A Design Method For Product Family Trade Studies Utilizing GVI and PFPF Metrics With Application to Robotic Ground Vehicles

Open Access
Bobuk, Aaron Michael
Graduate Program:
Mechanical Engineering
Master of Science
Document Type:
Master Thesis
Date of Defense:
August 02, 2010
Committee Members:
  • Timothy William Simpson, Thesis Advisor/Co-Advisor
  • design
  • product
  • family
  • trade studies
  • GVI
  • generational variety index
  • PFPF
  • product family penalty function
  • robot
  • robotic
  • ground
  • vehicle
  • ATSV
  • genetic algorithm
  • Simulink
  • Matlab
  • mathematical model
  • commonality
  • platform
  • optimization
Effective product platforms must strike an optimal balance between commonality and variety. Increasing commonality can reduce costs by improving economies of scale while increasing variety can improve market performance, or in our robot family example, satisfy a wide range of different missions. Two metrics that have been developed to help resolve this tradeoff are the Generational Variety Index (GVI) and the Product Family Penalty Function (PFPF). GVI measures the amount of product redesign that is required for subsequent product offerings to meet new requirements, whereas PFPF measures the dissimilarity or lack of commonality between design (input) parameters during product family optimization. GVI is examined because it is the most widely used metric applicable during conceptual development to determine platform components. PFPF is used to validate GVI because of its ease of implementation for parametric variety, as used in this example. This work describes a product family trade study that has been performed using GVI for a robot product family and compares the results to those obtained by optimizing the same family using PFPF. Additionally, this work provides a first attempt to validate the output of GVI by using a complementary set of results obtained from optimization. PFPF optimization is made possible by a fast, comprehensive, and accurate mathematical model that is developed as part of this work to calculate design parameters and functional capabilities of a robotic ground vehicle. Additionally, a design method for iteratively populating the trade space with robots using this model is presented. The results of this study indicate that while there are sometimes similarities between the results of GVI and optimization using PFPF, there is limited correlation between them. Moreover, the platform recommended by GVI is not necessarily the most performance-optimized platform, but it can help improve commonality. In the same regard, PFPF may miss certain opportunities for commonality. The benefits of integrating the two approaches are also discussed.