Effect of control model mismatch on model predictive control performance of sensible energy storage in dwellings with a heat pump and local electricity production

By introducing a hot water storage tank in a space heating system including a heat pump, the system gains flexibility towards the electricity grid. Electricity can be converted to heat at favourable times and used when required. Model predictive control of such a system requires a controller model which is able to represent the non-linear behaviour of the storage tank. In this contribution model predictive control of a building with heat pump and hot water storage tank with local electricity production, using a novel iterative linear optimal control approach is compared to the traditional ideal stratification formulation. Although the difference in model mismatch in a single open loop simulation is significant, the difference in total simulated energy cost or thermal discomfort when using model predictive control is limited due to the updating of the model states. As the ideal stratified linear program is solved significantly faster it can be used in simulated model predictive control. However for real-world model predictive control, where the solution time of both methods are acceptable, the iterative linear program results in a 2% decrease of energy costs while maintaining thermal comfort.status: publishe