Calibrating Energy Conversion Networks for utility optimization and risk management

We demonstrate modeling and parameter estimation of flexible energy systems used in campus and municipal scale utilities with complex energy requirements, multiple fuel sources, and requiring substantial operational flexibility. Useful models should accurately predict the work production of these flexible utilities. A framework for such a model is a class of bilinear models for estimating the efficiency of complex and flexible energy utilities [9], This framework, which we call Energy Conversion Networks (ECN), may be used to determine financially optimal operating conditions and opportunities for financial and operational hedging by the utility operator [8], [10]. Given an ECN model, we compute a unique input-output mapping from the decision variables and heat input to work output. Multiple network realizations may be possible from a given input/output model. Work output from these models are expressed as rational functions of entropy flux with parameters of temperatures, thermal conductances, and engine efficiencies. Using plant data from a report published by the California Energy Commission [6], we demonstrate fitting this data on several network realization examples. We use measured work output as a function of heat input to calibrate a model. Heat rate curves are the most common way these data are typically available. We developed a data fitting procedure to determine the parameters resulting in the best calibrated model given a particular network model and data set.