ESD. 71 Application Portfolio: Flexibility in the Product Design Process Alison Olechowski Department of Mechanical Engineering Massachusetts Institute of Technology 12/7/2010 Alison Olechowski ESD.71: Application Portfolio 1 ABSTRACT In this paper,we model fixed and flexible engineering design and perform a Monte Carlo simulationto explore differences inoutcomes. The typical point‐based design process consists of deciding at the project start on which specific design concept to pursue, based solely on first forecasts of uncertain customer requirement and technology performance. The alternative flexible set‐based design method uses parallel exploration and the delaying of decisions to achieve an improved pay‐off based on what is learned about both uncertainties through the development time. Results from the simulation reveal that the flexible design process results in a significantly higher expected net present value than the fixed process. Thus, the higher sales revenue resulting from the better design outweighs the high cost of carrying additional designs forward. Further simulation revealed that the set‐based design process is more valuable on a high‐risk project, such as an evolutionary product with new technology. On a low risk project, the marginally better resulting design is not worth the extra development cost. Alison Olechowski ESD.71: Application Portfolio 2 TABLE OF CONTENTS 1.0 INTRODUCTION.................................................................................................................................3 2.0 SYSTEM DESCRIPTION .......................................................................................................................3 3.0 SOURCES OF UNCERTAINTY .............................................................................................................. 5 3.1 Technology Uncertainty ................................................................................................................ 5 3.2 Customer Requirement Uncertainty............................................................................................. 6 4.0 SYSTEM DESIGN DESCRIPTION.......................................................................................................... 6 4.1 Deterministic Design ..................................................................................................................... 7 4.2 Flexible Design .............................................................................................................................. 8 5.0 SYSTEM DESIGN MODEL ...................................................................................................................9 6.0 SIMULATION RESULTS..................................................................................................................... 11 6.1 Interpretation of the Cumulative Distribution Function............................................................. 12 6.2 Performance Metrics ..................................................................................................................13 6.3 Break‐Even NPV Probability........................................................................................................ 14 6.4 Stability of Results....................................................................................................................... 14 6.5 Effect of Risk................................................................................................................................14 7.0 CONCLUSION................................................................................................................................... 15 8.0 REFERENCES ....................................................................................................................................16 9.0 APPENDICES ....................................................................................................................................17 9.1 One run of the point‐based simulation....................................................................................... 17 9.2 One run of the set‐based simulation .......................................................................................... 20 Alison Olechowski ESD.71: Application Portfolio 3 1.0 INTRODUCTION Many of today’s products are made up of mechanical, electrical and software components. The design of these new products can be a complex and costly process; there is interaction between components and modules, there are multipleengineering teams working on various facets of the product, there are a number of management decisions that need to be made throughout the proces s, and there aresignificant uncertainties affecting the design. New product design is influenced by uncertainties in technology, market/comp etition and customer requirement, just to name a few. These uncertainties are especially important to consider when designing revolutionary/radical (as opposed to evolutionary/incremental) products. Given the above characteristics, the system of product design was chosen for analysis in this application portfolio. 2.0 SYSTEM DESCRIPTION The product design decision‐maker is faced with a number of uncertainties throughout the life of the project. He/she will learn more information about technical performance and customer requirement as time goes on, when tests are run, prototypes are built, etc. but he/she must make decisions before that information is created. Consider the design of a product with multiple distinct features, for example, an isolator with ventilation, containment, and decontamination sub‐systems. This model will consider the design of one feature, from the point of view of one design team. As the team designs, they are receiving information about requirements from the other teams, since all features interact. At the same time, it’s possible that the customer has changing needs. Thus the performance requirement will change over the timeline of the project. The model presented will explore one specific feature requirement, for example, the required flow rate through the ventilation system. This is a measurable quantity that is very much dependent on the design of the other sub‐systems. A typical design process is made up of four phases, as shown in Figure 1. Consider a set of design review meetings occurring at the end of each phase, at which point there is
View Full Document