Smart Buildings Model Predictive Control of an Aquifer Thermal Energy Storage System (ST-16-C018)

A rapidly growing amount of office buildings in the Netherlands is using an Aquifer Thermal Energy Storage (ATES) system. An ATES system uses a well pump to extract cold groundwater for cooling. The returned warm water is injected and stored in a second well. During winter this warm water is used for heating and the returned cold water is injected again in the first well. An optimal functioning ATES system can significantly reduce energy use and CO₂emissions of an office building. An essential condition for optimal ATES operation is the thermal balance of the system. Office buildings typically store much more heat than cold, causing the entire underground slowly to heat up and causing cooling capacity problems on the long term. This is compensated by using cold outdoor air to store additional cold during the winter, called regeneration. In this research a methods were evaluated to keep the thermal storage in balance Model Predictive Control (MPC) is used to control the amount of regenerated cold to maintain the ATES balance. The key element in the method is the reference model to calculate the expected stored amount and use as model for MPC. A reference model was constructed based on a case study building and it contains three main blocks: ATES, Heating/Ventilation/Air-conditioning (HVAC) and load simulation. For the ATES system a lightweight finite element simulation method is developed, based on an axisymmetric grid. An additional method was developed to reconstruct the injected water temperatures and volumes, because these were not measured in the case study installation. The HVAC and load simulation models are based on logged building management system (BMS) data. The use of BMS data has the large advantage that models are easily configured and can automatically adjust to changes in the building. Using MPC it was possible to keep the ATES in balance over a simulated 20 years period. By using a slight cold surplus as target, the effect of exceptionally warm winters is minimal and extraction temperatures are very constant. For the case study building it can be concluded that MPC, using the developed reference model, is capable of automatically maintaining the ATES balance. Because the case study building type and size is comparable to the majority of the new Dutch office buildings, it is expected that large parts of the method are universally applicable.