A Method Of Defeaturing Error Estimation Based On Domain Unification

Simplifying an engineering design by removing geometric details(called defeaturing)that are unimportant to the engineering analysis accuracy before simulation can significantly accelerate the speed of mesh generation,improve the mesh quality,reduce the mesh generation failure,and reduce the complexity of downstream simulation tasks.However,the simplification process inevitably leads to engineering analysis error and loss of simulation accuracy,called defeaturing error here.Quantitative evaluation of the defeaturing error is of great significance to achieve the ultimate goal of physics-reliable geometry simplification.Due to the complexity of geometric models and physical problems and the lack of theoretical researches,effective error estimation of the defeaturing is still a bottleneck issue in seamless CAD/CAE integration.Defeaturing error is generally estimated a posteriori,and existing researches mainly focus on removing a single type of features,either boundary features or internal features.None of them works on both the boundary features and internal features,which raise various issues such as complex implementations and computations.In addition,previous approaches on internal feature only give a first-order defeaturing error estimation,and may lose the estimation accuracy.Approaches based on a concept of domain unification are proposed to tackle the above two issues.In summary,the main contributions of this paper include:1?A concept of domain unification is proposed for a posterior estimation of defeaturing error.It takes the process of defeaturing as a shape deformation,and thus unifies the computational domains before and after defeaturing.Based on this,a posterior estimation of the defeaturing error can be performed on a unified domain,where existing approaches can be borrowed.2?Based on the strategy of domain unification,a defeaturing error estimator is proposed that works both on the boundary features,either positive or negative,and on internal features.The defeaturing error is converted to the modeling error caused by different physical modeling problems,which are further estimated using the previous DWR method.3?A novel second-order defeaturing error estimator is proposed to estimate the defeaturing error for an internal feature using both the shape sensitivity analysis(SSA)and the topological sensitivity analysis(TSA),still based on the domain unification strategy.The method views the internal feature removal process as a deformation from feature to circular hole,and a circular hole removal process,which are respectively estimated using the second-order SSA and TSA.