Operational Control of a Multi-energy District Heating System: Comparison of Model-Predictive Control and Rule-Based Control

TL;DR

This paper compares rule-based and model predictive control strategies for a multi-energy district heating network, demonstrating MPC's superior efficiency and cost reduction capabilities in managing variable energy sources and storage.

Contribution

It introduces a Modelica-based co-simulation platform for implementing and validating complex MPC strategies in district heating systems.

Findings

MPC reduces operational costs compared to RBC.

MPC handles variable energy prices and solar intermittency more effectively.

Simulation runs within 20 minutes for yearly analysis.

Abstract

This study focuses on operational control strategies for a multi-energy District Heating Network (DHN). Two control strategies are investigated and compared: (i) a reactive rule-based control (RBC) and (ii) a model predictive control (MPC). For the purpose of the study a small scale district heating network is modelled using Modelica. The production plant combines a heat pump, a gas boiler and a thermal solar field on the production side with a storage tank for flexibility purposes. On the consumption side, the virtual buildings are aggregated into a single consumer. We use our co-simulation and control platform, called Pegase, to implement the studied strategies. For both strategies the goal is to meet the consumers' demand while satisfying technical constraints. In addition MPC has the objective to minimize the operational costs, taking into account variable electricity prices and availability of solar thermal resource. Different scenarios are also defined and compared to study the effect of the heat plant sizing and forecasting error. The operational cost is reduced when switching from RBC to a MPC. As can be expected, MPC is more efficient when dealing with variable energy costs, intermittent solar energy and storage capabilities. This study also demonstrates how our tools enable an easy coupling of Modelica-based simulation with various control strategies. It especially supports the implementation and validation of complex MPC strategies in an efficient way, and yearly simulations are performed within 20 minutes.