A Root Sum of Squares approach for determining uncertainties in three different ways of characterizing a surface temperature for a Hot Box Setup : With added U-value calculations and incorporated uncertainties

Abstract: The aim of this Master Thesis is to conduct an uncertainty analysis of three ways of characterizing a surface temperature of a Hot Box Setup (direct measurement, analytically calculatedand simulated). The purpose was to to provide insight into uncertainty analysis within Hot Box Setups. A Root Sum of Squares (RSS) method was used for the uncertainty analysis. The Hot Box setup consisted of a Hot Box with four interchangeable walls, two was subject to investigation- the XPS wall and a 1-pane window. Inside the Hotbox a heater and a fan was installed. The Hot Box was placed in a bigger chamber where the temperature can be expected to be relatively stable during the measurements. Main findings in this work were that the direct measurement of the outside surface temperature had the smallest uncertainty, (28.33 ±0.58) ◦C, while the analytically calculated surface temperature saw the biggest uncertainty,(27.69 ± 7.72) ◦C. The simulated surface temperature was found to be (29.11 ±0.7) ◦C, all being expanded uncertainty. The biggest uncertainties in the direct measurement of the surface temperature was from static errors in the thermocouples. In the analytically calculated surface temperature the biggest uncertainties propagated from the temperature measurementsof the inside air temperature and the inside XPS surface temperature. Good measures to consider for reducing uncertainties in the temperature analysis are to decrease uncertainties in the thermocouple measurements by having more accurate thermocouples, having more accurate air velocity measurements and to be careful in the examination of the k-value of the XPS material since it proved to affect uncertainties the most in the outside convective heat transfer coefficient, hoconv, and also significantly affect uncertainty in the ˙qtot for the XPS-wall. For the 1-pane window the U-value was found to be (5.27±0.82W/m2K) (expanded uncertainty). The biggest parameter affecting this uncertainty was from the h-value used for the U-value calculations.