All articles

MODEL PREDICTIVE CONTROL IN HYBRIDGEOTABS

Posted in: News

The GEOTABShybrid solution proposes the integration of GEOTABS and secondary heating and cooling systems using an advanced type of control, Model Predictive Control (MPC). In contrast to traditional Rule Based Control (RBC), which uses heuristic control rules that are tuned using trial-and-error to take into account the buffering capacity of the GEOTABS building, MPC uses a mathematical model of the building to predict the future behaviour of the building heating and cooling needs, and uses these predictions to optimise the control of the GEOTABS and secondary systems. MPC can be used for optimizing the building operation towards various objectives, such as energy efficiency, energy cost, grid flexibility etc., while maintaining the thermal balance of the geothermal field and the thermal comfort in the building.

The models used for model predictive control, can be trained from measurement data of the real building and simple physical principles (grey-box models), or they can be based on physical equations (white-box models) and physical properties of the buildings. White-box models have the benefit that they can be used as soon as the building operation starts (when no operational measurements are available yet). In the project, developments are mainly focusing on this white-box MPC approach, but the case-study building in Libeznice allows for comparison between both modelling approaches, since the original grey-box controller is modified to a white-box controller during the project.

While MPC is highly beneficial for reducing building operation costs and improving its energy efficiency as well as thermal comfort, the set-up and implementation has been an expensive and complicated process until now. The set-up of the model is a highly specialised job that requires case-by-case model development, and also the adaptivity, robustness and practical implementation of the MPC bring along quite some challenges. This leads to high engineering and commissioning costs. Therefore, a toolchain has been developed that allows to set-up the model for the MPC in a semi-automated way, which considerably reduces the required engineering effort. Additionally, a strategy to start up the MPC is developed and an approach to make the MPC adaptive to changes in the building over time will be developed. Moreover, the robustness of the MPC formulation is being studied to allow for a smooth control behaviour.

The theoretical MPC-developments and performance assessments by simulation are complemented with real-life testing of the MPC in a pilot building and 3 case-study hybridGEOTABS buildings. This enhances practical software and hardware implementation of the MPC in these operational buildings, monitoring of the actual performance of the building and systems before and after the MPC is implemented, and monitoring of the actual design and commissioning costs that the MPC implementation brings along.

In the summer of 2018, we are preparing the first MPC-implementations in the school of Libeznice (CZ) and the Infrax office building in Dilbeek (BE)!

In summary, this project removes the barriers that prevent MPC from being widely implemented, and demonstrates the performance and benefits of the integrated GEOTABShybrid concept with MPC.