Session 9, Paper 3, CIBSE ASHRAE Technical Symposium, Dublin, Ireland, 3-4 April 2014
Increasing studies imply that actual building energy use might significantly deviate from design expectation. This so-called “performance gap” becomes a daunting challenge for the involved professions, stimulating them to reflect on
how to investigate and better understand the reasons for the gap. Although finding the reasons is crucial, operational definitions and perspectives should precede it. This paper offers a perspective from which the performance gap
implies a model evaluation and comparison problem under uncertainty quantification framework. We argue that performance gap at the design stage is the result of uncertainties. Among different sources of uncertainties, inadequate model predictions are important contributors to the overall performance gap. It is thus desirable to evaluate model prediction capabilities and determine the best modelling strategy among proposed solutions. We develop metrics for model evaluation, taking uncertainties into account. The metric quantifies the discrepancies in model predictions as statistical confidence intervals.
Computational case studies are deployed to examine the explanation power of the metric. Design predictions from deterministic simulation are compared with those from uncertainty quantification.