Uncertainty Quantification for Digital Twins in Thermal Energy Distribution System

Digital twin (DT) models play a critical role in the management and control of the integrated energy system, where the predicted transients of energy systems can inform the dispatch optimization, supervisory control, and autonomous operations. The usefulness of DTs depends on the uncertainty of the DT models. This paper identifies the boundary conditions of inlet mass flow rates as the major sources of uncertainty in a Modelica-based DT model for a thermal energy distribution system. By calibrating the boundary conditions using sequential model-based optimization, this work significantly reduces the uncertainty in predicting the axial temperature distributions by more than 50%. This paper further discusses challenges in uncertainty quantification for DTs.