This paper presents a comparison of computersimulated and laboratory-measured thermal characteristics for three commercially available windows. The characteristics--thermal transmittance~ U, and shading coefficient, SC--werexperimentally derived under a guarded hot box thermal resistance test method and a recently developed solar-simulator-based test method. As a basis of comparison, analytical results were predicted for conditions corresponding to the laboratory test conditions.

Results indicate that simplified computer simulation programs may not adequately predict the thermal characteristics of all window frame configurations. Under certain conditions, a significant deviation in simulated Uvalues from laboratory results, up to 18%, may occur.

As commercially available glazing units include increasingly complex frame designs and material combinations, such computer programs may result in significant discrepancies. Complex designs and varying construction details due to manufacturing tolerances are accounted for in laboratory measurements. Consequently, to determine specific window characteristics, it is recommended that laboratory testing be undertaken and, if computer results are used, they should be refined and validated against experimental results.

Although the test specimens considered in this study are not representative of all currently available or installed products, this study may then serve as a basis for develop, ing a standard procedure to validate computer modeling intended to determine the thermal performance of glazing units.