Evaluation of optimization methods for the control of generators of a low carbon heat grid

Abstract

This study demonstrates an approach for the operational optimization of a district heating network supply system, consisting of multiple generators, including wastewater, air-source and ground collector heat pumps, a wood boiler, a combined heat and power plant, a natural gas boiler, and a photovoltaic system. To address the non-linear characteristics of the system, the optimization problem is formulated as a Mixed-Integer Nonlinear Programming (MINLP) problem and solved using a Nonlinear Programming (NLP)-based Branch-and-Bound (BB) method. Three NLP methods — Constrained Optimization BY Linear Approximation (COBYLA), Sequential Least SQuares Programming (SLSQP), and Trust-region algorithm for constrained optimization (Trust-Constr) — are applied and evaluated to assess their suitability for the optimization task. System models were developed in the Dynamic Modeling Laboratory (Dymola) environment to simulate and evaluate the optimization processes.

The results show that all three optimization methods significantly reduce operational costs and carbondioxid (CO₂) emissions compared to the reference algorithm. Among them, the COBYLA method exhibits the highest potential, achieving cost reductions of 9.06% and 7.39%, and CO₂ emission reductions of 2.49% and 5.65% for the two system scenarios (Red80 and WN40), respectively. Furthermore, the Pareto frontiers generated by the methods illustrate trade-offs between economic and environmental objectives, offering valuable insights for decision-making.

The proposed method lays the foundation for developing a flexible and adaptable MINLP optimization framework, which allows easy updates to system configurations and boundary conditions, making it suitable for various applications.