Paper Conference

Proceedings of Building Simulation 2009: 11th Conference of IBPSA



F. Domínguez-Muñoz, B. Anderson, J. M. Cejudo-López, A. Carrillo-Andrés

Abstract: Increasing attention is being paid to the application of uncertainty and sensitivity analysis methods to model validation and building simulation. The idea is to let users to apply uncertainty bands to their model input data. These bands are then propagated through the model to determine the uncertainty bands of the simulation results. Mathematical methods to deal with uncertainties in computer simulations are well developed. One of the main difficulties the practitioner finds when trying to apply these transparency to thermal radiation or type and pressure of the gas come into play (ASHRAE, 2005). For a given aged material sample, the average conductivity mainly depends on density (ρ), temperature (T) and water content (w, when the material is hygroscopic). However, these three factors do not fully explain the value of conductivity, and dispersion remains due to differences in raw material properties, manufacturing process, etc. Conductivity can then be written as an average value ( k ) plus a random deviation (δ): techniques to building simulation is the lack of information on the uncertainty that affects to typical k = f (ρ,T , w; other factors) = k (ρ,T , w) + δ (1) input variables (thermophysical properties of materials, internal gains, infiltration, etc.). This paper is a contribution to fill this gap. We present polynomial fits for the average thermal conductivity and its standard deviation as functions of density for typical insulation materials. These functions were obtained by processing a large experimental data set, which was compiled in a previous European project For example, figure 1 shows a set of 1340 measurements for the conductivity of expanded polystyrene (EPS) at 10ºC, and dry and aged material. The key independent variable is density. Dispersion around the average conductivity is clearly seen in this figure. 0.055 headed by the BRE Scottish Laboratory. To illustrate how these results can be used in practice, an example is discussed on the validation of the mathematical model of a solar thermal collector.
Pages: 1008 - 1013