A procedure was developed for systematically testing whole building energy simulation models and diagnosing the sources of predictive disagreement. Field trials of the method were conducted with a number of detailed state-of-the-art programs by researchers from nations participating in International Energy Agency (IEA) Task 12 and Annex 21. The technique consists of a series of carefully specified test case buildings that progress systematically from extremely simple to relatively realistic. Output values for the cases, such as annual loads, annual maximum and minimum temperatures, peak loads, and some hourly data are compared, and used in conjunction with diagnostic logic to determine the algorithms responsible for prediction differences. The more realistic cases, while geometrically simple, test the ability of the programs to model such combined effects as thermal mass, direct solar gain windows, window shading devices, internally generated heat, infiltration, sunspaces, earth coupling, and deadband and setback thermostat control. The more simplified cases facilitate diagnosis by allowing excitation of particular heat transfer mechanisms. The procedure was very effective at revealing bugs, faulty algorithms, and input errors in a group of building energy simulation programs that may be considered among the world’s best. The output data from the simulation programs can be used as reference ranges for comparing and diagnosing other detailed or simplified design tools.
Uncertainty Sources and Calculation Approaches for Building Energy Simulation Models
