The demonstration project has implemented, tested and demonstrated a fluid­borne heating system with a heat pump and a solar collector in a domestic building that makes optimal use of renewable energy from the grid.

This implied a test of model predictive control technologies that were able to benefit from probabilistic forecasts of heat load and energy prices. The controllers should be able to benefit from energy forecast services.

The final model structure has been established using statistical grey­box methods, and the parameters of the embedded model have been found using estimation techniques. The model has been formulated as a stochastic state­space model in continuous time, e.g. dynamics have been described as a set of stochastic differential equations.