Sensitivity analysis techniques for building thermal simulation programs

Three sensitivity analysis techniques, differential sensitivity analysis (DSA), Monte Carlo analysis (MCA), and stochastic sensitivity analysis (SSA), are appraised using three detailed finite difference simulation programs, ESP, HTB2, and SERI-RES. The applicability of the methods to simpler programs is considered. Domestic-scale, passive solar buildings are used as vehichles for testing the methods. The sensitivities, in both hourly and daily average predictions, due to the uncertainties in over 70 input parameters, are compared for DSA and MCA. The sensitivities of the predictions to changes in a reduced set of inputs are compared for DSA and SSA. It was found that in this case SSA had drawbacks. It is suggested that, at present, DSA is used to obtain the sensitivities of predictions to individual input parameter uncertainties and that MCA is used to obtain the total sensitivities in the predictions. With further work, it may be possible to extract individual sensitivities from MCA, which would make this the preferred technique.