OPTIMIZING BUILDING ENERGY SIMULATION MODELS IN THE FACE OF UNCERTAINTY

Abstract

Model-based optimizations usually neglect the fact that building parameters cannot be estimated with high ac- curacy. The highly nonlinear characteristics of building energy systems (e.g., on-off control and capacity limita- tions) makes it harder to quantify the impact of uncer- tainty of model parameters. In addition, buildings are not stationary as equipment ages and utilization changes over time, requiring adjustment of previously found optimal solutions. A methodology has been developed to over- come these shortcomings and is illustrated using a sim- ple example: a building using a solar thermal collector for heating and domestic hot water. The approach entails Monte Carlo sampling from the uncertain domain, with an embedded optimization routine. Conditional probabil- ity density functions (e.g., energy consumption and de- mand) quantify the difference between base case and op- timal scenarios in the presence of uncertainty.