| Abstract | Computer simulation of buildings is currently used to predict their performance and the effect of design changes. The effect on predictions caused by uncertain-ties in the input data is rarely assessed. However, if it is quantified the performance of the simulated building can be described as a range of possibilities given the inherent uncertainty in the input data. This paper takes the process further, assessing the effect of uncertainty for an initial model and a model incorpo-rating a design change. It then describes how to use sta-tistical tests to quantify the significance of the change in performance: i.e. has the design change produced a real difference to the buildings performance, or is the change in performance lost in the error band of the predictions. In this paper the method of quantifying the overall ef-fect of uncertainties on predictions and how to differen-tiate between a significant and insignificant change is de-scribed. This is then applied to two case studies to exem-plify the importance of quantifying the effect of uncertainties. |
|---|
