| Calibrated energy models are useful for commissioning of building systems, measurement and verification (M&V) of retrofits, and prediction of savings due to energy conservation measures (ECMs). Many energy simulation tools exist that are capable of producing calibrated energy models of buildings, but often require professional expertise to reduce user error, resulting in significant monetary and time investments. Automating the calibration process provides a solution to these challenges, by circumventing the need for a professional modeler. Currently, there are no tools capable of automatically calibrating energy models, providing the motivation for this research.;A tool based on MATLAB, Perl, and DOE2 has been developed, utilizing a genetic algorithm (GA) optimization technique to automatically calibrate energy models to utility data. Calibration is achieved through simulation of various building constructions and comparison of energy consumption outputs to utility data. Estimations of unknown building parameters and monthly adjustments to equipment schedules are used to develop a calibrated model. Three buildings serve as test cases for evaluation of the consistency, accuracy, and efficiency of the tool. The GA optimization proves to be less consistent than Particle Swarm Optimization (PSO), however provides similar levels of accuracy at reduced processing times. The resulting models are used as baselines for determining cost effective packages of ECMs through GA optimization. Testing shows ECM evaluation is sensitive to varying building constructions and calibration accuracies, as well as budget constraints and ECM costs. |
