Optimal System Design with Geometric Considerations.

Abstract

System design is tied to both functionality and geometric realization. The former is pertinent to system performance, and the latter is related to packaging. Packaging is an optimization process that finds a desirable placement for the system components within a given space. When the components do not fit into the allocated space at the packaging stage, the design engineers must make modifications that can affect the performance of the system. The modification of a component can also affect the geometry and positions of other components in the system. These changes might lead to an infeasible layout. Therefore, optimizing the system performance considering packaging is desirable. Packaging problems and solution methods have been studied in many applications, such as electrical circuit layout, glass or metal cutting, truck loading, trunk packing, rapid prototyping (RP), architectural floor plan layout, routing, and mechanical component layout. Packaging problems in a mechanical system design are more challenging than 2D applications such as circuit layout and the metal cutting problem; this is due to a larger design space and increased complexity of geometry. Complex 3D geometry leads to increased computational time for interference checking, which is inevitable for finding a feasible layout. Detailed 3D CAD models, however, are not required or not available at the preliminary design stage. Therefore, abstract representation of the components is necessary during the layout process. Abstract models should balance accuracy of geometry representation and rapid computation capturing designers’ intent. This dissertation presents a computational environment for addressing the combined packaging and optimal system design. The packaging problem also includes pipe generation because pipe routing is also important problems in mechanical system design. The simulation model of a thermal management system for heavy duty series hybrid electric vehicles is used to demonstrate the usefulness of the proposed framework.