Prediction of CAD model defeaturing impact on heat transfer FEA results using machine learning techniques

Essential when adapting CAD model for finite element analysis, the defeaturing ensures the feasibility of the simulation and reduces the computation time. Processes for CAD model preparation and defeaturing tools exist but are not always clearly formalized. In this paper, we propose an approach that uses machine learning techniques to design an indicator that predicts the defeaturing impact on the quality of analysis results for heat transfer simulation. The expertise knowledge is embedded in examples of defeaturing process and analysis, which will be used to find an algorithm able to predict a performance indicator. This indicator provides help in decision making to identify features candidates to defeaturing.