Modeling the product development process: the RIM case study

Modeling the Product Development Process: the RIM Case Study Ricardo Nuno de Oliveira Bastos Torcato January 2013 Thesis submitted to the Faculty of Engineering of the University of Porto for the degree of Doctor of Philosophy in Leaders for Technological Industries of the MIT-Portugal Program iii Supervisor António Augusto Fernandes Faculdade de Engenharia da Universidade do Porto, Portugal Co-Supervisors Ricardo J. Santos Faculdade de Engenharia da Universidade do Porto, Portugal Richard Roth Massachusetts Institute of Technology, USA v Abstract Early assessment of a material’s appropriateness and its inherent production technology applicability in a particular product is fundamental for it to be considered in further product development (PD) steps. Reaction Injection Molding (RIM) technology is noticeably under this paradigm. Often, the development team does not consider this technology in the selection process due to unawareness about it among the members of the team and qualified suppliers. This makes RIM a proper case study to contribute to a better understanding about the PD process, and particularly, the decision-making processes related to the adoption of new materials in the product and new technologies in the production process. As such, an empirical study was undertaken with two main objectives: i) model PD projects where new (unfamiliar) technology could be adopted to add empirical evidence about the factors and practices that benefit the performance of the PD process; ii) assess the suitability of Business Process Management (BPM) modeling approach to the PD process by the development of a modeling framework and test in the projects under study. This thesis presents two case studies of PD projects, modeled using the modeling framework developed in this research, based on the Riva method accompanied by textual narrative. The procedure for the implementation of the modeling framework was based on open-ended interviews to the experts involved in each project. These case studies provide important insights on the constraints that inhibit the consideration of other materials and technologies (and RIM in particular) in the PD process vi of rigid parts. Nevertheless, our analysis reveals scope for improvement in dealing with uncertainty and decision making by the introduction of the concept of postponement supported by practices and techniques that enhance flexibility. Moreover, we propose a modified framework of factors affecting the success of PD projects. Finally, we conclude that the Riva method supported by textual narrative proved to be a good modeling approach to understand the decision processes in these particular PD projects. vii Resumo A avaliação precoce da adequação de um material e da aplicabilidade da sua tecnologia de produção a um determinado produto é fundamental para que possa ser considerado nos passos seguintes de desenvolvimento de produto (PD). A tecnologia de moldação por injeção e reação (RIM) está visivelmente neste paradigma. Muitas vezes, a equipa de desenvolvimento não considera esta tecnologia no processo de seleção devido ao desconhecimento sobre a mesma entre os membros da equipa e fornecedores qualificados. Por esta razão um estudo de caso que considere a tecnologia RIM poderá contribuir para uma melhor compreensão sobre o processo de PD e, em particular, no que se refere aos processos de tomada de decisão relacionados com a adoção de novos materiais no produto e novas tecnologias no processo de produção. Neste sentido foi realizado um estudo empírico com dois objetivos principais: i) modelar projetos de PD, onde poderia ser adotada uma nova tecnologia (que a equipa de projeto não conhece) de forma a acrescentar evidência empírica sobre os fatores e práticas que beneficiam o desempenho do processo de PD, ii) avaliar a adequação da abordagem de modelação baseada na Gestão de Processos de Negócio (BPM) ao processo de PD através do desenvolvimento de um framework de modelação e aplicação nos projetos em estudo. Esta tese apresenta dois estudos de caso de projetos de desenvolvimento de produto, modelados com o framework de modelação desenvolvido nesta investigação, com base no método Riva complementado por narrativa textual. O procedimento para a aplicação do framework foi baseado em entrevistas abertas aos especialistas envolvidos em cada projeto. viii Estes estudos de caso evidenciaram perspetivas importantes a respeito de constrangimentos que inibem a consideração de outros materiais e tecnologias (e RIM em particular) no processo de PD de componentes rígidos. No entanto, a análise revela possibilidades de melhorias na gestão da incerteza associada à tomada de decisão pela introdução do conceito de postponement apoiado em práticas e técnicas que melhoram a flexibilidade. Propomos um quadro modificado de fatores que influenciam o sucesso de projetos de PD. Por fim, concluímos que o método Riva complementado por narrativa textual provou ser uma abordagem de modelação exequível para compreender a tomada de decisões nos projetos de PD estudados. ix Résumé L'évaluation précoce de la pertinence d'un matériel et son applicabilité technologique inhérente à la production d’un autre produit est fondamentale pour qu'il puisse être pris en compte au cours des diverses étapes de développement d’un produit (PD). C’est pourquoi Injection Molding Réaction (RIM) fait clairement partie de ce paradigme. Souvent, l'équipe de développement ne considère pas cette technologie dans le processus de sélection, généralement par méconnaissance de la part des membres de l'équipe et des fournisseurs qualifiés. Ainsi, RIM s’affirme comme une étude de cas, pouvant fortement contribuer à une meilleure compréhension du processus PD, en particulier, des processus de prises de décisions liés à l'adoption de nouveaux matériaux dans la fabrication du produit et à l’application des nouvelles technologies dans le processus de production. C’est pour cela qu’une étude empirique a été réalisée, visant deux principaux objectifs: i) modèle des projets de PD où une nouvelle technologie pourrait être adopté à additionner des données empiriques à propos des facteurs et des pratiques qui bénéficieraient la performance du processus de PD; ii) évaluer l'adéquation de l'approche de modélisation basée sur la Gestion des Processus Métiers (BPM) par le processus de PD et grâce au développement d'un cadre de modélisation et de test des projets en étude. Cette thèse présente deux études de cas de projets de développement de produits, qui ont été modélisés à l'aide du cadre de modélisation développé dans cette recherche; celle-ci basée sur la méthode Riva et accompagnée de narration textuelle. La procédure suivie pour la mise en œuvre du cadre de modélisation a été basée sur des interviews ouvertes aux experts impliqués dans chaque projet. xvi Figure 2.14 Core set of BPMN elements (Source: http://www.omg.org/bpmn/Samples/ Elements/Core_BPMN_Elements.htm) .............................................................. 53 Figure 2.15 IDEF0 box and arrow graphics (Source: http://www.idef.com) .................. 58 Figure 2.16 Petri nets graphical notation ........................................................... 60 Figure 2.17 Design Structure Matrix (Smith and Morrow 1999) .................................. 61 Figure 2.18 The RAD notation (Ould 2005) ......................................................... 66 Figure 2.19 The central problem of materials selection: the interaction between function, material, shape and process (Ashby 2005) ......................................................... 69 Figure 2.20 Materials in the design process (Ashby and Johnson 2002) ........................ 74 Figure 2.21 Four-step materials selection procedure: translation, screening, ranking, and supporting information (Ashby 2005) ................................................................ 75 Figure 2.22 Four-step process selection procedure. It is performed in parallel with the material selection. (Ashby 2005) .................................................................... 81 Figure 2.23 Scheme of a typical RIM process (Bayer 1995) ....................................... 83 Figure 2.24 Spectrum of PU systems ................................................................ 83 Figure 2.25 Example of a typical RIM parts assembly ............................................. 84 Figure 2.26 Qualitative positioning of RIM process (Vervacke 2009)............................ 85 Figure 3.1 Stage 1 research scheme ................................................................. 91 Figure 3.2 Stage 2 research scheme ................................................................. 98 Figure 4.1 Process modeling framework .......................................................... 104 Figure 4.2 Modeling procedure for the case studies research ................................. 105 Figure 4.3 Buddy electric vehicle .................................................................. 108 Figure 4.4 Standalone heat pump. ................................................................. 109 xvii Figure 5.1 Overview of the plastic industry from source to products (Rosato, Rosato, and Rosato 2004) .......................................................................................... 113 Figure 5.2 External value network of the RIM industry ......................................... 115 Figure 5.3 Positive and negative attributes of RIM process .................................... 118 Figure 5.4 RIM process flow ......................................................................... 121 Figure 5.5 Generic product development process at CEIIA .................................... 125 Figure 5.6 Handle a product development project: the Buddy case study .................. 127 Figure 5.7 Generic product development process at Bosch Termotecnologia. .............. 135 Figure 5.8 Handle a product development project: the water heater case study .......... 137 Figure 5.9 CEIIA’s flexible PD process based on activities overlapping, frequent iterations, testing and short milestones........................................................................ 145 Figure 5.10 Bosch’s quality gates and gatekeeper’s go/kill decisions ........................ 146 Figure 5.11 Bosch’s management control of the project’s KPIs ............................... 147 Figure 5.12 Modified Brown and Eisenhardt’s framework ...................................... 153 Figure 5.13 Bosch's material and technology selection activities ............................. 155 Figure B.1 Handle a product development project: the Buddy case study .................. 193 Figure B.2 Handle a product development project: the water heater case study .......... 195 xix List of Tables Table 2.1 Differences between a business process and an engineering design process (Vajna 2005) ..................................................................................................... 11 Table 2.2 Product development process with tasks and responsibilities (Ulrich and Eppinger 2008) ..................................................................................................... 23 Table 2.3 New product development best practices (adapted from Kahn et al. 2012) ...... 32 Table 2.4 Core initiatives for concurrent engineering programs (adapted from Swink 1998) ............................................................................................................ 37 Table 2.5 Methods and tools for each phase of DMADV (Ginn and Varner 2004) .............. 45 Table 2.6 Methods and tools for each phase of IDOV ............................................. 46 Table 2.7 Four categories of purposes for PD process modeling (Browning and Ramasesh 2007) ..................................................................................................... 57 Table 3.1 Characteristics of the two main research stages ...................................... 89 Table 3.2 RIM users studied in stage 1 of the research ........................................... 92 Table 3.3 Companies involved in stage 2 of the research ........................................ 95 Table 5.1 RIM tooling options ...................................................................... 122 xx Table 5.2 Factors that determined the selection of DCPD-RIM ................................ 132 Table 5.3 Factors that determined the selection of EPP ....................................... 143 Table A.1 Sample of RIM equipment manufacturers ............................................ 185 Table A.2 Sample of PU systems suppliers ....................................................... 187 Table A.3 Sample of DCPD systems suppliers .................................................... 188 Table A.4 Sample of RIM users ..................................................................... 189 xxi List of Acronyms BMC: Bulk Molding Compound BOM: Bill of Materials BPEL: Business Process Execution Language BPM: Business Process Management BPMI: Business Process Management Initiative BPMN: Business Process Model and Notation CAD: Computer Aided Design CAE: Computer Aided Engineering CAM: Computer Aided Manufacturing CE: Concurrent Engineering CFD: Computational Fluids Dynamics CP: Case process CSM: Composite Spray Molding CTQ: Critical To Quality xxii DCPD: Dicyclopentadiene DFA: Design for Assembly DFM: Design for Manufacturing DFSS: Design for Six Sigma DFX: Design for X DMADV: Define, Measure, Analyze, Design and Verify DMAIC: Define, Measure, Analyze, Improve and Control DOE: Design of Experiments DPD: Dynamic Product Development DSM: Design Structure Matrix ESD: Entrepreneur System Designer EPP: Expanded polypropylene FEM: Finite Element Method FMEA: Failure Mode and Effects Analysis (DFMEA: Design FMEA; PFMEA: Process FMEA) FRP: Fiber Reinforced Plastic GERT: Graphical Evaluation and Review Technique GRP: fiber-Glass Reinforced Polyester ICAM: Integrated Computer Aided Manufacturing ICOV/IDOV: Identify (requirements), Characterize/Design, Optimize and Verify (the design) IDEF: ICAM Definition / Integrated Definition IPD: Integrated Product Development IPDS: Integrated Product Delivery System KPI: Key Performance Indicators xxiii LCE: Life Cycle Engineering LFT: Long Fiber Thermoplastic LPD: Lean Product Development NDA: Non-Disclosure Agreement NPD: New Product Development OEM: Original Equipment Manufacturer PAD: Process Architecture Diagram PCP: Product Creation Phase PD: Product Development PDCPD: Polydicyclopentadiene PDD: Product Design and Development PDMA: Product Development and Management Association PDP: Product Development Process PERT: Program Evaluation and Review Technique PSC: Project Steering Committee PU: Polyurethane QFD: Quality Function Deployment RAD: Role Activity Diagram RASIC: Responsible, Accountable, Support, Informed, and Consulted RIM: Reaction Injection Molding RIMcop®: Reaction Injection Molding with Control of Oscillation and Pulsation RRIM: Reinforced RIM RTM: Resin Transfer Molding xxiv SBCE: Set-Based Concurrent Engineering SE: Simultaneous Engineering SIPOC: Supplier, Input, Product, Output, Customer map SMC: Sheet Molding Compound SME: Small and Medium Enterprise SRIM: Structural RIM STRIM: Systematic Technique for Role and Interaction Modeling TP: Thermoplastic TRIZ: Theory of inventive problem solving TS: Thermoset TTM: Time To Market UML: Unified Modeling Language UOW: Unit of work VOC: Voice of the Customer xBML: Extended Business Modeling Language 1 1 Introduction 1.1 Relevance and Motivation In the development process of products and its subsystems the development teams have to make several important decisions that have great impact in the project’s profitability. These decisions have a major influence in the development lead time and cost and also in the market success of the product (Cooper 2001; Cooper and Edgett 2005; Smith 2007). Early assessment of a material and inherent production technology applicability in a particular product is fundamental for it to be considered in further product development steps. The organizational pressure for speed and efficiency often results in conservative decisions, reducing the chances of studying and selecting other materials and technologies (Bayus 1997; Lukas and Menon 2004). Reaction Injection Molding (RIM) technology is noticeably under this paradigm. Due to the lack of knowledge about this technology, designers usually neglect it. However, given the proper conditions RIM may be the best option (important factors are functional requirements and expected production volumes). RIM technology for processing polyurethanes (PU) or dicyclopentadiene (DCPD) rigid parts is mostly used in enclosures (also called exterior panels or body of the product), which makes it a good case for study, because these components must usually be developed to differentiate the product, unlike many other components that can be purchased and/or shared (carryovers). In order to have a deeper understanding about the mechanisms behind the abovementioned scenario, it is necessary to first describe and analyze the product development Chapter 2 8 practices for PD. The second section addresses the descriptive approaches, namely the issue of modeling PD processes. The third section discusses materials and technology adoption methods and strategies. This is followed by a section on RIM technology. We conclude this chapter with the research questions, which prompted our research, and their relevance to our objectives. 2.1 Product Development Methods and Practices Literature on PD is vast. Several methodologies, methods, techniques and tools can be found in literature (Cooper 2001; Pahl et al. 2007; Ulrich and Eppinger 2008). All of these have mainly a prescriptive nature (Motte 2008; Gericke and Blessing 2011), that is to say they provide a procedural plan for the design process. They assist in identifying what has to be done, but they must be adapted to the specific PD projects. These are mostly based on a step-by-step problem-solving approach, a succession of tasks and decisions. As knowledge increases, the PDP reaches its objective (solution). Thus, they do not explain what and how things are done in practice; rather they give (mostly generic) orientation to the development team. Of course, those recommendations are based in practical experience and empirical evidences (Koskela 2000), yet they do not suffice to perform a PD project. 2.1.1 Product Development: an Overview Successful PD is essential in today’s competitive market as it is seen as a potential source of competitive advantage (Brown and Eisenhardt 1995). It is therefore not surprising that, over the years, noticeable standardization efforts (MacGregor et al. 2006) have emerged. In fact, while on the one hand PD process is understood as being unique, unlike other business and production processes, literature on PD relentlessly tends to provide recipes for successfully producing a product. Yet, within this trend, from the most recent literature it is possible to identify nonconformists, advocating that PD complexity and uniqueness cannot be tackled using the same models and practices, given the proven influence issues such as context and intervenients have. It is within these ambivalent bounds that we develop the points that follow. Product Development as a Process Various definitions of process suggest different managerial and technical perspectives, which can underpin this essential concept within business in general and PD in particular. Amongst the first authors to focus on the importance of reengineering the business Literature Survey 9 process, Hammer and Champy (1993) viewed it as a “collection of activities that takes one or more kinds of input and creates an output that is of value to the customer”. Similarly, Davenport (1993) also focused his attention on the “specific ordering of work activities across time and space, with a beginning and an end, and clearly defined inputs and outputs”. Curiously, at approximately the same time Ould (1995) argued that actual processes are not as orderly as Davenport’s and Hammer and Champy’s input–output view might suggest. Ould considered business processes as networks, where roles collaborate and interact to attain a given business goal. Thus, Ould (2005) defines a process as “a coherent set of activities carried out by a collaborating group to achieve a goal”. Later, Becker et al. (2003) defined a business process as “a special process that is directed by the business objectives of a company and by the business environment. Essential features of a business process are interfaces to the business partners of the company (e.g. customers, suppliers). Examples of business processes are the order processing in a factory, the routing process of a retailer or the credit assignment of a bank”. In all perspectives, except that of Ould, the focus is on the fact that a business process is repeatable and knowledge regarding its development is readily available. This contrasts with that of process within the realm of PD. In the latter, knowledge about a product is progressively generated as part is unknown in the beginning, which generates uncertainty. Design processes (as is the case of PD) encompass problem-solving characteristics (Lindemann, Maurer, and Braun 2009), which imply a degree of novel knowledge, common in the conception of new products. This uncertainty is gradually overcome through process iterations, however the fact remains that one of the main and unique characteristics of PD is its dynamic nature (see Section 2.1.2 and 2.2). The development of new products is a process which transforms opportunities into tangible products intended to produce company profits (Trott 2012). As the author acknowledges, the process itself is complex and therefore difficult to identify. Clark and Fujimoto established this idea of PD as a process in 1991. Drawing on six years of research conducted at the Harvard Business School on how different manufacturing firms around the world approach the development of new products, Clark and Fujimoto (1991) posit PD as a “process by which an organization transforms data on market opportunities and technical possibilities into information assets for commercial production”. In their point of view, PD not only considers concept generation, product planning and product engineering, but also process engineering, which creates the process. This pioneering Chapter 2 10 definition marked the beginning of the process approach in PD. Until then the engineering sector was responsible for PD, thus disregarding the integration of marketing activities, planning and product manufacture. Other authors (see for example, Hales 1993; Smith and Morrow 1999; Cooper and Edgett 2005; Browning, Fricke, and Negele 2006; Pahl et al. 2007) draw on this idea of PD as a complex process given the wide range of issues that must be addressed, as well as the diversity of people and the different organizational structures involved in the PD efforts. PD thus entails a considerable number of decisions in an extremely uncertain environment. Curiously, Ulrich and Eppinger (2008) amongst the most cited authors in product design and development define “Product development as a set of activities beginning with the perception of a market opportunity and ending in the production, sale and delivery of a product”, thus omitting the concept of process, opting for a generic approach to PD. Browning et al. (2006), drawing on their vast experience in PD, clarify some of the main concepts used as they consider processes [our emphasis] to be the key aspect of contemporary systems engineering theory and practice. They describe PD as “an endeavor comprised of the myriad, multi-functional activities done between defining a technology or market opportunity and starting production”, whose main objective is to create a “recipe” for producing a product (Browning et al. 2006 with credit given to Reinertsen 1997). The authors draw on Kline (1985) arguing that PD implies creativity and innovation and is nonlinear and iterative. By adopting Hammer’s (2001) view of process as “an organized group of related activities that work together to create a result of value” or Pall’s (2000) notion of process as “a network of customer-supplier relationships and commitments that drive activities to produce results of value”, Browning and his colleagues clearly establish PD as a complex process, asserting that any work (including work related to creativity and innovation) that is carried out in order to obtain a result has an underlying process. In sum, PDP can be defined as “a disciplined and defined set of tasks and steps that describe normal means by which a company repetitively converts embryonic ideas into saleable products or services” (Kahn, Castellion, and Griffin 2005). Based on the work of Vajna (2005) we summarize the main differences between business processes and PD processes in Table 2.1. Literature Survey 11 Table 2.1 Differences between a business process and an engineering design process (Vajna 2005) Business process Engineering design process  Processes are fixed, rigid, have to be reproducible and checkable to 100%  Results have to be predictable  Material, technologies, and tools are physical (e.g., in manufacturing) and/or completely described (e.g., in controlling)  Possibility of disruptions is low, because objects and their respective environments are described precisely  No need for dynamic reaction capability  Processes are dynamic, creative, chaotic; many loops and go-tos  Results are not always predictable  Objects, concepts, ideas, designs, approaches, trials (and errors) are virtual and not always precise  Possibility of disruptions is high, because of imperfect definitions and change requests  There is definitive need for dynamic reaction capabilities The Product Development Project Another concept that is ever present in PD literature is project. While the concept is widely used in PD, very few authors actually define it. Nonetheless, in other scientific areas, namely in Project Management, several authors have attempted to define the concept of project. Noticeably over the last years we have witnessed a shift in the notion of project (Shenhar, Levy, and Dvir 1997). One of the distinguishing characteristics between project and other kinds of organizations is its uniqueness. Mintzberg (1983) focuses on this distinguishing feature by defining project as “an organizational unit that solves a unique and complex task”. Weiss and Wysocki (1992) enumerate a series of characteristics they consider delimit the concept of project. Although they do not provide a clear definition, they concede a project must have the following characteristics:  Complex and numerous activities.  Unique: a one-time set of events.  Finite: with a begin and end date.  Limited resources and budget.  Many people involved, usually across several functional areas in the organization.  Sequenced activities.  Goal-oriented. Chapter 2 12  End product or service must result. The Project Management Institute’s definition focuses on its time dimension, stating that a project is “a temporary endeavor undertaken to create a unique product or service” (2000). In turn, the British Standards Institute’s guide to project management defines project more precisely, as “a unique set of co-ordinate activities, with definite starting and finishing point, undertaken by individuals or an organization to meet specific performance objective within defined schedule, cost and performance parameters” (BS:6079 2000). It is our belief, that the modern approach to PD process (see Dynamic Model Section 2.1.2) embeds all the characteristics of a project. Towards a Modern Product Development Process In 1986, Takeuchi and Nonaka acknowledged that the long-established sequential approach to developing new products was no longer adequate, advocating a holistic method, placing emphasis on companies’ need for speed and flexibility in the ever growing competitive world. The authors advocated the need to move from a PDP where one group of functional specialists passed the product on to the next group - known as waterfall or “over-the-wall” (see for example Otto and Wood, 2001) (Figure 2.1), to a PDP emerging from the interaction of a multidisciplinary team whose members work together throughout the entire process, engaging in iterative experimentation throughout all the phases of the development process - known as CE (see Section 2.1.3). Figure 2.1 Sequential (A) vs. overlapping (B and C) phases of development (Takeuchi and Nonaka 1986) Literature Survey 13 Takeuchi and Nonaka (1986) criticized the segmented and sequential phase-to-phase process: concept development, feasibility testing, product design, development process, pilot production, and final production, where functions were specialized and segmented, being carried out by different people at different times, separately. They defend an integrated approach, encouraging trial and error, which challenged the status quo. Nonetheless, the authors admitted that their holistic approach to PD may not work in all situations, due to its limitations, namely in terms of team effort; project specificities, in particular hi-technology or chemistry related projects; project size, where project scale limits extensive face-to-face discussions and “masterminded” projects, where one person makes the invention and hands down a well-defined set of specifications for others to follow. The traditional conception of PDP was based on functional specialization and standardization (see Taylor 1913), which used expertise as a means of efficiency in the organizational processes. Within the traditional approach, the steps for PD are predetermined, which help control and manage the project. As each step is completed before the next begins, experts can focus their skills and experiences on a specific step or set of tasks. Taylor (1913, cited in Koskela 2000) explained this idea as follows: “The work of every workman is fully planned out by the management at least one day in advance, and every man receives in most cases complete written instructions, describing in detail the task which he is to accomplish, as well as the means to be used in doing the work”. DesChamps and Nayak (1995) summarize the following characteristics of the traditional conception of the PDP, as barriers to effective product creation:  Compartmentalized organization and sequential working: Departments absorb and mold the skills of the people who compose them (engineering, production, marketing, finance, and so on). The PDP is seen and operated in a fragmented way, each group concentrating on their share of work. Communication problems arise because experts often do not understand what is requested. Working sequentially generates errors because knowledge may be withheld or delayed.  Needless iterative loops: The authors contend, compartmentalized organization and sequential working leads to needless iterative loops, as there is no common overall goal of product creation. Thus, if a problem occurs in manufacturing, they will very likely blame the designers. Chapter 2 14  Heavy-handed hierarchy: In a functional structure, employees think vertically, because they depend on command, control and integration of departmental superiors. In sum, we can say that traditional approaches use simple, precedence models and thus cannot capture the iterative nature of the PDP (Eppinger et al. 1994; Browning 1997). Nonetheless, these have two clear advantages: it is easy to control and predictable. However, as Takeuchi and Nonaka claimed, in order to face the challenges of fast changing marketplaces, intense competition and rapid technological evolution of today’s corporate world, companies need to break with the old approach to PD and embrace a more modern approach. Indeed, flexible PD processes, which take into account uncertainties, adjust to the process of creating new ideas with practical applications and therefore increase the likelihood of success of PD (Cooper 1990). Smith (2007) for example, drawing on the notion of agile within the software domain, proposes a flexible approach to PD, defending that "flexible processes are emergent" [emphasis in original], in the sense that there are many interdependencies working together, and thus are not predictable. Adopting a flexible approach to PD can therefore "anticipate the future better". Hence, we can consider two broad methodological approaches to PD: a traditional, sequential methodology and a modern, more iterative methodology. Drawing on this idea of a modern approach to PD, literature has grown considerably over the last decades. Some literature describes engineering design and PD in a broad and systematic way as a guide for improved implementation (Rosenau and Moran 1993; Otto and Wood 2001; Cleland and Ireland 2006; Pahl et al. 2007; Ulrich and Eppinger 2008). These authors base their findings on examples of PD experiences and advocate the use of a structured, multi-stage PDP to manage the competing technical and market risks. Other authors (Craig 2001; Leonard-Barton 2007; MacCormack, Baldwin, and Rusnak 2012) relate PD practices with architectural or corporate strategy. PD literature is, in fact, heavily based on studies of individual companies’ successful efforts to improve their PD approach, that range from software products (Cusumano and Selby 1995; MacCormack 2000) to the automobile industry (see authors such as Cusumano 1991, who compares US and Japanese automobile manufacturers processes, or Morgan and Liker 2006 and Ward 2007, who focus on the Toyota product development system). Literature Survey 15 Despite the array of studies that attest to the fact that rapid and innovative PD can provide critical competitive advantages to firms (Rosenau and Moran 1993; Ulrich and Eppinger 2008), as far as we are aware of, there are currently no established criteria for selecting or designing ideal PD processes, nor is any single process ideal for all circumstances and companies. On this, Krubasik (1988) draws attention to the fact that “one size does not fit all” [emphasis in original], as not all PD is alike. Nonetheless, often managers do not consider the context and “tend to fall back, by intuition or reflex, on a kind of generic ‘one size fits all’ approach to new product development”. As Otto and Wood (2001) explain: “Every company has a different development process out of necessity; there is no single ‘best’ development process; the design process and the product development process are misnomers. The sophistication of the product, the competitive environment, the rate of change of technology, the rate of change of the system within which the product is used: these and many other factors that shape a product development process change for different companies”. Even so, most modern PD approaches do have a common stem, which encompasses general actions, steps or stages. These are, however, non all-inclusive. In fact, companies, depending on their specificity, may have either additional or fewer steps. PD process enacts these actions, steps or stages in an organized way. Understanding the different approaches is an important step to guide future PD processes. “Technology-Push” and “Market-Pull” Product Development Models PD models in general include the definition and sequencing of phases or stages, the objectives of each within the model, and methods or tools that developers use to accomplish those objectives. Given the array of new products and contexts in which these are developed, it is not feasible to present one generic model. Nonetheless, the existing literature does delineate two different types of PD models: technology-push product development and market-pull product development. Technology-push PD starts with a given technology and then incorporates that technology into products that better satisfy market needs – thus the technology “pushes” the PD. The development team begins with a new technology and then finds an appropriate market. Chapter 2 16 On the other hand, in market-pull PD, the incentive for the product derives from customer needs. Market-pull uses technologies to create a new product or improve an existing product to meet customer needs, thus these “pull” the PD through to completion (Bishop and Magleby 2004). Leading literature in PD predominantly focuses the market-pull model as predominantly implemented by companies (Otto and Wood 2001; Kahn, Castellion, and Griffin 2005; Ulrich and Eppinger 2008), despite the inferences that a combination of a technology-push and market-pull PD model would be optimal. Bishop (2004) argues technology-push PD is generally considered more difficult and challenging, thus the reason most PD research has focused on market-pull development and why many researchers favor market-pull over technology-push development. Indeed, market-pull is still the most widely used PD model in companies that develop physical complex products (Ulrich and Eppinger 2008; Terwiesch and Ulrich 2009), as in this situation customer needs have to be understood before developing the technology that materializes the final product. 2.1.2 The Different Prescriptive Methods Whether technology-push or market-pull, PDP can, as stated above, be described through a number of identifiable steps or stages that represent the generic PD flow from the new idea to the final product, which provide an overview of the complete process. Over the years this process has been represented by numerous different models (some in constant evolution), which attempted to integrate its main activities. We will now discuss the prescriptive PD process models we consider most relevant. Pahl and Beitz Pahl and Beitz (2007) conceived the earliest and most important model in engineering design in the late 1970’s (originally published in German, Konstruktionslehre). The authors proposed plans and procedures “mandatory for the general problem solving process of planning and designing technical products”, that must be looked upon as “operational guidelines for action” adapted to each specific situation. They proposed a step-by-step procedure, which focuses on three main themes: optimization of principle, optimization of layout and optimization of production (see Figure 2.2). Literature Survey 17 Figure 2.2 Steps in the planning and design process (Pahl et al. 2007) The workflow proposed by the authors is based on the fundamentals of technical systems, on the fundamentals of the systematic approach and the general problem solving process. The authors focus on a systematic design as a complete process. For Pahl and his colleagues the four main phases in the planning and design process are: 1. Planning and task clarification: First it is important to clarify the task in detail, i.e. collect information about requirements, constraints and their importance. The result should be the specification of information in the form of a requirements list. Chapter 2 24 1. Discovery: Stage 0 is where pre-work designed to discover opportunities and generate new ideas occurs. 2. Scoping: Stage 1 consists of a quick and preliminary investigation and scoping of the project (mainly desk research) to narrow the field before Stage 2. 3. Building the Business Case: It is in Stage 2 that a more detailed research (market and technical) occurs leading to a Business Case, which includes product and project definition, project justification and project plan. 4. Development: Stage 3 incorporates the actual detailed design and development of the product, as well as initial testing. It is a lengthy stage that includes full production and market plans. The outcome of the stage is a tested product. 5. Testing and validation: In Stage 4 the proposed new product, its marketing and production plans are tested and validated through tests/trials in the marketplace, lab and plant. 6. Launch: Finally, in Stage 5, commercialization and full production, marketing and selling occur. Plans for market launch, production/operations, distribution, quality assurance and post-launching monitoring are executed. Figure 2.6 The Stage-Gate® Model (adapted from Cooper and Kleinschmidt 2001) Cooper and Kleinschmidt (2001) emphasize that careful attention should be paid to the early stages, where pre-development work must be done before PD begins. Furthermore, the authors claim the right organizational structure is a key factor in success, advocating the need for multidisciplinary, cross-functional teams dedicated to the projects, accountable for them from idea to launch, and led by strong leaders with top management support. The model underlines the role of information and knowledge, as it is crucial for decisionmaking at the different gates (Cooper and Kleinschmidt 1993). Literature Survey 25 Although the model depicts a simple, logical process, in practice the process is complex. In fact, no model is fixed given the existing multiple variables. As Cooper and Edgett (2005) acknowledge, there is a need to be flexible and adjustable to the scale, nature, magnitude and risk level of the different project types, and therefore there is no longer just one version of the Stage-Gate. The authors advocate a new NexGen Stage-Gate®, which integrates and embodies seven principles of Lean, Rapid and Profitable NPD – a broader, more holistic and multi-dimensional view of Lean PD (see Section 2.1.3) - namely:  Customer focused;  Front-end loaded;  Spiral development;  Holistic approach driven by multi-functional teams;  Metrics, accountability and continuous improvement;  Focused and effective portfolio management – a funneling approach and the resources in place;  NextGen Stage-Gate® process. Rozenfeld et al. More recently, and in an attempt to cover what the authors considered to be relevant but lacking in the PDP models thus far in the literature, Rozenfeld et al. (2006) developed a reference model for the PDP. The model is the result of the authors' experience and research results drawn from the literature in the area. The Unified Model of Reference combines Pahl and Beitz’s (2007) PDP model with the concept of CE, ensuring a model which agglomerates sequential steps (which may overlap), simultaneity and various adaptable tools and methods. The model (see Figure 2.7) is divided into three macro-phases: Pre-development; Development and Post-development. Each macro-phase is subdivided into stages (or phases), which, in turn, are subdivided into activities. The stages are determined by deliverables. The approval of these outcomes is formally carried out during stage reviews (gates), which intend to evaluate in detail the results obtained thus far in order to ensure (or not) the project’s continuity. Once approved, the deliverables cannot be changed, unless it is through a controlled process for incremental improvement of the PDP. Management has direct influence on the PDP as it controls information flow. Furthermore, the final Post-development phase clearly distinguishes this model from the ones presented before. The main objective of this phase is to accompany the product through the various stages of its life cycle. Chapter 2 26 Figure 2.7 PDP Unified Model of Reference (translated from Rozenfeld et al. 2006) The Unified Model contains phases and activities, similar to the abovementioned models, however in parallel, the authors provide a model to diagnose the maturity of the company’s PDP in order to identify the level at which the company is located and thus indicate priority intervention areas according to the company’s characteristics, its market and product. The main purpose of the model is to assist companies in the implementation of successful PDP. By considering the abovementioned models, we can state that most PDP models have an identical structure, while maintaining a degree of heterogeneity, given the specific industry or context to which they refer. Furthermore, and in line with Sharafi et al. (2010), most existing models do not contemplate simultaneous development, PD management and information management, although that is changing as was seen in Cooper and Rozenfeld. Summary of Methodological Approaches to Product Development In an attempt to summarize the vast methodological approaches to the PD process, Yazdani and Holmes (1999) proposed four generic models, which illustrates the evolution of the sequential, traditional methodology towards CE. The authors recognize the variety of methods, tools and techniques that companies employ to improve new PD. Furthermore, the authors acknowledge that the pressure companies are subjected to, namely in terms of quality, costs and time, as well as the context determine the different methodological approaches to PD. Literature Survey 27 Although current trends indicate the need to move towards CE and dynamic models, the truth is that some companies still employ more traditional approaches, while many employ none at all (Fitzgerald 1998). The Sequential Model The sequential model (see Figure 2.8) represents the traditional approach to PD. The product is designed and then individual functions contribute in a sequence of activities, with the process being repeated until a satisfactory result is achieved. This approach is functionally oriented with little integration, as opposed to process oriented. Changes normally occur in the manufacturing stage and, depending on the change, may imply the need to repeat part or the entire process. This has significant effects on costs and time. Furthermore, various management layers predominate in this type of organization. Figure 2.8 Sequencial engineering (Yazdani and Holmes 1999) The Design Centered Model Changes at each sequential stage increased costs considerably, thus proving that more life cycle considerations were needed at an earlier stage. The design-centered model shifts the focus of design analysis to the front end of the process (see Figure 2.9). Although the model does not require the participation of other departments (full integration), attention is given to the downstream activities, minimizing design changes. The process though is still mainly sequential, but there is a greater confidence in design detail. Although such approach improved quality and costs, need to reduce development Chapter 2 28 lead times continued. Furthermore, the emergence of more complex products hindered downstream activities. Figure 2.9 The design centered model (Yazdani and Holmes 1999) The Concurrent Definition Model Given the demand for shorter lead times, companies felt the need to incorporate specific expertise into the design stage, thus leading to a greater involvement of downstream activities. This was the beginning of concurrent product definition, characterized by the overlapping of the design and planning of process development (see Figure 2.10). Each stage of PD has a formal gate to confer the product’s evolution and conformity with downstream activities. The gates allow for many iterations and design changes to occur. At the end of each phase, the process is revised and becomes the basis for the subsequent phase. Figure 2.10 The concurrent definition model (Yazdani and Holmes 1999) Literature Survey 29 The emergence of multi-functional teams allows for informal and more intense information exchange. Within each phase, information acquires a dynamic nature. Product definition is thus more concurrent and not only relies on design analysis, but also on the expertise of each project team, who assumes greater decision-making responsibility. The Dynamic Model The dynamic model, a development of the concurrent definition model (see above), surfaced owing to the need for more intense communication amongst the multi-functional team from the beginning of each project. Thus, in this model, information becomes far more informal and intense - a necessity, since all activities begin at the same time. This enables lead time and cost reduction (see Figure 2.11). A fully dedicated project team must possess the necessary technical, business and project management skills in order to be able to make essential decisions at the working level. However, this is only possible in a flat organization model. Figure 2.11 The dynamic model of design definition (Yazdani and Holmes 1999) The dynamic model has been only identified in Japanese automotive companies, and it is also referred to in the literature as “set-based concurrent engineering” (Ward et al. 1995) which is one of the fundamental principles of “lean product and process development” (see for example, Morgan and Liker 2006; Ward 2007). We will continue this discussion in the next section. The Different Prescriptive Methods: Final Thoughts Yazdani and Holmes’s (1999) summary of the various methodological approaches to PD represents rather comprehensively the evolution models have undergone over the years. Chapter 2 30 Despite the differences in the various models presented, perhaps as they derive from different contexts and within different perspectives, all models have evolved towards a more flexible PDP, concurrent with the very nature of PD. Thus, despite the different forces driving each model, new tendencies denote greater degrees of flexibility. Nevertheless, models continue to seem sequential, focusing on the different process phases or stages in order to reach a satisfactory outcome. The process-based focus intends to improve PD effectiveness and efficiency, by drawing on best practices. A possible explanation may be the need for a graphical representation for easier communication and comprehension. We should however insist that the above mentioned evolution denotes that concepts such as iteration, overlapping activities, flexibility and agility are gaining prominence, to address PD uncertainty. 2.1.3 Product Development Practices Page and Schirr’s (2008) review of the literature published on the PD process reveals the field has grown tremendously with innumerable variables and intricate models being studied throughout the world. Substantial efforts have been made to determine which tools, techniques, and methods produce better results in practice. Nonetheless, most models presented, while based on best practices1, are prescriptive describing what should be done based on these proven practices. In order to understand the contemporary practices that differentiate highly successful companies, various studies have been undertaken to search for best practices such as the Product Development and Management Association’s (PDMA) Comparative Performance Assessment Study (Barczak, Griffin, and Kahn 2009; and more recently Kahn et al. 2012) and the American Productivity Quality Center NPD best practices study (Cooper, Edgett, and Kleinschmidt 2002, 2004a, 2004b). These studies provide a holistic perspective of all practices capable of promoting a successful PDP. Kahn et al. (2012) empirical study relied on the opinions of American, British and Irish PD practitioners to see what they perceive as best practice as opposed to the previous studies where best practices were identified and prescribed by researchers. The study characterized PD practice according to seven dimensions, as defined by the authors: 1 Kahn et al. (2012) draw on Camp (1989) and describe best practice as a “technique, method, process, or activity that is more effective at delivering a particular outcome than any other technique, method, process, or activity within that domain.” In the case of NPD practice, the authors claim, best NPD practices are those that promote greater success in developing and launching new products and services. Literature Survey 31  Strategy: “defining and planning of a vision and focus for research and development, technology management, and product development efforts at the strategic business unit, division, product line, and/or individual project levels”;  Process: “the implementation of product development stages and gates for moving products from concept to launch, coupled with those activities and systems that facilitate knowledge management for product development projects and the product development process”;  Research: “describes the application of methodologies and techniques to sense, learn about, and understand customers, competitors, and macro-environmental forces in the marketplace”;  Project climate: “the means and ways that underlie and establish product development intra-company integration at the individual and team levels”;  Company culture: “the company management value system driving those means and ways that underlie and establish product development thinking and product development collaboration with external partners”;  Metrics and performance measurement: “measurement, tracking, and reporting of product development project and product development program performance”;  Commercialization: “describes activities related to the marketing, launch, and post-launch management of new products that stimulate customer adoption and market diffusion”. Their analysis revealed that practitioners are clearly more aware of poor PD practices and have difficulty in identifying and expressing the good practices. Table 2.3 summarizes the characteristics, which practitioners perceived as a best practice across the seven dimensions. The authors conclude that whilst companies acknowledge these practices, they are difficult to prescribe, in particular when intending to encompass the multiple companyspecific PD situations. Chapter 2 32 Table 2.3 New product development best practices (adapted from Kahn et al. 2012) Dimension Best Practice Strategy Clearly defined and organizationally visible NPD goals. The organization views NPD as a long-term strategy. NPD goals are clearly aligned with organization mission and strategic plan. NPD projects and programs are reviewed on a regular basis. Opportunity identification is ongoing and can redirect the strategic plan real time to respond to market forces and new technologies. Process A common NPD process cuts across organizational groups. Go/no-go criteria are clear and predefined for each review gate. The NPD process is flexible and adaptable to meet the needs, size, and risk of individual projects. The NPD process is visible and well documented. The NPD process can be circumvented without management approval. Culture Top management supports the NPD process. Management rewards and recognizes entrepreneurship. Project climate Cross-functional teams underlie the NPD process. NPD activities between functional areas are coordinated through formal and informal communication. Research Ongoing market research is used to anticipate/identify future customer needs and problems. Concept, product, and market testing is consistently undertaken and expected with all NPD projects. Customer/user is an integral part of the NPD process. Results of testing (concept, product, and market) are formally evaluated. Metrics - Commercialization The launch team is cross-functional in nature. A project postmortem meeting is held after the new product is launched. Logistics and marketing work closely together on new product launch. Customer service and support are part of the launch team. A launch process exists. Literature Survey 33 In turn, Smith (2007) adverts that best practices in PD “say that one should plan the development project and then follow the plan”. Convergent with the dynamic nature of today’s reality, where uncertainty and change is the rule and not the exception, the author proposes a different approach. Smith advocates product development flexibility focused on people factors, as opposed to the more rigid process-based practices. The author defines product development flexibility as: “The ability to make changes in the product being developed or in how it is developed, even relatively late in development, without being too disruptive. The later one can make changes, the more flexible the process is. The less disruptive the changes are, the more flexible the process is” (Smith 2007). Although process and structure are necessary, too much process implies a higher degree of rigidity, which is often used as a measure to avoid uncertainty and risk. In contrast, Smith advocates people in flexible development, as, through experience, key decision-makers are able to judge and adjust to each specific situation. Although Smith draws on the "individuals and interaction over process and tools" as stated in the Agile Manifesto2, process is not neglected. Within this perspective, Smith proposes “customizable tools, techniques and approaches that will help accommodate and embrace change in order to promote a flexible PD”, namely:  Modular product architectures;  Front-loaded prototyping and testing techniques;  Set-based design to preserve options;  Frequent feedback from customers;  Close-knit project teams;  Collaborative decision making;  Framing decisions and anticipating the information needed to make them;  Rolling-wave project planning;  Development processes that maintain both quality and flexibility. Concurrent with Smith’s opinion that people factors are the most relevant in PD, the tools proposed by the author intend to foster flexible decision-making amongst the PD stakeholders, rather than a procedure to be replicated. 2 http://agilemanifesto.org/ Chapter 2 40  Knowledge Library: By keeping an organized knowledge system with projects developed, it becomes possible to sustain, apply, share and renew previously gained knowledge in order to enhance company performance and create value. Knowledge transfer between projects considerably reduces time-to-market (see Takeishi 2002; Hines, Francis, and Found 2006; Morgan and Liker 2006).  Set-Based Concurrent Engineering (SBCE): Concurrent engineering (where different activities and tasks are performed simultaneously and in parallel) is considered in terms of lean thinking. In SBCE, a broader set of alternatives (parallel and independent) is taken through the phases of the PDP by the development team which, then gradually eliminate alternatives until a better one is produced from a combination of systems, subsystems and components. In practice, this means delaying decisions (and increasing design team’s uncertainty), communicating and pursuing sets of possible solutions, rather than modifying a point solution, in order to reduce cost and time-to-market (see, for example Ward et al. 1995; Sobek II, Ward, and Liker 1999; Hines, Francis, and Found 2006; Doll, Hong, and Nahm 2010; Khan et al. 2011).  Standardization: Standardization is the basis for continuous improvement as it reduces variability and allows for the creation of predictable outcomes with quality. Standardization may be that of the project – the product, its components, materials and architecture – or that of the processes involved. The latter entails frequent tasks, their sequence and length, as well as the standardization of technical skills, which pertains to specific people’s skills and knowledge within the development team (see Cusumano and Nobeoka 1998; Liker, Collins, and Hull 1999; Liker and Morgan 2006; Morgan and Liker 2006; Emiliani 2008; Marksberry et al. 2010).  Cross-Functional Teams: Teams with members from different functional areas (such as marketing, production, etc.) work in an integrated fashion. The workload becomes leveled, management cadence is shortened, processes are synchronized across functional departments, and rework may be nearly eliminated (see Karlsson and Åhlström 1996; Cooper and Edgett 2005; Liker and Morgan 2006; Kim and Kang 2008).  Chief Engineer System/heavyweight team structure: The Chief Engineer4 is the one person in the PD project that pulls the different parts of the project together and creates a coherent whole. He is the person responsible for integrating the technical, process and customer knowledge in order to improve workflow. The 4 Also denominated as “entrepreneurial system designer” by Ward (2007). Literature Survey 41 Chief Engineer assumes not only the role of project manager, but also of leader and technical integrator (see, for example Kennedy 2003; Liker and Morgan 2006; Radeka 2012). In sum, as Karlsson and Åhlström (1996) discuss, the “collection of interrelated techniques including supplier involvement, cross-functional teams, SBCE, functional integration, use of heavyweight team structure and strategic management of each development” are essential to reach LPD. The authors stress the importance of a “coherent whole” as opposed to the implementation of individual techniques. They argue that implementing one technique does not mean LPD. Design for Six Sigma Six Sigma is a methodology that provides businesses with the tools to improve the capability of their business processes. Six Sigma is different from other quality initiatives, as it focuses on all aspects of business operation, seeking to improve key processes, not only product quality. It is therefore a process-based approach to business improvement, which revolves around a “do the right thing, and do things right all the time” rule (Yang and El-Haik 2003). Linderman et al. (2003) define Six Sigma as “an organized and systematic method for strategic process improvement and new product and service development that relies on statistical methods and the scientific method to make dramatic reductions in customer defined defect rates”. This definition underlines the importance of improvements based on what customers perceive to be a defect. The authors perceive as essential in the improvement effort being able to determine exact customer requirements, and then defining defects based on customer’s “critical to quality” parameters, instead of internal considerations. Therefore, data and objective measurement at each phase is crucial. The authors emphasize that the careful integration of tools with the methods is unique to Six Sigma. Tennant (2002) summarizes the concept of Six Sigma from three different perspectives: as a metric, as a methodology and as a philosophy. The philosophy underpinning Six Sigma is total customer satisfaction. The metric is related to the customer experience of quality and the costs associated with the delivery of poor quality (from the customer’s standpoint). The Six Sigma methodology links tools and techniques to foster successful process improvement, by identifying and eliminating defects. Chapter 2 42 The Six Sigma methodology was developed and pioneered at Motorola in the 1980s in order to reduce “quality costs” (i.e. the costs that derive from not doing things “right the first time”). The main aims were to reduce the number of mistakes/defects in the manufacturing process, thus leading to a reduction in manufacturing costs. Six Sigma is therefore considered the “ultimate measure of quality” (Antony 2002). The methodology was popularized by General Electric in the nineties and, by 2002 many of Fortune 500 companies claimed having implemented Six Sigma (Chowdhury 2003). Indeed, by placing the focus on customer needs and establishing quantifiable measures, Six Sigma projects leads to greater customer satisfaction, organizational performance and profitability (Jones, Parast, and Adams 2010). Zhang, Hill and Gilbreath (2011) consider five distinctive elements/principles of Six Sigma. These are:  Customer orientation: Customer orientation is an important principle of quality management. In Six Sigma, customer requirements are the main focus. Projects must clearly demonstrate their value to customers, and the benefits must be clearly visible from customers’ standpoint. Furthermore, customer orientation is used to decide on and prioritize projects.  Leadership engagement: Quality management effort entails strong top management support. In Six Sigma a mechanism is set forth to ensure that the leadership team is engaged and that Six Sigma is prioritized. Senior executives are directly involved in projects, which guarantees that the right projects are selected and receive organizational support. Full-time improvement project leaders (also known as Black Belts) are selected not only because of their technical knowledge but also their leadership skills.  Dedicated improvement organization: Six Sigma requires organizations to establish a dedicated organizational structure for improvement. This structure includes roles such as Green Belt, Black Belt, Master Black Belt, and Champions. Only the organization’s most dedicated and best employees can occupy Black Belt positions, which lead to effective improvements.  Metric focus: Six Sigma emphasizes the application of metrics and rigorous tracking of the metrics to guarantee benefits derive from improvement projects. All projects must have clearly defined goals, expressed in metrics. Each project is also inspected on its intended and realized benefits, usually in financial terms. Literature Survey 43  Structured method: Six Sigma is highly prescriptive, as each project must strictly follow a structured method, known as DMAIC. The DMAIC method breaks down as follows:Define the project goals, the customer (internal and external) requirements and the process.Measure the process to determine current performance.Analyze and determine the root cause(s) of the defects.Improve or optimize the process by eliminating defect root causes.Control future process performance. This allows for the structured exploration of root causes and structured control of the process in order to produce the desired output. By implementing a standard method, a common language is set across the organization, promoting knowledge creation and dissemination. In order to successfully implement the Six Sigma principles, organizations need more than an understanding of Six Sigma methodology, tools and techniques. Management commitment and involvement is needed. Six Sigma implies changes in the entire organizational infrastructure, based on cultural change and training (Wang 2008). Six Sigma does not address the original design of the product or process, it merely improves them. Antony (2002) and Chowdhury (2003) contend that for organizations to effectively reach the six-sigma quality level, they need to redesign their products, processes and services. Whereas Six Sigma focuses on reforming the production and business process to eliminate mistakes, improve morale and save money, organizations need to develop or redesign the process itself to prevent downstream errors. This can only be achieved by switching from “reactive improvement” to “proactive improvement” (Mader 2006), by implementing Design for Six Sigma (DFSS). DFSS is “the Six Sigma strategy working on early stages of the process life cycle” by utilizing “the most powerful tools and methods known today for developing optimized designs”, where the objective is to “design [our emphasis] it right the first time” (Yang and El-Haik 2003). Whereas Six Sigma is used to “react” to or fix unwanted events in the customer, design, or process domains, DFSS is used to “prevent” problems by building quality into the design process across domains at the thinking level (Smith 2001). Brue and Launsby (2003) define Design for Six Sigma as: “A systematic methodology using tools, training, and measurements to enable the design of products, services, and processes that meet customer expectations at Six Sigma quality levels. DFSS optimizes your design process to achieve Six Sigma performance and integrates Chapter 2 44 characteristics of Six Sigma at the outset of new product development with a disciplined set of tools”. In terms of systematic methodology, Tennant (2002) posits a structured methodology for PD that consists of a rigorous stage gate (“tollgate”) model, with deliverables that must be approved at the end of each stage, before the project continues. In regards to the implementation of tools, Mader (2002) concedes, “the power of DFSS is in the organization of the tools into a coherent strategy that aligns with the NPD process, not the individual tools themselves”. Thus, the careful selection and integration of a group of tools and techniques within a stage gate PDP, encompassing Six Sigma principles is regarded as the essence of DFSS. Kwak and Anbari (2006) draw on De Feo and Bar-El (2002) and summarize seven basic elements of DFSS as follows:  Customer-oriented design process with six sigma capability;  Design quality at the outset;  Top–down requirements flow down with flow up capability;  Cross-functional design involvement;  Quality measurement and predictability improvement in early design phases;  Process capabilities in making final decisions;  Process variances to verify that customer requirements are met. Unlike the DMAIC methodology applied in Six Sigma, the phases or steps of DFSS are not universally recognized or defined. Many companies and training organizations will present a tailored model, in order to suit their specific businesses. As a result, there are two widely recognized methodologies with several variants. In attempting to retain the link to DMAIC, one of the most popular DFSS methodology is DMADV (i.e. Define, Measure, Analyze, Design and Verify); the other is ICOV/IDOV (Identify, Characterize/Design, Optimize, Verify). We will present each briefly, as well as some of the associated tools for each phase. DMADV Breyfogle (1999) first refers to DMADV approach in opposition to DMAIC, advocating its appropriateness when a “product or process is not in existence and one needs to be developed. Or the current product/process exists and has been optimized but still doesn't meet customer/business needs”. The DMADV approach comprises five phases: Literature Survey 45  Define the design goals (consistent with customer expectations);  Measure the organization’s capability to produce the product and associated risks, as well as identify characteristics that are Critical To Quality (CTQ) based on the Voice of the Customer (VOC);  Analyze to develop and design alternatives and then evaluate each to select the best;  Design details, optimize the design, and plan for design verification;  Verify design performance, namely through pilot runs, then implement the production process. For each phase of the DMADV approach there are specific tools, which seek to foster the intended outcome. Given the vast array of tools mentioned in the literature (Kwak, Wetter, and Anbari 2006), we draw on Ginn and Varner’s (2004) Six Sigma roadmap, which concentrates on the DMADV phases of the DFSS process. Table 2.5 depicts the methods and tools proposed by the authors for each phase. Table 2.5 Methods and tools for each phase of DMADV (Ginn and Varner 2004) Phase Methods and tools Define Market analysis tools (Market forecasting tools, Customer value analysis, Technology forecasting and visioning); Process analysis tools (Control charts, Pareto charts); Traditional project planning tools (Work breakdown structures, Program Evaluation and Review Technique (PERT) charts, Gantt charts, Activity network diagrams); DMADV specific tools (Project charter, Inscope/out-scope tool, Organizational change plan, Risk management plans, Tollgate review forms). Measure Customer segmentation tree; Data collection plan; Customer research tools (Interviews, Contextual inquiry, Focus groups, Surveys); VOC table; Affinity diagrams; Kano model; Performance benchmarking; QFD matrix; CTQ risk matrix; Multistage plan; Tollgate review forms. Analyze QDF matrix; Creativity tools (Brainstorming, Analogies, Assumption busting, Morphological box); Pugh matrix; Tollgate review forms. Design QDF matrix; Simulation; Prototyping; Design scorecard; FMEA; Planning tools; Process management chart; Tollgate review forms. Verify Planning tools; Process analysis tools (Control charts, Pareto charts); Standardization tools (Flowcharts, Checklists); Process management charts. Chapter 2 46 IDOV For Yang and El-Haik (2003), DFSS comprises four phases, known as IDOV or ICOV:  Identify requirements: Draft project charter; Identify customer and business requirements;  Characterize the Design: Translate customer requirements to product/process functional requirements; Generate and evaluate design alternatives;  Optimize the design: Optimized design entity with all functional requirements released at the Six Sigma performance level;  Verify the design: Pilot testing and refining; Validation and process control; Full commercial rollout and handover to new process owner. As with the DMADV methodology, several authors have enumerated the various tools for each of the IDOV approach. Table 2.6 summarizes our findings based on Creveling, Slutsky and Antis (2003), Yang and El-Haik (2003), Woodford (2010) despite the slight variances found in the definition of each phase. Table 2.6 Methods and tools for each phase of IDOV Phase Methods and tools Identify Developing team charter; Gathering the VOC (Market/customer research and Affinity diagrams); Develop CTQs; QFD; Kano model; Risk Analysis; Benchmarking; Target costing; FMEA; SIPOC (Supplier, Input, Product, Output, Customer map); IPDS (Integrated Product Delivery System). Characterize the Design Concept generation and Design for X methods; Pugh concept evaluation and selection process; QFD; TRIZ; Axiomatic design; Robust design; DFMEA and PFMEA; Design review; CAD/CAE; Simulation; Process management; Materials selection; DOE (Design of Experiments); Systems engineering. Optimize Design/simulation tools; Transfer functions; DOE; Taguchi method for robust design, parameter design, tolerance design; Reliability-based design; Robustness assessment; Response surface methods; Optimization methods. Verify Process capability modeling; DOE; FMEA; Reliability testing; Poka-yoke, errorproofing; Confidence analysis; Statistical process control (Process control plan); Training; Disciplined new product introduction. Literature Survey 47 The tools and best practices associated with DFSS and the specific approaches mentioned above are what actually deliver results. Given that practices vary, and that Kwak, Wetter and Anbari (2006) have uncovered over 400 tools and techniques for Six Sigma, where only two thirds of which are in fact unique tools and not duplications with different names, it is feasible for us to state that Six Sigma in general, and DFSS in particular is rather complex in nature and perhaps seemingly rather disjointed given that its practical roots lay in the industry and the fact that most literature is based on best practice studies (Linderman et al. 2003). Product Development Practices: Final Thoughts PD practices vary according to different perspectives. Although the modern PD process is anything but linear and structured, in its majority PD best practices are still put forth as a set of procedures that must be followed and specific tools to use. Based on concrete practices identified in various types of organizations and different types of projects, most literature proposes a prescriptive, more structured approach to implementing what is perceived to be best practices to provide a successful PDP, where specific recipes can be applied. The idea being, if it has worked before, it can be reapplied. Perhaps, this is so as the focus is still very much on the process itself. Others, such as Smith (2007), propose tools for a flexible approach, considering the unique nature of the PD process and the fact that it is people centered. This idea is concurrent with the “one size does not fit all” defended by Krubasik (1988) and Otto and Wood (2001). Indeed, without questioning the prescriptive literature, we propose these best practices should always be adapted to each particular situation. By customizing procedures and tools and avoiding a direct and mechanistic implementation, we can foster a flexible PDP that supports uncertainty and adapts to changes which is a reality in today’s world. 2.1.4 Product Development Frameworks As mentioned previously, research on product development is diverse and fragmented, making it extremely difficult to understand, systematize and organize. Various authors have attempted to address this problem, among them Brown and Eisenhardt (1995) and Krishnan and Ulrich (2001), each within a different perspective. This section depicts the authors’ key research findings, as their synthesis is an important development and useful contribution to an in-depth understanding of PD. Gericke and Blessing (2011) state: Chapter 2 48 “A key weakness of design process models is their difficult application to real design problems. Most process models are too general to help project planning and to guide daily decisions. While most models of designing integrate an iterative component or the evaluation of the results they do not describe how to achieve fitness for purpose of the product”. Nevertheless, this important field of study already provided a set of principles that we can consider as the PD theoretical basis for our research. Several reviews organize research into frameworks. Brown and Eisenhardt (1995), for example, organize PD research depending on its methodological approach. They aggregate previous empirical findings into a framework of factors, which affect the success of PD projects (see Figure 2.12). Figure 2.12 The Brown and Eisenhardt (1995) framework An important observation of Brown and Eisenhardt is that: “… there are two relevant problem-solving models for organizing PD. One focuses on factors such as planning and overlap that are relevant for more stable products in mature settings, and the other focuses on experiential product design that is relevant for less predictable products in uncertain settings”. Literature Survey 49 Finally, the key construct behind the framework is that product financial performance is influenced by the actions of multiple players (that perform roles in the PD process). The authors argue that the project team, leader, senior management and suppliers affect process performance; the project leader, customers and senior management affect product effectiveness; and the combination of an efficient process, effective product and attractive market determines the financial success of the product. The PDP success factors synthesized by Brown and Eisenhardt are consistent with the principles and related practices defended by LPD, as explained in Section 2.1.3. This perspective clearly suggests that any PD process model should emphasize the roles involved in the process in parallel to the tasks and decisions. The model should enable a view of who is responsible for what and how they interact. As the next section will illustrate, most of the PD modeling approaches neglect this facet of the PD process concentrating almost exclusively on tasks. In real projects the “what to do” is translated in “how to do” by the roles involved in the PD process. We intend to understand this translation. Krishnan and Ulrich (2001) use a different approach. By combining different perspectives and disciplines, they identify two broad categories within the literature: one focusing on decisions that are made prior to the development project (which consider both strategic and organizational decisions); and a second category where the decisions are made within a development project (which include all the major phases in the development process). The authors classify research in clusters, which establish three essential enablers in PD decisions: product features (market and design), architecture-related issues (also encompassing organizational issues), and portfolio-selection decisions that attend to the strategic development aspects. Figure 2.13 illustrates their findings. Once more the referred model includes issues related with the tasks required to develop the product and, to a lesser extent, the roles involved in the development process (the “who”). The authors state: “Product planning decisions and development metrics seem particularly ad hoc in industrial practice. (…) Insights on customizing product development practices to diverse environments such as small entrepreneurial firms and varied industries should also help increase the relevance and applicability of the development literature” (Krishnan and Ulrich 2001). Chapter 2 56 Browning 2010), Browning and colleagues summarize the fundamental propositions that form the basis of PD process modeling theory (Browning, Fricke, and Negele 2006):  “The process of invention/innovation cannot be fully mechanized.  Nevertheless, the PD process has some repeatable structure.  Project management is facilitated by a structured approach.  Processes can be regarded and treated as systems that should be engineered purposefully and intelligently.  In many aspects, complex process behaviors can be better understood by examining their relatively simpler, constituent parts (actions) and those parts’ endogenous and exogenous relationships (interactions). We refer to this as the decomposition paradigm in process modeling.  There is always a gap between the real system and a model of it.  Process models are built for a purpose”. In an attempt to overcome the aforementioned difficulties, which derive from the uniqueness of PD, various modeling techniques specifically for PDP have been proposed, while other implemented techniques derive from BPM. Smith and Morrow (1999) reviewed some of the PD process modeling methodologies and evaluated them. The authors acknowledged the importance rigorous, tractable, and applicable modeling techniques, admitting that despite some existing rigor, if the techniques are not easy to use, their application will be non-existent. Finally, the authors concede PDP modeling is in an earlier stage of development when compared to, for example, management. 2.2.3 Analytical Approaches and Focuses Each modeling approach provides a different perspective of the process structure. Thus, there are modeling approaches that are best suited depending on the intention and focus. The variety inherent to the different needs and focuses is perhaps the reason for the plethora of PD modeling approaches in the literature. Recker (2006) stated a PhD student’s count of the process modeling techniques in use stopped at 3,000. The question then remains, how to best approach PDP modeling. Given that purpose drives the question of a suitable modeling technique (Recker 2006), Browning and Ramasesh (2007) provide a taxonomy of purposes for process modeling specifically in PD, reproduced in Table 2.7. Literature Survey 57 Table 2.7 Four categories of purposes for PD process modeling (Browning and Ramasesh 2007) Category Purpose 1. PD project visualization a. Actions, interactions, and commitments b. Customized “views” 2. PD project planning a. Making commitments b. Choosing activities c. Structuring the process d. Estimating, optimizing, and improving key variables (time, cost, etc.) e. Allocating resources 3. PD project execution and control a. Monitoring commitments b. Assessing progress c. Re-directing d. Re-planning 4. PD project development a. Continuous improvement b. Organizational learning and knowledge management c. Training d. Compliance Coming from a different approach and in order to propose a modeling framework for PDP considering its characteristics, Jun and Suh (2008) compare the most prominent modeling techniques referred in the literature and systematize them into two broad groups, graphbased techniques and matrix-based techniques. Graph-based modeling approaches include Petri nets and its variants, IDEF and its variants, GERT, among others. Within the matrixbased modeling approach, the leading technique is the task-based Design Structure Matrix, also known as DSM. The IPO paradigm Underlying most modeling approaches, both in business, management and in development processes, is the simple input-process-output (IPO) advocated by Hammer and Champy (1993) (see Section 2.1.1). The logic is that in modeling, engineers, for example, describe what they are doing (P) by describing relevant tasks and activities; what they need to accomplish this (I), such as documents and data files; and what they produce (O). The output of a process may be then used as input by other processes. This output-input relation embodies process interaction. Within this paradigm, the structure of the development process is in conflict with the structure of the information flow (Gartz 1997). Chapter 2 58 Most modeling techniques have in their genesis this IPO paradigm (see, for example, Fricke et al. 2000; Browning 2002; Browning and Ramasesh 2007, among others). We will now provide an overview of the most prominent modeling techniques in PD. Drawing on Jun and Suh’s (2008) work, we group the techniques into graph-based modeling approaches, where we will focus on IDEF and Petri nets; and within the matrix-based modeling approaches, we will focus on DSM and its variants. Graph-based modeling approaches IDEF The IDEF (stands for Integrated Definition) methodology comprises a group of modeling methods. It was originally created by the United States Air Force and is currently being developed by Knowledge Based Systems6. It was originally developed for the manufacturing environment, but since then, the more than fourteen methods (from IDEF0 to IDEF14) have been adapted for wider use. These methods are descendants of data flow diagrams with a functional, structured approach (Melão and Pidd 2001). Melão and Pidd claim IDEF0’s modeling constructs: inputs, activities, outputs, mechanisms and controls (see Figure 2.15) reveal its mechanistic perspective. Figure 2.15 IDEF0 box and arrow graphics (Source: http://www.idef.com) A closer look at IDEF3 (Integrated Definition for Process Description Capture Method) reveals that it complements IDEF0, as the former notation, which describes process flows, is capable of modeling the relationships between actions as well as the specific states a 6 see Knowledge Based Systems, Inc. homepage at: http://www.idef.com. Literature Survey 59 process undergoes (Kreimeyer 2009), thus capturing behavioral aspects (Knowledge Based Systems 2010). Although IDEF enables the view of organizational knowledge from multiple perspectives, the focus on roles/agents is weak and the prominence is on the sequential representation of activities (Turetken 2007; Knowledge Based Systems 2010). Recent studies using IDEF in PD advocate this technique alone is not adequate for complex processes (Chin et al. 2006) and should be conjugated with other techniques, such as colored Petri nets (see Petri nets, below). The authors state IDEF can give a clear description about input and output information, resources and constraints pertaining to the development process. They defend the approach can help streamline the overall process but only to a certain degree as IDEF0 is “static and qualitative and lacks mathematical rigor”. Other authors, such as, Kusiak and Belhe (1992), Zhang, Chen and Xiong (1999), and Fairlie-Clarke and Muller (2003) have also advocated the use of IDEF or IDEF-based methods (input-output) to model the PD process, as a rigorous approach capable of identifying a complete set of activities for any PD process. Petri Nets Carl Adam Petri invented Petri nets in 1962 as part of his Doctoral Thesis. Since then, Petri nets have been used to model and analyze processes, including protocols, hardware, and embedded systems to flexible manufacturing systems, user interaction, and business processes (van der Aalst 1998). This approach combines “visual representation using standard notation with an underlying mathematical representation” (Vergidis, Tiwari, and Majeed 2008), which makes this modeling approach unique (Peterson 1981). Petri nets use graphical and mathematical notations to represent a process. Activity flow is represented with place nodes and transition nodes, whereas arcs connect places with transitions (see Figure 2.16). Concepts such as service, goal and role are not explicitly supported. Aldin and Cesare (2009) claim Petri Nets have limited explicit expressivity given the reduced number of modeling elements. However, the authors believe it is an appropriate method for modeling systems with concurrency. Chapter 2 60 A numbers of extensions have been introduced to the classical Petri nets in terms of color (for literature on colored Petri nets see Jensen 1996), time (see for example Berthomieu and Diaz 1991) and hierarchy to solve some of the detected problems (van der Aalst 1998). Nonetheless, Aldin and de Cesare (2009) considered this modeling approach to be a non user-oriented technique, which is difficult to use without experience. Curiously, on this point, van der Aalst (1999) states Petri nets are “intuitive and easy to learn”. Figure 2.16 Petri nets graphical notation The use of Petri nets method has had some discussion in PD literature. The review of the literature by Chin et al. (2006), among others, allow us a glimpse of the most recent work in the area. Liu et al. (2002) applied workflow colored Petri nets to the PD workflow. Given the time factor in the PDP, Belhe and Kusiak (1993), Lin and Qu (2004), and Chin et al. (2006) mixed the Petri net extensions, introducing timed colored Petri nets in PDP modeling in order to analyze workflow time performance. Furthermore, authors such as Zu and Huang (2004) and Ha and Suh (2008) have also proposed the use of hierarchical timed colored Petri nets in PDP modeling and performance analysis of collaborative design. Literature Survey 61 Matrix-based modeling approaches DSM One of the most important matrix-based modeling approaches in PDP is that based on the Design Structure Matrix (DSM). This modeling approach was originally developed in order to analyze parametric descriptions of designs by Steward in 1981. Since then, a whole community has developed around this (see, among others Kusiak and Wang 1993; Eppinger et al. 1994; Smith and Eppinger 1997; Browning 2001; Cho and Eppinger 2001; Yassine and Braha 2003; Batallas and Yassine 2006; Huang and Gu 2006a). Recently Lindemann, Maurer, and Braun (2009) presented a compendium of a multitude of matrix-based approaches for managing complex data in product design. Underlying the basis of the DSM is the assumption that a design task is a predicable information-processing task with identifiable inputs and outputs and, as such, can be modeled using and creating information (Smith and Morrow 1999). As Smith and Morrow explain, the output information from one task then becomes the input information for another task. The relationships amongst the inputs and outputs may comprise cycles, which indicate the need for iteration. The DSM represents these relationships and dependencies (see Figure 2.17). DSM thus enables the modeling and analysis of complex processes, where entities and relations are known. However, drawing on Lewis and Cangshan (1997), Smith and Morrow concede DSM is relatively difficult to understand, especially in the beginning. Furthermore, the modeling approach, as originally proposed, is lacking in terms of sequencing and scheduling design tasks within iterative groups. Given the identified shortcomings, newer variations have since been proposed. Figure 2.17 Design Structure Matrix (Smith and Morrow 1999) B C A K L J F I E D H G B C X A X KXX L X X X X J X X X X X F X X I X X X E X X X D X X X H X X X X G X X Series Parallel Coupled Chapter 2 62 So as to propose a modeling framework for PD, Jun and Suh (2008) summarized the evolution of DSM. Since its development, Smith and Eppinger (1998) for example, proposed a variation of DSM, which draws on the strength of the dependencies and adds the execution time of each activity with diagonal elements. Kusiak et al. (1995) presented a matrix capable of representing the precedence relation between activities. In turn, Cho and Eppinger (2001) proposed a PDP modeling and analysis technique with the DSM and advanced simulation, considering the iterations that occur among sequential, parallel, and overlapped tasks. Huang and Gu (2006a), by drawing on PD as a dynamic process, developed a DSM in order to capture the interaction and feedback of design information. More recently, Danilovic and Browning (2007) proposed the evolution of DSM to domain mapping matrices, so as to allow the inclusion of more than one domain. Maurer (2007) took this approach further in order to model processes comprising multiple domains, each with multiple elements, connected by different types of relationships and dependencies, using a Multiple Domain Matrix (MDM). The underlying basis in all abovementioned extensions is the DSM, with just more details and different foci. 2.2.4 Limitations of the Modeling Techniques From our analysis above, we can state that, except for the DSM, business process modeling techniques focus the business processes in general, which significantly differs from the PDP. Consequently, despite the attempts to incorporate the various BPM modeling techniques in PDP, these fail to capture the overall essence of PD. Graph-base models are focused on the static control flow of the process, and thus fail to capture feedback loops and dynamic information exchange. They can, however, serve as basis for analyzing structural characteristics. Matrix-based approach DSM, on the other hand, has been developed specifically within the realm of PD. Although they are able to provide “simple and powerful manipulations of the PD process management” (Jun and Suh 2008), they too fail to capture various PD specific characteristics. Furthermore, they are difficult to comprehend, restraining understandability and usability, especially at the beginning (Browning 2001; Jarratt, Eckert, and Clarkson 2004). Although DSM is suitable to identify the entities and relations from any process, in practice, however, these types of models are often not as definite as they seem (Kreimeyer 2009), as modeling a PDP is, in essence a consolidation process of the different perspectives on an actual “as is” project. Additionally, most possess an activity-centric Literature Survey 63 view, emphasizing vertical, hierarchical relationships, while neglecting the various horizontal relationships, drivers of the deliverable flows that induce to the emergent behaviors of the overall process (Browning and Ramasesh 2007). DSM modeling approaches are not able to represent interactions across functional boundaries. Difficulties are encountered when attempting to represent relationships with three or more elements, or when dynamic relationships emerge (Crawley and Colson 2007). Finally, it is difficult to make a DSM at the conceptual design stage or for a new product that has never been designed before (Tang, Zhang, and Dai 2009). Indeed, the previous modeling approaches seem to fail to capture the overall uniqueness of the PDP. As Jun and Suh (2008) claim, modeling techniques have limitations in capturing PDP’s iterative, evolutionary, cooperative and uncertain characteristics. Given the limitations of the different modeling methods expressed in the literature, there is a clear demand for a modeling method that can represent typical PD characteristics holistically. This instigated our curiosity and our search that led us to the Riva Method, described below. 2.2.5 A Different Approach to Process Modeling: the Riva Method Given the specificities of PDP, and the shift from sequential, activity-focus process to people, dynamism and interactivity preached by the most recent methodological approaches to PD such as CE and LPD, perhaps a radical change is also needed in business process modeling. A shift in paradigm: from IPO models to a Role-modeling Approach The IPO process modeling paradigm, as mentioned above, focuses on activities and the flow of information and control across these. As we’ve seen, one of the problems with traditional IPO-based process models is they do not seem to capture interaction between human actors very well. Indeed, although these types of models seem to be compatible with and useful for procedure-based, production-oriented processes, they are not particularly adequate for less structured business processes such is the case of product development (Center and Henry 1993; Carlsen 1997). Furthermore, in processes primarily viewed as the transformation of input into output, stakeholders and their communications tend to be weakly represented (Ilgen et al. 2005). Therefore, to better represent business processes, especially development processes, it is crucial to include a richer description of the group dynamics involved, as well as the possible or intended human communication and interaction (Carlsen 1997). Chapter 2 64 Although IPO paradigm predominates in BPM, there are other modeling paradigms. Carlsen (1996) identified five main classes of process modeling languages, namely Input-ProcessOutput models (see above for a description); Conversation-based approaches; Rolemodeling approaches; Systems thinking and system dynamics; Constraints–based representations. We will focus on role-modeling paradigm, more specifically RAD - Role Activity Diagrams (Ould 1995, 2005) since we intend to surpass the mechanistic thinking of activity-oriented approaches and refocus on the responsibilities involved in PDP. The Riva Method Central to understanding the Riva Method is the notion that a business process is about people (see Section 2.1.1). It is about how people do business, how they think they do business, how they are supposed to do it, how they might improve it, and so on (Ould 2005). The author states a process:  Contains purposeful activity;  Is carried out collaboratively by a group;  Often crosses functional boundaries, and;  Is invariably driven by outside agents or customers. Ould (2005) argues: “It’s what people do, not what they do it to, that counts. A process is mainly about doing, deciding and cooperating, not data or things”. In Ould’s view, the social context perspective of an organization prevails. In order to incorporate these notions, and in a clear attempt to distance himself from the IPO paradigm, Ould (1995) developed the Riva Method, an extension of the formerly known STRIM (Systematic Technique for Role and Interaction Modeling), a modeling approach that explicitly integrates the process, organization and goal elements. Riva is based on the following basic and central concepts:  “A process is a coherent set of actions carried out by a collaborating set of roles to achieve a goal.  A role is a responsibility within a process.  An actor carries out a role.  A role carries out actions following business rules.  A role has props which it uses to carry out its responsibility. Literature Survey 65  Roles have interactions in order to collaborate.  A process has goals and outcomes” [emphasis in original] (Ould 2005). Riva uses two languages to represent processes:  Process Architecture Diagrams (PAD) are used to describe the arrangement of the organizational activity into individual processes; PADs may represent several or all of the business processes within an organization, and how these are interrelated.  Role Activity Diagrams (RAD) are used to describe an individual process. “A RAD shows the roles that play a part in the process, and their component actions and interactions, together with external events and the logic that determines which actions are carried out when. So, it shows the activity of roles in the process and how they collaborate” (Ould 2005). Given our interest in modeling the PDP, we will address RADs specifically as these allow business process to be diagrammatically modeled through the representation of roles, goals, activities, interactions and business rules (Melão and Pidd 2001). Furthermore, this technique is considered by some to be the most complete as it is able to represent the most relevant features of a process, namely goals, roles and decisions (Miers 1994). As mentioned above, the main concept in Riva is the role. A role is an area of responsibility. The concept may refer to concrete things such as organizational posts, functional units and job titles, as well as abstractions, such as Customers, for example. Everything that happens (whether activities or decisions) in a process happens in a role. An activity carried out by a role on its own is an action. Any collaborative act involving two or more roles is an interaction. Separate threads of activity within the same role represent concurrent activity. RADs illustrate the roles and their threads of activities, decisions and interactions; as well as triggers and goals or outcomes – also identified as intended states (Ould 1997). Figure 2.18 summarizes the notation for RADs. The gray, labeled box indicates the activities carried out by a role. On the top of the box, in one or more of the roles, the arrow describes the event that triggers the process. The RAD can then be read by following the vertical lines that derive from that event. A vertical line is a state the role can be in. A circle in the vertical line describes a state. A circle at the end of the thread indicates a process goal. The darker-shaded box corresponds to an activity carried out by a role. A question depicts decisions, followed by alternative paths, Chapter 2 72  4th step: Building a process-based cost modeling;  5th step: Life cycle analysis models;  6th step: Integrated comparison of economic and environmental life cycle performance. This method, besides aiding in selecting the most appropriate material based on requirements, considers total life cycle environmental impacts as well as costs, which are then mapped on a matrix. The visualization facilitates the design team’s final decision by providing them with an integrated comparison of fitness-to-purpose materials throughout the product life cycle. Most state-of-the-art literature on material and technology selection draws, directly or indirectly, on Ashby and his various publications although the shift in focus is notorious. Thus, Ashby’s contribution to our thesis is also pivotal. Dieter’s overview of the materials selection process (1997), for example, draws on Ashby’s work and that of Dixon and Poli (1995). Dieter summarizes Dixon and Poli’s material selection strategy as a four-level approach, explained as follows:  Level I: Decisions on whether the product (or component) should be made from metal, plastic, ceramic or composite based on critical properties.  Level II: Decisions on whether metal parts will be produced by deformation or casting; and if plastics will be thermoplastic or thermosetting polymers.  Level III: Narrow options to a category of materials: metals and plastics can be subdivided into specific categories, such as stainless steel (for metals) and polycarbonates (for plastics).  Level IV: Selection of a specific material according to grade or specification. Dixon and Poli’s four-level approach is funnel like, in the sense that the selection process is progressively narrowed down to a specific solution. They suggest the use of a guided iterative strategy to evaluate materials and technologies throughout the design process, in the form of a formalized handbook, which draws on charts and tables. Dixon and Poli (1995), similarly to Ashby (2005), defend a higher number of possible candidates increases design and manufacturing flexibility; an idea referred to as least commitment (a concept also present in SBCE, see Section 2.1.3). Towards a comprehensive approach of the implications and interconnections in material selection, Ashby (2005) states thinking about materials information is required at each Literature Survey 73 stage of the design process although the nature of that information differs greatly in terms of precision and depth. Drawing on Pahl and Beitz’s PDP model (see Section 2.1.2), the author clarifies:  In the first stage (Concept design) all options are open, thus there is a need for breadth and ease of access. At this time the designer should consider all the possibilities without restrains, both technical and aesthetic. Although the designer’s choice of concept will have repercussions on the overall configuration of the design, most decisions regarding material and form are left unanswered.  In the second stage (Embodiment design) a higher level of precision and detail is needed. It is at this stage that each potential concept is developed. Operations are analyzed and alternative choices in material and process are explored.  In the third stage (Detail design) a higher level of precision and detail is needed but for one or very few materials. The final step is a prototype to ensure design meets technical and aesthetic expectations. On this, Verganti (1998) posits early prototyping encourages uncertain information to be gathered in the early phases, without entering into detail design. This process for material selection does not end at this point, as failures in production or service should be analyzed and relevant information needs to be gathered and documented for future situations. Similarly to Dixon and Poli, Ashby proposes a refinement throughout the PD process, without implying that a decision has to be made at a specific stage. Figure 2.20 depicts Ashby’s thinking process about materials information. Within this thinking framework, Ashby et al. (2004) posit three main selection strategies. For the purpose of our study, we will explain these three strategies and then will focus on the free searching based on quantitative analysis method (referred to as selection by analysis in Ashby and Johnson 2002), as it involves a systematic procedure to material selection. Ashby bases his well-known book Materials Selection in Mechanical Design (2005) on this selection method. Free searching, based on quantitative analysis: This selection method is based on a quantitative analysis strategy that objectively and efficiently provides the requested information, if the inputs were precise, detailed and in a form that can be analyzed by standard engineering methods. In sum, this method requires the selection of a material from a database based on the expressed function, constraints and objectives. Each Chapter 2 74 combination of these three aspects leads to performance metric(s) containing a group of material properties or material index. These can then be compared. Figure 2.20 Materials in the design process (Ashby and Johnson 2002) Questionnaire strategy, based on expertise capture: Through the use of questionnaires, the user is able to be guided through a set of decisions previously constructed by documenting experts’ answers to a comprehensive set of specific questions and the subsequent questions until an unqualified answer is reached. This poses several problems, namely in terms of construction of the questionnaires as this implies time and patience. Furthermore, this selection strategy does not innovate, as it is not able to consider something the expert does not know, which might be the case of a new material or process. Inductive reasoning and analogy (involving past cases): This strategy has its roots on past cases, previous experiences, which need to be analyzed and documented. Requirements are expressed as the features of a problem. How this problem was solved is exploited, and knowledge is compared and tested for its ability to suit the requirements. In practice this means databases are searched for cases with similar features as the ones at hand. The analysis of the information retrieved is then adapted or combined in the new product. The main problem identified in this strategy is on how to structure and index the unstructured knowledge obtained from the various cases. Literature Survey 75 Ashby enumerates various problems with the last two strategies and seems to be more in favor of the Free searching, based on quantitative analysis method, stating that it is currently well developed and has been successfully applied. The author classifies it as fast, efficient, systematic and with potential for innovation. Aware of the advantages of this strategy, Ashby developed it further, and within this Analysis method, Ashby (2005) proposes a systematic procedure for materials selection which incorporates four main steps: translation, screening, ranking and supporting information (see Figure 2.21). Underlying this procedure is the basic principle that all materials should be considered as potential candidates for any application. Figure 2.21 Four-step materials selection procedure: translation, screening, ranking, and supporting information (Ashby 2005) The first step consists in translating, or examining the requirements in order to identify the constraints that these entail on the choice of material. This first step is a statement of product (or component) functions, constraints, objective and free variables allowed. After, and considering that all materials are candidates until proven otherwise, the screening process begins. This is where material choices are eliminated if the attributes cannot meet the constraints. Having limited the material choices, it is then necessary to rank the candidates by their ability to maximize performance, i.e. according to how well they can fulfill the objective. Finally, having arrived at a ranked short-list, it is then necessary to Chapter 2 76 search for a detailed profile of each candidate. This information may be descriptive, graphical or even pictorial. The final choice may occur at the detail design stage. Although it may seem like a very straightforward procedure, there are issues threatening the task, namely:  difficulty in tracking all the candidate materials,  difficulty in translating the requirements into material properties,  the tendency to benchmark similar products or applications and “copy” the material in order to reduce risk, and  the tendency to select the materials in early design stages to facilitate the remaining tasks. This four-step procedure for material selection proposed by Ashby is carried out throughout the PD process, beginning at the concept stage and concluded at the detail design stage. However, as also highlighted by the author and discussed in Section 2.1.1, PDP is complex and iterative, thus the various stages are interconnected and are revisited throughout the process. Identically, the material selection procedure may be triggered in any of the stages, depending on the material and the product component or subsystem in question. Thus, this procedure may be looked upon as a cyclical selection process to be applied and repeated until an optimal solution is achieved. Depending on the specific situation and the depth of the design team’s knowledge, the procedure may begin in the Concept stage, or Embodiment, and the cycle itself may stretch or be curtailed depending on the nature of the inputs. Ashby and Johnson (2002) claim there are various creative solutions for material selection in product design and, in addition to the essential requirements (an information structure and a selection method, explained above), propose complementary methods, as follows:  Selection by analysis (deductive reasoning): uses precise specified inputs and wellestablished modern design methods, through materials databases (this method was referred to as the Free searching, based on quantitative analysis in Ashby et al. 2004).  Selection by synthesis (inductive reasoning): is based on documented past experiences, when searching for equivalent features, intentions, perceptions or aesthetics. Literature Survey 77  Selection by similarity: draws on the attributes of existing materials to search for equivalent materials.  Selection by inspiration: draws on other products and materials for inspiration to suggest solutions. Although traditional analytical methods prevail, there are complementary methods in the sense that more than one method may be deployed in order to obtain an optimal solution. For example, we posit it is possible to employ selection by synthesis, selection by similarity and selection by inspiration within the screening step of the selection by analysis method. We cannot overlook the fact that these can be valuable sources of information and solutions and thus all help, not hinder, the material selection process, as combined they become an extremely powerful tool, more so than any one used individually (Ashby and Johnson 2002). Finally, we stress all material selection methods should be based on the premise that all materials are possible candidates, for as Ashby (2005) acknowledges, failure to start with a “full menu of material in mind (…) may mean a missed opportunity”. This means that we cannot simply rely on the extensive decision support tools available, as these may not be completely exhaustive given the constant changes, evolutions and introduction of new materials in the marketplace. Although the material selection process begins with product requirements, the final decisions will inevitably involve cost considerations, which, in many cases, will be the dominant criterion. In this regard, the most favorable solution may sometimes have to be discarded for the best available. Charles, Crane and Furness (1997) cite (Pick 1968) who has said: “Material (and process) selection always involves the act of compromise - the selection of a combination of properties to meet the conflicting technical, commercial and economic considerations”. This perhaps justifies the “but we've always done it this way” argument for material selection. 2.3.3 Technology Selection In regards to technology selection specifically, the term has different interpretations in different contexts. For example, the European Institute of Technology Management (EITM)9 considers technology selection as a Technology Management sub-process: 9 http://www-mmd.eng.cam.ac.uk/ctm/eitm/index.html Chapter 2 78 "Technology management addresses the effective identification, selection, acquisition, development, exploitation and protection of technologies (product, process and infrastructural) needed to maintain a market position and business performance in accordance with the company’s objectives". Thus, technology management attends to the various processes needed to deliver products and services to the market, which pertain to all aspects of integrating technological issues into decision-making. In the view of our thesis, our focus is on material selection and the production technologies, or processes, associated. Thus, this section of our literature survey will not approach the vast technology management realm; rather we will discuss the adoption of technology depending on and parallel to the material chosen, as these two decisions are “intrinsically interwoven” (Schey 1997). Riddle and Williams (1987) define technology selection as “the process of determining which (new or old) methods, techniques, and tools satisfy criteria reflecting a particular target community's requirements”. This selection process requires:  the ability to identify a set of options to be considered;  the ability to evaluate the options, either comparatively or in isolation; and  the ability to choose from amongst the various options, based on the evaluations. Although material selection will condition technology adoption and vice versa, it is a challenging process as technology alternatives are increasing and becoming evermore complex. The choice of the “right” technology is able to create significant competitive advantages as it fosters the development of competitive products, effective processes, or even completely new solutions (Torkkeli and Tuominen 2002). However, technology is often considered mainly in terms of the costs of the capital and labor involved (Bruun and Mefford 1996) for an estimated annual production volume (Ashby 2005; Eggert 2005). Process choice is thus greatly influenced by the total number of parts to be produced and by the rate of production (Schey 1997). Decisions are, consequently, usually made ad hoc, heavily based on company-established processes. In practice, this means technology selection is made in accordance with the existing manufacturing system, neglecting product requirements. Studies advocate (Kaplinsky and Posthuma 1994; Krishnan and Bhattacharya 2002) most organizations, in practice, are resistant to change as they are risk averse and therefore are Literature Survey 79 more reluctant to consider unproven technology. Indeed, the selection of the most appropriate technologies is amongst the most challenging decisions. Langley and Truax (1994) contend it is difficult to delimit technological adoption decisions, as these implicate three interacting processes: a strategic commitment process, a technology choice process and a financial justification process. Furthermore, the authors defend technology adoption decisions are often contextual and issue-oriented. Langley and Truax draw on the literature and propose three classes of process models to understand technology adoption:  Sequential models draw on the work of Mintzberg and Raisinghani (1976) and are based on the notion of technology adoption as a sequential decision process, comprised of a number of phases, where a group of activities is carried out. This approach has been heavily criticized for its simplicity and rigidness as it disregards other, more ambiguous interferences, such as social aspects.  Political models (Dean 1987) focus on how top managers are convinced to adopt new technology, through persuasion, salesmanship and negotiation. In these models personal credibility and political support are more important than financial or strategic positions.  Finally, serendipitous models draw on Mohr’s (1987) view of the non-deterministic nature of innovation processes. Mohr identifies routines, which may contribute to new technology adoption, among others, we underline: after recruitment if new employees have the knowledge; by imitation of what others are doing; for market survival; after a search; to maintain company status. Langley and Truax recognize the need to model the technology adoption process, as there is a lack of empirical research specifically in smaller firms. Various authors, amongst them Phaal and Muller (2009), defend that technology selection should start by considering the various functions that the product performs (or might need to perform) and the qualities (e.g. performance and reliability) that have to be achieved. Company-established processes influence the early stages of product development (product concept and product specifications), limiting options by creating an assumption that a determined technology is the only feasible solution. The focus on functions and qualities underlines user requirements, and enables broader thinking as to the possible solutions and technologies. Chapter 2 80 Phaal and Muller (2009) defend this shift in focus is possible through technology roadmaps, as these can support a range of different business aims, including product planning. The main benefit is the communication that is associated with the process and a common framework for thinking (Farrukh, Phaal, and Probert 2003). Johnson (2012), by applying the technology roadmaps to manufacturing technology, found that roadmapping is, in fact, “an effective tool for sharing, capturing and aligning information for developing and communicating the strategic vision, and for allocating investments”. While the author found that roadmapping to be effective at a strategic level, he argues is not an appropriate method for information on the manufacturing technologies required for a particular product being developed. We argue that perhaps the shift advocated by Phaal and Muller (2009) can also be possible using Ashby’s selection strategies. Similarly to the material selection process, Ashby (2005) posits it is important to choose the most appropriate process-route (manufacturing technology) for a material early in the PDP so as to avoid the costs associated with rework. The process to be undertaken will depend on the material selected, its size, shape and precision, as well as the production volume. Identically to materials, processes can be characterized by attributes. Its selection implies finding the best match between its attributes and design requirements. Ashby proposes a procedure for the selection of the most appropriate manufacturing technology parallel to that of material selection, illustrated in Figure 2.22. Once again, the underlying premise is that all manufacturing processes should be considered as viable until proven otherwise. The first step is then to translate the defined design requirements into constraints in terms of material, size, shape, tolerance, roughness and other relevant parameters. The constraints identified will then be the basis for the screening of the processes that cannot comply. For example, some processes are not economically viable unless substantial quantities are produced; some are incapable of producing large part sizes; while others cannot produce the desired geometric complexity (Eggert 2005). Diagrams or Process Information Maps, also known as PRIMAs (see Swift and Booker 2003), or other decision support tools (software) may be extremely useful for the screening process. The procedure then continues with the ranking of the processes by cost. On this, Ashby proposes four basic design guidelines to minimize the process cost: Keep things standard; Keep things simple; Make the parts easy to assemble; Do not specify more performance Literature Survey 81 than is needed. Consenting that “cost is one of the key strategic elements of product competitiveness and an early appreciation of the relationship between major design choices and the cost of the resulting product is a vital element of effective product development” (Clark, Roth, and Field 1997), Ashby proposes two possible approaches to rank the processes based on this criteria: economic batch size and technical cost modeling. While the former seen as a short-cut approach, the latter allows for a more precise insight because all inputs (materials, capital, time, energy and information) that contribute to the cost of the product have to be considered in the equation (Clark, Roth, and Field 1997; Ashby 2005). Figure 2.22 Four-step process selection procedure. It is performed in parallel with the material selection. (Ashby 2005) Having arrived at a feasible short-list, supporting information (such as details, case studies, availability, warnings, and so on) is then needed to help the final decision. As we’ve seen in the material selection process, choosing the most appropriate manufacturing technology is also complex. Nonetheless, a systematic approach instead of the mere reliance on past experience fosters better decisions. Chapter 2 88 accordance with the purpose and the focus of the model. In this study we are particularly interested in modeling the different activities performed by the different roles involved in the process and how they interacted. The roles involved in the PDP and how they interact are vital to the proper implementation of the PD principles and practices. The modeling framework developed in this research took this into consideration, i.e. to be able to model the activities but also how the roles involved in the project interact and make decisions in the different stages of the process. The framework is based on the Riva method that is, as far as we are aware of, being used for the first time to model the PDP of real projects to add empirical evidence based on a different modeling approach to PD literature. This study will also contribute to literature by evaluating the capacity of the Riva method for modeling real PD projects. Intending to, on the one hand, model PD projects where unproven technology could be adopted in order to contribute with experiential confirmation on successful PD factors and practices, and on the other hand, determine the suitability of a particular modeling approach within the unique realm of PD, this empirical study aims to answer the following research questions (RQ): RQ1. What are the constraints that inhibit the consideration of other materials and technologies in the PD process of rigid parts? RQ2. How is (and should be) the selection (decision-making) process? RQ3. Can we develop a set of principles and practices to facilitate better decisions in opposition to easier decisions? RQ4. Is the Riva method appropriate for modeling PD projects? 89 3 Research Methods and Procedure This study is based on the framing of qualitative research methodology (Neuman 2006; Merriam 2002). The research is divided in two main stages, as synthesized in Table 3.1. The first one is dedicated to the understanding of the current situation of the RIM industry, in particular for the production of three-dimensional (3D) rigid parts. The major players are identified along the value chain, from the suppliers of raw materials, the manufacturers of the molds, the injectors of the parts, to the final assemblers and owners of the products that incorporate the parts. Table 3.1 Characteristics of the two main research stages Stage 1: RIM industry analysis Stage 2: Case studies Approach Inductive Deductive Type of research Basic interpretive qualitative study Qualitative case study Data collection strategies Secondary analysis Networking Non-participant observations Semi-structured interviews Open-ended interviews Document analysis Validity/credibility Data triangulation Member checking Data triangulation Member checking Chapter 3 90 In this stage the research is mainly inductive (Merriam 2002; Neuman 2006), as there was no pre-existing theoretical framing due to the lack of available information about this industry and technology. The research design for this stage of the study is a basic interpretive qualitative study that is analyzed through descriptive methods. Secondary analysis, networking, non-participant observations and semi-structured interviews were used as data collection methods. At the end of this stage (see Section 3.1) it was possible to understand the structure of the value chain, the production processes and most of the applications where RIM has competitive advantage. Moreover, it allowed us to identify the key issue in this industry, which is the predominant role of the designer (i.e. the development team) of the product. This stage settled the context for the second stage of the research. With this stage it was possible to identify the focus of interest for this technology and consequently for this research. The second stage of the study (see Section 3.2) is devoted to the analysis of the PD process of products where RIM could have been a strong candidate for the production of one or several parts. The consideration whether RIM is a strong candidate or not is based on the criteria developed in the previous stage. In this stage two qualitative case studies were performed, one where RIM was selected and another where RIM was not selected. Both projects were considered successful with financial performance of the resulting products in agreement with the expectations of each owning company. Despite the fact that a case study inevitably results in new concepts and theory (which also occurred in this research, as will be explained in Chapter 5), in this stage the research is mainly deductive, because it intended to find in real projects practical evidence of the existing PD theory. The data collection methods used was open-ended interviews and document analysis. The data collected was analyzed with a combination of methods, namely textual narrative and process modeling. Later in this chapter, the justification for the modeling framework used in this stage of the study will be discussed. The PD process is modeled for each company, with particular attention to the concept development stage. The factors that favor a more proactive exploration of the development space are identified. The theoretical framing developed in the previous chapter is particularly important for the execution of this stage, because the goals of this research were not only to understand the Research Methods and Procedure 91 development process, but especially to add empirical evidence to theory and to validate the modeling framework used in the case studies. In both stages appropriate validation methods were employed, namely data triangulation and member checking, in order to ensure trustworthiness of the research (Kirk and Miller 1986; Denzin and Lincoln 1994; Creswell and Miller 2000). 3.1 Stage 1: RIM Industry Analysis The analysis started with a high level picture of the plastic industry, investigating the different current applications, the most important suppliers of the raw materials and the typical supply chains for each type of application. Figure 3.1 shows the research steps performed during this stage. Participation at Polyurethanes 2009 Technical Conference, in Washington DC, with the presentation of a poster (Torcato et al. 2009) made an important contribution in this stage. Afterwards, we focused on the investigation of the supply chain in the applications where RIM technology is used. Most of the companies were identified by web search, followed by email to confirm the main activity of the company. The email communication also proved to be a good source of information because some companies were willing to indicate their main competitors and suppliers. This step further improved the knowledge about why, where and how RIM technology is currently being used. Special attention was provided to the applications where 3D solid parts are needed, because of the interest of the research group in this type of application, as explained in the previous chapter. Figure 3.1 Stage 1 research scheme Finally, three companies that are representative of the traditional RIM technology user were studied. These RIM users are specialized in the production of 3D complex shaped 1st step - Plastic industry Secondary analysis Networking 2nd step - RIM technology Secondary analysis Networking 3rd step - RIM users Non-participant observations Semi-structured interviews Chapter 3 92 parts, including the injection, the finishing operations and, if demanded by the customer, some assembly operations. Table 3.2 exhibits the RIM users studied in this stage of the research. The field research involved the permanence in each company from one day to one week, depending on the conditions offered by each company and the budget available, in order to perform the following activities:  Deepen the understanding about the position of the company in the value chain.  Characterize the company.  Typify the portfolio of products and services.  Describe the production process, including the involvement of the company in the PD process, the mold design process, the pre-injection, injection and post-injection operations and assembly.  Identify the key success factors.  Validate the framework of applications where RIM has competitive advantage.  Validate the factors that inhibit the companies of achieving bigger sales and increasing the customers’ base. Table 3.2 RIM users studied in stage 1 of the research Armstrong Mold Corp. located in New York, USA Manufacturer of metal (aluminum and zinc) casting and plastic molding components in prototype and low-volume production quantities. RIM is mostly used for the production of PU covers for the medical sector. It represents around 30% of the business. RIM Manufacturing LLC located in Texas, USA Specialized in the production of custom PU parts, mainly for bezels and covers of a wide variety of products and industries. Production capabilities from injection to post-injection operations (finishing, painting), assembly of finished parts and quality control. RIMSYS located in Portugal A new company that started operations in the fourth trimester of 2009, located in the north of Portugal, specialized in RIM processing of DCPD for the production of exterior panels of vehicles. They have the knowledge and technology to produce and assemble enclosures for any kind of product. Research Methods and Procedure 93 The two US companies were visited during the researcher stay at MIT in the fall of 2009. They were very open and provided most of the data asked for, that allowed the development of grounded knowledge about the utilization of the technology, such as attributes of RIM, its qualitative positioning relative to other technologies, typology of applications and process flow. After the return of the researcher to Portugal close contact continued via email and telephone for feedback and further discussion. The stay in Armstrong Mold involved non-participant observation (Patton 1980; Babbie 2013) of the facilities and the operations performed in each step of the production process, interviews with the engineering manager and the operations manager of the plastics division, interviews with the director of marketing and documental analysis. An important facet of the visit was the non-participant observation of the beginning of a new project, namely the kick-off meeting where the part to be produced was analyzed and the planning of the project was discussed. The period spent in RIM Manufacturing allowed for a deeper analysis of the RIM process flow, including the post-injection finishing operations and some assembly operations. The visit involved interviews with the Chief Executive Officer and the Chief Operating Officer and non-participant observations of the production process. RIMSYS was visited in 2010 for observations, interviews with the Chief Technology Officer and follow-up discussions. As this company was in the first year of operations the follow-up discussions were not as productive as expected. This is, to some extent, understandable because the Chief Technology Officer was completely absorbed with the day-to-day operations due to lack of staff, which is typical in a start-up, leaving him little time available to participate in this research. Nevertheless, this company provided valuable information about RIM processing of DCPD systems, which is a material that many companies are not capable of transforming, including the two US companies studied. These companies were interested in the research because:  They recognize that research work involving RIM is beneficial for their business.  They want to understand the benefits of RIMcop®.  They are mainly in prototyping and low volume production, thus they are highly involved in the concept development stage of the PD process. As a result, they would like to learn more about the CE philosophy and the benefits of using its principles to support the decisions involved in the concept development stage. Chapter 3 94  They agree that the downstream processes are poorly considered in the development of product design concepts. The first criterion for the selection of the companies was the technology: the companies had to be specialized in RIM. Followed by the type of products or components manufactured, the companies had to be specialized in the production of 3D complex shaped rigid parts. Finally, researcher accessibility to the companies was determinant to finalize the selection process. The companies selected were the ones that showed more interest in the second step of this stage, were the ones that replied to our e-mails and eventually asked questions and contributed with suggestions. Furthermore, when invited to participate in the third step of this stage they accepted the solicitation with enthusiasm and offered their contribution for the planning and execution of the visits. The number of companies selected was determined by the above-mentioned selection process together with the budget available to perform this step. Thus, it was possible to study two companies in the USA and one company in Portugal. Given the objectives and characteristics of this step of the research, we consider three companies a fair sample of the population. We are confident that the collected information is representative of the reality of this industry and at least sufficient for this qualitative study. Notice that the companies are not homogeneous, they are very different in size and nature. Armstrong Mold is in the medium size of the SME with around 200 total employees and is not dedicated exclusively to RIM. RIM Manufacturing is a small size company, with less than 30 employees, specialized in the production of PU parts using RIM technology. RIMSYS is a start-up, with less than 10 employees, producing RIM manufactured PDCPD parts. An important issue in this step is that all the companies required that any information (for example in the form of reports, papers, etc.) resulting from this particular study had to be validated by them before becoming public. One of the companies even demanded the signature of a Non-Disclosure Agreement (NDA). Consequently, much information could not be explored in this thesis by imposition of the companies. However, we believe that this situation did not impair the quality of the thesis because the focus and scientific added-value are in the study of the PD process and its modeling, particularly the discoveries that were achieved in the two qualitative case studies explained below. We feel that, even without this contribution, we have achieved Research Methods and Procedure 95 the goals of this research and demonstrated the validity of the results and the contribution to the research area. This phase has the merit of allowing the researcher to deepen and systematize the knowledge about the RIM industry and technology, so as to have sufficient comprehension to competently perform the second stage of the research. Some of the results of this stage of the research were presented in two publications (Torcato, Santos, Dias, Roth et al. 2011; Torcato, Santos, Dias, Olivetti et al. 2011) and the feedback given by the scientific community in the conferences where the papers were presented had an important contribution to the design and execution of the remaining research. 3.2 Stage 2: Case Studies The field research performed in this stage is a qualitative case study (Merriam 2002; Yin 2009) where the units of analysis are two companies located in Portugal with vast experience in PD. Table 3.3 exhibits the companies involved in this stage of the research. They finished a development project of a product with enclosures suitable for RIM manufacturing prior to the involvement in this research. Table 3.3 Companies involved in stage 2 of the research CEIIA located in Portugal Centre for Excellence and Innovation in the Automotive Industry, a Portuguese Engineering Design Center specialized in product and process integrated development for the mobility industries. Bosch Termotecnologia located in Portugal A Competence Center located in Portugal specialized in products for domestic water heating that belongs to the multinational group Bosch. Responsible for the design and development of new devices and their production and marketing under the trademarks Bosch, Buderus, Junkers, Leblanc and Vulcano. The research approach is deductive (Babbie 2013) as there is a theoretical basis about the PD process and about the modeling approach that supported the execution of this stage of the research. Data collection is based on open-ended interviews (Flick 1998) to the experts working at each company, supported by documental analysis (Clarkson 2003) of digital and physical records with supervision of those experts. Chapter 3 96 The principal purpose of this stage of the research is to model the PD process of two particular projects where RIM could be an interesting solution for the enclosures of the product or family of products. We believe that these process models will allow a deeper understanding about the dynamics of PD projects and how companies manage these highly complex, uncertain and goal oriented processes. The focus of the modeling is on the decisions performed throughout the project; in particular the decisions related or that influence the selection of materials and technologies for the enclosures of the product. This focus in one particular subsystem of the product is justified by the technology that is under study in this research (RIM) and it’s appropriateness in that particular application. Moreover, this focus brings the benefit of facilitating the analysis, making it manageable for the researcher that, if required to analyze the decisions related with all the systems of the product, would not be able to finish the research in reasonable time. The findings are thus not generalizable to all the decisions performed in a PD process, as in any case study (Stake 1995). In many subsystems of a product, other factors enter the equation, such as: sharing components or subassemblies with other products; buying standard components or subassemblies from other companies; or developing a proprietary technology for the component or subassembly or for the production of the component or subassembly. The findings are applicable to the subsystems where a unique design is required and there is no preexisting interest in using or developing a proprietary solution. The projects under analysis are already completed and the products are already on the market, so there is no risk of the researcher interfering in the course of events. A drawback of this retrospective analysis is that the level of detail and personal involvement of the researcher is more limited than in an ethnographic analysis (LeCompte and Schensul 1999; Neuman 2006). However, given the objectives and the focus of the research, we think this is the best research design for the following reasons:  The projects had to be completed and with the materials selection already performed as the interaction with the researcher, which is interested in discussing a particular technology, could alert the development team to that solution and thus influence the selection process.  Only after the end of the project, are we able to have a higher level of certainty that RIM could have been an interesting solution to consider. Research Methods and Procedure 97  The projects had to be considered successful by the respective companies, to qualify for this study. It was only possible to have that assessment after product launch in the market.  Restrictions on the level of confidentiality would make the monitoring of projects in implementation phase extremely difficult, in some tasks impracticable. As was demonstrated in the visits to the RIM users. In the CEIIA project RIM technology was selected and in the Bosch Termotecnologia project RIM technology was not selected. This does not necessarily mean that in the former case the right decision was made and in the latter case the wrong decision was made. As will be demonstrated later on in the discussion, many factors contribute to the decisions and other technologies may be considered as well as RIM. Qualitative data was collected, synthesized and organized in order to enable the construction of the model. The process modeling procedure was performed with the support of narrative analysis (Riessman 1993; Labov and Waletzky 1997). The modeling framework will be discussed in the next chapter of this thesis. For validity purposes, data triangulation and member checking (Denzin and Lincoln 1994; Creswell and Miller 2000) were performed during the execution of the process model. Data triangulation was mostly performed by means of documental analysis (Clarkson 2003). Member checking was fundamental not only for validation of the information but especially for the development of the model. Because it was the feedback obtained during the member checking meetings that most contributed for the detail displayed in the models. These case studies are relevant for the overall research because:  They provide empirical evidence of the hypothesis developed in the research. In particular, the framing where RIM technology is competitive (technical requirements and production volumes) and the difficulty of the development team in making the switch to unfamiliar materials and/or production technologies due to risk aversiveness.  They also provide insights about the PD process adopted in these organizations, enabling the assessment of the level of implementation of CE principles and the mapping of the factors affecting the success of PD project.  The information collected will be very helpful to draw conclusions about the factors that contribute to the selection of a particular technology by the development team. As will be noticed, these do not rely on technical issues only. Chapter 4 104 The result of this work is two process models and narratives, which describe in rigor and detail the two real projects performed by these companies. The models are analyzed and the findings provide empirical insights about the PD process of these two projects. Ultimately we seek to give a contribution to PD and process modeling theory. The next paragraphs will explain the modeling procedure for the implementation of the framework in the case studies. A RAD of the research procedure is depicted in Figure 4.2. The modeling procedure is, itself, a process where the researcher interacts with the experts of the company in order to produce a process model and narrative of the PD project under analysis. Figure 4.1 Process modeling framework Thus, the “project case study” is the unit of work (UOW) of the researcher’s modeling procedure. Each project studied by the researcher is a case or instance of the UOW “project case study”. For simplification, one case was identified as “The CEIIA case study” and the other “The Bosch case study”. In both cases, the work performed by the researcher followed a standard set of activities, depicted in Figure 4.2 process model. Modeling Framework 105 In both cases the process was activated when the researcher and the company agreed on the terms and conditions of the collaboration. This negotiation process involved the definition of the research boundary and ownership, the selection of the PD project object of analysis, confidentiality management and the company’s roles that would collaborate in the modeling process. Figure 4.2 Modeling procedure for the case studies research The collaboration is based on a succession of interviews performed by the researcher to the above-mentioned roles (described in Figure 4.2 as interviewees). The first round of interviews allowed for a coarse representation of the PD process, this is the first iteration Chapter 4 106 of the process model. With this model in hand, a second round of interviews was then performed, this time accompanied by document analysis. For confidentiality reasons, the documental analysis was first reviewed by the interviewees to filter the information that could be released to the researcher. Document analysis is necessary for two reasons: to offer the details needed to write the textual narrative of the process; for validity purposes, allowing data triangulation with the information collected in the interviews (Patton 1980; Neuman 2006). This enhances the process model, as well as improves the robustness of the findings. Several documents related with the project under study are analyzed. Both companies have document management procedures and project dossiers organized under standardized rules, for traceability and compliance with quality management standards. The documents related with the periodic description of the project status in particular are very helpful to understand the evolution of the level of knowledge of the design team and the sequence of decisions performed on the PD process. Also, the technical documents, like 3D computeraided design (CAD) models, bill of materials (BOM), material technical sheets, CAE analysis and others are very useful to discover the underlying justification for some of the decisions performed by the design team. The analysis of the data collected in the second round of interviews and in the document analysis results in a significant improvement of the process model (this is the second iteration of the process model) and also enables the first iteration of the textual narrative of the process. Following this procedure, both the process model and the narrative are discussed with the company representatives. This member checking is a fundamental contribution for the validation of the data collected and also for the development of the model. The feedback obtained during the discussion meetings represents new information, both in terms of quantity and, especially, in terms of quality, because the discussions are based on a concrete model and narrative of the project, thus facilitating the identification of gaps and misunderstandings. Afterwards, a final version of the textual narrative and process model is proposed and integrated in a final report that is then delivered to the company for validation. This report also includes the discussion of the findings in comparison with theory and the preliminary conclusions and managerial implications. The report is discussed internally, at the company core, without interference from the researcher. This part of the procedure is done on a deferred basis. The discussion includes Modeling Framework 107 the validation of the contents in the report and the screening for confidential information. This member checking may result in a series of recommendations or corrections, if the company considers the report has incorrect information or is disclosing confidential information. If the researcher receives comments and corrections, it is then necessary to go back and rebuild the textual narrative and process model. Consequently, the final report is updated and resent to the company for validation. The researcher performs this rework until the company roles involved in the validation of the report are satisfied with the content, both in terms of trustworthiness and confidentiality. At this point, the company informs the researcher of the acceptance and validation of the report and, with this, the process modeling procedure is concluded. The next sections of this chapter introduce the case studies, give a synthetic explanation about the projects that were analyzed and modeled, particularly the product parts that were, or could have been RIM manufactured (the enclosures of the product), and conclude with a brief description of the specificities of the implementation of the modeling procedure in each case study. 4.1 The CEIIA Case Study The Buddy project was developed in CEIIA between March 2008 and February 2010 with the main objective of releasing a restyled model of Buddy, the Norwegian electric quadricycle vehicle produced by Buddy Electric AS, formerly known as Elbil Norge AS (Figure 4.3). The project involved the complete vehicle development, from style definition to functional prototypes manufacturing. This project is particularly interesting because DCPD material was selected for all exterior panels of the vehicle, which are manufactured with RIM technology. Furthermore, CEIIA and Elbil Norge had no previous experience with this material and technology. Also, the supplier of the material (Telene), despite the vast experience in the application of DCPD in the fabrication of exterior panels, was never involved in a similar experience, that of producing all body panels in the same material. All the companies involved in the project were exploring completely new territories; nevertheless they managed to reduce the uncertainty and risk associated with the decisions using several approaches that will be described and discussed in the next chapter of this thesis. Chapter 4 108 This was a very important project for both CEIIA and Elbil Norge. It was important for the former because both the product and the activities required were right at the core business of CEIIA, it was important for the latter because it was, and still is, the sole product of the company, representing the only survival option. Figure 4.3 Buddy electric vehicle The project was considered by CEIIA and Elbil Norge a success, both in terms of technological innovation and in terms of financial performance. Despite the high levels of uncertainty, the project finished on time and on budget. The product is selling according Elbil Norge’s expectations. The researcher was not allowed to record the interviews, which forced more discussion meetings in order to arrive at a satisfactory result. On the other hand, there was openness to provide technical information and some detail about the technologies studied by the development team. Generally speaking, the process modeling procedure was adopted as described in the previous section with no relevant setbacks except the one explained in the previous paragraph. The process model and textual narrative was produced and validated by CEIIA requiring only one rework cycle. 4.2 The Bosch Termotecnologia Case Study The water heater project was developed in Bosch Termotecnologia with the main objective of releasing the first model of a standalone heat pump (Figure 4.4). The project started in January 2010 and concluded in June 2011 when the product was launched under the trademarks Junkers, Buderus and Bosch. The first countries to receive this product were Modeling Framework 109 Belgium and Poland followed by France. The project involved the complete PD, from style definition through functional prototypes manufacturing and production ramp-up. This project is particularly interesting because EPP (expanded polypropylene) material (which is a direct concurrent of PU) was selected for the production of the enclosure of the heating module. EPP was far from being the natural choice, as will be explained in the next chapter. As in the previous project, Bosch was not familiar with this material and production technology. However, they found in their qualified suppliers’ database companies with experience in EPP application on exterior panels. Figure 4.4 Standalone heat pump. Reproduced with permission from Bosch Termotecnologia. Copyright 2012 Bosch Termotecnologia. Bosch Termotecnologia is particularly interesting because it has a history of success where the introduction of new materials and technologies in their products is concerned. The company and the multinational group that the company belongs to have, deeply routed in their core values, a culture and philosophy of innovation. The PD process is tailored to manage the risk associated with the exploration of new materials and technologies. The proof of this is the project under analysis, where EPP material was adopted for the first time in the enclosure of water heaters. Nevertheless, RIM was not considered in the selection process due to unawareness about this particular technology among the members of the design team and the qualified suppliers. This is consistent with the assumption that RIM technology is not well known in Chapter 4 110 the plastics industry and in the industries that develop and assemble products that can potentially incorporate plastic components. This lack of awareness, together with deficient prospection and dissemination activities of the RIM users, limits the opportunities for the introduction of this technology in new applications. Bosch considered the project a success, both in terms of technological innovation and in terms of financial performance, but particularly in terms of innovation, because it was the first standalone heat pump to be launched in the market by Bosch. Furthermore, it was the first to introduce the EPP material in the heating module enclosure. The researcher was allowed to record the interviews. Thus, the interviews were recorded, transcribed and validated by the company, facilitating the downstream activities required in the modeling procedure. On the other hand, there were some company-imposed restrictions, which limited the access to project documents and information, resulting in less detail than in the previous case study. A considerable amount of time was allocated to the discussion and revision of the model and narrative, fruit of the intense interactions needed to compile the required information. The process modeling procedure was adopted as described in the previous section, with no relevant setbacks except the one explained in the previous paragraph. The process model and textual narrative was produced and validated by Bosch after a very intense discussion and revision process, as mentioned above. 111 5 Findings and Discussion This chapter starts with the findings related with the current situation of the RIM industry, in particular for the production of 3D rigid parts. These build on the literature review carried out in Chapter 2, and add practical information regarding the technical and business description of RIM, retrieved during our interaction with the various stakeholders of the industry. The focus here is on the business description in order to overcome the gap identified in literature, which seldom discusses the issues in this area. The three RIM companies visited by the researcher provided valuable information about the production process, the current applications and the positive and negative attributes of the technology. The value of these findings is on describing the usage of RIM technology in real companies and identifying the difficulties, both technical and managerial, that they face. Thereafter, the findings related with the two qualitative case studies are presented. These case studies are two PD projects that have successfully completed. One project is from CEIIA and the other project is from Bosch Termotecnologia. The findings are presented in the form of two process models and narratives, which describe in rigor and detail the two real projects performed by these companies. The models are analyzed, providing empirical insights about the PD process of these two projects. The focus is on the development of the enclosures of the products, which were, in the case of CEIIA, or could have been, in the case of Bosch Termotecnologia, RIM manufactured. Chapter 5 112 In the discussion section we seek to give a contribution to PD and process modeling theory. Brown and Eisenhardt’s framework of factors that affect the success of PD projects are compared with the findings of the case studies, resulting in the proposal of a modified framework. The method used in each project to perform the materials selection is discussed and compared to the materials and processes selection strategies presented in theory, particularly Ashby’s systematic procedure. The degree of implementation of CE principles and lean development system principles in each project is also discussed. Finally, the effectiveness of the modeling framework developed for this research is discussed. The expectations created in the development of the framework are compared with the actual analysis and findings that the produced models allowed. 5.1 RIM Industry Analysis 5.1.1 Plastic Industry Plastic industry includes all businesses that rely on the utilization of any type of plastic material to create value. It is a pyramidal industry, with a small group of companies in the extraction of the feedstocks from the natural resources, namely petroleum, and production of the different monomers that are then transformed into the commercially available plastics ready for fabrication (see Figure 5.1). This part of the industry is typically process-based with emphasis on the chemical and process engineering disciplines. These plastics are used in an immensity of products including packaging, building and construction, electrical and electronic, appliance, automotive, and practically all markets worldwide (Rosato, Rosato, and Rosato 2004). Thus, a large number of companies and value chains operate in these markets, developing products that rely on the use of plastics and related fabrication technologies. The classification of materials into three main families is widely accepted: ceramics, metals and plastics. The family of plastics, which comprises more than 35.000 types, can be divided into two main classes: thermoplastics (TPs) and thermosets (TSs). TPs are, by large, the most widely used, with over 90wt% of all plastics. TPs are processed by applying heat to soften the material that then is pushed through pressure or another mechanical means to form the shape of the part. They can be repeatedly softened by reheating, thus can be processed several times. TSs can be processed by applying heat and/or pressure or by the reaction of two monomers. They flow to meet the shape of the mold and, at a certain temperature, they cure, solidifying after a Findings and Discussion 113 cross-linking chemical reaction of its molecules. In contrast with TPs, cured TSs cannot be re-softened. Figure 5.1 Overview of the plastic industry from source to products (Rosato, Rosato, and Rosato 2004) As explained in Chapter 2, RIM is a manufacturing technology used to process TSs, namely PUs and PDCPDs. Compared to the major manufacturing processes (extrusion, injection molding, blow molding and so on) RIM is a relatively new manufacturing technology, consuming a marginal weight percentage of all plastics. Nevertheless, due to its versatility and capacity to produce high quality parts, we consider this as an opportunity for growth rather than a disadvantage. 5.1.2 RIM Industry It is not an easy task to delimit and characterize the RIM industry. Actually, RIM is a manufacturing technology to produce plastic parts (see Chapter 2). It is a plastic molding process that, in theory, can be used in any industry for the production of any plastic component of a product. With this section we intend to introduce the concept of RIM industry, which we define as the system constituted by the companies specialized in the manufacture of plastic parts using RIM technology (RIM users), their suppliers and customers. The focus is on the RIM users, i.e. the companies that use RIM technology to produce their products, which usually are components or subassemblies to incorporate in the product of the customers. Chapter 5 120 many manufacturability issues. This problem was already identified in the preceding section. This underlines the importance of DFM and the advantages of stakeholders’ early involvement in concept development. There is no commercial software available for the simulation of the RIM injection process (equivalent to Moldflow for TPs). This tool would be very useful in the design of the part and the mold because the design decisions would be based on more accurate predictions. The experience of the two US companies is that if the production volume is more than 5.000 parts per year, RIM may not be competitive and the selection usually goes to lowpressure structural foam molding or TP injection molding. These companies already had production volumes of 1 part to 20.000 parts per year. The average and most common production volumes are between 100 and 1.000 parts per year. Likewise, the production lifetime can go from 1 single part to 15 years of production of the same part. In average, the production lifetime is 5 years. The parts produced by these companies are integrated in their customers’ products with a broad range of applications, such as medical equipment parts, automotive and truck parts, and agricultural, construction and utility machinery parts. These applications are in accordance with the general trend of the market, as explained in the literary review. RIMSYS is exclusively producing DCPD exterior panels for the automotive industry. Armstrong Mold works for several industries, the most important one being the medical devices manufacturers, followed by the electrical and telecommunications equipment. RIM Manufacturing has a wide variety of customers in several markets, including medical devices, automotive, industrial equipment, and others. Other technologies are in the market competing with RIM for the same applications. The most relevant ones are sheet metal, thermoforming, low pressure structural foam molding, TP injection molding, RTM, SMC/BMC, LFT, CSM and rotational molding. Customers may also select machining, if they are just making one part. The process flow observed in these companies for the production and delivery of the RIM parts is depicted in Figure 5.4. The activities that are always performed by the RIM companies are inside the grey area. They include the manufacture of the plastic parts, the assembly of the parts (if contracted with the customer), quality checks and packaging. The assembly operation usually involves several RIM parts produced by the company and may include other parts supplied by the customer or by external suppliers. Findings and Discussion 121 Some companies also have the capability to design and manufacture the RIM mold, which is the case of Armstrong Mold. RIMSYS and RIM Manufacturing do not have this capability but are highly involved in the design process of the RIM mold because most of the mold shops do not have experience in the production of molds for RIM. If the RIM users leave the decisions to the mold shops, the molds are designed based on the experience of TP injection, resulting in an unnecessary increase in cost. Figure 5.4 RIM process flow An important issue in mold design that greatly influences its cost is the mold material. Unlike TP injection where steel (or aluminum for a tool life of less than 100.000 parts) molds are a must, in RIM several possibilities can be considered depending on the tool life required. Table 5.1 exemplifies the tooling options these companies consider in each project. The selection of the mold material depends on the complexity and size of the part and the number of parts the customer needs. Naturally the mold material influences the quality of the produced part. However, the finishing operations can successfully convert a Chapter 5 122 poor quality part in a high quality part. Of course, better injected parts require less finishing, thus minimizing the post-injection costs. In the development phase of the final product, a decision is made to use a RIM manufacturable material in a particular part or in a set of parts. Afterwards, the development team designs the part, together with the rest of the product, generally with no involvement of the RIM user. Thus, frequently rework in the part design is necessary to facilitate mold and part manufacture. Table 5.1 RIM tooling options Mold material Part size and complexity Maximum nr. of parts Cost ($) Silicone rubber Small <1500 <10.000 Cast epoxy Small <3500 <20.000 Cut Ren shape Small <3500 <20.000 Cast aluminum or Kirksite (zinc-aluminum alloy) Small to big – Deep geometry >50.000 5.000 – 35.000 Cut aluminum Small to big >150.000 6.000 – 70.000 The part manufacture is divided in three steps: setup, injection and finish. Setup involves the fixation of the mold in the clamping unit, cleaning the mold, calibrating the temperature of the mold and the temperature of the monomers and finally injecting the first articles until the quality of the part is at the expected level. Then, the actual part production begins, which is basically the injection cycle. If processing a PU system release agent is sprayed, the mold is closed and the monomers are mixed and injected in the mold by the mixing head. After that, the curing starts involving the exothermic reaction of the monomers and subsequent solidification, forming the plastic part. Curing is the lengthiest stage of the injection cycle, representing between 30 and 70% of the cycle time. During this stage the operator deflashes the part injected previously. Finally, the mold is opened and the part is removed. Finishing involves all the operations required to put the injected part in accordance with the customer’s quality and functional requirements. It includes trimming, adding inserts or other hardware, cleaning, inspecting and repairing the part. Finally the part is painted, if contracted with the customer. Sometimes it is necessary to add hardware after painting. Findings and Discussion 123 The average reject rate in the studied companies is between 3 and 5%. As mentioned previously, part complexity increases reject rate. Most of the rejects are in the beginning of production (first parts) due to the manual adjustment of mixing parameters. These companies only use RIM systems. Armstrong Mold and RIM Manufacturing only use PU system, mainly Bayer systems. RIMSYS only use DCPD systems provided by Telene. None of these companies formulate their own systems. Nevertheless, there are competing companies operating in the market that buy the monomers and additives and formulate their own proprietary systems. Generally, the prices of the systems go from 3,3€/Kg ($2/lb) to 6€/Kg ($3,6/lb). The very low-density systems go up to 16,7€/Kg ($10/lb). The raw material weight required to produce a part is the part weight plus the material required for the runner, gate and flash, which is between 0,5 and 1 Kg (1-2 lb). The average scrap rate is 5% and depends on the dispensing machine being used. The scrap and the rejected parts are discarded and have no commercial value because the materials are thermosets, thus cannot be reprocessed. The main conclusions of this stage of the research are the following:  RIM is mostly applied in PD, prototyping and low to medium volume production;  In markets like automotive and medical equipment;  Where time-to-market is highly valued. In this study we were able to acknowledge that for the production of a RIM part, the part design features are usually the dominant factor in mold design, while the mold design characteristics, in turn, dominate the process plans of mold manufacture and molding operation. As a result, most of the costs of mold-making and molding operation, as well as the quality and reliability of molds and moldings are determined in the development stage of the product that the new RIM part belongs. Therefore, there is a need for a verified design that will provide the necessary insight into metrics such as development lead time and manufacturing costs to deal with the decisions required in early stage design in order to reduce the subsequent redesigns and reworks. Based on the characterization of the industrial practice of RIM, we believe the concurrent concept development is the most suitable approach (see Section 2.1.3). However, the concurrent development of RIM parts involves a substantial practical knowledge component (heuristic knowledge) of the relationship between part features, mold design requirements, mold-making process characteristics and the selection of production molding parameters. The designs and process plans involved are predominantly Chapter 5 124 based on the experience of designers. The processes rely heavily on engineers to define their designs in detail. Extensive mathematical analysis is often not used, as analytical models with sufficient accuracy and efficiency are not available. Calculations are limited to empirical rules. Hence, the designers of parts and molds are required to have a high standard of specific knowledge and judgment. Moreover, most decisions concerning the details of the design demand knowledge regarding the mutual influences between the various quantities. Changing one quantity in order to achieve better results, for example part design features, may have a negative effect on other influencing factors, for example the mold design and mold-making process. This implies that knowledge and expertise of more than one specific area are required to have an optimum solution. The inherent complexity and intensive knowledge requirements of this concurrent development problem are crucial aspects to consider. In synthesis, the PD process is the key for material and technology selection. If performed with proper levels of knowledge and concurrency the development team will select the optimum solution that, under certain conditions, may be PU or DCPD processed with RIM technology (see Section 2.3). RIM users alone have no capacity to influence the decisions of the PD team, as demonstrated in the last section. The development process is what needs to be improved so that the best choices are made, and thus the RIM users are protected by the process, because it is also of the interest of the development team to make the best choices (i.e. meet performance requirements at the lowest possible cost). The next section will present the findings of the two PD projects studied. One project is from CEIIA and the other project is from Bosch Termotecnologia. 5.2 Case Studies 5.2.1 The CEIIA Case Study CEIIA typically works on PD projects. These projects can be extremely diversified, ranging from particular PD activities (like styling, CAD modeling, CAE analysis, reverse engineering and prototyping), to a complete PD service. Thus, the “project” is CEIIA’s unit of work (UOW). Each project that CEIIA is working on is a case or instance of the UOW “project”. The work performed to take care of each case of the UOW is a case process (CP), which follows a standard set of activities (depending on the type of project). Findings and Discussion 125 In PD projects, the work is performed following the process described in Figure 5.5. This diagram is a generic view of the process intended to illustrate the most important activities performed and the terminology used in CEIIA. Note that CEIIA does not have a formal prescriptive representation of the PD process, this diagram was developed by the researcher to facilitate communication. The activities are chunked into three phases: Concept, Feasibility10 and Detail Design. Figure 5.5 Generic product development process at CEIIA11 Besides a list of phases and activities, not much information can be taken from Figure 5.5. A process model must show the coherence of actions and interactions carried out and the roles that perform or participate in them. The modeling framework adopted is based on the Riva method which uses the Role Activity Diagram (RAD) to describe an individual process (Ould 2005), as explained in Chapter 4. 10 Feasibility: part design with concept validation in relation to function, materials, manufacturing processes, assembly, maintenance, cost, perceived quality and other relevant aspects to the part. 11 BOM (Bill of Materials): components list that is the basis for design, purchase and manufacturing processes control. It can be an excel sheet or come from a data base. The components are allocated to functional group (e.g. chassis, powertrain, exterior body, etc.) and identified with a part number. The basic properties of each part are controlled with this list (e.g. quantity, weight, material, supplier, general dimensions, versions, etc.). Packaging: activity across the entire project in which it is defined the space occupied by each component. Generally follows a three-stage definition. The first definition is estimated with an interval of about 20mm from the final expected dimension and can be a simple geometric representation. In the final stage it has the necessary detail for the exact representation of the part. Teardown: decomposition of the reference vehicle or concept in all its components (listed in BOM format), allocation to functional group (e.g. chassis, powertrain, exterior body, etc.). Includes images to facilitate understanding. Basic properties like quantity, weight, material and supplier are included. It is also signalized if the component is carryover or tailored for the vehicle. Chapter 5 126 The process under study is a case process that deals with one case of a complete PD project. Name of the process:  Handle a product development project: the Buddy case study. This case process was activated or triggered by the Business Development Director after acceptance of the proposal by the Customer (which was Elbil Norge, the manufacturer of the vehicle). The goals of the process are the deliverables contracted with the Customer:  Complete project dossier including 3D models (engineering deliverable);  Ten prototype vehicles;  Production tools (molds) for series production of exterior and interior parts;  Production certificate issued by the homologation entity. Roles involved in the process:  Business Development Director;  Project Manager;  Prototype and Pre-series Manager;  Engineering Design Manager;  Engineering Team: 5 people, 1 team leader;  Style Team: 4 people, 1 team leader;  Quality Department;  Customer (Program Manager);  Telene (DCPD supplier). The RAD for the process “Handle a product development project: the Buddy case study” is depicted in Figure 5.6. The same model is depicted in Appendix 2 in A3 format for better visualization. The pages after the model are dedicated to the textual narrative of the process, as a means to obtain a detailed understanding of the process. Findings and Discussion 127 Figure 5.6 Handle a product development project: the Buddy case study Chapter 5 128 Textual narrative After triggering the Buddy project, the Business Development Director negotiated the necessary resources to perform the project with the Prototype and Pre-series Manager and the Engineering Design Manager. This negotiation was focused on assessing the staffing requirements to attain the deliverables and lead time contracted with the Customer, thus incorporating this project in CEIIA’s active portfolio of projects. The Project Manager was also selected. There was no planning nor were activities scheduled, because CEIIA does not follow a pre-determined process flow. Rather, milestones are defined to ensure focus but it is up to the team to decide what to do to accomplish those milestones. Having completed the negotiation, the Business Development Director nominated the Project Manager, who became the primarily responsible for the project. The start date for initiating the activities was defined and a plan of the project identifying and scheduling the milestones was discussed. These milestones are, in essence, the deadlines for style theme freeze, for architecture validation, to finish feasibility, to finish detail design and to deliver the prototype vehicles. With the information at hand, the Project Manager defined, together with the functional managers, the team members and the time allocation for the project. Given the complexity of the project, two teams were started (the Style Team with 4 members, and the Engineering Team with 5 members) by the Engineering Design Manager. On the start date, a Kick-off meeting was promoted by the Project Manager to discuss the project goals and to determine the activities for the following week. This meeting was attended by all the internal roles involved in the project. This meeting initiates the actual PD activities, in the sense that everything done up to the meeting can be considered as planning activities. After this, and till the end of the project, the Project Manager coordinates and prepares weekly assessment and planning reports (Design reviews). The model shows a lot of concurrency and collaboration throughout the process, especially after the Kick-off meeting12. During the life cycle of the project, one confidential design room was allocated exclusively to the project, so everyone involved was working in the same room. To perform the activity management required in the project, CEIIA used Excel 12 Kick-off meeting: all roles involved in the project are at this meeting to have a common and clear understanding about the background of the project, its goals, deliverables, and so on, as well as the functions and responsibilities of each. Findings and Discussion 129 spreadsheets, PowerPoint presentation software, Catia 3D CAD software, MSC CAE software, and so on. The activities performed in the concept development stage concerned the style theme definition and, in parallel, the architecture definition. The Engineering Team performed the latter and the Style Team, with deep involvement of the remaining roles, performed the former. The Engineering Team assessed feasibility of the different style theme proposals developed by the Style Team. The Project Manager coordinated this interaction and the functional managers followed the work progress. It was at this stage that the Customer was included in the PDP, with the responsibility of assessing the style theme proposals until satisfied. Thus, as everyone was involved in this interaction, the project could only move on after a style theme proposal was accepted by the Customer. Afterwards, the Style Team constructed the exterior style mockup using a 1:1 scale for Customer validation. Changes were made in the mockup until the Customer decided to freeze the style theme. Note that the Customer had a representative (Program Manager) in the design room almost permanently during the Concept and Feasibility stages. In parallel, the Engineering Team started three concurrent threads of activity, which mark the beginning of the Feasibility stage. They immediately started the homologation procedures that involved work till the end of the project. The other two threads started only after the style theme froze, because they involved interaction with the Style Team and the Customer, as well as the other roles. It is at this stage that Telene started its involvement in the project. In this major phase of the process, the Engineering Team validated the architecture and performed all the engineering work referred in Feasibility (see Figure 5.5) with the collaboration of many roles: the Project Manager coordinated the interaction and performed cost assessments; the Style Team was responsible for style refinements; the Prototype and Pre-series Manager assessed the producibility of the components that are not carryovers and, more specifically, manufactured the prototype mold to validate RIM technology; Telene assessed DCPD suitability for the body of the vehicle; and the Customer assessed the different solutions until satisfied. The most relevant and difficult decision points occurred in the Concept development stage and in the Feasibility stage. These decisions were made by the Customer but with deep involvement from the other roles, namely during the assessment of the style theme proposals and during the assessment of engineering solutions.