Method for Analysis of Front-End Technology Development Effectivity by Thomas P. Courtney M.S. Mechanical Engineering Santa Clara University B.S. Mechanical Engineering Rochester Institute of Technology AT THE MASSACHUSETTS INSTITUTE OF TECHNOLOGY JANUARY, 2001 @ 2001 Thomas P. Courtney, All rights reserved The author hereby grants to MIT permission to reproduce and distribute publicly paper and electronic copies of this thesis document in whole or in part. Signature of Author A J /Thomas P. Courtney System Design and Maqagement, January 2001 Certified by Cliff Whitcomb Associate Professor Ocean Engineering Thesis Advisor Accepted by Steven C. Graves LFM/SDM Co-Director Abraham Sieael Professor of Management Accepted by -- Paul A Lagace M/SDM Co-Director Professor of Aeronautics & Astronautics and Engineering Systems MASSACHUSETTS INSTITUTE OF TECHNOLOGY AUG 0 1 2002 LIBRARIES BARKER This thesis presents a method for analyzing the effectivity of an organization's fuzzy front-end processes. Several technology development frameworks were considered as the foundation of the proposed analysis method. Don Clausing's "An embedded process framework for Total Technology Development" was selected as the basis for the assessment. From this framework, a behavior driven questionnaire was developed. The interview format was selected as the method for obtaining data from the organization. The interview format specified and detailed questionnaire were developed to be generally applicable to any organization. The proposed assessment method was applied to one organization within Xerox Corp. Over 800 data points were gathered, summarized, sorted and discussed. Xerox uses the TTM (Time to Market) development process. The proposed assessment method focuses on assessing the actual processes implemented within the local organization. These processes may be different than the corporate documented TTM processes. Ten senior managers at Xerox were interviewed. Data from the interviews was summarized and sorted. Weaknesses and strengths of the organization were obtained from the sorted data. Initiatives may then be developed to improve upon the weaknesses. Strengths are highlighted and communicated to foster a learning organization. An example is given demonstrating the use of Monte Carlo analysis to improve upon one of the weaknesses obtained from the Xerox assessment. Further research into web-based assessment methods is proposed. Author: Advisor: Title: Date: Thomas P. Courtney Professor Cliff Whitcomb Method for Analysis of Front-End Technology Effectivity January 29, 2001 2 I am filled with gratitude to all those who have helped me on this invigorating journey at MIT these past three years. Through all the ups and downs, so many people have given so freely of their time, energy, ideas, and spirit. I am so thankful. First, I would like to thank my dearest companion Karen for her constant love and support. Her encouragement means so much to me. A warm hug and a loving prayer from her gave me the enthusiasm and stamina to be the best I could be. My SDM friends have taught me so much. Much of my learning came from each of you. I am thrilled to have been in the presence of so many dedicated, creative, energetic, brilliant and fun individuals. I would like to thank my incredible parents, Lucille and Richard Courtney, for all the lessons in life they have taught me. Their examples of dedication to excellence, compassion for what they do, and a thirst for continuous growth has shaped me into who I have become. God is the source of my strength. My travels to MIT provided me much more than an education. My travels to MIT placed people in my path that profoundly changed my life forever. My thanks goes out to all those who participated in the interviews. Thanks to Ken Altfather, Bob Burkett, Bill Hawkins, Juan Becerra, Terry Coggeshall, Mel Croucher, Mariano Freire, Louis Isganitus, and Dale Ims. Their candid discussions greatly added to the quality of this work. Thanks to my thesis advisor, Cliff Whitcomb. Cliff helped keep my thesis properly focused and provided valuable advice that greatly improved the quality of this thesis. Last, but not least, I would like to thank my children Aaron, Robert and Andrew. Your bright smiles and witty humor helped me over many bumps in the road. I am so pleased with the growth I have seen in each of you over these past years. Your youthful spirit and playfulness helped me to learn how to let the youthful child inside of me come out and play. 3 Table of Contents Chapter 1 Introduction................................................................................................6 Background ................................ ................................................................................. 6 Objectives of this Thesis............................................................................................. 6 The Problem Statement....................................................................................................7 Research M ethod ........................................................................................................ 8 Term inology ..................................................................................................................... 9 Chapter 2 Total Technology Development Framework .................... 11 Total Technology Development Framework Selection ............................................. 11 TTD Framework Overview......................................................................................... 12 Framework Phase 1 - Technology Strategy ............................................................... 12 Framework Phase 2 - Concept Generation and Enhancement ................. 13 Framework Phase 3 - Robustness Development and Analysis ................................. 14 Framework Phase 4 - Technology Selection, Transfer, and Integration.................... 15 Comparison to Xerox Time To M arket Process ........................................................ 16 Chapter 3 How to Analyze Upstream Technology Development Effectivity.....24 Introduction.................................................................................................................... 24 Process used to create the methodology. ................................................................... 24 Applicability to other companies............................................................................... 25 Generation of behavior driven questions .................................................................... 25 Selection of Interviewees........................................................................................... 26 How interviews were conducted................................................................................ 27 Chapter 4 Data Summaries and Key Points .......................................................... 28 Introduction .................................................................................................................... 28 Discussion of Lowest scoring behaviors.................................................................... 29 Discussion of Highest scoring behaviors....................................................................36 Discussion of behaviors having high standard deviations ........................................ 40 Chapter 5 Application of Results.............................................................................43 Chapter 6 Conclusions and Recommendations ...................................................... 57 Bibliography and References......................................................................................60 A ppend ix .......................................................................................................................... 62 Appendix 1: Phase 1 Process Steps .......................................................................... 62 Appendix 2: Phase 2 Process Steps .......................................................................... 65 Appendix 3: Phase 3 Process Steps .......................................................................... 69 Appendix 4: Phase 4 Process Steps .......................................................................... 71 Appendix 5 Source Code for Install W ater Loss M onte Carlo Analysis....................72 Appendix 6. Detailed Questionairre Responses.........................................................73 Appendix 7 Questionnaire........................................................................................ 75 4 Table of Figures Figure 1 TTD Framework .............................................................................................. 12 Figure 2 Xerox Business Model..................................................................................... 16 Figure 3 TTD Bubble Portfolio..................................................................................... 18 Figure 4 Market Strategy Allocation Model .................................................................. 19 Figure 5 Newness and Complexity vs Program Configuration.................. 22 Figure 6 Total Response Distribution ........................................................................... 28 Figure 7 Market Pull vs Technology Push.................................................................... 33 Figure 8 Truncated Normal Distribution....................................................................... 44 Figure 9 Quality Loss Function..................................................................................... 45 Figure 10 Monte Carlo Analysis Example.................................................................... 46 Figure 11 Useful Ink Histogram. Monte Carlo Output ................................................ 47 Figure 12 Factory % Water Normal Distribution.............................................................. 48 Figure 13 Transport LogNormal Distribution................................................................48 Figure 14 Evaporation Rate Emperical Table Distribution............................................ 48 Figure 15 Storage LogNormal Distribution .................................................................. 49 Figure 16 Rate at Environmental Storage .................................................................... 49 Figure 17 Installation Temperature Emperical Distribution ......................................... 49 Figure 18 Humidity Normal Distribution.......................................................................50 Figure 19 Pages/Day LogNormal Distribution ............................................................. 50 Figure 20 Sensitivity to Temperature Noise ................................................................. 52 Figure 21 Sensitivity to Humidity Noise....................................................................... 53 Figure 22 Sensitivity to Pages printed per day.............................................................. 54 Figure 23 Water Evaporation Rate Sensitivity.............................................................. 55 5 Chapter 1 Introduction Background How does one prosper in this high tech world? There is great leverage in the Front-End of the product development process. This fertile time is often called the Technology development phase, or Advanced Development phase. (Clausing, 1998). This thesis will refer to this process as the Fuzzy Front-End. (Khurana, 1997). This is a time of great uncertainty. Where is the market going? Which Technologies will become dominant in the future? Where are our competitors going? How will we make money? How should we be organized? What is more appropriate, Technology Push or Market pull? Several heuristics passed down over the centuries capture the importance of this phase. "The beginning is the most important part of the work." (Plato, 4 th Century B.C.). "All the serious mistakes are made in the first day." (Robert Spinrad, 1988). This early process is often confusing and chaotic. It is an artistic time. No one has the answers. Some organizations are fortunate to have a great system's thinker. These system's thinkers have the uncanny ability to merge and meld a plethora of disjointed and often confounding issues to create a set of technology elements which when integrated bring bountiful success to the organization. Most organizations however, do not possess such an individual. Somehow, ordinary people in ordinary organizations must navigate these waters and take their best guesses. This thesis suggests a better way. Objectives of this Thesis The goal of this thesis is to develop a method for assessing the strengths and weaknesses of an organization's implemented Front-End Technology Development process. Carefully note the word 'implemented' in the prior sentence. Many companies have documented processes and procedures, yet there can be significant differences between the corporate documented processes and the processes actually implemented within an organization. This thesis focuses on those processes actually implemented and practiced within an organization. The process will generate a list of areas for improvement. Once a list of weaknesses are derived using this method, the organization can create initiates targeted at improving their actual fuzzy front end processes. Where would one start to try to improve their actual processes? This thesis provides a 6 methodology for systematically understanding the strengths and weaknesses of an organization's Fuzzy Front-End Technology Development process. It is NOT the objective of this thesis to make specific prescriptions, although many prescriptive solutions are easily inferred from the proposed method. An example showing how the results can be used to create an initiative is given. Individuals inside or outside the organization may use the methodology proposed. The main objective is to get focus on the right problems. Management consultants may find this methodology extremely beneficial to help them pinpoint dysfunctional or absent processes for their clients. Different companies use different terminology and practices, therefore, an interview format for data collection was selected. Details of the interview procedure are given. Ultimately, people and organizations will use this material to make major changes within their organizations and reap significant success in their chosen markets. The Problem Statement Introduction of immature technologies is a leading source of schedule uncertainty. (Clausing, 1994). A distinct technology development phase can greatly reduce schedule uncertainty. (Clausing, 1994). Weak technologies even induce additional problems in product development programs. (Clausing, 1994) and (Altshuller, 1984). The lack of technology robustness has to be fixed in elaborate rework cycles during the production preparation phase and thus quality and time to market become uncertain. (Clausing, 1998). Cooper and Kleinschmidt's research indicated that successful projects spent twice the amount of money and almost twice the amount in man-days in the front end than did projects that failed. (Cooper and Kleinschmidt, 1988). Hence, there are many important reasons for seeking improvement in the fuzzy front end. There are many issues to consider. Technology Push or Market Pull? Technologists have a notion of what's possible. Business planners have a notion of what will sell in the marketplace. There should always be a comfortable tension between these two fields. Often, however, they live in two separate worlds. Their languages are different. Their motivations are different. Technologists want to see their latest invention incorporated because they believe it's the best thing since sliced bread. Product planners, on the other hand, are not impressed with under the hood wizardry. They are simply looking for features to delight customers at or below competitor's prices. The proposed methodology seeks to discover 7 and illuminate the most important areas in need of improvement in the fuzzy front end. How are we to spend this important time and money? Research Method An interview/questionnaire was developed to identify areas in need of improvement in the fuzzy front end. The questionnaire was developed based upon the Total Technology Development framework proposed by Clausing. (Clausing, 1998). I will refer to Clausing's Total Technology Development process as the TTD process. Several frameworks were considered. Other frameworks considered include Xerox's Time to Market Process, Michael Frauens "Improved Selection of Technically attractive projects using knowledge management and Net Interactive tools, (Fraeuns, 2000) and Robert Wirthlin's "Best practices in user needs/requirements generation." (Wirthlin, 2000). A discussion of the selection of Clausing's TTD process is included. Interviews were conducted with ten senior managers from one organization's division. The interview consisted of 82 questions. Respondent's scores were collected for each question. Subjective comments regarding each question were also recorded. Averages and Standard Deviations were calculated for each question. (Appendix XX). The average scores were sorted. A discussion of the lowest scored questions and highest standard deviation questions is given in section XX. Once a list of weaknesses are derived using this method, the organizaton can create initiates targeted at improving their actual fuzzy front end processes. 8 Terminology The following words or phrases show up throughout the thesis. They are given to help to reader interpret the thesis. System's Thinking "A way of thinking about, and a language for describing and understanding, the forces and interrelationships that shape the behavior of systems. This discipline helps us see how to change systems more effectively, and to act more in tune with the larger professes of the natural economic world." (Peter Senge, pp. 6-7). Technology Push Technology push can lead to great technical concepts, but they can dismally fail to meet some important customer needs. (Clausing, 1994). Market Pull The opposite of Technology push is market pull. Here, there are major customer needs and technologists attempt to deliver technologies which lack robustness, leading to disastrous results. (Clausing, 1994). Technology The word Technology derives from the Greek word technologia, meaning 'systematic treatment'. (Websters, 1996) Within this thesis, technology will be defined as: 'Technology is the structured application of scientific principles and knowledge to enhance the functional performance of a system. It has an overall structure and classification. A particular technology contains a body of formalized, transmittable scientific knowledge and has distinct attributes such as specialist techniques for investigation, measurement, and application.' (Clausing, 1998). Technology Readiness 9 Technology Readiness is the point in time in a program when the launch schedule, product quality and product cost can be estimated with reasonable confidence. It represents the point in time when a corporation makes a firm commitment of resources and money to develop and deliver a product. (Xerox TTM Process). Total Technology Development Phase For this thesis, this phase includes all development activities leading up to the point of Technology Readiness. I exclude basic research from this phase. Independence Axiom An acceptable design is achieved when the design parameters and functional requirements are related in a way that a certain design parameter can be adjusted to satisfy its corresponding requirement without affecting other functional requirements. (Suh, 1990). 10 Chapter 2 Total Technology Development Framework This chapter provides an overview of several frameworks that were considered for the foundation of this thesis. The rationale for selecting the Clausing TTD framework is discussed. The remainder of this section discusses some of the major differences between these frameworks and Clausing's TTD framework. New technologies are the life-blood of thriving companies. Generating and selecting winning technologies is a very difficult process. A process is needed to generate and assign these winning technologies to products. The process of generating and selecting these winning technologies is a very complex process. A.D. Wheelon stated it nicely: (Rechtin, E. Systems Architecting): "A System is successful when the natural intersection of technology, politics, and economics are found." How does one deal with these layers of complexity? The primary method for dealing with complexity is to decompose the larger problem into smaller manageable pieces. Hence, a framework is needed to decompose these activities down to manageable pieces. Total Technology Development Framework Selection Several frameworks were considered as the basis of the questionnaire. Frameworks considered include Clausing's "An embedding process framework for Total Technology Development" (TTD), (Clausing, 1998), Xerox's Time to Market Process, Michael Frauens "Improved Selection of Technically attractive projects using knowledge management and Net Interactive tools, (Fraeuns, 2000) and Robert Wirthlin's "Best practices in user needs/requirements generation." (Wirthlin, 2000). The Framework selected is titled, "An Embedding Process Framework For Total Technology Development," by Armin P. Schulz and Professor Don P. Clausing. They completed this work at the Center for Innovation in Product Development at MIT during 1998. Members from the Center for Innovation in Product Development, MIT and the Institute of Astronautics, Technical University of Munich cooperated on this work. Professor Clausing is the Xerox Fellow in Competitive Product Development at the Center for 11 Innovation if Product Development at MIT. This work is an enhancement of Clausing's earlier work in his book Total Quality Development (Clausing, 1994). There are several advantages for selecting this framework over the others: 1. The framework is comprehensive, incorporating all-important upstream and downstream influences. 2. The framework identifies many tools and techniques and their applications. This makes the framework very practical and useful. 3. The framework is flexible. Nearly any technology can benefit from these practices. 4. The framework encompasses aspects of human behavior. 5. The framework seeks to integrate the Voice of the Market with the Voice of Technology. 6. The concepts and tools are consistent with the training incorporated within MIT's System Design and Management curriculum. TTD Framework Overview The framework is subdivided into four Phases. Phase I Integrated Technology Strategy Phase II Concept Generation and Enhancement Z> Figure Phase III Robustness Development and Analysis Phase IV Technology Selection, Transfer and Integration 1 TTD Framework 12 Framework Phase 1 - Technology Strategy Phase 1 Overview Phase 1 systematically builds a solid foundation upon which superior Concepts are generated. This phase seeks to blend the Voice to the Market with the Voice of Technology. Phase 1 activities illuminate market and technology trends. Specific tools help translate market needs into technology parameters. This parameterization is essential. The parameterization speaks to the technologists in a language they can more easily understand. This phase also sheds light on the basic technology strengths that will be required to develop the new technology. One must determine whether these strengths are present in the company, need to be developed further, or whether outside partners should be considered. The success of this phase depends heavily on interactions between the technologists and business planners. (Phase 1 steps are described in Appendix 1) Framework Phase 2 - Concept Generation and Enhancement Phase 2 Overview Phase 2 is comprised of 4 main themes. The first is System Decomposition. This phase begins with the most important functional outputs determined from Phase 1. Functions are decomposed in solution neutral space. Triz concepts are applied to select functional solutions having high ideality. The second step is system re-integration. This is accomplished using a design structure matrix to aggregate related functions. This is the step where function is transformed into the form of potential solutions. The third step is concept enhancement. This is accomplished by defining critical failure modes and critical system conflicts. This important step uses breadboard testing to assess the significance of related failure modes. The 4th step is concept evaluation and selection. The goal of this step is to select the most promising solutions in terms of superiority and pre-robustness. (Phase 2 steps are described in Appendix 2) 13 Framework Phase 3 - Robustness Development and Analysis Phase 3 Overview Once several concepts are selected from Phase 2, the concepts now pass through robustness development. This phase is composed of two major steps. The first is robustness analysis and evaluation. The second is robustness design. Robustness analysis consists of defining metrics to measure robustness, selecting appropriate signalto-noise ratios, identifying critical parameter sets, identifying noise sources, establishing orthogonal array experiments and performing the experiments. Noises should be loud enough to cause the system to fail. "Make early prototypes Fail", (Courtney, 1998). It is only though observations of failures we gain a keen awareness of the system's robustness. The second step is robustness design. This phase involves calculating signal-tonoise ratios and selecting the best parameter settings. One major concern during this phase is how to handle interactions. If certain interactions are expected, a suitable orthogonal matrix should be used to capture the interaction. Preferably, the concept selected for robustness development will obey the independence axiom. (Suh, 1990). This will lead to an independent set of control factors. Independent control factors lead to independent design factors. This will eliminate the interactions among the control factors and is crucial for applying the additive model. (Clausing, 1998). (Phase 3 steps are described in Appendix 3) 14 Framework Phase 4 - Technology Selection, Transfer, and Integration Phase 4 Overview Winning technologies must meet four general aspects: 1. They must be Superior 2. They must be Robust 3. They must be Flexible 4. They must be Mature Superiority Superiority implies the satisfaction to cost ratio is higher. Satisfaction is viewed in light of both the market and technology based requirements defined in Phase 1. Robustness and Maturity A superior technology by itself won't guarantee success. A robust technology will be able to be productized through engineering with less rework cycles. Hence, the ability to predict schedules of robust technologies is greatly enhanced. Flexible Flexible technologies are able to easily adapt to variant products. Robustness development helps to make the technology more flexible. This selection criteria is similar to concept selection discussed in Phase 2. However, there is one distinguishing difference. Concept selection and enhancement seeks to enhance the concepts rather than to rule them out. Selecting winning technologies for incorporation into product programs is more heavily constrained in order to meet all the product program requirements. (Phase 4 steps are described in Appendix 4) 15 Comparison to Xerox Time To Market Process The Xerox Business Model is shown below. Xerox Management Model Xerox Business Architecture Tim*e Ch 2. eT IntrastrUCtur6 Time To Market Core Process -~ Define MarketAttackt Dsg 'Define ein DeiePrdcProduct Producf Product Customens e osrt Dmntae Deliver Delight Teechnnlogy Figure 2 Xerox Business Model Xerox's Time to Market Core process is subdivided into 7 Phases. 1. Market and Product Strategy and Vision 2. Phase 3.1 Define Market attack Plan and Technology 3. Phase 3.2 Define Product and Deliver Technology 4. Phase 3.3 Design Product 5. Phase 3.4 Demonstrate Product 6. Phase 3.5 Deliver Product 7. Phase 3.6 Delight Customers. This process encompasses all aspects of product and technology creation, delivery, and service. This thesis focuses on only one aspect of this end to end process, that being the front end technology development process. Hence, the portions of the Xerox process that relate to this thesis are only the first three, Market and Product 16 Strategy and Vision, Phase 3.1 Define Market Attack Plan and Technology, and Phase 3.2 Define Product and Deliver Technology. There are many similarities and differences between the Xerox process and the process proposed by Clausing's TTD process. Following are descriptions of the first three phases of the Xerox process followed by a description of the similarities and differences between the Xerox TTM (Time to Market) process and the processes proposed by Clausing's TTD process. Many corporations have documented processes. However, this doesn't mean they are used throughout the company. Whether a company implements its documented processes depends heavily on the amount of training and management support. Xerox Market and Product Strategy and Vision The output of this phase is a document referred to as the MPSV (Market Portfolio Strategy Vision document.) This document describes the current state of the market, defines the strategic market opportunities, and identifies vectors of differentiation that distinguish Xerox from competitors in the mind of the customer. Technology and value chain capabilities and strategies are identified and a risk assessment is provided. The details of this process are beyond the scope of this thesis. Similarities to Clausing's TTD process Both processes use various charts and graphs to capture the essential business and technical risks and opportunities. For example, Figure 2 shows Clausing's TTD proposed Bubble Portfolio Analysis of Market segments and products. Figure 3 shows Xerox's TTM Market Strategy Allocation Model. They share many similarities. Both indicate the market or business strength of each product offering. Both also indicate a notion of the market attractiveness or market growth potential. Both indicate the predicted future movement of each product. The Clausing TTD graph goes one step further by indicating the degree of competition for each product. 17 A k " * P3 40 0 * I- * 0 Pi: Product i Size of Circle represents revenue contribution. Arrow indicates trajectory. Shading indicates degree of competition 0 Weak below Avg Avg above Avg Strong Market Strength Figure 3 TTD Bubble Portfolio Figure 4 below describes Xerox's market strategy allocation model. Note the similarity to Clausing's TTD method. 18 2 4 Weak Medium Strong Business Strength Figure 4 Market Strategy Allocation Model In addition to the charts shown above, both Clausing's TTD and Xerox's TTM process propose similar charts for plotting risk-reward trade-offs. When selecting a new technology, one must be highly cognizant of the technology risks and they must be carefully weighed with the potential return on investment. When the technology is very new or the degree of innovation is new, Utterback suggests completely separating out the technology development from the organization. (Utterbach, 1991). Clausing also suggests cycling individuals from mainstream product development teams back into advanced technology development teams. (Clausing, Total Quality Development, 1994). This can ease the difficult of NIH, (Not Invented Here) syndrome so common when other people develop technology and throw it over the wall. Lowell Steele stated the following misconception: Misconception: The power of a new technology determines its success. Reality: The infrastructure required to support it is often the determining factor. (Steele, Managing Technology, p. 62) Many of the processes in Phase 3.0 seek to identify technologies and the underlying infrastructure required to enable the technology to be successful. 19 Management must be highly cognizant of this fact. The technology by itself is useless without the supporting infrastructure. Xerox Phase 3.1M The output of this phase is the Market Attack Plan (MAP). The MAP defines the market opportunities, defines market driven standards and coherence requirements, identifies a vector of differentiation, identifies value proposition and family products, creates a life cycle management plan and again performs risk assessment. Essential to this phase it the definition of a platform strategy to support a family of products over time. The Xerox phase goes into significantly more detail than Clausing's TTD process. For example, the Xerox process requires documenting where and how customers buy, and how customers decide to buy. One common theme is the identification of a Vector of Differentiation (Xerox Terminology) and identifying Superior technologies (Clausing Terminology.). Each process seeks to identify the superior technology in customer terms. This is important to assure technologies are selected for their market strength, not just their technological strength. This can be problematic, however. Some technologies are so new it is difficult to obtain accurate market research data simply because the customer is unable to express or relate to new discontinuous technologies. In these cases, Geoffrey Moore suggests a need to identify and target niche markets initially. He refers to this as the 'Bowling Alley.' (Moore, 1995mado). Pins in the bowling alley represent specific niche markets. As a company is successful knocking down a few of the pins, the others begin to drop as they hear and learn from the earlier successes. This can be very difficult for large corporations who tend to listen only to their closest customers. Their closest customers may NOT represent the best market for the company's new technologies. The disk drive industry is a good example of how listening to your closest customers can be bad advice. (Christensen, 1997). Both the Xerox TTM process and Clausing's TTD process should look carefully at their market definitions to assure they have focused on the best market segment. 20 There are many similarities between the two processes of this phase. Market trends are plotted, customer satisfaction targets are established, new technology trends are assessed, resource requirements are identified, reuse strategy created, cost assumptions accounted, value chain partners identified, make-buy decisions assessed, QFD matrix created, Failure modes are predicted. The Xerox process incorporates a technique known as Input-Output-Constraint charts. The IOC charts clearly define relationships between the subsystem modules. Clausing's TTD process does not explicitly define IOC charts, however, he does mention defining constraints and understanding subsystem dependencies through the interrelations matrix in the QFD House of Quality. Xerox Phase 3.1 P This output of this phase are specific Product proposals, a program economic case, value chain and channels plans for the program and a team is now chartered to begin development work. The elements of this phase are very synergistic with the Clausing TTD proposal. Some highlights of this phase include technology capability demonstration for all unproven elements. This is accomplished by charting the trajectory of critical parameters (Clausing refers to them as Design Parameters). This phase does NOT guarantee the technologies are Technology Ready. Technology readiness occurs at the completion of Phase 3.2. One significant difference between Xerox' and Clausing's TTD process is that Xerox defines newness and complexity ratings to the program. Depending upon the newness and complexity, the program is categorized as short, midlength, or long. Figure 5 below defines the Xerox newness and complexity configuration matrix. 21 Platform Mid-length Long Long Major Mid-length Mid-Length Long Minor Short Mid-Length Mid-length e0 Low Medium High Complexity Figure 5 Newness and Complexity vs Program Configuration Depending upon the assigned length of the program, benchmarks are defined and used to assign a time-table to the program. The assignment of program configuration also determines the number of review cycles required. For example, the Long program requires a total of 7 reviews while the short program only requires 5 reviews. Xerox Phase 3.2 There are 5 outputs from this important stage. 1. Specifications are complete, validated with customers, and under change control. 2. Value chain and channels plans are complete. Working agreements and contracts as appropriate have been signed. 3. The integrated program plan is complete and agreed to by all parties 4. Program business case is complete, agreed by all participants. 5. Technology and value chain readiness demonstration is complete and has verified readiness of the platform architecture and platform system elements for the program using the unproven elements. 22 Everything above is also included in Clausing's TTD process. Again, both Xerox and Clausing rely heavily on the use of the QFD House of Quality to track customer requirements directly to technology requirements. There is great similarity in one particular aspect, and that is the significance and importance of Technology Readiness. Both Xerox and Clausing state no Technology should be placed into a Product Development program until it has been proven, via simulation, experimental rigs, to be at a significant level of maturity. Clausing states, "The traditional lack of emphasis on robustness creates great schedule uncertainty during production preparation. The erratic performance that is caused by lack of robustness leads to many rounds of build-test-fix rework of product design." (Clausing, Total Quality Development, p 319). Therefore, Clausing strongly recommends only selecting robust technologies to move into product programs. Therefore, Clausing emphasizes Robust Design principles (Taguchi methods) much earlier than the Xerox process. Xerox views robustness development as something which occurs after Technology Readiness. Clausing believes this is too late. The Design Structure Matrix is proposed by both Xerox and Clausing to cluster subsystems based upon their degree of interactions. The Design Structure Matrix also acts to define the informal network of communications needed within the organization. In summary, several frameworks for fuzzy front end technology development have been studied. Clausing's TTD framework was selected for its completeness, flexibility and inclusion of the latest System Engineering tools. Clausing's TTD framework provides an excellent basis upon which to build an assessment methodology. 23 Chapter 3 How to Analyze Upstream Technology Development Effectivity Introduction This chapter provides a complete discussion of the thought processes that went into creating the assessment method. One of the goals of the thesis was to select a method that would be generic enough to apply to different companies or organizations. This goal has many implications on the development of this assessment tool. These issues and their bearing on the proposed assessment method are discussed. Process used to create the methodology. First, I researched various Technology development frameworks. I was looking for a framework with great flexibility. I wanted this work to be applicable to any company. I also wanted a framework that employed the latest tools and techniques in System Engineering. Clausing's paper "An embedding process framework for total technology development" (Clausing, 1998) met these two criteria very well. After selecting a framework, I needed a method to extract data from an organization seeking to improve its fuzzy front-end processes. I considered using a written form that could be mailed to each respondent. This method has the following shortcomings. First and foremost, the language used within different companies may be different. For example, one company may call a critical parameter a design parameter. Since the questions are asking whether or not specific behaviors were present or not present, I was concerned that respondents might answer they never perform a behavior, simply because they did not recognize the word or phrase. Second, written questionnaires do not generate any dialog. Without dialog, it is difficult, if not impossible, to ascertain the motives behind why a corporate process is not being followed. Hence, a written questionnaire would have been a poor method for obtaining data. Another method I considered was a roundtable discussion with all the respondents present. This method would have generated a great deal of dialog, but has one important shortfall. In a group environment, individuals may be reluctant to openly share their thoughts and feelings, especially in the presence of upper management. Hence, this method lacked the ability to get accurate 24 feedback from each individual. The method selected was the interview. This method overcomes the limitations of the above mentioned methods. The interviews consisted of 82 behavior driven questions. (see Appendix 7). Applicabilityto other companies Can this process be used by other companies? Clausing's TTD framework includes techniques which may not be familiar to all respondents. For example, DSM (Design Structure Matrix) method was not known by any of the Xerox respondents. However, the concept of the Design Structure Matrix is easily explained during the course of the interview. The respondent is asked when they employ any related processes that achieve the same end result. For example, when creating a task list DSM, some companies use a Gantt chart to understand the dependencies between tasks. In this case, the Gantt chart provides for some elements of the task list DSM. Here's another example. Some companies may not practice the principles of Robust Design. During the interview, if it clear the respondents are unfamiliar with Robust Design principles, a short overview of the method, its applicability and the potential benefits can be discussed. During the interview, the person giving the interview must be knowledgeable about Robust Design in order to give an overview of its value. This dialog is concluded by asking the respondent if they see any utility in the method. Positive responses from the respondents are noted. The value of this dialog is that awareness of the latest System's Engineering techniques are communicated to the respondents. If the respondents favorably respond to the techniques, this becomes an excellent starting point for bringing new initiatives and techniques into the organization. During my interviews with Xerox, this pattern became obvious. For example, when discussing DSM, all the respondents said they felt the DSM process was highly valuable and that the organization should try it out on select projects. Generation of behavior driven questions The questionnaire was developed a list of behaviors. The set of questions was developed with a one to one correspondence to Clausing's TTD process. For tools unknown to the respondents, I added verbiage to make the process more generic. For 25 example, the behavior surrounding TRIZ I said: TRIZ (or related methods) of conflict elimination methodology applied. This verbiage seeks to discover if the company utilizes some other process or technique which embodies the TRIZ principles. Again, the person conducting the interviews must be well versed in each of the tools and techniques in order to generate meaningful dialog during the interviews. The 5 possible responses for each question are: N- Never R- Rarely S - Sometimes F- Frequently A- Always These words were carefully chosen to span the maximum range of behavior space. Selection of Interviewees How were the respondents selected? The Xerox organization is a light-weight matrix organization. Respondents were selected from both the product side and the functional sides of the matrix. Furthermore, individuals were selected from each of the functional management teams in order to obtain a good cross section of the entire organization. One selected individual was not a member of a product team or a member of a functional department. This individual had a staff job as Technology Strategist. The ten respondents comprised one Vice President, one Product Manager, one Technology Strategy manager, four functional department managers, and three Principal technical contributors. Selection of individuals from other types of organizations should embody these principles. 1. Select managers representing each major subsystem of the product under development. This assures good homogeneity of the responses across the organization. 2. Select managers closest to the market. These individuals may have various titles such as Product planners, Product managers, Marketing managers. 26 3. Select a Technology Strategist, if one exists in the organization. 4. Include the most senior technical individuals. These individuals may have titles such as Principle Scientist, or Principle Engineer. How interviews were conducted I contacted each respondent and set up a two hour window. We met either at my office or at the respondent's office. I began the interview by explaining why I was conducting the interview, and the goal of the interview was to generate a list of issues upon which initiatives could be generated to improve the organizations fuzzy front end. I had a blank copy of the questionnaire available for the respondent to look at. I had second copy in front of me where I kept the scores. I also jotted interesting comments next to each question. For behaviors having low scores I would ask the respondents whether or not they thought the organization could improve its fuzzy front end by increasing a particular behavior. If the respondent said yes, I made an additional annotation next to that question. (Future improvements of the questionnaire should include this directly on the form). The average length of time was 1 interview. hours per I also had additional materials with me. For example, when discussing DSM, I had an simple DSM example with me to show to the respondent if they were unfamiliar with the method. I also had materials for TRIZ, House of Quality, and Robust Design available if the respondents were unfamiliar with the methods. In summary, an interview format for data collection was developed. Specific recommendations are given to make the process as generic as possible such that the assessment method could be applicable across many companies and organizations. 27 Chapter 4 Data Summaries and Key Points Introduction This chapter provides detailed data summaries of the most relevant findings from the interviews. First, the cumulative distribution of scores are presented and discussed. This is followed by discussions of the lowest ranking scores, followed by the highest ranking scores, followed by the scores having the highest standard deviation. Figure 6 below is the cumulative distribution of all answers from all respondents. Total Response Distribution 250 200 *E 150 50 C \ Figure 6 Total Response Distribution From Figure 6 above, approximately 4 % of the responses were Never, 26% of the responses were Rarely, 32% of the responses were Sometimes, 36% of the responses were Frequently, and 2% of the responses were Always. This result clearly indicates this organization does not fully embody the behaviors/processes outlined by Clausing, since only 30% of the responses were Never or Rarely. Assuming the behaviors/processes in Clausing's model are valid and have benefit, the data indicates there is substantial room for improvement within this organizations Front-end Technology development. 28 In order to improve the front-end technology development process for this organization, the following process is employed. Step 1: Calculate the average response for each question Step 2: Sort the responses from lowest average to highest average. Step 3: Eliminate any behaviors having less than 4 responses. Step 4: Critically assess the respondents verbal responses Step 5: Discuss the value of implementing the proposed behaviors. It is beyond the scope of this thesis to discuss all the ramifications of implementing process changes within an organization. This thesis presents the first step in that process, to identify beneficial behaviors lacking within the organization. Discussion of Lowest scoring behaviors. Lowest Score Behavior #1 PhaseiStep IQ IlndexlRi 21! 61 44A 1R2 1 1R3 1R4 11 1R5 2 1R6 2 1R 3r R8 1 1R9 2 IR10 I9 1, 1 .6 Apply Design Structure Matrix to re-allocate system elements and theirfunctions into modules, that is, highly interrelatedfunctions allocated into individual modules and work teams. Oddly, none of the respondents had ever even heard of or seen a Design Structure Matrix. During the interviews, the Design Structure Matrix concept was shown to each individual. Each respondent easily grasped the concept and felt it was an excellent methodology for mapping the interdependencies within the organization. One respondent noted the Design Structure Matrix can be utilized to assess the degree of risk in making certain decisions, or the risk to other subsystems if a decision is later modified. 29 One respondent commented how some individuals are located in separate buildings and that the Design Structure Matrix could highlight those groups or individuals that should be collocated. Lowest Score Behavior #2 Phs tpQldx R1R2 1 41 11 221 21 1R3 1R4 1R5 21 21 21 1R6 1R7 1R8 1R9 IR1 0 Avg 21 11 21 11 21 11 1.71 House of Quality builtfor each technology The respondents all answered this question consistently low. Again, this is a Xerox tool that is clearly documented in the TTM guide books. Why isn't this organization using House of Quality? House of Quality has gotten a culturally bad name for itself within this organization. The reason is because most managers have had to create a House of Quality to fulfill Xerox Phase gate criteria and have not obtained commensurate value. The reason is because they are creating the House of Quality too late in the process. The House of Quality has value during the Concept Phases, not at the completion of a phase. The House of Quality can be used as a communication vehicle between marketing and engineering. The cultural stigma from being forced to create a House of Quality to fulfill the TTM policing function has made it very difficult for anyone to sincerely practice using this tool. Another barrier has been the lack of a good House of Quality software tool for the creation and maintenance of the House. This is now changing. Recently a tool has been made available from QFD Designer from Qualisoft (see www.Qualisoft.com). This tool has a good user interface and can be learned quickly. Lowest Score Behavior #3 Phase Step Q Index R1 R2 2 9 2 58 2 2 8 3 56 2 2 9 1 57 2 R3 R4 2 3 R5 3 4 3 R6 2 1 2 R7 2 R8 2 2 2 R9 R10 Avg 1 1 1.9 2.2 1 1 2.0 30 TRIZ or related methods of conflict elimination methodology applied. Inventive problem solving methods such as ARIZ are applied to break system conflicts. Conflicts are identified to be eliminated. These two questions are highly related. At least fifteen individuals in this organization went through a one week TRIZ training workshop last year. Yet, evidence of TRIZ usage is nowhere to be found! What went wrong? It appears there was no management support for TRIZ, other than financing the training. What I found most disturbing however, is that the concepts embodied within TRIZ have not yet permeated this organization. This organization focuses greatly on Compromise rather than Conflict Elimination. Conflict elimination is a way of thinking. It is especially important during the Concept generation phase. Yet, this organization has not embraced this manner of thinking. This was highly evident from comments from respondents. Overall, most respondents agreed Conflict Elimination thinking would be beneficial to this organization. Their concern was the amount of time necessary to explore the alternative paths generated. Lowest Score Behavior #4 IPhaseiStep 2 -621 IQIJndexIRl 451 31 1R2- 1R3 JR4 I- 31 2 1R5 1 1R6 - JR7 31 1R8 21 3 IR10 JAvg 1R9 -21 I 1 2.21 Great care taken to assure the architecture has minimal interrelationsbetween modules. There was greater variance to this behavior from the respondents. Two of the respondents said the organization Never did this behavior. Four of the respondents said the organization Sometimes did this behavior. This issue has caused schedule problems for this organization. The firmware development for the inkjet printer was handed off to an outside firm. The firmware touches practically every subsystem within the printer. The interrelations between firmware and other modules is complex. The information 31 transfer requirements between firmware and the modules is high. The information conduit passed through one individual who tried to keep the firmware and testing requirements in order. It didn't work. The firmware became the critical path and caused the program to slip. The quality of the firmware also was in question. The architecture simply could not handle the two directional information pathways required between the modules and the firmware. A Design Structure Matrix would have made this obvious. 32 Lowest Score Behavior #5 Phase Step 1 1 1 4 Index R 1Q R2 R3 R4 R5 R6 R7 R8 R9 R10 Avg 1 4 2 3 3 3 3 2 2 2 22.6 23 2 3 2 2 2.3 Market Pull requirements clearly defined. Technologists and ProductMarketing work well together defining the best mix of technologies and markets. This organization makes decisions primarily based on Technology Push and less on Market Pull. Respondents were asked to place an 'X' where they thought we were on the Market Pull vs Technology Push continuum. And to place an '0' where they thought we should be. The results were unanimous. Market 0 X Pull Technology Push Figure 7 Market Pull vs Technology Push All respondents felt there needed to be more Market Pull and less Technology Push in decision making. There was fear of creating technologies the market wouldn't accept. The House of Quality can act as a speaking tool between these two forces. The left side (Rows) being the 'What' (Market Pull) and the vertical colurns being the 'How' (Technology Push). One manager felt having more technical people in the marketing function could help bridge this gap. The other major difficulty occurred when partnering with companies having different views of the market. This creates great confusion and chaos, resulting in delayed decision making and sometimes reverting decisions later in time. These political hurdles must be dealt with very early. If the differences in markets being served by the partners are too vast, highly flexible technology must be developed for the partnership to survive. 33 Lowest Score Behavior #6 IPhaseiStep IQ indexIJR1 21 61 11 351 1R2 31 1R3 21 1R4 31 IR5 1R6 1R7 21 21 21 1R8 21 IRio tAvg 1R9 21 31 21 2.31 Functions decomposed into sub-functions in solution-neutralspace, that is, without regard to any design concept. The respondents, who responded with 3- Sometimes, said this behavior occurs implicitly within the minds of the engineer. It isn't an explicit, written down process as suggested by Clausing. Most respondents were unsure of the value of explicitly performing this behavior. This process is the backbone of System' s Engineering. The value resides in the ability to decompose the function down to the essential functional elements prior to deriving a concept to map function to form. When an individual prematurely establishes a design concept, the brain tends to take hold of this idea and great energy is required to move the brain onto other concepts. The value of breaking down the functions to subfunctions in design neutral space is proposed to prevent the brain from adopting premature concept selections. Lowest Score Behavior #7 IPhase-Step IQI~ndexIR1 41741 82 1R2 1R4 jR3 2 jR5 3 1R6 1R7 3 1R8 1R9 31 21 JR10 IAvg 2 2 Promisingtechnologiesnot yet readyfor a productprogram areplaced back into robustness development. Many of the respondents indicated that promising technologies that are not yet ready for product programs are typically killed, rather than placed back into Robustness development. This organization does not actively staff resources to enable promising technologies to cultivate. The organization is so fixated on getting the next product out that managers are reluctant to place any resources on longer term, potential high gain, projects. This approach is short sighted. The organization needs to achieve a balance between getting products out the door, and maintaining a steady technology stream. This 34 is very difficult to achieve, because culturally, Xerox has a tendency to throw resources at problems. Managers feel constantly in a state of fire fighting. In this mode, all resources are thrown at the problem of the week to keep the fires from raging out of control. At this point, a senior manager, or VP needs to step in, assess the total picture, and make an appropriate judgement regarding allocation of resources. Lowest Score Behavior #8 IPhase Step IQ In-dex Ri 1 1 131 131 1R2 21 1R3 41 R4 IR5 3 31 R7 R6 2 21 1R8 1R9 2 31 R10 IAvg 2 2 2.5, Business Impact of technology assessed. This question generated lots of discussion. Because there is a lack of Market pull in this organization, technologists feel unsure of the value of their technical achievements. The business cases for recent product programs had not been shared with many of the senior managers interviewed. Hence, there was a lack of trust between technical managers and business planners. Business planners need to openly share their business models with the technical community. A sound business model can give the organization the needed motivation and personal security to know their efforts will bear fruit. Lowest Score Behavior #9 JRi 3 1R2 3 1R3 4 1R4 2 IR5 2 R6 1R7 1R8 1R9 JR10 IAvgI I1ndex PhaseiStep 1 2 IQ5 18 1 4 3 2 1 2.5 Reasonable timeline estimates created. Most managers agreed that unrealistic schedules are created. It is not clear whether this is good or bad for the organization. Many managers believe that creating realistic schedules is a poor idea because it takes the pressure off the people. They argue aggressive schedules drive and motivate individuals. There is a flip side to this issue. If schedule pressure is too great, individuals begin making poor decisions. They don't take 35 the time to properly gather the appropriate data for the decision. Excessive schedule pressure over prolonged periods of time can lead to individual bum-out. This organization is approaching burn-out stage as it continues to drive multiple parallel projects, each under aggressive schedule pressure. Lowest Score Behavior #10 IPhaseiStep IQolindexiR1 2 6 31 461 1R2 21 1R3 1R4 1 31 1R5 1R6 1R7 21 21 -31 1R8 21 1R9 IR1 0IAvg 31 21 42.61 Multiple System Architectures proposed, rather than keying in early on a single solution. Rarely is the first idea the best idea. Yet, respondents said very little time is spent developing multiple concepts! Pugh concept enhancement is a method proposed by Clausing which leverages the strengths of each concept to build new concepts. This behavior is grossly deficient in this organization. Most respondents agreed the organization should foster a more active concept generation phase. Discussionof Highest scoring behaviors. Highest Score Behavior #1 Phase Step Q Index R1 R2 3 123 69 4 3 122 68 4 R3 R4 4 R5 R7 R6 R8 4 3 2 4 4 4 R9 1 R10 Avg 4 4 Stdev 43.6 0.8 44.0 0.0 Design of Experiments. Environmental noises selected. Design of Experiments. Tolerance noises selected. 36 Most respondents indicated this organization does a good job defining environmental and tolerance noises. They indicated frequent testing at high and low temperatures, high and low humidities, and high altitude testing. Tolerance noises were also selected. This organization has painfully learned from past projects what happens when a product moves into production and production processes are not able to meet the tolerances specified. Because of this prior experience, the organization has become sensitized to discovering tolerance sensitivities early in the development process. There was one individual who did not agree tolerance noises were selected. This individual felt tolerance noises were selected, however, they were selected too late in the development process. Tolerance noises were considered something that could be worked out at a later stage in the program. Also, note that several respondents did not answer this question. This was because these individuals were at a higher level in the organization and did not understand this level of detail within their organization. Is this an indication that upper management isn't aware of what people lower in the organization are doing? Are these managers out of touch with the processes being used in their organization? Should these managers know whether or not environmental and tolerance noises are applied? Upon further probing in the interview, the respondents indicated they believed these noises were being applied, they simply didn't know the details of what specific noises were being applied. Highest Score Behavior #2 R2 Phase Step Q Index R1 1 1 1 1 1 8 R3 R4 R5 R6 R7 R8 R10 Avg R9 Stdev 10 4 4 5 3 4 4 2 4 3 4 3.7 0.8 8 3 3 4 4 4 4 4 4 4 43.8 0.4 Provide a technology perspective for a decision on which technologies to focus on. Technology Trends plotted. The respondents unanimously agreed that a very thorough technology perspective is used for decision making. Technology perspectives include benchmarking of 37 competitive technologies, studying patent literature, and in-depth technology risk assessment. However, the respondents felt that decision making was too focused on the technology perspective, and too weak on the market perspective, as discussed earlier (ref.Questionnaire Index #1). This organization is too heavily focused on technology push and lacks the market pull factors in decision making. Highest Score Behavior #3 Phase Step Q Index R1 R2 R3 R4 R5 R6 R7 R8 R9 R10 Avg Stdev 2 6 1 37 4 4 4 4 3 4 4 3 3 4 3.7 0.5 2 6 6 42 4 4 4 4 5 4 3 3 4 3.9 0.6 Target values establishedfor each design parameter. Appropriate design parametersdetermined for each subfunction. This question required some clarification during the interview. Xerox refers to Clausing's design parameters as critical parameters. This Xerox organization has been trained to apply Critical parameters to the design process. Management has supported and fostered the deployment of critical parameters throughout this organization. Note the consistent high scores across the organization. This is a good indication that training and management support has been effective at instilling the critical parameter design process. One of the lower scores came from an individual who worked in an area where it has simply been too difficult to define appropriate critical parameters. Many of the design practices in this group are driven by heuristics rather than strict critical parameter determination for each function. Attempts have been made to seek out the critical parameters and link them to function, yet the technology itself has been too difficult, even for senior level scientists, to extract the relevant critical parameters. Because of the lack of scientific knowledge linking critical parameters to function, they have had to rely on lots and lots of empirical testing to assure the design functions could be met. Highest Score Behavior #4 38 MPhase1 Step11 IQI11ndexIJRl 31 31 5 1R2 4 1R3 1R44 IR5 R6 4 1R7 4 1R8 1R9 51 IRl 41 ivg3. Idev09 21 Each competitor is well characterizedin the market. The scores were very high on this question, with only one major dissenter. Charts and graphs are frequently created showing the competitive strengths of each competitor. A competitive lab has even been established. This is a room dedicated exclusively to competitive products. The room was created to raise the awareness of the competition. Individuals within the organization are encouraged to visit the room and actually use the competitive products. Individuals may also "check out" competitive products for several weeks to use at their desktops. A great deal of analytic data is also collected from the competitive products. The one dissenter commented that he believe competitive information was collected, however, he felt it was not disseminated throughout the organization, and that only select individuals actually had access to the data. This individual happened to be located in another building. It appeared this was an example of how geographic location had left this individual out the mainstream flow of information. Again, the interview format was necessary to bring out this level of understanding. A strict questionnaire would not have elucidated this important insight. Highest Score Behavior #5 R2 Phase Step Q Index R1 4 54 1 8 2 4 48 8 1 2 4 4 4 4 3 3 R8 R7 R6 R5 R4 R3 4 4 3 3 R10 R9 4 4 4 4 Avg Stdev 0.5 3.8 0.5 3.8 FailureModes predicted. Lab experiments performed to assess failure mode impacts on System Performance. These scores were consistently high. Why? When this organization first builds a prototype, they have a test called the FMIT. This stands for Failure Mode Identification Test. The name itself assures failure modes will be found! This organization has 39 delivered two generations or products from their core technology set. They have learned through experience what important failure modes are necessary to evaluate. The organization does a good job of looking at failure modes both at a subsystem level and at the system level. Discussion of behaviors having high standard deviations Highest Standard Deviation Behavior #1 IPhase Step IQ Index Ri 1R2 1R3 1R4 IR5 1R6 1R7 1 241 171 51 41 31 41 4 5 1R8 1R9 41 R1 0 Avg IStdev 2 2 3.71 1.1 Resource Requirements determined. Very interesting results. The two lowest scores came from the technical function management level, where the two highest scores came from the managers responsible for resource planning! This implies the people responsible for resource planning feel like it's being done, yet the people who carry out the plans (the functional managers) don't feel like it's done, or they don't feel it is done well. During the interview, the functional managers asked whether the question meant whether resource requirements were 'accurately' determined. They believed resources were determined, however, they felt the resource requirements were not accurate, implying they are asked to do more with less people. Highest Standard Deviation Behavior #2 Phase Step Q Index R1 R2 R3 R4 R5 R6 R7 R8 R9 R10 Avg Stdev 1 2 6 19 4 4 4 2 2 4 4 4 2 2 3.2 1.0 1 2 5 18 3 3 4 2 2 1 4 3 2 1 2.5 1.1 Timeline estimates created. Budget requirements created. Each respondent asked this question. "Do you mean 'realistic' timelines are estimated?" The cynical tone of their voices indicated they believed timeline estimates 40 are created, however, most of the respondents indicated the estimates were not realistic. For consistency sake, I appended the word realistic to this question for each interviewee. During the interview two points of view on schedule creation emerged. One point of view was that schedules should be created more realistically, that is, add more time to the proposed schedules. They felt the organization had a better chance of hitting the schedule if was created more realistically. The other point of view was to create schedules that are more aggressive. They felt the same amount of slip would occur with either schedule! They felt a certain amount of schedule pressure was necessary to keep the organization pressured to deliver. This organization hadn't directly felt the impact of schedule slips in the past, so a cultural norm exists within the organization that tolerates schedule slips. Responses to budget requirements were similar to schedule requirements. A large portion of this organization's budget is engineering labor, therefore, schedule and budget estimates are linked. Highest Standard Deviation Behavior #3 Phase Step 1 1 1R2 indexiR1 4 4 3 1R3 3 IR5 1R4 2 4 1R6 2 1R7 3 1R8 4 R10 Avg Stdev 1R9 5 3 2 1.0 3.1 Key customer satisfactionparametersare identified for each market segment. The respondents giving low scores say management gives diverse answers as to what the markets are and what will succeed in each market. Why is there so much diversity of opinion? This organization has an alliance with another company. This partner is from another country and often has different opinions of the market and what the appropriate satisfaction parameters are for those markets. Having a partner with a different view of the market has destabilized this organization. They lack a consistent focus on which market they are attacking. This has created confusion as evidenced by the wide range of responses from Rarely to Always. In summary, 820 data points have been summarized, sorted and discussed. From these summaries, the organization has obtained a list of it's weaknesses and strengths. 41 These lists provide the bedrock upon which initiatives can be built to improve this organization's fuzzy front end technology development processes. 42 Chapter 5 Application of Results The previous chapter brought forward weaknesses within one organizations fuzzy front end processes. Now that a list of weaknesses has been created, initiatives can be put in place to create improvements. This chapter takes a look at one weakness found and proposes a method and technique to improve the shortfall. Question #13, 'Business impact of technology assessed', was one of the lower scoring questions. The organization lacked a method for translating technology variations to implications within the field. Monte Carlo Analysis is proposed as a technique this organization could incorporate to translate real world issues into business results. This section presents a Monte Carlo Analysis applied to the life cycle of an inkjet ink tank. Ink tanks represent the primary source of profits for inkjet printer vendors. Customer usage patterns determine how often they will purchase new ink tanks. Also, ink evaporates from ink tanks. Excessive evaporation of the water from the ink will cause the remaining ink in the ink tank to go bad. The Monte Carlo Analysis considers all these factors to obtain key information for business planners. Phase 3 Robustness Development and Analysis is lacking one important step. The proposed steps do not include a method for determining the system impact of normally distributed noises. Phase 3 Robustness Development and Analysis consists of a stepwise sequence for selecting appropriate control parameters to achieve the intended system functions AND to achieve the least sensitivity to noise factors. After optimum control factors are found it is desirable to determine the system sensitivity to the noise factors. However, these noises are almost always normally distributed and their contributions to system performance can not be assessed with a very high degree of accuracy using the proposed steps. Monte Carlo Analysis is proposed as an additional step to Phase 3. Monte Carlo Analysis will aid the system designer in understanding and quantifying the impact of noise distributions on system performance. The noise distributions may be of any shape. Most noises are normally distributed. However, some distributions are truncated normal 43 distributions. Truncated normal distributions occur when parts are screened on the production floor. Figure 8 below is an example of a truncated normal distribution. Inkjet Drop Mass Distribution 0.25 0.2 0.15 o I- 0.1 0.05 0 25 30 35 40 45 Drop Mass Figure 8 Truncated Normal Distribution In the above example, the production process is incapable of producing parts with a nominal drop mass of 35 ngrams and a maximum drop mass of 38 ngrams. Printheads having a drop mass greater than 38 ngrams place too much ink on the page and creates two failure modes. One, excess ink causes the ink drying time to increase, causing smearing of ink on the customer's hands. Second, excess ink causes ink to bleed through to the back side. Thus, production must suffer scrap costs. One must carefully weigh the scrap costs versus the cost of shipping an inferior product to the customer. This can be quantified integrating Phadke's Quality loss function. (Phadke,1989). 44 A typical Quality Loss function is shown in Figure 9 below. I Quality Loss Function 120 100 80 e + Seriesi 60 40 20 0 25 30 35 40 45 Drop Mass Figure 9 Quality Loss Function The Quality Loss Function relates the Total Cost to society. This may include service costs, market loss due to word of mouth from irate customers, customer down time costs, etc. This function can be extremely difficult to quantify. When entering a new market with excellent competitors, managers are very cautious about shipping inferior products, fearing repercussions in the marketplace. Hence, it is all too common for products to be launched into the market from truncated normal distributions. Robust design practices as described by Clausing, (Clausing, 1993), Phadke (Phadke, 1989), and Taguchi (Taguchi, 1993), seek to reduce the sensitivity to noise factors to allow for wider tolerances. Therefore, Robust design practices are essential to assure products are producible with acceptable tolerance latitude. Monte Carlo Analysis is best explained by an example. Consider the following system. The system is an inkjet inktank filled with color ink. The ink is comprised mostly of water (90%). During the life of an inktank, ink will evaporate from the inktank. If the amount of evaporation of water is too great, the concentration of the dye and co-solvents in the ink will increase. Printhead operation will become poor when the increase in concentration of non-aqueous components exceeds a certain limit. This 45 results in poor print quality which customers would judge as unacceptable. Therefore, the water concentration in the ink must remain above a certain threshold. Extend Software from Image That Inc. was used to create the Monte Carlo Analysis described below. Refer to Figure 10 below. OUTPUT %Useful Ink Factory% Water Wate rout Rate Transp or Transport Day Water ss Storage Water oss Rate at Env iron Inst Wat Install Pgs/Day S S te Storage Days A =A (Tem ,rH) Temp. Temp Install Te mp. Rate i xP rHRate at Environ rH Pages/Day printed Install rH Rate Figure 10 Monte Carlo Analysis Example Overview of Model An inktank ships with a fixed volume of ink. It is desirable that 100% of the ink in the tank remain at an acceptable water concentration over the life of the ink tank. The 46 simulation is set to simulate the life of 1000 ink tanks. The Monte Carlo analysis will calculate the percent of useful ink for each ink tank and display the results as a histogram. Each of the 1000 inktank simulations will be different and will experience different noise conditions according to the normal distribution noise factors. Figure 11 below is a sample histogram output of useful ink percent for 1000 ink tanks. Useful Ink Histogram 322 281.75 241.5 201.25 161 120.75 80.5 40.25 0 - Data Data 0.25 - 0.5 Usef ul Ink percent Data 0.75 1 Data Figure 11 Useful Ink Histogram. Monte Carlo Output The peak at 100% indicates 322 out of 1000 ink tanks had an acceptable level of evaporation over the installed life of those ink tanks. The bar at Useful ink = 50% and a count of 80 indicates that 80 out of 1000 tanks only had 50% useful ink over the life of those tanks. In this case, we expect those 80 customers would experience some print quality shortfalls after their tanks become half empty. The bar at Useful ink = 27% and a count of 10 indicates that 10 out of 1000 tanks only had 27% useful ink over the life of those tanks. In this case, those 10 customers would expect to see print quality defects after only 27% of the ink was used from their inktanks. This is clearly unacceptable. 47 What are the sources of noise in the Monte Carlo Analysis? There are 8 noise factors in the model. Different types of noise distributions are used. U Noise #1: Factory %Water. 1 2 3 Factory% Water This is a Normal Distributionof Distribution Plottr 6. the percent of water by weight in 4.1 the ink at the time the tanks are 3. ____ 0.565 Member Value 0.5475 0 .53 -smm% filled with ink. It has a very tight .I 1. 3 0.6 0.5825 distribution. Members Figure 12 Factory % Water Normal Distribution Noise #2: Transport Days in Truck Transport Days This is a LogNormal Distribution. Distribution Plotter 9.28 c-I This distribution represents the 6.9637 5 amount of time the tanks spend in 4.642 2.3212 5 transit in shipping. 2.2 63744 -mmm% 7.606081 12.94842 Member Value 23.63309 18.29075 Members Figure 13 Transport LogNormal Distribution Noise #3: Evaporation Rate in transit distribution ,--.::---.-.--. ---. -.- - , -... - -...--- Rate at Env iron . 19bb1b' This is a Empirical Table - 14713411 986896.7 - Distribution.The table is based 502452.4 18008.07 0.00 019 a 0.0009675 1mm inma 0.0025225 0.001745 Bin Range upon experimental rate loss lab 0.0033 experimental data. Den sity Figure 14 Evaporation Rate Emperical Table Distribution 48 Noise #4: Storage Days Distribution 17.655 .-m.-- Storage Days Plotter Distribution ~-~ p~ -~- This is a Log Normal 13.24125 Distribution.The distribution 8.8275 indicates the number of days a 4.41375, . tank may sit idle in a warehouse. 1739.258 1307.781 876.3026 444.8246 13.34662 Member Value % Members Figure 15 Storage LogNormal Distribution Noise #5: Rate at Environment. Storage Rate at Env iron - F . I . . P I . - 193939k- This is a EmpiricalTable 1456034 Distribution.The table is based 972673.2[-- upon experimental rate loss lab experimental data. 489312.8[ 5952.38100 0.00019 0.0009675 0.0033 0.0025225 0.001745 Bin Range Density Figure 16 Rate at Environmental Storage Noise #6: Installation Temperature Install Temp. 0.015 This is a EmpiricalTable 0.01127638 I ~KJ 0.007552761 Distribution.The table is an I estimate of the temperature 0.003829142- -' 0.0001055223 5 . Density 15 25 Bin Range J ~ 35 mm 45 distribution at the customer's site in deg C. Figure 17 Installation Temperature Emperical Distribution Noise #7: Installation Humidity 49 Install rH This is a Normal Distribution. Distribution Plotter 6.7 The distribution is an estimate of 5.032 5 the humidity distribution at the 3.35 1.677 5- customer's site. 0.04890791 0.6867333 0.8993418 % Members Figure 18 Humidity Normal Distribution Noise #8: Installation Payes/dav Install Pgs/Day 677 Distribution Plotter 8. 6.47 25 4.3 15 Distribution.The distribution -- -- This is a Log Normal indicates the number of pages a - - 0.4741248 Member Value 0.2615164 2.15 75 customer is likely to print per day. 2.363713 0.8969545 - 3.830471 5.29723 6.763988 Member Value % Members Figure 19 Pages/Day LogNormal Distribution There are three Calculation Blocks representing the three different modes of evaporation experienced by the tanks over their life cycle. Calculation Block #1: Transport Water Loss E Water K Days Rate WaterOut Transport Water Loss During each of the 1000 tank simulations, this block receives WaterIn, Days and Rate data points from the three Noise inputs. The block calculates the new water concentration and passes that information to the next block. 50 WaterOut=((GramsInkInTank*Watern)(Rateln*Daysln))/( Gramslnk~nTank -(RateIn*DaysIn)); Calculation Block #2: StorageWater Loss During each of the 1000 tank simulations, this block receives WaterIn data from the previous block, Days and Rate data points from the other two Noise inputs. The block calculates the new water concentration and passes Days Rate WaerOut I that information to the next block. WaterOut=((GramsInkInTank*Watern)- Storage (Rateln*Daysln))/( GramslnklnTank -(RateIn*DaysIn)); Water Loss Calculation Block #3: StorageWater Loss This block does more than the other blocks. Depending C waterin upon the customer's usage rate (Pages/Day), this block WaterOut pete calculates the estimated days to empty. It then calculates the water loss each day and determines whether the ink is 7 C page/Day n %sfu Daystill J good ink or bad ink. Appendix 5 lists the source code for this block. Installed Water Loss Auxilliary Calculation Block: Rate = f(Temperature and Humidity) This is a very important block. This block calculates the Rate =f(Temp,rH) Temp EU rate of evaporation as a function of temperature and humidity. k=6.3501; Rate 0 I.Pv = 0.098* exp(0.0603*TempIn); RateOut=k *( 0.9 - rHIn ) * ( Pv / (Templn+273.0)); 51 Sensitivity to Noise Factors. Once the Monte Carlo simulation is built, sensitivity to noise factors can easily be performed. This is in contrast to Robust design methodologies which can not predict the system response to real world distributions such as those described in the example above. Sensitivity Analysis #1: Temperature Noise impact on System Response. Plotter XY * Ma. 0.9 846597 0.9 193193 0. 853979 4 a 0.7 886387 a 0.7 232984 0. 657958 Mot 0.5 926177272774 nr .i 0.4619371 0.3965967 ---- -A l N 0.3312564 0.2659161 - % * a -no a a 0.2005758 0.1352354 0.06989512 0.004554796 5.757476 ~ *. 15.54281 * -- 25.32814 Temperature (C) ' 35.11347 ___ 44.8988 Figure 20 Sensitivity to Temperature Noise There is a strong dependence of temperature on Useful Ink Percentage. 52 Sensitivity Analysis #2: Humidity Noise impact on System Response. Plotter XY 1.05 0. 9848369 10 --- u~t~l~ now NON NIMV ass( x am IF - Post i I IC 0- N.%jL J a - 0. 8545106 s"o 0. 7893475 -- - 0. 7241844 -- * 0. 9196737 U' U a -- aa M * ...... 0. 6590212 0. 5938581- aW - a 0 .528695 0. 4635318 1- 0 ~ .. .n o a -a. - 0. 3983687 a 0. 3332056 - 0. 2680424 0. 1377162 - U 0.0 7255306 nnf7q' QOgW 0.2083549 4- - - I I 0.3116003 0.4148457 - - a * I 0.5180911 rH I - 0. 2028793 I 0.6213365 I 0.724582 I 0.8278274 Figure 21 Sensitivity to Humidity Noise There is a strong dependence of Humidity on Useful Ink Percentage, but not as strong as temperature. 53 Sensitivity Analysis #3: Pages printed per day Noise impact on System Response. Plotter XY 1.05 0.9847014 I'm U I I" 0.7888055 - Eu0 Ito 0.7235069 F m an writ, , a I quo U ma %"U Ib n a"BW as 0.3317152 0.26641657 -0.2011179 0.1358193I- qL VI U U 10i a Im a 0a MEN a - -0 I a 'mm MEN a 0 s - U p. I. U 0.4623124 0 aU - - U - m16U *-- 0.5929096 5-- WO- WE - 0 0 a IV- IIII I of ml 0.6582083 - - - - --- - - - 0.9606214 -- ~^- I 0.07052067 0 00522205 U - U U p~iI , U 0.8541041 0.3970138 I -w 0mm a 0.9194028 0.527611 1-- I 1.609049 2.257478 2.905906 Pages/Day Printed 3.554334 4.202762 4.85119 Figure 22 Sensitivity to Pages printed per day Useful ink percentage is function of pages printed per day up to about 4 pages per day. Above four pages per day there is no sensitivity to pages printed per day. 54 Sensitivity Analysis #4: Temperature Noise impact on System Response Plotter XY 1.05 r 0. 9849293 N 0. 9198586 0. 7897172 -N 0. 7246465- No 0 a 0. 6595758 - ~- 0. 5945051 - as 0. 8547879 ~ . I a 0. 5294344 0. 4643637 a ~.399293 W0 0. 3342223 0. 2691516 a 'iEU a a ~ - 0. 2040809 0. 1390102 - [it ~ N 0.0 7393953 0.0 0.0007690582 0.00282206 0.0 08981064 0.006928063 0.004875061 WaterEv apRate grams/day 0.01103407 0.01308707 Figure 23 Water Evaporation Rate Sensitivity The water evaporation rate sensitivity is very strong, as one would expect. Usefulness of Monte Carlo Simulation The example above shows how Monte Carlo Analysis can be utilized to more fully understand how real world distributions affect customer satisfaction. The model allows the organization to understand the business impacts associated with technology and market place variations. Monte Carlo Analysis is a practical tool. It has an intuitive appeal. It allows modeling of almost any input distribution. The software tool Extend, from Image That, Inc. is quite simple and easy to use. It has a complete library of the most common tasks. In addition, they provide tools for creating your own blocks, such as the calculation blocks shown in the above example. Understanding the business impact of technology requires a method for systematically translating technology inputs into market place outputs which directly 55 impact the business results. Monte Carlo analysis can be a useful tool for systematically mapping technology inputs and their sources of variation into business results. 56 Chapter 6 Conclusions and Recommendations An appropriate framework for total technology development was selected from the literature. A questionnaire was developed from this framework. The questionnaire was developed in a manner to make it generically applicable. The questionnaire was administered to ten individuals within one organization using an interview format. 820 data points were extracted from the interviews. The data was summarized, sorted, and discussed. The summarized data provided a list of the strengths and weaknesses within this organization. From the weaknesses, this organization can now begin to develop initiatives to improve it's fuzzy front end. An example is provided showing how Monte Carlo Analysis could be used to improve upon one of the lowest scoring questions. Many other conclusions and interesting observations were obtained during the course of this work. It is important to understand the difference between a corporation's documented processes and the processes actually implemented within an organization. Studying the corporation's documented processes would have been of limited value, because I found many discrepancies between the documented processes and the processes actually implemented at a local level. This is true of many organizations. The military has an acquisition process, however, implementation to the process is quite difficult. Why is this important to understand? When a company is faced with great competitive pressures, management may look to modify its processes in hopes of reaping significant returns. Management may look at the documented processes and try to build upon what is already in place. Yet, during the course of this work, I came to understand there can be great differences in what is actually implemented. Management first needs to determine what people are actually doing. For example, Xerox lists the Design Structure Matrix as one of the TTM tools. However, not a single individual I interviewed has even heard of the Design Structure Matrix. This strongly implies the need to understand what is happening locally before even beginning to overhaul a corporate process. Reacquainting the organization with existing corporate methods may be one way to improve upon defined weaknesses. For example, in the Xerox case the House of Quality is rarely used in this one organization, yet it is a standard corporate method. The 57 interview may create a new awareness of the value of the House of Quality as a communications vehicle, rather than documentation required to pass a phase gate. The interview is key to understanding the cultural dimensions associated with negative feelings toward a particular tool or technique. Given the cultural attitudes toward the House of Quality, I suggest locating someone new in the organization to act as a niche House of Quality activator. Once others see the value to be extracted, perhaps the tool will be pulled from the rest of the organization rather than pushed upon them. How applicable is this assessment method to other companies? Would this assessment method be effective within other companies? What difficulties might arise? I think there are two major difficulties which may limit the effectiveness of the proposed assessment method. First, language or nomenclature differences may hamper clear communication. If this occurs, additional time may be required for each interview. Second, if the organization already has a deeply engrained process, or the general feeling is that change is somehow bad, there may be negative reactions from management to 'fix' the organization. In this case, it is essential to find an upper management supporter. This supporter should have enough visibility and respect within the organization to foster an environment conducive to change. What difficulties were encountered during the interviews? How can they be improved? Several questions were duplicates of earlier questions, causing confusion. For example, the behavior 'Timeline estimates created', and 'Schedules estimated' are essentially the same. The subtle difference is that they refer to slightly different phases of the TTD process. The subtle distinction was difficult to clarify. Having knowledge of the companies process may make it easier to relate the questions to the respondents in their language. Therefore, it would be beneficial for the interviewer to understand the companies basic process prior to the interviews such that the interviewer could more easily relate the questions to the respondents. Another improvement to the process would be to add a check box next to each question. This check box would be checked when the respondent felt strongly the behavior was lacking and was important to the organization. Adding this to the interview sheet and data analysis would provide a list of 'low hanging fruit.' Low hanging fruit are 58 those actions that are easy to implement and typically yield quick improvements. When a group of managers all agree a behavior is lacking and that it is important, it will be much easier to get support for this initiative. A web-based questionnaire may be a viable method to perform an initial scoping of the issues facing the organization. However, this must be approached very cautiously, for many of the problems cited earlier regarding a written survey. A web-based questionnaire may need to be tailored to each organization to use the nomenclature consistent with the organization under study. Otherwise, the data obtained could be worthless! The web-based method could include 'education' buttons. For example, if the respondent was unfamiliar with DSM, a button could be pressed to take the respondent to an overview of the DSM method. Thus, the web-based method could act to educate as well as gather data. Low scores from the web-based survey should be followed up with interviews. This is critical. The interview is essential for understanding the social, political, and cultural dimensions to each behavior. High scores could be used to establish role model divisions or individuals. Identifying role models is an important element of learning organizations. In summary, the objectives of this thesis have been met. Recommendations to continue this work should focus on selecting and implementing initiatives to improve the defined weaknesses. I strongly feel the fuzzy front-end is a critical phase for high tech companies to master. I believe application of the techniques and methods proposed in this thesis will enable companies to move one step closer to realizing significant rewards in the marketplace. 59 Bibliography and References Altshuller, G. S. (1984), 'Creativity as an Exact Science', New York, Gordan & Breach Science Publishers. Christensen, C. (1997), 'The Innovator's Dilemma', New York, NY, HarperCollins Publisher. Clausing, D. (1993), 'Total Quality Development', New York, NY, ASME Press. Clausing, D. P. and Schulz, A (1998), 'An embedding process framework for total technology Development', Cambridge, MA, Massachusetts Institute of Technology, Center for Innovation in Product Development. Cooper, R. G. and E. J. Kleinschmidt (1988), 'Resource-Allocation in the New Product Process', Industrial Marketing Management 17(3): 249-262. Courtney, T. (1998), 'System's Architecture Principles Journal', Cambridge, MA, System's Architecture Class, Massachusetts Institute of Technology. Frauens, M. (2000), 'Improved Selection of Technically Attractive Projects Using Knowledge Management and Net Interactive Tools', Cambridge, MA, Massachusetts Institute of Technology. Hammer, M. and Champy, J. (1993), 'Reengineering the Corporation', New York, NY, HarperCollins Publisher. Khurana, A., and S. R. Rosenthal (Winter, 1997), "Integrating the Fuzzy Front End of New Product Development", Sloan Management Review. Moore, G. (1995), 'Inside the Tornado', New York, NY, HarperCollins Publisher. Phadke, M. S. (1989), 'Quality Engineering using Robust Design', New Jersey, Prentice Hall. Rechtin, E. and Maier, M. (1997), 'The Art of Systems Architecting', Boca Raton, FL, CRC Press. Senge, P. M. and Kleiner, A. and Roberts, C. and Ross, R. and Smith, B. (1994), 'The Fifth Discipline: The Art and Practice fo the Learning Organization', New York, NY, Bantam Doubleday Dell Publishing Group. Steele, L. W. (1989), 'Managing Technology', New York, NY, McGraw-Hill Book Company. Suh, N. P. (1990), 'The Principles of Design', New York, Oxford University Press. 60 Taguchi, G. (1993), 'Taguchi on Robust Design', ASME Press. Utterback, J. (1996), 'Mastering the Dynamics of Innovation', Cambridge, MA, Harvard Business School Press. 'Webster's New Universal unabridged Dictionary', (1996), New York, Barnes and Nobles Books. Wheelon, A. D. (1986), 'Collection of Student Heuristics in Systems Architecting', Los Angeles, CA, USC School of Engineering. Wirthlin, J. R. (2000), 'Best Practices in User Needs/Requirements Generation', Cambridge, MA, Massachusetts Institute of Technology. 61 Appendix Appendix 1: Phase 1 Process Steps Step la - Market and Product Assessment Inputs Market and product program data Outputs Significance of market segments and product programs Key Customer satisfaction parameters sets. Decisions: none Step lb - Technology and Technological Capability Assessment Inputs Technology and technological capabilities Key customer satisfaction parameter sets Outputs Significance of Technologies for each product Significance of Technological Capabilities and Technologies Decisions: none Step ic - Trend and Impact Assessment Inputs Market pull parameter set Technology push parameter set Outputs Market based trend analysis 62 Technology based trend analysis Gap Analysis Impact Analysis Decisions: none Step 2 - Develop Integrated Technology Strategy Inputs Significance of each market and product Significance of each technology and technological capability Gaps among market based trend and technology based trend Technological and economical impact of each technology Required target values for key parameter sets, derived from a market and technology perspective Available resources, budget, timeline Outputs Integrated strategy map Separate key parameter set for each product driven by market pull and technology push. Decisions Selection of technologies to be developed Selection of technological capabilities to be strengthened and pursed Selection of products programs to implement technologies Step 3 - Strategy Review Step 4 - Definition of needs for next generation technologies Inputs Integrated Strategy Map Market based trend models 63 Outputs Strategic House of Quality for each technology System Constraints and environment for each technology Selection Criteria Decisions Set target values for key parameter sets. Voice of Market Voice of Technology Step 5 - Review and Confirm Needs 64 Appendix 2: Phase 2 Process Steps Step 6a - Develop Functional Structure Inputs Parent Function and Criteria Outputs Subfunctions and subcriteria Subfunctional interrelations Decisions: none Step 6b - Deploy Current Function level to Alternative Solution Concepts Inputs Subfunctions and criteria Outputs Complete set of system structure matrices for each functional level and each alternative solution concept. System structure tree for each alternative solution concept. Decisions Selection of alternative solution concepts on each functional level. Step 6c - Deploy Initial Concept to System Architecture Inputs Set of system structure matrices for each alternative solution concept System structure tree for each alternative solution concept Outputs System architectures for each alternative solution concept 65 Hardware concept Decisions Most appropriate architecture for each alternative solution concept. Step 7a,b - Concept Evaluation and Selection Step 8a - Concept Failure and Conflict Analysis Inputs Alternative system structure matrices Alternative system structure trees Atlernative system architectures Alternative hardware concepts Outputs Failure modes and interrelations Critcal parameter sets Critical System functions System Conflicts Decisions None. Step 8b - Laboratory Work Inputs Failure modes and interrelations Critical parameter sets Critical System Functions System Conflicts Breadboard Model Outputs Significance of Failure modes 66 Significance of System Conflicts Decisions Select system conflicts to be eliminated Select failure modes to be reduced. Step 9 - Enhance Concepts Inputs Failure modes and their significance Critical parameter sets and their impact on system conflicts System conflicts and their significance System structure matrices System structure trees System architecture Hardware concept Outputs Enhanced alternative system architectures Enhanced alternative system structure trees Enhanced alternative hardware concepts Critical parameter setting and critical system functions. Decisions Select functions to be changed, eliminated, introduced Select solution concept to be changed, eliminated, introduced Step 10 - Concept Evaluation and Selection Inputs Alternative system structure matrices Alternative system structure trees Alternative system architectures Alternative hardware concepts 67 Outputs Superior and pre-robust concepts Decisions Select concepts to be proceeded into the next step or phase 68 Appendix 3: Phase 3 Process Steps Step 11 a - Define Functional Metric to Measure Robustness Selection of the proper functional metric is critical. The metric should capture the essential physics of the system function, incorporate the effects of noises, and should be based on the quality loss function concept. Maximizing the signal to noise ratio is equivalent to minimizing the sensitivity of the system to noises. (Taguchi, 1993). Step 1 lb - Determine and Select Critical Parameter sets Critical parameter are those parameters which when satisfied in the system will deliver the intended function. They are meant to be scientific quantities such as force or pressure as opposed to geometric quantities such as length or width. Many of these parameters were made self evident during Failure mode analysis from the prior concept selection phase. Step 12a - Design of Experiments 1- Select Alternatives for Evaluation Orthogonal arrays are utilized to structure an efficient experimental process for spanning the space identified by the control factors. Orthogonal arrays are balanced, meaning the equally distribute the effects of each control parameter equally against the other control parameters. Great care must be taken to manage interactions between control factors. Step 12b - Design of Experiments 2- Determine, Select and Impose Appropriate Noises All systems are subjected to a wide array of noises. Robust systems continue to operate properly even in the presence of these noises. Noises can be external such as temperature and humidity. Noises can also be internal such as tolerances on critical dimensions. Step 13 - Perform Experiments - Evaluate Performance of Selected Alternatives Step 14a,b - Calculate Signal-to-Noise Ratios and Select Best Parameter Settings Step 15 - Evaluate and Confirm Overall System Robust Performance 69 70 Appendix 4: Phase 4 Process Steps Step 16a - Criteria for Winning Technologies Four criteria are evaluated. Superiority, Robustness, Flexibility and Maturity. Superiority implies the satisfaction-to-cost ratio of the new technology is higher. Satisfaction is measured with respect to the market and technology based requirements defined during Phase 1. Robustness or maturity implies the technology can be developed on a schedule with reasonable accuracy and that all supporting manufacturing technologies readily available. Robustness or maturity also imply all legal or patent issues have been cleared through legal counsel. Flexibility implies the technology is adaptable to a wide range of applications. Flexibility links to Robustness. Robust designs tend to be less sensitive to variation, hence, are more easily adaptable to new environments. Step 16b - Defining metrics to evaluate Selection Criteria Criteria have been developed in the concept generation and enhancement phase. After Robustness optimization these criteria may have changed based on the learning during that phase. Step 17 - Selecting Winning Technologies Three outcomes are possible. One, the technology is selected for transfer into a product program, having met the requirements for Superiority, Robustness, Flexibility and Maturity. Two, the technology is rejected because of a lack in Superiority, Robustness, Flexibility or Maturity. Third, the technology is placed back into robustness development. This occurs when the technology is found to be superior, but lacks the necessary robustness and maturity to be placed into a product program on a reliable schedule. 71 Appendix 5 Source Code for Install Water Loss Monte Carlo Analysis daystoEmpty = int(Globall/Pages-dayIn); InkVolume = GlobalO; WaterVolume = WaterIn*InkVolume; PagestoFailOut = 350; J= 1; for i= 1 to days-toEmpty { WaterVolume = WaterVolume-RateIn(WaterVolume*Pages-dayin*0.0189/InkVolume); // second term is water printed each day InkVolume=InkVolume-RateIn-(Pages-dayIn*0.0189);!!- terms are water evap grams and ink printed grams per day if ((WaterVolume/InkVolume) > .5385) {J=J+ 1; }I/ add one good ink day. if (WaterVolume <0) { WaterVolume = 0; } }/I for i = 1 to days_to_Empty WaterOut=WaterVolume/InkVolume; WaterVolumeOut=WaterVolume; InkVolumeOut=InkVolume; PagestoFailOut = J * Pages-dayIn; UsefulInkOut = Pagesjto_FailOut / Global 1; DaysToFailOut=J; 72 Appendix 6. Detailed Questionairre Responses 1- Never, 2-rarely, 3-Sometimes, 4-Fre uently, 5-Always R2 Phase Step Q Index R1 4 1 1 1 1 2 2 1 12 5 1 1 3 3 14 4 3 1 4 5 5 1 1 1 6 6 4 7 3 1 17 3 18 8 1 9 4 1 9 10 4 1 1 10 111 11 1 4 1 112 12 13 2 1 113 4 1 14 1 2 3 1 22 15 5 23 16 1 17 5 1 2 2 5 18 3 1 19 4 1 26 1 2 7 20 31 21 1 2 1 4 1 22 4 2 23 1 1 4 3 24 1 4 1 44 25 26 4 1 4 5 4 1 4 6 27 28 1 1 47 29 1 48 30 1 51 5 2 31 4 1 32 2 1 53 4 1 54 33 34 3 1 5 5 2 6 1 35 3 2 6 2 36 3 6 1 37 4 2 38 4 2 6 2 2 6 3 39 4 6 4 40 2 2 4 4 3 2 2 3 3 4 4 4 3 4 3 2 2 4 3 4 4 2 2 R3 3 2 4 2 4 3 4 4 5 5 3 3 3 4 3 4 4 4 2 3 3 3 4 3 3 3 4 3 4 4 3 4 2 4 3 3 4 4 2 2 3 2 3 2 4 4 4 4 4 2 3 4 4 3 3 4 4 4 2 2 3 2 4 2 3 3 3 2 4 1 2 2 4 3 1 2 3 3 4 4 2 4 4 2 2 3 4 4 4 3 4 4 3 3 1 2 3 4 4 4 2 3 4 4 3 2 R9 R8 3 2 4 3 2 2 3 4 4 4 3 2 2 4 2 4 5 1 4 3 2 4 4 R7 R6 3 3 4 2 3 2 4 4 4 4 4 4 2 3 4 4 4 2 2 3 4 4 4 R5 R4 3 3 3 4 4 4 4 3 3 4 3 2 4 4 2 2 2 2 4 2 2 4 2 2 4 3 4 4 3 4 2 2 2 21 3 5 5 3 4 5 4 2 4 2 5 3 2 3 2 4 3 4 3 3 3 4 4 4 2 2 3 3 3 1 R10 Avg 2 2.6 3 2.7 2 3.9 2 3.1 33.0 2 2.9 2 3.4 4 3.8 3 3.5 4 3.7 3 1 2.8 2 4 3.5 2 2.5 2 2 3 3.1 3 3 2.8 4 4 3.6 2 3.7 2 1 2.5 2 2 2 3.2 3.3 3.4 3 2 1 1.7 2.3 3.5 2 4 3.3 3.8 4 4 4 4 3.7 2.0 3.8 4 3.5 2.8 4 1 3.0 3.7 2 2 4 3 3 3 2 4 3 3 3.4 3 3 3 2 2 2 4 4 4 3 2.3 3.2 3.7 3.4 2.9 1.7 73 Appendix 6 continued. Detailed Questionnaire Responses 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 3 3 3 3 3 3 3 3 3 3 3 3 4 4 4 4 4 4 6 6 6 6 6 6 7 8 8 8 8 8 8 8 8 8 9 9 9 10 10 10 11 11 11 12 12 12 12 12 13 14 14 15 16 16 16 16 17 17 417 4 17 5 6 7 1 2 3 1 1 2 3 4 5 6 1 2 3 1 2 3 1 2 3 1 2 1 1 1 2 3 4 1 1 2 1 1 2 3 4 1 2 3 4 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 2 4 4 2 2 2 4 4 1 3 3 4 3 1 3 2 4 2 2 1 2 2 5 2 3 3 3 2 4 2 1 2 2 4 4 2 3 2 3 3 3 3 2 3 3 3 2 2 2 2 4 3 2 3 2 3 3 2 4 2 3 3 2 2 2 3 2 2 3 3 3 2 3 1 3 3 2 3 3 2 3 3 1 2 2 1 1 4 3 2 4 3 4 4 2 4 4 4 3 2 2 2 3 2 _ 2 4 3 3 3 3 4 3 4 41 4 4 4 4 4 3 3 4 3 3 3 3 4 1 1 3 4 _ 2 4 3 3 4 3 3 4 4 4 3 3 4 4 3 3 4 3 3 3 4 4 3 4 3 3 4 2 4 3 4 4 3 3 3 3 2 3 3 1 2 2 4 2 2 4 4 4 2 3 3 2 4 3 2 2 2 4 4 2 3 2 4 3 3 4 3 4 3 2 4 1 1 3 3 21 2 3 4 3 2 1.8 3.9 3.0 1.6 2.2 2.6 ## 3.8 4 3 3.3 4 3.0 4 3.6 3 2.5 4 3.2 4 3.8 3.0 2.2 2.0 1 1.9 1 3.0 4.0 3 3.1 2.5 4 3.4 2.3 3.0 4 3.4 3 2.9 4 4.0 4 3.6 4 3.1 4 3.3 3 2.6 4 3.1 4 3.4 2.5 3.0 2.5 3.0 2 3.3 3 3.2 2.8 2.4 2 4 4 1 1 4 4 4 4 4 4 3 1 2 2 4 2 74 Appendix 7 Questionnaire The questions are intentionally 'behavior' driven. That is, the respondent reads the behavior and decides whether that behavior is practiced: Never Rarely Sometimes Frequently Always Wording was chosen to be reasonably unambiguous. Where there is ambiguity, I have included descriptive words in the Terminology portion of this thesis. Phase 1 Technology Strategy Step 1 1. Technologists and Product Marketing work well together defining the best mix of technologies and markets 2. Markets identified and well characterized. 3. Each competitor is well characterized in the market. 4. Key customer satisfaction parameters are identified for each market segment. 5. Provide market-perspective for a decision on which technologies to focus on. 6. Technologies identified which link directly to customer satisfaction parameters. 7. Core resource capabilities required defined for each new technology. 8. Provide technology perspective for a decision on which technologies to focus on. 9. Market trends plotted. 10. Technology trends plotted. 11. Technology limits are established and compared to market trends. 12. Technology Risk assessed. 13. Business Impact of technology assessed. Step 2 - Develop Integrated Technology Strategy 1. Technology elements mapped into a product platform strategy, identifying derivative products. 2. Reuse matrix developed showing which elements are reused in each product derivative 3. Make or buy decisions based upon company's technological ability and resource availability. 4. Resource requirements determined. 5. Timeline estimates created N R S F A N R S F A N R S F A N R S F A N R S F A N R S F A N R S F A N R S F A N R S F A N R S F A N R S F A N R S F A N R S F A N R S F A N R S F A N R S F A N R S F A N R S F A 75 6. Budget requirements estimated 7. Key parameters sets for each product are determined which will meet projected market needs. N N R R S S F F A A 1. Integrated technology strategy is reviewed and consistent with corporate strategy and vision N R S F A Step 4 - Definition of needs for next generation technologies 1. House of Quality built for each technology 2. Market Pull requirements clearly defined 3. Technology push requirements clearly defined 4. System constraints defined for each technology 5. Environmental constraints defined for each technology 6. Selection criteria defined for each technology. 7. Main function of each technology defined. 8. Target values for key parameter sets defined. N N N N N N N N R R R R R R R R S S S S S S S S F F F F F F F F A A A A A A A A N N N N N R R R R R S S S S S F F F F F A A A A A N R S F A N R S F A R S F A R S F A R S F A R S F A R S F A R R S S F F A A Step 3 - Strategy Review Step 5 - Review and confirm needs 1. 2. 3. 4. 5. Technology risks reviewed. Market place risks reviewed. Resource allocation confirmed Budget requirements confirmed Schedule estimated. Phase 2 Concept Generation and Enhancement Step 6a - Develop Functional Structure 1. Functions decomposed into subfunctions in solutionneutral space, that is, without regard to any design concept. 2. Important interrelations between subfunctions highlighted Step 6b - Deploy Current Function level to Alternative Solution Concepts 1. Appropriate design parameter determined for each N subfunction. 2. Design parameters checked against criteria and N constraints. 3. Likely System conflicts between design parameters N identified. 4. House of Quality or equivalent built to show linkage N between functional requirements and design parameters N 5. House of Quality or equivalent built to show harmful and useful interactions between design parameters 6. Target values established for each design parameter. N 7. Tolerance Range established for each design parameter. N 76 Step 6c - Deploy Initial Concept to System Architecture 1. Apply Design Structure Matrix to re-allocate system elements and their functions into modules, that is, highly interrelated functions allocated into individual modules and work teams. 2. Great care taken to assure the architecture has minimal interrelations between modules. 3. Multiple System Architectures proposed, rather than keying in early on a single solution. Step 7a,b - Concept Evaluation and Selection 1. See step Step 8a - Concept Failure and Conflict Analysis 1. Failure modes predicted 2. Critical parameters identified for each failure mode 3. Noise factors identified which may cause failure modes 4. System Conflicts are identified 5. New concepts developed to break conflicts, rather than compromise between conflicts. 6. Failure modes correlated to System Failures Step 8b - Laboratory Work 1. Lab experiments performed to assess failure mode impacts on System performance 2. Lab experiments performed to assess impact of system conflicts on system performance 3. Conflicts are identified to be eliminated Step 9 - Enhance Concepts 1. TRIZ (or related methods) conflict elimination methodology applied. 2. Inventive problem solving methods (such as ARIZ) are applied to break system conflicts. 3. Failure mode impacts reassessed for Enhanced concepts. Step 10 - Concept evaluation and Selection 1. Criteria for Concept selection created 2. Superiority criteria derived from market and technology based requirements. 3. Robustness and flexibility criteria included based upon breadboard models. Phase 3 Robustness Development and Analysis Step 11 a - Define functional metric to measure robustness N R S F A N R S F A N R S F A N R S F A N N N N N R R R R R S S S S S F F F F F A A A A A N R S F A N R S F N R S F A N R S F A N R S F A N R S F A N R S F A N N R R S S F F A A N R S F A N R S F A A 77 N R S F A N R S F A Step 1 lb - Determine and select Critical Parameter sets 1. Critical parameters selected for each functional metric. N R S F A Step 12a - Design of Experiments 1 - Select Alternatives for Evaluation 1. Orthogonal arrays constructed to span the critical parameter space. N R S F A N R S F A N N R R S S F F A A N R S F A N R S F A N R S F A N R S F A N R S F A N R S F A N R S F A N R S F A 1. Functional metrics to optimize selected based upon their ability to capture the essential physics of the system function. 2. Signal- to- noise ratios are selected based upon the type of quality loss function expected. Step 12b - Design of Experiments 2 - Determine, select and impose appropriate Noises 1. Noises selected broad enough to cause failure modes to appear. 2. Environmental noises selected 3. Tolerance noises selected. 4. Customer usage noises selected. Step 13 - Perform experiments - Evaluate Performance of selected alternatives 1. Experiments run with combinations of noises and control factors. Step 14 - Calculate signal-to-noise ratios and select best parameter settings 1. Signal-to-noise ratios are used to select best control factors 2. Adjustment control factors utilized to move the response to target. Step 15 - Evaluate and Confirm Overall System Robust Performance 1. The system is set up and run with the selected best control factors to verify performance. Phase 4 Technology Selection, Transfer, and Integration Step 16 - Criteria for Winning Technologies 1. Superiority criteria metrics defined based upon market and technology requirements and independence axiom. 2. Robustness criteria metrics defined 3. Flexibility criteria metrics defined. 78 4. Maturity criteria metrics defined based upon producibility, safety and legal issues. Step 17 - Selecting Winning Technologies 1. Technologies selected meeting established Market Quality goals. 2. Technologies selected meeting established cost criteria 3. Technologies selected meeting program schedule criteria. 4. Promising technologies not yet ready for a product program are placed back into robustness development. N R S F A N R S F A N R S F N R S F A A N R S F A 79