Advances in modeling and simulation have led to the development of highly detailed and complex building performance simulation (BPS) models. However, at the same time, they have become increasingly data hungry and computationally intensive tasks, for instance to assess the energy efficiency of early building designs or energy saving potential of retrofits. In the midst of it, not all the required datasets can be accessed or are available and also not all of the available ones can be obtained with exact certainty without any ambiguities. Thus the accuracy of building simulations based on single value assignments to its data inputs and parameters may be undermined. That points to the requirement of a paradigm shift in building simulation practices to present probabilistic predictions of energy consumption and as well as a mechanism to address and assess the impact of data uncertainties on the predictions. To this end, this book chapter first comprehensively summarizes the uncertainty assessment approaches in BPS by a systematic categorization in terms of quantified uncertain parameter, characterized and propagated uncertainty, applied uncertainty measures and the nature of BPS models used. Secondly, it describes two uncertainty measures: Morris and Sobol methods, respectively, one each from the class of qualitative and quantitative uncertainty assessment. The aim is to assess the impact of uncertainty in building geometry arising from its semi-/automatic modelling and reconstruction procedures on the building energy performance. Case study 1 on the application of Morris method has revealed a non linear nature of BPS model with respect to building geometry parameters and the position parameter with orientation being the most influential one. Case study 2 on the application of the Sobol method presents the uncertain range of predictions of hourly varying heating energy demands for a typical winter week specified by 95% interval bounds along with the corresponding calculated temporal sensitivity indices. The case studies have pointed out that building orientation and height are the most influential geometry parameters affecting energy predictions. Thus, it invites for further research into the development of more accurate techniques for building roof plane and ridge line extraction so that the uncertainty in the predictions can be reduced. The applications of methods have proven the demonstrated uncertainty assessment methodology to be generic and extensible, which has potential to be exploited for uncertainty quantification of parameters other than building geometry for instance, of building construction material and heating system.
CORE (COnnecting REpositories)
CORE (COnnecting REpositories)