The heat reduction effect of trees has been investigated through numerical simulations; however, there are still challenges to applying the scientific results to the planning process due to the model's complexity and the computational resources required. This study investigates a rapid spatial evaluation method for heat stress potential, measured by mean radiant temperature (MRT), by decomposing radiation into sub-radiation using a multilayer MRT model. This method also enables the reproduction of optimized layouts considering the effect of tree arrangement in residential buildings. Multi-objectives were achieved through an evolutionary algorithm, resulting in more effective design layouts combining tree types and arrangements, all within a standard budget. By adopting this study's approach, landscape designers can create climate-responsive tree layouts with reduced heat exposure and generate customized planting designs tailored to their preferences.