AN IMPROVED PROCEDURE FOR DEVELOPING CALIBRATED HOURLY SIMULATION MODELS

摘要

In order to improve upon previous calibration techniques,rnthis paper presents new calibration methods including arntemperature bin analysis to improve hourly x-y scatter plots, arn24-hour weatherdaytype bin analysis to allow for thernevaluation of hourly temperature and schedule dependentrncomparisons, and a 52-week bin analysis to facilitate thernevaluation of long-term trends. In addition, architecturalrnrendering is suggested as a means of verifying the buildingrnenvelope dimensions and external shading placement. Severalrnstatistical methods are also reviewed to evaluate therngoodness-ofincluding percent difference calculations, mean biasrnerror (MBE), and the coefficient of variation of the root meanrnsquared error (CV(RMSE)).rnThe procedures are applied to a case study buildingrnlocated in Washington, D.C. Nine months of hourlyrnwhole-building electricity data and site-specific weather datarnwere measured and used with the DOE2AD buildingrnsimulation program to test the new techniques. Use of thernnew calibration procedures were able to produce an hourlyrnMBE of -0.7% and a CV(RMSE) of 23.1% which comparernfavorably with the most accurate hourly neural networkrnmodels.