What are the most significant advantages of integrated control strategies over local controls for heat pump and PV system performance?

To maximize the energy-saving potential of coupled technologies like HPs and PVs, we need a “federated” (or integrated) control strategy. The core of this strategy is to capture the complex interplay between the outdoor environment, the thermal zone, and the equipment. By predicting how these factors work together, the model acts as the intelligence that drives the Model Predictive Control (MPC) engine. This engine uses these predictions to calculate and determine the most energy-efficient way to operate the HP in real-time.

The first significant advantage is the shift from reactive corrections (reacting to past changes) to predictive optimization (acting on future forecasts). A federated MPC uses predictive models—in our study, three interconnected Artificial Neural Networks (ANNs)—to forecast the future states of the thermal zone, the HP’s energy consumption, and the PV’s electricity generation over a given time horizon. This allows the controller to exploit favorable conditions before they happen. For example, on a morning with high solar radiation, our federated controller predicted the peak PV generation and strategically pre-cooled the space by lowering the setpoint. This effectively used the free solar electricity to store coolness in the building’s thermal mass, significantly reducing the need for grid-supplied energy later in the day. The local control, by contrast, simply missed this opportunity, waiting until the room was already hot to react – at which point it had to pull additional power from the grid.

The second key advantage is the intelligent modulation of the HP’s Part-Load Ratio (PLR). HPs are variable-speed machines capable of subtle modulation. However, local controls often force these sophisticated machines to act like blunt instruments. Relying on predefined rules—such as turning on only when the temperature exceeds a set limit—effectively reduces a variable-speed unit to a binary ‘on/off’ device. This creates an inefficient operating pattern that completely ignores the hardware’s ability to run continuously at partial loads. A federated controller, however, understands the non-linear dynamics of the HP and utilizes its full operational range intelligently. In our real-world test, during a high-cooling-load, low-PV-generation afternoon, the federated control employed a clever strategy: it circulated indoor air through the Direct Expansion (DX) coil without resorting to HP cooling operation (i.e., the compressor was off). This put the system into a minimum power mode—just enough to delay the IAT rise—drawing such minimal power (e.g., <100W) that it was completely offset by the low PV generation. It effectively maintained comfort at near-zero net energy consumption, a nuanced strategy a blind local controller cannot replicate.