(Dr. Farrokh Mistree, advisor)

"Product Platform Concept Exploration of Automobile Engines"

Abstract

Today’s market is characterized by globalization, volatility and intense competition.  Companies are adopting mass customization and open engineering systems (OES) “to achieve more with less” and be competitive in this market.  Product family design provides a method to mass customize a product providing more product variety in the market with a high level of design commonality between them.  In this thesis, the PPCEM (Product Platform Concept Exploration Method) is adopted as a framework to design a robust product platform for an automobile engine from which product variants of the family can be easily obtained.

To illustrate the product family design of automobile engines, top-level design specifications for automobile engine lubricated components (bearings, pistons, valves, etc.) are determined.  In this thesis the design using PPCEM results in performance (measured primarily in maintaining optimum oil film thickness and power losses in the sub-systems) that is fairly insensitive for different power ratings of the engines (products) and so achieves commonality.  The synthesis of the product platform using PPCEM involves the following activities:  Design space exploration (Market Segmentation and Design of Experiments), system modeling (engine friction simulation- EnFAS and system approximation using metamodels like Kriging and Response surface methods, RSM) and system synthesis (Compromise DSP using robust design concepts).  These methods help to increase computational efficiency, increase design freedom as well as design knowledge and provides satisficing solutions.  Since the product family design in PPCEM might depend on the system approximation, a comparative study is also done to assess the performance, merits and difficulties of the two different approximations (Kriging and RSM) in the product family development.  In addition research is also done to study the benefits of combining these two approaches and evaluating their performance in this case study.  This work will provide guidelines to help the designer make a better decision on the choice of the metamodel in other deterministic analyses.