Design Innovation Guided by Modeling Assumptions

In this paper we propose a method for innovation in design based on relaxing modeling assumptions. A design process is innovative if it can expand the design space it searches beyond the space defined by the initial problem formulation. Typically, an engineering design problem is formalized as a set of constraints that must be met on various properties of the resulting design (e.g., weight), and an objective function on these properties to be optimized. The properties are defined by giving a set of equations for deriving them, and the parameters in these equations define the search space. However, these equations normally are simplified models of the general underlying physics, based on various assumptions (e.g., radial symmetry) about the artifact being modeled. We present a method of expanding the initial design space, when it does not contain any acceptable designs, by introducing new parameters, based on a process of relaxing the assumptions underlying the given models of the physical properties. Our technique rationalizes the Dimensional Variable Expansion approach to innovative design found in the literature, and generalizes it.