A genetic algorithm for topology optimization of area-to-point heat conduction problem

This paper presents a way of solving the classical area (volume)-to-point heat conduction problem by the means of a simple Genetic Algorithm (GA) in square configuration. After a short description of the numerical method, the optimal solutions proposed for minimizing the peak or mean temperature of a domain are presented. The effects of the conductivity ratio and the filling ratio on the configurations of the conductive tree are also analyzed and discussed. A numerical benchmark is then established to assess the influence of mesh resolution and the reproducibility of the GA optimization. Results show that GA is capable of proposing solutions having almost the same cooling effectiveness for different mesh resolutions or random seed generators. GA is also relevant compared to other optimization techniques presented here. It can be considered as a simple, easy to adapt and robust but computation time consuming method for addressing the general area (volume)-to-point heat conduction problem.