Requirements analysis for Model Predictive Control in a decentralized district heating network

This study introduces a method to derive requirements for non-linear formulations in optimization problems for Model Predictive Control (MPC) of district heating networks. Those formulations become particularly relevant in decentralized networks where thermohydraulic effects stemming from pressure and temperature distribution impact the optimal dispatch schedule of producers. This is illustrated through a case study of the network in Weil am Rhein, Germany. Initially, a linear MPC formulation that neglects thermohydraulic dynamics was evaluated using one year of measurement data, revealing potential cost reductions of 14.3%. These savings primarily result from reduced operation of fossil fuel boilers and increased utilization of Combined Heat and Power plants. Subsequently, hydraulic simulations and monitoring data were analyzed, revealing that at least one of the production sites is unable to supply its installed capacity into the network during high-load scenarios due to hydraulic limitations. Furthermore, the analysis of thermal losses suggested that supply temperature optimization has an additional cost-saving potential of approximately 1.8%. The study concludes that future versions of the optimization framework require the consideration of pressure losses and pumping limitations to enhance operational reliability, while also recognizing additional improvement potential offered by supply temperature optimization. • Evaluation of thermohydraulic dynamics for improved model predictive control formulations. • Comparison of operating costs for conventional control vs. heat-based MPC strategies. • Impact of pressure losses and pump restrictions on MPC operation in decentralized networks. • Recommendations for optimizing operating strategies based on the analysis of monitoring data.