Dynamic Modeling of District Heating Network Based on Discrete Event Simulation

Abstract

Heating and cooling play important roles in saving energy and reducing emissions. Despite the potential of district heating (DH) systems to apply sustainable energy sources efficiently, DH accounts for only 8% of global heat consumption. The evolution of intelligent DH systems is constrained by insufficient digitalization, metering, and monitoring, which hinder effective optimization and planning. Addressing these challenges inevitably requires advanced dynamic simulation models that balance computational speed, accuracy, and adaptability for complex DH networks.

This research develops a flexible, accurate, and efficient dynamic DH network model based on the Lagrangian method. The model employs variable time step simulation within a discrete event simulation (DES) framework, offering key operational insights such as delivered temperature and energy, water transmission time, and heat losses. It can simulate complex meshed network topologies and diverse operational strategies, including "variable flow, variable temperature", making it suitable for real-world applications.

Real-world validations, including tests on a single pipe, a tree-shaped network, and a meshed network, verify the model’s high performance. For instance, an 85-day simulation of a meshed network with 186 pipes was completed in 0.29 seconds on a laptop (Intel Core i7-1185G7 CPU @ 3.00 GHz), achieving a mean residual standard deviation of 1.15 °C across 80 substations. These results highlight the DES model's potential for integration into future holistic system studies, empowering operators to optimize performance and advance the transition toward more sustainable and efficient heating systems.

The study establishes a robust foundation for variable time step simulation through critical sampling point identification and adaptive local updates based on lazy evaluation principles. The developed model dynamically adjusts temporal and spatial discretization to ensure accuracy, reducing computation time and minimizing numerical errors by avoiding unnecessary intermediate calculations. Additionally, a technique, called the tolerance threshold, by eliminating redundant sampling points improves computational efficiency without significantly compromising accuracy. This work not only provides a versatile tool for dynamic modeling of DH networks but also demonstrates the broader potential of variable time step simulations using the DES framework. By addressing key challenges in the development of the high-performance dynamic model, it will encourage the adoption of similar techniques across other energy system components.