Multi-objective constructal designs of a radiator with rectangular cooling channels and porous fins via composite-function, ANN and NSGA-II

A radiator model with rectangular cooling channels and porous fins is established in this study. With the premise of given volumes of cooling channels and porous fins, constructal design for the radiator is performed by varying the cooling channel aspect ratio ( γ c ) and ratio ( α ) of fin spacing to top fin width. A composite-function ( F SW ) consisting of dimensionless entropy generation rate ( S ˜ gen ) and pumping power consumption ( W ˜ P ) is minimized firstly. Furthermore, multi-objective-optimization focusing on S ˜ gen and W ˜ P is conducted. The results reveal that F SW reaches the twice minimum of 0.828, and the related optimal-construct of radiator is α opt = 2.26 and γ c , opt = 2.80 . Compared to their initial values, the optimized W ˜ p and F SW are reduced by 37.4% and 17.19%, respectively. For multi-objective-optimization, the Pareto-front shows a generally monotonic decreasing trend. The decision-making ways of LINMAP and TOPSIS provide the lowest deviation-index. The related optimal-construct is α opt = 2.11 and γ c , opt = 2.21 , which is adopted as the optimal design choice for the radiator. The innovation herein is setting up the radiator model with rectangular cold plate cooling channels and external porous fins for conventional scale cooling of hot fluid and completing its constructal design. Its flow channel and application scale and scenario are different from existing microscale researches.