Paper 072 – Impact of the weather forecast on a predictive controller performance: experimental studies with a residential heat pump for space cooling – 13th IEA Heat Pump Conference, Jeju, Korea
Heat pumps represent ideal systems to exploit the energy flexibility potential of buildings and their embedded thermal storage possibilities, particularly when they are operated by intelligent management systems such as model predictive controllers (MPC). The present study proposes to evaluate in an experimental setup the performance of an MPC controller used to control an air-to-water, variable speed heat pump for a residential building. In particular, the impact of the forecast quality is investigated, since MPC relies on the access to forecasts of the disturbances (weather, price, occupancy…) to perform its optimization. Two experiments of three days were thus repeated, the first one with a real weather forecast retrieved periodically from a commercial service, while the second one considered a perfect forecast. In cooling mode, the case with imperfect forecast resulted in an increase of the electrical energy use of the heat pump of 5.8%, corresponding to a cost increase of 11.2%, compared to the case with perfect forecast. These experimental results highlight the importance of the forecast quality on the performance of model predictive controllers for heat pumps.