Evaluation Of Analysis Error Caused By CAD Model Simplification

To achieve effective conversion from design models to analysis models and the seamless integration of CAD/CAE, the analysis oriented CAD model simplification plays a key role. However, current techniques for CAD model simplification can hardly fulfill the requirements of trusted analysis, due to the lacking of effective methods for quantifying the analysis error caused by CAD model simplification. To solve the problem, a physics-based approach to evaluate the analysis error caused by CAD model simplification is studied in this dissertation.The main works include:1. A framework for evaluation of analysis error caused by CAD model simplification is presented.According to the need of analysis oriented CAD model simplification and the principles that error evaluation must follow, we propose a framework for evaluation of analysis error caused by CAD model simplification.2. A physics-based error evaluation on detail removal is presented.In order to quantify the impacts of detail removal on analysis results, a physics-based error evaluation method is proposed. Specifically, we evaluate the error induced by detail removal based on the reciprocal theorem; and we quantify the impacts of each detail removal on the analysis results according to the changes of strain energy when each detailed feature is removed.3.A physics-based error evaluation on dimension reduction of thin regions is presented.In order to evaluate the impacts on the analysis accuracy due to the dimension reduction of thin regions in a CAD model, we propose a physics-based error evaluation method. We use the stresses changes on coupled interface to define the evaluation indicator. Local p adaptive method has been used on the thin regions to calculate the evaluation indicator. This method is able to quantify the impacts on the analysis results due to the dimension reduction of thin regions, while ensuring the analysis accuracy of the final model.4. An information reuse orientied method of boundary condition compensation for the simplified model is presented.In order to improve the accuracy of analysis results of the simplified model, we proposed an information reuse oriented method of boundary condition compensation for the simplified model. The boundary conditions of the simplified model are compensated by taking utility of the characteristics of the feature interfaces involved in computation of the evaluation indicator, which finally improves the analysis accuracy of the model.Based on above researches, we execute experimental validation of these approaches. At the same time, we give the relevant applications of the error evaluation indicator and its related analysis information on the generation of the analysis oriented multi-resolution model and the analysis based product detailed design.