As computational capabilities have increased, so too has our ability to apply engineering simulations to problems of ever increasing complexity. Unfortunately, while our desire to apply engineering analysis to problems of increasing complexity; the computational burden of these analysis tools often precludes their direct use on design problems where iteration is the norm. Not only does design entail iteration, the practice of design optimization relies upon iterative methods to arrive at an optimal and or robust design solution. Surrogate models, often called metamodels or models of models, are a common solution to this challenge. Numerous metamodeling techniques are available to the engineering designer. However, all metamodels face common issues in the acquisition of sufficient data to support the metamodel, algorithms to fit the metamodel to the data, and methods to exploit the metamodel for tasks such as design space visualization, analysis and optimization. This paper reviews the state of the art in metamodeling as a tool for product and process design. In addition to reviewing the various types of metamodels in use, discussions of current research issues in data acquisition, modeling the design space and using metamodels in product and process design are discussed. This review is intended to define the foundation for other papers within the conference session.

This content is only available via PDF.
You do not currently have access to this content.