Delta performance models for selection in decision-based design

The paper presents a modeling framework for rational selection between similar designs. Designers often make decisions between similar candidate designs such as deciding which off-the-shelf component should be used or whether to design that component from scratch. Such relative decisions between similar designs are common and important. A theoretical formulation of Delta Performance Models (DPM) is presented. DPM is an engineering design model that represents only the changes in performance that result from the small changes in competing designs. This method explicitly considers uncertainty and utilizes the fact that some uncertainty and modeling error is common to both candidate designs. DPM are intended to exploit the full resolution of engineering design models even if the models do not have high accuracy. The procedure is demonstrated on a simple design task using Monte Carlo simulations and a comparison is made with a traditional model.