TY - JOUR AU - Péan, Thibault AU - Lumbieres, Daniel Ramón AU - Colet, Alba AU - Bellanco, Ivan AU - Neugebauer, Martin Josef AU - Carbonell, Daniel AU - Iñarga, Jon Iturralde AU - García, Cristina Corchero AU - Salom, Jaume PY - 2022/05/22 Y2 - 2024/03/28 TI - Co-simulation studies of optimal control for natural refrigerant heat pumps JF - CLIMA 2022 conference JA - CLIMA VL - IS - SE - Digitization DO - 10.34641/clima.2022.432 UR - https://proceedings.open.tudelft.nl/clima2022/article/view/432 SP - AB - <p>The object of the present study is a natural refrigerant base (propane) heat pump system with a dual source/dual sink heat exchanger (air or ground-based) which is integrated into a centralized tri-generation system with PV and battery for a multi-family building located in Spain. To evaluate the performance of this complex system, a simulation environment was developed, connecting different software. The main program is TRNSYS, with the python package pytrnsys used to create the models and run the simulations, while a model predictive controller is externalized in a separate optimization software. The co-simulation environment enables to couple both software and operate the models in the simulation with the decisions made by the external controller. This environment was used to evaluate the considered system for three separate weeks of the year, each representative of the heating/cooling/DHW demands in winter, summer and intermediate seasons. For each of these weeks, the simulation was run once with a reference rule-based controller, and once with the advanced model predictive controller, to evaluate the additional benefits brought by the later strategy. The results were then extrapolated to the whole year, and revealed that the model predictive controller was able to provide cost savings of 12 to 20% (depending on the consideration or not of the cooling season which gave unexpectedly adverse results). This controller operated the heat pump more efficiently thanks to its prior knowledge of the best source to use at each moment (air or ground). It also managed the battery in a more economical way thanks to its prior knowledge of the time-varying electricity price, thus charging always at the cheaper hours of the day, and demonstrating the advantages of using forecasts and predictive optimization for HVAC control.</p> ER -