THE EFFECTS OF MODEL SIMPLIFICATIONS ON EQUIVALENT THERMAL PARAMETERS CALCULATED FROM HOURLY BUILDING PERFORMANCE DATA

Using high resolution metered data of building heating performance and environmental conditions, equivalent thermal parameters (ETPs) can be identified for a building. These parameters, which typically include an effective envelope conductance, an effective thermal capacitance, and effective solar apertures, can be extracted via several methods that have been developed and demonstrated in other studies. If regressed ETPs are physically meaningful, they can be adjusted to account for potential retrofits, and energy performance under different Iconditions can be estimated. This paper evaluates the ability of one multiple regression technique to identify physically meaningful parameters and to predict performance. Using computer simulations to generate pseudo data and eliminate uncontrolled variables, this paper reports the effects of driving forces not explicitly included in the model on the derived ETPs. Specifically, ET Ps are regressed for a simple, massless house with no solar or internal gains. The house is then modified to include various combinations of solar gains, internal gains, envelope mass, and ground coupling. ETPs are calculated for each scenario. The derived ETPs are then used to estimate consumption for comparison with the original simulations. The results indicate that the ETP regression model accounts for internal and solar gains reasonably well in a massless or very light mass building. When thermal mass is present in a building, the regression model tends to underestimate both the building heat loss coefficient (UA) and the derived equivalent horizontal solar aperture. The model produces the most accurate UA estimates in very cold winter months, and the least accurate estimates in the swing months of spring and autumn. The most accurate solar aperture estimates are produced in the hot summer months, and the least accurate estimates in the cold winter months. When thermal mass is present in a building, the regression model produces confounded UA, solar aperture, and unknown gain (the intercept term) parameters. The equivalent thermal mass and the equivalent mass-room air conductance parameters are largely unaffected. This cross-correlation, though always present, is explainable in terms of simple physics.