Data-Driven Co-Simulation Framework for Frost Prevention and Supply Air Boosting in MVHR Systems
Abstract
This study presents a data-driven co-simulation framework for real-time frost prevention and supply-air boosting in mechanical ventilation with heat recovery (MVHR) systems operating in cold climates. MVHR systems play a key role in reducing ventilation heat losses in airtight buildings; however, when exposed to cold outdoor air, frost formation on heat exchangers leads to increased energy use and reduced efficiency. In Stockholm, the defrosting load alone reaches approximately 162 MW, i.e., around 4% of the district-heating network’s peak capacity illustrating the importance of effective frost management strategies. The proposed framework integrates a frost-detection algorithm with a data-driven model predictive controller coupled to a neural state-space model of a full-scale MVHR unit at the KTH Live-In Lab. Implemented through a co-simulation framework, the controller dynamically adjusts fan speeds and heating-coil operation to maintain supply-air temperature while preventing frost accumulation. Simulation results confirm that the co-simulation framework eliminates frosting events, maintains stable operation, and improves energy efficiency without additional heating demand. The findings demonstrate the potential of datadriven co-simulation to enhance the resilience and performance of MVHR systems in cold-climate applications.
Keywords
Mechanical ventilation with heat recovery, co-simulation, model predictive control, frost prevention, data-driven modelling, energy efficiency