Assessing the replicability of optimal control for GSHP integrated with PV and battery storage in different European renovation scenarios

Residential buildings that combine a ground source heat pump (GSHP), photovoltaic (PV) panels and a battery can increase self sufficiency and provide services to the grid. In Europe, the variability of solar production together with time of use tariffs makes the operational cost of such installations highly sensitive to the control strategy. Current control strategies are either sub-optimal reactive rules that ignore future operation, or sophisticated model-predictive controllers that require an accurate plant model and must be re-tuned for every new building. Consequently, a simple, yet robust, control system that can be directly reused across many cases remains an open problem. A Genetic Algorithm (GA) is employed to generate a 24-hours schedule. Each chromosome contains 24 numbers x_t∈[−1,1]. The sign defines the action (discharge, idle, charge), the absolute value the fraction of the available power. The GA was applied to a pre-existing database of GSHP yearly energy consumption. For each of the 87 case studies (3 climates×6 retrofit levels×4 PV orientations for different building types) the algorithm minimises the electricity-purchase cost over a 24-h horizon. Compared with a rule-based baseline, the GA reduces the annual electricity bill by up to 30% and by around 250 €/y. The algorithm performs better in climates with more intermittent solar production. The terraced-house archetype shows the highest relative savings because PV production and demand are balanced. Deeper retrofits increase relative savings while absolute savings drop and larger intra-day price spreads drive higher relative savings, demonstrating the controller exploits price fluctuations.