Comparison of semi-quantitative and formal metrics for multizone air flow model quality assessment

Multizone air and contaminant flow models of actual buildings may produce significant deviations from measured conditions for a variety of reasons. The extent to which the well-mixed space assumption is valid is perhaps the primary reason, but other modeling approximations and input data uncertainties, e.g., the leakiness of building components and heating, ventilation, and air-conditioning system air flows also contribute to deviations. A method for calibrating multizone air flow models using a semi-empirical metric defined as the fraction of interzonal air flows that have the correct direction has been subjected to a number of field tests with encouraging results, i.e., the metric increases as a result of calibration. An important question is whether the improvement in this measure of quality is corroborated by the more rigorous-but also much more laborious to apply-ASTM Standard D5157 Standard Guide for Statistical Evaluation of Indoor Air Quality Models. To address this question, model calibration using the fraction of interzonal air flows metric and ASTM D5157 evaluations of carbon dioxide tracer gas releases at several measurement locations were conducted in two test buildings on a university campus. Quality in the ASTM standard is defined by the values of six different statistical quantities for each location considered, therefore its application to multizone systems is ambiguous. The principal approach adopted was to sum all metrics over all locations to obtain a single value representing the entire system. ASTM D5157 results also showed measureable reductions in deviations, which suggests that the semi-empirical metric is useful for its intended purpose. Consideration should be given to making ASTM D5157 more suitable for evaluating complex buildings.