5th InternatIonal ConferenCe on Additive technologies

5th International Conference on Additive Technologies - iCAT2014

Proceedings Vienna, Austria, 16. - 17. October 2014 DAAAM Specialized Conference

Editor: Igor Drstvenšek

Proceedings of 5th International Conference on Additive Technologies

Title: 5th International Conference on Additive Technologies iCAT2014 Type:


Editor:

Igor Drstvenšek (University of Maribor, Faculty of Mechanical Engineering, Slovenia)

Layout & Design:

Dušan Pogačar

Publisher:

Interesansa - zavod, Ljubljana

Impression:

Digital edition

Year of publishing:

2014

Disclaimer: The Organizing Committee of 5 International Conference on Additive Technologies – iCAT2014 accepts no responsibility for errors or omissions in the papers. The Organizing Committee shall not be liable for any damage caused by error or omissions in published papers. th

All rights reserved. This publication may not be reproduced in whole or in part, stored in retrieval system or transmitted in any form or by any means without written permission of the Organizing Committee of 5th International Conference on Additive Technologies –iCAT 2014. © 2014, The Organizing Committee of 5th International Conference on Additive Technologies – iCAT 2014 Publisher Interesansa - zavod, Ljubljana

CIP - Kataložni zapis o publikaciji Univerzitetna knjižnica Maribor 681.5(082) INTERNATIONAL Conference on Additive Technologies (5 ; 2014 ; Vienna, Austria) Proceedings [Elektronski vir] / 5th International Conference on Additive Technologies iCAT2014, 16-17. October 2014 ; editor Igor Drstvenšek. - Digital ed. Ljubljana : Interesansa-zavod, 2014 Način dostopa (URL): http://www.icat.rapiman.net/ icat2014/icat2014proceedings.pdf ISBN 978-961-281-579-0 1. Drstvenšek, Igor COBISS.SI-ID 79841793

2


Table of Contents I. Organisers

8

II. Sponsors

9

III. Committees

11

IV. List of Authors

13

V. Foreword: Is Additive Manufacturing a Mature Technology

14

VI. Invited Speakers

17

PAPERS BY CATEGORY 1. Additive Manufacturing in Medicine 2. Design for Additive Manufacturing 3. Metals in Additive Manufacturing 4. Polymers in Additive Manufacturing

1. Additive Manufacturing in Medicine Holger Freyer, Andreas Breitfeld, Stephan Ulrich, Rainer Bruns and Jens Wulfsberg 3D-printed elastomeric bellow actuator for linear motion

29

Lars-Erik Rännar and Åke Hamberg Design and manufacture of a titanium tibial reinforcement cage using electron beam melting

33

Andrey Koptyug, Lars-Erik Rännar and Mikael Bäckström Multiscale surface structuring of the biomedical implants manufactured in Electron Beam Melting technology: demands, advances and challenges

37

Patricia Lopes, Markus Fremmer, Christoph Bichlmeier and Peter Verschueren Patient-Specific Cardiovascular Models for Educational and Training Purposes

43

3


Sven Maricic, Ivan Dogan, Lado Kranjcevic, Ana Pilipovic and Daniela Kovacevic Pavicic The Application of 3D Modelling in Biofluid Mechanics

47

Radovan Hudak, Jozef Zivcak, Bruno Goban, Martin Lisy, Lukas Marincak, Andrej Jenca, Andrej Jenca Jr. and Miroslav Gajdos Additive Manufacturing, Verification and Implantation of Custom Titanium Implants – Case Studies

51

Tomaz Tomazic, Slavko Kramberger, Attila Szunyog, Breda Jesensek Papez, Tomaz Brajlih and Igor Drstvensek Personalized Shoulder Endoprosthetic for Glenoid Defect – A Case Report

56

2. Design for Additive Manufacturing Stamatios Polydoras, Christopher Provatidis, Theodoros Vasilopoulos, Evangelos Theodorou, Georgios Theodorou and Vassiliki Mitsopoulou Techniques and practices for the successful, cost effective reconstruction of skeletal elements of the last European elephant of Tilos with LOM and FDM Additive Manufacturing technologies: An interdisciplinary approach of AM for palaeontology

62

Mathias Bratl Customizable Personal Manufacturing

70

Kirsten Lussenburg, Natascha van der Velden, Zjenja Doubrovski, Jo Geraedts and Elvin Karana Designing With 3D Printed Textiles

74

Cees Jan Stam, Natascha M. van der Velden, Gerard Rubio and Jouke Verlinden Redefining the role of designers within an urban community using digital design and localized manufacturing of wearables.

82

Bogdan Galovskyi, Tino Hausotte, Dietmar Drummer and Meng Zhao Model of a Measurement Artifact for Additive Manufacturing

90

Robert Ian Campbell, Yudhi Ariadi and Mark Evans Facilitating Consumer Involvement in Design

96 4


3. Metals in Additive Manufacturing Dhirendra Rana In-Situ Property Improvements of Additive Manufactured Objects Using A CNC Integrated Pneumatic Hammer

101

Manuela Galati, Alessandro Salmi, Eleonora Atzeni and Luca Iuliano Simulation of Material State Change and Thermal Distribution in Electron Beam Melting

109

Vera Juechter, Adam Schaub, Marion Merklein and Robert F. Singer Titanium metal sheet structures of various wall thicknesses with additional functional elements prepared by selective electron beam melting in a powder bed

119

Marcin A. Królikowski Customization of 2D lattice structures as response of part load conditions for SLM manufacturing

123

Ralf Guschlbauer, Matthias Lodes and Carolin Körner Selective electron beam melting of pure copper: influence of energy input on surface roughness and dimensional accuracy

129

Dariusz Grzesiak and Marta Krawczyk Effects of the Selective Laser Melting process parameters on the functional properties of the Co-Cr alloy

133

Saeed Khademzadeh, Nader Parvin, Paolo Francesco Bariani and Federico Mazzucato Geometrical characterization of thin walls produced by micro laser sintering

137

Daniel Koutny, Radek Vrana and David Palousek Dimensional accuracy of single beams of AlSi10Mg alloy and 316Lstainless steel manufactured by SLM

142

Daniel Koutny, Radek Vrana and David Palousek Laser Beam Melting of Amorphous Metals

148

Daniel Koutny, Radek Vrana and David Palousek Laser beam melting of high strength aluminium alloys EN AW-6061 and EN AW-6082

153

5


4. Polymers in Additive Manufacturing Charoula Kousiatza and Dimitris Karalekas On the integration of fiber Bragg sensors as an in-process sensing system in additive manufacturing

160

Theresa Swetly, Jürgen Stampfl, Gero Kempf and Rainer-Michael Hucke Elastic properties of Additive Manufacturing materials for automotive applications

166

Stephanie Fanselow, Jochen Schmidt, Karl-Ernst Wirth and Wolfgang Peukert Production of micron-sized polymer particles by melt emulsification

174

Marius Sachs, Jochen Schmidt, Stephanie Fanselow and Karl-Ernst Wirth Tailoring melting behaviour of LBM powders

178

Damir Godec, Ivan Vidović and Maja Rujnić-Sokele Optimization of low budget 3D printing parameters

183

Miquel Domingo, Salvador Borrós and Guillermo Reyes A methodology to choose the best building direction for Fused Deposition Modeling end-use parts

189

Jannis Greifenstein and Michael Stingl Simultaneous optimization of build orientation and topology in layered manufacturing

194

Tomaz Brajlih, Matej Paulic, Tomaz Irgolic, Ziga Kadivnik, Joze Balic and Igor Drstvensek Study of the complementary usages of selective laser sintering during the high volume production of plastic parts

202

Maximilian Drexler, Dietmar Drummer and Katrin Wudy Selective laser melting of polyamide 12 - Interaction between time dependent exposure strategies and part positioning

208

Dietmar Drummer, Katrin Wudy and Maximilian Drexler Selective laser melting of polyamide 12: A holistic approach for the modeling of the aging behavior

216

Stefan Josupeit, Lavish Ordia and Hans-Joachim Schmid Development of a Basic Model to Simulate the Laser Sintering Cooling Process

222

6


Thomas Stichel, Tobias Laumer, Philipp Amend and Stephan Roth Electrostatic Multi-Material Powder Deposition for Simultaneous Laser Beam Melting

228

Thomas Stichel, Tobias Laumer, Philipp Amend and Stephan Roth Polymer Powder Deposition using Vibrating Capillary Nozzles for Additive Manufacturing

236

Tobias Laumer, Thomas Stichel, Stephanie Fanselow, Jonas Koopmann, Philipp Amend and Michael Schmidt Generation of multi-material parts with alternating material layers by Simultaneous Laser Beam Melting of polymers

242

Andreas Wegner, Ron Harder, Gerd Witt and Dietmar Drummer Influence of Process Parameters on the Part Properties of Laser Sintered Polyamide 11

248

Meng Zhao, Dietmar Drummer, Katrin Wudy and Maximillian Drexler Sintering study of polyamide 12 particles for selective laser melting

256

Dietmar Drummer, Ron Harder, Gerd Witt, Andreas Wegner, Katrin Wudy and Maximilian Drexler Long-term properties of laser sintered parts of polyamide 12 Influence of storage time and temperature on the ageing behaviour

262

7


Organized by Rapid Prototyping and Innovative Manufacturing Network

RApid Prototyping and Innovative MAnufacturing Network

in collaboration with:

University of Maribor, Faculty of Mechanical Engineering - Department of Production Engineering Association for Promotion of Automation and Robotics F-AR Danube Adria Association for Automation & Manufacturing

DAAAM Specialized Conference Danube Adria Association for Automation & Manufacturing

And with special contribution of:

Collaborative Research Center 814 - Additive Manufacturing at Friedrich Alexander University Erlangen - Nürnberg Institute of Production Engineering at Technical University Graz University of Applied Sciences Campus 02, Graz Faculty of Medicine, University of Ljubljana

← Back to Table of Contents

8


Sponsors Golden sponsors

VoxelJet www.voxeljet.com

EOS www.eos.info

Invited Lecturer’s Sponsor

Materialise www.materialise.com


Mark Medical www.mark-medical.com

9

always a layer ahead 3D printing systems from voxeljet:

innovative, fast, precise and economical ConvinCe yourself anD visit the voxel jet booth at iCat

Manufacture of 3D printing systems & service center for on demand production of sand molds and plastic models. voxeljet AG

info @ voxeljet.com www.voxeljet.com

10


Committees MEMBERS OF THE ORGANIZING COMMITTEE

iCAT REVIEW COMMITTEE FOR 2014

Drstvenšek, Igor (President)

Abdel Ghany, Khalid (EG)

Malisa, Viktorio (Vice-president)

Balc, Nicolae (RO)

Komenda, Titanilla (Vice-president)

Brajlih, Tomaž (SI)

Drummer, Dietmar

Campbell, Ian (UK)

Haas, Franz

Drstvensek, Igor (SI)

Ihan Hren, Nataša

Drummer, Dietmar (DE)

Jopling, Daniel

Ihan Hren, Natasa (SI)

Pogačar, Dušan

Poukens, Jules (NL)

Poukens, Jules

Tomazic, Tomaz (SI)

Schmidt, Michael

Wohlers, Terry (USA)

Tomažič, Tomaž Traussnigg, Udo Wohlers, Terry


11


iCAT SCIENTIFIC AND PROGRAMME COMMITTEE FOR 2014

Abdel Ghany, Khalid (EG)

Haas, Franz (AUT)

Pritrznik, Lidija (SLO)

Agarwal, Mukesh (IND)

Hadzistevic, Miodrag (SRB)

R. Vulicevic, Zoran (SRB)

Avellan, Lars (S)

Hagiwara, Tsuneo (J)

Ramani (IND)

Balc, Nicolae (RO)

Hodolic, Janko (SRB)

Reinhardt, Andrea (DE)

Balic, Joze (SI)

Homa, Johannes (AUT)

Rotaru, Horatiu (RO)

Bartolo, Paulo (PT)

Hudak, Radovan (SK)

Schaefer, Martin (DE)

Bernard, Alain (FR)

Ihan Hren, Natasa (SI)

Schermer, Scott (USA)

Bertrand, Phillippe (FR)

Iuliano, Luca (IT)

Schindel, Ralf (CH)

Bjork, Lennart (S)

Schmidt, Michael (DE)

Campbell, Ian (UK)

Janse van Vuuren, Michaella (SAR)

Cardoon, Ludwig (NL)

Jay, Olivier (DEN)

Sinkovic, Andreja (SI)

Cota, Vesna (CDN)

Junghanss, Michael (D)

Smurov, Igor (FR)

Cus, Franci (SI)

Katalinic, Branko (AUT)

Stamenkovic, Dragoslav (SRB)

Dean, Lionel T (UK)

Kosica, Olga (SLO)

Strojnik, Tadej (SI)

DeBeer, Deon (RSA)

Kostal, Peter (SK)

Sercer, Mladen (HR)

Dickens, Philip (UK)

Kramberger, Slavko (SI)

Tetsuzo, Igata (J)

Dolinsek, Slavko (SI)

Krenicky, Tibor (SK)

Tomazic, Tomaz (SI)

Drstvenšek, Igor (SI)

Kruf, Walter (NL)

Tromans, Graham (UK)

Drummer, Dietmar (DE)

Kyttanen, Janne (FI)

Tuomi, Jukka (FI)

Dybala, Bogdan (PL)

Levy, Gideon (CH)

Velisek, Karol (SVK)

Frost, Noel (AUS)

Malisa, Viktorio (AT)

Wimpenny, David (UK)

Fuzir, Bauer Gabrijela (SI)

Mateazzi, Paolo (IT)

Wang, Xiaohong (PRC)

Gerrits, Anton (NL)

Murphy, Michael (IRL)

Wong, Martin (HKG)

Gibson, Ian (SGP)

Ohigashi, Norihito (J)

Wood, Duncan (UK)

Godec, Damir (HR)

Ostojic, Gordana (SRB)

Yan, Yongnian (PRC)

Gonzales, David (ES)

Pogacar, Vojko (SI)

Yang, Dong-Yol (KOR)

Grossman, Bathsheba (USA)

Porter, Ken (AUS)

Yeung, Millan (CDN)

Grzesiak, Andrzej (D)

Poukens, Julles (NL)

Zhang, Qiqing (PRC)

Grzic, Renata (HR)

Preez, Willie, du (RSA)


Schwartze, Dieter (DE)

12


List of Authors Amend, Philipp Ariadi, Yudhi Atzeni, Eleonora Bäckström, Mikael Balic, Joze Bariani, Paolo Francesco Bichlmeier, Christoph Borrós, Salvador Brajlih, Tomaz Bratl, Mathias Breitfeld, Andreas Bruns, Rainer Campbell, Robert Ian Dogan, Ivan Domingo, Miquel Doubrovski, Zjenja Drexler, Maximilian Drstvensek, Igor Drummer, Dietmar Evans, Mark Fanselow, Stephanie Fremmer, Markus Freyer, Holger Gajdos, Miroslav Galati, Manuela Galovskyi, Bogdan Geraedts, Jo Goban, Bruno Godec, Damir Greifenstein, Jannis Grzesiak, Dariusz Guschlbauer, Ralf Hamberg, Åke Harder, Ron Hausotte, Tino Hucke, Rainer-Michael Hudak, Radovan Irgolic, Tomaz ← Back to Table of Contents

Iuliano, Luca Jenca, Andrej Jenca, Andrej Jr Jesensek Papez, Breda Josupeit, Stefan Josupeit Juechter, Vera Kadivnik, Ziga Karalekas, Dimitris Karana, Elvin Kempf, Gero Khademzadeh, Saeed Koopmann, Jonas Koptyug, Andrey Körner, Carolin Kousiatza, Charoula Koutny, Daniel Kovacevic Pavicic, Daniela Kramberger, Slavko Kranjcevic, Lado Krawczyk, Marta Królikowski, Marcin A. Laumer,Tobias Lisy, Martin Lodes, Matthias Lopes, Patricia Lussenburg, Kirsten Maricic, Sven Marincak, Lukas Mazzucato, Federico Merklein, Marion Mitsopoulou, Vassiliki Ordia, Lavish Palousek, David Parvin, Nader Paulic, Matej Peukert, Wolfgang Pilipovic, Ana Polydoras, Stamatios

Provatidis, Christopher Rana, Dhirendra Rännar, Lars-Erik Reyes, Guillermo Roth, Stephan Rubio, Gerard Rujnić-Sokele, Maja Sachs, Marius Salmi, Alessandro Schaub, Adam Schmid, Hans-Joachim Schmidt, Jochen Schmidt, Michael Singer, Robert F. Stam, Cees Jan Stampfl, Jürgen Stichel, Thomas Stingl, Michael Swetly, Theresa Szunyog, Attila Theodorou, Evangelos Theodorou, Georgios Tomazic, Tomaz Ulrich, Stephan van der Velden, Natascha Vasilopoulos, Theodoros Verlinden, Jouke Verschueren, Peter Vidović, Ivan Vrana, Radek Wegner, Andreas Wirth, Karl-Ernst Witt, Gerd Wudy, Katrin Wulfsberg, Jens Zhao, Meng Zivcak, Jozef 13


Is Additive Manufacturing a Mature Technology In 2011 3D printing became a hot topic in many journals and almost every decent journalist tried to cover the topic of Additive Manufacturing in their journals. Although public interest has mostly been triggered by the potential possibilities of using these technologies in medicine for printing »spare parts« for humans such kidneys, liver, etc or even for printing everything imaginable at home, this new interest has pushed many new applications into more conventional environments. Industry has been using these technologies for almost 30 years to overcome obstacles during the design and development processes. However, with the developments of new materials and machines the products of Additive Manufacturing became useful in everyday life although average humans were unaware of the existence of 3D printers. Spawned by the new interest of journalists, Additive Manufacturing soon moved into the agendas of many politicians and decision makers, thus enabling Additive Manufacturing to reach third place amongst the top ten Strategic Technology Trends for 2015. This evolution took almost thirty years and I am proud that our International Conference on Additive Manufacturing - iCAT has been a part of this process. iCAT conferences are products of the first Rapid Prototyping workshop held at the Faculty of Mechanical ← Back to Table of Contents

14


Engineering in Maribor, Slovenia in 2004 when the Rapid Prototyping and Innovative Manufacturing Network – RAPIMAN was launched by the participants from Loughborough University (Dr. Ian Campbell), the Technical University of Cluj-Napoca (Dr. Nicolae Balc) and the University of Maribor. Three years later the first iCAT was held in Celje, Slovenia and was then followed by 3 successful events that gathered together more than 400 experts from all over the world. This year at the 5th iCAT we are celebrating the 10th anniversary of RAPIMAN and I can know that our conference has always been one of the top scientific events that follows and introduces the more important current research topics. In addition, this year the Rapid Prototyping Journal celebrates its 20th anniversary as the very first journal dedicated to additive manufacturing. RPJ has been the vanguard of exciting and surprising developments from around the world, publishing a wide range of seminal articles along the way. The organizers of iCAT conferences are proud that RPJ has supported our conferences from the very beginning by providing the Best Paper Award and by its valuable panel of experts who have cooperated with us over all those years. Following published research in both RPJ as well as at the Proceedings of iCAT, we can clearly see the shift of interest from pure application-driven prototyping towards more production-driven demands. For many years experts have been trying to push the additive technologies where applicable into industrial environments in order to replace some conventional manufacturing processes. Unfortunately many attempts failed because of the extremely low reliability and efficiency levels of the available AM machines. It was soon clear that the AM industry was immature as yet and that it still struggled with very basic problems of running the machines. Fortunately with the recent popularity of AM, investors have started to support the industry’s efforts toward a more reliable machine base. The scientific papers also show this significant shift, which is highly obvious in the collated papers for this year’s iCAT. The majority of papers deal with problems that prevent the AM technologies from being used for industrial manufacturing. The described problems and solutions range from definitions of materials’ behaviour in regard to age-related complications and from working parameters to the accuracies of end-parts. The knowledge included within the presented solutions is very broad and interdisciplinary, thus making the solutions reliable and versatile. I can conclude by stating that we are facing Additive Manufacturing’s coming of age and the moment when many of the journalists’ prognoses and our secret wishes will come true. Prof. Dr. Igor Drstvenšek President of the organizing committee for iCAT 2014 ← Back to Table of Contents

15


Invited Speakers de Beer, Deon / Campbell, Ian / Cass, William J. / Christensen, Andy / Diegel, Olaf / Drummer, Dietmar / Homa, Johannes / Hudak, Radovan / Poukens, Jules / Van Straaten, Willie / Tomažič, Tomaž

Deon de Beer Technology Transfer and Innovation Support Office, North West University, South Africa Using AM to Revitalise Age Old Industries, Following a Sectoral Approach

Following some of Michael Porter’s theses, Deon discusses how age old industries can be turned around, using a fresh and innovative approach, using CAD, CAM, Reverse Engineering, Additive Manufacturing and Rapid Tooling. The paper focuses on the use of Additive Manufacturing (AM) platforms to develop an innovation support strategy for the Vaal Region, following an industry cluster approach. Importantly, the early results indicate that the local / regional approached taken, are equally applicable towards establishing national support strategies / platforms. In addition, both the impact on the host institution, as well as the region forms part of the review (following Porter’s theses). Of special importance, is the way in which the activities also created a foundation to develop a rural science and technology park development strategy, which may have significant impact on the future of science park development strategies for developing economies? Interestingly, the results also show that whilst incubation and new venture creation are important aspects of successful local and regional economic development strategies and obvious areas where the use of AM can play a significant role, it is as important to find support ← Back to Table of Contents

16


strategies for existing/conventional industries to maintain job opportunities and as such, help to “save” existing industries. In addition, the contribution made to AM development in South Africa by using a focused, industry supported cluster approach, has been highlighted.

Ian Campbell Loughborough University (UK) Facilitating Consumer Involvement in Design

This paper reports investigations into the potential for consumers to actively design their own desired products and thereafter to endorse them for manufacture. This idea emerged in anticipation of the rapid growth of low-cost fabrication technology, particularly 3D Printing. Recent developments in 3D Printing have led to renewed interest in how to manufacture customised products, and specifically, in a way that will allow consumers to create bespoke products more easily. However, the entry point to 3D Printing is typically a 3D computer aided design (CAD) model, and most CAD systems are typically difficult for non-experts to use. Consequently, to make 3D Printing more accessible to consumers, design systems need to be developed that are as easy to operate as are the 3D Printers themselves. This research reports on the development of a Computer-aided Consumer Design (CaCoDe) system that is designed to simplify the CAD process for non-designers. The analogy is giving the user a few blades to use rather than a whole Swiss Army Knife. The software, which has been developed using a Rhino and Grasshopper platform, is an easy-to-operate design system, where consumers interact with the dimensional parameters of pre-designed templates through on-screen slider-bars and pick-and-drag mouse movements. To determine the potential for consumer-led ← Back to Table of Contents

17


design, a range of product design templates were implemented using the software and then evaluated by non-designers. The results showed that the majority of the individuals found the system to be userfriendly and understandable. The results were used to define the final version of the software and to make recommendations for future developments in this area.

William J. Cass Cantor Colburn LLP, USA Additive Manufacturing and Intellectual Property

Understanding the challenges concerning the intellectual property (IP) rights associated with additive manufacturing is key to the development and use of this technology. The increased ability to replicate designs through various technologies (such as laser scanners) and convert them into CAD (design) files and STL (printing) files means that the technology has the potential to be a disruptive technology. The expiration of key patents, technical improvements and the changing legal landscape, mean that traditional ways of addressing IP need to be revisited. Global IP strategies through the Patent Cooperation Treaty, Berne Convention on copyrights, and the Hague System for the International Registration of Industrial Designs will be discussed. Bill Cass has tried cases as lead trial counsel before juries and judges involving millions in alleged damages for both plaintiffs and defendants involving all forms of intellectual property. Recently, Bill successfully defended Faro Technologies, Inc. in a two week jury trial in a case brought by Nikon Metrology, Inc., which involved laser scanners mounted on articulated coordinate measuring machines. His cases involve complex technology, including medical devices, circuitry, mechanical engineering, material science, chemistry, and computer software. ← Back to Table of Contents

18


Bill holds a mechanical engineering degree from Worcester Polytechnic institute, where he concentrated in robotics. He also holds a commercial multiengine instrument rating and advises clients in aviation technology. Bill has spoken on topics concerning intellectual property internationally at EuroMold, Rapid (SME), and also at a recent White House Symposium on Additive Manufacturing. Bill has appeared throughout the United States in various federal courts.

Andy Christensen 3D Systems, Vice President of Personalized Surgery and Medical Devices (USA) Personalized Surgery and the Future of Medical Applications for 3D Printing

Andy Christensen has been active in the additive manufacturing (AM) industry since the early 1990’s. From 2000 to 2014 he was the President and Owner of Medical Modeling Inc., a world-leading medical device AM service bureau based in Golden, Colorado. On April 2, 2014 Medical Modeling was acquired by 3D Systems Corporation. In his new role with 3D Systems Mr. Christensen will act as VP, Personalized Surgery & Medical Devices heading up this new business vertical. His educational background is a BS in Business from the University of Colorado at Denver. The vision that 3D Systems is now carrying out is to improve patients’ lives through use of AM technology and personalized solutions. Each day around the globe 3D Systems products are used to provide surgical teams more confidence and provide patients with excellence in reconstructive surgical care through use of patient-specific anatomical models, personalized surgical guides/instruments and AM-fabricated implants. Mr. Christensen has been actively involved with the Society of Manufacturing Engineers RTAM (Rapid Technologies and Additive Manufacturing) ← Back to Table of Contents

19


technical community for many years and is a past chair of the SME/RTAM Steering Committee. He has authored three book chapters and a number of articles on the use of AM technology for medical device applications. He is a recipient of the SME/ RTAM Industry Achievement Award, a prestigious award given for groundbreaking work in the AM industry. Mr. Christensen has a research interest in medical applications of additive manufacturing, metal additive manufacturing for implants, methods of surgical personalization and innovative biomaterials.

Olaf Diegel Faculty of Engineering Lund University, Sweden 3D Printing: Bridging the Creative Gap?

In the near future 3D printing will have a marked effect on how we order, design, and manufacture products. They will have a major Impact not just on products, but on our society, and how we live and do business. Besides its obvious technological and manufacturing benefits, 3D printing also has the potential to unleash innovation in an unprecedented manner. Though there is much research on the engineering and technical aspects of 3D printing, some of this research can almost be seen as constraining innovation, rather than encouraging it. Perhaps it is time to start looking beyond the engineering aspects of 3D printing, and look at how we can use it to become more creative? This presentation examines where 3D Printing technologies are from the point of view of stimulating and exploiting innovation, and how engineers should be encouraged to work more closely with artists and other innovators from outside the sphere of engineering in order to broaden their innovative potential. We also examine how additive manufacturing can be successfully implemented as part of the production chain and how it can be used, today, to successfully bring new products to market. ← Back to Table of Contents

20


Dietmar Drummer Friedrich Alexander University Erlangen Nürenberg, Institute of Polymer Technology, Germany Perception and reality of additive polymer processing

With almost unlimited freedom of design, additive manufacturing technologies open up new perspectives to achieve individual solutions. These types of manufacturing techniques barely set any limits to the spirit of innovation. Due to this fact additive manufacturing techniques follow the trend towards customized products and will allow for serial production in the future. Despite the high potential of additive processing of polymers, the step into serial production of highly individualized products was yet not realized. Nevertheless the medial driven hype related to additive processes is still increasing. But what is a realistic vision, which can be realized by scientific work, and what are pure dreams? Within this invited lecture reasons for these contrary trends are discussed. First an overview about different state of the art additive processes, using polymers, is given. Moreover typical process related potentials are discussed. Furthermore the invited lecture shows challenges scientists have to investigate for using the high potential of additive polymer processing. Overall the lecture tries to separate realistic visions from pure dreams related to additive manufacturing processes.


21


Johannes Homa Lithoz GmbH, Austria Lithography-based Ceramic Manufacturing – Additive Manufacturing of High-Performance Ceramics

In the field of ceramic processing there is a strong need for the introduction of Additive Manufacturing (AM) techniques. Tools for powder injection molding (PIM) are very expensive and require significant lead times which severely restrict the suitability of PIM for the production of small scale series or customized products; however, no adequate prototyping technology existed so far. The main reason for that are the high demands on high-performance ceramics – these materials are used where other materials fail, thus the quality and the reliability of the parts are crucial. This presentation focuses on a novel AM-approach, which is capable of producing strong, dense and accurate ceramic parts, namely the Lithography-based Ceramic Manufacturing (LCM). It is a slurry-based process, where a photocurable monomer system is mixed with ceramic powder and hardened through mask exposure to give the green body. In this state the generated photopolymer network acts as scaffold and binder for the ceramic particles. This polymer is later on removed at elevated temperatures and the resulting pores are subsequently eliminated during sintering. Thus, this method leads to highly dense ceramics with mechanical characteristics very similar to conventionally formed parts; for alumina a theoretical density of over 99.4 % and 4-point bending strength of over 430 MPa has already been realized. Moreover, due to its layer-by-layer approach, the LCM-technology provides the opportunity to shape highly complex and intricate geometries that cannot be realized by conventional means. Holes with a diameter of 200 μm and a wall thickness of below 150 μm could be realized by this technology to date. These characteristics render the LCM-technology a capable addition to conventional processing techniques in the field of ceramics.


22


Radovan Hudak CEIT Biomedical Engineering Ltd & Technical University of Kosice, Slovakia Additive Manufacturing, Verification and Implantation of Custom Titanium Implants

The paper focuses on three selected case studies of designing, manufacturing and application of custom implants. They include two cranial implants and one large maxillofacial implant. The implants were designed while minimising their weights and applying porous titanium structures. Direct Metal Laser Sintering Technology (DMLS) is one of the additive technologies currently used for the manufacture of custom implants made of Ti-6Al-4V titanium alloy. Technology (EOSINT M280, EOS GmbH, Germany), titanium material and used software applications (Materialise, Belgium) facilitate the manufacture of implants while maintaining the bionic principles (mechanic, rheological and anthropometric). For the verification purposes, in the manufacture process and in the postprocessing the metrotomography technologies (Metrotom 1500, Carl Zeiss, Germany) and the Scanning Electron Microscope (SEM) were used. The analysis to compare the current CAD model and the STL model, defectoscopy after the manufacture, and the analyses of implants after the thermal processing and surface finishing were carried out. Manufactured implants were successfully implanted at the Clinic of Stomatology and Maxillofacial Surgery of the Louis Pasteur University Hospital in Kosice, Slovakia, in cooperation with expert neurosurgeons and using the prepared navigation surgical guides. Implant quality and surgical intervention success are significantly influenced by a proper implant design based on the communication with the clinical staff, by setting the parameters of the manufacture process (DMLS), as well as the post-processing parameters.


23


Jules Poukens University Hasselt and Leuven, Belgium Need a New Skull or Mandible? 3D Print It!

Patients in the cranio-maxillofacial clinic often present with serious, complex, and potentially life- threatening or life-limiting medical conditions (e.g. tumor, trauma, aggressive osteomyelitis). Available treatments may not always give satisfactory results for patients and doctors. Therefore, complex problems ask for new solutions. An emerging technique in the medical field is Computer Aided Design (CAD), Computer Aided Manufacturing (CAM) by 3D printing. For successful implementation of CAD-CAM technology in the clinical practice doctors, dentists and engineers need to work together and share their expertise. This intense cooperation leads to 3D printing of custom patient specific implants. 3D printed implants are used for the treatment of skull defects, dental superstructures and world’s first 3D printed entire mandible replacement implant. Clinical cases will be highlighted. Prof. Dr. Jules Poukens is currently lecturer and researcher at the Biomed Research Institute of the University Hasselt and the University of Leuven in Belgium. He is a Cranio-Maxillofacial Surgeon at the Medical Center Sittard /Heerlen in the Netherlands. He was the leading surgeon that designed and implanted the world’s first 3D printed total mandibular implant and was one of the pioneers in using 3D printed skull implants. He has numerous publications and held numerous presentations in this field.


24


Willie Van Straaten !nventec, South Africa Exploring Additive Manufacturing through the lens of Value Innovation

There are major forces reverberating through our world today - impacting individuals, companies and industries alike, including those involved in additive manufacturing. The key questions to be addressed are:

Firstly, how to effectively navigate the trends and forces of change that are sweeping the business environment today; secondly, and perhaps more importantly, how to exploit these changes as opportunities within the realm of additive manufacturing; and thirdly, what can be done to advance the case of additive manufacturing as a disruptive and enabling technology. One perspective is to investigate challenges and opportunities facing additive manufacturing through a manufacturing lens. Such a view, however, may be too limiting to be practical and productive. An alternative point of view would be to apply a wider lens and see additive manufacturing as a link in a value chain, i.e. part of a sequence of activities aimed at progressively transforming a product idea into a market offering that is responsive to the needs and wants of the target customer. Such a perspective leads to the concept of value innovation - rooted in the capacity to think creatively and innovatively - a capability considered by leading companies around the globe as key to gaining competitive advantage and profitability. By placing additive manufacturing within the context of value innovation, the perspective of the practitioner can be broadened and attention can be directed and focused on the stages, phases and sequence of activities necessary to create product offerings that are competitive, unique and valuable. To highlight the potential benefits and implications of such an approach for the practitioner of additive manufacturing, value innovation and its underlying components will be discussed by answering questions such as: What is meant by value? How is value created, delivered and captured? How to generate and evaluate new product ideas? ← Back to Table of Contents

25


What is a customer value proposition (CVP) and how must it be defined in order to resonate with the target market? What is the interrelationship between creativity, invention, innovation and a business model? Practitioners of additive manufacturing are currently engaged in a wide array of activities along the value chain where they act, for example, as makers, hobbyists, marketers, product developers, suppliers and manufacturers. Surrounding and impacting upon these activities are major trends such as the growth and use of the internet, widespread social connectivity, rapid technological advances and fierce competition. Furthermore, the convergence of additive manufacturing with related technologies - intelligent software, novel materials, dexterous robots, and a range of web-based services, offer those involved in additive manufacturing a wide array of options to compete, innovate, and grow profitably. It is proposed that by embracing a value innovation perspective, practitioners of additive manufacturing will be better equipped to leverage the strengths of additive manufacturing, capitalize upon the new and emerging technologies and ultimately, create competitive and profitable products capable of delivering value and satisfaction to the customer.

Tomaž Tomažič TH Murska Sobota, University of Maribor, Slovenia Personalized Shoulder Endoprosthetic (PSE) for Defective Glenoids

A defective glenoid with pore bone stock is the most challenging problem in the shoulder endoprosthetic (EP). For these severe cases mostly a bone graft for restoration of glenoid shape and implant position in the functional plane is needed. Due to the new personalised resection guide technique developed with Slovenian technology, the determination of the optimal bone cuts, best anatomical axis and ← Back to Table of Contents

26


glenoid position in the available bone stock upon the preoperative 3D-CT joint model, is easier and more precise. Our goal is the most optimal bone fixation of the glenoid implant in the available bone stock, without stressing the primary on its retroversion and rather adapts the humerus head retroversion, which has been the primary rule for resection in the conventional technique. Due to this 3D preoperative model, we can create resection guides with optimum fits to the bony surface of the patient’s bone and determine the exact intraoperative bone cuts, the shape and size of the bony transplant, as well as the optimal positioning of the EP, which enables the best stability and kinematic of the prosthesis. The postoperative 3D-CT EP positioning has been analysed by comparing the planned and performed resections and the final joint biomechanics. The results showed an easier, quicker and more precise procedure, with less postOP-3D-CT mechanical axis outliers. Due to the optimal bone fit and the combined retroversion of both components, we expect optimal biomechanics, functions and prolonged EP survival.


27


ADDITIVE MANUFACTURING IN MEDICINE

CHAPTER 1.

Additive Manufacturing in Medicine


28

3 Proceedings of 5th International Conference on Additive Technologies


3D-printed elastomeric bellow actuator for linear motion H. Freyer, Institute of Machine Elements and Technical Logistics, Helmut Schmidt University Hamburg, Germany, [email protected] A. Breitfeld, Institute of Production Engineering, Helmut Schmidt University Hamburg, Germany, [email protected] S. Ulrich, Institute of Machine Elements and Technical Logistics, Helmut Schmidt University Hamburg, Germany, [email protected] R. Bruns, Institute of Machine Elements and Technical Logistics, Helmut Schmidt University Hamburg, Germany, [email protected] J. Wulfsberg, Institute of Production Engineering, Helmut Schmidt University Hamburg, Germany, [email protected] Abstract—This paper shows the design and production process of an elastomeric bellow actuator, which generates a specific lengthening by only using insides pressure changes. The design and development process was intensively influenced by the possibility of using the 3D-printing technology. Additionally measurement results of the manufactured bellow actuator with regard to the lengthening, the cycle time, the durability and the limitation of uses are shown. Summing up an electrorheological micro-actuator-system is presented. This system is a hydraulic system based on the property of electrorheological fluids. The integration of the micro-valves inside of the actuator enable to build a system which is completely leakage free. Keywords: elastomeric bellow, electrorheological fluid, linear motion, polyjet-printing

1. INTRODUCTION At present, the development of mechatronic systems is being enhanced and accelerated by the integration of Rapid Manufacturing technologies in multiple phases of the product lifecycle [1]. Regarding product design, modeling, construction and fabrication, Rapid Manufacturing Technology offers great opportunities. Reducing the time-to-market by accelerating the design process and ultimately reducing research and development costs [2,3]. Nevertheless, today’s Rapid Manufacturing technology mainly focuses on the production of rigid parts and components, neglecting elastomeric elements. The reasons can be found in limited material choice of traditional Rapid Manufacturing technologies such as SLS or FDM [4].

Today, elastomeric elements of mechatronic systems are produced using traditional molding procedures such as injection molding or vacuum casting. For these conventional manufacturing technologies expensive casting molds are needed. In early phases of the product development process, design and layout changes are common. Therefore the expenditures and time, which are needed to realize these changes need to be minimized in order to reduce development costs as well the time-to-market. A first step to lower the costs is to generate the casting molds by using rapid prototyping technologies with the development of multi-material Rapid Prototyping technologies such as Polyjet-Printing. Using this technology, rigid as well as elastomeric components can be produced. This Rapid Manufacturing technology of STRATASYS Inc. offers several elastomeric-like materials (fig.1), which can be combined with rigid materials in the same printing process. This printing technology can be used to accelerate the whole process by fabricating the elastomeric prototypes directly. Polyjet-Printing was used to design and produce an elastomeric bellow for limbless and linear locomotion by using only inside pressure changes. BACKGROUND - LOCOMOTION Limbless motion can be conducted by various locomotion strategies. One of them is the 1-dimensional peristaltic movement. This strategy is successfully used by earthworms and snakes. Travelling longitudinal bodywaves, opposite to the movement direction, shove the earthworm’s sections over the ground (Fig.1). 2.

This research is supported by the Research Institute for Materials, Fuels and Lubricants, Germany.


29



combined with section 2 generates half of the elongation of each wave. In this section the inner diameter (Di) is the most important dimension. The difference between the outer and inner diameter considering the wall thickness is defined as h. At the beginning of the design process (Figure 3 a), a reference bellow actuator with a constant wall thickness was defined in a CAD-System. From this initial position (Figure 3 (a) the following property and performance goals influenced the design process:

Figure 1. peristaltic body wave

The average speed of the system is directly proportional to the number of segments in the wave N. Considering the displacement of an increment W, the cycle time of a body wave T, the number of segments per wave N, the relative elongation and the initial length of a segment the average velocity can be defined to:

.



minimum stress in the material during the elongation,



maximum axial-elongation capability,



minimized expansion of outer diameter (section 1),



sufficient internal space for valve integration.

The first goal - to minimize the material stress is solved by using relatively high radii in every possible design element of the bellow actuator. To reach a high axial-elongation the sectional wall length needs to be maximized. This is realised by maximizing the relation of outer and inner diameter. This relation can be increased by designing indentations in the waves (b). However, in this case, the lateral stiffness is decreasing because of the minimized inner diameter.

(1)

3. DEVELOPMENT PROCESS OF THE BELLOW ACTUATOR After analysing the core principles of axial-elongation as conjugant phenomenon of material stretch, wave-number and wave geometry bending similar to flexure hinges, a CADmodel with an axial-elongation optimised wall and body geometry was created incrementally.

Figure 3. design steps of an bellow actuator

Another point is the expansion of the bellow actuator during its use with the inner pressure. There are several solutions to minimize the lateral expansion. Exemplary, stiffening rings can be integrated in section 1 of the bellow actuator. These rings could be separated or integrated into the bellow. Because of using the polyjet printing technology it is easy to use the integrated type of stiffening rings, because in this technology it is possible to generate rigid and elastomeric parts in the same manufacturing step. Both actuators have been produced and compared. Figure 2. cross section of an bellow actuator

Therefore the bellow actuator has to be divided into three sections (fig. 2). Section 1 is the connecting zone to the next actuator and can be defined by the outer diameter (Da) and the wall thickness (g1). Section 2 is defined by the outer diameter (Da) and the second wall thickness (g23). The third section


30



For all simulations previously determined material parameters of the prototyping material called Tango Black [7] have been used.

Figure 4. elastomeric bellow actuator with integrated rigid stiffness rings (left hand), comparison of bellow actuator with rigid and elastomeric stiffening rings (right hand)

Figure 4 shows the elongation of the bellow actuator with rigid and elastomeric stiffening ring at the outer diameter. Both actuators have the same reaction and there is no expansion in the lateral direction. However the connection between the rigid and the elastomeric material is not as good as it needs to be. At an inside overpressure of 900 mbar, the connection between the elastomeric material and the rigid material dissolved, even though that the two materials have been manufactured with overlapping slices. For this purpose and the fact, that the sustainable solutions should be produced in an commercial way, the decision have been made in favour of the solution with the elastomeric integrated stiffening ring, which is integrated into the innerside of the bellow actuator (Figure 3 c). By combining the two solutions b and c a bellow actuator was constructed, which has low stress in the material, negligible lateral expansion in the section 1 and sufficient space inside for the controlling valve (Figure 3 d). The dimensions of the final bellow actuator are shown in Table 1. TABLE I.

Figure 5. comparison of simulation and printed prototype version of the bellow

Figure 5 shows on the left hand a the simulated bellow actuator model in the starting position and in an elongated situation with an inside over pressure of 0.2 bar. Figure 5b displays the printed bellow actuator in the intial position with only the ambient pressure inside. On the right the elongated bellow actuator with an inside pressure of 0.22 bar can be observed. By comparing the different parts of the picture – especially a) and c) it is recognizable, that the results of the simulation and the reality are close together. However increasing the inside pressure the simulation shows a swelling section 3 until the maximum analyzed pressure of 1 bar is reached. The prototyped elastomeric material gets micro cracks at a level of 0.5 bar which defines the maximum useable inside pressure. Otherwise the micro cracks destroy the tightness of the bellow actuator. The arising micro cracks are the result of the connection between the manufacturing process caused support material and the elastomeric model material. These connections causes a rough surface structure at the elastomeric material, which makes it prone to cracking.

DIMENSIONS OF THE FINAL BELLOW ACTUATOR Bellow actuator dimensions

Description

Symbol

Value

Internal diameter

Di

5 mm

External diameter

DA

20 mm

Bellow wave height

h

5 mm

S1

1,5 mm

S23

1 mm

Lbellow

20 mm

Wall thickness, section1 Wall thickness, section 2 and 3 Overall length, w/o connection zone

This finalized version of the bellow actuator has run through several steps of simulation for the linear elongation. Especially because a design should be generated which could easily be scaled to other dimensions, with a specialized focused on the opportunity to vary the inner diameter to integrate the control valves for different needed force and pressure level applications.


Figure 6. performance chart of the final elastomeric bellow

The performance chart (Figure 6) for the 3D-printed final elastomeric bellow (Figure 2) shows that the bellow behaves very linear for frequencies up to 3 Hz. Elongations of over 6 mm result in a drastically reduced life span of the bellow. After checking the deformation of the final version of the designed and prototyped bellow actuator, a scaled bellow actuator with the aim to integrate an electrorheological twin valve has been designed [5]. Additionally, it is expected that by using conventional elastomeric materials, the maximum cycle

31


number of the actuator should increase significantly. Therefore a polyjetted moulding device has been manufactured.


5. OUTLOOK Now the bellow actuator can be glued together, so that the valve gets integrated into the bellow. By combining the electrorheological technology with a bellow actuator for linear motions or movements, a hydraulic system can be realized, which distinguished itself by a small space requierements, completely leakage free, noiseless and because of no relative movements an abrasion free usability [5].

Figure 7. moulding of the final bellow actuator

In this moulding the new bellow had been manufactured using a vacuum casting technology with a 2 - component polyurethane elastomeric casting material.

Figure 9. electrorheological bellow actuator system

The elongation of this system can be realized by simply changing the electrical field in the electrorheological microvalves [6]. In a next step this actuator system needs to be measured and at the end be implemented in the linear motion system. Figure 8. peformance chart of the scaled bellow actuator

Figure 8 shows the performance of the scaled bellow actuator manufactured with a PU elastomeric material. This material has a lower Shore hardness, which results at the same pressure a higher lengthening than the Rapid Prototyping material. However comparing with Figure 6 the deformation behaviour is nearly the same. Only hat higher frequencies and higher pressure – the polyurethane material has longer and more linear behaviour. This effect based on the structure in the printed elastomeric material. In there are very small rigid structures to define a specific Shore hardness. These structures are the cause for the different deformation behaviour. 4. CONCLUSION Summing up all the measurements it is obvious, that for small elongations, low frequencies and low stress in the material – which means low inside pressure, the elastomeric printed material is a very good possibility to check the reaction of the contour and the deformation behaviour. However the printed material is, regarding its structure and the tendency to obtain micro cracks, not recommendable for applications, where high material stresses and deformations are expected.


ACKNOWLEDGMENT This research is supported by the Bundeswehr Research Institute for Materials, Fuels and Lubricants, Erding, Germany. REFERENCES [1] [2] [3] [4] [5]

[6] [7]

Bertsche, B., 2007, ‘Development and test of innovative products Rapid Prototyping’, Springer, pp. 330–336. Goetz, W., 2014, , 'Ready to use in three days', www.industrieanzeiger.de, 12.03.2014. Yeong, W., Chua, C, Leong, K. and Chandrasekaran, M., 2004, ‘Rapid Prototyping in tissue engineering: Challenges and potential', in Trends in Biotechnology, vol. 22, Issue 12, New York: Elsevier. Sollier, E., Murray, C., Maoddi, P., Di Carlo, D., 2011,‘Rapid Prototyping polymers for microfluidic devices and high pressure injections’. in Lab Chip, vol.11, Issue 3752, 09.09.2011. Breitfeld, A., Freyer, H., Ulrich, S., Bruns, R., Wulfsberg, J., 2014, ‘Elastomeric Bellows Hydraulic Actuator with Integrated Electrorheological Control Valves’’, Acutator Proceedings, pp.20-34, in press. Freyer, H., Breitfeld, A., Ulrich, S., Schneider, S., Bruns, R., Wulfsberg, J., 'Electrorheological microvalves', WITpress,Advances in Fluid Mechanics , vol. 10. Seido-Systems, http://www.seidosystems.com/Product/DetailOffer?groupId=210316&productId=10572

32



Design and manufacture of a titanium tibial reinforcement cage using electron beam melting Lars-Erik Rännar, Dept. Quality Technology, Mechanical Engineering and Mathematics, Mid Sweden University. Östersund, Sweden, [email protected] Åke Hamberg, Dept. Orthopedics, Östersund Hospital, Östersund, Sweden, [email protected]

Abstract—Additive manufacturing (AM) of patient-specific implants and prosthetic devices has already proven to be valuable in many different areas, for example in orthopaedic surgery and in craniomaxillofacial surgery. Due to its nature, this manufacturing method is suitable for customized products and in this paper, we present a case study where AM, more specifically the Electron Beam Melting (EBM) method, has been used for the manufacture of a titanium tibial reinforcement cage. The aim of this work was to design and manufacture a tibial reinforcement cage, that mimicked the geometry of the inner cortical surface of the proximal tibia. The patient (who participated with informed consent) suffered from septic loosening of a total knee replacement (TKR). Due to compromised collateral ligaments, severe osteolysis and bone defects, bone impaction grafting and a rotating hinge prosthesis was chosen for reconstruction. Since the bony rim of the proximal tibia was deficient, a supportive structure, more rigid than the ones that are commercially available, was needed in order to achieve a stable fixation. Computed Tomography (CT) images were used to reconstruct the tibia and this model was then used as a reference for a surface model of the cortical proximal tibia. A thickness of 0.6mm was assigned to the 3D model; the model was designed with perforations (~500) with a diameter of 2mm in order to facilitate vascularization and bone ingrowth between the bone graft and the host bone. The reinforcement structure was thereafter manufactured in biocompatible Ti6Al4V alloy using the EBM method. After sterilization, the structure was used during revision surgery and bone impaction grafting was used in order to restore bonestock, The 15 and 52 week post-surgery follow-up showed satisfactory results with no signs of radiolucency or loosening. The patient was able to walk without walking aids. AM can bring about new opportunities in the treatment of complex bone defects. In comparison with ordinary rim mesh, customized reinforcement cages gives the orthopaedic surgeon more options regarding rigidity, designed surface topography enhancing bony ingrowth and also designed porosity for vascularization between the bone graft and host bone. Keywords - Total knee replacement, reconstruction, electron beam melting, bone defect, bone impaction grafting, uncontained defect, titanium cage, reinforcement ring

This work was partially funded by the Swedish Agency for Economical and Regional Growth (Tillväxtverket) and the European Regional Development Fund.


1. INTRODUCTION The world is facing a great challenge, with a population that is living longer than previous generations and for the orthopaedic field, this has increased the number of patients with severe bone defects such as osteolysis. However, new technologies such as Additive Manufacturing (AM) can provide new solutions where there is a need for implant components with complex patient-specific designs. AM has evolved considerably in recent decades; from being a method for building plastic prototypes to being a range of different methods for manufacturing parts in different materials such as paper, plastic, metal, ceramic and even food and living tissue. This work focuses on the additive manufacturing method of Electron Beam Melting, EBM (Arcam AB, Sweden), which was originally developed for processing tool steel [1, 2] but is now commonly used for processing different titanium alloys for aerospace and orthopaedic implants. The EBM method and the material Ti6Al4V have already proven to be valuable for customized implants. Initial studies in the early 2000s showed that EBM was promising for the manufacture of implants in titanium [3]. Reference [4] studied the possibility using EBM for the manufacture of bone plates for complex fractures and concluded that customized implants can be manufactured in a time span of 3 days from CT images to finished implant. Another study [5] showed that customized implants manufactured by EBM also had a better, that is more even, stress distribution, compared to conventionally manufactured implant components. The method have also been proposed as suitable for patient specific total knee implants due to the flexibility regarding the geometric design [6]. The Ti6Al4V material manufactured using the EBM method conforms to current material standards and has been tested in several metallurgical studies [7, 8], animal studies [9, 10] and has also been used for humans [11, 12]. During revisions with severe cortical defects in proximal tibia, it is often advantageous to restore the bonestock for best possible long term survivorship. Conventional methods include rim meshes in titanium or steel, where the bonestock is restored with morcellized bone and bone impaction grafting. For uncontained defects of the proximal tibia, fixation of a rim mesh presents a challenge since the reinforcement mesh must withstand the forces introduced during vigorous bone

33


impaction grafting. Fixation of the rim mesh is usually done with bone screws. The fixation is depended on the quality of the remaining bonestock, which usually is poor. Recently, augments in trabecular metal or titanium have been introduced on the market to replace deficient bone. They present possible solutions in specific situations but do not restore bonestock. The aim of this case study was to design and manufacture a tibial reinforcement cage in Ti6Al4V ELI for a revision surgery after the septic loosening of a total knee replacement (TKR). 2. MATERIAL AND METHODS 2.1. Study Design The patient in the study (who participated with informed consent) suffered from the septic loosening of a total knee replacement (TKR), see Fig. 1-2. Due to compromised collateral ligaments, severe osteolysis and bone defects, bone impaction grafting and a rotating hinge prosthesis were chosen for reconstruction. Since the bony rim of the proximal tibia was deficient, a supportive structure, more rigid than the ones that are commercially available, was needed in order to achieve a stable fixation.


2.2. Virtual Design CT examinations of the patient were performed and the DICOM (Digital Imaging and Communications in Medicine) files were then imported into the Mimics 14 software (Materialise, Belgium). Using automatic and manual segmentation, a 3D model (STL format) of the tibia was exported from Mimics and this model was later used as a reference when modelling the titanium cage, see Fig. 3. Since STL files generated from DICOM files are rather complex and large in file size, it was decided to use the STL as a reference for further CAD modeling in the Rhinoceros software (Robert Mc Neel & Associates). A modelling technique was used in which the STL file was sliced every 10mm in order to retrieve the inner contours of the tibia. A surface was then lofted between these contours and this surface model was transformed into a solid object using an offset of 0.6mm. In order to facilitate vascularization and bone ingrowth between the bone graft and the host bone, ~500 holes with a diameter of 2mm and an approximate array distance of 2.5mm were included in the model, see Fig. 4-5.

Figure 1. Pre-operative picture of the infected knee with original prosthesis, front.

Figure 3. The STL file of the tibia. Different views generated from the DICOM files.

Figure 2. Pre-operative picture of the infected knee with original prosthesis, side.


34



Figure 6. Several different cages prepared for manufacture. Support structures in grey/red.

Figure 4. Different views of the 3D model of the tibial cage.

It took 45 minutes to pump vacuum in the machine, followed by 9h 35min of manufacturing time. The build was cooled in the build chamber, after which it was removed from the machine and transported to the PRS (Powder Recovery System) where excess powder was removed using the same powder as during manufacturing. The supports used for manufacturing were removed using pliers and the cages were then put in plastic bags and sent to the hospital for sterilization. Cleaning and Sterilization The implants were ultrasonically cleaned with a standard cleaning solution (Branson 8510) at 60°C for 5 minutes, followed by washing (Getinge 46-4, Sweden) at 90°C for 60 minutes. The implants were then packed and standard steamsterilized (Autoclave Getinge HS6606-AC-1, Sweden) at 134°C for 4 minutes.

Figure 5. Different views of the planning of the tibial cage (red) in the tibia (grey).

2.3. Additive Manufacturing The 3D model of the tibial cage was exported to STL and then imported into the Magics software (Materialise, Belgium), where the model was prepared for manufacturing. Shrinkage compensation due to manufacturing was performed and support structures between the start plate and the cages (multiple cages with different designs were manufactured in the same build) were designed. A start plate measuring 170x170x10mm was used and the build height was 73.6mm, see Fig. 6. The files were finally prepared and sliced into 70µm layers in Build Assembler (Arcam AB, Sweden) before being sent to the manufacturing unit. The cages were manufactured using an Arcam EBM A2 machine using gas-atomized Ti6Al4V ELI powder with a particle size ranging from 45 to 100µm.


2.4. Clinical Application The reconstruction consisted of a two stage procedure. In stage one, the knee was debrided and the infected prosthesis replaced with a cement spacer loaded with antibiotics according to the specific pathogens in combination with supplementary systemic antibiotics. In stage two, the spacer was extracted, the wound thoroughly debrided and irrigated with pulsatile lavage. Curettes and bone rasps were used to adjust the form of the tibial canal. The titanium cage was then introduced into the tibial canal and had a solid pressfit fixation. No additional screws were necessary. Cancellous bone chips from fresh frozen femoral heads were prepared with rongeurs, soaked in bisphosphonate solution and rinsed in warm saline solution. Bone impaction grafting was then performed with the Ullmark instruments Knee Bone Grafting (Waldemar Link GmbH & Co.KG) followed by implantation of a cemented Endo Model Rotational Knee Prosthesis, (Waldemar Link GmbH & Co.KG). 3. RESULTS Plastic models of the tibia were additively manufactured prior to surgery in order to support surgical planning and discussions before the procedure. The patient was allowed immediate full weight bearing after surgery and the healing was uneventful. The 15 and 52 week post-surgery follow-ups showed satisfactory results with no signs of radiolucency or loosening, see Fig. 7-8. The patient was also able to walk without walking aids.

35



4. DISCUSSIONS AND CONCLUSIONS AM can bring about new opportunities in the treatment of complex bone defects. In comparison with ordinary rim mesh procedures, customized reinforcement cages give the orthopaedic surgeon more options regarding rigidity, designed surface topography (for enhancing bony ingrowth) and also designed porosity for vascularization between the bone graft and host bone. Another possible benefit is that the operation time could be reduced since there is less time needed for fixation and adaptation of the cage to the patient. REFERENCES [1]

Figure 7. Different views of the knee after operation.

Figure 8. X-ray taken at the 52-week post-surgery follow-up.

Cormier, D., Harrysson, O., West, H. 2004, ‘Characterization of H13 steel produced via electron beam melting’, Rapid Prototyping Journal, vol. 10, pp. 35-41. [2] Rännar, L-E., Glad, A., Gustafson, C-G. 2007, ‘Efficient cooling with tool inserts manufactured by electron beam melting’, Rapid Prototyping Journal, vol. 13, pp. 128-135. [3] Harrysson, O., Cormier, D. R., Marcellin-Little, D. J., Jajal, K. R. 2003, ‘Direct fabrication of metal orthopedic implants using electron beam melting technology’, Proceedings SFF Symposium, Austin, Texas USA, pp. 439-446. [4] Cronskär, M., Rännar, L-E., Bäckström, M. 2012, ‘Implementation of digital design and solid free-form fabrication for customization of implants in trauma orthopaedics’, Journal of Medical and Biological Engineering, vol. 32, n. 2, pp. 91-96. [5] Harrysson, Hosni, Y. A., Nayfeh, J. F. 2007, ‘Custom-designed orthopedic implants evaluated using finite element analysis of patientspecific computed tomography data: femoral-component case study’, BMC Musculoskeletal Disorders, vol. 8, n. 91. [6] Murr, L. E. et al. 2011, ‘Microstructure and mechanical properties of open-cellular biomaterials prototypes for total knee replacement implants fabricated by electron beam melting’, Journal of the Mechanical Behavior of Biomedical Materials, vol. 4, n. 7, pp. 13961411. [7] Christensen, A., Kircher, R., Lippincott, A. 2007, ‘Qualification of electron beam melted (EBM) Ti6Al4V-ELI for orthopaedic applications’, Proceedings of Medical Device Materials IV, Palm Desert, California, USA, pp. 48-53. [8] Murr, L. E. et al. 2009, ‘Microstructure and mechanical behavior of Ti– 6Al–4V produced by rapid-layer manufacturing, for biomedical applications’, Journal of the Mechanical Behavior of Biomedical Materials, vol. 2, n. 1, pp. 20-32. [9] Palmquist, A., Snis, A., Emanuelsson, L., Browne, M., Thomsen, P. 2013, ‘Long-term biocompatibility and osseointegration of electron beam melted, free-form–fabricated solid and porous titanium alloy: Experimental studies in sheep’, Journal of Biomaterials Applications, vol. 27, n. 8, pp. 1003-1016. [10] Petrović, V. et al. 2011. ‘A study of mechanical and biological behavior of porous Ti6Al4V fabricated on EBM’, Proceedings of VRAP 2011, 28 Sep – 01 Oct, Leiría, Portugal. pp. 115-120. [11] Dérand, P., Rännar, L-E., Hirsch, J.M. 2012, ‘Imaging, virtual planning, design and production of patient specific implants and clinical validation in cranio-maxillo-facial surgery’, Craniomaxillofacial Trauma and Reconstruction, vol. 5, n. 3, pp. 137-144. [12] Cronskär, M. 2014, ‘On customization of orthopedic implants – from design and additive manufacturing to implementation’, Mid Sweden University Doctoral Thesis 191, ISBN 978-91-87557-63-7. .


36



Multiscale surface structuring of the biomedical implants manufactured in Electron Beam Melting technology: demands, advances and challenges Andrey Koptyug, Department of Quality Technology and Management, Mechanical Engineering and Mathematics, Mid Sweden University, Akademigatan 1, SE 831 25, Östersund, Sweden, [email protected] Lars-Erik Rännar, Department of Quality Technology and Management, Mechanical Engineering and Mathematics, Mid Sweden University, Akademigatan 1, SE 831 25, Östersund, Sweden, [email protected] Mikael Bäckström, Department of Quality Technology and Management, Mechanical Engineering and Mathematics, Mid Sweden University, Akademigatan 1, SE 831 25, Östersund, Sweden, [email protected]

Abstract—paper discusses the challenges of additive manufacturing when multidimensional shape and surface feature control of the component on wide scale is essential, as it is for the manufacturing of the metallic biomedical implants. Paper also discusses most critical demands imposed by the biomedical implant manufacturing including implant surface roughness issues along with possible solution pathways, and gives some examples of the problems encountered and achievements reached in solving these challenges for the Ti6Al4V EBM®- manufactured components. Keywords- additive manufacturing; electron beam melting; EBM; metallic implants; surface roughness control

1. INTRODUCTION Additive Manufacturing (AM) of biomedical implants rapidly becomes an integral part of modern healthcare. This is due not only to the flexibility in design and manufacturing of the possible implant outlines and functionality, but also to the increasing availability of new materials and continuing work on the cost efficiency of manufacturing of complex, patientspecific implant structures. The capacity of AM technologies in producing components with complex shapes is often brought to the extreme in modern biomedical applications. To achieve desired functionality modern metallic implants often have sections of 3D-cellular structures combined with solid parts. This approach is driven not only by the desire to mimic the natural bone structure, but also by the fact that with AM technologies lightweight cellular structures are often cheaper and faster to manufacture. Significant advances in the integration of functional modeling into the design procedures and additive manufacturing yield very advanced implants with the geometry features successfully controlled at the scales from tens of centimeters to the fraction of a millimeter, manufactured in the single uninterrupted AM process such as for example in Electron Beam Melting (EBM®). But simultaneous control of the surface topography at much smaller scales in “as manufactured” components is still challenging. Also the question on what level of the implant surface

roughness is optimal is still widely studied and debated. Many specialists are now agreeing that different roughness scales may be important at different stages of the implant integration and “ideal” implant surface should most probably incorporate features with the dimensions down to nanometer scale. Thus AM technology of metallic implant manufacturing should be ready for such multidimensional component design and manufacturing through optimization of process parameters and by utilizing post-processing. The progress in manufacturing methods is complemented by the developments in the supporting computer-aided tools, yielding previously non-existent possibilities from the patient’s CT scan or MRI image through the functional design to the optimized patient-specific implant [1-5]. Practical medicine today becomes one of the key application areas for the Additive Manufacturing of metallic implants critical for the future of this rapidly developing segment of industry. So satisfying specific demands of the users in this application area is of high importance for additive manufacturing. Simultaneously, the process directed at satisfying specific requirements imposed by practical medicine to the metallic implants forces new developments in the additive manufacturing and related technologies. 2. CRITICAL DEMANDS TO THE METALLIC IMPLANTS The main requirements for the metallic implant are following from the fact that it should be working in the human body for a reasonably long period of time without degrading, when supporting broken bones or substituting some parts of them. Also such implants should be available for the patients and someone should be able to place them. Though general demands could be formulated with visible simplicity, the reality is infinitely more complex [6, 7]. Nevertheless one can roughly split these requirements into few major areas, namely the demands to the 3D geometry of the implants (shape structure etc.), their mechanical properties strongly related to implant functionality, and to the implant biocompatibility through its whole lifetime in the human body, and their

This work was partially funded by the Swedish Agency for Economical and Regional Growth (Tillväxtverket) and the European Regional Development Fund.


37


manufacturing cost. All these areas are directly relevant to the additively manufactured metallic implants [8, 9]. The functional and individual size and shape variability of the bones, together with numerous types of trauma when the reconstruction or fixation using implants could be needed require absolute flexibility in the available shapes of metallic implants. If one looks at it from the manufacturing point of view, the majority of the adult human bones are not exceptionally small or large, and consequently the majority of the needed implants would have the outline dimensions from some centimeters to some tens of centimeters, well within the routine possibility for the additive manufacturing of metallic components. Our experience with practical surgeons also shows that the required precision of the manufacturing in terms of outline dimensions is not very high and is commonly within millimeters or fractions of millimeters, which is again well within the capacity of AM methods [9-11]. Comparing the manufacturing costs of solid metallic implants one can conclude, that AM methods cannot (and most probably should not) compete with the standardized mass-produced items, but in some cases are already more cost-effective than other methods of manufacturing complex and customized ones [12]. Unfortunately such direct comparisons can also be misleading, as they commonly do not take into account other associated costs. For example, our experience shows that the operation time with the properly designed individualized fixation plate can be significantly less than in the case when a surgeon uses standard ones, which are to be cut and bent to fit during the operation [3]. So if we are only including into consideration the cost of corresponding surgical procedure (implant and operation together) it can be significantly smaller when using additively manufactured personalized implants than it is with even advanced ones produced with other methods. Taking into account the high costs associated with surgical procedure can already outweigh the difference in the costs of individualized additively manufactured implants and the ones produced by other technologies. Functional optimization of the implants that can be achieved using modern computer modeling and finite element analysis methods together with additive manufacturing can be significantly enhanced by varying not only the outline but also the structure of the implants. One of the most promising lines here is incorporation of the implant sections having different 3D-lattice structures together with solid parts [13, 14]. This approach is inspired in part by the desire of mimicking the trabecular (porous) bone structure. Varying the lattice cell type and beam thickness one can achieve different (often complex, anisotropic and changing through the structure) mechanical properties of the implant helping to optimize its functionality [13]. Modern additive manufacturing methods are perfectly suited for manufacturing metallic implants incorporating 3Dlattice structures with the lattice elements of millimeter and sub-millimeter sizes. In many cases such components are hardly possible to manufacture using traditional methods. Also 3D-lattice structures are cheaper to manufacture, than solid parts with the same outlines (process is faster and less material is consumed) when using AM methods. Biocompatibility of the implant and its longevity in human body is determined both by its material and the structure of its



exposed periphery [15, 16]. Though new promising materials for biomedical applications are constantly developed titanium and its alloys are still among the most widely used for metallic implants [6, 7]. It is determined by good mechanical properties, high stability at atmospheric conditions even at elevated temperatures and in water solutions, and as the result reasonably good biocompatibility (Ti and its alloys are commonly referred to as bioinert). Also titanium alloys, Ti6Al4V in particular, are processed quite well using AM methods. But when the implant material is chosen (which could be due to its formal approval for biomedical applications, possibility of existing AM equipment, availability and cost of proper source material etc.) it is time to think of the implant surface structure design and the capacity of the available AM systems to achieve the desired surface topography. And this is where one of the major challenges for the AM manufacturing lays at the moment [10, 11]. 3. IMPLANT SURFACE TOPOGRAPHY CHALLENGE One of the big questions related to the metallic implant surface topography is coming from the practical side: what kind of the surface should the implant have? The issue of the optimum roughness of implant surfaces is studied all over the world for quite some years. Possible answers are presented in numerous publications and intensely debated in both biomedical and engineering communities. But the results at the first glance seem to be completely confusing and the spread of the reported optimal feature dimensions is striking. Multiple papers report optimal feature sizes of 10 to 100 nanometers [17-20], of 0.1 to 1 micrometers [21-23], 0.5 to1.5 micrometers [24], 1 to 25 micrometers [25], hundreds micrometers [26, 27] and up to millimeters [28]. Of course one can expect certain spread in the reported “optimal feature size” values due to the complexity of surface topography and differences in the assessment methods of the surface roughness [29], but this can hardly explain such significant spread in reported values and especially the differences in trends for different surface exposure times (see, for example, [25]). Deeper analysis of the reported results shows that in-vitro (cell culture) studies tend to yield smaller “optimal roughness” values, in the micron and sub-micron range [17-19, 21-23, 25], with the results often differing for different cell incubation durations [17, 18] and different ways of assessment (monitoring initial cell attachment, cell spreading etc.). With the in-vivo studies implants are in contact with the live tissues and bones for much longer periods of time (multiple weeks, month and up to yearsin some human studies) as compared to the durations of cell culture experiments (implant exposure times with in-vitro experiments ranges from hours to weeks and up to a month). And the assessment methods with in-vivo studies are also very different (pullout tests, histology etc.). Corresponding reported “optimal surface feature sizes” for in-vivo experiments tend to be significantly larger, from microns, to hundreds microns [26, 27] and up to millimeters [28]). Taking into account that bone formation process is lengthy and involves many different stages it is hardly feasible that one can possible specify any single selected optimum average implant surface feature size. It is known, that nanometer scale surface topography features can impact very early stages of the process, from protein adsorption, through cell adhesion to cell

38


proliferation and differentiation [21]. So it is feasible, that nanometer size surface features should be most important at the earliest stages of the process, dominated by surface wettability and chemistry [30-32], and larger (possibly micron size) ones are becoming important when the primary cells (osteoblasts) are involved. It is also feasible, that at much later stages when the inner cell layers are masking the smallest implant surface features, only the larger ones can influence the processes. Largest size features (hundred micrometer to sub-millimeter ones) could clearly play significant role at much later stages with effective bone ingrowth into the pores and 3D-lattice structures helping to secure the ‘anchoring’ of the implant, and providing necessary space for effective transport of the needed biochemical components to the area of developing new bone. And it is why some of the researchers are stressing that a certain hierarchy of the implant surface feature sizes is important [12, 23, 30, 33-35]. Moreover, some studies also indicate that regular groove-like or dimension-gradient type structures of the implant surface may be helpful [28, 32, 33]. Thus a wide variety of the surface features is clearly important though no solid data is available at the moment weather it should be a surface feature size distribution clearly peaking at few dimensions (like, for example, with the BioHelix implants [23, 36] or regular channel-like features [28, 32, 33]), or more or less randomized one


functionality of the implant and clinical requirements of each individual case. 4. MULTISCALE IMPLANT SURFACE STRUCTURING

Additional dimension to the discussion on the optimal metallic implant surface topography is added by the opposite functional requirements from the permanent and temporary implants. Permanent implants should be functioning in the human body for considerably long time thus requiring best possible attachment to the surface or ingrowth of the newly formed bone. Temporary implants (like, for example, fixation plates in some cases) should be removed after certain period of time, and in many cases it means that the bone attachment to certain surfaces should be weak thus demanding much smoother surface. Consequently the manufacturing technology should ideally be able to adapt the surface topography to the

From the manufacturing point of view requirements discussed above mean that there is a need to achieve the control of periphery (“surface”) features of the metallic implants in the very wide scale, from nanometers to millimeters. Working with a powder-bed type AM methods, for example like EBM® [37], controlling the features down to millimeters is routine, and down to fractions of millimeter is quite possible with careful choice of process parameter settings. But controlling the features below the powder grain average dimensions (which is commonly about tens micrometers for Ti6Al4V used in EBM®) is quite challenging even for the solid (no 3D-lattice elements) structures with simple outlines [10, 11, 38, 39]. It is also quite challenging to keep the uniformity of the surface features having different orientation related to the melt pool surface [38, 39]. Although with certain care in process setting the orientation differences in the surface structure may still be acceptable, if reasonably wide feature distributions are requested. It is also quite noticeable that on the solid metal surface there is quite large number of powder grains partly fused or embedded into it (Fig. 1), which should yield a definite peak around the average size of the powder grain in the feature size distribution. Even preliminary analysis of the “as manufactured” side surfaces of the solid samples shows the presence of the features covering the scale from hundreds micrometers and down to sub-micrometer sizes (Fig. 2), and there are also some “elements of true porosity” as the partly fused grains have deep narrow crevices often surrounding them. Unfortunately the presence of loosely connected powder grains on the surfaces causes serious concern that they can somehow become loose during the lifetime of the implant in the human body, but there is a lack of reliable evidence on how dangerous it could be.

Figure 1. Powder grains partly fused and embedded into the surface of the solid Ti6Al4V EBM®-manufactured sample.

Figure 2. One of the loosely connected powder grains on the same surface shown in Fig.1


39


Figurre 3. Powder grains partly fused an nd embedded into the t surface of the ® beams of the 3D lattice Ti6Al4V EBM E -manufacturred sample.

Figurre 4. Powder graains partly fused an nd embedded into the surface of the ® beams of the 3D lattice Ti6Al4V EBM E -manufacturred sample.

Figuree 5. Changes in the t planar EBM®-m manufactured Ti6A Al4V sample withh increaasing etching timee. From left to righ ht: “as is” sample, and a samples from the same manufacturinng batch etched forr 5, 10, 15 and 25 minutes. m See [10, 11, 43] 4 for the etching g process details.



The surfaces s of thee beams of 3D D-lattice structtures made from Ti6 6Al4V using EBM® AM technology also have embedded d powder graiins (Figs. 3, 4). But the conntrol of the surface to opography esppecially in the cases with relaatively thin (less than n 0.5-0.25 mm m) beams becom mes more com mplex. With fine 3D-lattice structtures relativelly small mellt area is completely surroundedd by the sem mi-sintered poowder and precision control over the process becomes b tricky ky. Modern ARCAM EBM® machiines allow for quite flexible control of the meltiing process w with some stanndardized settiings better optimizedd for the latticee structures. But these settinggs are only “generallyy optimal” andd there are pron nounced differeences in the beam surrface topographhy for the diffferent parts off the lattice structure.. For example there t is a clearr difference deppending on how far are a the particullar beams from m the solid mettal sections or from the t periphery oof the structuree, what is the bbeam angle as related d to the workinng (melt pool) surface s etc. It iis clear that more reseearch into the process param meter optimizattion for the lattice strructures is needded. One of thee possible direections here can be in varying ennergy deposittion parameterrs for the manufactturing of the latttice structure beams b dependiing on their position in i the structure. Thou ugh some conttrol over the suurface topograaphy of the additivelyy manufacturedd components through varyiing process parameterrs is possible [[38, 39] it is raather limited esspecially in the case of o complex struuctures like 3D D-lattices. Addittional postprocessinng should be used to furthher modify thhe surface topographhy. Large varieety of differentt methods can be b used for the metalllic implant suurface topograpphy modificatioon [10, 11, 15, 20- 23, 25, 40-42] but chemiical and electtrochemical e to impllement [10, methods are by far the cheapest and easiest 11, 40]. For F example, ussing anisotropiic electrochemiical etching in viscous solutions onee can achieve certain surface ttopography control for f the solid Ti6Al4V T samp ples [10, 11, 43]. With increasing g etching timee larger and larrger surface feeatures and the majorrity of powderr grains looselyy connected too the metal surface are a removed [110, 11]. Similaar effect can be b achieved with chem mical etching [40]. Figs. 5-77 illustrate the changes in the planaar EBM®-mannufactured Ti66Al4V sample roughness parameterrs developing with w increasing g electrochemiical etching time. Theese changes aree quite visible in i the reflectedd light (Fig. 5). Also it i is easy to undderstand, that when w the smalleest features are etcheed away the remaining “w waviness” of the larger features still remains and thus Ra value changees are not Fig.6). So som me other param meters like strongly pronounced (F average peak-valley p disttance representt the situation some s better (Fig.7). Though T stylus profilometry iss fast and robuust it is not capable of o assessing thee overall surfacce and also cannnot resolve topographhy features sm maller than thee tip radius. Thus T other methods, like scanningg microscopy, optical interfeerometry or atomic foorce microscopyy may be prefeerred. Though electrochemica e al etching proccess can be freee from the aggressivve hydrofluoricc acid or its salts commonnly used in chemical etching of Ti aalloys it is pooorly suitable forr the lattice structuress. Electrochem mical etching is dependent on the electric field rapiidly decreasingg from the ouuter layers of the lattice structuress inwards. As the result periiphery of the lattice l (and especiallyy sharper beaam tips) is ovver-etched whhile deeper layers of the beams are hhardly etched at a all. Thus cheemical

40



Value,  140

Ra, X Ra, Y P‐V, X P‐V, Y

120 100 80 60 40 20 0 0 Fig gure 6. Changes in i the planar EBM M®-manufactured Ti6Al4V T sample rou ughness with increasing etching timee. Single sample prrofile traces are measured by the proofilometer Dektak 6M (stylus tip raddius 12 ). “As manuufactured” sample (bottom); sample from the same maanufacturing batch ettched for 10 minuttes (middle) and 255 minutes (top tracce). Traces are artificially offset. o

etchin ng in solutions of low viscosiity may be prefferable with thee 3D-laattice structurees. But even in i this case deense and deepp latticees would be quuite hard to treeat uniformly, as lattice cellss laying g deeper insidee the structure will be etched less efficientlyy than the ones closse to the stru ucture peripheery due to thee exhau ustion of etchiing solution [440]. Intense so olution mixingg and solution s injectinng into the struucture could heelp in this casee, but iff it would solvee the problem is to be further investigated. 5. CONCLUSION O A present leveel of technoloogy “as manuffactured” AM-At madee metallic impllants cannot saatisfy the compplete variety off the deemands to theiir functionality y. It seems that at the momennt the major m challeng ge is the compponent feature control at thee scaless smaller than average grain size of the ind dustrial powderr bed machines m relateed to the smalleest (down to naanometer scale)) featurres of the surfaace topography y. And thus certtain level of thee post-p processing is inndeed essentiall. W relatively simple solid temporary titaanium implantss With decreeasing the averrage roughnesss of the surfacees by chemicaal and electrochemica e al polishing may be already adequate. Buut for th he complex impplants, especiallly the ones coontaining latticee sectio ons, different post processsing strategiees should bee involved. Chemicaal and electrocchemical treatm ment (etching)) discu ussed above taargets the chan nges of surfacce topographyy, ratherr than its compposition. Manyy other methodds involving Ti alloyss surface modiification and vaarious coatingss are developedd (see, for example paapers [44-46]). For example electrochemica e al treatm ment producin ng titanium oxxide nanotubess (for examplee similaar to the one suggested in [446]) can be quite adequate forr the soolid permanentt titanium implaant post-processsing. Thus forr the nearest n future surface modiffication of the beams of thee densee and deep 3D-lattice titanium m structures onn one hand not underrmining the inttegrity of the structure itself (not ( destroyingg thin beams etc.) and a on the otther hand prooviding neededd meter surface features f presentts a real challen nge. nanom


5

10

15

20

25

30

35

Etching time, min

Figure 7. 7 Changes in thee planar EBM®-maanufactured Ti6Al44V sample roughnesss parameters with increasing etchingg time. Averaged Ra (Ra) and peak-valleey (P-V) distances for two orthogonaal directions on thee sample are given. Measurements M weree done with the styylus profilometer Dektak D 6M. Height pro ofiles were taken foor two orthogonal directions (X, Y) and resulting values were avveraged (see [42] for f more details).

REFEREN NCES [1]

Cronsskär, M., Rännar, L L. and Bäckström m, M., 2012, ‘Impleementation of digital design and sollid free-form fabbrication for custtomization of implannts in trauma orthopaedics’, J Med. Biol. Eng., vol. 322 no.2, pp.9196. [2] Cronsskär, M. and Bäcksström, M., 2014, ‘Modeling of fractuured clavicles and reeconstruction plattes using CAD, fiinite element anallysis and real muscu uloskeletal forces input’, WIT Transactions on Biom medicine and Healthh, no.17 , pp.235-2243. [3] Cronsskär, M., Rasmusssen, J. and Tinnsteen, M., 2013, ‘Coombined finite elemeent and multibodyy musculoskeletaal investigation of a fractured claviccle with reconstruuction plate’, Comp mput Methods Biom mech Biomed Enginn, 2014 [Epub aheaad of print] [4] Ciuranna, J., and Rezendde, R., 2013, ‘New w opportunities annd challenges for ad dditive manufacturring to produce bioomedical devices’, Management and Control C of Productiion and Logistics, vol. 6, no.1, pp. 283-288. [5] Fariass, T. P., Dias, F. L., Sousa, B. A.,, Galvão, M. S., Bispo,D. and Pastl, A. C., 2013, ‘Prototyping: major addvance in surgical planning and mizing prosthesess in patients with bone tumors of the head and custom neck’,, International Jouurnal of Clinical Medicine, M 2013, vol. v 4, no.7A, pp.1-77. [6] Frosch h, K.-H., and Stüürmer, K. M., 20006, ‘Metallic biiomaterials in skelettal repair’, Europeaan Journal of Trauuma, vol. 32, pp. 1149-159. [7] Grimm m, M. J., 2002, Chapter 37 ‘Selectiion of materials foor biomedical appliccations’, in Handbbook of Materials Selection, S Ed. Myyer Kutz, New York: John Wiley and S Sons, pp. 1165-11994. [8] Kopty yug, A., Rännar, L.-E., Bäckström m, M., Fager Fraanzén and S., Dérannd, P., 2013, ‘A Additive manufacturing technologyy applications targeting practical surggery’, Internationaal Journal of Lifee Science and Mediccal Research(LSM MR), vol. 3, no.1, pp. p 15-24. [9] Vandeenbroucke, B. andd Kruth, J.-P., 2007, ‘Selective laser melting of biocom mpatible metals fo for rapid manufactturing of medical parts’, Rapid Prototyping Journal, vool. 13 no. 4, pp.196 6-203. yug, A., Rännar, L.-E., Bäckström m and Klingvall, R. P., 2012, [10] Kopty ‘Electtron beam meltingg: moving from maacro- to micro- annd nanoscale’, Materrials Science Forum, vol. 706-709, pp p 532-553. [11] Kopty yug, A., Rännar, L.-E., Bäckströöm and Cronskärr, M., 2014, ‘Addiitive manufacturinng for medical and biomedical applications: advannces and challengees’, Materials Scieence Forum, vol. 783-786, pp. 1286-1291.

41


[12] Cronskär, M. , Rännar, L.-E. and Bäckström, M., 2013, ‘Production of customized hip stem prostheses: a comparison between machining and additive manufacturing’, Rapid Prototyping Journal, vol. 19, no.5, pp. 365-372. [13] Murr, L. E. , Gaytan, S. M. , Medina, F., Lopez, H., Martinez, E., Machado, B. I., Hernandez, D. H., Martinez, L., Lopez, M. I., Wicker, R. B. and Bracke, J. , 2010, ‘Next-generation biomedical implants using additive manufacturing of complex, cellular and functional mesh arrays’, Philos. Trans. R. Soc. Lond., A, vol.368, pp. 1999-2032. [14] Ryan, G., Pandit, A. and Apatsidis, D. P., 2006, ‘Fabrication methods of porous metals for use in orthopaedic applications’, Biomaterials, vol. 27, no.13, pp. 2651-2670. [15] Bauer, S., Schmuki, P., von der Mark, K. and Park, J., 2013, ‘Engineering biocompatible implant surfaces: Part I: Materials and surfaces’, Progress in Materials Science, vol. 58, no. 3, pp. 261-326. [16] Bauer, S., and Park, J., 2013, ‘Engineering biocompatible implant surfaces: Part II: Cellular recognition of biomaterial surfaces: Lessons from cell–matrix interactions’, Progress in Materials Science, vol. 58, no. 3, pp. 327-381. [17] Meredith, D. O., Eschbach, L., Wood, M. A., Riehle, M. O., Curtis, A. S. and Richards, R. G., 2005, ‘Human fibroblast reactions to standard and electropolished titanium and Ti-6Al-7Nb, and electropolished stainless steel’, J Biomed Mater Res A., vol. 75, no.3, pp.541-555. [18] Ballo, A., Agheli, H., Lausmaa, J., Thomsen, P. abd Petronis, S., 2011, ‘Nanostructured model implants for in vivo studies: influence of welldefined nanotopography on de novo bone formation on titanium implants’, International Journal of Nanomedicine, vol. 6, pp.3415-3428. [19] Osathanon, T., Bespinyowong, K., Arksornnukit, M., Takahashi, H. and Pavasant, P., 2011, ‘Human osteoblast-like cell spreading and proliferation on Ti-6Al-7Nb surfaces of varying roughness’, J Oral Sci., vol. 53, no.1, pp. 23-30. [20] Bressan, E., Sbricoli, L., Guazzo, R., Tocco, I., Roman, M., Vindigni et al, 2013, ‘ Nanostructured Surfaces of Dental Implants’, International Journal of Molecular Sciences, vol. 14, no.1, pp. 1918-1931. [21] Mendonça, G., Mendonça, D. B., Aragão, F. J., and Cooper, L. F., 2008, ‘Advancing dental implant surface technology-from micron- to nanotopography’, Biomaterials, vol. 29, no.28, pp. 3822-3835. [22] Wennerberg, A. and Albrektsson, T., 2010, ‘On implant surfaces: a review of current knowledge and opinions’, Int J Oral Maxillofac Implants, vol.25, no.1, pp.63-74. [23] Brånemark, R., Emanuelsson, L., Palmquist, A., and Thomsen, P., 2010, ‘Bone response to laser-induced micro- and nano-size titanium surface features’, Nanomedicine, vol. 7, no.2, pp. 220-227. [24] Shalabi, M. M., Wolke, J. G. and Jansen, J. A., 2006, ‘The effects of implant surface roughness and surgical technique on implant fixation in an in vitro model’, Clin Oral Implants Res., vol. 17, no.2, pp. 172-178. [25] Ponader, S., Vairaktaris, E., Heinl, P., Wilmowsky, C. V., Rottmair, A., Körner, C. et al, 2008, ‘Effects of topographical surface modifications of electron beam melted Ti-6Al-4V titanium on human fetal osteoblasts’, J Biomed Mater Res A., vol. 84, no.4, pp.1111-1119. [26] Cameron, H. U, 2981, ‘Essential design considerations for microporous implants: preliminary communication’, J R Soc Med., vol. 74, no. 12, pp. 887–891. [27] Baril, E., Lefebvre, L. P. and Hacking, S. A., 2011, ‘Direct visualization and quantification of bone growth into porous titanium implants using micro computed tomography’, J Mater Sci Mater Med., vol. 22, no.5, pp.1321-1332 [28] Fukuda, A., Takemoto, M., Tanaka, K., Fujibayashi, S., Pattanayak D. K., Matsushita, T., Sasaki, K., Nishida, N., Kokubo, T. and Nakamura, T., 2010, ‘Bone Ingrowth into Pores of Lotus Stem-Type Bioactive Titanium Implants Fabricated Using Rapid Prototyping Technique’, Bioceramics Development and Applications, vol. 1, pp. 1-3. [29] Wennerberg, A., Albrektsson, T., 2000, ‘Suggested guidelines for the topographic evaluation of implant surfaces’, Int J Oral Maxillofac Implants, vol. 15, no.3, pp. 331-344. [30] Ponche, A., Bigerelle, M. and Anselme, K., 2010, ‘Relative influence of surface topography and surface chemistry on cell response to bone implant materials. Part 1: Physico-chemical effects’, Proc. Inst. Mech. Eng. H J. Eng. Med., vol. 224, no.12, pp. 1471-1486.



[31] Anselme, K., Ponche, A. and Bigerelle, M., 2010, ‘Relative influence of surface topography and surface chemistry on cell response to bone implant materials. Part 2: biological aspects’ Proc. Inst. Mech. Eng. H J. Eng. Med., vol. 224, no.12, pp. 1487-1507. [32] Klymov, A., Prodanov, L., Lamers, E. Jansen, J. A., and Walboomers, X. F., 2013, ‘Understanding the role of nano-topography on the surface of a bone-implant’, Biomaterial Science, vol. 1, no.2, pp. 135-151. Vrana, N. E., Dupret, A., Coraux, C., Debry, C. and Lavalle, P., 2011, ‘Hybrid titanium/biodegradable polymer implants with an hierarchical pore structure as a means to control selective cell movement’, PLOS One, published: May 26, 2011, DOI: 10.1371/journal.pone.0020480. online: http://www.plosone.org/article/metrics/info%3Adoi%2F10.1371%2Fjou rnal.pone.0020480 [33] Cheng, A., Hyzy, S. L., Cohen, D. J., Boyan, B. D. and Schwartz, Z., 2014, ‘Osteoblast response to 3D porous titanium manufactured by laser sintering with multi-scale roughness mimicking trabecular bone’, in: Abstracts, Orthopaedic Research Society annual meeting, New Orleans, Louisiana, March 15-18, 2014, Abstract 279, online: http://prgmobileapps.com/AppUpdates/ors/Abstracts/abs279.html [34] Chen, J., Rungsiyakull, C., Li, W., Chen, Y., Swain, M. and Li, Q., 2012, ‘Multiscale design of surface morphological gradient for osseointegration’, J Mech Behav Biomed Mater., vol. 20, pp. 387-397. [35] Thomsson, M. and Esposito, M., 2008, ‘A retrospective case series evaluating Branemark BioHelix implants placed in a specialist private practice following 'conventional' procedures. One-year results after placement.’, Eur J Oral Implantol., vol. 1, no.3, pp. 229-234. [36] ARCAM AB, http://www.arcam.com/technology/additive-manufacturing/ [37] Ek, R., Hong, J. and Dejanovic, S., 2011, ‘Blood coagulation on electron beam melted implant surfaces, implications for bone growth’, in: Proc. 24th European Conference on Biomaterials, September 4th–9th, Dublin 2011, online: http://www.diva-portal.org/smash/get/diva2:578677/FULLTEXT01.pdf [38] Safdar, A., He, H. Z., Wei, L.-Y., Snis, A., and Chavez de Paz, L. E., 2012, ‘Effect of process parameters settings and thickness on surface roughness of EBM produced Ti-6Al-4V’, Rapid Prototyping Journal, vol. 18, no.5, pp.401-408. [39] Structures, J., and Wevers, M., 2013, ‘Surface Roughness and Morphology Customization of Additive Manufactured Open Porous Ti6Al4V’, Materials, vol. 6, no.10, pp. 4737-4757. [40] Alla, R. K., Ginjupalli, K., Upadhya, N., Shammas, N., Ravi, R. K and Sekhar, R, 2011, ‘Surface roughness of implants: a review’, Trends Biomater. Artif. Organs, vol. 25, no.3, pp. 112-118. [41] de Monserrat, I., Bernal, O., Risa, I., Hiroki, K., Ken-Ichiro, T., Naoko, I., et al, 2009, ‘Dental implant surface roughness and topography: a review of the literature’, J. Gifu. Dent. Soc, vol. 35, no.3, pp. 89-95. [42] Koptyug, A., Bergemann, C., Lange, R., Jaggi, V. E., Rännar, L.-E., and Nebe, J. B., 2014, ‘Osteoblast ingrowth into titanium scaffolds made by electron beam melting’, Materials Science Forum, vol. 783-786, pp. 1292-1298. [43] Garg, H., Bedi, G. and Garg, A., 2012, ‘Implant surface modifications: a review’, J Clin Diagn Res, vol. 6, no.2, pp. 319-324. [44] Goodman, S. B., Yao, Z., Keeney, M. and Yang, F., 2013, ‘The future of biologic coatings for orthopaedic implants’, Biomaterials, vol. 34, no.13, pp. 3174-3183. [45] Kim, H. S., Yang, Y., Koh, J. T., Lee, K. K., Lee, D. J., Lee, K. M. and Park, S. W., 2009, ‘Fabrication and characterization of functionally graded nano-micro porous titanium surface by anodizing’, J Biomed Mater Res B Appl Biomater., vol. 88, no.2, pp. 427-435.

42



Patient-Specific Cardiovascular Models for Educational and Training Purposes Lopes, P., Materialise, Leuven, Belgium, [email protected] Fremmer, M., Materialise GmbH, Gilching, Germany, [email protected] Bichlmeier, C., Materialise GmbH, Gilching, Germany, [email protected] Verschueren, P., Materialise, Leuven, Belgium, [email protected]

Abstract—Structural heart defects are relatively common abnormalities, and range from small connections between chambers to complex arrangements of the heart structures. Even though medical imaging techniques have been evolving rapidly in the last decades, challenges remain in the interpretation of some pathologies, impacting the decisions taken when planning a surgical treatment.

presence of complex congenital heart defects; for delivery and deployment of new medical devices, including artificial heart valves, catheters and other delivery systems, stent grafts and occlusion devices, among others; and for pre-surgical planning and training, particularly in cases of challenging anatomy or when recently launched devices are used.

The utilization of patient-specific physical models of the patient’s anatomy can not only contribute to improved communication between doctors and patients, but also to a better pre-surgical planning. Additive manufacturing offers the capability of producing highly-detailed, complex replicas of the patients’ anatomy in a relatively short period, providing valuable input for enhanced medical treatment.

2. MATERIALS AND METHODS

Keywords- additive manufacturing; cardiovascular; structural; anatomical replicas; communication; surgical training

1. INTRODUCTION Structural heart disease includes defects originated during embryological development, as well as disorders acquired during adulthood. Congenital heart defects affect 8 out of 1000 newborns [1]. While a large percentage of these defects have no serious consequences for the child’s life, some patients present life-threatening conditions that need to be carefully understood and treated. Also the number of patients with acquired disorders is expected to rise dramatically, as life expectancy increases worldwide. Multiple techniques are applied in the manufacturing of heart models for educational purposes. Nevertheless, these models are often generic representations of the anatomy and do not take into account the array of configurations that structural heart disease cases present. Additive manufacturing offers a flexible and powerful alternative to create accurate patient-specific replicas of the anatomy. Additive manufacturing technologies have been utilized in the medical field for more than two decades, especially in the orthopedic and craniomaxillofacial specialties [2-7]. The utilization of such techniques for the manufacturing of cardiovascular models is undoubtedly more recent, but has evolved substantially. Such models are commonly used for educational and communication purposes, especially in the


This paper describes the utilization of four additive manufacturing techniques for the creation of patient-specific cardiovascular models used in various applications. 2.1. Image Segmentation Medical images of the heart acquired with different modalities (CT, MR and echo) were obtained from multiple imaging centers. As the image quality varied considerably, a quality check was performed on each dataset to verify if all structures of interest could be adequately identified and segmented. The accepted datasets were imported into Mimics (Materialise NV, Belgium) and the cardiac structures of interest were manually or semi-automatically segmented. The segmentation process was initiated by selecting a range of gray values that adequately represented the blood pool, a method commonly known as “thresholding”. This operation results in the creation of a colored layer or mask on top of the original images. This layer can be made visible or invisible and modified without affecting the original images. The resulting segmentation mask was then cropped when it included adjacent structures not relevant for the anatomical model. The cropped mask was subsequently “region grown”, in order to include only regions of connected voxels. This workflow was repeated for different threshold values, when necessary, as the intensity of the voxels in some regions can differ from that of other regions, e.g., left chambers and right chambers, or lumen and myocardium. Mask editing tools performed both on the 2D images and 3D representations were then applied to separate the different heart chambers and vessels of interest. The segmentation process resulted ultimately in masks corresponding to some or all cardiac structures: right atrium (RA), including the superior and inferior vena cava, right ventricle (RV), left atrium (LA), left ventricle (LV), aorta (Ao) and pulmonary artery (PA). In some cases, and when relevant for the specific case, the

43


coronary arteries and the myocardium, as well as abnormal structures and implanted devices (e.g. artificial valves, shunts, etc.), were also included in the segmentation. The masks were ultimately converted to 3D representations by applying a marching cubes algorithm. Fig. 1 outlines the segmentation workflow. 2.2. Model Preparation The 3D objects resulting from the segmentation process were transferred to 3-matic (Materialise NV, Belgium) for optimization of the surface quality and correction of potential errors that influence the outcome of the 3D printing process. Optimization operations included global or local smoothing for filtering of image artefacts (Fig. 2) and removal of unnecessary internal surfaces. Models used for pre-surgical planning or testing purposes were hollow, in contrast with models used for educational purposes, which were solid. As the wall of the atria and main arteries cannot be extracted due to limited image resolution and contrast, a hollowing operation was applied when hollow models were required. The wall thickness applied was uniform and dependent on the additive manufacturing technique used and the final purpose of the model. The myocardial mass surrounding the ventricles is commonly visible in the images and can thus be segmented. Boolean operations were applied to unite the multiple components of the heart. Subsequently, a trim operation was used to open inlets and outlets. In some cases, the models were cut to provide visual access to the internal structures of interest.


Once the design was concluded, the 3D reconstructions were reviewed for potential errors, including overlapping or intersecting triangles, presence of unnecessary elements (shells) and holes. When in the presence of errors, these were automatically corrected in 3-matic. 2.3. Additive manufacturing Four different additive manufacturing technologies were used in the generation of the anatomical replicas. Solid models intended as didactic tools were ColorJet printed. Hollow models were manufactured using sterolithography, laser sintering and HeartPrint Flex. The interface between the computer model and the additive manufacturing machines was performed using Magics (Materialise NV, Belgium). 2.3.1. ColorJet Printing Models intended to be used as educational tools were rapid manufactured solid, using ColorJet Printing (CJP). This technology allows for the generation of colored reproductions, where each anatomical structure is represented by a unique hue. The 3D representations were exported in a file format that includes color information, such as colored STL files and VRML files containing texture. A ZPrinter 650 (ZCorporation, USA) was used to create models with a minimum resolution of 0.1mm. This machine uses standard inkjet technology, depositing a liquid binder, zb63 (3D Systems, USA) onto thin layers of powder, zp151 (3D Systems, USA). Print heads move over a powder bed upon which they print the cross-sectional area of the first slice of the object to be manufactured. The build platform is then lowered and a new layer is spread on top of the first layer. The process is repeated until the model is completed. Once the build was concluded, the parts were depowdered with low pressure air. Primary finishing included brushing and sanding the models. Subsequently, the models were dried for 24h at room temperature and cured. A secondary finishing further improves the surface quality. The last step consists in coating the models with transparent varnish to improve its resistance to light and appearance. 2.3.2. Stereolithography Stereolithographic models were produced when transparency, strength and/or water resistance were required. Such models are often used for the deployment of a device (e.g. stent graft, transcatheter heart valve or balloon), either in dry conditions or connected to a pump system with a circulatory fluid simulating the cardiovascular circulation.

Figure 1. Image segmentation process, starting with a) original CT images, b) general thresholding of the lumen, taking advantage of the contrast agent present in the blood pool, c) result of the segmentation process, where all main cardiac structures are represented by a different mask, and d) 3D reconstruction of the heart chambers, great arteries, coronary arteries and myocardium.


Mammoth stereolithography machines (Materialise, Belgium) were used for the creation of cardiovascular models of larger dimensions or when a larger number of parts was required. Tusk XC2700T material, a liquid photopolymeric resin, is cured when struck by a UV laser beam. This laser beam traces the first slice of the model, hardening the resin and originating the part to be manufactured. After each layer is concluded, a reservoir moves over the vessel and releases a film of liquid resin on the basin. The platform is lowered 0.1mm and the process continues until concluded.

44



Figure 2. Smoothing of the ascending aorta starting from a) original 3D reconstruction resulting from the segmentation process. b) Model after applying global smoothing, where it is possible to identify that the overall noise is excluded, but larger irregularities are kept. c) Model after applying smoothing only on selected regions, here represented on the right-hand side for clearer appreciation.

Due to the fact that a liquid polymer is used as manufacturing material, support structures are necessary to prevent overhangs from collapsing. These supports are printed in the same material as the final object, but in a lattice structure, optimally calculated to withstand the part during manufacturing and simultaneously consuming as little material as possible. Such support structures are automatically calculated in the Magics software and manufactured at the same time as the model rises in the building platform. Once the building of the part was concluded, the support materials were manually removed and the model was transferred to a post-curing UV chamber. Post-processing techniques included sanding and lacquering for enhanced transparency and surface finish. 2.3.3. Selective Laser Sintering Laser sintering was applied in the additive manufacturing of cardiovascular models when flexibility and tear resistance were required and transparency was not a requisite. An example of such application is testing device deployment under fluoroscopy imaging Layers of a polymeric powder material, TPU 92A-1, are sintered by a CO2 laser beam describing the cross section of the object. Once the first layer is finalized, the building platform lowers and a new layer is deposited over it. The temperature of the chamber is raised to a temperature just below the sintering point of the TPU material and the laser provides the final energy required to sinter the particles together. Powder material is self-sustainable and therefore does not require support structures. However, multiple objects can be manufactured over the complete extension of the building platform. The optimal distribution of parts in a building platform preventing collision, a process called “nesting”, was automatically performed in the Magics software. Once the building was concluded and the parts had cooled down appropriately, excess powder was removed. If the parts were intended to be used in a circuit with a circulatory fluid, a coating was applied to guarantee watertightness. 2.3.3. HeartPrint Flex HeartPrint Flex (Materialise, Belgium) was used when flexible translucent models of the anatomy were required. These models can be built in a single flexible material or including a rigid material to represent calcifications or implanted devices. In some applications, the models were selectively colored after production and cleaning, for


enhanced contrast of the region of interest, generally pathological structures. 3. RESULTS Multiple cardiovascular replicas of the anatomy of various patients were obtained from the four different additive manufacturing technologies described. Solid colored models additive manufactured included the four chambers, the great arteries and the coronary arteries, as well as abnormal structures, when present. The utilization of different colors for the heart structures allowed the comprehension of the spatial relation between the parts, leading to the identification of the corresponding defects. Such models were reported to provide significantly better learning results than corresponding virtual models [8]. Hollow transparent stereolithographic models were used for multiple purposes, namely testing the delivery and deployment of medical devices. Some of these rigid replicas were subjected to pressure values between 12-18atm, the typical pressure of balloon devices, without breaking. Some models were also made watertight to allow the performance of tests with circulating fluid. If water leakages were observed, these were shown to occur at the connections with rubber tubes, and not at the extent of the additive manufactured models. Wall thickness for these models varied between 1mm and 2mm and depended on the surface finishing required after the model built was concluded. When high transparency was required, a minimum wall thickness of 1.5mm was used, as intensive finishing was performed both on the internal and external surfaces. In models for which transparency was not critical, a 1mm wall thickness was used. If high pressure was to be applied onto the model, a 2mm wall thickness was chosen. Hollow opaque flexible laser sintered models were applied in cases where high resistance was required. These flexible replicas presented extremely high tear resistance, but limited distensibility. Flexibility was seen to be closely related to the wall thickness. This was, therefore, kept between 0.8mm and 1.2mm. For this reason, the representation of the myocardium was not included in models manufactured with this technology. Cases that did not entail the application of liquids or high pressure onto the model were commonly carried out with the HeartPrint Flex technology. The resulting translucent flexible models are considered realistic representations of the

45


cardiac structures as the texture and compliance resembles that of the heart. The Young’s modulus and distensibility of this material were found to be in the range of that of arterial tissue [9]. The minimum wall thickness used in HeartPrint Flex models was 1mm, usually in valve leaflets or small replicas of neonate vessels. Commonly, a wall thickness of 2mm was used for cardiac structures, unless deployment of a device was to be performed. In such cases, wall thickness values up to 5mm were applied. 4. DISCUSSION AND FUTURE WORK Although additive manufacturing of anatomical structures has been considerably common in orthopedic and craniomaxillofacial applications, the utilization of these technologies for the representation of cardiovascular structures is comparatively very recent. The latest advances in imaging techniques have provided the necessary image resolution for the reconstruction of anatomical structures with acceptable level of detail. But while rigid models can adequately represent bone structures, cardiovascular structures often require flexible replicas. Silicone models are largely used as training phantoms, though technical limitations prevent the generation of accurate patient-specific replicas. Furthermore, as the cost associated to the manufacturing of a mold is high, the manufacturing of a new mold for each patient is associated to a high investment.


Additive manufacturing is a suitable technology for the generation of patient-specific cardiovascular replicas in a cost- and time-effective manner. Whereas most additive manufacturing technologies currently build rigid models, HeartPrint Flex allows the creation of multi-material flexible models. Nevertheless, while the material properties can be compared to that of arterial tissue, the complexity of the cardiac tissue cannot yet be replicated with this technology. Tear resistance for tests carried out under pump is currently another limitation of this material. Finally, the transparency of this material could be further improved for enhanced visualization of devices deployed inside the models. Future efforts will be focused on the development of a material that presents mechanical properties comparable to that of the cardiac structures, while being tear-resistant, also when connected to a pump system with circulatory fluid. This material should also present high transparency while allowing selective coloring of structures. ACKNOWLEDGMENT The image data of the congenital heart defects were kindly provided by David Frakes, Heart In Your Hand, Phoenix, Arizona, USA. The image data of the double outlet right ventricle case was kindly provided by Dr. Shi-Joon Yoo, The Hospital for Sick Children, Toronto, Canada. REFERENCES [1] [2]

[3]

[4]

[5] [6]

[7] [8] Figure 3. Examples of the additive manufactured cardiac replicas. a) solid colored model of a tetralogy of Fallot case obtained with ColorJetPrinting. b) hollow rigid stereolithography model of an enlarged heart. c) hollow flexible laser sintered model of a double-outlet right ventricle case, on which the ventricles were virtually cut to allow visual access to the defect. d) hollow HeartPrint Flex model of the aortic root with rigid calcifications with a transcatheter aortic valve device deployed.


[9]

Hoffman, J. and Kaplan, S. 2002, ‘The incidence of congenital heart disease’, Journal of the American College of Cardiology, vol. 39. 12, pp. 1890–1900. Oris, P., Sitthiseripratip, K., Suwanprateeb, J. and Vander Sloten, J. 2002, ‘The cost-effective production of custom biocompatible implants by rapid prototyping – rapid tooling: an experience of 21 cases in Thailand’. Paper presented at the Progress in Rapid Prototyping and Rapid Manufacturing, Beijing, China, unpublished. Goffin, J., Van Brussel, K., Martens, K., Vander Sloten, J., Van Audekercke, R. and Smet, M. 2001. ‘Three-Dimensional Computed Tomography-Based Personalized Drill Guide for Posterior Cervical Stabilization at C1-C2’. SPINE. vol. 12., pp.1343-1347. Gelaude, F., Vander Sloten, J. and Lauwers, B. 2006. ‘Automated Design and Production of Carnioplasty Plates: Outer Surface Methodology, Accuracies and a Direct Comparison to Manual Techniques’. Computer-Aided Design & Applications. vol. 3, Nos. 14, pp. 193-202. Canter, H. et al. 2008. ‘Mandibular Reconstruction in Goldenhar Syndrome Using Temporalis Muscle Osteofascial Flap’. The Journal of Craniofacial Surgery. vol.9., No. 1. pp.165-170. Mavili, M., Canter, H., Saglam-Aydinatay, B., Kamaci, S. and Kocadereli, I. 2007. ‘Use of Three-Dimensional Medical Modeling Methods for Precise Planning of Orthognatic Surgery’. The Journal of Craniofacial Surgery. vol. 18., No. 4. pp.740-747. Lu, S. et al. 2009. ‘Rapid Prototyping drill guide template for lumbar pedicle screw placement’. Chinese Journal of Traumatology. vol.12. No. 3. pp.171-177. Fariha, E. et al. 2014, ‘Color-coded patient-specific physical models of congenital heart disease’, Rapid Prototyping Journal, vol. 20 . 4, pp.336-343. Baeck, K., Lopes, P. and Verschueren, P. ‘Material characterization of HeartPrint® models and comparison with arterial tissue’. Unpublished.

46


24 ADDITIVE MANUFACTURING IN MEDICINE

The Application of 3D Modelling in Biofluid Mechanics Ivan Dogan, Faculty of Engineering, University of Rijeka, Croatia, [email protected] Sven Maricic, Faculty of Engineering, University of Rijeka, Croatia, [email protected] Lado Kranjcevic, Faculty of Engineering, University of Rijeka, Croatia, [email protected] Ana Pilipovic, Faculty of Mechanical Engineering and Naval Architecture, University of Zagreb, Croatia Daniela Kovacevic Pavicic, Faculty of Medicine, University of Rijeka, Croatia

Abstract – This paper presents the usage of 3D modelling simulations in bio fluid mechanics. In the presented case, blood can behave as Newtonian fluid at high shear rates. The main aim is to compare the results of fluid flow through different types of blood vessels. A3D software is needed to model two different types of blood vessels. The first one represents a patient in a good health condition - with healthy blood vessels with clean section. The other vessel model represents a stage with significant plaque presence. A region of interest is placed in junction between two vessels. The presented method gives a better overview of blood behavior, allowing seeing health consequences of plaque in blood vessels. Keywords-Biofluid Mechanics, 3D flow numerical modeling, coronary flow

2. CFD IN BIOFLUID MECHANICS 2.1. Biological characteristics of the blood flow system The flow of blood in the circulatory system is of extremely pulsating character [1]. The cause of pulsation comes from the heart which pumps the blood through blood vessels by contracting and expanding, i.e. by pulsating. Blood vessels can be approximated by a model (Figure 1)of viscoelastic tube of variable diameter and physical properties of the wall. The pulsating blood flow is most pronounced in case of large arteries since they are the nearest to the heart. The more the blood flows away from the heart, the pulsation decreases, until it reaches the capillaries where the pulsation stops and the blood flow becomes uniform.

1. INTRODUCTION In modern engineering the development of computer simulations is essential. It enables control and prediction of behaviour of certain models even prior to their very implementation. This is especially important in the biotechnological field where computer models have been implemented for many years. Fast computer development has enabled more precise modelling and analysis of anatomical models and the presentation of different pathologies. The trends in modern science have been increasingly including teams of experts of various profiles – from e.g. experts in biomedicine to engineers and programmers of virtual applications. One of the topics that is getting more and more attention today is also the analysis of the condition of blood vessels and blood circulation in the human body in general. This topic is especially popular today with the modern ways of living dominated by large intake of fast, fatty oil-fried food. The deposition of plaque on the walls significantly reduces the blood flow. The consequences are various, and one of them is also the risk of an increase of various kinds of cardiovascular diseases. Blocking of blood vessels is one of the factors that may directly lead to heart attack. Computer software for 3D modelling such as SolidWorks or Blender can be used to develop models that may then be analysed by one of the standard computational fluid dynamics (CFD) software such as Fluent or OpenFoam. This paper describes the modelling and the analysis of the condition of blood vessels in two cases: a cross-section of the usual “healthy” blood vessel, and a cross-section of a blood vessel with deposits, the so-called plaque. The data have been sorted and the process has been presented in detail.


Figure 1. Numerical analysis of blood flow computed with ANSYS Fluent [2]. Material is an anisotropic hyperelastic tissue [1]

2.2. Application of CFD in coronary flow CFD modelling software can be used in many medical application, for example: numerical simulations of pulsatile flows and macromolecular (such as LDL) transport in complex blood vessels, including the cerebral artery 3, effect of using a full or partial clamp to control the blood flow streamlines and hence the location of stress concentration in a clean configuration of aorta 4, femoral anastomosis 5, simulation of the behavior of blood flow in microvessels 6, etc. It is known that Coronary Artery Disease (CAD) is responsible for most of the deaths in patients with

47



cardiovascular diseases. Stenosis severity is diagnostically proven by angiography analysis. Angiography gives main anatomical insight into the cardiovascular system. The functional or physiological significance is more valuable than the anatomical significance of CAD. Functional severity of the stenosis is usually diagnosed by the invasive clinical measurement of the pressure drop and flow. The most common clinical technique to evaluate the physiological significance of the coronary artery stenosis is the Fractional Flow Reserve (FFR) procedure. During the FFR procedure the guide wire is introduced into the coronary flow causing the uncertainty in pressure drop assessment for the moderate stenosis cases, usually underestimating it. In such cases Computational Fluid Dynamics proves to be useful noninvasive technique in assessment of the pressure drop and flow pattern analysis of the coronary flow. Govindaraju et al. [7] investigated CFD role in coronary flow diagnostics. Lin et al. [8] investigated complex pulsating causing wave propagation in a coronary system. Wu et al. [9] consider blood as a two-component (non-Newtonian and Newtonian fluid) mixture. Sher et al. [10] assume blood in narrow arteries to behave as a type of generalized Newtonian fluid called Carreau fluid. This work studied the flow pattern in the carotid artery. The carotid arteries are the blood vessels that carry oxygen-rich blood to the head, brain and face. They are located on each side of the neck. Carotid artery disease is the narrowing or blockage of these arteries (stenosis) due to plaque build-up (atherosclerosis). If a piece of plaque or a blood clot breaks off from the wall of the carotid artery it can block the smaller arteries of the brain. When blood flow to the brain is blocked, the result can be a transient ischemic attack (TIA), which temporarily affects brain function, or a stroke, which is permanent loss of brain function. Carotid artery disease is one of the most common causes of stroke. More than half of the strokes occur because of carotid artery disease. In Figure 2 normal carotid artery flow is shown. Figure shows velocity vectors shaded according to their magnitude. Carotid artery branches from the common artery to the external and internal artery. Computational fluid mechanics simulation shows precise blood flow pattern. In the Figure 2 maximal velocity areas show to be on the “inner” side of the external and internal carotid artery immediately after the bifurcation. Velocity profile obtained numerically relates well to on the Womersley number α = 14,628 (for the carotid artery) which is a measure of pulsating inertial force and viscous force [11]. Complex pulsating flow characteristics are introduced in the analytical models with the mentioned Womersley number which has specific values for the different vessels in the system [11]. Inner diameter of the common artery is taken to be DS = 8,6 mm (systolic) to DD = 8 mm (diastolic). Given dynamic viscosity of the blood is µ = 0,0036 Pas, density ρ = 1060 kgm-3 and volume flow Q = 10 mls-1, Reynolds number is calculated as Re = 4Q/Dπυ. Reynolds number takes values Re = 66 to 71, therefore flow is completely laminar.


Figure 2. Normal carotid artery

Flow in the same carotid artery but with stenosis is also modeled and shown in Figure 3. Velocity field comparison of the two cases shows quite different velocity pattern with strong pressure drop and velocity increase in narrow part with stenosis. This flow obstacle causes disturbed flow to propagate into both internal and external artery causing also change in volume flow ratio between the branching arteries. Application of the CFD tool gives as velocity and pressure field in the region of interest and with the desired detail, being totally noninvasive at the same time. With the use of CFD we can also predict disease propagation by numerically simulating the arterial stenosis grow in time and analyzing future change in flow pattern as the flow obstacle grows. This way clinicians can make decisions more easily and set the time frame for future procedures. This problem was modeled with Ansys Fluent but many other CFD numerical modeling software including OpenFoam prove to be more present and useful in clinical decisions. In mathematical and numerical modeling of the coronary flow main mitigating circumstance is fact that the flow is dominantly laminar with turbulent zones not far from the heart in large vessels (approx. 2 cm or more inner diameter). On the other complexity of the models is caused by non-newtonian nature of the blood, elasticity of the vessels and pulsating flow characteristics.

48


ADDITIVE MANUFACTURING IN MEDICINE tissue. The smallest arteries are arterioles, composed of several thin layers of muscle tissue and almost no connective tissue.

Figure 4. Numerica model of human vessel - section view of the artery structure [1]

Figure 3. Carotid artery with stenosis

Pulsating characteristics have been shown to be a result of two pumps. As the primary pump, the heart causes the blood flow to oscillate from zero to very high rates as the valves at the entrances and exits to the ventricles intermittently close and open with each beat of the heart. The second pump is a result of the respiratory and skeletal systems, which exert their greatest action on coronary flow. Complex pulse patterns are further propagated through the rest of the network [2].

Tunica intima is a thin innermost layer of an artery, as thick as one endothelial cell from which it is made. The endothelium plays an important role in the mechanics of blood flow, blood clotting and adhesion of leukocytes. Its task is letting water and tiny molecules through the blood vessel wall, as well as the secretion of vasoactive substances and the contraction and relaxation of smooth vascular muscles.

3. CIRCULATORY SYSTEM The circulatory system is a system of organs whose task is the transfer of substances from and into the cells. In case of humans, i.e. vertebrates, the circulatory system is a closed one. The closed circulatory system means that blood never leaves the system of blood vessels, but rather circulates constantly. The circulatory system consists of the heart and blood vessels (Figure 4). The heart is a muscular organ and the central organ of the circulatory system, whose contractions enable the circulation of blood through blood vessels. The aorta is a massive thick-walled artery which branches into the main arteries, then into arterioles and finally into the arterial capillaries that reach the intercellular space.

3.1. Arteries structure The arteries are wide, impermeable thick-walled blood vessels (Figure 5) and they make the arterial system. They carry away venous blood from the heart towards the lungs and the arterial blood to other parts of the body. There are three types of arteries that can be divided by their structure and size. Elastic arteries, such as the aorta, have a bigger diameter and a large number of elastic fibres. The muscular arteries are of smaller diameter than the elastic arteries and they contain a higher percentage of muscle tissue in relation to the connective


Figure 5. Artery - section view of the structure [1]

Tunica media is the middle layer, composed mainly of smooth muscle cells that make it alive and active. Its contraction and expansion change the artery diameter thus allowing the changes in the blood flow. Elastic fibres are also part of the middle vessel layer. Unlike muscle cells, the elastic tissue does not contain live cells, which renders it passive. The task of elastic fibres, apart from supporting the blood vessels, is to maintain continuous blood flow by the twitching of the expanded blood vessel after the pressure disappears. There is no tunica media in the capillaries. Tunica externa is the outer layer composed of connective tissue that includes passive elastic fibres, as well as of collagen, which is much stronger. Like tunica media, tunica externa is not present in the capillaries.

49


3.2. Blood as non-Newtonian fluid Blood is the basic example of biofluid. It consists of 40 to 45% of formed elements that contain red blood cells the erythrocytes, white blood cells, the leukocytes and platelets. Erythrocytes transport oxygen and carbon dioxide. Leukocytes are blood cells of immune system whose basic role is to protect the organism against microorganisms (bacteria, viruses, fungi and parasites). Platelets in blood are responsible for blood clotting due to their physical properties. The remaining 55 to 60% of blood is blood plasma, transparent yellowish fluid composed of 90% of water and proteins, electrolytes, hormones and nutrients. Blood forms 6 to 8% of human mass. Blood density is 1060 kg/m3, in contrast to the density of blood plasma which is similar to the density of water, 1000 kg/m3. This is because blood contains the previously mentioned blood cells. Viscosity depends strongly on the temperature, but since in humans the temperature is of constant 37ºC, the temperature dependent viscosity change is of no significance. What blood viscosity depends upon are: speed of deformation, hematocrit and the vessel diameter. Figure 6 shows the ratio of the shear stress and the angular velocity of deformation for blood. At the very beginning we can see the area of yield. This is the part where blood resists shear force and behaves like the Bingham fluid. This area is small and ranges from 0,0003 to 0,03 N/m3 after which the blood begins to flow. At the beginning of its flow blood behaves as Newtonian fluid up to the velocity of angular deformation of 100 s-1 after which the viscosity becomes constant and continues to behave as Newtonian fluid.

Figure 6. A relationship between shear stress and velocity

The layer of blood plasma along the vessel wall helps in reducing the viscosity; however, when the vessel diameter reaches the diameter of erythrocytes, there is no blood plasma layer anymore and the blood viscosity increases rapidly. For the blood flow in tubes with the diameter which is smaller than approximately 1 mm the viscosity is not constant with the tube diameter, which means that the blood in such vessels behaves as non-Newtonian fluid.


Clogging of blood vessels is one of the factors that can lead directly to heart attack. In this paper the carotid arteryflow pattern is numerically simulated. Flow simulation results in two similar carotid arteries – normal artery and artery with stenosis is compared. Velocity field comparison shows quite different velocity pattern with strong pressure drop and velocity increase in narrow part with stenosis. This flow obstacle causes change in volume flow ratio between the branching arteries. Application of the CFD tool gives as velocity and pressure field in the region of interest and with the desired detail. With the use of CFD, propagation of the coronary disease can be predicted by numerical simulation of the arterial stenosis grow in time and analysis of the future change in flow structure.By the use of CFD as a precise noninvasive tool,clinicians can make decisions more easily and set more precise time frame for future procedures. ACKNOWLEDGMENT This work is part of the research financed by the project IPA III c – Additive Technologies for the SMEs –AdTecSME. The authors would like to thank the European Union and the Ministry for the financing of this project. REFERENCES [1]

Dogan, I.2014., ‘Primjena 3D tiska u mehanici biofluida‘–final seminar, Faculty of Engineering, University of Rijeka [2] http://ansys.com, 15.08.2014. [3] Hong, J., Wei, L., Fu, C., Tan, W. 2008., ‘Blood flow and macromolecular transpost in complex blood vessels’, Clinical Biomechanics, vol. 23, suppl. 1, pp. S125-S129. [4] Ghodsi, S.R., Esfahanian, V., Shamsodini, R., Ghodsi, S. M., Ahmadi, G. 2013., ‘Blood flow vectoring control in aortic using full and partial clamps’, Computers in Biology and Medicine, vol. 43, issue 9, pp. 11341141. [5] O'Callaghan, S., Walsh, M., McGloughlin, T. 2006., ‘Numerical modelling of Newtonian and non-Newtonian representation of blood in a distal end-to-side vascular bypass graft anastomosis’, Medical Engineering & Physics, vol. 28, issue 1, pp. 70-74. [6] Jafari, A., Zamankhan, P., Mousavi, S.M., Kolari, P. 2009., ‘Numerical investigation of blood flow. Part II: In capillaries’, Communications in Nonlinear Science and Numerical Simulation, vol. 1, issue 4, pp. 13961402. [7] Kalimuthu Govindaraju, Irfan Anjum Badruddin, Girish N. Viswanathan, S.V. Ramesh, A. Badarudin, 2013., ‘Evaluation of functional severity of coronary artery disease and fluid dynamics' influence on hemodynamic parameters’, A Physica Medica, Volume 29, Issue 3, May 2013, Pages 225-232 [8] Bo-Wen Lin, Pong-Jeu Lu, 2014., ‘High-resolution Roeʼs scheme and characteristic boundary conditions for solving complex wave reflection phenomena in a tree-like arterial structure’, Journal of Computational Physics, Volume 260, 1 March 2014, Pages 143-162 [9] Wei-Tao Wu, Nadine Aubry, Mehrdad Massoudi, Jeongho Kim, James F. Antaki, 2014., ‘A numerical study of blood flow using mixture theory’, International Journal of Engineering Science, Volume 76, March 2014, Pages 56-72 [10] Noreen Sher Akbar, S. Nadee, Ain Shams, 2014., ‘Carreau fluid model for blood flow through a tapered artery with a stenosis’, Engineering Journal [11] Waite, L., Fine, J., ‘Applied biofluid mechanics’, 2007., McGraw-Hill

4. CONCLUSION Plaque deposition on the vessel wall significantly reduces the blood flow rate. The consequences are different, and one of them is the risk of various kinds of cardiovascular diseases.


50



Additive Manufacturing, Verification and Implantation of Custom Titanium Implants – Case Studies Hudak Radovan, Zivcak Jozef, Department of Biomedical Engineering and Measurement, Technical University of Kosice, Letna 9, Kosice, Slovakia, [email protected], [email protected] Goban Bruno, Lisy Martin, Marincak Lukas, CEIT Biomedical Engineering, s.r.o., Tolsteho 3, Kosice, Slovakia, [email protected], [email protected], [email protected] Jenca Andrej, Jenca Andrej Jr., Clinic of Stomatology and Maxillofacial Surgery, L. Pasteur University Hospital in Kosice, Rastislavova 43, Kosice, Slovakia, [email protected],[email protected] Gajdos Miroslav, Clinic of Neurosurgery, L. Pasteur University Hospital in Kosice, Trieda SNP 1, Kosice, Slovakia, [email protected] Abstract— Direct Metal Laser Sintering Technology (DMLS) is one of the additive technologies currently used for the manufacture of custom implants made of Ti-6Al-4Vtitanium alloy. Technology (EOSINT M280, EOS GmbH, Germany), titanium material and used software modules (Materialise, Belgium) facilitate the manufacture of implants while maintaining the bionic principles (mechanic, rheological and anthropometric). The article describes three selected case studies of designing, manufacturing and application of custom implants. They include two cranial implants and one large maxillofacial implant. The implants were designed while minimising their weights and applying porous titanium structures. For the verification purposes, in the manufacture process and in the post-processing the metrotomography technologies (Metrotom 1500, Carl Zeiss, Germany) and the Scanning Electron Microscope (SEM) were used. The analysis to compare the current CAD model and the STL model, defectoscopy after the manufacture, and the analyses of implants after the thermal processing and surface finishing were carried out. Manufactured implants were successfully implanted at the Clinic of Stomatology and Maxillofacial Surgery of the Louis Pasteur University Hospital in Kosice, Slovakia, in cooperation with expert neurosurgeons and using the prepared navigation surgical guides. Implant quality and surgical intervention success are significantly influenced by a proper implant design based on the communication with the clinical staff, by setting the parameters of the manufacture process (DMLS), as well as the post-processing parameters. Keywords- additive manufacturing, direct metal laser sintering (DMLS), cranial implant, maxillo-facial implant, verification, industrial computed tomography

1. INTRODUCTION In recent times, the additive manufacturing (AM) becomes well established also in the field of medicine. It facilitates the manufacture of products with high geometric complexity and uniqueness due to the low cost and quick manufacture time, compared to other manufacturing technologies [1]. The


manufacture of implants, small batch or custom-made, is currently carried out using the equipments applying various technological principles, such as direct metal laser sintering (DMLS), electron beam Melting (EBM), selective laser melting (SLM), selective laser sintering (SLS), stereolithography (SLA), polyjet or fused deposition modeling (FDM) [1, 3, 4, 6, 7]. These technologies use various types of biocompatible implantable materials, including commercially pure titanium cpTi, titanium alloy Ti-6Al-4V (Grade 5), chrome cobalt and polyetheretherketone (PEEK) [2, 7, 11, 12]. Specialized software applications and modules facilitate the use of medical imaging data, obtained mainly by the computed tomography (CT) and magnetic resonance imaging (MRI) to create a model of the damaged tissue, design an implant, including the attachments, and prepare an input file for the AM. The manufacture process is followed by the post-processing using several physical methods and technologies. The quality management process, including the verification of input data, manufacture systems, all processed, as well as the product as such, in connection with the medical AM, is currently a frequently discussed issue with an effort to establish the standards [5]. With each step, it is important to be certain that it complies with the requirements that protect both, a patient and a manufacturer. It is necessary to ensure the measurability of input and output data with the purpose to ensure the quality on individual manufacture levels. The process of verification of custom-made implants manufactured using the additive technology should integrate software and input data verification, input imaging data (CT, MRI) verification, manufacture equipment verification, product verification (geometric, material and mechanical verification), manufacturing workplace verification (microbial contamination), verification of processes, post-surgical verification. Implant verification from the biomechanical point of view, especially in case of strength destructive testing, increases the implant’s price, as more pieces must be manufactured. However,

51


a current status of the additive manufacturing repeatability could question also the results obtained in this manner. The solution is, for example, the computer-assisted strength analysis using the finite element analysis (FEA).Mechanical failure of an implant in a patient’s body can partially be predicted by the diagnostics using the industrial computed tomography that facilitates the analysis of, among others, the implant’s internal structure for the presence of pores, disruptions, or other material nonhomogeneity. Post-surgical implant verification or the diagnostics of patients in case of relapse (e.g. in patients with meningioma) particularly represent the issues frequently discussed by interdisciplinary teams. Titanium, or its alloys, are criticised for the impossibility to perform the diagnostics repeatedly using the CT or MRI, for example in case of cranioplasty. One of the solutions suggests the use of titanium or its alloys in the additive manufacturing in secondary reconstructions of large cranial defects formed in decompressive craniectomies following the trauma or bone infarction. Other materials, such as PMMA, should be used in primary reconstructions requiring the postsurgical imaging diagnostics [7]. Future studies in this segment might focus on the use of imaging phantoms with the application of implants of various sizes and subsequent artefact analysis. Certain disputes have also been raised in connection with thermal conductivity of titanium implants and the heat impact on the brain tissues, or the solution of re-surgeries in patients with cranioplasties. Recently, several case studies were published in journals, discussing the implants manufactured by the AM and subsequently implanted in patients. For example the EBM technology was used to manufacture hip implants, acetabular cups, including the manufacture of lattice structures [3, 4]. Drstvensek et. al. in 2008 produced mandibular implant by EOSINT M270 (EOS GmbH, Germany) machine out of Ti-6Al4V ELI material, certified for medical use. Machine builds metal parts using DMLS technology [13]. A drill guides were produced by same author in 2013 by rapid prototyping using SLS technology [14].


screws. A large bone defect was the result of the decompressive craniotomy. The source 3D data for the implant design were obtained using the computed tomography (CT) [10]. Software data were prepared using the In Vesalius package, SolidWorks and Magics (Materialise, Belgium) software applications. A medical biomodelwas builtusing a 3D Printer system (ZCorporation); 3D Zprinter 510 system, and materials ZP 130 (powder gypsum), ZB 58 (binder) and ZBond (resin).The prosthetic implant was fabricated by the direct metal laser sintering (DMLS) technique using the EOSINT M270 system (EOS GmbH, Germany). The material used was EOS Titanium Ti64 ELI (EOS GmbH, Germany), because it fulfils mechanical and chemical requirements of ASTM F 136 standard [10]. Following part of the paper describes 3 case studies of implants manufactured applying the DMLS technology by the CEIT Biomedical Engineering, s.r.o. company in cooperation with experts from the Department of Biomedical Engineering and measurement, Faculty of Mechanical Engineering of the Technical University of Kosice. The case studies describe two large cranial implants and one facial implant that have been implanted at the Clinic of Stomatology and Maxillofacial Surgery of the L. Pasteur University Hospital in Kosice supervised by surgeon prof. Andrej Jenca. 2. CASE STADY 1 - CRANIAL IMPLANT A patient aged 51 years experienced a skull fracture caused by the fall. The fractured bone was partially reconstructed, embedded at the original location and fixed to the adjacent hard tissues. This fracture was located in the area of parietal bone, on the right side. Reconstructed bone was fragile with the tendency to break. Moreover, after certain period of time it depressed, directed towards the brain, thus the risk of brain damage increased. Due to these circumstances, a custom-made medical device was indicated for the patient. The objective was to fabricate an implant that would cover critical and fragile parts and at the same time visually supplement the depressed skull, so that the patient could be fully integrated in the society.

In 2011, Dr. Joules Poukens and his team performed the total mandible replacement by implant produced using selective laser melting (SLM) technology in a female patient aged 83 years. The implant was coated with hydroxyapatite, a bio-active bone substitute [8]. Several studies indicate the use of additive technology in the manufacture of cranial custom implants that are applied in cranioplasty [9]. Cranioplasty is the procedure of choice for treating cranial defects commonly caused by trauma, tumor removal, or decompressive craniotomies. The main goal of cranioplasty is to protect the brain and alleviate psychological affliction caused by the defect and enhance social performance of the patients [7]. In 2014, A.L. Jardini et al. manufactured and implanted a cranial implant to a 28-year-old patient who experienced the trauma after a motorbike accident injury. The operation lasted approximately 3 hours. The implant manufactured by the additive manufacturing (AM) covered the area of 13.5 x 9.4 cm2.It was fixed to the right frontal bone of the skull using two


Figure 1. Modelling of cranial implant using mirroring – left, design of fixation elements with definition of the positions with length of particular screws - right;

The model was designed on the basis of the existing CT data of the patient by use of Mimics module (Materialise, Belgium). The implant design was created by a mirror reflection of the undamaged half of the parietal bone using 3-matic (Materialise, Belgium) (Fig. 1). Guide curves drawn on the basis of the reflected part were used to deduce the hypothetical lines of the implant so that the geometric continuity of the organ remained undisturbed. Concurrently, its margins adjoin to the margins of

52


adjacent bone structures to maintain smooth joins. Identification of proper shape required only 3 vertical and 3 horizontal lines. Eventually, a generated prosthesis area was drawn out to space with the specified final implant thickness of 1.5mm. The last crucial step was to design the fixation systems. Fixation points were chosen according to the bone thicknesses. Each of the fixation systems had 3 opening options to choose from for the fixation(Fig. 1). The thickness of the systems was 0.6 mm and they were smoothly emerging from the implant’s structure and adjoining to the adjacent bone. Final weight of the implant was 73 grams (Fig. 2).


3. CASE STUDY 2 - CRANIAL IMPLANT A patient aged 29 years experienced an injury after falling from a 4-meter height, which resulted in an extensive skull fracture and comminution on the left frontal side. Prior to the surgery, the patient had difficulties with motor movements of lower and upper limbs, including the holding function, and was mostly able to move only if accompanied with another person or in a wheelchair. His speech was difficult and he was joining words into sentences with frequent and long intervals, feeling tired after several sentences. After previous traumatological interventions on his cerebral cortex, the patient had a PMMA implant inserted in his cranial cavity (Fig. 4), which was not fixed laterally and was thus pushing directly onto his brain in case of changes in the pressure or during physical activities.

Figure 2. Implant made of Ti-6Al-4V alloy by DMLS technology – left, placement of the implant during the surgery – right.

To verify the dimensional and functional properties, plastic referential implant models and a mating component (skull with the defect) were fabricated during the modeling process to verify the shapes and dimensions. Plastic models were fabricated applying the Polyjet technology (Objet Eden 260V, Stratasys, USA). The implant as such was manufactured applying the DMLS technology, using the EOSINT M280 equipment (EOS GmbH, Germany) equipped with a 200 W fibre laser. The implant was made of Ti-6Al-4V (Grade 5), due to its biocompatible and mechanical properties. After the DMLS manufacture process, the implants had to be thermally treated, grinded, polished, and sand-blasted to acquire the required dimensions and surface properties.

Figure 4. Model of the patient’s skull defect with previous implant – left, placement of the new implant on the patient skull – right.

CAD modeling of the cranial implant was based on the patient’s CT images scanned by the medical CT in cross-section planes with the spacing of 1.5mm. The implant was modeled on the basis of mirror reflection of the healthy part of the skull and deduction of its curves to define the implant contours (Fig. 4).

The implant was successfully applied in May 2013 (Fig. 3) and after the recovery, the patient reports improved physical condition resulting from his self-confidence in everyday activities. The patient is now 15 months after the surgery in a good health condition (Fig. 3).

Figure 5. Large cranial implant – the final design – left, placement of the implant during the surgery – right.

Figure 3. Patient a week after the surgery – left, patient 15 months after the surgery with almost invisible scar - right.


Large custom-made cranial implant (Fig. 5) was manufactured applying the DMLS technology, using the EOSINT M280 equipment (EOS GmbH, Germany). The implant manufacture was followed by thermal treatment, subsequent grinding and sand-blasting to achieve the required surface parameters. The surgery process did not involve any complications, with regard to the size of the implanted part. After the recovery, the patient demonstrates fast convalescence with significant improvement in his motor function and speech abilities (Fig. 6).

53



4. CASE STUDY 3 - LARGE FACIAL IMPLANT

CT scans including over 60 anonymous subjects. The next step was to filter the data and choose the most appropriate ones to be used in our case, based on anthropometric dimensions. After sorting out the models, the 5 fitting ones were chosen and exported to 3D masks. In this case, the key selection factor was the upper jaw size as well as size and shape of orbital arcs of frontal bone. We used undamaged pterygoid plates as a reference to create the upper jaw arc. The only part of the implant virtually taken from the existing model was zygomatic and zygomatic process bone. The virtual arc fitted into zygomatic processes completed and settled the basic reference system of the future implant. Based on these references, out of 5 fitting scans the 3 best fitted were chosen (Fig. 7).

The patient was a 34 years old male, after the car accident. He experienced a severe injury and extensive soft and hard tissue damage (Fig. 7).

In the whole process, several plastic prototypes were manufactured by the PolyJet technology (Objet Eden 260V, Stratasys, USA) to check and verify the designing process.

Figure 6. Patient before the surgery – left, and 2 months after the surgery – right.

From the anatomical point of view, maxilla, vomer bone, nasal bone and approximately half of the lower jaw were completely missing. Heavily damaged were the palatine bone and the palatine process, frontal bone, left zygomatic bone and the remaining part of the lower jaw was assembled to one piece – not fully functional. Left eye is missing; the other one is fully functional. The patient had a few reconstruction surgeries in which part of his supraorbital foramen was reconstructed and cheek backing was added. All via bone grafts. The remaining part of the maxilla bone and the maxilla palatine bone was attached to the frontal bone. The patient could speak with poor articulation and he could taste food, but could not chew. Our aim was to restore the patient’s appearance in terms of achieving a condition closer to his original appearance and return to him his ability to speak and ingest normal food with some limitations. The project included 2 phases: 1. Reconstruction of facial bones including the functional upper jaw. 2. Reconstruction of the lower jaw. The main idea was to minimize the number of surgeries to avoid possible infections and other complications and design a 1-piece implant from the Ti-6Al-4V (Grade 5) material that meets all the requirements and standards. The implant should recover over 85% of the missing original facial bones. Its weight was lowered to the possible minimum, not endangering its mechanical properties, as in the end the implant will be mechanically loaded by chewing.

Figure 8. Final design of maxillofacial implant – left, implant after the production and postprocessing with porous structure and fixation holes.

The weight of titanium implant was 179 g, which was confirmed by biomechanical analysis as an acceptable weight. After postprocessing the weight was reduced to final 165 grams. Another step was the settlement of holes for teeth implants. Based on the existing 3D model of the used upper jaw and upon the consultation with dental technicians, the holes placement was proposed in the axis of the belonging teeth model. After the implantation and the healing process, dental bridge will be additionally created by dental technicians. The implant was produced by the additive manufacturing using the DMLS technology EOS M280 (EOS GmbH, Germany), and surgery was realized in June 2014. Two titanium implants were produced – one for the testing and one for the application. This is the biggest known maxillofacial titanium implant produced by the additive manufacturing in the history of this field (Fig. 8). It should integrate the subject back to the community and restore the lost vital functions. CONCLUSION

Figure 7. Defect of the maxillofacial area after the accident and first maxillofacial reconstruction – left, composition of the implant by 4 different source of data – right.

The skull shape of the subject was not symmetrical and due to damage of both sides of the face the mirroring features of the software was useless. We had to create the database of various


The article describes 3 successful case studies of the design, manufacturing, and implantation of two cranial implants and one unique maxillofacial implant. All the implants were manufactured applying the DMLS technology, using the EOSINT M280 equipment (EOS GmbH, Germany), from the Ti-6Al-4V titanium alloy (Grade 5). Dimensional and material parameters of the implant were verified applying the industrial computed tomography (Metrotom 1500, Carl Zeiss, Germany) and using the plastic referential model of the anatomical part with the bone defect.

54


Good conditions of patients after the surgery, enabling mental health and social inclusion of these patients, confirm the importance of applying the DMLS additive technology and the above mentioned titanium alloy in the manufacture of custommade implants. With certain types of implants and diagnoses, the use of material should be considered in order to ensure the possibility of postoperative monitoring of a patient with potential subsequent surgical intervention. The attention should also be paid to the process of monitoring and verification of additive manufacturing with gradual optimisation of parameters, provision for improved manufacture repeatability, and new materials development. As these implants are custom-made, the attention will also have to be paid to the process of mechanical and thermal verification of implants by computer-assisted simulations that can be integrated into the used software applications. ACKNOWLEDGMENT Presented manuscript was supported by project Research of new diagnostic methods in invasive implantology, MSSR3625/2010-11, Stimuls for Research and Development of Ministry of Education, Science, Research and Sport of the Slovak Republic, the project KEGA 036TUKE-4/2013, with title Implementation of new technologies in design and fabrication of implants in biomedical engineering and related scientific fields and project Center of excellence of biomedical technologies (ITMS: 26220120066) supported by the Research & Development Operational Programme funded by the ERDF. REFERENCES [1]

[2]

Ciurana, J., 2013, ‘New Opportunities and Challenges for Additive Manufacturing to Produce Biomedical Devices’, Department of Mechanical Engineering and Industrial Construction, University of Girona, Girona, 17003, Spain, Su, X., Yang, Y., YU, P., SUN, J., 2012 ‘Development of porous medical implant scaffolds via laser additive manufacturing’, Trans. Nonferrous Met. Soc. China 22 (2012) pp 181-187



[3]

[4] [5] [6]

[7] [8] [9] [10] [11]

[12] [13]

[14]

Koptyug, A., Rännar, Bäckström, M., et al. 2013, ‘Additive Manufacturing Technology Applications Targeting Practical Surgery’, in International Journal of Life Science and Medical Research, Feb. 2013, vol. 3, pp. 15-24. Petrovic, V., Haro, J.V., Blasco, J.R., Portolés, L., 2012,‘ Additive Manufacturing Solutions for Improved Medical Implants’in Biomedicine, ISBN 978-953-51-0352-3, pp.147- 181 Wai Yee Yeong and Chee Kai Chua, 2013, ‘A quality management framework for implementing additive manufacturing of medical devices’, Virtual and Physical Prototyping, 2013 Vol. 8, No. 3, 193–199, Jardini, A.L., Larosa M.A., Zavaglia, C.A., Bernardes, L.F., Lambert, C.S., Kharmandayan, P., Calderoni, D., Filho, R.M., 2014, ‘Customised titanium implant fabricated in additive manufacturing for craniomaxillofacial surgery’, Virtual and Physical Prototyping, 2014, Vol. 9, No. 2, 115–125,. Parthasarathy, J., 2014, ‘3D modeling, custom implants and its future perspectives in craniofacial surgery’, Ann Maxillofac Surg. 2014 Jan-Jun; 4(1): 9–18. Poukens, J., Laeven, P., Beerens, M., Koper, D., Lethaus, B., Kessler, P., Sloten, V., Lambrichts, I., 2010, ‘Custom surgical implants using additive manufacturing’, in Digital dental news, 2010, pp. 30-33 Kroonenburgh, I., Beerens, M., Engel, C., Mercelis, P., Lambrichts, I., Poukens, J., 2012, ‘Doctor and engineer creating zhe future for 3D printed custom made implants’, in Digital dental news, 2012, pp. 60-65 Jardini, A.L., Larosa, M.A., et al., 2014, ‘Cranial reconstruction: 3D biomodel and custom-built implant created using additive manufacturing’, in Journal of Cranio-Maxillo-Facial Surgery, pp 1-8 Jardini, A.L., Larosa, M.A., Bernardes, L.F., Zavaglia, C.A.C., Maciel Filho, R., 2011, ‘Application of direct metal laser sintering in titanium alloy for cranioplasty’, 6th Brazilian conference on manufacturing engineering, April 11th to 15th, 2011 – Caxias do Sul – RS – Brazil Neumann, A., Kevenhoerster, K., 2009, ‘Biomaterials for craniofacial reconstruction’, GMS Current Topics in Otorhinolaryngology - Head and Neck Surgery 2009, Vol. 8, ISSN 1865-1011, pp. 1-17 Drstvenšek, I., Ihan Hren, N., Strojnik, T., Brajlih, T., Valentan, B., Pogačar, V., Zupančič Hartner, T. ‘Applications of rapid prototyping in cranio-maxilofacial surgery procedures’. International journal of biology and biomedical engineering, 2008, vol. 2, iss. 1, pp. 29-38., ISSN 19984510. Merc, M., Drstvenšek, I., Vogrin, M., Brajlih, T., Rečnik, G., ‘A multilevel rapid prototyping drill guide template reduces the perforation risk of pedicle screw placement in the lumbar and sacral spine’. Archives of orthopaedic and trauma surgery, ISSN 0936-8051, 2013, vol. 133, no. 7, pp. 893-899

55



Personalized Shoulder Endoprosthetic for Glenoid Defect – A Case Report

Tomaz Tomazic, Orthopaedic department, Teaching hospital Murska Sobota, Slovenia, [email protected] Slavko Kramberger, Head of the Orthopaedic department, Teaching hospital Murska Sobota, Slovenia, [email protected] Attila Szunyog, Orthopaedics department, Teaching hospital Murska Sobota, Slovenia, [email protected] Breda Jesensek Papez, Head of the Institute of Physical and Rehabilitation Medicine, UKC Maribor, [email protected] Tomaz Brajlih, Faculty of Mechanical Engineering, University of Maribor, Smetanova. 17, Slovenia, [email protected] Igor Drstvensek, Faculty of Mechanical Engineering, University of Maribor, Smetanova. 17, Slovenia, [email protected]

Abstract - In severe shoulder arthritis cases with poor bone stock, thanks to the preoperative 3D-CT joint model, a use of personalised resection guides can help with the determination of the exact bone cuts, best anatomical axis and optimal component position due to the available bone stock. Especially in the glenoid defects the component positioning is a key factor for success of the shoulder arthroplasty, where the resection guide technique shows gut reliability and high precision. Keywords – reversed shoulder arthroplasty, personalised resection guides, JIGs, 3D-CT model, component positioning by glenoid defect, rotator cuff arthropathy 1.

advanced glenohumeral (GH) wear, with reduced subacromial space and severe loss of glenoid bone stock. The shoulder ultrasound showed a major rotator cuff tear. Because of the pain, the patient received intraarticular steroid injections and went to physiotherapy. Due to the ineffective conservative treatment the implantation of reversed total endoprosthesis (EP) of the right shoulder was indicated. 2.

PREOPERATIVE PLANNING

As part of the preoperative planning we performed a CT scan with 3D-CT reconstruction showing a severe GH arthritis with a large glenoid defect and the pathological glenoid retroversion (Figure 1).

INTRODUCTION

Due to the prolonged and severe pain in the right shoulder joint, accompanied by extremely limited mobility, the 79 year old female patient came on 5.2.2014 for the first examination to our clinic. The symptoms disabled the patient at her daily tasks and live skills. The active shoulder mobility was limited far below the horizontal line and due to the blocked rotation the functional movements to the neck and back were not possible. The passive shoulder abduction was up to 40˚, the anteflexion up to 60˚ and the rotations were blocked. X-ray showed


56



Figure 1: Preoperative 3D-CT scan with the determination of the pathological cup retroversion due to the scapular line

We used a JOINTcut method, which we started to develop in 2010 [1]. Due to the CT scans we managed to perform an individual preoperative 3D CT model of the affected joint based on the 2 mm slice thickness CT images. The images were stored in DICOM (Digital Imaging and Communications in Medicine) format and transferred to a workstation running EBS ver. 2.2.1 (Ekliptik, Slovenia) software to generate an automatic 3D reconstruction model for the joint (Figure2).

Figure 3: Preoperative virtual glenoid cup position due to the individual 3D-joint-model. Determination of the cup retroversion and the depth in the available bone socket

Upon this preoperative 3D-CT scan of a degenerative joint [3], we can create a virtual and individual 3D-joint-model in the EU-certificated polyamide material with the determination of optimal and exact joint resection levels for the endoprosthesis placement. Based on this model we can create individual resection guides (jigs), modified according to the surgeon’s suggestions. With the 3D printer we product the individual jigs, fitting optimal to the bony surface of the patient’s anatomy and helping the surgeon to perform the precise bone cuts for the EP placement, without changing the standard surgical approach (Figure 4).

Figure 2: Preoperative 3D-CT shoulder model present the retroversion and the large dorsal defect of the glenoid

Upon the 3D bony landmarks, transverse section of the scapula and shoulder joint and the principles of least-squares minimization, the system performed automatic segmentation and determined the transverse neutral scapular plane and the centre of shoulder joint rotation, responsible for the exact EP placement [1]. Through the centre of rotation of the humerus and glenoid it determined the mechanical axis of the shoulder due to the transverse scapular plane in the resection level, taking into account the combined retroversion (CRV) of 30° (CRV is the sum of glenoid and humerus retroversion). The preoperative determination of optimal size and EP position due to the inclination, retroversion of the glenoid and humerus [2], centre of rotation and especially the available bone stock of the glenoid are all crucial for the endoprosthesis placement and enables the best postoperative kinematic of the EP (Figure 3).


Figure 4: Personalized 3D-joint-model in the polyamide material with the humeral and glenoidal jigs

The advantages of the jigs are especially the accurate positioning of the glenoid entry point [4] and well determined inclination and retroversion of the glenoid component (optimal between 0° and 10° of retroversion). They determine also the depth of the central peg and the end position of the implants [5]. This helps to achieve maximum stability and the best possible kinematic of the EP. Our first goal is the most optimal bone fixation of the glenoid implant due to the available bone stock, without stressing the primary on its retroversion and rather adapts the humerus head retroversion, which has been the primary rule for the resection in the conventional technique [6]. This helps us to place the glenoid in the optimal position due to the available bone stock and ensure the best bony fixation of the component [7]. By that we can compensate the humerus retroversion within reasonable limits (optimal CRV of the glenoid and humerus is between 25° to 35°) and achieve the best possible positioning of all EP components. Due to the bone defect of the posterior glenoid wall we need a bony graft to compensate the defect under the implant [8]. The preparation of the graft with

57


conventional techniques is difficult, especially when the humeral head is damaged by positioning the conventional central drill needed for the resection of the head. In our resection guide technique the central drill is not needed, hence we do not damage the resected head and the preparation of the graft is, thanks to the additional resection line for the graft, much easier and more precise. Due to complete rotator cuff tear and severe osteoarthritis with bone defect the Reversed shoulder endoprosthesis was chosen. The optimal retroversion of the glenoid implant due to the best bony fit and the preoperative bone model was 10° and the associated humerus retroversion was 20° in order to achieve an optimal CRV of 30°. The glenoid inclination was determined to 5°. The preoperative measured humeral steam diameter was 15 mm and the cementless small-R glenoid with small-R liner for metal back was determined. The head diameter was 40 mm and the CTA head adapter for reverse humeral body with accompanying +3mm reverse liner was chosen. 3.


Figure 6: Position and the shape of the bone graft covering the dorsal glenoid defect

After the placement of the bone graft and glenoid base plate, the implant was additionally fixated with two screws guided through the extra jig. Intraoperative 16 mm diameter cementless finned steam with reversed humeral body and the Small-R cementless glenoid was implanted. Two 25 mm bone screws for glenoid fixation were placed. The M-insert and the 40 mm head were chosen. We didn’t need to change the preoperative determined retroversion of the glenoid or humeral component. After the implantation of all the endoprosthesis components (Figure 7a, 7b) we assessed the passive range of motion and the stability of the EP. After the surgery the arm was placed in abduction orthosis.

OPERATIVE TREATMENT

The first operative step after the preparation of the bony joint surface is the placement of the humeral jig required for the humeral head resection in the preoperative determined retroversion and inclination, as for the prepared glenoid bone graft resection. After the removal of the humeral head, the jig guides us to find the centre of humeral diaphysis and perform the socket for the humeral component. Using the glenoid jig (Figure 5) we can easily find the entry point of the glenoid centre needed for the placement of the central Kirschner wire (KW) as a guide for the reamer [9]. Figure 7a: Position of the humerus jig and humerus resection

Humeral and Glenoid Jig (Resection Guide)

Figure 7b: Placement of the bone graft and glenoid base plate Figure 5: Humeral jig with resection slot for the head and the Glenoid jig with the central entry point for the KW

Due to the jig, not only the direction, but also the depth of the component is determined, as the position and shape of the bone graft [10], covering the dorsal glenoid defect (Figure 6).


4.

POSTOPERATIVE TREATMENT

On the first postoperative day we performed an Xray (Figure 8), and 4 weeks after surgery an additional CT with the 3D reconstruction showing the position of the EP and the in-growing amount of the autograft (Figure 9). The postoperative retroversion of the glenoid implant measures 11° and the retroversion of humerus 20° [11]. The

58


positioning of the central peg is optimal to the available bone stock and the implanted bone graft is well glowed in. After the surgery the patient began with continues passive motion and gentle assisted active exercises below the shoulder level. She came to active rehabilitation a month after the surgery and after the postoperative CT -scan showed good bone graft integration, she begun with active exercises above the shoulder girdle. In the mean time she received sufficient analgesia and learned exercises were carried out in the home environment. The patient's mobility improved from 70° of abduction to 130° and after the second hospital stay the patient continued with intensive rehabilitation at the spa.

Figure 8a: Preoperative RTG image


anatomical axis and glenoid position in the available bone stock, thanks to the preoperative 3DCT joint model, is easier and more precise. Our goal is the most optimal bone fixation of the glenoid implant in the available bone stock without stressing the primary on its retroversion and rather adapts the humerus head retroversion, which has been the primary rule for resection with conventional techniques. Due to this 3D preoperative model, we can create resection guides with optimum fits to the bony surface of the patient’s bone and determine the exact intraoperative bone cuts, the shape and size of the bony transplant, as well as the optimal positioning of the EP, which enables the best stability and kinematic of the prosthesis. The postoperative 3D-CT EP positioning has been analysed by comparing the planned and performed resections and the final joint biomechanics. The first result showed an easier, quicker and more precise procedure, with optimal postOP-3D-CT mechanical axis and CRV. Due to the optimal bone fit of the central peg and the combined retroversion of both components, we expect optimal biomechanics, functions and prolonged EP survival.

Figure 8b: Postoperative RTG image

6.

REFERENCES

[1]

Figure 8: Preoperative and postoperative X-ray

Figure 9: The postoperative retroversion of the glenoid, the positioning of the central peg in comparison to the available bone stock and the autograft ingrowed

5.

CONCLUSION

A defective glenoid with pore bone stock is the most challenging problem in the shoulder EP. For these severe cases, mostly a bone graft for the restoration of the glenoid shape and implant position in the functional plane is needed. Due to the new personalised resection guide technique developed with Slovenian technology, the determination of the optimal bone cuts, best


Drstvensek I., Brajlih T., Tomazic T. Additive manufacturing in practical use. Journal of Trends in the Development of Machinery and Associated Technology. Vol. 17, No. 1, 2013, ISSN 2303-4009 (online), p.p. 9–16 [2] Churchill RS, Brems JJ, Kotschi H. Glenoid size, inclination, and version: an anatomic study. J Shoulder Elbow Surg 2001;10:327-32. [3] Friedman RJ, Hawthorne KB, Genez BM. The use of computerized tomography in the measurement of glenoid version. J Bone Joint SurgAm 1992;74:1032-7. [4] Nicholas J. Meyer, William T. Pennington, Dean W. Ziegler. The Glenoid Center Point: A Magnetic Resonance Imaging Study of Normal Scapular Anatomy. Am J Orthop. 2007;36(4):200-202. [5] Couteau B, Mansat P, Mansat M, Darmana R, Egan J. In vivo characterization of glenoid with use of computed tomography. J Shoulder Elbow Surg 2001;10:116-22. [6] Farron A, Terrier A, Buchler P. Risks of loosening of a prosthetic glenoid implanted in retroversion. J Shoulder Elbow Surg 2006;15:521-6. [7] Codsi MJ, Bennetts C, Powell K, Iannotti JP. Locations for screw fixation beyond the glenoid vault for fixation of glenoid implants into the scapula: an anatomic study. J Shoulder Elbow Surg 2007;16(3 suppl):S84-9. [8] Cofield RH. Bone grafting for glenoid bone deficiencies in shoulder arthritis: a review. J Shoulder Elbow Surg 2007;16(5 suppl):S273-81. [9] Anne Karelse, Steven Leuridan, Alexander Van Tongel, Iwein M. Piepers, Philippe Debeer, Lieven F. De Wilde. A glenoid reaming study: how accurate are current reaming techniques? J Shoulder Elbow Surg (2014) 23, 1120-1127 [10] Gilles Walch, Peter S. Vezeridis, Pascal Boileau, Pierric Deransart, M. Eng, Jean Chaoui. Three-dimensional planning and use of patient specific guides improve glenoid component position: an in vitro study. J Shoulder Elbow Surg (2014), 1-8 [11] Jonathan C. Levy, Nathan G. Everding, Mark A. Frankle,

59



Louis J. Keppler. Accuracy of patient-specific guided glenoid baseplate positioning for reverse shoulder arthroplasty. J Shoulder Elbow Surg (2014) 23, 1563-1567


60


DESIGN FOR ADDITIVE MANUFACTURING

CHAPTER 2.

Design for Additive Manufacturing


61



Techniques and practices for the successful, cost effective reconstruction of skeletal elements of the last European elephant of Tilos with LOM and FDM Additive Manufacturing technologies An interdisciplinary approach of AM for palaeontology Stamatios Polydoras, Rapid Prototyping & Tooling Laboratory, School of Mechanical Engineering, National Technical University of Athens (NTUA), Zografou-Athens, Greece, [email protected] Christopher Provatidis, Mechanical Design & Control Systems Division, School of Mechanical Engineering, National Technical University of Athens (NTUA), Zografou-Athens, Greece, [email protected] Theodoros Vasilopoulos, Mechanical Design & Control Systems Division, School of Mechanical Engineering, National Technical University of Athens (NTUA), Zografou-Athens, Greece, [email protected] Evangelos Theodorou, Mechanical Design & Control Systems Division, School of Mechanical Engineering, National Technical University of Athens (NTUA), Zografou-Athens, Greece, [email protected] Georgios Theodorou, Department of Historical Geology and Palaeontology, Faculty of Geology and Geoenvironment, National and Kapodistrian University of Athens (NKUA), Zografou-Athens, Greece, [email protected] Vassiliki Mitsopoulou, Department of Historical Geology and Palaeontology, Faculty of Geology and Geoenvironment, National and Kapodistrian University of Athens (NKUA), Zografou-Athens, Greece, [email protected]

Abstract— The paper presents interesting techniques and best practices followed within the ongoing EU-funded research project Thales - MIS380135, for the physical reconstruction of prototype skeletal element replicas, that will ultimately form an anatomically complete full-skeleton exhibit of an Elephas tiliensis, the last dwarf elephant to have lived in Europe. The skeletal elements constructed so far have been based on digital data acquired, processed and prepared from numerous actual fossils of several individuals recovered from excavations in the island of Tilos, Greece from 1971 to date. The project is an interdisciplinary approach, largely based on established innovative Mechanical Engineering principles, technologies, software and equipment, methodically put in use for palaeontological, natural heritage and educational purposes. Computer Tomography, Laser Scanning, Reverse Engineering, 3D CAD Modeling and Additive Manufacturing are involved and combined to achieve the end result in a repeatable, rapid and cost effective manner. The work has been scientifically supported by biologists and palaeontologists, who determine the skeletal elements’ final form, dimensions, analogies and level of detail, based on taphonomy, ontogenic observations, allometry, statistical analyses and mathematical modeling. Challenges regarding criteria and decision for the specimens’ AM distribution, limitations imposed by the nature and technical specifications of the LOM and FDM AM technologies and

equipment available to the research group, data and file size handling, fabrication cost and build time minimization and optimization, as well as, overall quality of the final result, had all to be considered and have been addressed. Specific techniques and critical process chain steps followed that successfully lead from original fossils to final skeletal element reconstructions, such as digital files generation and their AM preparation, segmentation, AM packing and orientation and final re-assembly methods, practices and parameters are demonstrated and discussed. Characteristic images of the results so far are presented and the introduced innovation and benefits already achieved for the field of palaeontology is assessed. Keywords- Additive Manufacturing; Rapid Prototyping; Laminated Object Manufacturing; Fused Deposition Modeling; Palaeontology; Elephas tiliensis; Digitization; Reconstruction

1. INTRODUCTION Additive/Layer Manufacturing (AM/LM), also recognized as Rapid Prototyping (RP) and by many simply as 3D Printing (3DP), has shown an impressive growth in the last two decades and can be definitely considered as a true modern technological revolution, [1]. This is due to the many different ways it has – during that time – been successfully implemented in many “closely related” technological fields, such as Industrial

This research has been co-financed by the European Union (European Social Fund – ESF) and Greek national funds through the Operational Program "Education and Lifelong Learning" of the National Strategic Reference Framework (NSRF) - Research Funding Program: Thales. Investing in knowledge society through the European Social Fund.


62


Design, New Product Design and Development, Plastics and Metals Molding, Casting and Manufacturing, Architecture, Medical and Bio-Medical Engineering. Lately though, the AM technological spectrum also gains ground in other, more “distant” scientific and application areas, e.g. GIS, Archaeology, Entertainment and others, by serving their needs in a really modern, fast, flexible and cost-effective manner, [2]. Such a diversified application of AM for palaeontology, a research work currently in progress within the frame of EUfunded research project Thales - MIS380135, is presented below, as an interesting interdisciplinary approach of great semantic importance for the European Natural Heritage. 2. BACKGROUND INFORMATION 2.1. Elephas tiliensis Greece lies exactly on the verge of three continents, Europe, Asia and Africa. Its location is therefore significant for historical, geographical, geological and palaeontological reasons and without doubt presents a unique natural and cultural environment. Over the years many archaeological and palaeontological excavations have revealed this rich heritage of Greece. Located on the Island of Tilos in Dodecanese island complex, the Charkadio cave is one of the richest fossiliferous sites studied by palaeontologists, hosting remains of many species of birds, reptiles and mammals; among them fossils of the last European dwarf elephant, Elephas tiliensis, have been unearthed, [3]. This insular form of elephant, ranging in adult height between 1.4 to 1.7m, was the last to inhabit a Mediterranean Island and the whole European continent in general. It migrated to Τilos 45,000 years ago and became extinct about 3,500 years BP (Before Present), probably due to the aftermaths of the major volcanic eruption of Santorini, so this elephant survived well into the Holocene period. The scientific and cultural importance of recreating, preserving and exhibiting a full-scale complete skeleton of a typical adult individual of this elephant species is obviously great not only for the palaeontology scholars globally, but also for the touristic public visiting the island of Tilos. Unfortunately, despite the large number of elephant fossils excavated from Charkadio since 1971, corresponding to approximately 77 individuals of different sex and age, few only anatomically associated skeletal elements have been found capable to define the skeletal configuration of a complete elephant, “Fig.1”. This makes the task of reconstructing a full Elephas tiliensis typical skeleton especially hard, without the assistance of modern digital methods and technologies.

Figure 1. A partial fossil-reconstruction of an E. tiliensis in 1994



2.2. Palaeontological team of the project - NKUA The palaeontology leg of the research team of this paper’s work consists of NKUA’s Professor G. Theodorou and his group of associates, who have been studying the findings of the Charkadio Cave and particularly the Elephas tiliensis for over 30 years. Professor Theodorou is the current Director of the Palaeontology and Geology Museum of NKUA and possesses valuable in situ experience, especially on prehistoric elephants [4]. 2.3. Engineering team of the project - NTUA The engineering leg of the research team consists of NTUA’s Professor C. Provatidis (Director) and researchers (faculty members, staff and research associates) of the Rapid Prototyping and Tooling Laboratory of NTUA’s School of ME. They all bear long experience in Engineering Analysis (FEA, FEM), 3D CAD Modeling, Reverse Engineering and Rapid Prototyping/Additive Manufacturing, with several relevant research work and applications conducted over the last two decades, [5, 6, 7, 8]. It is also to be noted that NTUA’s RP&RT Laboratory is one of the very first AM facilities in Greece, established in 1996. 2.4. Thales - MIS380135 Research Project The NKUA palaeontologists and NTUA engineers were originally brought together in 2008 with the idea to share experience and assist each other into the reconstruction of an anatomically complete Elephas tiliensis skeleton, both digitally, as well as, in the form of an actual full scale tangible exhibit. In order to cope with the inherent difficulties of such a task, especially the large number (more than 15,000) and the diversity of the fossils, but also the lack of some anatomically critical ones (e.g. skull bones) the project had to further be supported by experts from the Faculty of Biology of NKUA and the Faculty of Orthopedics from the University of Patras. The fully assembled research team submitted a proposal for the whole project to be funded within the frame of the Greece & EU funded Operational Program "Education and Lifelong Learning" of the National Strategic Reference Framework (NSRF) - Research Funding Program: Thales, Investing in knowledge society through the European Social Fund, in 2009. In the end of 2011 a grant was finally approved for the task to be realized (Thales Project MIS380135) and the team commenced to work. 3. BIOLOGICAL DATA PREPARATION More of thirty years of excavations in Charkadio, have produced about 15,000 bone fossils in total, identified to belong to about 77 individuals of E. tiliensis. Each skeletal element and in the current study long bones, were ranging in four major age classes according to osteological characteristics: (1) Infants, (2) Juveniles, (3) Sub-adults and (4) Adults. These fossils differ in terms of their morphology and size, with some of them broken, partial/fragmental or worn by external influences and others complete and in good overall condition. All are kept in the premises of the Palaeontology and Geology Museum at NKUA. It was naturally impossible to digitize all of these fossil sets. So a careful selection had to be made by the

63


palaeontologists for scanning-digitizing, according to a series of biometric criteria, in order for each typical bone of the skeleton (if available) to have a scanned set of all the above four age groups, in the best possible quality of the fossils kept. The skeletal elements finally selected were the ones best preserved, without cracks and retaining the diagnostic characteristic of their age (ossified or unossified ends, presence or absence of epiphyseal line). Utilizing the principles of taphonomy/stratigraphy, ontogenic observations, allometry and sexual dimorphism the palaeontologists would then be able to assign specific values for systematically defined, [9, 10], dimensional parameters on the digital CAD model of each bone prepared by the engineers, thus ultimately defining a properly formed, dimensioned and scaled typical skeleton of an average adult E. tiliensis individual. A brief definition of these principles and description of how they were methodically used for the presented project is given below. Taphonomy-Stratigraphy: Taphonomy refers to the way fossil remains (bones, skeletons etc.) become fossilized and are later found lying when recovered from their burial site. Stratigraphy deals with the different depths of distinct fossil-hosting rock layers of specific geological eras in an excavation site. During the first 10 years of the Charkadio excavations (1970s), emphasis was given to the stratigraphic analysis of the site. Two principal fossiliferous horizons were recognized. A lower strata, dated to 140,000 years BP, where fossils of the deer Dama dama dominate and an upper strata, with a height of 3.5m, dominated by dwarf elephant fossils. The elephantid fossil–richest stratum of the cave has been dated to 18,000 years BP, a period of low sea level. After the 1990’s, more detailed and taphonomic-oriented excavation techniques, with a standard grid system, were applied in Charkadio. Following this shift in methodology, the identification of several elephantid remains with anatomical association between them was made possible [3]. Approximately, eleven (11) different cases of such anatomically associated skeletal elements were recorded. Most are either forelimb or hindlimb elements; three cases of vertebrae in anatomical association have also been recorded. All of the above skeletal elements correspond to adult specimens. Ontogenic observations: For the purposes of the presented project, study of different ontogenetic stages of the skeletal elements has been done, by separating them into age groups (infant, adult, sub-adult, adult). As generally known, two ossifying centers exist in long bones, one at the proximal epiphysis and another at the distal epiphysis, that control the increase in the length of the long bones. An ossifying centre at the shaft (approx. the bone’s middle section) controls the bone’s width increase. Longitudinal growth dramatically decelerates with the fusion of the epiphyses. However, the ossifying centre at the shaft remains active after the fusion of the epiphyses, further increasing shaft width (respectively the bone’s width). In the majority of the long bones studied, it was observed that in different ontogenetic stages the longitudinal length and the width of the shaft were developed in different patterns according to the growth rate of the elephants.



Allometry: Highly correlated measurements (mainly maximal length and diaphysis length) of each separate long bone were chosen in order to create mathematical equations to estimate all the missing values appropriate for the project’s study. Statistical analysis through SPSS and the equation of simple allometry were utilized in order to estimate the size of fossils found in anatomical association [11, 12, 13]. These methods mentioned, describe allometric relationships of the measurements studied. Furthermore, dimensions of missing skeletal parts, fragmented skeletal material or badly preserved bones could be estimated, so as to complete the majority of the individuals’ measurements. The allometric growth pattern observed could also be influenced by sexual dimorphism. Long bones of extant male elephants are both longer and more robust than those of females, due to delayed epiphyseal fusion. With such a correlation, an estimation of dimensions and analogies of a typical dwarf elephant’s fore and hindlimb was made possible. Multiple vertebra samples have also been studied with the same methodology. So, during the first-launch period of the project, emphasis was given into the investigation of the appropriate techniques and process steps for the re-modeling (digital and actual) of the hind (i.e. back) limbs of the animal and some vertebral and thoracic elements. The exact skeletal elements initially selected for 3D CAD modeling and evaluation are included in Table1. In the case of symmetric bones (left and right hind limbs), only one would be scanned and modeled, with the other being simply a “mirror part” of it. TABLE I. SKELETAL ELEMENTS THE PROJECT’S INITIAL PHASE Description

SELECTED FOR MODELING DURING

Skeletal Element Project Code

Type

Quantity

Astragalus

472T

Left & Right

2

Tibia

T.3233

Left & Right

2

Fibula

T.158_98

Left & Right

2

Femur

T.97_52

Left & Right

2

Thoracic Vertebra

T.4682

Central

1

Rib

T.223_98

Left & Right

2

Cervical Vertebra

T.1551_98

Central

1

4. DIGITAL DATA PREPARATION & 3D CAD MODELING The major steps followed and evaluated for the Digital Data – 3D CAD models creation through the first set of the project’s fossils were: 

Digital data sets creation from Computer Tomography (CT)



Digital data sets creation from Laser Scanning



Surface-based 3D CAD models creation from the above data sets



Comparison of the integrity and quality of the 3D CAD models produced, according to their origin.

64


4.1. CT Scanning and data processing For the CT scans, the research group was kindly allowed to use a Philips Brilliance CT 64-slice tomograph, of the Konstantopouleio “Agia Olga” Neas Ionias General Hospital. As CT scan data can at any time be isolated and separately handled, for the sake of time compression and because of the limited availability of the tomograph, several bone fossil groups were simultaneously CT-scanned. For CT scanning, image size dimension was set to 512px × 512px, the pixel size was 0.793mm and the slice increment 0.399mm. In case of elongated bones the increment was doubled. Raw scan data were exported according to the Digital Imaging and Communications in Medicine (DICOM) CT native file format, [14]. A typical CT scan slice of a group of fossils can be seen in “Fig.2a”. The DICOM files of the CT scans were imported to the Materialise Mimics commercial software (licensed to NTUA) for processing. From the successive slices of each scan set, Mimics is able to assemble a 3D Point Cloud (X, Y, Z coordinates) of the scanned objects, enabling their further processing into 3D CAD models. As Mimics is mainly based in image processing, a thorough and careful boundary definition of the objects in each slice through “masks” is required, as shown in “Fig.2b”, in order to come up with an acceptable result in the 3D point clouds derived, “Fig.3”.

(a)

(b)

Figure 2. (a) CT Slice of group of fossils, (b) Boundary Definition through “masking”


4.2. Laser Scanning For comparison purposes, selected fossils were also individually scanned with a theoretically more accurate – but also more tedious – method, non-contact laser scanning, using NTUA’s FARO Platinum Laser Arm, a method that claims a scanning accuracy of approx. 50μm, “Fig.4”. Specific parameters set for laser scanning with the FARO Arm were: Filter angle = 75 degrees, Scan rate = 1/1, scan density = 1/1, exposure = 18. These scans resulted in point clouds of 1 to 4.5 million points per fossil, depending on their size and complexity. Despite the large size of the resulting files (in the order of hundreds of MBs), these “rich” sets of raw data would not be a problem, as in a later stage only about a 10% of each would be used for actual CAD modeling.

Figure 4. Laser Scanning with Faro Arm

4.3. Surface-based 3D CAD Modelling 3D Point clouds, either originating from CT scans and Mimics, or from laser scans with the FARO Arm, were in the next process step imported to another ME-oriented commercial software licensed to NTUA, Raindrop Geomagic Studio. Geomagic Studio is generally used mainly for Reverse Engineering (RE) applications, to convert 3D Point clouds, first into triangular meshes and finally into closed surface volumes (Solids). To effectively do so, several filtering options are available in the software, for the removal of points obviously alien to the scanned object (outliers), points falsely captured from deflections and other erroneous or redundant data. A proper lean and narrow data set can then serve, first for triangulation and then for surface fitting/patching, towards the end closed volume of a solid. The result can be exported to practically all existing 3D CAD software packages, via neutral standard exchange protocols, such as STEP and IGES. For the fossil data sets previously prepared (both from CT scans and laser scans), all these features of Geomagic Studio have been used and the process successfully yielded smooth and relatively accurate 3D solids from both methods, as shown in “Fig. 5”.

Figure 3. Point cloud generation from CT Slices


Figure 5. Geomagic Studio 3D Model of a Tibia bone.

65



3D and 2D Evaluation tools also exist in the Geomagic Studio environment, which verified that both CT scanning and laser scanning derive comparable 3D models, as shown in “Fig.6”, with accuracy deviations well acceptable for macroscopic observation of the skeleton under construction (where differences smaller than 1 mm would practically be negligible).

Figure 8. Scaled and oriented 3D CAD model of a Tibia in Solidworks

5. AM FABRICATION OF THE SKELETAL ELEMENTS The prerequisite for any AM/RP fabrication is the existence of a 3D Solid Model. With all 3D CAD models of the skeletal elements in Table 1 now available and finalized, the fabrication of their tangible replicas was made feasible. Figure 6. CT & Laser Scan 3D Model Comparison of Astragalus fossil (Deviations in mm)

4.4. Final 3D CAD Models’ dimensioning As mentioned already, the palaeontologists assisted by the group’s biologists, through measurements of actual fossils, “Fig.7”, taphonomy/stratigraphy, allometry, ontogenic observations, sexual dimorphism and proper mathematical/ statistical modeling currently under development within the MIS380135 project, are the ones who decide the final form, dimensions and analogies of the aforementioned 3D CAD Models developed. As this process of dimensional adjustment and finalization is utilized on specifically defined axes and planes on the bone geometry, [9, 10], it was convenient for the team’s designers to import these models in pure 3D CAD Software available at NTUA (i.e. 3DS Solidworks), where part-specific Coordinate Systems, Planes, Views etc., could be easily defined and manipulated for each and every bone already modeled. By applying the proper scaling factors in these planes or axes (uniformly or with a proper distribution) the exact final size and shape for the bone replicas was achieved, “Fig.8”, enabling the next stage of the overall process to take place, that be Additive Manufacturing.

Figure 7. Measurements on right (R) tibias of infant, juvenile and adult E.tiliensis (anterior & medial aspects)


5.1. Available AM equipment NTUA’s RP&RT Laboratory hosts two AM machines: (1) A rather old but fully operational Helisys LOM1015 machine, based on the Laminated Object Manufacturing AM technology, working with 0,1mm thick paper sheet as the raw material and with a building envelope of 380x250x360mm and (2) a small scale Fused Deposition Modeling machine, the Stratasys uPrint, that deposits ABS thermoplastic material in 0.254mmthick slices in an envelope of 200x150x150mm. Although both the above equipment are not considered today to be the pinnacle of AM in terms of speed, resolution or accuracy, nevertheless the dimensional stability, durability and cost-effectiveness of the parts they produce, made them highly suitable for the goals of the E. tiliensis project. The long experience of the Laboratory’s personnel on both technologies was also expected to optimize their utilization and maximize the benefit from their use, [15]. 5.2. AM technological limitations – Parts Distribution The LOM and FDM technologies differ fundamentally. The first begins with a solid material in the form of paper sheet rolls. It therefore only has to define the area of a fabricated part in each layer constructed and the rest is scrapped. The apparent benefits of the relatively low cost raw material (~500 € per 18lt of usable paper) and rather fast build rate for bulky parts, are to a certain extent shaded and limited by the high waste percentage (averagely measured to be 60-80%) and limitations in the constructed geometry’s complexity. The latter (entry level FDM) has less geometrical limitations for the parts, but is highly volumetric and with a build capacity limited for the uPrint machine typically to 425cm3 (quantity contained in a full filament roll). Therefore a part’s volume and overall dimensions heavily affect the total build times and cost, even the ability to be directly fabricated with the specific FDM machine. To compensate for that, the uPrint machine offers the option for sparse internal volume material deposition, which was adopted for this project. So the main criteria finally examined for the AM distribution and fabrication of the skeletal elements in Table 1, were identified and prioritized to the following:

66


size, compared to their full detail originals, were finally extracted for the appropriate AM fabrications to follow.



Part Volume (in mm3)



Part overall dimensions (in mm)



Part complexity and density (Part V/ Envelope V)



Part estimated build time



Part estimated build cost

Of course, the ability to segment and simultaneously “pack” several parts for a feasible and more cost effective fabrication in the same build on any of the available machines was taken into account, especially for the LOM1015. Based on the above, the skeletal elements were decided for building as shown in Table 2. TABLE II. Description

SKELETAL ELEMENTS AM DISTRIBUTION Skeletal Element Project Code

AM Machine

Segmented

Packed

Astragalus

472T

uPrint

No

No

Tibia (L&R)

T.3233

LOM

No

Yes

T.158_98

LOM

Yes

Yes

T.97_52

LOM

Yes

Yes

T.4682

LOM

Yes

Yes

Rib (L&R)

T.223_98

uPrint

No

No

Cervical Vertebra

T.1551_98

uPrint

Yes

Fibula (L&R) Femur (L&R) Thoracic Vertebra

a

Yes

a. Only the lower segments

5.3. STL file generation and manipulation The finalized original 3D CAD models existed in 3DS Solidworks format. Although, an STL file extraction from Solidworks is simple and straightforward, it was decided to export neutral STEP files of the skeletal elements, for import directly to NTUA-licensed dedicated commercial STL repair & manipulation software Materialise Magics RP. This way the full detail of the 3D CAD model could be preserved during transfer to Magics RP. According to their AM destination and the accuracy required, all bones could then be properly and efficiently “demoted”, smoothed, segmented and packed for LOM or FDM fabrication, via the STL files extracted from within Magics RP. The number of triangles contained in an STL file is directly proportionate to the size of that file. Any detail greater than the one required by the application is redundant and should be eliminated for practical (file storage and transfer capacity) and for specific essential technical reasons. Especially in the case of NTUA’s LOM1015 machine (1996 dated model), STL files must be kept under 30Mbytes (or approx.800,000 triangles), as its outdated operating Personal Computer (Intel Pentium, 4oo MB HDD) might significantly lag, or even crash over that size, thus discarding a build. For the FDM uPrint machine, such technical limitations are absent; nevertheless handy, downsized STL files were also attempted and achieved. Overall, by using the Magics RP software capabilities (triangle reduction iterations), STL files of averagely 1/20 triangle number and file



5.4. STL file segmentation, orientation and packing Having created the STL files of the bones of Table 1, according to the fabrication strategy of Table 2, once again in the Magics RP environment, segmentation of the files had to be accomplished, either for dimensional limitation reasons (e.g. Femur destined for LOM, Cervical Vertebra for uPrint) or for increased productivity and cost-effectiveness of each AM process followed (Fibula & Thoracic Vertebra for LOM). The key points in this process were, on one hand to split parts in a way that would ensure an easy and accurate reassembly and gluing of the final prototype bone replicas, leaving no residual degrees of freedom, and on the other hand ensure the maximum exploitation – waste minimization of the raw material after part-packing, especially for the LOM process (with Y-axis, the LOM paper roll width being the most critical parameter). Orientation of parts in the building platform also had to be addressed before any segmentation, in a manner that would ensure: (1) similar or exact alignment of parts to be later rejoined (for geometrical consistency and optical similarity), (2) minimum height (Z-dimension) for the sake of fast build times (mainly for LOM), (3) minimum “stair-stepping” effect on the parts, (4) minimum supporting material for the FDM produced parts and (5) feasible scrap removal for the LOM parts. After positioning parts in a Z-minimum position, splitting should then be performed in a way that would allow a fully defined, unidirectional sliding reassembly in the Z-axis direction for LOM, or in any direction for FDM. For the actual part splitting, several different configurations were considered and evaluated, but the ones preferred were: 

A combination of “Rounded Dove Tail” and Inclined plane for the LOM fabricated bones. This was achieved by two successive Cut operations (Teeth & Polyline), followed by two Boolean unifications in Magics RP SW. Among other advantages, the dove-tail shape allows for gravitational stresses during a suspension to be assigned to the own mass of the prototypes, not only to the glue, as possibly will be the case with the consecutive bone assembly of the limbs on the final skeleton exhibit.



A male-female boss & pocket with a 0.3mm clearance (to ensure a loose fit and proper room for glue) for the FDM fabricated Cervical Vertebra. This was easily done with only one Advanced Cut operation, again in Magics RP.

Graphic details of both segmentation techniques followed are illustrated in “Fig. 9a&b”, while a properly packed platform for LOM, containing the two lower segments (Left & Right) of the Femur is shown in Fig.10. The combination of proper orientation, part segmentation and packing allowed for reduced build times to be achieved, a waste reduction of approx. 20% for the LOM parts and a support material reduction of the same order for the FDM uPrint machine.

67


(a) Figure 9.


(b)

(a) “Dove tail – Inclined plane” spltting for LOM, (b) Boss-Pocket Segmentation for uPrint FDM

Figure 11. Side by side comparison of Tibia fossil and modified LOM model.

Figure 10. Packing of the symmetric (L & R) lower segments of the Femur for LOM fabrication.

5.5. Prototypes fabrication and final assembly Actual AM fabrication of the parts of Table 1, required:

Figure 12. Left and Right Hind Limbs AM Assemblies

(1) Six (6) runs of the LOM1015 machine, with a total of 124 machine operating hours, plus another 15 hours for pre- and post-processing and part separation (decubing). (2) Five (5) runs of the uPrint machine, with a total of 22 machine operating hours, plus another 7 hours for pre- and post-processing and part separation (from supports). The strictly AM-related total cost for all 12 bones of Table 1, fabricated by both AM technologies according to the strategy of Table 2, was estimated around 1.500 €, while they required about 15 full working days (as AM equipment also work overnight). This of course does not include hundreds to thousands man-hours and related cost for all appropriate previous research, analysis, scanning and modeling steps required.

Figure 13. LOM Model of Thoracic Vertebra

The reassembly of the segmented fabricated parts into single bones, as well as, of the bones themselves constituting the hind limbs proved to be immediate and accurate, leading to quite an impressive result, very similar to actual fossils even in terms of color and feel (for LOM), especially for the palaeontologists of the research team, who had no prior experience and interaction with Rapid Prototyping models. Selected images of the AM-made skeletal elements are shown in Fig.11, 12, 13 & 14.


Figure 14. Assembled FDM Cervical Vertebra

68


6. DISCUSSION & CONCLUSIONS Progress made so far within the Thales - MIS380135 project is truly impressive and especially for the nonengineering members of the research team above expectation. All stages followed have produced valuable techniques, practices and results and have identified specific effective and individual process steps that will be streamlined and systematically further exploited until the completion of the project and the final outcome, a complete E. tiliensis skeleton. The parts completed in the project’s initial phase are very important, as they were both large in dimensions and volume and also very typical of the difficulties the team had to face and overcome in every individual stage and discipline. For the palaeontologists of the research team, the interdisciplinary methodology being developed within the project itself, the time required, the corresponding estimated cost, the fidelity and accuracy produced and the overall process repeatability are an unprecedented breakthrough. They have nothing to do with conventional methods they have been previously following in similar palaeontological applications, thus creating a lot of optimism for the rest of the project and for their future applications. Beyond the specific project, this future could be even more promising, if contemporary, more advanced and larger AM machines are at some point to be used by the team for similar purposes (e.g. MCOR A4 paper-based Selective Deposition Lamination, large FDM machines, etc.) As for the parts made so far with the existing available LOM & small scale FDM equipment, they can be easily painted, finished, drilled, glued and assembled to the wishes of the palaeontologists for the final skeleton exhibit. But they can also very well serve, as master patterns for the skeleton’s reproduction in several copies, if needed, throughout well known established Soft/Silicone Rapid Tooling Techniques, [16]. The Digital 3D CAD assembly of the skeleton can also prove of major importance in the near future, as it will enable the easy proliferation, publicity and studying of the E. tiliensis among the palaeontology scholars and researchers worldwide, without having to visit Tilos and Charkadio, in the form of digital data and multimedia, even distantly over the Web. But it can also find other fruitful exploitation areas, such as the production of digital multimedia (movies and animations) for educational or recreational purposes, of full recreations of the animal with soft tissue, organs, skin and hair (digitally and physically), or even by just supporting the production of souvenirs for the visitors of the island of Tilos.


General Hospital, for their kind permission and technical assistance in using the Philips Brilliance CT 64-slice tomography of the hospital for the purposes of the project MIS380135. REFERENCES [1] [2]

[3] [4]

[5] [6] [7]

[8]

[9] [10]

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

Barnatt, C., 2013, 3D Printing: The next industrial revolution, Createspace Independent. Pantazis, G., Lambrou, E., Polydoras, S., & Gotsis, V., 2013, ‘3D Digital Terrestrial Model Creation Using Image Assisted Total Station and Rapid Prototyping Technology’, International Journal of Heritage in the Digital Era, vol.2, no.2, pp.245-262. Theodorou, G., Symeonidis, N., Stathopoulou, E., 2007, ‘Elephas tiliensis n. sp. from Tilos island (Dodecanese, Greece)’, Hellenic Journal of Geosciences,vol.42, pp.19-32. Theodorou, G. E., Roussiakis, S. J., Athanassiou, A., & Filippidi, A., 2010. ‘Mammalian remains from a new site near the classical locality of Pikermi (Attica, Greece)’, Scientific Annals, School of Geology, Aristotle University of Thessaloniki, sp.vol.99, pp.109-119. Provatidis, C. G., 2012, ‘Two-dimensional elastostatic analysis using Coons-Gordon interpolation’, Meccanica, vol.47, no.4, pp.951-967. Provatidis, C. G., 2013, ‘Simplified biomechanics for a possible explanation of the ancient Greek long jump using halteres’, Univers J. Eng. Sci., vol.1, no.1, pp.5-16. Polydoras, S., Sfantsikopoulos, M., & Provatidis, C., 2011, ‘Rational Embracing of Modern Prototyping Capable Design Technologies into the Tools Pool of Product Design Teams’, ISRN Mechanical Engineering, vol.2011, Article ID 739892, 12 pages Kaisarlis, G., Polydoras, S., Provatidis, C., 2011, ‘Evaluation of Geometrical Uncertainty Factors during Integrated Utilization of Reverse Engineering and Rapid Prototyping Technologies’, Proceedings of (DET 2011) 7th International Conference on Digital Enterprise Technology, Athens, Greece, September, pp.532-541. Smuts, M., & Bezuidenhout, A. J., 1994, ‘Osteology of the pelvic limb of the African elephant (Loxodonta africana)’, Onderstepoort Journal of Veterinary Research, vol.61, pp.51-66. Bezuidenhout, A. J., & Seegers, C. D., 1996, ‘The osteology of the African elephant (Loxodonta africana): vertebral column, ribs and sternum’, Onderstepoort Journal of Veterinary Research, vol.63, pp.131-147. Gould, S. J., 1966, Allometry and size in Ontogeny and Phylogeny, Biological Reviews, vol.41, pp.587-640. Jolicoeur, P., 1963, The multivariate generalization of the allometry equation, Biometrics, vol.19, no.3, pp.497-499. Huxley, J. S., 1932, Problems of relative growth, Methuen: London. Reprinted 1972, Dover Publications: New York. Mildenberger, P., Eichelberg, M., & Martin, E., 2002, ‘Introduction to the DICOM standard’, European Radiology, vol.12, no.4, pp.920–927. doi:10.1007/s003300101100 Polydoras, S., 2013, ‘Contribution to time and cost compression in New Product Development through Rapid Prototyping & Tooling technologies’, PhD Thesis, NTUA, Athens, Greece. Chua, C. K., Leong, K. F., & Lim, C. S., 2010, Rapid Prototyping: Principles and Applications. World Scientific.

ACKNOWLEDGMENT The authors would like to thank the administration and medical staff of the Konstantopouleio “Agia Olga” Neas Ionias


69



Customizable Personal Manufacturing Mathias Bratl, CAMPUS 02 UAS Automation Technology and Marshall Plan Foundation, Graz, Austria, [email protected]

Abstract—Past developments in the area of personal manufacturing, such as online platforms of CAD models, assist customers in printing their favorite 3D objects. Nevertheless, there are still many barriers, since customers do not have easy possibilities to customize 3D models. In order to fix these issues, this work deals with the development of a concept for customizable personal manufacturing, with a 3D-customization tool as the central element. The major output of this work is the software 3DCustomizer, which acts as a common product individualization platform for designers and customers. Designers are preparing their 3D models by using the software 3DCustomizer in combination with a CAD software. Many customization features can be integrated into a 3D model, such as signatures on 3D surfaces or formula-driven component dimensions. This whole information is stored in a configuration file. Customers are working on a simplified software interface. They are loading this configuration file and are then able to change the parameters, which were enabled by the designer. The 3D model will be updated by the CAD software in the background. Specialized converters are automatically converting the 3D geometry in several steps into a code for the 3D printing system. Prior to the software development, the theoretical foundations in the areas of CAD, 3D data conversion, additive manufacturing (AM) and programming interfaces have been developed. Some of these findings were determined by experiments. Different AM technologies were subjected to a comprehensive technology comparison. In addition, a fused deposition modeling (FDM) system has been tested with the aim to get the boundary conditions for the software development. Keywords-additive manufacturing; rapid prototyping; 3D printing; 3D CAD; mass customization; individualization

industry exceeded all expectations with new online tools for customers to individualize their favorite shoes and order them directly [2]. In 2013, Motorola presented their “Moto Maker”, where consumers are able to customize different properties of the smartphone, whereas 504 different variants are possible. In this work, it will be tried to blow up the limits of personalization options for customers. A completely new approach of decentralized product generation by using 3D printing systems will be developed, which runs on top of a new software platform, where customers and designers work together to customize products [6]. 2. INITIAL SITUATION AND TARGETS Additive manufacturing has already been established in the industry. It is especially important, when there is a need of prototypes with high complexity and when this need has to be covered in an easy and fast way without the possibility to realize it with conventional production systems. Different technologies in this innovative area of production have already been developed. There are AM technologies to print different types of synthetics, metal and ceramic-plaster mixtures. It is even possible to combine different materials within one print process or to print full-colored 3D objects [1]. Personal manufacturing focusses on inexpensive AM devices for customers. Past developments, like online platforms for CAD models are leading the way into a decentralized personal production. Nevertheless, there are still several barriers. On the one hand, customers are missing simple tools to individualize 3D models. On the other hand, the work process with inexpensive FDM devices is still very time-intensive.

1. INTRODUCTION “Consumers know what works best in everyday products and given their own manufacturing tools, can prototype it. […] Companies that ignore their customers’ talent for designing and making profitable, innovative products will lumber their way to obsolescence.” [3] Humans are striving for individuality. The external appearance, behavior, personal goals – these are attempts to stand out from the crowd. Nevertheless, in our consumer society material things are used for self-presentation. Several product areas have already taken advantage of this quest for individuality by breaking new ground in product commercialization. Various products are already based on modular systems, where customers can create their desired product from a pool of components. For example, several companies in the sports Marshall Plan foundation


Figure 1. Basic customization concept

This work deals with the elaboration of a new software platform, which eliminates the majority of these issues in the area of Personal manufacturing. Figure 1 shows the main concept, which illustrates an interconnection between different

CAMPUS 02 University of Applied Sciences

70



special areas. The software 3DCustomizer has to switch between different user groups, access 3D models within a CAD software and convert these into a readable code for a low-cost FDM system.

The results of the evaluation matrix have constituted a solid basis for further research, since it has revealed that low-cost systems in combination with 3D-customization possibilities provide a vast advantage for customers.

3. MATERIALS AND METHODS

Furthermore, a low-cost FDM system, Makerbot Replicator I [4], was comprehensively evaluated in several experiment stages. The findings concerning work processes, handling, automation possibilities and existing software solutions are providing an important base for the software development.

3.1. Mathematical relation of 3D models The heart of this new platform is the customizable 3D model. A 3D model, which was designed through a CAD software, is mathematically described through vectors and is easily parameterizable within this environment [5]. By changing one dimension or value in one part, it would influence all other related parts, respectively dimensions. Moreover, it is possible to make attributes dependable from each other and to control them via formulas. For example, the CAD software SolidWorks allows to access functions and variables via a programming interface. 3.2. Data preprocessing In order to produce this 3D model with a FDM machine, the data has to be converted in several steps, which is illustrated in figure 2. The mathematically described model is converted into a 3D mesh within the CAD software. A slicing tool is now calculating XY travel paths (g-code) for the FDM machine out of the 3D mesh [1]. After this process, the g-code needs to be converted into a machine-specified code.

4. PRACTICAL IMPLEMENTATION 4.1. Software concept After the elaboration of the theoretical fundament, the accomplishing of technology comparisons and experiments, a software concept was able to be developed, which illustrates all the important interconnections. The software builds upon existing data - a designer needs to prepare at least two assemblies, in order to be able to successfully connect the model to 3DCustomizer: 

mathematically related model assembly



linked print-layout assembly

Both assemblies consist of the same parts. However, the linked print-layout assembly is already arranged in order to be printable for a FDM system. As illustrated in figure 4, 3DCustomizer takes advantage of the findings described in chapter 3.1. It uses the interconnections and relations inside the CAD system, in order to perform complex 3D model adaptions. By adapting one dimension, all the related parts will change, as well as the linked print-layout will get rearranged and updated.

Figure 2. 3D data conversion

3.3. Experiments and comparisons If a customer has customized a 3D model, this virtual object will be printed with a FDM machine. Therefore, a comprehensive technology comparison should clarify the relevance of this manufacturing method. This technology comparison also includes the evaluation of the manufacturing accuracy, which is measured through structured light scanners. A target/actual-comparison of the 3D data provides a clear and objective statement about the accuracy and surface quality. This data is illustrated in a benchmark matrix (figure 3) with all the other findings.

Figure 4. Mathematical relation within the CAD model

Figure 3. 3D data conversion


4.2. Modularity and expandability This subchapter clarifies the attempt to keep the core of the software modular and expandable through array-based coding. Array-based coding means that all necessary code fragments are wrapped into arrays in order to be able to access every label, textbox, checkbox, value, etc. effectively through loops [7].

71


Otherwise, it would have been impossible to provide a modular and expandable software tool. Array-based coding keeps the core of the software as compact and efficient as possible. Furthermore, other features like saving and loading of configuration files need to access such arrays, as otherwise it would cause the code to burst at the seams. 4.3. Software functions Gradually, the main functions of the software, as illustrated in figure 5, have been integrated. The access to the CAD software SolidWorks was integrated as a first software building block. This function includes, among other things, the changing of dimensions and signatures, as well as special features, such as the automatic adaptions of 3D geometries to smartphones. This special feature was implemented, since the sample 3D model of this software development dealt with a virtual reality (VR) glasses for smartphones. “Smartphone VR glasses” is a project, which emerged out of the online community and makes it possible to use your smartphone in combination with selfprinted parts and lenses as a 3D gaming device [8]. Now, this smartphone assistant feature in the software 3DCustomizer allows users to automatically adapt the 3D model to their smartphone model. However, it is also quite easy for designers to integrate this feature into their 3D model since it is all about linking of values and variables. All these CAD functions are collected in the area “customization features”, which will be continuously extended.

Figure 5. Schematic overwiev of the main software functions

In order to generate the code for the FDM machine out of the 3D models, a converting function was implemented, which automatically acts in the background. The XML save management provides a defined interface between the different user groups. All the information that was set by the designer is stored in this XML configuration file. 4.4. Interface between user groups Personal manufacturing has to be distinguished into certain groups of users. There are several reasons for this, especially when it comes to an integration of customers without technical competencies. Thus, customers should…: 

… have an interface which is as easy to use as possible,



… not worry about parameters of the g-code,



… not worry about aligning and scaling of STL files in the building volume, and





… not worry about using support material or the observance of certain overhang angles.

These are work packages within the responsibility of the designers. In order to be able to distinguish in these two groups of people, designers and customers, clear and standardized interfaces have to be established. The fundamental idea is to make the customization features and the blocks inside the features activatable. This provides the designer with possibilities to choose their favorite features and the amount of blocks. Consequently, designers can individually shape the software for the 3D model. This activatable function enables a possibility to examine which properties have to be saved for the configuration files. However, the main advantage of this function is to be able to easily switch between designer and customer mode. All customization features and blocks which are not used, respectively activated, by the designer will not be shown in the customer mode. For example, a designer is able to choose between a great amount of customization features, which can be activated, edited and linked with a certain value of the 3D model. Now, if the customer opens the software with the configuration file of the designer, only the customization features, which were activated by the designer are visible. Furthermore, the interface of the software is reduced to a minimum, in order to reach a maximum of usability of the customer. 4.5. Converting of 3D models As described in chapter 3.2 the 3D model has to be converted so that the FDM machine is able to process these data. Therefore, 3DCustomizer accesses two conversion tools via different programming interfaces. Slic3r is a slicing tool, especially developed for AM purposes. The main goal of this tool is to create g-code files out of STL files. Slic3r offers accessibility via command-line interface” (CLI). This means, commands can be send to this tool, when wrapped into a certain code structure. The main usage of this tool is to create a g-code file, out of the converted printlayout. AM low-cost printers are different when it comes to gcode-files. Each printer needs special settings, like position of homing, extruder speeds, etc. In order to consider all these specific settings, Slic3r is able to deal with so-called initialization files (ini files), which stores all the information of the settings. Consequently, different machines could also be used with 3DCustomizer, unless initialization files are prepared. When dealing with a Makerbot Replicator as a printing device, a possibility for printing PC-autarkic exists. This means that g-code-files can be converted into machinespecified code (X3G or S3G), saved on a SD card and printed directly from there without the need of communication between a PC and the printer. Usually, the software on a PC translates each g-code command into machine specified code and sends it to the machine sequentially. In order to offer a CL interface, a plug-in was developed for ReplicatorG especially for this purpose. It is now possible to access the conversion feature of ReplicatorG with certain commands, without the need to load the graphical user interface. Thus, ReplicatorG is running in the background during the conversion process.

72


5. RESULTS The software 3DCustomizer performs satisfyingly in initial tests and experiments with 3D models. The current customization features cover the main purposes and desires of the designers. Especially, the powerful features (“design” and “smartphone assistant”, seen in figure 6) show the big potential of the 3DCustomizer. Gradually, new features will be added.


Also, the concept of the configuration files has already been established in initial experiments. Switching between different configurations runs very effortlessly. As a programmer, it is also quite easy and time-saving to add customization features or feature blocks, since everything is coded array-based and modular. 6. PROSPECTS Several improvements and new experiments in near future have to be done as well. Moreover, several experiments with gcode initialization files have to be done, since certain g-code settings must be adjusted as well as possible in order to get best printing results. Nevertheless, through the integration of new customization features, especially the powerful and focused ones, like smartphone assistant, the improvement of this tool will not stop in the future. If the first testing and coding period is over, a handful of designers will be invited to use 3DCustomizer for customization of their models. With their input and help, improvements can be realized as well as new features added. Thus, this next stage will be the first step towards crowdsourced customizable personal manufacturing. ACKNOWLEDGMENT

Figure 6. Excerpt from the graphical user interface

In addition, the rules for designers are comprehensible and clear. So, there are not that many boundary conditions for designers. Basically, they are able to design their models in the way they are used to, which is one of the biggest advantages within this new concept. Especially with the knowledge about external references between 3D geometries inside CAD systems, this leads to 3D models with high complexity and great possibilities for individualization, as illustrated in figure 7.

I would like to thank several persons and institutions who assisted me in personal and academic purposes during the elaboration of this project as well as the research stay in San Diego. Firstly, I am conveying my thanks to Marshall Plan Foundation, who have given me the opportunity of a scholarship. Without this great assistance, the research stay and project would not have been possible. Furthermore, I would also like to appreciate the great academic and personal assistance of the University of Applied Sciences CAMPUS 02 in Graz and therefore I would like to especially thank Univ.-Prof. Dipl.-Ing. Dr. techn. Franz Haas and FH-Prof. Dipl.-Ing. Dr. techn. Udo Traussnigg, who provided me with many opportunities to improve my technical and personal skills and assisted me a lot with the preparation of my research stay and with the elaboration of my thesis. REFERENCES [1] [2] [3] [4] [5] [6] [7] [8]

Chua, C. K.; Leong, K. F. & Lim, Chu-Sing D., 2010, Rapid Prototyping - Principles and Applications, 3rd ed., World Scientific Publishing Co. Pte. Ltd., Singapore. Anderson, C., 2012, Makers: The new industrial Revolution, Hanser, New York. Lipson, H.; Kurmann, M., 2010, Factory @ Home: The Emerging Economy of Personal Fabrication, pp. 55-60. Pettis, B.; Kaziunas France, A.; Shergill, J., 2013, Getting Started with MakerBot, O'Reilly Media, Inc., New York. Scheuermann, G. 2013, Grundlagen u. Methodik in zahlreichen Konstruktionsbeispielen, 4th ed., Hanser, Munich. Sloane, P. 2011, A Guide to Open Innovation and Crowdsourcing: Expert tips and advice, London. Doberenz, W.; Gewinnus, T., 2013, Visual C# 2012, Hanser, Munich. Jansen, J.-K., 2013, Android Smartphone als Virtual-Reality-Brille, in: c't, Heft 15/2013, pp. 64-65.

Figure 7. Example of assemblies, controlled by 3DCustomizer


73



Designing with 3D Printed Textiles A case study of Material Driven Design Kirsten Lussenburg, Industrial Design Engineering, TU Delft, Delft, The Netherlands Natascha van der Velden, Industrial Design Engineering, TU Delft, Delft, The Netherlands Zjenja Doubrovski, Industrial Design Engineering, TU Delft, Delft, The Netherlands Jo Geraedts, Industrial Design Engineering, TU Delft, Delft, The Netherlands Elvin Karana, Industrial Design Engineering, TU Delft, Delft, The Netherlands

Abstract— This paper describes the findings and results of a design project with the goal to design a wearable garment using 3D Printed textiles, which not only has functional or environmental superiorities, but also experiential ones. The approach that was adopted for this project is a recently developed method on Material Driven Design (MDD), which suggests a number of steps to design meaningful products when a chosen material is the point of departure. As this method has not yet been applied on a project involving additive manufacturing, another goal is to explore how the MDD method can be used in a project where AM is the primary production method. For MDD, this means that the material that is usually the starting point, should now be a combination of material, structure and process (MSP), and that it is important to understand how these aspects influence each other. The final MSP concept can be locally varied to create property gradients, which results in a range of slightly different MSP’s. These materials have been embodied in the design of a corselet, which utilizes the different properties of the MSP. A number of recommendations has been given for the development of future 3D Printed MSP’s. Additive manufacturing, textiles, 3D Printed textiles, Material Driven Design, garments

1. INTRODUCTION Until recently, applications of 3D Printing or Additive Manufacturing (AM) in the field of fashion have been limited to accessories and shoes, instead of garments. This could be explained by the limited set of materials available for AM that showed potential for comfortable garments. Pioneering work on AM fabrication of fabric-like materials was presented by Evenhuis and Kyttanen (2003), whose method included projecting a textile pattern onto a particular surface, for instance a piece of clothing, and generating a 3dimensional computer model of the pattern [1]. The result of this process is a complex model of interwoven links, which resembles chainmail structures as used for armour in the Middle Ages. Since then, the potential for creating textiles by means of AM has mostly been attributed to these structures. They are often called multiple assemblies, since in essence they consist of separate parts [2]. The only limiting factors attributed to these structures are the limitations of existing CAD modelling


tools, for instance the ability to “drape” the AM textile across a curved surface (such as the human body), which was extensively researched by Bingham et al [3], Crookston et al. [4] and Johnson et al. [5]. Proposed applications for these textiles were mainly functional, such as stab-resistant wearables [5] and high-performance or smart textiles [3]. However, the development of flexible materials suitable for AM seems to have renewed interest in other possibilities for the production of 3D Printed textiles. Mikkonen et al. [6] have tested the tensile strength of one of these flexible materials to determine whether it would be a suitable replacement for fabrics. At the same time, the possibilities of AM have not gone unnoticed in the world of fashion. The form freedom that AM provides has been utilised to create accessories that could not have been created without this technology. Only a few designers have tried their hand at making entire garments using AM. For example, Iris van Herpen, in collaboration with architects such as Beesley and Koerner and designer Neri Oxman, has designed and fabricated numerous sculptural AM garments [7]. These garments were 3D Printed using rigid and flexible materials developed especially for this purpose. The emphasis of such work ususally lies on finding a way to translate the vision of the designer and to make a statement, as is common for art, and not in creating objetcs for daily use. As a result, most of the 3D Printed garments illustrate groundbreaking developments but do not represent comfortable, ready-to-wear clothing. Considering the developments of AM, the potential of using 3D Printed textiles in ready-to-wear garments becomes apparent: they can be more comfortable, personalized and suitable for daily use. In this paper, the findings and results of a design project that adresses this gap are presented. The ambition of this project was to develop a wearable garment that not only has functional or environmental superiorities (e.g. comfort, personalised, no material waste), but also experiential ones, i.e. how a material/product is perceived by people (e.g. unique tactile experiences, the feeling brought on by unique garments). The approach is grounded on a recently developed method by Karana et al. (in review) on Material Driven Design (MDD),

74


which suggests a number of steps to design meaningful products when a material is the departure point. As this method has not yet been applied on a project involving AM, another goal is to explore how the MDD method can be used in a project where AM is the primary production method. 2. METHOD For this design project, a Material Driven Design (MDD) method was implemented [8]. The goal of the MDD method is to facilitate product design when a material is the main driver. The method is based on developing a thorough understanding of the material in order to reveal the unique qualities that can be emphasized in the final application. Karana et al. present four main steps in the MDD Method. The first step is centred on gaining an understanding of the material, by performing both a technical characterization as well as an experiential characterization. These can often be performed simultaneously, as they will complement each other. An important part of this step is playing or ‘tinkering’ with the material, to explore its limits. In the second step, a Materials Experience Vision is created. This vision expresses how a designer envisions the role of the material in product design, in relation to the user, product and context. The vision should be related to the unique functional and experiential qualities of the material, as well as to potential of the material for future, unforeseen applications. Such an abstract vision can be hard to relate back to formal material qualities. Therefore, in the third step, the designer can analyse the vision in order to obtain meanings (e.g. high-tech, feminine, cosy, and friendly) that can be translated to material properties using Meaning Driven Material Selection (MDMS) [9]. Finally, in the fourth step, the findings obtained in the previous steps are used to create material/product concepts. It is important to emphasize that the MDD method has been developed for material driven projects in which a particular material or material family (e.g. oak, cork, a smart composite, bio-plastics, etc.) is used as the point of departure for the design process. In this project, the intention is to explore how this method can be applied in a design project where not only the (type of) material is set, but where AM is defined as the primary production method. According to MDD 3D Printed textiles are classified as a semi-developed material [8]; a novel material of which the boundaries have not yet been determined. This material can be described as a combination of material, structure, and process (MSP), since these three factors influence each other and the properties of the 3D Printed textile. Therefore, all three are

a.

b.



important to the outcome of the final material. The results of the MDD method can be used to determine the boundaries of this MSP, to find a meaningful application and to give feedback for further development. 3. APPLICATION OF MDD METHOD IN DESIGNING WITH 3D PRINTED TEXTILES In this section, the application of the MDD method to a design project ‘Designing With 3D Printed Textiles’ is described. 3.1. Understanding the material In accordance with the MDD method, the first step of the process is understanding the material and characterizing it technically and experientially. It is encouraged to ‘tinker’ with the material, to get insights as to how it behaves. In order to gain an understanding of the MSP, a number of samples of 3D Printed textiles were obtained. Some samples were collected from AM service providers, designers, and open-source design databases, while others were specifically designed and 3D Printed for this project. Several samples are shown in Fig. 1. This was an iterative process, in which different possible designs were created and their feasibility as a 3D Printed textile was evaluated. It was found that different combinations of MSP result in different materials that can have different, meaningingful applications in different contexts. Three topics were important to understand the context of 3D Printed textiles: the process (i.e. 3D Printing), the product (i.e. textiles), and the MSP itself (i.e. 3D Printed textiles). Since the boundaries of the material had not yet previously been defined, it was necessary to analyse all three domains in order to find its limits and opportunities. This was done by means of literature studies, benchmarking and explorative sessions using the collected and fabricated samples. The most important results of this analysis are summarized below. a) 3D Printing Additive Manufacturing or 3D Printing is the collective term for all processes that can form a 3D product by means of adding material, rather than by subtracting material. The information for these products comes from a 3-dimensional computer-aided-design (CAD) model, which is sliced in discrete layers [3]. These slices correspond directly to the layers that are built by the AM process, allowing the production of virtually any geometry. Materials that can currently be processed by AM include polymers, metals and ceramics.

c. Figure 1. Samples of 3D Printed textiles

d.

75


The recent developments of AM have set high expectations for the future of this technology. As a result, the tone with which is spoken about 3D Printing in the media is one of excitement, anticipation and innovation, as is consistent with the Gartner Hype Cycle, where 3D Printing is now on its way to the top. Although the positive image of the AM process will most likely contribute to the acceptance of 3D Printed garments, it also means that if the product itself is perceived as not exciting or plain, it could clash with the expectations of the user, which may result in a negative attitude towards the product. b) Textiles Flexibility is the most important property for textiles, since without flexibility no wearable garment can be produced. However, there are more properties that are important for textiles, including warmth retention and absorption, softness and elasticity [10], [11], and that make them suitable to wear close to the skin. In order to understand why textiles have these properties, the structure and properties were analysed on four different levels: garment, textile, yarn and fiber. It is possible to distinguish a main structure for each level, which results in a hierarchical structure for the overall material. This hierarchical structure is responsible for most of the mentioned properties that are desirable in textiles, for instance warmth retention is caused by porosity in the structure [10]. However, although the hierarchical structure is important in order to create the desired properties, this brings challenges to the production of this structure. For each structural level, a different production process is necessary, of which the limitations and waste are accumulated across the chain. c) 3D Printed textiles Classification The main requirement for textiles created by means of 3D Printing was found to be flexibility, in order for them to be applied in wearable garments. Therefore, a classification is proposed based on the main source of the flexibility, as depicted in Fig. 2. Structure-based refers to the fact that the flexibility is obtained purely by the application of an appropriate structure, regardless of the material used. This kind of flexibility is obtained by means of discrete bodies that make up multiple assemblies. Material-based refers to the fact that the flexibility is obtained mainly due to the characteristics of the material, by the use of flexible materials such as elastomers. Finally, an overlapping category can be identified in which flexibility is obtained by a designed single body structure that incorporates variable thicknesses, which is named thin structures.

DESIGN FOR ADDITIVE MANUFACTURING Experiential characterisation To explore the experiential properties of the material, the collected samples were analysed by means of an explorative user study. In individual sessions >10 participants were shown the collected 3D Printed textile samples. While the participants were invited to touch and interact with the samples, their reactions, remarks, and interactions with the material were evaluated. It was found that the samples elicited movement in order to explore the flexibility of the material, by means of shaking, throwing and caressing the samples. ‘Playfulness’ and ‘surprising’ were found as pre-settled meanings, for which the flexibility of the material and the fact that they were 3D Printed contributed most. The latter also elicited a positive reaction, since the 3D Printing process is still perceived as new, exciting, and innovative. Although all samples were flexible, only one of the samples was explicitly described as a textile by the participants (Fig. 1a). The participants expressed that the fine structure of multiple assemblies made it feel softer and more drapable. The other samples were not seen as textiles. These results indicate that in order for the material to resemble a textile and obtain properties desired for textiles, such as softness and drapability, the macro-structure should be as fine as possible. In addition to the user studies, following the MDD method, a material benchmark was conducted in order to find examples of AM applied in the area of garments. Most application areas were found in the categories of jewellery and accessories (bags, shoes and hats). In the category of clothing, most application areas were dresses, underwear (corsets), swimwear (bikini) and more sculptural “armours”. 3.2. Creating a materials experience vision After the first step, according to the MDD method it is expected that the designer knows and understands the material. In order to find new, unique applications for the material, it is suggested to create a materials experience vision, which expresses the role of the material in the envisioned user experience, as well as the relation it has to the context [8]. In this design project, the vision is related to the findings from the analysis of the three domains mentioned earlier: 3D Printing, textiles, and 3D Printed textiles. The use of a new and innovative production process should be utilized to the fullest in order to be of most value. Personalization is one of the key aspects; garments can be produced to the exact measurements of people’s bodies, while still being economically viable. Also, the opportunities for including property gradients in the product to be printed (material or structural) are a unique benefit of the process. Looking at the current life cycle of garments, it becomes clear that it is driven by fluctuations in fashion, which often leads to the early disposal of garments. As a reaction to this, the trend of slow fashion is emerging. Slow fashion is centred on design for long term use and wear, with concern for the entire life cycle of the product [12]. It strives to achieve minimum impact and waste, by increasing the aesthetic, functional and emotional value of the garment [13]. This can be achieved by creating a timeless design that will withstand the influence of fashion. Personalization of a product can

Figure 2. Classification of 3D Printed Textiles


76


increase its emotional value [14], and thereby prolong its lifespan. From a functional perspective, the product should be a wearable garment that is not obtrusive or hindering in daily activities. On the material level, this means that the 3D Printed textile should on one hand be suitable for use in garments, and thereby withstand a number of technical requirements, such as flexibility, tear resistance, breathability and water resistance. On the other hand, there are a number of experiential qualities for the material that are related to creating a textile that is comfortable to use. Softness, smoothness, warmth, lustre and coarseness are examples of these qualities. In this case, a more abstract vision was desired, in order to go beyond the initial findings. The materials experience was formulated as the following statement: I want people to have an attachment to their 3D Printed garment in order to extend its life span, by creating a personally engaging experience, like the act of blowing bubbles. ‘Blowing bubbles’ is used as a metaphor, illustrating a simple, engaging act that is familiar to everyone. Making the biggest bubbles is a challenge, and watching the light react on them is a pleasure; they are engaging to make and engaging to watch. 3.3. Manifesting materials experience patterns According to the steps of the MDD method, the vision that was created in the previous step should be further analysed to obtain materials experience patterns [9]. These are obtained by analysing the vision and proposed interaction to distill ‘meanings’, which in turn can be translated into material properties by applying the Meaning Driven Material Selection (MDMS) method (see [15] for the application of the method). The meanings were distilled by means of analysing the metaphor and several brainstorm sessions. The two meanings that were thought to best fit the intended interaction were intriguing and familiar. The meaning intriguing is related to the engaging experience, which will keep being interesting and surprising over time, while the meaning familiar can be described as ‘a friendly relationship based on frequent association’, comparable to a favourite jeans that has been worn many times.


The meanings were translated to material properties by analysing them with the MDMS research. In this research, a number of participants was asked to find a material that fits a meaning, to provide an image of it (embodied in a product) and to rate it on a scale of a number of sensorial properties. 13 participants responded for the meaning intriguing, and 13 for the meaning familiar. The results of the MDMS research are clustered per meaning in Fig. 3 and Fig. 4. Intriguing materials were found to be surprising and unexpected, by having a different look than feel for instance. They were found to be playful and raise curiosity, versatile in their properties and pleasurable to feel. Selected products in which the materials were embodied were practical and functional, but were enhanced by the used material to make them more special and not standard. Familiar materials were found to be as expected and common. They are considered reliable, and have an air of nostalgia. Most of the selected materials were natural, and recognizable as such. They were embodied in functional, practical products that appeared archetypical for their category. As seen by the descriptions, certain aspects of the materials are contradicting (e.g. surprising versus expected), and some supplement each other (e.g. warm versus comfortable). Therefore, in some cases it might be possible for the material to be both familiar and intriguing at the same time, for other aspects it may be necessary to choose for one of the meanings. It is however important to understand how the meanings can be used to enhance and limit each other. It was decided that the interaction should be intriguing at first; by exploring and using the material it will become familiar and personal. This means that the material should be playful, unexpected, raise curiosity and invite to interact with. At the same time, the feeling of the material is important; both for the meanings as for the product category. The feeling of the material should be comfortable, playful and warm, and preferably be recognizable as a natural material. On a performance level, the material should be versatile and reliable, while on a product level it must be practical and functional, with an archetypical shape.

Figure 3. MoM of familiar materials


77



Figure 4. MoM of intriguing materials

3.4. Designing material/product concepts In the final step, the findings from the previous steps should be used to create material and product concepts. With the material requirements and the findings from the technical and experiential analysis in mind, a number of different MSP samples were created. One MSP seemed to be most promising to be used as a textile-like material, shown in Fig. 5. This was chosen to be used in the product concept creation. a) MSP development For the concept creation, two workshops were conducted, one with 13 students of fashion design and one with 12 students of industrial design. The ideas that arose from these workshops were analysed. It was found that the students did not regard the 3D Printed material as a textile; it was rather seen as a substitute for plastic parts that are normally rigid and solid (e.g. casts or braces). This led to the conclusion that for this MSP the structure and process were suitable, but the material was not: although the material (plastic) was familiar, it was not familiar for the context it was intended for. After all, although a lot of textiles are made of plastics, their structure prevents it from being recognized as a plastic. Therefore, a number of experiments were conducted with different materials. The material that showed the best results and had the best fit with the intended Materials Experience Vision was a mixture of cellulose fibers with a flexible acrylic, as shown in Fig. 6. This mixture can be printed using the AM technology Material Extrusion, in which the material is extruded through a (non-heated) syringe in the desired structure, after which it has to dry. The experiments were perfomed both by manually extruding the material through a syringe, as well as by mounting the syringe on a material extrusion printer Ultimaker Original. These initial experiments served as a proof of concept for the newly developed material, although more research is necessary in order to make it suitable for processing. However, the results do give an impression of what the material could be like in the future.


b) Product concept A concept was developed using this MSP. In order to do so, the unique properties of the MSP were analysed: its aesthetics are most prevalent, most notably the pattern that resembles lace and is somewhat prevailing. The structure can be varied with: it can have a square configuration or a hexagonal configuration, changing the appearance and openness of the material. It is also very suitable to make alterations in properties (i.e. making the pattern smaller and higher decreases the flexibility of the material), it makes sense to use it for applications where this quality could be used to the fullest. The product should fit in the category garments, as was part of the assignment. By means of several brainstorm sessions, the most valuable product direction was found to be brassieres. These contain a large number of different parts and functions, therefore they are extremely suitable to locally vary material properties and to integrate parts. The design is shown in Fig. 8. The choice was made to design a corselet, which is essentially a cross between a bra and a top, in order to demonstrate the versatility of the selected MSP. In the concept, the entire product is 3D Printed exactly to the size of the user and can be custom-made. This means that in theory the exact design can differ, depending on the needs and desires of the user. The design as presented here can be seen as a basis for further adjustments. It is printed at once, meaning there is no need for assembly. This also means that the MSP should fulfill all functions that are usually provided for by a number of different parts and materials. Two types of the pattern of the material are used: the hexagonal configuration for the cups, in order to accommodate the round shape, and the square configuration for the other parts. In order to provide for the supportive parts, gradients are applied to the material: a gradient in size and a gradient in thickness. Supportive, more structural parts have a smaller pattern size and are thicker (up to 1.5 mm), while the parts that do not have to provide support are thinner (~0.4 mm) and have a larger pattern size, which makes them softer and more pliant.

78



printer Ultimaker 2 using the material polylactic acid (PLA), which currently produces the most reliable results on this 3D Printer. It was found that the property gradients in the material worked well, although they could have been a little more pronounced by increasing the z-height. The smoothness of the underside of the MSP increased the skin comfort, although some of the edges were still rather sharp.

Figure 5. Chosen MSP

d) Life Cycle Analysis The fact that 3D Printing significantly reduces the number of process steps necessary to produce garments and the amount of waste material, means it has the potential to contribute to environmental sustainability. The total impact of the product was evaluated by means of a Life Cycle Analysis (LCA) (as explained in [16]), and compared to traditional manufactured textiles for 1 kg of textile. It was assumed that the 3D Printed textile was produced by the Fused Deposition Modelling (FDM) process of PLA. The results of the anaylsis, as shown in Fig. 9, were compared to those of traditional textiles, as analysed in [17]. It was found that the environmental impact of the 3D Printed textile is comparable to those of woven textiles with a yarn thickness of 300 dtex. The largest part of the costs is determined by the FDM process (51%), followed by the costs of the consumer transport by passenger car (31%). 4. DISCUSSION

Figure 6. Sample of cellulose fibers mixed with flexible acrylic

This paper has shown the application of an MDD method to a design process where AM is the primary production method. The goal of the design project was to create a menaingful application using 3D Printed textiles. Since the method was applied to not only a material, but a combination of material, structure and process, the process was somewhat different. For the first step, it was found that not only an understanding of the material is necessary, but an understanding of the MSP as a whole and of all separate aspects was necessary, including how they influence each other. It was also necessary to research and define the boundaries of the MSP, since this has not been done before. AM as a primary production process offers the opportunity of creating personalized products, which has influenced the final material concept. Rather than being one fixed material, the material can be locally varied to create property gradients, which results in a range of slightly different materials that all fit the intended vision.

Figure 7. Prototype of final design

c) Prototype A prototype of the design was built to test the application of the MSP, as shown in Fig. 7. A dress form was made, to which the product was fitted. It was printed on a material extrusion


The MSP and product that are created as a result of this process, demonstrate the potential for 3D Printed textiles. Even though the final material does not adhere to all the properties that are desirable for textiles, it has shown potential to be used as a 3D Printed textile for garments. Two main factors that should be improved before it can actually be worn are its tearresistance and the softness of the material, which is necessary if it is supposed to be worn close to the skin. The latter can be improved by either using a more compressible material or a material with a softer outer surface, such as the proposed cellulose material. The material as it is designed now is bound by current technological limitations. With improvements of current technologies, some advancements for the material can also be made, it would for instance be interesting to test the behaviour

79



Figure 8. Final design

of the material if the scale of the macro-structure can be decreased, to make it more resemblant of traditional textiles. The current material options were also found to be too limiting, which is why a new material blend was proposed. Although this blend has the potential to be printed, it would be interesting to see different possibilities for printing natural-based materials in the future. Therefore, for future applications for 3D Printed textiles, it is recommended that an AM process will be developed specifically to create textiles, rather than keep the focus on material development. In essence, textiles should be seen as an MSP: their properties are influenced by materials, structures

and AM processes. In order to be able to print textile-like materials, the materials, structures and process that are in place now should be thoroughly analysed and used as inspiration for new AM processes. Future work will first be focusing on testing the functionality of the MSP by means of the prototype. Next to that, the cellulose blend will need to be researched further in order to develop it for use in AM and for its function as a textile. REFERENCES [1] J. Kyttanen and J. Evenhuis, “Method and device for manufacturing fabric material,” WO2003082550 A22003. [2]

N. Hopkinson, R. Hague, and P. Dickens, Rapid manufacturing: an industrial revolution for the digital age. John Wiley & Sons, 2006.

[3]

G. a. Bingham, R. J. M. Hague, C. J. Tuck, a. C. Long, J. J. Crookston, and M. N. Sherburn, “Rapid manufactured textiles,” Int. J. Comput. Integr. Manuf., vol. 20, no. 1, pp. 96–105, Jan. 2007.

Figure 9. Results of the LCA analysis: impact of each stage of the life cycle of the product as percentage of total impact


80


[4]

J. J. Crookston, a C. Long, G. a Bingham, and R. J. M. Hague, “Finite-element modelling of mechanical behaviour of rapid manufactured textiles,” Proc. Inst. Mech. Eng. Part L J. Mater. Des. Appl., vol. 222, no. 1, pp. 29–36, Jan. 2008.

[5]

A. Johnson, G. a. Bingham, and D. I. Wimpenny, “Additive manufactured textiles for high-performance stab resistant applications,” Rapid Prototyp. J., vol. 19, no. 3, pp. 199–207, 2013.

[6]

S. Kivioja, H. Suonsilta, and O. Fi, “Printed material and fabric,” pp. 313–321, 2013.

[7]

I. van Herpen, “www.irisvanherpen.nl,” 2014. [Online]. Available: www.irisvanherpen.com. [Accessed: 20-Aug-2014].

[8]

E. Karana, B. Barati, V. Rognoli, and A. Zeeuw van der Laan, “Material Driven Design (MDD): An Illustrative Case on Designing With Waste Coffee Grounds.”

[9]

E. Karana, P. Hekkert, and P. Kandachar, “A tool for meaning driven materials selection,” Mater. Des., vol. 31, no. 6, pp. 2932–2941, Jun. 2010.

[10]

Y. Li, “the Science of Clothing Comfort,” Text. Prog., vol. 31, no. 1–2, pp. 1–135, Mar. 2001.

[11]

K. Slater, “Comfort Properties of Textiles,” Text. Prog., vol. 9, no. 4, pp. 1–70, Jan. 1977.

[12]

K. Fletcher, Sustainable Fashion & Textiles: Design Journeys. London: Earthscan, 2008.

[13]

S. Black, Eco-chic: the fashion paradox. London: Black Dog Publishing, 2008, p. 254.

[14]

R. Mugge, H. Schifferstein, and J. Schoormans, “Personalizing product appearance: the effect on product attachment,” Proc. Int. …, vol. 31, no. 0, pp. 1–15, 2004.

[15]

E. Karana, “Characterization of ‘natural’ and ‘highquality’ materials to improve perception of bioplastics,” J. Clean. Prod., vol. 37, pp. 316–325, Dec. 2012.

[16]

J. G. Vogtländer, LCA: a practical guide for students, designers and business managers. Second Edition. Delft: VSSD, 2012.

[17]

N. M. Van Der Velden, M. K. Patel, and J. G. Vogtländer, “LCA benchmarking study on textiles



made of cotton, polyester, nylon, acryl, or elastane,” Int. J. Life Cycle Assess., vol. 19, no. 2, pp. 331–356, Sep. 2014.

81



Redefining the role of designers within an urban community using digital design and localized manufacturing of wearables. Cees Jan Stam, Industrial Design Engineering, TUDelft, Delft, the Netherlands, [email protected] Natascha M.van der Velden, Industrial Design Engineering, TUDelft, Delft, the Netherlands, Gerard Rubio, OpenKnit, Barcelona, Spain Jouke Verlinden, Industrial Design Engineering, TUDelft, Delft, the Netherlands Abstract— The maker culture has created a dynamic in which designers are less responsible for the design and quality of the final product, but for the tools the consumer uses to create their own. While additive manufacturing (AM) is gaining acceptance among the general public, it is still seen as a prototyping tool instead of a high quality production technology. This limits its acceptance within co-design and maker culture. The research question is: How to create greater acceptance among the general public regarding the AM technology and its products? One way to create greater acceptance of digital design and manufacturing is to apply co-design principles on a local scale. By this means the public will be exposed and included in the design and production process, which will ensure the end product is better accepted. In time this could help spark a maker movement within the community. To validate these assumptions a test case was developed in which local design and production of simple wearables, small ready to wear garments like socks or hats, within an urban community will play a major role. During the research a digital design tool combined with a mobile digital knitting machine was developed to allow for a rapid codesign track. Wearables would be produced by the consumer themselves. The final design of the garment depends on the consumer’s choice of material, shape and pattern. A mobile setup provides the means to test the principle at different locations and allows the consumer to be intensively involved in the maker movement in their own neighbourhood. We implemented a small, low-cost knitting machine that was tested outdoors by park visitors. The anticipated results for this test case were: increased engagement in the production process, larger acceptance of digital design and an initial maker culture. Although the last result will be difficult to determine as it takes some time to develop. If successful, the maker culture will obtain greater exposure, acceptance and demand for digital design services and products. Even though the maker culture changes the role of the designer will definitely change, their importance to the design process will remain, not as a creator of designs but moreover as a guide to the making of consumer products. Keywords-component; Co-design, Digital manufacturing, Wearable’s, Maker Culture, Sustainablity, Local manufactoring

1. GENERAL INTRODUCTION Even though digital design has made big leaps in the last years, most consumers are still very much unaware of its potential. This is limiting the development as more users and


cases within the field will help mature the technology. In order to facilitate a greater awareness and eventually acceptance we need to look outside the current scope of the exposure of the technology. How to create greater acceptance among the general public regarding the AM technology and its products? This will be the main question addressed in this paper. There are several means to try and achieve greater acceptance among the general public however, not all are aimed towards this particular issue. When looking at the general knowledge about the production of user products most people are blissfully unaware. This creates a lot of preconceived notions about the difficulties and also possibilities during the production steps. In order to get a more realistic perception regarding AM it is therefore imperative to expose the general public to its difficulties and more importantly its opportunities. Even though the freedom created by digital design and manufacturing is not necessarily desired by the consumers, it also allows the design community to develop and define this design space through tools and methodology. This ensures that the users of AM facilitated design and production will be able to freely explore its possibilities without being overwhelmed. The chosen product group, wearables, was selected for it duality. While on one hand garments and other body orientated products are used as an expression of personal style and preference. Yet at the same time it also follows mass consumer behaviour. These two seem to be in direct conflict with each other. Another aspect in regards to wearables is product fit (Van Der Velden, Patel & Vogtländer, 2014). While no two people are exactly alike the consumers have to cope with standardized sizes. This in stark contrast to the fact that a correctly fitted product can greatly increase the product satisfaction. As such it is an area well suited to the possibilities of digital design and production, as it allows the users to design and wear made to measure or even bespoke tailored garments. Which in turn should result in a greater product attachment which carries value in the field of emotional sustainability. To ensure that consumers are aware of the possibilities granted by these technologies and expose them to it in a proper way is a challenge. This creates possibilities for designers to reshape their roles within this dynamic. As designers now get the opportunity to create the tools with which the everyday consumer could design and create their own wearables.

82


To achieve this the following activities were undertaken to test the public’s receptiveness to AM produced clothing. First of all a more detailed overview of the developments in local design & manufacturing will be given, as it defines the scope of the research. Followed by selecting a method of creating this acceptance, it will need to fit the context of not just creating acceptance but also instigating a maker spark in the consumer. This will then be applied to an interaction design to create the desired outcome. This interaction will be tested within a public area to validate its effect on the public. After which it will be evaluated and suggestions for adaptations to further improve its effectiveness are given. 2. STATE OF THE ART LOCAL DESIGN & MANUFACTURING 2.1. Redefining the role Within the current wearables market the interaction is based on the industrial production of clothing. This results in a gap between designer and end user, with producer, wholesalers and retailers as the stakeholders (Figure 1). This construction has benefits, since each stakeholder has a clear task which can be optimized and perfected. However it also results in a lot of global shipping and a gap between designer and end user. The process also relies on large production numbers in order to function creating the need for standardization. When looking at the new dynamic that is created though local design & manufacturing (LDM) we see that the stakeholders change and that they take on new roles. This contrasting dynamic creates new opportunities and benefits. The close proximity both physically and structurally allows for different design methods and interactions. It creates room for personal/ one off designs as well as small locally influenced series of products. This new dynamic however does ask for different design tools and methods. As such redefining the role of a designer in this context will be imperative. Localized manufacturing has other benefits regarding sustainability, when looking at shipping and emotional sustainability. As the involvement with the creation of the product grows so will the attachment.

Figure 1. Current interaction and the envisioned interaction between designer and consumer



2.2. Maker culture One of the main drivers behind localized manufacturing , and a redefining one for designers, is the rise of the maker culture. This cultural shift from mass produced to home-/selfmade products is driving a new wave of development in localized manufacturing and design tools and methodology. One of the ways designers can reinvent themselves is to tap into this movement and create the design tools needed for the general public to design and create their own products. While the early adapters have the skills to design and make what they come up with, this will not be the case for everyone. So this leaves a group to design for. The exposure of this maker culture is something else to consider, the current methods are aimed at the first group. They consist of several facilities/activities: a) Fablab The fablab principle is something that fits into the maker culture, as it allows makers to build more complex products for which they do not own the tools or have the expertise to build. These workplaces are stocked with digital manufacturing tools like; 3d printers, CNC machine and laser cutters. These are augmented by the more common tools like; drills, laves and band saws. These spaces are either open to the general public or are linked to educational or artistic institutes. While most major cities around the world have a fablab facility, most are hidden from public view due to location or lack of recognizable markings. This results in a lack of public knowledge about the facilities and as such fails to connect to the general public (http://fablab.org). b) Makerfairs These events are generally held in public areas/ buildings and generate more public awareness and attention. While still visited mainly by makers, they also attract people generally interested but not (yet) participating in maker culture. These event help to showcase, educate and create appeal for the results of maker culture. This has a great benefit in helping the movement to grow and develop. As fellow makers can meet and exchange ideas. This is augmented by the physical nature of the event in that the products and tools are there and can be used/touched and explored. Still most of the visitors are already interested in or connected to the movement, creating a new wave of makers from yet unengaged people is not the aim of these events. ( http://makerfaire.com/) c) Digital Maker Culture One of the effects of the digital design is the ability to share it using digital media. This does not exclude other nondigital designs as tutorials are also wide spread. This helps to create exposure for the products that can be made. While most users of digital and social media will come into contact with maker culture the effect of seeing a picture or movie is not the same as holding the actual product. This gap between exposure and contact is a limiting factor in creating attraction in regards to the final product (Doctorow, 2009).

83


3. METHOD 3.1. Co Creation In order to test the new dynamic between designer and end user it is important to redefine their relation. The freedom created by LDM also creates a larger design space. In order to help guide the end user in this process designers have an opportunity to lend their expertise by means of Co Creation. Within this design method designers are moving away from translating the needs of the end user into a product. Instead they are facilitating the creation of this product by the end user (Sanders & Stappers, 2008). This shift not only redefines the role of the designer but does the same for the role of the end user. Since They will have a greater influence on the front end of the design process, and as such on the final product This coincided with the change in dynamic envisioned for the application of LDM as the method allows for local influences to guide the design process. It is not just limited to the local users but also local materials and cultural heritage. This will be used in combination with the design of wearables, were a correct fit and integration of personal style is valuable. 3.2. Concept testing In order to evaluate the success of a localized manufacturing process concept testing will be used. The concept will be evaluated on several key aspects; general, features, product, durability and reliability. These aspects represent the desired overall qualities of the concept. By testing the concept using the intended target group as well as the intended context, the following data can be collected photographs, video and interviews. These will show the general public’s overall interaction with the concept as well as offer detailed accounts of individual interactions. These results will then be used to create a concept testing matrix. This will either validate or invalidate the concept as a means to achieve the desired goal of creating greater acceptance and interest.




The product coming out of the machine should require little to no extra actions, as close to ready to wear as possible.

In order to create clothing without directly using traditional methods there are several options. There are methods that work with regular yarn and use weaving/knitting techniques. Furthermore 3d printing clothing is being considered within the design community as a replacement of these traditional material and production techniques. However the aim of this research is to test ready to wear garments. While 3d Printing allows for great freedom in shape and construction it is seen more as an haute couture fashion technique for example the works of Iris van Herpen. This combined with the long production time makes it unusable for this research as the aim is to create more acceptance a more intermediate step is needed. As such the following possibilities were taken into consideration. Each will be shortly addressed and checked with the criteria. a) Knitic, manual knitting machine hack This system is a recent development, where by hacking the old manual knitting machines you are able to create new digital designs. The Knitic design couple is working with this technology using several interesting input signals to create uniquely patterned designs. The machines are reliable as they basically hack into an existing flat knitting machine. The main problem this creates is the sheer size and weight of these machines. The machine also is not able to knit full garments as it only allows for sheet knitting, this increases the manual workload after the initial knitting. Although possibly more reliable they are hard to modify. And while an interesting project it seemed unsuited for the current goal of local exposure.

4. DEVELOPING WALLY 120 4.1. Preperations In order to facilitate the localized manufacturing aspect of the test case, a mobile digital manufacturing tool was needed. In order to use the tool within the local context several criteria where listed:  It needs to be mobile, or light enough to be moved by a single person (less than 10 kg.)  Big enough to create small garments; socks, scarfs, hats.  Self-sustained when in use, no external power needed at the production location.  Allow for a made to measure approach, allowing the user to take his or her own measurements by adapting an existing template.  It needs to be reliable, as a minor error will ruin a garment. Figure 2. KNITIC digital design knitting machine & pattern example


84


b) Circular knitting machine When looking at the criteria most of the selected garments are tubular in shape. One of the fastest ways to knit in this fashion is using a circular knitting machine. These are very reliable as the knitting motion is never interrupted. They also allow for increasing and decreasing needles, which allows the knitting of heels. However no progress into digitizing this progress on a small scale has been made at this time. This is also likely related to the fact that sizes are only changeable by switching out the complete needle ring for one with more or less needles. So while good at what it does it can only do so much. The digital knitters are currently in use on an industrial scale but so far have not been scaled down for personal use.

Figure 3. Circular knitting machine manpowered

c) Openknit This project was created by Gerard Rubio, as part of his graduation thesis. The OpenKnit system is an open source project working towards creating a digital knitting machine. The designs and software are available for free and together with the bill of materials can be built anywhere in the world. The current design uses 3d printed parts, lasercut parts and some vendor parts. This enables anyone who lives close to an fablabs or has a small workshop at home to reproduce it and contribute to the further development of the device. This opens the project up for wide spread testing and exposure. It works by programming the pattern into Arduino which can be modified to the users specifications. The machine however is bulky and in its early stages of development. It also has some issues regarding reliability. The machine does offer the freedom to create several different types of garments. Currently ranging from dresses to sweaters to beanies. While not ready for complex patterns it does allow for different colours.



Figure 4. OpenKnit, open source knitting machine

4.2. Developing a mobile solution The OpenKnit system was selected as it offers a combination of open/digital design combined with an open source machine. This allows end users or communities to create their own machine while also allowing the designers to adapt them to their specific needs. However the current design of the OpenKnit system was not suited for mobile use, several adaptations would have to be made. Therefore the machine was redesigned to be smaller, lighter and sturdier. Several tests were executed to test the new components durability and reliability, this was done on the main machine. The main components were all tested and (partially) redesigned. This was mainly focused on the carriage and the rack & pinion. The carriage is responsible for both guiding the thread as well as controlling the motion of the needles. Where the rack & pinion is vital to the accuracy of the machine as it creates the input for the software to determine the carriage position on the needlebed. The resulting machine was mountable on any flat surface using two clamps, weight was reduced to 5,5 kg. Its needle beds have a total of 120 needles, 60 a side. This is sufficient to produce a small wearables. It is battery operated using a 12v battery and converter circuit to power both the stepper motor and servo’s which run on 12 and 5 volts respectively.

Figure 5. The redesigned OpenKnit machine “wally 120” also shown are the main components if the machine

85



5. ENGAGING LOCAL COMMUNITY 5.1. Test set-up When selecting the type of garment to make it was decided to make a beanie. This simple woolen cap design would allow for a quick turnaround during the sessions and would allow the machine to run without interruption for a long period of time without spending too much time on the production of the garment. In total it takes 35 minutes to create. The following set-up was used, it follows the shown scenario. The test scenario consists of several steps each with their own function. The first step is to set-up the machine on its location, this means attaching the machine to a supporting structure. This can be a structure available on site or one that is brought. Attaching Wally is done with clamps, which ensures a solid connection without damaging the structure. The second step is to hook up the electronics, this consists of connecting the battery and the USB cable to the laptop. For the third step an interested onlooker is approached, a short explanation is given about the purpose of the machine and the general theory. If the person is interested a short production track can be started. This started with the fourth step here we measure the size of the head in order to make a beanie made to measure. This data will be insert into the Arduino software upon which the needlebed is prepared. The fifth step is threading the machine and setting up the first two lines manually. Afterwards the Wally takes over and knits the garment to its desired length. An optional step is to cut and splice a different colour thread during the knitting process. For the last step the garment is closed manually while still on the machine and then removed from the comb and needlebed. It is then ready to wear (Figure 6). During the knitting of the wearable and afterwards, the users are asked to express the experience and they are given the opportunity to ask any questions they might have. Secondly they are asked to share their expectations and desires regarding the machine and its applications. This ranges from what they would use it for themselves, to what they would eventually want to be able to make with it. This will

Figure 6. The Test scenario


Figure 7. The set-up

give an insight into their standings in relation to the technology and might also illustrate the changes herein, as a result of this new experience. The sessions will be documented by both video and camera footage. This will later be used to analyse the effect of this new interaction that occurs within the public space. Important is to also document the range of interactions and steps the public goes through. 5.2. Test location In order to get enough exposure the test location is of significance. While the centre of Barcelona is lively and full of people, the tourist is not the target audience. While people are very open to new things while on vacation the main goal of the test is to see if the general public is willing to accept AM technology and maker culture. To this end the Parc de la Ciutadella was selected, this city park is visited mainly by the local population. The park is still crowded enough to have enough exposure while not including to many tourists into the test group. The mounting of the machine did limit the selection of the test site as most of the surfaces where unfortunately rounded and therefor unsuited for the chosen mounting system. The chosen site was located near a intersection of the walkways and the playground (Figure 7). Especially the proximity to children was useful lase their curiosity and lack of inhibitions will help pull in more people. 5.3. Results The results of the test session held in the Parc de la Ciutadella, show a great variety of interest. The session created a good crowd of people looking at the machine at work. The steps described in 4.2 were followed as closely as possible as the installation would allow (Figure 8). As expected the children were first to explore, drawing in their parents and later more bystanders joined to see what was going on. As can be seen in Figure 9 the public was rather mesmerised by the machine. The movement and sound creating interest and upon closer inspection questions start to arise about the project, the overall goals and future use.

86



Ranging from remarks about the look of the machine to the technical specifications used to create it. This wide range of interest was already very useful. The wonderment seemed to suggest that most people had never seen a machine like this in action, not to mention in the middle of a public park.

Figure 9. Some of the gathered audience

The answers to the questions were mixed and varied wildly in detail so in order to get a quantitative overview they were categorized as positive, negative or indifferent. This creates the following overview in regards to the questions, as seen in Figure 10. In total 26 people took the questionnaire during the session. The questions focused on the following aspects of the concept: general impression, available features, reliability of the machine, functionality of the product and the durability of the machine.

Figure 8. Testing the machine (top) measuring the size,(middle) knitting the beanie, (bottom) ready to wear


87



lower price (30%). This was in within the context of made to measure patterns. Were they can enter their measurements and select patterns/colours freely. 6. DISCUSION

Figure 10. Quantative overview of the questionaire positive/negative/indifferent

When discussing their possible future use of this technology, the responses were categorized as well. This question was aimed to determine the likelihood they would use this machine or something similar in the future. The responses ranged from; definitely, possibly & never. This resulted in the following overview:

Figure 11. Possibility of future use definitely/possibly/never

With regards to the type of desired use the following categories were given: small wearables/ Simple sweaters, dresses & vests/ Hoodies, buttoned & other complex garments. This was to see what type of garment they want to create should they use the machine themselves. They were told to ignore size or current technological limitations in this answer. The results were as followed:

When looking at the results of the engaging local community test case seems to check all the boxes in regards to the successful realisation of greater acceptance. However when looking at the test set-up and general several issues came to light. First, the test location in combination with the time the test was held at. The test was carried out during the late afternoon early evening 18:00 -20:15. This might have an effect on the results in that the public could be tired, on their way home. In order to exclude these and other factors from the results a second session at a different location and time of day would be needed. Secondly, while the machine performed well it struggled due to the method of placement. The gate it was attached to resulted in an off level position which created a greater strain on the system then initially anticipated. In order to prevent this in future tests either the mounting system or mounting location will need to be addressed. As the struggling machine has effect on the perception of durability and reliability as mentioned by one of the participants; “It seems to struggle a lot, especially going towards the edges of the beanie, does it always do this?”. In order to create a positive image for AM technologies the reliability will need to be increased. Another effect of the Wally 120 system that limits the testing at this point is the lack of interface design integrated into the system. In order to let the public use the machine by themselves, an interface will need to be developed. This also ties into the limitations currently attached to the machine as it is still not able to decrease needles, needed to be able to do short stitching. This is still under development and once completed will greatly increase the range of designs the Wally 120 could handle. Also there was the matter of language. Even though a Spanish native was present during the testing the researcher himself did not speak Spanish this created some difficulties explaining the machine and answering the questions. While this did not affect the general insights into the effect of the machine it did limit detailed discussions. 7. CONCLUSIONS

Figure 12. Percentages off desired use for Small/Simple/Complex garments

When asked why they would use the machine instead of buying readymade garments their response were mainly focused around the following properties; better fitting clothes (72%), more freedom in creation/personal style (45%) and


. The concept of introducing a localized manufacturing tool into an urban community resulted in a positive response from the general public. When looking at the results it is clear that when confronted with AM technologies in a urban context general interest is increased. Looking at the Attention Interest Desire Action or AIDA model the following can be concluded. The Attention was created, the Wally drew in a large crowd before it was even turned on. The Wally was considered intriguing, because of the colour, sound and overall shape that stood in stark contrast with its surroundings.

88


Attentiveness was high, 70% of the people that stopped to take a longer look asked questions, made pictures or were talking amongst themselves about the machine. When looking at the answers in regards to future use the crowd was positive. 64% of the participants of the questionnaire would use this machine if it would function similarly to the test conditions. 53% of the participants would use the machine for complex garments, while 87% of them would use it for small simple garments. This shows significant interest and desire in regards to using the machine. Action was not addressed in this test as the machine is still in development. However the initial responses regarding the Wally 120 were positive and several machines are currently under construction around the world. A important thing to notice is their need for better fitting clothes as 72% of the participants claim this as a reason to start using this type of clothing manufacturing. It seems that even though the standardisation of clothing is able to facilitate the industrial production of clothing it does not seem to fill the needs of the users. When looking at the role of the designer in this process, it can be concluded that this has been altered. The designer is no longer just creating products that fill the needs of the consumers. Instead we see a new task taking shape, designing and defining the tools and design space for the end user. This is partially done with a co creation process at this time. However this can be further developed to let the users freely design and manufacture their products without any direct contact. The contact between user and designer will then be through the design space created by the designer.



8. ACKNOWLEDGMENT Furthermore thanks go out to Prof. Dr.ir. J.C. Brezet for proof reading this paper, and adding his particular view as a soundboard during this research project. His focus on circular economy was a driving force during the entire process. Thanks also go out to the people at Arduino. There financial support to the OpenKnit project enabled to the fast creation of the Wally 120 machine. And allowed Cees Jan to prolong his stay in Barcelona further improving on the OpenKnit project in general. 9. REFERENCES [1]

Elizabeth B.-N. Sanders & Pieter Jan Stappers (2008) Co-creation and the new landscapes of design, CoDesign: International Journal of CoCreation in Design and the Arts, 4:1, 5-18 [2] Johansson, A., Kisch, P., & Mirata, M. (2005). Distributed economies–a new engine for innovation. Journal of Cleaner Production, 13(10), 971979. [3] Moore, W. L. (1982). Concept testing. Journal of Business Research, 10(3), 279-294. [4] Page, A. L., & Rosenbaum, H. F. (1992). Developing an effective concept testing program for consumer durables. Journal of Product innovation management, 9(4), 267-277. [5] Guljajeva, V. & Canet M. (2013, March 21). Knitic demo & tutorials now online!. Message posted to http://www.knitic.com/ [6] OpenKnit. (2014). Made In the Neighbourhood (ft. a clothing printer, OpenKnit) Retrieved March 8, 2014 from: http://vimeo.com/86987828 [7] Elissa, K., 2004,‘Title of paper if known’. Paper presented at the conference, place, unpublished. [8] Van Der Velden, N. M., Patel, M. K., & Vogtländer, J. G. (2014). LCA benchmarking study on textiles made of cotton, polyester, nylon, acryl, or elastane. The International Journal of Life Cycle Assessment, 19(2), 331–356. doi:10.1007/s11367-013-0626-9 [9] E.K. Strong (1925) Theories of Selling Journal of Applied Psychology, volume 9, pagina 75-86 [10] Doctorow, C. (2009) Makers. New York City, NY: Tor Book

89



Model of a Measurement Artifact for Additive Manufacturing Bogdan Galovskyi, Friedrich-Alexander-University Erlangen-Nuernberg (FAU), Institut of Manufacturing Metrology, Erlangen, Germany, [email protected] Tino Hausotte, Friedrich-Alexander-University Erlangen-Nuernberg (FAU), Institut of Manufacturing Metrology, Erlangen, Germany, [email protected] Dietmar Drummer, Friedrich-Alexander-University Erlangen-Nuernberg (FAU), Institute of Polymer Technology, Erlangen, Germany, [email protected] Meng Zhao, Friedrich-Alexander-University Erlangen-Nuernberg (FAU), Institute of Polymer Technology, Erlangen, Germany, [email protected]

Abstract— The aim of the current investigation is to develop a model of a measurement artifact and procedures of its manufacturing and testing. Several prototypes were manufactured concerning such influences as: location of the part in the powder bed, its size and typical shape deviation. The prototypes contain different structural elements for determining the tolerances of workpieces produced by SLS machines. Positional, run-out and shape deviations were estimated for the artifact with the lowest shape deviations. Reduction of the artifacts size due to the shrinkage effect was observed. Keywords- testing artifact; aditive manufacturing; measuring assurance; manufacturing resolution

1. INTRODUCTION The application field for additive manufacturing (AM) has been expanded during recent decades. Nowadays this technology is applicable not only in prototyping or scientific researches but also in mechanical engineering and medicine etc. Selective laser sintering (SLS) technology has a great potential due to the fast and easy adjustable manufacturing of complex parts with a wide range of sizes and materials. Intensive development of AM brings some limitations for its integration and use in industry. These are for instance: weak relation to the legal metrology and traceability to national standards. So called manufacturing resolution of the process is subjective value which depends on the diameter of laser beam, raw materials, location of the part during the sintering, temperature fluctuations during the process etc. A manufactured workpiece has to be measured and tested in accordance with the national and international standards, but these procedures are impossible because of the undeveloped metrological assurance in the field of AM. Measurement equipment provides control of conformity with specification for manufactured workpieces. In one of the standards regarding components and process testing for AM [1] is claimed: "...complete testing of all workpiece characteristics is possible only to a limited extent for reasons of cost effectiveness and technical feasibility...". In order to improve metrological assurance in the field of AM a number of research


projects are running [2]. Quality control, test methods for raw materials and process control are main research directions in the field of standardization for AM. Improvement of SLS process for plastic materials with the help of in-line measurements as well as post-process measurements is the research aim. Methods for quality control of SLS were listed and divided into the following directions [3]: testing of the raw powder material, process control, testing of the part quality. Investigations of the shape and position of the part (geometrical tolerances) have a higher priority [4]. Testing of the workpiece quality [5] refers to the testing procedures established for the traditional manufacturing approaches. Mechanical properties, shape, dimensional accuracy and surface quality are properties of the end product which should be confirmed in accordance with regular testing procedures. The quality of the process is of high importance for users and manufacturers of SLS machines. There are two basic approaches for evaluation of process parameters: 

reproducing with the help of SLS machine some testing part with numerous structural elements (testing part, testing artifact);



in-line monitoring of selected influencing parameters of the machine (temperature deviation inside the process chamber, power of the scanning laser, scanning speed etc.).

VDI 3405 establishes the second approach as the primary. Though some researchers of SLS process give the priority to the first method [5]. Both statements are based on numerous disadvantages: the approach with the testing artifact is subjective and represents process parameters indirectly; the monitoring of process parameters is too complex and cumbersome. It is necessary to determine the aim of the process testing before choosing appropriate approach. For manufacturers of SLS machines it is more important to monitor numerous parameters listed in VDI 3405. The testing program should be adapted for different machines and should be used together

90


with a lot of supplement measurement systems. But the expected result is uncertain because of numerous influential factors, their unpredictable behavior and mutual interaction. Any testing procedure will require significant time and financial investments. The approach based on testing artifact is more appropriate for users of SLS machine. This approach meets lower expenses and all errors are collected in the manufactured workpiece. After producing of the testing artifact, reproducibility of the SLS machine (manufacturing resolution) and errors of the manufacturing can be studied. Furthermore, not only geometrical parameters of the part can be evaluated, but also mechanical properties [6, 7]. The measurement artifact is an appropriate mean for calibration of measurement equipment. After the development materials and machines for AM, it was proposed to use a testing artifact in order to evaluate manufacturing capabilities of AM [5, 8]. Design of the testing artifact depends on raw material for SLS process, size of the process chamber etc. The artifact should contain features with the parameters for process testing and improvement. 2. DESIGN AND MANUFACTURING OF THE TESTING ARTIFACT 2.1. General considerations Depending on the final aim of research there are different testing parts and artifacts were designed for AM. For earlier investigations was developed and tested a number of simple objects, some examples of which are showed in Fig. 1. There are step height testing parts (Fig. 1a, 1b) that were developed in accordance with DIN EN ISO 5436-1:2000. The Siemens-Star testing workpiece was designed to improve the lateral resolution of SLS process [9]. Nevertheless it was necessary to get a universal part for the confirmation of numerous parameters. The experience and recommendations of other researchers were taken into account [5, 7, 8]. As it was mentioned [5, 10], the testing part should have various structural elements such as holes, tubes, parallel and perpendicular surfaces etc. The main rules for the artifact’s design were formulated and improved for developed measurement artifact:

a)

b)

c)

d) Figure 1. Example of testing parts





possibility to build A test area without support structures and post-process treatment;



manufacturing of the artifact should require low amount of raw material;



design of the artifact should be less affected by influence of systematic distortions during the manufacturing process;



structural elements should be measurable

Structural elements of the artifact were developed considering the need to investigate lateral and vertical manufacturing resolution of SLS process, internal deviation of the solid part, reproducibility of the simple geometries, and position deviation of the structural elements. 2.2. Structural elements of the masurement artifact In Fig. 2 general view of the designed artifact is presented with a scale factor of 1. The artifact has rectangular shape and one of the two large surfaces is the functional surface which contains most of structural elements. Sides of the measurement artifact have also structural element for the process properties evaluation. Holes which go through the artifact’s body are visible on the sides (Fig. 2b, 2c). These elements’ presence is aimed at the solution of two tasks. The first one is to get information about the deviation of the artifact’s body. Most part of these deviations occur during the temperature changes inside the process chamber (cooling of the sintered part) and are typical for parts which have big melting areas during the manufacturing process. Measurement of the axis deviation for each hole could be used for the evaluation of parts' geometry deviations and their localization. In previous work for such purposes was used the computer tomography measurements (Werth TomoCheck 200 3D) of the sintered parts [11]. There are also lateral elements on the sides of the artifact: circular and hexagonal elements. These elements allow the assessment of SLS process distortion during producing lateral structural elements with thin walls. The position distortion for such elements could be evaluated as well. The functional surface is the most important part of the measurement artifact. It was manufactured without any supporting structures. All structural elements should be described in detail (Fig. 3). All the mentioned values are given for the artifact with a scale factor of 1.

a)

b)

c)

Figure 2. General view of the measurement artifact

91


The central position of the functional surface is occupied with the vertical cylindrical structure of different diameters. The end of this cylinder contains a Siemens-Star structure (Fig. 3d). It is used for determination of the lateral manufacturing resolution of SLS process [9, 10]. The common axis of the two lateral cylinders with conical endings (Fig. 3j) defines one coordinate axis of the artifact. The orthogonal axis is defined by the other cylinder (Fig. 3j). In order to evaluate the position of deviations of the other structural elements these axes could be used. There are 3 sets of the step height standards. Positive (Fig. 3g) and negative (Fig. 3f) step height structures have 10 steps each and height (depth) difference of 50 μm between each other. The third set of step height standard (Fig. 3c) contains 20 steps (10 positive, 10 negative) with the height difference of 10 μm. Most users of SLS machines have experience with using recycled (refreshed) raw materials – powder which was in use but is not sintered, which has economic reasons. But because of the temperature stress the properties of powder are different from fresh one. For better results used powder should be mixed with the fresh one. Nevertheless, usage of the refreshed powder often leads to defects of manufacturing, such as very rough texture of the surface with the defects which are detectable without measurement instruments. Such defects are called "orange peel" and could be detected on the angled surfaces [12]. Structural element with angled surfaces is a part of the artifact (Fig. 3a). Functional angles are: between angled surfaces and the functional surface of the measurement artifact. The nominal angles are: 0.1 °; 0.5 °; 1 °; 5 °; 10 °; 15 °; 20 °; 30 °; 40 °; 45 °; 50 °; 60 °; 70ιǤ This structural element looks cumbersome, but it was designed taking into account properties of the measuring equipment. More traditional for such investigations are "open book" [13] structures, but they limit the application of the measuring equipment. There are number of cylindrical elements on the artifact which are represented as vertical and lateral, positive and negative (hollow) cylinders and hemispheres (Fig. 3b, 3e, 3h, 3k, 3l). a

g

b

h

c

i

d

j

e

k

f

l Figure 3. Functional surface of the artifact



They are intended for evaluation of position, cylindricity, roundness and perpendicularity of the reproduced structures. Lateral manufacturing resolution of the SLS machine depends on many parameters: one of which is spot diameter of the scanning laser on the powder bed. The smallest diameter is at the center of the manufacturing field and it increases closer to the edge of the powder bed. First of all it is a question of manufacturing strategy. But the artifact should help to evaluate the best lateral resolution for worst manufacturing conditions. One of the strategies should be the manufacturing of such artifact in different areas of the powder bed. The set of cylindrical structures (Fig. 3e, 3h) have diameter values nearly the same like the spot diameter of the scanning laser. The smallest cylinder has a diameter of 0.5 mm which allows to evaluate whether SLS machine able to reproduce one-point structure. Using of some periodical profile structures for testing of a surface measurement instruments is well known approach in manufacturing metrology. One of the good examples is using the Chirp standard [14]. Such structure has a periodical profile with scaled frequencies - the sequence of sinusoidal waves with special different wavelength values. Similar Chirp structure is located on the functional surface of the artifact Fig. 3i). The profile has a shape of a triangle periodic signal. There are 24 periods with the smallest value of 0.1 mm and the largest of 0.91 mm. The amplitude of the profile is constant and has the value of 0.1 mm. 2.3. Manufacturing of the testing artifact As it was mentioned above, after preparing a raw material, the manufacturing strategy should be developed. Nowadays this process to a great extent depends on the operator of SLS machine. There are many general recommendations which, nevertheless, are not always applicable for all types of materials or different parts’ design. In [15] were summarized experience together with manufacturing recommendations, but the development of normative documentation and standardization for AM is still in active phase. Some information about the manufacturing strategies, material and process preparation mentioned in the German standard VDI 3405 Part 2 [16]. But the mentioned standard regulates only quality assurance of the metal-based SLS process and there is no general regulation of the manufacturing strategy development. Orientation of the part inside the process chamber of SLS machine has a great influence. One of the important things is the orientation regarding the motion direction of the coating roller or blade. The long parallel contact of the coating roller with the sintering pool should be avoided as well as large sintered areas on the powder bed surface. SLS process creates three-dimensional parts, layer by layer from one of several powdered materials. Each layer of the workpiece is manufactured in the following way: system’s CO2 laser selectively heats up and fuses a layer of the powder to form a solid mass that matches the layer of the parts CAD design [17, 18]. The build temperature has to be above the crystallization temperature and under the melting temperature. For this reason a minimum temperature range for the melting range is desirable, i.e., the polymer should be fully melted upon reaching a temperature limit. The temperature interval between

92


the melting and recrystallization temperature should be set as large as possible. [18] However, polyamide 12 (PA12) is mostly employed because of its hydrophilic properties and further due to the big process window which offered to achieve reproducible manufacturing [19]. The quality of components, which are manufactured with SLS process, depends on various factors: shrinkage, shrink hole, material inclusions, fluctuations of the surface roughness and the functionality of the components [20]. For the current research was used PA12 (Type: PA 2200, EOS GmbH, Krailling, Germany). Powder was prepared in the following way: the mixture of 50 % new and 50 % aged powder was blended for 30 minutes in a rotary mixer at a revolution speed of 400 min-1 to reach a homogeneous basic material. Manufacturing of the artifact was done with the SLS machine DTM Sinterstation 2000 and the process had next parameters: layer thickness is 100 µm, laser power is 11 W, scanning speed is 279 mm/sec, and building chamber temperature is 175 °C. The first set of artifacts was designed with errors. The designed artifact was a solid body, so during the manufacturing process were presented large sintering areas. Since the temperature deviations after the manufacturing process, the workpiece was affected and the artifact with a curved body was produced (Fig. 4a – side view). Afterwards the design was changed: the artifact became hollow and we have added a lateral support structures for the functional surface of the artifact. Linear dimensions of the measurement artifact were the same: 100 mm x 100 mm x 30 mm. The DTM Sinterstation 2000 has the diameter of the powder bed around 254 mm and for tests were manufactured measurement artifact in original size and with scaling 1.5. During the manufacturing functional surface was parallel to the XZ- plane of the powder bed. Objects without scaling were manufactured the most successfully. The disadvantage of the

Scale||1.5:1 Scale|| 1:1


improved artifact is weakness of the construction caused by the cavities in the body. It affects the geometry of the part, especially scaled (Fig. 4b – side view). Nevertheless, the second set of artifacts has better quality of the functional surface. 3. MEASUREMENT RESULTS In order to measure the surface of the measurement artifact were used two systems: ATOS Compact Scan 2M – photogrammetric measurement system; Alicona Infinite Focus G4 – focus deviation measurement system. The artifact was contactless cleaned before measurements. As it was mentioned above, the best manufactured part has shape deviations. In order to inspect these deviations, as well as parameters of structural elements, were used software packages: PolyWorks and TalyMap Platinum. Measurement results of the whole artifact are unsatisfactory. The reason of deviations are obviously temperature changes during the cooling of the produced object. The presence of shrinkage should be taken into account. After the aligning of the measured data with the CAD model of the artifact, the color map of deviations was obtained (Fig. 5). In some areas the deviation of the geometry rising up to 1 mm, which indicates necessity to improve the design of the artifact and manufacturing strategy. Due to the reasons mentioned above there is no possibility to evaluate the run-out and position of features of the functional surface. But inspection of some features of the artifact is still important since its aim is to overview the ability of the machine to reproduce separate features of the functional surface. The first object of interest is structural element with angled surfaces (Fig 3). It is complicate to evaluate any measurement result because of the step effect during the manufacturing of surfaces with small angles and size. The smallest angle which could be detected was 1ι. Other surfaces were manufactured better but the standard deviation of angles manufacture is high (0.8ιሻ. DTM Sinterstation 2000 has a vertical resolution (Z axe direction) of 50 μm and lateral resolution of 750 μm. Structural elements with which is possible to confirm this values are the step height structures and the Siemens-Star.

a)

Scale||1.5:1 Scale|| 1:1

b) Figure 4. Manufactured parts


Figure 5. Deviation of the manufactured part

93


The central cylinder group (Fig. 3d) was manufactured with lower deviations (comparing to whole body). In order to determine the lateral resolution for SLS process it is possible to use well-known approach with the Siemens-Star, but taking into account 3 D manufacturing process: profile should be extracted using the circle; value of the profile depth should be close to the nominal value – 1 mm for the tested feature [11]. If the Siemens-Star is manufactured as separate object – it has better quality and the obtained diameter of the profile circle is 10.6 mm (when the element is the part of the artifact). It means the lateral resolution for this process is approximately 1.96 mm. The reason of such a worse lateral resolution is the pressure of the coating roller and small dimensions of the segments near the center of the Siemens-Star. Periodical topography can be observed near the center of the star (Fig. 6a). The smallest diameter with which the profile deviation is still detectable is 6 mm which gives us the lateral resolution of 1.2 mm. The measurement artifact gives another possibility to evaluate the lateral resolution of SLS machine. For this purposes the set of cylinders with different diameters are presented on the functional surface (Fig. 3h). Smallest cylinder has the diameter of 0.5 mm and it was successfully manufactured. The axe of the cylinder is orthogonal to the powder bed and it was one sintering point manufacturing. The body of the cylinder is distorted because it is single thin vertical structure but the next one with the diameter 1 mm has no distortions. It means that used SLS machine is able to produce parts with the declared lateral resolution. There are three sets of step height structures on the functional surfaces which are used to determine the vertical resolution of the SLS machine. Measurement results in Table 1 are also distorted because of the deviation of whole artifact’s body and the standard deviation for step height measurements achieves 0.3 mm. All structures are developed in accordance with [21].


TABLE I.

STEP HEIGHT MEASUREMENTS

Measurement results (mm) Nominal

Stair difference 50 μm

Nominal

Stair difference 10 μm

Height

Depth

Height

Depth

1.55

1.23

1.67

1.01

1.45

1.57

1.60

1.38

1.80

1.02

1.54

1.69

1.65

1.41

1.84

1.03

1.56

1.70

1.70

1.42

1.95

1.04

1.64

1.79

1.75

1.51

1.96

1.05

1.66

1.79

1.80

1.61

1.05

1.06

1.74

1.87

1.85

1.62

2.05

1.07

1.76

1.88

1.90

1.63

2.04

1.08

1.78

1.89

1.95

1.77

2.11

1.09

1.86

2.00

2.00

1.83

2.12

1.10

1.87

2.00

There are two structures for the step height and step depth (Fig. 3g, 3f) which have 10 stairs each with the difference in height (depth) of 50 μm. There is one structure (Fig. 3c) for both type of structures (height and depth) where stair have difference of 10 μm. But in this case it would be useful to overview the difference between the steps. Structures with the stair difference of 50 μm were manufactured with rising height and depth. It means that SLS machine is able to produce structures with the declared vertical resolution and high deviations which are caused by the distortion of the process parameters. The result for the structures with the stair difference 10 μm is not acceptable even to confirm declared vertical resolution. For better results the spacing between stairs should be increased. Comparing these step height measurements with the previous work [11] it should be noted that such structures as a part of the solid measurement artifact have better quality. 4. CONCLUSIONS

a)

b)

Figure 6. Measurements of structural elements


The necessity of using some measurement artifacts as well as development of the manufacturing strategies for different AM approaches has been discussed in number of publications. First standards regarding SLS process have a number of disadvantages and propose monitoring and investigation of the great amount of process parameters as a solution. Such approach is cost-intensive and could be an area of interest only for research institutes and manufacturers of these machines. At the same time some recommendations about preparing of the raw materials, CAD models, and orientation of the part in the powder bed are mentioned only in one part of the standard and only for metal based SLS process. The standardization process for AM should be more intensive in order to satisfy all needs of contemporary industry. Development of new standards should be provided with the experience of the research institutes and manufacturers of the AM machines. Development of a new design for the measurement artifact is caused by necessity to provide wider opportunities to evaluate the SLS manufacturing process. Presented design combines more structural elements for testing a manufactured surface’s quality, evaluating of

94


manufacturing resolution and providing computed tomography (CT) measurements. The first manufacturing of the artifact has showed a number of disadvantages in designed measurement artifact and was improved for the next set of workpieces. The functional surface of the artifact has lateral support structures which are located out of the measurement zones. Improved artifact is the hollow structure which is also provided with internal hollow measurement features. High distortion of the geometry nullifies the necessity in computed tomography measurements, because internal distortion is detectable without measurement equipment. At the same time the structural elements of the functional surface are inspected as separate features. The measurement artifact should be improved and investigated more profoundly.


[9]

[10]

[11]

[12]

ACKNOWLEDGMENT The authors want to thank the German Research Foundation (DFG) for funding the Collaborative Research Center 814 (CRC 814), sub-project C04. REFERENCES [1] [2] [3] [4] [5]

[6] [7]

[8]

VDI 3404:2009-12 ‘Additive fabrication - Rapid technologies (rapid prototyping) - Fundamentals, terms and definitions, quality parameters, supply agreements’. EU project SASAM, the support action for standardisation in additive manufacturing, 2014, ‘Roadmap for standardisation activities’, in http://www.sasam.eu/. VDI 3405 Part 1:2013-10, ‘Additive manufacturing processes, rapid manufacturing - Laser sintering of polymer parts - Quality control’. Buining, H., 2014, ‘Manufacturing trends. Standardisation for additive manufacturing’. Presentation at the IMS Workshop, Barcelona. Moylan, S., Slotwinski, J., Cooke, A., Jurrens, K., Donmez M., 2012, ‘Proposal for a standardized test artifact for additive manufacturing machines and processes’, Proceedings of the Solid Freeform Fabrication Symposium, Austin, TX, pp.902-920. DIN EN ISO 3167:2003-12, ‘Plastics - Multipurpose test specimens’. Wegner, A., Witt, G., 2013, ‘Ursachen für eine mangelnde Reproduzierbarkeit beim Laser-Sintern von Kunststoffbauteilen’(Reasons of low reproducibility for laser sintered workpieces), RTejournal - Forum für Rapid Technologie, vol. 2013 Scaravetti, D.,. Dubois, P., Duchamp, R., 2008, ‘Qualification of rapid prototyping tools: proposition of a procedure and a test part’, The


[13] [14] [15] [16] [17] [18] [19] [20]

[21]

International Journal of Advanced Manufacturing Technology, vol. 38 (7), pp.683-690. Weckenmann, A., Tan, O., Hoffmann, J., Sun, Z., 2009, ‘Practiceoriented evaluation of lateral resolution for micro- and nanometer measurement techniques’, Measurement Science and Technology, vol. 20. Campanelli, S.L., Contuzzi, N., Angelastro, A., Ludovico, A.D., 2010, ‘Capabilities and Performances of the Selective Laser Melting Process’, New Trends in Technologies: Devices, Computer, Communication and Industrial Systems, pp.233-252. Galovskyi, B., Fleßner, M., Loderer, A., Hausotte, T., 2013, ‘Systematic form deviations of additive manufactured parts - methods of their identification and correction’, Proceedings of the 11th International Symposium on Measurement and Quality Control., Krakau – Kielce, Poland. Yusoff, W.A.Y., Thomas, A.J., 2008, ‘The effect of employing an effective laser sintering scanning strategy and energy density value on eliminating “orange peel” on selective laser sintered part’, Proceedings of the International Association for Management of Technology.IAMOT. Castillo, L., 2005, ‘Study about the rapid manufacturing of complex parts of stainless steel and titanium’, TNO report with the collaboration of AIMME. Krüger-Sehm, R., Bakucz, P., Jung, L., Wilhelms, H., 2007 ‘Chirp calibration standards for surface measuring instruments’ Techisches Messen, vol.74(11), pp.572-576. Moylan, S., Slotwinski, J., Cooke, A., Jurrens, K., Donmez M., 2012, ‘Lessons Learned in Establishing the NIST Metal Additive Manufacturing Laboratory’, NIST TN - 1801. VDI 3405 Part 2:2013-08, ‘Additive manufacturing processes, rapid manufacturing - Beam melting of metallic parts - Qualification, quality assurance and post processing.’ Wohlers, T., 2014, ‘Wohlers Report 2014 – Rapid Prototyping – State of the Industry’. Gebhardt, A., 2012, ‘Understanding Additive Manufacturing’, Carl Hanser Verlag. Pham, D.T., Gault R.S.,1998, ‘A comparison of rapid prototyping technologies’. International Journal of Machine Tools & Manufacture, vol.38, pp.1257-1287. Harmann, W., Hausotte, T., Drummer, D., Wudy, K., 2012, ‘Requirements and constraints for the ues of optical measuring systems for in-line inspection additively manufactured parts’, RTejournal, vol.2012, Iss: 1. EN ISO 5436-1:2000, ‘Geometrical Product Specifications (GPS) Surface texture: Profile method; Measurement standards - Part 1: Material measures’.

95



Facilitating Consumer Involvement in Design R.I. Campbell, Design School, Loughborough University, UK, [email protected] Y. Ariadi, Design School, Loughborough University, UK, [email protected] M.A. Evans, Design School, Loughborough University, UK, [email protected]

Abstract— This paper reports investigations into the potential for consumers to actively design their own desired products and thereafter to endorse them for manufacture. This idea emerged in anticipation of the rapid growth of low-cost fabrication technology, particularly 3D Printing. Recent developments in 3D Printing have led to renewed interest in how to manufacture customised products, and specifically, in a way that will allow consumers to create bespoke products more easily. However, the entry point to 3D Printing is typically a 3D computer aided design (CAD) model, and most CAD systems are typically difficult for non-experts to use. Consequently, to make 3D Printing more accessible to consumers, design systems need to be developed that are as easy to operate as are the 3D Printers themselves. This research reports on the development of a Computer-aided Consumer Design (CaCoDe) system that is designed to simplify the CAD process for non-designers. The analogy is giving the user a few blades to use rather than a whole Swiss Army Knife. The software, which has been developed using a Rhino and Grasshopper platform, is an easy-to-operate design system, where consumers interact with the dimensional parameters of pre-designed templates through on-screen sliderbars and pick-and-drag mouse movements. To determine the potential for consumer-led design, a range of product design templates were implemented using the software and then evaluated by non-designers. The evaluation was undertaken using 40 individuals from a range of backgrounds and with a wide spread of ages. They were asked to use the software to design their own customised version of a product and then to evaluate its ease-of-use, specific user-interface features and the likelihood of them wanting to use it in future. The results showed that the majority of the individuals found the system to be userfriendly and understandable. The results were used to define the final version of the software and to make recommendations for future developments in this area. Keywords - design; consumer; shape manipulation; customisation

1. INTRODUCTION The main aim of this research was to determine the potential for consumers to become directly involved in product shape manipulation. This is a progression from the typical consumer interaction which is limited to product aspects such as options configuration or choosing colours. A key reason for this limitation is the inability for ordinary consumers to create or modify 3D models. A so-called Computer Aided Consumer Design (CaCODE) approach has been employed in this work, with a particular focus on designing products to be made using additive manufacturing (AM). The term “consumer design”


was coined because it refers to the conception, specification, design, or manufacture of products occurring with direct consumer input [1], a greater level of consumer involvement than is normally seen. The central requirement of CaCODE is that it must be relatively easy to use, since it will be operated by members of the general public with no existing CAD skills. A good analogy for CaCODE compared to conventional CAD is that the consumer is given a few blades to work with rather than an over-complicated Swiss Army Knife (Fig 1).

Figure 1. Swiss Army Knife analogy for conventional CAD

2. SOFTWARE DEVELOPMENT 2.1. Selecting a Development Platform The primary requirements of a CaCODE system are easeof-use (as already stated) and the ability to make changes to a 3D CAD model. This in turn leads to two components of the system, the user interface and the modelling kernel. There were four basic alternatives available for the development platform, a) use a conventional CAD package together with its own application programming interface (API), b) use a conventional CAD package with a special user-interface application, c) use a specially developed “easyCAD” system or d) develop a completely new system from scratch using a standard kernel such as ACIS. Options a) and d) were beyond the programming capabilities of the researcher so they were excluded. Exploring option c) led to some interesting investigations of the ODO software developed by Digital Forming [2]. This software uses a facet-based modeler to work on STL files. In the definition

96



stage, a designer works on an STL file of a product to create “features” and options that the consumer will be able to modify or select. This model is then published on the Internet were the consumer can access it, make changes to the product’s form or configuration (e.g. number of petal on a flower), and then ask for their design to be printed. The advantages of ODO are its ability to work with STL files and to create an Internet accessible model. The disadvantages are the time it takes to create the features and options and the inherent difficulties of working with a facet-based modeller compared to a conventional modeller. There are no standard features or parameters to work with and so everything must be defined by the user. This made the definition stage very time-consuming and so this development platform was rejected. Therefore, only one option remained, using a conventional CAD package with a special user-interface application. From previous research, the capability of using Grasshopper [3] to manipulate Rhino [4] CAD models was already known to be powerful and easy-to-use. Grasshopper is a graphical algorithm editor tightly integrated with Rhino’s 3-D modeling tools. It can be used to convert explicit Rhino models into parametric models, where parameter values can be typed in or selected by means of slider bars. These become the inputs to a procedural program that regenerates the Rhino model with new dimensions or other values. The program itself is created in a visually intuitive flow-diagram interface (see Fig. 2). Using Grasshopper, it is possible to generate a simple user-interface that can be used to directly modify the size and shape of a Rhino CAD model, in real-time. This is ideal for CaCODE and so this option was selected as the development platform.

Figure 2. Eample program created using Grasshopper interface

2.2. Developing the CaCODE system To keep the development and use of CaCODE simple, it was decided to cater for one particular product or application at a time. This would mean that the number of user inputs could be kept to a minimum, making the system easy enough to use so for someone with no prior CAD skills. Eventually, CaCODE variations were created for several products including plates, vases and beakers. All of these products were rotationally symmetrical, meaning that only a 2D profile had to be manipulated by the user. The aim of this research was to involve consumers in 3D design manipulation so what were effectively “2.5” dimensional components were considered too simple. Therefore, a slightly more complicated product was chosen as the vehicle for the first user trials of CaCODE. The product was a curved pen, as shown in Fig. 3. It was truly 3 dimensional and yet its shape could still be controlled fully using a reasonable number of parameters.


Figure 3. Curved pen design

The first stage of the CaCODE development process was to build a Rhino CAD model of the pen design. This was fairly straight-forward and only required the placement of a few points, four circular or elliptical profiles and a lofted surface. The designer then had to decide which parameters were to be fixed and which could be modified by the consumer. This is an important aspect of any consumer design system since there are certain elements of the design which must be “protected” for functional, safety or brand requirements [5]. The selected parameters were then given appropriate names to be used in two ways, firstly as variables inside the Grasshopper flowdiagram program and secondly, as slider bar inputs on the Grasshopper user-interface. The program was then constructed as a series of boxes either representing algorithmic procedures (e.g. calculations) or Rhino modelling functions. For example, X, Y and Z values were first used to generate a point at the top of the pen and this in turn was used with two radial values to generate an ellipse, which finally became part of the lofted exterior surface of the pen. Limits were set on the slider bars to ensure that no unreasonable values could be selected that would result in an unrealisable or self-intersecting model. The completed user-interface including graphics windows, is shown in Fig. 4. The consumer designer simply needs to move the sliders on the right and watch the pen design change in the three graphics windows.

Figure 4. Complete CaCODE interface for pen design

97



TABLE I.

3. USER TRIAL Following on from system development, the CaCODE software was presented to a reasonably representative sample of the general population and they were asked to design their own pens. Forty participants were recruited, none of whom had ever used a CAD system before. They were equally split between genders and their ages ranged from pre-teens to oversixties (see Fig. 5). The participants’ backgrounds ranged from research fellows through to homemakers. This represented quite a wide demographic range.

USERS’ EXPERIENCE OF SYSTEM Number of users

User Rating

Sliders

Click and drag

5 = enjoyable

13

9

4

7

15

3

19

15

2

-

1

1 = frustrating

1

-

Participants were then asked to rate the speed of learning the two systems on a scale of 1 to 5 where 1 was “very fast” and 5 was “very slow”. The results are shown in Table II. Here there is a marked difference between the systems with click and drag scoring much higher in the “very fast” category. It would appear that the majority of users found clicking on an object to change its shape and size quicker to learn than using slider bars, i.e. direct rather than indirect manipulation of parameters. TABLE II.

Number of users

Figure 5. Gender mix and age ranges of user trial participants

Prior to using the software, the participants were asked about whether they had ever wanted to make a consumer product or not and also whether they actually went on to make it or not. From these questions, three groups of participants were identified: Group 1 - 8 participants (20%) had at some time wanted to design a consumer product and then actually went on to design one Group 2 - 11 participants (27%) had at some time wanted to design a consumer product but had never actually done so Group 3 - 21 participants (53%) had never wanted to design a consumer product This result was initially disappointing as it seemed to indicate a less than half of the general public would be interested in consumer design. However, when looked upon with a “glass half full” attitude, it shows that there is actually great potential for consumer design to grow from its currently small base. Following this, the participants were introduced to the CaCODE software and asked to use it to create their own design of pen. They were actually asked to use two versions of the software, one with slider bars and the other were the pen parameters were modified through on-screen click and drag motions. One half of the participants used the slider bars first and the other half the click and drag version. 4. RESULTS Participants were first asked to rate their experience of using the two systems on a scale of 1 to 5 where 1 was “frustrating” and 5 was “enjoyable”. The results are shown in Table I. It can be seen that there was not a great deal of difference between the two systems, with the vast majority of users having a neutral to positive experience.


USERS’ SPEED OF LEARNING

User Rating

Sliders

Click and drag

5 = very fast

18

32

4

12

5

3

8

1

2

-

2

1 = very slow

2

-

Next, participants were asked rate how much they agreed with the statement “I found it is easy to change the shape” when using the two systems, again on a scale of 1 to 5 where 1 was “strongly disagree” and 5 was “strongly agree”. The results are shown in Table III. Once again, there is a marked difference between the systems with click and drag scoring much higher in the “strongly agree” category. It would appear that many users found clicking on an object to change its shape and size much more intuitive than using slider bars. TABLE III.

EASE OF MAKING DESIGN CHANGES Number of users

User Rating

Sliders

Click and drag

5 = strongly agree

11

25

4 = agree

17

11

3 = neutral

7

3

2 = disagree

5

1

1 = strongly disagree

-

-

Finally, the participants were asked some questions to see if using a consumer design system had possibly changed their attitude towards consumer products. The participants were asked if designing and/or making a product for themselves

98


would make them want to use it longer than a standard product. The number of users giving “yes”, “no” and “do not know” answers are shown in Fig. 6. The overwhelmingly positive response is an encouraging result in terms of the potential impact that Consumer design could have upon sustainability, i.e. a move away from “disposable products”. Some possible reasons for this from Mugge et al [6] were suggested to the 32 participants who responded “yes” and they were asked to agree or disagree with them. Their ratings are given in Table IV. All three reasons received strong positive answers.


manipulate until they reach a shape with which they are happy. As such, the system fits between “blank screen” CAD and configuration systems that allow the consumer to select from options but not to determine the shape of a product (see Fig. 7)

Figure 7. Where CaCODE fits into the spectrum of “user input”.

Figure 6. Users’ change of attitude towards consumer products TABLE IV.

REASONS FOR USING A PRODUCT LONGER Suggested Reasons Bonding with the product

They designed it themselves

The product expresses their identity

5 = strongly agree

14

16

24

4 = agree

12

16

13

3 = neutral

5

-

4

2 = disagree

1

-

1

1 = strongly disagree

-

-

-

User Rating

5. CONCLUSIONS This paper has reported work on a CaCODE system that enables ordinary people to get directly involved in product design. An important characteristic of the system is that an initial version of the product must be created by a designer, one who is proficient in CAD. This follows the philosophy that not everyone can be a designer [7], [8] and overcomes the “blank screen syndrome” that most consumers encounter when they are asked “what do you want your product to look like”? Instead, they are given a starting point product, which they can


Most of the participants who took part in the trial of the system found it to be user-friendly and understandable. They preferred click and drag operation as opposed to slider bars and believed that designing a product would make them want to use it for longer. All of these are important findings, which show consumer design could become a reality, provided the necessary tools are developed with the consumer’s needs firmly in mind. The results will used to help define future versions of CaCODE software. Current research issues being tackled at Loughborough University revolve around finding out which product characteristics consumers most want to modify, how personalization of a product can add value to it and the impact that such consumer-led design changes will have upon brand protection. REFERENCES [1] [2] [3] [4] [5]

[6] [7] [8]

M. Sinclair, “A Classification of Consumer Involvement in Industrial Design,” in Design School Research Student Conference (DeSReS), Loughborough University, 2011. http://www.digitalforming.com/ Accessed October 2014. http://www.grasshopper3d.com/ Accessed October 2014. http://www.rhino3d.com/ Accessed October 2014. M.A. Sinclair, R.I. Campbell, Y. Ariadi and M.A. Evans, “AM-enabled Consumer Design” in D.J. de Beer, and W. du Preez, (eds) Proceedings of 12th Annual RAPDASA Conference, Vanderbijlpark, South Africa, 2011. R. Mugge, J. P. L. Schoormans, and H. N. J. Schifferstein, “Emotional bonding with personalised products,” Journal of Engineering Design, vol. 20, no. 5, pp. 467-476, 2009. J. Duffy and A. Keen, “Can anyone be a designer?,” Fast Company, p. 116, Oct-2006. T. Wohlers, “Wohlers Talk: Most People Cannot Design,” 2008. [Online]. Available: http://www.wohlersassociates.com/brief03-08.htm. Accessed October 2014.

99


METALS IN ADDITIVE MANUFACTURING

CHAPTER 3.

Metals in Additive Manufacturing


100

1. Proceedings of 5th International Conference on Additive Technologies


In-Situ Property Improvements of Additive Manufactured Objects Using a CNC Integrated Pneumatic Hammer Dhirendra Rana, Department of Mechanical Engineering, Indian Institute of Technology Bombay, Mumbai, India, [email protected] K.P.Karunakaran, Department of Mechanical Engineering, Indian Institute of Technology Bombay, Mumbai, India, [email protected] Abstract— Residual stresses exist in metallic objects without the application of any service or external loads. These stresses can be devastating to stimulate the fatigue failure due to crack initiation. Hybrid Layered Manufacturing (HLM) is an Additive Manufacturing (AM) process for metallic objects using Metal Inert Gas (MIG) welding deposition. HLM objects are functionally hindered due to the preoccupation of residual stresses and porosity in it. To overcome these limitations, a new in house build in-situ property improvement technique using hammering is proposed in this paper.

Residual stress is defined as the stress resident inside a component or structure after all applied forces have been removed. A welded joint will contain high magnitude residual tensile stresses in the Heat Affected Zone (HAZ) adjacent to the weld. Conversely, the surface of induction hardened components may contain residual compressive stresses. Residual stresses influence fatigue life, dimensional stability, corrosion resistance and brittle fracture. Welding is more prone to residual stresses, the reasons of which are non-uniform thermal expansions and solidifications.

HLM process is capable of manufacturing most complex parts, but some manual operations are needed to make objects competent to conventionally manufactured objects. The authors describe the method of property improvement of metallic objects using an in-house hammer. After deposition of each layer, hammering operation is performed.

Due to the enormous heat generated during the welding process, the workpiece develops high amount of residual thermal stresses. Residual stresses often have a negative effect on the mechanical properties of a workpiece leading to service failure. Thus, they can adversely affect the life of the component manufactured by Hybrid Layered Manufacturing (HLM) process. Also they are frequently the cause for the occurrence of undesirable dimensional changes in the workpiece upon machining.

It was observed that the density of the sample after hammering (2.68 g/cm3) was almost equal to the original density of the sample which is 2.69 g/cm3. The samples produced without hammering showed the maximum percentage of closed porosity of 20.13% and 17.39%; whereas the closed porosity after hammering was found 2.78% and 0.08% measured in step over and weld direction respectively. HLM realizes near net shape of complex objects, further post processing namely machining and hammering is required. Keywords- Hammering; Additive Manufacturing (AM); Hybrid Layered Manufacturing (HLM)

1. INTRODUCTION Residual stresses are generated in metals by any operation which brings a non-uniform change in shape or volume throughout a work-piece. Such a change can be affected by operations that cause local plastic flow like heat treating, quenching, welding, casting, forming, machining, grinding. The stresses may develop directly by expansion or contraction, and may also result from changes in volume and coefficient of expansion that accompany metallurgical phase transformations.


Stresses can also be characterized as normal stresses that act perpendicular to the face of a material and shear stresses that act parallel to the face of a material. There is a total of 6 independent stresses at any point inside a material (3 normal and 3 shear stresses). The total stress experienced by the material at a given location within a component is given by the sum of the residual stress plus the applied stress. Therefore, knowledge of the residual stress state is important to determine the actual loads experienced by a component. In general, compressive residual stress in the surface of a component is beneficial. 1.1. Types of residual stresses The residual stresses can be categorized as: Type-I: Macro-stresses occurring over distances that involve many grains within a material. Type-II: Micro-stresses are caused by differences in the microstructure of a material. It occur over distances comparable to the size of the grain in the material. It can either occur in single-phase materials due to the anisotropic behavior of individual grains, or can occur in multi-phase material due to the presence of different phases.

101


Type-III: It exist inside a grain as a result of crystal imperfections within the grain. It can open cracks and increase crack propagation. The total residual stress at a given location inside a material is the sum of all 3 types of stresses. 1.2. Effect of residual stresses Solidification of weld bead from liquid phase to solid phase causes volumetric contraction. As a result of which the base metal, on which deposition is being done, gets stressed more and more after each layer. This results in the warping of the component. In the case of extreme warping it becomes mandatory to chop off the fixture bolts to get the final product out. Figure 1 shows the factors which imply the need of property improvement.

(a)


a. Post weld heat treatment (PWHT): PWHT means heat treating the workpiece after the welding process. Work or strainhardened metals exposed to the intense localized heat of welding tend to recrystallize and soften in the Heat-Affected Zone (HAZ). The admixture and deposited metal do not suffer recrystallization and usually remain as strong as the base metal [4]. Thus, while dealing with work or strain-hardened steel, failures usually happen in the HAZ. Martensite being brittle facilitates cracking, so to avoid martensite formation slower cooling rates are employed. Consequently, post-weld brittleness can be reduced by preheating the weld to slow the cooling rate and post-heating of the weld to facilitate slower cooling may also be necessary. The benefits of PWHT are: • It softens or tempers martensite or bainite that has formed in the HAZ. • It relieves stresses that can lead to cracking. • Proper heat treatment can change grain size, ductility, hardness, toughness, tensile strength.

(b)

Figure 1. Impeller with defects : (a) Impeller with warping, (b) Impeller with porosity defects

The residual stresses caused during welding are a major source of crack propagation. Contraction of metal along the length of the weld is partially prevented by the large adjacent body of cold metal. Hence, residual tensile stresses are set up along the weld [1]. Past work on modelling and analysis of weldments showed high residual stresses in welding [2, 3]. Fatigue is a surface related phenomenon, as the fatigue cracks usually initiate at the surface and grow from there into the material. Surface hardening process such as hammering is used to improve fatigue properties. The in-situ pressing of layers work hardens the surface layer and induces compressive residual stresses. The residual stress acts as an applied mean stress and a compressive residual stress will therefore retard fatigue crack initiation and growth. The work hardening results in an increased dislocation density, which hinders dislocation movements due to the fatigue load and suppresses localized plastic deformation which is a starting feature for crack initiation. 1.3. Stress relieving methods As residual stresses can adversely affect the performance of a component, it is important to reduce the magnitude of these stresses. Figure 2 lists the various stress reliving processes. The stress relieving processes can be mainly categorized into two groups based on the metallurgical changes.

Figure 2. Various stress relieving processes

b. Parallel heat welding: Parallel Heat Welding (PHW) reduces the residual stresses in weldments [5]. In the PHW process, a pair of parallel heating torches is attached on both sides of welding torch. These torches help in raising the temperature of the base plate. Thermal stresses in weldments occur due to thermal shock between the base plate and the weldments. The PHW of the base plate will reduce the thermal unevenness, thereby the residual stresses will be reduced.

Some of the techniques for stress relieving are discussed below.

1.3.2. Cold working Plastic deformation, which is carried out in a temperature region below the recrystallization temperature is called cold working. Under load, the grain size decreases with strain at low deformation. Stress relieving thorough cold working methods can be mainly divided into three different approaches:

1.3.1. Hot working Plastic deformation, which is carried out in a temperature region above the recrystallization temperature is called hot working. Stress relieving thorough hot working methods can be mainly divided into two different approaches.

a. Mechanical stress relieving (MSR): It is the method of enforcing an external load on the welded structure where a residual stress exists. Being superposed with the external load, the highly stressed part starts to behave plastically because the region cannot sustain the external load anymore. Consequently,


102


the residual stress can be reduced when the external enforcing load is removed. Some of the major methods are: Compressive mechanical loading is a method in which a compressive load is applied on the weldments. The residual stress relaxation behaviour, for the MSR treatment, is affected by the magnitude and direction of the mechanical load [6]. MSR treatment can decrease the tensile stress peak value of axial residual stresses and hoop residual stresses on the outer surface. Moreover, the peak value of the compressive residual stresses is reduced quickly, making the residual stress distribution more uniform [7]. Shot peening is a cold working process in which the surface of a part is bombarded with small spherical shots. Each piece of shot striking the material acts as a tiny peening hammer, imparting to the surface a small indentation or dimple. For the indentation to be created, the surface fibers of the material must be yielded in tension. Below the surface, the fibers try to restore the surface to its original shape, thereby producing a hemisphere of cold-worked material highly stressed in compression. Overlapping indentations develop an even layer of metal with residual compressive stress. It is well known that cracks will not initiate or propagate in a compressively stressed zone. The maximum compressive residual stress produced at or under the surface of a part helps in increasing resistance to fatigue failures, corrosion fatigue, stress by corrosion cracking, hydrogen assisted cracking, fretting, galling and erosion caused by cavitation [8]. Ultrasonic Impact Treatment (UIT) is a method in which residual stress improvement is achieved by using an ultrasonic wave vibration with pin impact. The UIT tool is smaller, lighter and easier to handle with less shaking and noise. UIT induced compressive stresses is comparable in magnitude to that achieved through shot peening. The depth of compressive stress layer due to UIT was about double the depth of shot peening [8]. Based on these results, it was suggested that UIT may have more beneficial effect on crack initiation life than shot peening. Laser Peening (LP) is a stress relieving process in which a laser beam is pulsed upon a metallic surface. A planar shock wave is produced that travels through the workpiece and plastically deforms a layer of material. LP is a more controlled and precise process that may provide a more robust fatigue enhancement. Various studies conducted showed that laser peening resulted in a substantial reduction in fatigue crack growth compared with the as welded condition and unwelded base material [9, 10]. b. Vibratory stress relieving: Induced vibration energy is used for stress reliving in weldments. By the application of the Vibratory Stress Relieving (VSR) obtain three primary benefits in metals are obtained [11, 12]. • Reduction in distortion during machining • Reduction in distortion over time • Reduction in weld cracking over time (increased service life) Based on the vibration frequency, vibration loading can be divided into two categories:



Harmonic vibration load process accomplishes stress relief by inducing a mechanical energy into a workpiece through harmonic vibration. An optimum energy level, that will cause stress relief, can be achieved by providing vibration. When the frequency of vibration is increased, the metal dissipates the induced energy through internal friction and results in lower amplitudes. The amount of energy being dissipated by the metal is its stress-relief potential. This dissipated energy reaches a maximum near the leading portion of the harmonic curve which is the optimum stress-relief vibration frequency. Beyond this range, the metal component cannot dissipate the induced energy and responds with a violent reaction (higher amplitudes), which is usually observed as bouncing, with high noise levels. This process of vibration stress relieving through harmonic vibrations is also available as a commercial process called “Meta-Lax” developed by Bonal Technologies, US. Random vibrational load is imposed during welding [13]. Residual stress relieving occurs mainly due to plastic deformation near the bead. Two kinds of random vibration, white noise and filtered white noise are used. c. Magnetic stress relieving is a method of using magnetic behavior for reducing stresses is a new and quick stress reducing method [14]. Magnetic treatment can increase the service life of machine tools by residual stress relief. Interlayer stress relieving was not useful in HLM as the influenced depth is face milled. If it is done after face milling, its effect is lost during the deposition of the next layer. Hence, only post-weld heat treatment is preferred in HLM. 2. HYBRID LAYER MANUFACTURING (HLM) PROCESS HLM realizes shapes using Gas Metal Arc Welding (GMAW) deposition. Such additively manufactured matrix inherently has porosity and residual stresses. Majority of the applications require homogeneous and dense structures. It is possible to achieve the desired density and dimensional stability by the appropriate selection of the HLM process parameters namely wire feed rate, welding current (Karunakaran et al. 2012). HLM is a Rapid Manufacturing (RM) process that combines the best features of CNC machining (subtractive method) and Rapid Prototyping (additive method) approaches. This process makes near-net shape of the object to be built; the near-net shape is then finish machined subsequently. Near-net shape building and finish machining happening at the same station is the unique feature of this process. This two-level approach focuses on material integrity during material addition and geometric quality during material subtraction. One may subject the near-net shape to stress relieving or heat treatment as required before finishing. Time and cost saving of this process can be attributed to reduction in NC programming effort and elimination of rough machining. It is envisioned as a low cost retrofitment to any existing CNC machine for making metallic objects without disturbing its original functionalities. Customized software generates the NC program for near-net shape building.

103


HLM is a 3-/5-axis CNC machine fitted with a GMAW deposition system as shown in figure 3. Existing technology in AM has enabled fabrication of fully functional non-metallic objects. But the AM of metallic parts is still in developing phase and fully functional metallic parts are yet to come. This paper discuss the property improvement of metallic AM parts so that they can endure the challenges same as conventionally manufactured parts.


Geometric Processing CAD Model, STL Format Slicing, CNC Toolpath

Building Near -Net Shape MIG Weld Deposition Surafce Milling

Stress Relieving

Finish Machining

Figure 4. HLM process flowchart

(a)

(b)

2.2. Organization of paper In this paper authors give introduction to various stress relieving methods and the HLM process. The current work focuses on property improvement by eliminating porosity and residual stresses with a CNC integrated pneumatic hammer. Authors observed mapping of residual stresses and the density variation in HLM components. 3. STRESS RELIEVING USING CLAMPING AND UNCLAMPING METHOD

(c)

(d)

Figure 3. Integration of HLM with conventional CNC machines: (a) 5 Axis CNC machine, (b) Close view of 5 Axis CNC HLM unit, (c) 3-Axis CNC machine, (d) Close view of 3-Axis HLM unit

The deposition rate and layer thickness of HLM are of the competing laser/electron beam based processes (100 g/min against 2-8 g/min; 2mm against 0.04-0.1mm). These deposition rate and layer thickness can be achieved by a firm clamping mechanism during deposition and machining. Since the substrate is mechanically constrained, extensive residual stresses are built up under such severe cyclic thermal conditions. When the mechanical constraints are removed, these stresses release themselves in the form of distortions. While the residual stresses shorten the fatigue life, the distortions lead to dimensional inaccuracies. In other words, residual stresses and distortions are the two interchangeable undesirable evils, the former more harmful as the latter can be absorbed into higher machining allowance. Therefore HLM is modified to minimize residual stresses by permitting them to be relieved or distributed on a wide area. This modification was done by integrating a pneumatic hammer to 3-axis CNC machine. 2.1. Methodology of HLM The stages involved in building the metallic object using HLM is shown in figure 4. a. Geometric processing b. Building the near-net shape c. Stress relieving d. Finish machining


Residual stresses are developed during weld-deposition due to the high amounts of non-uniform heat input. These stresses can be relieved either after every few layers (inter-layer stress relieving) or after the realization of the near-net shape. As HLM is a slow process, the stress relieving strategies will influence time and cost. Peening, vibration and heat treatment were identified as the three broad methods suitable for interlayer stress relieving. Vibration is actually stress redistribution rather than relieving. Peening too does not relieve the stresses, but only adds a compressive layer on the surface that arrests the crack propagation. Thus, complete stress relieving takes place only in heat treatment due to recrystallization. Hence, it is concluded that inter-layer stress relieving is redundant in HLM and only post-weld heat treatment is preferred. The reduction of heat input and management of its manifestations as follows: Reducing the heat input at the source: To reduce the heat input during weld-deposition, Cold Metal Transfer (CMT welding machine manufactured by Fronius was chosen to replace the older generation pulse synergic welding machines. As the name suggests CMT uses considerable less power than the conventional GMAW processes, thus reducing the overall heat input. Unconstrained substrate during weld-deposition: The residual stresses and distortions are the two interchangeable aspects of the non-uniform heat input. If the distortion is prevented by rigid clamping, it may lead to internal stresses. Hence it is preferable to deposit in unclamped condition so as to allow warping and surface mill in clamped condition. The distortion is absorbed into the machining allowance. This arrangement for clamping or unclamping is shown in figure 5 (a) and (b). The location pins with conical head ensure the positional accuracy. During milling the studs are pulled down by the pneumatic pads to withstand the face milling load. Figure 5 (c) and (d) shows this dynamic clamping in action. This arrangement is designed such that although it allows for the warping of the blank, it minimizes the positional dislocation.

104



hammer will vibrate at high frequency resulting in higher force. This may result in cracks on the sample and will be noisy for the operator. Optimizing percentage level for the process being performed is important for producing the desired effect on the surface. (a)

(b)

(c)

(d)

Figure 5. Dynamic clamping: a) Diaphragm actuator used for dynamic clamping, (b) Clamping devise mounted onto the table, (c) Soft clamping before pneumatic activation, (d) Rigid Clamping after pneumatic activation

(a)

(b)

4. PROPERTY IMPROVEMENT USING PNEUMATIC HAMMER The peening of layers by the CNC integrated pneumatic hammer is a hot working process. It is used to produce a compressive residual stress layer and to modify mechanical properties of HLM components. It entails impacting a surface with a round tip hammer with force sufficient to reach plastic deformation. Hammering on a surface spreads it plastically, causing changes in the mechanical properties of the surface. Depending on the part geometry, part material, hammering intensity, hammering coverage, hammering can increase fatigue life. Plastic deformation induces a residual compressive stress in a hammered surface, along with tensile stress in the interior. Surface compressive stresses confer resistance to metal fatigue and to some forms of stress corrosion. The tensile stresses deep in the part are not as problematic as tensile stresses on the surface because cracks are less likely to initiate in the interior. 4.1. Process description Different design options were taken into consideration during designing the holding fixture to the CNC machine. Figure 6 shows different arrangements for holding the pneumatic hammer in position and options to mount the assembly on the machine. The hammer was disassembled during milling and welding to avoid collision between the hammer and the work piece. The compressive residual stress is produced by the transfer of force from a vibrating hammer into the surface of a material with the capacity to plastically deform. The mechanics of the hammering involve properties like shape, and structure; as well as the properties of the work piece. Factors for the force transfer control of the pneumatic hammer are air pressure, the vibrating frequency, impact force and work piece properties. The percentage of the surface indented, is subject to variation due to the step over increment. Processing the surface with a series of overlapping passes improves percentage of indentation. Maintaining the distance between the hammer tip and the top surface of the object is an important parameter. This distance influences the force applied during hammering, if the distance between the tip of the hammer and the sample is too short the


(c)

(d)

Figure 6. Design alternatives for attaching the pneumatic hammer with the CNC machine: (a) First design option, (b) Housing and clamping for hammer, (c) Drilled holes to attach with machine, (d) Circular track to provide circular motion

The distortion is prevented by rigid clamping but it may lead to internal stresses. Hence it is preferable that peening should be done on each layer so as to allow the accumulated residual stress to be removed. The distortion that is occurred during deposition and peening, is absorbed into the machining allowance. Stresses in HLM can be relieved by in-situ hammering at the end of each layer, releasing the mechanical constraints during deposition and a post-build furnace treatment. Experiment was carried out for the first strategy. Figure 7 shows the experimental setup used for hammering operation. Figure7 (a) shows the 3 axis machine attached with hammer. Steps of hammering process are as follows, Weld-deposition: The required layer is built by moving the torch along the tool paths, thus depositing the material in the required places. The area-filling may follow a contour parallel or direction-parallel pattern. Normally the slices are horizontal. However, non-planar deposition is used in repair activities as well as building complex objects invariable axis. Surface milling: This step removes the scallops and ensures the desired layer thickness. It also removes the scale of the previous layer so as to provide a nascent surface for the weld-deposition of the next layer. In planar deposition, simple face milling is adequate to achieve this. In non-planar deposition, conformal milling using a ball end mill will be required.

105



Where,

Hammering: Residual stress are developed during welddeposition owing to the high amounts of non-uniform heat input. These stresses can be relieved after every few layers (inter layer stress relieving) by using the pneumatic hammer. Pneumatic hammer moved along the torch tool path. Also these stresses can be relieved after the manufacture of the near-net shape.

P = .07 kg/mm², D = 24mm, F = (.07×π×24²)/4 = 316.67 kg-F = 3.1 KN 4.2. Experiments results The following experiments were performed to observe the changes on the material after and before pneumatic peening.   

(a)

(b)

(c)

(d)

Figure 7. Experimental setup and sequence of operations: (a) 3-axis CNC machine with hammer, (b) Step 1: Weld deposition of each layer, (c) Step 2: Milling of each layer to get required layer thickness, (d) Step3: Hammering every layer after milling operation)

Two samples were produced one without pressing the layers and the other with in-situ pressing of each layer after deposition. Each layer was subjected to a compressive load of 3 KN using a numerically controlled hydraulic press. Measurements of residual stresses were carried out by X-Ray Diffraction. When the deposition was done with in-situ pressing of each layer, residual stresses were found to be significantly less. The in-situ peening of layers results a change from a complete tensile stress to a compressive stress. The force applied during hammering was calculated using the basic formula of single acting pneumatic cylinder. F = PA = P×π×D²/4 Where, F = force applied in kg-F P = Gauge pressure in kg/mm² D = inner diameter of the cylinder Now, F = P×π×D²/4


Variations in residual stress Variations in density Changes in porosity

4.2.1 Variations in residual stress Any mechanical and thermal activities that lead to residual stress have been studied. It was observed that high residual stresses remain in the component and lead to surface cracks. To solve this problem the in-situ peening has been applied. Two samples of Al-Si were fabricated using HLM with and without hammering of each layer of the sample. Two samples were prepared for surface residual stress testing and then same sample used for sub-surface residual stress after and before peening. As shown in table 1 a surface and sub-surface compressive residual stress was measured using x-ray diffraction. TABLE I.

Sample name

VARIATIONS IN SURFACE AND SUB-SURFACE RESIDUAL STRESS AFTER AND BEFORE PEENING

Residual Stress Testing position

Normal Stress (Mpa)

S1

Surface residual stress before peening

S2

Surface residual stress after peening

-20.9

Sub-surface Residual stress before Peening Sub-surface residual stress after peening

+59.9

S3 S4

+123.3

-19.6

Remark

Highly tensile, surface cracks Compressive residual stress, arrests surface cracks Accumulated residual stress Compressive residual stress, arrests surface cracks

It can be easily observed from table 1 that surface residual stress before peening is completely tensile (123.3 MPa) and thus leading to surface cracks. However the residual stress after peening reduced drastically to a completely compressive residual stress of -20.9 MPa which helps in arresting the initiation of surface cracks. Similarly the sub-surface residual stresses before and after peening showed tensile and compressive residual nature respectively. Figure 8 shows the variation of surface and sub-surface residual stress. It is important to note that the measured subsurface residual stress and surface residual stress after peening of each layer for sample 2 and 4 is almost equal. It can be

106



concluded that both surface and sub-surface residual stresses are relaxed due to the peening load.

During experiment in measuring the density of the sample, open and closed pores were taken into consideration. The following equations are driven and used in measuring the density of the sample by considering the volume fraction of open and closed pores.

𝐕𝐕𝐕𝐕𝐕𝐕𝐕𝐕𝐕𝐕𝐕𝐕 (𝐜𝐜𝐜𝐜𝟑𝟑 )

=

𝑾𝑾𝒅𝒅𝒅𝒅𝒅𝒅 − 𝑾𝑾𝒘𝒘𝒘𝒘𝒘𝒘 (𝒈𝒈) 𝐅𝐅𝐅𝐅𝐅𝐅𝐅𝐅𝐅𝐅 𝐃𝐃𝐃𝐃𝐃𝐃𝐃𝐃𝐃𝐃𝐃𝐃𝐃𝐃 (𝒈𝒈⁄𝒄𝒄𝒎𝒎𝟑𝟑 )

(1)

During the use of a surfactants or hot boiling water allows penetration of surface porosity for more accurate measurements, however, it must be noted that the temperature can affect the density of both the material and the fluid thus altering the final measured values. Furthermore, the wire supported the material to be measured must be very fine, otherwise error can be introduced into the wet weight of the piece in question. The experimental procedure during density measurement was explained as follows. Measure the dry weight of material


𝑽𝑽𝑽𝑽𝑽𝑽𝑽𝑽𝑽𝑽𝑽𝑽 𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇 𝒐𝒐𝒐𝒐 𝒐𝒐𝒐𝒐𝒐𝒐𝒐𝒐 𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑

=

𝑫𝑫𝑫𝑫𝑫𝑫 𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘 𝑫𝑫𝑫𝑫𝑫𝑫 𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘 − 𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺 𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘

(4)

𝑶𝑶𝑶𝑶𝑶𝑶𝑶𝑶 𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑 𝑽𝑽𝑽𝑽𝑽𝑽𝑽𝑽𝑽𝑽𝑽𝑽 𝒐𝒐𝒐𝒐 𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔

(5)

𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪 𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑 𝑽𝑽𝑽𝑽𝑽𝑽𝑽𝑽𝑽𝑽𝑽𝑽 𝒐𝒐𝒐𝒐 𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔

𝑽𝑽𝑽𝑽𝑽𝑽𝑽𝑽𝑽𝑽𝑽𝑽 𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇 𝒐𝒐𝒐𝒐 𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄𝒄 𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑 =

(6)

Based on the above formulas the open and close porosities are calculated and tabulated as shown in table 2.

MEASURED DENSITIES AND PERCENTAGE OF POROSITY WITH AND WITHOUT PEENING THE LAYERS

Manufacturing Method

Closed porosity (%)

TABLE II.

Hamm -erring

Test direction

Open porosity (%)

The Archimedes principle was applied to compare the density of the object before and after peening of each layer. Once the dry weight (mass) is obtained, the submerged weight is measured. The weight loss of the material is then determined and subsequently combined with the density of the fluid to determine the volume shown in Equation 1.

=

(3)

density

HLM has inherent defects such as porosities and nonuniform microstructure, which derives the need of post processing of HLM objects. Density is linked with many factors, including method of densification, shape and size of the part, particle size, its compressibility, and rate of hammering. Density can influence mechanical strength, hardness, electrical conductivity, and magnetic and gas permeability.

𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹𝑹 𝒅𝒅𝒅𝒅𝒅𝒅𝒅𝒅𝒅𝒅𝒅𝒅𝒅𝒅 𝒐𝒐𝒐𝒐 𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔

𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺 𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘 − 𝑫𝑫𝑫𝑫𝑫𝑫 𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘 𝛒𝛒𝛒𝛒

Vol. of closed pores (1E-3)

4.2.2 Variations in density and porosity Density is a measure of the atoms that make up the material and crystallographic packing configuration. The density of a material can be changed by changing the crystallographic packing even when the atom makeup of a material remains constant. Density measurement is a key tool in understanding the densification behaviour of materials via different processing routes.

=

(2)

Vol. of open pores

Figure 8. Measured surface and sub-surface residual stress: (a) Surface residual stress without hammering, (b) Surface residual stress with hammering, (c) Subsurface residual stress without hammering, (d) Subsurface residual stress with hammering

𝑽𝑽𝑽𝑽𝑽𝑽𝑽𝑽𝑽𝑽𝑽𝑽 𝒐𝒐𝒐𝒐 𝒐𝒐𝒐𝒐𝒐𝒐𝒐𝒐 𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑𝒑

𝑫𝑫𝑫𝑫𝑫𝑫 𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘 − 𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺𝑺 𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘𝒘 𝛒𝛒𝛒𝛒

Vol. of sample

(d)

=

soak wt.

(c)

𝑽𝑽𝑽𝑽𝑽𝑽𝑽𝑽𝑽𝑽𝑽𝑽 𝒐𝒐𝒐𝒐 𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔

Sus wt. of sample

(b)

Dry wt.

(a)

first to obtain to the mass of the part. After the dry weight has been accurately obtained, the part was then be placed in the holder and submerged in the water bath. The submerged or wet weight of the part was measured to determine the volume. The submerged part should be subsequently re-weighed five times to verify the validity of the results and generate an average wet weight value. The wet weight can then be used in the Archimedes equations to generate the density in grams per cubic centimeter.

1

Yes

Stepover

2.47

1.52

2.47

0.94

0.002

26.3

2.6

0.3

2.8

2

Yes

Along

3.35

2.10

3.35

1.25

0.004

0.94

2.6

0.3

0.0 8

3

No

Stepover

1.53

0.80

1.55

0.73

0.015

147. 3

2.0

2.1

20. 1

4

No

Along

1.20

0.64

1.21

0.55

0.014

97.0 4

2.1

2.5

17. 3

Sample name

It was observed that the density of the sample after hammering was almost equal to the original density of the sample which is 2.69 gm/cm3. However the density of the sample produced without peening was found to be less than the original one. The sample produced without hammering, tested in stepover direction and along the weld directions showed the maximum percentage of closed porosity of 20.13 % and 17.39 % respectively. The percentage of closed porosities (2.80 %) is high relative to the open porosities (0.3 %) in the sample built with peening of each layer and tested in stepover direction. The sample tested along the weld

107



direction the percentage of open porosities is higher than that of closed porosity, which was 0.3% and 0.08% respectively.



4.3. Advantages and limitations of the process The hammer integrated with CNC machine is in working after deposition. The features of the hammering process can be listed as,





4.3.1. Advantages Improves the feasibility of HLM for realization of objects for real world application like composite Dies.



It provides strong bond between layers and ensures better material integrity.



It improves the fatigue strength of components. 4.3.2. Limitations

  

The peening load may transfer to the machine bearings and may cause problem to machine performance. This process may not applied to complex shapes with internal cavities. Might be a challenge to integrate with 5-axis CNC machine, a complex and rigid support mechanism may be required.

5. CONCLUSIONS Hammering, vibration and heat treatment were identified as the three broad methods suitable for inter-layer stress relieving. Vibration is actually stress redistribution rather than relieving. Peening too does not relieve the stresses, but only adds a compressive layer on the surface that arrests the crack propagation. Thus, complete stress relieving takes place only in heat treatment due to recrystallization. Hence, it was concluded that inter-layer stress relieving is redundant in HLM and only post-weld heat treatment is preferred. In addition, attention should be given to reduction of heat input and management of its manifestations. Based on the experimental results, authors drawn the claims that, hammered objects have 70 % less open porosity than non-hammered objects. The density of the objects produced with hammering is 25 % more than non-hammered objects. Authors claims that, in- situ hammering process can be used for: 

Increase the density of weld deposited layers, results in compacting the grains of the metal.


To relieve residual stress from each layer to distribute the accumulated residual stress. To eliminate internal porosity and voids in each layer, improved mechanical properties.

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

[9] [10] [11] [12] [13] [14] [15]

Lin, Y.C., and Chen, S.C. 2003, ‘Effect of residual stress on thermal fatigue in a type 420 martensitic stainless steel weldment’, Journal of Materials Processing Technology, vol. 138 (1-3): 22-27. Sun, M.C., Sun, Y.H., and Wang, R.K. 2004, ‘The vibratory stress relief of a marine shafting of 35# bar steel’, Materials Letters, 58 (3-4): 299303. Sunar, M., Yilbas, B., and Boran, K. 2006, ‘Thermal and stress analysis of a sheet metal in welding’, Journal of Materials Processing Technology, 172(1):123-129. Hrivnák, I.1985, ‘A review of the metallurgy of heat treatment of welded joints’ International Journal of Pressure Vessels and Piping, 20(3): 223237. Lin, Y.C., and Chou, C.P. 1995, ‘A new technique for reducing the residual stress induced by welding in type 304 stainless steel’, Journal of Materials Processing Technology, 48 (1-4): 693-698. Yang, Y., and Lee, S. 1997, ‘A study on the mechanical stress relieving in a butt-welded-pipe’, International Journal of Pressure Vessels and Piping, 73(3):175-182. Xu, J., Chen, L., and Ni, C.2007, ‘A study on the mechanical stress relieving and safety assessment without post-weld heat treatment’, Materials Science and Engineering: A, 443(1-2): 107-113. Cheng, X., Fisher, J. M., Prask, H.J., Gnäupel-Herold, T., Yen, B.T., and Roy, S. 2003, ‘Residual stress modification by post-weld treatment and its beneficial effect on fatigue strength of welded structures’, International Journal of Fatigue, 25(9-11): 1259-1269. Luong, H., and Hill, M.R.2008, ‘The effects of laser peening on highcycle fatigue in 7085-T7651 aluminum alloy’, Materials Science and Engineering: A, 477(1-2): 208-216. King, A., Steuwer, A., Woodward, C., and Withers, P. J. 2006, ‘Effects of fatigue and fretting on residual stresses introduced by laser shock peening’, Materials Science and Engineering: A, 435-436: 12-18. Aoki, S., Nishimura, T., Hiroi, T., and Hirai, S.2007, ‘Reduction method for residual stress of welded joint using harmonic vibrational load’, Nuclear Engineering and Design, 237(2): 206-212. Warren, A., Guo, Y., and Chen, S. 2008, ‘Massive parallel laser shock peening: Simulation, analysis, and validation’, International Journal of Fatigue, 30(1): 188-197. Aoki, S., Nishimura, T. and Hiroi, T. (2005), ‘Reduction method for residual stress of welded joint using random vibration’, Nuclear Engineering and Design, 235(14): 1441-1445. Lu, A.L., Tang, F., Luo, X.J., Mei, J.F., and Fang, H.Z.1998, ‘Research on residual-stress reduction by strong pulsed magnetic treatment’, Journal of Materials Processing Technology. 74(1-3):259-262. Karunakaran, K.P., Bernard, A, Suryakumar, S., Dembinski, L., Taillandier, G. 2012, ‘Rapid manufacturing of metallic objects’, Rapid Prototyping Journal, 18:264–280

108



Simulation of Material State Change and Thermal Distribution in Electron Beam Melting (EBM) Manuela Galati, DIGEP, Politecnico di Torino, Torino, Italy, [email protected] Alessandro Salmi, DIGEP, Politecnico di Torino, Torino, Italy, alessandro [email protected] Eleonora Atzeni, DIGEP, Politecnico di Torino, Torino, Italy, [email protected] Luca Iuliano, DIGEP, Politecnico di Torino, Torino, Italy, [email protected]

Abstract—Process simulation could be a tool for decision-making and process optimization since virtual analysis facilitates the possibility to explore ‘‘what if’’ scenarios. In order to enhance the knowledge in this field, this study proposes a three-dimensional FE thermal model to simulate the transient heat transfer during EBM process. Several specific subroutines are developed to take into account beam position during scan and material state change from powder to liquid in the melting phase and from liquid to solid during cooling. The model is validated against literature data. The developed thermal model is able to forecast the evolution of the temperature inside the layers in the EBM process. Moreover, FE model properly predicts the geometry of melt pool and scan line. Keywords-additive manufacturing; electron beam melting; FEM; thermal model; temperature distribution; material change

1. INTRODUCTION In industry, there is a growing request of increasing flexibility and reducing production costs and lead-time. This result can be achieved by exploiting more efficient and effective manufacturing processes. Additive manufacturing (AM) technologies have the potential to respond to this request, revolutionizing the manufacturing approach [1-3]. Today additive processes allow the direct production of complex functional or end-usable metal parts [4]. In fact, AM guarantees long-term use of metal components, for the entire product life cycle or for a period as required. The additive manufacturing philosophy is that each object can be fabricated using a layerby-layer process. Three are the key phases: modeling the part using three-dimensional computer aided design (CAD) data, slicing the part into cross sections (layers), building the part by overlapping layers, according to CAD data. Additive techniques for metals are numerous and mainly differ in the construction method [3]. In particular, they can be classified into non-melting or melting processes, depending on which the heat input causes the partial or full melting of the raw material respectively. AM processes based on metal powders and high-energy beam, such as electron beam melting (EBM), belong to the category of melting processes. Each AM process can be associated with three main aspects: energy, material, and data or process control [5]. For instance, during EBM, electron beam (energy) melts metallic powders (material) to obtain a specify shape and solid massive part (data). Nowadays, the EBM process has many applications in the aerospace and medical fields. The aim is to fabricate complex parts of excellent


material. An example is the production of components of Titanium superalloys for which traditional methods are expensive and difficult to apply, because these metals have high mechanical properties and affinity with oxygen. Hence, EBM is not only a viable alternative to traditional processes but also a process with exclusive benefits. Despite the extensive advantages over conventional technologies, EBM still exhibits several process/part deficiencies. Especially, in order to obtain a defect-free part, beam characteristics, powder properties, and process control should be properly chosen by considering also their interactions. Today an empirical trial and error approach is typically adopted to identify a suitable combination of process parameters for a given metal powder. Alternatively, process simulation could be a suitable tool for decision-making and process optimization, since virtual analysis facilitates the possibility to explore “what if” scenarios [6]. Therefore, in order to optimize the quality of the part produced, the research community is studying and modelling the phenomena that occur during the EBM process. In order to analyze the influence of EBM process parameters researchers have proposed several FE thermal models [7-13]. Only one research about EBM thermo-mechanical model is available in the literature [14]. As regards the modelling of the powder, different approach are reported in the literature. Studies conducted by [8-14] are based on three-dimensional finite element numerical models and the powder is modelled as a continuum. On the contrary, in the research developed by Körner et al. [7], the powder is modelled as particle-to-particle with a two-dimensional mesoscopic approach. In this study, researchers developed a tool that considers the numerical behavior of each single particle of powder. They used an algorithm that simulates the sequential addition of layers of powder to generate a two-dimensional model composed of spherical particles. Even if this specific approach is more complex and limited to a symmetric scan line, this model was able to forecast the morphology of melting tracks and the balling effect. Another aspect to be considered is that the thermal behavior is different in areas corresponding to the powder layer, the melting pool and the bulk material, because of different physical properties of materials and heat transfer phenomena. In order to model powder properties, several approaches have been adopted. Zäh et al. [12] used the Zenher/Bauer/Schlünder model for density, specific heat and thermal conductivity. Shen and Chou [9] modelled the thermal conductivity according to the

109


studies of Tolochko et al. [15] and assumed the specific heat as that of the solid bulk material. Qi et al. [8] defined the solid thermal conductivity as a linear function of porosity. Sih and Barlow model [16] is used to model powder emissivity by [9, 10, 12-14]. Shen and Chou [9] assumed emissivity independent from temperature. All these studies ignored the heat dissipated by viscous forces and convection. Besides, in general it is assumed perfect wetting and the irradiation phenomenon is neglected [7, 8]. It is interesting to observe that, due to the lacking of experimental EBM data, many researchers validated their numerical models basing on the studies already carried out in EB welding. Further, some researchers took Gaussian volumetric distribution to model the heat source intensity like in selective laser melting (SLM) [9-14]. In all these studies, the researches underlined the complexity of describing the process by numerical models. Thus, only some phenomena of the EBM process were taken into account and some characteristics were not fully discussed. For example, in the literature the numerical implementation of the thermal behavior of the material after the beam passage and the cooling phase are not clearly described. Therefore, in this paper a comprehensive approach is adopted to develop a complete thermal model of EBM process considering heat source and main aspects and interactions of transient heat transfer. The numeric technique adopted to solve this thermal problem is the finite element method. This work includes: (1) a study of the electron beam impact on preheating powder bed through a Monte Carlo simulation, (2) a heat transfer analysis considering conduction and radiation phenomena, with (3) material thermal properties as a function of temperature, (4) the modelling of melting and cooling phases, and (5) a coupled analysis of phase change and material change. In order to validate the proposed numerical model, the simulation output is compared with experimental data available in the literature. 2. ELECTRON BEAM MELTING PROCESS The EBM machine is similar to an electron beam (EB) welding machine, but the scan process is like that of an electron

Electron Beam column Filament

Astigmatism lens Focus lens Electron Beam

Deflection lens

Heat shield Vacuum chamber

Powder hopper

Powder hopper

Rake Build tank

Figure 1. The ARCAM EBM machine.


Powder Start plate Build platform


Layer deposition

Table lowering

Raking AM part

Selective melting

Preheating

Figure 2. The phases of the EBM process.

microscope. In 1987, ARCAM AB developed the first EBM technology (Fig. 1). Within the EB gun, electrons are emitted from the cathode of a heated tungsten wire or of a single crystalline structure. The generated electrons are accelerated to a velocity between 0.1 and 0.4 times the speed of light, using an anode potential of 60 kV [7]. Two electromagnetic lenses control the EB. The first lens focuses the beam; the second one deflects the EB on the build table. In the EB gun no mechanical parts are in motion. This implies high accuracy and high scan speed. The beam current ranges between 1 and 50 mA and the beam diameter can be focused to about 0.1 mm [17]. Moreover, within the build chamber high vacuum is created in order to avoid electrons deflection by air particles. In fact, a turbomolecular pump evacuates the processing chamber [7]. The typical pressure of residual gases in an ARCAM EBM machine is 10–3 Pa in the vacuum chamber and 10–5 Pa in the EB gun [17]. In addition, two hoppers are used for powder dosing and a rake to level the layer. When the EB bombards the powder bed, electrons release their kinetic energy as heat. According to the beam energy density the powder is melted or heated [7]. Fig. 2 illustrates the EBM process, with the layer-by-layer part generation. The whole process occurs under vacuum, but during the melting phase inert Helium gas is blown at a pressure of about 10–l Pa to facilitate cooling and thermal stability [18]. The first powder layer is the part base. A preheating of the whole powder bed is required and takes place with a series of passages of the defocused beam at low power and high speed (i.e.15,000 mm/s and 30 mA [18]). Indeed, during the melt, the power is higher and the scan speed is lower, i.e. about 1,000 mm/s. After the selective melting phase, the build table is lowered by one layer thickness and additional powder is delivered from powder hoppers and raked. The process is repeated until the part is

110


completed. Typical layer thickness ranges between 50 and 400 µm. After building, the part cools down either under vacuum or helium flow. A soft agglomerate powder that adheres to the fabricated part, called breakaway powder, covers the part when it is removed from the building chamber [17, 18]. Cleaning of the part is done by sandblasting [19]. Because during the melting process there is no oxygen inside the building chamber, the unused powder can be recycled many times without altering its chemical composition and physical properties [17]. Thus, the material waste is minimized. 2.1. EBM Physical Mechanisms According to a default scan mode, the beam scans the top surface and leads to material melting. The heat transfer takes place by heat conduction between powder particles and between the powder bed and the bulk substrate, by irradiation from the powder bed to the chamber, and by convection between the powder bed and the surrounding environment. The electron kinetic energy is able to melt, sinter, heat, or vaporize material. During the process, material undergoes a first phase change from powder to molten metal and a second change when metal solidifies again as bulk material. The analysis of the process is complex because the scan speed is very high and the phase change occurs in a short time. Moreover, temperature distribution is inhomogeneous within the part/layer. These phenomena significantly affect the quality of the part in terms of tolerances and mechanical properties. In order to obtain parts without porosity and delamination, the molten pool should wet the previously consolidated material and the surrounding powder. In addition, when the melt pool solidifies, the top surface should be smooth enough to enable the spreading of the powder of the next layer and good surface finish [20]. Therefore, wetting characteristics of the solid phase by the liquid phase and capillary forces are very important for the process. To ensure correct wetting, energy beam must be able to melt both the top material and partially the bulk substrate. On the other hand, with the formation of the melt pool, the heat transfer increases and produces strong thermal gradients. The temperature distribution along the depth causes the generation of flotation forces, while the surface temperature distribution leads to turbulent flows such as Marangoni convection flows. Moreover, the phase change, i.e. melting or solidification of the material during the process, further affects the heat exchange. Fig. 3 shows a schematic view of physical phenomena involved in the EBM process. EB Radiation

Beam heat flux Vaporization

Phase change Material change Wetting Conduction

Marangoni convection Capillary forces

Figure 3. EBM physical phenomena (neglected phenomena in italic).



3. THERMAL MODEL OF THE EBM PROCESS 3.1. Heat Transfer Analysis In order to simplify the analysis, in this study only main phenomena are considered. Therefore, effects of capillary forces, wetting phenomena, Marangoni convection flows, and the vaporization of the material are neglected. According to the scientific literature, effects of Marangoni convection flows are irrelevant for scan speed higher than 100 mm/s that are common in EBM process [21]. a) Heat Transfer Equations Modeling the powder bed and the bulk substrate like a continuum, fundamental physical laws can be applied. Thus conservation of mass, conservation of momentum, and conservation of energy are used to obtain differential equations describing the thermal behavior during the process. The energy conservation is considered in order to calculate the temperature distribution. In a homogeneous body within which there is no exchange of matter, the temperature distribution T  x1 , x2 , x3  is expressed in Cartesian coordinates. In this model, the electron beam is perpendicular to the plane  X 1 , X 2  . Within an infinitesimal control volume of dimensions dx1 , dx 2 , dx3 the energy balance per unit volume and in differential form at a given instant of time can be expressed as  q  

De Dt

(1)

where q is the surface heat flux vector,    T  is the density and e is the thermal energy density. The  D Dt  notation is used to express the material or substantial derivate that is

D       x1  x 2  x3 Dt t x1 x2 x3

(2)

q is described by the Fourier's law as q    T

(3)

where    T  is the thermal conductivity. The thermal energy density e can be written as e  cT  h

(4)

where c is specific heat, T  T  x1 , x2 , x3 , t  is the temperature as a function of both space and time t, and h is the latent enthalpy defined as  L T  Tl   T T  s h  fs L L Ts  T  Tl  T T  l s   T  Ts 0

(5)

where Ts and Tl are the solidus and liquidus temperatures respectively, f s is the solid fraction, and L is the latent heat of fusion.

111



b) Modelling Domains and Boundary Conditions In order to completely describe heat exchange phenomena during EBM, equation (1) should be solved under appropriate boundary and initial conditions. The boundary conditions are

 D1  D2

 x , x , x   

3

: 0  x1  x1max , 0  x2  x2 max , 0  x3  tlayer  1

2

3

 x , x , x   

      

3

: 0  x1  x1max , 0  x2  x2 max , tlayer  x3  x3max  1

2

3

D  D1  D 2

 Ω  Σ

 x , x , x   D  x , x , x   D 1

2

3

1

1

2

3

2

: x3 t layer  : x3 t layer 

where D1 is the powder domain, D 2 is the bulk domain, D is the whole domain of the model, t layer is the powder layer thickness, Ω is the separation surface between the powder and the bulk domain for D1 , and similarly Σ is the separation surface between the bulk and the powder domain for D 2 . The initial conditions set is T ( x1 , x2 , x3 , 0)  Tpreheat with ( x1 , x2 , x3 )  D T ( x1 , x2 , x3 , 0)  Tr with ( x1 , x2 , x3 )  D T ( x1 , x2 , x3 , ) Tr with ( x1 , x2 , x3 )  D

T   q  x1 , x2 , x3 , v, t   qrad if  x1 , x2 , x3   Γ n Ω 

T n

S1  Ω

Ω

T   n Σ

T 0    n S 2  Σ

where Γ is assumed as the instantaneous surface where the heat flux q is supplied from an electron beam and S1 and S 2 are surfaces of D1 and D 2 domains respectively. Hence, heat flux is a function of coordinates, beam speed v and time t. Moreover, Tpreheat and Tr are respectively the temperature of preheating and the chamber temperature. Due to radiation, heat loss qrad can be written as



 qrad   T  T 4

4 r



(6)

where is    T  is the emissivity and  is the StefanBoltzmann constant which value is 5.67·10–8 W/(m2·K4). Heat loss by convection is negligible because EBM process is performed under high vacuum. 3.2. Material Properties The heat transfer analysis requires the knowledge of material thermophysical properties that are state variable. Namely, the


Temperature dependent values for the bulk material can be extracted from the technical databases. On the contrary, the properties of the corresponding powder material should be calculated from the respective bulk material by applying specific models. a) Modelling Properties of Powder Material Usually, powders used in the EBM process are produced by gas atomization and the obtained particles shape is spherical [22]. The main parameters that characterize powders particles are size distribution and relative density [23]. Supposing that the powder is approximated as packed bed of equal spheres with mean radius and connected by small circular necks, the relative packing density chosen is similar to the BCC atomic packing factor with porosity  equal to 0.32 [24]. Hence, the powder density can be defined by

1     bulk

(7)

Furthermore, the process is in vacuum, hence the Tolochko model is applicable to estimate the thermal conductivity of the powder [15]. This model takes into account the radiation heat through the pores and conduction heat between adjacent particles. In particular, powder thermal conductivity can be written as

Ω

T  n

Density Specific heat Latent heat of fusion Solidus temperature Liquidus temperature Thermal conductivity Emissivity

 powder

 qrad if  x1 , x2 , x3   Γ T  n

thermal behaviors of the powder and the bulk material significantly differ from each to other. Moreover, due to the large temperature range involved in the EBM process, the thermophysical properties must be expressed as function of temperature. In particular the required thermophysical properties are:

powder r  c

(8)

Where r is the effective thermal conductivity due to radiation and c is due to heat transfer through necks in the BCC lattice. The two terms of the second member of (8) can be computed as 16  T 3 3

(9)

c  3 bulk x

(10)

r 

where  is the mean photon free path between the scattering events, bulk is the bulk thermal conductivity, and x is the relative size of necks assumed equal to 0.09 according to experimental measurements of Tolochko et al. [15]. According to Sih and Barlow [16] the emissivity of the powder bed depends on the emission from the powder particle and the emission from holes in the powder bed. Consequently, the

112


emissivity can be written as

 powder  AH  H  1  AH   bulk

(11)

where AH is the surface fraction that is occupied by radiation emitting holes, and  H is the emissivity of the holes defined as

H 

 bulk

(12)

 bulk  f 1   bulk 

where  bulk is the bulk material emissivity and f is the fraction of total cavity surface that is cut away by emitting hole. In accordance with the assumed hypothesis of BCC packing and considering spherical holes, f is 0.25 and AH is 0.78. 3.3. Heat Source Modeling Before the melting phase, the powder bed is uniformly preheated. This step partially sinters the powder, reduces the distance between the particles forming circular necks and minimizes the powder spread. It is possible to analyze the impact of the electron beam on the powder using the Monte Carlo method by considering that the beam impacts the top surface perpendicularly. Monte Carlo method takes into account the mean free paths of electrons and the interaction probability for the phenomena that occur during collision. It estimates the trajectories of electrons within a sample. The factors that play a key role are material density, alloy elements and their atomic number, energy or accelerating voltage of the beam, and focus beam diameter.


lines refer to not back-scattered electrons. Grey lines in Fig. 4 illustrate the distribution of electrons along the radius and along the layer depth. The simulation results show two significant phenomena. The first one is that the electron energy rapidly decades as function of the depth and thus the depth of interaction volume is much smaller than the thickness of the layer [26]. In addition, the penetration depth is negligible if compared with powder particle size that is commonly smaller than 100 µm. Therefore, heat conduction effects between particles are more important than energy transferred by electrons. The second phenomenon is a lateral spread of the beam resulting in an effective diameter DE greater than the nominal focus beam diameter [27]. As a consequence, the electrons distribution along the radius is similar to a flattened Gaussian distribution as illustrated in Fig. 3. Thus, the energy of the electron beam can be simply modelled using a heat flux that is uniformly distributed on the surface  and it is defined as q  x1 , x2 , x3 , v, t  

UI S

(13)

where U is the acceleration voltage, I is the beam current, S   D 2 4 is the beam cross section considering the focus beam diameter D and  is a coefficient that takes into account the lateral spread and the efficiency of the beam control system. According to Alexander et al. [26] study and in order to ensure the power conservation  can be written as

  x1  x10 2   x2  x20 2   dS  exp     (14) 2   2 2 S 2 The simulation of electrons trajectories is run by CASINO   software [25]. The reported results are obtained for stainless steel 316L with an acceleration voltage of 60 kV and a beam where  x10 , x20  are the coordinates of the center of the electron diameter of 400 µm. Fig. 4 shows the trajectories of electrons beam focal spot and   D E 4 is the standard deviation of the into a cross section of a 316L solid sample. Especially backGaussian distribution of the electrons. scattered electrons are represented by red lines whereas blue 1.2·10–4 1.0·10–4 8·10–5 6·10–5 4·10–5 2·10–5 –200

–150

–100

–50

–250

–200

–150

–100

–50

0 Radius [µm] 0

50

100

150

200

50

100

150

200

0 250 250 0

0

5

5

10 15 20

Depth [µm]

–250

Hits (normalized) per area [1/µm2]

1

0

Hits (normalized) 2·10–3 4·10–3 6·10–3

10 15 20

Figure 4. Trajectories of electrons into a cross section of a 316L solid sample. Red lines are the trajectories of back-scattered electrons, while blue lines are the paths of not back-scattered electrons. Grey lines illustrate the distribution of electrons along the radius and along the layer depth.


113



4. FE THERMAL SIMULATION

TABLE I. CHEMICAL COMPOSITION OF STAINLESS STEEL 316L

A thermal simulation has been implemented to predict temperature, size and shape of the melt pool and scan line during EBM process. The three-dimensional finite element model is developed in Abaqus/Standard.

4.2. Heat Source In order to apply the heat flux on the top surface of the layer, whose location changes with time due to the movement of the beam, Abaqus DFLUX user subroutine is used. a) User Subroutine DFLUX DFLUX subroutine is used to define the flux distribution as function of position and time. User subroutine DFLUX is called at the beginning of each time increment and at each flux integration point. Then, it reads the simulation time and, according to the motion law defined in the user code, calculates the current position of electron beam center. Afterward, the domain Γ can be determined. Inner flux integration points of Γ surface, where the heat flux is applied, are selected according to the equation 2

2 D   x2  x02     2

Ni

Mo

wt %

0.017

17.0

12.8

2.1

Mn

Si

S

P

Fe

0.8

0.1

0.025

0.005

Bal

T (K)

333

755

1,396

 bulk 

0.78

0.86

0.96

 powder 

0.93

0.96

0.99

As indicated previously, the heat transfer analysis requires the implementation of thermophysical material properties as a function of temperature and material state, i.e. bulk or powder. Considering the coupled phase change and material state change, UMATHT, USDFLD and FILM user subroutines are used in Abaqus/Standard. 0.04

Bulk material Powder material

0.03 0.02 0.01 0

0

500

1,000

1,500

2,000

Temperature, T [K]

2

Figure 5. Thermal conductivity of bulk and powder materials.

(15)

where x1 is the beam scan speed, D is the beam diameter and  x01 , x02  are the coordinates of the starting point of the beam center. Equation (15) simulates the movement of the spot of the beam along X 1 axis. 4.3. Material Properties The material used in this study is stainless steel 316L and its composition is in Table I. Thermophysical properties of the bulk material are extracted from [8, 13] whereas powder properties are computed according to models detailed in §3.2a. The melting range of this alloy ranges between Ts = 1658 K and Tl = 1699 K and the bulk density is 7,970 kg/m3 at 293 K. The latent heat of fusion is 273 kJ/kg. The density of the powder is 5,400 kg/m3 at 293 K, considering an average particle size of 75 µm and a


Cr

porosity of 0.32, which are common values for commercial gasatomized powders of 316L. The abovementioned properties are shown in Fig. 5, Fig. 6 and Table II.

800

Specific heat, c [J/(kg·K)]

 x1  x01  x1t 

C

TABLE II. EMISSIVITY VALUES OF BULK AND POWDER MATERIALS

Thermal conductivity, λ [W/(m·K)]

4.1. Geometry and Mesh The FE model geometry consists of two parts: a substrate and one thin layer above. The substrate represents the solid bulk that was already processed and it is modelled as a part of 5 × 10 × 10 mm3. On the top of the substrate, a 0.1 mm thick layer is modelled as unsintered powder. The mesh consists of 8node linear heat transfer DC3D bricks according to Abaqus Documentation [28]. The aim of the used mesh strategy is to obtain detailed results within the powder layer in proximity of the incident electron beam. On the other hand, the size of mesh elements is chosen to avoid long running time and to ensure the absence of spurious oscillations in the solution caused by a numerical relationship between the minimum usable time increment, the element size, and the thermo-physical properties [28]. Hence, a finer mesh of 0.05 × 0.05 × 0.05 mm3 is used within a portion of the powder layer where the heat flux is applied. To simulate a single track, the electron beam flux moves along X 1 axis. Because of the symmetry along the  X 2 , X 3  plane only half of the workpiece is modelled.

Element

600 400 200 0

0

500

1,000

1,500

2,000

Temperature, T [K]

Figure 6. Specific heat of bulk and powder materials.

114


        

Thermal conductivity (S ) as a function of temperature Specific heat (c) as a function of temperature Latent heat Solidus temperature Liquidus Temperature Powder porosity Neck ratio Particle size Stefan-Boltzmann constant

At the start of the increment, the subroutine UMATHT reads the MAT_ID value and the temperature at each material calculation point and consequently calculates and updates thermal conductivity and specific heat. In addition, according to (5), subroutine UMATHT modifies the enthalpy function to take into account latent heat effects and solid fraction during melting and cooling phases. Finally, according to temperature solution and current value of MAT_ID, the subroutine UMATHT updates MAT_ID values, if required. b) User Subroutine USDFLD In order to update the density value from powder density to bulk density after melting, user subroutine USDFLD is used. This subroutine is linked to the subroutine GETVRM that reads the MAT_ID information and gives the corresponding density.

Subroutine UMATHT Read MAT_ID (material) and temperature

NO





(16)

This coefficient has units of film coefficient and it includes emissivity as a function of temperature. The required inputs are the emissivity  T  and the chamber temperature. 4.4. FE Model The flowchart in Fig. 7 shows model structure after the preheating phase. Initially for each material calculation point, MAT_ID and temperature values at the start of the time increment are read by the user subroutines UMATHT and FILM. Thereafter according to MAT_ID value, the temperature dependent properties are computed and density is updated. Then, the user subroutine DFLUX reads the current total time and applies the heat flux at each flux integration point according to the motion law of the beam. After that, the software solves the


YES

MAT_ID = 0

Set MAT_ID = 1

T ≤ Tsolidus

NO

YES Compute thermophysical properties as bulk material (MAT_ID = 1)

Compute thermophysical properties as powder material (MAT_ID = 0)

Heat source application Subroutine DFLUX Solution of heat transfer problem

END Figure 7. FE model structure (subroutines UMATHT and DFLUX) with material change procedure.

START Subroutine UMATHT

c) User Subroutine FILM In order to model heat radiation loss, a pseudo heat transfer coefficient hrad is defined as

h  T   T 2  Tr2 T  Tr  rad

Executed at each material calculation point

The thermophysical properties defined in this subroutine are specific heat, thermal conductivity, and phase change. The required inputs are listed below:

START

Read temperature

T ≤ Tsolidus

NO

Compute thermophysical properties as bulk material

YES

Executed at each material calculation point

a) User Subroutine UMATHT The user subroutine UMATHT is developed to define the thermal behavior as a function of temperature and material state. A solution-dependent state variable, named MAT_ID, works like an index of the material state at material calculation points of each element. Initially MAT_ID is set to 0 and powder properties are assigned to material calculation points of the elements. When MAT_ID changes to 1, bulk properties are assigned to material calculation points. MAT_ID value cannot more change back from 1 to 0.


Compute thermophysical properties as powder material

Heat source application Subroutine DFLUX Solution of heat transfer problem

END Figure 8. FE model structure (subroutines UMATHT and DFLUX) without material change procedure.

115



heat transfer problem. Successively, in the user subroutine UMATH the MAT_ID variable is updated, if necessary. When the temperature value at material calculation points is higher than the solidus temperature and the material is powder, the MAT_ID value changes from 0 to 1. The cycle is repeated for each time increment and until the simulation ends.

In order to validate the proposed FE model in terms of temperature distribution, melt pool size and width of the scan line, the outcomes of the numerical model are compared with experimental data available in the literature. To this aim, Zäh and Lutzmann [13] and Qi et al. [22] experiments are analyzed. Regarding the temperature distribution, the work of Zäh and Lutzmann [13] is considered. Their experiment consisted in the measurement of the local temperature at 0.3 mm below the powder layer, in the bulk substrate, by a thermocouple aligned with the center of the melt pool. In the experiment, the beam spot moves on a powder area adjacent to bulk material in the same layer, thus the top layer is modelled partially as powder and partially as bulk material. Therefore, there is no symmetry and the whole workpiece is modelled. According to experimental parameters, the simulation inputs are reported in Table III. In the numerical simulation, the temperature behavior is evaluated at five nodes located at 0.3 mm below the powder layer along the centerline. As illustrated in Fig. 9, nodes are equally spaced by a distance of twice the beam diameter. Fig. 10 shows the temperature evolution at these nodes as solid line. The experimental maximum temperatures measured by of Zäh and Lutzmann [13] during two tests are visibile in Fig. 10 as dashed lines for comparison. The circle markers on each solid line

x3 =

0.4

X3

xA1 = 0 mm xB1 = 0.8 mm xC1 = 1.6 mm xD1 = 2.4 mm xE1 = 3.2 mm

mm

Figure 9. Location of nodes where the temperature behavior is evaluated.

1,500 Temperature, T [K]

5. MODEL VALIDATION

X1

xA

xB

xC

xD

xE

1,450 EXPERIMENTAL

RESULTS

1,400 1,350 1,300

0

0.02

0.04

0.06

0.08

0.10

0.12

Time, t [s] Figure 10. Temperature evolution at five nodes as in Fig. 9 (PEB  150 W, v = 50 mm/s, and D  400 µm).

Maximum temperature, Tmax [K]

The flowchart in Fig. 8 shows the calculation method if the material change is not considered. The difference is significant. In fact, in this case the properties of the material are only temperature dependent. Correctly, the powder properties are such as corresponding liquid metal when the temperature is higher than melting point. Instead, when the temperature goes down, the properties are again those of the powder even if the material is actually bulk. This approach leads to unreliable numerical outcomes, because powder thermal conductivity is significantly different from that of the corresponding bulk material (Fig. 5).

X2

1,500

1,400

xC

xB

1,450

xD

xE

xA

1,350 1,300

x2 = 0 x3 = 0.4 mm 0

0.5

1.0

1.5

2

2.5

3.0

3.5

4.0

Distance from xA [mm] TABLE III. SIMULATION INPUT (ZÄH AND LUTZMANN) Parameter

Value

Beam diameter, D (µm)

400

Beam power, PEB  UI (W)

150

Beam efficiency, 

0.220

Surface flux, q (W/mm2)

263

Scan speed, v (mm/s)

50

Powder layer thickness, t layer (mm)

0.1

Preheating temperature, Tpreheat (K)

1353

Chamber temperature, Tr (K)

298


Figure 11. Maximum temperatures at nodes located at 0.3 mm below the powder layer along the centerline and at a variable distance from x A (PEB  150 W, v = 50 mm/s, and D  400 µm).

indicate the time instant in which the corresponding node on the top surface is the center of the heat flux area. Due to conductive effects, a time delay occurs from the application of the heat flux and the detection of the maximum temperature. Fig. 11 shows the maximum temperature detected in different nodes that are located at 0.3 mm below the powder layer along the centerline and equally spaced from each other by a distance equal to the beam radius. After the beam center moves of about one millimeter along X 1 axis, the graph shows a plateau around 1455 K. Comparing experimental and numerical results the observed maximum deviation is less than 3%.

116



TABLE IV. SIMULATION INPUT (QI ET AL.) Value

Beam diameter, D (µm)

200

Accelerating voltage, U (kV)

50 0.5

Beam efficiency, 

0.8

1.0

0.121

Scan speed, v (mm/s)

96.3

154.1

192.6

Preheating temperature, Tpreheat (K)

1353

Chamber temperature, Tr (K)

298

Considering the width of the scan line, the thermal model is validated through a comparison with the experiments conducted by Qi et al. [22]. In their experiments, single scan lines are fabricated under an acceleration voltage of 50 kV, a beam diameter of 200 µm, and variable beam current and scan speed. The material is stainless steel 316L with the same properties previously adopted. In order to resemble the test setup, the entire workpiece is modelled as powder material. According to experimental process parameters, the simulation inputs are reported in Table IV. Maintaining constant the accelerating voltage and the beam diameter, from (13) the intensity of the applied heat flux is directly proportional to beam current. Fig. 12 shows the width of single tracks obtained by simulation and experimentally measured by Qi et al. [22]. A deviation could be noticed at low scan speed. The reason could be ascribed to the effect of the Marangoni convection flows that occur at low scan speed. As mentioned previously, this phenomenon is neglected in the proposed numerical model. In addition, it should be considered that at low scan speed the track is more blurred [22]. Moreover, for low values of heat flux the balling effect occurs [13]. Thus, the sides of the track are not clearly recognizable and measurements of the width becomes more difficult. Moreover Qi et al. [22] work does not present the scattering of the measurements. The simulation results partially represent this behavior, because a uniform width of the track is only observed when the scan speed is equal to 5 mm/s. On the contrary, for lower scan speeds the line has not constant width. The magnitude of dispersion is comparable to the particle size. In general, the deviation between experimental and numerical results is lower than 15% when the scan speed is equal to 5 mm/s or the beam current is higher than 0.5 mA. 6. CONCLUSIONS In this work, in order to simulate the thermal distribution of EBM process during the scan of a single track, a threedimensional thermal FE model is developed. The approach adopted includes a comprehensive analysis of the heat source and of the main aspects and interactions of transient heat transfer. Primarily, the electron beam impact on preheating powder bed is studied. In fact, the kinetic energy of electrons is converted in thermal energy when electrons bombard the material. This analysis leads to model the energy source as


1.0 0.8 0.6 0.4 Experimental data Numerical results

0.2 0

2.3 3.3 5.0 2.3 3.3 5.0 2.3 3.3 5.0

0

1

2

3

4

5

6

5

6

5

6

Scan speed, v [mm/s]

(b) Beam current, I = 0.8 mA

1.2 Scan line width [mm]

Surface flux, q (W/mm ) 2


0.2 0

0

1

2

3

4

Scan speed, v [mm/s]

(c) Beam current, I = 1.0 mA

1.2 Scan line width [mm]

Beam current, I (mA)

Scan line width [mm]

Parameter

(a) Beam current, I = 0.5 mA

1.2


0.2 0

0

1

2

3

4

Scan speed, v [mm/s] Figure 12. Width of single tracks obtained by numerical simulation and experimentally measured (U  50 kV, D  200 µm).

uniform surface heat flux. Due to the high temperature gradient and the presence of different material states, i.e. powder and bulk, the thermophysical properties are modeled as function of temperature and material state. Properties of the bulk material are extracted from the technical literature whereas the material properties of the gas-atomized powder are calculated assuming BCC packing structure. In order to solve heat transfer problem Abaqus/Standard with several user subroutines are used. The user subroutine DFLUX is implemented to apply heat flux and move the electron beam spot. The user subroutines UMATHT, USDFLD, and FILM calculate the powder material properties. Moreover, conduction effects, radiation loss, melting and

117


cooling phases with latent heat effect, and a coupled phase change and material change are considered. For validation purpose, experimental data available in the literature are used as reference to test the response of the numerical model regarding temperatures distribution and widths of scan lines. A good agreement between simulations and experimental data is observed. Hence, the simulation adequately approximates the thermal phenomena occurring in the EBM process and their effects. Therefore, the developed thermal model could be used as a tool for process parameters optimization. Improvements to the existing model include the implementation of the mechanical properties of the material in order to have a complete coupled thermo-mechanical model for the evaluation of stresses and deformations induced by the beam passage. REFERENCES [1]

Atzeni, E., Iuliano, L., Minetola, P., and Salmi, A., 2010, ‘Redesign and cost estimation of rapid manufactured plastic parts’, Rapid Prototyping Journal, vol. 16(5), pp. 308-317. [2] Hopkinson, N. and Dickens, P., 2001, ‘Rapid prototyping for direct manufacture’, Rapid Prototyping Journal, vol. 7(4), pp. 197-202. [3] Wohlers, T., 2014, Additive Manufacturing State of the Industry (Wohlers Report 2014). Fort Collins, USA. [4] Atzeni, E. and Salmi, A., 2012, ‘Economics of additive manufacturing for end-usable metal parts’, The International Journal of Advanced Manufacturing Technology, vol. 62(9-12), pp. 1147-1155. [5] Milberg, J. and Sigl, M., 2008, ‘Electron beam sintering of metal powder’, Production Engineering, vol. 2(2), pp. 117-122. [6] Chiumenti, M., Cervera, M., Salmi, A., Agelet de Saracibar, C., Dialami, N., and Matsui, K., 2010, ‘Finite element modeling of multi-pass welding and shaped metal deposition processes’, Computer Methods in Applied Mechanics and Engineering, vol. 199(37-40), pp. 2343-2359. [7] Körner, C., Attar, E., and Heinl, P., 2011, ‘Mesoscopic simulation of selective beam melting processes’, Journal of Materials Processing Technology, vol. 211(6), pp. 978-987. [8] Qi, H., Yan, Y., Lin, F., and Zhang, R., 2007, ‘Scanning method of filling lines in electron beam selective melting’, Proceedings of the Institution of Mechanical Engineers, Part B: Journal of Engineering Manufacture, vol. 221(12), pp. 1685-1694. [9] Shen, N. and Chou, K., 2012, ‘Thermal Modeling of Electron Beam Additive Manufacturing Process: Powder Sintering Effects’, in ASME 2012 International Manufacturing Science and Engineering Conference collocated with the 40th North American Manufacturing Research Conference and in participation with the International Conference on Tribology Materials and Processing. American Society of Mechanical Engineers. [10] Shen, N. and Chou, Y., 2012, ‘Numerical thermal analysis in electron beam additive manufacturing with preheating effects’, in Proceedings of the 23rd Solid Freeform Fabrication Symposium, Austin, TX. [11] Soylemez, E., Beuth, J. L., and Taminger, K., 2010, ‘Controlling Melt Pool Dimensions over a Wide Range of Material Deposition Rates in Electron Beam Additive Manufacturing’, in Solid Freeform Fabrication Proceedings.



[12] Zäh, M., Lutzmann, S., Kahnert, M., and Walchshäusl, F., 2008, ‘Determination of Process Parameters for Electron Beam Sintering (EBS)’, in Excerpt from the Proceedings of the COMSOL Conference Hannover. [13] Zäh, M. F. and Lutzmann, S., 2010, ‘Modelling and simulation of electron beam melting’, Production Engineering, vol. 4(1), pp. 15-23. [14] Shen, N. and Chou, K., 2012, ‘Simulations of Thermo-Mechanical Characteristics in Electron Beam Additive Manufacturing’, in ASME 2012 International Mechanical Engineering Congress and Exposition. American Society of Mechanical Engineers. [15] Tolochko, N. K., Arshinov, M. K., Gusarov, A. V., Titov, V. I., Laoui, T., and Froyen, L., 2003, ‘Mechanisms of selective laser sintering and heat transfer in Ti powder’, Rapid Prototyping Journal, vol. 9(5), pp. 314-326. [16] Sih, S. S. and Barlow, J. W., 1995, ‘Emissivity of powder beds’, in Solid Freeform Fabrication Symposium Proceedings. Center for Materials Science and Engineering, Mechanical Engineering Department and Chemical Engineering Department, the University of Texas at Austin. [17] Biamino, S., Penna, A., Ackelid, U., Sabbadini, S., Tassa, O., Fino, P., Pavese, M., Gennaro, P., and Badini, C., 2011, ‘Electron beam melting of Ti–48Al–2Cr–2Nb alloy: Microstructure and mechanical properties investigation’, Intermetallics, vol. 19(6), pp. 776-781. [18] Gaytan, S., Murr, L., Medina, F., Martinez, E., Lopez, M., and Wicker, R., 2009, ‘Advanced metal powder based manufacturing of complex components by electron beam melting’, Materials Science and Technology, vol. 24(3), pp. 180-190. [19] Heinl, P., Rottmair, A., Körner, C., and Singer, R. F., 2007, ‘Cellular titanium by selective electron beam melting’, Advanced Engineering Materials, vol. 9(5), pp. 360-364. [20] Agarwala, M., Bourell, D., Beaman, J., Marcus, H., and Barlow, J., 1995, ‘Direct selective laser sintering of metals’, Rapid Prototyping Journal, vol. 1(1), pp. 26-36. [21] Gusarov, A., Yadroitsev, I., Bertrand, P., and Smurov, I., 2007, ‘Heat transfer modelling and stability analysis of selective laser melting’, Applied Surface Science, vol. 254(4), pp. 975-979. [22] Qi, H. B., Yan, Y. N., Lin, F., He, W., and Zhang, R. J., 2006, ‘Direct metal part forming of 316L stainless steel powder by electron beam selective melting’, Proceedings of the Institution of Mechanical Engineers Part B-Journal of Engineering Manufacture, vol. 220(11), pp. 1845-1853. [23] Krishnan, M., Atzeni, E., Canali, R., Calignano, F., Manfredi, D., Ambrosio, E. P., and Iuliano, L., 2014, ‘On the effect of process parameters on properties of AlSi10Mg parts produced by DMLS’, Rapid Prototyping Journal, vol. 20(6), in press. [24] German, R. M., 1989, Particle Packing Characteristics. Metal Powder Industries Federation. [25] CASINO, 2011, retrieved August 16th, 2014 from: http://www.gel.usherbrooke.ca/casino/. [26] Alexander, K., Andreas, B., and Carolin, K., 2014, ‘Modelling of electron beam absorption in complex geometries’, Journal of Physics D: Applied Physics, vol. 47(6), pp. 065307. [27] Mahale, T. R., 2009, Electron Beam Melting of Advanced Materials and Structures, PhD Thesis in Industrial Engineering, North Carolina State University. [28] Abaqus, ABAQUS Documentation. 2014, Dassault Systèmes: Providence (USA).

118



Titanium metal sheet structures of various wall thicknesses with additional functional elements prepared by selective electron beam melting in a powder bed Vera Juechter, Institute of Metals Science and Technology, Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany, [email protected] Adam Schaub, Institute of Manufacturing Technology, Friedrich-Alexander-Universität Erlangen- Nürnberg, Erlangen, Germany, [email protected] Marion Merklein, Institute of Manufacturing Technology, Friedrich-Alexander-Universität Erlangen- Nürnberg, Erlangen, Germany, [email protected] Robert F. Singer, Institute of Metals Science and Technology, Friedrich-Alexander-Universität Erlangen- Nürnberg, Erlangen, Germany, [email protected]

Based on the high flexibility in comparison to conventional production methods additive manufacturing technologies are becoming increasingly important for many industrial fields. The possibility of building highly complex parts with no need for tooling belongs to the many attractive features. For the processing of titanium alloys, which are prone to elements like oxygen and nitrogen, the selective electron beam melting SEBM process is particularly suitable. On the one hand a production under a vacuum atmosphere is possible. On the other hand the high building temperature allows a reduction of the residual stresses. A disadvantage of additive manufacturing techniques is the relatively long processing time due to the layer wise built up. To overcome the long cycle time a new and innovative approach is presented within this paper. Using a combination of conventional sheet metal forming and additive manufacturing methods both fabrication techniques are combined and their benefits merged. The first step to manufacturing parts in a reproducible way is to determine a suitable processing window. Consequently, in the present work additive structures are being built by SEBM with a variety of scanning speeds and line energies on Ti-6Al-4V sheets of various thicknesses, fabricated by rolling. Strategies will be discussed that result in dense parts with sufficient bonding between metal sheet and SEBM structure. Selective electron beam melting, sheet metal, Ti-6Al-4V

1. INTRODUCTION Additive manufacturing methods gain more and more interest because of their high geometrical flexibility. Beside processes for polymer materials, which exist for decades, in the last years great efforts were made to introduce metal based processes like selective electron beam melting SEBM [1]. The advantages of processing Ti-6Al-4V, as a light weight and biocompatibel material, in SEBM are the vacuum atmosphere and the high building temperature [2], [3]. So a reduced oxygen pickup is reached and also thermal stresses, caused by the high cooling rates, are reduced. The main challenge of all additive

We gratefully thank the German Research Foundation (DFG) for funding our research in the Collaborative Research Centre 814, project B5. ← Back to Table of Contents

manufacturing methods is the long build times because of the layer wise built-up. Here the traditional manufacturing processes like casting and forging are often less expensive due to less process time. Therefore, a combination of traditional manufacturing processes with additive manufacturing methods is the ideal way to benefit from the advantages of both methods. The idea in the present paper is to manufacture an 3D -shaped base component by traditional processes, on which another 3D-shaped part is then added by SEBM. As Ti-6Al-4V is the most common material for the selective electron beam process it was selected for the first experiments [4]. The first step to produce dense parts with a sufficient bonding is the evaluation of the processing window. In this vein adequate heating and melting strategies have to be developed as the process changes drastically by varying the base component geometry, in particular its wall thickness. 2. MATERIALS AND METHODS 2.1. Investigated material For this investigation gas atomized Ti-6Al-4V powder with a particle size distribution between 45 – 105 µm and a mean particle size of 68 µm was used. The particles show a mainly rounded morphology with a small number of satellites. The chemical composition lies within the standard DIN 17851:1990-11 [5]. As titanium sheet metal Ti-6Al-4V sheets with various thicknesses are used. The investigated material thicknesses are 1 mm and 5 mm. The sheet dimension is being kept constant at 150 mm x 150 mm. 2.2. Selective electron beam melting Selective electron beam melting SEBM is an additive manufacturing process, where parts are built up layer by layer in a powder bed by the use of an electron beam. Hereby the geometry of the part is given by a CAD-file, which is sliced

119



Figure 2. Schematic of the scanning strategy snake like hatching with a line distance of 100 µm and a rotation of the scanning pattern by 90°C after every layer.

Figure 1.

Schematic of the process steps, necessary to build one layer during selective electron beam melting [2]

into layers with a defined layer thickness. Typically layer thicknesses used in the SEBM-process vary between 50 µm and 200 µm. The SEBM-machine uses the CAD information for melting the areas which correspond to the solid parts finally. The process takes place at temperatures around 650 °C for Ti-6Al-4V. This is the temperature the start plate, on which the parts are built on, is heated up to before the process starts. To ensure a sufficient beam quality and to protect the powder from impurities a vacuum of 10-4 mbar to 10-5 mbar is applied. Additionally Helium is let into the chamber to reach a controlled vacuum of 2*10-3 mbar, which stabilizes the process. The SEBM-process consists of four main steps. First a powder layer is distributed by a rake across a 10 mm thick steel plate, see figure 1. The powder is preheated by a defocused electron beam to increase the mechanical stability and electrical conductivity. Afterwards the areas, which should be solid parts in the end, are molten by a focused electron beam. These steps are repeated until the last layer is built and the finished part can be removed of the machine. 2.3. Process setup An Arcam S12 machine is used to perform the experiments discussed in this contribution. In order to avoid distortion by thermal stresses the thin sheet is held down with screws on a 150 mm x 150 mm steel plate. On each sheet 9 samples with a dimension of 15 mm x 15 mm and a height of 5 mm are built with different melting parameters. A layer thickness of 50 µm was used. The melting strategy was a snake like hatching, where the electron beam is deflected in snake pattern with a line distance of 100 µm. In this way an overlap of the melt pools can be guaranteed. After every layer the melting pattern is rotated by 90° to enhance homogenous melting. Beside the melting pattern the scanning speed and the beam power define the melting. The scanning speed vscan and the power of the beam Pbeam determine the energy which is inserted in a line segment Eline, see Eq. 1.

The scanning speed varies between 0.2 m/s and 5 m/s and the power between 50 W and 1250 W, which leads to line energies between 150 J/m and 350 kJ/m. The exact values are given in table 1. After the build is finished the sheet with the additive parts on it has to be blasted with air and Ti-6Al-4V powder to remove the remaining sintered powder from the parts and the sheet. 2.4. Sample preparation For microscopical examination the additive parts are cut longitudinal in half, so a cross section of the additive part and the transition to the sheet is visible. The cross sections were embedded, grinded and polished. For microstructure analysis and the determination of the heat affected zone HAZ the specimens were etched using 8 % H2O2 in a 25 % KOH solution at 60 °C. For the examination of the heat affected zone HAZ five measurements per sample were taken. For the determination of the processing window the density of the samples were measured in the cross section by an optical examination. The criteria for good parts are a density higher than 99,5 %. If the energy input is to small bonding faults occur, which are characterized by a longitudinal shape perpendicular to the building direction and unmolten particles in between the void, see Fig. 3. TABLE I. Process parameters used for building the parts, whereas scanning speed , beam power and line energy are listed. Scanning speed [m/s]

Beam power [W]

Line energy [J/m]

0.2

50

250

0.2

60

300

0.2

70

350

1.0

200

200

1.0

250

250

1.0

300

300

5.0

750

150

5.0

1000

200

5.0

1250

250

(1)


120



Figure 3. Cross section of a dense and a porous part, where unmolten particles in between bonding faults are visible.

3. RESULTS 3.1. Processing window for different sheet thicknesses In this contribution sheet material out of Ti-6Al-4V was used. The processing window is defined by the parameters line energy and scanning speed. The criterion for dense parts is a relative density of 99.5 % or higher.

Figure 5. Microstructure of the sheet material, the heat affected zone and the SEBM-structure , whereas the morphology changes from elongated grains to equiaxed and columnar.

The processing window for additive parts built on a 1 mm and a 5 mm thick Ti-6Al-4V sheet are shown in Fig. 4. Increasing the beam power leads to reduced porosity, i.e. a minimum line energy is necessary to build dense parts. With increasing scanning speed the processing window, in terms of the minimum line energy, is shifted to lower values. The faster the beam is deflected the less energy is lost due to heat conduction. By varying the plate thickness the thermal conditions will be changed and the processing window is shifted to lower line energies. 3.2. Heat affected zone The heat affected zone HAZ is measured in the etched cross section of the sample, see Fig. 5. The base sheet shows an elongated microstructure in rolling direction. The microstructure of the SEBM-part is characterized by columnar grains oriented in building direction. The elongated grains consist of α- and β-phase lamellas. The heat affected zone shows equiaxed grains, which are formed when the material is annealed above above the β - transus temperature. At temperature, β grains grow until cooling sets in and the temperature drops below the β -transus temperature and α- and

Figure 4. Processing windows of additive Ti-6Al-4V parts build on a 1 mm thick (top) and a 5 mm thick sheet (bottom). With increasing sheet thickness the processing window is shifted towards higher line energies.


Figure 6. Heat affected zone of SEBM manufactured Ti-6Al-4V in dependency on the line energy for different scanning speeds. The HAZ grows with increasing scanning speed as well as with increasing line eenrgy.

121



Figure 7. Processing window (experimental data from Fig. 4 and a schematic representation of the heat affected zone thickness for additive parts built on a 1mm thick rolled base sheet. The heat affected zone increases with increasing line energy and scanning speed as well as the density of the additive parts does.

β lamellas precipitate. Due to the high cooling rates in SEBM-processes also martensite α´ may occur as it is well known from electron beam welding. But the high building temperature will lead to a transformation in α- and β lamellas. The heat affected zone shows a strong dependency on the process parameters and varies between 136 µm and 567 µm, see Fig. 6. At constant scanning speed the heat affected zone increases with higher line energies due to the higher energy input. Keeping the line energy constant, the heat affected zone grows with increasing scanning speed due to less thermal conduction loss. 4. DISCUSSION The processing window where dense structures are obtained by SEBM for Ti-6Al-4V was already determined in ref. [3] for standard building conditions on a 10 mm thick steel starter plate. The processing windows for additive parts on Ti-6Al4V sheets show in principal the same trends. In particular, the processing window drops to lower line energies by increasing the scanning speed due to less thermal heat loss. Due to different heat content when varying the sheet thickness the processing window will be shifted to higher or lower line energies. Decreasing the sheet thickness reduces the thermal mass and therefore the temperature of the sheet material increases. As a consequence, less energy input is necessary to melt the powder material and the processing window is shifted to lower line energies. The heat affected zone HAZ is dependent on the processing window. The HAZ increases with increasing scanning speed and line energy in the same way the processing window does, see Fig. 6. Reducing the HAZ as much as possible is important as the HAZ weakens the sheet material. For thick sheet materials the HAZ is small in relation to the sheet thickness. Decreasing the sheet thickness increases the absolute size of the HAZ and increases the ratio of HAZ to sheet thickness. In this case it becomes even more important to keep the HAZ


small. Taking the processing window into account the smallest HAZ occurs at low line energies. Therefore, problems with bonding faults may occur as the inserted energy isn`t sufficient to melt the particle layer. Bonding faults weaken the additive manufactured part and therefore they have to be avoided. The aim of the new technology suggested in this paper is to build dense parts with a good bonding and a small heat affected zone to achieve good properties. With decreasing sheet thickness it becomes more difficult to find suitable process parameters to meet the requirements. Knowing the processing window with the associated heat affected zone is a necessary tool to build dense additive parts on sheet material with a good bonding between part and sheet. ACKNOWLEDGMENT The authors like to thank Felix Knorr for the sample preparation, Thorsten Scharowsky for fruitful discussions and Jörg Komma for the technical assistance. We gratefully thank the German Research Foundation (DFG) for funding our research in the Collaborative Research Centre 814, project B5. REFERENCES [1] [2] [3] [4] [5]

U. Ackelid, M. Svensson, 2009, `Additive manufacturing of dense metal parts by electron beam melting` Materials Science and Technology Conference and Exhibition 2009, MS and T’09 4 P. Heinl, A. Rottmair, C. Körner, R. F. Singer, 2007, ‘Cellular Titanium by selective electron beam melting’ Advanced Engineering Materials, vol. 9 (5), pp. 360-364 V. Juechter, T. Scharowsky, R. F. Singer, C. Körner, 2014, ‘Processing window and evaporation phenomena for Ti–6Al–4V produced by selective electron beam melting’. Acta Materialia, vol. 76, pp. 252–258 A. Schaub, V. Juechter, R. F. Singer, M. Merklein, 2014, `Characterization of electron beam melted Ti-6Al-4V structures on sheet metal`Proceedings Esaform DIN 17851:1990-11, 1990, ‘Titanlegierungen; Chemische Zusammensetzung’ Deutsches Institut für Normung e.V., Beuth, Berlin.

122



Customization of 2D lattice structures as response of part load conditions for SLM manufacturing Marcin A. Królikowski Ph.D. Eng., Institute of Manufacturing Engineering, West Pomeranian University of Technology, Szczecin, Poland, [email protected] Abstract Paper presents discussion concerning application of SLM for manufacturing 2D lattice structures with third dimension assumed as linear. Structures are foreseen to be irregular according to specific requirements of application. Customization of lattice structures corresponding to load map is pointed out as a main issue in research work. A method of customization of lattice structure cells according to loads applied is presented. General methodology of customization of directional 2D lattice structures is also proposed. An experimental verification, as well as discussion on technological limitations and geometrical inaccuracy of SLM method are also presented. Paper concludes only 2D irregular structures manufacturing in order to maintain all process-related restrictions for further development. Keywords:

Metal additive manufacturing, Selective Laser Melting, Lattice structures,

1. INTRODUCTION In many applications, use of typical isotropic materials like metals are, is not sufficient. This is caused by contemporary modern design requirements to reduce weight and maximize effective use of material. A lot of solutions consider application of composites. A number of applications however demand metal structures, because of different factors as environment, temperature etc. As a solution for these requirements, since many years different cellular materials are being considered. From metal foams to cellular metal - metal structures and MMC’s (Metal Matrix Composites). There is a number of theoretical research dealing with lattice structures. A number of research work also concern additive manufacturing methods applied to rapid prototyping and manufacturing, Rehme et al. 2006, Rosen 2007, Williams et al. 2011, Hao et al. 2012, Hussein et al. 2011, 2013. All of mentioned research, which are accompanied by experimental investigations, concluding to real applications, discusses only regular lattice structures. Different types of regular structures are presented and discussed by Yadroitshev et al.2009, Hussein et al. 2011, 2013, Hao et al. 2012 or Santorinaios et al. 2006. There are general structures, not commonly applicable for direct metal melting methods. Rosen 2007, generated arbitral regular lattice structure basing on octet-truss, not everywhere applicable however for SLM. His research however generally solves problem of discretization continuous flat surface models with arbitral octet truss cell structure.


Rapid additive manufacturing of metal parts as quickly growing branch of modern technology using different methods – Kruth et al. 2005, offers possibility to generate complex geometry. There is a need to go further with lattice cellular structures applied to direct metal additive manufacturing such as selective laser melting (SLM). Research conducted and presented in this paper is a preliminary approach to this issue, applying non-regular lattice structures to loaded parts, as a response of load conditions. This approach enables to obtain specific inner structure of metal part oriented to application and load conditions of part / mechanism. As SLM technique is indeed a type of 2,5 axis numerically controlled manufacturing process, it was decided to start with 2D structure, where 3rd dimension is linear along vertical axis. It should provide clear analysis of technological influences on manufactured structures. 2. METHOD Method of generating non-regular cell lattice structures for different applications, elaborated in frames of the project mentioned in acknowledgements section is quite general but applies to strength resistance and stiffness criteria of the part. The flow chart of method is presented on Fig.1. First of all, the method applies technique based on the same idea as presented by Chen 2006, i.e. mesh based geometrical topology approach. The difference between method applied in this paper and Chen’s approach however is fundamental because in ours case this mesh is not directly used to build truss structure. Meshing executed by advanced CATIA mesher (or SIMULIA when applicable) is used to generate the map of required sensors, which can be displacement, energy, stress or any other resultant from FEM analysis. Native mesh geometry is not used for lattice structure topology as Chen 2006, mentioned in his research. Solid material part is meshed according to assumed meshing quality. Mesh generation applies use of typical and sophisticated tools to achieve appropriate meshing either oriented on geometry or parameters quality. After examination of meshing quality, typical analysis is carried out. A map of nodal stresses is generated. Type of cell geometry is selected and third dimension is filtrated resulting in 2D map of nodal stresses across the part. This map consist of coordinates of nodes as well as Tsai-Wu reduced stress values. Basing on this values either polynomial linear or surface interpolation is carried out. Interpolation curve or surface is

123


projected on the topology, when values of nodal stresses approximated are respectively referenced to assumed topology cells as cell parameters.


limitations for application of the methodology, from which the most important is required dimensional continuity of part (levers, bars etc.). One could of course apply a hybrid structure after local re-meshing of part but boundary conditions between different cell types are well known as problem of bonding lattice stiffeners together - Chen 2006, Hassan et al. 2011. 3. ANALYTICAL ANALYSIS Models were parameterized and generated in CAD/CAM system CATIA v5. As reference elements solid bars were used. Decision to take into account simple “X” box structure came from its simplicity and easiness to maintain in further investigations on manufacturability of lattice cell filled specimen bars. This is essential to focus on basic geometrical primitive structures to gain all process-related information with minimal geometrical complexity. “X” cell (fig. 3) could also fulfill another requirements besides of stiffness or strength resistance criteria, such as large radiation area for construction elements with for instance heating exchange functionality: 3.1. Input model Solid CAD model of bar was modeled as part in CAD/CAM system. A typical tetrahedral elements with linear edges was applied for meshing. Meshing was carried out using CATIA mesher with following parameters - element size 2mm with proportional sag 0,2. Mesh quality was focused on shape to achieve the best geometrical accuracy of mesh in relation to part. Boundary conditions were applied for simple bending with bending force 10[kN]. A map of nodes with reduced von Mises nodal stress values (fig.2) was exported to text file after typical static-case calculation. Maximum deflection is 0,13[mm].

Figure 1. Method of cell lattice structure application

This is the key point of method, to obtain discrete model of mesh nodes and stress values, then to apply common interpolation to get another discrete values located in cell nodes. These values correspond to cell size or cell stiffeners. After re-modeling of part with application of non-regular 2D structure (3rd dimension is assumed to be linear) analytical verification is carried out to verify expected effect of weight reduction in relation to other stiffness or stress criteria. This method can be used to strengthen the weakest points of structure or to homogenize given parameters across the 2D mean plane of the part. There is obviously a number of


Figure 2. Map of Huber – von Mises nodal stresses

3.2. Regular cell lattice model CAD model of elementary “X” cell was fully parameterized (fig.3) and used to multiply across the structure using macros, written in CATIA v5 script language. Technique implemented in this CAD/CAM system is called Power Copy for sequential multiplying solid elements bonding them to each other, but with different parameters at each execution. Power copy was executed from CATIA VBS script with wall values of 0,2mm. Elements were spread regularly across the part.

124



Figure 3. X” structure elementary cell

Regular cell lattice bar was meshed with 0,2[mm] elements with proportional sag 0,2. After static analysis with the same boundary conditions following deformation map was obtained (fig. 4).

Figure 5. Polynomial interpolation of nodal stress values

Figure 6. Non-regular "X" lattice structure

Bar was meshed with 0,2[mm] elements with 0,2 sag. After common static case calculations following displacement map was obtained.

Figure 4. Translational displacement on regular lattice element

Although maximum observed displacement caused by pressing force of bending area is over 1[mm], displacement measured in bottom fiber is approx. 0,35[mm]. 3.3. Non regular lattice model Results of nodal stress values gathered from preliminary solid model were filtered to obtain linear map of stress in neutral fiber of part. Only neutral fiber was taken into account in this research stage because of unknown manufacturability of these structures. It was expected to observe a big influence of SLM technology on part accuracy, mentioned in many publications for instance Kruth et al. 2005. It was assumed than complexity of analytical model is not proportional to expected manufacturing errors and to take into account only middle area of bar is enough. Further experimental verification of mass rise (tab. 2) proven that this assumptions were correct. Polynomial interpolation was carried out to obtain non-linear, continuous representation of stress across the part in neutral fiber (fig. 5). After re-scaling and adaptation to cell geometry, a spread of wall thickness from 0,2[mm] (minimum) to 0,5mm was calculated. A CATIA script was executed to build and bond “X” cellular elements according to estimated wall thickness. Adapted bar is presented at fig. 6.


Figure 7. Translational displacement of adapted non-regular lattice element

Tab. 1. summarizes theoretical displacement in relation to weight obtained by adaptation of lattice structure to nodal stress map. TABLE I. Type

ANALYTICAL EXAMINATIONS OF DEFLECTION M [g]

f [mm]

f/M *

BAR

39,00

0,13

0,0033

RX2D

17,00

0,53

0,0312

NX2D

28,00 0,16 0,0057 *M – mass, f – deflection - bottom fiber,

As can be seen from analytical examinations, weight loss of 30% of adapted non-regular (NX2D) bar is greater than decrease of stiffness, which is approx. 20%. A weight of

125


regular lattice cell bar (RX2D) obviously is 57% less than solid one but displacement of bottom fiber of regular bar is approx. 3 time larger (311%). One could of course play with these values, when required, adapting them to specific values by simple optimization. But, to be sure if technological errors of small, thin-walled structures do not cover results expected from analytical research, a series of samples was manufactured and experimental examination was carried out. 4. EXPERIMENTAL PART Research work concerning adaptation of 2D non-regular lattice structures has to be verified by experimental investigations. Experimental part consist of some stages, from which first is recognition of parameters or fully melted 1.4542 (EN 10088-3) steel. Parameters of SLM melted materials vary, depending on machine and process and have to be always verified, as presented by last work by Delgado et al. 2012. A number of specimen was tested and mechanical parameters were estimated. 4.1. Selective Laser Melting technique Rapid manufacturing of specimen was carried out in Rapid Prototyping & Manufacturing lab, in Institute of Manufacturing Engineering, West Pomeranian University of Technology, Szczecin, Poland. SLM machine from MCP HEK, is powered by 100W fiber laser. Preventive atmosphere is built by Argon gas with less than 1% O2. Scanning parameters were default for material file. Dimensions of samples were 10x10x50mm. Samples generated as dense STL files were prepared in Magics software, where appropriate support was applied at the bottom of samples. Process was run at SLM machine’s computer after slicing, hatching and estimating scanning parameters for this material. 4.2. Material applied Metallic powder used for experimental analysis is a prealloyed stainless steel in fine powder form. Its composition corresponds to European EN 10088-3. This type of steel is characterized by having good corrosion resistance and mechanical properties, especially excellent ductility in laser processed state, and is widely used in a variety of engineering applications. This material is applicable for many part-building applications such as functional metal prototypes, small series products, individualized products or spare parts. Standard processing parameters use full melting of the entire geometry. It can be machined, spark-eroded, welded, micro shot-peened, polished and coated if required. Typical applications of this steel powders: -

engineering applications including functional prototypes, small series products, individualized products or spare parts,

-

parts requiring high corrosion resistance and sterilisability,

-

parts requiring particularly high toughness and ductility.



4.3. Verification of manufacturability of thin-walled X structures wit application of SLM method Aiming at experimental verification of analytical research, a number of preliminary specimen was generated. There are 5 regular (R) specimen and 5 non-regular (N) manufactured. Specimen were removed from base plate and machined on CNC milling machine DMG 60 MonoBlock. HSC (High Speed Cutting) parameters were applied, in order to prevent thinwalled lattice cells (0,2 – 0,5mm) from deformations caused by traditional milling process. Specimen are shown on fig. 8 and fig. 9, after machining. The aim of machining was to check machinability of lattice cell structures, when there would be a need to apply them as a construction element. It was observed that although structure remain intact, jagged edges cover a part of cells.

Figure 8. Regular "X" 2D specimen after milling

Figure 9. Non-regular "X" 2D specimen after milling

It was observed, that thin walls of lattice structures, during manufacturing process, increased their dimensions. This phenomena, well known from publications is especially unwanted, when dimension of thin elements is getting closer to machine accuracy. Theoretically wall with thickness of 0,2mm is thick enough to be manufactured, what was the subject of our former research work however wall thickness increase depends strongly on manufactured geometry. In order to estimate real increase of weight in comparison to analytical estimations, specimen were weighted. Results are shown in table II. This phenomena strongly influences on correlation between design and real specimen dimensions & weight obtained by selective laser melting in this case. It should be

126


pointed out, that besides of manufacturer’s specifications practically in every case material is added to manufactured geometry. In former research work, concerned accuracy of SLM process, always additive deviations of dimensions were observed. TABLE II. No


Both irregular and regular specimen were tested for bending (fig. 11).

POST-SLM WEIGH INCREASE OF „X” SPECIMEN H [mm]

W [mm]

L [mm]

M_exp [g]

M_the [g]

RX2D 9,7

10,2

50,0

23,5

17,3

2

9,5

10,2

50,0

23,5

16,9

3

9,3

10,2

50,0

22,7

16,6

4

9,6

10,2

50,0

23,5

17,1

5

9,5

10,2

50,2

23,8

16,9

AVG

9,5

10,2

50,0

23,4

17,0

Δ

0,4

0,0

0,2

1,1

NX2D 1

9,4

10,2

49,3

31,6

27,2

2

9,6

10,2

49,2

32,1

27,8

3

9,4

10,2

49,1

31,7

27,2

4

9,3

10,2

49,3

31,2

26,9

5

9,2

10,1

49,2

30,8

26,6

AVG

9,4

10,2

49,2

31,5

27,1

Δ

0,4

0,1

0,2

1,3

Where – exp, means experimental, the – theoretical assumptions

Weight increase due to manufacturing process in case of regular “X” specimen (fig. 8) equals 38%, where non-regular, adapted specimen are only 16% heavier 4.4. Bending tests For assessment and verification of correlation between real behavior and analytical estimation, bending tests were realized. Equipment applied was Instron 8501 machine with free bending adaptor (fig. 10, 11).

Figure 11. Bending test

The objective of experimental bending tests was also to estimate real deviations of deflection caused by additional material applied to specimen, mentioned in chapter 4.3. Following graphical results were obtained for regular specimen (fig. 12 & 13). DEFLECTION [mm] -1,0

-0,8

DEFLECTION [mm]

1

-0,6

-0,4

-0,2

0,0 0

20

40

60

80

100

120

Time

Figure 12. regular specimen bending test graph. Colors correspond to specimen No.

-0,5

DEFLECTION [mm]

DEFLECTION [mm]

-0,4

-0,3

-0,2

-0,1

0,0 0

20

40

60

80

100

120

Time

Figure 10. setup for bending tests


Figure 13. non-regular specimen bending test graph. Colors correspond to specimen No.

127


Tests were carried out in linear cycle load – unload, reaching max bending force of 5[kN] in 60[s]. One can clearly see improvement of deflection parameters for non-regular, adapted specimen. Maximum deflection of regular specimen is approx. 1 mm, where non-regular is approx. 0,5. Table III summarizes maximum results of analytical experiments vs. experimental verification, takin of course into account increase of parameters of geometrical structure and increase of real weight of specimen. TABLE III. type

ANALYTICAL VS EXPERIMENTAL DEFLECTION

Analytical [mm]

Experimental [mm]

RX2D

0,27

1,01

NX2D

0,08

0,45

It could be clearly observed, that further investigations concerning SLM process, as mentioned before and noticed in many publications are needed. Analytical parameters were estimated for theoretical material parameters and ideal bending conditions without cracks, irregularities etc. It is also not obvious if lattice structures with thin walls could be calculated basing on experimental material verification with application of typical specimen (paddles, regular bars etc.). 5. RESULTS & CONCLUSIONS It could be clearly observed, that weight increase of specimen generated by SLM process correlates to complexity of lattice inner structure and rises with decrease of size of minimal geometrical elements. As assumed average accuracy of SLM process, at Realizer II machine is ± 0,1mm, getting closer to this value causes significant number of geometrical errors. Preliminary experiments concerning real accuracy and repeatability of lattice geometry generated by SLM let us consider rather + 0,1 – 0,2mm additive dimensions variations. In research on benchmarking of different SLM/SLS processes Kruth et. Al 2005 obtained additive character of geometrical deviations in MCP HEK SLM process. Values obtained by Kruth are over 1% larger than nominal both for solid geometry and holes. Geometrical inaccuracy and instability of lattice cellular elements also has been proven by other experimental investigations by Chunze et al. 2012 who manufactured Shoen gyroid structures, Santorinaios, et al. 2006, who met the limitations of cell size during SLM process, and others. Moreover is should be pointed out that wall thickness 0,2mm is below a limit for SLM process with this experimental conditions. It is suggested to have minimum 0,3 – 0,4 thick walls. Experiments leading to estimate a real correlation between material use and minimum geometry size, as wall thickness for lattice structures will be a subject of further investigations. Former research leaded mostly by Kruth, Yadroitsev, Hao, Hussein and Santorinaios, although very valuable and needful, currently are not sufficient for complex irregular lattice



structures. Observing results of experimental verification of bending, increase of material of specimen an its weight, one conclusion could be formed: very important for precise SLM manufacturing is to adapt scanning accuracy, machine parameters, bed size and scanning strategy to realized task. This could be achieved currently only on experimental way and takes a lot of effort. ACKNOWLEDGMENT Research work described in this paper is supported in frames of Polish national scientific project, No: N N504 492839, entitled Research on methods of manufacturing irregular lattice structures with application of SLM, financed entirely by National Centre of Science. REFERENCES [1] [2]

[3]

[4]

[5] [6]

[7] [8] [9] [10] [11]

[12]

[13]

Chen, Y. 2006. A mesh-based geometric modeling method for general structures. ASME Conference Proceedings: 9–281. Delgado J., Ciurana J., Rodriquez C. A., Influence of process parameters on part quality and mechanical properties for DMLS and SLM with ironbased materials, International Journal of Advanced Manufacturing Technology (2012) 60:601–610 Hao L., Raymont D., Yan C., Hussein A., Young P., Design and additive manufacturing of cellular lattice structures, P.J. Bartolo (Ed.) et al., Innovative Developments in Virtual and Physical Prototyping, Taylor & Francis Group, London (2012), pp. 249–254 Hussein, A., Yan, C., Everson, R., Hao, L., Preliminary investigation on cellular support structures using SLM process. In: Bartolo, et al. (Eds.), Innovative Developments in Virtual and Physical Prototyping. Taylor & Francis Group, London, pp. 609–612. 2011. Hussein A., Hao L., Yan Ch., Everson R., Young P., Advanced lattice support structures for metal additive manufacturing, Journal of Materials Processing Technology, 213 (2013) 1019 – 1026 Kruth J.-P., Vandenbroucke B., Van Vaerenbergh J., Mercelis P., Benchmarking of different SLS/SLM processes as rapid manufacturing techniques, Int. Conf. Polymers & Moulds Innovations (PMI), Gent, Belgium, April 20-23, 2005 Osakada K., Shiomi M., Flexible manufacturing of metallic products by selective laser melting of powder, International Journal of Machine Tools & Manufacture 46 (2006) 1188–1193 Rehme O., Emmelmann C., Rapid manufacturing of lattice structures with selective laser melting, Proceedings of SPIE The International Society for Optical Engineering, 6107 (2006), pp. 192–203 Rosen D.W., Computer-Aided Design for Additive Manufacturing of Cellular Structures, Computer-Aided Design & Applications, Vol. 4, No. 5, 2007, pp 585-594 Santos E.C., Shiomi M., Osakada K., Laoui T., Rapid manufacturing of metal components by laser forming, International Journal of Machine Tools & Manufacture 46 (2006) 1459–1468 Santorinaios M., Brooks W., Sutcliffe C.J., Mines A.W., Crush behaviour of open cellular lattice structures manufactured using selective laser melting, High Performance Structures and Materials, III 85, 481-490 Williams C.B., Cochran J.K., Rosen D.W., Additive manufacturing of metallic cellular materials via three-dimensional printing, International Journal of Advanced Manufacturing Technology, 53 (2011), pp. 231– 239 Yadroitsev I., Shishkovsky I., Bertrand P., Smurov I., Manufacturing of fine-structured 3D porous filter elements by selective laser melting, Applied Surface Science 255 (2009) 5523–5527

128



Selective electron beam melting of pure copper: influence of energy input on surface roughness and dimensional accuracy R. Guschlbauer, FAU, Joint Institute of Advanced Materials and Processes, Fürth, Germany, [email protected] M. A. Lodes, FAU, Joint Institute of Advanced Materials and Processes, Fürth, Germany, [email protected] C. Körner, FAU, Chair of Metals Science and Technology, Erlangen, Germany Abstract - In the presented work additive manufacturing of pure copper powder by selective electron beam melting is presented. Dense copper components could be processed which show that melting depth as well as cuboid width increase with growing energy input. In contrast, the surface arithmetic average roughness decreases with increasing energy input due to the incomplete melting of the powder. Keywords: Additive manufacturing (AM); selective electron beam melting (SEBM); pure copper powder; process development

1. INTRODUCTION Additive manufacturing (AM) is a manufacturing technique which creates parts layer upon layer. It is used for manufacturing of polymers, ceramics and metals. For all materials, a wide range of different process setups exist, each using specifically shaped raw materials, e.g. powder, liquid, foil etc. [1]. The advantage of AM is the almost unlimited geometrical flexibility of the processed parts in contrast to other manufacturing techniques like casting, machining or forging. The layer-wise component creation enables the production of undercuts and net structures e.g. with auxetic behavior [2]. Besides the geometrical aspect, AM has the advantage to shorten the production cycle by using rapid prototyping during product development. Additionally, product improvements can be implemented more efficiently than with conventional fabrication because AM does not need new, expensive and/or special tools for changing product details [3]. Thus, AM is advantageous for small series and especially for individual components like tailormade implants. Metal can be additively manufactured by several methods. Among powder-based techniques Selective Laser Melting (SLM) is the most common one. In the presented work Selective Electron Beam Melting (SEBM) is used. The setups of both systems are quite similar, with the main difference being the energy source (laser vs. electron beam). There are several advantages of SEBM, one being the beam deflection by electromagnetic lenses which allows a very high deflection speed of up to several thousand meters per second. Furthermore the manufacturing under low pressure vacuum/protective gas atmosphere inhibits powder contamination and offers processing of reactive or gas sensitive metals like titanium. In addition, the absorption physics are different, so that a higher energy input is possible using the electron beam. Usually very


low thermal residual stresses occur in SEBM, since high build temperatures are achievable due to the possibility of preheating the raw powder with a defocused electron beam. Copper exhibits excellent thermal and electrical conductivity. Considering metals, copper is only exceeded by silver which is much more expensive and has a 120 times worse price-per-conductivity ratio [4, 5]. SEBM of copper enables the potential to combine the high thermal and electrical conductivity with the geometric freedom resulting in e.g. components with complex internal cooling channels for automotive or aerospace applications or tailor-made net-structures for chemical engineering. Only little research on AM of copper has been done. Reagrding SLM it great difficulties to manufacture copper alloys due to the high reflectivity are reported [5–7]. For SEBM there are first attempts using copper with some content of oxygen [8] as well as pure copper [9]. Besides the mentioned reflection issue, the high thermal conductivity and according heat loss is challenging for the beam melting of copper. In comparison to other successfully used alloys in SEBM like Ti-Al6-V4, copper shows a 60 times higher thermal and electrical conductivity [10]. Here the high energy density of the electron beam as well as the protective gas atmosphere in SEBM are promising in terms of manufacturability requiring special adaption of process control. 2. THEORY 2.1. SEBM process The SEBM process starts with the construction of a virtual, three dimensional and computer aided designed (3D-CAD) file. The 3D-CAD file gets sliced in vertical layers afterwards. This 2D coordinates are addressed by the electron beam and powder is selectively molten layer by layer. The SEBM setup of an Arcam AB EBM machine is shown in Figure 2-1. It consists of two essential parts: the electron gun and the process chamber. The electron gun emits, accelerates, focuses and deflects electrons with a magnetic lens system. The building tank contains a moveable platform on which a massive start plate is located as base for the build. Supports connect the build to the start plate for good heat conduction and mechanical stability. The two powder hoppers on each side of the building tank feed the powder-layer creating rake.

129



Figure 2-2 shows the four-step production cycle of one layer. First, the powder is preheated with a defocused beam in order to create a slightly sintered surface for charge dissipation and mechanical stability. After preheating, the layer gets selectively molten with a focused beam. Then the movable process platform is lowered by a predefined layer thickness. In the last step, the rake moves over the building tank applying a new powder layer. 2.2. Beam parameters The emitted electrons are accelerated by a constant acceleration voltage UBeam. The beam intensity on the surface is controlled by the current IBeam, the focus offset and the deflection speed νBeam. The scan strategy is defined by the distance between the scan lines (line offset) and the route of the beam (snake, line order). For evaluating the influence of the electron beam on the component properties (e.g. density), usually experiments with various beam parameters are conducted.

To consider the beam deflection speed, the line energy El is established (2). It is defined as the beam power PBeam divided by the deflection speed νBeam and can be seen as energy input per melt line length. El= PBeam / νBeam [J/m]

(2)

2.3. Thermal diffusion length The thermal diffusion length can be described as the distance of heat dissipation from the input position where the initial heat flux is reduced to 1/e of its original magnitude. Equation 3 is the approximation of a solution of Fourier’s law of heat conduction valid for the used heating source and the examined axis of heat conduction perpendicular to the melt lines [12, 13]. It defines the thermal diffusion length Lth in dependence of the thermal diffusivity a and the return time treturn, which is the mean time till the heat source (electron beam) comes back to the starting point neglecting the shift through the line offset. Lth=2 (a·treturn)1/2 [m]

(3)

treturn is defined by the mean distance sreturn and the deflection speed νBeam (4): treturn= sreturn / νBeam [s]

(4)

3. MATERIAL AND METHODS 3.1. Raw material: Cu-powder The powder used is a 99.945 wt.% pure copper powder. It is gas atomized with nitrogen and shows a particle size distribution of DV10 = 45.6 µm, DV50= 68.0 µm and DV90= 95.8 µm. It features following measures: Tap density of 5.7 g/cm³, bulk density of 5.07 ± 0.01 g/cm³ and flow rate of 20.9 ± 0.18 s/50g. Figure 2-1: Schematic of the process chamber and the electron beam gun in a SEBM setup. The electron beam gun emits, focuses and deflects electrons with magnetic lenses. In the middle of the process chamber there is a building tank with a movable process platform inside. On the side of the platform there are two powder hoppers, feeding a rake which creates the powder layers [11].

3.2. Experimental setup The created 3D-CAD file for sample production consists of cuboids with the size of 10·15·15 mm³ (height · width · depth) which are connected to the start plate (15·150·120 mm³) with 5 mm long, cylindrical supports. This means that the process starts with building up the supports on the start plate, followed by the cuboids. The SEBM system used is an Arcam AB A2. The constant parameters for all experiments are shown in TABLE 1. The scan strategy during melting follows “snake setting”, which means the beam meanders over the surface with a line offset of 0.1 mm. The line energy and the deflection speed are modified in between 0.1 J/mm and 0.4 J/mm as well as 2 mm/s and 10 mm/s respectively. TABLE 1: CONSTANT PROCESS PARAMETERS Process pressure (Helium)

Figure 2-2: SEBM-process cycle for one layer [11].

The beam power PBeam is described by the beam current IBeam multiplied by the acceleration voltage UBeam (1): PBeam = IBeam · UBeam [W]


(1)

mbar

2∙10-3

Layer thickness

μm

50

Acceleration voltage

kV

60

Process temperature

°C

400

3.3. Sample preparation After SEBM processing, first slightly sintered powder around the components is removed by particle blasting. Then the samples can be disconnected from the start plate. Afterwards microsections out of the as-built components are prepared for analysis of porosity and melting depth as follows: Cutting

130



through the center, grinding to P4000, polishing to 1 µm, etching with H2O/EtOH/Fe(NO3)3 and etching with CuM1. 3.4. Characterisation The analysis of the prepared samples includes the size of the melting depth measured at microsections of etched samples with a Leica M205 C light microscope. The analysis of the surface (top view) roughness is conducted with an Olympus Lext OLS 4000 on two scan areas with 15.2 mm² each. 4. RESULTS AND DISCUSSION The process window for the production of the dense copper components is described elsewhere [14]. Briefly, the components are dense (rel. density is higher than 99.5 %) when the line energy exceeds 0.275 J/mm. This is independent from the beam velocity. Samples with higher or lower line energy show higher porosity. In comparison to the minimum line energy for dense components in Ti-Al6-V4, the required energy for dense copper parts is higher. For Ti-Al6-V4 the process window shows a minimum line energy of about 0.2 J/mm [15]. This difference can be explained by the higher thermal diffusion length of copper (see Table 2). 4.1. Melting depth The etched microsections display not only the SEBM-typical columnar microstructure, but also the melting depth (see Figure 4-1). The melting depth equals the thickness of the last molten layer. In Figure 4-2, the dependency of the melting depth measured at 5 positions for each sample is shown. The melting depth increases linearly with rising line energy. The measured melting depths are between 68 µm for a line energy of 0.2 J/mm and 138 µm for line energy of 0.4 J/mm. The single measurement points display a high scatter. For line energies below 0.2 J/mm the sample porosity increases to an extent that melting depth analysis is impossible.

Figure 4-2: Melting depth versus line energy. The melting depth rises linearly with increasing line energy from 68 µm to 138 µm.

4.2. Thermal diffusion length TABLE 2 shows the theoretical thermal diffusion length of copper and Ti-Al6-V4 at various beam velocities. The shorter the time till the beam returns is, the smaller the thermal diffusion length. In Ti-Al6-V4 for a beam deflection speed of 6 mm/s the thermal diffusion length is 0.1 mm, eleven times smaller than in copper (1.1 mm) due to the lower thermal diffusivity. This points out the high heat loss in SEBM of copper and shows the need of higher thermal input for melting copper in comparison to Ti-Al6-V4. 4.3. Dimensional accuracy Figure 4-3 shows the mean width of the cuboids in dependence of the beam power. The width was measured in xand y-direction. All cuboids are bigger than the specified size of 15 mm. A linear increase of the beam power results in linear growth of the cuboid width. The reason for this can be found by analyzing the beam path. The beam moves along the cuboids’ edges with the center of the beam spot on the specified outside. Since half of the beam spot is outside the specified geometry, energy is deposited outside of it and the cuboids grow bigger. Assuming a spot size of 400 µm, the measured width of the cuboid would be 15 mm plus two times the half spot size which is in good accordance to the minimum found cuboid width. The linear increase in cuboid width can be the result of an increasing beam spot size with rising beam current. Accurately sized components would be produced by undersized 3D-CAD files though. A second reason for the increasing cuboid width can be the higher thermal input per time as the return time decreases with an increase in deflection speed causing an enlargement of the melting zone of the beam.

Figure 4-1: Light microscopic image showing an etched sample for evaluation of the melting depth (arrow). The upper surface is the top surface of the cuboid samples giving an impression of surface roughness.


TABLE 2: THEROETICAL THERMAL DIFFUSION LENGTH OF COPPER AND TI-AL6-V4 AT VARIOUS BEAM DEFLECTION SPEEDS. Copper 3

Ti-Al6-V4

νBeam

m/s

6

sreturn

m

0.015

treturn

s

0.0025

a

m²/s

Lth

mm

0.000123 [16] 1.5

1.1

3

6

0.00003 [17] 0.2

0.1

131





Dense components (above 0.275 J/mm) show a roughness of about 15 µm.

6. ACKNOWLEDGMENT The authors wish to acknowledge financial support from the AMAZE Project, which is co-funded by the European Commission in the 7th Framework Programme (contract FP72012-NMP-ICT-FoF-313781), by the European Space Agency and by the individual partner organizations. Furthermore the authors would like to thank the Bavarian state for the funding of the Application center VerTec. Figure 4-3: Mean width of 42 cuboid samples in denpendency of the beam power. The width increases linearly with increasing beam power.

4.4. Surface arithmetic average roughness The surface roughness on the topside of the cuboids versus the line energy is displayed in Figure 4-4. It demonstrates that the surface arithmetic average roughness decreases with increasing line energy. Dense samples (line energy above 0.275 J/mm [15]) show a roughness of around 15 µm. Below 0.275 J/mm line energy roughness increases up to 45 µm due to incomplete melting which leaves some unmolten particles on the surface. One can assume that 10 % of the particles are bigger in diameter than Dv90 = 95.8 µm. If the beam is too weak for complete melting of the particles, the roughness is determined by the 10 % biggest particles with 95.8 µm diameter which are lying on top of the old melt surface in the new 50 µm thick powder layer. Thus the big particles top the layer thickness by 45.8 µm. which is in good accordance with the measured roughness for non-dense samples given in Figure 4-4.

REFERENCES [1] [2] [3] [4] [5] [6] [7]

[8]

[9]

[10] [11] [12] Figure 4-4: Surface arithmetic average roughness of the sample surface (top view) in dependency of the used line energy in the melting process. The threshold for dense samples is a line energy of 0.275 J/mm, below which an increase in surface roughness is evident (15 µm vs. up to 45 µm).

5. CONCLUSION 

The melting depth increases linearly with increasing line energy. The measured melting depths are between 68 µm and 138 µm for line energies of 0.2 J/mm and 0.4 J/mm respectively.



For dense manufacturing of copper a high energy density and fast melting is necessary due to the high thermal diffusivity resulting in thermal diffusion lengths in the range of millimeters



A linear increase of the beam power results in a linear increase of the cuboid width.


[13] [14] [15] [16] [17]

Gibson, I., Rosen, D.W., Stucker, B. 2010, Additive manufacturing technologies: Rapid prototyping to direct digital manufacturing. New York, Springer. Schwerdtfeger, J., Schury, F., Stingl, M., Wein, F., Singer, R.F., Körner, C., 2012, Mechanical characterisation of a periodic auxetic structure produced by SEBM, Phys. Status Solidi B ;vol. 249, pp. 1347–52. Zhai, Y., Lados, D.A., LaGoy, J.L., 2014, Additive Manufacturing: Making Imagination the Major Limitation, JOM, vol. 66, pp. 808–16. BGR, A.D. , 2012, Rohstoffsituationsbericht Zhu, H.H., Lu, L., Fuh, J., 2003, Development and characterisation of direct laser sintering Cu-based metal powder, Journal of Materials Processing Technology, vol. 140, pp. 314–7. Gu, D., Shen, Y., Fang, S., Xiao, J., 2007, Metallurgical mechanisms in direct laser sintering of Cu–CuSn–CuP mixed powder, Journal of Alloys and Compounds, vol. 438, pp. 184–9. Tang, Y., Loh, H.T., Wong, Y.S., Fuh, J., Lu, L., Wang, X., 2003, Direct laser sintering of a copper-based alloy for creating three-dimensional metal parts, Journal of Materials Processing Technology, vol. 140, pp. 368–72. Ramirez, D.A., Murr, L.E., Martinez, E., Hernandez, D.H., Martinez, J.L., Machado, B.I., Medina, F.,2011, Novel precipitate–microstructural architecture developed in the fabrication of solid copper components by additive manufacturing using electron beam melting, Acta Materialia ,vol. 59, pp. 4088–99. Frigola, P., Harrysson, O.A., Horn, T.J., West, H.A., Aman, R.L., Rigsbee, J.M., Ramirez, D.A., 2014, Fabricating copper components with electron beam melting, Advanced Materials and Processes, vol. 172, pp. 20–4. Anderl, P., 1989, Vergleich von Laser- mit Elektronenstrahlschweißen, Chemie Ingenieur Technik, vol. 61, pp. 767–74. Heinl, P., Rottmair, A., Körner, C., Singer, R.F., 2007, Cellular Titanium by Selective Electron Beam Melting, Adv. Eng. Mater., vol. 9 pp. 360–4. Carslaw, H.S., Jaeger, J.C., Conduction of heat in solids. 1986, New York, Clarendon Press, Oxford University Press. Helmer, H.E., Körner, C., Singer, R.F., 2014, Additive manufacturing of nickel-based superalloy Inconel 718 by selective electron beam melting: Processing window and microstructure, J. Mater. Res., pp. 1–10. Lodes, M.A., Guschlbauer, R., Körner, C., 2014, submitted. Juechter, V., Scharowsky, T., Singer, R.F., Körner, C., 2014, Processing window and evaporation phenomena for Ti–6Al–4V produced by selective electron beam melting, Acta Materialia, vol.76, pp.252–8. Laskar, J.M., Bagavathiappan, S., Sardar, M., Jayakumar, T., Philip, J., Raj, B., 2008, Measurement of thermal diffusivity of solids using infrared thermography, Materials Letters, vol. 62, pp. 2740–2. Komanduri, R., Hou, Z., 2002, On thermoplastic shear instability in the machining of a titanium alloy (Ti-6Al-4V), Metall and Mat Trans A, vol. 33, pp. 2995-3010.

132



Effects of the Selective Laser Melting process parameters on the functional properties of the Co-Cr alloy. Dariusz Grzesiak, West Pomeranian University of Technology, Szczecin, Poland, [email protected] Marta Krawczyk, West Pomeranian University of Technology, Szczecin, Poland, [email protected]

Abstract— The purpose of the present study is to investigate the impact of technological parameters on the mechanical properties of the Co-Cr alloy after the Selective Laser Melting (SLM) process. The energy density (ε) and the number of the laser beam passes on each melted layer were the variables. The static tensile test was used to evaluate the mechanical properties of the manufactured samples. The results of the study show significant differences in material characteristics resulting from different number of the laser beam passes over the programmed path. Keywords-SLM; CoCr; energy density; laser

1. INTRODUCTION In recent years a dynamic development of the modern manufacturing technologies enabling quality improvement and increase of geometric complexity of manufactured products and reducing the time required to put them on the market is observed. Among them the additive manufacturing technologies, classified at the beginning as the methods of rapid prototyping (RP) [1], which is often called 3D printing, deserve a special attention. Now they constitute a distinct group of methods more and more widely used in industry and being the subject of many scientific studies. One of the fastest growing manufacturing technologies is the method of Selective Laser Melting (SLM) [1 5]. SLM is a widely used method for manufacturing three-dimensional parts by the selective melting of metal powder layer by layer using a laser [2, 6, 7]. This method, compared to laser sintering (SLS), requires a higher energy level, which is usually achieved by using high power laser and thin layer of powder [2]. In the preparation by SLM technology the geometric models generated by the CAD software are directly used [1, 6, 8, 9]. Thanks to this, the method has almost unlimited possibilities with regard to the shape of the final product. High efficiency of the process, high density, good anticorrosive properties, good aesthetic appearance, the possibility of using many types of materials (i.a. metal matrix nanocomposites [10 13]), and the possibility to individualize the process influenced the application of this technique to the production of prosthetic restorations and attracted the attention of dentists [6, 14 17]. In prosthetic dentistry, most of the research on SLM method focused on CoCr alloys [9], from which the elements so far were obtained by casting [6]. It is worth adding that the Co-Cr alloys have good corrosion resistance, good wear resistance and excellent


biocompatibility [2, 8]. The use of SLM process for the production of implants entails many challenges, which can include obtaining high density, full control of the manufacturing process, appropriate hardness and wear resistance. This articles presents the results of research on the effects of technological parameters of the SLM process on the mechanical properties of the samples made of CoCr ASTM F75 alloy. 2. MATERIALS AND METHODS The tensile bars were manufactured by the REALIZER II 250 (MTT-Group) device, equipped with 100W Nd:YAG fiber laser. During the SLM process, the powder was applied in layers of 30µm. On each layer, the powder was melted by laser beam at the locations corresponding to the cross section of the sample at a given level. At this stage of the process, newly formed layer was permanently bonded to the so far prepared part of the manufactured element. During the melting of each layer, the laser beam, directed by the mirror system, passes the paths programmed by the control computer. Each path is composed of points corresponding to the single pulses of laser beam. The distance between the points and the paths composed of them, the time of the laser beam exposition at each point and the thickness of the powder layer, are the technological parameters. The values of those parameters have direct influence on the final effect of the SLM process. The SLM process scheme and the meaning of characteristic parameters are shown in Fig. 1. The samples were made of the CoCr ASTM F75 alloy powder supplied by the REALIZER device manufacturer. The chemical composition of the alloy is presented in table 1. During the experiment, twelve sets of samples were made. Each set contained six samples. Sets differed in the volumetric energy density delivered by the laser beam (ε [J/mm3]), which is expressed by (1) and depends on the following technological parameters of the SLM process: laser power – P [W], velocity of the laser beam movement on the path – V [mm/s], distance between the laser beam paths – h [mm], thickness of the melted layer – d [mm]. 

 = P / (V·h·d).



The velocity of the laser beam movement is expressed by (2) and depends on the following parameters: a [mm] – the

133



distance between the laser beam pulses, t [s] – the laser beam exposition time on a single point. V = a / t.





An additional sample variation was the number of the laser beam passes on computed paths. In the first case, the laser beam moved one time on each path, in the second case – two times. The full set of the samples is shown in table 2. After the SLM process was finished and specimen supports removed, the samples were finished according to the standard procedure.

TABLE II.

Titanium, Ti