基于自适应特征驱动法的散热结构拓扑优化设计

Topology Optimization Design of Heat Dissipation Structures Based on the Adaptive Feature-driven Method

拓扑优化技术已成为高性能散热结构的有效设计手段,典型散热结构优化结果常呈现出细小特征繁多、拓扑构型复杂等特点。然而传统拓扑优化方法由于设计结果难以导入到CAD系统和/或结果边界不清晰、不光滑,在对散热结构细小特征进行精确模型重构时存在很大的困难;特征驱动法、移动可变形组件法等能与CAD系统无缝衔接的新兴拓扑优化方法因具有初始布局依赖性,极难用于设计拓扑构型复杂的散热结构。鉴于此,借助仿生拓扑优化方法的树状分支结构层次生长技术,为散热结构的特征驱动优化提供合理的初始布局,发展出一种新型的自适应特征驱动法。该方法为尽可能扩大设计空间以获得更优异的结果,利用两个B样条曲线函数构建出具有足够变形能力的可弯曲特征;为在优化过程中反映仿生树状分支结构的层次关系,基于映射思想将分枝特征的起点创新性地定义于主枝特征的参数域内。以结构散热弱度最小化为优化目标,以给定的高导热材料体积分数为约束,对不同边界条件下的散热结构进行拓扑优化。数值算例表明,所提自适应特征驱动法能够使用相对很少的设计变量得到边界清晰光滑且性能优异的散热结构设计结果。

Topology optimization technology has become an effective design method for high-performance heat dissipation structures.The typical optimization results of heat dissipation structures often show the characteristics of many small features and complex topology configurations.However,the traditional topology optimization method has great difficulties in the model reconstruction of the small features of the heat dissipation structure because the design results are difficult to be accurately imported into the CAD system and/or the resulting boundary is not clear and smooth;New topology optimization methods,such as feature-driven method and moving morphable components method,which can seamlessly connect with the CAD system,are difficult to be used to design heat dissipation structure with complex topology configuration because of their initial layout dependence.In view of this,with the help of the hierarchical growth technology of the tree branch structure of the bionic topology optimization method,a new adaptive feature-driven method is developed to provide a reasonable initial layout for the feature-driven optimization of the heat dissipation structure.In order to enlarge the design space as much as possible to obtain better results,two B-spline functions are used to construct bendable features with sufficient deformation ability;In order to reflect the hierarchical relationship of the bionic tree branch structure in the optimization process,the starting point of the branch feature is directly defined in the parameter domain of the main branch feature based on the mapping idea.The topology optimization of the heat dissipation structure under different boundary conditions is carried out with the optimization goal of minimizing the thermal compliance and the given volume fraction of high thermal conductivity material as the constraint.Numerical examples show that the proposed adaptive feature-driven method can obtain the design results of heat dissipation structures with clear and smooth boundaries and excellent performance using relatively few design variables.