Calibration of an energy simulation with actual data has generally been considered too difficult to be part of the energy audit procedure. The purpose of this paper is to develop a systematic method using a “base load analysis approach” to calibrate a building energy performance model with a combination of monthly utility billing data and sub-metered data such as is commonly available in large buildings in Korea. The calibration procedure was specifically developed to be suitable for use in both the audit and savings determination procedure within a retrofit process. The procedure has been visualized using a logical flow chart and demonstrated using the simulation of a 26-story commercial building located in Seoul as a case study. The results indicate that the approach developed provided a reliable and accurate simulation of the monthly and annual building energy requirements of the case study building.

This paper describes a procedure for estimating weather-adjusted retrofit savings in commercial buildings using ambient-temperature regression models. The selection of ambient temperature as the sole independent regression variable is discussed. An approximate method for determining the uncertainty of savings and a method for identifying the data time scale which minimizes the uncertainty of savings are developed. The appropriate uses of both linear and change-point models for estimating savings based on expected heating and cooling relationships for common HVAC systems are described. A case study example illustrates the procedure.

An analytical model is developed to predict the annual variation of soil surface temperature from readily available weather data and soil thermal properties. The time variation is approximated by a first harmonic function characterized by an average, an amplitude, and a phase lag. A parametric analysis is presented to determine the effect of various factors such as evaporation, soil absorptivity, and soil convective properties on soil surface temperature. A comparison of the model predictions with experimental data is presented. The comparative analysis indicates that the simplified model predicts soil surface temperatures within ten percent of measured data for five locations.

Infiltration is customarily assumed to increase the heating and cooling load of a building by an amount equal to the mass flow rate of the infiltration times the enthalpy difference between the inside and outside air—with the latent portion of the enthalpy difference sometimes neglected. Calorimetric measurements conducted on a small test cell with measured amounts of infiltration introduced under a variety of conditions show convincingly that infiltration can lead to a much smaller change in the energy load than is customarily calculated; changes as small as 20 percent of the calculated value have been measured in the cell. The data also suggest that the phenomenon occurs in full-sized houses as well. Infiltration Heat Exchange Effectiveness (IHEE), ε, is introduced as a measure of the effectiveness of a building in “recovering” heat otherwise lost (or gained) due to infiltration. Measurements show that ε increases as: (a) flow rate decreases; (b) flow path length increases; (c) hole/crack size decreases. There is a clear correlation between large values of ε and large values of the exponent, n, so fan pressurization results may be useful in predicting ε for buildings.

Following several successful applications of feedforward neural networks (NNs) to the building energy prediction problem (Wang and Kreider, 1992; JCEM, 1992, 1993; Curtiss et al., 1993, 1994; Anstett and Kreider, 1993; Kreider and Haberl, 1994) a more difficult problem has been addressed recently: namely, the prediction of building energy consumption well into the future without knowledge of immediately past energy consumption. This paper will report results on a recent study of six months of hourly data recorded at the Zachry Engineering Center (ZEC) in College Station, TX. Also reported are results on finding the R and C values for buildings from networks trained on building data.