| Design of a structural system such as a building or a bridge involves one or more structural analysis procedures. Results of a structural analysis provide information about the performance of the structure, such as displacement, deformation, and stress of structural members. Such information enables a designer to design high-performance, cost-effective, and low-weight structures. However, the generation of an analysis model for a given design model is an inefficient and error-prone process. The major obstacle for automating the process is that the generation of an analysis model requires reasoning about what structural behaviors should be incorporated in the analysis model. Such knowledge can not be obtained by analysis programs, and is not provided in the design model. The goal of this research is to develop a new methodology to automatically generate an analysis model for a given design model. The methodology possesses the capability to reason about a design model and make assumptions about its structural behavior.;The input of the transformation process is a design model, and the output is an analysis model. The first step of the transformation process is to reason about what structural behaviors need to be incorporated in the generated analysis model. This is achieved by making engineering assumptions about the behaviors of the structure represented by the design model. An engineering assumption is an estimation about structural behavior based on the declarative knowledge about a design model. A logical reasoning system is employed to assist the selection of engineering assumptions. The next step is to decide how to simulate the assumed structural behavior, by selecting proper analysis-modeling actions. An analysis-modeling action is a unit operation that applies mathematical idealization and abstraction to the structure, such as selecting element types, applying constraints, or setting analysis options. The last step is to process the selected analysis-modeling actions. This step propagates analysis model entities, and hence generates a discretized analysis model.;A prototype software system is implemented based on the new methodology. The software architecture, specifications, and core algorithms are described. Examples are given to illustrate how analysis models are generated for given design models. |
