Improved genetic algorithm for pipe diameter optimization of an existing large-scale district heating network

The network optimization is vital for rational expansion of existing large-scale district heating systems (DHSs), which typically involves hundreds or even thousands of design variables, resulting in complexity and computational challenges when adopting traditional optimization methods. In this study, we developed an improved genetic algorithm (IGA) for pipe diameter optimization of existing large-scale DHSs. Only pipes with potential for profitability from reconstruction are selected as design variables. The index ΔP_max was used to assess the necessity of reconstructing each pipe. We found that selecting the critical specific frictional resistance was key for calculating ΔP_max, ensuring a balance between the stability and optimization ability of IGA. The IGA was applied to optimize an existing large-scale network in Xi'an City, northwest China. Compared to traditional genetic algorithm (TGA), the number of design variables was reduced sharply from 246 to 63. IGA achieved a reconstruction profit of 25.4 million CNY in <60 iteration steps, while TGA yielded a profit of 22.11–23.62 million CNY in >6000 iteration steps. Hence, IGA more efficiently identifies the optimum with lower computation cost and greater profits. The present study confirms that the IGA is an efficient tool for network optimization of existing large-scale DHSs.