Piping network optimization for district heating system using an enhanced Genetic Algorithm searching method

A.L.S. Chan

doi:10.1016/j.jobe.2024.110078

Piping network optimization for district heating system using an enhanced Genetic Algorithm searching method

2024-10-01

2

PRE

AI

A

A.L.S. Chan *

DOI：10.1016/j.jobe.2024.110078

Original

Original request for help

Save

Abstract

En 中文

Given the substantial construction and operational expenses associated with a district heating system (DHS), achieving an optimal layout for the piping network is paramount for its successful deployment. In this study, the optimization of DHS piping configuration has been explored. Traditional exhaustive search methods for network design are often impractical due to computational constraints, making Genetic Algorithm (GA) a preferred alternative for navigating the vast search space of optimization problems efficiently. However, GA, which relies on probabilistic heuristics, may not always pinpoints the most effective solutions, particularly those in close proximity to the current near-optimal configuration. To address this issue, an enhanced GA-based neighboring search method for identifying optimal or near-optimal piping layout for DHS more effectively has been developed in this study. When a local minimum is detected in the GA optimization process, the method initiates a neighboring search by selecting an elite candidate. The link with the highest piping cost from the elite candidate's configuration is identified. Subsequently, one of the other nodes in the configuration is connected to the nodes of the identified link to explore solutions beyond the local optimum. This new approach was firstly validated against an Optimal Communication Spanning Tree (OCST) benchmark problem, where it demonstrated superior efficacy by achieving higher success rate and quicker first-hit generation compared to conventional GA method. Specifically, this method showed a 60 % success rate with a minimum first-hit generation number of 140, outperforming the traditional GA's 40 % success rate and 263 generations. Further modification to the OCST benchmark to better simulate a real-world DHS application underscored the ability of this neighboring search method to exceed the previously established optimal configuration, uncovering a better' solution with superior fitness value. This promising methodology was subsequently applied to the hypothetical design of a DHS piping network, illustrating its potential to enhance the planning and implementation of efficient, costeffective district heating systems.

Keywords:

District heating system

Piping optimization

Genetic algorithm

Optimal communication spanning tree problem

Piping network optimization for district heating system using an enhanced Genetic Algorithm searching method

Abstract

Journal

Organization

Cited Papers