Comparative study on topology optimization of microchannel heat sink by using different multi-objective algorithms and objective functions

• Published in: Applied thermal engineering Vol. 252; p. 123606
• Main Authors: Wang, Jiahao, Melideo, Daniele, Liu, Xiaomin, Desideri, Umberto
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.09.2024
• Subjects: MCHS; Multi-objective algorithm; Thermal performance objective functions; Topology optimization design; Topology optimization design; Multi-objective algorithm; MCHS; Thermal performance objective functions
• ISSN: 1359-4311; 1873-5606
• DOI: 10.1016/j.applthermaleng.2024.123606

•SAW model and ε-constraint method are used to multi-objective topology optimization.•Thermal performance goals adopt maximizing heat transfer and uniformity respectively.•ε-constraint achieves better computational efficiency and convergence than SAW.•Adaptive-optimized microchannels adapt accurately to different objective requirements.•Topology optimization structure can produce optimal overall performance. To address the challenges of low computational efficiency, poor solution quality, and the difficulty in accurately and synergistically optimizing heat transfer and reducing flow loss in multi-objective topology optimization of microchannel heat sinks, this study innovatively proposes a multi-objective topology optimization model based on ε-constraint algorithm. Moreover, the multi-objective optimization functions are constructed using different heat transfer single-objectives: heat transfer amount JQ and temperature variance JTV. For model improvement methods, a double-interpolation concept improved on the q-parameterized interpolation function is used to alter the continuity distribution state of density design variable ξ. The adjoint-based discrete sensitivity model and Global Convergent Moving Asymptotic Algorithm are used to implement the iterative update of optimization structure. The result shows: the optimized structures and its performance parameters evolve regularly with the weight coefficients of multi-objective functions, revealing the optimization mechanism of microchannel and state variables, and the trade-off game between structure and performance; The convergence stability of ε-constraint algorithm is significantly improved compared to traditional normalized Simple Additive Weighting model, and the computational efficiency of the representative case is relatively improved by 40.4%. The ε-constraint algorithm effectively suppresses the grayscale area and intermediate density range, thereby achieving higher-quality solutions and the state variable distribution more consistent with physical laws. The optimization model responds significantly to different JQ and JTV, and corresponding optimized structures can achieve maximum heat exchange and optimal temperature uniformity under minimum fluid energy consumption, respectively.