ABSTRACT
Tests to establish the precision and bias of R-values for the central 2.44-m square part of 4.1-m square test panels were done using the large scale climate simulator (LSCS) at a U.S. national laboratory. Two different panels provided geometric variations. Air temperature in the climate chamber was varied from test to test, but was maintained near room temperature and above in the guard and metering chambers.
For most of the tests, the metering chamber required net heating to achieve steady temperatures. This is termed guarded hot box mode. For conditions where net cooling is required in the metering chamber, the LSCS is said to be in guarded cold box mode. A cooling loop was installed to provide slight overcooling of the metering chamber and allow control of temperature by the same heaters as used in guarded hot box mode.
In guarded hot box mode, intervals for 95% confidence ranged from ±1.8 to ±2.3% about linear regressions of the measured R-values with panel temperature. They increased slightly away from the respective average temperatures of all tests with each panel. Bias for the panel from layers of expanded polystyrene (EPS) over gypsum board and between and over wood joists varied from -1.3 to +2.4% of the R-value from a model using the panel’s geometry and independently determined thermal conductivities of its components. The other panel was a uniform slab of EPS. Unlike for the first panel, thermocouples were imbedded into the top and bottom surfaces and a layer of paint was applied to both surfaces. Bias for this panel ranged from -6.2 to -5.2% of the R-values from the full thickness divided by the measured thermal conductivity of the EPS. The bias is attributed to the imbedded thermocouples and the paint.
For the slab of EPS, R-values from a fit of its conductances, which are expected to be linear with temperature, were extrapolated slightly to the higher mean EPS temperatures of the guarded cold box tests. With optimum cooling, the LSCS produced R-values in guarded cold box mode within the 95% confidence interval for the guarded hot box tests.
