Advancement of Science and Technology in Sustainable Manufacturing and Process Engineering (Green Energy and Technology) [1st ed. 2024] 3031411722, 9783031411724

Table of contents :
Preface
Contents
Application of FACTS Devices for Transient Stability Enhancement in the Adama II Grid-Connected Wind Farm
1 Introduction
2 Literature Review and Basic Theories
2.1 Wind Turbines
2.2 Wind Farm Aggregation
2.3 General Assumptions for Modeling Wind Farms
3 Methodology
3.1 Minimization of Voltage Deviation
3.2 System Loss Reduction
3.3 Constraints of the Problem
3.4 Transformer Tap Setting Constraints
3.5 Case Study Test System Architecture
3.6 Data Collection
3.7 Data Analysis
3.8 Model of Adama II Wind Farm
3.9 Model of Adama II Wind Farm Integrated to Modified IEEE 14 Bus Test Systems
3.10 Optimal Placement of FACTS
4 Results and Discussion
4.1 Test System Analysis
4.2 Test System Analysis with UPFC Placement
4.3 Test System Analysis with IPFC Placement
4.4 Voltage Stability, Transient Stability, and Loss Reduction with FACTS Devices
5 Conclusion
References
Development of an Interpretable Deep Learning System for the Identification of Patients with Alzheimer´s Disease
1 Introduction
2 Materials
2.1 Subjects
2.2 Data Pre-processing
2.3 CNN Model
2.4 CNN Visualization
3 Results
4 Discussion
5 Conclusion
References
Digital Clinical Decision Support System for Screening of Eye Diseases
1 Introduction
2 Methods
2.1 Image Acquisition and Enhancement
2.2 Retinal Vessel Extraction
2.3 K-Means Clustering-Based Image Segmentation
2.4 Cup-to-Disk (CDR) Calculation
2.5 Intraocular Pressure Measurement
3 Results
3.1 Simulation
3.2 User Interface and Prototype
3.3 System Testing
4 Conclusion
References
Evaluation of Three Irrigation Management Tools for Improving Crop and Water Productivity of Wheat (Triticum aestivum) in Koga...
1 Introduction
2 Material and Methods
2.1 Description of the Study Area
2.2 Cropping Practices and Crop Production
2.3 Experimental Design and Treatment Setup
2.4 Land and Seedbed Preparation
2.5 Materials Used
2.6 Irrigation Water Productivity
3 Results and Discussion
3.1 Irrigation Water Used for Wheat Yield
3.1.1 Irrigation Water Applied at the Farm Level
3.1.2 Irrigation Water Used at Scheme Level
3.2 Wheat Yield
3.2.1 Wheat Yield at Farm Level
3.3 Irrigation Water Productivity
3.3.1 Irrigation Productivity at the Farm Level
3.3.2 Irrigation Water Productivity at Scheme Level
3.4 Effect of WFD and Chameleon Sensor at Block and Scheme Level
3.4.1 At the Block Level
3.4.2 At the Scheme Level
4 Conclusion
References
Design and Numerical Analysis of a Sorghum Reaper Machine
1 Introduction
2 Present Status of Sorghum Harvesting Methods in Ethiopia
2.1 Primary Data Collection
2.2 Manual Harvesting States in Raya Kobo and Azebo
3 Materials and Methods
3.1 Materials
3.2 Methods
3.3 Analytical Analysis of the Key Part for the Reaper of Sorghum Stalks
Design and Selection of Power Transmission System of a Machine
4 Result and Discussion
4.1 Analytical Analysis and Numerical Simulation
4.2 Design Specification of Root-Stalk Sorghum Reaper Machine
4.3 Comparison of Manual and Root-Stalk Sorghum Harvesting Results
5 Conclusion
6 Recommendation
References
Effects of Initial Moisture Content and Storage Duration on Physical and Chemical Characteristics of Stored Maize (Zea mays L....
1 Introduction
2 Materials and Methods
2.1 Experimental Design and Treatments
2.2 Chemical Composition
2.3 pH Value
2.4 Germination Test
2.5 Grain Damage Percent
2.6 Statistical Analysis
3 Results and Discussion
3.1 Stored Maize Grain Temperature
3.2 Ambient Temperature and Relative Humidity
3.3 Physical and Chemical Characteristics of Stored Maize
4 Conclusion
References
Experimental Analysis of the Effect of Exposure of an Electro-optic System to External Magnetic Field (Case Study: The Input L...
1 Introduction
2 Magnetic Field Effect on Electro-optical Devices
3 Experimental Setup and Method
4 Results and Discussion
4.1 Output Voltage Measurement
4.2 Comparison of Magnetic Field Effect on Output Signal
5 Conclusion
References
Identification of Dominant Lactic Acid Bacteria and Yeast Species from Teff Injera Dough Fermentation
1 Introduction
2 Materials and Methods
2.1 Raw Material Collection and Sample Preparation
2.2 Physicochemical Analysis of Teff Dough
2.3 Kinetics Growth and Enumeration of LAB and Yeast in Fermented Teff Dough
2.4 Identification of LAB
2.5 Identification of Yeast
2.5.1 Morphological Characterization of Yeast
2.5.2 Physiological Characterization of Yeast
2.6 Experimental Design and Data Analysis
2.7 Data Analysis
3 Results and Discussions
3.1 Physicochemical Analysis of Fermented Teff Dough Prepared Under Laboratory Conditions
3.2 Kinetics Growth of LAB and Yeast in Fermented Teff Dough
3.3 Morphological Characterization of LAB
3.4 Physiological Characterization of the Identified LAB Species
3.5 Biochemical Characterization Test for LAB
3.6 Identification of LAB
3.7 Morphological Characterization of Yeast
3.8 Physiological Characterization of Identified Yeast
3.9 Biochemical Characterization Test for Yeast
3.10 Identification of Yeast
4 Conclusion
References
Performance Evaluation of Grain Pro Collapsible Natural Convection Solar Dryer for Maize Grain Drying in Ethiopia
1 Introduction
2 Materials and Methods
2.1 Experimental Setup and Description
2.2 Experimental Procedures
2.3 Data Collection
2.3.1 Moisture Ratio
2.3.2 Drying Rate
2.3.3 Water Removal from Maize Grain
2.3.4 Total Energy Consumption
2.3.5 Drying Efficiency
2.4 Data Analysis
3 Results and Discussion
3.1 Temperature, Relative Humidity, and Solar Radiation of the Experimental Site
3.2 Temperature and Relative Humidity Variation of Air During Loading and Unloading of the Collapsible Dryer
3.3 Performance Evaluation of Collapsible Dryer
3.3.1 Moisture Ratio of Maize Grain
3.3.2 Drying Rate of Open Sun and Collapsible Dryer for Maize Grain Drying
3.3.3 Water Removal from Maize Grain
4 Conclusion
References
Quality Assurance Management Practices in Public Building Construction Projects: The Case of Bahir Dar City
1 Introduction
1.1 Definition of Quality
1.2 Quality Assurance
1.3 Benefits of Quality Assurance
1.4 Factors Affecting Project Quality Assurance
1.5 Mechanisms to Improve Project Quality Assurance
2 Research Methodology
2.1 Research Design
2.2 Sampling Technique
2.3 Data Collection Tools/Instruments
2.4 Data Measurement
2.5 Data Analysis
3 Result and Discussion
3.1 Current QA Management Practices in Public Construction Projects
3.2 Techniques Used for QA Management Practice
3.3 Major Factors That Affect QA Management Practices
3.4 Mechanisms to Improve QA Management Practices
3.5 Summary of Interview
3.6 Results of the Selected Case Studies
3.7 Discussion on the Findings
3.8 Construction QA and Management Framework
4 Conclusion
References
Review on Water Automatic Teller Machine (Water ATM) Technologies
1 Introduction
2 Methodology
2.1 Coin- and Note-Operated Water ATM Technologies
2.2 Smart Card Water ATM Technologies
2.3 Smart Card IoT-Based Water ATM Systems
2.4 Water ATMs´ Global Status
3 Conclusion
References
Yeast Bioremediation of Cr (VI) from Tannery Wastewater
1 Introduction
1.1 Background
2 Methodology
2.1 Chemicals
2.2 Sample Preparation
2.3 Characterization of Tannery Wastewater
2.4 Bio-removal of Cr (VI) in Liquid Media
2.5 Fourier Transform Infrared Spectroscopy (FTIR) Analysis of Bioaccumulation of Cr (VI)
2.6 Experimental Design for Bio-removal of Cr (VI)
2.7 Colony-Forming Unit Assay for Cell Viability Assessment
2.8 Numerical Optimization
3 Results and Discussion
3.1 Characteristics of Collected Tannery Wastewater
3.2 Statistical Analysis of Cr (VI) Removal Efficiency
3.3 Effect of Variables on the Removal of Cr (VI)
3.4 FTIR Analysis of Cr (VI) Removal
4 Conclusion
References
Production and Characterization of Ink from the Milky Fluid of Ficus vasta and Euphorbia abyssinica
1 Introduction
2 Materials and Methods
2.1 Equipment and Instrument
2.2 Chemicals and Reagents
2.3 Experimental Design
2.3.1 Experimental Setup and Description
2.4 Methods
2.4.1 Raw Material Collection
2.4.2 Raw Material Preparation
2.4.3 Phenolic Compound Test
2.4.4 Ink Formulation Procedure
2.5 Product Characterization
2.5.1 Determination of Viscosity
2.5.2 Determination of Drying Time
2.5.3 Measuring pH Value
2.6 Statistical Data Analysis
3 Results and Discussion
3.1 Characterization of Raw Materials
3.2 Phenolic Test
3.3 Effect of Solvent Concentration on Drying Time
3.4 pH Value of the Ink
3.5 Effect of Solvent on Viscosity
4 Conclusion
References
Assessment of the Causes of Non-revenue Water in Urban Water Distribution Systems: The Case of Bahir Dar City, Ethiopia
1 Introduction
2 Materials and Methods
2.1 Description of the Study Area
2.2 Methods
2.2.1 Data Collection and Preparation
2.2.1.1 Primary Data Collection
2.2.1.2 Secondary Data Collection
2.2.2 Sampling Method and Sample Size
2.2.3 Techniques of Data Analysis
3 Result and Discussion
3.1 Demographic Characteristics of the Respondents
3.2 Production of Water
3.3 Trends of Water Supplied, Billed Water, and NRW for Bahir Dar City from 2015/16 to 2019/20
3.4 Possible Causes of Non-revenue Water and Possible Solutions
3.4.1 Causes and Solutions of Real Losses
3.4.1.1 Cause of Real Losses
3.4.1.2 Solutions of Real Losses
3.4.2 Causes and Solutions of Apparent Losses
3.4.2.1 Causes of Apparent Losses
3.4.2.2 Solutions of Apparent Losses
4 Conclusion
References
Characteristics of Supply Chain Integration of Manufacturing Firms in Ethiopia
1 Introduction
2 Literature Review
3 Research Method
4 Results and Discussion
5 Conclusion and Outlook
References
Effect of Threshing and Storage Conditions on Mold Contamination of Stored Maize Grain (Zea mays L.)
1 Introduction
2 Materials and Methods
2.1 Materials
2.2 Experimental Procedure
2.2.1 Drying and Threshing of Maize Cob
2.2.2 Maize Storage
2.3 Sampling Procedure
2.4 Experimental Design
2.5 Preparation of Culture Media
2.6 Data Collection
2.6.1 Temperature and Relative Humidity of the Storage Room
2.6.2 Moisture Content
2.6.3 Mold Detection of Maize Grain
2.7 Data Analysis
3 Results and Discussion
3.1 Temperature and Relative Humidity of the Storage Room
3.2 Moisture Content of the Stored Maize Grain
3.3 Morphological Features of Fungi Identified in Maize Grain
3.4 Incidence and Frequency of Mold Fungi in Stored Maize Grain
4 Conclusion
Appendix 1
References
Formulation and Evaluation of Firefly and Artificial Bee Colony Algorithms for Maximum Power Extraction of Photovoltaic System...
1 Introduction
2 Review of Related Researches in Maximum Power Extraction of PV Systems
3 PV Characteristics
4 Partial Shagging Maximum Power Point Tracking
4.1 FFA Algorithm-Based Maximum Power Point Tracking
4.2 MPPT Using ABC Algorithm
5 Performance Evaluation
5.1 Specifications
5.2 Results and Discussion
6 Conclusion
References
Impact of Parametric Uncertainties on the Steady-State Performances of DFIG-Based WECS
1 Introduction
2 Parametric Uncertainty in DFIG Model Formulation
2.1 Model of DFIG-Based WECS in d-q Axes
2.2 Steady-State Model of DFIG-Based WECS and Its Performances
3 Results and Discussion
4 Conclusion
References
Participatory Evaluation and Demonstration of Handheld Maize Sheller
1 Introduction
2 Material and Methods
3 Results and Discussion
4 Suggestion from the Operators (Farmers)
5 The Economic Data and Cost Benefit Analysis
6 Conclusion and Recommendation
References
Index

Citation preview

Green Energy and Technology

Kibret Mequanint · Assefa Asmare Tsegaw · Zenamarkos Bantie Sendekie · Birhanu Kebede · Ephrem Yetbarek Gedilu   Editors

Advancement of Science and Technology in Sustainable Manufacturing and Process Engineering

Green Energy and Technology

Climate change, environmental impact and the limited natural resources urge scientific research and novel technical solutions. The monograph series Green Energy and Technology serves as a publishing platform for scientific and technological approaches to “green”—i.e. environmentally friendly and sustainable—technologies. While a focus lies on energy and power supply, it also covers "green" solutions in industrial engineering and engineering design. Green Energy and Technology addresses researchers, advanced students, technical consultants as well as decision makers in industries and politics. Hence, the level of presentation spans from instructional to highly technical. **Indexed in Scopus**. **Indexed in Ei Compendex**.

Kibret Mequanint • Assefa Asmare Tsegaw Zenamarkos Bantie Sendekie • Birhanu Kebede Ephrem Yetbarek Gedilu Editors

Advancement of Science and Technology in Sustainable Manufacturing and Process Engineering

Editors Kibret Mequanint Chemical and Biochemical Engineering Western University London, ON, Canada

Assefa Asmare Tsegaw Mechanical Engineering Bahir Dar University Bahir Dar, Ethiopia

Zenamarkos Bantie Sendekie Chemical Engineering Bahir Dar University Bahir Dar, Ethiopia

Birhanu Kebede Civil and Water Resources Engineering Bahir Dar University Bahir Dar, Ethiopia

Ephrem Yetbarek Gedilu Civil and Water Resources Engineering Bahir Dar University Bahir Dar, Ethiopia

ISSN 1865-3529 ISSN 1865-3537 (electronic) Green Energy and Technology ISBN 978-3-031-41172-4 ISBN 978-3-031-41173-1 (eBook) https://doi.org/10.1007/978-3-031-41173-1 © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors, and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer imprint is published by the registered company Springer Nature Switzerland AG The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland Paper in this product is recyclable.

Preface

It is a great honor for us to introduce the book Advancement of Science and Technology in Sustainable Manufacturing and Process Engineering, which includes chapters based on selected papers from the 10th edition of the EAI – International Conference on Advancements of Science and Technology (EAI ICAST 2022). EAI ICAST 2022 is an annual conference that takes place at Bahir Dar Institute of Technology, Bahir Dar University, Bahir Dar, Ethiopia. The conference covers topical science and technology issues and has brought together researchers, engineers, developers, practitioners, scholars, scientists, and academicians from around the world. The technical program of EAI ICAST 2022 was organized from seven main tracks: Track 1, Sustainable Processes for Green Technologies; Track 2, Artificial Intelligence and Digitalization for Sustainable Development; Track 3, River Basin Management and Trans-boundary Cooperation; Track 4, Agro-Mechanization and Manufacturing Systems; Track 5, Advances in Electrical and Computer Engineering; Track 6, Advances in Green Energy Technologies; and Track 7, Materials for Emerging Technologies. In this book, the Manufacturing and Process Engineering volume, 19 papers related to Agro-Mechanization and Manufacturing Systems and Sustainable Processes for Green Technologies, were accepted for publication. We sincerely appreciate the work of the Steering Committee Chair and members; the Organizing Committee Chair, Kibret Mequanint; the Organizing Committee Co-chairs, for their constant support and guidance that ensured the success of the conference. It was also a great pleasure to work with such an excellent Organizing Committee. We are grateful to the Technical Program Committee Chair and TPC Co-chairs; Zenamarkos Bantie, Abdulkerim Mohammed, Birhanu Kebede, Assefa Asmare, Teketay Mulu, Eshetu Getahun, and Addisu Alemayehu. The team performed exceptionally well to handle the peer-review process and design a highquality technical program. We are also grateful to the conference manager, Veronika Kissova, for her support and guidance throughout the process; EAI Publications Coordinator, Carlos Valiente for his determined work to facilitate the publication; and all the authors who submitted their papers to the EAI ICAST 2022 conference. v

vi

Preface

We believe that the EAI ICAST 2022 conference provided the platform for the scientific communities to discuss all science and technology aspects relevant to each track. We also expect that future EAI ICAST conferences will be as successful and stimulating, as indicated by the contributions presented in this volume. London, ON, Canada Bahir Dar, Ethiopia Bahir Dar, Ethiopia Bahir Dar, Ethiopia Bahir Dar, Ethiopia

Kibret Mequanint Assefa Asmare Tsegaw Zenamarkos Bantie Sendekie Birhanu Kebede Ephrem Yetbarek Gedilu

Contents

Application of FACTS Devices for Transient Stability Enhancement in the Adama II Grid-Connected Wind Farm . . . . . . . . . . Yosef Berhan Jember, Tewodros Gera Workineh, Tefera T. Yetayew, and Meresa Dinkayow 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Literature Review and Basic Theories . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Development of an Interpretable Deep Learning System for the Identification of Patients with Alzheimer’s Disease . . . . . . . . . . . Selamawet Workalemahu Atnafu and Stefano Diciotti 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Digital Clinical Decision Support System for Screening of Eye Diseases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Gizeaddis Lamesgin Simegn and Mizanu Zelalem Degu 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1

2 3 7 15 25 26 27 27 28 32 32 36 36 39 40 41 44 47 48

vii

viii

Contents

Evaluation of Three Irrigation Management Tools for Improving Crop and Water Productivity of Wheat (Triticum aestivum) in Koga Irrigation Scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Habtamu D. Tarekegn, Seifu A. Tilahun, Fasikaw A. Zimale, and Petra Schmitter 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Material and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

51

52 53 57 67 67

Design and Numerical Analysis of a Sorghum Reaper Machine . . . . . . . 71 Adino Amare Kassie, Hailu Shimels Gebremedhen, and Hailemichael Solomon Addisu 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 2 Present Status of Sorghum Harvesting Methods in Ethiopia . . . . . . . . . . 73 3 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 4 Result and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99 5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 6 Recommendation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 Effects of Initial Moisture Content and Storage Duration on Physical and Chemical Characteristics of Stored Maize (Zea mays L.) Grain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Habtamu Gebremichael Daba, Mulugeta Admasu Delele, Solomon Workneh Fanta, Nigus Gabbiye Habtu, Metadel Kassahun Abera, and Admasu Fanta Worku 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Experimental Analysis of the Effect of Exposure of an Electro-optic System to External Magnetic Field (Case Study: The Input Laser and the Output Detector Are Separately and Concurrently Exposed) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Alema Abraha and Getachew Alemu 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Magnetic Field Effect on Electro-optical Devices . . . . . . . . . . . . . . . . . 3 Experimental Setup and Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

109

110 110 112 116 116

119 119 121 122 124 130 131

Contents

Identification of Dominant Lactic Acid Bacteria and Yeast Species from Teff Injera Dough Fermentation . . . . . . . . . . . . . . . . . . . . . . . . . . Zinash Tadesse Bonger, Metadel Kassahun Abera, Takele Ayanaw Habitu, Agimassie Agazie Abera, Mesfin Wogayehu Tenagashaw, Deginet Teferi, Abebaw Teshome, Taddele Andarge, Sadik Jemal Awol, and Tadesse Fenta Yehuala 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Results and Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Performance Evaluation of Grain Pro Collapsible Natural Convection Solar Dryer for Maize Grain Drying in Ethiopia . . . . . . . . . Messenbet Geremew Kassa, Nigus Gabbiye Habtu, and Aynadis Molla Asemu 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Quality Assurance Management Practices in Public Building Construction Projects: The Case of Bahir Dar City . . . . . . . . . . . . . . . . Abrham Zegeye Getie, Bahiru Bewket Mitikie, and Solomon Melaku Belay 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Research Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Result and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Review on Water Automatic Teller Machine (Water ATM) Technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Talaksew Misganaw Agegnehu and Getnet Ayele Kebede 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Yeast Bioremediation of Cr (VI) from Tannery Wastewater . . . . . . . . . . Tessafa Abrham Ashagrie, Shegaw Ahmed, Asmare Tezera Admasie, and Mequanint Demeke Aynalem 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

ix

133

134 135 140 154 156 163

164 165 170 175 176 177

178 180 182 190 190 193 193 194 200 201 203

204 205 208 214 214

x

Contents

Production and Characterization of Ink from the Milky Fluid of Ficus vasta and Euphorbia abyssinica . . . . . . . . . . . . . . . . . . . . . . . . . Surafel Argahegn Agdew, Mequanint Demeke Aynalem, Yemsrach Mintesnot Melaku, Belete Adane Mandie, and Asmare Tezara Admassie 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Assessment of the Causes of Non-revenue Water in Urban Water Distribution Systems: The Case of Bahir Dar City, Ethiopia . . . . . . . . . . Yibeltal Fentahun Aycheh, Dagnachew Aklog Yihun, Chalachew Mulat Alemu, and Fasikaw Atanaw Zimale 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Result and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Characteristics of Supply Chain Integration of Manufacturing Firms in Ethiopia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tenaw Tegbar, Thoben Klaus, D. K. Nageswara Rao, and Bereket Haile 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Research Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Conclusion and Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Effect of Threshing and Storage Conditions on Mold Contamination of Stored Maize Grain (Zea mays L.) . . . . . . . . . . . . . . . Messenbet Geremew Kassa, Nigus Gabbiye Habtu, Admasu Fanta Worku, Biresaw Demelash Abera, and Aynadis Molla Asemu 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

217

218 218 222 226 226 229

230 231 235 247 248 249

249 252 254 254 257 258 261

262 263 268 278 279 279

Contents

Formulation and Evaluation of Firefly and Artificial Bee Colony Algorithms for Maximum Power Extraction of Photovoltaic Systems Under Partial Shade . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tefera T. Yetayew and Tigist D. Wudmatas 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Review of Related Researches in Maximum Power Extraction of PV Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 PV Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Partial Shagging Maximum Power Point Tracking . . . . . . . . . . . . . . . . . 5 Performance Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Impact of Parametric Uncertainties on the Steady-State Performances of DFIG-Based WECS . . . . . . . . . . . . . . . . . . . . . . . . . . . Endalew Ayenew, Getachew Biru, Asrat Mulatu, and Milkias Berhanu 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Parametric Uncertainty in DFIG Model Formulation . . . . . . . . . . . . . . . 3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Participatory Evaluation and Demonstration of Handheld Maize Sheller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Geta Kidanemariam, Worku Biweta, Seyfe Yilma, and Tekelgiorgis Mamuye 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Material and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Suggestion from the Operators (Farmers) . . . . . . . . . . . . . . . . . . . . . . . 5 The Economic Data and Cost Benefit Analysis . . . . . . . . . . . . . . . . . . . 6 Conclusion and Recommendation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

xi

283 283 284 286 288 291 295 297 299

299 301 308 316 317 319

319 321 322 325 325 326 327

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 329

Application of FACTS Devices for Transient Stability Enhancement in the Adama II Grid-Connected Wind Farm Yosef Berhan Jember, Tewodros Gera Workineh, Tefera T. Yetayew, and Meresa Dinkayow

Contents 1 2

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Literature Review and Basic Theories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1 Wind Turbines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Wind Farm Aggregation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3 General Assumptions for Modeling Wind Farms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1 Minimization of Voltage Deviation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 System Loss Reduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 Constraints of the Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4 Transformer Tap Setting Constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5 Case Study Test System Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.6 Data Collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.7 Data Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.8 Model of Adama II Wind Farm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.9 Model of Adama II Wind Farm Integrated to Modified IEEE 14 Bus Test Systems 3.10 Optimal Placement of FACTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1 Test System Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Test System Analysis with UPFC Placement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 Test System Analysis with IPFC Placement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4 Voltage Stability, Transient Stability, and Loss Reduction with FACTS Devices . . . 5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2 3 5 6 7 7 7 9 9 10 10 11 11 12 13 13 15 15 19 21 24 25 26

Y. B. Jember (✉) · T. G. Workineh Bahir Dar, Ethiopia T. T. Yetayew · M. Dinkayow Adama Science and Technology University, Adama, Ethiopia © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 K. Mekuanint et al. (eds.), Advancement of Science and Technology in Sustainable Manufacturing and Process Engineering, Green Energy and Technology, https://doi.org/10.1007/978-3-031-41173-1_1

1

2

Y. B. Jember et al.

1 Introduction Recently the power system network is referred as an integrated power system because all sectors are getting connected one another. Nowadays renewable energy are getting utilized as distributed generation or by integrating them with the main grid. Among the nonconventional sources of power, wind is becoming the most efficient and feasible type to be connected with the grid. A variety of disruptions are produced due to the integration of wind power or other sources into an existing grid. Some of which may have negative impacts on the network, such as blackouts or a loss of generator synchronism. These might have put the power system under more stress than it could handle [1]. It raises concerns about the power system’s transient stability. The issue of transient stability will arise if some generators are operating far from the load centers, which might pose a risk to the security of the supply and cause difficulties for grid operators in running the power system on a daily basis. Transient stability is the ability of a transmission line to withstand rapid changes in the environment of the power network, such as a three-phase fault or the sudden damage to large generating or load units, without losing stability. If no action is taken to stop the system oscillations in their nominal setting, they may disrupt the flow of electricity and may even cause the generators to become out of sync, which might cause a complete or partial system failure. The FACTS devices might be a way to provide this purpose without having to deal with the slowness and wear of electromechanical devices. By adjusting the characteristics of their series impedance, shunt impedance, current, voltage, and phase angle, FACTS can increase the stability of a network, such as the transient and small signal stability, and lower the flow of heavily loaded lines and support voltages. Controlling the power flows in the network results in a decrease in the flow of heavily loaded lines, an improvement in the system’s capacity to handle the load, a reduction in system loss, and an increase in system security [2–5]. The electrical grid can be viewed as a massive, interconnected, nonlinear system with a huge number of weakly dampened electromechanical oscillations. Longdistance power transfers are steadily rising as the power industry moves toward deregulation, surpassing the development of new transmission facilities and resulting in more weakly damped inter-area oscillations. When synchronous machines are disturbed in stable power systems, synchronism either returns to its initial state if there is no net change in power or reaches a new state without losing it, but synchronism is lost when there is a net change in power [6]. The operation of the system may be severely constrained if the damping of these modes decreases or turns negative. As a result, it is crucial to be able to identify their nature, locate their limits of stability, and, in many situations, utilize controls to stop their instability. This study deals with the application of FACTS for improving transient stability in a grid-connected wind farm called Adama II Wind Farm, installed around Adama city, Ethiopia.

Application of FACTS Devices for Transient Stability Enhancement. . .

3

2 Literature Review and Basic Theories FACTS controllers are power electronic devices that enhance power system operation through their control attributes and injection modes [7]. The devices are mainly grouped as: (a) Static synchronous series compensators, thyristor-controlled phase angle regulators, and thyristor-controlled series compensators (SSSC). (b) Static var compensator (SVC) and static synchronous compensator are two examples of shunt controllers (STATCOM). (c) Controllers that mix series-to-series and series-to-shunt flow, such as the interline power flow controller (IPFC) and the unified power flow controller (UPFC). A detail explanation of FACTS and their exclusive function is provided in Table 1. According to [8], power electronics rooted in FACTS devices can optimize power system problems. The dilemma of where to place flexible AC transmission system (FACTS) devices optimally is regarded as a challenge. Solving FACTS location problems typically involves the use of the metaheuristic methods. Following are the summary of some literature related to the use of FACTS in wind energy systems, and their impact on settling power system disturbances. Some review works are included to show prior studies conducted about the stability issues in the case study system, Adama II wind farm. According to the research in [9], the FACTS in a power system play a critical role in strengthening the stability, lowering losses, and lowering the cost of generation, as well as enhancing the system loading capacity by rerouting the power flow in the network. The location and type of the devices are taken into consideration in the proposed work using the sensitivity of the system loading factor, which corresponds to the real and reactive power balance equations with respect to the control parameters of FACTS, technique plus N-1 contingency criterion, and some considerations that fit with the formation of the network. Additionally, the nonlinear predictorcorrector primal-dual interior-point optimal power flow (OPF) algorithm is used to evaluate and identify the ideal position and parameters. The optimization findings show that placing FACTS devices in the proper location can improve loading capacity, or reduce losses in the system, and that the N-1 contingency criterion and Table 1 Comparison among FACTS controllers Name SVC SSSC STATCOM TCSC TCPAR UPFC IPFC

Type Shunt Series Shunt Series Series-shunt Series-shunt Series-series

Controller used Thyristor GTO GTO Thyristor Thyristor GTO GTO

Purpose Voltage control Power flow control Voltage control Power flow control Power flow control Voltage and power flow control Voltage and power flow control

4

Y. B. Jember et al.

the sensitivity of the system loading factor can be used to choose the optimal site for the FACTS devices. Static synchronous series compensator used in [10] is suggested to increase the voltage stability of three-phase wind farms. It is simulated in MATLAB/SIMULINK platform and addresses the function of SSSC in enhancing the voltage stability of grid-connected wind energy conversion system during three-phase grid fault. In [11], the optimal functioning of wind farms and FACTS devices is outlined in as a technique for lowering system hazards. Both a unified power flow controller (UPFC) and a thyristor-controlled series compensator (TCSC) have been taken into consideration for varying the temperature limit of transmission lines. The effects of the wind farm and the combination of the wind farm and FACTS devices on system economy were examined. Environments that are both regulated and unregulated have been selected to test the suggested strategy. Calculations of value at risk (VaR) and cumulative value at risk (CVaR) were used to assess system risk. Modified IEEE 14 bus and modified IEEE 30 bus systems were used for the task. In order to compare the effects of renewable integration in the regulated and deregulated power systems in terms of system risk and operating cost, a comparative study using the Artificial Gorilla Troops Optimizer Algorithm (AGTO), Honey Badger Algorithm (HBA), and Sequential Quadratic Programming (SQP) was conducted. The ideal placement and ratings for the FACTS devices were found by taking the system’s minimal generation cost into account. The inclusion of TCSC and UPSC when using a wind farm lowers the economic parameters of the system by lowering the system risk, as is clear from the data. In [12], static var compensator (SVC) and static synchronous compensator (STATCOM) research on wind power systems has been done by taking into account their broad range of applicability within techno-economic constraints. Nearly all of the regions where SVC and STATCOM have shown positive remedies in wind farms have been mentioned. The grid integration faces a variety of difficulties, including voltage dip, dynamic performance, stability, and fault ride-through capabilities, among other significant issues. In these circumstances, STATCOM and SVC’s proposed solutions have been succinctly presented. In order to examine its impact on the power system’s dynamic voltage stability, the voltage management capability of the wind farm was applied to Adama II’s wind farm during a short circuit fault at the Point of Common Coupling (PCC) [13]. Two cases are used in order to assess the consequences. These simulations run on the terminal voltage of wind generators that are both equipped with and without voltage controllers. The Adama II wind farm’s FTR (full turbine representation) is combined with STR (single turbine representation), and the generated model’s accuracy has been confirmed. Investigations were conducted into the wind farm’s active and reactive power characteristics as well as fault ride-through capability and voltage regulation at the PCC. The findings demonstrate that the wind farm’s integration into the electrical grid satisfies wind grid code standards. Different scenarios have been examined in the study by [14] that focus on the effect of integrating wind power on the performance of a power system’s transient stability. Utilizing simulation software called DigSILENT, the analysis and

Application of FACTS Devices for Transient Stability Enhancement. . .

5

modeling were carried out using information from the Adama II wind farm and Ethiopian Electric Utility (EEP). The Adama II wind farm’s 102 wind turbines are combined into a single turbine representation, and the model’s accuracy has been confirmed. Analysis of the transient stability impact has been done for a variety of factors, including generator technology and fault location. As a result, the critical fault clearance time (CCT) of the power system is positively impacted by the integration of wind power. However, this study aims to take things a step further by using FACTS devices connected at the proper points to introduce a three-phase fault that has an impact on a number of parameters in the network system, in addition to the fast control of active and reactive power in the power system, improve the power system transient stability, and dampen power system oscillation of Adama II wind farm integrated into the modified IEEE 14 bus test system. Numerous parameters in the proposed power network with and without FACTS devices have been researched for their effects on the improvement of the power system transient stability and the damping of power oscillations. Furthermore, this study’s goal is to examine the Adama II wind farm’s transient behavior while it is connected to the national grid. MATLAB/PSAT software is used to model and simulate the wind farm. A power flow study is performed by connecting the wind farm with the grid, which is represented by an IEEE 14 bus test system. The power flow analysis is followed by the placement and sizing of FACTS. The connected system is then simulated in order to see the improvements using IPFC and UPFC. The simulation is then run again with transient conditions and fault occurrences, and the outcomes are compared.

2.1

Wind Turbines

A wind turbine (WT) is a machine that captures air kinetic energy and feeds it to a generator to produce electricity. The wind turbine’s axis might be vertical or horizontal. While the blades of vertical axis wind turbines rotate on an axis perpendicular to the ground, those of horizontal axis wind turbines rotate on an axis parallel to the ground. In [5], it asserts that there are five different categories under which wind turbines can be divided: (a) Type 1: fixed speed WT with a step-up transformer-connected squirrel cage induction generator that is directly connected to the grid. (b) Type 2: wound rotor induction generator with variable speed limitation WT. (c) Type 3 is a variable speed WT with partial power electronic conversion and a doubly fed induction generator (DFIG). (d) Type 4: Variable speed WT with full power electronic conversion, which allows for the use of either induction or synchronous generators. (e) Type 5: variable speed with a mechanical torque converter connecting a synchronous generator to a low-speed and high-speed shaft.

6

Y. B. Jember et al.

Fig. 1 Schematic diagram of a DFIG-equipped WT

Because of the benefits listed below, DFIG is commonly used today: (i) It can operate at different wind speeds (around synchronous speed by 33%). Due to this, power can be generated even in windy conditions. (ii) Because the inverter uses only around 30% of the total power, it is less expensive. (iii) It has the ability to regulate the amount of reactive power that is delivered or absorbed from the grid, thereby regulating the power factor. (iv) Compared to traditional induction generators, it is more efficient. A DFIG-equipped WT’s schematic is shown in Fig. 1. The grid is directly coupled to the stator. Through slip rings and a back-to-back voltage source converter (VSC), the rotor windings are connected to the grid.

2.2

Wind Farm Aggregation

A wind power plant should be modeled to be as accurate as possible. However, it is impractical to represent hundreds of turbines and their associated branches [14]. It employs a condensed equivalent representation. A wind power plant is made up of numerous individual wind turbine generators (WTGs) that are connected to a medium-voltage collector system and, at the interconnection point, to the transmission system. Nameplate ratings for contemporary utility-scale WTGs range from 1 to 4 MW, and the terminal voltage is roughly 600 V. Each WTG is connected to a medium-voltage collector system that runs

Application of FACTS Devices for Transient Stability Enhancement. . .

7

between 12 and 34.5 kV by a step-up transformer, which is typically a pad-mounted device. One or more feeders that are connected to one another at a collector system station make up the collector system. The transmission system voltage is attained at the collection system station using one or more station transformers. An interconnection transmission line is required unless the collector system station is close to the connecting site. At the collector system station, reactive compensation may be provided in the form of mechanically switched capacitors and continuously variable equipment like STATCOMs or static var systems (SVS). Reactive compensation quantity and kind are determined by connectivity needs and collector system design factors, including voltage regulation and losses. A single turbine can be used to simulate the behavior of a whole wind farm (STR). The wind turbines are often connected to one another in a string (cluster) using subterranean cables, and the collector system is connected to a medium voltage of 12–34.3 kV voltage level. One or more groups of wind turbines could be present in each cluster. Each wind turbine has a unit transformer electrically connecting it to a medium voltage collector system, which increases the terminal voltage generated by the turbine. The central substation transformer is then linked to the collector system, which raises the medium voltage to the transmission voltage level [15].

2.3

General Assumptions for Modeling Wind Farms

The following assumptions will be taken into account while determining an equivalent circuit for the collector system [15, 16]. (a) It is assumed that the current injection from every wind turbine is the same size and angle. (b) The line capacitive shunt generates reactive power under the premise that the voltage at the buses is one per unit (1 p.u.). A typical topology and single turbine representation of a wind power plant are shown in Figs. 2 and 3, respectively.

3 Methodology 3.1

Minimization of Voltage Deviation

To prevent voltage collapses in problematic conditions, it is preferable to keep voltage variances within 5%. The voltages at the relevant buses may typically drop below 0.95 p.u. when the load requirements rise, necessitating the inclusion of additional voltage support at that specific bus. The voltage deviation will be reduced and the voltage stability will be improved by connecting the FACTS device

8

Y. B. Jember et al.

Fig. 2 Generic wind power plant topology

Fig. 3 Single turbine representation of a wind power plant (STR) [10–12]

to the proper bus and lines. The summing of voltage deviation must be kept to a minimum in order to have a consistent voltage profile. To reduce voltage deviation in this study, the voltage stability index is used as a criterion. The net voltage divergence of each bus in the network from unity is referred to as the voltage stability index (VSI) of the network. Mathematically, VSI in an n-bus system is stated as [9]: n

VSI =

j1 - V i j i=0

where Vi is voltage magnitude of ith bus.

ð1Þ

Application of FACTS Devices for Transient Stability Enhancement. . .

3.2

9

System Loss Reduction

Reactive power in a network is redistributed mostly as a result of transmission loss. Therefore, reducing transmission losses affects the actual power that the slack bus generates. The following is the mathematical formula for determining network losses [9]: nl

ploss =

Gl V 2i þ v2j - 2vi vj cos θij

ð2Þ

l=1

where Ploss is real power loss, Gl represents conductance of lth transmission line, nl represents the number of transmission line, Vi and Vj denote voltages at bus i and j, and θij denotes power angle.

3.3

Constraints of the Problem

The aforementioned optimization problem is subjected to a number of constraints, including restrictions on the FACTS device and the load flow, voltage, and reactive power generation limits. These restrictions can all be further broken down into equality and inequality restrictions. Equality Constraints The basic load flow equations are the equality constraints, and they are provided below [4, 9]. N

Pgi - Pdi =

jV i V j Y ij cos δi - δj þ θij j

ð3Þ

N

Qgi - Qdi =

jV i V j Y ij sin δi - δj þ θij j

where: Vi is the voltage magnitude at bus i, Vj is the voltage magnitude at bus j, Yij is the magnitude of the admittance of the line from bus i to bus, N is the number of buses. Pgi, Qgi are the real and reactive power generated at bus number i Pdi, Qdi are the real and reactive power demand at bus number i θi, θj are the bus voltage angle of the buses i and j respectively θij is the bus admittance angle of line connected between i and j

10

Y. B. Jember et al.

Inequality Constraints The operational boundaries for the system’s line flow (active and reactive power), voltages (generators and load buses), tap ratio (T ), FACTS parameter bounds, apparent power of transmission lines, and transformers are all governed by inequality constraints, and they can be represented as follows: additionally, they impose restrictions on the governing factors [9]. max Pmin gi ≤ Pgi ≤ Pgi max Qmin gi ≤ Qgi ≤ Qgi max V min gi ≤ V gi ≤ V gi

ð4Þ

V min ≤ V i ≤ V max i i max T min ij ≤ T ij ≤ T ij

where 0.9 ≤ Ti ≤ 1.0 and 0.8 < Vi < 1.2

3.4

Transformer Tap Setting Constraints

In an automatic transformer, the minimum tap setting should be 0 and the maximum tap setting should be 1. Similar to this, if tappings are present on a two-winding transformer’s secondary side, then 0 t n, where n is the ratio of transformation.

3.5

Case Study Test System Architecture

In order to be able to generalize the results of this study to all other networks, a combined interconnection of the Adama II wind farm and a modified dynamic IEEE 14 bus test system were chosen. The modified IEEE 14 bus system, which has 14 buses, 5 generators with AVR (Automatic Voltage Regulator), 11 loads, 3 tap-changing transformers, and 20 branches, is integrated with the wind farm and has a huge load. Bus number 3 has a larger load than the other buses. The 102 wind turbines in the wind farm are divided into 8 clusters, with 1 cluster having 11 turbines and the other 7 having 13 turbines each. The wind turbines are connected by 33 kV underground cables and overhead transmission lines, which vary in length and capacity depending on where each unit is located and how far away it is from the 33 kV collector bus. The entire test system is made up of a group of Adama II wind turbines that have been integrated into a modified IEEE 14 bus system at bus number 3. Of the four generators, two produce real power, while the other two, synchronous

Application of FACTS Devices for Transient Stability Enhancement. . .

11

compensators, and are only used to produce reactive power. The voltages are controlled by each generator’s AVR, with generator 1 serving as the reference generator. The device has an IEEE model-I, Type-II AVR. AVR is utilized to reduce electromechanical oscillations and enhance the system’s transient stability.

3.6

Data Collection

All of the information required for the paper study has been gathered. The power system parameters of the modified IEEE 14 bus test system, including all generator parameters, transmission line parameters, transformer parameters, and Adama II wind farm turbine parameters data, were collected. Figure 5 shows how this system is topologically organized. The Ethiopian Electric Utility’s Adama district has the static and dynamic data for the Adama II wind farm. The following is the Adama II wind farm’s integrated test system based on modified IEEE14: (i) 100 MVA is the apparent power. (ii) 50 Hz is the base frequency. (iii) The voltage bases for buses 01 through 05 are 33 kV, for buses 06 through 14 are 13.8 kV, and for bus 08 are 18 kV. Nominal ratings of devices are equivalent to system bases unless otherwise stated. Regarding the nominal ratings of the device, parameters are in per unit (p.u.).

3.7

Data Analysis

Based on the data gathered, the nature of the modified IEEE 14 bus test system integrated with the Adama II wind farm was analyzed. Under various conditions, its nature was examined and interpreted. The sources of transient stability are determined using the literature, and solutions to this issue are proposed. The power grid under examination in this study is a combined case study that looks at the connecting of a sizable wind farm to a modified IEEE 14 bus test system and the use of FACTS devices to reduce system instability. The system has a total real load of 259 MW and a reactive load of 81.4 Mvar. The system generates 272.58 MW and 108.89 MVAr in total in terms of actual and reactive power. System losses total 13.58 MW and 27.49 Mvar. Bus 01 serves as the slack bus and the reference angle for power flow analysis. Vmax = 1.20 p.u. and Vmin = 0.8 p.u. are utilized as voltage security limits for optimal power flow analysis. The primary and secondary sources’ data are combined, and transformed into MATLAB/PSAT model.

Y. B. Jember et al.

12

3.8

Model of Adama II Wind Farm

At an elevation of 1741–2173 m above sea level, the Adama II wind farm is situated in the heart of Ethiopia, around 90 km east of the capital Addis Ababa and 3 km North of Adama city. The feasibility study states that the wind park’s geographic center is located at 39° 12′ 10″E and 8° 34′ 18″N, and the area’s typical wind speed is between 8 and 9 m/s. The DFIG type has been employed in the farm. The wind farm’s total installed capacity is 153 MW, and each wind turbine has a rated voltage of 690 V. Each wind turbine is connected to a unit transformer that raises the generator’s terminal voltage (690 V) into the 33 kV level of medium voltage. Eight clusters made up of all 102 turbines are connected to the main substation via a 33 kV overhead wire. One cluster contains 11 turbines, while the other 7 clusters each include 13 turbines. Two transformers with a combined capacity of 90 MVA and a voltage level of 33/230 kV are present at the main substation. Figure 4 depicts the Adama II wind farm’s layout. A single 230 kV overhead transmission line connects the high-voltage terminal side to the Koka substation. The wind power plant is aggregated into a modified IEEE 14 bus system in this study.

Fig. 4 Layout of the Adama II wind farm

Application of FACTS Devices for Transient Stability Enhancement. . .

3.9

13

Model of Adama II Wind Farm Integrated to Modified IEEE 14 Bus Test Systems

Loss reduction, voltage profile improvement, and transient stability enhancement studies for the aggregated power network have been carried out in this chapter. A dynamic time domain simulation (TDS) was carried out to verify the impacts of the implemented FACTS controllers. Sub-transient models with AVRs were used to represent the generators at buses 01 and 02, as well as the synchronous condensers at buses 06 and 08. Between Bus 01 and Bus 14, a fault is added to provide a temporary instability. Buses 04 and 02 each experience a transient three-phase fault as a result of a sudden load change. Since system failures might happen anywhere in the system, the faulty bus is chosen at random. There may be issues with additional buses or generation units. Figure 3 displays a single-line diagram of the Adama II wind farm combined with a modified IEEE14 bus system. All generators are operated with constant mechanical input power and constant excitation, and two-dimensional loads are assumed to have constant impedance. The weakest bus is the one that is most likely to experience a voltage fall or experience a failure. A MATLAB/PSAT model of the IEEE-14 bus system is presented here, with a base MVA of 100 and a base voltage of 33 kV. Rotor angle, rotor speed, q-axis transient voltage, d-axis transient voltage, q-axis sub-transient voltage, and d-axis sub-transient voltage are the six state variables for the system. Two different generator types are included in the suggested test system: synchronous generators (on the grid side) and a single turbine that represents DFIGs from the Adama II wind farm. The synchronous generator does not run at full capacity during routine operations in order to account for the system’s power reserve. Synchronous generators react to system disruptions right away, whereas wind turbines take longer time to respond because of their intricate controls. As a result, additional systems, such as FACTS devices, are needed to help keep the electrical grid stable during and after a disturbance. Wind turbines can collectively have a big impact on a power system during a severe disruption like a nearby fault, even when individual turbines have little effect on the larger power system network. A reduced equivalent representation is necessary since it is not practical to simulate each unique wind turbine. For this study, the Adama II wind farm’s 102 wind turbines are combined into a single turbine. Figure 5 depicts the test system that is connected to a DFIG-based Adama II wind farm and contains a one-line design for a modified IEEE 14 bus test system.

3.10

Optimal Placement of FACTS

The lines carrying the highest active power line losses are identified for best FACTS device placement, and these lines’ beginning buses are chosen because of their voltage magnitude and related phase angle. Both voltage enhancement and an

14

Y. B. Jember et al.

Fig. 5 Adama II wind farm integrated to modified dynamic IEEE 14 bus test systems in MATLAB/ PSAT

increase in transmission line power flow capacity are achieved by using SSSC, IPFC, and UPFC. FACTS are installed at weak lines and weak buses to prevent transmission line overloading. This is primarily due to their ability to alter the total flow of reactive power in these lines. All of the system’s lines (aside from the line carrying the transformer) are chosen as potential sites for the placement of the IPFC and UPFC. The best location for UPFC and IPFC is found by selecting the transmission lines with the highest overall line losses. Every bus’s voltage profile as well as each line’s power flow has been obtained. Dynamic Model of Adama II Wind Farm Integrated to Modified IEEE 14 Bus Test System with UPFC UPFC is a combination series-shunt controller that can be set up in one of two ways: either as two independent series (SSSC) or as shunt (STATCOM) controllers that work as interconnected series and shunt components. Real power can be transferred between these two pieces via the power link once they have been united. These devices in this research have various compensation regimes, one of which is shunt and the other is series, hence the interconnected injection model has been used. According to Fig. 6, UPFC was attached to the lines with the highest line losses: Dynamic Model of Adama II Wind Farm Integrated to Modified IEEE 14 Bus Test System with IPFC The effectiveness of IPFC for enhancing transient stability was further evaluated using the dynamic model with an independent setup of the IPFC FACTS device. In Fig. 7, the MATLAB/PSAT Simulink configuration is displayed.

Application of FACTS Devices for Transient Stability Enhancement. . .

15

Fig. 6 Dynamic model of Adama II wind farm integrated to modified IEEE 14 bus test systems modeling in MATLAB/PSAT with fault and UPFC

Fig. 7 Dynamic model of Adama II wind farm integrated to modified IEEE 14 bus test systems modeling in MATLAB/PSAT with fault and IPFC

4 Results and Discussion 4.1

Test System Analysis

Case 1: Under Base Case Condition The Adama II wind farm’s traditional power flow analysis was completed using the Newton-Raphson (NR) method and merged with a modified IEEE 14 bus system in MATALB/PSAT. The primary benefit of the NR approach is its dependability in

16

Y. B. Jember et al.

Fig. 8 Dynamic model of Adama II wind farm integrated to modified IEEE 14 bus test systems modeling in MATLAB/PSAT under base case condition

terms of convergence. The voltage magnitude profiles were shown and examined after the power flow was simulated to identify the weakest bus or the greatest line losses. To choose the best location for FACTS devices, consider the bus with the lowest voltage profile magnitude and the lines with the biggest real power losses. Figure 8 shows the PSAT model of the test system in this case. The equivalent values of these system characteristics under base case conditions must be established in order to evaluate the effects of FACTS devices on system power loss, transient stability, and voltage deviation. According to Table 2, the line connecting buses 01 and 02, which was measured as 0.0431 p.u., and 04, which was measured as 1.0120 p.u. correspondingly, has the highest transmission line loss and the weakest bus voltage with a heavy load. Adama II wind farm’s modified IEEE 14 bus system is integrated with voltages that are limited by varying phase angles. Bus 01 is a slack bus, so its voltage and angle are both 1.06 p.u. At initial lambda, reactive power is 108.89 MVAr, whereas active power is 272.58 MW demand. As a result, the system is operating within its parameters and no unstable conditions are present. If load parameters change, power flow outcomes also alter. It may be inferred from Table 2 and Fig. 9 that bus number 04 is the least secure bus because its voltage is the lowest at each reactive load. As a result, this is the ideal place for the TDS STATCOM devices. It is evident from Table 3 that lines 11, 12, and 14 in the system experience the most active power loss. These lines are candidates for SSSC, IPFC, and UPFC devices because they connect to buses 01–02, 02–03, and 01–05, respectively. According to Table 3, the lines from bus 01 to bus 02, bus 02 to bus 03, and bus 01 to bus 05 had the largest line active power losses, with values of 0.0431 p.u.,

Application of FACTS Devices for Transient Stability Enhancement. . .

17

Table 2 Power flow results of IEEE 14 bus system for base case condition using power flow in MATLAB/PSAT Power Flow Results V [p.u.] Bus Bus 01 1.0600 Bus 02 1.0450 Bus 03 1.0100 Bus 04 1.0120 Bus 05 1.0160 Bus 06 1.0700 Bus 07 1.0493 Bus 08 1.0900 Bus 09 1.0328 Bus 10 1.0318 Bus 11 1.0471 Bus 12 1.0534 Bus 13 1.0470 Bus 14 1.0207

Phase [rad] 0.0000 -0.0871 -0.2227 -0.1785 -0.1527 -0.2516 -0.2309 -0.2309 -0.2585 -0.2622 -0.2590 -0.2664 -0.2671 -0.2802

P gen [p.u.] 2.3258 0.4000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

Q gen [p.u.] -0.1498 0.4882 0.2737 0.0000 0.0000 0.2251 0.0000 0.2516 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

P load [p.u.] 0.0000 0.2170 0.9420 0.4780 0.0760 0.1120 0.0000 0.0000 0.2950 0.0900 0.0350 0.0610 0.1350 0.1490

Q load [p.u.] 0.0000 0.1270 0.1900 0.0400 0.0160 0.0750 0.0000 0.0000 0.1660 0.0580 0.0180 0.0160 0.0580 0.0500

2

Fig. 9 Voltage profile magnitude for IEEE 14 bus system for base case condition

0.0234 p.u., and 0.0277 p.u., respectively. As a result, the line between buses 01 and 02 has the greatest line-active power loss. For TDS, the three FACTS devices (SSSC, IPFC, and UPFC) are best placed on this line. Case 2: Under Fault Condition Without FACTS Devices The dynamic IEEE 14 Bus Test System model in the PSAT toolbox has been used to inject a disturbance in the system. The fault time occurred at 1.00 s and was cleared at 1.25 s without the placement of FACTS devices. The test system PSAT simulation architecture is given in Fig. 10.

18

Y. B. Jember et al.

Table 3 Line flow results of IEEE 14 bus system for base case condition using power flow in MATLAB/PSAT Line Flows From bus Bus 02 Bus 06 Bus 12 Bus 06 Bus 06 Bus 11 Bus 09 Bus 09 Bus 14 Bus 07 Bus 01 Bus 03 Bus 03 Bus 01 Bus 05 Bus 02 Bus 05 Bus 04 Bus 04 Bus 08

To bus Bus 05 Bus 12 Bus 13 Bus 13 Bus 11 Bus 10 Bus 10 Bus 14 Bus 13 Bus 09 Bus 02 Bus 02 Bus 04 Bus 05 Bus 04 Bus 04 Bus 06 Bus 09 Bus 07 Bus 07

Line 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

P flow [p.u.] 0.4171 0.0804 0.0186 0.1827 0.0818 0.0456 0.0449 0.0872 -0.0627 0.2720 1.5712 -0.7112 -0.2308 0.7546 0.6019 0.5594 0.4569 0.1550 0.2720 0.0000

Q flow [p.u.] 0.0322 0.0312 0.0135 0.0974 0.0844 0.0640 -0.0047 0.0060 -0.0460 0.1619 -0.2046 0.0168 0.0669 0.0548 -0.0921 0.0159 0.1097 0.0280 -0.0651 0.2516

P loss [p.u.] 0.0092 0.0008 0.0001 0.0025 0.0011 0.0005 0.0001 0.0009 0.0010 0.0000 0.0431 0.0234 0.0040 0.0277 0.0048 0.0167 0.0000 0.0000 0.0000 0.0000

Fig. 10 MATLAB/PSAT model of IEEE 14 bus test system under faulty condition

Q loss [p.u.] -0.0080 0.0017 0.0001 0.0049 0.0024 0.0011 0.0002 0.0019 0.0020 0.0100 0.0731 0.0522 -0.0252 0.0613 0.0019 0.0112 0.0468 0.0127 0.0153 0.0094

Application of FACTS Devices for Transient Stability Enhancement. . .

19

Case 3: Under Fault Condition with FACTS Devices Placement The Time Domain Simulation (TDS) tool, which is a component of the MATLAB® PSAT toolbox, was used to run simulations for transient stability enhancement (TSE). The simulations were initially run under fault-free, base-case conditions without the use of FACTS controllers. Later, simulations with two FACTS controllers were also performed, with the outcomes being recorded. The study’s time domain simulations (TDS) and analysis used the generator power, rotor angle, and speed parameters. TDS has been used to assess the FACTS device models’ performance in this work. The disturbance created was a three-phase fault and a quick load change, with fault time happening at 1.00 s and clearing time at 1.25 s, both with and without FACTS.

4.2

Test System Analysis with UPFC Placement

Generator 1 Rotor Angle and Rotor Speed Responses In the graphs from Figs. 11, 12, 13, and 14, the green curve represents the responses of the generator and its parameter variation under the base case condition, the red curve represents the responses of the generators parameters under faulty conditions, and the blue curve represents the responses of the generators parameters under faulty conditions with FACTS device placement. The test systems’ responses to the synchronous generator 1’s rotor angle and rotor speed in the time domain are depicted in Figs. 11 and 12, respectively. The oscillations of the rotor angle and rotor speed responses of synchronous generators 1 remain unstable after the fault is cleared at 1.25 s by opening the circuit breaker at bus 02 on line from bus 02 to bus

Fig. 11 Generator 1 rotor angle responses

20

Y. B. Jember et al.

Fig. 12 Generator 1 rotor speed responses

Fig. 13 Generator 3 rotor angle responses

04, as shown by the red curve in Figs. 11 and 12, respectively, and exceed the simulations’ ending time of 200 s without the use of UPFC. The UPFC FACTS significantly dampened the oscillation of rotor angle at a time of about 150 s, as shown by the blue curve in Fig. 11, to a steady state value of 0.6 rad from the initial rotor angle of 0.3 rad when the dynamic model, fault applied at bus 04, and sudden load increase with the UPFC placement are simulated. As indicated by the blue curve in Fig. 12, the UPFC FACTS similarly greatly reduced the oscillation of rotor speed over a period of roughly 150 s to a steady-state value of 1 p.u. from the initial rotor speed of 1 p.u.

Application of FACTS Devices for Transient Stability Enhancement. . .

21

Fig. 14 Generator 3 rotor speed responses

Generator 3 Rotor Angle and Rotor Speed Responses The test systems’ responses to the synchronous generator 3’s rotor angle and rotor speed in the time domain are depicted in Figs. 13 and 14, respectively. The oscillations of the rotor angle and rotor speed responses of synchronous generators 3 remain unstable after the fault is cleared at 1.25 s by opening the circuit breaker at bus 02 on line from bus 02 to bus 04, as shown by the red curve in Figs. 13 and 14, respectively, and exceed the simulations’ ending time of 200 s without the use of UPFC. The UPFC FACTS significantly dampened the oscillation of rotor angle at a time of about 150 s, as shown by the blue curve in Fig. 13, to a steady state value of -0.08 rad from the initial rotor angle of –0.3 rad when the dynamic model, fault applied at bus 04, and sudden load increase with the UPFC placement are simulated. The blue curve in Fig. 14 shows how the UPFC FACTS significantly reduced the oscillation of rotor speed over a period of about 100 seconds, from the initial value of 1 p.u. to a steady state value of 1 p.u.

4.3

Test System Analysis with IPFC Placement

Generator 1 Rotor Angle and Rotor Speed Responses Figures 15 and 16, respectively, depict the time domain simulation of the test systems’ reactions to the synchronous generator 1’s rotor angle and rotor speed. The oscillations of the rotor angle and rotor speed responses of synchronous generators 1 after fault clearance at 1.25 s by opening the circuit breaker at bus 02 on line from bus 02 to bus 04 remain unstable and go beyond the simulations’ ending time set at 200 s, as indicated by the red curve in Figs. 15 and 16,

22

Y. B. Jember et al.

Fig. 15 Generator 1 rotor angle responses

Fig. 16 Generator 1 rotor speed responses

respectively, when a fault is applied at bus 04 and with a sudden load increase, without the placement of IPFC. The IPFC FACTS significantly dampened the oscillation of rotor angle at a time of about 40 s, as shown by the blue curve in Fig. 15, to a steady state value of 1.7 rad from the initial rotor angle of 0.3 rad when the dynamic model, fault applied at bus 04, and sudden load increase with the IPFC

Application of FACTS Devices for Transient Stability Enhancement. . .

23

Fig. 17 Generators 3 rotor angle responses

Fig. 18 Generator 3 rotor speed responses

placement are simulated. The blue graph in Fig. 14 shows how the IPFC FACTS considerably reduced the oscillation of rotor speed over a period of roughly 40 s, from the initial value of 1 p.u. to a steady state value of 1 p.u.

Generator 3 Rotor Angle and Rotor Speed Responses The test systems’ responses to the synchronous generator 3’s rotor angle and rotor speed in the time domain are depicted in Figs. 17 and 18, respectively. The oscillations of the rotor angle and rotor speed responses of synchronous generators

24

Y. B. Jember et al.

3 after fault clearance at 1.25 s by opening the circuit breaker at bus 02 on line from bus 02 to bus 04 remain unstable and exceed the simulation ending time set at 200 s, as indicated by the red curve in Figs. 17 and 18, respectively, when no IPFC is installed and a fault is applied at bus 04 along with a sudden increase in load. The IPFC FACTS significantly dampened the oscillation of rotor angle at a time of about 40 s, as shown by the blue curve in Fig. 17, to a steady state value of 1.3 rad from the initial rotor angle of –0.3 rad when the dynamic model, fault applied at bus 04, and sudden load increase with the IPFC placement are simulated. The blue curve in Fig. 18 shows how the IPFC FACTS considerably reduced the oscillation of rotor speed over a period of roughly 40 s, from the initial value of 1 p.u. to a steady state value of 1 p.u.

4.4

Voltage Stability, Transient Stability, and Loss Reduction with FACTS Devices

Significant transient stability improvement is shown by the dynamic model of the Adama II wind farm connected to modified IEEE 14 bus test systems with IPFC and UPFC findings. As shown in Table 4, from power flow results, in terms of loss reduction and voltage profile improvement, IPFC FACTS devices provide better

Table 4 Voltage profile magnitude and real power losses from PF reports

Bus Bus 01 Bus 02 Bus 03 Bus 04 Bus 05 Bus 06 Bus 07 Bus 08 Bus 09 Bus 10 Bus 11 Bus 12 Bus 13 Bus 14 Real power losses

Without fault and FACTS devices V [p.u.] 1.0600 1.0450 1.0100 1.0120 1.0160 1.0700 1.0493 1.0900 1.0328 1.0318 1.0471 1.0534 1.0470 1.0207 0.1358

With fault and without FACTS devices V [p.u.] 1.0600 1.0450 1.0100 0.9939 0.9992 1.0700 1.0324 1.0900 1.0076 1.0069 1.0326 1.0443 1.0339 0.9907 0.3454

With fault and UPFC devices V [p.u.] 1.0600 1.0450 1.0100 0.9942 1.0003 1.0700 1.0326 1.0900 1.0080 1.0073 1.0328 1.0443 1.0340 0.9910 0.2117

With fault and IPFC devices V [p.u.] 1.0600 1.0450 1.0100 0.9960 1.0028 1.0700 1.0337 1.0900 1.0094 1.0086 1.0335 1.0444 1.0343 0.9919 0.0970

Application of FACTS Devices for Transient Stability Enhancement. . .

25

Table 5 Time domain simulation results

Generator parameters Generator 1 rotor angle Generator 1 rotor speed Generator 3 rotor angle Generator 3 rotor speed Generator 1 real power Generator 2 real power Generator 1 reactive power

With no FACTS devices Settling time Goes beyond the simulation ending time set Goes beyond the simulation ending time set Goes beyond the simulation ending time set Goes beyond the simulation ending time set Goes beyond the simulation ending time set Goes beyond the simulation ending time set Goes beyond the simulation ending time set

With UPFC devices Settling time 150 s

With IPFC devices Settling time 40 s

150 s

40 s

150 s

40 s

100 s

40 s

100 s

40 s

80 s

20 s

50 s

20 s

power loss reduction and voltage profile improvement than UPFC. The two FACTS devices decrease power losses in underperforming networks by 0.2485 p.u. with IPFC and 0.1337 p.u. with UPFC. In this case study, IPFC was more effective than UPFC for applications involving the increase of transient stability, as indicated in Table 5.

5 Conclusion The wind farm considered for this study, Adama II, had 102 wind turbines combined into a single turbine representation, and it was integrated with a modified IEEE 14 bus system. Through PF results in PSAT, the ideal placement of FACTS has been successfully implemented. The line between buses 01 and 02 in the Adama II wind farm’s modified IEEE 14 bus system was determined to be the best location for UPFC and IPFC FACTS devices. On the Adama II wind farm integrated into the modified IEEE 14 bus system model, time domain simulations have been run under idealized conditions, faulty conditions without and with FACTS devices, and under base case conditions. The investigations were carried out using FACTS devices and other particular and comparative studies. Significant transient stability improvement is observed with the addition of IPFC and UPFC. IPFC devices offer better power loss reduction and voltage profile improvement than UPFC in terms of loss reduction. The system oscillations of the multi-machine have been controlled and the oscillation transient period has been decreased as a result of the individual compensation properties of UPFC and IPFC devices. Compared to the two FACTS devices, IPFC was found to be more effective at damping oscillations of the multi-machine

26

Y. B. Jember et al.

system and enhancing transient stability with the shortest transient period. Compared to UPFC, the power loss reduction and voltage profile support offered by IPFC FACTS devices are superior to UPFC. The two FACTS devices decrease power losses in underperforming points by 0.2485 p.u. using IPFC and 0.1337 p.u. with UPFC.

References 1. Khan, N., Ahmad, H., Khattak, A.: Transient stability enhancement of power system using UPFC (Unified Power Flow Controller). Int. J. Eng. Works. 4(2), 33–40 (2017) 2. Panda, B.K., Boini, S.K.: Design of SSSC to improve power system stability with fuzzy logic controller. Int. J. Adv. Res. Eng. Technol. (IJARET). 3(2) (2014) 3. Siva Sankar, A., Anjaneyulu, K.S.R.: Maintaining voltage stability by optimal locating and sizing by combined evolutionary algorithm. Int. J. Comput. Appl. (0975–8887). 84(12) (2013) 4. Latt, A.Z.: Transient stability analysis of 3-machine, 7-bus system using ETAP. sTA. 1, 1 (2019) 5. Heetun, K.Z., Abdel Aleem, S.H.E., Zobaa, A.F.: Voltage stability analysis of grid-connected wind farms with FACTS: static and dynamic analysis. Energy Policy Res. 3(1), 1–12 (2016) 6. Sutter, J., Nderu, J., Muriithi, C.: Placement of FACTS devices for voltage profile improvement and loss reduction. Int. J. Emerg. Technol. Adv. Eng. 5(10) (2015) 7. Sutter, J.R., Nderu, J.N., Muriithi, C.: Power system oscillations damping and transient stability enhancement with application of SSSC FACTS devices. Eur. J. Adv. Eng. Technol. 2(11), 73–79 (2015) 8. Devi, H.A., Padma, S.: Power system security enhancement using optimal placement and parameter setting of multi-facts devices with bbo algorithm. Int. J. Pure Appl. Math. 118(5), 785–804 (2018) 9. Lubis, R.S., Hadi, S.P., Tumiran: Selection of suitable location of the FACTS devices for optimal power flow. Int. J. Electr. Comput. Sci. IJECS-IJENS. 12(3), 38–49 10. Sunil, A., Shahin, M.: Improving the voltage stability of power system connected with wind farm using SSSC. In: 2021 International Conference on Communication, Control and Information Sciences (ICCISc), vol. 1, pp. 1–6. IEEE (2021) 11. Das, A., Dawn, S., Gope, S., Ustun, T.S.: A strategy for system risk mitigation using FACTS devices in a wind incorporated competitive power system. Sustainability. 14(13), 8069 (2022) 12. Tiwari, V.K., Gupta, A.R.: Application of SVC and STATCOM for wind integrated power system. In: Recent Advances in Power Electronics and Drives, pp. 181–192. Springer, Singapore (2021) 13. Anchinesh, M.: Analysis of dynamic voltage stability on the penetration of Adama II wind farm in Ethiopian grid. Thesis, Addis Ababa University (2017) 14. Megra, M.: Impact of large scale wind power integration on Ethiopian power system transient stability. Doctoral dissertation, ASTU (2017) 15. Patterson, J.: Final project report western electric coordinating council wind (2014) 16. Panumpabi, M.P.: Large wind farm aggregation and model validation. Thesis, University of Illinois (2011)

Development of an Interpretable Deep Learning System for the Identification of Patients with Alzheimer’s Disease Selamawet Workalemahu Atnafu

and Stefano Diciotti

Contents 1 2

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1 Subjects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Data Pre-processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3 CNN Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4 CNN Visualization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

27 28 29 29 30 31 32 32 36 36

1 Introduction Alzheimer’s disease (AD) is the most commonly occurring neurodegenerative disorder [1] that causes memory impairment at its initial stage and advances to a cognitive decline that can affect behavior, speech, visuospatial orientation, and motor system [2]. Early diagnosis is essential to plan treatment strategies that could slow down the disease progression and enhance the quality of life S. W. Atnafu (✉) Department of Electrical, Electronic, and Information Engineering “Guglielmo Marconi”, University of Bologna, Bologna, Italy Bahir Dar Institute of Technology, Bahir Dar University, Bahir Dar, Ethiopia S. Diciotti Department of Electrical, Electronic, and Information Engineering “Guglielmo Marconi”, University of Bologna, Bologna, Italy © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 K. Mequanint et al. (eds.), Advancement of Science and Technology in Sustainable Manufacturing and Process Engineering, Green Energy and Technology, https://doi.org/10.1007/978-3-031-41173-1_2

27

28

S. W. Atnafu and S. Diciotti

[3]. Diagnosis of AD needs a physician’s follow-up of the patient’s medical history by performing clinical assessments and neuropsychological test scores [3]. Neuroimaging tools, such as structural MRI (sMRI), functional MRI, and positron emission tomography (PET), are also used to confirm that the cognitive decline caused by AD is altering the brain structure. In the past, traditional machine learning methods have been frequently used to analyze neuroimaging data. Designing a machine learning system becomes a very long process due to the need to extract hand-crafted features. On the contrary, deep learning – a family of machine learning methods that has the ability to automatically extract features from complex data [4] – overcomes the main limitations of traditional machine learning approaches and hence became the current state-of-the-art technology in medical imaging, including neuroimaging. Convolutional neural networks (CNNs) are a type of deep learning models tailored for image processing applications [4]. A basic CNN model consists of convolutional layers, pooling layers, and fully connected layers. Numerous studies have employed CNNs for classifying sMRI of AD patients versus healthy subjects [5–10]. Most of these studies used MRI data obtained from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) dataset, and other few studies [10–16] applied deep learning techniques on the Open Access Series of Imaging Studies (OASIS) collection of brain images. Apart from their success in many applications, deep learning approaches have been criticized for producing highly non-interpretable models [17]. Interpretability is a requirement in many applications in which crucial decisions are made by users relying on a model’s outputs, such as in medical applications [18]. CNN visualization methods help in understanding the reasoning behind the model’s decisions. Several recent neuroimaging studies have integrated explainability tools in their CNN models to classify different neurological disorders [6, 7, 19–25]. Regarding AD classification, a few studies [6, 7, 23] employed CNN visualization techniques to highlight the features used by the model to make decisions. These studies used a public brain dataset of AD and healthy individuals, namely the ADNI. In this study, we propose an interpretable CNN for classifying sMRI scans obtained from the public OASIS dataset. The CNN model is trained based on a transfer learning technique by utilizing the weights of a pre-trained VGG16 network. Unlike the previous deep learning studies classifying the OASIS collection of brain images, our proposed model includes a wide range of visualization methods to confirm that the models focus on clinically defined AD regions.

2 Materials In this section, the datasets used, the model architecture, training, and validation schemes, and finally, the CNN visualization methods applied to the trained model and their interpretation are discussed.

Development of an Interpretable Deep Learning System for. . . Table 1 Demographic features of the dataset used in this study

2.1

Number of subjects Age (range, years) Age (mean ± SD, years) Gender (women/men)

29 Patients 100 62 – 96 76.70 ± 7.10 59/41

Healthy controls 100 59 – 94 75.50 ± 9.10 73/27

Subjects

In this study, we used the OASIS publicly available dataset of AD patients and healthy control (HC) subjects [26] which is freely available at https://www.oasisbrains.org/. The dataset consists of a cross-sectional collection of MRI scans of 416 right-handed subjects aged between 18 and 96 years. The scans were acquired using a 1.5 T scanner. A 100 AD patients [(59 women and 41 men, age 76.70 ± 7.10 years, mean ± standard deviation (SD)] and 100 HC subjects (73 women and 27 men, age 75.50 ± 9.10 years, mean ± SD) who have been previously selected by other authors [27] are included in our experiment. The difference in age was not significant ( p = 0.15 at t-test) across the two groups while, considering gender, a significant difference ( p = 0.04 at χ2-test) was obtained. Table 1 lists the demographic information of the dataset used in this study. In the OASIS dataset, the global Clinical Dementia Rating (CDR) score was used for AD diagnosis, as well as to determine the severity of the disease. CDR score was derived from individual CDR scores for the domains memory, orientation, judgment and problem-solving, function in community affairs, home and hobbies, and personal case [28]. A global CDR score of 0 represents subjects with normal cognition, while scores of 0.5 (very mild), 1 (mild), 2 (moderate), and 3 (severe) have all been labeled as AD.

2.2

Data Pre-processing

The OASIS dataset publicly provides pre-processed data, where gain-field correction, brain masking, and atlas-based co-registration [29] were applied to the raw MRI images resulting in a data matrix size of 176 × 208 × 176 and a voxel size of 1 mm × 1 mm × 1 mm [29]. We performed 2D image processing on such partially pre-processed 3D MRI volumes. The 2D image processing involves slicing the MRI volumes into 2D grayscale images using an axial anatomical plane, and performing slice selection based on entropy values. We followed an entropy-based slice selection similar to our previous paper [11]. Second, data were split into training and validation sets based on a five-fold CV scheme. In this step, all the slices of a single MRI volume were included in either the training or validation sets (subject-level split) to prevent the problem of data leakage [11]. Lastly, feature scaling was applied to normalize the images in both the training and validation sets based on the training set feature statistics (using mean and SD values). From each MRI scan, 10 slices are

30

S. W. Atnafu and S. Diciotti

selected based on their entropy values [27], producing a total of 2000 (1000 AD and 1000 HC) grayscale images. As compared to the number of parameters for building a CNN model, the size of the image dataset is insufficient to effectively train a CNN from scratch. To prevent the overfitting of a CNN model due to limited training samples, we employed a transfer learning technique by starting from a pre-trained VGG16 model and fine-tuning the model parameters on the MRI dataset. Since VGG16 is trained on colored RGB images, the grayscale MRI slices were converted to three-channel images by repeating the 2D image onto the three channels. By applying these pre-processing operations, we obtain an array of 2000 × 176 × 208 × 3 size.

2.3

CNN Model

The CNN model architecture is customized from the pre-trained VGG16 model. The fully connected (FC) layers of VGG16 are removed and replaced by a global average pooling (GAP) layer, and a last FC classification layer with a “softmax” activation is added (Fig. 1). During model training, three convolutional blocks were frozen to reduce the number of trainable parameters and to avoid overfitting. The rest two blocks of convolutional layers were fine-tuned along with the newly added FC layer. Model training was performed based on a five-fold CV scheme using an Adam optimizer with a learning rate of 1 × 10-4 and a learning rate decay of 0.5. Although Adam optimizer assigns a specific learning rate for each parameter, which is

Fig. 1 A customized VGG16 model consists of convolutional layers transferred from the pre-trained VGG16 model, a GAP (global average pooling layer), and two FC layers (FC-256 and FC-2)

Development of an Interpretable Deep Learning System for. . .

31

computed using the initial learning rate as an upper limit, to be sure that every learning rate update step does not exceed the initial learning rate, an exponential learning rate decay is assigned that lowers the upper limit. By doing so, the learning rate decay helps to prevent the loss from diverging after it decreases to a point. In our case, different values of decay were tried out, and the value 0.5 produced the best performance of the model over the validation set. The “categorical_crossentropy” was used as a loss function. By training the model for a different number of epochs with a batch size of 128 images, 90 epochs have produced the best performance of the model on the validation set. Four classification metrics, such as balanced accuracy, sensitivity, and specificity, were used to measure the performance of the model. The final results of the trained model are reported based on the average accuracy computed over the five folds on the validation set.

2.4

CNN Visualization

Model visualization methods enable understanding of the rationale behind a deep learning model’s decisions. For a CNN model, these interpretability approaches are applied to a trained model to inspect which image regions or features are given high importance for the prediction analysis. In this study, we employed four attributebased interpretability techniques (two gradient-based approaches, saliency maps and GradCAM, and two perturbation-based methods, SHAP and occlusion maps) for a classification problem of AD versus HC subjects. To emphasize the importance of these visualization tools, we performed two experiments. In the first experiment, a model is trained to classify subjects as AD versus HC, and visualization heatmaps highlight the brain regions that are used by the model to identify AD brain scans from healthy MRI images. This helps to check if our results are in line with the neural correlates of AD, which are defined in the previous AD studies. The aim of the second experiment is to highlight the potential of these visualization tools for identifying biased models producing highly inflated performances. Data leakage caused by slice-level split is one of the methodological pitfalls of applying 2D CNNs for the classification of 3D MRI data that result in a biased model outputting overestimated performance on the test set [8, 10, 11]. Following a similar procedure to the experiment explained in Sect. 3.2.1, in this study, we also trained two architecturally similar models on the same dataset using two data split methods. While the first model is trained by applying subject-level split, hence without data leakage, the second model is trained on data that is divided based on MRI slices introducing data leakage. Correctly classified AD test samples are then passed through the trained models, and visualization heatmaps generated from the two models are compared to check if reliable features are used by the two CNN models.

32

S. W. Atnafu and S. Diciotti

3 Results The results of Experiment 1 and Experiment 2 are presented in this section. In Experiment 1, the performance of our interpretable CNN model as measured by the average accuracy, sensitivity, and specificity values computed over the five folds on the validation set is reported (Table 2). Our model identifies AD MRI slices with an accuracy of 71.62%, sensitivity 71.85%, and specificity score 72.73%. The learning curve is also shown in Fig. 2. An example of the visualization heatmap images generated by passing MRI images of AD patients, which are predicted by the model taken from the validation set, can be seen in Fig. 3. In Experiment 2, the trained model introducing data leakage achieved a test set accuracy of 95.12% (Table 3). The learning curve and visualization heatmaps of the biased model are illustrated in Figs. 4 and 5, respectively.

4 Discussion In this study, a deep learning model customized from VGG16 is proposed for binary classification of AD versus HC subjects. The proposed CNN model was trained by employing a transfer learning technique to prevent model overfitting caused by the small size of the training data. Instead of training a shallow CNN from scratch, finetuning a deeper pre-trained model allows the trained models to achieve excellent results and to have better generalization ability since depth is a crucial parameter in the design of a good performing CNN model [30]. Hence a pre-trained VGG16 Table 2 Average model’s performance computed over the five folds on the validation set Training set Validation set

Sensitivity 0.8146 0.7185

Specificity 0.7686 0.7273

Fig. 2 Learning curves of the model on both the training and validation sets

Accuracy 0.7916 0.7162

Development of an Interpretable Deep Learning System for. . .

33

Fig. 3 CNN visualization heatmaps of MRI slices taken from four representative AD patients, who are correctly classified by the CNN model: (a) Grad-CAM images, (b) saliency maps, (c) occlusion maps, and (d) SHAP heatmaps Table 3 Average accuracy of the biased model computed over the five folds on the validation set Training set Validation set

Sensitivity 0.9996 0.9592

Specificity 0.9810 0.9450

Fig. 4 Learning curves of the biased model trained with data leakage

Accuracy 0.9923 0.9512

Fig. 5 CNN visualization heatmaps indicate a model producing a biased performance due to the presence of data leakage. Images on the left side are generated by the model trained on data split based on slices (with data leakage). For CAM, occlusion map, and SHAP, the heatmap represents a very low number (score close to 0), capturing the biased model. Rather, Grad-CAM is seen as less capable of identifying the biased model

Development of an Interpretable Deep Learning System for. . .

35

model is fine-tuned on our MRI dataset. The model was trained on a brain image collection of the OASIS dataset achieving an average accuracy of 71.62% on the test set. Compared to previous studies employing exactly the same T1-weighted MRI sub-samples taken from the OASIS dataset [10, 11], our model classifies AD and HC subjects with better accuracy. Although in other few studies [12, 14, 16] the authors reported higher accuracies, these results are due to the use of a larger number of subjects, multimodality, and the application of data augmentation to improve the performance of the model. Apart from reporting the model’s performance, neither of these studies included model visualization tools to ensure that the models focus on meaningful brain regions to perform the predictive analysis. On the contrary, our proposed model incorporates four different visualization methods, strengthening our system’s reliability. The results also showed that the interpretation techniques highlight features located around the frontal, temporal, and parietal lobe of the cerebral cortex and areas around the thalamus. The SHAP method outperformed the other methods in localizing the frontal lobe, while the cortical atrophy and alterations around the thalamus were better captured by the Grad-CAM method. The visualization outcomes by the CAM-based technique are very much distributed. Instead, the GradCAM method produced a more localized heatmap. Regarding the role of explainability tools in identifying a biased model, such as a model trained by introducing data leakage, all visualization techniques indicate the bias incurred during the model training procedure. Figure 5 illustrates representative visualization heatmaps produced by the two models by passing correctly classified AD slices. The pixels on the occlusion heatmap represent the classification probability of the input image with respect to the target class (in our case, to be classified as a slice of an AD patient). While the model trained without data leakage produces a heatmap with a probability range of [0.46, 0.53], the biased model generates pixels in a probability range of [6 × 10–6, 1 × 10-5], which is a very small value representing the pixel’s insignificance for the model’s prediction. Similarly, the heatmap generated by the SHAP method represents the SHAP values that explain the importance of each pixel on the input image. While for the non-biased model the SHAP values lie between [-0.0075, 0.0075], the values for the biased model range are within [-4 × 10–12, 4 × 10-12], illustrating that all parts of the input image have no importance for the prediction output by the model. Saliency mapping also produces gradient values (gradient of the model’s prediction probability with respect to the input images) in the range [0, 0.8] and [0, 0.000008] for non-biased and biased models, respectively, highlighting the reduced importance of input features to determine the model’s output in the case of a biased model. GradCAM, in contrast, showed less capability of capturing the biased model as it assigns gradient values between 0 and 0.8 for a biased model and 0 and 1 for the non-biased model. Since each interpretability approach has its own limitation, incorporating multiple visualization methods helps better understand deep learning-based predictive systems.

36

S. W. Atnafu and S. Diciotti

The main limitation of this study is that due to the small size of the dataset, the CNN model was fine-tuned insufficiently to achieve higher classification accuracy.

5 Conclusion In this study, we presented an interpretable 2D CNN model that performs a diagnosis of AD from sMRI data. The model was trained on the OASIS dataset of AD and HC subjects based on a five-fold CV scheme and achieved a classification accuracy of 71.62% on the test set. Beyond that, the model is characterized as an interpretable system since it allows for visualizing features or important brain regions that are given the highest importance by the model for the prediction task.

References 1. Selkoe, D.J., Lansbury, P.J.: Alzheimer’s disease is the most common neurodegenerative disorder. In: Basic Neurochemistry: Molecular, Cellular and Medical Aspects, vol. 6, pp. 101–102. Lippincott-Raven, Philadelphia (1999) 2. Kelley, B.J., Petersen, R.C.: Alzheimer’s disease and mild cognitive impairment. Neurol. Clin. 25(3), 577–609 (2007) 3. Small, G.W., Rabins, P.V., Barry, P.P., Buckholtz, N.S., DeKosky, S.T., Ferris, S.H., Finkel, S. I., Gwyther, L.P., Khachaturian, Z.S., Lebowitz, B.D., McRae, T.D.: Diagnosis and treatment of Alzheimer disease and related disorders: consensus statement of the American Association for Geriatric Psychiatry, the Alzheimer’s Association, and the American Geriatrics Society. JAMA. 278(16), 1363–1371 (1997) 4. LeCun, Y., Bengio, Y., Hinton, G.: Deep learning. Nature. 521(7553), 436–444 (2015) 5. Liu, M., Li, F., Yan, H., Wang, K., Ma, Y., Shen, L., Xu, M., Alzheimer’s Disease Neuroimaging Initiative: A multi-model deep convolutional neural network for automatic hippocampus segmentation and classification in Alzheimer’s disease. NeuroImage. 208, 116459 (2020) 6. Oh, K., Chung, Y.C., Kim, K.W., Kim, W.S., Oh, I.S.: Classification and visualization of Alzheimer’s disease using volumetric convolutional neural network and transfer learning. Sci. Rep. 9(1), 1–16 (2019) 7. Qiu, S.: Development and validation of an interpretable deep learning framework for Alzheimer’s disease classification. Brain. 143(6), 1920–1933 (2020) 8. Wen, J., Thibeau-Sutre, E., Diaz-Melo, M., Samper-González, J., Routier, A., Bottani, S., Dormont, D., Durrleman, S., Burgos, N., Colliot, O., Alzheimer’s Disease Neuroimaging Initiative: Convolutional neural networks for classification of Alzheimer’s disease: overview and reproducible evaluation. Med. Image Anal. 63, 101694 (2020) 9. Feng, W., Halm-Lutterodt, N.V., Tang, H., Mecum, A., Mesregah, M.K., Ma, Y., Li, H., Zhang, F., Wu, Z., Yao, E., Guo, X.: Automated MRI-based deep learning model for detection of Alzheimer’s disease process. Int. J. Neural Syst. 30(06), 2050032 (2020) 10. Yagis, E., De Herrera, A.G.S., Citi, L.: Generalization performance of deep learning models in neurodegenerative disease classification. In: 2019 IEEE International Conference on Bioinformatics and Biomedicine (BIBM), (pp. 1692–1698). IEEE (2019) 11. Yagis, E., Atnafu, S.W., de Herrera García Seco, A., Marzi, C., Scheda, R., Giannelli, M., Tessa, C., Citi, L., Diciotti, S.: Effect of data leakage in brain MRI classification using 2D convolutional neural networks. Sci. Rep. 11(1), 22544 (2021)

Development of an Interpretable Deep Learning System for. . .

37

12. Reference 13. Puente-Castro, A., Fernandez-Blanco, E., Pazos, A., Munteanu, C.R.: Automatic assessment of Alzheimer’s disease diagnosis based on deep learning techniques. Comput. Biol. Med. 120, 103764 (2020) 14. Saratxaga, C.L., Moya, I., Picón, A., Acosta, M., Moreno-Fernandez-de-Leceta, A., Garrote, E., Bereciartua-Perez, A.: MRI deep learning-based solution for Alzheimer’s disease prediction. J. Pers. Med. 11(9), 902 (2021) 15. Mehmood, A., Maqsood, M., Bashir, M., Shuyuan, Y.: A deep Siamese convolution neural network for multi-class classification of Alzheimer disease. Brain Sci. 10(2), 84 (2020) 16. Massalimova, A., Varol, H.A.: Input agnostic deep learning for Alzheimer’s disease classification using multimodal MRI images. In: 2021 43rd Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC), (pp. 2875–2878). (2021) 17. Linardatos, P., Papastefanopoulos, V., Kotsiantis, S.: Explainable ai: a review of machine learning interpretability methods. Entropy. 23(1), 18 (2020) 18. Lipton, Z.C.: The mythos of model interpretability: in machine learning, the concept of interpretability is both important and slippery. Queue. 16(3), 31–57 (2018) 19. Gao, J., Chen, M., Li, Y., Gao, Y., Li, Y., Cai, S., Wang, J.: Multisite autism spectrum disorder classification using convolutional neural network classifier and individual morphological brain networks. Front. Neurosci. 14, 629630 (2021) 20. Jimeno, M.M., Ravi, K.S., Jin, Z., Oyekunle, D., Ogbole, G., Geethanath, S.: ArtifactID: identifying artifacts in low-field MRI of the brain using deep learning. Magn. Reson. Imaging. 89, 42–48 (2021) 21. Zhang, Y., Hong, D., McClement, D., Oladosu, O., Pridham, G., Slaney, G.: Grad-CAM helps interpret the deep learning models trained to classify multiple sclerosis types using clinical brain magnetic resonance imaging. J. Neurosci. Methods. 353, 109098 (2021) 22. Tang, Z., Chuang, K.V., DeCarli, C., Jin, L.W., Beckett, L., Keiser, M.J., Dugger, B.N.: Interpretable classification of Alzheimer’s disease pathologies with a convolutional neural network pipeline. Nat. Commun. 10(1), 2173 (2019) 23. Lu, P., Hu, L., Zhang, N., Liang, H., Tian, T., Lu, L.: A two-stage model for predicting mild cognitive impairment to Alzheimer’s disease conversion. Front. Aging Neurosci. 14, 826622 (2022) 24. Iizuka, T., Fukasawa, M., Kameyama, M.: Deep-learning-based imaging-classification identified cingulate island sign in dementia with Lewy bodies. Sci. Rep. 9(1), 8944 (2019) 25. Sánchez Fernández, I., Yang, E., Calvachi, P., Amengual-Gual, M., Wu, J.Y., Krueger, D., Northrup, H., Bebin, M.E., Sahin, M., Yu, K.H., Peters, J.M.: Deep learning in rare disease. Detection of tubers in tuberous sclerosis complex. PLoS One. 15(4), e0232376 (2020) 26. Marcus, D.S., Wang, T.H., Parker, J., Csernansky, J.G., Morris, J.C., Buckner, R.L.: Open Access Series of Imaging Studies (OASIS): cross-sectional MRI data in young, middle aged, nondemented, and demented older adults. J. Cogn. Neurosci. 19(9), 1498–1507 (2007) 27. Hon, M., Khan, N.M.: Towards Alzheimer’s disease classification through transfer learning. In: 2017 IEEE International Conference on Bioinformatics and Biomedicine (BIBM), (pp. 1166–1169). (2017) 28. Morris, J.C.: The Clinical Dementia Rating (CDR): current version and scoring rules. Neurology. 43(11), 2412–2414 (1993) 29. Han, X., Kwitt, R., Aylward, S., Bakas, S., Menze, B., Asturias, A., Vespa, P., Van Horn, J., Niethammer, M.: Brain extraction from normal and pathological images: a joint PCA/imagereconstruction approach. NeuroImage. 176, 431–445 (2018) 30. Alzubaidi, L., Zhang, J., Humaidi, A.J., Al-Dujaili, A., Duan, Y., Al-Shamma, O., Santamaría, J., Fadhel, M.A., Al-Amidie, M., Farhan, L.: Review of deep learning: concepts, CNN architectures, challenges, applications, future directions. J Big Data. 8, 1–74 (2021)

Digital Clinical Decision Support System for Screening of Eye Diseases Gizeaddis Lamesgin Simegn

and Mizanu Zelalem Degu

Contents 1 2

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1 Image Acquisition and Enhancement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Retinal Vessel Extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3 K-Means Clustering-Based Image Segmentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4 Cup-to-Disk (CDR) Calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5 Intraocular Pressure Measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1 Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 User Interface and Prototype . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 System Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

40 41 42 43 43 44 44 44 44 44 45 47 48

G. L. Simegn (✉) Biomedical Imaging Unit, School of Biomedical Engineering, Jimma Institute of Technology, Jimma University, Jimma, Ethiopia AI and Biomedical Imaging Research Unit, Jimma Institute of Technology, Jimma University, Jimma, Ethiopia e-mail: [email protected] M. Z. Degu AI and Biomedical Imaging Research Unit, Jimma Institute of Technology, Jimma University, Jimma, Ethiopia Faculty of Computing and Informatics, Jimma Institute of Technology, Jimma University, Jimma, Ethiopia © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 K. Mequanint et al. (eds.), Advancement of Science and Technology in Sustainable Manufacturing and Process Engineering, Green Energy and Technology, https://doi.org/10.1007/978-3-031-41173-1_3

39

40

G. L. Simegn and M. Z. Degu

1 Introduction The human eye is a critical visual organ that responds to light and pressure providing sensation of sight, and allowing us to see and understand colors and object proportions. The structure of the eye is affected by a variety of factors, resulting in vision impairment and blindness. Glaucoma, cataract, and age-related macular degeneration (AMD) are the most common visual disorders and the primary causes of blindness globally [1]. Glaucoma is an illness characterized by particular optic nerve alterations and visual field abnormalities that correlate to areas of structural damage to the optic nerve. If not discovered and treated early enough, it could be a main cause of irreversible eyesight loss [2]. It is frequently linked to an increase in the intraocular pressure (IOP) of the fluid in the eye known as aqueous humor. High intraocular pressure is a common cause and the sole controllable risk factor for glaucoma [3]. A cataract is a clouding of the lens that causes a blurry image. Free radical-induced oxidative damage is thought to play a role in aging and the development of chronic illnesses, including cataract formation [4]. Cataracts are a leading cause of blindness around the world [5]. Drusen, or clumps of yellowish material, gradually form within and under the retinal pigmented epithelium (RPE) in age-related macular degeneration. The RPE cells may die, and as a result, the photoreceptors will no longer be able to function, resulting in vision loss in that area of the retina [6]. AMD is another prominent cause of vision loss in developed countries, with 288 million people expected to be affected by the year 2040 [7, 8]. Slit lamp microscopy, gonioscopy, ophthalmoscopy, and perimeter are among the most used instruments and procedures for diagnosing eye problems. Because of their high cost, most health facilities in the developing world do not have access to hightech instruments. Furthermore, typical tabletop high-tech imaging technologies are restricted due to their large design and requirement for expert operators. In low-resource settings, slit-lamp microscopy is the most common method for eye screening. Because automated devices are not available in low-resource settings, eye screening is usually performed by manual observation, which necessitates subjective decision-making. This process of subjective decision-making is prone to error, which may result in misdiagnosis. Few literatures have presented smartphone-based retinal imaging systems with wireless data transfer capabilities to gather diagnostic images for eye disease [9– 12]. Even though these methods make eye screening more affordable and accessible to a wider range of people, they have limitations in terms of image quality and processing power. Furthermore, for a thorough standalone eye disease diagnosis, critical factors such as intraocular pressure (IOP, which is the fluid pressure inside the eye) must be measured, which is not possible with smartphone-based imaging approaches. The balance between the rate of aqueous humor secretion and aqueous outflow determines IOP. It is used to determine the pressure inside the anterior eye based on a small portion of the cornea’s resistance to flattening. The normal IOP is

Digital Clinical Decision Support System for Screening of Eye Diseases

41

distributed in the range of 11–21 mmHg [13]. If the value is outside of this range, it is regarded as a suspect of abnormality. Glaucoma, uveitis, and retinal detachment can all be signs of abnormal IOP [14, 15]. In this work, a simple, cost-effective, and portable digital eye disease diagnosis support system that integrates an image processing unit and IOP monitoring system is proposed. This system allows the physician to save diagnosis results and analyze them later if necessary. This tool could be utilized in rural health centers, emergency departments, and eye clinics, among other places. Furthermore, because it is a digital system with image storage capabilities, it might be used for telemedicine applications.

2 Methods The proposed system includes image acquisition, image enhancement, vessel extraction, and IOP monitoring systems. To collect retina images, a 3D lens, which is often used in slit-lamp microscopy, is coupled with a high-resolution camera. After image acquisition, the images are denoised (enhanced) and segmented and vessels are extracted using the digital image processing unit. In addition, an infrared (IR) sensor is used in conjunction with an Arduino microcontroller to measure IOP. The user can easily obtain image and IOP data and enhance and extract the desired image feature, using a custom-made graphical user interface (GUI). Screening results are presented on the LCD as normal or abnormal according to the IOP measurement. An IOP greater than 22 mm Hg is the usual criterion for the presence of glaucoma [16, 17]. Furthermore, the cup-to-disk ratio (CDR) is derived from the segmented image as an additional means of glaucoma detection and diagnosis decision support. The overall features and flowchart of the proposed systems are shown in Figs. 1 and 2.

Fig. 1 General block diagram of the proposed system

G. L. Simegn and M. Z. Degu

42

Start

IR sensor module

Camera with 3D lens

Change in length (Δl )

Retina image

Enhancement IOP calculation σ=Δl ∗ε/l Segmentation

CDR= Cup diameter / Disc diameter

Segmented image

σ = 10 up to 22 mmHg

yes

Abnormal IOP value

No CDR >0.3

Yes

Glaucoma +ve CDR value

Normal IOP value

No Glaucoma -ve CDR value

Fig. 2 Flowchart of the proposed CDR determination and IOP measurement system

2.1

Image Acquisition and Enhancement

The image acquisition system is made up of a camera and a 3D lens (with 90 diopters) mounted on a well-designed wood stand and an adjustable chin support mechanism. The use of an adjustable chin support enables for image acquisition without sacrificing patient comfort. Adjusting the brightness and contrast of the acquired image is part of the image enhancement feature. A device’s display brightness is typically not a linear function of the applied voltage signal. As a result, the exact brightness variation of a digital image displayed will not be replicated. The discrepancy between how a camera collects content, how a display displays content, and how our visual system processes light is corrected via gamma correction [18, 19]. Thus, image enhancement is accomplished through histogram equalization, filtering, and gamma correction techniques.

Digital Clinical Decision Support System for Screening of Eye Diseases

2.2

43

Retinal Vessel Extraction

The morphology and branching pattern of retinal blood vessels alter as a result of numerous eye diseases [20]. The development of retinal vessels is a key sign for a variety of disorders, including diabetic retinopathy, stroke, hypertension, arteriosclerosis, and cardiovascular diseases [21]. To detect vessels, edge detection is a crucial image processing procedure. Edge detection algorithms seek out the most important edges in a picture or scene. The blood vessel extraction was performed by edge detection of the input retina image using Kirsch’s Templates [22] in various orientations, followed by thresholding algorithm.

2.3

K-Means Clustering-Based Image Segmentation

The process of partitioning an image into a set of non-overlapping sections whose union is the full image is known as segmentation. In this study, watershed transform and k-means clustering techniques were applied for image segmentation. A smoothing technique was used to reduce oversampling caused by the watershed transform. The cup and the optic disks were then segmented using the k-means clustering algorithm. K-means clustering is one of the unsupervised machine learning algorithms used to split data into k mutually exclusive groups known as clusters. A cluster is made up of objects that have a common feature. One cluster may have more than one region according to the distribution of objects in the space. Each cluster is described by its centroid and member function. The sum of distances between each object and its cluster centroid is minimized using the K-means algorithm, which is an iterative technique. The k-means algorithm produces as many well-separated and closely packed clusters as possible. The value of k (number of clusters) that is chosen has a significant impact on the outcome of the segmentation. The number k = 4 was chosen for our purposes because there are four sections in the acquired retina image: background, foreground, optic disc (OD), and blood vessels. The k-means algorithm produced four well-separated clusters. We choose the cluster with the highest intensity because the optic disc is the brightest region. Filtration of the optic disc was then performed to remove undesirable regions before it may be segmented. Filtering based on connected components is used to remove unnecessary regions and segment the optic disc. Connected component analysis (CCA) is a well-known image processing approach that scans an image and divides it into components based on pixel connectedness. The optic disk contains the highest-intensity cluster regions, which are separated by filtering these regions along major and minor axes. For detection, the centroid of the segmented optic disc was determined and compared to the ground truth border.

44

2.4

G. L. Simegn and M. Z. Degu

Cup-to-Disk (CDR) Calculation

The CDR (cup-to-disk ratio) is obtained after OD segmentation. The CDR compares the diameter of the optic disk’s “cup” component to the entire diameter of the disk [23]. The standard cup-to-disk ratio is 0.3. A high CDR could indicate glaucoma or another problem.

2.5

Intraocular Pressure Measurement

IOP is a crucial metric for predicting the progression of glaucoma, uveitis, and retinal detachment [14, 15]. The IOP of the eye was measured using an infrared sensor in this research. This was determined using the cornea’s Young’s modulus, the distance between the cornea and the eyelid, and the change in cornea length over time. The initial length is determined by multiplying the average distance between the sensor and the eyelid (10 cm) by the average distance between the cornea and the eyelid (0.03) [24]. For the cornea, a mean Young’s modulus of 7500.6 mmHg was used [25]. The real-time stress (pressure) was determined using the biomechanics principle σ = Δl*?/l, where σ is stress, Δl is change in length, ? is elastic modulus, and l is length. The values were evaluated using Arduino microcontroller and presented on LCD.

3 Results 3.1

Simulation

A simulation software (Proteus 8) was used to simulate the pressure measurement system. To replicate the pressure sensor, a Proteus IR sensor was employed. The Arduino was then supplied with the IR sensor’s output. The observed distance that is detected by the sensor is converted into pressure using the known Young’s modulus of the cornea. On the LCD, the output cornea distance and the converted pressure are displayed. The simulation’s results are demonstrated in Fig. 3.

3.2

User Interface and Prototype

To make the image processing module easier to use, a graphical user interface (GUI) was developed using MATLAB. Image capture and preview, image enhancement, vessel extraction, segmentation, and analysis are all part of this module. The image storage feature, which allows physicians to save images for later use, is also

Digital Clinical Decision Support System for Screening of Eye Diseases

45

Fig. 3 Simulation result of the IOP measurement

included. The GUI’s image acquisition, saving, image processing, and analysis sections are demonstrated in Fig. 4. In the enhanced and segmented image, the macula and optic disk can be clearly visualized. As illustrated in Fig. 4b, the segmentation method can distinguish between the optic disk and the macula. The vessel extraction reveals the central retinal artery and vein branches emerging from the optic disk, as well as other vessel structures. The computed CDR value for the provided images was 0.4, which is higher than the average normal value of 0.3 and could indicate glaucoma. However, research reveals that clinical evaluation of CDR values should not be utilized as the sole technique for detecting glaucoma development [26]. Additional investigations such as pressure may be required. Figure 5 shows the finished prototype, which comprises the IOP measuring and image processing components. The combined results of IOP, CDR value, and extracted vasculature provide significant information to the physician regarding eye diseases.

3.3

System Testing

The proposed system was tested for detection/determination accuracy, portability, and ease of use to guarantee that a system meets its planned specifications and other needs. The prototype (excluding the computer) costed 335.56 USD. This makes the design easily accessible, even in low-resource environments. Non-contact tonometry (or air-puff tonometry) was utilized as a reference for the accuracy test of the IOP

46

G. L. Simegn and M. Z. Degu

Fig. 4 A GUI demonstrating (a) parts of image acquisition and enhancement and (b) parts of the vessel extraction, segmentation, and analysis. The calculated CDR value for the given images shows 0.4 which may imply glaucoma

measurement. Five individuals took part in the experiment. Using our IOP measurement system, we were able to attain a 96.2% accuracy. The performance of the segmentation result was tested using locally acquired and publicly available annotated retinal images and found to be 93.3% accurate. Table 1 illustrates the tests performed and test results. In summary, the current approaches for diagnosing eye disorders in low-resource settings are manual, time-consuming, and physically uncomfortable for the clinicians. These manual diagnosis techniques are inherently subjective, making them prone to inaccuracy and misdiagnosis. The proposed method demonstrates a digital eye screening support technique that can be utilized to detect prevalent eye disorders. Because of its cost-effectiveness and robustness, the system could be used in low-resource contexts as a decision-support system in the diagnosis of eye disease.

Digital Clinical Decision Support System for Screening of Eye Diseases

47

Fig. 5 The final prototype of the proposed system Table 1 Testing plan and results No 1

Tests Accuracy (IOP)

2

Accuracy of cup and disc segmentation Screening time

3

Method Non-contact tonometry (or air-puff tonometry) as gold standard for 5 tests Test data (250 retinal images)

Result 96.2%

Time from recording to decision

40 s

93.3%

4 Conclusion The suggested method demonstrated its ability to capture and analyze images, extract vessels, and interpret the outcomes in combination with intraocular pressure measurement and cup-to-disk ratio calculation. Additionally, its image-saving capacity renders it well-suited for telemedicine applications. This proposed approach could serve as a valuable computer-aided diagnostic tool for extracting and interpreting fundus images. In particular, the extracted vessels and CDR outcomes could be useful in a computer-assisted diagnosis clinical setting, particularly in regions where both resources and specialists are scarce. Acknowledgment Tools and materials required for this study were supported by the School of Biomedical Engineering, Jimma Institute of Technology, Jimma University.

48

G. L. Simegn and M. Z. Degu

References 1. Yoong, W., Karavolos, S., Damodaram, M., Madgwick, K., Milestone, N., Al-Habib, A., et al.: Observer accuracy and reproducibility of visual estimation of blood loss in obstetrics: how accurate and consistent are health-care professionals? Arch. Gynecol. Obstet. 281(2), 207–213 (2010) 2. Weinreb, R.N., Aung, T., Medeiros, F.A.: The pathophysiology and treatment of glaucoma: a review. JAMA. 311(18), 1901–1911 (2014) 3. Coleman, A.L., Kodjebacheva, G.: Risk factors for glaucoma needing more attention. Open Ophthalmol J. 3, 38–42 (2009) 4. Lobo, V., Patil, A., Phatak, A., Chandra, N.: Free radicals, antioxidants and functional foods: impact on human health. Pharmacogn. Rev. 4(8), 118–126 (2010) 5. Correia, M., Das, T., Magno, J., Pereira, B.M., Andrade, V., Limburg, H., et al.: Prevalence and causes of blindness, visual impairment, and cataract surgery in Timor-Leste. Clin. Ophthalmol. 11, 2125–2131 (2017) 6. Bhutto, I., Lutty, G.: Understanding age-related macular degeneration (AMD): relationships between the photoreceptor/retinal pigment epithelium/Bruch’s membrane/choriocapillaris complex. Mol. Asp. Med. 33(4), 295–317 (2012) 7. Wong, W.L., Su, X., Li, X., Cheung, C.M., Klein, R., Cheng, C.Y., et al.: Global prevalence of age-related macular degeneration and disease burden projection for 2020 and 2040: a systematic review and meta-analysis. Lancet Glob. Health. 2(2), 106–116 (2014) 8. Westenskow, P.D.: Nicotinamide: a novel treatment for age-related macular degeneration? Stem Cell Investig. 4, 86 (2017) 9. Kim, T.N., Myers, F., Reber, C., Loury, P.J., Loumou, P., Webster, D., et al.: A smartphonebased tool for rapid, portable, and automated wide-field retinal imaging. Transl. Vis. Sci. Technol. 7(5), 21 (2018) 10. Suto, S., Hiraoka, T., Okamoto, Y., Okamoto, F., Oshika, T.: Photography of anterior eye segment and fundus with smartphone. Nippon Ganka Gakkai Zasshi. 118(1), 7–14 (2014) 11. Maamari, R.N., Keenan, J.D., Fletcher, D.A., Margolis, T.P.: A mobile phone-based retinal camera for portable wide field imaging. Br. J. Ophthalmol. 98(4), 438–441 (2014) 12. Myung, D., Jais, A., He, L., Blumenkranz, M.S., Chang, R.T.: 3D printed smartphone indirect lens adapter for rapid, high quality retinal imaging. J Mob Technol Med. 3(1), 9–15 (2014) 13. Sherwood, J.M.R.-T.E., Bertrand, J.A., Rowe, B., Overby, D.R.: Measurement of outflow facility using iPerfusion. PLoS One. 11(3), e0150694 (2016) 14. Acott, T.S., Kelley, M.J., Keller, K.E., Vranka, J.A., Abu-Hassan, D.W., Li, X., et al.: Intraocular pressure homeostasis: maintaining balance in a high-pressure environment. J. Ocul. Pharmacol. Ther. 30(2–3), 94–101 (2014) 15. Vranka, J.A., Kelley, M.J., Acott, T.S., Keller, K.E.: Extracellular matrix in the trabecular meshwork: intraocular pressure regulation and dysregulation in glaucoma. Exp. Eye Res. 133, 112–125 (2015) 16. Wang, Y.X., Xu, L., Wei, W.B., Jonas, J.B.: Intraocular pressure and its normal range adjusted for ocular and systemic parameters. The Beijing Eye Study 2011. PLoS One. 13(5), e0196926 (2018) 17. Mamikonian, V.R.: The upper limit of individual normal range of intraocular pressure--a personalized criterion for IOP evaluation. Vestn. Oftalmol. 130(6), 71–78 (2014) 18. Bull, D.R.: Chapter 4 - Digital picture formats and representations. In: Bull, D.R. (ed.) Communicating Pictures, pp. 99–132, 2014. Academic Press, Oxford 19. Farid, H.: Blind inverse gamma correction. IEEE Trans. Image Process. 10(10), 1428–1433 (2001) 20. Fraz, M.M., et al.: Retinal vessel extraction using first-order derivative of Gaussian and morphological processing. Advances in visual computing: 7th international symposium, ISVC 2011, Las Vegas, Proceedings, Part I 7. Springer Berlin Heidelberg, (2011)

Digital Clinical Decision Support System for Screening of Eye Diseases

49

21. Li, X., Wee, W.G.: Retinal vessel detection and measurement for computer-aided medical diagnosis. J. Digit. Imaging. 27(1), 120–132 (2014) 22. Kirsch, R.A.: Computer determination of the constituent structure of biological images. Comput. Biomed. Res. 4(3), 315–328 (1971) 23. Marjanovic, I.: The optic nerve in glaucoma. In: The Mystery of Glaucoma, 6th edn. Available from: https://doi.org/10.5772/19811 (2011) 24. van den Bosch, W.A., Leenders, I., Mulder, P.: Topographic anatomy of the eyelids, and the effects of sex and age. Br. J. Ophthalmol. 83(3), 347–352 (1999) 25. Elsheikh, A., Wang, D., Pye, D.: Determination of the modulus of elasticity of the human cornea. J. Refract. Surg. 23(8), 808–818 (2007) 26. Tatham, A.J., Weinreb, R.N., Zangwill, L.M., Liebmann, J.M., Girkin, C.A., Medeiros, F.A.: The relationship between cup-to-disc ratio and estimated number of retinal ganglion cells. Invest. Ophthalmol. Vis. Sci. 54(5), 3205–3214 (2013)

Evaluation of Three Irrigation Management Tools for Improving Crop and Water Productivity of Wheat (Triticum aestivum) in Koga Irrigation Scheme Habtamu D. Tarekegn, Seifu A. Tilahun, Fasikaw A. Zimale, and Petra Schmitter

Contents 1 2

3

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Material and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1 Description of the Study Area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Cropping Practices and Crop Production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3 Experimental Design and Treatment Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4 Land and Seedbed Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5 Materials Used . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.6 Irrigation Water Productivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1 Irrigation Water Used for Wheat Yield . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1.1 Irrigation Water Applied at the Farm Level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1.2 Irrigation Water Used at Scheme Level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Wheat Yield . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.1 Wheat Yield at Farm Level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 Irrigation Water Productivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.1 Irrigation Productivity at the Farm Level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.2 Irrigation Water Productivity at Scheme Level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

52 53 53 55 55 55 56 57 57 57 57 57 60 60 60 60 63

H. D. Tarekegn (✉) · F. A. Zimale Faculty of Civil and Water Resources Engineering, Bahir Dar University, Bahir Dar, Ethiopia S. A. Tilahun Faculty of Civil and Water Resources Engineering, Bahir Dar University, Bahir Dar, Ethiopia International Water Management Institute, Yangon, Myanmar P. Schmitter International Water Management Institute, Yangon, Myanmar e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 K. Mekuanint et al. (eds.), Advancement of Science and Technology in Sustainable Manufacturing and Process Engineering, Green Energy and Technology, https://doi.org/10.1007/978-3-031-41173-1_4

51

52

H. D. Tarekegn et al. 3.4

Effect of WFD and Chameleon Sensor at Block and Scheme Level . . . . . . . . . . . . . . . . . 3.4.1 At the Block Level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4.2 At the Scheme Level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

63 63 66 67 67

1 Introduction Water is among nature’s most essential requisites to sustain life for plants, animals, and humans. Besides various other uses of water, the largest use of water in the world is made for irrigating lands [6]. The predominant agricultural system is based on smallholder production and entirely dependent on rain-fed agriculture with limited areas currently developed under irrigation conditions. In all, smallholders account for 96% of the total area cultivated [7]. The Ethiopian farming system depends strongly on rain-fed smallholder farms as a means of food and income for its population; virtually all food crops come from rain-fed agriculture systems [15]. Rain-fed agriculture products are not sufficient to ensure food security and the market demand of society. Irrigation application plays a vital role in water and crop productivity, safe nutrient movement in the soil, and sustainable use of land resources [3, 4]. Despite Ethiopia's water resources’ potential for development and variable rainfall, the utilization of modern irrigation techniques remains underdeveloped, falling short of its inherent potential [5]. Presently, the imperative of water scarcity necessitates a compelling push toward the adoption of innovative strategies for optimizing areas under irrigated agriculture [16]. High water loss results in lesser yield and reduced irrigated areas, which are linked to ineffective water use [31]. Much of the losses and inadequacies of irrigation systems occur at the farm level. Improper on-farm irrigation management practices lead to poor water distribution, non-uniform crop growth, excessive leaching in some areas (leading to waterlogging), and insufficient leaching in others (leading to salinity buildup), all of which decrease the yield per unit of land area and per unit of water applied [23, 27]. Excess irrigation can lead to permanent loss of land resources and leaching out of nutrients through lateral flow and deep percolation [25]. Efficient scheduling of irrigation water applications gives the highest return for the least amount of water [12]. The goal of an effective scheduling program is to supply the plants with sufficient water while minimizing loss to deep percolation or runoff [32]. To increase the efficiency and productivity of the existing irrigation systems to optimize water use, i.e., less volume of applied water with greater production [8], On-farm water use can be reduced substantially without decreasing productivity through improved irrigation technologies and efficient water management systems [24]. Soil moisture sensors help to determine when to irrigate and how much irrigation water to apply to the field, and they can be easily understood by farmers

Evaluation of Three Irrigation Management Tools for Improving Crop. . .

53

[26]. Wise application of irrigation water is crucial to ensure efficient utilization of water and other resources [13]. The use of technologies helps to increase production from irrigated agriculture instead of expansion of irrigated agriculture by developing new water resources. Improving the on-farm performance of the existing irrigation system, and the surface irrigation system is an alternative sustainable developmental approach. In this study, the specific area in Ethiopia is called Mecha which is one of food insecurity of the rural population [29] and where a Koga irrigation scheme is installed by the government. Most farmers think that the more water they use to irrigate the crop, the higher the crop productivity will be. The shortage of irrigation water has become a source of conflict between irrigators and the Water User Association (WUA). The irrigation production during the dry period from the scheme is growing mainly wheat which is the fourth and the third most important food crop in terms of production (25,376,398 q) and productivity (17.46 q/ha), respectively, in Ethiopia [30]. Wheat is mainly grown in the highlands of Ethiopia, and the two main wheatproducing regions (Oromia and Amhara) account for about 85% of the national wheat production [2]. Ethiopia’s wheat yield was 29% below the Kenya average, 13% below the African average, and 32% below the world average [1]. The objective of this study is, therefore, to test the performance of two technologies that guide farmers when to irrigate so that the scheme saves water. One technology is a much simpler, more accurate, and affordable tool (Full Stop Wetting Front Detector (FSWD)) for soil moisture measurement that has been developed for practical use in the field. The FSWD shows how deep the water has penetrated the soil after irrigation. It can be used to find out if irrigation water is too little or too much, assist in the management of fertilizer and salts, and detect waterlogging [9, 11, 20]. The second technology is the Chameleon Soil Water Sensor which is a tool that shows different colors depending on how difficult it is for plants to take up water from the soil.

2 Material and Methods 2.1

Description of the Study Area

The study was conducted in the Koga watershed in the northwestern part of Ethiopia with a total area of 170,752 km2. The Koga watershed lies in the Blue Nile basin above the confluence of Gilgel Abay (Little Nile). The project area covers a total size of about 10,000 ha as shown in Fig. 1. The river terminates at its confluence with the Gilgel Abay just to the west of the town of Wettet Abay [29]. The catchment area of the dam is 170.9 km2 and extends to an altitude of 3200 AMSL [17]. The Koga watershed has a total area of 266 km2. The elevation of the watershed stretches from 1800 at the gauge station (11°22′12″ N latitude and 37°02′ 15″ E longitude) to 3000 m above sea level [14].

54

H. D. Tarekegn et al.

Fig. 1 Location map of Koga irrigation scheme

The monthly flow in the Koga River follows the rain pattern. The minimum flow occurs in April. An increase in flow follows in May due to the early rains and reaches a peak in August. From July to September, 70% of the flow occurs [33]. Because of these meteorological factors, the mean annual runoff varies from 0 to 35 l/s per km2 with the lowest flow occurring between December and March.

Evaluation of Three Irrigation Management Tools for Improving Crop. . .

55

The Koga irrigation scheme is designed to improve watershed management in the catchment area of about 22,000 ha of land and supply irrigation to about 7000 ha of command area [18].

2.2

Cropping Practices and Crop Production

Rain-fed agriculture and irrigated agriculture are the two types of agriculture in the study area. The major field crops generally grown in the Koga irrigation scheme are wheat, maize, teff, and barley. Crop yields used for household consumption and the market include maize, wheat, teff, potato, and pepper. The cash crops produced in the study area include wheat, potato, onion, cabbage, maize, and barley. Most farmers in the study area rely on rain-fed agriculture, and the crops that are mostly produced in the rainy season include maize, teff, potato, paper, barley, and wheat. Irrigated agriculture in the area includes potato, onion, cabbage, pepper, wheat, and maize. Out of these crops’ potatoes, onion, cabbage, maize, and wheat are produced for cash crops per marketable commodity.

2.3

Experimental Design and Treatment Setup

In the Koga irrigation scheme, there is 11-night storage, which uses to irrigate 12 blocks. From 12 blocks, 3 blocks were selected. Namely: Chihona, Adibera, and Teleta blocks at the head, mid, and tail, respectively. From each block, two tertiary canals have been selected: one at the head and the other at the tail. Again, from one tertiary canal, there are three treatments such as wetting front detector, Chameleon sensor, and farmers’ practice now called control. Three farmers were selected from each treatment in each tertiary canal from three blocks. So finally, 54 farmers were selected. The location (latitude, longitude, and elevation) of each quaternary outlet including each farmer’s field was taken by GPS. Since the study is not in a controlled system, the area in each block and treatment is different for each farmer’s field. Figure 2 shows the schematic overview of the experimental blocks and water management groups.

2.4

Land and Seedbed Preparation

The furrow length for one farm was the same for all farmers to ensure that irrigation timing and quantity is uniform as possible among farmers within one treatment group. In one experiment, wheat crops were planted within a 30 m furrow length and a 25 m furrow width. Since the system is not controlled, the proposed plot was selected based on the following criteria as a factor:

56

H. D. Tarekegn et al.

Control Chihona (Head)

Two Tc’s at head and tail

WF D CH

Three farm fields for each treatment

Control Koga Irrigation Scheme

Two Tc’s

Adibera

at head

(Mid)

and tail

Two

Teleta (Tail)

Tc’s at head

Three farm

WF D CH

Control trol WF D CH

fields for each treatment

A total of 54 farms

Three farm fields for each treatment

Fig. 2 Sample schematic overview of the experimental blocks and water management groups

All farmers use the same type and rate of fertilizer. All farmers can follow each treatment recommendation. All farmers use the same furrow width and length. All farmer’s plots should be on the same slope. In terms of agricultural practices, all farmers were instructed to use a similar type and amount of fertilizer. Farmers applied on a range of 80–120 kg per ha of urea (46% N) and 80–110 kg per ha of DAP (65% P and 18% N). The farmers were trained and instructed to adjust their irrigation quantity based on the signaling of both the shallow and deep detector: “if the shallow detector reacts sufficient water was available in the top 20 cm horizon with some percolation to lower horizons whereas if the deep detector reacted the likelihood of water moving below the root zone is high.” Hence, farmers were instructed to fill furrows and adjust the amount throughout the season based on the detector response after filling and also adjust their irrigation quantity based on the color of a Chameleon at the shallow, mid, and deep detector.

2.5

Materials Used

The materials used in the experiment include a shovel, trowel, measuring tape, wetting front detector, fertilizer, wheat yield, GPS, Soil Moisture Profile Probe, and V-notch. And also, the software used in the research are SPSS, MS Excel, and GIS.

Evaluation of Three Irrigation Management Tools for Improving Crop. . .

2.6

57

Irrigation Water Productivity

Irrigation water productivity is the total yield per unit quantity of irrigation water used. Different factors that affect water productivity are crop management, soil preparation, soil type, crop variety, and climate [21]. For this study, these factors will be constant for all three treatments. The number of observations used for water productivity was six for all three treatments, CH, WFD, and farmers’ local practice treatments, from each block. Within many irrigated areas, water is an increasingly scarce resource. Productivity of irrigation in terms of this scarce resource should be assessed from a variety of viewpoints [35]. The most common are: productivity in terms of actual evapotranspiration (Etc) and terms of the volume of water applied during the cropping period. The water productivity then is expressed as [19]: IWP =

Y V

where IWP = irrigation water productivity (kg/m3), Y = the yield of the harvested crop (kg/ha) and V = the total volume of water applied (m3)

3 Results and Discussion 3.1

Irrigation Water Used for Wheat Yield

3.1.1

Irrigation Water Applied at the Farm Level

The box plots of irrigation water applied for Chihona, Adibera, and Teleta blocks are shown Fig. 3a–c, respectively. The average irrigation amount for the WFD wheat field was 388 mm, 423 mm, and 428 mm for Chihona, Adibera, and Teleta blocks, respectively. The average irrigation amount for the Chameleon wheat field was 459 mm, 437 mm, and 375 mm for Chihona, Adibera, and Teleta blocks, respectively. The difference between the WFD and Control field at a 5% significance level is significant for the different treatments. Farmers irrigating more than 600 mm are 17% in the control group, while 6% in the WFD group and Chameleon group. But there is no significant difference within the same treatment at the different blocks. The analysis of the variance of the blocks and their interaction does not significantly influence the wheat water used. There was no interaction effect between blocks and field levels in the same treatment.

3.1.2

Irrigation Water Used at Scheme Level

The average irrigation amount for wheat fields was 488 mm, 413 mm, and 424 mm per season for control, WFD, and Chameleon fields, respectively, as shown in

58

H. D. Tarekegn et al.

Fig. 3 Box plot of the volume of water applied for (a) Chihona block (b) Adibera block, and (c) Teleta block

Evaluation of Three Irrigation Management Tools for Improving Crop. . . Table 1 Volume of Irrigation water(mm) used at scheme level

Treatment Control WFD Chameleon

Statistical descriptors Min Max Avg Scheme 247.68 687.60 488.25 259.44 867.60 413.16 259.20 630.72 424.05

59

SD

CV%

118.25 132.83 99.30

24.22 32.15 23.42

Fig. 4 Boxplot of water applied for the scheme

Table 1. And the boxplot of water applied for the scheme is shown in Fig. 4. On average, 19%, 56%, and 6% of irrigation water were saved at Chihona, Adibera, and Teleta blocks, respectively, when the farmer used the WFD irrigation scheduling method compared to the control. The difference was significant at a 5% significance level. Similarly, on average 4%, 45%, and 25% of irrigation water were saved at Chihona, Adibera, and Teleta blocks, respectively, when the farmer used the WFD irrigation scheduling method compared to the control. The difference was significant at a 5% significance level. The coefficient of variation is highest at the WFD treatment (32%) while for control and Chameleon sensor treatment, the coefficient of variation is almost 24%. Teleta blocks were less efficient in irrigation than the other two blocks. The difference is significant at a 5% significance level for the Chihona block. There was no significant difference between a block in the same treatment.

60

3.2 3.2.1

H. D. Tarekegn et al.

Wheat Yield Wheat Yield at Farm Level

The average wheat yield of the three blocks for the WFD field was 3533 kg/ha, 3167 kg/ha, and 2956 kg/ha for Chihona, Adibera, and Teleta blocks, respectively. Similarly, the average wheat yield of the three blocks for the Chameleon field was 3113 kg/ha, 3748 kg/ha, and 3604 kg/ha for the corresponding blocks. For the control field, the average yields were 2413 kg/ha, 2543 kg/ha, and 3408 kg/ha for Chihona, Adibera, and Teleta blocks, respectively, as shown in Fig. 5a–c. Additionally, boxplots of scheme control, WFD, and Chameleon treatment are shown in Fig. 6a–c, respectively. In this study, the yield for control and WFD was 2600 ± 800 kg/ha and 2600 ± 900 kg/ha [22]. According to Haileslassie [34], the wheat yield in Koga was 1600 kg/ha. In 2011 the wheat yield in the Amhara region was recorded as 1700 kg/ha, and the overall range in Ethiopia was 700–2800 kg/ha [36]. This study shows that the wheat yield is 2900 ± 1900 kg/ha and 3430 ± 1230 kg/ha for WFD and Chameleon treatment, respectively. In this study, the maximum yield is 4800 kg/ha in the WFD treatment. The boxplot of scheme wheat yield for different treatments is shown in Fig. 7, and wheat yield (kg) at scheme level is shown in Table 2.

3.3 3.3.1

Irrigation Water Productivity Irrigation Productivity at the Farm Level

The average irrigation water productivity for the control wheat field was 0.40 kg/m3, 0.51 kg/m3, and 1.0 kg/m3 for Chihona, Adibera, and Teleta blocks, respectively. The average irrigation water productivity for the Chameleon wheat field was 0.7 kg/m3, 0.87 kg/m3, and 1.04 kg/m3 for Chihona, Adibera, and Teleta blocks respectively. Similarly, the average irrigation water productivity for the WFD wheat field was 0.91 kg/m3, 0.78 kg/m3, and 0.86 kg/m3 for Chihona, Adibera, and Teleta blocks, respectively. Boxplots of Chihona, Adibera, and Teleta blocks water productivity in kg/m3 are shown in Fig. 8a–c, respectively. Similarly, the boxplots of scheme control, WFD, and Chameleon sensor water productivity in kg/m3 are shown in Fig. 9a–c, respectively. In the control group, 50% of the farmers had water productivities below 0.5 kg/m3, whereas in the WFD and Chameleon treatments, it covered only 11% and 6%, respectively. Water productivity greater than 1 kg/m3 for the control group covers only 11% while 44% and 28% for WFD and Chameleon treatments. The earlier study shows that in the Chameleon group, 10% of farmers had water productivities below 0.5 kg/m3. Water productivity exceeding 1 kg/m3 was reached by 10–20% of the farmers with irrigation scheduling technologies or information, while only 2% of the control group achieved the same results [28].

Evaluation of Three Irrigation Management Tools for Improving Crop. . . 5

Yield(103kg/ha)

4.5 4 3.5 3 2.5 2 1.5 1 Control

WFD

Chameleon

Treatments a) 5 4.5

Yield(103kg/ha)

4 3.5 3 2.5 2 1.5 1 0.5 Control

WFD

Chameleon

Treatments b)

5 4.5

Yield(103kg/ha)

4 3.5 3 2.5 2 1.5 1 0.5 Control

WFD

Chameleon

Treatments c)

Fig. 5 Boxplot of wheat yield for (a) Chihona block, (b) Adibera block, and (c) Teleta block

61

62

H. D. Tarekegn et al. 5 4.5

Yield(103kg/ha)

4 3.5 3 2.5 2 1.5 1 Chihona

Adibera

Teleta

Blocks a)

5.5 5

Yield(103kg/ha)

4.5 4 3.5 3 2.5 2 1.5 1 0.5 Chihona

Adibera

Teleta

Blocks b)

5 4.5

Yield(103kg/ha)

4 3.5 3 2.5 2 1.5 1 Chihona

Adibera

Teleta

Blocks c)

Fig. 6 Boxplot of scheme (a) control, (b) WFD, and (c) Chameleon treatment

Evaluation of Three Irrigation Management Tools for Improving Crop. . .

63

5.5 5

Yield(103kg/ha)

4.5 4 3.5 3 2.5 2 1.5 1 0.5 Control

WFD

Chameleon

Treatments

Fig. 7 Boxplot of scheme wheat yield for different treatments Table 2 Wheat yield (kg) at scheme level Treatment Control WFD Chameleon

3.3.2

Statistical descriptors Min Max Avg Scheme 2000 4400 2788 1000 4800 3219 2200 4667 3489

SD

CV%

770 1061 659

27.6 33.0 18.9

Irrigation Water Productivity at Scheme Level

The average water productivity obtained in this study was 0.64 kg/m3, 0.85 kg/m3, and 0.87 kg/m3 for control, WFD, and Chameleon sensors, respectively, as shown in Table 3. And this value was found in the range of water productivity determined by the other investigations (0.72 kg/m3 for WFD, 0.81 kg/m3 for Chameleon, and 0.51 kg/m3 for control farmers) in the Koga irrigation scheme [28]. The gains in water productivity for WFD and Chameleon-guided farmers were provided with the introduction of on-demand delivery of irrigation water supplies which resulted in maximum yield as compared with control farmers.

3.4 3.4.1

Effect of WFD and Chameleon Sensor at Block and Scheme Level At the Block Level

If all farmers in the Chihona block use Chameleon sensor technology as a treatment method, then the amount of irrigation water saved by the Chameleon sensor field

H. D. Tarekegn et al.

64 1.4 Water productivity(kg/m3)

1.2 1 0.8 0.6 0.4 0.2 0 Control

WFD

Chameleon

Treatments a)

Water productivity(kg/m3)

1.4 1.2 1 0.8 0.6 0.4 0.2 0 Control

WFD

Chameleon

Treatments b) 2

Water productivity(kg/m3)

1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0 Control

WFD Treatments c)

Chameleon

Fig. 8 Boxplot of (a) Chihona, (b) Adibera, and (c) Teleta block water productivity (kg/m3)

would be 86,714 m3 (45%), and with this amount of water, 456 ha of the additional wheat field could be grown. Similarly, if all farmers in this block use WFD as a treatment method, then the amount of irrigation water saved by the WFD field would

Evaluation of Three Irrigation Management Tools for Improving Crop. . .

65

2

Water productivity(kg/m3)

1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0 Chihona

Adibera Blocks

Teleta a)

1.4

Water productivity(kg/m3)

1.2 1 0.8 0.6 0.4 0.2 0 Chihona

Adibera

Teleta

Blocks b)

Water productivity(kg/m3)

1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0 Chihona

Adibera

Teleta

Blocks c)

Fig. 9 Boxplot of scheme (a) control, (b) WFD, and (c) Chameleon sensor water productivity (kg/m3)

66

H. D. Tarekegn et al.

Table 3 Wheat water productivity (kg/m3) at scheme level

Treatment Control WFD Chameleon

Statistical descriptors Min Max Avg Scheme 0.32 1.78 0.64 0.18 1.288 0.85 0.50 1.448 0.87

SD

CV%

0.36 0.33 0.29

56.88 39.16 32.83

Table 4 Effect of WFD and Chameleon sensor at block and scheme level

Block Chihona Adibera Teleta Scheme

Percentage increment WFD Chameleon 56 45 19 4 6 25 26 11

An additional area to be irrigated WFD Chameleon 724 ha 456 ha 68 ha 13 ha 36 ha 134 ha 2080 ha 690 ha

The additional volume of water (103 m3) WFD Chameleon 109.02 86.71 61.80 13.74 6.37 28.47 491.17 197.96

be 109,021 m3 (56%), and with this amount of water, 724 ha of the additional wheat field could be grown. In the same fashion as the Chihona block, if all farmers in the Adibera block use the Chameleon sensor as an irrigation management method, then the amount of irrigation water saved would be 13,740 m3 (4%), and with this amount of water, 13 ha of the additional wheat field could be grown by Chameleon sensor irrigation technique. Similarly, if all farmers in the Adibera block use WFD as an irrigation management method, then the amount of irrigation water saved by the WFD field would be 61,800 m3 (19%), and with this amount of water, 68 ha of the additional wheat field could be grown by WFD irrigation technique. Again, for the Teleta block, if all farmers use Chameleon sensor technology as an irrigation management method, then the amount of irrigation water saved would be 28,468 m3 (25%), and with this amount of water, 134 ha of the additional wheat field could be grown by Chameleon sensor irrigation technique. Similarly, if all farmers use WFD as a treatment method, then the amount of irrigation water saved by the WFD field would be 6373 m3 (6%), and with this amount of water, 36 ha of the additional wheat field could be grown as it was for the Chameleon sensor. 3.4.2

At the Scheme Level

If all farmers in the Koga irrigation scheme use the Chameleon sensor as an irrigation management method, then the amount of irrigation water saved would be 197,961 m3 (11%), and with this amount of water, 690 ha of the additional wheat field could be grown by Chameleon sensor irrigation technique. Similarly, if all farmers use WFD as an irrigation management method, then the amount of irrigation water saved by the WFD field would be 491,167 m3 (26%), and with this amount of water, 2080 ha of the additional wheat field could be grown by WFD irrigation technique. The effect of WFD and Chameleon sensor at block and scheme level is shown in Table 4.

Evaluation of Three Irrigation Management Tools for Improving Crop. . .

67

4 Conclusion This study investigated the effects of different irrigation treatments on the yield, amount of water applied (irrigation depth), and irrigation water productivity. The study shows that using irrigation water management tools (WFD and Chameleon irrigation scheduling) at the farm level results in a significant 26% and 11% significant level ( p < 0.05) saving of irrigation water, respectively. These increments can create an opportunity to irrigate an additional irrigation area of 2081 ha and 691 ha with WFD and Chameleon, respectively. However, the saving differs among the various blocks at the farm level. Thus, these instruments can solve the water shortage and avoid major problems and conflicts between upstream and downstream irrigators. The average yield in the scheme is 2790 kg/ha, 3220 kg/ha, and 3490 kg/ha for control, WFD, and Chameleon sensor, respectively, and hence there is no significant difference at a 5% significance level in yield. In the same fashion, the average water productivity of the scheme level was 0.64 kg/m3, 0.85 kg/m3, and 0.87 kg/m3, respectively, which is significant at a 5% significance level. This indicates that water productivity was higher in WFD and Chameleon sensor groups as compared to the control group, and for this, there was a statistically significant difference between them. The higher the water productivity, the lower the volume of water applied to keep the yield constant. In conclusion, the wetting front detector and Chameleon can save a substantial amount of water without yield reduction. The reduction in irrigation water usage did not show any negative effect on yield. It could be concluded that farmers with irrigation scheduling technologies had significantly higher water productivity compared to the control group. Acknowledgements This research was carried out as a collaborative partnership between the International Water Management Institute (IWMI) and Bahir Dar Institute of Technology-Bahir Dar University as part of the WaPOR project (https://www.fao.org/in-action/remote-sensing/ publications/wapor-publications/ru/) accessed on 21 December 2021. We acknowledge the support given by institutions to our data collection efforts and successful implementation of this research work.

References 1. Abate, G.T., de Brauw, A., Minot, N., Tanguy, B.: The impact of the use of new technologies on farmers’ wheat yield in Ethiopia: evidence from a randomized controlled trial. IFPRIDiscussion Papers (2015) 2. CSA, E. Population projection of Ethiopia for all regions at wereda level from 2014–2017. Central Statistical Agency of Ethiopia 1, 167–176 (2013) 3. Ali, M.H., Talukder, M.S.U.: Increasing water productivity in crop production—a synthesis. Agric. Water Manag. 95(11), 1201–1213 (2008)

68

H. D. Tarekegn et al.

4. Awulachew, S.B., Merrey, D., Kamara, A., Van Koppen, B., Penning de Vries, F., Boelee, E.: Experiences and Opportunities for Promoting Small-Scale/Micro Irrigation and Rainwater Harvesting for Food Security in Ethiopia, vol. 98. IWMI (2005) 5. Awulachew, S.B.: Irrigation potential in Ethiopia: constraints and opportunities for enhancing the system. Gates Open Res. 3, 22 (2019) 6. Bakker, M.: Multiple Uses of Water in Irrigated Areas: A Case Study from Sri Lanka, vol. 8. IWMI (1999) 7. Berhane, G., Dereje, M., Hoddinott, J., Koru, B., Nisrane, F., Tadesse, F., Taffesse, A.S., Worku, I., Yohannes, Y.: Agricultural Growth Program (AGP) of Ethiopia–Baseline Report 2011. ESSP/EDRI Report. International Food Policy Research Institute, Addis Ababa (2013) 8. Beshaw, K.: Evaluation of alternate and surge flow furrow irrigation methods for onion production at humbo, southern Ethiopia. MSc. thesis, Haramaya University, Haramaya (2011) 9. Demelash, N.: Deficit irrigation scheduling for potato production in North Gondar, Ethiopia. Afr. J. Agric. Res. 8(11), 1144–1154 (2013) 10. Dorosh, P., Rashid, S.: Food and Agriculture in Ethiopia: Progress and Policy Challenges. University of Pennsylvania Press, Philadelphia (2013) 11. Drechsel, P., Heffer, P., Magen, H., Mikkelsen, R., Wichelns, D.: Managing Water and Fertilizer for Sustainable Agricultural Intensification. International Fertilizer Industry Association (IFA), International Water Management Institute (IWMI), International Plant Nutrition Institute (IPNI), and International Potash Institute (IPI), Paris (2015) 12. Endrie, B.: Evaluation of Water Saving, Yield, and Water Productivity of Potato (Solanum Tuberosum L.) by Using Wetting Front Detector and Different Levels of Fertilizer Application at Koga Irrigation Scheme in Amhara Region, Ethiopia. Haramaya University, Dire Dawa (2017) 13. Etissa, E., Dechassa, N., Alamirew, T., Alemayehu, Y., Desalegne, L.: Irrigation water management practices in smallholder vegetable crops production: the case of the central Rift Valley of Ethiopia. Sci. Technol. Arts Res. J. 3(1), 74–83 (2014) 14. Gebrehiwot, S.G., Gärdenäs, A.I., Bewket, W., Seibert, J., Ilstedt, U., Bishop, K.: The longterm hydrology of East Africa’s water tower: statistical change detection in the watersheds of the Abbay Basin. Reg. Environ. Chang. 14(1), 321–331 (2014) 15. Hordofa, T., Menkir, M., Awulachew, S. B., & Erkossa, T.: Irrigation and rain-fed crop production system in Ethiopia, impact of irrigation on poverty and environment in Ethiopia, 27–36 (2008). 16. Iglesias, A., Garrote, L.: Adaptation strategies for agricultural water management under climate change in Europe. Agric. Water Manag. 155, 113–124 (2015) 17. MacDonald, M.: Koga irrigation and watershed management project. Interim Report (2004) 18. Marx, S.: Large-scale irrigation in the Blue Nile Basin: chances and obstacles in implementing farmers’ self-management: a case study of the Koga irrigation and watershed management project in Amhara Region. IWMI, Ethiopia (2011) 19. Molden, D., Oweis, T., Steduto, P., Bindraban, P., Hanjra, M.A., Kijne, J.: Improving agricultural water productivity: between optimism and caution. Agric. Water Manag. 97(4), 528–535 (2010) 20. Palada, M., Bhattarai, S., Wu, D.L., Roberts, M., Bhattarai, M., Kimsan, R., Midmore, D.: More Crop Per Drop: Using Simple Drip Irrigation Systems for Small-Scale Vegetable Production (Issue BOOK). AVRDC-The World Vegetable Center (2011) 21. Sahin, U., Kiziloglu, F.M., Angin, I.: Changes in some quality properties after different storage periods of potato tubers grown under well and deficit irrigation conditions. Bulgarian J. Agr. Sci. 12(5), 673 (2006) 22. Schmitter, P.S., Haileslassie, A., Dessalegn, Y., Chali, A., Langan, S.J., Barron, J.: Improving on-farm water management by introducing wetting-front detector tools to smallholder farms in Ethiopia (2017) 23. Shahnazari, A., Liu, F., Andersen, M.N., Jacobsen, S.-E., Jensen, C.R.: Effects of partial rootzone drying on yield, tuber size, and water use efficiency in potato under field conditions. Field Crop Res. 100(1), 117–124 (2007)

Evaluation of Three Irrigation Management Tools for Improving Crop. . .

69

24. Singh, R., Kundu, D.K., Bandyopadhyay, K.K.: Enhancing agricultural productivity through enhanced water use efficiency. J. Agric. Phys. 10(2), 1–15 (2010) 25. Sorando, R., Comín, F.A., Jiménez, J.J., Sánchez-Pérez, J.M., Sauvage, S.: Water resources and nitrate discharges about agricultural land uses in an intensively irrigated watershed. Sci. Total Environ. 659, 1293–1306 (2019) 26. Stirzaker, R.J., Hutchinson, P.A., Mosena, M.L.: A new way for small farm irrigators to save water. 6th International Micro-Irrigation Congress (Micro 2000), Cape Town, South Africa, pp. 1–9, 22–27 Oct 2000 27. Strelkoff, T.S., Clemmens, A.J., El-Ansary, M., Awad, M.: Surface-irrigation evaluation models: application to level basins in Egypt. Tran. ASAE. 42(4), 1027 (1999) 28. Svedberg, E.: Impact on Yield and Water Productivity of Wheat by Access to Irrigation Scheduling Technologies in Koga Irrigation Scheme, Ethiopia. Uppsala University, Uppsala (2019) 29. Tesfaye, M., Fassil, K.: Suitability of Koga watershed for irrigated sugarcane and onion production in the blue nile basin, Ethiopia. J. Dry Lands. 4(2), 325–332 (2011) 30. Tsegaye, D.: Profitability of contractual bread wheat seed production in Mecha district of Amhara region, Ethiopia. J. Cent. Eur. Agric. 13(1) (2012) 31. Wheater, H., Evans, E.: Land use, water management, and future flood risk. Land Use Policy. 26, S251–S264 (2009) 32. Zotarelli, L., Dukes, M.D., Romero, C.C., Migliaccio, K.W., Morgan, K.T.: Step-by-step Calculation of the Penman-Monteith Evapotranspiration (FAO-56 Method). UF/IFAS Extension. The University of Florida, Gainesville (2016) 33. Gebrehiwot, S. G., Taye, A., & Bishop, K. Forest cover and stream flow in a headwater of the Blue Nile: complementing observational data analysis with community perception. Ambio, 39, 284–294 (2010). 34. Haileslassie, A., Agide, Z., Erkossa, T., Hoekstra, D., Schmitter, P. S., & Langan, S. J. On-farm smallholder irrigation performance in Ethiopia: From water use efficiency to equity and sustainability. International Livestock Research Institute. (2016) 35. Kijne, J. W. Water productivity under saline conditions. In Water productivity in agriculture: limits and opportunities for improvement (pp. 89-102). Wallingford UK: CABI Publishing. (2003). 36. ETH (The Federal Democratic Republic of Ethiopia Central Statistical Agency), Report on Area and Production of Major Crops. Agricultural sample survey 2011/2012. Vol I. The Federal Democratic Republic of Ethiopia Central Statistical Agency (2012).

Design and Numerical Analysis of a Sorghum Reaper Machine Adino Amare Kassie, Hailu Shimels Gebremedhen, and Hailemichael Solomon Addisu

Contents 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Present Status of Sorghum Harvesting Methods in Ethiopia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1 Primary Data Collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Manual Harvesting States in Raya Kobo and Azebo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1 Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 Analytical Analysis of the Key Part for the Reaper of Sorghum Stalks . . . . . . . . . . . . . 3.3.0 Design and Selection of Power Transmission System of a Machine . . . . . . . . 4 Result and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1 Analytical Analysis and Numerical Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Design Specification of Root-Stalk Sorghum Reaper Machine . . . . . . . . . . . . . . . . . . . . . . 4.3 Comparison of Manual and Root-Stalk Sorghum Harvesting Results . . . . . . . . . . . . . . . 5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Recommendation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

72 73 74 75 77 77 78 79 93 99 100 103 105 106 106 107

A. A. Kassie (✉) · H. S. Addisu School of Mechanical and Industrial Engineering, Institute of Technology, Dire Dawa University, Dire Dawa, Ethiopia H. S. Gebremedhen College of Electrical and Mechanical Engineering, Addis Ababa Science and Technology University, Addis Ababa, Ethiopia © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 K. Mequanint et al. (eds.), Advancement of Science and Technology in Sustainable Manufacturing and Process Engineering, Green Energy and Technology, https://doi.org/10.1007/978-3-031-41173-1_5

71

72

A. A. Kassie et al.

1 Introduction Ethiopia has a population of about 110 million people, with over 85% relying on rain-fed agricultural and livestock production [1]. Agriculture, which involves important sources in each area of Ethiopia, has a crucial role in gross national food security and socio-economic development. Geographically, Ethiopia has a diverse range of agro-climatic zones. According to the Ethiopian government program, agrarians occasionally divide the climatic conditions into three groups, which are Dega (cold climatic), Woina Dega (temperate climatic), and Qolla (hot climatic) (low land; or warm climatic). This diversity makes it a favorable region for growing a variety of crops. Ethiopia’s agriculture grain is complex, including substantial differences in grains grown across the country’s regions. Five main cereals (teff, wheat, maize, sorghum, and barley) make up most of Ethiopia’s agricultural yields, contributing an interesting proportion to the economy. Of the five cereals sorghum (Sorghum bicolor) is abundant in different regions of Ethiopia [2]. Sorghum grain is one of the world’s most significant food grains, ranking fifth in terms of covered field areas and production [3]. Sorghum output continues to be highest in the USA, accounting for 25% of total global production, followed by India (21.5%), Mexico (11%), and China (9%) [4], although production has increased in Africa and Europe. Ethiopia is Africa’s second-largest sorghum producer, after Sudan [5]. The crop is largely planted as a food crop for making leavened bread (injera) and other food products, as well as for preparing locally manufactured alcoholic beverages (for example, Tela and Areke). There are three major agro-ecologies in which sorghum crop is grown. It is the most important crop in the arid lowlands, accounting for more than 60% of cultivated land [6]. Using sorghum crop yields, three prominent production zones (dry and wet lowlands) have been identified in North Wollo, including those in the Amhara (Raya-kobo), Oromia (Mieso, Babile, and Chiro), and Benishangul Gumuz (Assosa) regions, which are well-known to residents of Ethiopia [7]. Harvesting is widely recognized as a critical and significant process for sorghum crop yield, quality, and production cost. In Ethiopian locations, all sorghum is still harvested by hand with a sickle. Reaping, laying, gathering, transporting, and stacking the cut crop are all part of the sorghum-harvesting process. Seed stage sorghum harvesting by hand can be a time-killer and an expensive process. Laborers are migrating to industries as a result of industrialization in many emerging countries, resulting in agricultural manpower shortages [8]. Furthermore, a survey of planted sorghum lands indicated that overall post-harvest losses of sorghum crops at farm level in Raya Kobo residences Girana agricultural institution Woreda ranged from 25% to 32%. Harvesting is normally delayed during the peak reaping period of matured sorghum crops owing to labor shortages, resulting in a shattering loss of subsequent crops. Mechanized agriculture makes it easier to boost the productivity and profitability of farming operations by maximizing the use of farm power and machinery. Finally, the goals of this research have been set as a guideline for designing a modern sorghum reaper machine to reduce drudgery and optimize stalk-based cereal crop outputs in Ethiopian regions for future harvesting.

Design and Numerical Analysis of a Sorghum Reaper Machine

73

2 Present Status of Sorghum Harvesting Methods in Ethiopia In the agricultural operations in Ethiopia, harvesting plays an important role in the production cost of teff, sorghum, and maize. In most of the regions of Ethiopia, sorghum is manually harvested by hand with wooden sickles. There are two main reaping techniques for harvesting sorghum crops as per the climate conditions. The first reaping technique is used when the mature crop is harvested, and it involves removing the stalk’s ear-heads first, then the noncrop stalks. Then, after a week, noncrop sorghum stalks are harvested for cattle feed, and temporary booths are built for social events such as weddings. The stalks are chopped and stored for use as dry season animal fodder, house fence, and firewood, particularly in Raya, Kobo, and Raya Azebo homes. There would be two major drawbacks if the sorghum crop does not dry out correctly with its stalk after reaping. Because the harvested crop is stored in the underground of wells, according to a Raya Kobo, Girana Agricultural Research Report, the crop rots swiftly during that time, and sorghum injera is tested as bread when consumed. In addition, some peasants use another harvesting technique in selected highland areas. They gather the sorghum seeds using a basket after shattering the sorghum stalk. This strategy, on the other hand, results in grain loss and the improper use of stalks for another use; in terms of benefits, most Ethiopian farmers do not prefer this practice (Fig. 1).

Fig. 1 Tall varieties of sorghum stacked stalks in the residence of Raya Kobo fields

74

A. A. Kassie et al.

Sorghum stems have grown tall enough that they no longer stay erect for long periods of time, and when the stalk of the sorghum stalk dries out, unseasonal winds can easily destroy it. As a result of the hand harvesting, around 25–32% of the sorghum grains are discarded. It is customary for women in the Raya district to collect the scattered crop by carrying the basket on their backs for a long time through the farm of the whole sorghum land. According to Melese Melku’s interview response (Head Officer of Raya-Kobo agricultural research), there were 59,763 farmers involved in sorghum harvesting from 2020 to 2021. Based on the promising findings of this study, the farm is extremely productive, but the hand labor effort is extremely ineffective. At present, most of the developing countries are involved with industrial development, especially in mechanization of rice and wheat harvesting [9, 10]. This is because rural laborers are migrating to higher-paid jobs in the industry for an entire year. It is a serious challenge to resist the pace of food production for the exponential growth of the Ethiopian population in the next few years. To obtain knowledge as input data, this research needs to interview farmers and research officers in Raya Kobo sorghum regions about the manual harvesting techniques used.

2.1

Primary Data Collection

Source types were used as primary data in the North Wollo, Amhara, Ethiopia and Raya Kobo districts. Grain yield and farmers’ perceptions of sorghum harvesting, average sorghum height, and the number of men and women farmers who participate in sorghum cultivation each year are among the major statistics. Between 2010 and 2021, a total of 16,000 and 30,771 hectares of sorghum grain cultivation farms were documented in Raya Azebo and Raya Kobo respectively. Stalk samples were taken, with moisture content ranging from 15% to 25% (Fig. 2).

Fig. 2 Measuring of sorghum stalk in the north wollo-waja, farmers’ land

Design and Numerical Analysis of a Sorghum Reaper Machine

75

Table 1 Measured values of the length of seed-stage sorghum stalks obtained from primary data Sample intervention residences of sorghum lands Merto Gederu Dydagal Araqote Golesha Zoble Aradum Gobye Mendefera Girana

Maximum height of sorghum (m) 3.20 3.10 3.30 3.20 4.20 2.90 3.00 2.80 3.40 3.00

Minimum height of sorghum (m) 2.40 2.80 2.00 2.60 3.10 2.30 2.70 2.20 2.95 2.40

The average height of sorghum (m) 2.8 2.95 2.65 2.9 3.65 2.6 2.85 2.5 3.175 2.7

Measurement of Sample Sorghum Heights The sorghum stalk was measured with a meter and the average diameter of ten sample stalks of sorghum was recorded as 0.021 mm (Table 1). The cut cross-sectional areas of the sorghum stalk 90° circular axis were calculated from the major diameter and length samples using the following geometric formula: As, max =

π × d 2 max , 4

ð1Þ

where, As, maxis the maximum area of the stalk, dmax is the maximum diameter of the stalk.

2.2

Manual Harvesting States in Raya Kobo and Azebo

The mechanical qualities of sorghum stalks, the effects of hand harvesting, and the losses caused by manual harvesting per season were studied in the first study to meet the goal of this research. As a result, a single sorghum field from one of Ethiopia’s regions has been chosen for the study. The core basic primary data were from interviews with the local farmers of South Wollo, Raya Kobo, and Alamata who have sorghum grain-cultivated landholdings in which we enquired about the harvesting trends of matured sorghum grains (Table 2). According to the data obtained, 21 people were required to reap 1 ha of sorghum fields (Table 3). ETB = Ethiopian birr Based on the information obtained from a literature review of primary data 316,305,897.6 ETB is required for reaping 47,069.33 ha of sorghum-cultivated

76

A. A. Kassie et al.

Table 2 Primary data report in Raya Kobo Descriptions in number 59,763

Items Total number of farmers engaged in a sorghum field The average number of total hectares present, ha

47,069.33

Reference Girana Research Office from 2017 to 2020 Report from Raya-Kobo, Girana Research Office for 2017–2020

Table 3 Report on the manual harvesting cost of 1 ha Item Costs expended for 21 people in 1 ha

Total labor cost of reaping 1 ha

Description in numbers Lunch 45 × 21 Dinner 55 × 21 Daily labor cost 220 × 21 6720 ETB

Reference Direct interview Direct interview Direct interview

35 Loss of sorghum crop per year 30

% loss per year

25 20 15 10 5 0 2012

2013

2014

2015 2016 2017 Year (2012–2019)

2018

2019

Fig. 3 Manual harvesting losses of crops from 2012 to 2019 in fields. (Report from the Raya Kobo, Girana Research Office for 2012–2019)

fields per year in both Raya Kobo and Raya Azebo regions. Sorghum crops become overripe, causing crops to fail on farmed land. Such issues have been present for several years. As primary data, losses of 5 consecutive years were documented, as shown in Fig. 3.

Design and Numerical Analysis of a Sorghum Reaper Machine

77

3 Materials and Methods Scientific analysis can address such vexing issues as why, where, and what methods the author utilized to achieve the principal goal of the research. The type of material developed for each type of reaper machine is also examined in detail.

3.1

Materials

The materials used in the various designs have been carefully chosen to ensure efficient operation and a long machine life. While selecting the machine components the following important factors were considered: maintainability and reliability of materials, strength, weight, and size of the materials, friction and wear resistance of materials, and material availability on Ethiopia’s various markets. Mild steel is commonly used in bridges, buildings, and machine shafts. The world’s most popular form of steel has a wide range of applications, is used on a daily basis, and it can be cut and adjusted as needed [11]. High-carbon steel (HCS) is an engineering metal alloy material that is constructed from iron and carbon elements. The internal structure of HCS is formed from dual-phase microstructures of metastable retained austenite and martensitic, which contribute to the material’s high strength and exceptional abrasion resistance of the material and addresses the hardness of each phase and the overall microstructure in HCS with or without a post-tempering heat treatment [12]. Rubber is elastic, but at lower temperatures it becomes hard [13]. Engineering sheets of steel that have been zinc-coated are known as galvanized iron sheets (Table 4).

Table 4 Engineering material used for reaper machine components The sorghum reaper machine components Frame of a reaper Stalk divider of the reaper Star wheels of the reaper Reaping bar (cutter bar) Sprockets of the reaper Chain of the reaper Flat and v-belts of the reaper Shafts and pulleys

Materials used Galvanized milled steel Galvanized iron sheet metal Polycarbonate plastic High carbon steel Milled steel High carbon steel Rubber Milled steel

78

A. A. Kassie et al.

Identify problem

Modeling reaping system Modeling the conveying system

Replacement of sickle by reaper machine

Data collection about size and properties of matured sorghum stalks

Experimental testing on seed stage of sorghum stalk sample to identify mechanical properties of stalk

Conceptual development and selection based on physical structure (height and size) of sorghum stalks

Modeling the driving system Compare analytical and numerical analysis using finite element method

If components are not safe

If components are safe

Ready to manufacture

Synthesized the desired motion of the reaper machine

Material selection for each member of reaper machine Design the key parts of a reaper machine system

Modeling the whole assembly system of a reaper machine based on design specification

Fig. 4 General flow charts of methods

3.2

Methods

It is easy to recognize the needs of sorghum farmers, such as how much they need to reduce their labor force, based on information from the literature, and their desire to focus on modernizing the country’s agricultural system within the next few years. As a result, this study chose to create a sorghum reaper machine utilizing the following method (Fig. 4). Experimental Preparation for Testing Seed Stage of Sorghum Stalk The main goal of this experiment was to acquire some basic data on the force and energy needs for types of stalk deformation when reaper machines were used. Shear and bending were calculated using a universal testing machine (UTM; Figs. 5 and 6). The length of the stalk was 75 mm, the diameter was 21 mm, and the samples were taken based on standard preparation of hump stalks. Using the universal testing equipment, the following results were obtained from a seed-stage sorghum stalk sample. The stalk’s maximum load resistance is around 10.14 KN with a maximum tensile strength of 29.276 MPa. To find the shearing behavior a stalk equation was applied [11].

Design and Numerical Analysis of a Sorghum Reaper Machine

79

Fig. 5 Specimen of a sorghum stalk for tensile testing

Fig. 6 Specimen of a sorghum stalk for testing bending

σs =

F , 2×A

ð2Þ

where σ s = maximum shear strength of the stalk, F = resistive force, N, A = area of the stalk, m2.

3.3

Analytical Analysis of the Key Part for the Reaper of Sorghum Stalks

Analytical thinking requires researchers to learn how to use the language of physics and mathematics to find a correct solution to a well-defined problem in a given knowledge domain [12, 13]. Inertial forces, acceleration and deceleration, fatigue,

80

A. A. Kassie et al.

load and structural design, bearing loads, and so on, are all covered [14, 15]. Based on the authors’ knowledge, this study’s endeavor to improve the sorghum reaper machine could be safe from failure. Selection of the Cutting Blade Cutters only perform cutting off of the root stalk of crop plants. The reaping cutter blade and fixed noncutter frame are used in the design-cutting device. Here are the selection criteria for cutter blade dimensions based on the Standard Size parameter [12, 16, 17]. • The clamping angle obtained between the stalk and sharp edge of the cutter α varies from 22 to 310, with the height of the cutter blade section should not exceed 120–130 mm for most of the stalk-based cereals. • To avoid cutting off the stalks, the cutter section should be serrated. The pitch of the serrated cutter blade should not exceed 1–2.3 mm. • Clearance between the cutter bar and fixed cutter plates is settled between 0.5 and 1.0 mm to give the best efficiency. • The width of a reciprocating type cutter bar standard cutter blade does not exceed 76–80 mm. As most Ethiopian farmers sow crops of sorghum on the ground randomly, the root stalk of the plant grows without a line. Fodder reaper investigators use 1.30–1.50 m of the width of the cutter [18]. Therefore, owing to the tall height of the stalk, this research determined the length of the reaping bar to be 1.30 m. Then, the number of cutter blades of the reaper machine can be calculated as follows: Bc =

Lbr , Lw

ð3Þ

where Bc is the number of cutter blades, Lbr is the length of the reaping or cutter bar, Lw is the width of one section blade. The increase in cutting angle may reduce the cutting resistance, but too large a cutting angle would influence the clamping stability [16] (Fig. 7). H=

Vm × π 30 × Vm = , ϖ n

where H is the advance distance, m, Vm is the orward speed of a reaper machine, m/s, ω is the angular speed of a crank, rad/s, n is the rotational speed of the crank, rpm, where S is the cutter blade’s stroke, m, H is the advance distance, m, H is the height of the movable edge blade, C is the width of the fixed blade, m.

ð4Þ

Design and Numerical Analysis of a Sorghum Reaper Machine

81

Fig. 7 Cutting diagram of a single movable blade cutter

When the cutter finishes a stroke from left to right, the reaper machine will move forward by advanced distance H. Cutting areas covered by the right edge of the cutter blade include the letters A, B, C, and D. Then, the cutter blade continues to finish a stroke from right to left, and the machine moves forward by an advanced distance once again. Cutting areas covered by the left edge of the cutter include F, E, H, and G. It is obvious that if the repeated cutting area is too large, repeated cutting would lead to a waste of power; if the failure to cut an area or the blank area is too large, some sorghum stalks may be pushed over and lead to miss-cutting. When the advanced distance increases, the repeated cutting area increases, whereas the failure to cut the area will decrease, or vice versa. The relation between the speed of the cutter blade and the forward speed of a reaper machine can be expressed by the cutting speed ratio [19]: R=

Vf S × n=30 S = = , H Vm H × n=30

where R is the cutting speed ratio,

ð5Þ

82

A. A. Kassie et al.

Vf is the average speed of the cutter blades, m/s, S is the cutting stroke of a single movable blades cutter, m. If the cutting speed ratio, R, is too small, it leads to missing cutting and unstable cutting quality of sorghum stalks; if the cutting speed ratio R is too large, repeated cutting of stalk may occur, then it results in power waste [20–23]. According to the planting pattern and planting density of the sorghum stalk and the analysis of the cutting diagram of the single reciprocating blades cutter, the propelling forward speed of the sorghum reaper is preliminarily selected. For a checkup, the comparison of the forward speed of the reaper with reality, the normal walking speed of a human ranges from 0.7 to 0.8 m/s and the average walking speed of an adult man is about 0.9144–1.219 m/s [24]. Therefore, 0.75–1.3 m/s is taken as the forwarding speed, to reduce miss-cutting of the sorghum stalk and the waste of power consumption during the reaping time. the speed cutting ratio ranges between 1.3 and 1.4 m/s since the design of rack angle cutter blades ranges from 220 to 310 [16]. Therefore, calculating the rotational speed of a crankshaft, when Vf = 0.975 m/s and S, 76 mm, when Vm, 0.75 m/s and R, 1.3. V f = V m × R0:75 × 1:3 = 0:975 m=s, when V m = 1:3 m=s and R = 1:4 V f = 1:3 × 1:4 = 1:82 m=s,

V f = ðS × n=30Þ, → n = V f × 30=S when V f = 0:975 m=s and S = 76 mmn = ð0:975 × 30=0:076Þ = 384:868 ffi 400rpmwhen V f = 1:82 m=s and S = 76 mm, n = ð1:82 × 30=0:076Þ = 718:42 ffi 750rpm,

ð6Þ

ð7Þ

Hence, the rotational speed of the root stalk sorghum reaper crank is about 400–750 rpm. Analysis of Reaping Power for Seed-Stage Sorghum Stalks The average diameter of seed-stage sorghum stalk was 21 mm, the shear strength of the sorghum stalk that tested from the UTM was 9.232 MPa, according to this research design the number of cutter blade sections was calculated as 17, and based on Eq. (1) the area of the sorghum stalk was found to be: As, max =

π × d 2 max π × 0:02114792 = = 0:00035m2 4 4 σ s × As

Fr =

9=:232 × 106 × pa × 0:000335m2 3231:2 Fr = N ffi 3:23 kN:

ð8Þ

The area under the curve of the load and the displacement that the shear test performed in the cross-section of sorghum stalk is equal to the energy required to cut a single sorghum stalk [25].

Design and Numerical Analysis of a Sorghum Reaper Machine

Er = Rf × F r × AS , F Rf = P Fa

83

ð9Þ

where Eris the energy needed to reap one single sorghum stalk, FP is the peak force under load vs. distance curve in the UTM, is the average cutting force load versus displacement curve in the UTM, Rf is the reaping average to peak force ratio, Fr is the reaping force for a single stalk of sorghum. Using Eq. (9), and the force versus displacement curve on the UTM, find the reaping average to peak force ratio. Rf =

6:36KN = 0:627: 10:14KN

The length of the cutter stroke (S), which passed through a single sorghum stalk, is equal to the diameter of a stalk. Then, the sample average diameter of sorghum is about 0.021 m. Er = 0:627 × 3:2 kN × 0:021 m = 42:134 N:m Pr = Er × f cr ,

ð10Þ

where fcris the reaping frequency of the cutter blade in one step of the forwarding speed, Pr is the reaping power for one single stalk of sorghum. ω = 2π × f cr ,

ð11Þ

where n and ω are the rotational speed and angular speed of a crank respectively. Here, from earlier analysis the rotational speed of a reaper cutter blade crank ranges from 400 to 750 rpm. Let us, take the maximum rotational speed of the crank n, which is 750 rpm. 750 = 78:5398 rad=s 60 78:5398 = 12:5 hz f cr = 2π Nm Pr = 42:134 × 12:5 = 526:675 = 526:67 W: s ω = 2π ×

Therefore, the power that is needed to reap one single stalk of seed-stage sorghum is 526.675 W.As there are 17 cutter blades in a reaper, let us assume that 17 stalks are reaped in one step of forward speed. The total power needed for reaping 17 stalks in one step of the forwarding speed would be:

84

A. A. Kassie et al.

Pr = 17 × 526:675 = 8:9535 kw£12 HP: Design of the Star Wheel and Bunch Sorghum Stalk Divider The lugged flat belt conveyer transports the cut sorghum stalk to the right side of the reaper with the help of the star wheel for easy bundling. Some basic enhancements were made to the current rice star wheel when constructing the sorghum stalk star wheel. • The horizontal side conveying the velocity of the star wheel and the lugged flatbelt should be greater than or equal to the forward speed of the reaper machine. • It is acceptable to be 12° < α > 22°, as the star wheel is involved with the side of the angle for rice reapers; make sure that this sorghum reaper is 45° < α < 65°. The above ideal improvements not only increased the conveyer’s performance and transport reliability but also significantly decreased metal consumption. The desired horizontal inclination of rice and wheat crop star wheels empirically arrived at 45°–65° [25]. The minimum required speed of the star wheel is found to be as Vs =

Vm , cos α

ð12Þ

where, Vs is the average star wheel velocity, Vm is the machine forward velocity, α is the angle of inclination of the star wheel. Thus, for a 45–65° angle the above expression simplifies as the following (Fig. 8): V s > 1:315 × V m > 1:9 × V m :

ð13Þ

As the height of the sorghum stalk is an average of 3.17 m, then to protect the free inlet of the sorghum stalk from the cutter blade before reaping, the star wheel design inclination angle ranged between 450 ≤ α ≤ 650. Therefore, the star wheel velocity is ranges from 1.06 to 3.076 m/s (Table 5). Fig. 8 Star wheel inclination for the root stalk of the sorghum reaper

Design and Numerical Analysis of a Sorghum Reaper Machine

85

Table 5 Standard design-selected star wheel specification [26–28] Particular Inside radius of the star wheel, Ri Addendum circle radius of the star wheel, Ra Pitch circle radius of the star wheel, Rp Dedendum circle radius of the star wheel, Rd Thickness of the star wheel, ts Material of the star wheel

Specification 10 mm 150 mm 115 mm 75 mm 8 mm Plastic

The total number of star wheel =

length of cutter bar : crop divider size

Then, six star wheels can be assembled in this sorghum reaper machine. The star wheels that are driven by the flat-belt conveyor lug have a linear speed at the peak equal to the conveyor v-belt, and the star wheel circulates with its axis; therefore, the angular speed of the star wheel can be calculated as follows: V s = r a × ωs ,

ð14Þ

where Vsis the linear speed of the star wheel at the peak, m/s, ra is the addendum circle radius of the star wheel. The length of the star wheel lug can be calculated from the relation of its star wheel addendum and dedendum radiuses (Fig. 9): Ls = Ra - Rd = 150 - 75 = 75 mm,

ð15Þ

where Lsis the length of the star wheel lug, mm. Design of Cut Bunch Sorghum Stalk Fla- Belt Conveyor A fork-type lugged flatbelt conveyer transports the cut stalk of sorghum to the right side of a reaper at an angle of 90° to make bundling. When the two pulleys’ centers are far apart, a flat belt is preferable [29, 30]. (a) Rotational speed flat-belt conveyor pulley V cf = π × Dfp ×

N fp , 60

ð16Þ

where Vcf is the speed of a flat-belt conveyor, which is equal to the peak speed of a star wheel: Dfp = 630 þ 10 þ 10 = 650mm:

86

A. A. Kassie et al.

Fig. 9 Star wheel drafting in CATIA software

According to Yardley and Stace [31], the diameter of the header pulley is appropriate for the power motor, which is less than 50 kW and 20 mm (10 + 10) spare diameter is provided owing to the lagging of the pulley: N fp =

3:076 × 60 = 90:38 rpm, π × 0:65

where Dfpis the diameter of a flat-belt conveyor pulley, m, Nfpis the rotational speed of a flat-belt conveyor pulley, rpm. (b) Length of the flat belt conveyor. From the geometrical relation in Fig. 10, the center-to-center distance the can be calculated as: C = Lbr þ Rpl þ Rpr ,

ð17Þ

where Lbris the length of the reaping bar, Rpl is left-side conveyor tail pulley radius, Rpr is right-side conveyer header pulley radius. To protect blockage of the cut stalks in the vertical position of the front header the angle of the two pulley assembly techniques should be 90°, the reaping width of a single-section cutter is 76 mm, and there were 17 cutter blades [32, 33].

Design and Numerical Analysis of a Sorghum Reaper Machine

87

Fig. 10 Drafting of a lugged flat-belt conveyor

17 × 76 = 1292 ffi 1300mm Lbr = 1292 mm, Rpl þ Rpr = 650 mm C = 0:325 þ 1:292 þ 0:325 = 1942 mm, Lcf = 2C þ

Dpl - Dpr π × Dpl þ Dpr þ , 2 2

ð18Þ

where Lcfis the length of the flat-belt conveyor: Dpl , Dpr - diameter of the left tail and right header pulley, Lcf = 2 × 1:792 þ

π 0:5 - 0:5 × ð0:5 þ 0:5Þ þ = 5:926 m: 2 2

(c) Analysis of tension and the angle of contacts in the flat-belt drive of the conveyor θ = 1800 - 2α Dsp - Dlp 2C Dsp = Dlp = 650 mm sin α =

C = 1942 mm ð0:650 - 0:65Þ sin α = =0 2 × 1:942 θ = ð180∘ - 2 × 0Þ = 180∘ = 3:14 rad: In order to calculate the tension of the flat belt on the tight side (T1) and on the slack side (T2), let us apply the following expression with the pre-calculated design power of 0.46 kW.

88

A. A. Kassie et al.

2:30 × log =

T1 T1 = μ × θ × cosec β and log × = 1:1965, T2 T2

ð19Þ

where μ is the friction coefficient between the flat belt and groove pulleys (0.3 for the rubber belt). θ is the angle subtended arc, along which the flat belt contacted the pulley, at the center, β is the angle of the flat belt groove (take 2β = 40°). From the antilog table four-digit of 1.1965: T1 = 17:72, T 1 = T 2 × 15:72 T2 Prc = ðT 1 - T 2 Þ × V cf T 1 = 10:24 × 15:72 = 161:09 N: (d) Lug pitch length of the flat-belt conveyor According to Rahman and Butt [33] the pitch length of the flat-belt conveyor lugs can be found. Pl = π ×

Ds 300 =π× = 157 mm, Ln 6

where Pl is the lug pitch length, Ds is the outer diameter of the star wheel, Ln is the number of lugs in the star wheel. To prevent blocking, the bunch of conveyed stalks should be greater than or equal to the number of cut stalks in each cycle of reaping [34]. Qscon ≥ Qscut ,

ð20Þ

where, Qscon is the stalk conveyed output (a sorghum stalk that has been harvested and moved to the side by a conveyor belt) per unit time Qscut is the cut-stalk output V cf × h × wr × ρ2 ≥ V m × wr 2 × ρ1 V × w r 2 × ρ1 h≥ m V cf × wr × ρ2 h≥

wr V m × Z V cf

ð21Þ

Design and Numerical Analysis of a Sorghum Reaper Machine

89

Table 6 Material properties of milled steel for the conveyor shaft Particulars Material Material density Maximum tensile strength Safety factor (Sf) Yield strength (σ s) Bending stress (σ bc)

Specification Milled steel 7861 kg/m3 500 MPa 1.5 250 MPa σs S :f = 167 MPa

ρ ρ2 AA1 1 m2 = = 2= = 7:217 ρ1 AA2 ρ1 0:14 m2 1:3 1:292 h≥ × ≥ 0:0756m: 7:217 3:076

Z=

Therefore, the height of the flat-belt conveyor lug is 75.6 ffi 76 mm, where, h is the height of the flat-belt conveyor lugs, ρ1 is the density of stalks in the field, ρ2 is the density of cut crops on the vertical platform, Z is the gathered crop parameter, AA1 is 1 m2 of the advanced area in the field, AA2 is the area of the circle of the bunched cut stalks from 1 m2 of the area in the field, its closeness being similar to the cut stalks on the vertical header platform. Design Analysis of the Shaft for Flat-Belt Conveyors The right side of the conveyor shaft is created with a length of 2000 mm to transmit the proper torque and sustain the rotating pulleys (Table 6). a twisting moment, the diameter of the conveyor shaft obtained by using the following equation: π × τc × d c 3 16 π × σ bc × d c 3 , Mc = 32 Tc =

ð22Þ ð23Þ

where Tc is the turning effect of the force (torque) acting on the conveyor shat τc is the torsional shear stress of the conveyor shaft, σbc is the bending stress of the conveyor shaft, Mc is the bending moment of the conveyor shaft, dc is the diameter of the conveyor shaft. As the reaper conveyor shaft is subjected to a combined bending and twisting moment, Guest’s theory can be applied:

90

A. A. Kassie et al.

σ bc max = Mec =

1 2

1 1 σ þ 2 bc 2

1 2

σ bc 2 þ 4 τc 2 1 2

Mc þ Mc2 þ T c2

Tec = ðK m × M c Þ2 þ ðK t × T c Þ2

=

π × σ bc × dc 3 32

ð1=2Þ

ð24Þ

,

where Tec = equivalent torsion of propeller shaft. (e) Km = fatigue shock factor for combined bending of the shaft Kt = fatigue shock factor for combined torsion of the shaft.

Tc =

Pc × 60 = 49:02 Nm 2π × N cf

M c = ðT 1 þ T 2 Þ × Ls = 342:66 Nm Applied loads for heavy shakes, Kt, Km = 3 and 2.8 respectively, Mec =

1 2

Mc þ Mc2 þ T c2

1 2

=

π × σ bc × dc 3 32

965:05 = 16395186:66 × dc 3 d c = 0:03888 m = 38:88 mm at 40 mm ðstandardÞ: The maximum shear stress can be determined using the shear stress principle in combination with the torque and bending moment: τp max =

16 × M c 2 þ Mec 2 πd c 3

1=2

=

16 × 965:022 þ 342:662 π0:043

1=2

= 8149133511:45 ≈ 81:49 Mpa:

ð25Þ Analysis of Cut Stalk Conveying Power and Stability The cut sorghum stalks are automatically captured by a star wheel and a finger-type flat-belt lug, and have conveyed horizontally into the right side of the front reaper machine. The force of the cut sorghum stalks in the conveying process is shown in Fig. 11, as well as the condition of horizontal conveying in vertical plate form. F1 þ F2 þ F3 ≥ f 2 þ f 3 þ f 4 , where, F1 is the acting force by the lugs of the lower conveyor flat belt, F 2 is the acting force by the lugs of the middle flat - belt conveyor,

ð26Þ

Design and Numerical Analysis of a Sorghum Reaper Machine

91

Fig. 11 Analysis of sorghum stalks on the flat-belt conveyor lug

F 3 is the acting force by the lugs of the upper flat - belt conveyor, f 1 is the frictional resistance between the root stalk of the sorghum and the cutter blade,

f 2 is the frictional resistance between the lug of the lower conveyor and the stalk,

f 3 is the frictional resistance between the lug of the middle conveyor and the stalk, f 4 is the implication force between the cut sorghum stalks: If the reaped stalk of the sorghum bundle is to be conveyed transversely standing in a vertical position, the sum of the torques that act on the stalk bundle should be zero [34]. M 1 = F 1 × L1 þ F 2 × L3 þ F 3 × L5 - f 2 × L2 - ðf 3 þ f 4 Þ × L4 = 0 M 2 = ðF 4 - F 5 Þ × L2 þ ðF 6 - F 7 Þ × L4 = 0, where F4 is the bracing force of the lower lug in a flat - belt conveyor,-

ð27Þ

92

A. A. Kassie et al.

F 5 is the pressure of compressed spring on the cut sorghum stalk, F 6 is the bracing force of the upper lug of the flat - belt conveyor F 7 is the extrusion force of the cut sorghum stalks π - θ the lower conveyor to the cutter blade, 2 L2 is the vertical dimension from point of action, f 2 , to the cutter blade,

L1 is the vertical dimension from lugs of

L3 is the vertical dimension from the lugs of the middle conveyor to the cutter blade, L4 is the vertical dimension from the point of action, ðþÞto the cutter blade, L5 is the vertical dimension from the lugs of the upper conveyor to the cutter blade:

The height of the sorghum stalks was about 2200–4200 mm, which had been taken from 10 sample sorghum lands from the southern region of Amhara, Raya Kobo. The height of the matured sorghum stalks was an average of 3170 mm. According to preliminary analysis, the lower hemp-conveying chain was about 70 mm from the cutter, the middle stalk of the sorghum-conveying belt was about 400 mm from the lower stalk-conveying belt, and the upper stalk-conveying belt was about 630 mm from the middle stalk-conveying belt, and the total height of the reaper header was 1200 mm (Table 7). Hence, F1 = F2 = F3 = Fc = 50.245 N Pcl = F c × V cv Pcl = 50:245 N ×

3:076 m = 154:55 W, s

where Pcl is the lower conveying power, Fc is the tension force of the conveyers that acts on the cut stalks, Vcv is the conveying speed of each of the three conveyor belts. As the same force acts on all three conveyors, the conveyor power could be:

Table 7 Specification of header height from cutter blade of the reaper No. 1 2 3 4 5 7 9 10

Particulars L1 L2 L3 L4 L5 Height of the front header Material of the header frame Material of the upper frame

Specifications 0.07 m 0.27 m 0.47 m 1.0 m 1.5 m 2m Q235A milled steel Aluminum alloy series

Design and Numerical Analysis of a Sorghum Reaper Machine

93

Pc = Pcl þ Pcm þ Pcu = 154:55 þ 154:55 þ 154:55 = 463:66 W, where Pcm is the power of the medium conveyor, Pcu is the power of the upper conveyor. Selection of Power Source for a Reaper Machine The main source of power required to operate the root-stalk sorghum reaper was computed by adding up the full load of working hours of the machine (cutting with driving power and conveying with driving power). Depending on the performance requirements of the reaper, the Variable Compression Ratio engine can work at different compression ratios. Pe =

full‐load ðconveying power þ cutting powerÞ , η

ð28Þ

where Pe is the engine shaft output power, η is the overall efficiency of an engine by taking the power losses of the bevel gear-box, pulleys, belts, sprockets, and couplings into account [35, 36]. Take 92%: =

ð8953:475 þ 463:66Þ = 10:236 kw ffi 14 HP: 0:92

Design and Selection of Power Transmission System of a Machine Cutting power is 8.954 kW, and conveying power is 0.464 kW; hence, a 14 hp diesel engine is required. Design of the V-Belt Drive for the Engine Output Shaft to the Propeller Shaft The v-belt has excellent absorbency of vibration and tends to transmit power with a long trouble-free life. Standard tables and design expressions were taken from Ahorbo and Rajput [37, 38]. Pulleys with a radius of less than 100 mm are constructed without arms, and pulleys with a radius of 100 to 300 mm are made with four arms. (a) Peripheral speed of the v-belt and pulley diameter Vsvb =

π × Dsp × N sp π × 0:140 × 2500 = = 18 m=s, 60 60

where Vsvb is the peripheral speed of the v-belt small pulley, m/s, Dspis the diameter of the small pulley, m, Nsp is the rotational speed of the small pulley, rpm.

94

A. A. Kassie et al.

To get reduction speed rpm at the propeller shaft of a reaper, the diameter of the driven pulley should be greater than that of the driving pulley. Dlp = 2 × Dsp = 280mm, which is standard: V lvb =

π × Dlp × N lp , 60

where Vlvbis the peripheral speed of the large v-belt pulley, Dlp is the large diameter of the pulley, Nlp is the rotational speed of the large pulley. (b) Length of the v-belt and the angle of contact

Lvb = 2C þ 1:157 × Dlp þ Dsp þ

Dlp - Dsp 4C

2

ð29Þ

C = 2 × Dlp = 2 × 0:28 = 0:56 m Lvb = 1:61469 ≈ 1694 mm, where, Lvb is the pitch length of the v-belt, C is the center distance between the larger pulley and the small pulley. (c) The angle of contact for the open v-belt drive θ = ð180∘ - 2αÞ Dsp - Dlp sin α = 2C θ = 180∘ - 2 × ð- 7:9786∘ Þ = 195:957∘ : (d) Analysis of v-belt tension To analyze the tension of the v-belt on the tight and slack sides, the expression with the pre-calculated design power of 10.297 kW was applied (Fig. 12): 2:30 × log

0:3 × 3:42rad × cosec 17∘ T1 = μ × θ × cosec β = T2 2:3 log ×

T1 = 1:52575 T2

T 2 = 18:88 N, T 1 = 580:63 N:

Design and Numerical Analysis of a Sorghum Reaper Machine

95

Fig. 12 V-belt drive system between the engine and the propeller shaft

Fig. 13 Cutter system modeling of a reaper using CATIA software

Modelling Parts of a Reaper Using CATIA Software 3D modeling entails the use of software tools to produce 3D digital representations of machine parts and structures, such as computer-aided design applications [39, 40]. Modeling of the Cutter System A cutter bar, cuter guard, riveted fasteners, bolts, nuts, crank, and reciprocating connecting rods make up the reaping system (Fig. 13).

96

A. A. Kassie et al.

Fig. 14 Subassembly of conveyor modeling using CATIA software

Modeling of the Conveyor System Using CATIA Software In this system, there are header frames, upper, medium, and lower conveyors, pulleys, flat belts, v-belts, lugs, bolts, nuts, shaft, crop dividers, star wheels, and chain drives (Fig. 14). Assembly of the Sorghum Reaper Machine In this article, the reaper’s 3D modeling is made up of 891 parts, including bolts and nuts. Every component of a machine is designed with specific constraints in mind before it is modeled. As a result, a reaper of modeling has been clearly created (Fig. 15). Finite Element Analysis Finite element methods are utilized to mathematically model structural components of products and solve extremely difficult problems computationally [41–44]. Kinematics Analysis of the Reciprocating Position of the Cutter Bar At a rotating speed of 718.4 rpm, the kinematics of the reciprocating crank mechanism of the reaper is examined: 2π × N = 75:23 rad=s, ðθR Þ = wt 60 r2 r2 þ r × cos × ðwt Þ þ × cosð2 × wt Þ : X ðθ R Þ = 1 4 × L1 4 × L1 ω=

Design and Numerical Analysis of a Sorghum Reaper Machine

97

Fig. 15 Full assembly modeling of a reaper using CATIA software

a

b Reciprocating cutter bar position of a reaper x value

crank position

40

300 30 250

Displacement (in mm)

Displacement (in mm)

20 200 150 100 50 0 –50 0

10 0 –10 –20 –30

1

2 3 4 5 Angle of position (radians)

6

7

–40 0

1

2

3

4

5

6

7

Angle of position (radians)

Fig. 16 (a) Reciprocating cutter versus crank angle and (b) crank X and Y displacement versus the crank angle

(a) MATLAB result for crank position analysis in x and y coordinate systems (see Fig. 16b): Sx ðθR Þ = R × cosðθR Þ Sy ðθR Þ = R × sinðθR Þ: (b) MATLAB result for cutter bar velocity analysis (see Fig. 17a):

98

A. A. Kassie et al.

a

b

cuter bar Velocity

4000

crank velocity 3000

3000 2000 Velocity in mm/sec

Velocity in mm/sec

2000 1000 0 –1000

1000

0

–1000

–2000 –2000

–3000 –4000 0

1

2 3 4 5 Angle of position (radians)

6

–3000

7

0

1

2

3 4 5 Angle of position (radians)

6

7

Fig. 17 (a) Cutter bar velocity versus crank angle and X and Y component of (b) crank velocity versus crank angle

a 2

x10

b

cuter bar Acceleration X value

5

2.5

1.5

x10

crank acceleration

5

acceleration (mm/sec2)

Velocity in mm/sec2

2 1 0.5 0 –0.5 –1 –1.5 –2 –2.5 –3 0

1.5 1 0.5 0 –0.5 –1 –1.5 –2

1

2

3

4

5

6

7

–2.5

0

Angle of position (radians)

1

2

3

4

5

6

7

Angle of position (radians)

Fig. 18 (a) Cutter bar acceleration versus crank angle and (b) X and Y components of crank acceleration vs crank angle

V ðθR Þ = - ϖ × r × sinðϖt Þ þ

r × sinð2ϖt Þ : 2 × L1

(c) Crank velocity analysis in x and y coordinate systems (see in Fig. 17b): V x ðθR Þ = - ϖr × sinðθR Þ and V y ðθR Þ = - ϖr × cosðθR Þ: (d) Result for the cutter bar acceleration analysis (see Fig. 18a):

AðθR Þ = - ω2 × r × cos × ðwt Þ þ

r × cosð2 × wt Þ : L1

Design and Numerical Analysis of a Sorghum Reaper Machine

4

x107

99

Shaking force x value alonge x and y coordinates

3

2

force [N]

1

0

–1

–2

–3

–4 0

1

2

3

4

5

6

7

Angle of position (radians)

Fig. 19 The x and y component shaking force versus crank angle

(e) Crank acceleration analysis in x and y coordinate systems (see Fig. 18b): Ax ðθR Þ = - ϖ 2 × r cosðθR Þ and Ay ðθR Þ = - ϖ 2 × r sinðθR Þ: (f) Simulations for a mass-balanced study of the cutter system (see Fig. 19). Effective mass at the crank MA) = M2a + M3a = 2.66 kg + 1.33 kg = 3.99 kg. Effective mass at the slider cutter MB) = M4 + M3b = 1.27 kg + 4.55 kg = 5.8 kg:

(lumped bar

F x ðθ R Þ = M A × - r × ω2 × cosðθR Þ - M B - ω2 × r × cos × ðθR Þ þ - M A × - r × ω2 × r sinðθR Þ :

(lumped

mass mass

r × cosð2 × θR Þ L1

F y ðθ R Þ =

4 Result and Discussion The outcomes of this article are classified into two groups. The first was that each component of the reaper machine could be validated with senior similar articles by comparing analytical and numerical analyses to ensure that the components were

100

A. A. Kassie et al.

Fig. 20 Maximum shear stress result of (a) a propeller shaft and (b) an engine shaft

safe. The generic design reaper, on the other hand, should be compared with manual harvesting if the machine is used in sorghum fields in order to observe the considerable benefits of agricultural machinery over manual harvesting.

4.1

Analytical Analysis and Numerical Simulation

To determine whether the crank is in pure rotation and the slider cutter is in pure translation, a MATLAB simulation has been implemented. Simulation Result on the Dynamic Behavior of the Reciprocating System The MATLAB Program database calculates the position, velocity, and acceleration of the cutter bar. Figures 16, 17, 18, 19, 20, 21, and 22 show these functions for this reciprocating system of a reaper, in the MATLAB program, as plotted for a constant crank (ϖ) over two revolutions. The acceleration curve in Fig. 18a shows the effects of the second harmonic term most clearly because the term’s coefficient is larger than its correspondent in either of the other two functions. The fundamental (- cos ϖt) term gives a purely harmonic function with a period of (360∘). This fundamental term dominates the function as it has the largest coefficient. The flat top and slight dip in the positive peak acceleration of the Fig. 18a is caused by the cos(2ϖt) second harmonic adding or subtracting from the fundamental. Note that every value of the peak acceleration of the cutter bar and crank, even at a moderate speed, is a moderately sized reciprocating reaper with a 76-mm stroke distance and(r/Ll) = 1/ 3 mm. This article also looked at the dynamic behavior of a reaper’s reciprocating mechanism using an approximate kinematic model. The combination of masses (lumped) parameter model of the reaper reciprocating system of the cutter bar crank is shown in Fig. 19; the summation of the masses of the crank M2a and the

Design and Numerical Analysis of a Sorghum Reaper Machine

101

Fig. 21 Maximum shear stress simulation result of a conveyor shaft

Fig. 22 Comparison of manual and reaper harvesting (a) losses and (b) costs

portion of the connecting rod M3a is MA at the contact point B; two masses are also lumped, the reaper reaping bar mass M4 and the reaming portion of the connecting rod masses M3b (their sum is MB). The MATLAB program calculates the shaking forces at a constant crank speed of 718.18 rpm for a combination of linkage parameters input into it as shown in Fig. 19, which also shows the shaking force along the x and y planes above and below the wt values. The lumped mass A and lumped mass B are almost 95% equal. Then, based on this information from the

102

A. A. Kassie et al.

Table 8 Maximum shear stress analysis of reaper shafts

Engine shaft Propeller shaft Conveyor shaft

Torque (turning effect of the force) inNm 39.33

Bending moment in Nm 89.93

Equivalent bending moment in Nm 282. 12

Maximum shear stress in MPa 55.85

78.66

719.4

2164.63

69.8

49.02

342.66

965.05

81.49

Table 9 The maximum shear stresses (τmax) result obtained by ANSYS R.19.2

Engine shaft Propeller shaft Conveyor shaft

Me applied on each cross-section of shafts in Nm 282.12 2164.63 965.05

Maximum shear stress in MPa 55.14 66.953 77.28

MATLAB database, the reciprocating system is running well at a constant speed of 718.18 rpm. Owing to this significant result, the reaper machine of the reaping system can work well without creating noise and vibrations. Maximum Shear Analysis on the Engine and Propeller Shaft The engine and transmission shafts were designed with lengths and diameters of 1200 mm with 55 mm and 800 mm with 30 mm respectively. Because reaper shafts spin, they may be subjected to a twisting moment as well as coupled stresses. The analytical results for both the propeller shaft and the propeller are listed in Table 8. Numerical Simulation Results of Shafts Obtained from ANSYS Figures 20 and 21 illustrate the acquired maximum shear stress findings for each analyzed cross-section shaft of a reaper as an exemplary numerical simulation analysis using ANSYS R.19. Thus, Table 9 gathers the maximum shear stresses (τmax) result obtained with the ANSYS software. Comparison of Analytical and Numerical Simulation Results The conveyor shafts and those obtained by using the finite element method software (τmax) are presented in Table 10. As shown in Table 9 the difference %error of τmax is lower than 0.055. Therefore, based on Murawski [44] the calculated changes of analytical and numerical maximum shear stresses of percentage errors do not exceed 8%. Therefore, this result highlights that the analytical design of each transmission shaft has been confirmed nearly with numerical simulation and the percentage error disparities between analytical and numerical are statistically significant. Then, for each shaft of the

Design and Numerical Analysis of a Sorghum Reaper Machine

103

Table 10 Comparison between analytical and numerical analysis

Engine shaft Propeller shaft Conveyor shaft

Equivalent bending moment in Nm 282.12

Maximum shear stress results using analysis in MPa 55.85

Maximum shear stress results using FEM in MPa 55.14

Error (%) 1.27

2164.63

69.8

66.953

4.08

965.05

81.49

77.28

5.16

FEM = Finite element method

Table 11 Specifications of the sorghum reaper machine Descriptions Machine name Manufacturer Model Overall dimensions (L × W × H ) Total weight Cost of the total reaper Power source Types of fuel used Width of cutter bar Forward speed in full load Vm Cutting speed, Vf Conveying speed, Vs

Details Sorghum root-stalk reaper Any medium-sized metal workshop BiT SR @ HA1 ( 3.30 × 2.00 × 2.30), m 750 kg 183697.26 ETB 14 hp Diesel 1292 mm 1.3 m/s 1.82 m/s 3.076 m/s

reaper machine, the design specifications of each parameter collected from journals and e-books were made safe.

4.2

Design Specification of Root-Stalk Sorghum Reaper Machine

The sorghum crop reaper design specification was produced using experimental results based on the mechanical properties of seed-stage sorghum stalks. As the machine has not yet been manufactured, the manual and reaper evaluation of harvesting has been carried out theoretically using the concept of Gore et al. [45]. During the theoretical comparison, the following parameters were measured: the machine’s forward speed is 1.3 m/s, and let us take a 50-m distance in the field (Table 11). ETB = Ethiopian birr

104

A. A. Kassie et al.

(a) Travel speed of the sorghum reaper, t t= =

mf Vm

ð30Þ

50m = 38:46 s, 1:3m=s

where t is the time of travel. (b) Speed of operation of the sorghum reaper A time of 38.46 s (0.0107 h) was needed to travel 50 m (0.05 km): the speed operation sorghum reaper ðV t Þ =

0:05 km = 4:67km=h: 0:107 h

(c) Field capacity Theoretical field capacity, ha=h =

V t × r w 4:67 × 1:292 = = 0:1206, 50 50

where rw is the reaping width of the reaper, Vt is the speed of travel, km/h. (d) Sorghum crop loss estimation in the field of 1 m2 According to data from the research office in Raya Kobo, Girana, the average sorghum crop production 2100 kg/ha (Table 12). 2:5 g × 10,000m2 = 25,000 g=ha = 25Kg=ha

25 × 100% = 1:19%: 2100

The analytical harvesting loss by the root-stalk reaper for the sorghum crop was found to be 1.19%. (e) Cost-related information on the sorghum reaper The sorghum reaper also has the same rental infrastructures for supporting Ethiopian farmers during the harvesting season of sorghum and maize crops, according to the rules and regulations of plow machine rental information available in Humera. When the machine is set up in micro and small enterprise facilities in Table 12 Annual observation of income and loss of the sorghum reaper Item The average yield of sorghum per year Loss of crops from 17 carriage sorghum 1 hectare

Descriptions 2100 kg/ha 2.5 gm 10,000 m2

Design and Numerical Analysis of a Sorghum Reaper Machine

105

Table 13 List of sorghum reaper machine rentals per hour Item Money needed to reap sorghum for 1 h Total money needed for one hectare

Description 1500 ETB 8.29 h × 475 = 3938.6 ETB

each of Ethiopia’s sorghum-growing districts, such as Raya, Humera, Chiro, Assosa, and Benshagul Gumuz, it will contain the following account details for each farmer if presented in the form of rent: for 1 h = 0:1206 ha: Therefore, the reaper machine needs 8.29 h to complete 1 ha of sorghum stalk fields (Table 13). ETB = Ethiopian birr Therefore, 185, 389, 152.2 Birr is needed to reap 591, 200, 230 kg/h of sorghumcultivated fields per year.

4.3

Comparison of Manual and Root-Stalk Sorghum Harvesting Results 28:7 þ 30 þ 31:5 þ 27:5 þ 26:5 þ 25:7 þ 25 þ 26:7 8 = 27:7Total manual harvesting losses in 47069:33 haT ml = 27:7% × 47069:33 = 12708:71 ha,

Aml % =

where Amlis the average loss of manual harvesting per year in 1 ha Tmlis the total losses from 47,069.33 ha due to manual harvesting. Owing to mechanization, the harvesting losses of the sorghum reaper are calculated at 1.9% theoretically from 1 ha of field. Based on Manjunatha et al. [46], this loss value (1.9%) is significantly accepted: T rl % = 1:9 × 47069:33 = 894:3 hectare, where Trl% is the total losses from 47,069.33 ha due to reaper harvesting. Using the root stalk sorghum reaper, it is possible to save up to 11,814.4 ha out of a total of 47,069.33 ha. Rayian residents spend about 316, 305, 897.6 Birr per year to reap seed-stage sorghum by laborers who come from regions such as Jemdo Mariam, Muja, Tata, Mezgeramba, Lasta, Tekulesh, Sekota, Geregera, Stayish, Korem Maychew, and Zobl, according to this study article. As a result, if it swiftly develops this sorghum

106

A. A. Kassie et al.

harvesting equipment and distributes it to farmers through a small micro-enterprise, they can save up to 130, 916, 745.4 Birr every year on 47,069.33 ha of sorghum lands.

5 Conclusion The key conclusions of this study are as follows: agriculture in most Ethiopian regional states is characterized by low production owing to a lack of current technology practice, management, and effective control of recurring crop losses due to manual harvesting as well as labor crises. The delayed harvest is attributed to labor shortages as a result of changes in occupation from agriculture to other fields, resulting in higher annual grain yield losses in 1 ha of sorghum fields because of the over-maturity of grains, whereas it adds up to 889, 609.7 Kuntal of sorghum grains wasted from the Raya Kobo region’s sorghum fields. According to the findings of this study article, farmers who utilize this new sorghum reaper machine can save up to 130.9 Million Birr per year on a total of 316.3 million Birr labor costs on 47,069.33 ha of sorghum lands. Using a sorghum reaper machine, grain losses caused by over-maturity could be reduced by up to 11,814.4 ha out of a total of 47,069.33 ha. Rather than minimizing losses and manpower shortages, the reaper plays an important role in decreasing human drudgery and saving time. As a result of these considerable benefits, manual harvesting should soon be replaced with a modern root-stalk sorghum harvester.

6 Recommendation The adoption of appropriate modern sorghum and maize harvesting practices that have been designed in this research paper is urgently needed to increase crop productivity and economic emancipation in the Ethiopian regional states. The detailed design and analysis of sorghum crop reaper which includes propelling and reaping system, assembly, since the cutting mechanism involves reciprocating motion, the dynamic behavior of the system is chucked out by using MATLAB, the resulting graph recommended that the crank is in the right rotation and the cutter bar is in the exact translation. The resulting graph recommended that the crank is in the right rotation and that the cutter bar is in the exact translation. The analytical designs of the flat-belt conveyors, power transmission system, and engine shaft were compared with numerical simulations using ANSYS. As the numerical result in ANSYS demonstrates, the errors from shafts are less than 8%, indicating that the shaft design is nearly precise. If it was more than 8%, as past scholars have stated, the design is not acceptable; however, if it is less than 5.5%, the design is acceptable. Beyond that 2D and 3D models of the root-stalk sorghum reaper components have been generated using CATIAV5, R19 software to clarify the physical structure for

Design and Numerical Analysis of a Sorghum Reaper Machine

107

manufacturers during the manufacturing period in the future. As a result, this study recommends that Ethiopian government stakeholders take steps to manufacture this root-stalk sorghum reaper as soon as possible, based on the design specifications established in this study, in order to avoid the previously mentioned losses and manual labor crises during each sorghum-harvesting season.

References 1. Araya, A., Keesstra, S., Stroosnijder, L.: A new agro-climatic classification for crop suitability zoning in northern semi-arid Ethiopia. Agric. For. Meteorol. 150(7–8), 1057–1064 (2010) 2. Wondimu, Z., et al.: Agro-morphological diversity of Ethiopian sorghum [Sorghum bicolor (L.) Moench] landraces under water limited environments. 67(8), 2149–2160 (2020) 3. Rashid, S., et al.: The barley value chain in Ethiopia. 3(169), 169 (2019) 4. Visarada, K., Aruna, C.: Sorghum: a bundle of opportunities in the 21st century. In: Breeding Sorghum for Diverse End Uses, pp. 1–14. Elsevier, Hyderabad, India (2019) 5. Demeke, M., Di Marcantonio, F.J.G.O.R.: Analysis of incentives and disincentives for sorghum in Ethiopia. Analysis of incentives and disincentives for sorghum in Ethiopia 3(6), 34 (2019) 6. Solomon, H., et al.: Promotion of improved sorghum technologies through large-scale demonstration in Gololcha Woreda, Arsi zone of Oromia regional state, Ethiopia. Am. J. Plant Sci. 12(03), 366 (2021) 7. Seyoum, A., et al.: Defining Sorghum Product Concept from Production to Processing (2019) 8. Lee, K., Malerba, F., Primi, A.: The fourth industrial revolution, changing global value chains and industrial upgrading in emerging economies. J. Econ. Policy Reform. 23(4), 359–370 (2020) 9. Hasan, K., et al.: Impact of modern rice harvesting practices over traditional ones 8, 89–108 (2020) 10. Li, L., et al.: Effect of corrosion and hydrogen embrittlement on microstructure and mechanical properties of mild steel. Constr. Build. Mater. 170, 78–90 (2018) 11. Chattopadhyay, P., Pandey, K.: Mechanical properties of sorghum stalk in relation to quasistatic deformation. J. Agric. Eng. Res. 73(2), 199–206 (1999) 12. Katz, R.: Integrating analysis and design in mechanical engineering education. Procedia CIRP. 36, 23–28 (2015) 13. Richards, K.L.: Design Engineer’s Handbook. CRC Press, Technion City, Haifa, Israel (2012) 14. Sahoo, A.U., Raheman, H.: Development of an electric reaper: a clean harvesting machine for cereal crops. Clean Techn. Environ. Policy. 22(4), 955–964 (2020) 15. Pandey, M., Devnani, R.: Design, development and field evaluation of Vertical Conveyor Reaper Windrower. Agric. Mech. Asia Afr. Lat. Am. 16(2), 41–52 (1985) 16. Huang, J., et al.: Design and test of two-wheeled walking hemp harvester. Int. J. Agric. Biol. Eng. 13(1), 127–137 (2020) 17. Norton, R.L.: Design of Machinery: An Introduction to the Synthesis and Analysis of Mechanisms and Machines. McGraw-Hill/Higher Education, china (2008) 18. Robert, L.N.: Design of Machinery. McGraw-Hill Education (2019) 19. Dimitriadis, C.I.: The Design of an Improved Efficiency Lavender Harvester. Cranfield University (2005) 20. Shen, C., et al.: Experimental analysis on mechanical model of ramie stalk. Trans. Chin. Soc. Agric. Eng. 31(20), 26–33 (2015) 21. Song, Z., et al.: Experiment on cutting characteristics of cotton stalk with double supports. Trans. Chin. Soc. Agric. Eng. 31(16), 37–45 (2015) 22. Tsegaye, A.: Adaptation and Performance Evaluation of Engine Operated Reaper

108

A. A. Kassie et al.

23. Fitzpatrick, K., Brewer, M.A., Turner, S.: Another look at pedestrian walking speed. Transp. Res. Rec. 1982(1), 21–29 (2006) 24. Zareishahamat, E., Maysami, M.: Design, construction and performance evaluation of a reaper for small farms (a case study: Iran). Agric. Eng. Int. CIGR J. 21(3), 104–113 (2019) 25. Ling, B.: Wheat and paddy reapers in China. In: Small Farm Equipment for Developing Countries: Proceedings of the International Conference on Small Farm Equipment for Developing Countries: Past Experiences and Future Priorities, 2–6 Sept 1985. Int. Rice Res. Inst (1986) 26. Kyriacos, D.: Polycarbonates. In: Brydson’s Plastics Materials, pp. 457–485. Elsevier, Bai Ling, China (2017) 27. Ogunlowo, Q., Olaoye, J.: Development and performance evaluation of a guided horizontal conveyor rice harvester. Agrosearch. 17(1), 66–88 (2017) 28. Hainsworth, S.: Power transmission by flat and ‘v’ belts. J. Text. Inst. Proc. 26(1), P3–P10 (1935) 29. Olutomilola, E.O., Ayodeji, S.P., Adeyeri, M.K.: Finite element analysis of a washing and preheating unit designed for plantain flour process plant. Int. J. Eng. Technol. IJET. 5(4), 117–127 (2019) 30. Bassey, J.E., Bala, K.C.: Development of an automatic mini-conveyor system for product monitoring. In: IOP Conference Series: Materials Science and Engineering. IOP Publishing, South East Asia (2018) 31. Yardley, E.D., Stace, L.R. (eds.): Belt Conveying of Minerals. Elsevier, Ota, Nigeria (2008) 32. Belay, A.: Design and Analysis of Manually Driven and Engine Powered Wheat Crop Reaper for Broad Bed Furrows 33. Rahman, Z., Butt, K.: The tractor-mounted reaper in Pakistan. In: Small Farm Equipment for Developing Countries: Proceedings of the International Conference on Small Farm Equipment for Developing Countries: Past Experiences and Future Priorities, 2–6 Sept. 1985. Int. Rice Res. Inst. (1986) 34. Sjuchro, D.W., et al.: The Role of Community Radio in Participation of Development of Various Fields in Rural Communities. J. Posit. Sch. Psychol. 7505–7513 (2022) 35. Legros, L.A.: The transmission of power from the engine to the road wheels in motor vehicles. Proc. Inst. Automob. Eng. 3(1), 334–382 (1908) 36. ESCAP, U.: ReCAMA regional directory of agricultural machinery manufacturers and distributors (2021) 37. Ahorbo, G.K.: Design of a throw-in axial flow rice thresher fitted with peg and screw threshing mechanism. Int. J. Sci. Technol. Res. 5, 171–177 (2016) 38. Rajput, R.: A Text Book of Automobile Engineering. Firewall Media, Pasfic Asia (2008) 39. Billah, M., et al.: Selection of V-belt for power transmission in agricultural machinery. Progress. Agric. 19(2), 187–194 (2008) 40. Buswell, R.A., et al.: Freeform construction: mega-scale rapid manufacturing for construction. Autom. Constr. 16(2), 224–231 (2007) 41. Russell, T.F., Wheeler, M.F.: Finite element and finite difference methods for continuous flows in porous media. In: The Mathematics of Reservoir Simulation, pp. 35–106 . SIAM (1983) 42. Bro-Nielsen, M., Cotin, S.: Real-time volumetric deformable models for surgery simulation using finite elements and condensation. In: Computer Graphics Forum. Wiley Online Library, UK (1996) 43. Duarte, C.A., Babuška, I., Oden, J.T.: Generalized finite element methods for three-dimensional structural mechanics problems. Comput. Struct. 77(2), 215–232 (2000) 44. Murawski, L.: Shaft line alignment analysis taking ship construction flexibility and deformations into consideration. Mar. Struct. 18(1), 62–84 (2005) 45. Gore, A., Kasal, Y.G., Shinde, S.: Field evaluation of self propelled reaper binder in wheat crop. Plant Arch. 18(1), 551–554 (2018) 46. Manjunatha, M., et al.: Field performance evaluation of vertical conveyor paddy reaper. Karnataka J. Agric. Sci. 22(1), 140–142 (2009)

Effects of Initial Moisture Content and Storage Duration on Physical and Chemical Characteristics of Stored Maize (Zea mays L.) Grain Habtamu Gebremichael Daba, Mulugeta Admasu Delele, Solomon Workneh Fanta, Nigus Gabbiye Habtu, Metadel Kassahun Abera, and Admasu Fanta Worku

Contents 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1 Experimental Design and Treatments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Chemical Composition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3 pH Value . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4 Germination Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5 Grain Damage Percent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.6 Statistical Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1 Stored Maize Grain Temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Ambient Temperature and Relative Humidity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 Physical and Chemical Characteristics of Stored Maize . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

110 110 111 111 111 112 112 112 112 113 113 113 116 116

H. G. Daba (✉) Kulumsa Agricultural Research Center, Asella, Ethiopia Faculty of Chemical and Food Engineering, Bahir Dar Institute of Technology, Bahir Dar University, Bahir Dar, Ethiopia M. A. Delele · S. W. Fanta · N. G. Habtu · M. K. Abera · A. F. Worku Faculty of Chemical and Food Engineering, Bahir Dar Institute of Technology, Bahir Dar University, Bahir Dar, Ethiopia © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024 K. Mequanint et al. (eds.), Advancement of Science and Technology in Sustainable Manufacturing and Process Engineering, Green Energy and Technology, https://doi.org/10.1007/978-3-031-41173-1_6

109

110

H. G. Daba et al.

1 Introduction Maize is the predominant crop and a vital source of calories in Ethiopia [1]. According to [2], Ethiopia produced 9,635,735 tons of maize in 2019 with a yield of 42.4 quintals per hectare. Maize grain storage is essential for sustaining food supply in households and for selling at a higher price during the lean season since maize is mainly produced seasonally in developing countries [3]. Maize storage in Ethiopia lasts 8–10 months [4]. Farmers’ traditional storage structures let maize grain deteriorate through the effects of biotic and abiotic factors in the maize grain storage ecosystem [5]. Factors affecting maize grain quality during storage are categorized into grain storage environment characteristics, storage duration, storage structure type, and the initial grain condition before storage [3]. During the maize post-harvest supply chain, storage is known to be the critical loss point [5]. There are different underlying factors of maize bio-deterioration during storage which reduce the economic and nutritional value of maize. Among them, moisture and temperature play a crucial role in the quantity and quality loss of maize during long-term storage [3, 6, 7]. In rural areas, farmers tend to store maize grain with a moisture content of 13–15% [8]. Farmers in Ethiopia commonly store maize kernels in woven polypropylene sacks [3, 9–11]. Many studies [3, 9, 10, 12] on the physical and chemical characteristics of stored maize are limited to 6 months of storage time and do not elaborate on the effect of Ethiopia’s rainy season on stored maize grain physical and chemical characteristics in woven polypropylene sacks. Moreover, little information is available regarding the storability of maize grain in hot and humid climatic conditions at 15% initial moisture content while maintaining its physical and nutritional quality within a woven polypropylene sack. Research on nutritional quality deterioration of stored maize is scanty and a research gap exists in identifying the effects of storage duration and maize initial moisture content on the physical and chemical characteristics of stored maize in woven polypropylene sack. Such knowledge will enable concerned bodies to make informed decisions on maize storage duration and initial moisture content. Thus, this study aimed to evaluate the effects of storing maize at different initial moisture content (12% and 15% wet basis) and storage duration at 1800 m above sea level (a.s.l.) on the physical and chemical characteristics of maize (Zea mays L.) grain.

2 Materials and Methods The experiment was carried out for an 8-month storage period with maize grain (Bako-Hybrid 661 variety) harvested in December 2020 and stored at Bahir Dar (1800 m.a.s.l.). Bahir Dar is a city located in the Amhara National Regional State, Ethiopia; about 578 km northwest of Addis Ababa, Ethiopia. The city lies between latitude of 11°35′ 37″ N and longitude of 37°23′26″ E [13].

Effects of Initial Moisture Content and Storage Duration on Physical. . .

2.1

111

Experimental Design and Treatments

A completely randomized design (CRD) was employed for this study. Two factors were studied, viz., initial moisture content (IMC) (12% and 15% w.b) and storage duration (0 month, 4 months, and 8 months). The experiment was arranged in a completely randomized design (CRD) with three replications. To achieve the desired initial moisture content of the experiment, maize grain was conditioned based on Eq. (1) [14]. Q = Aðb–aÞ=ð100–bÞ

ð1Þ

where Q = amount of water to be added (ml), a = IMC of the sample (%), A = initial weight of the sample (g), b = final moisture content of the sample (%) The conditioned maize was then tightly packed in an aluminum cover and put in 5 °C refrigerator for 7 days, homogenized manually. The storage ambient temperature and relative humidity (RH) was measured daily using Vicks® portable digital thermo hygrometer (KAZ Incorporated, New York, NY, USA). Meanwhile, the grain temperature was measured at an interval of 1 h for 8 months at Bahir Dar (1800 m.a.s.l.) using a HOBO® UX100–011 temperature/RH data logger (Onset Computer Corporation, Bourne, Massachusetts, USA). Twelve kilogram of maize grain (each 2 kg was conditioned to 12% and 15% IMC) was stored at Bahir Dar Institute of Technology (1800 m.a.s.l.) for an 8-month storage duration. A hand sampling technique was carried out based on the International Seed Testing Association [15]. The maize samples were thoroughly homogenized by inversion and representative samples were reduced through the conning and quartering method as described in [16]. Four hundred gram of maize grain was sampled randomly at baseline (before the start of the experiment), after 4 months and 8 months from the top, bottom, sides, and center of maize grain in a woven polypropylene sack and transported immediately to the laboratory. The samples were then stored at 4 °C in a refrigerator until laboratory analysis. Measurements were done in triplicate [13].

2.2

Chemical Composition

Maize grain chemical compositions such as oil content, protein content, starch content, and moisture content of maize samples were measured using near infrared transmittance using the Foss Infratec 1241 grain analyzer (Tecator, Hoganas, Sweden). Ash content was determined as described in [17].

2.3

pH Value

The pH value was determined from a filtrate of 2 g ground maize sample (80 mesh size) mixed with 20 ml of distilled water. The solution was then stirred for 1 min with a stir bar and then the pH value was recorded using a glass electrode pH meter (HI 2211, Hanna instruments, Woonsocket, USA) [18].

112

2.4

H. G. Daba et al.

Germination Test

Maize grains were placed on plastic trays filled with a 2-cm thick sand layer. Four hundred maize grains (100 maize grains per plastic tray) were placed apart so that they did not touch each other or the walls of the plastic tray. Then, the kernels were covered with a 0.5 cm layer of sand without compacting to ensure adequate air supply. Before planting of the maize grains, the sand was sterilized for 1 h at 200 °C and moistened with distilled water. Then the samples were placed at room temperature and maize kernels that germinated normally were counted after 7 days [19].

2.5

Grain Damage Percent

Maize grain samples were divided using the conning and quartering technique until a final sample of 100 g obtained. From 100 g of maize kernels, insect-damaged and undamaged kernels were separated, counted, and weighed. Insect damaged kernels were visually identified based on holes made by insects. Grain damage percent was calculated using Eq. (2) [20]: Damage ð%Þ = ½Number of damaged kernels ÷ Number of kernels in 100 grams × 100

ð2Þ

2.6

Statistical Analysis

Statistical analysis of the physical and nutritional quality parameters was done using Minitab® (Minitab version 17 statistical software; Minitab Inc., Pennsylvania, USA). Homogeneity of variance was tested using Levene test in Minitab statistical software. Grain damage percent data were transformed to square root before statistical analysis, and then the treatment means were converted back to the original format for presenting the results obtained. Analysis of variance (ANOVA) was carried out. Tukey test was used to separate the means when there were significant differences among treatments at 5% level of significance.

3 Results and Discussion Storage duration significantly affected (P < 0.001) the moisture content of stored maize within a woven polypropylene sack. The moisture content of maize stored with 12% initial moisture content increased (P < 0.001) from 11.1% (4 months) to

Effects of Initial Moisture Content and Storage Duration on Physical. . .

113

Table 1 Ambient temperature and relative humidity data of the experimental location Location/Altitude (asl) (m)

Month

1800

February–March April–May June–July August– September

Temperature (°C) Min. Max. Mean 22.2 29.7 25.9 21.2 31.8 25.4 20.6 31.7 23.8 20.9 28.1 25.9

Relative humidity (%) Min. Max. Mean 24 38 30.6 34 48 43 52 79 65.5 60 83 71.5

12.89% (8 months), while maize stored with 15% initial moisture content increased (P < 0.001) from 11.23% (4 months) to 12.51% (8 months) of storage in a woven polypropylene sack. On the other hand, the initial moisture content of stored maize for 4 months had no significant impact (P = 0.228) on the stored maize moisture content (11.1%) at 12% initial moisture content and moisture content (11.23%) at 15% initial moisture content within a woven polypropylene sack at 1800 m above sea level. The higher relative humidity during the rainy season, after 4 months of storage (Table 1), increased the moisture content of stored maize in woven polypropylene sacks. The permeability of the woven polypropylene sack lets the stored maize grain moisture content fluctuate with the surrounding storage environment [21].

3.1

Stored Maize Grain Temperature

Mid-March, mid-May, and early August 2021 were the periods of the highest maize grain temperature at 1800 m.a.s.l., but the lowest grain temperature at 12% and 15% IMC maize were observed in July 2021 (see Fig. 1).

3.2

Ambient Temperature and Relative Humidity

Monthly ambient temperature and relative humidity data at 1800 m.a.s.l. in 2021 is presented in Table 1.

3.3

Physical and Chemical Characteristics of Stored Maize

Data of maize grain physical quality and chemical composition during 8 months of storage at 1800 m.a.s.l. with 12% and 15% IMC are presented in Tables 2 and 3, respectively.

Temperature inside storage structure (degree Centigrade)

114

H. G. Daba et al.

Temperature changes in 12 % mc, 14 % mc, 15 % mc conditioned maize 29

Variable 12 % mc 14 % mc 15 % mc

28 27 26 25 24 23 22 21 Feb

Mar

Apr

May

June

July

August September

8 storage Months (Feb 2021 - Sept 2021)

Fig. 1 Temperature changes (°C) of conditioned maize grain in woven polypropylene sack

Stored Maize Grain Damage and Germination [12, 22–25] stated the role of increased ambient air temperature and ambient relative humidity in increasing grain moisture content and grain damage through higher insect invasion and mold proliferation, which agrees with the study results (Tables 2 and 3). However, maize grain germination percentage decreased with an increase in storage duration (Tables 2 and 3), which is in agreement with the reports of [5, 12, 21, 23, 24, 26]. Stored maize oil content There was a decrease in oil content with storage (Tables 2 and 3). [27] noted the role of grain respiration, oxidation, and enzymatic activities in stored maize for the relative decrement of oil content after prolonged storage duration, which coincides with the study results. Stored Maize Protein Content There was a decrease in protein content with storage period (Tables 2 and 3). [3, 9, 10, 25, 26, 28, 29] observed a similar trend. The reduction in protein content is attributed to the higher insect feeding on stored maize grain and also the higher mold proliferation on damaged grain [3, 9, 26, 27]. Stored Maize Starch/Carbohydrate Content In the present study, starch content of stored maize had slightly decreased as storage period increased from 0 month to 4 months (Tables 2 and 3). [3, 10, 26] reported a decline in carbohydrate content as storage duration increased. [30] reported a decrease in carbohydrate content of stored maize due to insect infestation and feeding on the endosperm of maize grain, which in turn results in a relative increment of protein, fiber, and moisture content of stored grain. [28] reported a decrease in starch and protein content of stored maize due to the increase in grain damage by Sitophilus zeamais feeding.

Effects of Initial Moisture Content and Storage Duration on Physical. . .

115

Table 2 Mean (±SD) of physical and chemical characteristics of stored maize (12% IMC) Moa 0 4 8 F2, 8 P value

Oil (%) 5.22 ± 0.03c 4.98 ± 0.09c 4.53 ± 0.25d 15.55 0.004

Protein (%) 9.28 ± 0.09c 8.57 ± 0.06d 8.86 ± 0.26d 14.44 0.005

Starch (%) 70.62 ± 0.14d 69.47 ± 0.02e 71.99 ± 0.03c 692.23 F” less than 0.05 indicating that the model is significant. The significant models of this analysis are A, B, C, AB, A2, and B2 while the “lack of Fit, F-value” is 0.3284 implying that the lack of fit is not significant relative to the pure error [11]. Analysis of variance was used to check the accuracy of the developed model where the R-squared value of the developed model is 0.9219 implying that 92.2% of the total variation in the removal efficiency of Cr (VI) is attributed to the experimental variables studied [9]. The “pre-R-squared” value of 0.8732 is also in reasonable agreement with the “Adj R-squared.”

3.3

Effect of Variables on the Removal of Cr (VI)

The full factorial design was used to analyze the effect of variables on Cr (VI) removal efficiency with the help of contour surface and perturbation plots. The bioremediation process of Cr (VI) was significantly affected by the individual variables such as incubation time (A), pH (B), and initial Cr (VI) concentration

210

T. A. Ashagrie et al.

Table 3 Result of ANOVA for response surface-reduced quadratic mode Source Model A-time B-pH C-initial con. of Cr (VI) AB A2 B2 Residual Lack of fit

Sum of squares 11414.2 911.47 4596.73 1504.79 306.72 1351.88 2023.39 1037.92 647.06

DF 6 1 1 1 1 1 1 7 0

Mean square 1902.37 911.47 4596.73 1504.79 306.72 1351.88 2023.39 22.08 32.35

Fvalue 86.14 41.27 208.5 68.14 13.89 61.22 91.62 – 1.2

Pure error Core total

390.86 12452.12

7 3

14.48 –

– –

Prob > F