A data-driven approach for simulating the energy performance of closed-loop hydronic heating systems

<p dir="ltr">Water supplies in closed-loop hydronic heating systems often contain varying amounts of impurities impacting system’s energy performance and are often poorly maintained, resulting in degraded performance. Traditional first-principles models inherently do not account for such problems. This study aims to develop a data-driven model to predict the impact of low-quality water supplies on the energy performance of such systems. Controlled experiments in dedicated testing facilities were conducted to collect energy-performance-related data to train Artificial Neural Network (ANN), Random Forest (RF), Gradient Boosting Machine (GBM), and Support Vector Machine (SVM) models. An experimental rig was installed in a laboratory to replicate the operation of a closed-loop hydronic heating system and control the parameters that influence its energy performance. Results suggest that the RF model demonstrates the lowest prediction errors in testing (mean absolute error of 0.95), followed closely by the ANN model. Both models outperformed a first-principles model which recorded a mean absolute error of 2.26. The findings suggest that machine learning models can facilitate more accurate and near-real-time estimations of pumping energy demand under degraded conditions, by identifying inefficiencies in the system overlooked by first-principles models.<br><br><b>Key Innovations</b></p><p dir="ltr"><br></p><ul><li>Integration of a water quality parameter regarding sediment buildup in hydronic heating sys?tems into machine learning (ML) models for estimating pump power consumption.</li><li>Rigorous comparison of ML models for pump power consumption prediction to justify the selection of appropriate method for QA/QC protocols</li></ul><p><br></p>