Feature-driven Structural Topology Optimization Theory And Method

Topology optimization is an effective design method of innovative structural patterns,which aims at maximizing the load-bearing potential of materials and attaining high-performance and weight-reduction design.Feature-based design is a fundermental design method of CAD communities and the basis of CAD-CAM integration.The two design methods have been independent of each other for a long term due to different design views.At present,the mechanical-then-feature design mode is generally used in engineering applications of topology optimization.Based on finite-element model and density-based method,structural topology optimization only considers the mechanical performance design of structures.A detailed feature-based design process is carried out to fullfill the requirements of mechanical features.Nevertherless,the two-step procedure not only elongates the design cycle,but also cannot gurantee the simultaneous optimum of mechanical performance and feature requirements.In recent years,more and more researchers begin to focus on feature design in topology optimization process.This thesis is dedicated to developing a feature-driven structural topology optimization method,including versatile feature modelling,mechanical analysis of feature model,sensitivity analysis of feature design variables,topology optimization design of structures with freeform design domain and pressure loads on moving boundaries.In this work,the following aspects are discussed.(1)An engineering feature driven topology optimization method is proposed.In this method,multiple engineering features are regarded as basic design primitives of topology optimization.Each feature is implicitly described in terms of level-set function(LSF).Topology optimization is achieved via the translation,rotation,scaling,deformation and Boolean operations of features within the design domain.It is indicated that different formulations of level-set functions directly influnce the configuration of optimized results and distribution of gray materials.In this sense,feature models described with signed distance function can obtain clear optimized configuration.Thus,a versatile approximation method of signed distance function is developed.Firstly,1-order approximation of signed distance function is deduced based on Tylor's expansion fomula.Secondly,the bounded normalization of KS function is mathematically demonstrated,which guarantees the quasi-normalization of level-set function for engineering features and structures.(2)A feature-driven topology optimization method is developed for structures with freeform design domain.The method sufficiently considers the complexity of design domain boundary.The freeform design domain modeler(FDDM)acts as a working window that clipping the solid and(or)void feature based topology variation modeler(TVM)via Boolean operations and achieving the topology optimization within the freeform design domain.A unified topology variation modeler is constructed with mixed solid-void features.Finite cell method is used for high-accurate structural analysis.It is demonstrated that sensitivity analysis with boundary integral scheme is independent of the mathematical expressions of the level-set function.Finally,the effectiveness of the presented method is illustrated with several typical examples including structures with freeform design domains,periodic structures and cyclic symmetry structures.The effects of different design models defined with pure-solid,pure-void and mixed solid-void features on topology optimization results are also studies.(3)A concurrent shape and topology optimization method is developed with implicit B-spline features.For simultaneous variation of moving pressure boundary and inner topology boundary in optimization problems with design-dependent pressure loads,implicit B-spline curves are introduced in the form of level-set functions(LSFs)to act as common design primitives for the description of both moving pressure boundary and inner topology boundaries of a structure.By updating the centers' position and controling radii of implicit B-spline curves,the pressure boundary and inner topology boundaries evolve simultaneously.An artificial non-designable offset domain of small bandwidth is introduced along the moving pressure boundary to reasonably ensure the continuity of pressure boundary and the imposition of pressure loads on solid materials.Implicit B-spline curves combine the advantages of parametric B-spline curves and implicit level-set functions.Firstly,the boundary can evolve freely.Secondly,finite cell method with fixed mesh can be used to achieve high-accurate structural response analysis.