Model-Based Support for Mutable Parametric Design Optimization

Traditional methodsfor parametric design optimization assumethat the relations between performance criteria anddesign variables are known algebraic functions with fixed coefficients. However, the relations maybe mutable, i. e. , the functions and/or coefficients maynot be knownexplicitly because they depend on input parameters and vary in different parts of the design space. Wepresent a model-based reasoning methodologyto support parametric, mutable, design optimization. First, wederive event modelsto represent the effects of the system’s parameterson the material that flowsthroughit. Next, weuse these models to discover mutablerelations betweenthe system’s design variables and its optimization criteria. Wethen present an algorithm that searches for "optimal" designs by employingsensitivity analysis techniques on the derivedrelations.

Authors

• Ravi Kapadia
• Gautam Biswas
• Vanderbilt University

Keywords

No keywords are indexed for this paper.