三维对流散热结构超大规模变密度拓扑优化方法研究

It is of great importance to ensure efficient convective heat-transfer capability when designing lightweight and compact industrial products. Large-scale topology optimization is a promising technique for designing heat conductive structure as well as cooling channels, which may bring tremendous opportunities in enhancing the heat-transfer capability of traditional designs. However, traditional SIMP based topology optimization approaches considering the engineering heat transfer model have difficulties in applying correct convective thermal load over the heat exchanging surface during the optimization process. Besides, those with fluid models lack robustness and suffer from high computational cost in solving the highly nonlinear fluid equations. Hence, traditional approaches can neither guarantee meaningful results nor ensure robust computational processes. They are not suitable for large scale parallel computation. The focus of this project is to propose brand new large-scale topology optimization solutions to design lightweight and manufacturable structures with optimized convective heat-transfer capability. First, a structural outer boundary dependent thermal load model is proposed to characterize convective heat transfer, which ensures that no convective load is applied over the internal boundaries if appeared. The long-standing issue that the conductive material tends to stick to the boundaries of a pre-defined design domain is investigated and avoided. Moreover, a new linear potential flow model based topology optimization approach is developed to optimize the cooling channel layout. The potential model in calibrated by turbulent flow model and efficient optimization strategies are proposed. Furthermore, by utilizing parallel computing and large scale topology optimization techniques, conductive structures and cooling channels of above 10 million degrees of freedom are optimized. This project will reveal the relationship among design freedoms, optimized layouts of convective heat transfer structures and the heat dissipation performance. The proposed approaches are validated by designing conductive structures and cooling channels of the battery pack of elective vehicles. The corresponding additive manufactured prototypes are obtained and their thermal behaviors are tested via experiments.

现代工业产品紧凑轻量化设计目标迫切需求高散热构型设计。利用超大规模拓扑优化技术设计三维导热结构和冷却流道，有望实现产品对流散热性能跃升。针对传统变密度拓扑优化方法中存在的动态边界对流热载荷施加难、流体优化不稳健、效率低的难题，本项目基于工程对流散热仿真，研究动态边界对流热载下的三维导热结构变密度拓扑优化模型，探明导热材料边界附着效应的产生机理及消除方法，提出导热结构构型优化方案；研究三维液冷管散热的等效渗流散热仿真模型及流体物理参数标定方法，建立基于渗流散热仿真的三维冷却流道变密度布局优化模型及优化方案；结合超大规模并行优化技术，实现面向千万级以上设计自由度的三维导热结构和冷却流道拓扑优化，探明设计自由度对结构构型影响机制，揭示结构构型对散热性能影响规律；以电动汽车电池组导热结构和冷却流道设计为例，开展面向增材制造的实证研究；为设计散热性能优异、面向增材制造的三维对流散热结构提供新方法。

现代工业产品紧凑轻量化设计目标迫切需求高散热构型设计，利用拓扑优化技术设计三维导热结构和冷却流道，有望实现产品对流散热性能跃升。本项目首先研究动态边界对流热载下的三维导热结构变密度拓扑优化模型，探明导热材料边界附着效应的产生机制，发展了从单元密度向相邻单元密度差连续转换的对流热载插值新模型及其拓扑优化新方法，有效解决了导热材料附着于设计域边界、阻碍流体流入的关键问题，确保了对流散热结构优化设计的合理性；进一步提出基于子结构凝聚法的导热结构拓扑优化新模型，顺利实现以较低计算成本优化设计尺寸小、排布密的高散热结构拓扑。其次，通过研究管内液冷散热快速等效仿真及流道布局优化模型，提出基于等效线性渗流散热模型的冷却流道拓扑优化新方法，实现了高散热冷却流道的快速、稳健设计；发展基于可变单纯复形的流道拓扑优化方法，实现了边界清晰光滑且具有优异流体力学性能的流道拓扑优化设计；进一步提出考虑含金属承载结构、多尺度点阵复合结构、散热流道及孔洞的多材料、多尺度、多学科拓扑优化方法，实现了散热流道-承载结构构型的快速拓扑优化设计。在此基础上，开发了基于分布式数据存储、消息传递接口技术的有限元仿真及优化并行求解算法，建立了三维对流散热结构超大规模拓扑优化设计平台；开展了不同自由度与优化参数下的散热结构优化设计及性能对比，探明了设计自由度对优化结构构型及对流散热性能的影响规律，实现了千万级以上设计自由度的三维导热结构与流道拓扑优化；以电动汽车电池组导热结构和冷却流道设计为例，验证了所提方法的有效性。项目研究建立了一系列面向三维对流散热结构的高效、鲁棒拓扑优化模型及方法，解决了对流散热结构拓扑优化中存在的结构动态边界对流热载荷施加难、非线性流体优化不稳健、效率低的关键难题，相关理论与技术手段可应用于新能源动力电池、热加工装备等工业产品的散热结构及冷却流道拓扑优化设计。