Demand Modeling for Enterprise-Driven Product Design
Author: Kumar, Deepak
Supervisor: Chen, Wei
Institution: Northwestern University, Evanston, Illinois
In a highly competitive global market, it is becoming increasingly important to view product design as an enterprise-level profit-driven activity. Demand models have become central to enterprise-driven design approaches since they help estimating the economic impact (e.g., revenues, profits, and cost) of a product‘s design. The primary contribution of this dissertation is the development and integration of demand models into three major product design formulations. First, a multilevel optimization formulation of Decision-Based Design (DBD), the enterprise-driven design approach used in dissertation, is developed to better address the hierarchical nature of decision-making in industry and the associated computational complexities. The approach treats enterprise-level product planning and engineering-level product development as two separate optimization problems, and links them through a demand model. The approach is demonstrated through the design of a vehicle suspension system. Next, the Market-Driven Product Family Design (MPFD) methodology is developed to enable designers to systematically and simultaneously examine market considerations (market share, revenues, price, etc.) and cost benefits due to shared manufacturing resources. In contrast, existing methods focus exclusively on commonality benefits. The MPFD approach helps designers make decisions on what the optimal number of products in the line should be, how to position each of them in appropriate market niches, and how best to allocate manufacturing resources. The design of a family of electric motors is used to demonstrate the approach. Finally, a hierarchical demand modeling approach, together with a model fusion technique, is developed to incorporate preference models at system, sub-system and component levels, and pool preference-models from multiple data sources. The motivation arises from the need to deal adequately with the complexity of large design artifacts or exploit the availability of preference-data from multiple sources (e.g., sales records, surveys). Vehicle package design is used to articulate the research issues as well as to demonstrate the approach. It is hoped that the methodologies and tools developed in this dissertation will help designers include market considerations more explicitly in problem formulations and adopt an enterprise-driven approach to design. Towards that goal, the applicability of the design approaches presented in this dissertation is demonstrated through industrial and numerical case studies.