| Built-in heat exchange channels are widely used in a variety of heat exchange equipment and devices,but their designs still mostly adopt parallel branch or serpentine structures,which makes it difficult to meet the multi-objective requirements of flow and temperature.The maturity of additive manufacturing technology makes it possible to manufacture and apply complex channel designs,but there are still many problems in the physical field analysis and structural optimization methods of complex heat exchange channels that need urgent exploration.This thesis takes the built-in net-based heat transfer channel as the research object and investigates multiphysics solving,topology optimization of complex channel structures,and other related topics.(1)A thermal-fluid model based on the pipe-net method and the finite difference method is established,which offers low computational cost and allows for quick acquisition of flow and temperature fields of the net-based heat exchange channel.Furthermore,we compare the model to experimental data from the literature and find that the mean absolute percentage deviation(MAPE)of the maximum temperature is about4.06% under different Reynolds numbers.(2)A method to solve the local loss coefficient calculation problem for net-based channels is proposed,which can calculate the local loss coefficients of three-way and multi-way pipes.This solution method is based on CFD simulation data and employs intelligent calculation methods.The solution method for the three-way local resistance coefficient is trained using Neural Networks,achieving a MAPE of approximately 6.55%when compared to the simulation data.In addition,we investigate the effects of Reynolds number,pipe length-diameter ratio,and multi-pass pipe joints on local loss coefficient.A multi-branch vector synthesis method is developed based on the aforementioned three-way pipe research to calculate the local loss of multi-way pipes with three or more branches distributed in the plane or space.The results obtained through this method exhibit a MAPE of approximately 15.32% when compared to the CFD simulation data.(3)A topology optimization method is developed for designing heat exchange channels in the design domain of plane and curved surface.In this method,a net-based thermal-fluid model is utilized to predict the physical field,and a metaheuristic algorithm genetic algorithm(GA)without gradient information is employed to efficiently obtain optimized heat exchange channels that are complex and clear.The optimization results are studied with respect to the initial density of structural branches and the choice of objective functions.It is found that the optimization results are not highly dependent on the initial density of structural branches,and that comprehensive multi-objective optimization can better meet engineering requirements.Additionally,the topology optimization research of the heat exchange channel in the curved surface is carried out.The research objects include the solar cavity receiver and its heat exchange channel,as well as internal cooling channels in turbine blades.In the receiver optimization,a non-uniform variable heat source that considers heat loss is adopted.Compared to the reference design,the multi-objective optimization results indicate that the temperature uniformity and thermal efficiency were increased by 72% and 0.2%,respectively,and the pressure drop was reduced by 88%.Regarding the optimization of the blade internal cooling channel with compressible and variable gas media,a multi-objective optimization research is conducted under different numbers of inlets and outlets.The results show that excellent temperature uniformity can be attained with low total pressure drop through the selection of a suitable combination of multi-inlet and multi-outlet configurations,provided that the temperature meets the design requirements.(4)A multi-level optimization method that comprehensively considers topology,shape and size optimization variables is developed and applied to the design of heat exchange channels in cooling plate.A hybrid optimization algorithm combining GA and Method of Moving Asymptotes(MMA)is proposed to carry out multi-level optimization at a relatively low computational cost for the net-based heat exchange channels.The computational time of the hybrid optimization is reduced by approximately 90.9%compared to fully GA-based multi-level optimization.The multi-level optimization using the hybrid optimization algorithm is also compared with the conventional density method,demonstrating that the former can obtain clear and complex channel distributions in significantly less time. |
