what has been consumed from the downstream inventory since the last cycle. The fact that the quantity of any SKU produced on the wheel will be based on actual consumption means that each spoke can be slightly larger or slightly smaller than designed and can differ from cycle to cycle. The wheel is said to “breathe,” to be constantly seeing slight changes in the size of each spoke. This is one of the key advantages of wheels, that they will adapt to changes in demand from cycle to cycle. Because you can’t be sure that the plusses and minuses will balance out on every cycle, it is wise to build some “breathing room” into every cycle, to accommodate situations where the increases outweigh the decreases.

Product Wheel Design The best way to understand product wheels is to walk through the wheel design process. 1) Begin with an up-to-date, reasonably accurate Value Stream Map (VSM) The design of product wheels requires two general classes of information: one describing the throughput and performance of the equipment, and the other describing the product line-up, including demand history and trends and product characteristics affecting changeovers. A VSM gives the former. It provides a holistic view of how material flows through the process and how it is transformed from raw materials into the end product. It includes data such as reliability, yield, throughput, demand, and changeover times, that indicate how smoothly material is flowing. More importantly, it provides perspectives and specific data which are very useful in deciding where in the process product wheels would be appropriate. Another feature of the VSM is that it shows the flow of information from customer orders through forecasting, demand management, and planning

52  ◾  Scheduling Strategies

to the creation of schedules and how they reach the production floor. The understanding this provides is vital to a scheduling transformation program. 2) Determine which process steps should be scheduled by product wheels Any step in the process, any piece of equipment or process system, which has appreciable changeover times or losses should be examined as a candidate for a product wheel. “Appreciable” in this context means any changeover long enough or experiencing enough material loss that it influences the scheduling of the step. If the time, difficulty, or material loss is sequencedependent, it is even more likely that product wheels will be beneficial, because of the sequence-determining part of the methodology. A product wheel can be applied to a single independent step in a process or an entire coupled production line. Wheels can be run on more than one process step along the material flow, provided that there is enough inventory between any of those steps to decouple wheels that are following different sequences. 3) Analyze product demand variability: Identify candidates for MTO The next step in product wheel design is to examine the average demand and the demand variability of each product to be made on a wheel, to decide whether that product is best made to order or made to stock. High-volume products with low variability should be made to stock, while low-volume products with high variability are best made to order (Figure 6.1). One requirement for making to order is that the total manufacturing lead time and shipping time, from the MTO point to arrival at the customer, must be less than the delivery lead time committed to the customer. 4) Assign product families to specific production lines When multiple pieces of equipment or production lines are capable of making the same products, grouping products that run well together into families, and assigning families to production equipment can simplify changeovers and operations. This is similar to the lean concept of cellular manufacturing (see Chapter 25). In many cases, this assignment alone has been shown to increase manufacturing capacity by 10% or more, as in the Nature’s Bounty case cited in that chapter. 5) Calculate the shortest wheel time possible (available time model) Once products have been assigned to specific equipment or production lines, the next step is to focus on each specific line. One of the most important

PRODUCT VARIABILITY

Repetitive Scheduling Strategies  ◾  53

LOW DEMAND

HIGH DEMAND

HIGH VARIABILITY

HIGH VARIABILITY

MAKE TO ORDER (MTO)

MTO? MTS?

LOW DEMAND

HIGH DEMAND

LOW VARIABILITY

LOW VARIABILITY

MTO? MTS?

MAKE TO STOCK (MTS)

PRODUCT DEMAND Figure 6.1  MTO/MTS decision matrix.

decisions to be made for each line is how long the wheel cycles should be. One approach is to maximize flexibility, customer service, and responsiveness by finding the shortest cycles that will fit within the available capacity. The available time model considers some period of time, perhaps a week. It then computes the amount of time required to produce the full customer demand for that period and subtracts that from the total available time. The difference is the amount of time that could be available for changeovers. The total changeover time for one cycle of the wheel is calculated by adding up all the individual product changeovers. That sum is then divided into the total time available for changeovers to indicate how many cycles could be completed per week. The wheel time can be the number of hours per week divided by the number of cycles.

Wheel cycles per period =

Total available - Total productiontime å Changeover times per cycle

This doesn’t necessarily give the most appropriate wheel time, but it does provide a useful perspective. 6) Estimate economic optimum wheel time (the EPQ model) The available time calculation only provides the shortest wheel possible under the current conditions, but not necessarily the best. The EPQ (Economic

54  ◾  Scheduling Strategies

Production Quantity) model takes into account the economics involved and thus gives another perspective, and may be more appropriate for a business challenged by the cost of manufacturing. The EPQ is a classic way to calculate lot size or campaign size. It approximates the best balance between total changeover cost and inventory carrying cost. Larger campaigns, with their longer product wheels, require more inventory to supply customers or downstream operations while specific products are not being made, and smaller campaigns require more changeovers. The specific equation for the quantity that results in the lowest total cost is:

EPQ =

2 ´ COC ´ D D ö æ V ´ r ç1 ÷ è PR ø

where COC = changeover cost, D = demand per period, V = unit cost of the material, r = % carrying cost of inventory per period, and PR = production rate in units per period. This equation is sometimes called economic order quantity, or EOQ. It must be noted that the EPQ calculation is only an approximation, which takes into account the factors that typically are most significant, but ignores some factors that usually have a lower impact. The inventory included, for example, is only the cycle stock required, that is, inventory to satisfy average demand between production spokes for a material. It ignores safety stock, the inventory needed to protect against normal random variation in supply and demand. There are more advanced algorithms that can be applied that take safety stock into account, and also recognize that in-family changeovers are less time-consuming and costly than family-to-family changeovers. The better advanced wheel design packages include these models. 7) Determine the wheel time (making the choice) Because the available time model and the EPQ analysis will generally yield different results, a choice must be made. If the business wants to run the very shortest wheel to be most agile and responsive, the available time answer should be chosen. Similarly, if the time needed for production consumes almost all available capacity, the available time model defines the lower boundary for overall wheel time. Conversely, if there is available capacity and the business wants the most economic design, then the EPQ-based analysis should be the basis for design. The EPQ model will give a somewhat different answer for each product, so some degree of rounding and grouping into fixed time frames that are multiples of each other must be done. For example, all products may be put

Repetitive Scheduling Strategies  ◾  55

into one-week, two-week, or four-week cycles. Again, advanced wheel design software usually has the capability to do this rounding and grouping. 8) Calculate inventory requirements With MTO products, no designed inventory is needed to support the product wheel. As product orders are received, they are loaded onto the wheel schedule. After they are produced on the wheel, they flow through the downstream steps, through to packaging and truck loading. They may stop temporarily at a buffer inventory, but for reasons other than requirements imposed by the wheel schedule. They may also wait in finished product inventory for a truckload to be produced, staged, and shipped. For MTS products, inventory will be required, proportional to wheel time. For these products, there must be inventory downstream, either as WIP or as finished product, to satisfy the needs for each material during the period when other products are being produced on the wheel. Figure 6.2 tracks the inventory for a single product made on a wheel. An amount, P1, of that product is produced at the beginning of wheel cycle 1 and flows through downstream process steps into the inventory graphed in Figure 6.2. Demand for that product during the remainder of the wheel will consume portions of that inventory until the next production, P2, arrives at the inventory. This rising and falling of inventory in a saw tooth pattern repeats for every cycle of the wheel. If demand were always exactly equal to the average, only cycle stock, the average demand during one wheel cycle, would be needed. However, demand will typically vary in a random fashion, sometimes greater than

P2 D1

P3 D2

D3

Safety Stock

Cycle Stock

P1

WHEEL TIME

WHEEL TIME

WHEEL TIME

CYCLE 1

CYCLE 2

CYCLE 3

D = DEMAND DURING THE WHEEL CYCLE P = PRODUCTION TO BEGIN A WHEEL CYCLE

Figure 6.2  Inventory required to support the wheel (cycle stock and safety stock).

P4

56  ◾  Scheduling Strategies

average and sometimes less. To avoid stockouts during periods of higher demand, safety stock is usually carried. Safety stock can also provide protection from stockout in cases where the wheel, due to production problems, doesn’t return to this spoke at the scheduled time. The demand D2 in Period 2 is higher than average and the safety stock provides material to supply downstream needs and customers until production P3 arrives. The fact that we are replenishing what has been consumed, cycle stock plus or minus some safety stock, rather than what the forecast specifies is what makes this “pull.” It should be apparent that shorter wheels require less cycle stock and longer wheel times more cycle stock. Safety stock is also somewhat proportional to wheel time; shorter wheels mean that the variability causing the need for safety stock will be less (this is true if the variables follow a normal distribution). The calculation of cycle stock is straightforward. It is the average demand for that product for one wheel cycle time, times the cycle frequency if the product is not made every cycle. If a product is made every third cycle, for instance, the cycle stock must support demand over three wheel cycles. Cycle Stock = Average Demand during One Wheel Cycle × Cycle Frequency The calculation of safety stock is somewhat more complex. There is a statistical formula often recommended, and it does provide some useful guidance. However, it only calculates the safety stock required to meet a cycle service level, the expected percentage of cycles without a stockout, but not the quantity of the stockout. What most businesses target is a desired fill rate, which is the percentage of volume ordered that can be filled. It is a much more difficult calculation, but the better wheel design software packages include it. In addition to cycle stock and safety stock, any additional product that must be made to fill minimum batch size requirements should be included, as should any inventory required for quality hold times. 9) Assign lower-volume products to specific cycles The wheel time selected in step 7 will be appropriate for all the high-volume products but may be inappropriate for lower-volume products, so they should be made only on every second cycle, every fourth cycle, or at whatever frequency makes the length of that run worth the changeover it takes. The EPQ model from step 6 gives very good guidance in selecting that frequency for each product. Thus we have all the high-volume products made on each cycle along with half of the two-cycle products and one-fourth of the fourcycle products. Care should be taken when assigning lower-volume products to specific cycles to continue the changeover reduction effort; for example,

Repetitive Scheduling Strategies  ◾  57

if a problematic allergen is contained in only two products and they’re both four-cycle products, they should be assigned to the same cycle so that changeover has to be done only once every four cycles. 10) Determine the optimum sequence Determining the optimum sequence to cycle through the products made on any production line or major piece of equipment is one of the most important aspects of wheel design. Find the best sequence, and wheels will run at their best; settle for a less optimum sequence, and wheel performance will suffer. Sequences should group products with similar changeover or cleanup requirements together to reduce changeover time and cost. On a wheel with more than one unique cycle, each cycle must be sequenced. 11) Review with stakeholders It is critical that everyone who is affected by production scheduling understands how this transformation will change their part of the operation. This includes schedulers, production personnel, maintenance mechanics and supervisors, test lab operations, and accounting, warehousing, and marketing personnel. It also includes members of management who value increased throughput, those accountable for customer delivery performance, and those responsible for inventory policy. Because wheel design strives to find the optimum balance among those three variables, the leadership responsible for them must participate actively in setting objectives, design freedom, and constraints. The more that these people are kept informed during the design, the less you will need to repeat prior steps. 12) Revise the current scheduling process All formal scheduling processes must be examined and modified as appropriate to accommodate product wheel scheduling. Coordinating wheel schedule patterns with the ERP system, be it SAP, Oracle, JDA, or something else, is critical for successful wheel operation. Chapters 10–13 cover this in more detail. A product wheel for one of the salad dressing packet lines described in Chapter 5 is shown in Figure 6.3. It packages eight final SKUs comprising four dressings in three packet sizes. The products had been allocated to the seven packet lines in a way that all kosher products were run on only a few lines, as were the organic dressings. Thus packet size, allergen group, and dressing were the only things to consider in defining the sequences for this line. The primary wheel time selected was two weeks, with three of the lowervolume products made every second cycle, two on cycle 1 and one on cycle 2 (Figure 6.4). There were no MTO products assigned to this line (very few

3.4 fl oz Caesar 2 fl oz Caesar 2.5 fl oz Caesar 2.75fl oz Caesar 2 fl oz Gr Yogurt Caesar 2.5 fl oz Lite Caesar 2.5 fl oz Lite Caesar - Canada Available Time

3.4 fl oz Caesar 2.5 fl oz Caesar 2 fl oz Gr Yogurt Caesar 2.5 fl oz Lite Caesar - Canada 204/2.5fl UL/TF LT CAESAR 2 fl oz FF Caesar Available Time

90.0 18.1 78.8 20.8 20.1 28.4 6.1 17.7

90.0 78.8 20.1 28.4 6.1 26.3 30.3

Product Description

857356 750816 191862 270006 58191 125000

857356 85998 750816 99090 191862 270006 58191

Total Unit Forecast

2 2 2 2 2 4

2 4 2 4 2 2 2

Selected Wheel Time (weeks)

90.0 78.8 20.1 28.4 6.1 26.3 30.3

90.0 18.1 78.8 20.8 20.1 28.4 6.1 17.7

Hours per Cycle

60 60 38 60 45 100

60 120 60 100 38 60 45

Average Shelf Life Days (available on plant)

Figure 6.3  Product wheel design for a salad dressing packet packaging line.

Cycle 2 PP92 PP83 PP49 PP84 PP43 PP27

Cycle 1 PP92 PP41 PP83 PP72 PP49 PP84 PP43

Product

Hours per Cycle

3.4fl 2.5fl 2fl 2.25fl 2.5fl 2fl

3.4fl 2fl 2.5fl 2.75fl 2fl 2.25fl 2.5fl

Size

3"X6.5" 3"X6" 3"X6" 3"X6" 3"x6" 3"X5.5"

3"X6.5" 3"X5.5" 3"X6" 3"x6" 3"X6" 3"X6" 3"x6"

Packet Size

NO NO NO NO NO NO

NO NO NO NO NO NO NO

NO NO NO NO NO NO

NO NO NO NO NO NO NO

Kosher Organic

1,2,8,10 1,2,8,10 1,2,8,10 3, 7, 8 3, 7, 8 3, 7, 8

1,2,8,10 1,2,8,10 1,2,8,10 1,2,8,10 1,2,8,10 3, 7, 8 3, 7, 8

Allergen

YES YES YES YES YES YES

YES YES YES YES YES YES YES

Particulate

C384 C384 C420 C615 C615 C186

C384 C384 C384 C384 C420 C615 C615

Dressing Formula

58  ◾  Scheduling Strategies

Repetitive Scheduling Strategies  ◾  59

CYCLE 1 - WEEK 1 & 2 PP43

CYCLE 2 - WEEKS 3 & 4

FREE

PP84

FREE PP92

PP27

PP92

PP43

PP49

PP84

PP72 PP41

PP49

PP83

PP83

Figure 6.4  Pie charts for the two salad dressing wheel cycles.

of the packet products are MTO), but if there were, they would have been included. A spoke for each MTO would be shown at the appropriate point in the sequence on each cycle; because the shipping lead time for all MTOs is less than four weeks, there must be an opportunity to make them on each two-week cycle . Note that with the assignment of products to packaging lines we did, this packet line has some time not needed for production or changeovers, shown as available time and sometimes called “breathing room.” That covers the major steps in wheel design. There are additional steps required to prepare for start-up on the wheels and then the “Go Live” period and finally to ensure the gains are being sustained that are all covered in Chapter 28, the implementation roadmap. Before starting wheel design, opportunities to reduce changeover time and difficulty using SMED techniques (Chapter 23) should be explored. Shorter changeovers will enable more effective wheels to be designed.

Synergy with Lean Product wheels are very much in alignment with lean principles and concepts: ◾ A Value Stream Map is the recommended starting point, to understand Flow. ◾ Wheels are an excellent way to achieve Heijunka. ◾ Assignment of products to specific production lines is a component of cellular manufacturing. ◾ The design process will identify where SMED will have the greatest impact and quantify its value. ◾ Inventories are replenished by Pull, not by any MRP forecasting logic.

60  ◾  Scheduling Strategies

◾ Finished product inventory is managed by Kanban principles. ◾ Wheel design determines appropriate Supermarket quantities. ◾ And finally, the entire process engages representatives from across the workforce and is in fact dependent on their active participation.

Benefits of Product Wheels Product wheels almost always increase OEE and usable capacity and therefore increase throughput which often leads to increased sales and revenue. Changeovers become easier and less costly. Product wheels right-size campaign lengths based on demand and frequently reduce the number of changeovers for low-volume products. This is one of the main factors that draw companies to product wheels. Companies like The Bountiful To summarize, Product wheels: Company and Appvion have saved • Increase usable capacity and millions of dollars without having throughput, resulting in increased to spend a single dollar in capital revenue or reduced cost. investment by moving to a product • Tend to level production as a natuwheel strategy. ral behavior. Product wheels tend to level • Optimize production sequence. production as a natural behavior. • Add structure and predictability to Because cycle stock is based on high-variety operations. customer demand over the period • Add stability to scheduling and of one or more wheel cycles, proreduce chaos and firefighting. duction naturally flows to match • Provide a basis for informed decicustomer orders. The product wheel sions about production sequence design methodology tends to drive and campaign length. mixed model scheduling to the • Provide a basis for informed decismallest practical campaign size and sions about MTO for appropriate thus achieves the lean objective of products. Heijunka. • Optimize transition cost versus Product wheel design forces a inventory carrying cost. structured analysis of the various • Provide a structured basis kinds of changeovers to optimize for determining cycle stock the sequence of production. People requirements. sometimes assume that they are fol• Provide a structured basis for callowing an optimized sequence, but culating safety stock requirements. a data-based analysis often reveals • Dramatically simplify the weekly that improvements can be made. scheduling process; key decisions All key decisions about sequenchave been made in the design ing, campaign lengths, and producphase. tion frequencies are made in the • People perform better following an design phase, so they don’t have to established routine. be repeated week by week during

Repetitive Scheduling Strategies  ◾  61

the ongoing scheduling process. This simplifies the weekly process and frees up schedulers’ time to make better decisions about non-routine things. Product wheels add predictability to high-variety operations, operations with a high degree of product differentiation. Everyone associated with an operation scheduled by product wheels knows what is going to be produced and at what time. Any special tools, materials, or personnel required for changeovers can be scheduled in advance. If the start-up of the next material places additional loading on the test lab, this can be planned for. The design methodology allows a clear understanding of tradeoffs between changeover costs versus inventory costs. The benefits of SMED activities (see Chapter 23) can be quantified. Justification for capital improvements that would improve flexibility can be determined. A fixed, repeatable schedule allows cycle stock requirements to be explicitly known. A fixed span between production cycles of a given material fixes the time at risk from supply or demand variability so that safety stock requirements can be calculated. By implementing product wheels we were able to move from fill rates of ~75% to over 99% reliably in a three-month timeframe. (David Kaissling – Formerly Chief Supply Chain Officer – Shearer’s Snacks) Finally, and perhaps most importantly, people tend to perform better if they follow a routine. We all have small routines we go through in our daily lives, like how we dress in the morning and how we start our car, buckle our seat belts, and adjust our mirrors. Following these routines makes things easier and more efficient, and product wheels bring this same efficiency to the workplace; schedulers and operators become more and more comfortable as they follow established patterns. For this to be fully effective, they should have a hand in designing those patterns; their perspectives always lead to more practical patterns and sequences. Morale is improved because operations people would rather spend time making products than performing changeover tasks.

Repetitive flexible Supply (RfS) RfS is a scheduling strategy developed by Ian Glenday and described in his book Lean RfS. RfS has objectives and procedures somewhat similar to product wheels but comes at them from a different direction. It begins with an A–B–C product classification, calling A products “greens,” B’s “yellows,” and C’s “reds.” The classification is generally based on volume. If a plant has several production

62  ◾  Scheduling Strategies

lines, it assigns greens to one or more lines, yellows to other lines, and reds to still other lines. Thus when analyzing the full portfolio and deciding what products to assign to each line, RfS is decidedly different from product wheel logic. Because RfS assigns products primarily on volume, rather than on families that run well together, it will generally result in a higher number of hours lost to changeovers. However, they are more similar when deciding on frequencies and campaign sizes for the products on a specific line; rather than using EPQ calculations or EPQ-based algorithms, it continues the green–yellow–red classifications if there are combinations of them on a given line. Greens run every cycle, and yellows and reds run less frequently. Reds and perhaps yellows are not even put in specified cycles but are produced when there is a need. This green–yellow–red classification generally gets you to nearly the same place as the EPQ-based logic, because the products EPQ suggests run every cycle are going to be the A’s, every second cycle products will be most of the B’s, and so on. Another significant difference is that product wheels produce the quantity needed to replenish what has been pulled from inventory since the last cycle, thus the time for any SKU could be slightly longer or shorter than the designed time, and the wheel is said to “breathe.” This is aligned with the lean principle of pull: Replace only what has been consumed. In its initial maturity levels, RfS operates on fixed time periods rather than target quantities, so if an SKU is to be produced for 12 hours, it is produced for 12 hours on every assigned cycle. Thus on a good day, you will overproduce, and on a day when things aren’t running so well, you’ll underproduce. In lean terms, this is push replenishment. Glenday proposes running a fixed plan for eight weeks before making any adjustments, which has the advantage of greater repeatability and predictability but will require more safety stock to cover the risk of poorer-performing weeks. If the basic cycle is two weeks, an SKU on a product wheel needs only enough safety stock to mitigate two weeks of variability because you can make adjustments every two weeks. But if you follow RfS and stay with planned quantities for eight weeks, you need safety stock to protect for that period, i.e., statistically twice as much. His safety stock calculations seem to confirm that. Another key difference is that RfS strives to run the shortest primary cycle possible rather than the most economical. It does expand the cycle to fit within convenient time periods, like weeks. But it doesn’t seem bothered by the reality that very short cycles may require a high number of very costly changeovers and thus may degrade operating cost and profitability. It leans towards the available time model described above but arrives there by trial and error rather than following a mathematical formula.

Repetitive Scheduling Strategies  ◾  63

A major similarity is a focus on determining the optimum sequence, to reduce non-productive time and increase useful capacity. But his method relies much more on experience and gut feel than on any analytical process. This will work well with just a few SKUs and changeover parameters but becomes difficult with a larger number of SKUs and changeover parameters. The RfS process has five steps: 1) Fixed Sequence, Fixed Volume – Establish a pattern and establish economies of repetition; create push flow 2) Faster Fixed Sequence, Fixed Volume – Maximize the impact of repetition and increase the flow rate 3) Fixed Sequence, Unfixed Volume – Begin to respond to demand, with some pull, continue to rely on repetition 4) Unfixed Sequence, Fixed Volume – Move away from repetition, pull using small volumes 5) Unfixed Sequence, Unfixed volume – Maximum flow and complete pull To get to phases 4 and 5, you have to reduce changeover times to the point where they are no longer a concern. That doesn’t happen often in process operations, thus many RfS implementations likely end with phase 3, which is the phase most like product wheels. At the Kimberly Clark plant described in Lean RfS, the primary cycle was four weeks at the start, then reduced to two weeks as changeover times got shorter. It’s not clear if they got changeovers down enough to move away from any sequencing.

Rhythm Wheels Josef Packowski explains the Rhythm Wheel concept in Lean Supply Chain Planning. It is very similar to the product wheel methodology we’ve been describing with these features in common: ◾ An optimal sequence is determined, which is followed cycle after cycle. ◾ Higher-demand products are made every cycle, while lower-demand products are made on a fixed subset of the cycles. ◾ The amount of a specific product made on each cycle is to replenish actual consumption rather than always producing the designed quantity, thus following lean pull replenishment principles. ◾ Make-to-order (MTO) products can be made on the same wheel with make-to-stock (MTS) products. Spokes are allocated to them in the appropriate place in the sequence, at a frequency to satisfy MTO lead time requirements. If there are no orders for them in this cycle, the spoke is skipped.

64  ◾  Scheduling Strategies

Fixed Sequence Variable Volume (FSVV) In Liquid Lean, Ray Floyd describes a production scheduling technique called Fixed Sequence Variable Volume (FSVV), which is very similar to product wheel methodology. As its name implies, finding the optimum sequence to minimize changeover time and cost is of paramount importance, just as it is with wheels. Like wheels, FSVV follows pull replenishment principles, where what is produced during any campaign is based on actual consumption rather than on any designed amount. And like wheels, it produces lower-volume products on a frequency less than every cycle. The key difference is that rather than setting a fixed wheel time, FSVV allows total cycle time to float. The primary reason is that it has generally been used with complex chemical polymerization reactions which were extremely capacity limited. Therefore, with high demand and difficult changeovers, they tend to run long cycles. The key to shortening the cycle time is reducing changeover time by finding a better sequence, but determining the optimum sequence is very difficult and is therefore a focus of continuous improvement activities. Incremental sequence improvements are being made regularly, so effective capacity is almost continuously increasing while cycle time is almost continuously decreasing. While Product Wheels and RfS tend to use time periods for the complete cycle that make logical sense to the operation, for example, completing a cycle every day, every week, or every two weeks, FSVV accommodates operations where the cycle times and capacity don’t allow rounding to weeks or days. Although overall cycle time is allowed to float, it is consistent enough from cycle to cycle that cycle stock requirements can be determined reasonably well. All in all, the two scheduling methodologies are far more alike than not and offer similar benefits. Floyd reports that the Exxon polypropylene plant in Baytown Texas was able to reduce changeover losses by 90% and increase reactor effective capacity from 50% of the nameplate to more than 85% using the FSVV methodology.

Summary Although product wheels are probably the most widely used, any of these structured, repetitive scheduling strategies will almost always outperform a more traditional scheduling process. A well-designed sequence offers greater throughput by reducing changeover difficulty and time. Following the same plan cycle after cycle, changeover after changeover, allows people to get accustomed to the repeating patterns. Practice does make perfect (or at least better and better).

Repetitive Scheduling Strategies  ◾  65

They enable the business to make conscious decisions about the trade-off between Cash (throughput), Cost (inventory), and Service (fill rate) rather than let them fall wherever they may from a scheduling process that doesn’t place a priority on those factors. And, as will be covered in the next chapter, a structured scheduling ­framework provides a better foundation to manage the disruption that almost all manufacturers are facing. You shouldn’t try to adhere to the planned patterns perfectly, but make whatever adjustments are required and follow the plan as well as is practical. That will avoid the chaos that usually ensues if you abandon the plan and create ad hoc schedules day by day. As we’ve said before, agility is far more important than perfection. Because product wheels are among the most comprehensive of these structured methodologies, that’s what we’ll use in the examples throughout the book.

Chapter 7

Dealing with Disruption The Nature of Disruption All manufacturers are facing a high degree of disruption today, and indeed have been off and on since the dawn of the industrial revolution, including plagues, labor unrest and strikes, political unrest, material shortages, and weather events. We are faced with all the same disruptions today, but at an unprecedented level in recent world history: pandemics; labor shortages as workers re-evaluate their needs against their willingness to compromise their lifestyle; political unrest and uncertainty; material shortages caused by systemic supply chain weaknesses exposed by Covid; and weather events, becoming less predictable and more severe due to climate change. We hope that by the time you read this, Covid-19 and all of its various mutations will be under control, but the conditions that allowed Covid to be so devastating are still with us, so we are still vulnerable. The net effect of these on society in general and manufacturing specifically is likely far less than it was decades or centuries ago; we now have systems and structures that make society, supply chains, and manufacturing more robust and resilient. But those same systems and structures can be less tolerant and more rigid than the ad hoc processes of yesteryear, so they can appear to be making things worse. Thus in the face of these disruptions, rather than trying to adapt the systems and structures to the changing conditions, there is a temptation to abandon them and revert to unstructured decision-making based on the events of any specific day. The more mature manufacturers have realized that some structures and patterns of doing things may have to be loosened a bit to match the emerging realities, but that some degree of structure will be better than none at all. That is all true of the disciplined, structured scheduling strategies we’ve described.

DOI: 10.4324/9781003304067-9

67

68  ◾  Scheduling Strategies

Manufacturers of packaged food products, for example, have seen demand spike and forecasts become very unreliable. The available workforce is uncertain from day to day. Certain ingredients are unavailable from time to time.

Ability to Deal with Disruption The advice we are giving to companies in this situation is rather than abandoning the product wheel structure and reverting to ad hoc day-by-day scheduling to make whatever MRP says is needed and can be made, take several major steps: 1) Temporarily rationalize the SKU line-up. Remove a number of products from the catalog based on a collaborative process involving business leaders, marketing representatives, and operations people. The criteria should include a balance of factors: – Low-volume products – Low-margin products – Products needing hard-to-get ingredients – Products causing difficult changeovers, especially those with unique allergens or particulates 2) Remove all of those from the wheel designs. Redesign the wheels to allow more time for the manufacture of high-volume and high-margin products, and to add breathing room (unallocated space on the wheel) to allow for more flexibility and agility. DISRUPTIVE FORCES RAW VOLATILE MATERIAL DEMAND SHORTAGES

MANAGEMENT SUPPORT

SUPPLY CHAIN ISSUES

WEATHER EVENTS

STABLE, AGILE STRUCTURED SCHEDULE DESIGN

ENGAGED WORKFORCE RELIABLE EQUIPMENT

CROSS TRAINED WORK FORCE

STAFFING ISSUES

CONTINGENCY PLANNING

RATIONALIZED SKUs

COUNTERMEASURES

Figure 7.1  Disruptive forces and countermeasures.

EXECUTABLE SCHEDULES ACCEPTABLE DELIVERY PERFORMANCE

ALTERNATE MATERIALS & RECIPES

STANDARDIZED PACKAGING MATERIALS

Dealing with Disruption  ◾  69

3) Plan for disruption. Anticipate all likely causes of disruption and decide in advance how to behave in each case. 4) Run the wheels as close to the design as possible. Other than deviations absolutely necessary, stay with the designed patterns. 5) Treat the removed products like low-priority MTOs. If they are ordered, and the necessary raw materials are available, capacity is available, and they can be fit into the sequence efficiently, make them. If not, don’t produce them. 6) Learn from your experience. After some period, adjust the wheel patterns and/or the contingency plans to better suit whatever realities you have to deal with. But do this as a deliberate, coordinated activity, not as a knee-jerk reaction to some crisis. With this approach, lines will not run as efficiently as before the disruption, but some or most of the efficiency gains will be held. And life will seem less chaotic, certainly not back to normal, but much better than a new fire drill every day (Figure 7.1). This approach concentrates the available labor force and working equipment on products that are likely to be able to be manufactured, products that will reduce required changeovers so that there is additional capacity to provide a degree of manufacturing flexibility, and products that will maintain or increase operating revenue. The COVID shock sent our global supply chain into a collective tailspin from which we have been slow to recover. Had more of our manufacturing organizations been built on this structured scheduling methodology, I believe our response would’ve been stronger, quicker, and far less painful to our manufacturing teams, sales teams, customers, and consumers. (Dave Rich – VP, Strategic Sourcing and Fulfillment, Litehouse Foods) There is a tendency to lose faith in disciplined processes and structures and revert to old habits and behaviors in the face of what can feel like total chaos. That’s why it is critically important for senior leaders to continue to exhibit their commitment and support. It’s also important to communicate that they don’t expect perfection, that customer orders will be missed, and that as long as people are following the newly defined policies, everything will be OK. People in the trenches, those making the dozens of small decisions every hour to keep the operation running and material flowing, need to be reassured that they have a safety net, that they won’t be criticized or penalized for deviating from prior rules and practices. They should instead follow, critique, and work to improve the new rules and standards.

70  ◾  Scheduling Strategies

None of this works without the will and commitment of both leaders and workers. Those can be very fragile in times of stress and thus deserve significant attention. But with confidence in the revised plan and the commitment to follow it, things will operate more smoothly. The key is to maintain the discipline of the process and focus on the short-term variations that will help maintain some degree of the planned rhythm and lead to longer-term stability and success.

An Example: The Story of P&G Luvs Diapers In the 1980s, Luvs Diapers launched a highly successful market innovation for boy- and girl-specific diapers. Volume increased by 30% almost overnight, and the business was running out of manufacturing capacity. Fortunately, the test markets had correctly predicted the business volume increase, and we had stockpiled inventory ahead of the introduction to give us time to react to volume uncertainty, but we found ourselves rapidly running out of inventory. We used this as a catalyst to try out some ideas with which we had been experimenting, short-cycle manufacturing, regular repeating cycles, and changeover improvement. The idea was to improve manufacturing efficiency and reduce inventories simultaneously, so that we could keep up with the demand increase, and wouldn’t have to devote precious capacity to building stock back to higher levels. Our early research had been encouraging. We had developed a simulation of our production lines, distribution system, and customer order patterns. The simulation showed that no amount of inventory was as effective as short production cycles in achieving high customer service. To make short cycles possible, we had been exploring minimizing changeover losses by dedicating production lines by size. Although we had improved other changeovers, a size changeover involved removing and replacing major sections of the line. It took most of a production day, and it wasn’t something we would be able to improve quickly, or in time to help our current situation. We’ve talked about long changeovers as a characteristic of the process industry in Chapter 2, and Chapter 25 will cover grouping into virtual work cells to simplify changeovers and increase throughput. Luvs diapers were produced at six sites across North America. Previously, our strategy had been for each site to be self-sufficient on product mix, to minimize distribution costs. We ran the numbers on the potential efficiency gain of eliminating size changeovers against the estimated interplant costs to balance capacity and this was positive too. Since we didn’t have any better alternatives to avoid the upcoming out of stocks, we decided to put the plan into action. We dedicated lines by size, avoided size changeovers entirely, and accepted interplant shipments to balance capacity as a tradeoff for avoiding lost sales. Some lines were even dedicated by size and count. This new strategy had the advantage of matching flow almost exactly. Daily production of each size could almost exactly match average daily demand.

Dealing with Disruption  ◾  71

The next part of the strategy was to produce every size and language variation every week, in a repeating cycle. On a daily level, the higher volume commodities used in every product had level flows. The size-specific components had level requirements every day. It was only the packaging materials that had variability, but they also matched average demand on a weekly basis, and there was a regular pattern or sequence within the week. The line operators now had stable sequences and got really good at the remaining changeovers and startups from changeovers. Our suppliers could fall in line with this production; they received a steady, consistent, and predictable volume of orders. We avoided line shutdowns from missing materials. From a finished product inventory standpoint, there was no longer cycle stock for the sizes or the high-volume product count variations. Production occurred every day to match demand. The minor volume count and language variations required a cycle stock, but it was only to cover one week. A major component of safety inventory is uncertainty during the time when not producing, sometimes called time at risk. Since high-volume products were produced every day, there was no time at risk to protect, and little safety stock was needed. For the lower-volume variations, the time at risk was never more than a week. The net result was that customer service remained high and out of stocks were avoided. Production efficiency improved from the 70% range to the high 80% range, and inventory turns tripled. Figure 7.2 shows the trend of Luvs diaper inventories.

Luvs Diaper Inventory in Days on Hand

1988/89 Average

Aug-85 Sep-85 Oct-85 Nov-85 Dec-85 Jan-86 Feb-86 Mar-86 Apr-86 May-86 Jun-86 Jul-86 Aug-86 Sep-86 Oct-86 Nov-86 Dec-86 Jan-87 Feb-87 Mar-87 Apr-87 May-87 Jun-87 Jul-87 Aug-87 Sep-87 Oct-87 Nov-87 Dec-87 Jan-88 Feb-88 Mar-88 Apr-88 May-88 Jun-88 Jul-88 Aug-88 Sep-88 Oct-88 Nov-88 Dec-88 Jan-89 Feb-89 Mar-89 Apr-89 May-89 Jun-89 Jul-89

Luvs Diaper Inventory in Days on Hand

Figure 7.2  Reduction in diaper inventories due to short-cycle production, numerical value of days on hand omitted.

72  ◾  Scheduling Strategies

While we had been continuously improving inventory performance, it shows a step change down to a lower consistent level when the shortcycle strategy was implemented. Note the management of overall inventory during 1988 and 1989 was in a consistent bandwidth, as described in Chapter 9 on the role of inventory.

SCHEDULING PROCESSES, SYSTEMS, AND SOFTWARE

3

Chapter 8

The Role of Forecasting Our purpose in this chapter is not to go into detail on forecasting methods and software but to talk about how to think about the forecast and evaluate its effectiveness and impact on scheduling. Every business needs some level of forecasting to commit to manufacturing capacity, staffing, ingredient, material, and component purchases. For example, even with a pull replenishment strategy, replacing what has been used or sold carries the implicit forecast that the future will be like the past, with an amount of variability that can be buffered by inventory. For a maketo-order business, there is a limit to the horizon of customer orders, but the business must continue beyond this horizon. For most businesses, the customer’s expected lead times will be shorter than the time required to procure all materials and produce; therefore, some level of forecasting is required to procure materials. For the longer term, to sustain any business, its managers must project future requirements for capital equipment, supply base, and staffing. A simple forecast might be a moving average of shipments over some period of time, or perhaps a weighting or smoothing of the history. For a seasonal business, it might be the orders from last year for this time of the year. This is often called a naïve forecast. However, if a business can do better than a simple naïve forecast, it may be able to produce more economically or serve more customers. In a maketo-stock business, it will be able to have a better assortment of products on hand. A make-to-order business would be able to tailor its staffing and material purchases to the level of demand. Anticipating periods of high demand can improve customer service, and anticipating low demand can minimize inventories and obsolescence.

DOI: 10.4324/9781003304067-11

75

76  ◾  Scheduling Processes, Systems, and Software

Forecast Value Add The difference between a simple naïve forecast and a managed forecast is called forecast value add. Evaluation of forecast value add is important because there are many cases of forecasts being worse than the underlying data; in other words, negative forecast value add. The authors worked with a food manufacturer where we showed them that the magnitude of their forecast error was higher than that of a simple 12-month moving average of sales by SKU. Some interventions in forecasting degrade value. A manufacturer of expensive boats asked each of their regional sales representatives to forecast their sales for the next year by the specific size and model. For many of these boats, history showed that only a few would be sold for the entire country. How likely is it that a sales rep could guess whether one of their customers would decide to purchase a specific size and model next year? A simple analysis of the sales variability of any of these boats at the regional level would have shown that their demand was not statistically forecastable. On the other hand, it’s possible that through contact with their customers, a rep might know that one of them has decided on a specific size and model, and is just waiting to place their order. A proper value-added forecasting process would differentiate when a forecast enhancement is adding value from when it is just a guess. So, what should this boat manufacturer do? While it may not seem so, it is a good example of the diverging flow of a typical process manufacturer, as described in Chapter 2. The basic components of resins, composite substrates, fabrics, paints, varnishes, engines, marine electronics, and control systems can be made into many different models and sizes of boats, and at the consolidated level, is much more likely to be forecastable. Since there was only one manufacturing site, the regional variation of the forecast was not really important; it was simply a sales tool to hold sales reps accountable.

Bias and Accuracy A common framework for analyzing forecast value add is to look at bias and accuracy. Bias is one of the best and simplest tools for analyzing forecast accuracy. Is the forecast consistently under-predicting or over-predicting the actual demand? Accuracy is more difficult to define. Its importance is twofold, measuring improvement, and knowing the amount of error that the inventory, planning, and scheduling systems must be able to absorb to meet the business’s customer service goals. We often hear the statement that the forecast is always wrong. While this is true, it overlooks or downplays how useful the forecast can be and the

The Role of Forecasting  ◾  77

importance of understanding its accuracy and accounting for ranges of possibilities in business planning, scheduling, and inventory management. Let’s consider the analogy of forecasting the single roll of one die. How would you like to be the dice forecaster if you were evaluated on the accuracy of your guess as to what the next roll will show? You have a one in six chance of being right. But that position overlooks what we do know. We can say with certainty that any roll of the die will be between one and six, and we can account for this range of error once we understand it. If we always have six available and replenish our inventory to this level, we will have perfect customer service barring some other supply chain upset. We know that if we have more than seven, we are overprotected. What about rolling two dice together? Even though the results are completely random, we can suddenly make a much better forecast. However, one of the authors often starts out with a set of loaded dice when demonstrating this point in their supply chain classes. The distribution of a roll of two dice is shown in Figure 8.1. While rolls of 12 and 2 are possible, they are rare events. In the business analogy, we can start to make informed decisions about the level of customer service that we wish to protect with the inventory. Players in backgammon do this all the time when judging which pieces to move forward and balancing moving further forward against protecting current gains. Another example is the weather forecast. Everyone likes to say that it is always wrong. This may be true if we expect it to tell us exactly when it will rain two days from now. Instead of absolute numbers, we need to think about trends, ranges of possibilities, and contingencies. Current weather technology is quite accurate when looking at trends of the weather, with the understanding that the timing of the weather events might be slightly off and that a probability is not a guarantee. It can be very helpful in choosing between the activities one might plan to do next weekend and knowing what to be prepared for.

Probability distribution Rolling 2 dice 25

20

20

Probability (percent)

Probability (percent)

Rolling 1 die 25

15 10 5 0

1

2

3

4

5

6

15 10 5 0

2

3

4

5

6

Outcome

Figure 8.1  Probability distribution for one and two dice.

7 Outcome

8

9

10

11

12

78  ◾  Scheduling Processes, Systems, and Software

Coefficient of Variation For accuracy, the evaluation method that we recommend is the coefficient of variation (CV), which is the standard deviation divided by the mean. It has the advantage of being a dimensionless number: Thus, the variability of different items with different levels of volume can be compared. For example, if we have SKU A with a standard deviation of 100 units and SKU B with a standard deviation of 10 units, B might look like the more stable, predictable product. But if the monthly sales of A are 2,000 units and B are 10 units, then A has a CV of .05 while B is 1.0. Thus, CV provides a useful way to compare products with widely differing demands. As another example, in analyzing the well-known bullwhip effect, we could compare the variability of finished product shipments to that of the distribution center receipts, the production schedule, and component orders (Figure 8.2). Another advantage of the coefficient of variation is that it can be used directly in inventory targeting calculations. Other common accuracy measures are mean average percentage error (MAPE), weighted average percentage error (WAPE), and similar. They have the advantage of being concepts that are easy to understand and calculate compared to the more abstract CV, which uses standard deviation. But they cannot be used directly in inventory targeting, and they have other flaws. The first is that the amount of under-shipping a forecast is limited as shipments can’t be less than zero, while the amount of over-shipping is unlimited. Therefore, they have an intrinsic bias. The second is that depending on how they are calculated, zero shipments in any measurement period can result in a divide-by-zero error. The calculation of standard deviation for its use in CV is easy with modern software; Excel has several standard deviation functions available. Nowadays, most plant personnel are familiar with statistical process control and can understand the concept of standard deviation.

The Bullwhip Effect Supplier Manufacturer Distribution Center Customer

Figure 8.2  The bullwhip effect.

The Role of Forecasting  ◾  79

Going back to our food manufacturer, when we translated their forecast error from MAPE to a coefficient of variation and compared the CV of their forecast to the CV of their shipments, it was easy to see that the variation or error was higher for their forecast than it was for their shipments. And that a simple moving average of their shipments was more accurate than their forecast (Figure 8.3).

Timing and Aggregation There are at least two other dimensions to assessing forecast accuracy: the forecast’s timing and its aggregation.

Figure 8.3  MAPE compared to the CV of forecast and CV of shipments.

80  ◾  Scheduling Processes, Systems, and Software

The measured forecast period and its time aggregation should relate to the timing of critical business decisions. It should use the forecast that was available at the time the decisions were made. The food manufacturer that we talked about earlier allowed changes to the forecast right up to the week of manufacturing and measured their accuracy against the updated forecast. Yet they committed to the production schedule, staffing, and component materials three weeks in advance. Clearly, their accuracy measure didn’t relate to the timing of their critical business decisions. A better accuracy measure would compare the forecast to shipments with a three-week lag. These kind of accuracy charts are often called waterfall charts. Assessing forecast accuracy with a one-, two-, three-, or four-week time lag may be more meaningful. Not all business decisions will be made at the same timing, and therefore there is not a one-size-fits-all approach for the forecast lag. A business may have different kinds of production lines, serving different customers, with different commitment zones for their production schedule. Some components can be received on short lead time. Others, perhaps with overseas sourcing, have a much longer lead time. Waterfall charts, or forecast error calculations with different time lags, can help you assess the amount of error that must be accounted for when making each of these decisions. Sales are fulfilled by the individual product code at the final point of shipment to the customer. The task of the supply chain is to get the right product to the right place at the right time. Therefore, getting the total amount of sales correct, but the product mix wrong, or having the product mix correct, but the products in the wrong place are examples of aggregation errors.

Different Forecast Goals The demand forecast will be used by several disciplines in the company, with different goals and requirements. For example, manufacturing and purchasing must plan and schedule the specific equipment, staffing, and components at the manufacturing plant sites. Finance is using the forecast to assess profitability and may roll up products with similar cost and profitability implications. Management wants to know overall volumes by geography, split along the lines of sales, marketing, or profit and loss accountability, and may roll it up by quarters and fiscal years. The distribution organization uses it to plan warehousing and transportation capacity, but will translate the forecast to volumes, weights, and flows in each lane and warehouse, and perhaps major product characteristics like refrigeration. The question to ask is whether your business has a forecast that is aligned with the aggregation needed for each function. Specifically related to our topic of production scheduling is the demand forecast at the right level for scheduling, or has it been designed more to meet the needs of finance and management?

The Role of Forecasting  ◾  81

To resolve this dilemma, forecasting software should be able to easily aggregate and disaggregate the forecast along different lines of products, geographies, and time. The base forecast should be at a level of products and time that will allow each part of the business to meet its needs. For example, forecasting at the national level may be good enough for financial planning but will be inadequate for manufacturing, warehousing, and shipping.

Choice of Demand Forecasting Unit The demand forecasting unit (DFU) is the lowest level of the forecast and should be the point within the product aggregation hierarchy where it makes the most sense to develop the forecast. Forecasts at levels above the DFU can be derived by aggregating the forecasts. Forecasts below the DFU are derived by breaking down the forecast mathematically. An example of a DFU might be a specific size and flavor of dressing at Blue Lakes. Over time, the artwork for the bottle and container will change, there will be formula upgrades to improve its flavor, cost, or shelf life, and there may be promotional offers. But the total demand for the DFU will be relatively constant, and its volume trajectory can be tracked and statistically forecasted. The specific SKUs are the artwork variants, product upgrades, and promotional packs. However, it could be that the best place to forecast is higher in the aggregation. At Blue Lakes, a dressing can be produced in jars, tubs, different sizes of bottles, and packets. The jars, tubs, and bottles are all standard across many different dressings. Only the printed labels and printed packets are different by SKU. Perhaps it makes sense to forecast at the level of the major mix variants, because that’s where the greatest accuracy is gained for capacity planning and ingredients. This would become the demand forecasting unit. The size and packaging variations would use a standard percentage breakdown for their volume estimate and for ordering labels and packets. The standard jars, bottles, and tubs could be stocked with a buffer inventory or Kanban.

Product Transitions Product transitions, sometimes called phase-ins and phase-outs, are a common forecasting business case, and typically of great concern for schedulers. Most forecasting software has features to allow forecasting transitions and tweaking how the product to be phased out ramps down and the new product ramps up. New product introduction is another dimension of aggregation. When a new SKU is a replacement, we might expect that the total demand for the new plus old is relatively constant, perhaps with a bump in demand to reflect the

82  ◾  Scheduling Processes, Systems, and Software

improved new product and a decline for the old product because of its product lifecycle. Often there will be a distribution pipeline impact where the supply chain needs to run out of the old product ahead of its discontinuation, followed by a quick ramp-up and bump in demand as the new product is stocked. Some products may be discontinued without a replacement, and some new products may be introduced without a current equivalent product. Capable forecasting software will have provisions to model all of the above phase-in and phase-out scenarios. The transition problem in scheduling can be greatly different depending on whether the timing of the product transition is fixed, or based on the depletion of the old item and the availability of the new item. A simple example could be a running product upgrade to improve performance or reduce manufacturing costs compared to a holiday promotion. The holiday promotion must be available on a certain date to meet marketing commitments; after the holiday, any leftover products must be scrapped, sold at a discount, or held for next year. A sudden unexpected peak of demand in September could result in Christmas items on the shelf in October unless more of the regular product can be quickly produced. In contrast, the strategy for the running upgrade is to introduce it as soon as it becomes available, but the customer has no firm expectations of exactly when they will get it. In the running upgrade, what counts is getting the total forecast for the two items correct. In the seasonal variation, managing the split and timing is more important.

Product Segmentation for Forecasting Not all products are equal. They differ in their amount of variability and forecastability, value or margin, and importance to the business. In the example below, products are segregated along two dimensions. The first is their amount of variability and, therefore, whether their demand is forecastable or not. The second is their value to the business. This creates four categories of products that the business can use to adjust its forecasting, production, and inventory strategy (Figure 8.4). Business Value

Forecastable

Not Forecastable

High Value

Focus Effort Here Collaborative Forecasting Market Intelligence

Difficult Enrich the Forecast

Low Value

Low Touch Use Statistical or Simple Forecast

Buffer with Inventory Convert to Make to Order

Figure 8.4  Product segregation by value and forecastability.

The Role of Forecasting  ◾  83

To put this diagram in perspective: ◾ High-value forecastable products provide the most opportunity and payoff for effort and improvement. ◾ Forecastable products of low value or low business value will have a low payoff for improvement. Minimize their forecasting effort or put their forecast on automatic. ◾ Non-forecastable products of low value or low business value can be protected through inventory buffers. Consider changes to ordering policies like producing to order or extending the customer lead times. ◾ Non-forecastable products of high value or high business value pose the most creative challenge. For example: – Is there other related information that can be used to improve the forecast? – Can the forecast be moved up to a higher level to where its accuracy is better, then use the higher-level forecast to plan component materials and flexible capacity? – Can the differentiation to specific products be delayed by using postponement or Finish-to-Order strategies?

Summary All forecasts are wrong, but most are useful. To know how useful a forecast is, you must understand how it was created, if bias is present, and have some idea of its accuracy. If the forecast is very inaccurate, a naïve forecast based on an average of past shipments or other simple calculations may serve you better. For a forecast to be useful, it must have the appropriate timing or latency. A forecast of what we need to produce next week or the week after is not very helpful if we had to order raw materials and commit to the line schedule a month ago. Forecasts can be aggregated at different levels for different purposes. For planning additional plant capacity, a forecast aggregated at the full product line-up may be appropriate. But for planning finished product inventory requirements, safety stock needs, and projecting fill rates, the forecast should be at the individual SKU level. Finally, the level of effort to improve the accuracy of the forecast should be influenced by the value of the product and its value to the business.

Chapter 9

The Role of Inventory

Figure 9.1  A sign in Roanoke, Virginia, USA. We hope that this book helps you to do better. Photograph by the author.

Determining appropriate inventory levels is generally one of the most important and challenging tasks that operations managers face: Carry too much, and you’ve got more money tied up in working capital than you need. Carry too little, and you face the possibility of unacceptable stockouts and disruption of your manufacturing process. Basically, inventory exists anywhere in a process where flow is not continuous, where material stops moving for any reason for any length of time, or where material may not be immediately accessible for use. The longer the material is stopped or the longer it may be inaccessible, the greater the resulting inventory. This includes Work in Process (WIP), finished goods waiting for shipment to customers, stock in quality inspection, stock in transit, deployed inventory at distribution centers, and raw material inventory, where raw materials are purchased in lot sizes greater than what can be immediately consumed.

DOI: 10.4324/9781003304067-12

85

86  ◾  Scheduling Processes, Systems, and Software

Components of Inventory These components of inventory can be broken down and compartmentalized to allow its calculation and visualization (Figure 9.2). 35

Aggregate Inventory Components

30

Business Protection

25

Cycle

20

Non Performing

Safety Goods Issue

15

Goods Recipt

10

Quality Hold Distribution

5 0

Inventory

Figure 9.2  Components of inventory.

The main inventory components that affect production planning and scheduling strategy are safety stock, cycle stock, and business protection. Other components for goods receipt processing, goods issue, quality inspection, quality hold, or non-performing inventories tend to be constant and affect planning by exception, when their levels spike or are different than expected. For example, a large quality hold. Transportation and distribution inventories lie outside the site and outside the scope of this book but are an important component for managing the business’s total inventory level. Cycle stock is, as its name implies, inventory that will cycle from lower levels to higher levels as stock is produced and depleted. Safety stock covers variability in demand and supply. The stock of any single material will never be constant, and it will change with phases in a product’s production cycle and fluctuations in its demand or line efficiency. However, the aggregate stock of all products at a site should be relatively constant and should be managed to an upper and lower limit. We will cover cycle stock and safety stock later in the chapter and start with a discussion of managing inventory to control limits. The upper limit of total site inventory is the point at which production should be curtailed or additional demand should be accepted or created. Inventories higher than this limit tie up capital without incurring additional benefits, and risk obsolesce or expiration. The lower limit of aggregate inventory is the point at which the site can no longer produce economically because there is not enough time to complete the efficient production sequence or to finish one thing before something else is needed. Production scheduling becomes like a game of “whack a mole.” At Procter & Gamble, we used to call it the death spiral. Inventories are too low

The Role of Inventory  ◾  87

to produce on an efficient production cycle and disruption increases, and the shorter cycles and disruptions cause less efficient production which leads to even lower inventories, which lead to even shorter cycles, even less efficiency, and more disruption. An example of managing inventories is shown in Figure 9.3, from the example of Luvs Diapers in Chapter 7 on Dealing with Disruption, this time redrawn with inventory target limits. Note that overall inventory was managed within the limits and shows control even during the period when inventory was being steadily reduced. The lower limit was the point at which the business would lose the ability to provide customer service while maintaining its weekly cycle through the language and count variations. Above the upper limit was considered too much inventory, and production was curtailed. Luvs Diaper Inventory with Average and Limits

Jul-89

Jun-89

Apr-89

May-89

Mar-89

Jan-89

Feb-89

Dec-88

Oct-88

Nov-88

Lower Limit Sep-88

Jul-88

Aug-88

Jun-88

Apr-88

Upper Limit May-88

Feb-88

Mar-88

Jan-88

Inventory Target Dec-87

Nov-87

Oct-87

Sep-87

Aug-87

Jul-87

Jun-87

May-87

Apr-87

Feb-87

Mar-87

Luvs Diaper Inventory in Days on Hand

Figure 9.3  Managing inventory within limits.

Managing Inventories Once inventory targets have been established, the business must manage them. The inventory of any individual material will be constantly changing as orders are received and material is produced. Carrying safety stock is pointless if it is never used. Consequently, the inventory of any single material can vary between zero and a high number. It’s not bad management if some products are below safety stock or above their targets. However, the aggregate inventory of a site, business, or family of products should be relatively constant. Some products will be produced, while others are declining. Some

88  ◾  Scheduling Processes, Systems, and Software

products will be over-shipping, while others are under-shipping. But it is a problem if many products are low or high on inventory at the same time; therefore, the aggregate or total inventory must be managed between upper and lower limits. If too many products are low at once, it will overtax the production lines. If too many products are high at once, working capital is not being used efficiently, there may be insufficient storage space, handling and damage costs will increase, and the risk of obsolescence and shelf-life expiration increases. Typically, a site has a limited number of production resources, and it can only produce a limited assortment of products at one time. Low inventories disrupt the efficient sequence, since there isn’t time to produce a reasonable size batch of each product in the sequence before something else becomes a priority. Most product groups have an ideal sequence of production, and efficiency will be lost if they are produced out of sequence. Allergens and colors are two examples. A sequence that adds allergens will minimize cleaning, since the process must be thoroughly cleaned as soon as an allergen is subtracted. Colors may be able to flow from light to dark, or in a color wheel sequence, with little or no cleaning between them. Let’s consider how to set the limits. The lower limit of inventory for efficient production is readily calculable. It’s the sum of safety inventories, average cycle stock, quality inspection inventories, staging or transportation inventories, distribution inventories, and held, dead, or other non-performing stock. The allowances for these other stocks can be calculated in several ways. One is simply to look at how much stock is currently tied up in each category, or to develop an average for each over time. Another is to take a more data-based view, by asking questions like: How long should it take for an item to clear quality inspection? How long should it take for a material to be ready for use from the time it is received? How long should it take us to move the finished product through the plant to the shipping dock? How long should it take us to work off or disposition held product and what is the rate of products put on hold? The answers to these questions can be converted from time to quantity by estimating an hourly, daily, or weekly volume. The upper limit is harder to calculate and more subject to judgment, it is the point at which production should be curtailed or additional demand should be accepted or created. Inventories higher than this limit tie up capital without incurring additional benefits. The costs of handling, damage, obsolesce, and expiration increase out of proportion with any offsetting benefit of protecting service or minimizing changeovers. Often physical space, working capital limits, or obsolescence can put a hard ceiling on the maximum. In the absence of constraints, the question becomes one of looking at the range of variability in the site’s total demand over time. The distance between the upper and lower limits should allow for inventory to bounce between the two limits given natural variability, as seen in Figure 9.3. Beyond the upper limit

The Role of Inventory  ◾  89

or below the lower limit, it should be unlikely that inventory will return to the range without corrective action. Statistically, it’s possible to calculate the combined variability of all products at the site; however, there is often a correlation between products and external causes like events, weather, or competitive actions that cause many products to move together, rendering the statistics invalid. Therefore, judgment, hard limits, or the review of demand history may be the best method. Regardless of the difficulty in setting overall targets, choosing almost any informed set of targets and managing against them will improve results. Next, let’s consider the actions to take when approaching or outside the limits. If you’re below the lower limit, production will have to run harder to catch up, efficiency may decrease, and disruption might increase. In the extreme, it will result in the death spiral discussed in the opening paragraphs, where disruption leads to lower efficiency, which leads to more disruption and even lower efficiencies. It’s important to stop the cycle before it starts. When inventories are trending dangerously low, production should be added, or demand should be managed. If a site ever finds itself in the death spiral, the only way out is to manage demand or increase capacity. Recently, COVID and supply chain disruptions put many sites in this area. A typical response was to prioritize customers and reduce the product assortment to allow more efficient production, as recommended in Chapter 7. The upper limit is easier to manage, but often more painful internally. It’s easier because one solution is just to stop producing. It’s painful because no one likes to shut down lines and employees can become demoralized. Often it conflicts with how results are measured, and how people are rewarded. Plant accounting gets upset if they have to absorb what are often called idle mill charges. This can be mitigated by planning in advance for what to do with downtime and using balanced measures for results and reward systems. An alternative to curtailing production is to accept additional demand or generate demand, often through promotion and pricing; however, these actions typically have a longer lead time than production curtailment and must be put in motion before the inventory limits are reached.

An Inventory Management Example A large integrated paper company had a wood pulp and a diaper division. One grade of internally produced wood pulp was used in the manufacture of the diaper absorbent core. On average, the output of the pulp plant equaled the requirements of the diaper division; however, there was variation in the rate of pulp production and in the demand for diapers.

90  ◾  Scheduling Processes, Systems, and Software

Company management believed there was a cost advantage to using all the diaper pulp production internally and was reluctant to sell pulp on the outside market when diaper demand was low or when the pulp division was running at high efficiency. When the pulp division had production upsets, or when diaper demand was high, there was a reluctance to purchase pulp from the outside market until the last minute, resulting in crisis purchasing of pulp to avoid shutdowns of diaper lines. History showed a cyclical stockpiling of large quantities of pulp in high cost outside warehousing, followed by periods of low inventory with expedited purchases of premium-cost pulp from the outside market. A simulation was created to show that if buying and selling were done mechanically, at certain inventory levels, the total amount of buying and selling was relatively small, inventory stayed within limits, and the overall result was more profitable. A meeting was called with the top management of the diaper and pulp divisions. The meeting kicked off by explaining that even though demand and supply were equal on average, historically there were periods where the volume mismatch was significant, both high and low. The simulation was used to educate management that there were times when it was more profitable to sell pulp than to store it, and other times when a minimum stock should be maintained by purchasing pulp from the outside market in an orderly fashion. The rules were followed in the subsequent years and inventory remained within limits. The diaper lines ran better because they were supplied with fresher pulp from consistent sources.

Cycle Stock and Safety Stock Cycle stock is the amount of a specific product to be made during the production cycle, to satisfy demand over the full cycle including the portion of the cycle when other products are utilizing the asset. For example, if the production process is based on a total production cycle of seven days, the cycle stock for Material A would be seven days. If Material A occupies one day of the cycle, at the end of its production day there must be six days of material in the finished goods warehouse, or in downstream process steps and headed for the warehouse. That material is needed to satisfy the demand for product A in the six-day interim until Material A will be made again. So the cycle stock for Material A includes the one day that was consumed while Material A was being produced, and the six days to satisfy demand during the rest of the cycle. Safety stock is material held to satisfy demand in cases where actual demand is higher than expected, or where the next cycle was late in starting. Figure 9.1 shows a profile of inventory versus time for a single SKU in a case where cycle stock and safety stock are present. In production period P1, cycle stock is produced, to raise the level to A. Demand during the next

The Role of Inventory  ◾  91

D1

P2

D2

P3

D3

P4

Safety Stock

Cycle Stock

cycle, D1, is equal to the average demand, so the cycle stock is consumed, but the safety stock is not. Production P2 raises total inventory back to level A. Demand during the next cycle, D2, is higher than average so, in addition to consuming all the cycle stock, some of the safety stock is needed. This would also be the case if it took longer than average for the process to complete its cycle and return to making this material. Thus, the safety stock will protect flow against either variation in demand or variation in supply lead time. Production P3 must be greater than average in order to replace cycle stock plus the amount of safety stock that was consumed (Figure 9.4).

14 days (approx)

14 days (approx)

14 days (approx)

Wheel Time EPEI Figure 9.4  Cycle stock and safety stock.

Cycle stock is based on the average demand expected. This can be based on either demand history or on a forecast. More detail on cycle stock and how to calculate it in various situations can be found in several texts, including Lean for the Process Industries, 2nd edition (Productivity Press, 2019). But safety stock and its calculation deserves a closer look.

Calculating Safety Stock To repeat, safety stock is inventory carried to prevent or reduce the frequency of stockouts and thus provide better service to customers. Safety stock, sometimes called buffer stock or reserve stock, can be used to accommodate: ◾ Variability in customer demand or in demand from downstream process steps (where demand history is used to set cycle stock or order points)

92  ◾  Scheduling Processes, Systems, and Software

◾ Forecast errors (where forecasts are used to set cycle stock targets or order points) ◾ Variability in supply lead times ◾ Variability in supply quantity If these variabilities are random and are reasonably normally distributed, the following calculations will result in appropriate safety stock levels. Even if not, they will still give some guidance and will be better than the rules of thumb, for example, that safety stock be set at 20% of cycle stock, or “we’ll be OK if we keep two weeks on hand.” According to the Central Limit Theorem of probability, when random variables are summed up, their sum will tend towards a normal distribution even if the individual variables are not normally distributed. For example, the roll of a die is not a normal distribution, it’s flat, with six possible outcomes, each an equal probability. But the results of multiple rolls of dice quickly converge on a normal distribution.

Variability in Demand To understand how we can avoid stockouts in the face of variable customer demand, a short lesson in statistics is in order. Figure 9.5 is a histogram, a plot showing the number of cycles at which each demand range occurs. If we consider rolls of a specific grade of paper made on a paper-forming machine, with an average demand of 130 rolls per weekly production cycle, the histogram shows how many weeks the true demand was within each range. The .

Number of weeks with that demand

Of 52 weekly cycles

12 Weeks 8

8 Weeks 5

5

3

3

2

2 75 85 95 105 115 125 Rolls Rolls Rolls Rolls Rolls Rolls

135 145 155 165 175 185 Rolls Rolls Rolls Rolls Rolls Rolls

Average Demand 130 rolls

0 Demand

Figure 9.5  A histogram of weekly demand.

The Role of Inventory  ◾  93

histogram shows that, for the 52 production cycles within a year, the demand was very close to the average for 12 of those weeks. In this plot, the width of each bar represents 10 rolls; so on these 12, the demand was between 125 and 135 rolls. It was somewhat higher, 135–145 rolls, during 8 weeks, and 145–155 rolls during 5 weeks. As the range of demand values goes higher, the number of weeks within that range decreases. There is a similar pattern on the other side of the average; for 8 weeks, the demand was between 115 and 125 rolls, and between 105 and 115 for 5 weeks. This bell-shaped curve is typical of While standard deviation, many demand patterns. sigma, is a very good meaSome products will have little variability sure of absolute variability, it and thus a very narrow histogram, while is a poor indication of relative others will have higher variability and a variability. In the prior example wider histogram. The width of the curve and with a demand of 130 rolls per the underlying variability can be characterweek and a sigma of 28 rolls, ized by a statistical property called stanif we didn’t have the 130 roll dard deviation and symbolized by sigma, σ. number, we wouldn’t know if While the calculation of standard deviation is a sigma of 28 rolls was very beyond the scope of this discussion, undermodest or very high. However standing σ can help us calculate how much if we express the variation as safety stock we need to give us various levthe ratio of sigma to average els of protection against demand variability. demand, we get a value of If we carry no safety stock and have 0.21, or 21%, which gives us only the 130 rolls of cycle stock, that will be an indication of how signifienough to satisfy all demand for this prodcant the variation is. This ratio, uct on half the cycles; half the time demand σ/D is called the Coefficient of will be at 130 rolls or less, and half the time Variation (CV) and is considgreater. With no safety stock, we will be ered a very good indication of vulnerable to stockouts on half the cycles. relative variability. Statistics teaches us that if we carry extra stock equal to 1σ, that will be enough to cover demand on 84% of all cycles, as shown in Figure 9.6. Sigma is 28 rolls for this product, so if we carry 28 rolls of safety stock in addition to the 130 rolls of cycle stock, that should be sufficient to prevent stockouts on 84% of the cycles, about 44 weeks. If we carry safety stock equal to 2σ, that should cover 98% of the cycles, as shown in Figure 9.7. Thus the key to determining safety stock is deciding on the tolerance for stockouts and then using that to determine how many sigma’s of variability you need to cover. For example, if you decide that you can tolerate stockouts on no more than 2% of the cycles, that sets the cycle service-level goal at 98%, and we saw in Figure 9.3 that that requires 2σ of safety stock, or 56 rolls. The percentage of cycles you hope not to have stockouts is called cycle service level (CSL), and the number of sigmas required to achieve that is called the service-level factor or the Z factor.

94  ◾  Scheduling Processes, Systems, and Software 50%

 84%

Cycle Stock

0

28 rolls

Safety Stock

Inventory = Cycle Stock + Safety Stock Demand

Figure 9.6  Safety stock equal to one standard deviation covers 84% of the cycles. 50%

 98%

Cycle Stock

Safety Stock 56 rolls

0 Demand

Figure 9.7  Safety stock equal to two standard deviations covers 98% of the cycles.

The general equation for safety stock required to cover demand variability is shown below. From our discussion of forecasting in the previous chapter, if you follow our recommendations and measure your forecast error in Coefficient of Variation (CV), the standard deviation can be quickly calculated by multiplying the standard deviation by the mean of the demand:

Safety Stock  Z  D

The Role of Inventory  ◾  95

Many businesses will use inventory modules that will make these calculations automatically or multi-echelon inventory optimization systems (MEIO), which will make recommendations about where inventory should be placed within the supply chain to minimize overlap and total inventory. However, it is helpful for a planner or scheduler to understand the concepts behind these programs. Figure 9.8 shows the relationship between Z and service level. As can be seen, the relationship is highly non-linear: Higher service-level values, i.e., lower potential for stockout, require disproportionally higher safety stock levels. Statistically, a 100% service level is impossible.

3.5 3

Z FACTOR

2.5 2 1.5 1 0.5 0 80

85

90

95

100

105

DESIRED SERVICE LEVEL Figure 9.8  The relationship between service factor and service level.

Typical service-level goals are in the 90%–98% range, but good inventory management practice suggests that rather than using a fixed Z value for all products, Z should be set independently for groups of products based on strategic importance, profit margin, dollar volume, or some other criteria. Doing this will place more safety stock in those SKUs with greater value to the business, and less safety stock in the products believed to be less important to business success. The above equation assumes that the standard deviation of demand is calculated from a data set where the demand periods are equal to the lead time or production cycle length. If not, an adjustment must be made to the standard deviation value to statistically estimate what the standard deviation would be if calculated based on the periods equal to the total lead time. As an example, if the standard deviation of demand is calculated from weekly demand data, and

96  ◾  Scheduling Processes, Systems, and Software

the lead time is two weeks, the standard deviation of demand calculated from a data set covering two-week periods would be the weekly standard deviation times the square root of the ratio of the time units, or 2 . Bowersox and Closs, in Logistical Management, use the term performance cycle (PC) to denote the total lead time. If we let T1 represent the time increments from which the standard deviation was calculated (1 week in the above example), PC to represent the total lead time or production cycle length, then

Safety Stock  Z  PC

T1

 D

When procuring raw materials, the performance cycle includes the time to: ◾ Decide what to order (order interval or review period) ◾ Communicate the order to the supplier ◾ Manufacture or process the material ◾ Deliver the material ◾ Perform a store-in Inside our own manufacturing facility, the performance cycle includes the time to: ◾ Decide what to produce ◾ Procure materials that are not in stock ◾ Manufacture the material ◾ Release the material to the downstream inventory ◾ Return to the next cycle ◾ If we are carrying inventory in a finished product warehouse, and customers allow a delivery lead time greater than the time needed to deliver to the customer, then the remaining customer lead time can be subtracted from the performance cycle The performance cycle can be considered the time at risk, i.e., the time between making a determination on how much to produce and the time to make the next determination and have it realized. If cycle stock has been calculated from historical demand, then the variance used in the safety stock calculation should be based on past demand variation. If forecasts are used to set cycle stock, then the thing requiring protection is forecast error. The standard deviation of forecast error would replace the standard deviation of past demand in the safety stock formula, which would become:

Safety Stock  Z  PC

T1

  Fcst Err

The Role of Inventory  ◾  97

It is critical that in these calculations, the same time units (days, weeks, etc.) be used for all variables. If there is bias in the forecast, efforts must be made to improve the forecasting process to reduce and then eliminate the bias. Forecast bias will cause you to underestimate or overestimate the safety stock needed. It must be emphasized that the PC/T1 factor is a statistical adjustment to approximate the standard deviation of demand over the time period of the performance cycle and is just an approximation. It gives reasonable results in cases where the performance cycle is greater than the data collection time period but can give very poor results going in the other direction, where PC is less than T1, especially when the time parameters are small, in going from weeks to days for example. If PC is much less than T1, you should try to measure demand variability or forecast error on a more frequent basis, to reduce T1 to a frequency closer to PC. Ideally, PC = T1 so that no adjustment is needed.

Seasonality If seasonality is a significant cause for demand variability, it should be recognized and used to periodically adjust safety stock to reflect the forecast demand during the various high and low periods. In other words, you are looking for the variability around the seasonal profile. If not recognized, and treated as normal demand variability, it could cause a very high level of safety stock while still not providing enough material to cover demand in the peak season.

Variability in Lead Time The equations in the preceding section calculate the safety stock needed to mitigate variability in demand or forecast error. If variability in lead time is of concern, the safety stock needed to cover that is:

Safety Stock  Z   LT  Davg

The average demand term (Davg) is in the equation to convert the standard deviation of lead time expressed in time units into production volume units (such as cases, gallons, pounds, rolls). Note that this equation needs no adjustment for the performance cycle.

Combined Variability If both demand variability and lead time variability are present, the safety stock required to protect against each can be combined statistically, to give a lower total safety stock than the sum of the two individual calculations. If

98  ◾  Scheduling Processes, Systems, and Software

demand variability and lead time variability are independent, i.e., the factors causing demand variability are not the same factors influencing lead time variability, and if both variabilities are reasonably normally distributed, the combined safety stock is Z times the square root of the sum of the squares of the individual variabilities:

Safety Stock  Z 

PC 2 2  D   2LT Davg T1

The reasoning behind this is that if the two variabilities are independent, it is very unlikely that demand extremes will occur at the same time as very long lead times. If σD and σLT are not statistically independent of each other, this equation can’t be used, and the combined safety stock is the sum of the two individual calculations.

   PC   Safety Stock   Z       Z   LT  Davg  D     T1   

Cycle Service Level and Fill Rate The equations in the preceding sections will predict the safety stock needed so that a certain percentage, say 95%, of the replenishment cycles will be completed without a stockout. This is often called cycle service level (CSL). Business leaders often want to control the percentage of the total volume ordered that is available to satisfy customer demand, not the percentage of cycles without a stockout. The former quantity is called fill rate and is often considered to be a better measure of inventory performance. Figure 9.9 illustrates the difference. Where cycle service level is an indication of the frequency of stockouts, without regard to the total magnitude, fill rate is a measure of inventory performance on a volumetric basis. The specific calculations of safety stock required to achieve a desired fill rate are very complex and beyond the scope of this chapter. An excellent discussion can be found in Chopra and Meindl’s Supply Chain Management. However, some observations are in order. With stable demand patterns and supply behavior (that is, low standard deviations of demand and lead time), fill rate will generally be higher than the cycle service level, as illustrated in Figure 9.10. Although stockouts will occur, with low supply and demand variability the magnitude of each stockout will be small. With high variability in either demand or lead time or both, the opposite will usually be found. Figure 9.11 illustrates a case where demand variability is high, where the standard deviation of demand is half of the average demand. Although there are few stockouts (because of the safety stock being

The Role of Inventory  ◾  99 1400

Finished Product Inventory

1200 1000 800 600 400 200 0 -200

1

7

13

55

64

-400

CSL reflects the FREQUENCY of stockouts, the percentage of cycles where stock-outs occur

FILL RATE reflects the AMOUNT stocked out, the percentage of Demand not met

Figure 9.9  Cycle service level and fill rate. 1200 1000 800 600 400 200 0 -200 -400

1

4

7

10

13

16

19 22

25

28 31

34

37 40

43 46

49

52 55

58

61 64 67

Stockouts will occur, but with low demand variability, material shorted will be very low So Fill Rate will exceed CSL

Figure 9.10  Inventory profile with low demand variability (CV = 0.2).

carried), the magnitude of any stockout can be quite high. Thus, in this case, the fill rate is actually less than the CSL.

Safety Stock and Lot Size Impact Although it is counterintuitive, the combination of lot sizing with infrequent demand often means that low-volume products don’t require safety stock as

100  ◾  Scheduling Processes, Systems, and Software 2000 1500 1000 500 0 1

7

13

19

25

31

37

43

49 52

61

67

-500 -1000

With high demand variability, there may be few stockouts, but the amount can be quite large! So Fill Rate can be lower than CSL

Figure 9.11  Inventory profile with high demand variability (CV = 0.5).

traditionally calculated per the previous sections. Lot sizes require making more production than ideal most of the time, since the production quantity must be rounded up to fit the lot size. Therefore, there is an overlap between the lot size impact and safety stock that should be accounted for. The overlap can be approximated by subtracting half of the lot size from the safety stock. Additionally, the minimum lot size for low-volume products often lengthens the production frequency, reducing exposure to service level and fill rate incidents. Even for higher-volume products, lot sizing can reduce safety stock requirements. Higher-volume products tend to have more stable demand. As shown in the previous section, more stable demand results in a lower fill rate impact during service-level disruptions. In combination, the rounding to lot size and lower fill rate impact may significantly reduce the amount of safety stock required to meet fill rate targets of stable products. While the interactions of safety stock, lot sizing, demand stability, production frequency, service levels, and fill rates can be analyzed statistically, the formulas are complex. Simulation may be the best way to account for all the interactions and provide confidence in the result. The better scheduling packages have this capability; if configured properly, they will automatically take all of these variables into account and predict the necessary safety stock level for each SKU.

Summary Inventory is created by what we produce, and what we produce is controlled by what we schedule, so any discussion of scheduling must include a focus

The Role of Inventory  ◾  101

on and an understanding of the inventory created. Any scheduling process, including the underlying strategy and the supporting tools, must have the goals of creating enough inventory to satisfy customer demand at the desired fill rates, create as little working capital as is practical, fit within the warehouse, and not require current production capacity limits to be exceeded. And yet inventory management often doesn’t get the attention it deserves because its importance to business success is often misunderstood. Inventory management sometimes comes down to two actions: ◾ The business decides that working capital is too high and dictates that inventories be cut by XX%. ◾ Marketing becomes concerned about less-than-ideal fill rates and dictates that inventories be increased by YY%. So inventory fluctuates between these emotionally set limits, a clearly unsustainable situation. Well-managed companies realize that determining the most appropriate inventory levels and then managing them is critically important to manufacturing effectiveness and to business profitability. If inventories are higher than ideal, excess working capital is tied up and the risk of damage, obsolescence, and product expiration increases. If total inventories are lower than needed, it may overtax the production facilities. Inventory of any material, be it a raw material, WIP (work in process), or a finished product SKU, consists of several elements: ◾ Cycle stock, needed to supply downstream demand when the facility is producing other products or between shipments of raw materials ◾ Safety stock, needed to maintain flow in the face of higher-than-expected demand or longer lead times ◾ Quality holds, material being held awaiting release for quality testing ◾ Staging inventory, material released for shipment but not yet loaded on a truck and dispatched ◾ Inventory produced beyond that needed, to fill batch size requirements ◾ Obsolete, slow moving, or dead inventory The most complex and least understood of these is safety stock. Safety stock can be calculated to meet a cycle service-level target or a fill rate target, and it’s important to understand that distinction. There are commonly cited statistical formulas that can be used to calculate the safety stock required to meet a service-level target, and while they are far better than rules of thumb, they are approximations. The calculation of safety stock needed to meet a fill rate target is much more complex and is best done by computer simulation.

102  ◾  Scheduling Processes, Systems, and Software

Once targets are set, inventory levels must be monitored against those targets. And when the inventory for any specific SKU or for the entire portfolio falls outside the expected bands, corrective action must be taken. This includes curtailing demand or increasing production capacity when inventories are too low and by curtailing production or increasing market share when inventories get too high.

Chapter 10

Typical Scheduling Process Steps The Planning and Scheduling Process This chapter is the first of three to discuss the scheduling process in detail. In this first chapter, we will talk about scheduling at a single operational level, as represented by the Blue Lakes packing areas. In the next chapter, we will address the complexity of scheduling multi-level or multi-step operations and how the scheduling process changes depending on the location of the operation’s constraints and how much freedom or inventory is allowed between the levels. The third chapter will cover additional complexities of tank management and specific flow paths between operations. As we discussed in previous chapters, a quality plan is a prerequisite to a good schedule. Therefore, we will start by reviewing the overall planning and scheduling cycle, as shown in Figure 10.1.

Figure 10.1  The high-level planning and scheduling cycle.

DOI: 10.4324/9781003304067-13

103

104  ◾  Scheduling Processes, Systems, and Software

Exception Management At the start of the scheduler’s day, the first thing that should be done is an exception review, sometimes called response planning. Not only does it make common sense to find and fix problems as quickly as possible, but response planning also sets the stage for the remainder of the planning and scheduling work. Otherwise, the new schedule will be built on a foundation that will change, making it obsolete before it even starts. Lead time, which includes reaction time, is a major factor in determining inventory levels. From our inventory discussion in the previous chapter, safety inventory is a function of the square root of the lead time. Reaction time is an important part of the lead time and something that the business can directly control through the work process and better planning and scheduling software. A focus on the agility to find problems and quickly resolve them will minimize inventory. In fact, reaction time and lead time are two of the few things not subject to the law of diminishing returns. Because of the square root function, the returns get greater the more that lead time is reduced. Figure 10.2 shows the relative costs of holding safety stock as response time gets shorter. It’s based on an example product with sales of 10,000 cases per week, a weekly CV of 30%, a cycle service target of 98%, a value of $25 per case, and an annual holding cost of 25%. The shape of the curve will be similar regardless of the assumptions for holding cost, demand, and case cost. Early in my career, our team developed a simulation of our production lines, distribution network, and customers which showed that no amount of

Value of One Day Faster Response

$16,000.00 $14,000.00 $12,000.00 $10,000.00 $8,000.00 $6,000.00 $4,000.00 $2,000.00 20

18

16

14

12

10

8

6

Response Time, Days

Figure 10.2  The increasing value of faster response.

4

2

0

$-

Reduction in Annual Holding Costs

$18,000.00

Typical Scheduling Process Steps  ◾  105

inventory could protect customer service as effectively as shorter production cycles and faster reactions to demand changes.

Preparing to Plan Preparing to plan is a last-minute check that everything is in order before starting the planning process. For example, is all the master data in place and accurate within the planning and scheduling horizon? Typical things to look for would-be new products that might have a planning bill of materials, but not one that has been approved for production. Has downtime been updated with the latest maintenance plans? There should be checks that production, receipt, and inventory transactions are up to date and synchronized, that all of the interfaces have run correctly, and that there is no missing or corrupted data that will prevent the calculation of the plan and schedule. In many systems, these checks are automated and will give a warning to the scheduler. For example, in SAP, exception groups 5 and 8 show that the MRP run has failed for a material. The cause of the errors should be fixed and the plan should be rerun for any products within the scheduling horizon before proceeding. If the production process is running significantly ahead or behind schedule, completion dates and times should be adjusted. However, this should not be micro-managed. There will be a normal variation in production rates and process stops. Unless the adjustments are automated, it may be more work and noise than necessary to continually adjust the times. When adjusting timing, it’s often best to focus on adjusting the start date and time of the next order to be run, rather than trying to change an order already in progress. From a practical standpoint, there may be some overlap between exception management and preparing to plan, but the end result should be that emergencies have been dealt with and that everything is in place and ready to start creating the plan and schedule. For example, if the line is running significantly ahead or behind schedule, this might have been dealt with in exception management, or it might be a routine adjustment that is captured in the prepare-to-plan step.

Creating the Production Plan The production plan is the foundation for scheduling. With a good plan, the schedule will fit. The scheduling task becomes one of choosing the best timing and sequence for the planned orders to maximize customer service and minimize changeover times and costs.

106  ◾  Scheduling Processes, Systems, and Software

Without a good plan, no amount of scheduling effort will make the schedule fit. Even the most advanced and expensive advanced planning and scheduling system will be unable to create a schedule that satisfies customer demand and is feasible to execute. The following are key considerations for a quality production plan to ensure that your scheduling and execution will be successful: ◾ The plan requires no more than the available capacity, resources, or materials and components, after taking all losses, such as changeovers, maintenance, team meetings, process improvement time, or scrap into account. ◾ Use realistic or demonstrated production rates for every product in the plan. ◾ All products required to be produced are included, in the correct quantities, and in the correct time periods. ◾ Agreed-upon demand from all sources is included. ◾ Inventory is at target levels at the end of each planning period. ◾ When multiple production levels are involved, lot size criteria are used at each level.

Creating the Detailed Schedule We will cover creating a detailed schedule later in this chapter.

Communicating the Plan Once scheduling is complete, the schedule must be communicated to the shop floor, maintenance, purchasing, and production operations. At Blue Lakes, communicating the plan would include the schedule for the packing lines, directions to mixing on the timing and quantity of mixes, and directions to the spice room and liquid prep that would alert them to thaw frozen ingredients and prepare the spices and liquids in time for mixing. It provides the officially released schedule to the lines, with the expectation that the work has been done in planning and scheduling so that they will be able to execute it as written. In other words, it fits within demonstrated capacity, places products on the right lines, avoids constraints of staffing, tanks, and other resources, and directs production to specific flow paths when necessary.

The Packing Line Schedule A typical packing line schedule gives the details of each run by the production line and its product code and description, start time, end time, and

Typical Scheduling Process Steps  ◾  107

quantity. It might include things like allergens, contaminants, and color information so that the lines can understand and prepare for the upcoming changeovers and cleanouts. It may include notes, for example, whether the run must be to an exact quantity or instructions on what the line should do when running ahead or behind schedule. For example, if the line is running poorly, should they cut the run short at the end of the allotted time or must they continue until the run is completed? Therefore, scheduling software should have the ability to capture product attributes and convey them in a flexible way to the operations, and the ability to write notes about orders. The notes should persist when orders are rescheduled and adjusted.

ERP and Shop Floor Systems Often, the communication of the schedule to the floor will be done by interfacing the orders and notes to the ERP or shop floor systems; however, other systems may lack the ability to convey the details of product attributes, processing notes, or other critical information. Therefore, it’s common to supplement the schedule in the ERP or shop floor systems with a direct schedule issued by the scheduling system, containing more information and notes.

The Mixing Schedule The mixing schedule at Blue Lakes could be created in different ways, depending on whether mixing is a constraint to the operation. If mixing is not a constraint, the instructions can be simpler and list the mix required, its attributes, the quantity to be mixed, the time that it is required to be ready for packing, and the lines that will be using it. They should include directions on quantities to be offloaded into totes for packet production or later production in bottles, and tubs. With this information, the mixing operation can use their experience to choose the best available tank and when to start each mix. When mixing is a constraint, the instructions would be more detailed about exactly when to start each mix and which tank to use.

The Spice and Liquid Prep Rooms Since the spice and liquid preparation rooms are not constraints, they would be given a view of the mixing plan for upcoming days, which would allow them time to prepare the kits of liquids and spices. To thaw the frozen ingredients requires four to seven days’ notice; therefore, this implies a scheduling commitment zone, since once ingredients are thawed, they must be used.

108  ◾  Scheduling Processes, Systems, and Software

Preparing for Tomorrow Preparing for tomorrow updates master data and parameters with the things the planners and schedulers have learned during the day. It includes longer term less urgent work on the strategy and parameters beyond the current planning window and managing the long-term plan.

The Detailed Scheduling Process In the case of Blue Lakes, during the summer outdoor cooking season, the demand for BBQ sauce and ketchup increases significantly, making the packing lines more capacity limited. For the rest of the year, packing has excess capacity, and it’s more efficient to optimize the mixes to minimize washouts and cleaning. The totes and tanks at Blue Lakes illustrate different amounts of freedom or inventory between the levels. The packets are filled from totes containing the mix. From the value stream map, we can see there are 120 individual 300-gallon totes that can be used to feed the packet lines; therefore, there is considerable flexibility to schedule the packet lines independently of mixing. The bottle, jar, and tub lines can be fed directly from the tanks. Additional work for handling and cleaning can be avoided by feeding them directly whenever possible. However, feeding directly from the tanks adds scheduling complexity. For example, a tank can only hold one mix at a time. There are 12 packaging lines but only 8 tanks; therefore, packing can’t run more than 8 mixes at a time directly from the tanks. The mixing must finish before packing can start. The tank must be emptied completely before the next mix is added, and emptying a tank quickly will free it up for the next mix. If this isn’t enough complexity, consider what would happen if the packing lines were distributed in two different buildings, and each building had its own independent mix system. Now it’s not enough to know that we need a thousand island dressing mix at a specific date and time, we also need to know which line will pack it, and therefore, which set of tanks to use. We will explore this in Chapter 12 when we get into the additional complexities of scheduling tanks and flow paths. The process of creating the detailed schedule is shown in Figure 10.3. This chapter will focus on the first three steps, scheduling the constrained process, evaluating the schedule’s Key Performance Indicators (KPIs), and adjusting the constrained schedule. In the next chapters, we will cover the multi-level processes of aligning other levels to the constraint and adjusting all levels as necessary to create a fully feasible schedule across all levels. The first step of creating the schedule is to sequence the orders on the constrained step of the process to maximize its efficiency or other Key Performance Indicators (KPIs). Since the constraint controls the throughput

Typical Scheduling Process Steps  ◾  109

Figure 10.3  The detailed scheduling process.

and performance of the entire process, the other steps will be aligned to the constraint after it has been scheduled. For a single-level process, its only level is the constraint. For multi-level or multi-step processes, the technique used to create the remainder of the schedule depends on the location of the constraint, the degree of freedom between the stages, and the possible flow paths through the system. We will discuss this in more detail in the next chapter. The scheduling problem generally falls into three broad classes: single-level processes or closely coupled multi-level processes, downstream constrained, and upstream constrained. Since the best sequence may be different across different stages, the degree of freedom determines whether the stages must follow together in lockstep, or whether there is enough inventory between the steps to provide some ability to schedule each level independently. For example, on the Blue Lakes packing lines, changing and adjusting the bottle size is more difficult and takes more time than an allergen, organic, color, or particulate change; Figure 10.4 shows the changeover priority matrix.

110  ◾  Scheduling Processes, Systems, and Software

Priority

Change

1

Size/Bottle

2

Organic/Allergen

3

Color

4

Particulates

5

Viscosity

6

Neckband

7

Label

8

Cap

Figure 10.4  Priority of changeovers on the Blue Lakes bottling lines.

Therefore, when bottling is the process constraint during the summer BBQ season, the strategy is to run all the variations in one bottle size, before changing to a different bottle and rolling through the dressing variations again. However, upstream, the mixing lines don’t have bottles, neckbands, labels, or caps; therefore, their schedule is driven by minimizing allergen, organic, color, particulate, and viscosity changes. Therefore, the ideal sequence for Blue Lakes is different when there is adequate capacity in packing and mixing is the constraint.

Scheduling the Constraint There are four basic options for creating the schedule at the detailed level. Manual scheduling, just-in-time scheduling, scheduling with a repeating sequence, and scheduling to maximize or minimize a set of KPIs.

Manual Scheduling We’ve witnessed many sites using Excel or ERP modules with graphical scheduling software that allows drag-and-drop sequencing of orders. Their drawbacks have been discussed in subsequent chapters on the role of ERP systems (Chapter 13) and scheduling in Excel (Chapter 14).

Just-in-Time Scheduling Scheduling just-in-time will minimize inventory, but at the cost of additional changeovers, since it is unlikely that order due dates will fall in the best sequence to maximize production efficiency. A just-in-time schedule will have implications for suppliers, since the demand for ingredients, components, or

Typical Scheduling Process Steps  ◾  111

raw materials will be less predictable. Either the just-in-time schedule must be locked in with sufficient time for suppliers to follow along, or the difference must be buffered with raw material inventory.

Repetitive Sequence Scheduling The repetitive scheduling strategies described in Chapter 6 generally offer the most efficient and cost-effective lot sizes and sequences. They also offer the advantage of developing the schedule once and then having to adjust only to accommodate changes in demand and unexpected events. They may, however, require more finished product inventory than other scheduling approaches.

KPI-Based Algorithms and Solvers On paper, a KPI-driven scheduler will create the best possible sequence. It will follow closely, but not exactly a repeating sequence, since they both have similar objectives. However, its sequence at any one time will be dependent on the orders in the queue; therefore, it won’t repeat from period to period. Some of the benefits of repeatability and learning curves in changeover and startup will be lost. It’s unlikely that a scheduler will be able to figure out in their head the best sequence at any one time for a group of orders unless their products and product attributes are extremely simple. Therefore, this strategy requires the assistance of sophisticated scheduling software. As we’ve covered before, scheduling is an extremely complex problem to solve, and the absolute best schedule is almost impossible to find. The goal is to be close enough or good enough. Optimizations are normally not suited for scheduling problems. Most KPI-driven scheduling software is essentially a rapid computer-assisted form of trial and error, moving orders to different places, assessing the KPIs, and then trying other combinations to see if they can find one that’s better. They will use techniques to narrow the search and home in on the best options to explore. The KPIs need to be carefully designed and tuned to get good results. We’ve seen cases where they have not been maintained, with the result that the scheduler mechanically runs the automated scheduling and then goes about inspecting and moving every order with drag and drop. The typical goal of a scheduling solver or algorithm is to minimize changeover times and costs, without running out of inventory, without running out of resources, and without exceeding upper inventory limits. The drawbacks are that the scheduler will often be able to find a place in the schedule that they can improve with a few simple swaps. They may not trust the results or understand how the system arrived at the best schedule.

112  ◾  Scheduling Processes, Systems, and Software

Change management can help to successfully implement software that relies on complex algorithms. The goal is to quickly deliver an automated schedule that’s good enough to achieve most of the business results and be executable in practice. The schedulers must be able to see the big picture, that overall, the schedule will be better, even if there are a few places that are not optimum. That the benefit of being able to quickly reschedule and react to disruption far outweighs any slight losses in other areas. They must be upskilled to be able to understand and tune the KPI algorithms, and to know when they need adjusting. Management must support and reward keeping the system maintained to produce good results with minimal intervention.

Resources A resource is something that is required by the production process but that doesn’t go into the product itself. Therefore, they are not consumed by the process and are sometimes called non-consumable resources. A typical example is staffing. We can define how many people are required to produce each product on each line. But once the operation finishes, the people are free to move on to producing the next item. Refrigerated finished goods storage is another example of a resource at Blue Lakes. Products requiring refrigeration go into storage as they are produced, and out as they are shipped. To start, we need to know which products require refrigeration, and the amount of space required for each one, based on some common unit like pallet spaces or cubes. We can then create a time-phased total of the refrigeration space requirements by adding production converted to refrigeration units and subtracting customer and forecast demand. We can compare the requirements to the available space in a continuous time series or a graph, visible during the scheduling process. Figure 10.5 shows an example of a production line schedule showing staffing or crewing resource requirements. The red cell shows crewing above a defined alert limit. The production schedule has been color coded by the crewing requirements. Products in purple require 10 people, and those in orange require 15. You may be asking why the schedule didn’t avoid the constraints in the first place. It may be that there is no schedule that can avoid out of stocks while respecting staffing limits. Since out of stocks have a

Figure 10.5  Example of a line schedule showing crewing requirements and limits.

Typical Scheduling Process Steps  ◾  113

higher weighting in this business’s KPI formulation, the assumption is that the business will increase staffing if necessary to avoid out of stocks. If staffing the KPIs were weighted more heavily, the algorithm would have arrived at a schedule that stayed within staffing limits but instead would have customer service issues. There is a risk in any KPI-driven algorithm or optimization that setting too many limits will result in an impossible situation that no schedule can satisfy; therefore, it’s important to give the program a safety valve, even if it’s at a very high cost, that will allow it to create a schedule that the planners can use as a starting point to manually adjust and modify.

Evaluating and Adjusting the Schedule As we’ve seen above with staffing out of limits, no schedule is likely to be perfect on its first pass through the scheduling strategy, whether manual, JIT, a repeating sequence, or KPI driven. Some compromises will need to be made. The goal is to find the issues in the schedule and either resolve them or communicate the issue to others so that they are not taken by surprise and can mitigate the impact. For example, advance notice of the high staffing would allow moving people from other areas of the site on a planned basis, bringing in contractors for the shift, or asking operators to work overtime. If the choice was to rearrange orders to stay within staffing limits, advance notice to customer service would allow them to negotiate delivery dates and quantities with customers. We will cover details of how specialized software can help with schedule evaluation and adjustments in Chapter 15.

Releasing Firm or Committed Orders The last step in the detailed scheduling process is to commit or release orders to production. The form that this will take depends on systems and processes, but the principle is the same. In some systems, it’s a change in the order’s status from a planned order to a production or process order. It carries the authorization to produce the item and begin the steps necessary for production. Prior to authorization, the schedule is just a projection, and might be changed. After authorization, production is allowed to take action. For example, Blue Lakes would start thawing ingredients. Once started, they must be used or scrapped. The authorization to produce feeds into the higher-level process of Communicating the Plan, which we discussed earlier in this chapter.

Chapter 11

Multi-Level Scheduling This chapter will discuss scheduling techniques for multi-level scheduling; the following chapter will cover additional complexities like tanks and specific flow paths between operations and equipment. The first consideration in multi-level scheduling is to decide which of the levels or steps in the process should be scheduled. For example, Blue Lakes has three levels of the process that appear in the value stream map, the spice room, and liquid prep, the mixing tanks, and the packaging operations. Starting upstream, the spice room and liquid prep can prepare whatever spices and liquids are needed as long as they receive the mixing schedule with enough advance notice to prepare everything for mixing. Therefore, they don’t need to be directly scheduled; they can be handled through a work process that gives them the mixing schedule in advance. However, as we’ve noted before, it requires four to seven days to thaw frozen ingredients, and once they are thawed, they must be used quickly. This puts a lower limit on the advance notice period and requires that the schedule be firmed and released at least seven days in advance. Moving downstream to the mixing decks: How much attention do we need to pay to scheduling them? At the simplest level, we know that there are only eight tanks and one hundred and twenty totes available to store the mixed dressings. Therefore, there must be a flow balance between the mixing operations and the filling/packing operations that keep the intermediate inventory greater than zero and less than the total volume of the totes and tanks. From an efficiency point of view, we will reduce handling labor and waste by pulling directly from the tanks instead of offloading them to totes. We will discuss the complexity of scheduling individual tanks in the next chapter. In this chapter, we will treat the tanks and totes as an aggregate inventory and flow balancing problem. At the downstream end of the process, the packet lines and the bottling lines produce the customer-identifiable finished product; therefore, they must

DOI: 10.4324/9781003304067-14

115

116  ◾  Scheduling Processes, Systems, and Software

be scheduled to respond to customer demand and avoid finished product inventory that would expire or exceed warehouse capacity limits.

Product Mix and Moving Bottlenecks The product mix changes with the seasons. The demand for ketchup and BBQ sauce peaks during the summer and causes packaging to become a constraint because of the additional products. Therefore, during the summer the process is downstream constrained, and the packing lines are scheduled first to maximize their throughput. To support them, mixing rotates more frequently through the formulas, making smaller batches, and incurring more washouts and cleaning, which increases the cost to produce but avoids lost sales. For the remainder of the year, the process is upstream constrained; mixing is scheduled first to maximize its efficiency and minimize cleaning while the packing lines follow along. The additional bottle changes aren’t consequential, because there is enough capacity in packaging, and no sales are lost. Larger mixing batches, longer continuous mixing runs, and fewer washouts reduce the amount of wasted ingredients, resulting in lower manufacturing costs. However, at any time, there could be periods where either level becomes the constraint. Sometimes, it’s impossible to optimize either level of the schedule, the best schedule in packing can’t be executed in mixing, and the best schedule in mixing can’t be executed in packing; therefore, the overall schedule is a compromise. Often this is referred to as a floating bottleneck, which will be discussed further in Chapter 26. The techniques in this chapter can help you to sort out floating bottlenecks. The general rule is to schedule the most constrained level first, then synchronize the other levels to the constraint, and finally compromise whenever synchronization to the ideal can’t be achieved. Downstream is the flow towards the customer, and upstream is the direction towards components, raw materials, ingredients, and suppliers (Figure 11.1).

Figure 11.1  Upstream and downstream.

Multi-Level Scheduling 

Types of Scheduling Problems Scheduling problems can be classified by whether they are single or multilevel, the degree of freedom or flexibility between the levels, and the location of the constraint. Levels of Production: ◾ Single level ◾ Multi-level Degree of Freedom: ◾ Unconstrained ◾ Loosely coupled ◾ Closely coupled Location of the Constraint: ◾ Upstream constrained ◾ Downstream constrained

Degrees of Freedom between Levels An example of a multi-level schedule with unconstrained intermediate inventory is a paper plant making tissue or towels. The papermachines make large parent rolls that are slit and converted to tissue and towels. Since the parent rolls are discrete, any combination can be stored, they can be supplied to the production lines whenever needed, and there is enough storage space to accommodate the most reasonable schedules on both the upstream paper machines and the downstream converting lines. Many pharmaceutical and nutraceutical manufacturers are similar, with intermediate products stored in racks, bins, and totes. Automotive paint production follows the same pattern, with resins stored in stainless steel totes prior to tinting. The unconstrained case is the easiest multi-level process to schedule, because the schedule for each level can be completed independently, but in practice much is given up by scheduling them independently. The tissue lines will run better when they are fed fresh paper, so the paper plant’s schedules should be coordinated between the levels. The same is true in the production of house wrap between forming and bonding. In the pharmaceutical or nutraceutical example, coordination between levels will reduce intermediate inventories, reduce lead times, and decrease the risk of shelf-life expiration. Blue Lakes is an example of a loosely coupled intermediate inventory. There are a fixed number of tanks feeding the tub, bottle, and jar lines, there are more packing lines than tanks, and only one mix can be in a tank at once. A schedule that requires too many mixes at once isn’t feasible

◾  117

118  ◾  Scheduling Processes, Systems, and Software

unless more mixes have been stockpiled in totes. A schedule that packs products too slowly will run out of tanks and totes to store the mixes. A schedule that plans the wrong mixes will not support customer service. And a schedule that runs too many packing lines at once will run out of mix. The bottling lines at Blue Lakes are an example of a closely coupled process. A line consists of several pieces of equipment: bottle unscrambling, air jet cleaning, bottle filling, capping, sealing, neck banding, labeling, case packing, and palletizing, all of which can run at different rates. The line should be scheduled based on the run rate of the slowest piece of equipment for each SKU. That may be bottle filling for the larger bottles, labeling for the smallest bottles, and carton erection for very small cases going to convenience stores.

Impact of the Constraint’s Location In general, downstream-constrained scheduling processes are the easiest to schedule, and upstream constrained are the hardest. Close coupling between the levels actually makes the scheduling problem easier, because the other levels must follow the constraint. In many cases, closely coupled operations with a clear constraint can be scheduled as if they are a single level. Upstream-constrained processes with some amount of freedom, but not enough that the levels can be scheduled independently, are the hardest; sometimes, the schedule that will work best for the upstream constraint is impossible to follow in the downstream operations.

More Than Two Levels When there are more than two levels, scheduling should still start from the constraint and then move upstream and downstream level by level. It’s when compromises need to be made that it gets more complex, since there are more levels to consider. As we’ve discussed briefly in Chapter 4 in our overview, a production plan that is workable across all the levels can greatly simplify the scheduling process by ensuring that the total production is feasible across all the levels and that only the sequence of orders at each level needs to be optimized. This is especially true as the number of levels increases. We will cover this in more detail in Chapter 21 on capacity planning.

Batch and Lot Size Restrictions Batch and lot size restrictions are common in the process industry. At Blue Lakes, the mixes are made in 300- or 600-gallon tanks. Formulas that specify the quantity of each ingredient, the mixing time, and the mixing conditions

Multi-Level Scheduling 

Figure 11.2  Batch rounding example.

have been tested and approved only for these specific batch sizes. Therefore, the downstream requirements from packaging must round to 300 or 600 gallons for each mix. The same mix may feed multiple packing lines; perhaps it goes into jars and packets simultaneously. On the packing lines, we want to make complete unit loads to minimize damage in storage and broken stack loss. Therefore, the challenge is to find a common denominator of pallet quantities for each unique finished product that will satisfy its customer demand, but total across all the packing lines into a multiple of 300 or 600 gallons of mix. An example of a frozen food plant’s schedule is shown in Figure 11.2. The upstream primavera sauce, material number 34, must be made in 3,000-kilogram increments. There are three packing materials using the primavera sauce within the same scheduling period, products 1143, 1133, and 1129. In total, they require 13,000 kilograms of primavera sauce, which isn’t an increment of 3,000. Therefore, one of the packing runs must be increased by 2,000 cases or reduced by 1,000 cases. The choice depends on whether one of them can be reduced while still satisfying customer requirements, or whether one can be increased without fear of expiration.

Distribution Rules Multi-level scheduling is influenced by how production and consumption are distributed. For example, the mix can’t be pulled from the tank and packed until the mixing is completely finished. Typical distribution options are at the start of the operation, at the end of the operation, specific points during the process, or evenly distributed. Let’s discuss this with a couple of examples. For the start of the operation, consider the liquids and spices that have been prepared for mixing. We can imagine several different scenarios that you might be familiar with from your own cooking. There are ingredients

◾  119

120  ◾  Scheduling Processes, Systems, and Software

that are added at the start of the recipe, and others that are added at specific points in the cooking or preparation process. There might even be some like frosting or sauce that are added at the very end, after cooking. However, for practical purposes at Blue Lakes we can consider that the liquids and spices must be available at the start of mixing. For the output of mixing, the mix isn’t ready for bottling until it has completed its processing and testing. However, in the pet food plant that we will discuss in the next chapter, pet food is continually placed into the bins by the extrusion process, and it’s ready for packing immediately. The intermediate pet food, called kibble, is pulled from the bins by the packing process, and the two can be operated simultaneously. This is an example of a continuous distribution. The diagrams in Figures 11.3 and 11.4 show the impact on intermediate inventory with different distributions of upstream production and downstream consumption. The most conservative rules are production receipts at the end of the upstream process, and consumption at the beginning of the downstream process. But they won’t reflect reality in a continuous process like the pet food plant described in the next chapter. Continuous processes should normally use a continuous production or consumption distribution.

Figure 11.3  Production distribution rules.

Figure 11.4  Consumption distribution rules.

Multi-Level Scheduling 

Logical Relationships between Levels Some multi-level scheduling problems can be solved by the application of logical rules and timing relationships between the levels. The most obvious is start to start, where at a minimum the downstream operation can’t start before at least one of the upstream operations. Finish to start is the case at Blue Lakes between mixing and packing. The mixing is made in batches, one tank at a time, and the dressing or sauce can’t be packed until the tank is finished and checked for quality. A finish-to-finish relationship might apply when the downstream equipment consumes the intermediate faster than the upstream can make it. Therefore, the upstream operation needs a head start, and the operations should finish together to allow the process to be cleaned and changed over to the next item (Figure 11.5). There may also be a time dimension to the relationship. For example, the upstream operation needs a 30-minute head start to convey the product through the plant to the downstream. Or on a finish-to-finish relationship, it might not be critical that all the operations end at once, but there could be a policy that the system should be cleaned and turned over within three hours of the upstream finish. The location of the constraint may make a difference in the application of logical relationships. When the process is upstream constrained, it may be more appropriate to use a start-to-start relationship to avoid blocking the upstream operations and losing capacity on the constraint. In this case, it’s not a problem if the downstream operation shuts down periodically due to starvation. When the process is downstream constrained, a finish-to-finish or finishto-start relationship may be more appropriate. This gives the upstream a head start and establishes a buffer inventory between the operations, to avoid losing capacity downstream if there is an upstream upset.

Figure 11.5  Logical relationships between levels of production.

◾  121

122  ◾  Scheduling Processes, Systems, and Software

Linking between activities Some scheduling software allows activities to be linked, often with the specific precedence and timing relationships discussed above. Linking between activities has advantages and disadvantages. If one part of the linked operation is moved, the others automatically move with it, and other operations will flex around them, or be bumped earlier or later, while also maintaining their linkages and relationship rules. When it works correctly, it can minimize the work to realign production when the schedule is adjusted. However, sometimes, moving linked activities is like a chain of dominoes. Moving one activity causes a cascade, and the whole schedule explodes. In practice, we’ve often found that it’s best to create the constrained schedule first, without any linking enabled to other levels. Then activate the linking to align the schedule between levels and follow the precedence rules between the operations. If the constraint needs to be rescheduled, it’s often best to deactivate the linking before doing this.

The Multi-Level Scheduling Process To start our discussion of multi-level scheduling, the easiest multi-level schedule is one that is downstream constrained with enough intermediate inventory storage space that the stages can be independently scheduled. For these cases, we start by scheduling the downstream operation and then scheduling the upstream operations using dependent requirements and MRP logic, working our way upstream to the last operation closest to the incoming ingredients or materials. We are using the term MRP broadly; the calculations can be performed by MRP or ERP systems, other supply chain planning software, or Excel. MRP logic multiplies the scheduled downstream quantities by their BOM explosion factors for the intermediates and sets due dates that meet the downstream requirements. The MRP calculations will not result in the most efficient upstream sequence, but this is OK since the upstream operations are not the constraint. Less sophisticated systems may only resolve planning to the daily bucket level. If not careful, this could lead to the upstream being scheduled at 15:00 when the downstream requirement is at 08:00. To compensate, a time buffer could be added to the MRP logic, for example, to make the upstream orders one day in advance of the downstream requirements. If a more sophisticated scheduling system is available, logical precedence rules or drag-and-drop scheduling can be used to make sure that the upstream orders are produced in time for the downstream requirements. When upstream constrained, the first step is still a planning run using MRP logic or a planning Bill of Materials to transfer customer demand through the

Multi-Level Scheduling 

levels of production up to the constrained level. In the Theory of Constraints language, this is Drum, Buffer, Rope. This is an analogy for translating enditem demand to the constraint. The demand for the end items is the Drum, and tying the constraint’s schedule to the end items is the Rope. Next, the requirements suggested by MRP must be grouped and sequenced to create the most efficient schedule at the upstream constraint, as covered in the previous chapter. Then the downstream operations must be aligned to the constrained upstream schedule, and the schedule at the last stage must be checked to see that it still meets demand. As the Theory of Constraints states, the output or throughput of the constraint governs the entire process. Therefore, the planning and scheduling task is to optimize the throughput of the constraint, making sure that its output isn’t wasted by an inefficient changeover sequence, or by producing unneeded products and creating unnecessary inventory. It will be helpful to have logic that groups the upstream requirements during their MRP run in a way that matches their production strategy. For example, at Blue Lakes, if our mix production cycle was weekly, we might automatically combine all the orders for the same mix within the week, and round them to 300- or 600-pound batches. Unfortunately, simple MRP logic won’t ensure consistent rounding across levels. As we discussed earlier in the chapter on batch and lot size restrictions, all the levels must round to their own lot sizes, or else things won’t add up and there will be intermediate products left over at the end of the run that must be carried over or scrapped. The scheduler may have to make manual adjustments in run quantities for consistency. The best planning optimization software can calculate consistent quantities and timing across all the levels, round them to batch and lot sizes, and substantially reduce the amount of work to be done in scheduling.

Scheduling with Inventory Constraints between Levels Specialty scheduling systems provide additional capability for coordination between production levels compared to Excel or ERP/MRP systems. Their visual nature and continuous timeline will show the relationship between the upstream and downstream items. Intermediate inventory can be calculated at any point in time with checks and warnings to highlight issues when the downstream is planned before the upstream, when inventory is outside of high or low limits, or when the flow rates between the processes are unbalanced. They can enforce the logical rules that we’ve talked about earlier in this chapter. Figures 11.6–​11.8 show a simplified view of coordinating upstream and downstream production using one upstream line and one downstream line, showing the instantaneous graph of the intermediate inventory.

◾  123

124  ◾  Scheduling Processes, Systems, and Software

Upstream Run

Upstream Line

Intermediate Inventory

Downstream Line

Packing Too Early Intermediate Inventory Below Zero

Packing Run

Figure 11.6  Downstream operation planned too early.

Upstream Line

Max

Upstream Run

Intermediate Inventory Over Max

Zero Downstream Line

Packing Run

Packing Too Late

Figure 11.7  Downstream operation planned too late.

Upstream Line

Max

Upstream Run

Intermediate Inventory

Zero Downstream Line

Packing Run

Figure 11.8  Properly synchronized upstream and downstream production.

Chapter 12

Tanks, Bins, and Flow Paths This chapter will discuss some of the key steps and features necessary to coordinate tank or bin storage. In many multi-level processes, aggregate inventory visibility, resource constraints, and logical relationships between the operations will be sufficient to create a good schedule that is executable across all the levels and meets the customer’s required quantities and due dates. However, some operations with storage tanks, bins, and specific flow paths between the operations will require more detail and visualization of the flows between the operations and the inventory in each tank or bin at any point in time. Blue Lakes has enough tanks, and the ability to offload their contents into totes, so the scheduling of the mix tanks themselves is unlikely to be necessary. However, Blue Lakes has a pet food division, with a plant in Kingston, Ohio, to make dry pet food, close to the suppliers of its main ingredients of meats and grains. Two processing operations called extrusion make the pet food kibble and place it into four large bins. Eight packing lines draw directly from the bins to produce the finished pet food. There is no provision to offload the bins into totes or other storage. Unlike the Blue Lakes PA plant, where a batch must finish before it can be used, packing at Kingston can start as soon as enough product is in a bin to feed the lines (Figure 12.1). With only four bins and eight lines pulling intermediates, the synchronization between kibble extrusion, the products in each bin, and the packing lines pulling from each bin must all be carefully coordinated. Each bin can only hold a few hours of production from one extruder, and the typical extrusion run creates far more volume than a bin can hold. The kibble has different formulas for cats and dogs, different kibble sizes, and variations of ingredients, for example, corn, rice, beef, lamb, or chickenbased. Packages come in small, medium, and large sizes, and since the product is exported, there may be several language variations of each product and size.

DOI: 10.4324/9781003304067-15

125

126  ◾  Scheduling Processes, Systems, and Software

Figure 12.1  The Kingston pet food plant.

There is a best sequencing strategy for the major constraint of kibble extrusion, starting with a weekly cycle of each meat and other main ingredients. It requires a complete cleanout of the system between them, as some pets will be allergic to some of the ingredients. This drives the site’s production schedule. Each kibble size will be produced as it goes through the ingredient cycle. Packing follows the extrusion cycle. Since the packing equipment is less capital intensive than the extrusion process, the plant is designed with excess packing capacity; however, the packing lines must still be carefully scheduled to synchronize with the extruders and meet consumer demand for the different packing sizes and language variations. Collectively, the usage of the packing lines pulling from each bin can’t exceed the rate at which the upstream extrusion process makes kibble or else they will be starved and will have to shut down, but they must draw the kibble fast enough that the bins won’t overflow, or else the extruder will be blocked and will shut down. At the end of a kibble run, the bin must be completely emptied. It’s a tricky flow-balancing problem. Several years ago, Kingston coordinated kibble production and packaging using a complex Excel spreadsheet. The planner would spend most of the day creating a schedule for the next 24 hours. But because of the simplifications required to represent the Kingston process in an Excel spreadsheet, it was common for the extrusion process to shut down because the bins were overflowing or not emptied in time for the next extrusion product, and for the packing lines to shut down because too many of them were trying to run at once using the same kibble formula. Recently, Kingston implemented an advanced planning and scheduling system with the capability to correctly model the full complexity of their operation, with a continuous schedule that evaluates the bin inventory at any point in time and an optimization layer that recommends coordinated quantities of each kibble and each packaging variation to satisfy customer demand. After implementation, the effort to produce the schedule each day is less than an hour, the kibbling process and packaging operations seldom stop due to blocking or starvation, and efficiency has increased by about 5%.

Tanks, Bins, and Flow Paths  ◾  127

Tank and Bin Scheduling A tank or bin is a storage location with some special characteristics. For the remainder of this chapter, we will use the terms tank and bin interchangeably. The same concepts can also be used to define things that are not a tank or a bin. For example, if it is necessary to carefully manage the in and out flow, the refrigerated storage space at the Blue Lakes salad dressing plant could be scheduled in detail using tank management. The steps for configuring a tank are as follows: ◾ Define which products or materials the tank or bin is allowed to hold. ◾ Define the maximum capacity of the tank or bin, in common units across all the materials that are allowed in it. ◾ Set high and low warning limits. Most specialized software with tank management features will allow setting warning limits for high and low capacity. ◾ Define whether the tank or bin can hold only one product or multiple products at the same time. It may seem counterintuitive that a tank can hold multiple products at the same time. For example, we wouldn’t want to mix different salad dressings at Blue Lakes in the same tank. However, we’ve used tank management to represent the constraints of rack storage areas that held many different intermediate products for repacking, and conveyor belts that could be overloaded with certain production mixes. ◾ Define which upstream processes or lines are allowed to put products into the tank. ◾ Define which downstream processes or lines can pull products from the tank. Typically, specialized software will allow you to define the tanks or bins as above and assign upstream runs to feed the tank and downstream runs to pull from the tank. It will track and display a continuous inventory of the tank’s contents, with warnings if high or low limits are exceeded. Schedules can be adjusted to keep the tank within limits by drag-and-drop or scheduling algorithms.

Tank Scheduling Example A simplified tank or bin scheduling example is shown below. One upstream extruder is feeding a bin, and two downstream packing lines are pulling intermediate products from the bin. The scheduler’s task is to arrange the upstream and downstream production to match the flows in and out of the tank or bin to keep inventory above zero and below maximum capacity. Figure 12.2. shows the downstream packing production scheduled too early, with the packing line being starved and forced to shut down.

128  ◾  Scheduling Processes, Systems, and Software Packing Too Early, Bins Below Zero 00:00

Extruder

01:00

Packing Line 2

03:00

04:00

Chicken Rice

05:00

06:00

Chicken Corn

Below Zero

Bin 1 Packing Line 1

02:00

Below Zero

Large Chicken Rice

Large Chicken Corn Small Chicken Corn

Small Chicken Rice

Figure 12.2  Downstream packing too early. Packing Too Late, Bins Overflow 00:00

Extruder

01:00

02:00

03:00

Chicken Rice

04:00

Packing Line 1

Over Max

Large Chicken Rice

Packing Line 2

06:00

Chicken Corn

Over Max

Bin 1

05:00

Large Chicken Corn

Small Chicken Corn

Small Chicken Rice

Figure 12.3  Downstream packing too late. Balanced Flow 00:00

Extruder

01:00

02:00

Chicken Rice

03:00

04:00

05:00

06:00

Chicken Corn

Bin 1 Packing Line 1

Large Chicken Corn

Large Chicken Rice

Packing Line 2

Small Chicken Rice

Small Chicken Corn

Figure 12.4  Balanced flow.

Figure 12.3 shows the packing line scheduled too late; the bin will overflow and block the extruder, and the extruder will shut down with no place to put its intermediate product. Figure 12.4 shows a balanced flow into and out of the tank; the tank inventory stays between the minimum and the maximum level.

Tanks, Bins, and Flow Paths  ◾  129

The task is more complex when multiple processes are feeding multiple tanks or bins, and multiple downstream operations are pulling from the tanks and bins, but the thought process is the same. Balancing upstream and downstream flows with tank limit warnings will help to guide you.

Specific Flow Paths The techniques for modeling specific flow paths are similar to tanks and use many of the same concepts. Similar rules can be used to describe which operations are allowed to feed the flow path and which operations are allowed to consume from it. Figure 12.5 shows the packing stages of a site that makes tissue and towel products in two different store shelf sizes, a small count and a large count. The paper-converting lines, Line 1 through Line 5, make rolls of tissue and towel finished products. Small counts are shipped with the tissue and towel shelf units packed into cardboard cases. For large counts, the same tissue and towel units are bundled together and overwrapped with a large plastic bag, and the large bag becomes the shelf unit. Unlike the Blue Lakes example of packing in the previous chapter, each line does not have its own dedicated packing equipment.

Converging Flows Within this process are examples of flows that can converge; for example, Line 1 and Line 2 could both be running into Wrapper 1. This would require that both are making the same base product and that the rate of Wrapper 1 can handle the output of both lines. You will find other examples of flows that can merge: Lines 2 and 3 can both run into Wrapper 2. Lines 4 and 5 can run into Wrapper 3.

Figure 12.5  Specific flow paths in a paper plant.

130  ◾  Scheduling Processes, Systems, and Software

Lines 4 and 5 can both run into Case Packer 3, but the case packers do not have sufficient speed to handle the output of two converting lines. Therefore, this is not possible unless Line 4 diverts some of its output to Wrapper 3.

Diverging Flows There are also examples of flows that can split or diverge. Line 2 can feed both Wrapper 1 and Wrapper 2, which would allow making two similar SKUs from the same basic tissues and towel rolls in different languages or package count variations. While this was occurring, Line 1 would run to Case Packer 1, and Line 3 would run to Case Packer 2 to make products packed in cardboard cases.

Before and after APS Implementation Before implementing an advanced planning and scheduling (APS) system with the ability to encode the flow paths into its master data, the scheduler only had the view shown in Figure 12.6 in their ERP system’s scheduling module. They had to remember all the restrictions and attempt to avoid them. For example, they had to know not to schedule Lines 4 and 5 to run two products requiring cardboard cases at the same time. It would either be completely infeasible if they were different products or partly infeasible if they were the same product unless they also adjusted the rate or diverted some of the flow into a large count wrapped product. In the ERP scheduling system, the rate differences were handled by complex routings or recipes. For example, there would be an unrestricted routing for cardboard case products when Line 4 or Line 5 fed Case Packer 3 on their own, and a restricted rate routing when the lines fed Case Packer 3 together. The scheduler would have to manually choose which routing to select, and manually divert part of Line 3 or 4’s production to a routing that used Wrapper 3 when merging flows. After implementing an APS system, they were able to model and encode the physical flow paths between the operations. When the schedule put too ERP Graphical Schedule Line

Descripon

12/1

12/2

Line 1

Packing Line 1

12345

23456

45678

56789

Line 2

Packing Line 2

10123

23456

80234

56789

Line 3

Packing Line 3

Line 4

Packing Line 4

50123

Line 5

Packing Line 5

50123

20123

12/3

12/4

56123

80234

40345

62789 30234

Figure 12.6  ERP system view of the same process.

12/5

12/6

12/7 67890

34567 71234 90123

50678 60123 60123

Tanks, Bins, and Flow Paths  ◾  131

much load on a packing system, they received capacity overload warnings, prompting them to divert some of the production to other case packers or wrappers. The typical configuration for flow paths in an APS system is similar to tanks and bins; it defines: ◾ Which upstream lines can feed the flow path. ◾ Which downstream lines can receive products from the flow path. ◾ Which products the flow path can convey. ◾ Whether the flow path can convey multiple products at the same time.

Simplifying the Complex The common theme of tank management and the specific flow paths is that the rules of scheduling can be very complex. We’ve toured paint plants where milling operations fed dispersion tanks, which fed intermediate storage tanks, and then filling lines. Not everything was interchangeably connected, and each operation and each pipe connecting each operation had a color wheel to follow, or else a major cleanout was required. If both parts of a two-part catalyzed mix were sent through the system sequentially, without two other products in between, the entire system would solidify, requiring a major rebuild. A system that can represent and encode the physical reality in the scheduling system will allow the planners to see the consequences of their decisions, prevent them from scheduling combinations that are impossible for the operation to follow, and free them from having to remember the rules. This can be especially important for business continuity. It makes it possible for a less experienced or substitute scheduler to cover vacations, vacancies, and absences. A well-constructed Value Stream Map will illustrate the connections and limitations between operations. Specialized scheduling software with tank management and flow path features will allow the scheduler to encode the rules in the system. This makes it more likely that the schedule will follow the rules and be executable on the shop floor. A note for process improvement. There may be changes that could be made to the plant layout to make the scheduling process simpler and increase the site’s agility, efficiency, and throughput. For example, adding some additional wrapping machinery to the tissue/towel plant would improve agility and simplify operations. We hope that this book may let you see some of the possibilities and understand the business case.

Chapter 13

The Role of ERP Systems in Planning and Scheduling Most businesses have an ERP (Enterprise Resource Planning) or an MRP (Manufacturing Requirements Planning) system. In this chapter, we’ll use them interchangeably. If used properly, the ERP system can set the stage for successful scheduling. It can be the system of record for master data and transaction data and perform the tasks that a scheduling system is not equipped to do, things like quality management, lot tracking and tracing, finance, regulatory compliance, and purchasing. If its master data and demand are properly managed, it can create regular repeating orders to support a repetitive scheduling strategy. However, ERP systems have some critical limitations for scheduling. They typically assume infinite capacity, their lowest resolution for inventory planning and alerts is at the daily level, and they consider every product and intermediate product to be independent.

Assumption of Infinite Capacity ERP systems usually assume infinite production capacity when they create their plans. This is not necessarily a bad thing; an infinite capacity plan tells you what you should be doing in the ideal world. However, it will create problems for procurement and scheduling if not managed well. Most ERP systems have a planning time fence, a boundary where only the planner can create or change orders. Outside the planning time fence, the ERP system is free to create and change orders as needed to satisfy demand. The planning time fence should correspond to the period in which the scheduler will take control of the orders, as we covered in Chapter 4 in our discussion of planning and scheduling time zones.

DOI: 10.4324/9781003304067-16

133

134  ◾  Scheduling Processes, Systems, and Software

With an ERP system, whenever demand is not managed to be capacity feasible, it will create problems for procurement and scheduling. The schedule can’t use more than the available capacity, or else it won’t be feasible to execute on line. But since the ERP system is not constrained by capacity, whenever demand is higher than what is scheduled, the ERP system will continue to recommend new orders to fulfill the unmet demand. The new orders will accumulate at the planning time fence, with messages to move them earlier, since the ERP system can’t create new orders inside the planning time fence. In this situation, the schedulers will need to look through every order at the planning time fence, to decide which ones to produce and which ones not to produce, since there isn’t enough capacity to schedule them all. The ERP system will be recommending purchase orders for materials, ingredients, and components to support the orders that are accumulating. However, only some of them will be produced. If procurement is automated, orders will be sent to suppliers for materials that can’t be used, increasing working capital, tying up storage space, and risking obsolescence. To prevent overordering, purchasing will have to somehow determine which products won’t be produced, link them to the appropriate materials, and make sure these materials are not overordered. Note that just deleting the infeasible orders in this situation does no good. The ERP system will simply put them back, just outside the planning time fence, with messages to move them earlier. The right way to manage this is with a Sales and Operations Execution process that balances capacity with demand and constrains demand when necessary to be capacity feasible. An example of a business that did this well is a food manufacturer that we know. Recently, many food manufacturers in North America have been limited by the availability of starch. This company managed the shortage by prioritizing their customer orders and constraining them by the amount of starch available. They only released demand to ERP and scheduling that could be covered by the starch that they expected to receive.

Daily Time Resolution Typically, the inventory levels are only calculated at the daily level in an ERP system. A mismatch within the day, for example, a shipment scheduled for 08:00 and production at 15:00, would not be visible in the inventory calculations or warnings. Neither would a multi-level schedule with upstream production later than the downstream requirements.

Assumption of Independence An ERP system’s assumption of independence can be illustrated at Blue Lakes; by the way it would calculate bottling runs for finished products. An

The Role of ERP Systems in Planning and Scheduling  ◾  135

ERP system will recommend production for each finished product when it breaks its safety level. It will round the run to the finished product’s lot size criteria. In the case of Blue Lakes, the finished product rounding is to pallet quantities. The ERP run will not consider that all of the finished product runs for the same dressing mix, for example, thousand island dressing, whether in packets, bottles, or tubs, must round to mix batches of 300 or 600 gallons. It will not consider that all the thousand island orders for the week should run at the same time. The best multi-level planning and scheduling systems will have these calculations built in. At the planning level, they can enforce consistent batch sizes at all levels. In scheduling, they can automatically link related runs of upstream intermediates and downstream products, alert the scheduler to batch rounding issues, and allow the user to easily adjust either the upstream or the downstream quantities to meet the batch requirements. Beyond the rounding to batches, there are additional economies by grouping similar mixes, for example, by organic and not organic, allergens, and colors. The typical ERP system lacks the logic to consider that similar items should run together and should be recommended with the same timing.

ERP Scheduling Modules Some ERP systems have scheduling modules that can resolve the detailed timing of the schedule, but in our experience, there are complex scenarios where they fall short, or usability issues, especially in multi-level planning and tank management, which were covered in Chapters 11 and 12. An example of usability in a typical ERP system is the visibility of product attributes during scheduling. For example, at Blue Lakes there are eight attributes that affect bottle sequencing. However, many ERP systems can only display the material’s number and description, or perhaps a few attributes. Our experience is that it usually takes seven to eight attributes to understand which products should be scheduled together and to calculate the changeover difficulty. With most ERP scheduling modules, the scheduler must remember the product attributes and picture the resulting changeovers in their head. The best scheduling systems will allow visibility of an unlimited number of product attributes. They can automatically calculate changeover times and costs based on attributes and color code the schedule by the attributes to help the user to understand the schedule. We’ve worked with many companies that have ERP systems, yet they use Excel spreadsheets for their planning and detailed scheduling instead of their ERP system’s features. In most cases, you will find complex Excel spreadsheets at any site with multi-level production unless they are using specialized scheduling software suited for multi-level scheduling. For reasons that we will cover in the next chapter, Excel is not a satisfactory solution either.

136  ◾  Scheduling Processes, Systems, and Software

Repetitive Scheduling in an ERP System Period lot size strategies in an MRP system can create orders for repetitive scheduling when configured with the right lot sizes, frequencies, and inventory targets. Most ERP systems will have period lot size logic, which will create a series of evenly spaced orders at a selected frequency; for example, every day, every three days, every week, every two weeks, or every month. At Blue Lakes, if we wish to run all of the Italian dressing mixes together, we might decide to run Italian every week, because it is a big seller, but Italian Parmesan only every second cycle because of its lower volume and the additional cleanout that its dairy component adds. We would set a period lot size of one week or seven days for Italian and two weeks or fourteen days for Parmesan Italian. The logic for a period lot size within a planning system is as follows: ◾ Find the first time period that inventory will reach its minimum. Create an order to produce the product in this time period. ◾ Calculate the order’s quantity by looking ahead for the length of the period lot size and summing the requirements. For example, if the lot size is seven days, sum the next week’s forward requirements. ◾ Round the order as needed to meet the material’s minimum lot size and rounding parameters. ◾ Find the next time period in which inventory will reach the minimum and calculate the order quantity per the rules above. ◾ Continue each time inventory is projected to reach the minimum for the remainder of the planning horizon. Close readers of this logic will see that it doesn’t guarantee a plan that makes any product on a specific frequency, it will come close, but the rounding-up effect of lot sizes can cause the interval to be extended. Since it looks for when inventory reaches the target, the start of the cycle may not be at the beginning of a week or on a specific day, and each product’s cycle will be calculated independently. When demand is stable, the lot size itself can also be used to approximate an order frequency. With stable demand, a lot size calculated to equal a week’s average demand will result in orders spaced roughly every week. Inventory targets are important in a repetitive scheduling strategy and should be encoded in the ERP system. With insufficient safety stock, you won’t really have a repeating schedule. To maintain customer service at target, you will be continually breaking into the repeating sequence to make products that are at risk of out of stocks. We discussed this in Chapter 9 on the role of inventory.

The Role of ERP Systems in Planning and Scheduling  ◾  137

Quality Management To respond to usability issues and unique scheduling requirements due to manufacturing equipment configuration, scheduling systems tend to go through a faster development cycle than ERP systems. In cases where a business must be in compliance with Good Manufacturing Practices (GMP) or similar validation, requiring that the scheduling system be validated will put unnecessary roadblocks in the way of improving its usability and efficiency. Therefore, all transactions requiring validation should take place in the ERP or other validated systems, and not in the scheduling system. The creation of a schedule is not something that regulators see as requiring validation. Therefore, the scheduling system itself does not need to be validated, if care is taken in the interface to the ERP system to be sure that the scheduling system never performs a task that requires validation. This can be handled by distinguishing between orders that are being planned and scheduled but are not approved for execution, from those that have been released to the shop floor for execution. For the planned and scheduled orders, a scheduling system can have free reign, as long as care is taken in the ERP configuration to prevent such an order from ever being produced. When an order is approved and released for production, the scheduling system should be restricted in the changes that are allowed. Typical tasks that should only be performed in a validated system are assigning and tracking batch numbers and selecting the recipes, bills of material, or formulas to be used in production. Tracking data during the batch’s production should be performed in a validated system and not in the scheduling system.

System of Record A fundamental principle is that there should be only one source of data. The ERP system can be this system of record, the single source of truth for master data and transaction data. The scheduling system should pull its data from the ERP system, create and visualize the schedule in ways that are not possible in most ERP systems, send the completed schedule back to the ERP system for execution, and performing the tasks that require validation.

Chapter 14

Excel as a Finite Scheduling Tool Many companies use Excel for production scheduling. Some believe that it is effective, while some realize that it’s cumbersome and ineffective but don’t know where to turn. Compared to specialized software, Excel is inefficient for both the scheduler’s time and the company’s production resources. Dependency on Excel for scheduling exposes a company to continuity risk. In Chapters 10–12, we talked about the process of production scheduling. It’s almost impossible to build all of the features required to perform them efficiently into an Excel model. So why do people still continue to use Excel for production planning and scheduling? We believe that it’s a combination of several factors. As we discussed in the previous chapter on the role of ERP systems, most ERPs lack some of the critical features needed for detailed scheduling, for example, the resolution of the schedule and its sequence within the daily bucket and the coordination of schedules between product families and across stages of production. Even when the features may be available in the ERP program, they may be difficult to use, or have not been properly configured. Often, we find that master data and transaction data have not been maintained with the quality necessary for detailed scheduling in the ERP system, therefore the scheduler does it themselves in Excel, outside the ERP system, where they have control of the data and can enter the correct values. There may be a lack of knowledge about the availability of better scheduling software and the need for its features, or a lack of understanding by the management of the importance of scheduling and its contribution to the bottom line. There is a reluctance of management to pay for proper scheduling software, often because they don’t understand the critical role of scheduling and its impact on their operations. Unfortunately, many of the costs of inefficient

DOI: 10.4324/9781003304067-17

139

140  ◾  Scheduling Processes, Systems, and Software

schedules are not visible unless one is looking for them, and they are often attributed to other causes. In Excel, a planner can spend all day making a poor plan. For example, at a large pet food manufacturing site, the scheduler took most of the day to create the schedule in Excel. Pet food making is a two-stage operation; a large processing unit makes kibble and puts it in bins, and packing lines pull the kibble from the bins and pack it into different sizes and language variations. It’s a tricky flow balancing problem; the schedules created by Excel frequently overfilled the bins causing the upstream processing unit to shut down, or starved the bins forcing the packaging lines to shut down. This is similar to the cereal packaging bottleneck described in Chapter 26. By implementing specialized scheduling software, the pet food scheduler could create schedules that would hold up in practice, by 11 am. Production efficiency increased, and scheduling effort was reduced by 30%. Powdered detergent is similar to pet food in its scheduling characteristics. A large detergent plant in the Mideast saw the same benefits in planner productivity and reduction of unplanned stops, increasing production efficiency by approximately 5%. Think about how much money a 5% production increase could be worth to your site or business. We hope that this book helps, but the seemingly high cost of specialty planning and scheduling programs and the inability of management and IT to understand how important usability and features are to a planner and scheduler make the approvers reluctant to purchase proper software. The general availability of Excel means that the planner or scheduler doesn’t have to ask anyone to buy or develop the features that they need. However, the use of Excel for scheduling comes with a number of drawbacks: ◾ Business continuity and the ability of others to understand the scheduling spreadsheet. ◾ Prone to error, hard to find or correct an error. ◾ Not good for time-phased planning. Businesses have differing lead times for different materials, sometimes in hours, sometimes in days. ◾ Scheduling takes a long time versus specialty software. ◾ Difficult to visualize inventory, capacity, and sequence simultaneously. ◾ Manipulation of the schedule by trial and error in blocks of time versus a continuous timeline. ◾ Trial-and-error versus auto-generated schedules.

Business Continuity For business continuity, the typical scheduling spreadsheet is complex, with multiple sheets, macros, and lookup functions. No one except the person who wrote it can completely understand it. A new scheduler must either write

E xcel as a Finite Scheduling Tool  ◾  141

their own spreadsheet or continue to use one that they don’t fully understand. I have been there. I had a complex Lotus 123 (before Excel) threedimensional macro-driven spreadsheet that planned 70 diaper lines. I trained my replacement to use it; a month later when I checked in on them, they had scrapped my spreadsheet and developed their own. It’s difficult to train a substitute scheduler for vacation or absence coverage. Planners are stressed because they are afraid to let the organization down if they take time off. I’ve been there too, I once had to cover the medical leave of the company’s planner for diaper packaging. I figured out that I only cost the company a couple hundred thousand dollars in extra packaging obsolescence during my substitute time. We did work last year for a plant making mouthwash and other personal care products. They had been looking for something to replace their Excel scheduling workbook; the manager explained to us that their schedulers often did things unintentionally with very expensive hidden consequences.

Critical Features of Scheduling Software Let’s look at some of the critical things planning software needs to do in order to be effective and consider how well Excel stacks up against them. Some typical features of a graphical drag-and-drop scheduling program are: ◾ Alerts when capacity or component materials are not available ◾ Alerts for low or negative inventories ◾ Provides product attribute visibility ◾ Calculates changeover times based on from and to product attributes. For example, allergens, colors, contaminants, bottle and cap sizes, or case packer rails ◾ Automatically places orders in the best sequence to reduce changeover time ◾ Provides simultaneous visibility of inventory and capacity ◾ Provides visibility of critical constraints, for example, people and materials ◾ Rounds production quantities to lot sizes ◾ Offsets production receipts and component availability for transportation, quality inspection, goods receipt, or goods issue processing time ◾ Coordinates multiple levels of production

Issues with Excel Every one of the tasks above is difficult to perform in Excel. The best graphical scheduling systems allow an immediate understanding of the schedule at a glance. In one view the scheduler can see capacity

142  ◾  Scheduling Processes, Systems, and Software

utilization, the sequence of product attributes, changeover time and losses, downtime, relationships of upstream and downstream schedules, resource usage, and alerts for component and inventory shortages. In contrast, there is no easy way to create similar graphics and warnings in Excel. Most Excel-based scheduling programs use time-bucketed cells with complicated lookup functions and complex formulas. While conditional color coding of the cells might be possible, for example, based on inventory levels, production capacity, component, or resource shortages, consider the complexity of the formulas that would be required to perform the conditional lookup calculations (Figure 14.1).

Figure 14.1  Excel compared to scheduling software.

E xcel as a Finite Scheduling Tool  ◾  143

Moving production from cell to cell and experimenting with different scenarios is not something that can be done with one mouse movement by drag and drop in Excel; dragging and dropping would move the whole cell along with its formulas, lookups, and color coding. Instead, the scheduler typically types in the quantity directly and must know lot sizes and rounding values for the product in question. Moving production from cell to cell requires deleting and retyping.

Visibility of Attributes and Sequencing What about the visibility of attributes while scheduling? Our experience is that it requires about seven or eight attributes to properly schedule and sequence a typical product. For example, at Blue Lakes, we need to know about allergens, organic requirements, religious rules, color, bottle size and shape, particulates, viscosity, neckband, label, and cap. Sounds like either they need a complex lookup function to find product attributes and display them or else the knowledge must be kept in the scheduler’s head. In most businesses, there is a best sequence in which to make products, and changeovers are not the same in either direction; the changeover from A to B is different than the changeover from B to A. For allergens, a building chain is a minor change, but as soon as an allergen is removed, it requires a major cleanout. For example, at Blue Lakes, a dressing could start with a milk or dairy base and then add eggs, soy, wheat, and nuts without a washout. However, as soon as one is removed, the process must be completely cleaned. How do you write a formula in Excel that considers the incoming and outgoing product’s attributes, calculates the correct changeover time, and subtracts it from the time available in the planning period? Sounds like something that a planner must know in their head and then make allowances to factor down the production in a period where there is a major changeover. In contrast, the best scheduling programs will typically have algorithms to consider product attributes and changeover times, select a best sequence to minimize changeovers, calculate the appropriate changeover times, and subtract them from the time available.

Time Offsets Offsetting for delivery, quality, goods receipt, and quality inspection will require an inventory formula that references the addition of inventory from production in an earlier cell, offset by several periods. This might work OK if the columns are in days and the time offset is an even number of days and always the same, but what if the time offset is different for different materials? Think of the complexity of creating and checking all of these formulas so that the time offset is correct for each material and for each period.

144  ◾  Scheduling Processes, Systems, and Software

If the spreadsheet is simplified to use weekly buckets for production, and the offset is 4 days, something produced on Monday can be used in the same week. But if produced on Friday, it can’t be. How can this be represented in an Excel formula? Because of this, schedulers working in Excel tend to work in buckets, for example, days, weeks, or shifts, whereas the real world is continuous and the time required to do tasks are not even days, weeks, shifts, or months.

Lot Sizing and Multi-Level Scheduling Consider a common business case where there is a minimum lot size to make the production setup worthwhile, but then once the line is running, there are discrete production rounding quantities that make sense. It could be the size of a processing unit, a certain number of hours between required process cleaning or maintenance, or increments in which critical components are received, for example, rolls or bundles of packaging materials. Think about trying to write these formulas into an Excel cell, where the parameters might be different for different materials. Because of the difficulty, usually the scheduler just has to know, for example, something like “I have to run a minimum of 2500 cases of this material, and the rounding is 1250 per batch.” At a typical dry laundry or pet food plant, a large processing unit makes a bulk product and places it into bins or buggies. Several packaging lines pull the bulk and pack it into different variations, for example by size or language. The packing lines can’t start before the proper bulk is produced. Their aggregate offtake must be fast enough to keep the bins and buggies from being overfilled, but slow enough that they won’t run out of bulk product. Think of the complexity of writing these formulas into Excel.

Summary All of the above leads to a business continuity risk of using Excel for scheduling. Things like considering product attributes in the sequence, accounting for changeover times, accounting for receipt offsets, and coordinating upstream and downstream production rely heavily on the planner’s experience and skill in Excel. The resulting spreadsheets are complex and prone to error. Trying to use and understand a spreadsheet that someone else developed is almost impossible. This is why lead planners are afraid to take a vacation and time off, and why their managers worry about coverage for accidents or illness. In contrast, a well-designed process using standard software can allow any planner to plan any product. We’ve seen cases where planners from other sites and other businesses substituted for an absent scheduler, and where contractors quickly stepped in to fill coverage gaps for schedulers that needed to attend training.

Chapter 15

Software Designed for Production Scheduling With previous chapters as a background, let’s recap the requirements for planning and scheduling software necessary to create a schedule that can be executed by the production operations and that maximizes their efficiency while meeting customer requirements. Specialized scheduling software helps to create the schedule quickly, providing the agility and flexibility to meet disruption. Typically, its use reduces scheduling effort by about one-third and increases the site’s production efficiency by 5%–30%, resulting in greater throughput, which usually more than justifies the cost of the new software.

Supporting Processes Just implementing better scheduling software may not be sufficient to achieve success. A more comprehensive approach to systems and work processes may be needed. Scheduling is the end of the line, the point at which plans meet reality. Scheduling may be blamed for issues handed down to it from the longer-term and higher-level processes. Even the best scheduling software will have difficulty sorting out an overloaded schedule or one with insufficient staffing and materials to execute. Scheduling takes place at too low a level of detail, and there are too many options to consider. For efficiency in scheduling, the higher-level processes should create and maintain an executable plan with the following characteristics, which we will discuss in more detail in Chapter 21 on production and capacity planning: ◾ The production plan fits within available capacity, using realistic or demonstrated production rates, and taking losses such as changeovers, maintenance, and scrap into account. ◾ There are enough resources, for example, staffing to execute the plan. DOI: 10.4324/9781003304067-18

145

146  ◾  Scheduling Processes, Systems, and Software

◾ The critical materials and ingredients are available. ◾ It includes all the products required to be produced and captures the agreed-upon demand from all sources. A recent example is that many food manufacturers in North America have been limited by the availability of starch. A company that managed the shortage well prioritized their customer orders and constrained them by the amount of starch available. They only released demand to scheduling that could be covered by the starch that they expected to receive. We’ve seen others who did not manage their capacity well. When the schedule didn’t fit, the scheduler had to look at the inventory situation of every product, decide which of many products were in the best inventory position, and delete their respective production orders.

Scheduling Requirements Software designed for production scheduling should have these attributes to create an efficient schedule: ◾ Visibility of capacity required, capacity available, and capacity loading. ◾ Visibility of which lines are capable of making a given product and alternative lines. ◾ Visibility to the demand situation for a product and the pegging of demand back to customers and upstream intermediates. ◾ Calculation of inventory on a continuous timeline. ◾ Alerts for low or negative inventories. ◾ Easy ability to move between products that require the same kind of capacity, for example, a bottling line, and evaluate their inventory and demand situation. ◾ Visibility of lot size and rounding constraints for each product. ◾ Visibility of the production sequence on a continuous timeline. ◾ Visibility of product attributes critical to scheduling, our experience is that seven to eight are typically required, with somewhat fewer in chemical operations and even more in some packaging operations. ◾ Visibility of critical material, component, and ingredient requirements and their availability. ◾ Visibility of critical resource constraints and their availability, for example, staffing. ◾ Alerts when capacity, resources, or materials are not available. ◾ Automatic recalculation of changeover losses based on the current sequence. – The calculations should be able to cover the case where the change from product A to product B may not be the same as from B to A.

Software Designed for Production Scheduling  ◾  147

– If your business introduces things like allergens or contaminants, the calculation should understand that there are low changeover costs for a sequence of additive chains and high changeover costs for removal and cleaning, or running out of sequence. ◾ Offsetting of production receipts and component availability for quality inspection, goods receipt, or goods issue processing time.

Repetitive Scheduling Requirements For those who wish to use the repetitive scheduling strategies that we recommend in this book: ◾ When scheduling, there should be a method to automatically place new orders into their best position in the production sequence while meeting their due dates. ◾ There should be a way to group products that run well together and schedule them consecutively. ◾ Consistent master data should be used for repetitive sequence design and its scheduling. ◾ It should be easy to transfer a completed repetitive schedule design to the production scheduling software.

Multi-Level Requirements For those with multi-level processes, there are additional requirements: ◾ Instantaneous calculation of intermediate inventories on a continuous timeline, with warnings of when intermediate inventories exceed userdefined high and low limits. ◾ Visibility of the connection between upstream intermediates and the downstream orders that consume them. ◾ The ability to synchronize orders across multiple levels based on precedence and timing relationships. ◾ If necessary, capabilities for tank management and specifying flow paths between operations.

Software Selection The software selection process should start with a vision of how the scheduling process could change, what the new process would look like, and the business results that this would achieve. With this unifying vision, you can

148  ◾  Scheduling Processes, Systems, and Software

begin looking at your current scheduling process, the gaps between your current process and the vision, and the scheduling features and functions necessary to fill the gap. This is often called an “as is, to be” analysis. From there, the next step is to develop business scenarios and stories about what the software must do to efficiently schedule your business, with your equipment. We’ve often asked the schedulers to submit their most difficult scheduling cases and then worked with them to understand the requirements to properly handle each case and how their current system falls short. Each item was cross-referenced by the site and type of product that required it, the value, and the criticality of the requirement. With this preparation, we were able to develop demonstration scripts, interview vendors, and conduct demonstrations. We could fully explain our requirements when necessary during discussions with vendors and understand whether the alternatives they proposed would work. A demonstration of scheduling software is critical to assess its usability in practice. A Request for Proposal (RFP) is not sufficient. The requirements in an RFP typically can’t convey the full complexity of the request behind each requirement. And the requirement can easily be misinterpreted, answered incorrectly, or answered in a simplistic way. A demonstration script can give the vendor more context about the requirement, and a demonstration will show the usability of the software in meeting the requirement in practice. It’s unlikely that any software will meet every requirement; therefore, it’s important to gain a further understanding of the requirements that can’t be met. Does the software vendor understand the requirement, have they thought about it, and do they have a solution that they would be willing to develop? Is it critical to have the feature now, or is it something that could wait for maturity and continuous improvement? Would it be possible to delay the implementation of the sites and businesses that must have the feature? For multi-site implementations or implementations across multiple product lines at the same site, we have often phased our implementations by the business value of better scheduling, prioritizing the sites and product lines with the highest value, and adjusting the plan by the availability of critical features. New software will work differently from the way you are used to doing things. Be careful of hard and fast requirements. Often the software may do what you are asking in a different way, or provide more efficiency in other areas that will make up for small losses in some areas. We had a business that refused to implement new scheduling software because it couldn’t automatically move orders from one bottling line to another during automated detailed scheduling, even though it was proven to be more efficient at the rest of their multi-level scheduling process. As we will cover in our discussion of virtual cellular flows in Chapter 25, moving orders from line to line during scheduling is not always a best practice. Once we got the business to understand and implement the software, they found that they could move orders for swing products from line to line during capacity

Software Designed for Production Scheduling  ◾  149

planning to balance demand. And the new software allowed them to create a better schedule faster, because it could automatically synchronize the multilevel schedule between the upstream and downstream operations, which had to be done by hand in their older software. To close this chapter, we must give a warning. The planning and scheduling software is the planner’s window into the world. Any change to it will be uncomfortable, regardless of how difficult their current system is to use and how good the new system is. They have learned to use their old system with excellence, and any change will be threatening. We will cover change management topics, for example, readiness criteria, leadership roles, and success factors later in this book.

Chapter 16

Critical Ingredients, Raw Materials, and Components It almost goes without saying that raw materials, components, and packing materials are needed to produce a finished product. Therefore, the scheduling process should take their availability into account. But at the point that availability affects scheduling, it’s to some extent too late. Effective component and supplier management strategies can go a long way towards ensuring that materials will be available to support the production plan.

Availability Checking Let’s start with the discussion of availability checking. While it might seem best to check the availability of every material before committing to a schedule, the reality is that there are practical problems with this approach. First, from a computing power, interface, and data maintenance standpoint, this will require many more calculations and many more data points to an interface, which may result in sluggish performance and lag times when computing schedules. From a constraints standpoint, it may introduce so many constraints that no production plan would be considered feasible. Consider that the current material plan was designed to support the old schedule; therefore, by definition, the new schedule will need different materials, many of which will not already be on order or available. The real task of availability checking is separating the things that can’t be changed from those that can be changed in time to support the new plan. Therefore, during scheduling the checking effort should be concentrated on a few critical materials that may not be able to react to the proposed schedule, and the remainder should be managed by techniques that will assure that they are available.

DOI: 10.4324/9781003304067-19

151

152  ◾  Scheduling Processes, Systems, and Software

Critical Materials Here are some ways to think about the materials that might be critical for your business: SKU-specific items related to a single SKU or a narrow group of SKUs. These items usually have the most variability in their requirements from day to day and week to week, and it’s obvious that an SKU can’t be made without these items. Pre-printed packaging materials are an example. At Blue Lakes, these would be the labels for the jars, bottles, and tubs. Long lead time items, but not SKU specific. At Blue Lakes, the example would be some of the dry ingredients and spices. Near-term schedule changes that would increase total volume in the short term, for example, the decision to run overtime or over a weekend, should be checked that there is sufficient stock and expected receipts to cover the increase in production. Long lead time and SKU-specific items. An example at Blue Lakes could be the dressing packets, where the printed plastic packet film requires a longer lead time than the paper labels for the jars and tubs. Anything with an overseas source is likely to fall into this category or the one above. These are the items most likely to cause scheduling difficulty. They require an inventory and business strategy decision. Either the frozen schedule for the finished product needs to be extended to their lead time or the potential variation in the finished product schedule must be covered by buffer inventory.

The Firm Zone Strategy Since extending the frozen finished product schedule causes additional inventories of finished products, increased obsolescence risk, and reduced responsiveness, this becomes a cost and business strategy tradeoff. Multilevel Inventory Optimization (MEIO) packages have been designed to address these issues; alternatively, they can be analyzed in spreadsheets and by simulation. However, in most cases, it’s better to buffer a few components than to extend the finished product frozen scheduling period. We can frame this discussion about component inventory versus firming or freezing the schedule by talking about two endpoints. For ultimate flexibility and responsiveness, we could carry sufficient component inventories to allow us to change the schedule at any time, without worrying about their availability. On the other end, if we extended the firmed production schedule to the horizon of the longest lead time component, we would not need to carry any

Critical Ingredients, Raw Materials, and Components  ◾  153

raw material inventory, since we would know our dependent requirements from production with certainty before we ordered any materials. Neither of these endpoints is likely to be practical. Even if we had the ability to store unlimited component inventory, it’s unlikely that in the process industry we could have a completely flexible schedule without affecting throughput. We discussed this in Chapter 2 when we talked about the characteristics of process operations, in Chapter 6 on the advantages of repetitive schedules and stability, and in Chapter 7 on dealing with disruption in a structured way. On the other end, freezing or firming the schedule for the length of the longest lead time components, so that we have certainty in their requirements, is also impractical in most businesses. This would require too much finished product inventory, and conditions will change by the time we reach the end of the frozen schedule. We will miss opportunities to adjust our schedule to capitalize on new volume opportunities and we will over-produce some of the items that have fallen out of favor. Normally, there is a sweet spot that enables a long enough firm or frozen zone to provide production stability, order most of the materials with certainty, and still respond to the ups and downs of customer demand. At Blue Lakes, we might synchronize everything to the packaging materials, since they are specific to SKUs and carry a lead time of about a week. Therefore, we would freeze our production schedule for the next week. This also allows the time to thaw frozen ingredients, which requires four to seven days, with the certainty that once we thaw them in a specific quantity, what we thaw out will be used. Some of the dry, refrigerated, and bulk ingredients may have longer lead times than one week; however, they go into many products, they can be stored, and unlike packaging materials, which can only be used for a specific item, they won’t go obsolete. We can afford to carry a buffer inventory to be sure they are in stock. And since they go into many items, their usage variability from week to week isn’t great, and consequently, it doesn’t take much buffer inventory to provide protection from variability.

Strategy Examples We will talk about a German plant in Chapter 21 that is an example of how not to do it. During an advanced planning and scheduling implementation, the site was complaining that their new software didn’t work and could not create a schedule. We started by understanding and listing the business rules that had been implemented in the software: ◾ Customer service should be at least 98%, and no backorders were allowed.

154  ◾  Scheduling Processes, Systems, and Software

◾ The customer order lead time was three days. ◾ There was to be little or no finished product inventory. ◾ There was to be no raw material or component inventory, and shipments of components were to arrive just in time for production. ◾ The product production schedule was to be frozen for three weeks, because printed packaging materials were specific to the end-item SKU and had a three-week lead time. It’s obvious that this strategy is not going to work in practice. A business cannot expect to freeze its schedule for twenty-one days because of a long lead time component material, while expecting to accept customer orders on three-day notice, while carrying no finished product inventory. We interviewed the planners to find out what they actually did and found that they broke the rules all the time. On a daily basis, they broke into the fixed production schedule and changed it to adjust for customer orders and component delivery issues. But then they created other problems because they couldn’t get the packaging materials in time, they stressed their supplier capacity with the new orders, they created inefficient internal schedules that reduced capacity, and they ended up receiving component materials that were no longer needed. The reality was that no one was getting what they wanted. Customer service was not meeting the target, the finished product and component inventories were higher than expected, and they were out of production capacity. The situation was improved by recognizing that the rules didn’t add up, many were not real constraints, and inventories were insufficient to support the business strategy. Once the problem was understood, the team refocused on planning against the real constraints: protecting the three-week lead time for packaging materials and producing efficiently. They decided to keep the finished product inventories low and reduced the frozen scheduling period to three days, to correspond with customer order lead times. This increased responsiveness and made low-finished product inventories possible. A buffer inventory of the long lead time packaging material was created to allow production flexibility. Other components could be procured from local suppliers within the three-day frozen period, and their orders were scheduled to arrive slightly before needed, to protect against transportation delays and production rate variability. At Toyota in Georgetown Kentucky, I witnessed the coordination between the production of cars and the suppliers of their seats. The car plant locked in their schedule one day ahead of time and transmitted the schedule to the seat supplier. The seat supplier produced in a reverse sequence and loaded the seats into semi-trailers so that the first seats needed were at the back of the trailer, and the last seats needed were at the front.

Critical Ingredients, Raw Materials, and Components  ◾  155

A liquid detergent manager and its bottle supplier work in a similar way. For liquid detergent, the contents of the bottle are mixed from commodities stored in tanks and specific dyes and perfumes. A relatively few commodities, dyes, and perfumes can make any product. But the bottle shapes, sizes, and colors are unique to a few products. The schedule for the next shift is finalized by the production line and transmitted to the bottle supplier just down the road. On each line, production quantities are rounded to the nearest quarter truckload of bottles. The bottle supplier blows the appropriate quantity of each bottle size, shape, and color. They load them into specific trucks, destined for each production line, loaded in reverse order so that the first bottles required for the next shift are at the back of each truck assigned to a specific line.

Summary To recap and to provide a few words of warning, what we are discussing here can only be executed if production and suppliers are reliable. ◾ The business should measure its production schedule achievement and its supplier delivery performance and continuously improve both. ◾ Schedules should be frozen or firmed for a period that allows for stability and efficient production while accommodating the lead time of most components, and the ordering lead time of customers. ◾ Those components that cannot be ordered and received within the frozen or firm period should be buffered and should be checked for availability when creating a new schedule for the next period. ◾ When executing the processes to deal with disruption, the availability of certain critical materials should be checked before accepting any change to the schedule. As discussed in Chapter 7, products that require hardto-get or long lead time materials should be considered for removal from the catalog or treated as opportunistic MTOs.

Chapter 17

Scheduling Software: Security and Privacy Many companies come from an environment where scheduling is performed “locally.” For example, some use Microsoft Excel to create production schedules and store these schedules on local hard drives or shared network drives. In these companies, there are no additional IT considerations for scheduling. The other type of “local” refers to other legacy software installed in local data centers. While this dramatically increases hardware cost and IT overhead, security and privacy are typically well understood and covered by the data center’s overall security and privacy. In contrast, modern software platforms are often cloud software because of the inherent advantages in cost, agility, and management of cloud software, but this introduces a new set of considerations for IT departments in the areas of security and privacy. This chapter aims to help IT stakeholders evaluate cloud software for security and privacy effectively. Businesses often use lengthy “Security Questionnaires” to ask these questions of their vendors. Below is the essence of what is vital for security and privacy in our view. We find the terms “Security” and “Privacy” are often used interchangeably, but they mean slightly different things. Security is the safety of the data and system from malicious actors. In contrast, privacy usually refers to how your confidential company and personal data are handled under regular operation. In both cases, the sensitivity of the specific data placed into the cloud software is a factor. Certifications such as SOC 2 cover many of the considerations below. We like SOC 2 specifically for certifications as it covers common considerations that manufacturing businesses care about in cloud software. On the other hand, formal certifications are sometimes tricky for smaller software companies to perform. SOC 2 comes in two flavors: Type I describes a vendor’s systems and whether their design is suitable to meet relevant trust principles. DOI: 10.4324/9781003304067-20

157

158  ◾  Scheduling Processes, Systems, and Software

Type II details the operational effectiveness of those systems. A third-party auditor performs SOC 2 certifications.

Security For security, the two questions to answer for yourself are: 1. How likely is it for cloud software to be compromised by a hacker? 2. If it does get compromised, how would it affect my company? In both cases, a number of additional questions apply: ◾ How experienced is the development team, and have there been past breaches? ◾ Are there frequent releases, and how is third-party software kept up to date? ◾ Are third-party security audits regularly performed where a trusted third party attempts to gain access to the system? Are summarized findings available of the most recent audit? Again, if available, a SOC 2 or similar certification provides a shortcut that answers most of these questions. When considering how a hypothetical compromise may affect your business, there are two dimensions: (1) How sensitive is the data and (2) how tightly coupled is the cloud software to your other systems. Concerning sensitivity, scheduling systems often contain product characteristics that are not very sensitive. Forecast demand and planned production may be very interesting to competitors for a short time but quickly goes out of date. Some planning systems contain financial data, which can be more sensitive depending on the details. Concerning coupling, the colloquial term is “blast radius.” Specifically, the critical question is: Would compromising the cloud software that handles planning and scheduling affect other parts of your company? For example, does the system interface with your ERP through well-defined flat files that are logged and well-specified, or does the cloud software have a direct database connection through your firewall right to your SAP database? The latter is not recommended! Of course, there is a continuum, but ultimately for an interface, it’s important that (1) the data is as constrained as possible and (2) as visible as possible, and (3) there should be an on–off switch that you control. Being constrained refers to the range of ways your system may interpret each data field – to be constrained means there’s no more flexibility than needed for the integration. Visibility means that activity (ongoing and past) is apparent with minimal effort.

Scheduling Software  ◾  159

Privacy Good privacy is important because it helps avoid unintended disclosure of your data by the vendor and because good privacy controls are another essential layer of security. In addition, from a legal point of view, privacy considerations may be necessary for some jurisdictions for specific data, like certain demographic data. Here are some questions to ask when evaluating the privacy of cloud software: ◾ Does the vendor’s customer service team have access to your data? Under what circumstances would they have access? ◾ What data is stored in the system, and is any more data than necessary stored? ◾ Are the actions of your users and customer service logged and visible to users of the application? ◾ Does the vendor have well-defined procedures for dealing with customer data? ◾ Are the relevant policies of the vendor available for review? ◾ Is two-factor authentication in use? The SOC 2 certification includes privacy considerations. Nevertheless, it can be beneficial to answer these questions precisely.

PREREQUISITES TO GOOD SCHEDULING

4

Chapter 18

The Role of the Plant Leader Describing the full role of the plant leader is beyond the scope of this book. In this chapter, we focus on establishing the conditions for repetitive scheduling methods to achieve their maximum benefit, and to protect that value, in particular by neutralizing the impacts of disruption as much as possible. Scheduling must aim to support operations by scheduling production to meet its service, cost, and cash goals in the most efficient way. And when disruptions occur, the scheduling process should enable the plant to respond quickly, effectively, and efficiently, without triggering unintended consequences. Done well, this will be a key enabler for agility and resilience.

Future Proof the Plant A US manufacturer of OTC and consumer healthcare goods experienced severe disruption during the pandemic. With their diverse range of raw materials, they experienced frequent material shortages, sometimes for long periods. They were subject to the same packaging disruptions that many companies experienced. The lack of spares impacted equipment availability. They were unable to run all the lines due to labor issues. Transportation was a constant headache. Customers delayed orders or increased the volume. In these conditions, they were forced to change the schedule frequently. This was done with limited ability to foresee the consequences on other customers, resulting in unintended consequences. And plant output suffered. Capacity planning and scheduling were done in Excel. The constant changes to orders and resources played havoc with Excel’s ability to keep up. Maintaining a repetitive production cycle was challenging, to say the least. However, the plant team was competent, disciplined, and creative. Despite the pressure, they took immediate steps to counter the external chaos. This

DOI: 10.4324/9781003304067-22

163

164  ◾  Prerequisites to Good Scheduling

Future-proof the Plant

The Role of the Plant Leader

Deal with Disruption Reinforce repetitive patterns of production

Figure 18.1  The Role of the Plant Leader.

chapter describes some measures they and other companies took to insulate the plant as much as possible from outside disturbances. These are some of the structural changes that enabled the OTC manufacturer to withstand external disruptions by simplifying products and production. At the start of the pandemic, the lack of standardization resulted in unnecessary complexity, which greatly complicated planning and scheduling. Most of these steps cannot be done by the plant alone. In general, the plant leader will need to collaborate with the corporate functions to make the structural changes described below. There is a great deal of work to do, which the plant leader should champion. The plant leader should ensure that sufficient plant resources participate so that the in-depth equipment and product knowledge of plant personnel is captured in the measures taken.

Raw Material Supply Risk While there are many external causes of supply variability, there is only so much a manufacturer can do to mitigate these. However, internal changes can be made to reduce or eliminate single points of failure like raw materials supply by identifying alternate suppliers, and configuring planning and ERP systems accordingly so they can be used at a moment’s notice. Some customers have responded to actual or potential global shortages of a given ingredient and have proactively added product recipes and labels to use alternative raw materials should the need arise. The planning, scheduling, and procurement systems have been configured so they can immediately switch to the selected alternative suppliers or materials. This enables them to respond faster than their less-prepared competition to grow market share.

Standardizing Packaging Raw Materials Adopting standard bottles, caps, and label sizes for multiple product families will simplify procurement, warehouse operations, and product changeovers on the lines.

New Product Development Involvement New product development will be responsible for reformulating and testing recipes for new SKUs that utilize alternative ingredients. New labels must be developed for standard container sizes and reformulated products. Have them ready to be deployed at short notice.

The Role of the Plant Leader  ◾  165

Transportation Risks Transportation issues can be proactively mitigated by planning alternative routes and harbors, e.g., in case of adverse weather events that result in port or road closure, advance planning to use rail instead of trucks, and so on.

Labor Risk To reduce the risks of labor shortages, some companies have intensively cross-skilled shop floor personnel to enhance flexibility, where a given individual can do several functions.

Simplifying the Product Portfolio Simplification is a core strategy. Which customers should we prioritize if we’re forced to make choices? Should we suspend or permanently eliminate lowvolume SKUs? Sales will fiercely resist, at least until the savings have been quantified. Savings will result from fewer changeovers and reduced business complexity; risk mitigation measures can be more focused; planning and replanning can be much faster. Manufacturing has longed to cut the “SKU tail” as many low-volume SKUs are unprofitable and consume valuable line time and plant resources.

Selective Automation Selective line automation can reduce wasted time and skills to do certain tasks such as boxing products, reducing changeover times, etc. Structural changes like these will help to insulate the plant from external disruptions.

Improve Changeovers Changeover times can be significantly reduced using the SMED techniques described in Chapter 23. Prioritize the work according to the changeovers that most impact throughput.

Example The OTC supplier implemented these changes. The standardization of components reduced complexity, simplifying the tasks of maintaining supply, tracking the components, provisioning the lines, and so on. Changeover times were also reduced. The alternative ingredient and packaging suppliers addressed many headaches due to shortages. Rationalizing the SKU lineup led to longer runs and fewer changeovers.

166  ◾  Prerequisites to Good Scheduling

The result: A more straightforward operation with increased plant output and service levels, despite continued disruptions.

Dealing with Disruption Collaboration The scheduler is constantly faced with making quick decisions that significantly impact plant performance. This is particularly so when unexpected snags arise that demand a rapid schedule change. Planning and scheduling have become more tightly integrated than before. One food manufacturer now replans and reschedules most weekdays to try to stay ahead of the stream of problems they experience. The relationship between the planner and the scheduler is much closer than before. They are at the hub of the daily triage process, in which a cross-functional team meets to decide how best to meet their commitments to customers despite the issues they face. Since some people may be working from home, meetings are a mix of in-person and virtual. A leading Canadian generics pharmaceutical company that distributes its product to 60 countries was experiencing severe service issues. Service levels were below 55%. By applying cross-functional virtual teams focused on demand, supply, and manufacturing, enabled by business social media tools and end-to-end visibility, service levels recovered to over 95% in nine months. The teams focused on the flow of materials to the customer, responding in near real time to disruptions and demand changes. Each plant had a virtual team that included the planner and scheduler: When the scheduler was faced with a choice, they would consult the team by posting a description of the issue and the options to the online message system. Quality, warehouse, maintenance, and other SMEs would respond if they had some information that might better inform the situation, thereby influencing the decision. This immediate access to the collective wisdom of the team and status significantly improved the quality of scheduling decisions. The example below helps to understand how these virtual teams should be designed, established, and managed.

Physical Triage Meetings In the plant, we are used to cross-functional teams tackling equipment breakdowns, for example. Consider this example. The operator stops a high-speed filler on a packaging line because fill levels are approaching the lower limit. A cross-functional team (instrumentation, quality, electrical, mechanical) quickly meets at the filler with the operator and gathers information about the problem. They apply problem-solving methods

The Role of the Plant Leader  ◾  167

like Five Why. If, for example, it is due to wear on a part, the mechanic immediately assumes the responsibility to replace it and get the filler restarted with the operator. If they cannot fix the problem, it will be escalated right away, armed with the detailed information the team has collected. The purpose of the cross-functional team is clear: It must keep the plant equipment running well and producing at the target rate. First, let’s analyze the filler breakdown response. 1. A cross-functional team is formed with members with the skills to diagnose and fix most problems. 2. Their collective purpose and individual roles are well-defined. 3. The cross-functional team members quickly flow to the problem (the stopped filler). 4. Those members who can contribute to the diagnosis and repair do so. The rest will remain on the scene only as long as they might be needed. Then they go back to their work but are alert in case they are required. 5. Each breakdown is different, as are the actions to diagnose and repair it. Team members work spontaneously, using their knowledge and experience to resolve the issue. 6. Formal lean techniques like Five Why problem-solving are used. All the team members are trained in advance. 7. The problem is escalated when the team realizes it can’t resolve it directly. 8. Issue and resolution information is captured for later reporting, analysis, and improvement purposes. A few things are worth emphasizing. 1. There is no process describing each action. Instead, there is a flexible framework that guides their work. 2. The team members figure it out as they go – and they do it very effectively. 3. These activities are not managed directly.

Implementing a Virtual Team in the Plant The outline of physical triage teams above is helpful when thinking about virtual teams. In many ways, the only difference is that collaboration takes place in cyberspace using business social media messaging tools. The mission is not to keep the equipment running, but instead it is to make maximum use of the plant assets through effective plans and schedules despite unexpected internal and external issues. Companies like Unilever, Spotify, and others have adopted an agile business model, at the heart of which are virtual teams called squads and tribes.

168  ◾  Prerequisites to Good Scheduling

These are formally constituted teams responsible for business outcomes. They are modeled on similar principles per the filler example above. Virtual teams are increasingly how business is done, with people working from home. A helpful reference is Reinventing Organizations by Frederic Laloux. It provides guidance to successfully implement what he refers to as the Teal organization model. This requires more fundamental change than agile. Triage processes and virtual teams naturally extend beyond the plant. A materials planner in the plant can work in a messaging stream with corporate procurement to address the urgent need for equipment spares. In long, multi-plant supply chains, companies have established a planning community comprising the network planner and plant planners and schedulers, together with quality, logistics, and other personnel from each plant. Messaging tools eliminate the time and distance between the participants. We’re all used to doing this in our personal lives, so it is natural to follow suit at work. Collaboration maturity has become a competitive weapon, harnessed by large and small companies to respond quickly and better than their competitors: enhanced supply chain and plant resilience, flexibility, and agility result.

What Is Needed of the Plant Leader? Plant leaders should organize virtual teams using the filler example earlier in this chapter as a reference model enabled by social media tools. The team needs a clear purpose, to fulfill a critical business outcome that is valuable to customers. The disciplines required to achieve the mission must be in the team. Roles must be clear. Team members must act in real-time in response to an event. Swarm the problem. Depend on the wisdom of the individuals and the team to get it done. Provide guiding principles that define the boundary conditions, and the rules within them. Then step aside and let the collective ingenuity, deep understanding of their domains, and the magic of collaboration do its work. Make sure that decision-making authority in each area is clear. The plant leader plays a critical role in championing these changes and ensuring that the right level of attention is paid to the details. These changes impact ways of working, collaborative problem-solving under pressure, and rapid decision-making with limited information. Your support for the team is essential – not only to support them but also to coach them when they stumble. Your guidance and coaching will be the catalyst for their success.

Reinforcing Repetitive Patterns of Production This is more comprehensively addressed in the Sustainment section of Chapters 27 and 28.

The Role of the Plant Leader  ◾  169

In summary, those include: 1. Measure schedule adherence, inventory compliance, and On Time In Full (OTIF) performance. Drive improvements. 2. Ensure data governance is healthy by tracking governance process adherence. 3. Periodically assess master data integrity using checklists. Address exceptions by eliminating root causes, like new product data not being correctly updated. 4. Regularly review and update the contingency plans that you put in place earlier, guiding decision-making when disruptions occur. 5. Ensure that schedulers are appropriately trained. A common problem is when a scheduler is replaced: A noticeable drop in plant performance often follows because of a lack of training and plant knowledge.

Summary Some of the steps described above should be led and implemented at the corporate level in conjunction with the plant. Their success will significantly depend on the plant leader driving the right practices and behaviors, and supporting and coaching the teams as they transition to new ways of working. Middle management will also need your guidance: The authority they have depended on might dwindle if they cannot lead in a more transparent, knowledge-driven environment. The plant leader has never had a more challenging role. It is a time of change. Leading and managing the changes at the plant to achieve improved agility and resilience is the challenge – all while maintaining production performance.

Chapter 19

Scheduling Readiness Criteria Implementing a new scheduling system is a significant project. The scheduling system is the planner’s “window on the world.” They have developed a mental model of their site and production process, aided by their current scheduling system. Any change will be uncomfortable until they learn a new way of thinking. To illustrate this point, many years ago, we were conducting a software search for a new scheduling system. Our search team developed a rating scale for the degree to which prospective software met each element of our demonstration. Within this business, there was a crusty old planner by the name of Joe. The rating was on a numerical scale, zero through five. The project team had internal descriptions of each rating, the description of a five was that “even Joe would like it.” However, when the new scheduling software was introduced to Joe’s site, which happened to be the lead site, Joe thought it was awful. According to Joe, the new graphical scheduling system would never work, it had all kinds of problems, and his old text-based system was better and easier to use. However, we proceeded to testing, parallel runs, and startup. Six months later, Joe was our lead trainer and an outspoken advocate for the benefits of the new system. The Gartner Group’s Hype Cycle, and its “Trough of Disillusionment,” Figure 19.1, comes to mind. Every project, including scheduling projects, will encounter roadblocks and issues somewhere during its lifetime. For it to succeed, there must be a commitment from leadership to the concept and process and the will to stay the course when it hits the inevitable speed bumps. Therefore, without adequate attention to readiness and supporting processes, a planning and scheduling project will fail. When the business benefits of better scheduling are understood, there will be initial enthusiasm and a call to action, but the reality is that the initial momentum will be hard to sustain through the inevitable problems and roadblocks without supporting business processes and culture.

DOI: 10.4324/9781003304067-23

171

172  ◾  Prerequisites to Good Scheduling

Expectations

The Hype Cycle

copyright 2018 Gartner, Inc Innovation Trigger

Peak of Inflated Trough of Expectation Disillusionment

Slope of Enligntenment

Plateau of Productivity

Time

Figure 19.1  The Gartner hype cycle. Source: Gartner, August 2018.

It’s tempting to get caught up in the excitement and proceed with a project that shows promise and a compelling business case, despite lacking some of the fundamentals that we will talk about in this chapter. By this point in a project, everyone understands the potential and is eager to proceed, and the business promises to improve before implementation. Every time one of us has gone ahead and ignored those little warning voices in our head, the result has been a failed or delayed project, with the end result being that the project took longer than if the work to achieve the prerequisites had been done upfront. Our proposed readiness criteria for planning and scheduling projects are listed below. They are derived from and are a subset of the Oliver Wight ABCD checklist for Operational Excellence, based on the author’s experience with the things that will derail a planning and scheduling project. It should be noted that there are more things that must be in place for a successful scheduling transformation than are covered here; they are detailed in the Roadmap and Critical Success Factors (Chapters 28 and 29). This chapter focuses on those more specifically related to the scheduling process itself. It is always tempting to rationalize the lack of a prerequisite, accept the business’ promises to improve before implementation, and decide to proceed with the project. At this point in a project, everyone understands the potential and is eager to proceed. We are well up the hype cycle and heading towards the peak of inflated expectations. But every time I’ve ignored the small warning in my head and proceeded despite my misgivings, there has been a problem with the project that resulted in a major misstep. And likely the project

Scheduling Readiness Criteria  ◾  173

took longer than if the work to achieve the prerequisites had been done in the first place.

Readiness and Sustainability While we present it as project readiness criteria. Many of the items are also sustainability criteria. After implementation and throughout the system’s life, they should be reviewed periodically to ensure that the system isn’t degrading (Table 19.1). The data elements will be discussed in more depth in the next chapter on accessible, accurate, and complete data. Data accuracy and timeliness are critical. Without accurate data, even the best scheduling system will produce schedules that can’t be followed on the shop floor and won’t deliver customer orders. In one of our early scheduling projects, the production planners always complained that the most common reason for them to change the schedule, and the reason why their plans would not hold up, was that inventories were inaccurate, especially at distribution centers. At the time, this was just accepted as the way things were. It wasn’t until a later implementation of MRP II that we realized that this state of affairs was unacceptable. Inventory record accuracy policies and audits gave us the tools to maintain accurate inventories and build the foundation for a reliable schedule. The timeliness of data should be a red flag. When reviewing a site’s data, the presence of any overdue element is a strong warning that there isn’t a culture of timely data updates and an understanding of the importance of accurate and timely data. We’ve talked about capacity management briefly in Chapter 4 on the planning processes, and will cover it in more depth in Chapter 21 on effective production and capacity planning. For now, the analogy of a jigsaw puzzle may be useful. Demand flow and capacity management are like making sure that the table is big enough to hold the finished puzzle and that all the pieces are there before you start work. Scheduling is like assembling the pieces. Without a big enough table, and if some of the pieces are missing, nothing that the scheduler can do will make everything fit together properly.

Project Roles Two of the most important elements of project success are leadership alignment and the appointment of a strong project leader, who has the necessary authority and credibility to break down barriers when problems arise. Problems will likely arise during any project and a scheduling project is no exception.

174  ◾  Prerequisites to Good Scheduling

Table 19.1  Planning and Scheduling Project Readiness Criteria Inventory record accuracy

The quantity and location information is correct within a tolerance.

Overdue in transit orders

Overdue in transit orders indicates a lack of attention and understanding of the importance to update data on a timely basis.

Overdue quality management or inspection stock releases

Overdue quality or inspection stock releases indicate that something should have been released, according to the expected timing, but it hasn’t been. Therefore, when will the material be available? If the planners and schedulers can’t count on the data, they will create additional orders to cover. Instead of going overdue, the dates should be adjusted.

Bill of material accuracy

Bills of material include all planned components, with the correct quantity, scrap rates, and effectivity dates.

Routing/recipe accuracy

The schedule uses realistic rates and the correct machines or processing units for each item.

Product master data accuracy

Lot sizes, min/max/rounding values, inventory targets and policies, and production frequency.

Schedule adherence

Percentage of released production orders completed within a quantity and timing tolerance. This is the proof that the scheduling process is based on accurate data and is creating schedules that can be reliably executed.

Capacity planning

Capacity planning should be done on all constrained resources for the full planning horizon. The planned capacity load should match the actual load within tolerance.

Overdue production or process orders

Overdue production and process orders also indicate a lack of attention to detail and understanding of the importance of timely data entry. All overdue production orders should be rescheduled or canceled during the response planning process.

Downtime, uptime, and calendar maintenance

Planning calendars, downtime, and uptime have been maintained. The overall available time for each production process is accurately reflected in the planning systems.

Planner qualification

Production and material planners have been trained and qualified at the appropriate level to perform their tasks.

Demand planning/ demand flow

Demand from all sources is flowing to the planning system.

Capacity feasible plan

RCCP and capacity requirements planning are used to ensure a feasible plan over the entire planning horizon.

Project Management Prerequisites Site management

Site management has been briefed on the project and is supportive. (Continued)

Scheduling Readiness Criteria  ◾  175

Table 19.1  Planning and Scheduling Project Readiness Criteria (Continued) Key user

An experienced key user(s), who understands the complete planning and scheduling process, has been assigned to the project. They have the time available to participate in sprints and testing, and their management is committed to their involvement.

Business leader

A strong business leader, who has the ability and organizational standing to break down roadblocks and resolve critical issues, has been assigned to the project.

Management commitment

Upper management has been briefed on the project, they understand the business case, the cost and resource requirements, and they are committed to the project’s success.

Project manager

A qualified project manager has been assigned to the project.

Training and qualification

Resources to produce documentation and conduct the training and qualification have been assigned to the project. Users will be given adequate time to train and qualify.

Business process redesign

There is a commitment to business process redesign in order to support the new work processes.

Business case

A business case for the project has been developed.

A strong business case and vision for the project will be necessary to gain leadership alignment. The company’s upper management must be briefed on the project, they must understand the business case, and the cost and resource requirements; they must also be committed to the project’s success. They must understand that new work processes and business processes will be required. They must be willing to assign and support a few critical roles that the project will require: a strong business leader, a key user, a project manager, and a training and qualification leader. Depending on the size of the company or the site, the roles may be part time or combined, but the requirement for success is people who have the time to devote to the project without neglecting their other daily tasks. The corollary to this is that a scheduling project must develop a strong business case to make the need for change compelling. When the project meets the inevitable roadblocks, the business leader will have the ability and the organizational standing to resolve the critical issues and remove them. They will support the changes in work processes and procedures that will be needed to make the project successful. The key user is an experienced planner or scheduler who understands the complete planning process and has the time available to participate in the development, testing, and training.

176  ◾  Prerequisites to Good Scheduling

The training and qualification leader ensures that the project provides adequate company-specific documentation to protect business continuity, such that a new scheduler or substitute scheduler can be quickly and properly trained. New schedulers just out of training only have book learning. The qualification process ensures that skills are properly transferred, and it gives the schedulers confidence that they are prepared for their roles.

Readiness Examples To close the chapter, here are two examples that we talk about in other places but are viewed here from a readiness criteria standpoint. In our discussion of managing critical components, we talked about a plant that was an example of how not to do it, and we will come back to them again in a later chapter on capacity management. However, for this chapter on readiness criteria, what should have warned us was their lack of production schedule achievement. We had a failed project on our hands; the site was complaining that their new software didn’t work and could not create a schedule. We started by understanding and listing the business rules that had been implemented in the software, and it was obvious that they were not going to work in practice. When we interviewed the planners, we found that before the new software, they broke the rules all the time. But the software wasn’t allowed to break the business rules; therefore, it couldn’t create a schedule. The project was recovered by recognizing that the rules didn’t add up and changing the strategy to one that could be successfully executed. In another case, a snack food plant attempted to implement an advanced planning and scheduling program when its demand was running at 120% of the plant’s capacity. They complained that the software didn’t work because it couldn’t create a schedule that met customer service objectives. They weren’t doing any overall capacity management, and it wasn’t until we pointed out that the plant was over capacity that they understood why the scheduling system appeared to be ineffective. And that they had a problem that couldn’t be solved by just scheduling better. We temporarily stopped the project and asked them to come back when they had constrained their demand to be equal to demonstrated capacity. They did this, and their next startup was successful. The project team learned to ask assessment questions about demand management, and whether there was a process in place to ensure a feasible plan over the entire horizon. From these and other examples, we learned to stop or delay projects early when the business didn’t meet the assessment criteria. Continuing without these criteria being met meant that we would end up with a failed project or a delay later in the project.

Chapter 20

Accessible, Accurate, and Complete Data The foundation of a production schedule is accurate and timely data. Without it, it will not be possible to create a schedule that can be followed on the factory floor or that won’t be frequently changed. Inaccurate or delayed data in source systems is a common reason that planners and schedulers develop their own Excel spreadsheets rather than relying on corporate systems. It allows them to override inaccurate data with corrections, but it comes at a cost. When planners build their own spreadsheets, the discussion of any scenario starts with a question about the assumptions, and a reconciliation of various viewpoints on the correct starting data, and each of the participants may have a different opinion. Often, the model has aged between the time it was created and the time of the final decision, and the reconciliation must start all over again. We can think back to the countless times we’ve updated scenario models and discussed plans based on spreadsheets with names like Initiative Option 4.3A, where there were earlier Initiative Option versions 2, 3, 4, 4.1, 4.2, 4.3, and finally 4.3A, each painstakingly updated to reflect the data as of the meeting. In contrast, proper planning and scheduling software, which we’ve discussed in earlier chapters, is always up to date and recalculated based on the latest data, with warnings, action messages, and volume at risk recalculated against today’s reality. If based on accurate and timely data, it will be an outlook that everyone can agree upon and rally around.

Master Data and Transaction Data Data is typically divided into two types, but both need to be accurate and timely: transaction data and master data. Master data changes infrequently. DOI: 10.4324/9781003304067-24

177

178  ◾  Prerequisites to Good Scheduling

For scheduling purposes, it describes things like products, production lines, and ingredients. Transaction data for planning and scheduling gives the timing and quantity of demand and supply elements, for example, current inventories, expected purchasing receipts, quality management receipts, and customer orders.

Examples of Data Accuracy and Timeliness Problems Let’s consider inventory record accuracy. Before the importance of supporting data was generally understood, we implemented a graphical finite scheduling system across several sites based on a regularly repeating sequence of production, what we now call a product wheel. The production schedulers objected; they said, “how can we keep to a repeating pattern when we don’t have accurate inventories out in our distribution centers? We keep getting surprised by inventory shortages that cause us to break into our schedules.” Recently, we’ve worked with a customer where the in-transit records followed the financial flows of inter-company billing but often didn’t keep pace with the velocity of the actual shipments. This caused a lengthy pause to investigate every projected inventory shortage to see whether it was real or just a delayed transaction and whether the planners and schedulers needed to react.

Data Audits or Checking Practices The best practice is to establish a data ownership and governance process before implementing a planning and scheduling project and to audit the key planning parameters required to support efficient, low-touch planning and scheduling on a regular basis. We’ve seen scheduling projects that never reached their value realization or were derailed and delayed due to inaccurate or incomplete master and transaction data. All good data quality review processes have these characteristics: ◾ The process is documented. ◾ The actual data is checked against a standard. ◾ The results are measured and tracked against a goal. ◾ There is a root cause analysis of gaps. ◾ There is visibility to leadership.

Documenting the Process The purpose of documenting the process is to ensure that there is measurement consistency across the people performing the data quality reviews over

Accessible, Accurate, and Complete Data  ◾  179

time. The amount of detail provided in the documentation should be justified by the importance of the measure, but at a minimum, it should cover the key questions of who, what, where, when, why, and how: ◾ Who: What is the role of the person who checks the data, and what qualifications should they have to ensure that their conclusions are valid? ◾ What: Identifies the scope of the check, and the specific values or attributes to be checked. ◾ Where: The location of the standards used for measurement and where the results are maintained. ◾ When: What are the frequency and the timing of the check? ◾ Why: Why is this measure important? This should justify the choices for the data to verify. ◾ How: Describes the checking process at a level of detail that would enable a person with the identified qualifications to perform the data quality check reliably from beginning to end.

Checking Data against a Standard The standard for checking data can be: ◾ A single value, for example, always use this entry. ◾ A reference to another document, for example, the values are documented in the most recently published Operating Strategy located at …. ◾ Another measurable relationship that clearly provides the person performing the check with a way to confirm that the actual values are correct. The best standards are those that allow for simple verification and do not require a significant amount of interpretation of the results.

Measuring and Tracking Results against a Goal Measuring and tracking provide the means to identify trends and set action limits. It also provides a means to justify an increase or decrease in the checking frequency or scope. The choice of a goal should always be based on the needs of the business. The best practice for reporting results is that the target should be identified as a percent of correct material codes vs. the number checked. This means that if 100 materials were reviewed and 99 were correct, but one code was found to have three errors, the score would be 99/100. The site may choose to root cause each error individually, but the score should be based on the percentage of correct material masters.

180  ◾  Prerequisites to Good Scheduling

Analyzing the Root Cause of Gaps Gaps against targets should be identified, and root cause analysis should be used to focus the improvement efforts. This is critical to driving improvement. Results should be tracked, and gap analysis performed until results are regularly within predefined control limits. When results are at target, the scope and/or frequency of the data quality checks may be reduced.

Leadership Visibility The results should be reported as part of the KPI reviews at the site and business level for visibility. If data accuracy is not important to leadership, it will degrade over time, and the checks will go out of use.

Planning and Scheduling Data Some common master data and transaction data elements that the schedule depends on are listed below: ◾ Products ◾ Product attributes ◾ Lot sizes and rounding values ◾ Production rates and efficiencies ◾ Uptime and downtime ◾ Changeover times ◾ Demand from all sources ◾ Customer orders ◾ Inventory targets and methods of calculating inventory targets ◾ Resource requirements ◾ Raw material, ingredient, or component requirements when materials are a constraint ◾ Lead times for material delivery, quality inspection, goods processing, or movement through the plant ◾ Production line assignments A frequent issue for schedulers is keeping the line calendars and available time up to date. The best practice is to establish a process for requesting and confirming downtime for all except emergency issues. In many cases, checking transaction data is similar to checking master data. For example, inventory can be counted and compared to the records. Some, like supplier delivery performance, are measured with a time and

Accessible, Accurate, and Complete Data  ◾  181

quantity tolerance. Did the supplier deliver the requested quantity, within a quantity tolerance, and on time, within a timing tolerance? Frequent overdue transactions should be a red flag that something is wrong and that accurate and timely data is not considered important or rewarded. For example, a receipt from a supplier that was to arrive last week, a production order that should have been completed yesterday, or a receipt from quality inspection that was due two days ago. If they have been received, they should be properly recorded and closed. If they have not been received, their dates should be updated.

Summary The proof of accurate data is production schedule achievement. Production schedules will hold up and be executable if you have correctly modeled your production lines, their capabilities, and your products when combined with accurate and timely transaction data.

Chapter 21

Effective Production and Capacity Planning An effective production and capacity planning process is an important prerequisite to scheduling. It may be helpful at this point to make a distinction between planning and scheduling. Planning is deciding which products to run, and in what quantities, during each time period. Scheduling is choosing the best sequence in which to make the products for each time period, in a way that set-up costs are minimized, without running out of stock or missing customer due dates. The time periods used for planning depend on the products and the capability of the equipment. It should roughly correspond to the time required to cycle through the production of most products. For a highly flexible business that can make anything required on any day, the planning time period could be daily. But for many businesses, it may take a week, several weeks, a month, or even a quarter to rotate through most of their items. Even if the business could cycle through its products every three days, the human mind doesn’t work very well in thinking about three-day periods, and a weekly planning time period would probably be appropriate.

The Importance of Planning Without an adequate plan, it’s impossible to create a good schedule. If we use the analogy of a puzzle, capacity planning is like getting ready to put a puzzle together and scheduling is like assembling the pieces. The planning process is making sure that the table or work area is big enough and that all of the pieces are there. If the table is too small, nothing that the scheduler can do will make all the pieces fit onto the table. Some are going to fall off, and it will be difficult to decide which should be left out. Making sure that

DOI: 10.4324/9781003304067-25

183

184  ◾  Prerequisites to Good Scheduling

we have all the pieces is ensuring that demand from all sources is properly flowing and that we have created requirements to satisfy the demand. To give an example, a large European plant that supplied snack food to a global market implemented an advanced planning and scheduling (APS) system. At the time of the implementation, the business was booming and customer orders were at 120% of the plant’s capacity. They thought the APS system would solve all their problems. But at startup, the site couldn’t create a schedule that met customer service objectives, and the plant complained that the new software didn’t work. Of course, you can see the issue. If orders are higher than capacity, somebody will not be served. Simply throwing the issues on the production scheduler, the shop floor, or planning and scheduling software will not solve the problem. The project team stopped the startup and asked the business to constrain demand before resuming the project. To guide future projects, the project team learned to ask assessment questions about demand management and whether there was a process to ensure a feasible production plan over the scheduling horizon.

Resolving Overloads In the case above, with such a large overload, it was relatively clear that the problem could not be solved internally and that it would involve prioritizing and managing demand. However, in many cases, capacity issues are not so clear-cut. The planner must analyze whether overloads in some periods can be solved by shifting production to other machinery, other sites or contractors, advancing orders to earlier periods, or delaying orders to later periods. Each of these choices comes with a tradeoff. Moving to other production lines in the same period impacts inventory and customer service the least of any alternative. But the ability of the alternate lines to produce efficiently at target quality must be evaluated. Producing at contractors and other sites brings in a host of cost and distribution factors to be evaluated. Producing earlier adds inventory; therefore, the ability to hold the inventory, the cost of working capital, and the impact on shelf life must be evaluated. When the available capacity is much earlier than the overloaded period, the planner must decide how much inventory pre-build is too much and how early is too early. Producing later may be a reasonable alternative for businesses that carry safety stock, but it increases the risk of customer service issues. Asking customers whether they can delay shipments or take partial shipments may be an option in some cases. This kind of analysis is much better done at the planning level than at the scheduling level. As we’ve discussed previously, the time-bucketed

Effective Production and Capacity Planning  ◾  185

calculations in planning are mostly simple math, but scheduling is a form of the traveling salesperson’s problem that is known to be difficult or impossible to solve. Scheduling takes place at too low a level of detail to see the big picture. When the schedule doesn’t fit, it can be overwhelming to assess the business and financial impact of each possible production run and prioritize. The scheduler may resort to the first option that works rather than the best option to satisfy business priorities. In cases where the scheduler will run out of stock no matter what they do, everything may look equal, and the decision may be arbitrary. If the overload is passed to the shop floor, the decision may be to run whatever is most convenient or runs best, not what is most important. By developing an adequate plan, we reduce the scope of the scheduling problem and make it more likely that a good sequence will satisfy the prioritized customer demand and fit within capacity, component, and other resource constraints. Modern planning software will continuously calculate the forward-looking inventory of every product against its inventory bands while simultaneously evaluating the capacity of the production equipment and the availability of other constraints such as staffing and component materials.

Automated Planning It’s relatively easy to automate planning to develop an optimization guided by priories that match business objectives. A typical set of priorities might highly penalize inventories below zero, moderately penalize inventories below a low inventory zone, and lightly penalize inventories above maximum while respecting capacity constraints. Just these simple constraints can often produce a viable plan that will follow common business rules and be easy to manage in the detailed scheduling process (Figure 21.1). For a multi-level process, adding constraints of batch sizes at each level and intermediate inventory limits makes the optimization more complex but brings a better result to guide scheduling. However, it’s tempting to try to model every constraint in a planning optimization program, but this comes with a risk. The more the constraints,

Inventory Level

Optimization Penalty

Above Maximum Above Target Between Target and Minimum Below Minimum Below Zero

Low No Penalty Low High Very High

Figure 21.1  Optimization priorities.

186  ◾  Prerequisites to Good Scheduling

the more difficult it will be to understand the solution and why the software chose what it did. Further, there is a risk that there is no possible solution that satisfies all of the constraints. In this case the optimization may fail, take a long time to solve, or take a nonsensical creative approach that’s impossible in the real world. If the software is going to run out of stock and incur high penalty costs no matter what it does, all answers may look similar, and the result may be arbitrary. The best practice is to focus on a few main constraints and to always give the optimization an “out” or exit strategy. This ensures that the optimization delivers a result that the planner can review and modify, instead of a failure and no answer at all. Examples include the ability to have inventories below zero at a high cost, in other words, backorders, or the purchase of constrained materials at a high cost from an alternate supplier of last resort, even if no such supplier exists in the real world.

Planning Example During an advanced planning and scheduling implementation at a large plant in Germany, the site was complaining that their new software didn’t work and could not create a schedule. We started by understanding and listing the business rules that had been implemented in the software: ◾ Customer service should be at least 98%, and no backorders were allowed. ◾ The customer order’s lead time was three days. ◾ There was to be little or no finished product inventory. ◾ There was to be no raw material or component inventory, and shipments of components were to arrive just in time for production. ◾ The product production schedule was to be frozen for three weeks, because printed packaging materials were specific to the end-item SKU and had a three-week lead time. It’s easy to see that this so-called “strategy” had no basis in reality – it was based on unreasonable rules. It’s impossible to allow customers to place their orders on a three-day notice, stock no component or finished product inventory, and freeze the production schedule for three weeks while delivering near-perfect customer service. We interviewed the planners to find out what they actually did and found that they broke unreasonable rules all the time. On a daily basis, they broke into the fixed production schedule and changed it to adjust for customer orders and component delivery issues. But the software was required to follow the rules and couldn’t do this; therefore, it was impossible to create a schedule in the system.

Effective Production and Capacity Planning  ◾  187

This is something we’ve seen often. Management asks for impossible rules; we often call them strategies that don’t add up, yet the planners try their best to meet them. Planning against unreasonable rules is like overdriving your headlights in a car. You can get away with it for a while, but sometimes things happen. The result of unreasonable rules is long-term performance worse than if the business had used reasonable rules to start with. In the example above, the reality was that no one was getting what they wanted. Customer service was not meeting the target, finished product and component inventories were higher than expected, and they were out of production capacity as the frequent schedule changes and disruptions compromised production efficiency. Whenever the schedule was changed, unnecessary component inventories were created due to items already en route, and supplier capacity was stressed to add orders for the new items added to the schedule. The team fixed this by recognizing three things: 1. The planners were breaking unreasonable business rules all the time. 2. The rules didn’t add up, and many were not real constraints. 3. Inventories were insufficient to support the business strategy. Once the problem was understood, the business rules were changed. The team refocused on planning against the real constraints, maintaining an efficient production sequence to maximize capacity, and protecting the threeweek lead time for packaging materials. They decided to keep the finished product inventories low but reduced the frozen period to three days, to correspond with customer orders’ lead times. This increased responsiveness and made low-finished product inventories possible. A buffer inventory of the long lead time packaging material was created to allow production flexibility. Most other components could be procured from local suppliers within the three-day frozen period, and their orders were scheduled to arrive slightly before needed, to protect against transportation delays and production rate variability. Thus, adopting logical business rules provided the ability to create practical plans, which in turn allowed effective schedules.

Characteristics of a Good Production Plan The following are the key considerations for a quality production plan to ensure that scheduling and execution will be successful: ◾ The plan requires no more than the available capacity, resources, or materials, after taking all losses, such as setups, maintenance, team meetings, process improvement time, or scrap into account.

188  ◾  Prerequisites to Good Scheduling

◾ Realistic or demonstrated production rates are used for every product in the plan. ◾ All products required to be produced are included, in the correct quantities, and in the correct time period. ◾ Agreed-upon demand from all sources is included. ◾ Inventory is at target levels at the end of each planning period. ◾ When multiple production levels are involved, lot size criteria are used at each level.

Managing Inventory Targets and Constraints Let’s elaborate on the point about managing inventory targets during planning. It is critically important for a business to maintain the capability to cycle through the efficient production of all its products. When this critical level of inventory is not maintained, the efficient production cycles will be disrupted, decreasing capacity and decreasing inventory further. The new lower inventory levels will cause more breaks in the efficient sequence, leading to more capacity losses and even lower inventories, and so on. Procter & Gamble used to call this the death spiral, and a recent food customer has expressed this as the inventory roller coaster. The only way to break the death spiral or inventory roller coaster is to return to an efficient mode of operation, which requires managing demand and the product portfolio to return to a workable level of production efficiency and inventory. We saw this during the recent COVID disruption. Many of the food producers that we worked with reduced their assortment of products offered for sale. This allowed them to produce more efficiently with the staffing available and focus their procurement efforts on fewer materials and critical materials. For example, starch was in short supply nationally. One manufacturer aligned their production to the availability of starch and made sure that the starch was converted first into critical products to supply their most important customers. As we discussed in Chapter 9 on the role of inventory, a business should calculate the critical level of inventory required for efficient production of their product portfolio and then balance capacity and demand to maintain inventory above the required level. The critical level is roughly equivalent to the sum of the safety stock, lot size impact, and average cycle stock for each SKU, plus an allowance for quality inspection stock, dead stock, and other frozen stock. The general rule of thumb is that the inventory required to protect the business increases exponentially once capacity utilization exceeds about 90%. Therefore, a business should maintain at least a 10% margin between average customer demand and production capacity, and it must be prepared to limit

Effective Production and Capacity Planning  ◾  189

sales whenever inventory approaches the critical level necessary to maintain efficient production.

Summary The points that we want you to take away from this chapter are: ◾ There is a distinction between planning and scheduling. ◾ Most business and prioritization decisions are easier to manage at the planning level. ◾ The business rules that are used to make decisions must add up and be feasible. ◾ The planning process balances demand and capacity to maintain target inventories and the ability to produce in an efficient cycle. ◾ Planning can reduce the scope of the scheduling problem. The task of scheduling is more straightforward when given a plan that fits within available capacity and resources.

Chapter 22

Workforce Engagement Selling the Idea For any significant initiative to succeed, especially one where you are transforming your scheduling process in a way that alters your entire manufacturing managing processes, you must have the support of the majority of the workforce. They must have faith in what you’re trying to accomplish, a belief that it will succeed if properly designed and executed. They must understand that it presents a better and more satisfying way to do their jobs, and are motivated to help it succeed. The first step in engagement is open, honest communication about what is planned and why it is important. The benefits to the business, the plant, and each individual employee must be clearly stated. Putting the story together shouldn’t be hard, because everyone who touches production scheduling or is affected by it should benefit. ◾ Schedulers should find that this makes their jobs easier, less stressful, and more meaningful. The designed patterns should make all the normal, routine orders easy to schedule: Just follow the patterns. That provides more time to focus on any upsets or special situations that may have just arisen; this additional time should allow for more deliberation and more consultation with others and lead to better decisions. All that should make the job more satisfying. We implemented a new scheduling process, including product wheels and improved scheduling software, at a pudding plant a while back. Before we started, the scheduler had announced her retirement date; she was planning to leave because the high stress was getting to her. They were finding it hard to train a replacement because the current process was largely undocumented, and only in the scheduler’s head. (We find this dependence on “tribal knowledge” very common in small- to medium-sized companies.) With the system we provided, they found it DOI: 10.4324/9781003304067-26

191

192  ◾  Prerequisites to Good Scheduling

very easy to train a replacement because all the “rules” and all the preferences were now documented, but then the current scheduler decided to stick around longer because her job had gotten so much easier. ◾ Production line workers quickly get into the rhythm that the sequence and regular cycles create. People tend to be creatures of habit and like to work in some degree of structured routine, where they know what they will be doing tomorrow and the next day. They also enjoy that they are spending more time on valuable, productive tasks and less on non-productive things like changeovers and minor stops. While this (product wheel implementation) was taking place it became a much more enjoyable place to work for all. Associates were able to know what their day would look like; it was predictable … and not chaotic anymore. (Dean Bordner, prior Senior VP of Operations, Nature’s Bounty) ◾ Wheels and other repetitive scheduling strategies set up regular patterns for raw material and packaging component needs, so procurement clerks have greater visibility on what they will need to order and can put more structure and predictability in their ordering patterns. They can provide suppliers with better forecasts, which should take some of the stress out of supplier discussions. ◾ As the beginning and end of each run are now more predictable, any start-of-run and end-of-run QA tests become easier to schedule. To the extent that wheels on different lines can be coordinated, it becomes possible to smooth out the loading on the QA lab, making the technicians’ jobs more routine and less stressful. ◾ The time that a line can be available for Preventative Maintenance (PM) can be better scheduled, smoothing out the maintenance mechanic’s workload. Specific times for PMs can be built into the regular pattern, at whatever frequency is appropriate for that equipment. We put product wheels on several extruders at an ethylene co-polymers plant in Texas and built PMs into the regular wheel patterns. There were six extruders, each running a seven-day wheel. Production on each extruder took about six days, and the seventh day was dedicated to PMs. The wheels on the extruders were staggered so that each started its cycle on a different day, and so each came to its PM on a different day, leveling the load on the maintenance crew. We proposed that a major vitamin tablet company implement product wheel scheduling at one of its packaging facilities a few years ago to increase throughput and reduce cost. The final decision to proceed was up to the

Workforce Engagement 

Chief Supply Chain Officer; he was intrigued by the concept but said that he wanted to put his resources instead on initiatives that would have a direct benefit on his people. When we explained how product wheels would do that, he very quickly authorized our work.

Designing the New Process ◾ Include representatives of all affected groups in the design. Their experience and perspectives will lead to better designs, and their inclusion will increase buy-in and ownership. This approach to working with people on the floor is key to managing the change needed to stabilize manufacturing, and reap huge benefits. (David Kaissling – former Chief Supply Chain Officer – Shearer’s Snacks) ◾ The numbers don’t always work out the way you want them to, so compromises have to be made. For example, you may not have the capacity to put all organic products on one line so a few will have to go on a line with non-organic products. Preview that compromises will sometimes have to be made so people don’t get discouraged but realize that even with compromises these techniques can bring most of the benefits.

Executing the New Process ◾ Promote value for standard work, for following the standards, for not being reactive when things appear not to be working, for following the defined contingency plan if one exists for this situation, and for working to create one if not. ◾ Explain and honor the safety net principle, that as long as people are following the new rules and standards, they will not be punished if things go wrong. ◾ Define metrics and KPIs that reflect what you are trying to accomplish. Make sure everyone understands how these metrics capture the new objectives. It is a mistake to expect everyone to follow a new, better process but then hold them to old and obsolete measures. Product wheels will almost always increase throughput on a weekly or monthly basis, but throughput on some days may be lower because of a difficult

◾  193

194  ◾  Prerequisites to Good Scheduling

product mix; so old metrics that prized maximum daily throughput should be revised to value longer-term throughput. If a line has idle time that could be used to run more changeovers, shorten cycles, and reduce inventory, that will lower OEE. Where that may be a good thing from an overall economic viewpoint, OEE targets should be relaxed to avoid that being a barrier.

Chapter 23

Changeover Reduction – SMED If product changeovers are long or costly, there is a tendency to run long campaigns before changing to the next product, to minimize the overall penalty incurred. The EPQ logic reinforces this; long or costly changeovers drive the EPQ answers to long campaigns, resulting in longer product wheel cycles and higher inventories. A product wheel strategy will work in this situation and will improve operations, but better wheels can be designed if the changeovers can be shortened. Thus any product wheel initiative should include an analysis of changeover times and how they can be reduced.

SMED and Its Origins Toyota recognized the value of faster changeovers in the early 1950s as the Toyota Production System (aka Lean Manufacturing) was beginning to evolve. One of the most time-consuming changeovers they faced was the replacement of the dies on the large presses used to stamp out auto body parts, which was taking several hours. Shigeo Shingo, an industrial engineer consulting with Toyota, developed a methodology for examining all set-up operations and modifying the set-up process to reduce the overall time. Using Shingo’s techniques, Toyota was able to shorten the die changes from three hours to fifteen minutes by 1962, and to an average of three minutes by 1971. In recognition of this tremendous accomplishment, Shingo’s methods and techniques have become the standard for changeover reduction and have come to be known by the acronym SMED, for Single Minute Exchange of Dies.

DOI: 10.4324/9781003304067-27

195

196  ◾  Prerequisites to Good Scheduling

SMED Concepts Four of the fundamental ideas that SMED promotes are shown graphically in Figure 23.1: 1. Identify tasks that could be done outside of the changeover time: Recognize that some of the tasks normally done during a changeover can be done before the equipment is turned off and production stops. These are called “external” tasks and include things like bringing all the necessary tools, new parts, and packaging materials to the equipment. As the setup is nearing completion, moving any parts removed from the equipment, housekeeping, and cleanup tasks can often wait until after the equipment is turned back on. These external tasks can consume a lot of time so moving them outside of the time window when the machine is not producing can shorten set-up time dramatically. 2. Move all “external” tasks outside of the changeover: Move the tasks identified in step 1 so they happen either before the equipment is taken off-line or after it comes back up.   3. Simplify the remaining internal tasks: Use dowels, locating pins, fixtures, and visual marks to speed up the time required to get new parts in place. Standardize bolts where possible to minimize the number of wrenches required. Use quick-disconnect fasteners where possible. Use poka-yoke (mistake-proofing) techniques to ensure that the apparatus cannot be installed incorrectly. “Center lining” is a methodology often employed to ensure that mechanical adjustments are done correctly. PRODUCT A

PRODUCT A

CHANGE OVER

CHANGE OVER

PRODUCT B

CHANGEOVER INTERNAL TASKS

EXTERNAL TASKS

EXTERNAL TASKS

INTERNAL TASKS

PRODUCT B

EXTERNAL TASKS

PRODUCT A

EXTERNAL TASKS

INTERNAL TASKS EXTERNAL TASKS

PRODUCT B

INTERNAL TASKS EXTERNAL TASKS

Figure 23.1  Major SMED improvement steps.

Identify tasks which can be external Move external tasks outside the changeover window

PRODUCT B

EXTERNAL TASKS

PRODUCT A

PRODUCT B

EXTERNAL TASKS

INTERNAL TASKS

PRODUCT A

PRODUCT C

Simplify internal tasks

Perform internal tasks in parallel

Changeover Reduction – SMED  ◾  197

4. Where feasible, perform internal tasks in parallel: If several operators can perform tasks concurrently, the time can be reduced without increasing the total labor content of the setup. While the steps appear simple, the real challenge lies in how to discover the improvements needed to achieve these steps. Developing a detailed process map and timing diagram is a good way to start. Video recording a changeover also provides a valuable view of what actually happens during the changeover and where savings might be possible. It is imperative that the people who regularly perform the changeover play a key role in the discovery/improvement process; they very likely have insights into things that can be done to reduce time. Their experience and perspectives can lead to reductions and their involvement will build their support for the new process. After the changeover process has been revised and tested, it must be documented, standardized, and audited on an ongoing basis so that theimprovements can be sustained.

Process Industry Changeovers In assembly plants, changeovers generally consist of mechanical and/or electrical modifications to the equipment, subsequent calibration and adjustment steps, and often the creation of a test part to check dimensions against acceptable tolerances. In the process industries, we also see tasks of this nature, such as resetting the width of the die in a sheet casting process or adjusting conveyor rails for different bottle sizes in our salad dressing plant. However, it is frequently the case that more of the time is spent cleaning out the ingredient feed systems and the processing equipment to prevent crosscontamination. In many food processing plants, for example, equipment is shared among several product varieties, which may or may not contain allergens, such as peanuts. This can pose very stringent requirements for cleaning between products. As another example, the tinting tanks used in paint manufacturing require thorough cleaning during color changes. The tasks performed during these cleanups are well suited to conventional SMED analysis. In extrusion, sheet goods, and batch chemical processes, much of the time lost is the time required to bring the product to the appropriate temperature, pressure, speed, or thickness, after all the mechanical tasks have been performed, as shown in Figure 23.2. Therefore, the SMED process must also deal with these so that the total time for the changeover including the time for process conditions to stabilize is reduced. Because these components of the transition are more dependent on the process chemistry or physics than on manual tasks, more technology-related solutions are often employed. These may include techniques like adaptive process control to speed up ramping back to first-grade product.

198  ◾  Prerequisites to Good Scheduling

CHANGEOVER

PRODUCT A

MANUAL TASKS

BEGIN PRODUCT B PHYSICS AND CHEMISTRY REACH EQUILIBRIUM

FIRST QUALITY PRODUCT B

Figure 23.2  Components of some process changeovers.

In some process operations, a non-trivial part of the changeover is the test lab response time. Some food processing operations require testing after an allergen clean to ensure complete decontamination. Most continuous flow chemical processes require testing to ensure that all properties are on aim conditions after an ingredient, catalyst, temperature, or pressure change. In these cases, SMED should investigate ways to reduce lab turnaround time by focusing on flow through the lab, and on lab scheduling processes. Preparing a VSM or flow diagram of the testing lab would lead to a better understanding. You may learn that samples are batched and gathered into groups before testing, causing delays; in that case, single-piece flow should be evaluated. The benefits of reducing lab turnaround time may justify buying more analyzers or hiring additional technicians.

Automotive Fluids Packaging BG Products is a manufacturer of brake fluids, transmission fluids, and other automotive care products, packaged in 6, 12, 32, and 64 oz bottles and cans, and larger containers for repair shops. We developed product wheels for several of their packaging lines several years ago. While designing a wheel for their rotary packaging line, we recognized that we could build a much better wheel if we could reduce the time of the longest changeovers. Their most complex changeover took an average of eight hours, which seemed excessive. By applying SMED techniques we were able to get it down to three hours, and with repeated improvements achieved a complete changeover in 2 ½ hours. We started by building a cross-functional process map (aka swim lane chart) of the entire changeover. In a brainstorming session with operators and the lead mechanic, we discussed potential improvements and captured the more practical ideas on a future state swim lane map. We had a large copy of the map printed and hung on an easel in the area, and as they went through the next changeover, they captured comments and additional ideas on the map and recorded the actual times for each step. Later that day, we held a post-mortem and discussed what went well and what needed further improvement. All of that was captured on the next version of the future state

Changeover Reduction – SMED  ◾  199

map. That map was printed for the next changeover and the process was repeated. It took very few cycles to reach the 2 ½ hour milestone. The improvement ideas included changing the orientation of some access doors to keep them out of the way when open, making the movement of tools and change parts external, and doing more tasks in parallel. We also improved the communications protocols so that everyone involved was alerted when this class of changeover was upcoming. Despite the success, they continued to print the latest version of the map for each changeover, track each step and time, and conduct a post-event review. That turned out to have a lot of power: It provided data needed for ongoing continuous improvement of the changeover, and it kept everyone reminded of the importance of quicker changeovers. The result was that we were able to design a product wheel for the rotary line with a primary cycle of 14 days, which wouldn’t have been possible without the changeover reduction.

Diaper Manufacturing As illustrated in this example, many of the steps of SMED are simple and common sense. Solutions are readily found once crews understand the importance of reducing changeover time and focus their attention on it. Procter & Gamble was struggling to keep up with the demand for Luvs Diapers. As described in Chapter 6, they turned to product wheels to improve efficiency but also started to implement SMED. By applying the wheel principles of family grouping and cellular manufacturing, they eliminated the most difficult changeover between the small-, medium-, and large-size diapers and began focusing on the next most difficult changeover between the boy/girl gender. The changeover required replacing the absorbent core-forming dies on a large forming wheel. One of the authors received permission to use a line and crew for a day with the objective of reducing the changeover time to ten minutes or less. The first step of the SMED process is to watch and evaluate the current state. With video cameras recording, the line crew performed a normal changeover. Since they all knew they were being recorded, they wanted to show their best. The line stopped, and everyone jumped to put their locks on the disconnect switches so that work could start, but then they realized that the line had stopped in the wrong place. They all quickly removed their locks, the startup horn sounded, and the line went through its required delay to make sure everyone was clear. It started again, diverting product into the scrap bin as it came up to speed. This time it was stopped in the right place, and everyone jumped again to place their locks. Next, the crew retrieved the changeover carts from the storage room at the head of the line and wheeled them out to the line. Then they started

200  ◾  Prerequisites to Good Scheduling

removing dies from the large core-forming wheel using regular right-angle Allen wrenches turned by hand. Some of the Allen wrenches were rounded over so a crewmember was dispatched to the storeroom to get some new wrenches. The replacement dies and their bolts were all jumbled in the changeover cart drawers. Many of the bolts had bad heads because they had been removed using the rounded-over Allen wrenches, so another crew member was dispatched to get some new bolts. The changeover took about an hour. Once the line was back up and running, the crew convened to review the video of the changeover and brainstorm improvement ideas. They immediately came up with the following list which corresponds to many of the SMED concepts above: 1) Someone would watch the line during the shutdown, make sure it was stopped in the right place, and signal everyone when it was OK to place their locks. 2) The changeover carts would be moved to the lines before the line shut down. 3) The new forming dies would be laid out on tables next to the forming wheel with their bolts in place. 4) All the bolts would be checked ahead of time and replaced as necessary. 5) All the tools would be checked for condition and replaced as necessary. 6) The crew went to the hardware store and procured some ball head screwdriver-handled Allen wrenches to quickly start each bolt. 7) Electric drivers with Allen wrenches in their chucks were placed on the line to rapidly remove the old dies and tighten the new bolts on the new dies. 8) The crew would work in teams, one team removing the old dies with an electric driver in reverse. Another team had one person starting the bolts and another tightening them with an electric driver. In the afternoon, the crew re-did the changeover. Everything was staged and ready, the line stopped in the right place on a signal from the person watching, and the change took about 15 minutes. Subsequent improvements reduced the time to ten minutes and the procedure was standardized and rolled out to all of the Luvs Diaper lines.

SMED Beyond Product Changes Although SMED can be quite valuable in optimizing product changeovers, it has additional uses in process plants. In many process operations, equipment must be taken out of service nightly, weekly, or bi-weekly for cleaning and/ or sanitation. While these are not literally changeovers and are required for

Changeover Reduction – SMED  ◾  201

reasons other than a product change, they do steal productive capacity from the operation and should be accomplished as quickly as possible. SMED has been effective in a number of these situations.

A Non-Manufacturing Example Many of the books and articles on SMED use a racecar pit crew as an example of a setup done well. Procter & Gamble, who sponsored the Tide Racing Team, brought in their pit crews to give talks and made an internal video called “Changeovers Ain’t the Pits.” The talk often started with a question for the audience, how long do you think it would take you to go out and get four new tires for your car? Then they are shown a video of a NASCAR pit stop. Anyone who has watched a professional automobile race can relate to pit crew operations, to the precision, the coordination, and the purposefulness of everything that’s being done. The pit crew operation provides a strong visual image of SMED principles at work: ◾ All tasks that could possibly be done externally are. The tools are ready, the new tires are in place, and everything is prepared for the moment when the car enters the pit. ◾ All tasks have been thoroughly analyzed, simplified, and structured to be done in the shortest possible time. ◾ All tasks are done in parallel. All four tires are replaced simultaneously, while the gas tank is being filled. ◾ Technology has been appropriately applied, for example, to the mechanism used to lift the car for tire changes. ◾ Everyone understands his or her role and has practiced it to get the time down to the absolute minimum. ◾ All pit stops continue to be timed, and there is an intense ongoing effort to further reduce time in the pit. Everyone involved understands that these races are often won or lost in the pits and is highly motivated to contribute to the potential victory. The more that manufacturing teams understand that operational success is likewise dependent on fast, effective changeovers, the easier it will be for the operation to become lean.

SMED Applied to Blue Lakes Packaging When embarking on a SMED initiative, it’s important to begin with the equipment or lines that can get the most benefit. For example, while the Blue Lakes packet line described in Chapter 6 would benefit from SMED, it should

202  ◾  Prerequisites to Good Scheduling

be down the priority list. It has only one packet size change and one allergen clean on the first two-week cycle, and two packet changes and one allergen clean on cycle 2, so the capacity increase would be small. And the line has adequate capacity, so throughput increases are not an immediate concern. The three bottling lines are much more appropriate targets; bottle size changes are very time-consuming, and they have more organic changes to schedule and more allergen combinations requiring extensive cleaning. And the lines are very heavily utilized so any throughput improvement will have value.

Summary Most of the equipment found in a process plant must produce a variety of products. In some cases, the changeover from one product to the next is fast and inexpensive. In others, the changeover can be long and/or costly, driving schedulers to longer production campaigns to minimize the overall changeover cost. To shorten changeovers and thereby promote shorter campaigns, Shigeo Shingo developed an effective work practice that has come to be known as SMED, which is effective at reducing the time required for all mechanical tasks and adjustments required by the changeover. And SMED has proven to be very effective in process operations. However, many process steps experience an additional set of changeover challenges to get the product back to first-grade specifications after the changeover. These require that SMED broaden its focus to include improving the process chemistry and/ or physics to reduce variation and reach aim conditions more quickly and to shorten test lab response time. SMED can offer benefits beyond product changeovers; it has also proven to be quite effective in improving nightly or weekly sanitation cycles.

Chapter 24

Production Stability For product wheels or any scheduling strategy to be practical, the manufacturing process must be stable and capable. Without a reasonable degree of operating stability, it is foolish to expect that any schedule could be regularly and reliability followed. The relationship between production stability and scheduling is a two-way street. There are things a repeatable scheduling strategy can do to enhance operating stability and things that stability can do to make following the schedule much more likely. The most frequent causes of operating instability are (in no particular order, but numbered for ease of reference): 1) Personnel call-outs due to illness 2) Lack of raw materials 3) Yield losses or high scrap rates 4) Insufficient inventory 5) Capacity mismanagement 6) Lack of operating discipline 7) Equipment failures 8) “Minor stops” after a changeover 9) Unexpectedly long changeovers 10) Running at reduced rates due to equipment issues 11) Insufficiently trained operators Number 1 is a very real issue, made worse by the pandemic. Not an easy problem to solve, but it is HR’s responsibility to do whatever can be done. Number 2 falls in Procurement’s lap, but a regular and predictable schedule can make it easier to forecast needs and collaborate with suppliers and perhaps achieve a higher likelihood of materials being available when needed. The third item is traditionally considered the quality function’s or the technical group’s responsibility; here again, a structured schedule that sequences DOI: 10.4324/9781003304067-28

203

204  ◾  Prerequisites to Good Scheduling

products to simplify changeovers can help reduce the portion of material lost on changeovers. So there are a number of things that the scheduling strategies we’ve been promoting can do to help increase operating stability. The fourth cause, insufficient inventory, can be due to a lack of schedule achievement or inventory inaccuracy. Implementing and emphasizing the KPIs recommended here will drive better schedule achievement and stronger operating discipline. Lack of appropriate capacity management can be the strongest derailer of all. Realistic capacity must be understood and must be discounted for the OEE factors described later in this chapter. And the scheduling process must not schedule beyond effective capacity. Capacity management was discussed in Chapter 21. Most of the remaining factors are related to the equipment and how well it is maintained, and that’s the focus of this chapter. If those things are done well, it goes a long way towards creating a stable operation that can execute reasonable schedules.

Total Productive Maintenance Total productive maintenance (TPM) is an effective process to do that; it’s a philosophy, a set of principles, and specific practices aimed at improving manufacturing performance by improving the way that equipment is maintained. It was developed in Japan in the 1960s and 1970s, based on preventive maintenance and productive maintenance (PM) practices developed in the United States. But where PM is focused on the maintenance shop and mechanics, TPM is team based and involves all parts and all levels of the organization, including plant managers, supervisors, and operators. It drives towards autonomous maintenance, where the majority of maintenance is done by those closest to the equipment, the operators. If operators do more of the routine maintenance, it frees the maintenance group up to focus on equipment modifications and enhancements to improve reliability. The goal of TPM could be described as the development of robust, stable value streams by maximizing overall equipment effectiveness (OEE). Some key elements of TPM are: ◾ Preventive maintenance: time-based maintenance, maintenance done on a schedule designed to prevent breakdowns before they can occur ◾ Predictive maintenance: condition-based maintenance, using instruments and sensors such as vibration monitors to try to anticipate when equipment is about to break down so that it can be fixed before failing ◾ Breakdown maintenance: repairing the equipment after a breakdown occurs ◾ Corrective maintenance: ongoing modifications to the equipment to reduce the frequency of breakdowns and make them easier to repair

Production Stability  ◾  205

◾ Maintenance prevention: design equipment that rarely breaks down and is easy to repair when it does fail ◾ Autonomous maintenance: team-based maintenance done primarily by plant floor operators Autonomous maintenance is perhaps the most important element. It provides a much more timely response to problems, faster solutions, and therefore less downtime per event. But for autonomous maintenance to be effective, operators have to be trained, both in their normal job functions and in the tasks required to keep the equipment running smoothly. Unfortunately, that is not always the case; many plants don’t have any formal operator training programs and rely on experienced operators to mentor the new ones. That can be very effective, and can promote comradery within the workforce, but has the danger of passing bad habits and practices from generation to generation. So manufacturing operations should drive towards autonomous maintenance to the full extent possible but also ensure that procedures are documented and operators have the training required to be effective. The biggest challenge of TPM is often the culture change required, moving from a mindset that the maintenance group owns accountability for equipment performance to one where everyone in the plant has that accountability. Plant leadership has the responsibility for creating and promoting that culture and for actively participating in maintenance and reliability discussions and decisions. This is essential; in plant cultures where equipment reliability is viewed as strictly a maintenance function, no matter how good a PM program the maintenance group develops, the necessary time for PMs may not be allocated. Although operations managers may support the concept of PMs, when the time comes they may view making today’s production target a higher priority. The “productive” aspect of TPM is that all of the preceding should be done in a way that is economical and effective. It is not the intent of this chapter to explore TPM in great depth but to illustrate the things that can be done to create a stable production platform so that planned schedules can be expected to be achieved.

TPM Relevance in Process Industries Asset productivity is more important to the effectiveness of a process plant than it is to a discrete parts plant. The equipment tends to be larger, more expensive, and less cost-effective to duplicate. Therefore the process industries have an even greater need to do everything within reason to improve the reliability and uptime of the equipment and a greater need for the benefits that TPM can bring. And because the equipment is more often the bottleneck to increased flow than labor is, equipment breakdowns generally result in a loss of throughput

206  ◾  Prerequisites to Good Scheduling

and therefore revenue. Any time that asset utilization reaches the 90%–95% range, achieving high equipment reliability becomes even more critical. In the words of TPM for Every Operator, “The busier you are, the more you need TPM.”

TPM Saves Money There is a common misconception that reliability and availability could be increased by increasing the maintenance budget, by hiring more mechanics, and by stocking more spare parts. TPM breaks that paradigm by enabling reliability and availability to be raised without budget increases. In fact, TPM frequently reduces overall maintenance costs even while increasing equipment uptime. Because operators are now doing most of the routine maintenance, fewer mechanics may be needed. With fewer frequent breakdowns, fewer repair parts are consumed. The higher equipment uptime resulting from an effective TPM process enables smoother, more stable material flow with fewer hold-ups and faster and more cost-effective product wheels.

Overall Equipment Effectiveness (OEE) One of the most widespread measures used to gauge the stability of an operation and the underlying effectiveness of the TPM effort is OEE. The reason for its popularity is that it captures in a single metric all the factors that detract from optimum equipment performance. OEE is the product of three factors: ◾ Availability ◾ Performance ◾ Quality

Availability Availability captures all downtime losses, including breakdown maintenance, preventive maintenance, minor stops, and time spent in setup or changeover. Availability is calculated as actual operating time divided by planned production time: Actual Operating Time Planned Operating Time Planned Operating time - Breakdowns - PMs - Changeover time - Minor stops = Planned Operating Time Availability =

Production Stability  ◾  207

Note that set-up time or changeover time should not include the time getting properties back within specification after the changeover, because that loss is captured in the quality factor in OEE. Alternatively, the time getting properties back within specification could be included in changeover time, if the material losses during that period are not counted as quality losses. Either way is correct; just be sure you are not penalizing for those losses twice.

Performance Performance captures the loss in productivity if equipment must be run at less than the design throughput rate because of some equipment defect. For example, chemical batches can take longer to heat up or react if residue has built up on vessel walls, thus impeding heat transfer. Rotating machinery, paper winding equipment, or plastic film processing equipment may have to be run at slower speeds to prevent excess wobbling when bearings are worn. A line making frozen meals may have to be slowed down if the mashed potato or the gravy dispenser is partially clogged. Performance is calculated as actual throughput divided by rated throughput:

Performance =

Actual Throughput Rated Throughput

A caution on performance calculation: In the process industries, there are often rate limitations due to the requirements of the material being processed, not due to any equipment defect. For example, when heat-treating sheet goods to set properties in, some products may require the heat treater to run slower to allow for more time at temperature than required for other products. Likewise, some paint resin formulas may require more “cook time” to completely react compared to other resins run on that equipment. A salad dressing cup filling line may be able to run at a high rate when filling viscous dressings like ranch or Caesar but may have to run at a slower rate when filling less viscous vinaigrette dressings to prevent sloshing and spilling. Conversely, the Blue Lakes bottling lines have to run slower with more viscous dressings to prevent air entrapment, bubbling, and overflowing. Because these rate limitations are due to product characteristics rather than equipment performance, the performance metric should not be penalized. (For equipment that must run at different rates for different products, rated throughput should be a weighted average based on the actual product mix.)

Quality Quality captures the loss in equipment productivity when out-of-specification product is being made, including material that must be scrapped, material

208  ◾  Prerequisites to Good Scheduling

that must be reworked to be acceptable, and yield losses during startup or when going into or coming back from a product changeover. Quality =



Quantity of First Grade Material Total Quantity produced

OEE is then calculated as: OEE = Availability × Performance × Quality



Figure 24.1 diagrams the buildup of OEE from its component factors.

SHIFTS NOT PLANNED NO DEMAND

PM's

ACTUAL OPERAT'G TIME

AVAIL ABILITY

SET UP TIME

ACTUAL THROUGH PUT

PERFORMANCE

EQUIV TIME AT REDUCED RATE

SCRAP, YIELD LOSS

FIRST GRADE MATERIAL

Figure 24.1  Components of the OEE calculation.

QUALITY

PLANNED OPERATING TIME TOTAL MATERIAL PRODUCED

RATED THROUGHPUT

TOTAL TIME

UNPLANNED DOWNTIME

OEE

Production Stability  ◾  209

Calculation of OEE As an example of the calculations, consider one of the Blue Lakes mix tanks. The relevant data and the calculation of OEE are: ◾ The plant is operated twenty-four hours, seven days per week. Not all mix tanks nor all packaging lines run that many shifts, but this tank does. ◾ Thus, the planned operating time is (24 × 7) = 168 hours/week. ◾ There are, on average, four hours per week of unplanned downtime. ◾ PMs are done on Saturdays and take about four hours. ◾ This tank typically runs 28 different dressing formulas per week, with a one-hour clean between each different dressing. On average eight of those cleans are allergen cleans and take two hours. ◾ Actual operating time is 168 − 4 − 4 − 20 − 16 = 124 hours/week. ◾ Availability = 124 hours ÷ 168 hours = 74%. ◾ Each mix tank is agitated, and the agitator bearings can wear, particularly when running very viscous dressings. The bearings on this agitator were replaced a few months ago, so the agitator can run at the rated speed and there is no rate reduction to include. ◾ All of the dressings are ultimately within all quality specifications and salable, but four batches per week require an average of two extra hours to enhance uniformity or blend in additional ingredients to bring some property within specs. ◾ Quality = (168 − 8) / 168 = 95%. ◾ OEE = 74% × 100% × 95% = 70%. Now that we have that understanding, we can decide what knobs we can turn to improve mix tank throughput. If we could cut our first pass off-spec rate in half, it would give us four additional hours of capacity each week. That might require better tank temperature control, better agitator speed control, or more precise metering of sensitive ingredients. A much richer target would be the cleans. Each clean must be done thoroughly and completely, but should they take one and two hours? If the regular and allergen cleans could be cut to 30 minutes and one hour, respectively, it would provide another 18 hours of throughput weekly. SMED, covered in the previous chapter, might yield an improvement in that range.

VSM Data Boxes: OEE If you are preparing a Value Stream Map (VSM) to provide data and insights for your scheduling strategy design (and you should!), the current value of OEE should be included in the data box for each process step. Because a major function of a VSM is to capture barriers to smooth flow, and all OEE losses are flow barriers, its inclusion is critical to a full understanding of flow.

210  ◾  Prerequisites to Good Scheduling

To further that understanding, the components of OEE (availability, performance, and quality) should also be shown. It is often helpful to show the changeover times as separate factors because they have such a strong influence on scheduling. In what is considered one of the landmark books on TPM, Introduction to TPM, Seiichi Nakajima lists six big equipment losses:

1. Equipment failure 2. Setup and adjustment 3. Idling and minor stoppages 4. Reduced speed 5. Process defects 6. Reduced yield

The OEE metric captures all of Nakajima’s six big losses. Another benefit of OEE is that it gives operators, who are responsible for most of the routine equipment maintenance under TPM, an indication of how well they are meeting that responsibility. Therefore, OEE and its components should be a prominent part of any visual management activity.

Non-Standard OEE Metrics Although there are clear standards on what losses are covered in the OEE metric and how to calculate it, a lot of plants don’t follow the standard; we’ve encountered a number of different measures called OEE. There are several possible reasons: 1) Not having a full understanding of what should be included. 2) Doing it properly is perceived as too difficult. They don’t currently collect data on all the relevant losses and don’t feel the effort to do that would be worthwhile. 3) Deliberately using a different formula to hide problems and make themselves look better than they really are. 4) Deliberately using a different formula to emphasize factors they think are most important to drive improvement. The first two can possibly be overcome by better education on the value of OEE to drive productivity improvements and what should be included. The third is the least excusable reason and requires a change in corporate culture to overcome. The message from the top has to be that OEE is not a metric to drive punishment for poor performance. It is a metric that should be used to understand where and how you can improve your current performance.

Production Stability  ◾  211

The fourth can be appropriate if it is not being done simply to hide embarrassing data, the factors included are where the operation most needs to make improvements, and everyone involved understands what is actually being measured. We developed product wheels for a nutraceutical plant that doesn’t include Quality when measuring OEE on its packaging lines. Three to four percent of the product fails quality standards, but it is rarely because of the product itself; it is almost always that the documentation, vitally important in any regulated industry, has not been completed in time. It may delay release of the product, but the product rarely gets scrapped, so no time is lost on the line. OEE is used to focus attention on changeovers, minor stops, and run rate issues, which are much more significant losses for that plant. Quality is addressed every day in the morning rack-up, so it is not being ignored, and programs have begun to highlight operator awareness of the importance of timely documentation. Another bad practice I have seen is with multi-plant companies where each plant uses a different version of OEE. They do this because they know that corporate is measuring their plant performance against their sister plants and don’t want to look bad in comparison, so they use an OEE calculation designed to emphasize their strengths. Not only is the leadership following a very bad practice, but it is also unfair because different plants have different generations of equipment and more or less difficult product mixes and shouldn’t be expected to perform equally well. Everyone involved with collecting data, analyzing results, or deciding on actions to take based on OEE should understand that there are sometimes very appropriate actions that can lower the OEE value. For example, if a piece of processing equipment or a full production line has modest utilization, say 60%, some of the available time could be used to do more changeovers, reduce production cycles, reduce inventories, and make the operation more agile and responsive. While it may be a very good thing to do for all of those reasons, it will reduce the OEE value. Changeover time is an OEE loss, while idle time is not. In summary, OEE can be a very useful metric to drive improvements in operating performance, which is the ultimate goal of TPM. It does this by highlighting the various losses that impede smooth flow. But it must be done with a full understanding of what is being measured and how to interpret the results.

Summary Proper equipment maintenance is important to the performance of any manufacturing operation and is required to provide the operating stability and predictability to support any scheduling methodology; if you can’t rely on your equipment, you can’t rely on your schedules. Process equipment tends to be

212  ◾  Prerequisites to Good Scheduling

expensive, which means that providing excess capacity (to mitigate equipment downtime) is rarely an economic option, nor is it a good business practice. Therefore, you have to find ways to get the most out of what you’ve got. TPM is a set of principles and practices that have proven to be effective in improving maintenance and, more importantly, improving overall equipment performance. TPM begins to create a culture that involves all levels of the organization in the total maintenance process and puts direct responsibility for much of the maintenance work in the hands of those closest to the equipment: the operators. TPM emphasizes teamwork and operator involvement, and the need for vigorous, visible commitment and leadership from upper management. OEE measures overall equipment performance and indicates the effectiveness of a TPM program. While OEE is a very powerful and useful metric, it must be used appropriately. It should be used to drive improvement in equipment performance and effectiveness, not to punish plants with poor OEE numbers.

Chapter 25

Cellular Manufacturing Before embarking on the scheduling process design, you should explore cellular manufacturing and its fit with your process equipment footprint. Applying cellular manufacturing to process layouts where it is applicable can greatly simplify manufacturing and scheduling. The cellular concept separates material flow paths and product groupings into much more manageable subsets, meaning that each wheel will have fewer products and fewer variables to manage. Changeovers almost always become simpler and quicker even before the structured scheduling strategy takes those benefits further.

Typical Process Plant Equipment Configurations Cellular manufacturing only applies where you have similar equipment in parallel at one or more points in the process. It is particularly beneficial where there is parallel equipment at several steps, like the configuration shown in Figure 25.1, a pattern typical of many process industry plants. There is a small number of key processing steps, in this case, four, and there are a few (three or more) production lines, machines, tanks, or reaction vessels in parallel at each step. The parallel machines are quite similar, and often a specific material can be processed by any one of them. Occasionally, the machines or vessels have some unique capabilities such that some materials must go to a specific machine or vessel. Process plants usually require this array of equipment to handle the high volume of material to be produced. Practical equipment size limitations prohibit the design of a single machine or vessel large enough to process the full throughput required. Product mix considerations, i.e., the high degree of product variety, would encourage the use of many small machines to give more flexibility, but economies of scale have deterred plant designers from going in that direction. As capital cost optimization has traditionally

DOI: 10.4324/9781003304067-29

213

214  ◾  Prerequisites to Good Scheduling

MATERIAL FLOW

Step 1 Tank 1

Step 1 Tank 2

Step 2 Reactor 1

Step 2 Reactor 2

Step 3 Machine 1

Step 4 Machine 1

Step 1 Tank 3

Step 4 Machine 2

Step 1 Tank 4

Step 2 Reactor 3

Step 3 Machine 2

Step 2 Reactor 4

Step 3 Machine 3

Step 4 Machine 3

Step 4 Machine 4

Figure 25.1  Typical process industry equipment footprint – functional configuration.

overridden good lean thinking in process plant design, the result is a few large vessels or machines at each process step. This equipment configuration is highly valued for the flexibility it offers. If a batch of material is leaving step 1, and one of the step 2 machines is down for maintenance, there may be others available to process the material. The result is that flow paths are often as shown in Figure 25.2. All of the flexibility in the system is exploited, but generally with more negative than positive consequences. There is frequently a belief that utilizing this flexibility maximizes asset utilization, although the opposite is usually true. This mode of operation brings a number of problems. Because flow is not well coordinated, material doesn’t always flow directly from one step to the next but may be sent to storage. Thus large WIP storages can be created. Flow becomes unsynchronized, is difficult to visualize, and is even more difficult to manage. Because each piece of equipment can process most or all of the product types, each scheduling sequence would have the entire product line-up to consider, making the optimum solution difficult, if not impossible, to find. Quality may suffer for two reasons. There may be a significant time lapse between each process step, so any defects or out-of-spec material may not be discovered for some time, making all of the intermediate materials suspect. Even with this simple-looking arrangement, there are 192 (4 ´ 4 ´ 3 ´ 4) possible flow path combinations. Since no two machines or vessels are perfectly alike, they may not produce exactly the same product, thus providing 192 different ways that process variabilities can add up. A Statistical Process Control (SPC) specialist would tell you that you don’t have a process, you have 192 different processes. With so many variables, root cause analysis of product defects can become difficult.

Cellular Manufacturing 

Step 1 Tank 1

Step 1 Tank 2

Step 2 Reactor 1

Step 2 Reactor 2

Step 3 Machine 1

Step 4 Machine 1

Step 1 Tank 3

Step 2 Reactor 3

Step 3 Machine 2

Step 4 Machine 2

Step 1 Tank 4

Step 2 Reactor 4

Step 3 Machine 3

Step 4 Machine 3

Step 4 Machine 4

Figure 25.2  Typical process industry flow patterns.

Because there are thought to be alternate paths available whenever a piece of equipment fails, there is far less urgency to maintain the equipment appropriately. Thus, with time, equipment performance as measured by OEE deteriorates. The biggest problem from our perspective is that it makes any of the repetitive scheduling applications much more difficult because each production line or machine may have to process most or all of the full product line-up.

Cellular Manufacturing Applied to Process Lines In discrete parts manufacturing processes, cellular manufacturing has traditionally required relocating the equipment into U-shaped or L-shaped patterns, to provide shorter paths for operators to travel. It has therefore been thought that cells weren’t applicable to most process operations, because the equipment is most often very large and expensive to relocate. But if one considers that the primary benefit of the U or L arrangement is that one worker can operate several machines as a way to optimize labor productivity and that all the other benefits of cells can be attained without rearrangement, it becomes apparent that a process plant can reap these benefits by managing flow in a cellular fashion, without any equipment relocation. The key is to think in terms of flow rather than geography.

◾  215

216  ◾  Prerequisites to Good Scheduling

In World Class Manufacturing, Richard Schonberger describes cellular manufacturing and cites industries that are well suited to this arrangement. He mentions light assembly, such as electronics and machining industries, but makes no mention of process industries. This may have been appropriate in 1986 when the western world was just beginning to adopt lean concepts and it was important to publicize the applications with the most obvious benefit. What is somewhat disappointing is that during the next 30 years, little was published to expand the view of the applicability of cellular techniques. The basic concept is straightforward: Start by grouping all process materials or products into families requiring similar process conditions. Then identify the process equipment required by each family, but instead of creating a work cell by rearranging the equipment, create virtual work cells by defining the acceptable flow patterns. Figure 25.3 shows what this would look like for the process diagrammed in Figures 25.1 and 25.2. Again, no equipment would have to be moved; the new, more limited flow patterns would simply have to be defined and followed. It must be recognized that the numbers don’t always work out perfectly but that reasonable compromises can usually be found which will give most of the benefit. In the case shown, since there are only three machines at step 3, one must be shared between cells 2 and 3. If that machine didn’t have enough capacity to process the total throughput of the two cells, additional compromises would have to be made. But even with necessary compromises, flow patterns tend to be much simpler.

Step 1 Tank 1

CELL 1

Step 1 Tank 2

Step 1 Tank 3

CELL 2 Step 2 Reactor 1

CELL 3 Step 2 Reactor 2

Step 3 Machine 1

Step 4 Machine 1

Step 2 Reactor 3

Step 3 Machine 2

Step 4 Machine 2

Figure 25.3  Grouping into virtual work cells.

Step 1 Tank 4

CELL 4 Step 2 Reactor 4

Step 3 Machine 3

Step 4 Machine 3

Step 4 Machine 4

Cellular Manufacturing 

It should be noted here that although process plants have the logical process configurations shown, they are not generally arranged that way geographically. The equipment is often dispersed in what appear to be illogical arrangements. One reason for that is that many process plants are 60 years old or more and have been re-engineered several times as products became obsolete and new ones replaced them. Because process equipment is expensive, there has been a tendency to try to reuse as much of it as possible, to add new equipment only when absolutely necessary, and to locate it wherever it would fit, not where it would facilitate smooth flow. So as plants evolve, the equipment arrangements typically become more and more scattered. The point is that virtual work cell patterns are not as obvious as these diagrams would make them appear. Although the new virtual flow paths are the more readily apparent aspect of cellular manufacturing, the more significant benefit is that the product portfolio can be divided up into families, so that each piece of equipment must process far fewer product types. Having fewer products on each wheel significantly simplifies wheel design and operation. For example, if all the salad dressings assigned to a specific bottling line have the same allergen group, then allergen cleans go away for that line. Or if the products assigned to a specific film sheet annealer are all heat-set within a narrow range of temperatures, the time required to change the temperature on product changeovers is far less. The advantages of the virtual work cell concept are: ◾ Flow becomes far easier to understand, visualize, and manage. ◾ Flow tends to be better synchronized and more continuous, with less material being transported to storage, so WIP and material handling are reduced. ◾ Quality improves because feedback is much more immediate. ◾ As depicted in Figure 25.3, we now have only four possible flow paths instead of the 192 we had before, so product variability is reduced. ◾ Each cell is generally processing a subset of the full product mix with similar requirements, so changeovers become far easier, and usable capacity increases. ◾ And most importantly, it becomes far more practical to apply a structured schedule because of the limited number and clearly defined flow paths and a smaller, more tightly knit product family.

Synthetic Sheet Manufacturing Example Several years ago we applied product wheels to the forming machines making large rolls of synthetic sheeting for weatherproofing houses and making

◾  217

218  ◾  Prerequisites to Good Scheduling

hospital gowns and curtains. But before we began the wheel design, we analyzed the process for cellular applicability. Figure 25.4 shows the process, with four machines that form the sheet, four bonders that heat-set properties into the film sheeting, three slitters that cut the 10–12-foot-wide rolls into narrower rolls, and three choppers that cut the rolls across the width into shorter rolls or sheets. The footprint was well suited to cellular flow, resulting in the cellular layout in Figure 25.5. There were only three slitters, so one had to be shared between cells 1 and 2. That posed no problem because the slitters could easily outrun the upstream equipment by a factor of two. Changing the positions on the rotary slitting knives to go to a different cut pattern was a very quick operation, only three to five minutes, so the shared slitter added little complexity as it was shifted from one cell to the other. And there was enough WIP capacity upstream of the slitters to provide the buffering required. The choppers had to be shared among the four cells because each was geared to a specific slit width range. Even with those compromises, the flow is restricted to twelve possible combinations, where before the formation of virtual cells, 144 possible paths (that is, 4 × 4 × 3 × 3) existed. Thus, the key advantages of cellular manufacturing apply. This virtual cellular pattern and the allocation of product families to specific cells allowed for much simpler product wheels to be applied to the forming machines.

Roll Forming Machine 1

MATERIAL FLOW

Roll Bonder 1

Roll Forming Machine 2

Roll Forming Machine 3

Roll Bonder 2

Roll Slitting Machine 1

Roll Bonder 3

Roll Slitting Machine 2

Chopper 1

Chopper 2

Packaging

Figure 25.4  Sheet goods process.

Roll Slitting Machine 3

Chopper 3

Roll Forming Machine 4

Roll Bonder 4

Cellular Manufacturing 

Roll Forming Machine 1

Roll Forming Machine 2

Roll Forming Machine 3

Roll Forming Machine 4

Roll Bonder 1

Roll Bonder 2

Roll Bonder 3

Roll Bonder 4

CELL 3

CELL 1

Roll Slitting Machine 1

Chopper 1

Roll Slitting Machine 2

CELL 2 Chopper 2

Chopper 3

Roll Slitting Machine 3

CELL 4

Packaging Figure 25.5  Sheet goods process with cellular flow.

Virtual Cell Implementation in a Synthetic Rubber Production Facility We wanted to apply product wheels to a synthetic rubber production process but found it difficult because of the uncoordinated flow paths and lack of any defined product structure. The footprint looked very well suited to cellular flow, and it worked out more advantageously than usual because a configuration was found where the demand, capacities, and numbers of pieces of process equipment could be almost equally balanced. Figure 25.6 depicts the process, which begins with three tanks where monomers and other ingredients could be weighed so that the correct quantities could be fed into one of six polymerization kettles. After the batch polymerization was completed, the polymer could be fed to one of three tanks where a chemical process called emulsion stripping could take place. The stripped emulsion was then stored in tanks (not shown) and later cast onto one of six freeze rolls where the emulsion would solidify into a thin sheet. The sheet was then gathered into a rope and cut into pellets, which were bagged on one of three bagging lines, and then the bags of pellets were shipped to the finished product warehouse. Customers of this synthetic rubber would use it as received or compound it with other materials to make high-pressure hoses, industrial belting, and gaskets and seals for refrigerator doors and car trunk lids. The product line-up consisted of three major families: type F, type J, and type R. The volume was distributed among the families in the approximate

◾  219

220  ◾  Prerequisites to Good Scheduling

Weigh Tank 1

Polymerization Kettle 1

Polymerization Kettle 2

Emulsion Strip 1

Casting Roll 1

Polymerization Kettle 3

Weigh Tank 3

Polymerization Kettle 4

Polymerization Kettle 5

Emulsion Strip 2

Cutter 2

Bagging line 1

Casting Roll 3

Casting Roll 5

Cutter 3

Polymerization Kettle 6

Emulsion Strip 3

Casting Roll 2

Casting Roll 4

Cutter 1

Weigh Tank 2

Cutter 4

Casting Roll 6

Cutter 5

Bagging line 2

Cutter 6

Bagging line 3

Figure 25.6  Synthetic rubber manufacturing configuration.

ratio of 45% F, 35% J, and 20% R. Each family comprised 8–15 individual grades. The pre-cellular scheduling process was: ◾ A monthly sales and operations planning (S&OP) process would determine the production needs for the coming month, by specific grade within each family. ◾ The plant production scheduler would determine the grade with the most immediate customer due date, say grade J-43. ◾ The entire production facility was set up to make grade J-43. Production would continue until the full monthly requirement for J-43 had been produced. ◾ The scheduler would then determine the grade with the next most immediate customer due date, say F-6. ◾ The entire facility would be reconfigured to produce F-6. The entire monthly requirement of F-6 would be produced. ◾ This scheduling process would be repeated until all production requirements for the month had been met. Operating in this manner created several problems: ◾ The polymer area, the stripping area, the casting and cutting area, and the packaging area were run independently, each with its own area

Cellular Manufacturing 

supervisor. Each supervisor was directed to optimize flow and productivity in their area, with little regard to the full operation. Thus there was little coordination between process areas. Consequently, flow was very nonsynchronous. Batches often had to wait for hours in the storage tanks between steps and often got out of sequence. Asset productivity suffered as a result. ◾ Because flow patterns were confusing, and because there was no single individual responsible for overall flow, batches occasionally got pumped to the wrong place and subsequently had to be “ditched” (sent to the waste sewer). ◾ Large quantities of WIP usually resided in the storage tanks as a result of the flow discontinuities. ◾ Because each grade was produced only once per month, large quantities of finished product inventory had to be maintained in the warehouse. ◾ Transitions from one family to another were complex and time-consuming. ◾ After major transitions were mechanically complete, and the process restarted, it could take several hours for viscosity, the most important product characteristic, to come within customer specifications. This created significant yield losses because of material waste. This time required to reach aim (i.e., target) conditions would further deteriorate asset productivity. ◾ It was extremely difficult to implement product wheels on any major step because every piece of equipment had to process the full line-up. To overcome these problems, a cellular configuration was designed for the rubber process (Figure 25.7). The numbers worked out remarkably well in this case. Because there are three major product families, it seemed logical to create three virtual cells. And because there are either three or six pieces of equipment at each significant process step, dividing the equipment into cells was straightforward. Even though product demand wasn’t equally distributed across families, with type F seeing 45% of the total demand, the productivity improvement resulting from virtual cell implementation enabled even the type F cell to have enough capacity to produce to demand. The results of operating following the cellular plan were dramatic. The most significant benefits resulted from the reduction in yield losses while getting back within viscosity specifications after a changeover. In the old scheme, it would take several hours, perhaps four or more, to get back to aim conditions after a change from F to J, J to R, or R to F. Because the new cells would each produce only grades within a type, target viscosity could be reached rapidly, typically within two to four minutes. This not only improved yield significantly, it also improved asset productivity because the time previously spent getting on aim could now be spent producing first-grade product.

◾  221

222  ◾  Prerequisites to Good Scheduling

Type F CELL

Type J CELL

Weigh Tank 1

Polymerization Kettle 1

Polymerization Kettle 2

Polymerization Kettle 3

Emuls Strip 1

Casting Roll 1

Polymerization Kettle 4

Bagging line 1

Polymerization Kettle 5

Casting Roll 3

Casting Roll 5

Cutter 3

Polymerization Kettle 6

Emuls Strip 3

Casting Roll 2

Cutter 2

Weigh Tank 3

Emuls Strip 2

Casting Roll 4

Cutter 1

Type R CELL

Weigh Tank 2

Cutter 4

Bagging line 2

Casting Roll 6

Cutter 5

Cutter 6

Bagging line 3

Figure 25.7  Synthetic rubber virtual cell configuration.

The organizational structure was also changed, to be based on flow rather than on function. Instead of an area supervisor for each process step, there was now a flow manager for each cell, responsible for all the equipment within that cell and for synchronizing flow within the cell. Specific benefits recorded by the business were: ◾ Scrapped material was reduced by 28%. ◾ Finished product variability, measured by standard deviation of viscosity of first-grade product, was reduced by 15%. ◾ Lead time through the complete process was reduced by 28%. ◾ The average changeover time was reduced from eight hours to three hours. ◾ The average time to reach aim viscosity targets was reduced from five hours to five minutes. ◾ Usable capacity was increased by several million pounds per year. ◾ Finished product inventory was cut in half. It should be emphasized that no equipment was relocated. No significant equipment modifications were required. There was some process piping removed, and some valves were locked out so that the cellular boundaries couldn’t be crossed inadvertently. Lines were painted on the floor to designate cell assignment, and signs were hung over each process vessel or machine to indicate its cell. Thus, there was some slight cost involved, but minuscule compared to the benefits. With these flow improvements and product family allocations, it became straightforward to design a very effective product wheel for each cell.

Cellular Manufacturing 

Would Cellular Flow Apply to the Salad Dressing Operation? There are a number of manufacturing operations where the normal configuration is already cellular. Most packaging operations fit in this category: Each line is self-contained and relatively close-coupled, with no interconnection with the other lines. We applied product wheels to each of the twelve packaging lines at a Nature’s Bounty tablet packaging plant in Florida, and there was no reason to consider cells because each line was completely independent of the other lines. More on that later. There are other situations where you might think cells would apply, but in reality, the results would be more negative than positive. The Blue Lakes salad dressing plant is a good example. It may appear that you could apply cells by combining each mix tank with a few of the packaging lines, but that would actually make the changeover situation worse and reduce throughput. To provide the required dressings, the mix tanks must run at a very high utilization. For maximum utility, the large mix tanks must process the highvolume products, and the smaller tanks all of the lower-volume dressings. Within that split, the dressings are assigned to tanks grouped by allergen content, particulate content, and then by specific dressing. Thus each specific dressing must be made in a designated mix tank. Because the same dressing may go to retail bottles, tubs, and packets of various sizes, there are many crisscrossing flow lines between the tanks and the packing lines.

Group Technology But there is a subset of cellular manufacturing which has been applied here, Group Technology (GT). GT is the engineering term for the process of allocating segments of the entire product line-up to specific pieces of equipment. (The APICS dictionary defines group technology as: “an engineering and manufacturing philosophy that identifies the physical similarity of parts (common routing) and establishes their effective production. It … facilitates a cellular layout.”) What we just described for the allocation of dressings to mix tanks is an example of GT, where we’ve assigned a well-bounded subset of the full lineup to each mix tank. We have done the same thing for the packaging lines, by grouping all packet products first by width, then by length, and then by allergen content, and assigning each group to a packet line. This provides the same benefits that cellular flow brings to a complete manufacturing footprint, but on a narrower scale, to specific pieces of equipment or self-contained production or packaging lines. And it simplifies the product wheels running on those assets, with each having a narrow slice of the full line-up to schedule and with quicker and less expensive changeovers.

◾  223

224  ◾  Prerequisites to Good Scheduling

Reduction in bottle sizes per packaging line

12 10 8 6 4 2 0

1

2

3

4

5

6

7

No of bottle sizes per line - prior

8

9

10

11

12

13

No of bottle sizes per line - new

Figure 25.8  Reduction in bottle sizes per line.

The Nature’s Bounty plant we mentioned is a dramatic example of the benefits of GT. The plant in Florida packaged tablets, capsules, and gel caps into bottles of various sizes and colors. There is a lot of variety in the packaging; in addition to bottle size (12 variations), there are variations in bottle color (4), cap type (child-proof, logo cap, gear-shaped, flip-cap), allergens (8, in 14 different combinations), and presence or absence of cotton or desiccant. Of all these, bottle size had the most time-consuming changeover, requiring all rails, guides, and equipment heights up and down the line to be adjusted. The planning manager had made an attempt at GT, but without a defined allocation process, had trouble making it work. Consequently, during any given month a packaging line may be faced with nine or ten different sizes. Before embarking on product wheel design, we implemented GT, with the results shown in Figure 25.8. The average number of sizes run per line dropped by 56%, with three of the lines getting down to a single bottle size! Prior to GT, no line had run fewer than four sizes. This allocation of bottle sizes to specific lines gave them a four to five percentage point improvement in OEE. That and the subsequent application of product wheels gave them a total throughput increase of 35%! As we implemented the scheduling wheels we started to see immediate and lasting benefits. We set up lines by bottle size since that was our longest changeover event and then would run numerous

Cellular Manufacturing 

SKUs of the same bottle size and only do label changes which were our quickest changeover. (Dean Bordner, prior Senior VP of Operations, Nature’s Bounty)

Summary Where the equipment configuration lends itself to cellular manufacturing, it should be designed and implemented before the scheduling strategy is developed. Repetitive schedules are far easier to put in place with the small number of flow paths that a cellular arrangement requires, and more importantly, with a reduced number of products to be processed on each piece of equipment. And where the configuration does not lend itself to cells, the concept of group technology should be applied wherever there is similar equipment in parallel. Not only does this reduce the number of products on each flow path, but a grouping can usually be found which reduces changeover time and cost, leading to much more effective production sequences.

◾  225

Chapter 26

Managing Bottlenecks and Constraints Many production lines have bottlenecks or constraints, steps in the process that don’t have enough effective capacity to handle the demand or cannot handle the demand under certain conditions. That can be because the step doesn’t have the inherent capacity required, that OEE factors (see Chapter 24) reduce the effective capacity to an insufficient level, or that some product mix and timing combinations exceed their capacity. For simplicity, we will call them bottlenecks in this book. Bottlenecks are more likely to be a problem in process manufacturing than in discrete parts assembly. Equipment can be very expensive, so there is a tendency not to build much excess capacity, and with sales volume growth and demand variability, bottlenecks arise. For process plants that run 24 × 7, which is frequently the case, adding extra shifts or scheduling overtime is not an option. Goldratt’s Theory of Constraints suggests using inventory to protect the bottleneck, but this is not often a practical option. And that only protects the bottleneck against upstream and downstream outages but does nothing to open the bottleneck up. Where bottlenecks exist and overtime and extra shifts are not options, possible relief can be gained by: ◾ Accelerating TPM and other OEE improvement programs to increase effective capacity. ◾ Capital equipment improvements to increase capacity. ◾ Rationalizing the portfolio to eliminate low-margin SKUs or SKUs requiring disproportional changeover times, i.e., SKUs whose contribution doesn’t cover the revenue lost in their changeovers. ◾ Adopting the structured scheduling strategies we’ve been recommending to reduce capacity lost in changeovers. This is generally the lowest cost option and one that provides the best business case, as few or no products have to be cut from the catalog. DOI: 10.4324/9781003304067-30

227

228  ◾  Prerequisites to Good Scheduling

Poor Scheduling Can Cause Bottlenecks As an example of how poor scheduling can create a bottleneck, let’s look at a cereal plant that manufactures two families of cereal, one formed into thick shapes like stars and circles and the other formed into relatively flat flakes of various shapes. The plant can be divided into three major areas (Figure 26.1): shape manufacturing, flake manufacturing, and packaging, which includes bagging, boxing, cartoning, and palletizing. From the data presented on the Value Stream Map, packaging has a utilization of only 75%, even though it takes the full output of both cereal production areas. However, in real life, the storage silos often became full and forced a production line to go down. Analysis revealed that although the packaging area appeared to have excess capacity, it became a flow constraint because the two manufacturing areas and the individual lines within those areas were being scheduled independently of each other with no coordination. An improved scheduling process, that coordinates all manufacturing lines, should smooth out the flow to packaging and thus open the bottleneck up. Implementing a structured, repetitive scheduling process like product wheels will allow all the manufacturing lines to be synchronized in a way that levels flow to packaging.

Moving Bottlenecks Identifying and managing bottlenecks can be difficult in process plants because the bottleneck may be at a different process step for one material being produced than for another material; the bottleneck may move as the process cycles through the various products being made. As an example, consider one sheet-forming and one bonding machine from the process mapped in the last

SHAPE MANUFACTURING STORAGE SILOS FLAKE MANUFACTURING

Figure 26.1  Cereal manufacturing.

PACKAGING

Managing Bottlenecks and Constraints  ◾  229

chapter, shown in Figures 25.4 and 25.5. As can be seen from the data boxes in Figure 26.2, each has an effective capacity greater than the demand, so neither appears to be a bottleneck. However, the values shown in the data boxes represent the averages taken across the full product line-up at the typical mix. When forming a sheet with high basis weight, the forming machine must run at much slower linear speeds, so for that product, the capacity will be less than demand and the machine becomes a bottleneck, as depicted in Figure 26.3. With other products, forming may have excess capacity while bonding may become the bottleneck. For products that must be bonded at a higher temperature, the line speed must be slower to allow the sheet to be in contact with the heated bonding roll long enough for complete heat transfer from the roll to the sheet. So when making products requiring high bonding temperatures, bonding becomes the bottleneck, as illustrated in Figure 26.4. Some people will say that in the long run, it really doesn’t matter, that it all averages out over some period of time. And that’s true if you have enough upstream and downstream inventory to completely buffer through the various flow limitations. But that is not always the case, and so it really does matter. Traffic on any major road or highway may be fairly light averaged over a 24-hour period but can be a nightmare during morning and evening rush hours. The instantaneous behavior is far more relevant than the average behavior. SHEET FORMING 1

BONDING 1

6

1.5

1.5

AVERAGE OF ALL PRODUCTS

Figure 26.2  Forming and bonding utilizations based on average effective capacity. SHEET FORMING 1

BONDING 1

6

1.5

PRODUCT 432A

Figure 26.3  Product that causes forming to be a bottleneck.

1.5

230  ◾  Prerequisites to Good Scheduling

SHEET FORMING 1

BONDING 1

6

1.5

1.5

PRODUCT 4516F

Figure 26.4  Product that causes bonding to be a bottleneck.

STEP 1

STEP 2

STEP 3

STEP 4

Figure 26.5   A simple example of a moving bottleneck.

Scheduling Moving Bottlenecks The fact that the bottleneck may move during the production cycle must be recognized so that appropriate bottleneck management strategies can be used with all process steps that can be bottlenecks. The most appropriate way to deal with the moving bottleneck problem is to approach it not as a moving bottleneck challenge, but as a scheduling challenge: “Which steps do I schedule, and what’s the best way to do that?” Figure 26.5 is a very simple example, a four-step process where step 2 is the bottleneck for some SKUs and step 4 is for others. Rather than focusing on the fact that the bottleneck is different for different SKUs, focus on run rates. For some SKUs the run rate is limited by step 2, and by step 4 for others. Calculate the effective run rate for each SKU based on its bottleneck step and the number of line hours needed to produce the forecast demand and build the schedule on that basis. Figure 26.6 illustrates this situation in a ketchup plant. The line shown receives ketchup from the kitchen where it is temporarily stored in a surge tank and then bottled in 14 oz, 20 oz, 28 oz, 32 oz, 44 oz, and 64 oz bottles. When running the large bottles, the rotary bottle filler is the bottleneck. With smaller bottles, the labeler is generally the bottleneck; but with products going to convenience stores in small cartons, the carton erector–case packer becomes the bottleneck. But the fact that the bottleneck can be in three different places is not an issue if the line is scheduled properly. The line run rate for the large bottles is calculated based on the filler rate for those sizes, as is the weekly number of production hours needed for each of the large bottle SKUs. Similarly, the run rates and weekly hour requirements are calculated for the other sizes based

SHRINK WRAP TUNNEL

Figure 26.6  Bottlenecks in a tomato ketchup plant.

SURGE TANK From KITCHEN

BOTTLE FILLING MACHINE

BN #1 (FOR 44 AND 64 OZ BOTTLES)

BOTTLE CAPPER

CASE PATTETIZER

ACCUMULATING BOTTLE CONVEYOR

PALLET STRETCH WRAPPER

BOTTLE LABELER

BN #2 (FOR 14 OZ BOTTLES) CARTON ERECTOR / CASE PACKER

BN #3 (FOR 14 OZ BOTTLES)

Managing Bottlenecks and Constraints  ◾  231

232  ◾  Prerequisites to Good Scheduling

on either the labeler or the carton erector–case packer. The schedule is then set to meet those production hour requirements with the sequence designed to maximize throughput. We resolved a very similar situation in packaging vitamin tablets. The bottle filler was the bottleneck with large bottles, but the capping machine became the bottleneck with very small bottles. So we developed product wheels based on the run rates and line hours required for each of the SKUs. Figure 26.7 depicts a more complex situation, with three main process steps and very little in-process inventory to de-couple them. (If there were significant in-process inventory, each step could be scheduled independently; the sheet forming process cited above is a good example.) The specific process shown is vitamin tablet manufacturing, where each of the three major steps, Blending, Compression, and Coating, can be a bottleneck depending on what is being produced. The first decision is where the pacemaker, the step to drive the schedule, should be. The schedules for the other steps should then be slaved to the pacemaker. It’s very helpful for whatever scheduling tool is being used to have multi-level scheduling, sometimes referred to as multi-echelon scheduling capability. In this case, we chose Compression to be the pacemaker and scheduled each press using product wheels and then slaved blending and coating to the press schedules. Figure 26.8 shows a very simplified version of a potato chip production line. When making chips that go into large bags, the fryer is the bottleneck. If filling all small bags, the baggers become the bottleneck. The solution is to COMPRESSION 1

BLENDER 1

COMPRESSION 2

COATER 1

BLENDER 2

COMPRESSION 3

COATER 2

BLENDER 3

COMPRESSION 4

COATER 3

COMPRESSION 5

Figure 26.7  A complex example of a moving bottleneck.

Managing Bottlenecks and Constraints  ◾  233

SEASONER / BAGGER 1 SEASONER / BAGGER 2

POTATO CHIP FRYER

SEASONER / BAGGER 3

SEASONER / BAGGER 4 SEASONER / BAGGER 5 Figure 26.8  Bottlenecks in a potato chip line.

schedule the baggers so that a combination of large and small bags are being filled at any time. This is practical because very little product differentiation occurs in the fryer area; the only major differences in that part of the process are in the cut: flat, wavy, or ripple. The major product distinctions come in the seasoners (sour cream and onion, barbeque, ghost pepper, etc.) and bag size. Thus in this plant, we made bagging the pacemaker and designed product wheel schedules for each of the baggers. The bagger schedules were coordinated so that all were packaging ripple cut or flat or wavy at the same time and that then drove the fryer schedule. The Blue Lakes packet filling lines had been bottlenecks. The combination of group technology (Chapter 25), grouping packets into families by size and assigning each family to a packaging line, and product wheel scheduling created enough additional capacity to eliminate the constraints.

Summary Identifying bottlenecks and potential bottlenecks is important in any operation, but often more so in process manufacturing operations. Many of these lines run around the clock, seven days per week, at or near full capacity, so there is no extra time available to create additional capacity. Where

234  ◾  Prerequisites to Good Scheduling

bottlenecks exist in your process, and extra shifts and capital improvements are not viable options, improved scheduling is the most practical route to increasing throughput. It is a less costly alternative to capital projects, so the idea deserves significant consideration in capacity-limited process operations. Process bottlenecks are just as often due to reliability problems and equipment downtime as they are to inherent capacity limitations, so where that is the case, TPM and OEE improvement programs should be begun or accelerated. If the bottleneck is at the step in the process being scheduled, a structured, repetitive strategy should open it up significantly. If the bottleneck is elsewhere in the process, the new schedule should be designed in a way that maximizes the effective use of the bottleneck. In complex process configurations with little in-process inventory to buffer between steps, a pacemaker should be chosen and the upstream and downstream steps slaved to it. With this level of complexity, a design tool with multi-level scheduling capability is very helpful.

GETTING TO SUCCESS

5

Chapter 27

Leading Scheduling Improvements to Drive Value: Five Steps for Leaders As discussed in Chapter 6, in most process manufacturing plants, there is a strong argument that repetitive scheduling will deliver manufacturing efficiency and stability. Rather than addressing each form of repetitive scheduling methodology, we focus on product wheel scheduling in this chapter. Chapter 28 will outline the project steps to implement a new scheduling process and software system. That chapter targets mainly those who are responsible for the implementation, the project manager and the sponsor of the project, and the core project team. This chapter is a guide for plant leaders to lay solid foundations for the formal scheduling project. It emphasizes the people and process aspects and the changes necessary for project success. Both chapters are necessary and complementary. The relative timing of the leadership and the project processes will depend on plant maturity and circumstances. For less mature plants, the sequencing below may make sense. That way, the plant team will be better prepared for the project so that it will run more smoothly, the risk of failure will be lessened, and the time to value will be reduced (Figure 27.1). However, it is quite feasible to run the two in parallel. Chapter 27: Leading Scheduling Improvements to Drive Value: Five Steps for Leaders IMPROVEMENT GOALS & PLAN

PERFORMANCE CULTURE

IMPROVE SCHEDULING

TAKE STOCK

SUSTAINING THE GAINS INITIAL PREP

STRATEGY DESIGN

SCHEDULE DESIGN

FINAL PREP

SUSTAINING

Chapter 28: A Roadmap to Project Success

Figure 27.1  Leadership and project tasks.

DOI: 10.4324/9781003304067-32

237

238  ◾  Getting to Success

Laying the Foundations for Effective Scheduling Scheduling must aim to support operations by planning production to meet its service, cost, and cash goals in the most efficient way. And when disruptions occur, the scheduling process should enable the plant to respond quickly, effectively, and efficiently, without triggering unintended consequences. Done well, this will be a key enabler for agility and resilience. Resilience and agility are ultimate behaviors shaped by culture. So, behaviors and culture must change to realize the full benefits of good scheduling. There is no guarantee that a good schedule will be followed, particularly if production has had latitude in the past to deviate from the schedule. For example, if a packer is jamming when running a specific SKU, production may switch to a product that runs well rather than fixing the problem. Often that may be sensible: Maintenance may not be available to fine-tune the machine settings. But it’s important to consider the impacts of switching to another product – see the section later in the chapter that discusses dealing with disruption. Discipline and performance management must be in place to ensure alignment of planning and execution. The people in the plant need to understand why this is important, and how it will improve the plant’s performance and reduce the stress they experience daily in the plant. They need to feel comfortable and enthusiastic, so they change their behaviors. Effective management of change is critical for success.

Five Steps to Value for Leaders This chapter targets manufacturing and plant leaders and focuses on change management and skill development. It lays out a five-step progressive learning and improvement path to establish the foundational knowledge and behaviors for the scheduling project (Figure 27.2).

Step 1: Layout the Improvement Goals and Plan 1. Develop a Tangible Vision The plant leader should establish a vision of how she intends to transform plant performance, and what that will take. Refer to the insert for a sample vision.

Leading Scheduling Improvements to Drive Value  ◾  239

IMPROVEMENT GOALS & PLAN 1. Develop a tangible vision 2. Communicate to leaders and other stakeholders 3. Identify supporters and cheerleaders 4. Develop an incremental implementation plan 5. Develop a change plan

PERFORMANCE CULTURE 6. Bring the voice of the customer in the plant 7. Improve shop floor discipline 8. Implement weekly reporting and drive improvement 9. Freeze the frozen horizon 10. Dealing with schedule disruptions

IMPROVE SCHEDULING 11. Implement simple product wheel scheduling as a Team 12. Drive further Improvements 13. Celebrate successes 14. Align Procurement, the Warehouse, and Quality to the wheel rhythm

TAKE STOCK

SUSTAINING THE GAINS

15. Review Progress

22. Ownership is key

16. Lessons Learned

23. Establish sustaining practices early

17. Decide on full plant rollout 18. Select scheduling software with IT support 19. Select an implementation consultant 20. Get budget approval 21. Plan the implementation project

24. Verify sustainment practices are working 25. Formalize training, qualification, and coaching 26. Track key benefits 27. Leverage software improvements 28. Implement a Planning Community of Practice

Figure 27.2  Five steps to value.

Sample Improvement Vision In the future, our lines will run more smoothly, with less firefighting and schedule changes. We will establish a consistent way of running the lines, to maximize efficiency and throughput. We will help our planners and schedulers to put together better schedules based on our collective experience applied to product wheel scheduling methods. We will follow the schedule meticulously, and operations on the shop floor and in the warehouse will improve. The lines will run more smoothly and efficiently because there will be fewer changeovers. We will set safety stocks to deal with demand fluctuations without changing the schedule. And when an external event forces us to change the schedule, we will respond quickly with a well-thought-through plan based on a deeper understanding of the options and consequences of each option. We will work in different ways and improve our decision-making. Our culture and the way we work will become more agile. We will start today to enroll everyone in the journey. This will reduce our stress levels and improve our work-life balance.

The vision should be tangible and meaningful and presented in a form that will resonate with the plant personnel. It should stress the positive impacts on the key stakeholder groups.

2. Communicate to Leaders and Other Stakeholders The shift to the new scheduling process will be a big change. Leadership and communication will determine the ease and speed with which the organization changes.

240  ◾  Getting to Success

Communication should be digestible, and not feel overwhelming. Set the journey in stages of change with corresponding improvements. Initially, focus on the early stages: Once they’re onboard, the participation and learning will make it easier to absorb more of the picture. Describe a realistic picture of the next steps, and how that will lead them to a better place. Emphasize that this is a plant journey. If Lean is part of the plant language, describe product wheel scheduling as lean scheduling, i.e., a comprehensive form of Heijunka, which will be implemented as part of the overall lean initiative. When communicating with the leadership team, ask each of them to develop a plan for the changes needed within their functions.

3. Identify Supporters and Cheerleaders As the initiative launches, identify people who “get it,” and are supportive. Ideally, they will be influencers in the plant. During the early stages, you could pilot the initiative in an area whose leader is enthusiastic, and where resistance to change is less. It’s important that the scheduler is supportive and has the ability to understand and adopt the new scheduling principles immediately. Product wheel scheduling is a more structured version of common sense, of what every scheduler tries to do. Build on that and give the scheduler the chance to shine during the wheel design and implementation process. Encourage the network of supporters and cheerleaders to evangelize the new approach. Depending on the size of the plant, consider formalizing a Centre of Excellence (CoE) with the key disciplines and influencers, to become a resource for the rollout of the product wheel scheduling process. After going live, the COE should retrain new schedulers and shift leaders and continually drive adherence and improvements.

4. Develop an Incremental Implementation Plan People learn by doing. It’s easier to make small changes than big ones. The small improvements that result from meeting an incremental set of goals feel good and become fuel to do better. Change is less intimidating, and life for the operator will slowly improve as production starts to stabilize, and firefighting lessens. “Eat the elephant one bite at a time” as the saying goes. The plan should be broken into stages of change and improvement: up to two to three months each. Each stage should have clear objectives and targets. When implementing, ensure the plan focuses the teams on maintaining previous performance targets, and on achieving the next set of targets. Decide where to start. Ideally, this would be a relatively standalone set of lines that make one or more product families that are not made on other lines. The area supervisor should be a competent leader who has shown he can drive change. The scheduler is a critical resource, since he is the gearbox between planning and production, and he must do a good job of wheel

Leading Scheduling Improvements to Drive Value  ◾  241

scheduling. Lastly, there should be considerable opportunity for improvement, so that results are rapid and convincing.

5. Develop a Change Plan Today many manufacturing companies have adopted a change methodology. If this is true for your company, use the resources and standard process to develop a change plan. If you don’t have a methodology, the Prosci Change methodology is widely used and recommended. It is a practical, common-sense approach to change management. It is also customizable, so it can be scaled to any situation. Like most change methodologies, it includes stakeholder analysis and the development of a Communication Plan based on a component of “what’s in it for me?” for each stakeholder group. A Training Plan is also needed.

Step 2: Work on the Culture Many plants have been on a lean journey for some time. Some of what is outlined below may already have been done. Adjust the suggestions below to fit into your existing framework.

6. Bring the Voice of the Customer into the Plant The sales organization or customer service in some companies is acutely aware of service problems. They hear the customer every day – particularly when the customer is not satisfied. Bringing these perspectives and specific issues can galvanize change. Breaking down the silos leads to awareness and understanding of the importance of sticking to the plan, and maintaining quality at all times. It also gives more meaning to the work done in the plant. In a multi-tiered supply chain, the downstream plant is the upstream plant’s immediate customer. Frequent communication is vital, so both plants can better understand the impacts of their actions on the other. In more mature supply chains, there will likely be a collaborative process across all the plants involved in making a product family. This could be focused, e.g., on a triage process where all participants help determine how to maintain service levels if a disruption occurs. Indeed, since COVID, most product supply chains have been forced to work this way. COVID has forced more communication up and down the supply chain, and service has been raised in priority. Unfortunately, the KPIs for the supply chain relate to On Time in Full (OTIF) service performance, yet the plants are often rewarded on efficiency and cost. These misalignments are becoming more apparent, and we’ve seen some companies starting to align their KPIs better.

242  ◾  Getting to Success

7. Improve Shop Floor Discipline As the plant leadership team becomes more aware of the need to put customer service ahead of most other objectives, the plant leader needs to lay out a simple plan to improve OTIF, including improving discipline to meet the schedule adherence goals and hit inventory compliance targets.

8. Implement Weekly Reporting and Drive Improvement To prepare the ground for improved scheduling, the plant must establish the discipline to execute schedules and track reasons for non-compliance consistently. Later, an inventory compliance process will be implemented. 1. Improve Schedule Adherence: If schedule adherence is not currently measured, get a simple, practical reporting process into place. Initially set a stretch target of, say, doubling the current schedule adherence figure: Typically, schedule adherence is as low as 30%–40%. So set a target of 60% schedule adherence to be achieved within 45–60 days. Make sure the goal is cascaded to the line supervisors and the operators, and hold them accountable to hit the goal. Track the reasons for deviating from the schedule. It is essential to be supportive: Often, the reasons for changing the schedule are caused outside production. The raw materials or labels are not available in the needed quantity. Sales call with an urgent demand. At this point, supervisors and leaders must coach and remove causes – not blame. Later, capture two-tier reasons for all schedule misses, such as “raw material shortage” and “Quality rejection on received raw material batch.” Report schedule adherence with deviation reasons at the line level at the end of each shift and day. Integrate this into the existing plant performance management system. Roll up the line-level reports to the daily/weekly area manager meetings and up to the plant manager level. Accountability to meet the targets exists at each level, and collaboration to meet the targets is essential at each level and across the levels. The continuous improvement team could lead the problem-solving process to eliminate the most significant causes, particularly triaging across the levels to implement systemic changes that eliminate all but essential schedule changes. If not already in place, the supervisors and schedulers should be trained on the root-cause analysis and elimination process so that the continuous improvement mindset becomes embedded in planning. 2. Using a similar approach, track end-of-week total inventory, and later start to track at the individual SKU level, initially using your existing

Leading Scheduling Improvements to Drive Value  ◾  243

inventory policies: Later, when we design product wheels, the safety and cycle stocks will be carefully calculated. 3. Similarly, improve Fill Rate Compliance or OTIF for the plant. 4. Leadership should pay keen attention and ensure the necessary support is provided to reach the targets.

9. Freeze the Frozen Horizon! The problem: A colleague claims that decisions made outside the plant cause 75% of waste in the plant. The same might be said for production: Procurement or warehouse lapses, for example, result in wasted time and effort inside production. Tracking the reasons for schedule changes will pinpoint some of the shop floor’s challenges. Consider this example. Many smaller companies are growing because of their commitment to customer service. Sales will routinely call the plant and demand that shipments be prioritized to meet a rush order. The resulting schedule change typically results in unintended consequences: shorting other customers, for example, or asking suppliers to jump through hoops. So further schedule changes become necessary. Before long, the schedule is only valid for a very short time. Firefighting becomes the order of the day – the result: stressed-out production teams, low OEE, and poor customer service. The cost of goods is significantly higher than needed, and plant capacity is wasted with unnecessary changeovers. Most frozen horizons aren’t truly frozen. The schedule issued at the start of the frozen period could be substantially different from the actual production record at the end. The solution: First, ensure that the length of your frozen period is justified, for example, by the lead time of critical components that can’t be buffered, or by the decisions that need to be made to support production. For example, at Blue Lakes, frozen ingredients take seven days to thaw and must be used as soon as they have thawed. Once the frozen period is decided, it must be enforced. The lines must be insulated from external factors as much as possible within the frozen period. This is discussed in the next section.

10. Dealing with Schedule Disruption There will always be valid reasons to change the schedule. An opportunity with a new strategic customer could rightfully take priority. A successful launch that exceeds the forecast could cause an immediate supply challenge. Schedule changes are inevitable, and a standard practice is needed to ensure that the number of changes is minimized and that the implications of essential changes are understood. Response options should be compared

244  ◾  Getting to Success

and conscious, deliberate decisions made, so the response does not result in unintended consequences down the line. At the same time, speed is of the essence. Good scheduling includes a process for rapid but structured decisions when a schedule change becomes unavoidable. It is common to make time in the schedule available for unexpected prioritization of orders without disrupting the overall weekly schedule. This is “breathing room” in product wheel terminology. It isn’t fair to hold line supervisors accountable for hitting schedule adherence targets when sales routinely ride rampant over production plans. So, establish policies in advance that inform and speed decision-making. 1. Establish rules that determine how proposed schedule changes are triaged and decided. The scheduler and the supervisor can determine some. Others may require the plant manager’s approval. Specific rules should guide what new orders are accepted inside the frozen zone and who must authorize them. 2. For example, when sales call, wanting their customer’s order to be prioritized, refer them to the plant manager so that production is shielded from these external pressures. The plant manager can weigh the options with help from the team to make the best business decision. 3. Longer-term planning is also a frequent culprit, for example, misalignment of planning and scheduling parameters or an incomplete solution to a planning problem that is passed to the plant scheduler to resolve. The plant manager needs to help them understand why it is essential for them to improve planning. 4. Capture all instances where the frozen window is violated. Analyze them to eliminate the causes wherever possible. COVID has forced other, more effective approaches to deal with disruption. During COVID, people were often forced to work from home. Emails and conference calls are clunky and ineffective. To foster speed and agility, some companies have successfully implemented new ways to collaborate supported by software that instantly connects the people who need to triage disruptions of all types, not just schedule-related ones. An external event like a raw material delivery delay is immediately communicated to those with more context and the impacts of one response option vs. another. The ensuing interactions – typically using messaging – quickly inform all involved. In many cases, the best response quickly becomes apparent, including the detailed actions to get it done right away. The transparency and immediacy of communication foster agility and resilience: Indeed, at scale, it can be transformative. This is similar to the lean process where the trades gather with the operator at a failed machine to discuss the symptoms, diagnose the root cause, and quickly get it up and

Leading Scheduling Improvements to Drive Value  ◾  245

running again. Electronic collaboration tools allow people in different locations to meet and work effectively in cyberspace to solve problems.

Step 3: Improve Scheduling It is sometimes beneficial to manually implement product wheel scheduling before doing a more rigorous design and implementation project supported by specialized software. In more mature plants, it would likely make sense to implement the project as described in Chapter 28 immediately. For completeness, we’ll describe manual scheduling below.

11. Implement Simple Product Wheel Scheduling as a Team Now we’re ready to start improving scheduling. Initially, you can adopt a rapid, keep-it-simple approach involving team members, so they understand what is being proposed and become invested in the choices through their involvement and understanding. Although any repetitive scheduling method could be used, it is better to decide which approach you favor so that it becomes integral to the future state plant vision. As mentioned, we assume the ultimate objective is to establish dynamic product wheels to simplify the text. However, this approach will work for any of the repetitive methods. In working sessions with schedulers and supervisors, design simple product wheels based on, for example, high-volume “A” products being produced weekly, medium-volume products every two weeks, and low-volume products every four weeks. MTO products would be assigned a frequency based on their lead times and spread across the cycles. Assign the products to specific default lines based on operators’ knowledge of which products run best on each line and try to balance the schedule across lines and cycles. Involve maintenance and quality as well. Use tribal knowledge to establish the best sequence on each line. Don’t sweat it too much: This process will inevitably result in increased throughput and more schedule stability. The project (Chapter 28) will use a more rigorous but similar thought process. Some spare time in the wheel schedule (our friend “breathing room”) should be reserved to accommodate unscheduled needs should they arise so the normal schedule is not affected. Modify the planning parameters in the ERP or planning system, to match the wheel design, including run length (batch size) and frequency. Ensure planning master data, including run rates, match plant reality. If the supply chain organization calculates inventory settings, ensure they and the wheel design are aligned. If it’s up to the plant, the wheel design process mentioned above will calculate suitable targets, which should be configured in the planning system.

246  ◾  Getting to Success

12. Drive Further Improvements The resulting schedule will be better since it is based on the team’s collective knowledge. There will be fewer changeovers. There will be fewer unplanned changes to the schedule. The increased throughput will provide more breathing room to accommodate essential schedule changes. The tendency for firefighting will reduce. Schedule compliance will improve, as will plant OTIF. Sales will see fewer missed orders. Planning will realize they need to respect the frozen window and plan accordingly. Some examples are related in Chapter 29. Raise all the metrics. Continue to instill the discipline to not deviate from the schedule. If not already in place, add the second tier reason codes for exception reporting for deeper analysis and cause elimination.

13. Celebrate Successes OTIF and customer service will likely improve. Plant throughput will rise. Celebrate the successes and recognize the key contributors. The work–life balance for the shop floor teams should improve noticeably: Unplanned Friday night overtime could become the exception rather than the norm. The plant manager will see her results improve. The head of the supply chain will see the improvements in your plant and encourage others to follow suit.

14. Align the Plant to the Wheel Rhythm The repeating sequence each week will result in a pattern of raw materials flowing into the plant, through the stages of production to the warehouse, to distribution from the plant. This represents an opportunity for procurement to align the contracts to take advantage of the improved predictability for supply by the suppliers. Longer term, as the plant stabilizes, the supply contract could take advantage of this to secure better pricing or demand better delivery performance since the supplier should face fewer last-minute order changes. Because of an expected reduction in the number of changeovers, the warehouse will have fewer SKU changes to service – stocking the line and moving the produced goods back to the warehouse. There will be fewer situations when warehouse personnel must accommodate last-minute schedule changes. The more predictable sequences on each line could impact storage locations to improve picking efficiency. Quality could take advantage of similar opportunities, having more predictability on when beginning-of-run and end-of-run testing will be required. In some instances, their patterns of material flow through the plant could result in logistics patterns into and out of the plant.

Leading Scheduling Improvements to Drive Value  ◾  247

The plant manager should encourage his team to exploit the opportunities that the wheel cadence presents.

Step 4: Take Stock 15. Review Progress Examining the journey and resulting improvements will likely reveal more opportunities.

16. Lessons Learned Capture the learnings: These will be useful at the next stage when/if you decide to do the full scheduling project.

17. Decide on the Full Plant Rollout It is likely that the performance improvements that have been achieved at this stage, while being impressive, are a fraction of what is possible in all but simple plants. The more complex the product line-up and the changeover characteristics between products, the higher the number and complexity of stages of production – all these factors weigh into the potential for further improvement. In our experience, a full implementation as described in Chapter 28 will yield at least twice the benefits achieved so far.

18. Select Scheduling Software Select the scheduling software that is suitable for the repetitive scheduling method you selected. Sometimes, your existing software may suffice, be it Excel, your ERP system, or your planning system. If you decide to look further afield, the investment in specialized scheduling software will typically offer an excellent ROI. In any event, some investigation is needed. Once that is complete, if you decide to look at other software, initiate a software selection process. Involve IT since they will have to support its implementation and support. This can be a time-consuming process so get an early start.

19. Select an Implementation Consultant You may have developed the capabilities to implement the full project without external support. Or you may feel that external expertise would result in greater benefit. Or you may not have the spare resources to scale to a full rollout. In any event, it’s advisable to decide early and select a suitable partner if that is how you want to go.

248  ◾  Getting to Success

20. Get Budget Approval Set a timeline for the full project, and have your team develop the budget and resourcing needs including IT support. This step may well require that the organization as a whole has to make a choice if standardization of processes and software is your policy across the plants. In this case, you have amassed solid evidence that the central manufacturing or supply chain organization can use to support a broader rollout to all plants.

21. Plan the Full Implementation Once the budget is approved, do more detailed planning. Doing this well in advance, incorporating the learnings your plant has accumulated, will pay dividends later – and ensure that the path you have forged is built on to realize your original vision. Much of this book is about the next steps – implementing mathematically optimized methods of repetitive production supported by software solutions. The journey described above will have established the conditions for success for a more advanced project. Teams will be more disciplined. Firefighting will have abated. Belief in the plant leader’s vision and ability to make it happen will be cemented. It’s time to move to the next stage!

Step 5: Sustaining the Gains It’s a fact that many improvement programs deliver improvements that erode over time. We’ve seen this be the case with scheduling projects. It is critical to sustain and further improve the practices that have resulted in the improvement after going live.

22. Ownership Is Key The key to sustaining benefits is ownership: ownership of the outcomes and the practices that lead to success. Good leaders demonstrate this firsthand, and they instill it in the people working with them. That said, there is no issue with seeking outside help. But the help must be just that: “Give us advice and help us figure out how to do this thing.” But handing the keys to the consultant is inevitably a recipe for failure. It’s a delicate balance. A good consultant has one goal: to build the client’s capabilities to continue without the consultant and to achieve and sustain the total potential value.

23. Establish Sustainable Practices Early Many initiatives result in an improvement bump that is not sustained. The focus that existed during the initiative fades, and results wane.

Leading Scheduling Improvements to Drive Value  ◾  249

The new scheduling practices must be built back into the standard work, into the plant operating model. For example, dealing with schedule (and other) disrupting events must be entrenched into daily practice. It should be how work is done, day in and day out. Likewise, the tracking of schedule adherence (as well as inventory compliance and OTIF) with exception reason codes must be built into the standard meeting and performance management process. It’s far easier to build in the sustainment practices during the project than to retrofit them later when bad habits have been formed. That way, sustainable practices are used to execute the project. Consider a recent example. A large company had implemented a global, end-to-end planning solution. A team was put together to improve master data quality. A data governance process was designed but not implemented, so data integrity was degraded. When it came time to go live, the master data quality did not support the planning process. Result: The data exercise had to be redone, this time with the data governance process in place. Data ownership was established upfront, and the data owners cleaned the data. With some nurturing, data management will become standard work.

24. Verify That Sustainment Practices Are Working Verify process adherence: If the shop floor starts to take liberties and change the schedule, this must be identified early and corrected. So tracking is needed of schedule adherence, inventory compliance, and adherence to other processes – and verification that reason codes are being used to drive improvements. Similarly, data governance process adherence should be monitored. Analyze master data periodically to confirm it is in good shape: This could sound like a waste of time to some. Why should scheduling master data degrade over time? Well, the continuous improvement team efforts might reduce changeover times and increase line rates. SKUs are constantly added and removed from the portfolio. New personnel may not fully understand how to configure the BOM for a new product. There are many reasons why master data will naturally degrade over time. So, the process of actually assessing master data integrity will highlight that practices are not being followed, such as new product introduction master data management.

25. Formalize Training, Qualification, and Coaching People may come and go. Performance will suffer if new people aren’t adequately trained, coached, and monitored over time to ensure they get up the learning curve. Implement a formal training and qualification system. The training should focus on tasks the schedulers will need to accomplish and include exercises

250  ◾  Getting to Success

and what-if cases. Qualification, based on demonstrating critical tasks, will give the trainee confidence that they have learned the skills necessary to succeed or point out areas they need to revisit. The same is true for other roles, like data stewards.

26. Track the Key Benefits If OTIF or service levels are gradually declining, take a thorough look at why. This may be a good use of external experts who can look with a fresh eye to identify what might be causing the decline and advise what to do about it.

27. Take Advantage of Vendor Software Improvements Maintain a good relationship with the scheduling software product vendor. Listen to the new features they have introduced and adopt them if they will drive more improvement or reduce cost.

28. Implement a Planning Community of Practice (COP) In large plants, a COP can play a key role in improving planners’ skills. It can share learnings across plants to improve performance overall. A COP can work with the software vendor to implement software improvements to better schedule their manufacturing processes, for example. It can be a source of expertise across the network. We suggest that many of these sustaining practices should be implemented before going live, not afterward. That way, they can reduce implementation costs, drive value realization faster, and sustain and improve resulting benefits.

Chapter 28

Where to Begin: A Roadmap to Project Success If your current scheduling strategies, processes, and software don’t match what we’ve recommended, and you recognize that there would be benefits in transitioning to something better, there is a well-traveled roadmap you can follow. You can start at the beginning and follow the whole map, or you can start at some later point depending on how mature and robust your manufacturing and transactional processes are and your readiness for change. The steps laid out below provide a very logical flow and each step lays the best foundation for each subsequent step. However, there are many cases where improved scheduling is urgently needed or there is a significant benefit in more effective scheduling that it makes sense to skip some steps and move ahead as rapidly as possible. Steps 8–10 are examples; they should certainly be done but can wait until the new scheduling system is operational. The key steps are: 1) Prepare for the transformation. Make sure that leadership is prepared to fill their role. Select and train the team. Map the current product and information flow. Make appropriate improvements to manufacturing flow. 2) Select the production strategy. Decide what adjustments should be made to the planning and scheduling systems. Select and configure any new software. Gather all needed data. Decide which steps should be scheduled. 3) Design the new scheduling strategy, i.e., the new repetitive process. Ensure that inventories and changeover frequencies meet business targets. 4) Make the final preparations for Go Live. Make sure that inventories are in place to carry you through the first cycles, and that everyone has

DOI: 10.4324/9781003304067-33

251

252  ◾  Getting to Success

acquired the necessary capability. Test the new procedures and software. Go Live. 5) Ensure that all systems and processes are in place for all benefits to be sustained, and expanded to other lines. It’s important to note that the leadership activities described in Chapter 27 on dealing with change (Figure 27.1) are complementary and overlap in time with the activities described here. This chapter focuses on the specific project tasks (Figure 28.1), while the previous chapter focused on the steps that leaders need to do to ensure these will be successful.

Initial Preparation 1) Make sure that the leadership is fully engaged. Ensure that leadership at the highest levels see this as a requirement for maximum business success, that they understand the need for this transformation and how valuable it will be to the bottom line, understand their responsibilities, are exhibiting the right behaviors, and are committed to assigning the needed people and funding. 2) Establish a project steering team to guide and coach the team doing the design work, ensure that all needed resources are available and engaged, and remove barriers. Membership should include the operations director, the plant manager, and the planning manager. The team should meet regularly during the preparation, design, scheduling, and final preparation stages. The team should be kept in place during the sustaining phase unless its functions align well with the normal leadership processes.

INITIAL PREP 1. Engage Leadership 2. Establish Steering Team 3. Form the Project Team

STRATEGY DESIGN

SCHEDULE DESIGN

11. Select a Production Strategy

21. Determine Scope of the First Phase

12. Design the Planning Process Flow

22. Define Product Families

13. Write a Demo Script

23. Determine Production Frequencies

4. Conduct a Readiness Assessment

14. Select Scheduling Software

5. Train the Team

15. Train the Schedulers

6. Develop a VSM

16. Collect Data

7. Map the Planning Process Flow

17. Decide which Steps to Schedule

8. Analyze Bottlenecks

18. Consider Multi-Level

9. Reduce Changeover Times

19. Consider Cellular Flow

10. Evaluate Reliability

20. Install the Software

FINAL PREP 29. Confirm Training 30. Select Go-Live Date 31. Conduct a Parallel Run 32. Calculate Starting Inventory

24. Calculate Required Inventory

33. Create Contingency Plans

25. Design Production Cycles

34. Develop Auditing Plans

26. Coordinate Raw Material Planning

35. Make the Go No Go Decision

27. Review with Stakeholders

36. Go Live

28. Configure the Scheduling and ERP Software

Figure 28.1  Implementation roadmap.

SUSTAINING 37. Continue Steering Team function 38. Follow the Auditing Process 39. Evaluate Contingency Plans 40. Celebrate Success 41. Roll the Transformation Out

Where to Begin  ◾  253

3) Form the project team. Decide who will be involved in the transition. Make sure they understand their roles and responsibilities and have the appropriate background and training to fulfill them. This includes understanding the details of the strategy options and the implications for various parts of the organization. A key requirement is to help them understand how this will make their jobs easier and more effective. Production stability and reduction in unplanned changes will have a big impact on their quality of life; people who can follow a routine feel less stress and have higher job satisfaction. Get the team organized and aligned. Determine the level of participation needed from each participant, the degree of training required, and set a preliminary schedule. 4) Conduct a readiness assessment and create a plan to fill the gaps. Readiness criteria are covered in more detail in Chapter 19. It’s never too early in a project to begin to assess readiness and correct the issues you find. Moving forward with a project that is not ready usually results in a failed or delayed project, one that takes more time than if the issues had been addressed upfront. 5) Conduct any training needed. Schedule the training, get it on people’s calendars, and decide who will do the training and what material to use. 6) Develop a Value Stream Map of the entire operation. Identify the process steps that are currently scheduled and those that simply follow the schedules of other steps. Note the key changeover times of each step. Identify all WIP locations and typical contents in volume and days. List the utilization (demand/capacity) of each step and identify bottlenecks and near-bottlenecks. 7) Map the flow of planning data. On the Information Flow portion of the VSM, show all key steps and systems currently involved in developing the daily or weekly production schedules, including Customer Order Management, Forecasting, Demand Management, Capacity Management, Planning, and Scheduling. The VSM in Figure 5.1 is a good example. It’s important at this point to think about the data necessary (step 16) for scheduling and where it can be sourced in the systems. 8) Analyze the VSM for bottlenecks. Determine how to open up or manage any bottlenecks identified from the VSM. 9) Use SMED (Chapter 23) to aggressively pursue the reduction of time and material loss of all changeovers, especially those that influence scheduling in any way. If your facility uses Kaizen Events to design and execute improvements, this would be an ideal candidate. 10) Evaluate equipment reliability. Analyze OEE metrics to determine how reliable the equipment is and if it can be relied on to execute a defined schedule. Implement TPM (Chapter 24) to the fullest practical extent. If not already in place, begin any operator training required to move toward autonomous maintenance. Address the cultural challenges

254  ◾  Getting to Success

of having mechanics let go of some of their responsibilities and having operators embrace responsibility for equipment performance.

Scheduling System Design 11) Select a production strategy. Determine which aspects of product wheels/rhythm wheels and RfS will be components of the new strategy. Chapter 6 described the options and their advantages and disadvantages in more detail. 12) Design the new planning process flow and assess the gaps between the new process and the current process. 13) Write a software demonstration script. Identify scheduling scenarios from your business and develop requirements based on the scenarios. Classify the requirements by site, production line, products, value creation, and criticality. A demonstration of scheduling software is critical to assess its usability. 14) Demonstrate and select the scheduling software. Select the software based on your assessment of its usability in satisfying the highvalue and critical requirements. New scheduling software typically faces an uphill battle for financial justification versus what are perceived to be low-cost solutions like the corporate ERP system or a Microsoft Excel workbook. This selection should be made as soon as practical, to allow time for IT and other stakeholder groups to provide their input and participate in the selection. If you are considering keeping your current software, it’s important to put it through the same scenario, requirement, and usability assessment. We often called this an integrated walk-through, mocking up and following the new process in the software to assess its usability. 15) Train the schedulers. If the software is a change from current practice, make sure that all planners and schedulers are trained not only in how to use it but also in the underlying strategy and logic so that they don’t create unintended consequences. Line up technical resources to provide guidance and additional training if difficulties arise. 16) Collect all needed data. In some cases, it is very straightforward to gather the data needed to inform all decisions and design tasks. In others, it can be quite time-consuming; the required data may be scattered across many databases with no single person having access to all of it. The data includes: – All product characteristics including allergen content and organic requirements – Packaging characteristics including container size and shape and case and pallet counts – Run rates

Where to Begin  ◾  255

– Forecasts, forecast error, bias, and variability – Is the product typically MTS or MTO? – Other factors like Quality Assurance hold times, shelf life, fill rate target, etc. 17) Decide which steps should be scheduled. Analyze the VSM to understand which steps would be good candidates for a product wheeltype scheduling strategy. 18) Decide if multi-level scheduling is needed. Decide if there are process steps in series whose schedules must be coordinated. Does one production asset feed several downstream operations, such as packaging? Is there sufficient inventory between steps that they can be scheduled independently? 19) Decide if cellular flow is needed. Decide if the equipment footprint suggests cellular manufacturing (see Chapter 25), and if so, create the virtual flow paths and allocate product families to the cells. This will simplify changeovers as well as any upstream–downstream coordination needed. 20) Install the software. Install the new software, connect it to other systems that will supply data and receive the schedule, per the information flow developed in step 7 and the process flow in step 12. Test the interfaces. This is sometimes called the technical cutover.

Strategy Design These steps should be done by a cross-functional team, with planners, schedulers, and operations people, with some input from maintenance, procurement, and quality assurance. 21) Determine the scope of the first phase. Some plant managers choose to begin with a single line or a small group of lines to prove the concept, demonstrate the benefit, work out all practical details, and work through any initial challenges before moving through the entire plant. If a single line or a small group is chosen, they should be lines that completely encompass all product families involved. In other cases, it makes more sense to go with a major part of or the full plant. 22) Define/Assign product families. Divide the set of products included in this phase into groups with very similar characteristics, to minimize changeover complexity on any line. Assign each group to a line. Using the weekly hours for each product, make sure that no line is overloaded. That may require some compromising, like breaking a large family into pieces if it is too large to fit on a single line. 23) Determine the optimum frequency at which each product should be made. Economic Production Quantity (EPQ) calculations, which balance

256  ◾  Getting to Success

the cost of changeovers with the cost of carrying inventory can provide some guidance here. There are some advanced algorithms (Silver, Jackson) that can provide more accurate answers than the traditional EPQ formula. Product shelf life and minimum batch sizes must also be taken into account. The ideal frequencies recommended will not likely give the same numbers for all products, so some compromises must be made. You must select a single frequency for all the high-volume products, which closely matches each one’s ideal. As most plants prefer to do things in weekly increments, it usually works best to select an even multiple of weeks for the primary frequency. The lower-volume products should be made on some multiple of the basic frequency. For example, if the highest volume products should be made every two weeks, then lower-volume products should be made every four weeks, eight weeks, and for very low volume products, every 16 weeks. 24) Calculate the total inventory required to support the wheels as designed, based on average forecast and selected frequency. Include safety stock, which can be calculated from each product’s frequency and forecast error; include a factor for any manufacturing variability. Include an allowance for any time that product must be held awaiting quality test results. Include an allowance for situations where more product than needed must be produced to fill minimum lot size requirements. The more capable wheel design software products can do these calculations. 25) Assign products to cycles, balance loading, and sequence cycles. Select the cycles on which the lower-volume products will be made. Try to group these products to minimize difficult changeovers on as many cycles as possible. For example, if a few products have a very problematic allergen, try to group them on the same cycle to minimize the number of cycles on which that allergen clean must be done. Make sure that no cycle on any line is overloaded and make any adjustments necessary. Determine the optimum sequence for each cycle, one that minimizes the total changeover time or difficulty. 26) Coordinate raw material planning. Work with Procurement to align raw material processes with the new scheduling patterns. 27) Review the final design with all stakeholders. The person responsible for inventory must agree with the calculated inventory requirements. Operations must agree with the production frequencies and sequences. Operations and perhaps maintenance must confirm that the changeover timings are reasonable. The more these people are involved as the design progresses, the smoother the review discussion will go. If the design is done as a five-day Kaizen Event or Rapid Improvement Team activity (this is often the case), reviews should be done at the end of each day for any stakeholders not in the teamwork activities. 28) Configure the scheduling and ERP software. Since the planners and schedulers have been trained, data has been collected, and the

Where to Begin  ◾  257

scheduling process and strategy have been designed, you are now ready to encode the strategy into the software to be used for the scheduling process. If an ERP system will be used to create the orders to be scheduled, it should be updated with the inventory targets, frequencies, and lot sizes from the design.

Final Preparation 29) Confirm all training. Make sure that all who participate in the weekly scheduling process are fully trained in the selected scheduling concept and the use of all software involved. If a new software package is being used to better support and simplify execution of the new scheduling patterns, all planners and schedulers should be oriented and trained in its use. 30) Select a Go Live date for the selected set of lines. 31) Conduct a parallel run or acceptance test. This will validate the scheduler’s training and give them confidence that the new process and software can create a good schedule. Scheduling is mission critical and you should not cutover unless you are confident in the new process and the people who will run it. 32) Calculate starting inventory requirements. Determine the inventory each product will need at the start, keeping in mind that some products will not be made until late in the first set of cycles. Prepare an inventory build plan, and ensure that it’s practical to build that quantity before Go Live. If not, you can still begin on the selected date, but recognize that the first few cycles will not be able to fully execute the designed schedule, and that it will take several cycles for all inventories to ramp up or down to the required levels. 33) Create contingency plans. Things will not always go according to plan. There will be occasions when it becomes necessary to deviate from the planned schedule. Rather than facing those disruptions one at a time as they arise, it is better to plan for them and decide in advance the best course of action for each type of disruption, whether it be raw material shortages, labor shortages, machine failures, unexpected demand surges, or anything else you can anticipate. Discussing and agreeing on the appropriate steps to take in these situations avoids after-the-fact second-guessing and finger-pointing. Having a Contingency Plan for each expected disruption, with full alignment of all involved, minimizes the consequences of the problem and facilitates getting back on the schedule sooner. 34) Develop an auditing plan. As the scheduling strategy is being used on a daily or weekly process, there are three categories of information that should be analyzed.

258  ◾  Getting to Success

– Are you actually following the designed patterns? When circumstances require a deviation from the accepted sequence, there should be a convenient way to record the reason for the deviation along with other useful information: line, product, time, and duration. A Pareto analysis should be done periodically so that chronic “wheel breaks” can be addressed. – Are you getting the expected results, in throughput increases, time spent in changeovers, and inventory required? If not, and wheel breaks are not a significant contributor, the design should be re-examined. – Is it time to refresh the design? After a few weeks or a few months, it is usually beneficial to re-evaluate the design in light of current conditions. Demand for some products may have dropped off, while demand for others has grown and new products may have been added. Continuous improvement projects may have increased run rates or reduced changeover times. Shelf life constraints may have gotten tighter or been relaxed. The quality testing lead time may have been reduced. A product wheel design has a degree of tolerance and will continue to perform well in spite of some degree of change, but there are usually performance enhancements to be gained by taking advantage of any improvements. 35) Make the go/no-go decision. Before going live evaluate everything, including an update of the readiness criteria, to be sure that you are confident that everything is in place and that the new process and software will create a good schedule that can be executed on the production lines. 36) Go Live. Although there can be a lot of anticipation and excitement, if all of the preparation has been well done, this is usually uneventful.

Sustaining Some of the things necessary to sustain and expand the benefits of the new scheduling strategy, scheduling process, and software have already been touched on. But there are additional things that should be done. 37) Continue the steering team function. Some form of the steering team should be kept in place, to ensure that all defined processes are being followed, that challenges are being appropriately met, and that priority continues to be given to the appropriate KPIs. In the best of cases, this migrates to the normal leadership managing processes, with representation from planning and scheduling. 38) Follow the auditing process. If the patterns are frequently broken, look for and correct any chronic problems. If the expected benefits are

Where to Begin  ◾  259

not being seen, analyze why not. Refresh the wheel patterns when conditions have changed noticeably. 39) Evaluate contingency plans. When disruption occurs, are the contingency plans being followed? Are they working? The initial plans are generally a good starting point, but your experience can often suggest improvements to make them more effective. 40) Celebrate success. Recognize and reward those who are doing their part to make the new scheduling process successful. 41) Roll the transformation out. If a single line or small group of lines was chosen for the first phase, develop a plan to roll the process out across the rest of the plant. Once the entire plant has been transformed, look to other plants in the enterprise that could benefit from the improvements made here. Even if tangible benefits are not obvious in the other plants, there is a very strong benefit in getting all plants on the same process and systems.

Summary When embarking on a process redesign or complete transformation like this, it helps to have a roadmap to follow, a guide to explain all key steps along the way, to ensure that nothing gets overlooked. The steps presented here are based on many years of practical experience which gave us an education in what works well and what doesn’t. It’s all about preparation, preparation, and more preparation; doing the initial preparation before any design or alternative selections are done saves time and minimizes the need to loop back and revisit decisions. The sustaining tasks may be the most important; no matter how successful the earlier steps have been, without the appropriate attention and “care and feeding” it will all be for naught.

Chapter 29

Critical Success Factors Scheduling Strategy Critical Success Factors For any of these scheduling strategies, whether product wheels, rhythm wheels, RfS, or FSVV, to be sustained, there are three crucial business processes and capabilities that must be in place: 1. Finished product and raw material inventories must be properly calculated and built to allow the business to reliably complete the planned sequence, considering the normal variability of demand, supplier delivery performance, and production reliability. For example, unless the Product Wheel sequence can be reliably completed most of the time, it’s not really a Product Wheel. 2. Despite the best efforts to properly design inventories, there will be unplanned exceptions. The business must learn to react in a systemic way, evaluating whether a deviation from the designed pattern is the right thing to do. When a deviation is approved, the emergency or highpriority orders must be placed where they will have the least impact and then return to the wheel structure to complete the remainder of the sequence. Without deliberate reaction, emergencies can cascade, making it difficult to get the schedule back on track. Contingency Planning, a process to anticipate the most possible schedule disruptors, determine the appropriate response, and get consensus from all stakeholders, is a critical task. 3. There must be a process to assess when business conditions have changed, and a commitment to redesign the repetitive patterns when needed. Often, everyone knows that conditions have changed, but the effort for redesign seems too high, and the business decides by default to muddle through, resulting in lost efficiency and unmet business objectives.

DOI: 10.4324/9781003304067-34

261

262  ◾  Getting to Success

Scheduling System Critical Success Factors When evaluating a scheduling system, look closely at its ability to execute these activities: 1. As families of products are being assigned to production lines, there must be visibility to the available capacity and current loading of each line so that no line becomes overutilized. 2. The system should provide a straightforward, convenient way to place new orders into their best position in the sequence while meeting their due date. 3. An effective strategy will cause high-volume or high-value products to be produced more frequently than low-volume or low-value products; therefore, the system must provide a way to visualize and track multi-cycle wheels, where low-volume products are not made on every cycle. 4. All product attributes influencing changeover cost or difficulty must be visible during the design of the structured, repetitive patterns, during the normal scheduling process, and when changes must be made to respond to some interruption. In our experience, many well-known scheduling systems fail to provide adequate visibility of a product’s characteristics and their implications for changeovers and sequencing. Batch chemical processes typically have three or more relevant attributes, food packaging lines can have eight or more, and nutraceutical lines can have more than a dozen. Lacking proper visibility to attributes, planners must memorize them; this makes backfill difficult, and the sequence is often sub-optimized. 5. Changeover costs and time, based on product attributes and production sequence, should be visible and recalculated whenever the schedule is changed. 6. Inventory calculations should consider customer service goals, and they should be based on fill rate targets rather than cycle service levels if that’s what the business drives. Lot sizing, planning lead time, goods receipt time, quality hold requirements, and demand and production reliability should also be taken into account. 7. Consistent master data should be used for design, scheduling, and execution. There should be a single source of truth for any required data, and the path to obtain it should be clearly understood and documented. 8. A straightforward way to transfer or activate a completed schedule design to production scheduling must be provided.

Cultural and Behavioral Critical Success Factors 1. Leadership must own the transition to the new scheduling process. Good leaders clearly and frequently explain why it is needed, and how it will

Critical Success Factors  ◾  263







benefit the business, the operation, and the entire workforce. They coach their teams to respect the frozen window and to use the Contingency Plan when disruptions occur. Crucially, they drive and coach their teams to meet stretch goals for schedule adherence and inventory compliance. 2. They must actively lead the effort, provide all needed resources and work to remove barriers, and resolve any conflicts. 3. With any new initiative or significant program, there can be both proponents and skeptics in the workforce. The proponents must be true disciples for the new process if it is to succeed. If they believe that this is a better way to do their jobs, they owe it to themselves to try to convince the skeptics. They need to listen patiently to all concerns and tactfully pose convincing counter-arguments. They need to realize that the skeptics may have valid reasons for their hesitation and be respectful, not judgmental. 4. Respect for and adherence to standard work is essential. Everyone throughout the organization must understand the need for standard work and the value, stability, and consistency it provides. Some plants tolerate a “cowboy mentality” where some think they have the latitude to decide what is best in any situation and take whatever action they feel is necessary. They sometimes do what looks like a good thing without a full understanding of all the ramifications, some of which may be detrimental. Unforeseen consequences have a way of rearing their head later, compounding the problem, and edging the plant back to the death spiral. We have seen companies that reverted to their traditional firefighting mode – and lost ten points of OEE when high efficiency is most needed. What frequently follows is an inventory roller-coaster, where the loss of throughput results in falling inventory for some SKUs. The following week there is more to do, and more disruptions, and before long inventory has dropped dangerously low – which in turn causes more schedule changes. While ingenuity is valuable and should be encouraged, it should be practiced within good Management of Change (MOC) practices, where new ideas and suggestions are evaluated from all perspectives and adopted if they appear to be an improvement. Documentation and communication of the new work practice are key pieces of MOC. 5. Typically, the path from the status quo to excellent scheduling is nonlinear. Sales will inevitably continue to push back, believing the only way to get what they want out of production is to apply pressure to make what their customers need, now. Planning, not fully realizing the impact of their actions, and often being accountable for keeping inventory between limits, prioritizes that over sticking to the schedule, thus compounding the problem. It takes strong leadership to patiently protect the schedule and allow the process to work. When sales and planning experience a slow improvement of production predictability, coupled

264  ◾  Getting to Success

with better service levels, then they will start to permanently adapt their perspective. 6. Leadership must recognize that their job is not done when the new scheduling process is in place and operating. There will be speed bumps along the way, so they must continue to champion the process, provide guidance, make sure the needed resources continue to be available, and ensure that the root causes of recurring problems are understood and corrected.

Chapter 30

Success Stories: Examples of Scheduling Best Practices We feel that the most compelling case we can make for the value of these practices is to let some of our clients and professional colleagues speak directly to you.

Dean Bordner – Nature’s Bounty Formerly Senior VP of Operations, Nature’s Bounty (Now The Bountiful Company) The heart of any business is its scheduling and we had just had a heart attack and needed triple bypass surgery! When you stop shipping to Costco for three weeks straight it is never a good thing. This was the reason I ended up hiring on with NBTY and taking over responsibility for their Deerfield Beach packaging location as well as other locations. One walk through the facility and it was easy to see problems and deficiencies. Out of stocks to our retailers, unscheduled downtime due to waiting on materials or changes in immediate priorities, unplanned and poorly executed changeovers, unproductive labor costs, excessive idle capacity, and non-existent preventative or predictive maintenance. The solutions to these issues however were not as readily available. We were also trying to implement HighPerformance teams but due to the unending amount of changes and disruptions in the schedules, we were not gaining traction. During this critical time, I was introduced to Noel Peberdy and Peter King who discussed a new concept to me of product wheels. The concept of product wheels, as laid out in Chapter 6, was DOI: 10.4324/9781003304067-35

265

266  ◾  Getting to Success

innovative but also intuitive. While I had not done this in production, a similar concept of ABC analysis in warehouses used similar principles to put your biggest movers into the closest travel lanes thus reducing travel time. Noel and Peter used analytics to determine which of our 1,000 SKUs needed to be run on a defined frequency and on which of the ten packaging lines we had to choose from. They also looked at the types of changeovers we had (full line, bottle size and color, label, allergen, etc.) to best align our product flow and reduce changeover time and unnecessary changeovers. This was an incredibly complex set of variables and would not have been possible without the introduction of scheduling wheels. Peter had to collect all the data some of which did not exist, find the truth in the data (how long did each type of changeover actually take versus hearsay) and put that data into a meaningful framework that could be understood and actionable. Fortunately, they have now incorporated all of this and many more positive innovations into their Phenix system to make this more streamlined and beneficial. Needless to say that as we implemented the scheduling wheels we started to see both immediate and long-term benefits. We set up lines by bottle size since that was our longest changeover event and then would run numerous SKUs of the same bottle size and only do label changes which were our quickest changeovers. We put in a Kanban system to ensure we had the materials ready and staged for the next three runs since we were now getting predictable schedules which eliminated most of our unplanned downtime. As we would now know which lines would be running, we could start preventative and then predictive maintenance which became an upward spiral of productivity. Free capacity was unlocked that didn’t need any capital dollars and with the labor that was turned from unproductive waiting or doing unnecessary changeovers we could run additional lines and cut into our backlog of orders thus producing massive gains in productivity and increases in sales revenues. While this was taking place it became a much more enjoyable place to work for all. Associates were able to know what their day would look like; it was predictable and growing and not chaotic anymore. Management time moved from fire-fighting to planning and strategy to unlock new areas of productivity. Our high-performance teams started to take off as we could keep the teams stable and have the time to work with them. The numbers told a great story – Service levels rose significantly. Capital needs for new lines for existing products were reduced to zero. Downtime dropped dramatically. Production costs were reduced by $1.5 Million per year

Success Stories 

($10 Million per year across all the sites where we implemented product wheels), while output was increased by 35%. I am happy to say that these and other improvements at the site enabled the Deerfield Beach site to win our site of the year award just a few years after implementing the concepts laid out inside this book. To add some background detail, one of the first things we did was assign bottle sizes to specific packaging lines, as described in Chapter 25. That alone added four to five points of OEE and increased throughput by more than 10%. The throughput increase from that and the product wheel implementation allowed two full lines to be idled, which led to the $1.5 million annual savings referenced.

Mike Evans – Bellisio Foods VP of Operations Scheduling the facility has always been a strength of our company. We found ourselves in a position in 2019 where our changeovers were becoming inefficient due to product mix and more SKUs than we could efficiently schedule. In addition, the plant scheduler for almost 30 years was retiring. We then partnered with Zinata and began the Product Wheel discovery and installation. The 13 models presented gave us the opportunity to best pick the operating model that would leverage plant efficiencies and changeovers, and optimize run time against the appropriate MAPE. As our legacy scheduler retired, the new scheduling team used the Product Wheels efficiently to optimize inventory levels while scheduling the plant efficiently. The Wheels also drove process improvement in our business by way of improved sales forecasting, fewer changeovers, and increased efficiencies. Albeit we did pivot during the pandemic to a less efficient scheduling model, we have been able to quickly move back to the product wheels learning how to run them in a disruptive environment. Would highly recommend the use of wheels in any production scheduling environment to reduce changeovers, optimize inventory levels and drive a more efficient S&OP process. An international pandemic, labor compression, and record inflation have really made plant scheduling even more challenging than it has ever been. Product Wheels have been the backbone of which to “grab on to” for these difficult environments.

◾  267

268  ◾  Getting to Success

Dave Rich – Litehouse Foods Vice President, Strategic Sourcing & Fulfillment At the time Zinata entered my life, the growth and success of Litehouse had been largely due to a sales-driven strategy, accompanied by an enthusiastic ownership culture. We had productive and engaged teams in our facilities but had reached a point where the proliferation of SKUs and machines had made scheduling extremely complex. With limited tools, we knew we were not optimizing efficiency on our lines but were continuing to plow ahead, adding new machines to get the capacity we suspected we could find in our lines if we could schedule more efficiently. We engaged King, Peberdy, and Nall to help us improve our scheduling process through product wheel projects, first in our Michigan facility, then in our Utah plant. We immediately found that one of the keys to success was their including plant personnel to determine what an efficient schedule really looked like. We allowed the experts on the floor to determine downtime drivers and provide their input on ranking product attributes for allocation across lines and sequencing within a line. Once we had a handle on our downtime drivers, it became relatively simple to implement product wheels across the lines in the plants. We were able to significantly reduce the number and severity of changeovers, minimizing intensive allergen washes, bottle size changes, etc. One aspect of this process that I hadn’t anticipated was its comprehensiveness. I went into this project thinking it would help us to allocate, schedule, and sequence our packaging lines better, and it certainly delivered on this promise. My hope was that we could find hidden capacity on our lines and delay the purchase of additional lines and additional staffing, and this mission was accomplished. What I hadn’t understood, however, was the impact this change would have on other functions in our company, from R&D to sales to customer service to inventory control to warehouse to logistics and transportation. We discovered changes that had to be made in what we asked of our customers in terms of SKU management and lead times. In the process of understanding what drives downtime, we made changes in product formulation and aligned our QA practices across facilities. We had to examine our entire process from soup to nuts, and in that examination also had to take a hard look at our culture and the way we went to market. Scheduling lies at the core of what a manufacturing business is and does, and Zinata’s thorough reexamination and improvement of scheduling practices at Litehouse had prerequisites and implications that were surprisingly broad and impactful.

Success Stories 

I am grateful for what these concepts and tools brought to our company, and would encourage anyone looking for hidden capacity on their lines to implement the practices outlined in the book, and to do so with eyes wide open to the potential impacts across your organization.

James Overheul – BG Products Formerly Operations Director The process for creating the product wheels allowed us to see why we were having issues in our production environment. The product wheels gave us a process so that we can respond to the changing needs of our customers yet not lose our way on the routine items. The reason that Overheul engaged us to implement wheels was to reduce the chaos and bring some structure to their scheduling process. We did that, with these added benefits: ◾ Assigned groups of bottles by size to specific packaging lines, thus simplifying changeovers and increasing throughput ◾ Reduced working capital by several hundred thousand dollars by moving some Make to Stock products to Make to Order ◾ Applied SMED (Chapter 23) to one of their most automated packaging lines, reducing the time for the most complex changeover from eight hours to less than three hours Prior to the introduction of product wheels, BG had been able to satisfy customer needs at very high service levels, largely through the efforts of a number of highly dedicated, conscientious, experienced employees who took initiative to meet immediate needs. However, the lack of structured work processes and standard practices required constant monitoring of market and operational conditions to ensure continued high levels of customer performance. BG operations leadership recognized this and initiated programs to bring more structure and rigor to the operation, better definition of work practices, and higher adherence to work standards. Value Stream Mapping and product wheels were key components of those programs. As a result of wheel implementation on the Rotary Filling Line and three other packaging lines, BG has seen a significant reduction in inventories while continuing its very high levels of customer service. But perhaps the greater benefit has been the stability and predictability this has brought to the operation.

◾  269

270  ◾  Getting to Success

Ryan Scherer – Appvion Former Organizational Excellence and Capacity Manager There are now 11 machines on wheels: two sheeters, five coaters, and four paper machines. All of the Appleton wheels, regardless of the type of equipment on which they were applied, have seen similar benefits: significantly lower inventories, shorter lead times, reduced changeover losses due to improved sequencing, greater schedule stability and predictability, thus resulting in improved customer performance with fewer BSPs [Broken Service Promises]. The on-going benefit of Appleton’s Lean Six Sigma efforts, including product wheels and pull replenishment strategies, has been $20–$30 million annually, each year …. Total inventory has been reduced by 21% and Cash Conversion Days by 17%. Implementing wheels on the coaters allowed us to level load our large runs to eliminate the peaks and valleys in the schedule, and thus reduce overtime, create better flow, reduce WIP, and allow for a more predictable production schedule. Before the wheels were implemented, the same sheet parts would be running on two different sheeters at the same time, changeovers between cut sizes were very high, and finished goods inventory was out of control due to the unpredictable replenishment lead time. These difficulties were increased by the use of a forecast and “push scheduling”. … Again, this effort brought predictability to the schedule, significantly reduced cut size changeovers, and began to demonstrate a positive impact on finished goods inventory. Because the results were so positive from the four wheels implemented on two coaters and two sheeters, our goal in 2009 was to get our three thermal coating machines on wheels. The thermal coaters were certainly more of a challenge than the carbonless coaters, but offered even greater opportunity due to high raw materials costs, and growing finished goods inventories due to the use of our ERP/MRP system and forecast mentality or “push” scheduling process. Again, we saw many of the results achieved earlier: improved schedule predictability, reduced and more efficient changeovers due to optimized sequencing, improved lead times, and fewer BSPs.

David Kaissling – Shearer’s Snacks Formerly Chief Supply Chain Officer for Shearer’s Snacks, With 40 Years in CPG as Head of Supply Chain for Fortune 500 Companies By implementing product wheels when I was head of Supply Chain we were able to move from fill rates of ~75% to over 99% reliably in

Success Stories 

a three-month timeframe. King’s approach to working with people on the floor captured in this book is key to managing the change needed to stabilize manufacturing and reap huge benefits. The wheel implementation made life on the floor for thousands of people better, particularly for folks off shift in the middle of the night. Having a predictable cycle of changeovers is huge to improving performance and improving morale on the factory floor. The reason we were brought in was not to improve throughput or fill rates, although we did both. The prime driver was to harmonize finite scheduling processes across the manufacturing sites. The business had recently transitioned daily production scheduling from the corporate office to the eight individual plants, and each had a somewhat different process with different business rules. The product wheel practices we installed provided that standardization.

Raymond Floyd – Exxon Mobil SVP Suncor Energy (Retired), Current Member of Manufacturing Hall of Fame, The Shingo Academy, and the Baldrige Award Board of Overseers In continuous processing, especially reactive continuous processing, effective production scheduling is a critical tool to optimize productto-product transitions and one of the most critical factors to achieve truly effective use of your production resources. King made a significant contribution to understanding and improving production scheduling in his first book. I have personally used his concepts with great benefit. As plant manager of the Exxon Mobil plant in Bayport, TX, Floyd led the implementation of Fixed Sequence Variable Volume, a repetitive scheduling strategy described in Chapter 6. That reduced changeover losses by 90% and increased reactor effective capacity from 50% of nameplate to more than 85%. With the success of that and other lean initiatives, Floyd and his plant were awarded the coveted Shingo Manufacturing Prize. Floyd described those improvements in Liquid Lean, which was awarded a Shingo Research Prize.

Ethylene Co-Polymers – Sabine, TX We designed wheels for a plant in Texas that makes extruded polymer pellets sold to customers who mold them into various plastic parts. The plant has four extruders, and a line-up of more than 60 products which can be made on any extruder.

◾  271

272  ◾  Getting to Success

To begin, the plant process engineer applied Group Technology to the product line-up for Extruders 1 and 2, which had shared 31 products, and was able to assign 15 and 16 products to Extruder 1 and 2, respectively. The number of product families run on Extruder 1 dropped from 8 to 2, dramatically simplifying changeovers. The EOQ analysis and the other factors analyzed suggested that a sevenday wheel would be appropriate for all four extruders. Each wheel included unassigned time of 24 hours or greater, and by staggering the start of the wheels, we were able to spread these times over the week to smooth out the requirement for preventative maintenance resources. For example, Extruder 1’s wheel started on Tuesday morning, so Monday was available for maintenance, product qualification runs, and continuous improvement activities. Extruder 2’s wheel started on Wednesday morning, so its PM time came around on Tuesday. The results were impressive. Extruder 1, for example, had been on a 21-day cycle, so the seven-day wheel allowed a significant inventory reduction, saving almost $400,000 in working capital. Even with the high inventory that had been available, the 21-day cycle had to be broken frequently to prevent stockouts. Because the methodology described in previous chapters was followed to design the seven-day wheel, it was more robust and more stable, resulting in more predictability, far fewer wheel breakages, and far less frustration on the plant floor. And the fewer and easier changeovers from the Group Technology product line-up reduced yield losses by more than $2 million per year. Here’s what the team leader presented to site management in the project review (Figure 30.1).

Yield Related

$2.4MM/yr

Yield - Autoclave impact

$306M/yr

Working Capital (one time)

$388M

Operating Utility (2% benefit - soft)

$287M/yr

More predictability for PM, less frustrations on shop floor

Priceless

Figure 30.1  Benefits of product wheels at the TX plant.

Success Stories 

Martin Fernandes – Dow Chemical Director of Supply Chain Innovation Fernandes told me in 2012 that Dow had made 5S, product wheels, and demand-driven pull systems a major focus of their continuous improvement initiatives. They were applying wheels across all process areas, from the upstream continuous chemical reactions, which may produce only a few products, to the downstream batch operations, which convert those few materials into perhaps several dozen product variations. The application of wheels and demand-driven pull had at that time resulted in 10%–20% lower inventories, 30%–40% shorter manufacturing lead times, greater operational stability and predictability, and more consistent lead times.

Dave Stauffer – Advanced Food Products Director of Supply Chain Advanced Food Products, a maker of cheese sauces and puddings, was dealing with operating practices that couldn’t support its volume growth strategy, and production scheduling knowledge that was left to a single person. By implementing Product Wheel scheduling using specialized software, line changeovers were reduced, and very time-consuming upstream feed system swaps decreased from several per week to several per year. They were able to increase annual volume by 6% using 2% less production hours. The most important change related to overtime. The site’s policy is to complete the week’s required production before leaving at the end of the shift on Friday evening/Saturday morning. Overtime was frequent to complete the runs. Saturday runs were reduced by 67% after implementing the repetitive scheduling strategies. It became a policy to include instructions for which additional products should be produced on the Friday midnight shift when running ahead of schedule. The reduction in overtime and the predictable, repetitive schedule patterns increased team morale. The level of chaos was so significantly reduced that the scheduler decided to postpone her retirement and continue in what had become a more enjoyable job. The specialized software chosen captured all the run rules and tribal knowledge making training of a new scheduler feasible. The concept in this book along with the Phenix planning tools allowed us to move very complex scheduling rules from head knowledge into a cloud-based system. It has improved our speed of scheduling and the consistency of scheduling to our established rules. It allowed us to hire and train a new scheduler with greater ease and success.

◾  273

Index Note: Locators in italics represent figures and bold indicate tables in the text. Accuracy, 76–81, 173 Ad hoc/poor scheduling, 6 Advanced planning and scheduling (APS) system, 130–131, 184 Aggregate inventory, 86, 86, 87, 115 Agility, 5, 23, 27, 50, 65, 104, 131, 145, 157, 169, 238, 244 Appleton Paper, 24, 50 APS system, see Advanced planning and scheduling system Assembly manufacturing, 9, 20 Attributes, visibility of, 143 “A” type process, 16, 17, 19, 20 Automated planning, 185–186 Automotive fluids packaging, 198–199 Autonomous maintenance, 204, 205, 253 Availability checking, 151 Batch and lot size restrictions, 118–119, 123 Batch chemical process equipment, 10 Behavioral critical success factors, 262–264 Bias in forecast, 76, 78 Bin scheduling, 127 Blast radius, 158 Blue Lakes salad dressing plant, 39 Blue Lakes value stream map, 42 Bordner, Dean (Nature’s Bounty), 265–267 Bottlenecks, 227 moving bottlenecks, 116, 228–233 poor scheduling causing, 228 in potato chip line, 232, 233 The Bountiful Company, 49 Breakdown maintenance, 204, 206 Breathing room, 51, 59, 68, 244–246 Bullwhip effect, 78, 78 Business continuity, 131, 140–141 Business imperatives, 3 scheduler’s world, 3–4

scheduling, challenge of, 4–5 scheduling, importance of, 5–6 scheduling as a foundation of manufacturing performance, 6–7 Business Planning, 27, 29, 77 Capacity, 19, 20, 28, 35, 64, 70 effective, 227, 229, 271 infinite, 133 planning, 28, 163 Cellular flow, 31, 218, 223 Cellular manufacturing, 213 applied to process lines, 215–217 Group Technology (GT), 223–225 salad dressing operation, application to, 223 synthetic sheet manufacturing example, 217–218 typical process plant equipment configurations, 213–215 virtual cell implementation in synthetic rubber production facility, 219–222 Center lining, 196 Centre of Excellence (CoE), 240 Cereal manufacturing, 228 Change management, 112, 238, 241; see also Enterprise Resource Planning (ERP) system; Leaders, steps to value for; Leadership; Multi-level scheduling; Planning; Scheduling; Scheduling processes and software; Typical scheduling process steps Changeover difficulty, 10 improvement in, 165 reduction, 195 automotive fluids packaging, 198–199 diaper manufacturing, 199–200 275

276  ◾ Index non-manufacturing example, 201 process industry changeovers, 197–198 SMED and its origins, 195 SMED applied to blue lakes packaging, 201–202 SMED beyond product changes, 200–201 SMED concepts, 196–197 starting up after, 10–11 times, 63, 135, 143, 165, 258 Coefficient of variation (CV), 78–80, 94 Collaboration, 166 Combined variability, 97–98 Committed orders, 113 Communicating the plan, 106 ERP and shop floor systems, 107 mixing schedule, 107 spice and liquid prep rooms, 107 packing line schedule, 106–107 Communication between systems, 41 Company management, 90 Complex scheduling, 131 Constraint, scheduling, 110 just-in-time scheduling, 110–111 manual scheduling, 110 repetitive sequence scheduling, 111 Constraint’s location, impact of, 118 Converging flows, 129–130 Corrective maintenance, 204 Covid-19, 3, 7, 67, 89, 241, 244 Cowboy mentality, 263 Critical materials, 152 Critical success factors, 261 cultural and behavioral, 262–264 scheduling strategy, 261 scheduling system, 262 CSL, see Cycle service level Cultural challenges, 45–46 Cultural critical success factors, 262–264 CV, see Coefficient of variation Cycle service level (CSL), 93, 98–99, 99 Cycle stock, 86, 90–91, 91 Daily time resolution, 134 Data, 177 analyzing the root cause of gaps, 180 checking data against a standard, 179 documenting the process, 178–179 leadership visibility, 180 master data and transaction data, 177–178 measuring and tracking results against a goal, 179 planning and scheduling data, 180–181

Data accuracy and timeliness problems, examples of, 178 Data audits/checking practices, 178 Degrees of freedom between levels, 117–118 Demand, variability in, 92–97, 92 Demand forecasting unit (DFU), 81 Demand management, 32, 176, 184 Detailed schedule, creating, 106 Detailed scheduling process, 108–110 DFU, see Demand forecasting unit Diaper manufacturing, 199–200 Discrete parts assembly manufacturing, 9 Disruption, 67 ability to deal with, 68–70 dealing with collaboration, 166 necessity of plant leader, 168 physical triage meetings, 166–167 virtual team, implementing, 167–168 example, 70–72 nature of, 67–68 Disruptive forces and countermeasures, 68 Distribution Requirements Planning (DRP) software, 32, 32 Distribution rules, 119–120 Divergence vs. convergence, 16–19 Diverging flows, 130 Documenting the process, 178–179 Downstream constrained process, 116 Downstream-constrained scheduling processes, 35, 36, 116, 118 DRP software, see Distribution Requirements Planning software Economic order quantity (EOQ), 54, 272 Economic Production Quantity (EPQ) model, 53–54 Effective capacity, 227, 229, 271 Effective production and capacity planning, 183 automated planning, 185–186 good production plan, characteristics of, 187–188 importance of planning, 183–184 inventory targets and constraints, managing, 188–189 overloads, resolving, 184–185 planning example, 186–187 Effective scheduling, laying the foundations for, 238 Enterprise Resource Planning (ERP) system, 107, 133, 139 daily time resolution, 134

Index  ◾  277 independence, assumption of, 134–135 infinite capacity, assumption of, 133–134 mixing schedule, 107 quality management, 137 record, system of, 137 repetitive scheduling in, 136 scheduling modules, 135 spice and liquid prep rooms, 107 EOQ, see Economic order quantity EPQ model, see Economic Production Quantity model Equipment spares, 3 ERP system, see Enterprise Resource Planning system Ethylene co-polymers (Sabine, TX), 271–272 Evans, Mike (Bellisio Foods), 267 Example process, 39 cultural challenges, 45–46 information flow, scheduling, 41 process, 39–41 product differentiating characteristics, 43–45 products, 43 Excel, 33, 139 business continuity, 140–141 issues with Excel, 141–143 lot sizing, 144 multi-level scheduling, 144 scheduling software, 141, 142 sequencing, 143 time offsets, 143–144 visibility of attributes, 143 Exception management, 104–105 Fernandes, Martin (Dow Chemical), 273 Fill rate, 98–99, 99 Finish-to-finish relationship, 121 Finish-to-order (FTO) environment, 51 Firm zone strategy, 152–153 Fixed Sequence Variable Volume (FSVV), 24, 50, 64 Flow paths, 129 advanced planning and scheduling (APS) implementation, 130–131 converging flows, 129–130 diverging flows, 130 Floyd, Ray, 24, 64 Floyd, Raymond (Exxon Mobil), 271 Forecasting, 75 bias and accuracy, 76–77 coefficient of variation (CV), 78–80 demand forecasting unit (DFU), 81 goals, 80–81

product segmentation for, 82–83 product transitions, 81–82 Forecast value add, 76 FSVV, see Fixed Sequence Variable Volume FTO environment, see Finish-to-order environment Gaps, analyzing the root cause of, 180 Gartner Group’s hype cycle, 171, 172 Glenday, Ian, 61, 62 GMP, see Good Manufacturing Practices Goal-seeking algorithms, 34–35 Good Manufacturing Practices (GMP), 137 Good production plan, characteristics of, 187–188 Good scheduling, 6, 23 Graphical drag-and-drop scheduling program, 141 Group Technology (GT), 223–225 GT, see Group Technology Heijunka methods, 49–50 Independence, assumption of, 134–135 Infinite capacity, 133 Infinite capacity, assumption of, 133–134 Information flow, scheduling, 41 Integrated Planning, 27 Introduction to TPM (Seiichi Nakajima), 210 Inventory, 85 calculating inventory requirements, 55–56 components of, 86–87, 86 constraints, between levels, 123 cycle service level (CSL) and fill rate, 98–99, 99 cycle stock and safety stock, 90–91, 91 demand, variability in, 92–97, 92 example, 89–90 lead time, variability in, 97–98 managing, 87–89, 87 managing inventory targets and constraints, 188–189 safety stock, calculating, 91–92 safety stock and lot size impact, 99–100 Just-in-time scheduling, 110–111 Kaissling, David (Shearer’s Snacks), 270–271 Kaizen Event, 256 Key Performance Indicators (KPIs), 34, 108, 111–112 Kibble, 125–126 King, Jennifer, 24

278  ◾ Index King, Pete, 6 King, Peter, 24 Kingston pet food plant, 125, 126 KPIs, see Key Performance Indicators Labor risk, 165 Leaders, steps to value for, 238 improving scheduling, 245 aligning the plant to the wheel rhythm, 246–247 driving further improvements, 246 simple product wheel scheduling as a team, 245 successes, celebrating, 246 layout the improvement goals and plan, 238 change plan, developing, 241 incremental implementation plan, developing, 240–241 leaders and other stakeholders, communicating to, 239–240 supporters and cheerleaders, identifying, 240 tangible vision, developing, 238–239 sustaining the gains, 248 establishing sustainable practices early, 248–249 formalizing training, qualification, and coaching, 249–250 implementing a planning community of practice (COP), 250 key benefits, tracking, 250 ownership, 248 vendor software improvements, taking advantage of, 250 verifying that sustainment practices are working, 249 taking stock, 247 budget approval, getting, 248 deciding on full plant rollout, 247 full implementation, planning, 248 implementation consultant, selecting, 247 lessons learned, 247 review progress, 247 scheduling software, selecting, 247 work on the culture, 241 bringing the voice of the customer into the plant, 241 frozen horizon, 243 schedule disruption, dealing with, 243–245 shop floor discipline, improving, 242

weekly reporting and drive improvement, implementing, 242–243 Leadership, 46, 252, 262 and project tasks, 237, 237 visibility, 180 Lead time, 97–98, 104 Lean implementing, 6 synergy with, 59–60 Lean Manufacturing, 195 Lean RfS (Ian Glenday), 24, 61, 63 Liker, Jeffrey, 49–50 Limited extra capacity, 20 Limited resources, balancing, 12–16 Linking between activities, 122 Liquid Lean (Ray Floyd), 64 Logical relationships between levels, 121 Lot size, 99–100, 144 Lower-volume products, 56–57 Luvs diapers, 70–72 Maintenance prevention, 205 Maintenance windows, 6 Make-to-order (MTO) environment, 51, 52, 53, 57–59 Make-to-stock (MTS) environment, 51, 53, 55 Management of Change (MOC) practices, 263 Manual scheduling, 110 Manufacturing performance, 6–7 Manufacturing processes, 9 Manufacturing Requirements Planning (MRP) system, 4, 20, 25, 34 Manufacturing strategy, 5 MAPE, see Mean average percentage error Master data, 177–178 Master production scheduling, 28 Master roll, 19 Mean average percentage error (MAPE), 78, 79 MEIO, see Multi-echelon inventory optimization systems MEIO packages, see Multi-level Inventory Optimization packages Minor stops, 10 Mixing schedule, 107 MOC practices, see Management of Change practices Money, total productive maintenance saving, 206 Moving bottlenecks, 228–229 complex example of, 232

Index  ◾  279 scheduling, 230–233 MRP system, see Manufacturing Requirements Planning system MTO environment, see Make-to-order environment MTS environment, see Make-to-stock environment Multi-echelon inventory optimization systems (MEIO), 95 Multi-level Inventory Optimization (MEIO) packages, 152 Multi-level scheduling, 115, 119, 144 batch and lot size restrictions, 118–119 bottlenecks, moving, 116 distribution rules, 119–120 inventory constraints between levels, scheduling with, 123 linking between activities, 122 logical relationships between levels, 121 problems, 121 process, 122–123 product mix, 116 scheduling problems, types of, 117 constraint’s location, impact of, 118 degrees of freedom between levels, 117–118 when there are more than two levels, 118 Multi-step manufacturing, 11–12 Nakajima, Seiichi, 210 New process designing, 193 executing, 193–194 New product development involvement, 164 New scheduling system, implementing, 171 Non-consumable resources, 112 OEE, see Overall equipment effectiveness On Time in Full (OTIF) service performance, 241 Optimum sequence, determining, 57 OTIF service performance, see On Time in Full service performance Overall equipment effectiveness (OEE), 6, 7, 46, 206, 208 availability, 206–207 calculation of, 208, 209 metrics, 46, 60 non-standard OEE metrics, 210–211 performance, 207 quality, 207–210 Value Stream Map (VSM) data boxes, 209–210

Overall Operating Efficiency, 6 Overheul, James (BG Products), 269 Overloads, resolving, 184–185 P&G Luvs diapers, 70–72 Packing line schedule, 106–107 Packowski, Josef, 24, 63 PC, see Performance cycle Performance cycle (PC), 96–97 Phase-ins and phase-outs, see Product transitions Physical triage meetings, 166–167 Plan, communicating, 106 ERP and shop floor systems, 107 mixing schedule, 107 spice and liquid prep rooms, 107 packing line schedule, 106–107 Planning automated, 185–186 example, 186–187 importance of, 183–184 and scheduling data, 180–181 and scheduling process, 103 Plant leader, role of, 163, 164 disruption, dealing with collaboration, 166 necessity of plant leader, 168 physical triage meetings, 166–167 virtual team, implementing, 167–168 future proof the plant, 163 changeover times, 165 example, 165–166 labor risk, 165 new product development involvement, 164 product portfolio, simplifying, 165 raw material supply risk, 164 selective automation, 165 standardizing packaging raw materials, 164 transportation risks, 165 repetitive patterns of production, reinforcing, 168–169 Plant making salad dressings, 20 PM, see Preventative Maintenance PM practices, see Productive maintenance practices Poor scheduling, 6, 228 Pre-cellular scheduling process, 220 Predictive maintenance, 204 Preparing to plan, 105 Preventative Maintenance (PM), 192, 204 Privacy of software, 159

280  ◾ Index Probability distribution, 77 Process industries, 9 Process industry changeovers, 197–198 Process industry manufacturing, 9 Process operations, characteristics of, 9 changeover, starting up after, 10–11 changeover difficulty, 10 divergence vs. convergence, 16–19 limited extra capacity, 20 limited resources, balancing, 12–16 multi-step manufacturing, 11–12 product differentiation points, 19–20 sanitation cycles, 11 shelf-life constraints, 11 Procter & Gamble, 188, 199, 201 Product attributes, 34, 107, 111, 135, 142, 143, 262, 268 Product differentiating characteristics, 43–45 Product differentiation points, 19–20 Product families, assigning, 52 Production planning, 28–31, 105–106 Production scheduling, 5 Production stability, 203 overall equipment effectiveness (OEE), 206, 208 availability, 206–207 calculation of, 209 non-standard OEE metrics, 210–211 performance, 207 quality, 207–210 Value Stream Map (VSM) data boxes, 209–210 total productive maintenance (TPM), 204–205 relevance, in process industries, 205 saving money, 206 Production strategies, 23–25 Production supervisors, 46 Productive maintenance (PM) practices, 204 Product mix, 116 Product portfolio, simplifying, 165 Products, 43 Product segmentation for forecasting, 82–83 Product transitions, 81–82 Product wheel design appreciable changeover times/losses, 52 current scheduling process, revising, 57–59 Economic Production Quantity (EPQ) model, 53–54 inventory requirements, calculating, 55–56 lower-volume products, 56–57 make-to-order (MTO), 52

optimum sequence, determining, 57 product families, assigning, 52 stakeholders, reviewing with, 57 value stream map (VSM), 51–52 wheel time, 52–55 Product wheels, 24, 50 benefits of, 60–61 design, 51–59, 58 lean, synergy with, 59–60 Project success, roadmap to, 251 final preparation, 257–258 initial preparation, 252–254 strategy design, 255–257 sustaining, 258–259 system design, scheduling, 254–255 Prosci Change methodology, 241 Push scheduling, 270 Quality management, 137 Rapid Improvement Team activity, 256 Raw material standardizing packaging raw materials, 164 supply risk, 164 Reaction time, 104 Readiness criteria, scheduling, 171 project roles, 173–176 readiness and sustainability, 173, 174–175 readiness examples, 176 Record, system of, 137 Releasing orders to production, 113 Repetitive flexible supply (RfS), 24, 50, 61–63 Repetitive scheduling, 35 in ERP system, 136 requirements, 147 Repetitive scheduling strategies, 49 Fixed Sequence Variable Volume (FSVV), 64 product wheels, 50 benefits of, 60–61 design, 51–59, 58 synergy with lean, 59–60 repetitive flexible supply (RfS), 61–63 Rhythm Wheels, 63 Repetitive sequence scheduling, 111 Request for Information (RFI), 37 Request for Proposal (RFP), 37, 148 Resources, 112–113 Reverse osmosis (RO) system, 45 RFI, see Request for Information RFP, see Request for Proposal Rhythm Wheels, 24, 50, 63

Index  ◾  281 Rich, Dave (Litehouse Foods), 268–269 RO system, see Reverse osmosis system Rough cut planning, 28 S&OE, see Sales & Operational Execution S&OP, see Sales & Operation Planning Safety stock, 87, 90–91, 91, 98–100 Salad dressing production, 12, 15 Sales & Operational Execution (S&OE), 28 Sales & Operation Planning (S&OP), 27, 220 Sanitation cycles, 11 Schedule adherence, 242 Schedule evaluation and adjustments, 113 Scheduler, 3–4, 30, 33–35, 81, 95, 112, 134, 144, 166, 176, 180, 191 Scheduling, 31 challenge of, 4–5 as a foundation of manufacturing performance, 6–7 importance of, 5–6 process, 35–36 Scheduling problems, types of, 117 constraint’s location, impact of, 118 degrees of freedom between levels, 117–118 Scheduling processes and software, 27 goal-seeking algorithms, 34–35 production planning, 28–31 repetitive scheduling, 35 scheduling, 31 scheduling process, 35–36 scheduling software, 33–34 software selection, 37 supporting processes, 31–33 Scheduling software, 33–34 critical features of, 141 vs. Excel, 142 Scheduling strategy critical success factors, 261 Scheduling system critical success factors, 262 Scherer, Ryan (Appvion), 50, 270 Schonberger, Richard, 216 Seasonality, 97 Security of software, 158 Selective automation, 165 Sequencing, 44, 143 Service-level factor, 93 Service-level goals, 95 Sheet goods process, 218, 219 Shelf-life constraints, 11 Shingo, Shigeo, 195 Shop floor systems, 107

Single Minute Exchange of Dies (SMED), 59 applied to blue lakes packaging, 201–202 automotive fluids packaging, 198–199 beyond product changes, 200–201 concepts, 196–197 diaper manufacturing, 199–200 improvement steps, 196 and its origins, 195 non-manufacturing example, 201 process industry changeovers, 197–198 SKUs, see Stock Keeping Units SMED, see Single Minute Exchange of Dies Software, scheduling, 157 privacy, 159 security, 158 Software designed for production scheduling, 145 multi-level requirements, 147 repetitive scheduling requirements, 147 scheduling requirements, 146–147 selection of software, 147–149 supporting processes, 145–146 Software selection, 37 SPC, see Statistical Process Control Spice and liquid prep rooms, 107 Spotify, 167 Stakeholders, reviewing with, 57 Standard deviation, 93, 94 Staple fiber manufacturing, 12, 14 Statistical Process Control (SPC), 214 Stauffer, Dave (Advanced Food Products), 273 Stock Keeping Units (SKUs), 5, 9, 18, 18, 152 Strategy design, 255–257 Strategy examples, 153–155 Success stories, 265 Bordner, Dean (Nature’s Bounty), 265–267 ethylene co-polymers (Sabine, TX), 271–272 Evans, Mike (Bellisio Foods), 267 Fernandes, Martin (Dow Chemical), 273 Floyd, Raymond (Exxon Mobil), 271 Kaissling, David (Shearer’s Snacks), 270–271 Overheul, James (BG Products), 269 Rich, Dave (Litehouse Foods), 268–269 Scherer, Ryan (Appvion), 270 Stauffer, Dave (Advanced Food Products), 273 Supply chains, 3 Supporting processes, 31–33 Synthetic rubber manufacturing configuration, 219–222, 220, 222

282  ◾ Index Synthetic sheet manufacturing example, 217–218 System design, scheduling, 254–255 Tank scheduling, 127–129 Teal organization model, 168 Theory of Constraints (Goldratt), 123, 227 Time at risk, 61, 71, 96 Timeliness problems, examples of, 178 Time offsets, 143–144 Tomorrow, preparing for, 108 Total productive maintenance (TPM), 204–205 relevance, in process industries, 205 saving money, 206 Toyota Production System, 195 The Toyota Way (Jeffrey Liker), 49–50 TPM, see Total productive maintenance Transaction data, 177–178 Transportation risks, 165 Trough of Disillusionment, 171 Two-cycle product wheel, 24, 24 Typical process plant equipment configurations, 213–215 Typical scheduling process steps, 103 communicating the plan, 106 ERP and shop floor systems, 107 packing line schedule, 106–107 detailed schedule, creating, 106 detailed scheduling process, 108–110 evaluating and adjusting the schedule, 113 exception management, 104–105 KPI-based algorithms and solvers, 111–112 planning and scheduling process, 103 preparing for tomorrow, 108 preparing to plan, 105 production plan, creating, 105–106 releasing firm/committed orders, 113 resources, 112–113 scheduling the constraint, 110 just-in-time scheduling, 110–111 manual scheduling, 110

repetitive sequence scheduling, 111 Unilever, 167 Upstream-constrained schedule, 35, 36, 36, 116, 116, 118 Value-added forecasting process, 76 Value for leaders, see Leaders, steps to value for Value stream map (VSM), 42, 51–52, 131, 209–210 Variability in lead time, 97–98 Virtual cell implementation in synthetic rubber production facility, 219–222 Virtual team, implementing, 167–168 Virtual work cells advantages of, 217 grouping into, 216 Visibility of attributes, 143 Vitamin tablet manufacturing, 12, 13 Volatility, level of, 3 Volume planning, 28 VSM, see Value stream map “V” type process, 16, 17, 18–20 WAPE, see Weighted average percentage error Waterfall charts, 80 Weighted average percentage error (WAPE), 78 Wheel time, 52–55 WIP, see Work in Process Workforce engagement new process designing, 193 executing, 193–194 selling the idea, 191–193 Work in Process (WIP), 85 World Class Manufacturing (Richard Schonberger), 216 Z factor, 93