RSS

White-box Model Predictive Control: Optimal Control and System Integration of Heat Pumps

Increased use of renewable energy sources and heat pump-based technologies are making heating, ventilation and air conditioning of buildings more complex. This White Box Model can help optimize control and system integration of heat pumps.

Building Heating, Ventilation, and Air Conditioning (HVAC) systems are becoming more complex due to the integration of renewable energy sources and heat pump-based technologies. We are evolving from a society where the building demand determines heating and cooling loads to a society where the availability of heat and cold, through price signals, determines when to heat or cool a building. Furthermore, renewable energy sources tend to use lower temperatures for heating and higher temperatures for cooling since heat pumps operate more efficiently at smaller temperature differences, and direct geothermal cooling is simply not available at temperatures lower than the soil temperature. Smaller temperature differences result in smaller thermal powers, meaning that sudden power peaks have to be spread over longer periods. In order to reach the building comfort, set points in time, this typically means that heating and cooling have to start sooner, depending on the emission system inertia.

This discussion first illustrates the need for adequate HVAC system selection and component sizing during the design stage. There no longer exists one correct design but rather a multitude of feasible designs that consider the local context and potential of the building and the sustainability ambitions of the stakeholders in the construction process.

Model Predictive Control (MPC) is a predictive control methodology that relies on a mathematical model of a system to control that system, in this case, a building and its HVAC. The model considers weather forecasts, occupancy, the building envelope, and the HVAC devices that are connected to the building envelope. The model predicts the impact of the current control actions on the building energy use, emissions, and comfort and on operational constraints during the coming days. Control signals are sent to the existing building management system, and 15 minutes later, the optimization is repeated. While this summarizes the main functionality of an MPC, various approaches exist for implementing the model, which has to be custom developed for a building. Differences are related mostly to 1) the number of components and 2) the level of detail of those components. Such models generally have a strong mismatch between the model and the physical system, making it impractical to couple the model to the physical building.

Our goal is to capture the full complexity of each building and exploit the full potential of its HVAC system. Therefore, we use models that are both detailed in the represented physics and in the spatial representation of the rooms of the building. Thanks to this, the model is aware of the full system complexity. One of the demonstration buildings is the 4-story, 3000 m2 office building of Fluvius and Boydens engineering, part of Sweco in Brussels. For the relatively colder bottom floor, MPC managed to shift a larger fraction of the heating load to the Concrete Core Activation, which allowed the Air Handling Unit flow rates and the supply air temperature for the whole building to decrease. The strength of our detailed MPC implementation is that it was able to deal with these complex interactions of multiple zones and components, while at first sight, we, in fact, believed that the elevated condenser temperatures were a bug.

Filip Jorissen, Damien Picard, Wim Boydens, Lieve Helsen, Belgium
The text has been shortened by the HPC team

Download the full article from the Publication database here >
Read the full text in the HPT Magazine here >

Subscribe to Heat Pumping Technologies Magazine and Newsletter