Machinability of Fibre-Reinforced Plastics 9783110292251, 9783110292220

Table of contents :
Preface
Contents
List of contributing authors
1 Laser material machining of CFRP – an option for damage-free and flexible CFRP processing?
1.1 Introduction
1.2 State of the art of machining of CFRP
1.2.1 Cutting CFRP
1.2.2 Surface pre-treatment of CFRP
1.2.3 Shape cutting of CFRP
1.3 Laser material interaction
1.4 Laser material machining of CFRP
1.4.1 Laser cutting of CFRP
1.4.2 Laser surface pre-treatment of CFRP
1.4.3 Laser ablation of CFRP
1.5 Conclusion
2 Rotary ultrasonic machining of CFRP composites
2.1 Introduction
2.1.1 CFRP composites
2.1.2 Rotary ultrasonic machining
2.1.3 Purpose of this chapter
2.2 Rotary ultrasonic machining system set-up
2.2.1 Ultrasonic power supply
2.2.2 Ultrasonic transducer
2.2.3 Ultrasonic amplitude transformer (horn) and tool holder
2.2.4 Cutting tool
2.3 Input variables and output variables in RUM
2.3.1 Machining variables
2.3.2 Cutting tool variables and cooling variables
2.3.3 Workpiece properties
2.3.4 Output variables
2.4 Effects of input variables on output variables
2.4.1 Effects on cutting force
2.4.2 Effects on torque
2.4.3 Effects on cutting temperature
2.4.4 Effects on edge quality
2.4.5 Effects on surface roughness
2.4.6 Effects on burning of machined surface
2.4.7 Effects on tool wear
2.4.8 Effects on MRR
2.4.9 Effects on power consumption
2.4.10 Effects on feasible regions
2.5 Summary
3 High-speed robotic trimming of CFRP
3.1 Introduction
3.2 Machinability of CFRP
3.2.1 Evaluation of the cutting force
3.2.2 Assessment of the machinability of CFRP under high-speed robotic trimming
3.2.3 Cutting forces for robotic trimming experiments
3.2.4 Quality of robotic trimmed specimens
3.2.5 Surface quality
3.3 Conclusion
4 Numerical modeling of LFRP machining
4.1 Introduction
4.2 Orthogonal cutting
4.2.1 2D modeling
4.2.2 3D modeling
4.2.3 Thermal effects
4.3 Drilling
4.3.1 Comparison between simplified and complete drilling models
4.3.2 Thermal model of drilling
4.4 Conclusions
5 Delamination in composite materials: measurement, assessment and prediction
5.1 Introduction
5.2 Mechanisms of delamination
5.2.1 Peel-up delamination
5.2.2 Push-out delamination
5.3 Measurement of delamination
5.3.1 Visual methods
5.3.2 Image processing
5.3.3 Acoustic emission
5.3.4 Scanning acoustic microscopy (SAM)
5.3.5 Ultrasonic C-scan
5.3.6 Radiography
5.3.7 X-ray computerized tomography
5.3.8 Shadow moiré interferometry
5.4 Assessment of delamination
5.4.1 Delamination factor/conventional delamination factor
5.4.2 Delamination size
5.4.3 Two-dimensional delamination factor (Fd)
5.4.4 Damage ratio
5.4.5 Delamination factor
5.4.6 Adjusted delamination factor
5.4.7 Equivalent delamination factor
5.4.8 Refined delamination factor (FDR)
5.4.9 Shape circularity ( f)
5.4.10 Minimum delamination factor
5.5 Delamination in milling
5.6 Numerical prediction of delamination
5.6.1 Regression analysis
5.6.2 Artificial neural network (ANN)
5.6.3 FE simulation methods
5.7 Summary
6 Drilling of high impact polystyrene composites materials
6.1 Introduction
6.2 Materials and Manufacturing
6.3 Experimental work
6.4 Response surface methodology-based modeling of process parameters
6.5 Results and Discussion
6.6 Conclusions
7 A review on investigations in drilling of fiber reinforced plastics
7.1 Introduction
7.1.1 Delamination and delamination mechanism
7.1.2 Fabrication of polymer matrix composites
7.2 Drilling process
7.2.1 Delamination assessment
7.2.2 Effect of various machining parameters on delamination
7.3 Role of digital image processing in delamination assessment
7.4 Summary
Index

Citation preview

Advanced Composites Davim ∙ Machinability of Fibre-Reinforced Plastics

Also of Interest

Series: Advanced Composites J. Paulo Davim (Ed.) Published titles in this series (ISSN 2192-8983): Vol. 3: Metal Matrix Composites (2014) Ed. by Davim, J. Paulo Vol. 2: Biomedical Composites (2013) Ed. by Davim, J. Paulo Vol. 1: Nanocomposites (2013) Ed. by Davim, J. Paulo/ Charitidis, Constantinos A. Nanomaterials in Joining Constantinos A. Charitidis (Ed.), 2015 ISBN 978-3-11-033960-4, e-ISBN 978-3-11-033972-7

Nanoparticles Raz Jelinek, 2015 ISBN 978-3-11-033002-1, e-ISBN 978-3-11-033003-8

Nanocarbon-Inorganic Hybrids Next Generation Composites for Sustainable Energy Applications Eder, Schlögl (Eds.), 2014 ISBN 978-3-11-026971-0, e-ISBN 978-3-11-026986-4

Functional Materials For Energy, Sustainable Development and Biomedical Sciences Leclerc, Gauvin (Eds.), 2014 ISBN 978-3-11-030781-8, e-ISBN 978-3-11-030782-5

Nanotechnology Reviews Kumar, Challa (Editor-in-Chief) ISSN 2191-9089, e-ISSN 2191-9097

Machinability of Fibre-Reinforced Plastics | Edited by J. Paulo Davim

Editor Prof. Dr. J. Paulo Davim University of Aveiro Department of Mechanical Engineering Campus Santiago 3810-193 Aveiro, Portugal [email protected]

ISBN 978-3-11-029222-0 e-ISBN (PDF) 978-3-11-029225-1 e-ISBN (EPUB) 978-3-11-038887-9 Set-ISBN 978-3-11-029226-8 Library of Congress Cataloging-in-Publication Data A CIP catalog record for this book has been applied for at the Library of Congress. Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available on the Internet at http://dnb.dnb.de. © 2015 Walter de Gruyter GmbH, Berlin/Boston Cover image: gettyimages/thinkstockphotos, Abalone Shell Typesetting: PTP-Berlin, Protago TEX-Produktion GmbH, Berlin Printing and binding: CPI books, GmbH, Leck ♾ Printed on acid-free paper Printed in Germany www.degruyter.com

Preface Currently, fibre-reinforced plastics (FRPs) present great development to application in modern industry due to their excellent properties. FRPs have high tensile strength and stiffness, lightweight, corrosion resistance, non-magnetic properties. Therefore, FRPs are particularly attractive to replace conventional materials for many technological applications. Therefore, FRPs have the potential to replace conventional materials in various fields of application such as automotive, aerospace, marine and construction and rehabilitation of structures as well as in others advanced industries. As result of these potentials applications, exits a great necessity to understand the problems associates with the machinability of these group materials. Machinability refers to the relative ease with which a group the materials can be machined using appropriate tooling and cutting parameters. The present volume aims to provide recent information on machinability of fibrereinforced plastics – in seven chapters. The chapter 1 of the book provides information on Laser material machining of CFRP. Chapter 2 is dedicated to rotary ultrasonic machining CFRP composites. Chapter 3 described high-speed robotic trimming of carbon fibre reinforced polymers. Chapter 4 contains information on numerical modeling of machining of LFRPs. Chapter 5 described delamination in composite materials (measurement, assessment and prediction). Chapter 6 described drilling of high impact polystyrene composites materials. Finally, chapter 7 is dedicated to a review on investigations in drilling of fiber reinforced plastics. The present volume can be used as a research book for final undergraduate engineering course or as a topic on manufacturing with FRPs at the postgraduate level. Also, this book can serve as a useful reference for academics, researchers, manufacturing, mechanical and materials engineers, professionals in machining of FRPs and related industries. The interest of scientific in this book is evident for many important centers of the research, laboratories and universities as well as industry. Therefore, it is hoped this book will inspire and enthuse others to undertake research in this field of machining of fibre-reinforced plastics. The Editor acknowledges De Gruyter for this opportunity and for their enthusiastic and professional support. Finally, I would like to thank all the chapter authors for their availability for this work. Aveiro, Portugal, March 2015

J. Paulo Davim

Contents Preface | V List of contributing authors | XI F. Fischer, S. Kreling, D. Blass, R. Staehr, S. Bluemel, P. Jaeschke, K. Dilger 1 Laser material machining of CFRP – an option for damage-free and flexible CFRP processing? | 1 1.1 Introduction | 1 1.2 State of the art of machining of CFRP | 2 1.2.1 Cutting CFRP | 2 1.2.2 Surface pre-treatment of CFRP | 3 1.2.3 Shape cutting of CFRP | 5 1.3 Laser material interaction | 7 1.4 Laser material machining of CFRP | 8 1.4.1 Laser cutting of CFRP | 8 1.4.2 Laser surface pre-treatment of CFRP | 18 1.4.3 Laser ablation of CFRP | 21 1.5 Conclusion | 25 W. Cong, F. Ning 2 Rotary ultrasonic machining of CFRP composites | 31 2.1 Introduction | 31 2.1.1 CFRP composites | 31 2.1.2 Rotary ultrasonic machining | 33 2.1.3 Purpose of this chapter | 33 2.2 Rotary ultrasonic machining system set-up | 34 2.2.1 Ultrasonic power supply | 34 2.2.2 Ultrasonic transducer | 34 2.2.3 Ultrasonic amplitude transformer (horn) and tool holder | 36 2.2.4 Cutting tool | 38 2.3 Input variables and output variables in RUM | 38 2.3.1 Machining variables | 38 2.3.2 Cutting tool variables and cooling variables | 38 2.3.3 Workpiece properties | 39 2.3.4 Output variables | 40 2.4 Effects of input variables on output variables | 40 2.4.1 Effects on cutting force | 41 2.4.2 Effects on torque | 50 2.4.3 Effects on cutting temperature | 52

VIII | Contents 2.4.4 2.4.5 2.4.6 2.4.7 2.4.8 2.4.9 2.4.10 2.5

Effects on edge quality | 57 Effects on surface roughness | 59 Effects on burning of machined surface | 61 Effects on tool wear | 62 Effects on MRR | 65 Effects on power consumption | 66 Effects on feasible regions | 73 Summary | 77

M. Slamani, J.-F. Chatelain 3 High-speed robotic trimming of CFRP | 83 3.1 Introduction | 83 3.2 Machinability of CFRP | 85 3.2.1 Evaluation of the cutting force | 85 3.2.2 Assessment of the machinability of CFRP under high-speed robotic trimming | 87 3.2.3 Cutting forces for robotic trimming experiments | 90 3.2.4 Quality of robotic trimmed specimens | 93 3.2.5 Surface quality | 95 3.3 Conclusion | 99 H. Miguélez, N. Feito, C. Santiuste, J. Díaz-Álvarez, M. Rodríguez-Millán, X. Soldani 4 Numerical modeling of LFRP machining | 103 4.1 Introduction | 103 4.2 Orthogonal cutting | 105 4.2.1 2D modeling | 105 4.2.2 3D modeling | 110 4.2.3 Thermal effects | 117 4.3 Drilling | 121 4.3.1 Comparison between simplified and complete drilling models | 122 4.3.2 Thermal model of drilling | 126 4.4 Conclusions | 134 J. Babu, J. Philip, T. Zacharia, J. Paulo Davim 5 Delamination in composite materials: measurement, assessment and prediction | 139 5.1 Introduction | 139 5.2 Mechanisms of delamination | 140 5.2.1 Peel-up delamination | 140 5.2.2 Push-out delamination | 140

Contents | IX

5.3 5.3.1 5.3.2 5.3.3 5.3.4 5.3.5 5.3.6 5.3.7 5.3.8 5.4 5.4.1 5.4.2 5.4.3 5.4.4 5.4.5 5.4.6 5.4.7 5.4.8 5.4.9 5.4.10 5.5 5.6 5.6.1 5.6.2 5.6.3 5.7

Measurement of delamination | 141 Visual methods | 141 Image processing | 141 Acoustic emission | 142 Scanning acoustic microscopy (SAM) | 143 Ultrasonic C-scan | 143 Radiography | 144 X-ray computerized tomography | 145 Shadow moiré interferometry | 145 Assessment of delamination | 147 Delamination factor/conventional delamination factor | 147 Delamination size | 149 Two-dimensional delamination factor (Fd ) | 149 Damage ratio | 150 Delamination factor | 151 Adjusted delamination factor | 151 Equivalent delamination factor | 152 Refined delamination factor (FDR ) | 154 Shape circularity ( f ) | 154 Minimum delamination factor | 155 Delamination in milling | 156 Numerical prediction of delamination | 157 Regression analysis | 157 Artificial neural network (ANN) | 158 FE simulation methods | 159 Summary | 160

K. Palanikumar, T. Srinivasan, K. Rajagopal, J. Paulo Davim 6 Drilling of high impact polystyrene composites materials | 163 6.1 Introduction | 163 6.2 Materials and Manufacturing | 165 6.3 Experimental work | 167 6.4 Response surface methodology-based modeling of process parameters | 169 6.5 Results and Discussion | 171 6.6 Conclusions | 176 R. Pramod, S. Basavarajappa, J. Paulo Davim 7 A review on investigations in drilling of fiber reinforced plastics | 179 7.1 Introduction | 179 7.1.1 Delamination and delamination mechanism | 180 7.1.2 Fabrication of polymer matrix composites | 181

X | Contents 7.2 7.2.1 7.2.2 7.3 7.4

Drilling process | 181 Delamination assessment | 184 Effect of various machining parameters on delamination | 186 Role of digital image processing in delamination assessment | 190 Summary | 191

Index | 195

List of contributing authors J. Babu Department of Mechanical Engineering St. Joseph’s College of Engineering and Technology Choondacherry, 686 575 Kerala, India e-mail: [email protected] Chapter 5 S. Basavarajappa Department of Mechanical Engineering University B. D. T. College of Engineering Davanagere, 577 004 Karnataka, India e-mail: [email protected] Chapter 7 D. Blass Institute of Joining and Welding Braunschweig Technical University Langer Kamp 8 38106 Braunschweig, Germany Chapter 1 S. Bluemel Laser Zentrum Hannover e. V. Hollerithallee 8 30419 Hannover, Germany Chapter 1 Jean-François Chatelain Department of Mechanical Engineering 1100 Notre-Dame Street West Montreal (Qc), H3C 1K3, Canada e-mail: [email protected] Chapter 3 Weilong Cong Department of Industrial Engineering Texas Tech University Box 43061 Lubbock TX-79409, USA e-mail: [email protected] Chapter 2

J. Paulo Davim Department of Mechanical Engineering University of Aveiro Campus Universitário de Santiago 3810-193 Aveiro, Portugal e-mail: [email protected] Chapters 5, 6, and 7 J. Díaz-Álvarez Department of Biomedical and Aerospace Engineering University Carlos III of Madrid Avda. de la Universidad 30 28911 Leganés, Madrid, Spain Chapter 4 K. Dilger Institute of Joining and Welding Braunschweig Technical University Langer Kamp 8 38106 Braunschweig, Germany Chapter 1 N. Feito Department of Mechanical Engineering University Carlos III of Madrid Avda. de la Universidad 30 28911 Leganés, Madrid, Spain. Chapter 4 F. Fischer Institute of Joining and Welding Braunschweig Technical University Langer Kamp 8 38106 Braunschweig, Germany e-mail: [email protected] Chapter 1 P. Jaeschke Laser Zentrum Hannover e. V. Hollerithallee 8 30419 Hannover, Germany Chapter 1

XII | List of contributing authors S. Kreling Institute of Joining and Welding Braunschweig Technical University Langer Kamp 8 38106 Braunschweig, Germany Chapter 1 H. Miguélez Department of Mechanical Engineering University Carlos III of Madrid Avda. de la Universidad 30 28911 Leganés, Madrid, Spain e-mail: [email protected] Chapter 4 Fuda Ning Department of Industrial Engineering Texas Tech University Box 43061 Lubbock TX-79409, USA Chapter 2 K. Palanikumar Department of Mechanical Engineering Sri Sai Ram Institute of Technology Sai Leo Nagar, West Tambaram Chennai, 600 044 Tamil Nadu, India e-mail: [email protected] Chapter 6 Jose Philip Department of Mechanical Engineering St. Joseph’s College of Engineering and Technology Choondacherry, 686 575 Kerala, India Chapter 5 R. Pramod Department of Mechanical Engineering Amrita Vishwa Vidyapeetham University Bangalore, India e-mail: [email protected] Chapter 7

K. Rajagopal Department of Chemical Engineering Sri Sairam Institute of Technology Sai Leo Nagar, West Tambaram Chennai, 600 044 Tamil Nadu, India e-mail: [email protected] Chapter 6 M. Rodríguez-Millán Department of Mechanical Engineering University Carlos III of Madrid Avda. de la Universidad 30 28911 Leganés, Madrid, Spain Chapter 4 C. Santiuste Department of Continuum Mechanics and Structural Analysis University Carlos III of Madrid Avda. de la Universidad 30 28911 Leganés, Madrid, Spain Chapter 4 X. Soldani Department of Mechanical Engineering University Carlos III of Madrid Avda. de la Universidad 30 28911 Leganés Madrid, Spain Chapter 4 Mohamed Slamani Department of Mechanical Engineering Faculty of Technology University of M’Sila BP 05 cité Ennasr M’sila, 28000, Algeria e-mail: [email protected] Chapter 3 X. Soldani Department of Mechanical Engineering University Carlos III of Madrid Avda. de la Universidad 30 28911 Leganés Madrid, Spain Chapter 4

List of contributing authors | XIII

T. Srinivasan Department of Mechanical Engineering Sri Sairam Institute of Technology Sai Leo Nagar, West Tambaram Chennai, 600 044 Tamil Nadu, India e-mail: [email protected] Chapter 6

R. Staehr Laser Zentrum Hannover e. V. Hollerithallee 8 30419 Hannover, Germany Chapter 1 Tom Zacharia Department of Mechanical Engineering St. Joseph’s College of Engineering and Technology Choondacherry, 686 575 Kerala, India Chapter 5

F. Fischer, S. Kreling, D. Blass, R. Staehr, S. Bluemel, P. Jaeschke, K. Dilger

1 Laser material machining of CFRP – an option for damage-free and flexible CFRP processing? Abstract: With the increased implementation of carbon fiber reinforced plastics (CFRP), modern technologies for automated, highly productive and cost-efficient processing of CFRP are being sought, as is the maintenance of CFRP-based goods. Recent progress in laser system technology in terms of average power, beam quality and beam guiding systems enables innovative techniques for machining carbon fiberreinforced plastics. This work compares laser-based material machining such as laser cutting, laser material ablation and laser surface pre-treatment with the state-of-theart mechanical methods.

1.1 Introduction Today, carbon fiber reinforced plastics (CFRP) are widely used for aerospace, automotive and transportation applications. Induced by both major industrial projects such as the Airbus A350 (with a CFRP content of about 50 %) or the BMW i3, the first volume production vehicle with a 100 % CFRP passenger safety cell to improve the vehicle’s energy consumption and a planned annual output up to 40 000 vehicles, and medium-term issues such as electro-mobility, the use of lightweight materials is expected to move towards high volume production. However, with the increase in production volumes technologies for more productive and cost-efficient CFRP processing are being sought as is the maintenance of CFRP-based goods. Laser technology provides reliable high-speed processes in these industries. In contrast to isotropic materials like metal, CFRPs are anisotropic materials that consist of two parts: matrix and reinforcement. In CFRP, the reinforcement is carbon fiber; it provides the strength. The matrix is usually a thermoplastic or thermoset polymer, such as epoxy, which binds the reinforcements. Because of the carbon fiber reinforcement, composite materials are distinguished by their extremely high strength and rigidity. Low density, excellent damping properties and a high resistance to impact combine with exactly modifiable thermal expansion to complement the complex profile of characteristics. Due to the anisotropic properties, different approaches for cutting, surface pre-treatment and repair are required. Even this material group’s locally anisotropic behavior such as absorption, ablation threshold and thermal conductivity as well as the thermal sensitivity of the polymer matrix have been obstacles to industrial implementation of laser technology until now.

2 | 1 Laser material machining of CFRP – damage-free?

1.2 State of the art of machining of CFRP 1.2.1 Cutting CFRP One established technique to cut CFRP is milling, a mechanical machining process with rotating cutters. Milling of metals is a standard process with many years of industrial use in series production. Although the milling of composites is already widespread within the production of CFRP parts, a continuous need to improve the milling process is present. As a contact process, the milling of composites has several disadvantages and challenges. Two of these challenges can be directly related to the interaction of the cutter and the CFRP. Both the inhomogeneous and anisotropic characteristics as well as the build-up of composites can lead to damage, delamination and reduced quality [1]. A proper machining quality is especially challenging while machining unidirectional (UD) composites [2]. The force brought into the material by the milling process depends on several factors. Different studies describe the force as dependent on the feed rate; however, slow feed rates do not necessarily lead to a reduction of delamination. Hocheng et al. have shown that an undesirable roughness is generated for process parameters with increased cutting speed and a low feed rate [3]. This can be explained by heat generation during the process, which is increased by the existent low heat conductivity. In the studies, temperatures in the range of T = 200 °C to T = 500 °C were measured at the surface of the cutter [1, 4]. These can be reduced by increasing rotating speeds; this can also reduce the resulting forces. Machining composites becomes challenging because of the difference in hardness of the matrix and fibers. Tool wear is especially related to the carbon fibers. The existence of free fiber endings in the work area increases the friction, and loose fragments increase abrasion. Generally, it can be stated that tool wear depends on the mechanical process and build-up and volume fiber content of the material [1]. Tool life is connected to wear and is based on different parameters, but cutting speed has the biggest influence. Caprino and Langella stated that material removal by traditional techniques is still a subject of research, and the demand on developments in the areas of tools, machines and strategy is still high [1]. More research is especially needed in effective dissipation of heat and removal of particles during dry cutting. Among the unconventional laser processing methods such as ultrasonic, electric discharge, plasma or laser machining, the only processing method to have reached a certain industrial application level is waterjet machining or abrasive waterjet (AWJ) machining. Waterjet machining uses the principle of erosion, ablating the work piece material by means of a highly accelerated water jet. The intense acceleration up to speeds of ∼ 900 m/s is achieved by feeding pressurized water through a small orifice of ∼ 0.1–0.4 mm diameter. To enhance the ablation capabilities and ablation rate of this technique, abrasive particles, e.g. garnet, aluminum oxide or silicon oxide, are added after acceleration in a mixing chamber. The mixture is then directed by a mix-

1.2 State of the art of machining of CFRP |

3

ing/focusing tube onto the work piece. Feed motion is usually performed by using gantry robots or articulated arm robots. Typical advantages of AWJ cutting are versatility regarding the cutting of different material combinations, suitability for thick laminates, low cutting forces, the environment-friendliness and absence of thermal effects. Typical phenomena limiting AWJ cutting are the trailback effect, striations on the kerf surface, tapered geometry of the cutting kerf, delamination in layered structures and abrasive particle embedment in the kerf. Despite problems leading to a poor work result, other disadvantages include contaminated surfaces of the work piece which complicates further handling, the noise of the water jet and its limited accuracy [5–10].

1.2.2 Surface pre-treatment of CFRP For all processes requiring adhesion between any kind of coating and fiber-reinforced plastic (FRP) material (e.g. adhesive bonding), painting a surface as a pre-treatment process is generally required to ensure good adhesion. The main reason for this is due to surface contamination resulting from the use of release agents in the manufacturing process, which are required to ensure good release properties and demoulding without damage to the FRP part. However, residues of these may remain on the surface and thus affect adhesion of the adhesive or paint layer. State-of-the-art methods for surface pre-treatment are generally based on a mechanical removal of the matrix resin surface layer, including contaminations, or on a physical modification or activation.

Mechanical surface pre-treatment The most common methods for surface pre-treatment of FRP are grinding, grit blasting or the application of so-called peel-plies. These methods can be categorized as mechanical surface treatments, because the major mode of action is based on the mechanical removal of contaminations, together with the top resin layer. For most repair applications and also low-volume parts in the automotive industry, manual grinding – an easy-to-use process with a low investment – is the most common method. The contaminations are removed together with the top resin layer using sand paper or grinding fabrics. Depending on the intensity of the grinding process, reinforcing fibers, which are close to the surface, can also be damaged or partly removed. Figure 1.1 shows a scanning electron microscope (SEM) picture, magnified 200 times, of a grinded surface which also vividly demonstrates damaged or respectively cut fibers at an untreated area and thus the disadvantages of this process. These are basically a high dependency of the outcome on the worker’s skill, difficulties with regard to process control, low process speed and thus high manual effort and the ne-

4 | 1 Laser material machining of CFRP – damage-free?

5 kV

X200 100 μm 0674 1 fs 2012

Fig. 1.1: SEM picture of a grinded CFRP surface.

cessity of cleaning the surface after the grinding process [11]. In some applications, wet grinding is performed to prevent the occurrence of dust; however, this leads to the necessity of drying the cleaned parts afterwards. The characteristics and also the disadvantages named above may be transferred more or less directly to surface pre-treatment using conventional grit-blasting processes. Additionally, encapsulation is necessary in these processes to prevent contamination of the surroundings by grit and blasting residues. This leads to a challenge in regards to automated blasting process using industrial robots. In the aerospace industry, so-called peel-plies are a widespread method for surface pre-treatment. These are fabrics that are laminated into the FRP surface before curing the matrix resin. During curing, the matrix resin flows into the fabric, allowing one to ply it off the surface and remove it together with the topmost resin layer, leaving a comparatively reproducible, rough surface structure. The surface thus created depends on the fabric, because the structure is directly casted by the remaining resin layer. A microscopic picture of a typical peel-ply structure is shown in Fig. 1.2. Numerous researchers (e.g. [12–16]) have taken a closer look at the influence of the fabric and material of the peel-ply on the adhesion behavior of the resulting surface. Some of these publications show significant amounts of contamination, resulting both from residues of release agents originating from the peel-ply surface or fibers of the peel-ply itself on the surface, which show a negative impact on adhesion properties. As a result, peel-ply surfaces are often abraded in practice after the removal of the peel-ply.

Fig. 1.2: Microscopic picture of a typical peel-ply surface structure.

1.2 State of the art of machining of CFRP |

5

Physical surface treatment Physical treatment methods are characterized, contrary to the mechanical ones, by a direct physical interaction with the polymeric surface. The most common representatives of these methods are the atmospheric and low-pressure plasma processes. The effect of these methods is based on breaking bonds in polymer-bonds caused by highenergy particles in the plasma. This leads to the surfaces’ micro abrasion as well as a physical activation of the surface by the creation of reactive bonds. These again can interact directly or, respectively, form covalent bonds with the adhesive or base paint layer, enabling or increasing the bondability of low surface energy polymers, like most thermoplastic materials (e.g. PEEK and PA). In the context of pre-treating FRP, a key question about plasma processes is whether the residues of the release agent on the part surface can be completely removed by the plasma process, or whether a chemical activation of the release agent occurs which enables adhesion between the release agent and the adhesive or paint. In the latter case, the question is again, whether the adhesion between the release agent and FRP surface is sufficient to provide a strong bond that is also stable in the long term. For low-pressure plasma processes, a complete removal of surface contaminations is described, e.g. in [13, 17]. Contrasting results gathered with atmospheric plasma shown in [15] demonstrate adhesion failure if the amount of initial surface contamination is too high. Another primary question in relation to plasma surface activation is based on the long-term stability of the surface activation. This is due to the reactive bonds created by the plasma process interacting with the atmosphere, and thus the activation is reduced with time. A decrease of bond strength of carbon fiber-reinforced PEEK of about 60 % is shown in [17] if the time span between plasma activation and adhesive bonding is four hours. This result vividly demonstrates that the plasma surface must be considered more as a surface activation than a surface cleaning process.

1.2.3 Shape cutting of CFRP The shape cutting or ablation of single composite layers becomes necessary when a composite is damaged and has to be repaired [18]. Due to their properties, with CFRP it is not possible to replace the whole damaged area, the way it is done for metallic parts where the damaged area is excluded and a representative is inserted and welded to the original part. In contrast to the repair techniques of these parts, which are often very coarse and also in contrast to the cutting of CFRP mentioned above, the shape cutting of CFRP has to be very sensitive because of its application for scarf repairs (Fig. 1.3) [19]. In this and any other structural composite repair technique, it is the primary objective to restore the strength and stiffness of the damaged part [20] by a single ply-based

6 | 1 Laser material machining of CFRP – damage-free? Low velocity blunt impactor

Subsurface damage

(a) Scarf angle

Straight taper for scarf repair

(b) Stepped taper for stepped scarf repair

(c)

Over-ply

Scarf repair patch (d)

Over-ply

Stepped scarf repair patch (e) Fig. 1.3: Bonded composite repair. Presentation of typical damages (a) and different repair approaches (b)–(e) [1].

approach [21]. To reach this goal, it is necessary to achieve a sufficient load transfer through the repair patch [22]. The most efficient approach is to build up the patch with the same composite lay-up as the preliminary structure and then bond the repair patch in the same specific ply angle as this structure [22, 23]. Therefore it is necessary to be able to expose single layers and stop between single plies during the ablation process. With every mismatch in the fiber orientation of preliminary structure and repair patches, the load transfer through the repaired area and thus in the mechanical properties of the repaired composite part are weakened [24]. During industrial service processes, the shape cutting of CFRP is mostly done with tools, which were already mentioned in the section about cutting CFRP. It is mostly done by conventional machining processes like milling, grinding or sanding [18]. Based on the equal operations such as the cutting of CFRP mentioned above, challenges are also relevant for the ablation of single layers (e.g. heat-caused delamina-

1.3 Laser material interaction

|

7

tion, fiber pullout, etc.). Above all, there are challenges related to the adhesive-based repair process. The most important fact is that almost all conventional processes have to be done with an additional cooling fluid to reduce the heat transfer in the CFRP [18]. These cooling fluids have different negative aspects for the bonding process. First, the exposed fibers carry the fluid into the CFRP; and second, the cooling fluids consist of lubricants [25]. Both lower the adhesion of the bonding area and can later diffuse in the adhesive which then reduces its cohesive strength [26]. In addition, there are some challenges resulting from the most common repair applications and the repair process itself. With common tools, shape cutting is done manually (mostly with handheld machining tools), and the quality of the scarf is highly dependent on the technician’s skills [27]. This leads to the fact that even on flat surfaces, shape cutting is a time-consuming procedure, when a sufficient quality of the scarf in terms of angle and geometry is required. With conventional methods, shape cutting of typical damage takes around ten hours. When it comes to more complex geometries this time will be largely increased. So it has to be concluded that the available shape cutting techniques are not very efficient in terms of the related repair process of CFRP [18]. Therefore the main challenges are the sensitivity of the process to exposing single fiber layers and reducing process time [24].

1.3 Laser material interaction Due to its heterogeneous and anisotropic character, composite materials such as CFRP are known to be very difficult to machine [66]. Due to the robust fiber material, a strong wear on the deployed expensive diamond-coated tools causes high costs in production. Even non-contact laser machining is problematic, since a very high intensity in the range of MW/cm2 to GW/cm2 is required for a thorough degradation of carbon fibers, while the polymeric matrix material has a low ablation threshold and shows high transparency for a wide range of relevant laser wavelengths [52]. The transmission spectra of thermoset matrix materials displayed in Fig. 1.4 explain this phenomenon. Near the typical infrared (IR) wavelength of solid-state lasers (λ = 1000 nm), transmission is very high, which means that a defined ablation of bulk material is not possible. Close to the mid wavelength IR as emitted by CO2 -lasers (λ = 10 600 nm), transmission drops but the strong thermal characteristics of the radiation creates a high risk of composite damage such as delamination [61]. The photothermal behavior of this laser source’s derives from the low photon energy of approximately 0.12 eV (λ = 10 640 nm). This energy is not sufficient for direct intermolecular bond-breaking of the treated materials, since the typical bonding energy of polymeric matters (e.g. C–N: 3.04 eV or C–C: 3.62 eV) exceeds this quantum drastically [67]. Thus, for this

8 | 1 Laser material machining of CFRP – damage-free? 308nm 355nm 1064nm

10600nm

1.0 0.9 0.8

Absorption [–]

0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 250

500

750

1000

1250 1500 1750 Wavelength [nm]

8000

9000 10000 11000

Fig. 1.4: Transmission characteristic of a thermoset matrix material (epoxy resin) with indication of relevant laser wavelengths.

wavelength regime, the major degradation impact on CFRP is induced by heat, which causes carbonization and delamination within the laminate. Accordingly, UV-laser radiation with a wavelength of λ < 400 nm features a low transmission and therefore good results for precise material ablation. With photon energy of 3.49 eV (λ = 355 nm), the process is predominantly characterized by a photochemical mechanism, since most polymeric bonds can be cracked by a single photon. Industrially available UV-laser sources provide a low average power of < 50 W. Hence, to insure the high intensity necessary for sublimation of carbon fibers, short-pulsed radiation (τR = 30 ns) is required.

1.4 Laser material machining of CFRP 1.4.1 Laser cutting of CFRP In contrast to the fabrication of CFRP parts, laser processing is already common in metal parts fabrication. The major benefits that promote laser use in production are flexibility for 3D machining and non-contact operation. Hence, material-related tool wear is no issue with lasers, and they offer a process free of forces. Laser processing, cutting in particular, of CFRP is a challenging task. In applying conventional technologies, which are mainly based on contour cutting strategies, cut-

1.4 Laser material machining of CFRP |

9

ting geometry is passed once by use of a laser cutting head. This cutting head mainly consists of optics for collimating and focusing the laser beam onto the work piece and provides an assisting gas stream that is responsible for blowing out the molten and sublimated material inside the cutting kerf. The laser head is guided by robots or axis systems. Basically, the kind of process management is similar to that of milling and abrasive waterjetting. If CFRP parts and components in a thickness range of a few mm have to be cut by applying contour cutting strategies, laser systems in the multi-kw range need to be used. Such high power lasers allow high feed rates of several m/min but, as a thermal process during the cutting procedure, heat-induced damage appears at the cutting edge of processed CFRP (compare Fig. 1.8). This heat-affected zone (HAZ) of the work piece is prevalent for CFRP due to different melting temperatures and thermal conductivities of the polymeric matrix material and the reinforcing carbon fibers. The formation of a HAZ starting at the cutting edge and extending into the bulk material can be observed for both thermoset and thermoplastic matrices. Contour cutting provides the highest energy input in terms of applied line energy, resulting in highest temperatures within and in close proximity to the laser material interaction zone. Consequently, the most extensive damage at the cutting edge is observed for such processing methods. Apart from contour cutting, there is also multi-pass cutting. While using this technology, the laser radiation is guided by mirrors or by optical fibers to a galvo scanner system; it focuses the laser beam by means of specialized optics (F-Theta-optics) and deflects the beam across the CFRP surface by means of two movable mirrors (compare Fig. 1.3 or Fig. 1.4). Due to the application of mirrors, the laser beam can be guided at a maximum speed of up to several m/s across the surface of the work piece along the geometry to be cut. In contrast to contour cutting, this geometry is passed several times by the laser beam at high speeds in multipass cutting. In this case, lower line energies are applied, resulting in a reduced thermal impact at the interaction zone and at cutting edge. As a result of this, the HAZ decreases and damage to fiber matrix adhesion is minimized. Between two passes of the laser beam, the CFRP cools down, which again reduces the process temperatures and therefore the critical temperatures inside the bulk material. In addition to the applied laser cutting strategy, the laser source is of great importance with respect to achieving cutting quality and processing speed. As it is known from the micro machining of metals and non-metals, ultra short-pulsed laser systems that provide pulse durations in the range of picoseconds and nanoseconds offer excellent damage-free processing qualities. Emission wavelengths of such laser sources are available in the near infrared or frequency converted in the visible and ultraviolet range of the electromagnetic spectrum. Due to very short laser-material interaction times, a nearly cold ablation occurs. However, such laser sources are commercially available only up to medium output powers of a few hundred watts, which is quite low for cutting applications in the macro range. When using such laser sources, only low feed rates can be achieved, typically in the range of some cm/min. Additionally,

10 | 1 Laser material machining of CFRP – damage-free? the beam delivery and forming is complex and, typically, laboratory conditions are required. With the increasing need for new cutting technology for macro machining, it is becoming increasingly important for composite production lines to take advantage of industrially established laser sources. Therefore, the following detailed analysis of the laser cutting process, composition of the HAZ as well as the correlation between mechanical properties of laserprocessed CFRP samples and the extent of the HAZ focuses on the use of continuous (cw) and pulsed (pw) emitting laser systems with wavelengths in the near infrared was performed. For both laser types, medium output powers in the kW range are available. A summary of the cutting principles that can be applied using a galvo scanner system can be found in Fig. 1.5, while Fig. 1.6 shows a typical setup with a galvo scanner. A quality evaluation of laser cut CFRP can be conducted by measuring the HAZ using cross-section specimens, scanning electron microscopy (SEM), dye penetrant tests and conducting mechanical testing on laser cut specimens. In optical cross-section micrographs, the HAZ can be distinguished by visible areas with fully degraded polymer matrix (to almost exclusively carbon) and charred fibers (AHAZ1 ), and by areas with partially degraded polymer matrix and undamaged fibers (AHAZ2 ). For thermoplastic matrix materials, a third HAZ can be defined (AHAZ3 ) in which the melting temperature of the matrix was exceeded while outgassing of volatile components (e.g. water) took place, causing porosity and bloating [28]. The different HAZs are illustrated using carbon fiber-reinforced polyphenylene sulphide Contour cutting – Only one pass to achieve a full cut – Slow scanning speeds

Laser beam Evaporated material

HAZ

Multipass cutting – Multiple repetitions on same contour to achieve a full cut – High scanning speeds

Continious wave laser (cw) – Combined melting and evaporting – Higher cutting rates – More distinctive HAZ Evaporated material Plasma

Molten material

Pulsed laser Laser – Mainly evaporation beam – Lower cutting rates – Smaller HAZ Plasma

HAZ

Fig. 1.5: Schematic principles of laser cutting strategies and laser-material interaction for different operation modes.

1.4 Laser material machining of CFRP |

Galvoscanner

11

Collimator

Optical fibre Laser source F-Theta optics

Laser beam

CFRP sheet Fig. 1.6: Schematic setup of a cutting system with a galvo scanner.

Unmodified material

Cutting edge

βHAZ1 βHAZ2

b

βHAZ3

HAZ (a)

1000 μm

AHAZ1

AHAZ2

AHAZ3

(b)

1000 μm

(c)

2000 μm

Fig. 1.7: Appearances of the HAZ for different materials. (a) CF/PPS, (b) CF/PEI, (c) CF/Epoxy.

(CF/PPS, Fig. 1.8 (a)), carbon fiber-reinforced polyetherimide (CF/PEI, Fig. 1.8 (b)) and carbon fiber-reinforced epoxy (CF/epoxy, Fig. 1.8 (c)) as an example. Since heat conduction is highly dependent on fiber orientation, the extent of the HAZ can vary from layer to layer (Fig. 1.8, right). An average width of the HAZ βHAZ can be calculated by dividing the area AHAZ by the material thickness b (equation (1.1)): βHAZ =

AHAZ b

(1.1)

SEM allows for the examination of the kerf area or the kerf surface. It mainly reveals the appearance of thermal cracks, exposed fibers due to evaporated matrix as well as general shape of the fiber ends. Several publications have shown characteristic cone-shaped fiber ends, as well as thermal cracks, exposed fibers or cavities for laser processing strategies and parameter combinations accompanied by large heat input (Fig. 1.8 (a)). Such disadvantageous strategies or parameter combinations usually include contour cutting or multi-pass cutting using slow speeds and no time periods between passes (Fig. 1.8 (a) and (b)). Improvements are observed when increasing the number of passes (together with the

12 | 1 Laser material machining of CFRP – damage-free?

(a)

(b)

(c)

(d)

CF/PEI, Contour

CF/PEI, Multipass, tΔ=0s

CF/PEI, Multipass, tΔ=3s

Cone-shaped and exposed fiber ends

Fig. 1.8: SEM captures of laser cut CFRP.

scanning speed) or by adding time periods between the passes (Fig. 1.8 (c)). Besides SEM, dye penetrant testing can help to reveal cracks and cavities. Figure 1.9 shows a dye penetrant applied on CFRP specimens cut with different parameters. Red colored areas represent regions with cavities or cracks. It is clearly visible that an increase in scanning speed and number of passes leads to an improvement in the cutting kerf quality.

(a)

(b)

CF/PPS, Multipass, vs=0.6m/s, n=6, tΔ=0s

CF/PPS, Multipass, vs=3m/s, n=75, tΔ=7s

Fig. 1.9: Dye penetrant test on laser cut structures.

The quality of the cutting edge based on the criteria described in the previous section can be influenced by several factors. The general factors will be explained based on the example of cutting with a continuous wave laser. Accordingly, the characteristics and differences for cutting with a pulsed laser will be explained separately. The first approach for the cutting of CFRP was the contour cutting with a laser cutting head, where an airstream is ejected coaxially to the laser beam through a nozzle. Several investigations have shown that cutting speed is a major factor in the size of the HAZ [29–32]. The size of the HAZ can be reduced by increasing cutting speed vc , whereas, for increased laser power PL , the cutting speed can be further increased. In this case, the minor drawbacks of additional laser power are compensated for by the more distinctive positive effects of the increased cutting speed. A factor describing the relation between laser power and cutting speed is the line energy EL . Line energy is

1.4 Laser material machining of CFRP |

13

defined as the energy brought into the material per unit length and can be calculated by equation (1.2): P (1.2) EL = L . vc The minimum required line energy can be identified by determining the maximum cutting speed for a certain laser power. Doing so for one laser power, the maximum achievable cutting speed for a different laser power can be estimated. The minimum required line energy depends on the material, optical setup and laser. Cutting experiments with a disk laser emitting at a wavelength of λ = 1030 nm with a laser power of PL = 16 kW have shown that HAZ cannot be avoided entirely with a contour cut (Fig. 1.10). The area of the HAZ decreases steadily for increasing cutting velocities until it converges asymptotically to a minimum.

Average width of HAZ [mm]

0,5 0,4 0,3 0,2 0,1 0 0

10

20

30

40

50

60

70

80

Cutting speed [m/min] Fig. 1.10: Area of the HAZ depending on the cutting speed for CF/PPS (d = 1.2 mm) using a PL = 16 kW cw laser.

In order to realize a further reduction of the HAZ, the multi-pass strategy can be used [28, 33–35]. For this strategy, the laser is guided multiple times at high scanning speed vs over the contour in order to achieve a complete cut. By knowing the required line energy and modifying equation (1.2), the necessary number of passes n can be estimated. Equation (1.2) is modified by adding the number of passes n and exchanging cutting speed and scanning speed (equation (1.3)): EL =

PL ⋅ n . vs

(1.3)

High velocities with high precision are best achieved by using a galvanometer scanner to move the laser beam relative to the surface of the CFRP. The multi-pass strategy at

14 | 1 Laser material machining of CFRP – damage-free? high scanning speeds reduces the laser material interaction time for a specific point along the cutting edge. Additionally, the emerging time period between two successive passes of a point along the cut allows for the heat to dissipate out of the cutting region. High scanning speeds can have the opposite effect on the HAZ if the time period between two passes is too small. Here the time period depends on the scanning length ls and the scanning speed vs . This effect can be seen in Fig. 1.11. All data points for each laser source were taken at constant line energy and scanning length; only scanning speed and number of passes were varied. For cuts with a cw laser at a laser power of PL = 4 kW, the maximum width of the HAZ increases for scanning speeds over vs = 10 m/s and the corresponding number of passes. This effect can be explained by heat accumulation within the cutting region; heat is unable to dissipate before new energy is introduced into the material. Using a pulsed laser source with an average laser power of PL, avg = 750 W and a pulse duration of tp = 30 ns, this effect is unrecognizable for scanning speeds up to vs = 60 m/min (Fig. 1.12). Pulse overlap [%]

Average width of HAZ [mm]

100 2,5

99

98

97

96

95

94

93

Cw laser Pulsed laser

2,0 1,5 1,0 0,5 0,0 0

10

20

30

40

50

60

Scanning speed [m/min] Fig. 1.11: Average width of the HAZ depending on scanning speed and pulse overlap for a cw laser and a pulsed laser source.

The effect of heat accumulation for high scanning speeds or short scanning length can be compensated for by inserting an additional time break tΔ between repetitions in order to increase the time period between two passes. The same strategy can also be used to further decrease the size of the HAZ; this can be achieved for a specific laser power and scanning speed. Figure 1.12 shows the influence of cutting length on the size of the HAZ for investigations with a nanosecond pulsed laser using an average laser power of PL,avg = 750 W and a scanning speed of vs = 8.16 m/s. A visualization of the influencing factors laser power, break time, cutting speed and number of passes is shown in Fig. 1.13.

1.4 Laser material machining of CFRP |

0

100

Cutting length [mm] 300 400 500

200

600

700

15

800

Average width of HAZ [mm]

2,5 2,0 1,5 1,0 0,5 0,0 0

1

2 3 Loop time Δtc [s]

4

5

6

Fig. 1.12: Average width of the HAZ depending on the loop time and cutting length, respectively.

500 μm

CF/PPS, PL=1kW, Contour cutting, v8=0.1m/s

500 μm

500 μm

CF/PPS, PL=1kW, Multipass cutting, n=6, v8=0.6m/s, tp=0s

500 μm

500 μm

CF/PPS, PL=1kW, Multipass cutting, n=6, v8=0.6m/s, tp=3s

500 μm

CF/PPS, PL=1.5kW, Multipass cutting, CF/PPS, PL=1.5kW, Multipass cutting, CF/PPS, PL=1.5kW, Multipass cutting, n=6, v8=1.25m/s, tp=3s n=12, v8=3.15m/s, tp=3s n=24, v8=6.2m/s, tp=3s

Fig. 1.13: Influence of laser power, cutting strategy, break time between passes and cutting speed on cutting quality.

Generally, the factors influencing the achievable quality are identical for the pulsed and continuous wave laser processing of CFRP. In addition to the laser power and scanning speed, a process with a pulsed laser source can also be described by the parameters pulse repetition rate and pulse duration. Additionally, the parameters pulse overlap and pulse energy are used to define and describe a laser process. Pulse over-

16 | 1 Laser material machining of CFRP – damage-free? lap is defined as the percentage area two successive laser spots overlap and it depends on focal diameter, scanning speed, pulse duration and repetition frequency. Several investigations came to the conclusion that a decrease in pulse overlap, which can be achieved by increased scanning speed or decreased repetition rate, influences the generation of a HAZ positively [36, 37]. This effect is exemplified in Fig. 1.13 for cuts with a nanosecond pulsed laser. The pulse energy is calculated by the quotient of average laser power and repetition rate. Investigations with a nanosecond pulsed laser have shown that increasing the pulse energy will increase the HAZ [38]. As described before, a coaxial gas stream is used during laser cutting with a cutting head. This provides two additional factors that could influence cutting edge quality. These factors are the gas pressure p and the type of assisting gas. The most common assisting gas is nitrogen, which is used because of its inert effects [39, 40]. Negarestani et al. suggest that the use of a gas mixture containing approximately 88 % nitrogen and 12 % oxygen at a pressure of p = 8 bar improves cutting quality, for example, in terms of less fibers pulled out [35]. Inert gases such as nitrogen are also used for many remote cutting studies. Two different approaches are commonly used. The first approach is to use cross jets or other nozzles to blow nitrogen over the surface of the CFRP within the working area of the scanner. Second, the cutting process is performed in processing chambers that are flooded with an inert gas. Sato et al. investigated nanosecond laser cutting within an argon ambiance and stated that this procedure might be a useful method to reduce the HAZ [41]. Although processing within a flooded process chamber seems to improve cutting edge quality for certain materials and laser sources, this approach has its drawbacks; they are related to practicality for big parts and 3D processing with handling systems such as robots and axis systems. Minimizing the damage (HAZ) by using multi-pass cutting strategies with high beam deflection speeds and a high number of repetitions leads to heat accumulation in the material, especially in thick laminates. Such processes require long cooling periods between passes to ensure a high cutting quality level which, in turn, leads to high total processing times. To meet the demand of the industry for efficient processes and in order to reduce process times, several options for improving efficiency can be chosen: 1. Depending on the size of the structures and the number of parts needed, the simultaneous processing of multiple parts can be achieved if the size and geometry of more than one part fit into the limited working field of the scanning system. 2. If the size of multiple parts or even a single part exceed the working field of a scanner, tiling a contour in combination with a linear stage system allow for an extended working field. In this case, a single pass would be applied before moving the working field of the scanner on to another position using the linear stage and applying the next pass in this position (see Fig. 1.15).

1.4 Laser material machining of CFRP |

17

3.

Optimized cooling strategies can help dissipate heat and reduce processing times especially for long processes with heat accumulation. 4. In order to ensure a constant cutting quality level, concurrent process temperature monitoring and parameter control based on non-contact temperature measurement can be implemented [42]. As a final proof of quality, laser cut structures especially need to demonstrate their compliance to the mechanical requirements of an end user. Only few approaches including the correlation of the HAZ and the mechanical properties have been published. Many publications similarly suggest that a general trend exists for decreasing strength properties with increasing heat input and therefore increasing the zone affected by heat. Nevertheless, minimizing heat input by using adapted processing parameters and strategies leads to marginal differences in mechanical properties if compared to reference processing methods such as milling or waterjet cutting. By using special sample geometries or specimen sizes apart from international standards, the sensitivity of a specimen to the heat-affected zone can be increased and visualized in the strength properties [28, 32, 43–45].

Fig. 1.14: Example of a cutting layout of multiple structures on CF/Epoxy with a thickness of b = 3.2 mm.

Aside from process quality and efficiency improvements, it was shown that the use of optimized strategies help to reduce hazardous emissions originating from the laser cutting process. As shown in Fig. 1.14, multi-pass cutting strategies with additional breaks were found to decrease the emission of organic volatile components, carbon monoxide and particles for CF/epoxy as well as for CF/PPS [46].

18 | 1 Laser material machining of CFRP – damage-free? 70 Concentration [mg/m3]

60 50

CO VOCs Particles

40 30 20 10 0 Contour

Multipass Multipass with breaks

Contour

CF/PPS

Multipass Multipass with breaks CF/Epoxy

Material & Cutting strategy Fig. 1.15: Overview of the concentrations of hazardous substances measured during laser cutting of CF/epoxy and CF/PPS using different cutting strategies [46].

1.4.2 Laser surface pre-treatment of CFRP Based on the disadvantages of the pre-treatment methods mentioned above together with the increasing use of composite parts, especially in the aerospace industry since the late 1980s, the interest in using lasers as free-of-wear tools with high reproducibility for composite pre-treatment has increased significantly. Hence numerous research projects have been initiated and the results have been published [47, 48]. Different wavelengths, from the UV [49–53] to visible green [54] and the near IR-range (∼ 1064 nm) [55], which is emitted by conventional fiber-lasers up to the mid IR-range (10 600 nm) [56, 57], and their influence on surface morphology and achievable bond strength have all been considered. The following section covers comparative investigations in regard to the laser material interaction and influence of laser pre-treatment on adhesive bondability, which are published among others in [58]. The investigations shown here were performed using a XeCl Excimer laser with a wavelength of 308 nm, a fiber laser with a wavelength of 1064 nm and a q-switch CO2 -laser with a wavelength of 9250 nm. All three lasers emit laser pulses with a length in the ns regime. The CFRP samples are manufactured from an aerospace grade prepreg material; a typical 120 °C curing epoxy film adhesive was used for the bonding tests. The most commonly applied test for the evaluation of the bondability of a surface after pre-treatment is the single lap shear test. For this test, two specimens with a size of 100 × 25 mm2 are pre-treated and then adhesively bonded together with an overlap of 12.5 mm. Figure 1.16 gives a summary of the bond strength achieved in the lap-shear test after pre-treatment with three different laser sources. Besides bond strength, the fracture mode is depicted using abbreviations according to DIN EN ISO 10365,

1.4 Laser material machining of CFRP |

19

50% AF; 50% CF

100% CSF

100% CSF -Matrix

0

1000% CF

25

60% CF; 40% AF

50

5% AF: 90% CF; 5% CSF

75

90% CF; 10% CSF

Lap shear strength / reference strength [%]

100

manual 308 nm 308 nm 1064 nm 1064 nm 9250 nm 9250 nm abrading 0,6 J/cm2 0,6 J/cm2 7 J/cm2 4 J/cm2 0,5 J/cm2 0,66 J/cm2 2 Pulses 16 Pulses 3 Pulses 3 Pulses 24 Pulses 30 Pulses

Fig. 1.16: A summary of bond strength achieved in the lap-shear test after pre-treatment with three different laser sources.

CF meaning cohesion failure, AF adhesion failure and CSF cohesion substrate failure or delamination inside the adherent, respectively. To increase the comparability, first the bond strengths re referred to as the ones determined for manually abraded specimens that predominantly fail inside the adhesive. This is the theoretically achievable maximum, because the cohesive strength of the adhesive used limits the strength of the whole compound. Second, the accumulated energy are calculated for each of the laser pre-treatment parameters. This parameter is given by the multiplication of the spot fluence with the number of pulses that interact with each aerial increment of the surface. The latter size can be calculated considering the spot overlap in feed and hatch direction (compare Fig. 1.20). These values are given in Tab. 1.1, they correspond to the parameters used for single lap-shear tests shown in Fig. 1.16. Besides the bond strength achieved, a comparison of laser parameters and thus an investigation of the laser-material interaction can be performed considering the surface appearance after pre-treatment. A selection of typical surface appearances is given in Fig. 1.17. Combining the results from Figs. 1.17 and 1.18 with the accumulated surface energies and appearances in Tab. 1.1 enables an assessment of the occurring laser-material interaction and gives the possibility to attain a fundamental understanding of the effects. Using 308 nm laser wavelengths, for which matrix absorption is about 95 % (compare Fig. 1.4), it can be assumed that laser radiation only interacts with the top resin layer, causing photo-thermal and possibly (because of the comparatively high photon energy) partial photo-chemical removal of the top matrix resin layer. In this case, the

1000um

20 | 1 Laser material machining of CFRP – damage-free?

20kV

X50 500μm 0745 ifs 2012

Resin remaining after pre-treatment with 308nm

20kV

X50 500μm 0738 ifs 2012

Fibers exposed after pre-treatment with 308nm

1000um

Resin remaining after pre-treatment with 1064nm

30kV

X50 500μm 2069 ifs 2010

Fibers exposed after pre-treatment with 1064nm

Fig. 1.17: A selection of typical surface appearances.

Tab. 1.1: Summary of the laser parameters shown. Wavelength [nm]

Pulse fluence [J/cm2 ]

No. of pulses [—]

Accumulated energy [J/cm2 ]

Surface appearance

308 308 1064 1064 9250 9250

0.6 0.6 7 4 0.5 0.66

2 16 3 3 24 30

1.2 9.6 21 12 12 19.8

Resin remaining Fibers exposed Visible fiber damage Resin remaining Resin remaining Fibers exposed

matrix is removed from the top and also selective matrix removal without fiber damage is possible if the pulse fluence is lower than the ablation threshold of the carbon fibers [52, 53]. This method of removal is vividly demonstrated by the upper and lower left surface images in Fig. 1.18, which also correspond to the surfaces created with twoand 16-pulse pre-treatment with 308 nm; these were adhesively bonded and tested in the lap-shear test afterwards. Evaluation of the fracture surfaces enables an inference on the surface state after pre-treatment. The specimens that were treated with only two pulses show a significant amount of adhesion failure, which means failure between the adhesive and specimen surface. This hints on residues of a release agent on the surface. However, the bond strength achieved is in the same magnitude as the abraded specimens and the ones treated with 16 pulses and thus corresponds to an accumulated energy that is eight times higher.

1.4 Laser material machining of CFRP |

21

The surface appearances as well as the resulting bond strength after pre-treatment are similar for the M-IR laser radiation with a wavelength of 9250 nm. However, the specimens that were treated with the lower intensity again show a significant amount of adhesion failure and in this case also reduced bond strength. Considering the pulse fluence, which is in the same magnitude as for the 308 nm excimer laser, but the higher amount of pulses or respectively accumulated energy required to achieve a surface with exposed fibers, it can be assumed that a different kind of interaction occurs due to the lower photon energy. For the solely thermal ablation at the M-IR wavelength, a higher amount of energy is required to completely remove the surface contaminations. However, due to high matrix absorption and comparatively low pulse fluence, no visible fiber damage occurs and also no material damage can be observed. In contrast to the UV and M-IR wavelength, the absorption of matrix resin is low for the N-IR wavelength of 1064 nm; this results in surface appearances as shown in the upper and lower right image in Fig. 1.18. Due to the low absorption, the laser radiation directly interacts with the carbon fibers and causes a mostly thermal effect in the interface between fiber and top matrix layer. In the single lap shear tests, the surfaces with the closed matrix layer remaining show very low bond strength and delamination of this matrix layer. If the intensity or accumulated energy is increased respectively, the matrix layer delaminates from the surface. However, residues of the resin remain on the surface (compare Fig. 1.16, lower right image). The specimens on which bonding is performed on this surface show delamination failure underneath the top fiber layer. However, the strength of the bond is in the same magnitude as the abraded ones. Nevertheless, for most applications this failure mode cannot be accepted because the energy absorption during the fracture is significantly lower. The results from the comparative bonding tests presented here show a high influence of wavelength on the result of achievable strength and failure mode. This is, on the one hand, caused by the dependence of matrix resin absorption on wavelength and thus the mode of interaction. On the other hand, it is dependent on whether photo-thermal or photo-chemical ablation is the predominant ablation mechanism. However, the results vividly demonstrate that with the use of wavelengths that have a high absorption in the matrix and appropriate parameters, an efficient pre-treatment of composites is especially possible and no damage is caused to the material.

1.4.3 Laser ablation of CFRP As mentioned before, the structural repair of CFRP is one of the main challenges to establish CFRP parts in structural applications especially for the aerospace industry, and the ablation of single fiber layers or plies has been the increasing focus of industrial and scientific research.

22 | 1 Laser material machining of CFRP – damage-free? Due to the fact that CFRP’s repair process is a relatively new topic, several approaches based on machining methods for classical materials were investigated. One of them is to use laser radiation to ablate single plies, which was based on the experience in the cutting of CFRP in the 1990’s [59]. The advantages mentioned such as non-contact operation, high flexibility and accuracy are also highly relevant to promote laser radiation as a tool to ablate CFRP [18]. The tools presented for the cutting process (scan heads, etc.) can also be implemented for the ablation process; there is a different approach in the laser source’s application [60]. As mentioned above, cutting CFRP was and is still done with IR lasers, because this process needs high average laser powers to work sufficiently. The approach for the ablation of CFRP is to use laser sources that emit radiation in the ultraviolet range of the electromagnetic spectrum (typically three times frequency converted solid state laser with a wavelength of 355 nm) [52]. This approach is based on different requirements for the ablation process compared to the cutting process, which promotes these laser sources. The repair process is not integrated in a normal production process and therefore short process times are not necessarily needed. For this reason it can be accepted that the average power of such UV-lasers is much lower (typically 20 to 400 W) than those of lasers that are used to cut CFRP. In addition, these laser sources are promoted because the interaction with the material is less thermally effected than those of the IR-laser [61]. Taken into account that the exposed and later on repaired area has to transfer load, a big HAZ within this area has to be avoided. While it may be adoptable for edges of cut composite parts, the HAZ and its weakening effects on the inter- and intra-laminar composite strength cannot be accepted in load transferring areas; it must be as small as possible. Compared to IR-laser sources, the HAZ caused by UV-laser sources is much smaller, even with the same average output power [62]. This can be explained with the photo-chemically dominated interaction between laser radiation and material. Figure 1.18 already shows the correlation of the HAZ size and the laser source respective to its decreasing wavelength. In this study, epoxy-based CFRP specimens were treated with laser radiation from UV to IR. The maximum size of the HAZ caused by IR-radiation (1064 nm) is almost doubled compared to processing with a laser radiation of 532 nm. In addition, this effect is even more developed when the wavelengths become shorter – in the range of UV-radiation. So the small HAZ and the photo-chemically dominated ablation process are the main factors which qualify UV-lasers for the ablation of CFRP [62]. The ablation process itself is normally realized by hatching the area with parallel lines. Due to the fact that there is a small dependency between the hatching direction and the fiber orientation for the ablation depth (especially along and across to the fiber orientation), the specific fiber orientation has to be considered to achieve a solid ablation process. To reduce the influence of this dependency, there should be a balanced approach in hatching direction for each cycle. For example, Fischer et al. in-

1.4 Laser material machining of CFRP |

23

Average power [W] 0

2

4

6

8

10

12

14

16

18

20

60 1064 nm 532 nm

50

HAZ [μm]

40 30 20 10 0 0

10

20

30 40 50 Peak fluence [J/cm2]

60

70

80

Fig. 1.18: Correlation of fluence and HAZ-size for different wavelengths [62].

y – Hatch direction

vestigated the ablation of fabric thermoplastic CFRP. To achieve a stable process, the hatch direction was alternated 0–90° in each cycle [52]. Furthermore, the ablation process is then based on the correlation between ablation depth and treatment cycle (considering the specific laser parameters, the ablation depth per cycle is in the range from one to some tens of μm with common UV-lasers). For a dimensional ablation, another relevant parameter is the hatch distance; it represents the distance between two consecutive treatment lines (Fig. 1.19).

Spot diameter

V Hatch distance

x – Feed direction Fig. 1.19: Prescinded hatching of an area.

The ablation depth per cycle normally decreases with an increase in hatch distance. This effect is non-linear and based on the increased laser energy per area. It can be correlated with thermal interactions of laser radiation and composites. Especially when the hatch distance is smaller than the spot diameter, thermal effects become highly relevant and the ablation depth per cycle is increased (Fig. 1.20). This primary pos-

24 | 1 Laser material machining of CFRP – damage-free? 5 Hatch cycles 15 Beam deflection 400mm/s Spot diameter 25μm

Ablation depth [mm]

4

3

2

1

0 0

20

40

60

80

100

Hatch distance [μm] Fig. 1.20: Typical correlation of ablation depth and hatch distance [52].

itive effect has the disadvantage that the ablation process becomes more and more unstable with the increasing influence of thermal effects. Considering this correlation, it is possible to use lasers as tools for precise layerby-layer removal. With this laser-based ablation process, it is possible to expose single fiber layers (Fig. 1.21) without damaging the fibers, as it is done by conventional approaches for composite repair.

100 μm

Fig. 1.21: Laser-based excavated fibers [52].

This effect includes some advantages for the bonded repair of CFRP structures. With conventional methods, there is still an amount of matrix resin between exposed fibers (Fig. 1.21). Focusing the adhesive bonding during the repair process is favorable. First, it is positive because the direct adhesion between adhesive and fibers provides a better load transfer through the repair patch compared to matrix resin residues that may have a weakened adhesion to the fibers [63]. Second, consistent fiber exposure is positive because the adhesion for every single ply-to-ply joint is predictable, while it is hard to predict if there are some areas with a high amount of matrix resin residues and some without any residues. Based on the application of CFRP ablation in the bonded composite repair process, there are typical indicators to judge the quality and efficiency of an ablation

1.5 Conclusion

|

25

technique. Two typical parameters are the production time and strength of a repaired specimen [64]. Here it can be mentioned that the laser-based approach can reduce production time. Voelkermeyer et al. were able to reduce the time for a complex repair process from 40 h (conventional mechanical approach) to 12 h by applying a laser-based approach. It can be justified by the automated process and also by the effect that the laser-based approach is able to fabricate smaller scarf ratios. They were able to achieve a scarf ratio of 1 : 10 with the laser, while due to the machining tools only a ratio of 1 : 20 was reachable with conventional mechanical tools [65]. This leads to a much smaller area for the ablation process and therefore to a reduction of processing time. As the second indicator, the ablated and repaired specimens should be mechanically tested and compared to undamaged specimens. In the publications cited, the laser ablated specimens were able to reach the strengths of the undamaged samples. Fischer et al. proved the same shear strength (according to DIN 65148) for the laser-ablated specimens as the reference specimens within the margin [65]. Even the samples that were repaired based on the smaller scarf ratio performed on the same level. Therefore the laser-based CFRP ablation could be characterized as an efficient tool to solve some of the main challenges for structural composite repair: 1. The application of UV-lasers offers the possibility to expose and ablate single fiber layers without matrix resin residues; 2. The ablation process can be automatically performed, which allows the reduction of process time; 3. Compared to mechanical methods, the laser ablation allows smaller scarf ratios and different scarf geometries; 4. Therefore a repair process for areas is possible where the structural repair with conventional methods is limited.

1.5 Conclusion Recent progresses in laser system technology and process development enable new technologies for machining CFRP. Adapted strategies, in particular multi-pass cutting strategies combined with highly brilliant laser sources or pulsed laser sources with high output powers have immense potential for future development. Apart from improvements in laser cutting techniques, innovative cooling strategies need to be developed for improved heat dissipation, resulting in smaller HAZs and better mechanical capabilities as well as improved process efficiency. Scanning systems used for remote cutting are limited by their working fields; thus a combination of scanning systems and robotic systems or linear stage systems is one of the next logical steps in development. Within this field, the simultaneous movement of scanner mirrors and the stage or robot needs to be developed. The layer-by-layer removal of damaged composite material provides a cavity for refilling with repair plies and allows for an automated fast repair concept for CFRP

26 | 1 Laser material machining of CFRP – damage-free? without thermal damages. In addition, the layer-by-layer removal offers a technique without following process steps like cleaning or activating. The selective removal of the matrix material under consideration of the absorption behavior with a laser offers a practical method to improve the surface pre-treatment of CFRP laminates for adhesive bonding, as one example. The variation in the ablation result such as a complete excavation of the fibers or just an ablation of a few microns of matrix material enables adapted bonding strategies and thus more efficient processes and also a bond strength as high as the reference. All techniques presented achieve a reliable and automatable handling. Moreover, the new processes avoid maintenance costs from tool wear, are fast enough and can be expected to have a successful economic implementation.

References [1]

[2] [3] [4] [5] [6] [7] [8]

[9] [10]

[11] [12] [13]

[14]

Hocheng H, ed. Machining technology for composite materials, principles and practice. Philadelphia, Pa.: Woodhead Pub., 2012. 1 online resource (xv, 472. (Woodhead Publishing in materials). ISBN: 978-0-85709-030-0. Hintze W, Hartmann D. Modeling of Delamination During Milling of Unidirectional CFRP. Procedia CIRP 2013;8:444–449. Hocheng H, Puw HY, Huang Y. Preliminary study on milling of unidirectional carbon fibre-reinforced plastics. Composites Manufacturing 1993;4:103–108. Masuda M, Kuroshima Y, Chujo Y. Failure of tungsten carbide-cobalt alloy tools in machining of carbon materials. Wear 1993;169:135–140. Ahmad J, ed. Machining of Polymer Composites. Boston, MA: Springer US;2009:165. Hashish M. Trimming of CFRP Aircraft Components. In: WJTA-IMCA, ed. American WJTA-IMCA Conference & Expo;2013. Hashish M. Abrasive-Waterjet Machining Of Composites. WaterJet Technology Association Jet News 2008;(December):2–18. Kalla DK, Dhanasekaran PS, Zhang B, Asmatulu R. Abrasive waterjet machining of fiber reinforced composites: A review. In: The 4th Manufacturing Engineering Society 2012: AIP;2012: 535–542. Snider D, Hashish M. AWJ trimming of composites and cutting of other materials using 6-axis robots. In: WJTA-IMCA, ed. American WJTA-IMCA Conference & Expo;2011. Shanmugam DK, Nguyen T, Wang J. A study of delamination on graphite/epoxy composites in abrasive waterjet machining. Composites Part A: Applied Science and Manufacturing 2008;39: 923–929. Wachinger, G et al. New Trends in CFRP Treatment and Surface Monitoring for Automated Structural Adhesive Bonding,ICCM 2009. Bénard Q, Fois M et al. Influence of the Polymer Surface Layer on the Adhesion of Polymer Matrix Composites. Journal of Thermoplastic Composite Materials 2009;22:51. Chin JW, Wightman JP. Surface characterization and adhesive bonding of toughened bismaleimide composites. Composites Part A: Applied Science and Manufacturing 1996;27(6): 419–428. Klapprott DK. Key factors of the peel ply surface preparation process. Henkel – Technical Data Sheet;2009.

References |

27

[15] Holtmannspötter J, v. Czarnecki J, Wetzel W, Dolderer D, Eisenschink C. The use of peel ply as a method to create reproducible, but contaminated surfaces for structural adhesive bonding of carbon fiber reinforced plastics. J. Adhesion 2013;89:96–110. [16] Hart-Smith LJ, Redmond G, Davis MJ. The curse of the nylon peel ply. In: Proceedings of the 41st SAMPE Meeting, Anaheim, California 1996:303–317. [17] Davies P, Courty C et al. Surface Treatment for Adhesive Bonding of Carbon Fibre-Poly(etherether ketone) Composites. J. Mat. Sci. Letters 1991;10:335–338. [18] Caprino G, Tagliaferri V. International Journal of Machine Tools and Manufacture. 1988;4: 389–398. [19] Sheikh-Ahmad JY. Machining of polymer composites. New York: Springer;2009. [20] Bazargan M. Airline operation and scheduling. Farnham: Ashgate Publishing Limited;2010. [21] Charalambides MN, Hardouin R, Kinloch A J, Matthews F L. Adhesively-bonded repairs to fibrecomposite materials I. Experimental. Composites Part A Applied Science and Manufacturing 1998;29(11):1371–1381. [22] Armstrong KB, Cole W, Bevan G. Care and repair of advanced composites. London: SAE International;2005. [23] Wang CH, Gunnion AJ. Optimum shapes of scarf repairs. Composites Part A Applied Science and Manufacturing 2009;40(9):1407–1418. [24] Gardiner G. Primary structural repair: the quest for quality. High Performance Composites 2011 November. [25] Shanmugam DK, Nguyen T, Wang J. A study of delamination on graphite/epoxy composites in abrasive waterjet machining. Composites Part A Applied Science and Manufacturing 2008;39:923–929. [26] Markatos DN, et al. The effects of manufacturing-induced and in-service related bonding quality reduction on the mode-I fracture toughness of composite bonded joints for aeronautical use. Composites: Part B 2012;45:556–564. [27] Ferreira JR, Coppini NL, Miranda GWA. Machining optimization in carbon fibre reinforced composite materials. Journal of Materials Processing Technology 1999;92:135–140. [28] Staehr R, Bluemel S, Hansen P, Jaeschke P, Suttmann O, Overmeyer L. The influence of moisture content on the heat affected zone and the resulting in-plane shear strength of laser cut thermoplastic cfrp. Plastics, Rubber and Composites 2014, DOI: http://dx.doi.org/10.1179/ 174329814Y.0000000114. [29] Bluemel S, Jaeschke P, Suttmann O, Overmeyer L. Comparative Study of Achievable Quality Cutting Carbon Fibre Reinforced Thermoplastics Using Continuous Wave and Pulsed Laser Sources. 8th International Conference on Laser Assisted Net Shape Engineering LANE 2014 2014;56:1143–52. (http://www.sciencedirect.com/science/article/pii/S1875389214001746). [30] Bluemel S, Jaeschke P, Wippo V, et al. Laser Machining of CFRP using a high power laser – Investigation on the heat affected zone. In: Proceedings of the 15th European Conference on Composite Materials;2012. [31] Goeke A. Laserstrahltrennen von Faserverbundkunststoffen. Dissertation. 1st ed. Göttingen, Niedersachsen: Cuvillier Verlag;2011. ISBN: 978-3-86955-618-5. [32] Jaeschke P, Kern M, Stute U, Kracht D, Haferkamp H. Laser processing of continuous carbon fibre reinforced polyphenylene sulphide organic sheets–Correlation of process parameters and reduction in static tensile strength properties. Journal of Thermoplastic Composite Materials 2014;27:324–337. [33] Klotzbach A, Hauser M, Beyer E. Laser Cutting of Carbon Fiber Reinforced Polymers using Highly Brilliant Laser Beam Sources. Physics Procedia 2011;12 Part A:572–577. [34] Leone C, Genna S, Tagliaferri F, Tagliaferri V. Fibre laser cutting of CFRP thin sheets by multipasses scan technique. Optics and Lasers in Engineering 2014;53:43–50.

28 | 1 Laser material machining of CFRP – damage-free? [35] Negarestani R, Li L, Sezer HK, Whitehead D, Methven J. Nano-second pulsed DPSS Nd:YAG laser cutting of CFRP composites with mixed reactive and inert gases. International Journal of Advanced Manufacturing Technology 2010;49:553–566. [36] Bluemel S, Brede S, Jaeschke P, Suttmann O, Overmeyer L. Applying a DOE model for the determination of appropriate process windows for nanosecond laser processing of CFRP. In: Proceedings of ICALEO 2014;2014. [37] Pagano N, Leone C, Lopresto V, De Iorio I. Laser cutting of CFRP sheet by pulsed Nd:YAG source: Influence of pulse energy, pulse duration and overlapping on kerf geometry and HAZ. In: 6th IPROMS virtual conference;2010. [38] Freitag C, Hafner M, Onuseit V, Michalowski A, Weber R, Graf T. Diagnostics of Basic Effects in Laser Processing of CFRP. In: Proceeding of International Symposium on Laser Processing of CFRP and Composite Materials;2012. [39] Jaeschke P, Fischer F, Kern M, Stute U, Kracht D. Laser cutting of CETEX thermoplastic composites using high-power multimode fibre laser. In: Proceedings of Sampe 2011;2011. [40] Abedin F. Review on Heat Affected Zone (HAZ) in Laser Machining. In: Proceedings of the 6th Annual GRASP Symposium;2010:63–64. [41] Sato Y, Tsukamoto M, Matsuoka F, Takahashi K, Masuno S. NANOSECOND LASER PROCESSING OF CFRP IN AR GAS AMBIENCE FOR HAZ REDUCTION. In: Proceedings of ICALEO 2014;2014. [42] Bluemel S, Staehr R, Jaeschke P, Stute U. Determination of Corresponding Temperature Distribution within CFRP during Laser Cutting. Physics Procedia 2013;41:408–414. [43] Fürst A, Rose M, Klotzbach A, Hauptmann J, Heber T, Beyer E. Influence of Laser Irradiation on the Failure Behaviour of Unidirectional Fibre Reinforced Polymers. In: ECCM, ed. Proceedings of the 16th European Conference on Composite Materials;2014. [44] v. Bestenbostel W, Kolb M, Fürst A, Wetzig A. Damage-Property Correlation After Laser Cutting of CFRP. In: ECCM, ed. Proceedings of the 16th European Conference on Composite Materials; 2014. [45] Harada Y, Suzuki T, Nishino M, Niino H. INVESTIGATION ON THE TENSILE STRENGTH OF CFRP/CFRTP MANUFACTURING USING HIGH-POWER LASERS. In: Proceeding of International Symposium on Laser Processing of CFRP and Composite Materials;2012. [46] Walter J, Hustedt M, Staehr R, et al. Laser Cutting of Carbon Fiber Reinforced Plastics – Investigation of Hazardous Process Emissions. Physics Procedia 2014;56:1153–1164. [47] Caprino G, Tagliaferri V, International Journal of Machine Tools and Manufacture 1988:28(4): 389–398. [48] Cenna AA, Mathew P, International Journal of Machine Tools & Manufacture 1997:37(6); 723–735. [49] Tönshoff HK, Hesse D, Mommsen J, CIRP Annals – Manufacturing Technology 1993;42(1): 247–251. [50] Bénard Q, Fois M, Grisel M, Laurens P. Laser surface treatment of composite materials to enhance adhesion properties. In: Adhesion – Current Research and Application, W. Possart (ed.), pp. 305–318, WILEY-VCH, Weinheim, Germany;2005. [51] Buchman A, Dodiuk H, Rotel M, Zahavi J. Durability of laser treated reinforced PEEK/epoxy bonded joints. In: Polymer Surfaces and Interfaces: Characterisation, Modification and Application. K. L. Mittal and K. W. Lee (eds.), pp. 37–69, CRC Press, Boca Raton, FL;1997. [52] Fischer F, Romoli L, Kling R. Laser-based repair of carbon fiber reinforced plastics. CIRP Annals – Manufacturing Technol. 2010;59:203–206. [53] Fischer F, Kreling S, Dilger K. Surface structuring of CFRP by using modern excimer laser sources. Physics Procedia 2012;39:154–160.

References |

29

[54] Silvain JF, Hiroyuki N, Shigeyuki O, Shinya N, Akira Y. Surface modification of elastomer/carbon composite by Nd+:YAG laser and KrF excimer laser ablation. Appl. Surface Sci. 1999;141: 25–34. [55] Büchter E. Weniger ist mehr! – Moderne Reinigungsverfahren für den Leichtbau. Reinigen und Vorbehandeln 2008;62:14–15. [56] Fischer F, Kreling S, Frauenhofer M, Dilger K. Laser surface pre-treatment of CFRP for adhesive bonding in consideration of the absorption behavior. J. Adhesion 2012;88:350–363. [57] Hartwig A, Hunnekuhl J, Vitr G, Dieckhoff S, Vohwinkel F, Hennemann OD. Influence of CO2 laser radiation on the surface properties of poly(etheretherketone). J. Appl. Polym. Sci. 1997; 64:1091–1096. [58] Kreling S. UV- bis M-IR-Laserstrahlung zur Klebvorbehandlung von CFK, Dissertation, TU Braunschweig;2015. [59] Toenshoff HK, Hesse D, Mommsen J. Micromachining Using Excimer Lasers. CIRP Annals – Manufacturing Technology 1993;42(1):247–251. [60] Voelkermeyer F, Kling R, Denkena B. Flexible and Damage Minimized Machining of Fiber Plies and CFRP by UV-laser Radiation. Proceedings of CFK-Valley Stade Convention, Germany 2007: 135–137. [61] Herzog D, Jaeschke P, Meier O, Haferkamp H. Investigations on the Thermal Effect Caused by Laser Cutting with Respect to Static Strength of CFRP. International Journal of Machine Tools & Manufacture 2008;48:1464–1473. [62] Wolynski A, Herrmann T, Mucha P, Haloui H, L’huillier J. Laser ablation of CFRP using picosecond laser pulses at different wavelengths from UV to IR. Physics Procedia 2011;12: 292–301. [63] Fischer F, Kreling S, Jaeschke P, Frauenhofer M, Kracht D, Dilger, K. Laser Surface Pre-Treatment of CFRP for Adhesive Bonding in Consideration of the Absorption Behaviour. Journal of Adhesion 2012;88;350–363. [64] Voelkermeyer F, Fischer F, Stute U, Kracht D. Laser-based approach for bonded repair of carbon fiber reinforced plastics. Physics Procedia 2011;12;537–542. [65] Fischer F, Keck R, Kaden M, Romoli L, Kling R. Laser as an innovative tool for laminates repair. Paris: Sampe Europe;2010. [66] Hu NS, Zhang LC. Some observations in grinding unidirectional carbon fibre-reinforced plastics. Journal of Materials Processing Technology 2004;152(3):333–338. [67] Chen S, Kancharla Y. Lu: Laser-based microscale patterning of biodegradable polymers for biomedical applications. Int. J. of Mat. & Prod. Tech. 2003;18:457–468. – ISSN 0268-1900

W. Cong, F. Ning

2 Rotary ultrasonic machining of CFRP composites Abstract: Drilling is a major machining process performed for assembly purposes in CFRP composite applications. Traditional methods of drilling CFRP composites (including twist drilling and its relative methods) have many machining problems such as short tool life, poor hole quality and low machining efficiency. There are critical needs to develop time-efficient and cost-effective processes for drilling CFRP composites. Rotary ultrasonic machining (RUM) can be one of them. This chapter presents RUM of CFRP composites, including definitions, features, machine elements (system set-up) as well as experimental and theoretical investigations. These investigations include effects of input variables (tool rotation speed, feed rate and ultrasonic power) on cutting force, torque, cutting temperature, edge quality, surface roughness, burning of machined surface, tool wear, material removal rate (MRR), power consumption and feasible regions during RUM of CFRP composites.

2.1 Introduction 2.1.1 CFRP composites 2.1.1.1 Properties and applications of CFRP composites Fiber reinforced plastic composites can be classified into different categories based on the type of fibers. The fibers include glass fiber, carbon fiber (graphite fiber), polyethylene fiber, boron fiber, ceramic fiber and kevlar fiber [36]. Carbon fiber reinforced plastic (CFRP) composites consist of two materials: carbon fibers and polymer. Within CFRP composites, the carbon fibers are surrounded in one or more orientations in a polymer matrix and can be in the form of small particles, whiskers or continuous filaments. The carbon fibers are used to support the load, while the polymer matrix is used to bind and protect the fibers and transfer the load to the reinforcing fibers [10, 23, 28]. The polymers used as the matrix of CFRP composites are generally classified as thermoset (epoxy, phenolic, polyester, etc.) or thermoplastic (polyether-ether-ketone, polyimide, etc.) resins [53]. CFRP composites have a variety of attractive properties, including low density (lower than aluminum, providing light-weight engineering solutions); high stiffnessto-weight ratio (stiffer than titanium); excellent fatigue, corrosion and wear resistance; outstanding toughness and damage tolerance (enabled using proper fiber orientations); high dimensional stability; excellent chemical resistance (controlled by the polymer matrix); and low friction coefficient, thermal expansion, electrical conductivity [10, 21, 37, 41, 51, 53].

32 | 2 Rotary ultrasonic machining of CFRP composites Due to the combination of these superior physical, thermal and mechanical properties, CFRP composites serve as the most commercial engineering materials and more and more of them have been widely used in place of metals in many different applications, such as aerospace and commercial aircraft, automobile, marine, medical prosthesis, sports goods, laptop computers, robot arms, maglev train guide-ways, bridges, chemical containers and fishing rods [25, 39, 43, 47, 53]. Especially in aerospace and the commercial aircraft industry, the usage of CFRP composites grew remarkably. For instance, in Boeing 787, the Boeing Company’s newest generation of commercial aircraft, nearly half of the materials in the frame are carbon fiber reinforced plastic and other composites, leading to weight savings of 20 % [5, 45]. Besides, as aircraft is engaged in corrosive environments, composites would not be easily subject to fatigue and corrosion damage due to their good fatigue and corrosion resistance, resulting in saving on maintenance costs [26].

2.1.1.2 Machining of CFRP composites Although CFRP composites can be fabricated to near-net shape, machining is needed as a post-processing operation to create some features, including turning [46], milling [7], drilling [2, 8] and surface grinding [48, 50]. Besides, some unconventional machining processes have also been adopted for machining CFRP composites like laser machining [31], electrical discharge machining [30] and abrasive waterjet cutting [34]. Among all the machining processes, drilling is a major machining process performed for assembly purposes in CFRP composite applications. For example, a large number of holes need to be drilled on CFRP laminates for assembly of Boeing 787 aircraft [6, 29]. Twist drilling and its derivative methods and core grinding are often used in CFRP composites’ hole making. However, they have many drawbacks when drilling CFRP composites, such as short tool life, poor hole quality and low machining efficiency [16, 55]. To improve drilling performance, Park et al. [41] presented grinding for composites’ hole making using a core drill bit with metal-bond PCD particles. As a result, the thrust force decreased compared to twist drilling, leading to less delamination. However, some drawbacks of grinding still exist including tool degrading and bad surface finish [56]. Moreover, high-speed drilling is a promising drilling operation due to its higher productivity and delamination reduction, but such a drilling operation is very expensive as it has to be performed in a high-speed drilling machine system that requires a large amount of energy. Therefore, developing time-efficient and cost-effective drilling processes becomes very important in the assembly of CFRP composites parts.

2.1 Introduction |

33

2.1.2 Rotary ultrasonic machining 2.1.2.1 Principle of RUM RUM is a hybrid machining process that combines material removal mechanisms of diamond grinding and ultrasonic machining, as shown in Fig. 2.1. The cutting tool is a core drill with metal-bonded diamond abrasives. During RUM, the rotating tool vibrates axially at an ultrasonic frequency (typically 20 kHz) and moves along its axial direction towards the workpiece. Coolant pumped through the core of the tool washes away the swarf and prevents the tool from jamming and overheating.

Rotation Coolant flow in

Ultrasonic vibration

Coolant flow out

Workpiece

Feeding

Coolant flow out

Abrasive portion

Fig. 2.1: Illustration of rotary ultrasonic machining.

2.1.2.2 Advantages of RUM RUM is a relatively low-cost, environment-benign process that easily fits in the infrastructure of the traditional machining environment. Other advantages of this process include high hole accuracy, superior surface finish, high material removal rate, low tool pressure and low tool wear rate in machining of other materials [9, 24, 44].

2.1.3 Purpose of this chapter The purpose of this chapter is to present RUM of CFRP composites, including definitions, features, machine elements (system set-up), experimental and theoretical investigations. These investigations include effects of input variables (tool rotation speed,

34 | 2 Rotary ultrasonic machining of CFRP composites feed rate and ultrasonic power) on cutting force, torque, cutting temperature, edge quality, surface roughness, burning of machined surface, tool wear, material removal rate (MRR), power consumption and feasible regions during RUM of CFRP composites.

2.2 Rotary ultrasonic machining system set-up The rotary ultrasonic machine mainly consists of two major systems: an ultrasonic spindle system and a coolant system, as illustrated in Fig. 2.2. The major components in an ultrasonic spindle system include an ultrasonic power supply, an ultrasonic spindle, an electric motor, a feeding device and a control panel. In the ultrasonic spindle, as shown in Fig. 2.3, there are an ultrasonic transducer, an ultrasonic amplitude transformer (horn) and tool holder, and cutting tool. The motor atop the ultrasonic spindle supplies the rotational motion of the tool and different speeds can be controlled by adjusting the motion speed controller on the control panel. The feed rate can be controlled by the compressed air-motivated hydronic system. Compressed air is provided by the air compressor. The feed rate controller on the control panel was used to adjust feed rate. There are two feeding motions: constant feed rate (with different cutting force) and constant pressure (with a different feed rate). The RUM system was originally designed with constant pressure that is similar to the ultrasonic machining (USM) system. In the last two decades, the constant feeding system was developed. Since MRR directly affects the machining performance and can be controlled with a constant feeding motion, such set-ups have been widely used. The coolant system is comprised of a pump, coolant tank, pressure regulator, flow rate and pressure gauges, and valves. Such a system could provide coolant to the spindle and the interface of machining.

2.2.1 Ultrasonic power supply Ultrasonic power supply converts conventional electrical supply (typically 50 Hz or 60 Hz) to high frequency electrical energy. Frequencies of 20–40 kHz are the most commonly used [3, 49]. The high frequency electrical energy is supplied to the ultrasonic transducer. Changing the output level of ultrasonic power can adjust the ultrasonic vibration amplitude if other parameters are kept the same.

2.2.2 Ultrasonic transducer The ultrasonic transducer converts high-frequency electrical energy into high-frequency mechanical motion. The length and strength of the transducer material can also control the vibration amplitude [3, 49]. There are two major types of transducers –

2.2 Rotary ultrasonic machining system set-up |

Ultrasonic spindle system

Feeding device

Spindle motor

3000

Ultrasonic spindle Pressure gauge

Control panel

Power supply

Valve Tool

Air compressor

Coolant system

Workpiece Coolant tank Fixture

Pump

Machine table Pressure regulator Flow rate gauge Fig. 2.2: Rotary ultrasonic machining (RUM) system set-up.

Ultrasonic transducer Ultrasonic transformer

Connector to the motor Fan

Bearings

Tool holder Fig. 2.3: The structure of an ultrasonic spindle.

Stators

Valve

35

36 | 2 Rotary ultrasonic machining of CFRP composites piezoelectric transducers and magnetostrictive transducers. Piezoelectric transducers, mainly used in RUM processes, are made of quartz crystals or polycrystalline ceramics, as illustrated in Fig. 2.4. Quartz crystals have volumetric shrinkage and swelling if they are subjected to high-frequency electrical energy. Volume change due to ultrasonic electrical energy leads to up and down ultrasonic mechanical vibration in an axial direction. The maximum amplitude of vibration can be found at the resonant frequency of the crystal. In this case, the length of the crystal must equal half of the wavelength of the ultrasound wave in the crystal [3, 4, 49]. The transducer made of polycrystalline ceramics has a sandwich structure that is stacked up ceramic disks, with a high density base and a low density block [3]. Such transducers can convert 96 % of electrical energy to mechanical motion without cooling [3, 4, 49]. Radiating face Low density block Disks of quartz crystals or polycrystalline ceramics High density base

Bolt

Fig. 2.4: Piezoelectric ultrasonic transducer (after [4]).

2.2.3 Ultrasonic amplitude transformer (horn) and tool holder The vibration amplitude generated from a transducer is not adequate for machining. An ultrasonic amplitude transformer is used to increase the vibration amplitude. The length of the transformer is integral times of half the ultrasound wavelength in transformer material. The reduction of cross-sectional area results in the increase of vibration amplitude [3, 49]. The increase in vibration amplitude can reach 600 %. The tool holder not only mounts the tool, but also increases the amplitude with a similar function as an ultrasonic amplitude transformer. In other words, the tool holder partially acts as an acoustic resonance transformer. The ultrasonic amplitude transformer can be in different shapes, including amplifying transformer (stepped, exponential and tapper tool holder) and non-amplifying transformer (hyperbolic tool holder). The tool holder was usually integrated with the ultrasonic transformer, as illustrated in Fig. 2.5 [52]. The transformer is used to increase the intensity of transducer vibration so that the tool can be driven to perform the cutting operation. It consists of a hard and non-

2.2 Rotary ultrasonic machining system set-up |

Stepped

Exponential

Tapper

Hyperbolic

37

Fig. 2.5: Different shapes of cylindrical tool holder (after [52]).

magnetic steel that could be easily machined. The excellent fatigue strength is similar with that of K-Monel, metal bronze and mild steel. The length of horn (L) is equal to one-half of the wave length of sound in the steel. Take the exponential horn with circular cross section for example. The transformation ration K could be obtained by K=

S Dl =√ 0, Ds S1

(2.1)

where: Dl is the larger diameter; Ds is the smaller diameter; So is the area of the larger section; and S1 is the area of the smaller section. Taper index for an exponential horn could be expressed as β =

ln(K) . L

(2.2)

The value of K varies under different machining conditions. It is often set from 3 to 4 for fine machining and from 4 to 5 for rough machining. The length of the horn L is usually selected as either half wave (n = 1) or full wave (n = 2) length in horn material. Length of the exponential horn is given by 2

L=

ln(K) nC √ } . 1+{ f 2π n

(2.3)

And the variation of D is given by (also called Law of change of shape) D = Dl exp−β X .

(2.4)

Operating frequency must exceed critical frequency fc fc =

βC . 2π

(2.5)

The commonly used materials for ultrasonic amplitude transformers and tool holders include monel, titanium alloy (Ti-6Al-4V), stainless steel (AISI304), aluminum and aluminum bronze [52].

38 | 2 Rotary ultrasonic machining of CFRP composites 2.2.4 Cutting tool The cutting tool used in RUM is a super-abrasive core drill with a cup-shaped grinding wheel, as illustrated in Fig. 2.6. The detailed information of cutting tool variables will be shown in Section 2.3.2.

Do Di

Abrasive portion

Tu ni

ng

le ng

th

Tool connection portion

Fig. 2.6: Illustration of grinding tool used in rotary ultrasonic machining (RUM).

2.3 Input variables and output variables in RUM Input variables in RUM processes could be classified under three aspects. They are machining variables, cutting tool variables and cooling variables, respectively. These input variables were determined according to the experience from preliminary experiments and limitations of the experimental set-up.

2.3.1 Machining variables Machining variables of RUM are related to variables of both abrasive grinding and ultrasonic vibration, as RUM is the hybrid of these two processes. Thus, main machining variables are tool rotation speed, feed rate, ultrasonic power (ultrasonic vibration amplitude) and ultrasonic vibration frequency. The machining variables in the chapter are listed in Tab. 2.1.

2.3.2 Cutting tool variables and cooling variables Cutting tool variables and cooling variables are listed in Tab. 2.2.

2.3 Input variables and output variables in RUM

|

39

Tab. 2.1: Machining variables. Variables

Definition

Tool rotation speed Feed rate Ultrasonic power Ultrasonic vibration frequency

Rotational speed of tool Feed rate of tool Percentage of power from ultrasonic power supply Being fixed at 20 kHz on the machine

Tab. 2.2: Cutting tool variables and cooling variables. Categories

Variables

Definition

Cutting tool

Abrasive type Abrasive size Abrasive concentration

Diamond, CBN Mesh size or diameter of the abrasive The number of abrasives in a unit volume

Cooling

Coolant type Coolant flow rate Coolant pressure

Cold air, cutting fluid Flow rate of cutting fluid Pressure of cold air or cutting fluid

2.3.3 Workpiece properties Based on carbon fiber structures, CFRP composites can be classified into four types: wide yarn woven, thin yarn woven, flake and uni-directional continuous [19]. The fiber structures are illustrated in Fig. 2.7. Specifications of these structures are shown in Tab. 2.3. The type of CFRP workpiece that is most used in the investigations of this chapter, as illustrated in Fig. 2.8, consisted of wide yarn woven structured carbon fibers and epoxy resin matrix. The carbon fiber yarn in the woven structure had an orientation of 0/90 degrees, a thickness of 0.2 mm and a width of 2.5 mm. The workpiece contained 42 layers of carbon fibers and was sized 200 mm × 150 mm × 16 mm. Some specific properties of CFRP composites are shown in Tab. 2.4. Tab. 2.3: Specifications of carbon fiber structures of different CFRP types. CFRP

Fiber structure

Orientation

Fiber size

Number of Layers

Thickness

#1

Wide yarn woven

0/90°

42

16 mm

#2

Thin yarn woven

0/90°

20

7 mm

#3

Flake

N/A

N/A

12 mm

#4

Uni-directional continuous

45°

2.5 mm by 0.2 mm fiber yarn 0.35 mm by 0.1 mm fiber yarn 0.1 mm thickness fiber flake 0.2 mm fiber layer

24

18 mm

40 | 2 Rotary ultrasonic machining of CFRP composites

Wide yarn woven (#1)

Thin yarn woven (#2)

Flake (#3)

Unidirectional continuous (#4)

Fig. 2.7: Fiber structures of CFRP composites.

Carbon fiber Epoxy resin

Fig. 2.8: Illustration of fiber structures in CFRP composites.

2.3.4 Output variables The major output variables in RUM in this chapter are listed in Tab. 2.5.

2.4 Effects of input variables on output variables The RUM experiments have been conducted under a wide range of input variables to assess the effects of input variables on output variables.

2.4 Effects of input variables on output variables

| 41

Tab. 2.4: Properties of CFRP composites. Property

Unit

Value 3

Density of CFRP Hardness (Rockwell)

kg/m HRB

1550 70–75

Density of epoxy matrix Poisson’s ratio of epoxy matrix Tensile strength of epoxy matrix Elastic modulus of epoxy matrix Fracture toughness of epoxy matrix (Energy/Gc )

kg/m3 — GPa GPa J/m2

1200 0.3 0.13 4.5 500

Density of carbon fiber Poisson’s ratio of carbon fiber Tensile strength of carbon fiber Elastic modulus of carbon fiber Fracture toughness of carbon (Energy/Gc )

kg/m3 — GPa GPa J/m2

1800 0.3 5 230 2

Tab. 2.5: Output variables. Output variables

Unit

Cutting force Torque Cutting temperature Edge quality Surface roughness Burning of machined surface Tool wear Material removal rate (MRR) Power consumption Feasible regions

N N⋅m °C μm μm % mg mm3 /min W ⋅ h/mm

2.4.1 Effects on cutting force The data acquisition system, as shown in Fig. 2.9, was employed to measure cutting force in the axial direction and torque. The dynamometer (Model 9272, Kistler Inc., Switzerland) was used to measure cutting force in the axial direction, and the electrical signals from it were amplified by the charge amplifier (Model 5070A, Kistler Inc., Switzerland). Then the A/D converter transformed the electrical signals into digital signals, which were collected by a data acquisition card (PC-CARD-DAS16/16, Measurement Computing Corporation, Norton, MA, USA) with the help of a DynoWare software package (Type 2815A, Kistler Inc., Switzerland) on a computer. After setting the sensitivity of the dynamometer into the DynoWare, the cutting force can be obtained. The sampling rate was 20 Hz during all experiments.

42 | 2 Rotary ultrasonic machining of CFRP composites Workpiece Fixture Dynamometer

Machine table

Amplifier

A/D converter

Computer

Fig. 2.9: Data acquisition system of cutting force during RUM.

The cutting force measured fluctuated with time within a certain range. A typical curve of the cutting force signal in the time domain during RUM of CFRP can be observed in Fig. 2.10. Cutting force for drilling each hole was represented by the maximum cutting force value (Fz), and average cutting force was used in mechanistic predictive cutting force model development in Section 2.4.1.4. 160 Maximum cutting force

Cutting force (N)

120

80

40

0 0

10

20 Time (s)

30

40

50

Fig. 2.10: Typical relationship between cutting force and time (in RUM).

2.4.1.1 Effect of tool rotation speed A comparison of cutting force among RUM with cutting fluid coolant, RUM with cold air coolant, twist drilling and abrasive grinding [16, 40] when tool rotation speed changed is shown in Fig. 2.11. In this section together with Section 2.4.1.2, the plot of twist drilling (solid line) will follow the range of right vertical axis and the other three plots (dash lines) will follow the left one in order to compare all the items in the figures. It can be observed that for all the methods, cutting force decreased with the increase of tool rotation speed. Cutting force in RUM with each type of coolant was sig-

2.4 Effects of input variables on output variables | 43

Cutting force (N) – Dash lines

RUM with cutting fluid RUM with cold air Grinding Twist drill

350

850

300

700

250

550

200

400

150

250

100 1000

Cutting force (N) – Solid line

1000

400

100 5000

2000 3000 4000 Tool rotation speed (rpm)

Fig. 2.11: Cutting force comparison among the four methods with different levels of tool rotation speed.

nificantly smaller than that in twist drilling. Cutting force in RUM with cutting fluid coolant was also smaller than in grinding at almost all levels of tool rotation speed. In RUM, using cutting fluid could lead to smaller cutting force.

2.4.1.2 Effect of feed rate A comparison of cutting force among RUM with cutting fluid coolant, RUM with cold air coolant, twist drilling and abrasive grinding when feed rate changed is shown in Fig. 2.12. 750 RUM with cutting fluid RUM with cold air Grinding Twist drill

210

600

170

450

130

300

90

150

50 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Cutting force (N) – Solid line

Cutting force (N) – Dash lines

250

0 0.8

Feed rate (mm/s) Fig. 2.12: Cutting force comparison among the four methods with different levels of feed rate.

44 | 2 Rotary ultrasonic machining of CFRP composites For RUM with cold air coolant, twist drilling and abrasive grinding, cutting force increased with the increase of feed rate. In contrast, during RUM with cutting fluid coolant, cutting force decreased first when feed rate increased from 0.1 to 0.2 mm/s. Beyond this point, a further increase in feed rate increased cutting force. Cutting force in RUM with each type of coolant was significantly smaller than in twist drilling, and cutting force in RUM with cutting fluid coolant was also smaller than in grinding at almost all levels of feed rate. Interestingly, at some feed rate settings (such as 0.1, 0.7 and 0.8 mm/s), cutting force was about the same for both types of coolant using RUM, while for the other feed rates, using cutting fluid could lead to smaller cutting force.

2.4.1.3 Effect of ultrasonic power A comparison of cutting force between cold air coolant and cutting fluid coolant using RUM when ultrasonic power changed is shown in Fig. 2.13. It can be seen that, for both types of coolant, cutting force decreased with the increase of ultrasonic power. At all settings of ultrasonic power, cold air did not have the lubricating effect that cutting fluid had, resulting in larger cutting force. 200

Cutting force (N)

160 120 80 40 RUM with cutting fluid RUM with cold air

0 0

20

40

Ultrasonic power (%)

60

80

Fig. 2.13: Comparison of cutting force using different types of coolant.

2.4.1.4 Mechanistic predictive cutting force model A mechanistic predictive model considering ultrasonic machining as the predominant process was developed by Cong et al. (2014). It was established by analyzing one particle first and then integrating all active abrasive partials that were engaged in machining to find out the relationship between input variables and cutting force. Machining of CFRP composites was based on brittle fracture as chip separation occurred by brittle fracture [32]. The main model developing steps are shown in Fig. 2.14. All the input variables in this model are listed in Tab. 2.6.

2.4 Effects of input variables on output variables

| 45

Cutting force (F)

Input variables Workpiece properties

Machining and tool variables

Max impact force, (Fi)

Material removal rate (MRR) CFRP micromechanics process

Number of active abrasives (n)

Material removal analysis

E, v, Kc Removed volume for one abrasive (V1)

Indentation depth, (δ)

Fracture volume factor (Kv)

Fig. 2.14: Model developing processes.

Tab. 2.6: Input variables in the cutting force model for CFRP composites. Categories

Input variables

Unit

Elastic modulus, E Poisson’s ratio, 𝜈 Fracture toughness, Kc

MPa

Workpiece Properties

Tool variables

Machining variables

MPa ⋅√mm

Outer diameter, Do Inner diameter, Di Abrasive concentration, C Abrasive size, d

mm mm

Amplitude, A Frequency, f Feed rate, Fr Tool rotation speed, S

mm Hz mm/s rpm

mm

In this model, the maximum impact force for one abrasive grain taking part in cutting was expressed as: F 81/2 Ed1/2 δ 3/2 F1 = i = , (2.6) n 3(1 − 𝜈2 ) where Fi is the maximum impact force between tool and workpiece, δ is the indentation depth. It could be expressed as:

46 | 2 Rotary ultrasonic machining of CFRP composites 2 1/3

δ =[

9 (Fi /n)2 1 − 𝜈2 ( ) ] 16 d/2 E

.

(2.7)

Elastic modulus E and Poisson’s ratio vare different in longitudinal and transverse directions of the fiber. In RUM of CFRP, the machining load is applied on the transverse direction of the fiber. There is an assumption that a cylindrical fiber is treated as a rectangular one, as illustrated in Fig. 2.15.

(2)

(1)

Fig. 2.15: Cylindrical fiber simplification method.

Rule of mixtures (ROM) formula [1, 27] is used to predict the elastic modulus of CFRP in the fiber longitudinal direction (E1 ): E1 = Ef Vf + Em Vm ,

(2.8)

where Ef is the elastic modulus of fiber material in CFRP; Em is the elastic modulus of matrix material in CFRP; Vf is the volume fraction of the fiber which can be calculated by of fiber Vf = volume ; total volume Vm is the volume fraction of the matrix which can be calculated by of matrix Vm = volume . total volume There is an assumption in this formulation which states that strains in the longitudinal direction of the fibers are the same as those in the matrix [1, 27]. The Poisson’s ratio of CFRP composite materials in the longitudinal direction is 𝜈12 = 𝜈f Vf + 𝜈m Vm ,

(2.9)

where 𝜈f is Poisson’s ratio of fiber material in CFRP and 𝜈m is Poisson’s ratio of matrix material in the CFRP.

2.4 Effects of input variables on output variables

| 47

The inverse rule of mixtures (IROM) formula [1, 27] is used to determine the elastic modulus of CFRP in the fiber transverse direction (fiber perpendicular to load) (E2 ): Vf V 1 = + m. E2 Ef Em

(2.10)

The main assumption of this determination is that the stress is the same in the fiber and the matrix. This assumption is needed to maintain equilibrium in the transverse direction and implies that the fiber-matrix bond is perfect [1, 27]. The elastic modulus used in this model is E = E2 =

Ef Em

(2.11)

Vf Em + Vm Ef

There is a relationship between elastic modulus and Poisson’s ratio in a different direction: 𝜈ij 𝜈ji = (with i ≠ j) . (2.12) Ei Ej In this case, the Poisson’s ratio in transverse direction can be calculated by 𝜈 = 𝜈21 =

𝜈12 E2 . E1

(2.13)

The number of active abrasive grains, n, on the end face of cutting tool can be obtained by: 2

n=[

2

2

0.88 × 10−3 3 Ca 3 0.88 × 10−3 Ca 3 ] A0 = [ ] ( 3 ) A0 3 (π /6) × 100 (π /6)d ρ 100 d ρ 2

= 6.561 × 10

−4

(2.14)

3 C ( 3a ) A0 , d ρ

where Ca is the abrasive concentration; ρ is the density of abrasive material, g/mm3 , ρ = 3.52 × 10−3 g/mm3 for diamond; A0 is the area of the cutting tool end face, mm2 , A0 = π (D2o − D2i )/4, (Do and Di are the outer and inner diameters of cutting tool, respectively, mm). Assuming the abrasive grains are rigid, the impulse in terms of maximum impact force (Fi ) during one cycle of ultrasonic vibration is Impulse = ∫ Fi dt =∼ Fi Δ t ,

(2.15)

cycle

where Δ t is the time during which an abrasive particle is penetrating into the workpiece (effective contact time), s. In RUM, the cutting tool oscillates up and down and rotates simultaneously. Therefore, an abrasive particle at the end face of the tool moves along a sine wave that can be observed from machined surface, as shown in Fig. 2.16. A is the amplitude of

48 | 2 Rotary ultrasonic machining of CFRP composites

Cutting tool

z=Asin(2πft)

Mean position A O

O δ Work-piece

z

t1

t2

t

A A–δ A z

δ Δt

Fig. 2.16: Calculation of effective cutting time Δ t (after [42]).

ultrasonic vibration. It takes an abrasive particle a time of t/2 to move from z = (A − δ ) to z = A. Δ t can be calculated using the following equation: Δ t = 2(t1 − t2 ) =

1 π δ [ − arcsin (1 − )] , πf 2 A

(2.16)

where f is the frequency of ultrasonic, Hz. The impulse in terms of the cutting force F during one cycle of ultrasonic vibration is F Impulse = , (2.17) f where F is the cutting force measured during the experiments in RUM of CFRP, N. Cutting force F could be expressed as 1 1 1 1 δ δ 81/2 nEd1/2 δ 3/2 F = n [ − arcsin (1 − )] F1 = [ − arcsin (1 − )] . (2.18) 2 π A 2 π A 3(1 − 𝜈2 ) Moreover, the material removal volume by one abrasive partial could be calculated by regarding material removal as the result of brittle fracture. At the interface of the abrasive grain and workpiece surface, the abrasive grain has a maximum indentation value δ in a specific period of time. This phenomenon results in the change of lateral crack length CL and lateral crack depth CH , as illustrated in Fig. 2.17. Lateral cracks will form and propagate during the indentation of the material. The material will be removed from the workpiece if there are two adjacent indentations generating the lateral cracks. The fracture zone for one grain can be illustrated in Fig. 2.18. The fracture zone can be simplified as the volume of a half ellipse with three half-axes length of CL , CH and 2L . The material removal volume V1 by one abrasive could be determined as follows by simplifying the fracture zone: V1 =

1/2 F 3/4 1 δ DS π [ − arcsin (1 − )]} , KV π ( 1 ) (dδ − δ 2 ) { 3 KC 60f 2 A

(2.19)

2.4 Effects of input variables on output variables

| 49

where KV is fracture volume factor that is a proportionality parameter. KC can be calculated by [38] Kc = (E2 GC )

1/2

≅ [2E2 (Gf vf + Gm vm )]1/2 .

(2.20)

F1 Abrasive grain CL δ

CH

Lateral crack Workpiece Deformation zone

Median crack

Fig. 2.17: Material removal mechanism by one abrasive partial.

Material removal rate (MRR) could be calculated either based on material removed volume and feed rate, or through summation of the MRR of all abrasive particles on the end face of the cutting tool. The equations could be expressed as follows, respectively: π (D2o − D2i ) Fr ; 4 1/2 F 3/4 nπ S (Do + Di ) ( 1 ) (dδ − δ 2 ) MRR = nf V1 = KV 180 KC δ π ⋅ [ − arcsin (1 − )] . 2 A

MRR = Fr A0 =

(2.21) (2.22)

Abrasive path Lateral crack 2CL

δ

L CH

Fig. 2.18: Theoretical volume of fracture zone.

50 | 2 Rotary ultrasonic machining of CFRP composites After equaling equations (4.16) and (4.17), the obtained equation can be associated with equation (4.18): π (D2o − D2i ) 1/2 n1/4 π 2 S F 3/4 { Fr = KV (Do + Di ) ( ) (dδ − δ 2 ) { { { 4 180 K { C { { { { { 1 1 δ 1/4 ⋅ [ − arcsin (1 − )] { 2 π A { { { { { { { { 81/2 nEd1/2 δ 3/2 1 1 δ { [ − arcsin (1 − )] . F= 2 { 2 π A 3(1 − 𝜈 )

(2.23)

In this simultaneous equation, the indentation depth δ and cutting force F are unknowns.

2.4.2 Effects on torque Similar to the measurement procedure for cutting force, torque for drilling each hole was represented by the maximum torque value (Mz) as well.

2.4.2.1 Effect of tool rotation speed A comparison of torque among RUM with cutting fluid coolant, RUM with cold air coolant, twist drilling and abrasive grinding when tool rotation speed changed is shown in Fig. 2.19. It can be seen that for all the methods, torque decreased with the increase of tool rotation speed. The RUM with each type of coolant had smaller torque than twist drilling when tool rotation speed was set at 2000, 3000 and 4000 rpm, which was similar to that of cutting force. Torque in the RUM with cutting fluid coolant was smaller than that in grinding at all levels of tool rotation speed and it also generated smaller torque than with cold air. 1.5

Torque (N·m)

1.2

RUM with cutting fluid RUM with cold air Grinding Twist drill

0.9 0.6 0.3 0 1000

2000 3000 4000 Tool rotation speed (rpm)

5000

Fig. 2.19: Torque comparison among the four methods with different levels of tool rotation speed.

2.4 Effects of input variables on output variables

| 51

2.4.2.2 Effect of feed rate A comparison of torque among RUM with cutting fluid coolant, RUM with cold air coolant, twist drilling and abrasive grinding when feed rate changed is shown in Fig. 2.20. With the increase of feed rate, torque increased in all methods. At all levels of feed rate, torque in RUM was significantly smaller than that in twist drilling. Torque in RUM with cutting fluid coolant was smaller than that in grinding at all levels of feed rate. Using cutting fluid coolant in RUM led to smaller torque compared to using cold air. Again, torque had a similar trend with cutting force when feed rate increased. 1.2 RUM with cutting fluid RUM with cold air Grinding Twist drill

Torque (N·m)

0.9 0.6 0.3 0 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

Feed rate (mm/s)

Fig. 2.20: Torque comparison among the four methods with different levels of feed rate.

2.4.2.3 Effect of ultrasonic power A comparison of torque between cold air coolant and cutting fluid coolant using RUM when ultrasonic power changed is shown in Fig. 2.21. Torque decreased with the increase of ultrasonic power for both types of coolant. The torque was about 0.5 N ⋅ m for both types of coolant when there was no ultrasonic vibration. When ultrasonic power increased from 20 % to 80 %, using cold air resulted in larger torque than when using cutting fluid. The trends of cutting force and torque as ultrasonic power increased were the same. 0.7 0.6

Torque (N·m)

0.5 0.4 0.3 0.2 0.1

RUM with cutting fluid RUM with cold air

0 0

20 40 Ultrasonic power (%)

60

80

Fig. 2.21: Comparison of torque using different type of coolant.

52 | 2 Rotary ultrasonic machining of CFRP composites 2.4.3 Effects on cutting temperature In this section, two cutting temperature measurement methods (thermocouple method and fiber optic sensor method) were compared to investigate cutting temperature in RUM of CFRP. A blind hole (with a diameter of 1.5 mm) perpendicular to the tool feeding direction and along the radius direction of the hole was drilled into the workpiece. A K-type thermocouple (Model SC-GG-K-30-36, Omega Engineering, Inc., Stamford, CT, USA) and a fiber optic sensor were positioned inside the blind hole and touched the end of the blind hole. The signals from the thermocouple were collected by a digital thermometer (HH147U, Omega Engineering Inc., Stamford, CT, USA). The signals from the fiber optic sensor were demodulated through an optical sensing analyzer (OSA Si720, Micron Optics, Atlanta, GA, USA). The digital data were recorded and displayed on a computer using software (MatLab 7.13, MathWork Crop., Natick, MA, USA). The sampling rates for both measurement methods were set at 1 Hz. The thermocouple method is based on the principle of an thermoelectric effect [54], and the fiber optic sensor method is based on the Fabry-Perot (FP) principle. The schematics of these two methods are shown in Fig. 2.22 and Fig. 2.23, respectively. The effects of three input variables (tool rotation speed, feed rate and ultrasonic power) on cutting temperature using both methods were analyzed. Tool Thermocouple

Workpiece

Fiber optic sensor

Fiber core

Single-mode fiber

Fig. 2.22: Illustration of temperature measurement.

Borosilicate glass Fabry-perot cavity

Multi-mode fiber

Fig. 2.23: The structure of the fiber optic sensor.

2.4 Effects of input variables on output variables | 53

2.4.3.1 Effect of tool rotation speed The cutting temperature versus machining time curves under different levels of tool rotation speed for both methods are shown in Fig. 2.24. The temperature-time curves obtained by the thermocouple method were smoother than those obtained by the fiber optic sensor method. The maximum cutting temperatures measured by the thermocouple method were located between 20 and 40 s. However, the maximum cutting temperatures measured by the fiber optic sensor method for different levels of tool rotation speed were located in different time.

80

4000 3000 2000 1000

3

60 40

1

20 0 0

(a)

120

2

20

40 60 Time (s)

80

2

4000 3000 2000 1000

100 Temperature (°C)

Temperature (°C)

100

100

3

80 60 40

1

20 0 0

20

40 60 Time (s)

80

100

(b)

Fig. 2.24: Cutting temperature versus machining time for different levels of tool rotation speed (feed rate = 0.3 mm/s; Ultrasonic power = 40 %). (a) Thermocouple, (b) Fiber optic sensor; 1: The cutting tool started cutting the workpiece; 2: The end face of the cutting tool passed through the measured position; 3: The cutting tool cut through the workpiece thickness.

Effects of tool rotation speed on maximum cutting temperature are shown Fig. 2.25. It can be seen that, when tool rotation speed increased from 1000 to 3000 rpm, maximum cutting temperature increased. However, a further increase of tool rotation speed from 3000 to 4000 rpm led to a decrease of maximum cutting temperature. Compared with what was measured by the thermocouple method, maximum cutting temperature measured by the fiber optic sensor method was higher by about 20 °C under each level of tool rotation speed. This trend is different from observations reported for RUM of titanium by Cong et al. [13]. They found that, in RUM of titanium, cutting temperature decreased sharply as tool rotation speed increased from 1500 to 3000 rpm, but gradually increased as tool rotation speed increased from 3000 to 6000 rpm. When feed rate was kept the same, an increase of tool rotation speed led to an increase of sliding distance between the diamond and workpiece within a fixed period of time [11, 35]. This might be the reason why cutting temperature increased as tool rotation speed increased from 1000 to 3000 rpm. Compared with 3000 rpm, a tool rotation speed of 4000 rpm caused a lower depth of cut for individual diamond grains on the tool end face, leading to lower cutting force [11], which might cause the temperature to decrease.

54 | 2 Rotary ultrasonic machining of CFRP composites 120

Maximum temperature (°C)

100 80 60 40 20

Fig. 2.25: Effects of tool rotation speed on maximum cutting temperature (feed rate = 0.3 mm/s; Ultrasonic power = 40 %).

Fiber optic sensor Thermocouple

0 1000

2000 3000 Tool rotation speed (rpm)

4000

2.4.3.2 Effect of feed rate The cutting temperature versus machining time curves under different levels of feed rate for both methods are shown in Fig. 2.26. The temperature-time curves obtained by the thermocouple method were smoother than those obtained by the fiber optic sensor method. For both methods, the maximum cutting temperatures for different levels of feed rate (0.1, 0.3 and 0.5 mm/s) were located around 75, 40 and 30 s on the temperature-time curves, respectively. Effects of feed rate on maximum cutting temperature are shown in Fig. 2.27. As feed rate increased, maximum cutting temperature decreased for both methods. Compared with that measured by the thermocouple method, cutting temperature measured by the fiber optic sensor method was higher. Also, as feed rate increased, the differences between the temperature measurements by the two methods decreased. This trend

Temperature (°C)

100

2

80

2 3

3 3

60 40 1 20 0 0

(a)

2

0.1 0.3 0.5

50

100 150 Time (s)

200

250

120

300

2

2

100 Temperature (°C)

120

23

3

0.1 0.3 0.5

3

80 60 40 1 20 0 0

50

100 150 Time (s)

200

250

300

(b)

Fig. 2.26: Cutting temperature versus machining time for different levels of feed rate. (Tool rotation speed = 3000 rpm; Ultrasonic power = 40 %). (a) Thermocouple, (b) Fiber optic sensor; 1: The cutting tool started cutting the workpiece; 2: The end face of the cutting tool passed through the measured position; 3: The cutting tool cut through the workpiece thickness.

2.4 Effects of input variables on output variables | 55

Maximum temperature (°C)

150 120 90 60 30 0 0.1

Fiber optic sensor Thermocouple 0.3 Feed rate (mm/s)

0.5

Fig. 2.27: Effects of feed rate on maximum cutting temperature. (Tool rotation speed = 3000 rpm; Ultrasonic power = 40 %)

is different from observations reported for RUM of titanium by Cong et al. [13]. They found that, in RUM of titanium, when feed rate changed from 0.02 mm/s to 0.04 mm/s, there was nearly no change in cutting temperature. However, when the feed rate increased from 0.04 mm/s to 0.06 mm/s, cutting temperature increased dramatically. When tool rotation speed was kept the same, a lower feed rate led to a longer grinding time which then generated more heat. This might be the reason for the higher temperature when the feed rate was lower.

2.4.3.3 Effect of ultrasonic power Cutting temperature versus machining time curves under different levels of ultrasonic power for both methods are shown in Fig. 2.28. Labels 1, 2 and 3 in Fig. 2.28 indicated when the cutting tool started cutting the workpiece, when the end face of the cutting tool passed through the measured position and when the cutting tool cut through the workpiece thickness. The temperature-time curves obtained by the thermocouple method were smoother than those obtained by the fiber optic sensor method. The fiber optic sensor method was more sensitive to temperature than the thermocouple method. The maximum cutting temperatures measured by the thermocouple method were located between 20 s and 40 s on the temperature-time curve. However, the maximum cutting temperatures measured by the fiber optic sensor method were located between 30 s and 50 s on the temperature-time curve. Effects of ultrasonic power on maximum cutting temperature are illustrated in Fig. 2.29. In Fig. 2.29, the data points are the maximum values on the temperature-time curve. When ultrasonic power changed from 0 to 20 %, maximum cutting temperature did not change much for both methods. When ultrasonic power increased from 20 % to 60 %, maximum cutting temperature dramatically increased. This trend is different from observations reported for RUM of titanium by Cong et al. [13]. They found that, in RUM of titanium, cutting temperatures with ultrasonic vibration (ultrasonic

180

140 120 100 80 60

60% 40% 20% 0%

2 3

40 20 0

1

0

20

40 60 Time (s)

80

60% 40% 20% 0%

3

120 90 60 1

30 0 0

100

(a)

2

150 Temperature (°C)

Temperature (°C)

56 | 2 Rotary ultrasonic machining of CFRP composites

20

40 60 Time (s)

80

100

(b)

Fig. 2.28: Cutting temperature versus machining time for different levels of ultrasonic power (tool rotation speed = 3000 rpm; Feed rate = 0.3 mm/s). (a) Thermocouple, (b) Fiber optic sensor; 1: The cutting tool started cutting the workpiece; 2: The end face of the cutting tool passed through the measured position; 3: The cutting tool cut through the workpiece thickness.

power = 20 %, 40 % and 60 %) were lower than those without ultrasonic vibration (ultrasonic power = 0 %). When ultrasonic power increased from 20 % to 60 %, there was an obvious (but not dramatic) increase in cutting temperature. Compared with those measured by the thermocouple method, cutting temperatures measured by the fiber optic sensor method were higher, especially, when ultrasonic power was high. At this point in time, the authors did not fully understand why relatively big differences in measured temperature between the thermocouple and fiber optic sensor methods existed. However, they had several hypotheses in mind and will conduct further research to test them. The authors noticed larger fluctuations in measured temperature by the fiber optic sensor method. One possible reason is that the measuring point (micrometers in size) of the fiber optic sensor was much smaller than that (millimeters in size) of the thermocouple. A smaller measurement point was more sensitive to a change 180

Maximum temperature (°C)

150 120 90 60 30 0 0%

Fiber optic sensor Thermocouple 20% 40% Ultrasonic power

60%

Fig. 2.29: Effects of ultrasonic power on maximum cutting temperature (tool rotation speed = 3000 rpm; Feed rate = 0.3 mm/s).

2.4 Effects of input variables on output variables

| 57

of temperature. The authors plan to conduct further investigations to understand the large temperature fluctuations. Ultrasonic power determines vibration amplitude. As ultrasonic power increases, the vibration amplitude increases [12]. This will increase the penetration depth of diamond grains into the workpiece material, increasing the interaction force between the diamond grain and the workpiece material [35]. This could result in an increase of cutting temperature.

2.4.4 Effects on edge quality Machined holes under different machining parameters are shown in Fig. 2.30. Delamination is caused by splitting or separating a laminate into layers. It is one of the principle damages that affect edge quality. It can be seen that there is no delamination, fiber pull-out or chippings in the edge at hole entrance, and the edge conditions will remain regardless of machining conditions. But the edge quality at hole exit is different from that at the entrance. A few splits occur at the connection of two holes on the top laminate in a brittle way (areas 1 and 2 in Fig. 2.30). All the holes were machined without taper (the diameters at the entrance were the same as those at the exit). The variation coefficient (standard deviation divided by mean) of diameter between all the produced holes was less than 1 %.

1 2

Entrance

Exit

Fig. 2.30: Observation of the machined holes on the carbon fiber-reinforced epoxy workpiece [22].

The hole quality in RUM is mainly affected by edge chipping of the drilled hole at the hole exit, and it can be evaluated using two criteria including chipping size and chipping thickness on the machined rod, as illustrated in Fig. 2.31. The lower values of chipping size and chipping thickness will lead to improved hole quality. Figure 2.32 shows a machined rod and edge chipping observed under a digital microscope in the top to bottom direction (VHK-1000, KEYENCE, Japan). It can be seen that only edge chipping consisting of three laminated layers exited on the wall without any fiber-pullout or fracture of the material when the rod was trim and neat. The first layer is made up of fibers in clockwise direction (area 1 in Fig. 2.32). The second layer

58 | 2 Rotary ultrasonic machining of CFRP composites Chipping size Machined hole

Machined rod Edge chipping 3D view

Chipping thickness 2D view

Fig. 2.31: Illustration of chipping size and chipping thickness in RUM.

is the matrix layer with a clear brittle fracture edge (area 2 in Fig. 2.32). The last layer is the fibers in counter-clockwise direction (area 3 in Fig. 2.32). Chipping size of the last layer (from a to b) is about 600 μm and is nearly twice the size of the second layer. The edge chipping was further analyzed in three-dimensions (3D) using the microscope (VHK-1000, KEYENCE, Japan) as shown in Fig. 2.33. Measuring software (VHKH3M, KEYENCE, Japan) was used to measure chipping thickness on the 3D topography. The measuring section plane was along the direction from a to b in Fig. 2.33 and interacted with all three layers of the composite material (areas 1, 2 and 3 in Fig. 2.33). It is seen that the chipping thickness of layer 3 was 207.9 μm. The thickness of layer 2 was not uniform, but was smaller in the edge and big near the rod. However, compared to the size of the hole, the proportion of the chipping size and chipping thickness to radius and length of the machined hole was only 6.25 % and 1.89 %, respectively. From Fig. 2.32 and Fig. 2.33, layer 2 was mainly the remaining material not machined completely by the abrasive tool that removed part of the epoxy matrix in layer 2 until layer 3 was fractured from the composite and the whole rod was pressed out of the hole.

3

b

2 Top View direction

Bottom

a 1

100 μm

Fig. 2.32: Observation of machined rod and chipping size [22].

2.4 Effects of input variables on output variables

| 59

3D measurement Measure Result Height [A-B] 207.9 μm Width [C-D] 513.9 μm

Top Edge chipping

List >>

Top

2

3

View direction

Measuring section plane

Bottom

1

500.0 D

C

Measuring outline

207.9 μm A

a

B

1

b

2

3

Fig. 2.33: Observation and measurement of chipping thickness [22].

2.4.5 Effects on surface roughness The measurement of surface roughness Ra started at a location near the hole entrance and moved along the axial direction of the hole, as shown in Fig. 2.34. Four measurements were performed with 90° between two adjacent measurements. Each measurement was repeated twice, leading to eight Ra values in total for each hole. Average surface roughness Ra was chosen to evaluate the machined hole surface conditions with the tested range and cut-off length being set as 4 mm and 0.8 mm, respectively.

Four measurements 0°

90°

Probe Entrance

270°

Exit 180°

Fig. 2.34: Illustrations of surface roughness measurement.

60 | 2 Rotary ultrasonic machining of CFRP composites

Surface roughness (μm)

2.4.5.1 Effect of tool rotation speed A comparison of surface roughness among RUM with cutting fluid coolant, RUM with cold air coolant, twist drilling and abrasive grinding when tool rotation speed changed is shown in Fig. 2.35. Surface roughness decreased as tool rotation speed increased for all the methods. RUM with each type of coolant had smaller surface roughness than twist drilling when tool rotation speed was set at 2000, 3000 and 4000 rpm. When tool rotation speed was set between 1000 and 2000 rpm, RUM with cutting fluid had larger surface roughness than grinding; however, with the increase of tool rotation speed from 2000 to 5000 rpm, surface roughness in grinding was larger. At all levels of tool rotation speed, using cutting fluid led to smaller surface roughness than using cold air.

16

RUM with cutting fluid RUM with cold air Grinding Twist drill

8 4 2 1 1000

2000 3000 4000 Tool rotation speed (rpm)

5000

Fig. 2.35: Surface roughness comparison among the four methods with different levels of tool rotation speed.

2.4.5.2 Effect of feed rate A comparison of surface roughness among RUM with cutting fluid coolant, RUM with cold air coolant, twist drilling and abrasive grinding when feed rate changed is shown in Fig. 2.36. Surface roughness increased with the increase of feed rate for all of the methods. Surface roughness in RUM was significantly smaller than that in twist drilling at all settings of feed rate. The smallest surface roughness could be obtained in RUM with cutting fluid and grinding, both of which had nearly the same values. Also, using cutting fluid in RUM led to smaller surface roughness than using cold air.

2.4.5.3 Effect of ultrasonic power A comparison of surface roughness between cold air coolant and cutting fluid coolant using RUM when ultrasonic power changed is shown in Fig. 2.37. With the increase of ultrasonic power, surface roughness increased when using cold air, but did not change much when using cutting fluid. Using cutting fluid led to smaller surface roughness at all settings of ultrasonic power.

2.4 Effects of input variables on output variables

| 61

Surface roughness (µm)

6 RUM with cutting fluid RUM with cold air Grinding Twist drill

4 2 0 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

Feed rate (mm/s) Fig. 2.36: Surface roughness comparison among the four methods with different levels of feed rate.

3.0 RUM with cutting fluid RUM with cold air

Surface roughness (μm)

2.5 2.0 1.5 1.0 0.5 0.0 0

20 40 Ultrasonic power (%)

60

80

Fig. 2.37: Comparison of surface roughness using different type of coolant.

2.4.6 Effects on burning of machined surface During machining, the epoxy matrix could be burned under certain conditions due to machining-induced heat. Burning ratio (=

Burning area on machined hole surface ) Total area of machined hole surface

was used to describe the severity of burning on the machined hole surface. In the chapter, burning ratio was estimated by the ratio between the number of fiber layers that had burning and the number of fiber layers that had no burning. Burning of the machined surface did not occur when using cutting fluid in RUM under any of the test conditions. In contrast, burning occurred when using cold air in RUM under some conditions. Table 2.7 shows the results on burning ratio when using cold air. Table 2.7 shows the effects of ultrasonic power on the burning ratio. When the feed rate was 0.1 mm/s, higher ultrasonic power (80 %) caused a burning of machined surface. When the feed rate was 0.5 mm/s, burning did not occur no matter what the

62 | 2 Rotary ultrasonic machining of CFRP composites Tab. 2.7: Effects of ultrasonic power on burning ratio using cold air. Feed rate (mm/s)

Ultrasonic power (%) 0

20

40

60

80

0.1 0.5

0 0

0 0

0 0

0 0

10 0

* Tool rotation speed = 3000 rpm.

Tab. 2.8: Effects of tool rotation speed on burning ratio using cold air. Feed rate (mm/s)

Tool rotation speed (rpm) 1000

2000

3000

4000

5000

0.1 0.5

0 90 %

0 50 %

0 0

0 0

0 0

* Ultrasonic power = 40 %.

Tab. 2.9: Effects of feed rate on burning ratio using cold air. Feed rate (mm/s) 0.1 0

0.2 0

0.3 0

0.4 0

0.5 0

0.6 0

0.7 10 %

0.8 20 %

* Ultrasonic power = 40 %; Tool rotation speed = 3000 rpm.

ultrasonic power was. Table 2.8 shows the effects of tool rotation speed on the burning ratio. When the feed rate was 0.5 mm/s, the burning ratio became higher as tool rotation speed decreased (to 1000 or 2000 rpm). When the feed rate was 0.1 mm/s, burning did not occur no matter what the tool rotation speed was. Table 2.9 shows the effects of feed rate on the burning ratio. The burning ratio became higher when the feed rate was too high (0.7 and 0.8 mm/s).

2.4.7 Effects on tool wear Figure 2.38 shows pictures of a brand new tool and a used tool in RUM. It can be seen that the used tool was shorter than the new one. The used tool had been used to drill more than 200 holes and its abrasive portion decreased by 0.9 mm in length. At this wear rate, one tool with 7 mm length of abrasive portion can drill more than 1400 holes. Figure 2.39 compares tool wear (i.e. cumulative tool weight loss) between the two types of coolant. After each drilling test, tool weight loss was measured after removing the tool from the RUM machine. The weight was measured with a scale (APX-200,

2.4 Effects of input variables on output variables

Used tool

| 63

New tool Fig. 2.38: A new RUM tool and a used RUM tool after drilling more than 200 holes.

Denver Instrument, USA). For the first ten holes, both types of coolant had similar tool wear. After ten holes, as more holes were drilled, tool wear increased steadily when using cold air, but did not change much when using cutting fluid. Differences in tool weight loss between the two types of coolant increased as more holes were drilled. After 30 holes were drilled, the difference was around 8 mg.

Cumulative weight loss (mg)

15 RUM with cutting fluid RUM with cold air

12 9 6 3 0 0

5

10

15

Number of holes

20

25

30

Fig. 2.39: Comparison of tool wear.

The digital microscope (Model BX51, Olympus Inc., Tokyo, Japan) with magnification from 50 to 200 was employed to study tool wear mechanisms in RUM by examining the end face and lateral face of the tool. To guarantee that the same area of the tool’s surface was observed every time, a special fixture was applied to hold the tool. The appearance of the diamond grains on the tool’s lateral face did not visibly change after 30 holes were drilled, which means only a small amount of wear of diamond grains occurred on the tool’s lateral face. However, wear of the diamond grains at the end face was obvious. Figure 2.40 shows the tool topographies in the same place of the end face before and after 30 holes were drilled. Attrition wear can be observed on the diamond grains and significant wear of the sharp cutting edges can be also observed on grains 1 and 2. Figure 2.41 shows the tool topographies in the same place of the end face before and after 30 holes were drilled. It can be seen that the diamond grain is completely

64 | 2 Rotary ultrasonic machining of CFRP composites

1

1

2

2

200 μm Before

200 μm After

Fig. 2.40: Attrition wear of the diamond grain on the tool end surface [22].

pulled out of the metal bond, forming a hole on the tool end face. After the drilling tests, it was found that only one grain was pulled out and others remained the same. This is due to the fact that the grain was pulled out prematurely before its working life was effectively completed. There is no visible grain in Fig. 2.41 after drilling tests. Grooves on the metal bond surface after drilling are lower than those before drilling with a brighter metal bond. These suggest that no diamond grain in the machining process would result in wearing the metal bond.

200 μm Before

200 μm After

Fig. 2.41: Grain pulled out on the tool end surface [22].

The relation between tool weight loss and the number of drilled holes is shown in Fig. 2.39. It can be seen that the most weight was lost (less than 5 mg) after the first 5 holes were drilled. The difference of the tool length before and after 30 holes were drilled changed little (less than 1 μm). Tool weight loss increased slightly with an increase in the number of holes drilled.

2.4 Effects of input variables on output variables

| 65

2.4.8 Effects on MRR Material removal rate (MRR) was calculated as the volume of removed material divided by machining time. It can be expressed by the following equation: 2

MRR =

π ⋅ [(D/2)2 − (Dr /2) ] ⋅ h T

,

(2.24)

where D is the diameter of machined hole, h is the thickness of workpiece, T is the time for drilling the hole and Dr is the diameter of the machined rod. It was also measured by the vernier caliper.

2.4.8.1 Effect of tool rotation speed A comparison of the MRR among RUM, twist drilling and abrasive grinding when tool rotation speed changed is shown in Fig. 2.42. In this section, in order to compare all the items in the figures, the plot of the twist drill will follow the range of the right vertical axis and the other two plots will follow the left one. It can be seen that at all levels of tool rotation speed, MRR almost remained constant for all the three methods. The MRR in twist drilling was higher than those in RUM and grinding. 15

2.94

14.8

2.88

14.6

2.82

14.4

2.76 2.7 1000

14.2

RUM Grinding Twist drill

2000

3000

4000

Material removal rate (mm3/s) – Solid line

Material removal rate (mm3/s) – Dash lines

3

14 5000

Tool rotation speed (rpm) Fig. 2.42: MRR comparison among the four methods with different levels of tool rotation rate.

2.4.8.2 Effect of feed rate A comparison of MRR among RUM, twist drilling and abrasive grinding when feed rate changed is shown in Fig. 2.43. In these three methods, MRR increased linearly with the increase of feed rate. However, the increasing rate of twist drilling was larger than those of the other two that were almost the same. In addition, the values of MRR in twist drilling were higher than those in RUM and grinding at all levels of feed rate.

66 | 2 Rotary ultrasonic machining of CFRP composites

Material removal rate (mm3/s)

21

RUM Grinding Twist drill

18 15 12 9 6 3 0 0

0.1

0.2

0.3 0.5 0.4 Feed rate (mm/s)

0.6

0.7

0.8

Fig. 2.43: MRR comparison among the four methods with different levels of feed rate.

2.4.9 Effects on power consumption Power consumption of the entire RUM system and each component under different settings of ultrasonic power, tool rotation speed, feed rate and CFRP type was studied. The power consumption presented in this chapter was the electricity energy (W) consumed when drilling a hole in the workpiece material divided by the workpiece thickness (mm). Note that the CFRP workpieces have different thicknesses.

2.4.9.1 Effect of tool rotation speed Effects of tool rotation speed on power consumption are shown in Fig. 2.44. As tool rotation speed increased, power consumption of ultrasonic power supply decreased, power consumption of spindle motor increased dramatically, power consumption of coolant pump and air compressor kept unchanged and power consumption of the entire RUM system increased slightly. Power consumption percentages of each component under different settings of tool rotation speed are shown in Fig. 2.45. For different settings of tool rotation speed, the power consumption of the coolant pump always had the largest percentage. As tool rotation speed increased from 1000 to 5000 rpm, the power consumption percentage of the ultrasonic power supply decreased slightly from 11 % to 8 %, the power consumption percentage of the spindle motor increased from 1 % to 15 %, the power consumption percentage of the coolant pump decreased from 76 % to 67 % and the power consumption percentage of the air compressor decreased slightly from 12 % to 10 %.

2.4 Effects of input variables on output variables

Power consumption (w·h/mm)

0.5

| 67

Ultrasonic power supply

0.4 0.3 0.2 0.1 0

Power consumption (w·h/mm)

0.5

Spindle motor

0.4 0.3 0.2 0.1 0

Power consumption (w·h/mm)

4 Coolant pump 3 2 1 0

Power consumption (w·h/mm)

0.5 Air compressor 0.4 0.3 0.2 0.1 0

Power consumption (w·h/mm)

5 Total

4 3 2 1 0 1000

2000 3000 4000 Tool rotation speed (rpm)

5000

Fig. 2.44: Effects of tool rotation speed. (Ultrasonic power = 30 %; Feed rate = 0.5 mm/s; CFRP #1).

68 | 2 Rotary ultrasonic machining of CFRP composites

Air compressor 12%

Ultrasonic power supply 11%

Spindle motor 1%

Ultrasonic Air compressor power supply 9% 11% Spindle motor 9%

Coolant pump 76%

Coolant pump 71%

(a)

(b)

Air compressor 10%

Ultrasonic power supply 8% Spindle motor 15%

Coolant pump 67% (c) Fig. 2.45: Power consumption percentage of each component under different settings of tool rotation speed. (Ultrasonic power = 30 %; Feed rate = 0.5 mm/s; CFRP #1). (a) Tool rotation speed = 1000 rpm; (b) Tool rotation speed = 3000 rpm; (c) Tool rotation speed = 5000 rpm.

2.4.9.2 Effect of feed rate The effects of feed rate on power consumption are shown in Fig. 2.46. As the feed rate increased, the power consumptions of the ultrasonic power supply, spindle motor and coolant pump decreased dramatically, whereby the power consumption of the air compressor remained the same and the power consumption of the entire RUM system increased remarkably. The power consumption percentages of each component under different settings of feed rate are shown in Fig. 2.47. As feed rate increased from 0.1 to 0.7 mm/s, the power consumption percentage for the air compressor increased from 2 % to 14 %, the power consumption percentage of the coolant pump decreased from 79 % to 67 % and power consumption percentages for the ultrasonic power supply and spindle motor did not change much and remained at approximately 9 %.

2.4 Effects of input variables on output variables

Power consumption (w·h/mm)

0.5

| 69

Ultrasonic power supply

0.4 0.3 0.2 0.1 0

Power consumption (w·h/mm)

0.5

Spindle motor

0.4 0.3 0.2 0.1 0

Power consumption (w·h/mm)

4

Coolant pump

3 2 1 0

Power consumption (w·h/mm)

0.5 0.4

Air compressor

0.3 0.2 0.1 0

Power consumption (w·h/mm)

5 Total

4 3 2 1 0 0.1

0.3 0.5 Feed rate (mm/s)

0.7

Fig. 2.46: Effects of feed rate. (Ultrasonic power = 30 %; tool rotation speed = 3000 rpm; CFRP #1).

70 | 2 Rotary ultrasonic machining of CFRP composites

Air compressor 2%

Ultrasonic power supply 10%

Air compressor 7%

Ultrasonic power supply 9%

Spindle motor 9%

Spindle motor 9%

Coolant pump 79%

Coolant pump 75% (b)

(a)

Air compressor 11%

Ultrasonic power supply 9%

Spindle motor 9%

Air compressor 14%

Spindle motor 10%

Coolant pump 67%

Coolant pump 71% (c)

Ultrasonic power supply 9%

(d)

Fig. 2.47: Power consumption percentage of each component under different settings of feed rate. (Ultrasonic power = 30 %; tool rotation speed = 3000 rpm; CFRP #1). (a) Feed rate = 0.1 mm/s; (b) Feed rate = 0.3 mm/s; (c) Feed rate = 0.5 mm/s; (d) Feed rate = 0.7 mm/s.

2.4.9.3 Effect of ultrasonic power The effects of ultrasonic power on power consumption for the entire RUM system and each component are shown in Fig. 2.48. When ultrasonic power increased from 0 to 80 %, the power consumption of the ultrasonic power supply increased slightly, the power consumption of the spindle motor decreased significantly, the power consumption of the coolant pump and air compressor was constant and the power consumption of the entire RUM system almost remained constant. The power consumption percentages for each component under different settings of ultrasonic power are shown in Fig. 2.49. For different settings of ultrasonic power, the power consumption of the coolant pump always had the highest percentage (about 70 % of the entire RUM system power consumption) and the power consumption percentage of the air compressor remained unchanged at 11 %. As the ultrasonic power

2.4 Effects of input variables on output variables

Power consumption (w·h/mm)

0.5

| 71

Ultrasonic power supply

0.4 0.3 0.2 0.1 0

Power consumption (w·h/mm)

0.5

Spindle motor

0.4 0.3 0.2 0.1 0

Power consumption (w·h/mm)

4

Coolant pump

3 2 1 0

Power consumption (w·h/mm)

0.5

Air compressor

0.4 0.3 0.2 0.1 0

Power consumption (w·h/mm)

5 Total

4 3 2 1 0 0

20 40 Ultrasonic power (%)

60

80

Fig. 2.48: Effects of ultrasonic power. (Tool rotation speed = 3000 rpm; Feed rate = 0.5 mm/s; CFRP #1).

72 | 2 Rotary ultrasonic machining of CFRP composites increased, power consumption percentage for the ultrasonic power supply increased from 0 to 16 %; in contrast, the spindle motor power consumption percentage decreased from 20 % to 3 %. The effects of ultrasonic power on other output variables (including cutting force, torque and surface roughness) had been studied [17]. When ultrasonic power increased from 0 % to 80 %, cutting force and torque decreased. The decrease of cutting force and torque were about 20 % and 40 %, respectively. When ultrasonic power increased from 0 to 80 %, surface roughness decreased first and then increased. Compared with RUM without ultrasonic power, the decrease of surface roughness was about 10 %. It is noted that ultrasonic power in RUM can reduce cutting force, torque and surface roughness without increasing power consumption.

Air compressor 11%

Spindle motor 20%

Air compressor 11%

Coolant pump 69%

Ultrasonic power supply 9%

Spindle motor 9%

Coolant pump 71%

(a)

(b)

Air compressor 11%

Ultrasonic power supply 16% Spindle motor 3%

Coolant pump 70% (c) Fig. 2.49: Power consumption percentage of each component under different settings of ultrasonic power. (Tool rotation speed = 3000 rpm; Feed rate = 0.5 mm/s; CFRP #1). (a) Ultrasonic power = 0 %; (b) Ultrasonic power = 40 %; Ultrasonic power = 80 %.

2.4 Effects of input variables on output variables

| 73

2.4.9.4 Effect of CFRP type The effects of CFRP type on power consumption are shown in Fig. 2.50. The CFRP type significantly affected power consumption of the ultrasonic power supply and spindle motor. The power consumption of the ultrasonic power supply was the highest when machining CFRP #1 (with wide yarn woven fiber structure) and the lowest when machining CFRP #3 (with flake fiber structure). In contrast, the power consumption of the spindle motor was the highest when machining CFRP #3 and the lowest when machining CFRP #1. For different types of CFRP, the power consumption of the coolant pump and air compressor remained unchanged. The power consumption for the entire RUM system did not change much for these different CFRP types. The power consumption percentages of each component for RUM of different CFRP types are shown in Fig. 2.51. When the CFRP type changed, the power consumption of the coolant pump always had the highest percentage (71 % ∼ 73 %). The power consumption percentage for the air compressor remained at 7 %. The power consumption percentages for the ultrasonic power supply and spindle motor changed slightly.

2.4.10 Effects on feasible regions 2.4.10.1 Feasible regions of tool rotation speed and feed rate Figure 2.52 shows the feasible regions of tool rotation speed and feed rate. The range of tool rotation speed was from 500 to 6000 rpm and the feed rate range was from 0.1 to 0.9 mm/s. The ultrasonic power was set at 40 %. Figure 2.52 (a) shows the feasible region with cold air pressure of 40 psi. When tool rotation speed was 1000 rpm or lower, dry machining was not feasible at any level of feed rate. Tool blockage (indicated by letter “T” in Fig. 2.52) happened under all of these conditions. Burning (indicated by letter “B” in Fig. 2.52) could be observed on most of the machined holes. Holes drilled with a higher feed rate always had delamination (indicated by letter “D” in Fig. 2.52). When the tool rotation speed was 4000 rpm, dry machining was feasible at all levels of feed rate. At other levels of tool rotation speed (instead of 500, 1000 and 4000 rpm), dry machining was feasible at some levels of feed rate. For example, dry machining was feasible when tool rotation speed was 3000 or 5000 rpm and feed rate was from 0.1 to 0.7 mm/s, when tool rotation speed was 2000 rpm and feed rate was 0.1 or 0.3 mm/s, as well as when tool rotation speed was 6000 rpm and feed rate was 0.5 or 0.7 mm/s. Burning was the primary limiting criterion under these conditions, although workpiece delamination and tool blockage also occurred under some conditions. Figure 2.52 (b) shows the feasible region with cold air pressure of 50 psi. The feasible region was larger than that with cold air pressure of 40 psi. When tool rotation speed was from 3000 to 5000 rpm, dry machining was feasible at all levels of feed rate. However, when the tool rotation speed was 1000 rpm or lower, dry machining

74 | 2 Rotary ultrasonic machining of CFRP composites

Power consumption (w·h/mm)

0.5

Ultrasonic power supply

0.4 0.3 0.2 0.1 0

Power consumption (w·h/mm)

0.5

Spindle motor

0.4 0.3 0.2 0.1 0

Power consumption (w·h/mm)

4

Coolant pump

3 2 1 0

Power consumption (w·h/mm)

0.5

Air compressor

0.4 0.3 0.2 0.1 0

Power consumption (w·h/mm)

5 Total

4 3 2 1 0 0

#1

#2 #3 Type of CFRP

#4

Fig. 2.50: Effects of CFRP type. (Ultrasonic power = 30 %; Tool rotation speed = 3000 rpm; Feed rate = 0.3 mm/s).

2.4 Effects of input variables on output variables

Air compressor 7%

Ultrasonic power supply 9%

Air compressor 7%

Ultrasonic power supply 9%

Spindle motor 9%

Spindle motor 11%

Coolant pump 75%

Coolant pump 73% (b)

(a)

Air compressor 7%

Ultrasonic power supply 8%

Spindle motor 14%

Air compressor 7%

Ultrasonic power supply 8%

Spindle motor 13%

Coolant pump 72%

Coolant pump 71% (c)

| 75

(d)

Fig. 2.51: Power consumption percentage of each component under different CFRP types. (Ultrasonic power = 30 %; Tool rotation speed = 3000 rpm; Feed rate = 0.3 mm/s). (a) CFRP #1; (b) CFRP #2; (c) CFRP #3; (d) CFRP #4.

was not feasible when the feed rate = 0.1 mm/s and tool rotation speed = 1000 rpm). When the tool rotation speed was 2000 rpm, dry machining was feasible only when feed rate was from 0.1 to 0.5 mm/s. When tool rotation speed was 6000 rpm, dry machining was feasible at all levels of feed rate except 0.1 mm/s.

2.4.10.2 Feasible regions of ultrasonic power and feed rate Figure 2.53 shows feasible regions of ultrasonic power and feed rate. Ultrasonic power was changed from 0 to 100 % with an interval of 20 %, feed rate was changed from 0.1 to 0.9 mm/s with an interval of 0.2 mm/s and tool rotation speed was set at 3000 rpm. With a cold air pressure of 40 psi, dry machining was feasible when the ultrasonic power ≤ 60 % and the feed rate ≤ 0.7 mm/s. A combination of a high feed rate

Feed rate (mm/s)

76 | 2 Rotary ultrasonic machining of CFRP composites

Feed rate (mm/s)

Feasible Not feasible 0.9 B/D/T B/D/T B/T B/D B 0.7 B/D/T B/T B/T 0.5 B/T B/T B/T 0.3 B/T B/T 0.1 B/T T 500 1000 2000 3000 4000 5000 Tool rotation speed (rpm) (a)

Feed rate (mm/s)

Feasible Not feasible 0.9 B/D/T B/D/T B/T 0.7 B/D/T B/T B/T 0.5 B/T B/T 0.3 B/T B 0.1 B/T 500 1000 2000 3000 4000 5000 (b) Tool rotation speed (rpm)

0.9 0.7 0.5 0.3 0.1

B/T

0%

Feed rate (mm/s)

(a)

20% 40% 60% Ultrasonic power

Feasible 0.9 0.7 0.5 0.3 0.1

B B 6000

B 6000

Fig. 2.52: Feasible regions of tool rotation speed and feed rate. (a) Cold air pressure = 40 psi; (b) Cold air pressure = 50 psi.

Not feasible

B 80%

B B/T B/T B/T 100%

Not feasible

B/T

0% (b)

Feasible B B

B/D

20% 40% 60% Ultrasonic power

B 80%

B 100%

Fig. 2.53: Feasible regions of ultrasonic power and feed rate. (a) Cold air pressure = 40 psi; (b) Cold air pressure = 50 psi.

(0.9 mm/s) and low ultrasonic power (≤ 40 %) or a combination of a low feed rate and high ultrasonic power would cause burning of machined surface or tool blockage. The feasible region became large when cold air pressure was increased from 40 to 50 psi.

2.5 Summary |

77

Ultrasonic power

2.4.10.3 Feasible regions of tool rotation speed and ultrasonic power Figure 2.54 shows the feasible regions of tool rotation speed and ultrasonic power. The range of tool rotation speed was from 500 to 6000 rpm with an interval of 1000 (except 500 from 500 to 1000), the range of ultrasonic power was from 0 % to 100 % with an interval of 20 % and the feed rate was fixed at 0.5 mm/s. When the tool rotation speed was ≥ 3000 rpm, dry machining was feasible under most conditions except when ultrasonic power = 100 % and tool rotation speed = 3000 or 6000 rpm and when ultrasonic power = 0 and tool rotation speed ≥ 5000 rpm (these combinations would result in burning of the machined workpiece). When tool rotation speed was ≤ 2000 rpm, dry machining was only feasible when tool rotation speed = 2000 rpm and ultrasonic power = 60 % or 80 %. Other conditions would cause burning of the machined surface, or burning and tool blockage, or burning and tool blockage and workpiece delamination. The feasible region became large when cold air pressure was increased from 40 to 50 psi. With cold air pressure of 50 psi, dry machining was feasible when tool rotation speed ≥ 2000 rpm (at all levels of ultrasonic power).

100% B/T 80% B/T 60% B/T 40% B/T 20% B/T 0% B/D/T 500

Feasible Not feasible B/T B B B/T B/T B/T B/T B/T B/T B/T B/T B 1000 2000 3000 4000 5000 Tool rotation speed (rpm)

100% B/D/T 80% B/T 60% B/T 40% B/T 20% B/T 0% B/T 500

Feasible Not feasible B/T B/T B/T B/T B B 1000 2000 3000 4000 5000 Tool rotation speed (rpm)

Ultrasonic power

(a)

(b)

B

B 6000

6000

Fig. 2.54: Feasible regions of tool rotation speed and ultrasonic power. (a) Cold air pressure = 40 psi; (b) Cold air pressure = 50 psi.

2.5 Summary This chapter presents the introductions of rotary ultrasonic machining (RUM) including machine components, input variables, output variables and its machinability in the RUM of CFRP composites. The structure of the chapter is shown in Fig. 2.55.

78 | 2 Rotary ultrasonic machining of CFRP composites

Rotary ultrasonic machining of CFRP composites RUM system ∙ Principle and features of RUM ∙ Ultrasonic spindle system – Ultrasonic spindle – Ultrasonic power supply – Ultrasonic transducer – Ultrasonic amplitude transformer and tool holder – Cutting tool – Electric motor – Feeding device – Control panel ∙ Coolant system ∙ Data acquisition system

Input variables ∙ Machining variables – Tool rotation speed – Feed rate – Ultrasonic power – Ultrasonic vibration frequency ∙ Cutting tool variables – Abrasive type – Abrasive size – Abrasive concentration ∙ Cooling variables – Coolant type – Coolant flow rate – Coolant pressure

Output variables ∙ Cutting force ∙ Torque ∙ Cutting temperature ∙ Edge quality ∙ Surface roughness ∙ Burning of machined surface ∙ Tool wear ∙ Material removal rate (MRR) ∙ Power consumption ∙ Feasible regions

Fig. 2.55: Framework of this chapter.

Acknowledgement This work is supported by the Foundation of the Whitacre College of Engineering and the Office of Vice President for Research at Texas Tech University. The authors also want to extend their acknowledgements to Elsevier and Sage Publications who granted the permission to use copyrighted figures in this chapter.

References [1] [2] [3] [4] [5] [6] [7] [8]

Barbero EJ. Chapter 4 Micromechanics, Introduction to composite material design. 2nd ed. Boca Raton, FL, USA: CRC Press, Taylor & Francis Group, LLC.;2010:91–142. Bhatnagar N, Naik NK, Ramakrishnan N. Experimental investigations of drilling on CFRP composites. Material and Manufacturing Process 1993;8(6):683–701. Boothroyd G, Knight A. Nonconventional machining processes in fundamentals of machining and machine tools. Boca Roton, FL, USA: CRC press;2006:467–517. Bradford JD, Richardson DB. , Production engineering technology. 3rd ed. London: Eacmillan; 1980. Boeing Co. website. Boeing 787 from the ground up. 2006. Available at: http://www.boeing. com/commercial/aeromagazine/articles/qtr_4_06/article_04_2.html. Boeing Co. website. 787 Dreamliner Program Fact Sheet. 2014. Available at: http://www. boeing.com/commercial/787family/programfacts.html. Chao PY, Hwang YD. An improved Taguchi’s method in design of experiments for milling CFRP composite. International journal of production research 1997;35(1):51–66. Chen WC. Some experimental investigations in the drilling of carbon fiber-reinforced plastic (CFRP) composite laminates. International Journal of Machine Tools and Manufacture 1997; 37(8):1097–1108.

References |

79

[9] Cleave D. Ultrasonic gets bigger jobs in machining and welding. Iron Age 1976;9:69–72. [10] Chung DDL. Composite materials science and applications. 2nd ed. London: Springer London Ltd;2010. [11] Churi NJ, Pei ZJ, Treadwell C. Rotary ultrasonic machining of titanium alloy: effects of machining variables. Machining Science and Technology 2006;10(3):301–321. [12] Cong WL, Pei ZJ, Mohanty N, Van Vleet E, Treadwell C. Vibration amplitude in rotary ultrasonic machining: a novel measurement method and effects of process variables. Journal of Manufacturing Science and Engineering 2011;133(3):034501-1–6. [13] Cong WL, Feng F, Pei ZJ, Deines TW, Treadwell C. Experimental study on cutting temperature in rotary ultrasonic machining. Proceedings of NAMRI/SME, 39, 369–376, Corvallis, OR, USA, June 13–17,2011. [14] Cong WL, Feng F, Pei ZJ, Deines TW, Treadwell C. Dry machining of carbon fiber reinforced plastic composite by rotary ultrasonic machining: effects of machining variables. Proceedings of the ASME 2011 International Manufacturing Science and Engineering Conference (MSEC 2011), Corvallis, OR, USA, June 13–17, 2011. [15] Cong WL, Pei ZJ, Deines TW, Treadwell C. Rotary ultrasonic machining of CFRP using cold air as coolant: feasible regions. Journal of Reinforced Plastics and Composites 2011;30(10): 899–906. [16] Cong WL, Pei ZJ, Feng Q, Deines TW, Treadwell C. Rotary ultrasonic machining of CFRP: a comparison with twist drilling. Journal of Reinforced Plastics and Composite 2012;31(5):313–321. [17] Cong WL, Feng Q. Pei ZJ, Deines TW, Treadwell C. Rotary ultrasonic machining of carbon fiber reinforced plastic composites: using cutting fluid versus cold air as coolant. Journal of Composite Materials 2012;46(14):1745–1753. [18] Cong WL, Zou XT, Deines TW, Wu N, Wang XW, Pei ZJ. Rotary ultrasonic machining of CFRP composites: an experimental study on cutting temperature. Journal of Reinforced Plastics and Composite 2012;31(22):1516–1525. [19] Cong WL, Pei ZJ, Deines TW, Srivastava A, Riley L, Treadwell C. Rotary ultrasonic machining of CFRP composites: a study on power consumption. Ultrasonics 2012;52(8):1030–1037. [20] Cong WL, Pei ZJ, Sun X, Zhang CL. 2014, Rotary ultrasonic machining of CFRP: A mechanistic predictive model for cutting force. Ultrasonics 2012;54(2):663–675. [21] Davim JP, Reis P. Drilling carbon fiber reinforced plastics manufactured by autoclave-experimental and statistical study. Materials and Design 2003;24(5):315–324. [22] Feng Q, Cong WL, Pei ZJ, Ren CZ. Rotary ultrasonic machining of carbon fiber-reinforced polymer: feasibility study. Machining Science and Technology 2012;16(3):380–398. [23] Gay D, Hoa SV, Tsai SW. , Composite materials design and applications. 4th ed. Boca Raton, FL, USA: CRC press; 2003. [24] Graff K. Ultrasonic machining. Ultrasonic 1975;5:103–109. [25] Guu YH, Hocheng H, Tai NH, et al. Effect of electrical discharge machining on the characteristics of carbon fiber reinforced carbon composites. Journal of Materials Science 2001;36(8): 2037–2043. [26] Yancey R. IndustryWeek. Website. How composites are strengthening the aviation industry. June 11, 2012. Available at: http://www.industryweek.com/none/how-composites-arestrengthening-aviation-industry. [27] Kaw AK. , Chapter 3 Micromechanical analysis of a lamina, Mechanics of composite materials. 2nd ed. Boca Raton, FL, USA: CRC Press, Taylor & Francis Group, LLC.;2006. [28] Kinet D, Mégret P, Goossen KW, et al. Fiber bragg grating sensors toward structural health monitoring in composite materials: Challenges and Solutions. Sensors 2014;14(4):7394–7419.

80 | 2 Rotary ultrasonic machining of CFRP composites [29] Krishnaraj V, Prabukarthi A, Ramanathan A, Elanghovan N, Senthil Kumar M, Zitoune R, Davim JP. Optimization of machining parameters at high speed drilling of carbon fiber reinforced plastic (CFRP) laminates. Composites Part B: Engineering 2012;43(4):1791–1799. [30] Lau WS, Wang M, Lee WB. Electrical discharge machining of carbon fibre composite materials. International Journal of Machine Tools and Manufacture 1990;30(2):297–308. [31] Lau WS, Yue TM, Lee TC, Lee WB. Un-conventional machining of composite materials. Journal of Material Process Technology 1995;48:199–205. [32] Lazar MB. Cutting force modeling for drilling of fiber reinforced composites. Doctoral Dissertation, Swiss Federal Institutes of Technology in Lausanne, Switzerland;2012. [33] Legge P. Ultrasonic drilling of ceramics. Industrial Diamond Review 1964;24(278):20–24. [34] Lemma E, Chen L, Siores E, Wang, J. Study of cutting fiber-reinforced composites by using abrasive water-jet with cutting head oscillation. Composite Structures 2002;57(1):297–303. [35] Liu DF, Cong WL, Pei ZJ, Tang, YJ. A cutting force model for rotary ultrasonic machining of brittle materials. International Journal of Machine Tools and Manufacture 2012;52(1):77–84. [36] Mallick PK. Fiber-reinforced composites: materials, manufacturing and design. New York, NY, USA: CRC press, Marcel Dekker Inc.;1993. [37] Mallick PK. Composite engineering handbook. New York, NY, USA: CRC Press, Marcel Dekker Inc.;1997. [38] Matthews FL, Davies GAO, Hitchings D, Soutis C. Finite element modeling of composite materials and structures. Boca Raton, FL, USA: CRC Press LLC.;2003. [39] Mazumdar S. Composites manufacturing: materials, product and process engineering, New York, NY, USA: CRC Press, Marcel Dekker Inc.;2001. [40] Ning FD, Cong WL, Pei ZJ, Tang, YJ, Treadwell C. Rotary ultrasonic machining of CFRP: A comparison with abrasive grinding. Submitted to Journal of Reinforced Plastics and Composites;2015. [41] Park KY, Choi JH, Lee DG. Delamination-free and high efficiency drilling of carbon fiber reinforced plastics. Journal of Composite Materials 1995;29(15):1988–2002. [42] Pei ZJ, Prabhakar D, Ferreira PM, Haselkorn M. A mechanistic approach to the prediction of material removal rates in rotary ultrasonic machining. Journal of Engineering for Industry 1995; 117(2):142–151. [43] Peters ST. Handbook of composites. London, UK: Thompson Science;1998 [44] Petrukha P. Ultrasonic diamond drilling of deep holes in brittle materials. Russian Engineering Journal 1970;50(10):70–74. [45] Quilter A. , Composites in aerospace applications. IHS White Paper 2001;444:1–3. [46] Rajasekaran T, Palanikumar K, Vinayagam BK. Experimental investigation and analysis in turning of CFRP composites. Journal of Composite Materials 2011;46(7):809–821. [47] Sadat AB. Delamination and other types of damage of graphite/epoxy composite caused by machining. Applied Mechanics Division and the Materials Division: Machining of Advanced Materials. Los Angeles, CA: June 28–30 1995:41–52. [48] Sasahara H, Kikuma T, Koyasu R, Yao Y. Surface grinding of carbon fiber reinforced plastic (CFRP) with an internal coolant supplied through grinding wheel. Precision Engineering 2014; 38(4):775–782. [49] Singal RK, Singal M, Singal R. Fundamentals of machining and machine Tools. New Delhi, India: I.K. International Publishing House Pvt. Ltd.;2008. [50] Soo SL, Shyha IS, Barnett T, Aspinwall DK, Sim W.M. Grinding performance and workpiece integrity when superabrasive edge routing carbon fibre reinforced plastic (CFRP) composites. CIRP Annals-Manufacturing Technology 2012;61(1):295–298. [51] Strong AB. Fundamentals of composites manufacturing: materials, methods and applications, 2nd ed. Dearborn, MI, USA: SME;2008.

References |

81

[52] Thoe TB, Aspinwall DK, Wise MLH. Review on ultrasonic machining. International Journal of Machine Tools and Manufacture 1998;38(4):239–255. [53] Tong L, Mouritz AP, Bannister MK. 3D fibre reinforced polymer composites. Oxford, UK: Elsevier Science Ltd;2002. [54] Van Herwaarden AW, Sarro PM. Thermal sensors based on the Seebeck effect. Sensors and Actuators 1986:10(3–4);321–346. [55] Wong TL, Wu SM, Croy GM. An analysis of delamination in drilling composite materials. Proceedings 14th SAMPLE Technology Conference 1982, Atlanta, GA, USA:472–483. [56] Yadav RN, Yadava V. Influence of input parameters on machining performance of slottedelectrical discharge abrasive grinding of Al/SiC/Gr metal matrix composite. Materials and Manufacturing Processes 2013;28(12):1361–1369.

M. Slamani, J.-F. Chatelain

3 High-speed robotic trimming of CFRP Abstract: The demand for aircraft components with high mechanical and physical properties such as high strength, high stiffness, low weight, durability and extreme corrosion resistance has increased the popularity of CFRP. CFRP parts are usually produced by molding or near net shape processing. In some applications, however, trimming, milling and drilling are still required to bring CFRP parts to their final shapes and sizes. While the trimming of carbon fiber is generally carried out with CNC machine tools, however, the process is relatively slow due to the low flexibility of machine tools, their high up-front costs and their higher operating costs. Thanks to their adaptability, programmability, high dexterity and good maneuverability, industrial robots offer cutting-edge and lower-cost solutions than machine tools for bringing molded CFRP parts to their final shapes and sizes. However, for such applications, the robot must be very stiff for the machining operation to be generated. If the robot is not sufficiently stiff, deviations in the shape and position of the workpieces will occur. In this chapter, the machinability of CFRP with high-speed robotic end milling under varying cutting conditions is analyzed and evaluated. Cutting force, trajectory deviation and surface roughness are used as the machinability criteria.

3.1 Introduction CFRP are gaining ever-wider acceptance in aerospace applications, thanks to their desirable mechanical and physical properties. CFRP parts are usually produced by molding or near net shape; in some applications, trimming, milling and drilling are still required to bring CFRP parts to their final shapes and sizes. While the trimming of carbon fiber is generally carried out with the CNC machine tool, this is however a relatively slow process due to the machine tool’s low flexibility, its high up-front costs and its higher operating costs. Thanks to their adaptability, programmability, high dexterity and good maneuverability, industrial robots offer cutting-edge and lower-cost solutions than machine tools for bringing molded CFRP parts to their final shapes and sizes. Indeed, these robots have already been introduced to many industrial applications, including welding, painting and assembly and have produced excellent results. They are relatively cheaper when compared to machine tools, are flexible and have a large working envelope. However, under a heavy cut and given their serial structure, such robots are susceptible to errors of many sources, with the most essential being geometrical errors, servo errors, as well as the end-effector deflections caused by cutting forces and torques.

84 | 3 High-speed robotic trimming of CFRP Without contact, the static accuracy of an industrial robot is mostly affected by geometric errors caused by mechanical-geometrical imperfections, such as link parameter errors and non-geometric errors due to gravity, joint compliance, gear errors and backlash. In general, the static accuracy of an industrial robot can be improved by an appropriate calibration technique. Unfortunately, static accuracy does not consider the real mode of the robot’s operation. The industrial robot consists of several links, each of which is connected to one servo system. Furthermore, servo vibration and dynamic errors also affect the dynamic performance of the robot, especially at high TCP speeds (feed rate), and thermal expansion can affect repeatability and accuracy considerably, albeit only when long periods of time are considered (e.g. several hours of operation after a cold start). It is well known that in many automated manufacturing systems, higher speed is a key to productivity enhancement. High accuracy trajectory performance is also a requirement in many industrial robot operations and should be provided by the servo mechanism. Many investigative directions have been proposed to increase the dynamic accuracy of industrial robots, such as motion control and trajectory planning, for example. Unfortunately, a huge gap exists between the sophisticated control algorithms developed in robotics research and the simple control techniques used in industrial robotic applications. A major problem with the servo systems of industrial robots is trajectory deviation and contour error. When the robot speed is relatively low, the trajectory deviation and contour error caused by the servo system are usually acceptable. However, once high speed and high accuracy are demanded, as in waterjet cutting, laser cutting, gluing, dispensing and high-speed trimming, for example, trajectory deviation and contour errors will have a significant effect. On the other hand, gravitational loading causes structural deformation and compliance errors, which can occur not only as a result of the robot manipulating a heavy object, but also due to the links supporting their own weight. A similar problem is also present during machining, where high cutting forces are generated and act on the robot structure. Usually, in robotic machining, this force causes deflections that directly affect the quality of machined parts. To overcome this limitation and significantly reduce cutting forces, the feed rate, depth of cut and cutter diameter must be maintained at small values. Furthermore, the robot deflection problem can also be solved by modifying the robot model or the robot control program. In practice, the latter approach is considered more realistic. However, it requires an accurate stiffness model and a precise cutting force model. To date, robotic machining has been applied on soft materials (like plastics) and low hardness materials (such as aluminum alloys). Robotic machining of heterogeneous materials such as CFRP remains a serious challenge in performing manufacturing operations.

3.2 Machinability of CFRP |

85

3.2 Machinability of CFRP The first theoretical work on the orthogonal machining of CFRP was reported by Everstine and Rogers [1] in 1971. They presented a theoretical analysis of machining of materials reinforced with strong fibers. They formulated a model for the prediction of the cutting force for fiber-reinforced materials for a special case in which the leading edge of the tool is perpendicular to the direction of relative motion between the tool and the workpiece. A few years later (1980), Koplev [2, 3] studied CFRP orthogonal cutting. He found that chip formation was strongly affected by fiber orientation and occurred during a series of successive ruptures. He also concluded that surface quality and the delamination factor were strongly influenced by cutting forces and tool geometry. Another somewhat related study was later conducted by Takeyama and Iijima [4] on unidirectional glass fiber reinforced plastics specimens. They evaluated the machinability of glass fiber reinforced plastics in orthogonal cutting with varied fiber angles. A methodology to estimate the average cutting force and roughness in reference to the fiber angles was also proposed. Hocheng et al. [5] investigated the machinability of unidirectional carbon fiber reinforced plastics. They found that the chip forms observed reveal that the cutting mechanism involves fracture rather than plastic deformation and that the chips produced can be hazardous to humans. They also found that the milled surface parallel to the laminate is quite smooth for different cutting conditions. Their results supported the findings of Koplev [2]. Recently, Hintze et al. [6] investigated the occurrence of delamination of the top layers during the machining of CFRP tape, with the focus being on the contour milling process. The occurrence and propagation of delamination were studied by milling slots in unidirectional CFRP specimens having different fiber orientations. They found that delamination is highly dependent on fiber orientation and on tool sharpness. As mentioned above, the majority of the studies that have been conducted on the machining of CFRP laminates are on CNC milling and drilling; with the exception of a few works [7, 8], the number of studies on the robotic trimming of CFRP are quite limited. Furthermore, the assessment of the machinability of CFRP in high-speed robotic trimming has never been investigated; in fact, this is one of the major issues in robot machining that must be addressed in order to extend it to more applications. Furthermore, in order to assess the robotic trimming accuracy of CFRP, the correlation between robot deflection cutting forces and machining parameters must be established through robust experiments.

3.2.1 Evaluation of the cutting force The development of process control schemes to avoid delamination by controlling and regulating the cutting process parameters require an accurate prediction of the cutting forces [9]. Cutting forces represent an important factor of machinability eval-

86 | 3 High-speed robotic trimming of CFRP uation and their size will directly influence the quality of machined parts. They are related to many factors, such as cutting parameters, workpiece materials and tools. A 3D finite element (FE) model of drilling in CFRP composite laminate has been developed by Phadnis et al [10]. The effects of cutting speed and feed rate on the thrust force and torque was studied using numerical simulations. The FE model was validated using experimental results for a combination of drill feed and cutting speed; it was later used to predict the thrust force and torque for varying process parameters. They concluded that both the thrust force and the torque may be reduced by using a combination of low drill feed rate and high cutting speeds (spindle speed). Rao et al. [11] used the finite element method (FEM) to predict the chip formation mechanism in the orthogonal machining of unidirectional glass fiber reinforced polymer (UD-GFRP) composites. The effect of fiber orientation, tool parameters and operating conditions on fiber and matrix failure and chip size was also investigated. The orthogonal cutting of unidirectional fiber-reinforced polymer composites was analyzed by Arola and Ramulu [12] using the finite element method. Predictions for the cutting forces from numerical simulations were verified with experimental measurements for the orthogonal trimming of unidirectional graphite/epoxy. Mkaddem et al. [13] established another finite element model to investigate the orthogonal machining of Unidirectional Glass Fiber Reinforced Plastics (UD-GFRP). A simulation scheme entailing fiber orientation, depth of cut and tool rake angle was also constructed for investigating the cutting and thrust force developed during machining. Jahromi and Bahr [14] developed a theoretical model using the energy method to predict the machining forces for the orthogonal machining of unidirectional polymer-matrix composites (PMCs) for fiber orientations ranging from 90° to 180°. The validity of the proposed model was verified experimentally using tools with rake angles of 5°, 10°, 15° and 20°. They concluded that their model works well when the fracture plane angle is between 90° and 180°. Haiyan et al. [15] used an analytical cutting force model based on mechanistic modeling techniques to simulate cutting forces in the helical milling of CFRP. In addition, the cutting force coefficients were corrected according to the experimental data and the established model was tested through cutting experiments. They found that the resultant radial and axial cutting forces decrease with an increase in the cutting speed and increase with an increase in the feed rate per tooth and axial feed rate. Kalla et al. [9] developed a methodology that combines the mechanistic modeling techniques from metal machining and neural network approximation in order to obtain a predictive cutting force model for helical end milling of carbon fiber reinforced polymers (CFRP). They concluded that the mechanistic modeling approaches from metal cutting are valid for machining CFRP. Furthermore, model predictions were compared with experimental data and were found to be in good agreement in cutting unidirectional laminate, but less so in the case of a multidirectional laminate. More recently, Karpat and Polat [16] proposed a mechanistic model for milling multidirectional CFRP laminates using double helix milling tools. In the model, cutting and edge coefficients are calculated based on the laminate fiber direction. Issues related to surface quality and tool wear are also

3.2 Machinability of CFRP |

87

investigated. Zaghbani et al. [17] presented a comprehensive analysis of the instantaneous cutting forces at play during the trimming process of unidirectional laminates. They developed an empirical model for cutting forces, using a high mechanistic order model. They found that for machined laminates, the fiber orientation does not significantly influence the profile of the tangential and radial forces, but that it influences their amplitude.

3.2.2 Assessment of the machinability of CFRP under high-speed robotic trimming Because the compliance error is highly dependent on the manipulator configuration during trimming, it is very important to pay particular attention to the selection of the best robot configuration during trimming. Slamani et al [18] developed a robust methodology based on the design of experiments to study the effect of the robot configuration and cutting conditions on the accuracy of CFRP parts during the high-speed robotic trimming of 84 specimens of 33 mm length. Their tests were performed on a 6-axis KUKA KR 500-2 MT industrial robot mounted on a 13-foot linear rail and manipulating a heavy spindle HSD Mechatronic ES 789 delivering spindle speeds of up to 26 000 rpm. Two placements (configurations) of the robot were tested during the analysis. The first placement (Fig. 3.1) was characterized by a relatively stretched configuration and a trimming direction parallel to the linear axis (parallel to the y-axis of the cell). The second placement (Fig. 3.2) was characterized by a relatively folded configuration and a trimming direction perpendicular to the linear axis (parallel to the x-axis of the cell). The distance between the robot base and the tool was 2669 mm for the first placement and 1816 mm for the second placement. They found that the optimal configuration providing the best part accuracy was the second placement (with the relatively folded configuration). The optimum cutting conditions related to this optimal configuration were achieved at a low feed rate (0.2 mm/rev) and a moderate cutting speed (400 m/min). In this chapter, a similar test was performed on the same robot using the second placement to trim 9 CFRP specimens: 12-inch long and using cutting conditions smaller than those used in [18]. Next, the machinability of the CFRP was evaluated via parameters such as form error, dimensional error and surface quality. The laminates used in these machining tests were prepared in a controlled aeronautical environment using pre-impregnated technology. The stacks were autoclavecured and the plies were oriented such as to ensure that the laminate had quasiisotropic properties. The 24-ply laminate was 3.68 mm thick, with a fiber volume fraction of 64 %. Before the first trimming test was started, the laminates were pre-drilled for tightening on a machining fixture, as shown in Fig. 3.3. The aluminum back plating system (Fig. 3.4), which uses 42 screws and a torque wrench to secure the laminate, was designed to trim 5 slots under different cutting conditions.

88 | 3 High-speed robotic trimming of CFRP

Linear axis

Trimming direction

Y X

m

9

6 26

m

Fig. 3.1: First placement, with a relatively stretched configuration.

Y X Trimming direction

Linear axis 1816 mm

Fig. 3.2: Second placement, with a relatively folded configuration.

The subassembly (laminate and back plate) was tightened to a 3-axis Kistler 9255B type dynamometer table. The assembly was subsequently installed on the KUKA DKP-400 2-axis positioning table (Fig. 3.5), situated in the working space of the robot. The positioning table and the linear axis supporting the robot were static during the trimming tests.

3.2 Machinability of CFRP |

89

Fig. 3.3: CFRP laminate, back plate and 3-axis Kistler table installed on 2-axis positioning table KUKA DKP-400.

Clearance holes (reference holes)

42 holes to attach the CFRP plate on the template

Clearance holes (reference holes) Fig. 3.4: Aluminum back plating system.

The tool used in these tests was a 3/8 inch diameter Onsrud PCD end mill with two straight flutes, having a 20° rake angle, a 10° relief angle and a 5 μm cutting edge radius. The cutter was inspected prior to the machining operation. As shown in Tab. 3.1, different combinations of cutting parameters were tested. The cutting depth was supposed to be constant along the cut (3.68 mm). Two laminates were trimmed in this test. The cutting forces were measured during the test in the x, y and z directions with a 3-axis dynamometer table. The cutting force data were then recorded for further analysis and evaluation. After the trimming of the nine specimens was completed, each one was then inspected with a Mitutoyo CRYSTA coordinate measuring machine (Fig. 3.6).

90 | 3 High-speed robotic trimming of CFRP

Z

Y X Fig. 3.5: Photo of 6-axis KUKA KR 500-2 MT industrial robot with a relatively folded configuration.

Tab. 3.1: Cutting conditions of robotic trimming tests. Test

Cutting speed (m/min)

TCP speed (feed rate) (mm/rev)

1 2 3 4 5 6 7 8 9

150 150 150 200 200 200 300 300 300

0.15 0.20 0.25 0.15 0.20 0.25 0.15 0.20 0.25

3.2.3 Cutting forces for robotic trimming experiments Knowing the cutting forces components during high-speed robotic trimming processes is crucial as they are deemed to constitute the most important indicator of machining state and part quality. Figure 3.7 shows the cutting force components produced during the robotic trimming of 12-inch length specimens. Figures 3.8 and 3.9 express the evolution of the feed and normal cutting forces versus the TCP speed and cutting speed during the robotic trimming of 12-inch specimens. From Fig. 3.8, it can be seen that the feed force increases as the cutting speed and the TCP speed increase. This is explained by the fact that when the feed rate increases, the laminate becomes more rupture-resistant and requires greater effort. Hence, the

3.2 Machinability of CFRP |

Y

X

Fig. 3.6: Inspection of trimmed specimens on Mitutoyo CRYSTA coordinate measuring machine.

e rc fo l a rm

No

Axial force

Feed force

Fig. 3.7: Cutting force components during high-speed robotic trimming.

91

92 | 3 High-speed robotic trimming of CFRP

Feed force (N)

–150 –100

–50

0 0.15 0.20 TCP sp 0.25 eed (m m/rev )

300 n) i /m (m d ee 150 sp g n tti Cu 200

Fig. 3.8: Feed force as a function of TCP speed and cutting speed.

cutting force increases as the feed rate increases. The results presented in Fig. 3.9 illustrate the evolution of the normal cutting forces according to the TCP speed and cutting speed. They show that when the TCP speed increases, the normal cutting forces increase and become significant. However, increasing the cutting speed is found to slightly increase the normal cutting forces. The rate of change of the cutting forces with the cutting speed is believed to be associated with the cutting temperatures. Similar results were obtained when machining GFRP [19].

80 60

20

sp ee d( m/ mi n)

40 300

0

200 0.15

0.20 0.25 TCP speed (mm/rev)

150

Cu tti ng

Normal force (N)

100

Fig. 3.9: Normal force as a function of TCP speed and cutting speed.

3.2 Machinability of CFRP |

93

3.2.4 Quality of robotic trimmed specimens During robotic machining, a heavy cut would generally produce more inaccurate components than would a light cut. Furthermore, during robotic machining of carbon fiber reinforced polymer, the anisotropic and highly abrasive nature of CFRP, combined with the higher cutting forces and the lower stiffness of the robot, lead to high levels of vibration. This in turn results in numerous machining problems, such as rapid tool wear, fiber pull-out, fiber fracture, delamination, trajectory deviation, poor quality and in some cases, rejection of machined parts.

3.2.4.1 Form quality Figures 3.10 and 3.11 show the trajectory deviations at a TCP speed of 0.20 mm/rev and cutting speeds of 200 m/min and 300 m/min, respectively, for up-cut milling of two selected specimens of 12 inches in length. Because the industrial robot lacks dynamic stiffness, three different types of mechanical vibration are present during the high-speed robotic machining process, which is comprised of the tool and tool holder, the workpiece and the robot itself. The vibration of the robot itself is the most important source of errors that affect the machined part’s quality. As illustrated in Figs. 3.10 and 3.11, the trajectory deviations are not similar and are affected by the cutting conditions. This is shown by different wavy paths that are clearly visible. Results also show that at high cutting conditions, dynamic errors become a significant source of errors, which affect the path accuracy. This is manifested in high amplitude vibration along the measured path, which strongly affects the accuracy of the trimmed part. This behavior is explained by the variations of the cutting force during machining and the poor rigidity resulting from flexibility in the joints, which induces vibrations in the end-effector. It is important to note that the dynamic performance of an industrial robot is even less homogeneous than its static performance. Obviously, the less the main joints (especially joint 1) are displaced, the better the dynamic performance of the robot. Figure 3.12 shows that the profile deviations vary as a function of the cutting conditions. They increase slightly with an increase in the feed rate and cutting speed. As expected, the cutting speed effect is not as significant as the TCP speed effect, but a higher cutting speed and higher TCP speed seem to result in the worst profile deviation. A better profile is obtained at lower cutting speeds and higher TCP speeds. The results also show that the profile deviations vary from 0.14 mm to 0.24 mm. These results are much better than those obtained in [18]. These improved results are achieved thanks to the appropriate selection of the cutting parameters chosen according to Slamani et al. [18], who proposed the best operational conditions including robot configuration, feed rate and cutting speed ranges for similar tool type and composite materials.

94 | 3 High-speed robotic trimming of CFRP 0.15

Actual profile error

Profile error [mm]

0.1

Desired profile error

0.05 0 –0.05 –0.1 0

50

100 150 200 Trimmed length [mm]

250

300

Fig. 3.10: Profile error at 200 m/min cutting speed and 0.20 mm/rev TCP speed.

0.15

Profile error [mm]

0.1 0.05 0 –0.05 –0.1 0 0

50

100 150 200 Trimmed length [mm]

250

300

Profile deviation (mm)

Fig. 3.11: Profile error at 300 m/min cutting speed and 0.20 mm/rev TCP speed.

0.25 0.2 0.15 0.1 0.05 0 300 200 150 Cutting speed (m /min)

0.25 ) 0.20 rev m/ 0.15 m ( ed spe TCP

Fig. 3.12: Profile errors as a function of the cutting speed and TCP speed.

3.2 Machinability of CFRP |

95

Dimensional error (mm)

3.2.4.2 Size quality Figure 3.13 presents the dimensional errors as a function of the cutting speed and feed rate. We see from this figure that the dimensional errors are strongly affected by the TCP speed, increasing when the TCP speed increases. The results also show that the dimensional errors are sensitive to the cutting speed variation, but no clear trend is observed. A numerical evaluation shows that the dimensional errors range from 0.084 mm to 0.245 mm. Finally, the best dimensional error was obtained at a low TCP (0.15 mm/rev) speed and moderate cutting speed (200 m/min).

0.25 0.2 0.15 0.1 0.05 0 300 200 150 Cutting sp eed (m/m in)

0.25 ) 0.20 /rev m 0.15 (m ed e sp TCP

Fig. 3.13: Dimensional errors as a function of the cutting speed and TCP speed.

3.2.5 Surface quality Over the past decade, many studies have considered the roughness of machined surfaces of fiber reinforced plastics, with a view to predicting the roughness from machining conditions. Boudelier et al. [20] proposed a methodology to optimize process parameters for trimming applications. Their methodology was then applied in the composite trimming process, with a diamond abrasive cutter. They found that grit size and feed per revolution are key factors in ensuring required surface roughness and improving productivity, respectively. Davim et al. [21] referred to the ANOVA method and the design of experiments for evaluating the surface roughness parameters Ra with respect to the milling conditions of glass fiber reinforced plastic (GFRP). They selected two cutting parameters as variables: feed rate and cutting speed. Their studies led to the conclusion that the roughness Ra increases with an increased feed rate and decreases with an increased cutting velocity. As well, Davim and Reis [22] conducted a similar study and arrived at the same conclusions regarding the machining of carbon fibers. Similar results were also found by Palanikumar [23], who used the Taguchi method and Pareto ANOVA analysis to study the average roughness parameter Ra during the turning of GFRP composites. An approach for optimizing the ma-

96 | 3 High-speed robotic trimming of CFRP chining parameters in the milling of glass fiber reinforced plastic (GFRP) composites was proposed by Jenarthanan and Jeyapaul [24]. They conducted milling experiments based on the Taguchi technique, using a solid carbide cutting tool. The machining parameters, such as spindle speed, feed rate, helix angle and fiber orientation angle were optimized by multi-response considerations, namely, surface roughness, delamination factor and machining force. They found that a lower fiber orientation angle, a lower helix angle, a moderate spindle speed and a lower feed rate constitute the ideal conditions for machining GFRP composite plates. Slamani et al. [25] studied the machinability of autoclave-cured 24-ply CFRP laminate under varying cutting conditions using a CVD diamond-coated carbide tool with six straight flutes. They found that the optimum cutting conditions are achieved at lower feed rates and higher cutting speeds. Palanikumar et al. [26] studied the influence of cutting conditions on surface roughness parameters in the turning of glass fiber reinforced composite materials. They found that the surface roughness increases with an increase in the feed rate and decreases with an increase in the cutting speed. They also developed empirical models to correlate the machining parameters with surface roughness. More recently, Zhang et al. [27] studied the effect of cutting parameters, such as cutting velocity, cutting depth, cutting width, as well as feed rate, on the surface roughness of CFRP, using PCD tools. Unlike what was found in previous results, they found that the cutting depth has the most significant effect on surface roughness, followed by the cutting velocity and the cutting width, while the feed rate has little effect. Palanikumar [28] built a prediction model based on the fuzzy modeling of Ra and Rt . The same method was used by Rajasekaran et al. [29] for the turning of carbon/epoxy composites. For both studies, the authors neglected the effect of fiber orientation. However, other studies have considered the ply orientation as a variable in their analysis of the surface roughness. Eriksen [30] found that surface roughness was independent of fiber orientation, but only for short fibers. In another study, a mathematical model with regression analysis and ANOVA was realized by Palanikumar and Davim [31], leading to the conclusion that a low fiber orientation angle generates a better surface finish. This effect was confirmed by Sarma et al. [32] in a study in which the roughness parameter Ra was evaluated in a second-order model, based on four machining parameters. The roughness was measured with a vision system using a digital camera, in order to avoid contact with the machined surface. Their studies were conducted on GFRP and did not take into account any negative fiber angles. However, Jahromi et al. [33] studied the effects of all fiber orientations on surface damage occurring during the machining of unidirectional composites. They concluded that the worst case lies between an 80° and a 135° fiber angle and the best case at 0° or 180°. Chatelain et al. [34] used a PCD tool comprised of two straight flutes to trim 32-ply carbon fiber laminates. They found that the fiber angle is an important parameter affecting the roughness profile. They also showed that each ply orientation has its own “typical” profile, regardless of machining conditions. In addition, they showed that the cutting speed effect was not as significant as the

3.2 Machinability of CFRP |

97

feed rate effect on surface roughness, but that a higher cutting speed leads to better surface finishes in most cases. Regarding the effects of cutting parameters on the surface roughness, most studies conclude that the surface quality is improved with a low feed rate, high cutting speeds and low depths of cut. The feed rate appears to be the most significant factor of influence on the surface finish, followed by the cutting speed, while the depth of cut has less of an effect than the two others. In general, the authors base their analyses on statistical indicators such as Ra , Rt or Rz , which are efficient for isotropic metallic materials. In the case of metallic materials, the probe location has a low impact on the measurement results. That is not the case for composite materials, where the location of the probe with respect to the ply orientation can significantly affect the measurements. It is therefore very important to pay particular attention to the probe location during a surface roughness measurement of composite materials. Furthermore, surface roughness is considered as the quality index of the machined parts. Unstable cutting conditions during high-speed robotic trimming cause many effects on the surface quality of finished products. Therefore, assessing surface roughness based on machining parameters is a crucial exercise for cost-effective operations.

3.2.5.1 Surface quality of robotic trimmed specimens Characterizing the surface quality of trimmed parts is quite complex. This is due to various factors affecting the surface roughness during the high-speed robotic machining process. Figures 3.14 and 3.15 respectively show the surface profiles for +45° and −45° ply orientations obtained in previous experiments under the same cutting conditions. We can observe that the fluctuation in −45° ply orientations is more significant compared with that of +45° ply orientations. Poor surface quality was seen at the −45° ply orientation, where the surface roughness was irregular and had high protrusions (peaks) and very deep valleys. This could be explained by a fiber bending which occurs during machining and that would be different for each ply orientation. The results in Figs. 3.14 and 3.15 also show that the difference in magnitude between the +45° to −45° ply orientations is clearly visible and is more than 4 times higher in the case of a −45° ply orientation. Figure 3.16 shows the effect of cutting speed on surface roughness (Ra ) with different feed (TCP speed) values. We can observe that the surface roughness increases as the value of the cutting speed increases. From Fig. 3.17, it can be noticed that an increase in feed increases the surface roughness. Therefore, it can be concluded from the experimental analysis that better surface characteristics can be obtained by using lower feeds in combination with lower cutting speeds.

98 | 3 High-speed robotic trimming of CFRP 6 4

Height [μm]

2 0 –2 –4 –6 0

0.5

1

1.5 2 2.5 Distance [mm]

3

3.5

4

4.5

5

Fig. 3.14: Roughness profile for +45° ply orientation, at 0.2540 mm/rev feed and 400 m/min cutting speed.

20

Height [μm]

10 0 –10 –20 –30 0

0.5

1

1.5 2 2.5 Distance [mm]

3

3.5

4

4.5

5

Fig. 3.15: Roughness profile for −45° ply orientation, at 0.2540 mm/rev feed and 400 m/min cutting speed

0.2540 mm/rev 0.3048 mm/rev 0.3556 mm/rev 0.4064 mm/rev

1.1

Ra [μm]

1 0.9 0.8 0.7 0.6 0.5 200

400 Cutting speed [m/min]

650

Fig. 3.16: Variation of surface roughness as a function of cutting speed.

3.3 Conclusion

| 99

200 m/min 400 m/min 650 m/min

1.1

Ra [μm]

1 0.9 0.8 0.7 0.6 0.5 0.254

0.3048 0.3556 0.4064 Feed rate [mm/rev]

0.4572

Fig. 3.17: Variation of surface roughness as a function of feed rate.

3.3 Conclusion High-speed trimming with industrial robots is one of the modern technologies that, in comparison to CNC machining, improve efficiency and decreases costs and machining time. This chapter presents the findings of an experimental investigation of the effect of feed rate and cutting speed on cutting forces and part quality during high-speed trimming of CFRP. The quality of the machined part was evaluated via parameters such as form error, dimensional error and surface quality. Based on the experimental results, the following observations were made: – The TCP speed strongly affects the accuracy of the machining process. – Results showed that the feed force increased with an increase in the TCP speed and cutting speed. The normal forces proved to be less sensitive to the cutting speed, but increased as the TCP speed increased. – Results also showed that a high-speed robotic trimming process produces significant profile deviations; they increase slightly with an increase in the TCP speed and cutting speed. – Dimensional errors are more severely affected by the machining parameter; they increase with an increase in the TCP speed and cutting speed. – The experimental results show a significant correlation between the surface quality and the ply orientation. The fact that the −45° ply orientation clearly represents the worst case should be noted and is probably due to fiber bending during machining. – An analysis of roughness errors in terms of the machining parameters leads to the conclusion that a lower TCP speed and lower cutting speed give a better surface quality.

100 | 3 High-speed robotic trimming of CFRP

References [1] [2] [3] [4] [5] [6]

[7] [8]

[9] [10]

[11] [12] [13]

[14] [15]

[16]

[17]

[18] [19] [20]

Everstine G, Rogers T. A theory of chip formation of FRP composite materials. Journal of Computational Materials Science 1971;5:94–106. Koplev A. Cutting of CFRP with single edge tools. Proceedings of the third international conference on composite materials. Paris, 1980;2:1597–1605. Koplev A, Lystrup A, Vorm T. The cutting process, chips and cutting forces in machining CFRP. Composites 1983;14:371–376. Takeyama H, Iijima N. Machinability of glass fiber reinforced plastics and application of ultrasonic machining. Annals of CIRP 1988;37(1):93–96. Hocheng H, Puw H, Huang Y. Preliminary study on milling of unidirectional carbon-fibre reinforced plastics. Composites Manufacturing 1993;4(2):103–108. Hintze W, Hartmann D, Schutte C. Occurrence and propagation of delamination during the machining of carbon fibre reinforced plastics (CFRPs): an experimental study, Composites Science and Technology 2011;71(15):1719–1726. Dumas C, Boudelier A, Caro S, Garnier S, Ritou M, Furet B. Development of a robotic cell for trimming of composite parts, Mechanics & Industry 2011;12:487–494. Slamani M, Gauthier S, Chatelain J-F. A study of the combined effects of machining parameters on cutting force components during high speed robotic trimming of CFRPs. Measurement 2015; 59:268–283. Kalla D, Sheikh-Ahmad J, Twomey J. Prediction of Cutting Forces in Helical End Milling Fiber Reinforced Polymers. Int. J. Mach. Tools Manuf 2010;50:882–891. Phadnis AV, Roy A, Silberschmidt VV. Finite element analysis of drilling in carbon fiber reinforced polymer composites. Journal of Physics, Conference Series 2012;382:doi:10.1088/ 1742-6596/382/1/012014. Rao GVG, Mahajan P, Bhatnagar N. Machining of UD-GFRP composites chip formation mechanism, Compos Sci Technol 2007;67(11–12):2271–2281. Arola D, Ramulu M. Orthogonal cutting of fiber-reinforced composites: a finite element analysis, Int J Mech Sci 1997;39(5):597–613. Mkaddem A, Demirci I, Mansori ME. A micro-macro combined approach using FEM for modelling of machining of FRP composites: cutting forces analysis, Compos Sci Technol 2008; 68(15–16):3123–3127. Jahromi, AS, Bahr B. An analytical method for predicting cutting forces in orthogonal machining of unidirectional composites. Composite Science and Technology 2010;70:2290–2297. Haiyan W, Xuda Q, Hao L, Chengzu R. Analysis of cutting forces in helical milling of carbon fiber-reinforced plastics. Proc. Inst. Mech. Eng. Part B-J. Eng. Manuf 2012;doi:10.1177/ 0954405412464328. Karpat Y, Polat N. Mechanistic force modeling for milling of carbon fiber reinforced polymers with double helix tools. CIRP Ann-Manuf. Technol 2013;http://dx.doi.org/10.1016/ j.cirp.2013.03.105. Zaghbani I, Chatelain J-F, Berube S, Songmene V, Lance J. Analysis and modelling of cutting forces during the trimming of unidirectional CFRP composite laminates. Int. J. Machining and Machinability of Materials (IJMMM) 2012;12(4):337–357. Slamani M, Gauthier S, Chatelain J-F. Analysis of trajectory deviation during high speed robotic trimming of carbon-fibre reinforced polymers, Robot. Comput. Int. Manuf 2014;30 (5):546–555. Sheikh-Ahmad JY. Machining of Polymer Composites, Springer Science+Business Media, LLC, DOI:10.1007/978-0-387-68619-6. Boudelier A, Ritou M, Garnier S, Furet B. Optimization of process parameters in CFRP machining with diamond abrasive cutters. Advanced Materials Research 2011;223:774–783.

References |

101

[21] Davim JP, Reis P, Antonio CC. A Study on Milling of Glass Fiber Reinforced Plastics Manufactured by Hand-lay up Using Statistical Analysis (ANOVA). Composite Structures 2004;64(3): 493–500. [22] Davim P, Reis P. Damage and dimensional precision on milling carbon fiber-reinforced plastics using design experiments. Journal of Materials Processing Technology 2005;160(2):160–167. [23] Palanikumar K . Cutting Parameters Optimization for Surface Roughness in Machining of GFRP Composites using Taguchi’s Method. Journal of Reinforced Plastics and Composites 2006; 25(16):1739–1751. [24] Jenarthanan MP, Jeyapaul R. Optimisation of machining parameters on milling of GFRP composites by desirability function analysis using Taguchi method, International Journal of Engineering, Science and Technology 2013;5(4):23–36. [25] Slamani M, Chatelain JF, Hamedanianpour H. Comparison of two models for predicting tool wear and cutting force components during high speed trimming of CFRP. Journal of Material Forming 2014;DOI: 10.1007/s12289-014-1170-2. [26] Palanikumar K, Mata F, Davim JP. Analysis of surface roughness parameters in turning of FRP tubes by PCD tool. Journal of materials processing technology 2008;204:469–474. [27] Zhang YD, Xiao HH, Teng Y. Study of Surface Roughness in Milling Carbon Fiber Reinforced Plastics Using PCD Tool. Key Engineering Materials 2013;531–532:63–66. [28] Palanikumar K. Surface Roughness Model for Machining Glass Fiber Reinforced Plastics by PCD Tool using Fuzzy Logics. Journal of Reinforced Plastics and Composites 2009;28(18): 2273–2286. [29] Rajasekaran T, Vinayagam, BK, Palanikumar K, Prakash S. Influence of Machining Parameters on Surface Roughness and Material Removal Rate in Machining Carbon Fiber Reinforced Polymer Material. Frontiers in Automobile and Mechanical Engineering (FAME) 2010:75–80. [30] Eriksen E. Influence from Production Parameters on the Surface Roughness of a Machined Short Fiber Reinforced Thermoplastic. International Journal of Machine Tools and Manufacture 1999;39:1611–1618. [31] Palanikumar K, Davim JP. Mathematical model to predict tool wear on the machining of glass fibre reinforced plastic composites. Materials & Design 2007;28(7):2008–2014. [32] Sarma PMMS., Karunamoorthy L, Palanikumar K. Surface roughness parameters evaluation in machining GFRP composites by PCD tool using digital image processing. Journal of Reinforced Plastics and Composites 2009;28(13):1567–1585. [33] Jahromi AS, Gudimani G, Kalla DK, Bahr B. Effect of High RPM Machining and Fiber Orientation On Subsurface Damage in Machining of Unidirectional Composites. SAMPE 2011 State of the Industry: Advanced Materials, Applications, and Processing Technologies, Long Beach, CA. [34] Chatelain J-F, Zaghbani I, Monier J. Effect of Ply Orientation on Roughness for the Trimming Process of CFRP Laminates. World Academy of Science, Engineering and Technology 2012;68: 1204–1210.

H. Miguélez, N. Feito, C. Santiuste, J. Díaz-Álvarez, M. Rodríguez-Millán, X. Soldani

4 Numerical modeling of LFRP machining Abstract: This chapter presents some considerations for modeling long fiber reinforced polymer (LFRP) composites machining. The composite components are usually made to the desired product’s final size. However, some machining operations are required to achieve dimensional tolerance and assembly requirements. LFRP composites are considered difficult to cut due to the presence of hard fibers, being especially vulnerable to machining-induced damage. Numerical modeling of machining can help in analyzing the process; however, it is a relatively new approach to the study of composite cutting. In this chapter, main contributions dealing with LFRP modeling of cutting are summarized. Recent contributions by authors in this field are included, giving the reader information on the development of numerical models that cover mechanical and thermal aspects of different cutting processes.

4.1 Introduction Long fiber reinforced polymer (LFRP) composites have been extensively used in structural components. Their attractive properties – fatigue and corrosion resistance combined with light weight, high specific stiffness and strength – made this family of materials suitable for a wide range of applications in aeronautical, automotive, marine and sporting industries. The problems associated with precision and efficiency in cutting LFRP composites have become important issues in the manufacturing field. Although the components are usually made to the final size of the desired product, machining processes (mainly trimming, milling and drilling) are needed to achieve dimensional tolerance and assembly requirements. LFRP composites are considered difficult to cut materials due to the presence of hard fibers, being especially vulnerable to damage, mainly delamination, fiber pull-out and matrix thermal degradation [1]. Machining-induced damage has significant importance in industrial applications; for instance, poor hole quality in composite drilling accounts for an estimated 60 % of all parts rejection. The workpiece’s surface integrity is critical for the subsequent assembly stage and, of course, during the service life of the component. In fact, recent calls for international projects deal with this unsolved problem: drilling holes in high responsibility composite components, hole quality improvement, non-destructive inspection techniques and other aspects showing the importance of high value machining operations previous to final assembly of the components.

104 | 4 Numerical modeling of LFRP machining Experimental investigation of the machining of low machinability materials, such as LFRP composites, is time-consuming and expensive. Moreover, health hazards associated with fiber inhalation and skin contact reduces the possibility of carrying out extensive experimental research on LFRP composites machining. On the other hand, the shape of the composite, usually thin-walled, also complicates the execution of the tests, which are commonly performed at low cutting speeds achieved after a complex machine-tool positioning process (see for instance the works [2–4] focused on orthogonal cutting of Glass and Carbon LFRP composites). On the other hand, finite element (FE) is a powerful method that offers the possibility to complete experimental studies while avoiding technical problems and elevated cost. Another advantage is the possibility to uncouple influencing parameters in problems depending on a large number of variables such as machining processes. The interest of modeling composite machining has motivated the development of a recent review with an overview of machining models of all composite types [5]. Compared to the large number of scientific works dealing with metal cutting, a limited amount of technical literature exists on FE analysis of cutting LFRP composites, mainly focusing on orthogonal cutting. Although orthogonal cutting is not an industrial process, it is the objective of numerical analysis due to its simplicity, even in the case of metal cutting [6, 7]. Some examples of the modeling strategies and damage criteria used in scientific literature for orthogonal cutting of composites can be found in [8–11]. Most of the literature cited in these works is based on two-dimensional (2D) modeling of orthogonal cutting and assume the hypothesis of plane stress. Cutting forces and intra-laminar damage can be predicted using this simple approach. However, this technique is not suitable to predict delamination, one of the most important defects induced during machining and related to out-of-plane stresses. On the other hand, only unidirectional laminate can be modeled, while quasi-isotropic laminates are commonly used in structural applications due to their higher performance. Recently, 3D modeling of orthogonal cutting has been developed in order to simulate delamination phenomena and reproduce different stacking sequences [12, 13]. Simplified quasi-static three-dimensional (3D) models of drilling, which consider the drill as a punch for delamination prediction, are also found in scientific literature. For instance, the approach used by Durao et al. in [14, 15] and Singh et al. [16] when studying glass fiber reinforced polymer composites (GFRP) drilling: they showed the influence of the drill point angle in the damage induced. In addition, analytical efforts have been done with the objective of relating delamination with the applied trust force [17] that seems to be one of the most influential factors in the generation of out-of-plane failure during drilling [18]. As far as the simulation of the complete process of chip removal is concerned in the case of real industrial machining processes of composites (mostly drilling and milling of the contour) that involve complex tool geometries and oblique cutting, only simplified modeling tools have been developed up to date. Modeling drilling processes

4.2 Orthogonal cutting | 105

is very difficult, because of the need to simulate drill rotation and feed using both damage and erosion criteria at the elements leading to high computational cost [19]. The objective of this chapter is to give an overview of work done in the field of LFRP machining including recent contributions by the authors. This chapter is structured in several sections covering main contributions in modeling different LFRP cutting processes. First, modeling orthogonal cutting is described, including both 2D and 3D approaches. Thermal damage is also commented on. Drilling is included in the following section. Finally, conclusions of the chapter are presented.

4.2 Orthogonal cutting The analysis of orthogonal cutting has been focused on different aspects of the process. The use of 2D models limits the approach, since unidirectional composites should be assumed and inter-laminar damage cannot be modeled. The 3D approach allows the analysis of multi-directional composites and the study of delamination. Mechanical damage is significant; however, it is not the sole problem when machining composites. The temperature around the machined zone should be controlled in order to avoid matrix degradation. In this section, 2D, 3D and thermal effects when modeling orthogonal cutting are commented on.

4.2.1 2D modeling Macro-mechanical approaches, modeling the composite as an anisotropic homogeneous material, are the mostly common methodology in the literature. In general, unidirectional fiber orientation was assumed in two-dimensional approaches. Arola and Ramulu [8] presented the first study of orthogonal cutting of unidirectional composites based on a finite element. Ramesh et al. [20] studied the influence of orientation for four different FRP. Failure mechanisms based on fiber–matrix cracking were implemented; the conclusion was that the load needed to induce failure was dependent on fiber orientation. Mahdi and Zhang [21, 22] used both 2D and 3D approaches to simulate composite cutting. 2D analysis reproduced an equivalent homogeneous material predicting cutting forces of FRP as a relation to fiber orientation. 3D approaches simulated a composite cell based on different perfectly bonded constituents (fiber and matrix). Arola et al. [23] focused on the influence of tool geometry on cutting forces and sub-surface damage for orthogonal machining of unidirectional FRP. It is generally shown in these studies that good approximation to experimental cutting forces has been obtained while the thrust forces were poorly predicted. The same problem has been observed in numerical simulations of metal cutting due to the difficulty of reproducing the complex interaction between the tool and workpiece [24].

106 | 4 Numerical modeling of LFRP machining Structural applications for aeronautical components need excellent mechanical properties usually achieved with CFRP composites. The enhanced mechanical properties that characterize carbon fiber influences chip generation and induced sub-surface damage. The authors developed a numerical analysis comparing the behavior of GFRP and CFRP [11]. A plane stress model was developed using the commercial finite element code ABAQUS/explicit. A dynamic explicit analysis was carried out with: plane stress; quadrilateral, linearly interpolated elements; reduced integration and automatic hourglass control (CP4SR in ABAQUS/explicit notation [25]). Geometry and boundary conditions of the numerical model are shown in Fig. 4.1. The values for cutting parameters and tool geometry (depth of cut, cutting speed, rake angle, clearance angle, and edge radius) were defined in coherence with those used in [3] in order to validate predicted numerical results. The aim of the model was to analyze not only chip initiation phenomena, but also the evolution of the cutting process. Thus, a cutting length equal to 2 mm was considered – enough to reach steady state conditions during cutting process. The model was simply based on a 2D approach and its details can be found in [11]. Rake angle=5° α Cutting speed

Tool

H=1 mm

Feed=0.2 mm

R=0.5 mm Clearance angle=6° β

UX =0

Workpiece UX =0

Fiber orientation Ɵ

UX =Uy =Rz =0 L=3 mm

5 μm

Fig. 4.1: Model geometry, boundary conditions and detail of mesh around the tool tip [11].

The damage initiation criteria for fiber reinforced composite were based on the Hashin theory [26, 27] that includes four failure models (fiber tensile/compressive and matrix cracking/crushing). The formulation is showed in Tab. 4.1 where where σ11 denotes the stress in fiber direction, σ22 the stress in transverse direction, τ12 the in-plane shear stress, XT the longuitudinal tensile strength, XC the longitudinal compressive

4.2 Orthogonal cutting | 107

Tab. 4.1: Hashin model formulation. Failure mode

Hashin formulation

Fiber tension (σ11 > 0)

dft = (

Fiber compresion (σ11 < 0) Matrix cracking (σ22 > 0) Matrix crushing (σ22 < 0)

2 σ11 2 τ ) + α ( 12 ) ≤ 1 XT SL σ11 2 c df = ( ) ≤1 XC 2 2 σ τ t dm = ( 22 ) + α ( 12 ) ≤ 1 YT SL 2 2 2 Y σ σ τ c dm = ( 22 ) + [( C ) − 1] 22 + ( 12 ) ≤ 1 2ST 2ST YC SL

strength, YT the transverse tensile strength, YC the transverse compressive strength, SL the longitudinal shear strength and ST the transverse shear strength. Failure occurs when dij reaches the value 1. The characteristics of the materials used in the simulations are summarized in Tab. 4.2. Main conclusions obtained from the study are summarized in the following. CFRP were considered more brittle composite materials while GFRP were catalogued as more ductile composite materials. Numerical parameters representing the degree of ductility exhibited by the composite during cutting was the energy needed to complete breakage of the elements. This fact was coherent with the behavior observed when testing these two types of composites under dynamic loading. Tab. 4.2: Characteristics of materials used in simulations [11]. Mechanical properties

Glass FRP

Carbon FRP

Longitudinal modulus, E1 (GPA) Transverse modulus, E2 (GPa) In-Plane shear modulus, G12 (GPa) Major Poisson’s ratio, 𝜈12 Longitudinal tensile strength, XT (MPa) Longitudinal compressive strength, XC (MPa) Transverse tensile strength, YT (MPa) Transverse compressive strength YC (MPa) Shear strength, S (MPa)

48 12 6 0.28 1200 800 59 128 25

126 11 6.6 0.28 1950 1480 48 200 79

Machining induced sub-surface damage experienced by CFRP was far more reduced than that observed in GFRP (see Fig. 4.2). This behavior could also be related to the energy absorbed up to element breakage. GFRP exhibited large deformation before breakage and extended damage beneath and ahead of the tool tip. However, in the case of CFRP, the damage was located in the zone of the uncut chip in front of the interface, as this zone was removed with the chip formation. The chip formation mech-

108 | 4 Numerical modeling of LFRP machining t = 3, 10 –3 s

t = 6, 10 –3 s

Glass fiber

t = 1, 10 –3 s

HSNMCCRT (Avg: 75%)

Carbon fiber

+1.000e+00 +9.167e–01 +8.333e–01 +7.500e–01 +6.667e–01 +5.833e–01 +5.000e–01 +4.167e–01 +3.333e–01 +2.500e–01 +1.667e–01 +8.333e–02 +0.000e+00

(a)

Carbon fiber

Glass fiber

t = 1, 10 –3 s

t = 3, 10 –3 s

t = 6, 10 –3 s

HSNMTCRT (Avg: 75%) +1.000e+00 +9.167e–01 +8.333e–01 +7.500e–01 +6.667e–01 +5.833e–01 +5.000e–01 +4.167e–01 +3.333e–01 +2.500e–01 +1.667e–01 +8.333e–02 +0.000e+00

(b) Fig. 4.2: Evolution of matrix damage – (a) crushing and (b) cracking – during chip formation: comparison between CFRP and GFRP, fiber orientation 45° [11].

anisms and induced damage strongly depends on fiber orientation. The effect of the bending moment (involving both cracking and crushing damage) was observed for fiber orientations close to the cutting speed direction in CFRP. The increment of angle orientation led to increased crushing damage beneath the machined surface. As it was demonstrated, material parameters clearly influence the results obtained during cutting simulation. The influence of the model’s numerical parameters (such as mesh size and shape) has been extensively analyzed in different works focusing on metal cutting (see for instance [28–30]). Although macroscopic results such

4.2 Orthogonal cutting

| 109

as cutting forces are not considerably affected by these parameters, the local results, which depend on localization phenomena, are strongly dependent on mesh size and shape. Despite the interest in properly defining the size and mesh of the numerical model, it is not easy to found works analyzing the effect of numerical parameters when simulating composite cutting. The influence of the mesh size and orientation were recently studied in [31] using the 2D approach described previously. The authors analyzed the influence of the numerical parameters involved in the modeling process in cutting forces, chip morphology and damage. In addition, the influence of the level of energy needed to reach the element’s complete breakage was studied. The statement of this level of energy is crucial to simulate the material behavior: when a low level of energy is implemented, the element is eroded just after the onset of damage, while a high level of energy allows for a high deformation of the element before total breakage. Thus it is possible to distinguish between more ductile composite materials showing progressive failure; and more brittle composite materials, presented sudden breakage, not only in machining processes but also when experiencing other dynamic loads. Although it could not be considered a purely numerical parameter because it is representative of mechanical composite behavior, it is difficult to find precise information about its value. Therefore a parametric study of the influence of this parameter is very interesting for future research. The main conclusions obtained from the work in reference [31] are summarized below. Cutting and thrust forces were found to be robust output variables that can be considered independent of the numerical parameters studied in the paper. The influence of rake angle and cutting edge radius were also analyzed. Lower values of rake angle made the cutting process more difficult; this led to enhanced force values. The increment of the cutting edge radius also produced an increment of the forces. In fact, the rounded cutting edge could be interpreted as a decreased effective rake angle. Chip morphology was strongly influenced by the level of energy needed to reach complete element breakage (see Fig. 4.3). Moreover, chip morphology was also dependent on geometrical tool parameters. The rake angle interacted with the fiber orientation, resulting in a modified inclination relative to the rake surface. The rounded cutting edge had a slight effect on chip morphology. Damage was the most sensible factor. Not only mesh size, shape and orientation also influenced the predicted fields of damage. The damage distribution also changed significantly with the variation of the energy level needed for breakage: the lower was the energy, the more brittle did the composite behave. In consequence, the damage tended to be reduced to a small area surrounding the tool and beneath the machined surface (Fig. 4.3).

110 | 4 Numerical modeling of LFRP machining

Energy = 400

Energy = 600

Energy variable

Matrix crushing

Matrix cracking

Energy = 200

HSNMTCRT (Avg: 75%) +1.000e+00 +9.167e–01 +8.334e–01 +7.500e–01 +6.667e–01 +5.834e–01 +5.001e–01 +4.168e–01 +3.334e–01 +2.501e–01 +1.668e–01 +8.347e–02 +1.454e–04 HSNMCCRT (Avg: 75%) +1.000e+00 +9.167e–01 +8.333e–01 +7.500e–01 +6.667e–01 +5.833e–01 +5.000e–01 +4.167e–01 +3.333e–01 +2.500e–01 +1.667e–01 +8.333e–02 +1.321e–08

Fig. 4.3: Influence of energy for element breakage in matrix damage [31].

4.2.2 3D modeling The development of 3D models is required because only unidirectional laminates can be modeled using a 2D approach, while quasi-isotropic laminates are commonly used in structural applications due to their higher performance. Moreover, 2D modeling is not suitable for predicting delamination, one of the most important defects induced during machining and related to out-of-plane stresses. The interest of improving the depth of knowledge concerning composite cutting motivated a recent work by the authors [12]; it focused on the analysis of out-of-plane failure in orthogonal composite cutting. A three-dimensional finite element (FE) model for orthogonal composite cutting was developed in order to evaluate the validity of the assumptions involved in the formulation of 2D approaches and when analyzing the influence of a stacking sequence. Results provided by 3D and 2D analysis were compared. The 2D model was similar to the one previously described (Fig. 4.1). 3D dynamic explicit analysis was carried out using 8-node brick elements with reduced integration (C3D8R in ABAQUS/explicit notation [25]). A scheme for the numerical model showing geometry and boundary conditions is presented in Fig. 4.4. The geometric characteristics of the model (tool geometry, feed and cutting length) were the same as those used in a 2D approach. The laminate thickness was stated to be 0.1, 0.4 and 0.8 mm in order to analyze the influence of this parameter. Mechanical behavior of LFRP was implemented in a VUMAT subroutine, where damage was predicted using a Hou formulation [32] that included four failure criteria (fiber failure, matrix cracking, matrix crushing and delamination). Under a given load, the stresses at each integration point were computed in the user sub-routine. Then, each failure

4.2 Orthogonal cutting | 111

Cutting speed Laminate width (0.1–0.8)

Tool

Feed Workpiece Ux=0

Ux=0 Fiber orientation Ɵ Ux=Uy=Rz=0

Fig. 4.4: Scheme of 3D model [12].

criteria was computed as a function of stresses and if any failure occured, the material properties at that point were degraded according to the mode of failure. When the failure criterion was reached, the stresses in the damaged area were reduced to reproduce the properties degradation. The Hou model [32] presented in Tab. 4.3 consists of three failure criteria that consider out-of-plane stresses. For the fiber failure criterion, Hou proposed the same formulation under tensile and compressive loading, including longitudinal shear stresses σ12 and σ13 . For the matrix cracking criterion, Hou included the transverse shear stress σ23 . Hou also introduced a delamination criterion for tensile out-of-plane stresses (σ33 > 0). Parameters in Tab. 4.3 are the following: σ11 , σ22 and σ33 are the stresses in longitudinal, transverse and through-the-thickness direction, respectively; σ12 , σ23 and σ31 are the shear stresses; XT and XC are the tensile and compressive strengths in longituTab. 4.3: Hou damage criteria. Failure mode Fiber tension Fiber compression Matrix cracking Matrix crushing Delamination

Hou formulation σ2 + σ2 σ11 2 ) + ( 12 2 13 ) XT SL 2 2 σ12 + σ13 σ11 2 2 ) +( ) dfc = ( 2 XC SL 2 2 2 σ σ σ 2 dmt = ( 22 ) + ( 12 ) + ( 23 ) YT SL ST 2 2 Y σ σ σ σ 1 2 dmc = ( 22 ) + ( C 222 ) 22 + ( 12 ) 4 ST YC SL 4ST σ33 2 σ23 2 σ13 2 2 ddel = ( ) +( ) +( ) ZT ST SL df2t = (

112 | 4 Numerical modeling of LFRP machining dinal direction; YT and YC are the tensile and compressive strengths in the transverse direction; ZT is the tensile strength through-thickness direction; SL is the longitudinal shear strength; ST is the transverse shear strength (failure occurs when di j reaches the value 1). Material properties used in the simulations are provided in the Tab. 4.4. Tab. 4.4: Material properties [12]. Property of carbon epoxy T300/914

Value

Longitudinal modulus, E1 (GPa) Transverse modulus , E2 (GPa) In-plane shear modulus, G12 (GPa) Major Poisson’s ratio, 𝜈12 Through thickness Poisson’s ratio, 𝜈23 Longitudinal tensile strength, XT (MPa) Longitudinal compressive strength, XC (MPa) Transverse tensile strength, YT (MPa) Transverse compressive strength, YC (MPa) In-plane shear strength, S12 (MPa) Longitudinal tensile strength, ZT (MPa) Longitudinal shear strength, SL (MPa) Transverse shear strength, ST (MPa)

136.6 9.6 5.2 0.29 0.4 1500 900 27 200 80 27 80 60

Although cohesive elements are suitable for delamination prediction, they were not used in this work in order to avoid increasing the complexity of the model: smaller elements would have been required to simulate the interface; this would have led to and increase in computational cost. Several conclusions were derived from the work. 2D simulations assuming a plane stress hypothesis mainly reproduced the phenomena experienced by the external plies of the laminate. 3D simulations showed that the thinner the unidirectional HSNMCCRT (Avg: 75%) +1,000e+00 +9,167e–01 +8,111e–01 +7,500e–01 +6,667e–01 +5,811e–01 +5,000e–01 +4,167e–01 +1,111e–01 +2,500e–01 +1,667e–01 +8,111e–02 +0,000e+00

2D model

3D model – external ply

3D model – internal ply

Fig. 4.5: Damage model (matrix crushing) for fiber orientation 45° and 2D vs. 3D (thickness of 3D laminate t = 0.8 mm) [12].

4.2 Orthogonal cutting | 113

[45/ –45/0/90]s

45º

[90/0/45/ –45]s

laminate, the more similar the 3D and 2D results (see Fig. 4.5). These observations indicate the importance of considering out-of plane effects during cutting and should be taken into account when analyzing results obtained with 2D modeling. The 3D approach was applied to the simulation of quasi-isotropic laminate, and the influence of the stacking sequence on the development of damage was demonstrated. When the plies are located at the free surface, they tend to develop larger damaged zones than when they are located at the inner zone of the laminate (see Figs. 4.6 and 4.7). This conclusion is important for applications such as bolted joints that establish a preferable stacking sequence for the dominant loading case. The design of the joint should be developed in a way that accounts for both service loads and the machining process needed by the composite component.

90º

–45º

0º

90º HSNMCCRT (Avg: 75%)

0º

45º

–45º

+1,000e+00 +9,167e–01 +8,111e–01 +7,500e–01 +6,667e–01 +5,811e–01 +5,000e–01 +4,167e–01 +1,111e–01 +2,500e–01 +1,667e–01 +8,111e–02 +0,000e+00

Fig. 4.6: Damage (matrix crushing) in each ply of both laminates considered (stacking sequences at 45/−45/0/90° and 90/0/45/−45°) [12].

Matrix failure

Delamination

[90/0/45/–45]s

[45/–45/0/90]s

Fibre failure

Fig. 4.7: Fiber compression, matrix crushing and delamination (stacking sequences at 45/−45/0/90° and 45/−45/0/90°) [12].

HSNMCCRT (Avg: 75%) +1,000e+00 +9,167e–01 +8,111e–01 +7,500e–01 +6,667e–01 +5,811e–01 +5,000e–01 +4,167e–01 +1,111e–01 +2,500e–01 +1,667e–01 +8,111e–02 +0,000e+00

114 | 4 Numerical modeling of LFRP machining The results obtained in the work were improved by using cohesive elements for delamination modeling. Cohesive elements have demonstrated their ability to simulate delamination. Delamination prediction is one of the most important defects when machining LFRP. Different authors have focused attention on the experimental analysis of machining-induced delamination. Davim et al. [33] investigated the influence of cutting parameters and the matrix on the specific cutting force, delamination factor and surface roughness. The feed parameter had the most influence on the delamination factor. Abrao et al. [34] checked the influence of tool geometry and cutting parameters on delamination during drilling glass FRP. It was shown that the strong influence of the drill geometry in competition with the thrust force was commonly assumed to be the most influential factor. In this work, it was demonstrated that the drill responsible for the highest thrust force caused the second smallest delaminated area. A novel technique for measuring the delamination factor after drilling FRP laminates, using digital analysis, was presented in [35]. The experimental results indicated that the use of digital analysis is suitable for controlling the drilling-induced damage in carbon fiber reinforced plastics. This technique was successfully applied to damage control in thehigh-speed drilling of glass FRP in [36]. Delamination decreased as the cutting speed increased within the cutting range tested, probably due to an increased cutting temperature with spindle speed; this leads to enhanced softening of the matrix and less delamination. Krishnaraj et al. [37] explored the high-speed drilling of carbon FRP and demonstrated the greater influence of feed rate on the thrust force, push-out delamination and diameter of the hole. Lower feed rates reduce thrust force and push-out delamination, while higher feed rates lead to hole diameters closer to the nominal value. Spindle speed is one of the major determinants for the circularity of the drilled hole. On the other hand, spindle speed and feed rate had negligible influence on peel-up delamination within the tested range. In a recent review dealing with drilling of composites [38], the authors summarized main contributions in the literature concerning drilling-induced delamination (mainly experimental and analytical approaches), highlighting the importance of this phenomenon and showing some techniques, tools and operations developed to minimize the occurrence of delamination. Studies of milling carbon fiber reinforced polymer (CFRP) composites are somewhat limited compared to drilling. Hintze et al. [39] focused on milling CFRPs experiments and observed that the occurrence of delamination is closely related to tool wear and the top-layer fiber cutting angle. Lopez de Lacalle et al. [40] studied the performance of multi-tooth cutting tools during the trimming process of CFRP. The authors demonstrated the superior performance of multi-tooth cutting tools with TiAlN coating in comparison to the behavior of straight edge polycrystalline diamond (PCD) tools for finishing operations. Davim and Reis [41] established a model using multiple re-

4.2 Orthogonal cutting | 115

gression analysis between cutting velocity and feed rate with the surface roughness and damage in a CFRP composite material. Feed rate was the most influential parameter on both the surface roughness and delamination factor. Similar behavior concerning the influence of feed rate on workpiece integrity indicators was observed by Davim et al. [42] during milling glass FRP. Karpat et al. [43] developed a mechanistic cutting force model for milling CFRPs based on experimentally collected cutting force data during slot milling of unidirectional CFRPs laminates using two different polycrystalline diamond cutters. No delamination at the sides of the slot for fiber angle equal to 45° was observed. Delamination damage occurs not only in composite cutting, but also in other dynamic processes such as impact loading (see for instance [44]). It is accepted in the scientific community that the prediction of delamination requires the use of cohesive interactions modeling at the interface between plies [45]. This type of interaction has been successfully applied in the modeling of dynamic loading of composites (see for instance [46]), and in the simplified models of drilling assuming the process to be similar to punching. Cohesive interactions were used to model out-of-plane failure induced during cutting in complete simulations including chip removal in a recent work of the authors. Tensile stress t0n(t0s,t0t)

Gcn(Gcs,tct) Separation

Fig. 4.8: Traction separation response for cohesive interactions: tensile stress vs. separation.

A modification of the 3D model described above was performed, using cohesive interactions between plies for simulating delamination [13]. Modeling delamination by cohesive interaction allows for the reproduction of delamination propagation as a fracture mechanics phenomenon, instead of stress-based models as in the case of Hou criteria. The surface-based cohesive interaction available in ABAQUS was used to predict the delaminated area. Figure 4.8 shows a typical traction-separation response using cohesive interaction [47]; a linear elastic behavior is considered until delamination onset is verified. Thus, a failure criterion is required to predict damage. In this work, delamination onset was predicted by a quadratic stress criterion considering out-ofplane stresses (equation (4.1)). 2

{

2

2

t t tn } + { 0s } + { 0t } = 1 , tn0 ts tt

(4.1)

where tn is the normal traction stress, ts and tt are the shear traction stresses. tn0 , ts0 and tt0 are maximum admissible values of stresses.

116 | 4 Numerical modeling of LFRP machining Once damage is verified, ABAQUS cohesive interaction requires the definition of a damage evolution law. The behavior of the cohesive interaction after damage onset was defined by the Benzeggagh-Kenane (BK) damage model [47]. The BK model is based on the energy dissipated due to failure considering a traction-separation response according to equation (4.2) where GS = Gs + Gt and GT = Gn + GS . The variables Gn , Gs and Gt refer to the work done by the traction and its conjugate relative displacement in the normal direction and the first and the second shear directions, respectively. The quantities Gcn , Gcs andGct refer to the critical fracture energies required to cause failure in the normal, the first and the second shear directions, respectively; η is a cohesive property parameter. The BK model is especially useful when the critical fracture energies during deformation are only along the first and the second shear directions and the same; i.e. Gcs = Gct . GCn + (GCs + GCn ) (

GS η ) = GC . GT

(4.2)

The values implemented in the model were coherent and equivalent to those used for the Hou out-of-plane failure model. A 3D model was applied to simulate orthogonal cutting of quasi-isotropic laminates considering two stacking sequences [45/−45/0/90°]S and [90/0/45/−45°]S . Cohesive properties are provided in the Tab. 4.5. Tab. 4.5: Matrix cohesive characteristics [13]. Property

Value traction stress tn0 traction stress ts0 traction stress tt0

Maximum normal (MPa) Maximum normal (MPa) Maximum normal (MPa) Critical fracture energy in normal direction GCn (kJ/m2 ) Critical fracture energy in normal direction GCs (kJ/m2 ) Critical fracture energy in normal direction GCt (kJ/m2 ) Matrix cohesive property parameter η

55 68 68 0.3 2.023 2.023 1.75

The evolution of chip formation predicted during cutting is shown in Fig. 4.9. Chip morphology is not uniform through the laminate thickness, since the composite is a non-homogeneous material. Each ply with a given orientation of the fiber presents a different evolution of chip and damage extension; the discontinuity between plies is one of the reasons causing delamination. Moreover, out-of-plane loads – neglected when plain stress state is assumed, which is the common hypothesis in 2D analysis – enhance delamination. In the case of unidirectional laminates using the 3D approach, the analysis of the influence of fiber orientation gives similar results to those obtained with the 2D model when the laminate is thin enough to ensure the validity of a plane stress assumption, as has been previously explained.

4.2 Orthogonal cutting | 117

t = 0 ms

t = 0.04 ms

t = 0.09 ms

t = 0.13 ms

t = 0.17 ms

t = 0.19 ms

Fig. 4.9: Evolution of chip generation with cutting time (stacking sequence 45/−45/0/90°]S ) [13].

Cohesive interactions implemented in the 3D model have shown important differences in the prediction of the damaged area when compared to a classical formulation. The Hou approach for delamination damage undervalued the levels and extension of this phenomenon. A qualitative comparison with experimental results from the literature with significant levels of damage has shown more realistic predictions in the case of implementing cohesive interactions. The strong influence of the staking sequence on delamination damage induced during machining was demonstrated. The largest delaminated areas were observed at the interface of 0/90° and 0/−45° (see Fig. 4.10). Laminate architectures with these interfaces should be avoided at the free surface where the plies tend to develop larger damaged zones.

4.2.3 Thermal effects Mechanical delamination is not the unique damage mechanism involved in composite machining. The relatively low glass transition temperature (around 180 °C for a typical epoxy resin in CFRPs) explains the susceptibility of the composite to suffering thermal damage [48]. Matrix degradation could produce delamination related to a fur-

118 | 4 Numerical modeling of LFRP machining

–45/45 Interface 45/–45 Interface

(a)

Classic interactions

–45/45 Interface 45/–45 Interface

Cohesive interactions

Cohesive interactions

0/90 Interface

+1.000e+00 +9.900e+01 +1.500e+02 +0.000e+00

Classic interactions

+1.000e+00 +9.900e+01 +1.500e+02 +0.000e+00 Cohesive interactions

Classic interactions

90/90 Interface

Classic interactions

0/45 Interface

90/0 Interface

Cohesive interactions

(b)

Fig. 4.10: (a) Delamination predicted using cohesive interaction (left column) and Hou model (right column). Stacking sequence [90/0/45/−45°]S . (b) Delamination predicted using cohesive interaction (left column) and Hou model (right column). Stacking sequence [45/−45/0/90°]S [13].

ther strength reduction. On the other hand, cutting temperature affects not only the surface quality of the workpiece, but also influences tool wear evolution. Despite the importance of analyzing thermal phenomena involved in machining composites, only few authors have focused on this problem. Several authors have focused on obtaining indirect information through the measurement of temperatures at the cutting tool using a thermocouple. Chen [49] measured the temperature at the flank surface of the drill during drilling and obtained a significant increment of temperature (120–300 °C) when the cutting speed increased from 40 to 200 m/min. The temperature obtained during drilling and orbital milling of hybrid components Ti/CFRP/Al was evaluated by Brinksmeier et al. [50] using a thermocouple embedded at the tool tip. Orbital drilling involved lower temperatures and better surface quality than drilling (82 °C in CFRP with acutting speed equal to 40 m/min). In the studies reported in references [49, 50], temperature measurement was carried out by installing a thermocouple inside the tool, giving indirect information about the temperature level at the workpiece. In a recent work [48], a complete experimental work involving direct temperature measurement, both at the workpiece and at the tool, was reported. Milling of CFRP was conducted with a carbide end mill in dry conditions. Three routes for measurement were used: the tool-workpiece thermocouple method, an infrared thermo-graph camera for end mill surface temperature measure-

4.2 Orthogonal cutting | 119

ment and an embedded K-type thermocouple in the CFRP for measurement of the temperature at the machined surface layer. At a cutting speed equal to 300 m/min, the temperature at 0.3 mm beneath the machined surface reached 104 °C. This temperature was much lower than that measured at the cutting point (tool–chip contact point). Haddad et al. [51] analyzed high-speed trimming of a multidirectional CFRP using unused and used burr tools. The influence of machining parameters (feed speed, cutting speed and cutting distance) on the cutting forces, machining temperature and the machined surface quality was studied. Machining with unused tools resulted in recorded temperatures that were lower than the glass transition temperature of the composite, while the temperature was higher than this threshold when machining with used tools; SEM analysis of the machined surface showed damaged areas. Machining parameters had a strong influence on the variation of the machined surface quality and cutting forces. In a recent work [52], an experimental study on cutting temperature in rotary ultrasonic machining of carbon fiber reinforced plastic using two measurement methods was presented: a novel method based on a fiber optic sensor and a well-known technique based on thermocouples. Both methods were compared, and the relations between input variables (ultrasonic power, tool rotation speed and feed rate) and cutting temperature were experimentally determined. As it was mentioned above, the measurement of damage is expensive and sometimes requires destructive techniques, thus it is desirable to develop simulation tools that integrate thermal issues that are able to reproduce damage mechanisms induced during machining. Numerical models based on the finite element method appear to be useful tools for thermal phenomena analysis, since they allow quantifying temperature at zones that are difficult to access. Numerical models are rarely developed that account for thermal phenomena in machining. The authors explored the ability of numerical modeling based on finite elements for the simulation of temperature enhancement during cutting. The previously mentioned model of orthogonal cutting was modified in order to account for thermal effects [53]. The model is used for the prediction of intralaminar damage, delamination and thermal damage in terms of the temperature level beneath the machined surface. Thus the subroutine VUMAT was modified including heat conduction (see thermal properties in Tab. 4.6 obtained from scientific literature). It should be noted that the anisotropic nature of the workpiece, with strong differences in thermal properties between matrix and fiber, resulted in no uniform heat propagation depending on fiber orientation. The heat generated due to material deformation was assumed to be negligible. It is worth noting that CFRP composites exhibit elevated elastic modulus with small deformations even when breakage is initiated. Due to this brittle behavior, heat generation produced by plastic deformation is assumed to be negligible. This behavior is quite different from that observed in metal cutting with large amounts of plastic work con-

120 | 4 Numerical modeling of LFRP machining Tab. 4.6: Thermal properties of composite [53]. Workpiece (CFRP) Longitudinal thermal conductivity (W/mK) Transverse thermal conductivity (W/mK) Longitudinal thermal expansion (10−6 /K) Transverse thermal expansion (10−6 /K)

6 0.5 1 26

verted into heat leading to a very high temperature at the primary shear zone. Since deformation energy is neglected, the unique heat source considered in the model was friction at the interface. The degradation of composite properties induced by the temperature increment was not accounted for. The onset of thermal damage was considered to appear when the critical temperature was reached. Some authors consider the critical level equal to the glass transition temperature (around 180 °C for a typical epoxy resin). The threshold of temperature indicative of thermal damage was assumed to be equal to 150 °C, being a temperature value between the initiation of mechanical properties degradation (around 100 °C [54]) and the resin transition temperature. Despite the limitations of this preliminary study, some interesting conclusions were derived. First, the strong influence of the fiber orientation on the temperature distribution was observed because of the different values of thermal conductivity exhibited by the matrix and the fiber (see Fig. 4.11). On the other hand, the friction increment at the interface resulted in increased thermal (for orientations of 0° and 45°) and mechanical (for orientation −45°) damage. As is well known, friction can be reduced significantly by using a proper tool coating. The increment of cutting speed produced a rise in temperature; however, the increment was limited in the model by the coupled mechanical phenomena involving element erosion and avoiding heat transmission. An interesting result was that in some analyzed cases, the depth of the thermally affected layer was larger than the penetration of mechanical damage. This fact corroborates the importance of accounting for thermal phenomena when modeling composite cutting. As mentioned above, this work was preliminary. Thus, further analysis should improve the limitations of the model, for instance including thermal damage in the failure model, since thermal effects would affect the material’s mechanical response.

4.3 Drilling

121

Cutting time = 5 μs

90° Laminate 0° Laminate –45° Laminate 45° Laminate

Cutting time = 1 μs Cutting time = 3 μs

|

Fig. 4.11: Temperature fields for different ply orientation (90/0/−45/45°). Temperature contours range from 150 °C to 100 °C. Cutting speed 100 m/s, friction coefficient 0.5 [53].

4.3 Drilling Drilling operations are required before composite components are joined mechanically [33]. A significant percentage of component rejection in aircraft manufacturing is due to delamination that is induced during drilling. Delamination is one of the undesired effects of machining when using unappropriate cutting parameters or worn drills and has received extensive attention in the literature (see previous section). The most important contributions in the field of composite drilling are summarized in a recent review [38] including techniques, tools and operations developed to minimize the occurrence of delamination. A significant number of works have been commented on in the previous section. Finite element modeling of the complex drilling process was recently achieved in [55, 56]. In these works, drilling of CFRP was successfully reproduced including drill penetration in the workpiece, material failure and elements erosion. Good agreement between the measured and predicted torque, thrust force and delamination extension was shown. Previously, simplified models of CFRP drilling were developed, having the advantage of a very reduced computational cost. Modeling drilling processes involves an elevated difficulty, because of the need of simulating drill rotation and feed movement. As it was commented in previous sections, a common assumption in simplified models is that the drill acts like a punch that pierces the laminate; see for instance [16, 57–59].

122 | 4 Numerical modeling of LFRP machining The interest of modeling drilling processes is clear due to the importance of this operation for the industry. There are two groups of available simulation tools: simplified models that treat the problem as a punching process, with efficient computational cost and reduced geometrical complexity; or complete models that reproduce the drilling process rigorously simulating rotation and feed movement of the tool (including penetration of the drill in the workpiece and element erosion, leading to elevated computational cost). The implementation of machining models in industry is still a challenge, probably because of the complexity of the simulation tools and required computational time. The availability of simple models leading to reasonable predictions could help in the wide implementation of simulation tools in industry.

4.3.1 Comparison between simplified and complete drilling models In a recent work by the authors, a comparison of predictions given by a simplified and a complete drill model based on finite elements is provided [19]. The aim was to give an overview of the main advantages of both types of modeling, analyze their accuracy when predicting delamination and study the influence of some parameters involved in delamination. Two numerical models that respectively simulated complete drilling and simplified punching of tape laminate were developed using the commercial finite element code ABAQUS. A scheme of the models that shows boundary conditions is presented in Fig. 4.12. In the simplified model, the drill acts like a punch pushing the laminate (two stages of drill penetration are simulated); see Fig. 4.12 (a). The complete model reproduces the complex 3D process with rotary and feed movement; see Fig. 4.12 (b). The drill and the laminate characteristics were the same for both models and were obtained from [55] for model validation. Both models include restriction to displacement in z direction in the base of the workpiece except for an inside a circumference with a diameter of 16 mm where the z displacement is free. On the other hand, displacements were not allowed in the contour of the workpiece. The drill was assumed to be rigid, with a diameter of 3 mm and a tip angle of 120°. The workpiece characteristics were similar to those stated for experimental validation (see [55]): CFRP composite (UD T300/LTM45-EL) composed of tape plies with a 2 mm thickness consisting of 16 plies with a stacking sequence [04 /904 ]S (each ply was 0.125 mm thick). Each ply was modeled at the zone close to the drill entrance using the solid elements C3D6R with six nodes (1 element along the thickness). The use of wedge elements (a prismatic element with a triangular section) minimizes the dependence of the results with mesh orientation in the laminate plane. The minimum element size was 0.25 mm. Far from the drill entrance zone, hexagonal elements C3D8R

4.3 Drilling |

123

U3

ω U1

U2

Fr

Tf

U1=U2=0

U1=U2=0

(a)

(b)

Fig. 4.12: (a) Scheme of the simplified model (Tf = trust force). (b) Scheme of the complete model (ω = rotation speed, Fr = feed rate) [19].

with 8 nodes and reduced integration were used, with a minimum element size of around 1 mm [25]. The anisotropic composite was modeled using an elastic behavior up to failure. Elastic properties of the composite are presented in Tab. 4.7. Tab. 4.7: Material ply properties. Property

Value 3

Density ρ (kg/m ) Longitudinal modulus E1 (GPa) Transverse modudlus E2 (GPa) In-Plane shear modulus G12 (GPa) Out-of-Plane shear modulus G23 (GPa) Poisson’s ratio 𝜈12

1600 127 9.1 5.6 4 0.31

Inter-laminar failure was modeled with the use of cohesive elements. A small thickness was assigned to the interface (5 μm) in order to improve the numerical behavior when elevated deformations were reached during calculation. Meshing strategy in the planes 1–2 was the same as that used in the ply, with one element along the thickness (direction 3). Delamination modeling requires the establishment of a damage initiation criterion and a damage evolution law as described above. The interaction between workpiece and tool was modeled using the algorithm surface-node surface contact that is available in the ABAQUS/explicit. The contact was defined between the drill surface and the composite plate nodes in the region adjacent to the contact area. In addition, a self-contact condition was used to avoid penetration between the eroded composite elements. A constant coefficient of friction equal to 0.3 at the tool/workpiece interface was assumed.

124 | 4 Numerical modeling of LFRP machining The complete model allowing for chip removal involved a dynamic analysis including geometric non-linearity and large deformation options. The problem was solved using an explicit integration scheme (ABAQUS/explicit). A compromise between accuracy and computational cost was achieved when selecting the element size (previously commented) directly involved in the time step. The rotary movement of the drill around the y-axis at a constant spindle speed and feed rate in direction y were imposed. A section of the drilled hole simulated with the complete model is presented in Fig. 4.13; it shows the entrance and exit of the drill. The cutting parameters were stated to be equal to those provided in [55] used for model validation. The rotary velocity was equal to 2500 rpm and feed rate was equal to 2.5, 5 and 8.3 mm/s. The calculation time for simulation ranged from 4 days to 3 weeks in a work station with 16 CPUs. The complete model was validated by comparing experimental results provided in a recent work [55] dealing with drilling of tape carbon-epoxy LFRP composite. The model was validated in terms of the delamination factor at the entrance of the drill, torque and thrust force. The delamination factor was calculated as the ratio between the maximum diameter of the delaminated area and the nominal diameter of the drill. Reasonable accuracy was reached when predicting these parameters (see details in [19]).

(a)

(b)

Fig. 4.13: Section of the hole during penetration of the drill simulated with complete model: (a) entrance and (b) exit of the drill [19].

The simplified model involved a dynamic analysis that also used an explicit integration scheme (ABAQUS/explicit). No rotation was imposed to the drill. A constant thrust force (obtained from the complete model previously validated) was applied at the top of the drill, corresponding to the level of penetration of the drill at the simulated stage. The elements corresponding to the pre-drilled volume were removed from the model, and the drill contacted the workpiece at a depth equal to H (see Fig. 4.14). Two stages of drill penetration across the workpiece were simulated: H equal to 1 mm and 1.625 mm (corresponding respectively to 8 and 13 plies drilled respectively). The first stage corresponded to the maximum level of thrust force observed experimentally and in sim-

4.3 Drilling |

125

Ф=3mm

H 2 mm Fig. 4.14: Scheme of the pre-drilled hole in the simplified model [19].

ulations with the complete model. The second stage was selected because only a few plies were not drilled and the origin of delamination was commonly observed to be close to the drill exit. Efficient computation was achieved for simulation in the order of several minutes of calculation time in a work station with 16 CPUs. The simplified model needs the value of thrust force as an input (obtained at the drill penetration simulated), and main output is the prediction of the delaminated area. The complete model also provides delamination, but the input data are rotary velocity and feed rate. Figure 4.15 (a) and (b) show the delamination factor predicted with both models for both stages of penetration considered (clamping was applied in direction z at the bottom of the plate, except for a free circular surface with a diameter of 16 mm). 1.8

1.6

Simplified

1.4

1.4

1.2

1.2

Delamination factor

Delamination factor

1.6

1.8

Complete

1 0.8 0.6 0.4 0.2

Simplified

1 0.8 0.6 0.4 0.2

0

0 fr: 2.5 mm/s fr: 5 mm/s fr: 8.3 mm/s Tf: 165.5N Tf: 211.5N Tf: 240.9N

fr: 2.5 mm/s fr: 5 mm/s fr: 8.3 mm/s Tf: 121.6N Tf: 125.2N Tf: 152.4N 13 plies penetration

8 plies penetration (a)

Complete

(b)

Fig. 4.15: Predicted and experimental delamination factor: (a) penetration of the drill through 8 plies; and (b) penetration of the drill through 13 plies [19].

126 | 4 Numerical modeling of LFRP machining It is possible to observe the overestimation of a delamination factor predicted with the simplified model; in particular, it is slightly larger than that obtained with the complete model. It is important to highlight this result; the simplified model is conservative when predicting delamination. The simplified model slightly overestimates the delamination factor value, giving a conservative prediction of damage. This result is especially significant when the complexity and calculation cost of both models are compared. Simplified simulations can be solved in several minutes while the complete model needs several days; this methodology could be considered when designing drilling processes with different drill geometries and cutting parameters. The implementation of numerical models in industry to help in the design process requires simple and fast simulation tools. During drill penetration, it is also interesting to analyze intra-laminar damage. Matrix and fiber failure result in element erosion as the drill penetrates though the workpiece. Intra-laminar damage is observed beneath the drill tip as it advances though the composite or pushes it, depending on the model considered. Intra-laminar damage is observed in zones with a lower diameter than that of the drill. Thus, the damaged zone is eroded as the drill penetrates the composite plate. As a consequence, only the inter-laminar damage, i.e. delamination, is analyzed. The simplified model was used to study the influence of several factors involved in the drilling process. First of all, the influence of thrust force was analyzed. It is well known that the delamination factor increases as force is increased. A plateau that limited this increment of delamination factor was found at high enough thrust forces. The maximum level of delamination was coincident with that induced by complete perforation of the composite plate with a punch with a geometry similar to the drill. Although perforation is not commonly used in hole manufacturing in composites, the delamination factor in this process can be used as an upper limit for conventional drilling, giving valuable information for the designer. The influence of clamping was also studied. Here, a significant increment in delamination factor was found when the diameter of the free surface ranged from the value equal to the drill diameter to higher values up to five times the drill diameter. For higher values, negligible influence of clamping was observed. Finally, the simplified model was used to simulate other stacking sequences corresponding to laminates [45/−45/0/90°]4 and [90/0/45/−45°]4 . The quasi-isotropic laminates presented a similar delamination factor, in both cases lower than the one exhibited by the laminate [04 /904 °]S .

4.3.2 Thermal model of drilling As mentioned above, thermal effects influence surface integrity when drilling LFRP. Thermal damage, related to low glass transition temperature (around 180 °C for a typical epoxy resin in CFRP), is in the basis of matrix degradation, and it is also involved in

4.3 Drilling

|

127

plies separation. Despite the potential risk of thermal effects in machining CFRP, they have only been analyzed in few works in the literature, mostly based on experimental work. The lack of information related to the numerical prediction of thermal issues in drilling processes has motivated the development of a work focused on the estimation of heat amount from variables easy to measure in process: torque and thrust force [60]. A simple analysis based on energy balance was presented to obtain the heat generated at the interface indirectly from experimental tests. As far as the bibliographic revision was carried out, this approach has not been applied to the problem of drilling; however, analytical models of impact in composites based on energy balance (impact processes have common characteristics with drilling, for instance, the penetration of the projectile/drill into the target/workpiece) in [61] are interesting. Once the amount of heat is available, it is possible to establish a temperature distribution with a simple numerical model accounting for thermal conductivity in the composite. The detection of critical levels of temperature at certain zones can be used for assessment during the definition of the drilling process of high-responsibility components, avoiding the risk of thermal damage. The establishment of critical levels of torque and thrust force would allow for the detection of excessive tool wear. The aim of the paper and its practical application are illustrated in Fig. 4.16.

Definition of drilling process Measured torque and thrust force

Analytical model Energy balance: Estimation of heat ammount at tool/chip interface

Numerical model Temperature distribution

ΔWT + WF = Ef + Em + Ec + ΔQ

Corrective actions Detection of tool life related to surface integrity Optimization of cutting parameters

Analysis of critical zones

Fig. 4.16: Relationships between proposed experimental, analytical and numerical steps [60].

Estimation of frictional heat from energy balance The drilling process is performed at a constant feed and rotary velocity. The control of the machine tool maintains these parameters, and the resultant torque and thrust force measured at the spindle are a consequence of cutting parameters, material properties and characteristics of the cutting tool, including contact behavior at the interface. Figure 4.17 illustrates the entrance of the drill in the composite during cutting at a certain cutting time. Thus, a simplified method to estimate the heat generated at the

128 | 4 Numerical modeling of LFRP machining Lcut

Lcut =

Rcut

Effective cutting edge length during cutting stages

cos α

dθ Lcut fcut t1

t2

t3

t

α

Differential volume drilled during dt

Fig. 4.17: Scheme of the drilling process: differential volume removed during the differential time dt and effective cutting edge at the different stages of drilling from entrance to drill exit [60].

interface has been developed assuming the hypotheses of constant feed and rotary velocity. The differential work involved in torque (dWT ) and thrust force (dWF ) during a differential time increment dt involved in a differential turn angle of the drill dθ is the result of several contributions, as summarized in equation (4.3): the energy required for breakage of composite (dEf ), the kinetic energy transferred to the chip (dEc ) and the amount of heat generated at the interface due to friction (dQ). dWT + dWF = dEf + dEc + dQ .

(4.3)

The terms corresponding to work developed by torque and thrust force were obtained from experiments. The torque and thrust force recorded at each time increment were derived from the discretization of the evolution with cutting time of both variables, measured during the drilling test. The term corresponding to energy involved in composite breakage can be calculated considering the differential volume removed by the drill during a differential time dt corresponding to a differential drill turn dθ . dEf (θ ) = wf dVf (θ ) ,

(4.4)

where wf is the specific energy for the woven composite breakage, and dVf the differential volume associated to a differential turn of the drill dθ . The specific energy can be estimated as wf = 2(0.5Xεf ) where X is the strength of the woven composite (the same in direction 1 and 2 because of the nature of the

4.3 Drilling

|

129

woven architecture of the composite) and εf is the ultimate strain of the composite. The approach was similar to that used by Artero-Guerrero et al. [62] when modeling the impact on a woven composite. The differential volume considered is represented in Fig. 4.2. From this figure, the volume can be calculated as equation (4.5) demonstrates; Lcut is the effective cutting edge length and fcut is the feed rate. dθ (4.5) 2π It is worth noting that the volume of material removed depends on the stage of the drilling process. It is possible to distinguish three different stages illustrated in Fig. 4.17: the entrance of the conical zone; the cut performed with the complete edge; and the exit of the drill. For the geometry of the drill used in the machining tests (described in the next section), the stages indicated in the figure correspond with the following values of time and effective length of the cutting edge where Rcut is the drill radius. dVf (θ ) = π L2cut fcut

Entrance stage, time 0–t1 :

Rcut t ⋅ cos α t1 R = cut cos α R t − t2 )] = cut [1 − ( cos α t3 − t2

Lcut1 =

(4.6) (4.7)

Steady stage,

time t1 –t2 :

Lcut2

Exit stage,

time t2 –t3 :

Lcut3

(4.8)

The kinetic energy of the chip, once separated from the workpiece, can be neglected due to the small mass of the chip and the velocity range involved in cutting. A first estimation showed that the level was negligible compared to the rest of the terms in equation (4.3) (approximately 0.005 % of the energy required for breakage of composite (dEf )). Thus, the heat generated at the interface can be estimated as: dQ = dWT + dWF − dEf .

(4.9)

The expression presented in equation (4.9) was applied to real experiments that involved the drilling of woven carbon composite. The specific conditions of the experiments are presented in the next subsection, involving both new and worn tool geometries.

Application to experiments In order to apply the energy balance formulated in equation (4.9) to a real case of CFRP drilling, experimental tests were performed on a CFRP composite. The material used in this test was a woven CFRP composite. Each ply was manufactured by Hexcel Composites with an AS-4 carbon fiber and epoxy matrix. The specimens were cut in plates with a stacking sequence of 10 plies with the same fiber orientation in all of them, and a total thickness of 2.2 mm. The mechanical properties of this material are listed in Tab. 4.8.

130 | 4 Numerical modeling of LFRP machining Tab. 4.8: Mechanical properties of AGP 193-PW/8552 composite material. Property

Value 3

Density ρ (kg/m ) Resin content (%) Longitudinal modulus E1 (GPa) Longitudinal modulus E2 (GPa) Poisson’s ratio 𝜈21 Longitudinal tensile strength XT (MPa) Longitudinal compressive strength XC (MPa) Transverse tensile strength YT (MPa) Transverse compressive strength YC (MPa) In-Plane shear strength ST (MPa)

1570 55.29 68 68 0.21 880 880 880 880 84

The cutting tests were carrying out in the machining center B500 KONDIA shown in Fig. 4.18. The machining center was equipped with a Kistler (9123C) dynamometer for the measurement of cutting forces and torque (see Fig. 4.18). The drill (diameter 6 mm, point angle 118°) was recommended by the manufacturer GUHRING for CFRP drilling. Drilling tests were performed with a new drill and also with a worn tool exhibiting flank wear (this wear mode is commonly observed to be dominant in drilling CFRP). A fresh tool and severe wear (flank = 0.3 mm [61]) were tested in order to study the different conditions of cutting forces and torque, and in consequence different levels of generated heat. Due to the difficulty in obtaining controlled worn geometries directly from wear tests, the flank at the clearance surface was artificially generated by grinding.

Fig. 4.18: Machining center used in the experiments equipped with the dynamometer and acquisition system. Also with a system for chip aspiration [60].

The drilling experiments were conducted without coolant with a feed of 0.1 mm/rev and cutting speed 50 m/min. The characteristics of the workpiece and drill stated for the experiments allowed the calculation of time intervals defined in equations (4.4)– (4.6). Accounting for a drill tip angle equal to 118°, the thickness of the composite plate and the drill radius were equal to 6 mm, t1 = 0.38 s, t2 = 0.88 s and t3 = 1.26 s.

4.3 Drilling

|

131

Power [W]

120 New tool - Total power New tool - Thrust direction power New tool - Peripherical direction power

100 80 60 40 20 0 0 –20

0.5

1.5

2

Time [s]

(a) 120 100 80 60 40 20 0 0 –20 –40 (b)

1

Worn tool - Total power Worn tool - Thrust direction power Worn tool - Peripherical direction power

0.5

1

1.5

2

Time [s]

Fig. 4.19: (a) Power due to the drilling operation (spindle velocity 2653 rpm and feed 0.1 mm/rev) for a new tool. (b) Power due to the drilling operation (spindle velocity 2653 rpm and feed 0.1 mm/rev) for a worn tool (with flank = 0.3 mm) [60].

The evolution of power consumed in the trust and cutting movement (obtained from measured thrust force and torque, respectively) is presented in Fig. 4.19 (for a cutting speed of 50 m/min and feed of 0.1 mm/rev). This kind of graph was obtained with a fresh tool in Fig. 4.4 (a) and worn tool in Fig. 4.4 (b). From the recorded signal, the amount of heat generated due to friction was obtained as follows. The power consumed by peripheral friction was obtained as the average value of total power once exit stage (t > t3 ) was reached. During entrance stage (t < t1 ), the peripheral friction power (due to contact between the drill body and the hole wall) is zero, while during the steady stage (t1 < t < t2 ) it is linear from 0 to the exit stage value. The power consumed by the cutting edge friction was estimated according to equation (4.9) as the total power minus the energy consumed by composite breakage (equations (4.6)–(4.8)), and minus the peripheral friction. The discretization of the curve of heat vs. cutting time allows for the determination of thermal flux to the workpiece and the analysis of heat propagation in a finite element code. The heat propagation in the workpiece was analyzed using the finite element model.

132 | 4 Numerical modeling of LFRP machining Numerical modeling of temperature distribution The numerical model was developed using the commercial finite element code ABAQUS/explicit. The aim of the model was to analyze heat propagation during the drilling process in order to observe temperature distributions in the workpiece to identify critical levels and zones. First of all, the assumptions formulated should be explained. The model did not account for chip removal, and, in order to save computational time, axial symmetry is considered. The model has been discretized; each drill revolution corresponded to one depth of penetration equal to the feed (this depth of penetration per revolution will be used as a time increment per step in applying the loads). The frictional heat in each time increment was calculated from the analysis explained in the previous section. The heat partition between the tool and workpiece was assumed to be 50/50 %, being a common hypothesis in the simulation of cutting (see for instance [21]). The scheme of the model is shown in Fig. 4.20, including boundary conditions and geometry. The model was meshed with 70 000 linear triangular elements with a size of 25 μm. This mesh showed an adequate balance between computational cost and precision. The mechanical properties of the workpiece are summarized in Tab. 4.8. Thermal properties for CFRP in the scientific literature are given in a wide range, thus the values used in this work (thermal conductivity 5 W/mK and specific heat 1100 J/kg K) were averaged from several references covering different applications [63–66]. The procedure of the simulation is described in the following. In a generic time step (defined by the time that it takes for a drill revolution), the amount of heat along the cutting edge and lateral wall was calculated using the measured thrust force and torque by applying the analytical model, as was explained in the previous section. At

Elements to be removed in the following step Surface heat temperature due to the machining process

Convection flux

Convection flux Fig. 4.20: Scheme of the numerical model [60].

Different drilling stages

4.3 Drilling |

133

the end of each step, the layer of elements corresponding to the chip area removed in one revolution of the drill was eliminated, and the heat corresponding to the subsequent step was applied. The model was applied to three real cases of drilling with a fresh tool and two levels of wear. The numerical model showed the maximum temperature occurring at the hole wall close to the exit of the drill; this is a zone where mechanical delamination was commonly observed (see Fig. 4.21). The occurrence of thermal damage, in the case of excessive wear, enhances the risk of a defect at the exit of the hole. 9 mm 0.69 mm

1.76 mm

2.2 mm Temp (Avg: 75%)

250 μm 150 μm

(a)

+4.530e+02 +4.397e+02 +4.263e+02 +4.130e+02 +3.997e+02 +3.863e+02 +3.730e+02 +3.597e+02 +3.463e+02 +3.330e+02 +3.197e+02 +3.063e+02 +2.930e+02

325 μm 275 μm

(b)

Fig. 4.21: Predicted temperature (K) for tests developed at cutting speed 50 m/min and feed 0.1 mm/rev (grey zone represents temperature higher than 180 °C, 453 K): (a) fresh tool; (b) worn tool [60].

The model is simple and very efficient under the computational point of view. The problem of using realistic models of drilling, including penetration and cutting movement as well as elements erosion is the computational cost. The implementation of simulation tools in industry for assistance during manufacturing requires rapid response. The model proposed could be easily implemented to detect excessive levels of torque due to no proper cutting parameters or excessive wear of the tool.

134 | 4 Numerical modeling of LFRP machining

4.4 Conclusions In this chapter, main contributions in the field of numerical modeling (based on finite element) of LFRP cutting were summarized. The anisotropic nature of the composite and its non-homogeneous architecture based on the superposition of layers had to be accounted for. In most analyzed cases, the mechanical behavior of the composite was modeled using stresses-based failure sets of criteria, such as Hashin or Hou failure criteria [26, 27, 32]. Most works in the literature focus on orthogonal cutting, being the simplest machining process. 2D modeling has been developed that gives valuable information about intra-laminar damage, cutting forces and chip morphology. Only the cutting of unidirectional laminates can be simulated; the analysis is limited to phenomena occurring in the layer. 3D analysis is required in order to simulate different stacking sequences and interlaminar phenomena. Delamination was modeled using cohesive elements combined with Hou failure criteria for intra-laminar damage prediction; this proved its ability to predict cutting-induced damage. The hypotheses of the 2D model were analyzed in a comparison between 2D and 3D approaches; this showed the influence of layer location in the damage. Even in the case of orthogonal cutting, the 3D approach has been poorly developed in the literature. Actual industrial machining involves complex tool geometry, exigent parameters definition, and tool wear control. Drilling is one of the most important operations, and it has received attention with the development of simplified models (considering drilling similar to punching) and, recently, complex models allowing for chip removal. The comparison between both models has shown the most accurate predictions of the complete model; however, the computational cost is much higher. Thermal issues have rarely been accounted for in the literature. Few efforts have shown a significant extension of thermal damage. Thermal degradation of the matrix also influences delamination and thus mechanical damage. Despite different authors’ efforts in developing numerical models of LFRP cutting, there is a lack of these kinds of predictive tools in the literature, especially concerning actual industrial processes (for example, models of milling are not available). The implementation of modeling in an industrial environment is still far from usual procedures for process definition. However, numerical modeling is a promising technique for helping in this important industrial aspect. In fact, recent calls for international research projects highlight the skills required to be able to perform process modeling. It is worth noting the elevated cost of experimentation in composite cutting; this corroborates the importance of developing validated tools that are able to accurately simulate cutting processes.

References |

135

Acknowledgement The authors acknowledge the financial support for the work given by the Ministry of Economy and Competitiveness of Spain under the project DPI2011-25999 and FPI subprogram with the reference BES-2012-055162.

References [1] [2]

[3]

[4] [5] [6] [7] [8] [9] [10]

[11] [12] [13] [14] [15] [16] [17] [18]

Grilo TJ, Paulo RMF, Silva CRM, Davim JP. Experimental delamination analyses of CFRPs using different drill geometries. Compostes Part B 2013;45:1344–1350. Nayak D, Bhatnagar N, Mahajan P. Machining studies of uni-directional glass fiber reinforced plastic (UD-GFRP) composites part 1: effect of geometrical and process parameters. Mach Sci Technol 2005;9:481–501. Bhatnagar N, Nayak D, Singh I, Chouhan H, Mahajan P. Determination of machining-induced damage characteristics of fiber reinforced plastic composite laminates. Mater Manuf Process 2004;19(6):1009–23. Wang XM, Zhang LC. An experimental investigation into the orthogonal cutting of unidirectional fiber reinforced plastics. Int J Mach Tools Manuf 2003;43:1015–22. Chinmaya R. Dandekar,Yung C. Shin, Modeling of machining of composite materials: A review. Int J Mach Tools Manuf 2012;57:102–121. Miguélez MH, Soldani X, Molinari A. Analysis of adiabatic shear banding in orthogonal cutting of Ti alloy. Int J Mech Sci 2013;75:212–222. Molinari A, Soldani X, Miguélez MH. Adiabatic shear banding and scaling laws in chip formation with application to cutting of Ti–6Al–4V. J Mechan Ph Solid, 2013;61:2331–2359. Arola D, Ramulu M. Orthogonal cutting of fiber-reinforced composites: a finite element analysis. Int J Mech Sci 1997;39:597–613. Mahdi M, Zhang L. A finite element model for the orthogonal cutting of fiber reinforced composite materials. J Mater Process Technol 2001;113:368–372. Mkaddem A, El Mansori M. Finite element analysis when machining UGF reinforced PMCs plates: chip formation, crack propagation and induced damage. Mater Des 2009;30(8): 3295–3302. Santiuste C, Soldani X, Miguélez MH. Machining FEM model of long fiber composites for aeronautical components. Compos Struct 2010;92:691–698. Santiuste C, Miguélez MH, Soldani X. Out-of-plane failure mechanisms in LFRP composite cutting, Compos Struct 2011;93:2706–2713. Santiuste C, Olmedo A, Soldani X, Miguélez MH. Delamination prediction in orthogonal machining of carbon LFRP composites. J Reinf Plastic Comp 2012;31(13):875–885. Durao LMP, de Moura MFSF, Marques AT. Numerical simulation of the drilling process on carbon/epoxy composite laminates. Composites: Part A, 2006;37:1325–33. Durao LMP, De Moura MFSF, Marques AT. Numerical simulation of the drilling process on carbon/epoxy composite laminates. Compos Part A: Appl Sci Manuf 2006;37:1325–1333. Singh I, Bhatnagar N, Viswanath P. Drilling of uni directional glass fiber reinforced plastics: experimental and finite element study. Mater Des 2008;29:546–553. Hocheng H, Dharan CKH. Delamination during drilling in composite laminates. J Eng Ind 1990; 112:236–239. Upadhyay PC, Lyons JS. On the evaluation of critical thrust for delamination free drilling of composite laminates. J Reinf Plast and Comp 1999;18:1287–1303.

136 | 4 Numerical modeling of LFRP machining [19] Feito N, López-Puente J, Santiuste C, Miguélez MH. Numerical prediction of delamination in CFRP drilling. Comp Struct 2014;108:677–683. [20] Ramesh MV, Seetharamu KN, Ganesan N, Shivkumar MS. Analysis of machining of FRPs using FEM. Int J Mach Tools Manu 1998;38:1531–1549. [21] Mahdi M, Zhang L. An adaptive three-dimensional finite element algorithm for the orthogonal cutting of composite materials. J Mater Process Technol 2001;113:368–372. [22] Mahdi M, Zhang L. A finite element model for the orthogonal cutting of fiber reinforced composite materials. J Mater Process Technol 2001;113:368–372. [23] Arola D, Sultan MB, Ramulu M. Finite element modelling of edge trimming fiber-reinforced plastics transactions of the ASME. J Eng Mater Technol 2002;124:32–41. [24] Díaz-Álvarez J, Cantero JL, Miguélez MH, Soldani X. Numerical analysis of thermomechanical phenomena influencing tool wear in finishing turning of Inconel 718. Int J Mech Sci 2014;82: 161-169. [25] Hibbit, Karlson. Sorensen Inc. ABAQUS user’s manual 6.4-1;2003. [26] Hashin Z, Rotem A. A fatigue criterion for fiber-reinforced materials. J Compos Mater 1973;7: 448–464. [27] Hashin Z. Failure criteria for unidirectional fiber composites. J Appl Mech 1980;47:329–334. [28] Barge M, Hamdi H, Rech J, Bergheau J-M. Numerical modelling of orthogonal cutting: influence of numerical parameters. J Mater Process Technol 2005;164–165:1148–1153. [29] Miguélez H, Zaera R, Rusinek A, Moufki A, Molinari A. Numerical modelling of orthogonal cutting: influence of cutting conditions and separation criterion. J Phys IV 2006;134:417–422. [30] Hortig C, Svendsen B. Simulation of chip formation during high-speed cutting. J Mater Process Technol 2007;186:66–76. [31] Soldani X, Santiuste C, Muñoz-Sánchez A, Miguélez MH. Influence of tool geometry and numerical parameters when modeling orthogonal cutting of LFRP composites. Composites: Part A 2011;42:1205–1216. [32] Hou JP, Petrinic N, Ruiz C, Hallett SR. Prediction of impact damage in composite plates. Compos Sci Technol 2000;60(2):228–73. [33] Davim JP, Reis P, Antonio CC. Drilling fiber reinforced plastics (FRP) manufactured by hand lay up influence of matrix (VIAPAL VUP 9731 and ATLAC 382 05). J Mater Process Technol 2004; 155–156:1828–1833. [34] Abrão AM, Campos Rubio JC, Faria PE, Davim JP. The effect of cutting tool geometry on thrust force and delamination when drilling glass fibre reinforced plastic composite. Mater Des 2008;29:508–513. [35] Davim JP, Campos Rubio JC, Abrão AM. A novel approach based on digital analysis to evaluate the delam ination factor after drilling composite laminates. Compos Sci Technol 2007;67: 1939–1945. [36] Campos Rubio JC, Abrão A, Faria P, Correia AE, Davim JP. Effects of high speed in the drilling of glass fibre reinforced plastic: evolution of the delamination factor. Int J Mach Tools Manuf 2008;48:715–720. [37] Krishnaraj V, Prabukarthi A, Ramanathan A, Elanghovan N, Senthil Kumar M, Zitoune R, Davim JP. Optimization of machining parameters at high speed drilling of carbon fiber reinforced plastic (CFRP) laminates. Compos Part B: Eng, 2012;43(4):1791–1799 [38] Liu DF, Tang YJ, Cong WL. A review of mechanical drilling for composite laminates. Compos Struct 2012;94:1265–1279. [39] Hintze W, Hartman D, Schütte C. Occurrence and propagation of delamination during the machining of carbon fibre reinforced plastics (CFRPs) an experimental study. Compos Sci Technol 2011;71:1719–1726.

References |

137

[40] Lopez de Lacalle N, Lamikiz A, Campa FJ, Valdivielso A, Etxebarria I. Design and test of a multitooth tool for CFRP milling. J Compos Mater 2009;43:3275–3290. [41] Davim JP, Reis P. Damage and dimensional precision on milling carbon fiber reinforced plastics using design experiments. J Mater Process Technol 2005;160:160–167. [42] Davim JP, Reis P, Antonio CC. A study on milling of glass fiber reinforced plastics manufactured by hand lay up using statistical analysis (ANOVA). Compos Struct 2004;64:493–500. [43] Karpat Y, Bahtiyar O, Deger B. Mechanistic force modeling for milling of unidirectional carbon fiber reinforced polymer laminates. Int J Mach Tools Manuf 2012;56:79–93. [44] Zaera R. Ballistic impacts on polymer matrix composites, composite armor, personal armor in impact engineeringof composite structures. In: Abrate S (ed.) Impact engineering of composite structures. New York, NY: Springer Vienna;2011. [45] Wisnom MR. Modelling discrete failures in composites with interface elements. Compos: Part A: Appl Sci Manuf 2010;41:795–805. [46] Aymerich F, Dore F, Priolo P. Prediction of impact induced delamination in cross ply composite laminates using cohesive interface elements. Compos Sci Technol 2008;68:2383–2390. [47] Benzeggagh ML, Kenane M. Measurement of mixed mode delamination fracture toughness of unidirectional glass/epoxy composites with mixed mode bending apparatus. Compos Sci Technol 1996;56:439–449. [48] Takeshi Y, Takayuki O, Hiroyuki S. Temperature measurement of cutting tool and machined surface layer in milling of CFRP. Int J Mach Tool Manuf 2013;70:63–69. [49] Wen-Chou C. Some experimental investigations in the drilling of carbon fibre-reinforced plastic (CFRP) composite laminates. Int J Mach Tool Manuf 1997;37:1097–1108. [50] Brinksmeier E, Fangmann S, Rentsch R. Drilling of composites and resulting surface integrity. CIRP Ann-Manuf Technol 2011;60:57–60. [51] Haddad M, Zitoune R, Eyma F, Castanié B. Machinability and surface quality during high speed trimming of multi directional, CFRP. Int J Mach Mach Mater 2013;13:289–310. [52] Cong WL, Zou X, Deines TW, Wu N, Wang X, Pei ZJ. Rotary ultrasonic machining of carbon fiber reinforced plastic composites: An experimental study on cutting temperature. J Reinf Plastics Compos 2012;31:1516–1525. [53] Santiuste C, Díaz-Álvarez J, Soldani X, Miguélez MH. Modelling thermal effects in machining of CFRP composites. J Reinf Plastic Comp 2014;33(8):758–766. [54] Saafi M. Effect of fire on FRP reinforced concrete members. Compos Struct 2002;58:11–20. [55] Phadnis VA, Makhdum F, Roy A, Silberschmidt VV. Drilling in carbon/epoxy composites: experimental investigations and finite element implementation. Compos A Appl Sci Manuf 2013; 47:41–51. [56] Isbilir O, Ghassemieh E. Numerical investigation of the effects of drill geometry on drilling induced delamination of carbon fiber reinforced composites. Compos Struct 2013;105:126–33. [57] Durão LMP, Gonçalves DJS, Tavares JMRS, Albuquerque VH, Aguiar Vieira A, Torres Marques A. Drilling tool geometry evaluation for reinforced composite laminates. Compos Struct 2010;92: 1545–50. [58] Durão LMP, de Moura MFSF, Marques AT. Numerical prediction of delamination onset in carbon/epoxy composites drilling. Eng Fract Mech 2008;75:2767–78. [59] Durão LMP, de Moura MFSF, Marques AT. Numerical simulation of the drilling process on carbon/epoxy composite laminates. Compos A Appl Sci Manuf 2006;37:1325–33. [60] Díaz-Álvarez J, Olmedo A, Santiuste C, Miguélez MH. Theoretical estimation of thermal effects in drilling of woven carbon fiber composite. Materials 2014;7:4442–4454. [61] López-Puente J, Varas D, Loya JA, Zaera R. Analytical modelling of high velocity impacts of cylindrical projectiles on carbon/epoxy laminates. Compos Part A: Appl Sci.Manuf 2009;40: 1223–1230.

138 | 4 Numerical modeling of LFRP machining [62] Artero-Guerrero JA, Pernas-Sánchez J, López-Puente J, Varas D. On the influence of filling level in CFRP aircraft fuel tank subjected to high velocity impacts. Compos Struct 2014;107:570–577. [63] Santiuste C, Barbero E, Miguélez MH. Computational analysis of temperature effect in composite bolted joints for aeronautical applications. J Reinf Plast Compos 2011;30:3–11. [64] Yashiro T, Ogawa T, Sasahara H. Temperature measurement of cutting tool and machined surface layer in milling of CFRP. Int J Mach Tools Manuf 2013;70:63–69. [65] Villière M, Lecointe D, Sobotka V, Boyard N, Delaunay D. Experimental determination and modelling of thermal conductivity tensor of carbon/epoxy composite. Compos. Part A Appl Sci Manuf 2013;46:60–68. [66] Gralewicz G, Owczarek G, Więcek B. Investigations of single and multilayer structures using lock-in thermography—Possible applications. Int J Occup Saf Ergon 2005;11:211–215.

J. Babu, J. Philip, T. Zacharia, J. Paulo Davim

5 Delamination in composite materials: measurement, assessment and prediction Abstract: Delamination of composites is the dominant machining defect, and its elimination by choosing proper machining parameters can bring in benefits in terms of service reliability and cost reduction. Drilling-induced delamination usually occurs at the start of drilling – called peel-up delamination – and at the exit of the drill, called push-out delamination. The common methods to measure the extent of delamination are briefly described in this chapter. These measurements are used to develop assessment factors, both dimensional and non-dimensional, for comparing delamination damage resulting from different machining methods and parameters. The delamination factors proposed by different researchers are explained while pointing out their merits and demerits. A universally applicable delamination measure is yet to be developed. Modeling techniques such as regression analysis, artificial neural network and FE simulation used by different researchers for predicting the delamination damage are also described.

5.1 Introduction Superior properties such as low density, high strength and stiffness as well as chemical and corrosion resistance widen the field of application for composite materials in various industries such as aerospace, automobile, defense, transport and power generation. These materials are difficult to machine compared to conventional homogenous materials, because of their non-homogeneous, anisotropic nature, being reinforced with abrasive constituents. Delamination is one of the most critical defects that can occur in machining operations of the composites; this can severely affect part performance. Delamination is defined as the separation of plies from each other. It occurs at the entrance of the hole when the drill bit comes in contact with the workpiece and the shearing action forces the first plies to separate from those immediately below. This is called peel-up delamination. When the drill bit nears the exit of the hole, the force exerted by the drill bit on the remaining few plies causes them to deform and separate from each other, which is called push-out delamination. Drilling is the machining operation required to facilitate assembly of composite structures. The rejection rate can be as high as 60 percent because of drilling-induced delamination [1]. Researchers conclude that feed rate is the most influential factor on delamination, and there is a critical thrust force below which there would be no delamination. However, this value differs for different drill bit geometries and materials.

140 | 5 Delamination in composite materials: measurement, assessment and prediction

5.2 Mechanisms of delamination Drilling-induced delamination mainly occurs at the start and the finish of the drilling operation. That is, it occurs at the entrance and exit of the hole, termed peel-up delamination and pull-out delamination respectively. Intermediate layers are not considered to be affected in this way. Both these mechanisms associated with drilling composite laminates are reported in the literature. Figure 5.1 shows a schematic representation of peel-up and push-out delamination.

5.2.1 Peel-up delamination Peel-up delamination occurs at the entrance side of the hole when the cutting edge of the drill abrades the top layers of the laminate, making them to move upwards, and tends to pull these abraded laminates along the flute of the tool. Thus this material spirals up further as drill bit progress downwards. This leads to a force acting upwards which causes separation of initial plies from the uncut laminate plies which are being pushed away by the thrust force of the drill.

5.2.2 Push-out delamination Push-out delamination occurs at the exit side of the hole. Workpiece experiences compressive force in drilling operation. Laminates under the drill tend to be drawn away from inter-laminar bond around the hole. When drilling operation is nearing the end the drill is cutting the last laminate/laminates, the uncut thickness value being small, the low bond strength of this uncut laminates results in separation of plies under the applied deformation force exerted by drill, causing the delamination. In general, the severity of push-out delamination is higher than peel-up delamination. Peel-up delamination

Push-out delamination

D

Fig. 5.1: Schematic representation of peel-up and push-out delamination.

5.3 Measurement of delamination |

141

5.3 Measurement of delamination In order to assess the extent of delamination, the nature and dimensions of the delamination zone must be measured. These measurements are used to develop delamination assessment factors that facilitate the comparison of the influence of the machining parameters on the delamination of different composite materials. In the following sections, the measurement methods of delamination damage on composite materials are described. The methods that are often used by researchers and practitioners to measure delamination damage are: 1. visual methods; 2. image processing; 3. acoustic emission; 4. scanning acoustic microscopy; 5. ultrasonic C-scan; 6. radiography; 7. X-ray computerized tomography; 8. shadow moiré laser interferometry.

5.3.1 Visual methods Toolmaker’s microscopes are generally used to measure the delamination zone. Magnifications of 5× to 30× [2] are standard. A strong light at the rear makes the segregation of the damaged zone possible. Sometimes a dye penetrant is used to clearly demarcate the damaged area. The diameter of the drilled hole and maximum extension of delamination can be directly measured, but the area of damage requires additional techniques such as etching and grid development. The simplicity of this measurement technique and its low cost are the reasons for the wide use of this method for delamination assessment.

5.3.2 Image processing Visual methods clearly fail when delamination of materials such as CFRP are measured due to the difficulty in distinguishing the damaged zone from the rest. Digitization of the visual image and its processing with the aid of a computer vastly improve measurement accuracy. A CCD camera with a built-in digitizer is used to photograph the drilled specimen’s delaminated surface. The digital image data are fed into a computer and manipulated in terms of magnification and sizing. Commercial software such as Corel Draw [2] and others in the public domain can be used for this purpose. The area of the damaged zone is usually assessed by pixelization (division into pic-

142 | 5 Delamination in composite materials: measurement, assessment and prediction ture cells) and multiplying the number of pixels in the damaged zone with the area of a pixel. Stereo-microscopy in conjunction with computer based image processing allows for three-dimensional (3D) visualisation of the damage. A stereomicroscope has a special built-in illumination which reflects off of the uneven surface of the delaminated area. An integrated camera takes the visual images and stores them in a desktop computer. These images are then processed by using commercial software such as “AnalySIS” [3] to measure the delamination. Another method to obtain a digitized image that is widely accepted is to use a high resolution (800 dpi) flat-bed scanner [4]. This acquired image is transported to a computer and processed using any of the accurate image-processing software; commercially available Corel Draw software can give an accuracy of 10−3 mm [2] for measurements. The damaged region can be easily distinguished from the undamaged area as a shadow zone; its measurement can be obtained by using the processing software. An important aspect of computerized image processing is the selection of parameter values for brightness, noise suppression (de-speckle), image enhancement and edge detection [4]. Correct selection of these threshold parameter values is very important to ensure the proper demarcation of the damaged zone and its measurement. Thus the four stages in digital image analysis are: 1. image acquisition using visual imaging cameras, scanning or X-rays; 2. image segmentation by applying proper parameter threshold values to clearly distinguish the delaminated zone from the rest; 3. image processing using processing software in a computer to obtain the measurements of the damaged region such as perimeter, area, texture and other dimensions; 4. pattern recognition using strategies to clearly identify the regions of interest in delamination assessment.

5.3.3 Acoustic emission When materials fracture under stress, the energy release occurs as an emission of acoustic waves. The root mean square (RMS) value of the acoustic waves so released is the measure of the stress severity. Thus a correlation can be established between the thrust force of drilling composite parts and the acoustic emission, an abrupt peak in the RMS value of acoustic emission indicating the occurrence of delamination [5]. The number of acoustic pulses can be monitored to predict the extent of delamination. This feature allows real-time monitoring of machining operations to control the delamination of composite materials. Experiments by different researchers show a linear relationship between the area of delamination damage and acoustic energy released [6]. For different composite materials under specific machining parameters, a quantitative assessment of delamina-

5.3 Measurement of delamination

|

143

tion can be made by measuring the acoustic emission signal strength (RMS value) once the proportionality constants are determined experimentally. However, this is not considered to be an accurate method for measuring delamination damage [5], though it has the advantage of on-line measurement.

5.3.4 Scanning acoustic microscopy (SAM) The ultrasonic C-scan method has the limitation of high-energy dissipation due to absorption in the matrix and scattering by fibers of the composite material. It also suffers from the shadow effect due to delamination of the surface shadowing defects located further away [7]. Acoustic microscopy is more sensitive than the C-scan method in this respect. There are two types of acoustic microscopes: the scanning laser acoustic microscope (SLAM) and scanning acoustic microscope. The SLAM uses a through-transmission ultrasonic method, where the signal that comes through the specimen is scanned by the scanning laser detector. The detector provides the images of the internal features such as defects and discontinuities based on the strength levels of the detected ultrasound. The working principle of the SLAM is illustrated in Fig. 5.2. Scanning laser detector Fluid couplant Sample

Ultrasound Fig. 5.2: Working principle of SLAM.

A fine ultrasonic beam is used to scan the specimen line by line in the case of SAM. A lens is used to focus the ultrasonic waves on a spot in the specimen, and the reflected waves are detected. These received echoes are analyzed and stored; this creates an image of the area scanned. From these stored data, a complete image is developed, as in a scanning electron microscope. The working principle of the SAM is illustrated in Fig. 5.3.

5.3.5 Ultrasonic C-scan Ultrasonic C-scan is a frequently used method for measuring delamination damage in composite parts caused by drilling [8]. The specimen to be inspected is impinged with ultrasonic waves. The attenuation of the waves as they pass through or reflected by the specimen depends on the impediments the waves encounter, which is a function

144 | 5 Delamination in composite materials: measurement, assessment and prediction Input

Echo

Transducer

Fluid couplant Specimen

Fig. 5.3: Working principle of SAM.

of its structure and configuration. The ultrasonic waves from the transmitter scan the specimen; their echo signals are received at the receiver with a digital oscilloscope and the attenuation of the waves are recorded. These data are then processed in a computer to reconstruct the delamination damage. Fig. 5.4 shows a schematic arrangement for ultrasonic C-scan. Composite materials have a heterogeneous structure that causes the ultrasonic waves to attenuate and scatter. This calls for very careful C-scan data interpretation. By proper selection of the ultrasonic probe and other settings of the emitter and receiver, good results of the size and shape of the delamination can be obtained with ultrasonic C-scanning [9]. Ultrasonic probe

Computer

Printer

Specimen

Fig. 5.4: Schematic arrangement for ultrasonic C-scan.

5.3.6 Radiography Non-destructive inspection by obtaining images of the parts using X-ray or 𝛾-ray radiation is a very common method to measure delamination damage on composite materials [9]. Usually the component to be inspected is dipped in or brushed with a chemical that is opaque to radiation to improve the image’s contrast. Di-iodomethane and tetrabromoethane are two such penetrant chemicals [10]. Radiographic inspection is carried out in a protective enclosure such as a lead container, and the acquired images are digitized for measurement by processing in a computer. Figure 5.5 shows a schematic arrangement for radiography. Images of

5.3 Measurement of delamination

Lead container

|

Computer

Printer

Image processing

Results

145

Image

Specimen

Digitization

X-ray emission

Image acquisition

Fig. 5.5: Schematic arrangement for radiography.

the component before and after drilling are obtained so that pre-existing defects can be eliminated before the delamination damage is inspected. A threshold value of the grey scale is chosen either manually or automatically for which many techniques are available, so as to demarcate the damaged portion [10]. Using commercially available image-processing software, the required dimensions of the delamination can then be obtained for calculating the delamination factor.

5.3.7 X-ray computerized tomography X-ray computerized transverse axial scanning (tomography) is a widely used medical diagnostic tool that provides images of different slices of the body part being examined [11]. Such equipment can be adapted to attain images of drilled composite specimens that can then be digitally processed for delamination data [8]. A schematic diagram of an X-ray computerized tomography arrangement is shown in Fig. 5.6. A common frame carries the X-ray tube, detectors and collimators. The specimen is X-rayed through the edges of the series of its slices, rotating the frame 180° in steps of 1°, the radiation absorption (attenuation) being detected and recorded at each step [11]. These data are digitally processed to develop a density profile of the slice. The image can be viewed on the computer screen, its dimensions assessed and the image and its data can be printed. X-ray computerized tomography is an accurate and quick method for visualizing and measuring peel-up, push-out and internal delamination that can occur due to drilling composite components [3].

5.3.8 Shadow moiré interferometry Moiré means light and dark bands. Shadow moiré is an interferometric method where a reference grating and its shadow on the test specimen cause interference of the light rays falling on them; this results in light and dark fringes [12]. This fringe pattern depends on the distance of the specimen surface from the grid and thus provides the con-

146 | 5 Delamination in composite materials: measurement, assessment and prediction

180° rotation in steps of 1°

Equipment frame

e ub yt a r X-

Computer

Printer

en

Direction of slicing

cim pe

S

Image processing

Delamination image/data

rs cto

te

De X-ray screening

Fig. 5.6: Schematic arrangement for X-ray computerized tomography.

tours of the surface [13]. The fringe order of the pattern at any point is proportional to the deviation of the point from the plane of the specimen surface. This principle is used to measure the delamination resulting from machining of composite components. The fringe pattern can be assessed by introducing a phase-shifting device such as a piezo-electric transducer in the light path and measuring the light intensities of the pattern on the specimen surface [14]. This can then be developed in a threedimensional contour, thus giving data on the extent of the delamination. The arrangement for a shadow moiré phase-shifting interferometry is shown in Fig. 5.7. The fringe pattern is photographed using a CCD camera, and the image data is fed to a computer to analyze and develop the three-dimensional contour of the specimen surface. The out-of-plane region of the surface represents the delamination. The area of delamination is calculated from the number of pixels in the delaminated region. CCD camera Laser

Collimating lens Master grating Phase shifting piezoelectric transduser Drilled specimen (showing fringe pattern)

Fig. 5.7: Schematic arrangement for shadow moiré phase-shifting interferometry.

5.4 Assessment of delamination |

147

5.4 Assessment of delamination Different researchers used different techniques to quantify delamination damage. Abrao A. M. et al. [15] first presented the principal parameters used by researchers to evaluate delamination quantitatively. One group of researchers used dimensional parameters: delamination area or length. Another group of researchers used dimensionless parameters: ratio of maximum diameter of the delamination zone to the hole diameter; ratio of drill radius to the delamination radius; ratio of damaged area to the hole area; and similar ratios. The ratio concept allows comparisons among the results obtained by researchers; Fig. 5.8 presents the principal parameters used by various researchers to evaluate delamination. The methods used by them for the assessment of delamination are tabulated in Tab. 5.1. The various methods of assessing delamination are given in detail in the following sections. Assessment of delamination

Dimensional parameters

One dimensional parameters • Difference between the maximum damage radius and drilled hole radius • Difference between the maximum damage diameter and drilled hole diameter • Average of two perpendicular meassurements of the diameter of the damage • Sum of the lengths of the internal cracks

Two dimensional parameters Delamination area

Non dimensional parameters

Single feature

Combined features

• Ratio of drill radius to the delamination radius

• Sum of diameter and area ratios

• Ratio of damage area to hole area

• Ratio of damaged area to square of its perimeter

• Ratio of maximum diameter of the delamination zone to the hole diameter

• Diameter ratio modified with weighted severity of damage

• Ratio of equivalent diameter to the hole diameter • Ratio of diameter of minimum enclosing area to the hole diameter

Fig. 5.8: Principal parameters used for assessment of delamination.

5.4.1 Delamination factor/conventional delamination factor Chen [16] was the first to propose a comparison factor for evaluating and analyzing the extent of delamination damage in carbon fiber reinforced composite laminates. This ratio was called delamination factor (Fd ). Delamination factor is defined as the ratio of maximum diameter (Dmax ) in the delaminated zone to the nominal diameter of the hole (D), which is generally taken as the nominal diameter of the drill bit as shown in

148 | 5 Delamination in composite materials: measurement, assessment and prediction Tab. 5.1: Different methods of delamination assessment. Sl. Evaluation of No. delamination factor

Formula used

Author, reference

Dmax DO

1 Conventional delamination factor (Fd )

Fd =

2 Delamination size

Delamination size = Rmax − R

3 Two-dimensional delamination factor (Fa )

Fa = (

4 Damage ratio

DRAT =

5 Delamination factor

Fd =

6 Adjusted delamination factor (Fda )

Fda = Fd +

7 Effective delamination factor (Fed )

Fed =

De ; D0

8 Refined delamination factor (FRd )

FDR =

Dmax DO

9 Shape’s circularity 10 Minimum delamination factor

Chen [16] U. A. Khasaba [17]

Ad )% Anom

Faraz et al. [18]

DMAR AAVG

Mehta et al. [19]

Ad A

Mohan et al. [20] Ad (Fd2 − Fd )

Davim et al. [4]

Amax − AO 0.5

De = [

4(Ad + A0 ) ] π A

Tsao et al. [21] A

+ 1.783 ( AH ) + 0.7156 ( AM ) O

+ A f = 4π 2 P D Fdmin = min D0

A 3 0.03692 ( A L ) O

O

2

V. A. Nagarajan et al. [22] Durão et al. [23] Duarte Nuno Rodrigues da Silva [24]

Fig. 5.9. Equation (5.1) represents the formula for the delamination factor. D (5.1) Fd = max D Several authors used this formula to assess the delamination damage because of its simplicity and ease of calculation; recent research papers also used this method. There is no validation of this method with the experimental strength values of the joints made after drilling. The explanation can be that the strength of the joint may depend on the maximum crack length in the vicinity of the joint; this is the maximum diameter (Dmax ) in the delaminated zone. Though it is a good measure of delamination in

D max D Fig. 5.9: Schematic representation of conventional delamination factor.

5.4 Assessment of delamination |

D max

D max

D

D max

D

(a)

(b)

D max

D

(c)

D max

D

(d)

149

D

(e)

Fig. 5.10: Schematic representation of various delamination damages with the same conventional delamination factor.

general, it can lead to wrong assessments in certain cases. Figure 5.10 (a) to (e) illustrates different delamination damages, all giving the same value for the delamination factor. Obviously, damage shown in Fig. 5.10 (e) is more severe than others.

5.4.2 Delamination size U. A. Khasaba [17] assessed the damage around the hole by delamination size, which is defined as the difference between the radius of maximum damage and that of the drilled hole as shown schematically in Fig. 5.11. Equation (5.2) represents the formula to determine delamination size. It is a dimensional quantity and has units, generally measured in mm. Delamination size = Rmax − R (5.2)

R max R Fig. 5.11: Schematic representation of delamination size.

This method is also simple and easy to calculate, and it differentiates delamination damages with different Rmax values; the greater the value, the greater the damage. This method, which is calculated from the same measurements as the delamination factor, has the same disadvantage.

5.4.3 Two-dimensional delamination factor (Fd ) The two methods – namely conventional delamination factor and delamination size – may not give complete information of delamination damage, because the extent of

150 | 5 Delamination in composite materials: measurement, assessment and prediction delamination caused by a few fibers pushed down or peeled up to a particular width may not give the real delamination zone around the drilled hole. This was pointed out by Faraz et al. [18], and they proposed the two-dimensional delamination factor (Fa ) to assess the level of delamination damage. Equation (5.3) represents the formula to determine the two-dimensional delamination factor, which is schematically shown in Fig. 5.12. A Fa = ( d ) % (5.3) Anom Ad

D Fig. 5.12: Schematic representation of two- dimensional delamination factor.

This method considers the damaged area in the assessment of delamination damage. The damaged area (Ad ) indicates the extent of delamination damage. This is a refinement over the previous methods for assessing delamination, as it distinguishes the severity of the delamination damages with different damage areas. But it cannot differentiate the delamination damage between one that is uniformly deformed in the vicinity of the hole and another that has deeper and longer cracks, even though both have the same damage area as shown in Fig. 5.13. Both the figures show the same area of delamination and hence the same Fa values, but the Fig. 5.13 (b) specimen, which has deeper and longer cracks and thus a lower strength of the joint, is likely to fail at a lower load application than the specimen shown in Fig. 5.13 (a). The effect of crack length on delamination damage is not included in this measure, as only delaminated area is used for calculating the ratio. Ad 1

Ad 2

D

(a)

D

Ad 1 = Ad 2 (b)

Fig. 5.13: Schematic representation of different delamination damages with the same two-dimensional delamination factor.

5.4.4 Damage ratio Another method similar to the two-dimensional delamination factor was suggested by Mehta et al. [19]. This is damage ratio DRAT , which is defined as the ratio of hole periph-

5.4 Assessment of delamination

|

151

eral damage area DMAR to nominal drilled hole area AAVG . This delamination damage is evaluated by measuring the area with the use of digitized images of the damaged region. Equation (5.4) represents the formula for calculating the damage ratio. DRAT =

DMAR AAVG

(5.4)

The only difference in this method from the two-dimensional delamination factor is that it is expressed as a factor instead of a percentage. This method also gives better results for delamination damages that have different damage areas. But it cannot differentiate between delamination damages that have same area but in which one is uniformly deformed in the vicinity of the hole and the other has deeper and longer cracks as mentioned above.

5.4.5 Delamination factor N. S. Mohan et al. [20] suggested a similar ratio named as the delamination factor (Fd ). It is defined as the ratio of delaminated area to nominal hole area. Equation (5.5) represents the formula for calculating the delamination factor. Fd =

Ad A

(5.5)

The variation of this method with a two-dimensional delamination factor and damage ratio is in the calculation of the Ad value. The authors defined Ad as the area of the envelope of damage area, including the hole area. This method may also suffer from the same disadvantages as those of two-dimensional delamination factors and damage ratios, as these methods consider only the damaged area to evaluate the delamination factor.

5.4.6 Adjusted delamination factor As mentioned earlier, the conventional delamination factor and delamination size are not appropriate because the size of delamination may not fully represent the severity of damage. They do not include the damage area and two- dimensional delamination factor, and the damage ratio methods also do not fully represent the magnitude of damage – these methods do not consider the crack length while assessing the delamination factor. Taking both the crack length and the damage area into consideration, Davim et al. [4] proposed a novel approach to measure the delamination factor and called it the adjusted delamination factor (Fda ). Equation (5.6) represents the formula to determine the adjusted delamination factor, which is shown schematically in Fig. 5.14. The first part of equation (5.6) represents the maximum crack length contribution (conventional delamination factor and delamination size). The second part of

152 | 5 Delamination in composite materials: measurement, assessment and prediction equation (5.6) represents the damage area contribution (two-dimensional delamination factor and damage ratio) in delamination. Fda = Fd + Amax A0

Ad (Fd2 − Fd ) Amax − AO

(5.6)

= Delamination area related to Dmax = Drilled area of D.

Ad D max D Fig. 5.14: Schematic representation of adjusted delamination factor.

This method provides a better approach when compared to the methods mentioned above, as it takes care of both the damaged area contribution and maximum crack length contribution in the assessment of delamination. This method differentiates delamination damage for the cases shown in Fig. 5.15 (a) and (b), and Fda values will be quite different for the two cases, even though Fd values are the same, as the areas of delamination damage are different. This method also differentiates delamination damage for the cases shown in Fig. 5.15 (c) and (d); Fda values will be quite different for the two cases, even though Fa value and damage ratio are the same (as the area of delamination damage is the same). This is due to different Dmax /Rmax values which effect the value of Fda and hence change the Fda values. This method may not differentiate the delamination damages shown in Fig. 5.15 (a) and (e). Figure 5.15 (e), which consists of very deep cracks may constitute very negligible area; therefore the values of adjusted delamination factors for both Fig. 5.15 (a) and (e) can appear almost the same. But the Fig. 5.15(e) specimen will be more prone to fail at much lower load applications than the Fig. 5.15 (a) specimen during service conditions as it is more severely damaged with deeper cracks. In this method, the higher the damage on Ad , the higher the effect on Fda .

5.4.7 Equivalent delamination factor It is advantageous to use the adjusted delamination factor for practical applications, as it takes into consideration both the crack length and damage area contribution in the assessment of delamination. The additive nature of the Fda formula leads to a higher value of the delamination factor than the actual as explained by C. C. Tsao et al. [21]. Therefore, they proposed a new approach to determine the delamination factor that is called the equivalent delamination factor (Fed ). The schematic representation of

5.4 Assessment of delamination |

Ad ~ 0

Ad

D max

Ad 1

D max

D

D max 1

D

(a)

(b)

Ad 2

Ad ~ 0

D max 2

D

D max

D

(c)

(d)

153

D

(e)

D max 1 < D max 2 Ad 1 = Ad 2 Fig. 5.15: Schematic representation of adjusted delamination factor with different damage profiles.

the equivalent diameter is shown in Fig. 5.16. Equation (5.7) represents the formula to determine Fed , where equivalent diameter (De ) is calculated as given in equation (5.8), and D0 is the nominal hole diameter. Fed =

De D0

De = [

4(Ad + A0 ) ] Π

(5.7) 0.5

(5.8)

Ad De

Do Fig. 5.16: Schematic representation of equivalent delamination factor.

Fed is a more refined method for assessing delamination when compared to the conventional delamination factor and delamination size. This method is similar to the two-dimensional delamination factor and damage ratio because it considers the area of delamination damage in the assessment of the delamination factor. The main difference between this method and the two-dimensional delamination factor or damage ratio is that this method converts the two-dimensional area into equivalent onedimensional diameter to calculate the assessing parameter. This method differentiates clearly different delamination damages that have the same Fd values but different Ad values. This method cannot differentiate between the delamination damages shown in Fig. 5.17 (a) and (b). Figure 5.17 (a) and (b) represent the same delaminated area and therefore will have the same equivalent diameter De ; hence Fed values are also the same, but the Fig. 5.17 (b) specimen is severely damaged with more cracks and therefore more prone to fail at lower load application than the Fig. 5.17 (a) specimen. This method only considers the damaged area contribution on delamination damage. Crack contribution on delamination damage is not taken into consideration in this method.

154 | 5 Delamination in composite materials: measurement, assessment and prediction Ad 1

Ad 2

De 1

De 2

Do

Do

Ad 1 = Ad 2 (a)

De 1 = De 2 (b)

Fig. 5.17: Schematic representation of different damages with the same equivalent delamination factor.

5.4.8 Refined delamination factor (FDR ) The methods described so far do not consider the severity levels of delamination damage in assessing delamination. Therefore, V. A. Nagarajan et al. [22] proposed a method that takes the severity levels of the damage into account. This method divides the total damage area into three different damage zones: heavily damaged area (AH ); medium damaged area (AM ); and low damaged area (AL ). By using Buckingham’s π theorem, they expressed the refined delamination factor (FDR ). Equation (5.9) represents the formula to determine refined delamination factor. This method uses neural networks in MATLAB for processing the captured delamination images in order to identify different damage zones and to calculate the various parameters mentioned in the equation. FDR =

Dmax A A 3 A 2 + 1.783 ( H ) + 0.7156 ( M ) + 0.03692 ( L ) DO AO AO AO

(5.9)

This method is more refined than other methods, as it considers all the influencing factors to assess delamination. The authors validated their results by conducting strength tests on the joints made after drilling. However, they limited their validation tests to support their method at a spindle speed of 1400 rpm only. It would have been better if the authors had conducted more strength tests with samples drilled at all the selected spindle speeds to validate their method of assessing the delamination factor. It is difficult to differentiate three zones within the delaminated area, and the parameters used in assessment of delamination are more sensitive to the selected threshold values in the processing stage using digital image processing.

5.4.9 Shape circularity ( f ) Another way of assessing the delamination factor by considering the shape of delamination damage was suggested by Durão, Tavares, Albuquerque, et al. [23]. They defined a new term, shape circularity, which is the shape’s compactness compared to a circle of equal perimeter. Equation (5.9) represents the formula to determine shape circularity. A f = 4π 2 (5.10) P

5.4 Assessment of delamination |

155

The value of shape circularity will be nearly 1 if the damage pattern resembles a circle. When the value approaches 0, the damage pattern will be an elongated polygon. They concluded from their research that the circularity of the damaged region was higher and closer to a circular shape for holes drilled by HSS drills; there was no influence of tool geometry on the shape parameter for holes drilled by using carbide drills. Finally, they concluded that there was no correlation between circularity and damage extension on bearing strength. This gives an indication that shape circularity may not be an appropriate method of delamination assessment.

5.4.10 Minimum delamination factor Another approach similar to Chen’s method to quantify the drilling-induced damage proposed by Duarte Nuno Rodrigues da Silva [24] is the minimum delamination factor (Fdmin ). He believed that the concept of the delamination factor was to satisfy the need for an easy comparison of different delamination damages that resulted from different drilling methodologies and conditions. For the evaluation of the damage quantity, the damaged shape is irrelevant. His main focus of delamination measurement was in assessing the smallest area that contained all of the damage that resulted from the drilling process. This concept is very similar to the one devised by Chen, but the main focus is to minimize the calculated factor by considering the smallest area enclosing the delamination damage which can be obtained by neglecting the concentricity between the drilled hole and the damage-enclosing area. Enclosing the area’s center should coincide with the affected area’s center, which may or may not be the center of the drilled hole. The minimum delamination factor is defined as a ratio between the diameter of the minimum enclosing area (Dmin ) and the pretended drill area (D0 ). This is schematically shown in Fig. 5.18. Equation (5.11) represents the formula for calculating the minimum delamination factor. Fdmin =

Dmin D0

D min Do D max Fig. 5.18: Schematic representation of minimum delamination factor.

(5.11)

156 | 5 Delamination in composite materials: measurement, assessment and prediction Fdmin gives the minimum drilling-induced damage, and at the same time it allows a better comparison between different drilling damages resulting from different drilling conditions. When the damage is a regular crown, the center of the crown will be very close to the drilling center; then Fdmin and Fd values do not deviate much from each other. But this method gives better results than Chen’s results for irregular areas. This method also did not consider the damage area contribution in the assessment of delamination, and thus it suffers from the same disadvantages as those of the conventional delamination factor and delamination size. One can conclude from the available literature that several authors used different approaches to assess the delamination factor. It can also be seen that most of the authors used the conventional delamination factor for the assessment of delamination, because it is easy to understand and calculate. Most of the empirical models for drilling-induced delamination by various researchers were obtained by using linear regression analysis for the conventional delamination factor (Fd ). The methods listed at [6–8, 10] appear to be refined methods for assessing the delamination when compared with the methods [1–5, 9], but each one has its own merits and de-merits. There is no common agreement and consistency of use for the various methods. Further refinements in delamination measures can be expected until a universally accepted method is achieved.

5.5 Delamination in milling Milling is one of the machining/finishing operations for removing extra material from the workpiece to obtain a desired surface quality with dimensional tolerances. Milling of composite materials produces defects such as surface delamination; its severity depends on the cutting parameters used during the milling process. Delamination in milling is assessed by a delamination factor [25], which is defined as the ratio of the width of maximum damage, (Wmax ), and the nominal width of the cut (W) as shown schematically in Fig. 5.19. Equation (5.12) represents the formula for calculating delamination factor. W Fd = max (5.12) W Wmax = width of maximum damage in mm W = nominal width of the cut in mm. Sometimes milling can also be used to make holes in composite materials. Recent research [26] shows that helical milling of CFRP results in much lower delamination than that in other processing methods including high-speed drilling. This is because helical milling reduces the thrust force, and more space is available to expel the chips. This results in less friction between the workpiece and tool. Helical milling is a more recently developed hole-making process that involves the rotation of a cutting tool around its own axis and simultaneously about a central axis set off from the axis of

5.6 Numerical prediction of delamination

Damage

|

157

W

W max

Feed Fig. 5.19: Schematic representation of delamination factor in milling.

the cutting tool; this causes the cutting tool feed to be along a helix [27]. The delamination factor can be determined using the same equation (5.1) as the conventional delamination factor used in drilling, viz. D Fd = max D

5.6 Numerical prediction of delamination Delamination can be predicted by numerical analysis using regression analysis, neural networks/fuzzy logics and finite element simulation methods. Regression analysis and artificial neural networks use the experimental output data of the delamination factor in the selected input parameters to establish an empirical model to determine delamination in terms of the process parameters. Simulation methods use the workpiece and cutting tool properties and machining conditions.

5.6.1 Regression analysis Regression analysis is a statistical technique that gives the relation between dependent variables and independent variables. In machining composite materials, input parameters can be speed, feed, tool bit geometry and material, workpiece thickness and material as well as fiber orientation. Empirical models developed for the prediction of delamination are classified as: 1. with one input parameter; 2. with two input parameters; 3. with three input parameters; 4. with four input parameters.

158 | 5 Delamination in composite materials: measurement, assessment and prediction 1.

Initial empirical models for delamination were given in terms of average thrust force FZ ) [16]. 2. Later empirical models for delamination were given in terms of two variables: cutting speed and feed rate in a linear form [28], in an exponential form [29] and with the consideration of the interaction effect [30]. 3. Other empirical models for delamination were given in terms of three variables: cutting speed, feed rate and diameter of the drill bit [31]/tool pre-wear [32]. Later empirical models for delamination were given in terms of three input parameters and interaction effects of the parameters: cutting speed, feed rate and drill bit geometry (point angle) [33]. 4. Recent empirical models for delamination were given in terms of four input parameters: cutting speed, feed rate, diameter of the drill bit, fiber orientation angle and the interaction effect of these parameters [34].

5.6.2 Artificial neural network (ANN) The artificial neural network is based on the functional aspects and structure of biological neurons. This method can be effectively used to model complex non-linear relationships between input and output parameters. Forward neural network (FNN) with multiple layers can be used for modeling the drilling of composites. To estimate delamination in composite materials, input parameters can be feed rate and spindle speed, and output can be the delamination factor. Fixed/flexible sigmodal functions can be used; a flexible sigmodal function has better stability and faster learning ability when compared to a fixed sigmodal function. But a fixed sigmodal function gives less converging time. A delamination prediction/estimation in CFRP by using three-layer ANN is presented by Karnik et al. [35] with three input neurons (spindle speed, feed rate and drill bit point angle), 12 hidden layer neurons and one output neuron (delamination factor). The model is trained by an error-back propagation algorithm using an experimental database as an input and output pattern. It has been observed that the maximum error between the experimental and predicted delamination factor was 1.6 %. A delamination prediction/estimation in GFRP by using three-layer ANN is presented by Mishra et al. [36] with four input parameters: spindle speed; feed rate; drill diameter; and drill point geometry. They use 22 hidden layers and experimental values of input and output parameters for training and testing the network. They find that the error to be less than five per cent for most of the cases, and the maximum and minimum absolute error for training patterns to be 12.7 % and 0.1 %, respectively. The modeling and control of the entire drilling of composite material are well presented in a review paper by Amrinder Pal Singh et al. [37].

5.6 Numerical prediction of delamination |

159

5.6.3 FE simulation methods Finite-element techniques with an appropriate numerical model, which could give complete descriptions of the drilling process in composite materials, can predict the delamination damage and other parameters such as thrust force and torque. It is necessary to use exact experimental conditions while constructing and meshing the model. It is also necessary to consider some assumptions in order to simplify the model to save cost and time for analysis. This may be the cause of the observed variation between experimental and simulation output values. To increase the precision of results, selection of an accurate model, mesh size and shape are needed. Initial numerical models [38–42] and the FE model [43, 44] use linear elastic fracture mechanics (LEFM) theory to determine the critical thrust force at the onset of delamination during drilling; no delamination occurs below this force. The first model was presented by Hocheng and Dharan [38]; later, other models were developed [39–42]. These analytical models consider only inter-laminar fractures during drilling. Most of the models assumed the delamination shape to be elliptical. FEM models that represent a numerical prediction of delamination onset in carbon/epoxy composite drilling were presented by Duarao LMP et al. [44] using mixed-mode damage propagation. Recent models with a FE simulation of drilling of composite laminates could predict the delamination factor fairly accurately. Ozden Isbilar [45] could predict the delamination factor in CFRP composite laminates by using a progressive failure model (intra-laminar failure) and progressive delamination model (inter-laminar failure). They used Hashin’s theory [46] for intralaminar damage initiation criteria and assumed initiation behavior to be orthotropic. Fiber compression and tension, matrix compression and tension were considered to be damage-initiation mechanisms. They used a zero-thickness, surface-based cohesive behavior approach which is similar to the cohesive element approach proposed by Alfano and Crisfield [47]; they found that simulation value of the delamination factor was 1.2 against experimental value of 1.3, with an error of 8 %. V. A. Phandis et al. [48] developed a (3D) finite element model of drilling in composite laminates, accounting for complex kinematics at the drill-workpiece interface by using cohesive zone elements to simulate inter-ply delamination in a composite. The values of the delamination factor at the entry of drilling predicted by the FE simulation method are very close to the experimental values, whereas in the case of exit the FE simulation values are over-estimated when compared to experimental values. Ongoing research in this field is expected to further fine-tune modeling for a more accurate prediction of delamination in composites.

160 | 5 Delamination in composite materials: measurement, assessment and prediction

5.7 Summary Delamination is a critical drilling defect on composite materials that can lead to serious service failures. Many methods that vary in simplicity, accuracy and cost are employed to measure delamination damage. They range from visual methods, computerized image processing and radiography to acoustic emission monitoring, computerized X-ray tomography and laser interferometry. Familiarity with the different methods allows one to select a suitable measurement technique for a particular application. The various delamination assessment factors proposed by different researchers provide the basis for comparing machining parameters for drilling different composites. An understanding of the type and severity of damage reflected by their values is necessary for their proper implementation. The predictive models for delamination are useful in determining machining parameters.

References [1] [2] [3] [4]

[5]

[6]

[7] [8]

[9] [10] [11] [12] [13]

Wong TL, Wu SM, Croy GM, An analysis of delamination in drilling composite materials. In: 14th National SAMPE technology conference. Atlanta, GA, USA;1982:471–483. Khashaba UA. Delamination in drilling GFR-thermoset composites. Compos Struct 2004;63: 313–327. Faraz A, Biermann D, Weinert K. Cutting edge rounding: an innovative tool wear criterion in drilling CFRP composite laminates. Int J Mach Tools Manuf 2009;49:1185–1196. Davim JP, Campos Rubio JC, Abrão AM. A novel approach based on digital image analysis to evaluate the delamination factor after drilling composite laminates. Compos Sci and Technol 2007;67:1939–1945. Xiaojiang Cai, Sheng Qin, Hongzhou Zhang, Shu Han and Ming Chen,. Experimental analysis of on delamination damage by acoustic emission in high speed drilling of carbon fibre reinforced plastics. Key Engineering Materials 2014;589–590:287–292. Hocheng H, Tsao CC. Micro-mechanisms of delamination in composite materials induced by drilling. Wiley Encyclopedia of Composites, Second Edition. Edited by Luigi Nicolais and Assunta Borzacchiello. John Wiley & Sons, Inc,2012:1–16 Gao SL, Kim JK. Scanning acoustic microscopy as a tool for quantitative characterisation of damage in CFRPs. Composites Science and Technology 1999;59:345–354. Tsao CC, Hocheng H. Computerized tomography and C-Scan for measuring delamination in the drilling of composite materials using various drills. Int J Mach Tools Manuf 2005;45: 1282–1287. Duarte S, Teixeira P, Machado CM. Methodology analysis for evaluation of drilling-induced damage in composites. Int J Adv Manuf Technol 2014;71:1919–1928. Chen WC. Some experimental investigations in the drilling of carbon fibre reinforced plastics (CFRP) composite laminates Int J Mach Tools Manuf 1997;37(8):1097–1108. Hounsfield GN. Computerised transverse axial scanning(tomography) ; Part 1; Description of system. British Journal of Radiography 1973;46:1016–1022. Robert FA. An introduction to Moiré method with applications in composite materials. US Army Materials Technology Laboratory;August 1992. Wang WC. Moiré Method, University of Washington, Lecture ME 557:1–77.

References | 161

[14] Seif MA, Khashaba UA, Rojas-Oviedo R. Measuring delamination in carbon/epoxy composites using a shadow moiré laser based imaging technique. Compos Struct 2007;79:113–118. [15] Abrao AM, Faria PE, Campos Rubio JC, Reis P, Davim JP. Drilling of fiber reinforced plastics: a review. J Mater Process Technol 2007;186:1–7. [16] Chen WC. Some experimental investigations in the drilling of carbon fiber-reinforced plastic (CFRP) composite laminates. Int J Mach Tools Manuf 1997;37(8):1097–1108. [17] Sonbaty E, Khashaba UA, Machaly T. Factors affecting the machinability of GFR/epoxy composites. Compos Struct 2004;63:329–338. [18] Faraz A, Biermann D, Weinert K. Cutting edge rounding: an innovative tool wear criterion in drilling CFRP composite laminates. Int J Mach Tools Manuf 2009;49:1185–96. [19] Mehta M, Reinhart TJ, Soni AH. Effect of fastener hole drilling anomalies on structural integrity of PMR-15/Gr composite laminates. Proc. of the Machining Composite Materials Symposium, ASM Materials Week, 1992:113-126. [20] Mohan NS, Kulkarni SM, Ramachandra A. Delamination analysis in drilling process of glass fiber reinforced plastic (GFRP) composite materials. J Mater Process Technol 2007;186: 265–271. [21] Tsao CC, Kuo KL, Hsu IC. Evaluation of a novel approach to a delamination factor after drilling composite laminates using a core-saw drill. Int J Adv Manuf Technol 2012;59:617–622. [22] Nagarajan VA, Selwin Rajadurai J, Annil Kumar T. A digital image analysis to evaluate delamination factor for wind turbine composite laminate blade. Composites:Part B 2012;43:3153–59. [23] Durão LMP, Tavares MRS, de Albuquerque VHC. Damage evaluation of drilled carbon/epoxy laminates based on area assessment methods. Compos Struct 2013;96:576–583. [24] Duarte Nuno Rodrigues da Silva “Image processing methodology for assessment of drilling induced damage in CFRP” Thesis 2013, Faculdade de Ciências e Tecnologia, Universidade Nova de Lisboa. [25] Davim JP, Pedro R. Conceico Antonio C. A study on milling of glass fiber reinforced plastics manufactured by hand-lay up using statistical analysis (ANOVA). Compos Struct 2004;64: 493–500. [26] Qin X, Wang B, Wang G, Li H, Jiang Y, Zhang X. Delamination analysis of the helical milling of carbon fiber-reinforced plastics by using the artificial neural network model. Journal of Mechanical Science and Technology 2014;28(2):713–719. [27] Ni WY. Orbital drilling of aerospace materials, SAE International,2007;1;3814–3822. [28] Davim JP, Reis P. Drilling carbon fiber reinforced plastics manufactured by autoclave – experimental and statistical study. Mater Des 2003;24:315–24. [29] Sardinas RQ, Reis P, Davim JP. Multi-objective optimization of cutting parameters for drilling laminate composite materials by using genetic algorithms. Compos Sci Technol 2006;66: 3083–1088. [30] Krishnaraj V, Prabukarthi A, Ramanathan A, Elanghovan N, Kumar MS, Zitoune R, Davim JP. Optimization of machining parameters at high speed drilling of carbon fiber reinforced plastic (CFRP) laminates Composites: Part B 2012;43:1791–1799. [31] Tsao CC, Hocheng H. Taguchi analysis of delamination associated with various drill bits in drilling of composite material. Int J Mach Tools Manuf 2004;44:1085–1090. [32] Khashaba UA, EI-Sobaty IA, Selmy AI, Megahed AA. Machinability analysis in drilling woven GFR/epoxy composites: part II – effect of drill wear. Composites: Part A 2010;41:1130–1137. [33] Gaitonde VN, Karnik SR, Campos Rubio JC, Esteves Correia A, Abrao AM, Paulo Davim J. Analysis of parametric influence on delamination in high-speed drilling of carbon fiber reinforced plastic composites. J Mater Process Technol 2008;203:431–438. [34] Rajamurugan TV, Shanmugam K , Palanikumar K, Analysis of delamination in drilling glass fiber reinforced polyestercomposites. Mater and Des 2012;45:80–87.

162 | 5 Delamination in composite materials: measurement, assessment and prediction [35] Karnik SR, Gaitonde VN, Campos Rubio JC, Esteves Correia A, Abrão AM, Davim JP. Delamination analysis in high speed drilling of carbon fiber reinforced plastics (CFRP) using artificial neural network model. Mater Des 2008;29:1768–1776. [36] Mishra R, Malik J, Singh I. Prediction of drilling-induced damage in unidirectional glass-fibrereinforced plastic laminates using an artificial neural network. Proc Inst Mech Eng Part B: J Eng Manuf 2010;224(5):733–738. [37] Singh AP, Sharma M, Singh I. A review of modeling and control during drilling of fiber reinforced plastic composites, Composites: Part B 2013;47:118–125. [38] Hocheng H, Dharan CKH. Delamination during drilling in composite laminates. J Engng Indus 1990;112:236–239. [39] Piquet R, Lachaud F, Ferret B, Swider P. Étude analytique et expérimentale du perçage de plaques minces en carbone/époxy. Mech Indus 2000;1:105–111. [40] Zhang LB, Wang LJ, Liu XY. A mechanical model for predicting critical thrust forces in drilling composite laminates. Proc Inst Mech Engrs B 2001;215:135–146. [41] Jain S, Yang DCH. Effects of feed rate and chisel edge on delamination in composites drilling. J Engng Indus 1993;115:398–405. [42] Jung JP, Kim GW, Lee KY. Critical thrust force at delamination propagation during drilling of angle-ply laminates. Compos Struct 2005;68:391–7. [43] Durao LMP, de Moura MFSF, Marques AT. Numerical simulation of the drilling process on carbon/epoxy composite laminates. Composites A Appl Sci Manuf 2006;37:1325–33. [44] Durao LMP, de Moura MFSF, Marques AT. Numerical prediction of delamination onset process on carbon/epoxy composite drilling. Engineering Fracture Mechanics 2008;75:2767–2778. [45] Isbilir O, Ghassemieh E. Numerical investigation of the effects of drill geometry on drilling induced delamination of carbon fiber reinforced composites. Composite Structures 2013;105: 126–133. [46] Hashin Z. Failure criteria for unidirectional fiber composites. J Appl Mechm1980;47:329–34. [47] Alfano G, Crisfield MA. Finite element interface models for the delamination analysis of laminated composites. Int J Numer Method Eng 2001;50:1701–36. [48] Phadnis VA, Makhdum F, Roy A, Silberschmidt VV. Drilling in carbon/epoxy composites: Experimental investigations and finite element implementation. Composites: Part A 2013;47:41–51.

K. Palanikumar, T. Srinivasan, K. Rajagopal, J. Paulo Davim

6 Drilling of high impact polystyrene composites materials Abstract: Thermoplastic polymer matrix composite materials are widely used in various industrial and domestic applications. They have low processing costs and are a well-suited material for improved machining characteristics. Glass fiber reinforced thermoplastic (GFRTP) composites have become attractive among a variety of composite manufacturers and users due to their good properties such as high damage tolerance and higher moisture resistance. Drilling glass fiber reinforced high impact polystyrene (GF/HIPS) composite is significantly varied from other conventional composite materials due to its impact properties. Drilling processes cause delamination damage and a roundness error in the holes drilled in the laminates. The main aim of the present investigation is to study the effect of machining parameters such as feed rate, drill diameter and spindle speed on the thrust force in drilling high impact polystyrene composites. The past literature has indicated that delamination damage is related to the thrust force in the drilling of composite materials. An empirical relation has been established to correlate the drilling parameters and their interaction with thrust force. Box-Behnken design-based response surface methodology (RSM) is used for developing the model. The effectiveness of the models are checked by using analysis of variance (ANOVA). The factors and their interaction in drilling GF/HIPS composite material is presented in detail.

6.1 Introduction Fiber reinforced thermoplastic matrix composites are finding increased application in light-weight design aerospace and automotive industries. Glass fiber reinforced thermoplastic composite materials are widely used in different fields of science and technology due to its exceptional and desirable properties. The result of these properties and potential applications is a strong demand to understand the issues associated with fabricating and machining composite materials. The drilling of (GFRTP) composites is a very important and unavoidable operation for the assembly stage of mechanical joints. Ramulu et al. [1] have studied the polymer matrix composites (PMC) materials that play a key role in applications for aerospace, automotive, chemical and thermal industries. Chawla [2] has reported that GF/polyester composites materials are well suited for many engineering applications due to their high strength-to-weight ratio, corrosion resistance and high impact strength properties. Bachitar et al. [3] have reported the processing characteristics and mechanical properties of sugar palm fiber rein-

164 | 6 Drilling of high impact polystyrene composites materials forced high impact polystyrene (HIPS) composites. Hocheng and Puw [4] have studied the mechanics of the thermoset and thermoplastic composites materials and also the importance of drill bit materials and geometry for drill hole quality. Eriksen [5] investigated the surface quality of the hole of short fiber reinforced composites. Wakeman et al. [6] have highlighted the role of process parameters, mechanical performance and identifying the quality of GFRTP laminate composites. Vikki Franke [7] concluded that the drill edge geometry has a large impact on the performance of a cutting tool during drilling. Bhatnagar et al. [8] studied the drilling quality and surface finish, which depends on the orientation of the fiber and is influenced by the type of fiber embedded in the composite materials. Palanikumar et al. [9] have conducted drilling operations by using various types of drill bits and have developed the model by using RSM and analysis of variance (ANOVA) for analysis. Mohan et al. [10] have investigated the influence of drilling parameters on the thrust force and torque. The main aim is to find out the main interaction factors that influence the drilling process and obtain minimal thrust force and torque. Henninger and Friedrich [11] have considered the comprehensive parameter study of process technology and operating efficiency, which they discussed elaborately in their paper. Kishore et al. [12] have reported that the drill geometry is an important parameter and have established a strong relationship between the geometry of the drill point and the induced damage in drilling. The forecast of thrust force in drilling remains an important issue. Naisson et al. [13] have developed a new analytical model for finding thrust force and torque based on the differentiation of the cutting geometry during drilling. Srinivasan and Palanikumar [14] have reported the influence of cutting parameters on thrust force in the drilling of GFR/PP composite materials; significant parameters are explained with the help of 3D graphs. They have asserted that the increase of feed rate and drill diameter increases the thrust force when drilling glass fiber reinforced thermoplastic (GFR/PC) composites. Tsao and Hocheng [15] conducted the experiments and studied the thrust force associated with the peripheral drilling moment. Palanikumar and Srinivasan [16] investigated the proper selection of cutting parameters like feed, speed and drill diameters. They have found that they are dependent on the GF/PP composite, which is very important in the drilling process. Murthy et al. [17] optimized drilling experiments on GFRP composite plates with a solid carbide drill bit and have determined that the thrust force is decreased efficiently by decreasing the applied feed rate or increasing the rotation spindle speed of the drill bit. Madhavan and Balasivanadha prabu [18] developed the model for drilling CFRP composite; results demonstrated that thrust force is higher for the HSS drill when compared to the carbide drill. Subramanian and Senthilvelan [19] compared the static and dynamic joint strength performance of glass fiber reinforced thermoplastic (polypropylene) leaf springs with a steel plate. Jayabal and Natarajan [20] presented the mathematical model for the comprehensive study of the interactions of drilling parameters and their influence on thrust force and torque. Palanikumar [21] analyzed

6.2 Materials and Manufacturing |

165

the drilling parameters and identified the influence of parameters that affect the machining characteristics. Palanikumar and Davim [22] developed the mathematical model for finding the tool wear and significant parameters on the machining of GFRP composites using the analysis of variance. From the analysis of the results, it is important to identify the connection between different controllable parameters and also to identify the various important parameters that influence the quality of drilling. Srinivasan and Palanikumar [23] have considered a delamination model for GFR/HIPS composite laminate. The delamination is measured using a Tool maker’s microscope. The delamination in drilling is evaluated in terms of drilling forces, tool diameter and spindle speed for the respective hole quality. The results indicated that an increase in spindle speed reduces delamination, whereas an increase in feed rate increases delamination; they also indicated that a ‘Brad and spur’ drill is suitable for drilling composite materials. Box Gep [24] reported the detailed analysis of variance (ANOVA) of the experimental data, which gives important information about significant factors of thrust force and torque. Wu et al. [25] evaluated molding variables and their interactions based on the design of experiment (DOE) and solved the regression equation, which is obtained from a Box-Behnken design (BBD); they then analyzed the response surface plot. Prakash and Palanikumar et al. [26] investigated the influence of cutting parameters using Box-Behnken design on surface roughness of composite material. Design of experiments (DOE) is one of the important statistical approaches to identify the effect of both unknown and known process variables. The present paper investigates the relative significance of drilling parameters such as feed rate, drill diameter and spindle speed on the thrust force using response surface methodology (RSM). Very little research has presented drilling on a thermoplastic matrix with glass fiber reinforcement. The recent development in thermoplastic composites on structural materials has been driven by technological advantages such as higher toughness, impact resistance and the ability to be repeatedly shaped by the application of heat and relatively low forces [27]. In the present work, glass woven mat fabric reinforcements with high impact polystyrene matrices (in film/sheet form) are used. The influence of cutting parameters on thrust force and torque in drilling are carried out and presented in detail with images from a scanning electron microscope and 3D response plots.

6.2 Materials and Manufacturing Composite materials are identified as an emerging material for applications in various engineering fields such as automotive, aerospace, spacecraft and sports goods. The selection of composite materials is based on mechanical, chemical and functional properties. To achieve the shape of a composite structure, it is manufactured during the initial stage of fabrication. The dimensional tolerance and final shape of the compos-

166 | 6 Drilling of high impact polystyrene composites materials ite structure are maintained in various machining operations. Among them, drilling is one of the important machining operations that is required to join two different composite structures together. The fabrication of composite laminates takes place as injection molding, resin transfer molding, pultrusion and compression molding, etc. Among them, compression molding is selected to be the most suitable. The GFR/HIPS composite laminates are manufactured by a process of film stacking and hot pressing in a hydraulic compression molding press. The thermoplastic matrix – high impact polystyrene (HIPS) in the form of film and reinforcement as a glass fiber (GF) woven mat have an approximate weight ratio of 1 : 1. The layers of GF woven mat and HIPS films are alternatively arranged (with the high impact polystyrene sheets outside on both faces) and placed in between the mould of 250 mm × 250 mm size platens of the press. The press platens are electrically pre-heated to the glass transition temperature and then up to 5 bar pressure is applied. The layup is fabricated between steel templates and aluminum sheets coated with silicone spray; this acts as a releasing agent at the surfaces to prevent the composite laminate surface from sticking to the mold. The arrangement is maintained under the platens of a compression-molding unit, and it is heated at a uniform rate to 220 °C, which is maintained for about 30 minutes. The laminates are then permitted to cool in the mold at room temperature; then they are removed from the mold. Figure 6.1 shows the step-by-step procedure of a fabrication of GF/HIPS composite laminates. Further, glass fiber reinforced high impact polystyrene (GF/HIPS) composite laminates are prepared as test specimens, as are the specified dimensions for finding various mechanical properties and strength, as per ASTM standards. The fabricated glass fiber reinforced thermoplastic composite (GFRTP) laminates are tested for tensile (ASTM 3039), flexural (ASTM 790) and impact properties (ASTM 256) as per the

Upper mold

Glass fiber mat

Heat + Pressure

Thermoplastic sheet Cleaning & Trimming Thermoplastic sheet

(a)

Lower mold

(b)

Glass fiber + High impact polystyrene (c)

(d)

(e)

Fig. 6.1: Fabrication process of fabrication of GF/HIPS composites. (a) Stocking arrangement; (b) Stocking arrangement; (c) Compression molding machine; (d) Secondary operation; (e) Final product.

6.3 Experimental work

|

167

ASTM standard method of procedures. The tensile test and flexural test are carried out using a computer-controlled universal testing machine (UTM), and the impact test is conducted using an Izod impact testing machine. Three identical samples are tested for tensile, flexural and impact strength. The experiment’s results are listed in Tab. 6.1. Tab. 6.1: Experimental results for mechanical properties. Composites

Max. force

GFR/high impact polystyrene

Strain

(N)

Tensile strength (MPa)

(%)

Flexural strength (MPa)

Impact value (J/mm2 )

4459.449

44.596

2.8

42.545

2.87

6.3 Experimental work The experiments were carried out using a Box-Behnken design. Box-Behnken design is among the most important statistical methods used in designing experiments. These designs are an economic alternative to central composite design. Box-Bhenken designs are normally used to analyze a factor at three levels. These designs are rotatable. Montogomery (1997) indicated that Box-Bhenken design does not contain fractional factorial design, so the results are easily interpreted. Box-Behnken design is an independent quadratic design that does not contain an embedded factorial or fractional factorial design. In this design, the treatment combinations are at the midpoints of edges of the process space and at the center. These designs are rotatable (or near rotatable) and require three levels of each factor. Figure 6.2 illustrates a Box-Behnken design for three factors; see George Box and Donald Behnken (1960). For three factors, the Box-Behnken design offers some advantages by requiring a fewer number of runs. Box-Behnken design is used to conduct the experiments in the present study.

1

C 1 B –1 –1

A

1

–1 Fig. 6.2: Box-Behnken design for three factors.

168 | 6 Drilling of high impact polystyrene composites materials Straight shank Groove edge

Brad point

Full length heat treatment Fig. 6.3: Brad and spur drill used for the experiments.

Cutting spurs

The drilling experiments are carried out on glass fiber/high impact polystyrene matrix composite laminate specimens on a vertical machining center (VMC) using a solid carbide ‘brad and spur’ type drill bit with three different diameters shown in Fig. 6.3. A suitable fixture or clamping system is used for holding composite laminates in the VMC; three composite laminates are drilled at each condition. The thrust force causes many problems during drilling. The thrust force calculation is very useful to predict damages such as fiber breakage, matrix cracking, debonding, fiber pull-out, fuzzing, spalling, thermal degradation and delamination. Thrust force is measured with the help of a Kistler piezo-electric dynamometer. The experiments are conducted at three levels by prefixing the cutting parameters. The experimental set-up is presented in Fig. 6.4. The machining parameters used, their designation and their levels are presented in Tab. 6.2. The typical thrust force measured using the Kistler dynamometer is presented in Fig. 6.5. Figure 6.5 (a) shows the typical thrust force curve observed in drilling fiber reinforced composites; the thrust force increases steeply during the drilling process until drilling is completed, then it falls to zero. The mapping of thrust force (Fz ), torque (Fx ) and other component (Fy ) is presented in Fig. 6.5 (b). The complete experimental design and the result obtained in drilling fiber reinforced composite material is presented in Tab. 6.3.

f

M

Fy

Fx Ff

Drilling machine with dynamometer

Charge amplifier

DAQ system

Computer with dynoware software

Fig. 6.4: Experimental set-up for measuring thrust force using a dynamometer.

6.4 Response surface methodology-based modeling of process parameters | 169

Tab. 6.2: Machining parameters used and their levels. Parameters

Parameter designation

Drill diameter, mm Feed, mm/min Spindle speed, rpm

d f N

Levels 1

2

3

6 100 1000

9 200 3000

12 300 5000

280 240

300

200

Fz [N]

120 80 40 0 –20

(a)

5

10

15

Fx [Nm],Fy [N],Fz [N]

250

160

200 150 100 50 0 –50

Time [s]

5

10

15

20

Time [s]

(b)

Fig. 6.5: Typical thrust force (a) and thrust force, torque and other forces (b) observed in drilling composites.

6.4 Response surface methodology-based modeling of process parameters In many response surface methodology problems (RSM), significant parameters such as main and interaction variables are unknown. The main responsibility of RSM is to find a suitable approximation for the original true functional relation between response and the independent variables [28]. Kwak [29] reported that response surface methodology (RSM) is the collection of statistical and mathematical techniques. The methodology is useful to develop the modeling and analysis of engineering problems. This method was initially used for model fitting of physical experiments but later applied to design the experiments of RSM, which quantifies the relationship between the controllable input parameters and obtained response surfaces. Box and Draper [30] developed the model for physical experiments and later implemented it in machining problems. The objective of the response surface analysis is to determine the global optimization of process parameters. In response surface methodology, the relationships between the control parameters and the responses are given in equation (6.1) as: Y = f (X1 , X2 , . . . , Xk ) + ε ,

(6.1)

170 | 6 Drilling of high impact polystyrene composites materials Tab. 6.3: Box-Behnken experimental design with experimental results. S. No.

Run order

Drill diameter (mm)

Feed rate (mm/min)

Spindle speed (rpm)

Thrust force (N)

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

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

6 12 6 12 6 12 6 12 9 9 9 9 9 9 9 9 9

100 100 300 300 200 200 200 200 100 300 100 300 200 200 200 200 200

3000 3000 3000 3000 1000 1000 5000 5000 1000 1000 5000 5000 3000 3000 3000 3000 3000

163.33 181.81 245.67 281.14 215.11 204.15 192.85 238.03 169.92 251.24 190.57 270.46 252.45 233.54 236.43 244.56 242.78

* Average of three results

where Y is the response variable and X1 , X2 , . . . , Xk are independent variables. The function f is called the true response function. The residual ε measures the experimental error. In this experiment, the input variables are the feed rate, drill diameter and spindle speed. The second order response equation used in this study is given in equation (6.2) as: Y = β0 + β1 x1 + β2 x2 + ⋅ ⋅ ⋅ + βi xi + ⋅ ⋅ ⋅ + ε ,

(6.2)

where Y is the response (thrust force), β0 , β1 , β2 , . . . , βi are regression coefficients, x1 , x2 , . . . , xi represent the predictor variables such as drill diameter, feed rate and spindle speed and ε represents the error associated with the model. In response surface analysis, normally, quadratic response functions are used. The quadratic response function is represented as: k

k

i=1

i=1

y = β0 + ∑ βi xi + ∑ βii xi2 + ∑ ∑ βij xi xj + ε .

(6.3)

i j>i

The general polynomial response equation in matrix form is given in equation (6.3) as: Y = Xβ + ε . (6.4) The regression coefficient β expressed in matrix form is given in equation (6.4) as: B = (X T X)−1 X T Y .

(6.5)

6.5 Results and Discussion

| 171

To determine the value of the response variable, a quadratic mathematical model is developed by using experimental values. The statistical analysis and the adequacy of the model are checked with analysis of variance (ANOVA) at a 95 % confidence level. The fitness of the developed model is verified with the coefficient of determination (R2 ) value. Based on response surface methodology, the prediction models are established between the machining parameters and thrust force: Thrust force = + 79.12806 + 15.31100 d + 0.53497 f − 0.017684 N + 9.36667E −003 d f + 2.74417E −003 d N − 6.21250E −006 f N − 1.20742 d2

(6.6)

− 5.66925E −004 f 2 − 7.91063E −007 N 2 . In the above model, d represents drill diameter, f represents feed and N represents spindle speed. The coefficient of correlation tested for the above model is 0.9091, which is very close to 1 and demonstrates that the developed model is adequate at a 95 % confidence level. Further, analysis of variance is carried out for thrust force in the drilling of GFR/HIPS composites. The results indicate that the model is significant. The factors d, f , f 2 , df and dN are significant model terms. The analysis of variance results also indicate that the lack of fit is not significant, which shows that the model is effective in analysis of thrust force in the drilling of GFR-HIPS composites. The analysis of variance carried out for the model predicting thrust force is presented in Tab. 6.4. The normal probability plot for thrust force in the drilling of GFR/HIPS composites with respect to the externally studentized residual is presented in Fig. 6.6 (a). The normal probability plots are used to arrive at the normality assumption in the experimental run and model, which is based on the central limit theorem. In the figure, the data are spread in a straight line, which shows the correlation that exists between the experimental and predicted values. Further, Fig. 6.6 (b) shows the correlation between the experimental values and the predicted values by the developed model. From the analysis of the figure, it has been asserted that the predicted values of thrust force are close to the actual values; thus the models developed for predicting thrust force are very useful for predicting thrust force when drilling GFR/HIPS composites.

6.5 Results and Discussion High impact polystyrene (HIPS) matrix material is widely used in many industrial applications such as making prototype models, roof sheets (low-strength structural applications), food processing containers, etc. HIPS matrix material is one of the polymer matrix materials (PMC) among the thermoplastic polymer composites. It has high toughness, high strength and high impact strength; limitations are low transparency, low electrical properties and high moisture absorption. HIPS have an excellent dimen-

172 | 6 Drilling of high impact polystyrene composites materials Tab. 6.4: Analysis of variance (ANOVA) for thrust force in drilling GFR/HIPS composites. Source

Sum of squares

DF

Mean square

F value

Prob > F

Model d f n d2 f2 N2 df dN fN Residual Lack of Fit Pure Error Cor Total

14046.67 976.82 11181.85 33.58 31.58 1084.38 6.18 497.21 135.33 42.16 1404.36 54.10 1350.26 15451.03

9 1 1 1 1 1 1 1 1 1 7 3 4 16

1560.74 976.82 11181.85 33.58 31.58 1084.38 6.18 497.21 135.33 42.16 200.62 18.03 337.56

7.78 4.87 55.74 0.17 0.16 5.41 0.031 2.48 0.67 0.21

0.0065 0.0631 0.0001 0.6947 0.7034 0.0530 0.8657 0.1594 0.4385 0.6606

significant

0.053

0.9815

not significant

260 240 Predicted

Normal % probability

280 99 95 90 80 70 50 30 20 10 5 1

220 200 180 160

–2.00

–1.00

0.00

1.00

2.00

160

Externally studentized residuals

(a)

180

200

220

240

260

280

Actual

(b)

Fig. 6.6: Correlation graph for thrust force. (a) Probability of residuals; (b) Actual vs. predicated.

sional stability, high impact resistance, good aesthetic qualities, high machinability at a low cost. The impact of drilling is one of the major crises met by all kinds of manufacturing industries. By optimizing drilling parameters, this eases the work in industry and for researchers. During drilling, delamination has a huge impact on composite materials. It is mainly caused by the thrust force that is developed due to an improper selection of cutting parameters such as spindle speed, drill diameter and feed rate. Figure 6.7 shows the analysis of thrust force in the drilling of glass fiber reinforced high impact polystyrene composites, which is carried out through single effect graphs. Figure 6.7 (a) shows the effect of feed rate on thrust force in the drilling of GFR/HIPS composites and indicates that an increase in feed rate increases thrust force. Figure 6.7 (b) shows the effect of drill diameters in drilling GFR/HIPS composites; the results show that the increase in drill diameter increases thrust force due to

| 173

6.5 Results and Discussion

300

300

280

280

260

260 Thrust force (N)

Thrust force (N)

increased contact between the tool and workpiece. Figure 6.7 (c) shows the effect of spindle speed in drilling GFR/HIPS composites; the result indicates that spindle speed is not a significant factor that affects thrust force. Feed rate and drill diameter are the main influential parameters that determine thrust force when drilling GFR/HIPS composites.

240 220 200 180

240 220 200 180 160

160 100

150

200

250

6

300

8

9

11

12

Drill dia (mm)

Feed (mm/min.) (a)

(b) 300 280

Thrust force (N)

260 240 220 200 180 160 1000

2000

3000

4000

5000

Spindle speed (rpm) (c) Fig. 6.7: Effect of machining parameters with respect to drill diameter feed rate and spindle speed.

The influence of interaction between the parameters is analyzed by using 3D response graphs. Three-dimensional response graphs show the effect of two varying parameters by keeping the third parameter at a constant middle level. Figure 6.8 (a) shows the effect of drill diameter and feed rate in drilling thermoplastic composite materials. The results indicate that an increase in drill diameter slightly increases thrust force in drilling GFR/HIPS composites. Figure 6.8 (b) shows the effect of drill diameter and spindle speed in drilling GFR/HIPS composites with respect to thrust force. Figure 6.8 (c) shows the effect of feed rate and spindle speed on thrust force in drilling GFR/HIPS composites. The results show that an increase in feed rate slightly increases

Thrust force (N)

174 | 6 Drilling of high impact polystyrene composites materials

300 280 260 240 220 200 180 160 300 250 Feed

12 200 150 (mm /min .)

11 9 100

6

8

Thrust force (N)

(a)

l dia Dril

) (mm

300 280 260 240 220 200 180 160 5000 4000 Spin 3000 2000 dle spee d (rp 1000 m)

Thrust force (N)

(b)

12 9

8 6

11

) (mm l dia Dril

300 280 260 240 220 200 180 160 5000

(c)

4000 Spin 3000 2000 dle s pee 1000 d (rp m)

12

100

250 200 n.) 150 /mi m d (m Fee

Fig. 6.8: 3-D response graph for thrust force in drilling GFR/HIPS composites. (a) Effect of feed rate and drill diameter on thrust force; (b) Effect of spindle speed and drill diameter on thrust force; (c) Effect of spindle speed and feed rate on thrust force.

thrust force in drilling of GFR/HIPS composites as mentioned above. The results indicate the same trends as discussed above. The experimental results indicate that the feed rate and drill diameter have a high influence in affecting thrust force when drilling thermoplastic composite materials. Feed rate is a highly influential parameter that affects thrust force. It has been inferred from the above discussion that the proper implementation of cutting parameters reduces the thrust force when drilling GFR/HIPS composites. The scanning electron micrograph of the drilled surface is presented in Fig. 6.9; it shows the drilled hole and magnified portion of the hole drilled. Figure 6.9 (a) shows the drilled hole surface and adjacent cut surface. Figure 6.9 (b) clearly indicates the surface of the drilled hole. The result indicates that there is an insufficient material

6.5 Results and Discussion

| 175

distribution observed in the surface of the hole; this may be due to the drilling process in which the thrust may damage the hole and an insufficient distribution of resin or matrix material is observed. Figure 6.9 (c) and (d) shows the enlarged view of the drilled surface observed in the drilling of GFR/HIPS composites. The GFR/HIPS composite topographical surface image shows voids in the matrix; these are caused by a drilling process in which changes occur in the cross section of the drilled surface.

Wall

Drilled area

10.0kV 61.4mm x 10 SE (a)

Wall

5.00mm

Insufficient distribution of mareix materials

10.0kV 57.9mm x 100 SE (b)

5.00um

Voids

10.0kV 59.1mm x 50 SE (c)

1.00mm

10.0kV 57.8mm x 100 SE (d)

Fiber and matrix damage 500um

Fig. 6.9: SEM micrograph of drilled hole.

Figure 6.10 shows the atomic force microscopy (AFM) image of the typical drilled hole for GFR/HIPS composites. In atomic force microscopy (AFM), a mechanical probe is utilized to generate magnified images of drilled area surfaces down to a micrometer/nanometer resolution. In order to obtain a 3D topographical graph, an AFM probe is scanned at a constant force between the probe and the GF/HIPS surface. The implementation of atomic force microscopy (AFM) has gained a phenomenal acceptance over a variety of research and science applications. Systematic usage of AFM would enable its characteristics to be used as a helpful instrument for the assessment of tool surface texture. In Fig. 6.10 (a), the extent of the surface images was precisely measured by the widest scan size of 40 μm. In the AFM image, fiber bonding appears as a clear spot on a dark background, which is formed by the high impact polystyrene matrix.

176 | 6 Drilling of high impact polystyrene composites materials 20 18 16 14

μm

12 Counts

4.0 3.0 2.0 1.0 0 0

10 8 6

μm

5 40

10

4

35

15

30

20

2

25 25

20

0

15

30

10

35 40

5

0

0

0.5

1.0

μm

(a)

2.0 μm

2.5

3.0

3.5

4.0

(b)

40

4.0 3.5

30

4.0 3.300

35

A

X=20.24 μm Y=4.014 μm

3.5

3.0

3.0

25

2.5 2.5 2.0 μm

2.0

μm

20 μm

1.5

15

1.5

10 XS1 (line 47)

1.0

1.0

0.5

0.5

5 0

1.5

0

0 0

5

10

15 20 μm

(c)

25

30

35

40

0

(d)

5

10

15 20 Plane, μm

25

30

35

40

Fig. 6.10: AFM surface profile of GFR/HIPS composite.

Figure 6.10 (b) shows the histogram of atomic force micrography result. The result shows a maximum of 20 counts for a working distance of 40 μm. The two-dimensional image collected in a digital format for the particular surface is presented in Fig. 6.10 (c); when using this image, a wide variety of image manipulations are possible for AFM data. Quantitative topographical information, such as lateral spacing, step height and surface roughness are readily obtained. The variation in profile observed on the line of the image is shown in Fig. 6.10 (d). The figure indicates that the maximum variation occurred at 20.24 μm in the x direction, where the maximum obtained reading is 4.014 μm. The figures clearly indicate the variation in surface at the particular section.

6.6 Conclusions In the present investigation, thrust force in drilling glass fiber reinforced thermoplastic (GFRTP) composite materials is carried out by using a brad and spur drill bit. Based on the investigation, the following conclusions resulted:

References |

–

–

–

177

the experimental result and analysis indicate that the increase in feed rate and drill diameter led to an increase in thrust force when drilling thermoplastic composites, whereas an increase in spindle speed does not show any appreciable change; the empirical model was established between cutting parameters and thrust force by using the response surface method. The results indicate that the model is well suited for predicting thrust force when drilling GF/HIPS composites; the surface topography of the composite specimen were analyzed by using SEM and AFM images. The surface indicated an insufficient matrix distribution, pits and small fiber damage.

References [1] [2] [3]

[4] [5]

[6]

[7] [8]

[9] [10]

[11] [12]

[13]

Ramulu M, Branson T, Kim D. A study on the drilling of composite and titanium stacks. Composite Structures 2001;54:67–77. Chawla Krishan K. Composite materials science and engineering. New York: Springer; 2001. Bachtiar D, Salit MS, Zainudin E, Abdan K and Zaman K, Dahlan HM. Effects of Alkaline Treatment and a Compatibilizing agent on Tensile properties of Sugar Palm Fibre Reinforced High Impact Polystyrene Composites. Bio Resources 2011;6(4):4815–4823. Hocheng H, Puw H. On drilling characteristics of fiber-reinforced thermoset and thermoplastics. International Journal of Machine Tools and Manufacture 1992;32(4):583–592. Eriksen E. The influence of surface roughness on the mechanical strength properties of machined short fibre-reinforced thermoplastics. Composites Science and Technology 2000;0: 107–113. Wakeman MD, Rudd CD, Cain TA, Brooks R, Long AC. Compression moulding of glass and polypropylene composites for optimized macro- and micro-mechanical properties. Part IV: Technology demonstrator – a door cassette structure, Composites Science and Technology 2000;60:1901–1918. Franke V. Drilling of long fiber reinforced thermoplastics – Influence of the cutting edge on the machining results. Manufacturing Technology 2011;60:65–68. Bhatnagar N, Ramakrishnan N, Naik NK, Komanduri R. On the machining of fibre reinforced plastic (FRP) composite laminates. International Journal of Machine Tool Manufacturing 1995; 35(5):701–16. Palanikumar K, Prakash S, Shanmugam K. Evaluation of delamination in drilling GFRP composites. Materials and Manufacturing Process 2008;8:858–864. Mohan NS, Ramachandra A, Kulkarni SM. Influence of process parameters on cutting force and torque during drilling of glass-fiber polyester reinforced composites. Composite Structures 2005;71:407–413. Henninger F, Friedrich K. Thermoplastic filament winding with online-impregnation. Part A: process technology and operating efficiency, Composites 2002;A(33):1479–1486. Kishore RA, Tiwari R, Rakesh PK, Singh I and Bhatnagar N. Investigation of drilling in fibrereinforced plastics using response surface methodology. Proceedings of the Institution of Mechanical Engineers, Part B: Journal of Engineering Manufacture 2011;1:225–453. Naisson P, Rech J, Paris H. Analytical modeling of thrust force and torque in drilling. Proc IMechE Part B: Journal of Engineering Manufacture 2013;227(10):1430–1441.

178 | 6 Drilling of high impact polystyrene composites materials [14] Srinivasan T, Palanikumar K, Rajagopal K. Influence of Thrust Force in Drilling of Glass Fiber Reinforced Polycarbonate (GFR/PC) Thermoplastic Matrix Composites Using Box-Behnken Design. Procedia Materials Science 2014;5:2152–2158. [15] Tsao CC, Hocheng H. Effects of peripheral drilling moment on delamination using special drill bits. J Mater Proc Technol 2008;201(1–3):471–476. [16] Palanikumar K, Srinivasan T, Rajagopal K, Latha B. Analysis of Thrust Force in Drilling of GF/Polypropylene (GFR/PP) Composites. Materials and Manufacturing Process; DOI.10.1080/ 10426914.2014.961478. [17] Murthy BRN, Rodrigues LLR, Devineni A. Process Parameters Optimization in GFRP Drilling through Integration of Taguchi and Response Surface Methodology. Research Journal of Recent Sciences 2012;1(6):7–15. [18] Madhavan S, Balasivanadha Prabu S. Experimental investigation and Analysis of Thrust Force in Drilling of Carbon Fibre Reinforced Plastic Composites using Response Surface Methodology. International Journal of Modern Engineering Research 2012;2:2719–2723. [19] Subramanian C, Senthilvelan S. Effect of reinforced fiber length on the joint performance of thermoplastic leaf spring. Materials and Design 2010;31:3733–3741. [20] Jayabal S, Natarajan U. Influence of cutting parameters on thrust force and torque in drilling of E-glass/Polyester composites. Indian Journal of Engineering & Materials Sciences 2010;17: 463–470. [21] Palanikumar K. Studies on Machining Characteristics of Glass Fiber Reinforced Polymer Composites. PhD thesis, Anna University, Chennai, India;2004. [22] Palanikumar K, Davim JP. Mathematical model to predict tool wear on the machining of glass fibre reinforced plastic composites. Materials and Design 2007;28:2008–2014. [23] Srinivasan T, Palanikumar K, Rajagopal K. Delamination in Drilling of GFR/High Impact Polystyrene composites. Advanced Materials Research2013;622–623:1271–1274. [24] Box GEP, Hunder WH, Hunder JS. Statistics for experiments. New York: John Wiley & Sons;1978. [25] Wu L, Yick KL, Ng SP, Yip J. Application of the Box-Behnken design to the optimization of process parameters in foam cup molding. Expert Systems with Applications 2012;39:8059–8065. [26] Prakash S, Palanikumar K, Mercy JL, Nithyalakshmi S. Evaluation of Surface Roughness Parameters (Ra, Rz) in Drilling of MDF Composite Panel using Box-Behnken Design (BBD). Int. J. Design Manuf. Technol. 2011;5:52–62. [27] Lakshmi Narayanan T, Madhavan S, Sinha A. Investigation of Thrust, Torque and Delamination on Drilling of Glass Fabric/Polypropylene Matrix Composite. International Conference on Mechanical and Electrical Technology 2010;1:99–103. [28] Montgomery DC. Design and analysis of experiments. 6th ed. New York: John Wiley & Sons, Inc.;2005 [29] Kwak JS. Application of Taguchi and response surface methodologies for geometric error in surface grinding process. International Journal of Machine Tools and Manufacture 2005;45: 327–334. [30] Box GEP, Draper NR. Empirical Model Building and Response Surfaces. New York: John Wiley & Sons;1987. [31] Montogmery C. Design and Analysis of Experiments. 4th ed. New York: John Wiley & Sons; 1997. [32] Box G, Behnken D. Some new three level designs for the study of quantitative variables. Technometrics 1960;2:455–475.

R. Pramod, S. Basavarajappa, J. Paulo Davim

7 A review on investigations in drilling of fiber reinforced plastics Abstract: In the past few decades, polymer based fiber reinforced composites have been the prime materials chosen over the conventional materials in many industries, whereby drilling has been the secondary manufacturing process adopted for structural integration. Many specific characteristics of composites such as marked anisotropy, structural non-homogeneity, abrasiveness and lack of plastic deformation determine the behavior of composites during machining; this has makes it different than machining conventional materials in many respects. The principal objective of this work is to present a detailed literature review on drilling polymer composites with a focus on delamination. Aspects such as delamination mechanisms, fabrication methods, the type of drilling process adopted by various researchers, cutting parameters employed during drilling, mathematical delamination modelling, effect of thrust force, spindle speed, thermal loads, tool wear, surface roughness, tool geometry and tool materials on delamination and hole quality are summarized. In addition, the role of digital image processing in delamination assessment was also studied. Considerable efforts have been made by various researchers to comprehend delamination mechanisms, and there is still a demand for characterizing delamination with the implementation of nano-filler materials, the use of blended epoxy resins, reduced wear on drill tools, real-time drilling monitoring, the effect of stack supports and fiber stacking order, higher quality hole production and avoiding de-burring.

7.1 Introduction Polymer-based fiber reinforced composite materials are rapidly supplanting conventional materials because they possess unique characteristics such as low density, a low coefficient of thermal expansion, high specific strength with better fatigue resistance and a specific modulus [1]. The market has witnessed a surge in demand for these materials in most of the high production-rate industries like aerospace, spacecraft, marine, sports, recreation and automobiles. Structural components integration often requires machining even after it has been precisely fabricated. Drilling is an inevitably complex stochastic process as the machining behavior of composite materials differ from conventional materials machining; they are characterized by marked anisotropy, structural heterogeneity, abrasiveness and deficit plastic deformation. The drilling operation is always accompanied with induced defects such as fiber breakage, de-bonding, pull-out, stress concentration, thermal damage, matrix crack-

180 | 7 A review on investigations in drilling of fiber reinforced plastics ing, matrix fuzzing, matrix burning, fiber-matrix debonding, undesired eccentricity, fiber fuzzing, spalling, micro-cracking, delamination and so on [2]. Among the induced defects, delamination acts as a limiting factor for reliability. This reduces the composites’ assembly tolerance, bearing strength, residual strength and strength against fatigue, thus deteriorating the structural integrity and long-term performance of composite materials [3]. Apart from delamination burrs generated due to tool damage, tool wear also reduces assembly tolerance [4]. It is estimated that during assembly of an aircraft, 60 % of part rejections is due to drilling-associated delamination [5]. De-burring accounts for an increase in production cost by 25 %. Though an adequate amount of work has been carried out in analyzing delamination due to drilling processes using various types of drill tools with different tool geometries and tool coatings from an experimental and analytical point of view, there is still lies a thrust on characterizing the delamination under the addition of micro and nano fillers as reinforcements in the polymer matrix and by providing metallic stacking supports during drilling. Composition and fiber stacking has an influence on burr formation, delamination, surface quality and tool life. The specific criteria for optimal drilling performance in fiber-reinforced materials include the lowest delamination and fiber breaking, minimum hole shrinkage and good surface quality. During machining, tool wear affects tool life and surface finish of the machine component. In the case of drilling, wear is categorized as flank wear, chisel wear, corner wear and crater wear. Wear on the drill has a definitive influence on hole quality, dimensional accuracy and the tool life of drill bits.

7.1.1 Delamination and delamination mechanism Delamination is an inter-ply failure phenomenon induced during drilling, and it is a limiting factor for PMC component life. Delamination will occur and develop in two phases along the fiber orientation. (a) Phase I – chisel edge action: when the critical value of thrust force for the remaining plies at the exit when the reduction in laminate thickness is exceeded by the thrust force on the chisel edge, the inter-ply bonding strength is reduced. This causes push-out delamination. A larger amount of thrust force is generated due to a negative rake angle during the machining of laminates by chisel edge. (b) Phase II – cutting edge action: delamination is caused due to chaffing by the cutting edge of the drill at the entry of the laminate. This causes an increase in thrust force that results in a peeling effect on laminates. This type of delamination is known as peel-out delamination; it can be avoided with a lower feed control. Most delamination during machining is on account of push-out delamination [6].

7.2 Drilling process | 181

7.1.2 Fabrication of polymer matrix composites Many researchers used different types of fabrication methods to create the composite laminates. The summary can be found in the Tab. 7.1. Tab. 7.1: Details of the fabrication method adopted for manufacturing polymer matrix composites [7–30]. (1)

Fabrication method

Hand lay method [75 %], resin transfer molding [10 %], vacuum assisted resin transfer molding [10 %], autoclave method [5 %]

(2)

Fiber material

Glass fibers [50–60 %], carbon fibers[20–35 %], kevlar [5–10 %]

(3)

Resins used

Epoxy resin [70 %], polyester resins [30 %]

(4)

Laminate thickness

03 mm [30 %], 5 mm [20 %], 6 mm [40 %], 10 mm [5 %], 12 mm [5 %]

(5)

Volume fraction

Fiber [30–45 %], matrix [55–70 %]

(6)

Type of fibers

Chopped, long fiber with continuous or cross windings and woven mat type

(7)

Fiber orientation

Unidirectional and bi-directional/woven

(8)

Other reinforcements

Aluminum metal sheet for fiber metal laminae, micro fillers and nano fillers such as SiO2 , CaCO3 , alumina, graphene/ CNT, TiO2 , CaSiO3

(9)

Reinforcement particle sizes

100 μm to 15 nm

(10)

Percentage of reinforcement

1–3 % of volume fraction

7.2 Drilling process Aircraft, automotive and marine industry orders are expected to grow exponentially over the next few years, while machining applications for composite parts may double by 2016. The drilling process has become an inevitable secondary machining operation for structural integrity in most component assemblies. The drilling process can be carried out using non- traditional machining such as laser drilling and water jet machining, or with traditional machining such as conventional drilling. The design of the drilling process on composite laminate can be summarized in Fig. 7.1. We can see various types of drilling processes and machining parameters that were adopted by various researchers as represented in Tabs. 7.2 and 7.3, respectively.

182 | 7 A review on investigations in drilling of fiber reinforced plastics

Selection of fiber and matrix

Selection of fabrication method

Selection of cutting tool parameters and machining parameters

Delamination assessment and delamination characterization

Instrumentation setup to characterize delamination and tool wear

Machining centre setup and workpiece setup

Delamination assessment and optimization of cutting tool and machining parameters

Fig. 7.1: Drilling process employed for polymer matrix composites [6–46].

Tab. 7.2: Summary of drilling process employed by various researchers [2–38]. (1)

Types of drilling

Conventional drilling, high-speed drilling, vibration-assisted drilling, cryogenic drilling, laser drilling, grinding drilling

(2)

Drill tool materials

High-speed steel, coated and uncoated cemented carbide, polycrystalline diamond, cobalt

(3)

Types of drill tool

Step, brad and spur, brad point, slot drill bit, core drill bit, end mill cutters, straight flute drill bit, one shot drill bit, jo drill bit, saw drill bit, candle stick drill bit, multi-facet drill bit, step-core drill, step-core-candlestick drill, core-candle stick drill

(4)

Drill point angle

Twist drill Step drill Two-facet drill bit One shot drill bit Three-facet drill bit Four-facet drill bit

90–118, 120, 155, 175, 185, 185/178° 118° 133.4° 30° 130° 133.4, 120°

(5)

Helix angle

Twist drill Step drill Two-facet drill bit One shot drill bit Three-facet drill bit Four-facet drill bit

30° 30° 25° 35, 40° 30° 25, 30°

(6)

Drill bit diameter

3, 5, 6, 6.4, 8, 10 mm

(7)

Cooling method

Dry, mist, flood and cryogenic

7.2 Drilling process |

183

Tab. 7.3: Summary of cutting parameters adopted for drilling [2–26]. (1)

(2)

Spindle speed

Feed

Conventional drilling High-speed drilling Vibration-assisted drilling

Grinding drilling Cryogenic drilling

500 to 2500 rpm 4000 to 40 000 rpm 100 to 5000 rpm Frequency range: 100 Hz with amplitude 1 to 30 μm 1500 to 5000 rpm 500 to 2000 rpm

Conventional drilling High-speed drilling Vibration-assisted drilling Grinding drilling Cryogenic drilling

10 to 500 mm/min 1000 to 9000 mm/min 10 to 80 mm/min 10 to 60 mm/min 10 to 300 mm/min

Several instrumentations have been used by various researchers to analyze delamination. For a complete assessment of drilling, the required instrumentation systems are shown in Fig. 7.2.

Dynamometer [piezo/strain gauage type]

Thrust force & torque

Surface profiler

Surface roughness

Digital image acquistion & processing system

Cutting edge radius & delamination

Optical & tool makers microscopes

Burr generation

CMM machine

Diameter of the drilled hole roundness

Profilometer

Subsurface quality & tool wear

Fig. 7.2: Schematic diagram of instrumentation systems for drilling process characterization.

184 | 7 A review on investigations in drilling of fiber reinforced plastics 7.2.1 Delamination assessment The delamination occurring around a drilled hole is usually assessed by several dimensional and non-dimensional parameters. The dimensional parameters used to assess are: (a) area of delamination; (b) average sum of the length of internal cracks; (c) difference between drilled hole radius/diameter and maximum damage radius/diameter; and (d) average of perpendicular measurements of damage area diameter. The non-dimensional parameters used are: (a) two-dimensional delamination factor: ratio of the damage [Adamage ] area to the hole area [Ao ]: Adamage − Ao ) % ; and (7.1) Fa = ( Ao (b) delamination factor [diameter based]: ratio of maximum diameter [ Dmax ] of the delamination zone to the hole diameter [6] [Do ]: Fd =

Dmax . Do

(7.2)

A one-dimensional delamination factor may lead to confusion as some loose peeled out fibers may give the value of Dmax inaccurately. Hence two-dimensional delamination factors obtained by image analysis helps in increasing delamination assessment accuracy. To further evaluate drilling on the basis of digital image analysis, Davim [4] suggested an adjusted delamination factor including crack size and damage area contributions: D A Fda = α max + β max (7.3) Do Ao or Ad Fda = Fd + (7.4) (F 2 − Fd ) , (Amax − A0 ) d where α = (1 − β ) and β =

Ad (Amax −A0 )

and where α and β are weights in parts.

(c) Delamination factor [radius-based] is the ratio of drill radius to the delamination radius [22]: R Fd = max Ro FDR =

Dmax A 2 A A 3 + 1.783 ( H ) + 0.7156 ( M ) + 0.03692 ( L ) , D0 A0 A0 A0

where FDR = refined delamination ratio, AH = heavily damaged area, AM = medium damaged area, and AL = light damaged area.

(7.5)

7.2 Drilling process |

185

Tab. 7.4: Empirical models based on linear regression analysis for delamination caused by drilling. (1)

Sardinas et al.

Standard twist drill [carbide] Fd = 1.93f 0.1429 Vc0.1022

(2)

Tsao et al.

Standard twist drill [HSS] Fd = 1.961 − 10.955f − 1.81 × 10−4 S − 1.77 × 10−2 d Brad Point drill [HSS] Fd = 1.539 − 2.274f − 7.81 × 10−6 S − 1.7 × 10−2 d Slot Drill Bit [HSS] Fd = 1.508 − 3.385f − 8.681 × 10−6 S − 1.49 × 10−2 d

(3)

Gaitonde et al.

Standard twist drill [carbide] Fd = −0.810444 − 0.001889Vc − 0.109957f + 0.03454θ + 0.000011Vc f − 0.000009Vc θ + 0.00167f θ + 0.00000Vc2 + 0.00553f 2 − 0.000115θ 2

(4)

Davim et al.

Standard twist drill [carbide] Fd = 0.966 + 1.085 × 10−3 Vc + 0.134f Brad Point drill [carbide] Fd = 1.006 + 1.980 × 10−4 Vc + 0.021f

(5)

Erol Kilickap et al.

Standard twist drill [HSS] Fd_entrance = 0.53125 + 0.0329A + 0.83B + 0.193C − 0.000083A2 − 1.075B2 − 0.03575C 2 − 0.0005AB − 0.00005AC − 0.025BC Standard twist drill [HSS] Fd_exit = 0.44 + 0.06895Vc + 0.5525f + .074θ − 0.000199Vc2 − 0.975f 2 − 0.01225C 2 + 0.0005Vc f + 2.0296E −17 Vc θ + 0.15f θ

(6)

Khashaba et al.

(7)

Krishnamoorthy et al.

Standard twist drill [carbide] Fd = 1.482 + 1.44 × 10−3 Vc + 3.143f + 0.0193 W Fd_Entry = 1.08419 − 2.78E −5 ϑ + 5.76389E −3 d − 1.51667E −3 f + 1.02917E −5 ϑ d + 5.83333E −7 ϑ f − 7.5E −5 df − 5.29E −8 ϑ 2 − 6.39E −4 d2 + 2.058E −5 f 2 Fd_Exit = 1.118 + 4.45E −7 Vc + 3.334E −4 d − 2.26667E −3 f + 3.75E −7 Vc d − 5.334E −8 Vc f + 4.58333E −6 df − 3.334E −9 Vc2 + 2.0834E −5 d2 + 2.7667E −5 f 2

(8)

S. R. Karnik et al.

Fd = −0.810444 − 0.001889Vc − 0.109957f + 0.034546θ + 0.000011vf − 0.000009Vc θ + 0.001670f θ + 0.000003Vc2 + 0.00530f 2 − 0.000115θ 2

Vc : cutting speed or spindle speed in rpm f : feed rate in mm/rev Fd : elamination factor W: tool pre-wear in grams θ : point angle in degrees d: diameter of the drill bit in mm Fd_exit and Fd_entrance : delamination factor at the exit and entrance of the laminate, respectively

186 | 7 A review on investigations in drilling of fiber reinforced plastics Several empirical models for delamination caused by drilling have also been derived using linear regression analysis. The empirical models developed are based on different drilling conditions adopted by individual researchers; some of the empirical models are represented in Tab. 7.4.

7.2.2 Effect of various machining parameters on delamination The quality of the hole drilled is affected by thrust force; it acts as a key parameter to assess delamination. Hocheng and Dharan developed the analytical model based on linear elastic fracture mechanics relating the critical thrust force and energy release rate. Many works have shown that delamination occurs when the thrust force developed during machining crosses the critical thrust force in magnitude. Several researchers have developed analytical models based on Hocheng and Dharan for various types of drill bits that consider the material to be elastic and isotropic. They are summarized in Tab. 7.5. Machinability aspects that consider the effects of various machining process parameters can be carried out by developing mathematical models that are a function of composite material, torque, thrust force, spindle speed, feed rate, point angle, number of drilled holes, tool pre-wear, effect of coolants, tool material and tool geometry. Drilling-induced delamination is directly affected by an increase in thrust force; this was established in several works. By optimizing cutting parameters affecting the thrust force and by minimizing the thrust force, we can thus achieve delamination-free holes.

7.2.2.1 Effect of spindle speed on thrust force It was evident in the works carried out by various investigators that a gradual increase in spindle speed or cutting speed resulted in a decrease in thrust force and torque. In some drill geometries, the effect of spindle speed on torque were negligible [27–33]. At higher spindle speeds, the decrease in thrust force and torque is due to a softening of the matrix caused by a temperature increase and low coefficient of thermal conductivity of the fibers [29]. With worn out drill bits, the thrust force increased with an increase in spindle speed; a reversed effect was observed with new drill bits. A variation in spindle speed in various types of drilling processes can be found in Tab. 7.3.

7.2.2.2 Effect of feed rate on thrust force The effect of feed rate is remarkable. As feed rate increases, thrust force increases in conventional drilling processes. In vibration-assisted drilling process, as feed rate increases with a high spindle speed thrust force decreases. Lower feed rates with higher spindle speeds favor laminate drilling in conventional drilling by minimizing the loss

7.2 Drilling process | 187

Tab. 7.5: Thrust force models for drilling-induced delamination [28–33].

(1)

Twist drill bit

FCT = π (

(2)

Slot drill bit

FCSL =

(3)

Brad point drill bit

FCBP =

(4)

Core step drill

FCC =

(5)

Step drill bit

8GIC Eh3 ) 3 (1 − ϑ 2 ) 1

1/2

√1 + α 2 (1 − 2S2 + S4 ) 1+α √1 + α 2 (1 − 2S2 + S4 )

FCT FCT

β (2 − β ) 4

√ [1 − (1 − β ) ] − 0.5S2 [1 − (1 − β )6 ]

FCT

FCST = 2

[(1 − ϑ ) + 2 (1 + ϑ ) ε 2 ] } √2 { { 2 2 3 2 2 1 − ϑ (1 + ϑ ) [2 (1 − ϑ ) (1 + 2ϑ ) − 12 (12 − 4ϑ + 3ϑ + 3ϑ ) ε − 8(1 + 3ϑ )ε ln ε } { } (6)

E ϑ GIC h Rt Rdl a S b t Roc

Twist drill bit

For orthotropic materials

FZ = 8 × π × (

D =

1 (3D11 8

D󸀠 =

D + D66 D11 + D22 + 12 2 3

GIC × D 1 3

−

󸀠

D 8D

1/2

FCT

1 2

)

+ 2D12 + 4D66 + 3D22 )

elastic modulus poisson ratio critical strain energy release rate (in mode I fracture) uncut-plies thickness under drill bit radius of drill bit radius of delamination ratio between concentrated load (P1) and peripheral circular load (P2) = Roc /Rdl = t/Roc thickness of core drill bit outer radius of core drill bit

of mechanical strength, but lower feed rates may also result in thermal degradation of the matrix. At higher feed rates, fibers are pushed instead of shearing because the rake angle becomes negative. At higher feed rates, the occurrence of delamination is also less, but inter-laminar cracks are found around high-density areas [18–31].

7.2.2.3 Effect of thermal loads on thrust force The effect of thermal load during machining is more than the mechanical loads, hence the use of coolant through mist or flooding is a must to avoid the thermal damage due to the fibers’ reduced thermal conductivity. The friction between the drilled hole and the drill causes thermal damage. Use of cryogenic liquids [39] has a better effect in re-

188 | 7 A review on investigations in drilling of fiber reinforced plastics ducing protruding, fuzzy and uncut fiber removal. If thermal loads are not addressed, there can be radial expansion in the hole size due to the expansion in drill diameter. Though the tool wear rate, temperature and thermal damage caused by cutting force reduces when coolant is applied, thrust force and torque increases due to the increase in material modulus under coolant action. Applying coolant would to some degree reduce the cutting temperature and thus reduce the defects related to thermal damage, as well as lower the tool wear rate. However, since materials tend to have a higher modulus under coolant, thrust force and torque might elevate [17, 19, 26, 31].

7.2.2.4 Effect of tool wear on delamination Tool wear has a high economic impact during drilling because it increases the thrust force inducing a delamination and dimensional errors in the size of the bore hole. In addition, due to tool wear, frequent tool changes are necessary. Tool wear mechanisms in drill tools are classified as: (a) adhesion tool wear; (b) micro-chipping; and (c) abrasive tool wear. Tool wear during drilling are caused by abrasive fibers, abridged thermal conductivity, frictional heat, matrix fuzzing and clogging [32, 33]. Understanding wear mechanisms and the effect of input factors on tools helps in increasing tool life and the surface quality of the drilled hole. Among tool wear mechanisms, abrasive wear is seen as predominant. Flank wear and rake wear are two modes of tool wear [11]. Rake wear is very small in most of the tools, and the level of wear in a tool is based on flank wear. Tool wear increases with an increase in feed rate, spindle speed and point angle. When drilling multiple holes, wear increases rapidly in initial stages, and then later one can witness steady wear. As the number of holes drilled increases, there is an increase in the thrust force as there is an increase in tool wear [40]. Tools made of PCD and carbide as well as coated tools yield longer tool life than HSS tools. Spindle speed Point angle

Thrust force

Number of drilled holes Feed rate Tool wear

Fig. 7.3: Effect of various input parameters on thrust force.

7.2 Drilling process | 189

7.2.2.5 Effect of tool geometry and tool materials on delamination Damage such as elliptical hole formation due to the alternate cycles of torsion and compression of fibers before shearing is caused by the drill geometry that is associated with the angle between cutting edge, point angle and the fiber orientation of the laminates. Drill geometry also influences thrust force as studies suggest a lower point angle results in a decrease in thrust force [38]. With the use of multi-facet drill bits with more cutting edges, tool wear decreases. The chisel edge should be reduced as far as possible, and the tool should have a small rake angle. As the tool diameter increases, shear area also increases; this results in a higher thrust force and torque. Smaller tip diameter drill bits with medium tool lengths help reduce vibration produced during machining of laminates with 4–8 mm thickness. The cutting edge radius has a direct effect on delamination and surface quality, while the corner wear of the tools results in burr generation. Tool geometry has an influence on both thrust force and drillinginduced delamination. Therefore a better thought process should be applied to tool selection in order to reduce the effect of indentation by the chisel edge; this results in a minimization of delamination [37]. Among the tool materials, WC, PCD, cobalt and HSS have been used by many researchers. WC and cobalt tools seem to be favorable for a small series of holes with lesser tool damage and less frequent tool changes. Much tool geometrical diversity can be seen among HSS and WC tools. The friction coefficient is lower in HSS. WC drill bits have smaller thrust forces than PCD, but PCD tools can be operated with higher spindle speeds [38]. Among the different types of tools, step, brad and drill bits were better among the remaining tools in a comparison as shown in Fig. 7.4. Twist drills have been found to be better in reducing delamination in the presence of pre-drilled pilot holes. A pilot hole reduces the effect of chisel edge during machining [35]. But pre-drilling when using tools like brad and dagger has no added advantages, and more space is required by dagger tools at the exit. Apart from tool geometry, even the ply material, uncut ply thickness and individual ply orientation also play a key role in delamination. Brad

Dragger

Step

Twist 85

Twist 120

Core drill bit

Slot drill

Candle stick drill

Saw drill

Delamination factor

Tool wear

Surface roughness

Fig. 7.4: Effect of various parameters on tool geometry.

Thrust force

190 | 7 A review on investigations in drilling of fiber reinforced plastics 7.2.2.6 Surface roughness The surface roughness of a hole increases with an increase in spindle speed and feed rate. The parameters that influence surface roughness are tool insert radius, feed rate and the interaction between feed rates and insert radius. Surface roughness is better among laminates manufactured by a hand-lay process. The surface quality of holes are superior when a pilot hole is present and the tool has multi-cutting edges [35]. The goal of manufacturers considering productivity is to achieve better hole quality with a good surface roughness and avoid operations such as reaming or any other postdrilling operations to make the process cost effective [41–43].

7.3 Role of digital image processing in delamination assessment Image processing techniques ahve evolved as the new tool for the analysis and evaluation process of delamination during drilling. Image processing helps to quantify the magnitude of damage during machining operations. Image acquisition is the most important step in the image processing techniques. Various environments and techniques used in the acquisition of images [44] are shown in Fig. 7.5.

Image acquisition systems

Visible light

Conventional & enhanced radiography Ultrasonic C-Scans

X-Rays

Image acquisition

Infrared Video & digital cameras X-Ray computer tomography

Microwaves Video & digital cameras (a)

(b)

Fig. 7.5: (a) Image acquisition environment and (b) techniques used for image acquisition.

7.4 Summary |

191

After capturing images of delaminated regions, the proper evaluation criterion should be applied for damage assessment. The image processing carried out by various researchers is summarized in Fig. 7.6. During the morphological processing of the acquired image, all of the drilled hole’s geometrical entities and image components required for delamination assessment are extracted. Later, segmentation is carried out to extract the lines, areas and curves around the drilled surface and delaminated holes. Several soft computing techniques such as artificial neural network, fuzzy logic, etc. are used in conjunction with image processing after segmentation for better assessment of delamination [45, 46].

Fabrication of laminates

Drilling of laminate

Image acquisition of drilled hole

Image enhancement

Image restoration & compression

Morphological processing

Segmentation of image around drilled hole area

Segmentation of image around delaminated area

Segmentation of image around cutting edge & flank

Delamination and tool wear assessment

Assessment of surface finish

Create knowledge base

Fig. 7.6: Steps in image processing for assessment of drilling-induced delamination.

7.4 Summary The use of composite laminates has been increasing at an exponential rate and steadily replacing conventional materials in industries such as aircraft, automobile and marine. Compared to conventional materials, machining composite materials differs in many aspects. The present review presents an insight into several developments in the drilling process and the effect of parameters such as feed rate, spindle speed, drill geometry and drill tool material, thermal loads on thrust force, tool wear, etc. A summary of various methods of drilling, fabrication methods and instrumentation systems for delamination assessment has been presented. Apart from the conventional twist drill, many new types of special drill bits with different tool geometries have been adopted to reduce delamination. Studies have shown that flank wear is more critical then rake wear, and tool wear will lead to diminution of hole

192 | 7 A review on investigations in drilling of fiber reinforced plastics surface quality and increase the cost of machining. Several empirical models have been developed for a better understanding and quantification of delamination due to drilling. Among the cutting parameters, high spindle speed with a lower feed rate is better for a conventional drilling process. More work is yet to be carried out with high-speed drilling. Applying coolant must be done cautiously. Several studies show the implementation of image processing techniques along with soft computational techniques for delamination assessment and surface quality assessment.

References [1] [2] [3] [4] [5]

[6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17]

Teti R. Machining of composite materials. CIRP Annals – Manuf Technol 2002;51:611–634. Tagliaferri V, Caprino G, Diterlizzi A. The effect of drilling parameters on the finish and mechanical properties of the GFRP composites. Int J Mach Tools Manuf 1989;30(1):77–84. Stone R, Krishnamurthy K. A neural network thrust force controller to minimize delamination during drilling of graphite-epoxy laminates. J Mach Tools Manufact 1996;36(9):985–1003. Davim JP, Reis P, Antonio CC. Experimental study of drilling glass fiber reinforced plastics (GFRP) manufactured by hand lay-up. Compos Sci Technol 2004;64:289–297. Mishra R, Malik J, Singh I, Davim JP. Neural network approach for estimate the residual tensile strength after drilling in uni-directional glass fiber reinforced plastic laminates. Mater Des 2010;31:2790–2795. Hocheng H, Tsao CC. The path towards delamination-free drilling of composite materials. J Mater Process Technol 2005;167:251–264. Rawat S, Attia H. Characterization of the dry high speed drilling process of woven composites using machinability maps approach. CIRP Annals – Manuf Technol 2009;58:105–108. Chen WC. Some experimental investigation in the drilling of carbon fiber reinforced plastic (CFRP) composite laminates. Int J Mach Tools Manuf 1997;37:1097–1108. Tsao CC, Hocheng H. The effect of chisel length and associated pilot hole on delamination when drilling composite materials. Int J Mach Tools Manuf 2003;43:1087–1092. Tsao CC, Hocheng H. Taguchi analysis of delamination associated with various drill bits in drilling of composite material. Int J Mach Tools Manuf 2004;44:1085–1090. Tsao CC, Hocheng H. Effect of tool wear on delamination in drilling composite materials. Int J Mech Sci 2007;49:983–988. Tsao CC, Hocheng H. Parametric study on thrust force of core drill. J Mater Process Technol 2007;192–193:37–40. Tsao CC. Prediction of thrust force of step drill in drilling composite materials by Taguchi method and radial basis function network. Int J Adv Manuf Technol 2008;36:11–18. Tsao CC. Experimental study of drilling composite materials with step-core drill. Mater Des 2008;29:1740–1744. Tsao CC, Hocheng H. Effects of special drill bits on drilling-induced delamination of composite materials. Int J Mach Tools Manuf 2006;46:1403–1416. Piquet R, Ferret B, Lachaud F, Swider P. Experimental analysis of drilling damage in thin carbon/epoxy plate using special drills. Composites, Part A 2000;31:1107–1115. Gaitonde VN, Karnik SR, Campos Rubio J, Esteves Correia A, Abrao AM, Davim JP. Analysis of parametric influence on delamination in high-speed drilling of carbon fiber reinforced plastic composites. J Mater Process Technol 2008;203:431–438.

References |

193

[18] Karnik SR, Gaitonde VN, Campos Rubio JC, Esteves Correia A, Abrao AM, Davim JP. Delamination analysis in high speed drilling of carbon fiber reinforced plastics (CFRP) using artificial neural network model. Mater Des 2008;29:1768–1776. [19] Rahme P, landon Y, Lachaud F, Piquet R, Lagarrigue P. Analytical models of composite material drilling. Int J Adv Manuf Technol 2011;52:609–617. [20] Arul S, Vijayaraghavan L, Malhotra SK, Krishnamurthy R. Influence of tool material on dynamics of drilling of GFRP composites. Int J Adv Manuf Technol 2006;29:655–662. [21] Nagarajan VA, Rajadurai JS, Kumar TA. A digital image analysis to evaluate delamination factor for wind turbine composite laminate blade. Composites, Part B 2012;43(8):3153–3159. [22] Kilickap E. Determination of optimum parameters on delamination in drilling of GFRP composites by Taguchi method. Indian Journal of Engineering and Material Sciences 2010;10:265–274. [23] Quiza Sardiñas R, Reis P, Davim JP. Multi-objective optimization of cutting parameters for drilling laminate composite materials by using genetic algorithms. Composites Science and Technology 2006;66:3083–3088. [24] Khashaba UA. Delamination in drilling GFR-thermoset composites. Composite Structures 2004;63:313–327. [25] Krishnamoorthy A, Boopathy SR, Palanikumar K, Davim JP. Application of grey fuzzy logic for the optimization of drilling parameters for CFRP composites with multiple performance characteristics. Measurement 2012;45:1286–1296. [26] Karnik SR, Gaitonde VN, Campos Rubio JC, Correia AE, Abrão AM, Davim JP. Delamination analysis in high speed drilling of carbon fiber reinforced plastics (CFRP) using artificial neural network model. Materials and Design 2008;29:1768–1776. [27] Hocheng H, Tsao CC. Comprehensive analysis of delamination in drilling of composite materials with various drill bits. J Mater Process Technol 2003;140:335–339. [28] Hocheng H, Tsao CC. Effect of special drill bits on drilling-induced delamination of composite materials. J Mach Tools Manuf 2006;46:1403–1416. [29] Zhang LB, Wang LJ, Liu XY, Zhao HW, Wang X, Luo HY. Mechanical model for predicting thrust and torque in vibration drilling fibre-reinforced composite materials. Int J Mach Tools Manuf 2001;41:641–657. [30] Hocheng H, Dharan CKH. Delamination during drilling in composite laminates. ASME J Eng Indust 1990;112:236–239. [31] Zitoune R, Collombet F. Numerical prediction of the thrust force responsible of delamination during the drilling of the long-fibre composite structures. Composites, Part A 2007;38: 858–866. [32] Palanikumar K, Davim JP. Assessment of some factors influencing tool wear on the machining of glass fibre-reinforced plastics by coated cemented carbide tools, Journal of Materials Processing Technology 2009;209:511–519. [33] Uysal A, Altan M, Altan E. Effects of cutting parameters on tool wear in drilling of polymer composite by Taguchi method. Int J Adv Manufacturing Technology 2012;58:915–921. [34] Gómez MP, Hey AM, Ruzzantea EJ, D’Attellis CE. Tool wear evaluation in drilling by acoustic emission. International Congress on Ultrasonics, Universidad de Santiago de Chile, January 2009. Physics Procedia 2010;3:819–825. [35] Won MS, Dharan CKH. Chisel edge and pilot hole effects in drilling composite laminates. American Society of Mechanical Engineers 2002;124:232–245. [36] Shyha IS, Aspinwall DK, Soo SL, Bradley S. Drill geometry and operating effects when cutting small diameter holes in CFRP. International Journal of Machine Tools and Manufacture 2009; 49:1008–1014.

194 | 7 A review on investigations in drilling of fiber reinforced plastics [37] Durão LMP, Gonçalves JSD, Tavares JMRS, de Albuquerque VHC, Aguiar Vieira A, Torres Marques A. Drilling tool geometry evaluation for reinforced composite laminates. Composite Structures 2010;92:1545–1550. [38] Velayudham A, Krishnamurthy R. Effect of point geometry and their influence on thrust and delamination in drilling of polymeric composites. Journal of Materials Processing Technology 2007;185:204–209. [39] Velayudham A, Krishnamurthy R. Evaluation of tensile strength of woven carbon/epoxy composite laminates at cryogenic temperatures using the open hole specimens. ASTM J Test Eval 2011;39(4):690–695. [40] Tsao CC, Hocheng H. Effect of tool wear on delamination in drilling composite materials. International Journal of Mechanical Sciences 2007;49:983–988. [41] Palanikumar K, Mata F, Davim JP. Analysis of surface roughness parameters in turning of FRP tubes by PCD tool. Journal of Materials Processing Technology 2008;204:469–474. [42] Tsao CC, Hocheng H. Evaluation of thrust force and surface roughness in drilling composite material using Taguchi analysis and neural network. Journal of Materials Processing Technology 2008;203:342–348. [43] Fetecau C, Stan F. Study of cutting force and surface roughness in the turning of polytetrafluoroethylene composites with a polycrystalline diamond tool. Measurement 2012;45: 1367–1379. [44] de Albuquerque VHC, Tavares JMRS, Durão LMP. Evaluation of Delamination Damage on Composite Plates using an Artificial Neural Network for the Radiographic Image Analysis. Journal of Composite Materials Online First, November 9, 2009. [45] Davim JP, Rubio JC, Abrao, AM. A novel approach based on digital image analysis to evaluate the delamination factor after drilling composite laminates. Compos Sci Technol 2007;67: 1939–1945. [46] Nagarajan VA, Sundaram S, Rajadurai JS. A novel approach based on digital image analysis to evaluate refined delamination factor for E-Glass 21xK43 Gevetex/LY556/DY063 epoxy composite laminates. Proceedings of the Institution of Mechanical Engineers. Part B, Journal of Engineering Manufacture 2011;225:1977.

Index 2D analysis, 105 3D machining, 8 3D model, 110

equivalent delamination factor, 152 etching, 141 exposed fibers, 7

ablation depth, 23 abrasive waterjet (AWJ) machining, 2 absorption, 19 acoustic emission, 142 adhesion failure, 19 adjusted delamination factor, 151 analysis of variance, 171 anisotropic, 1 artificial neural network, 158 assessment of delamination, 147 assisting gas, 16

finite-element technique, 159 flat-bed scanner, 142 free-of-wear tool, 18 fringe pattern, 146

bond strength, 18 brad and spur, 168 Buckingham’s π theorem, 154 carbon fiber reinforced plastic (CFRP) composite, 31 coaxial gas stream, 16 cohesion failure, 19 cohesion substrate failure, 19 cohesive strength, 19 composite material, 139, 163 composite repair, 5 conventional delamination factor, 149 critical thrust force, 139 cutting kerf, 12 cw laser, 14 damage ratio, 150 damage zone, 154 delamination, 21, 139 delamination factor, 151 delamination size, 149 dimension-less parameter, 147 dimensional error, 95 dimensional parameter, 147 drill model, 122 drilling, 31, 121, 163 energy absorption, 21 epoxy film adhesive, 18

glass fiber, 163 grinding, 3 grit blasting, 3 heat accumulation, 14 heat-affected zone (HAZ), 9 helical milling, 156 high impact polystyrene (HIPS), 164 image processing, 142 intensity, 7 IR-laser, 22 laser material interaction, 9 line energy, 12 linear elastic fracture mechanics, 159 long fiber reinforced polymer, 103 maximum crack length, 148 measurement method, 141 mechanical vibration, 93 milling, 2, 156 minimum delamination factor, 155 multi-pass cutting, 17 multi-pass strategy, 13 numerical analysis, 157 orthogonal cutting, 105 peel-ply, 3 peel-up delamination, 140 phase-shifting, 146 photo-chemical, 8, 19 photo-thermal, 19 photon energy, 7 photothermal, 7 pixel, 142

196 | Index plasma process, 5 polymer matrix material, 171 polyphenylene sulphide, 10 pre-treatment, 3 pulse duration, 15 pulse repetition rate, 15 pulsed laser, 14 push-out delamination, 140 radiography, 144 refined delamination factor, 154 regression analysis, 157 response surface methodology (RSM), 165 – Box-Behnken design, 165 – design of experiments (DOE), 165 robotic trimming, 90 rotary ultrasonic machining (RUM), 31 – grinding, 33 – machinability, 77 scanning acoustic microscope, 143 scanning laser acoustic microscope (SLAM), 143 scarf ratio, 25 severity of damage, 160 shadow moiré phase-shifting interferometry, 146 shape circularity, 154 shape cutting, 5 simultaneous processing, 16 single fiber layer, 24

single lap-shear test, 19 stereo-microscopy, 142 surface, 95 surface activation, 5 surface energy, 5 surface quality, 97 thermal ablation, 21 thermal crack, 11 thermal damage, 117 thermoplastic, 9, 163 – GFRTP laminate, 164 thermoset, 7, 9 threshold parameter, 142 thrust force, 168 transmission spectra, 7 trimming, 87 two-dimensional delamination factor, 150 ultrasonic C-scan, 143 ultrasonic probe, 144 UV-laser, 22 vertical machining center (VMC), 168 visual method, 141 waterjet machining, 2 X-ray computerized tomography, 145