Paper Conference

Proceedings of uSim Conference 2020: 2nd uSim Conference of IBPSA-Scotland


Balancing accuracy and computation burden – an evaluation of different sensitivity analysis methods for urban scale building energy models

Pamela Jane Fennell, Paul Anthony Ruyssevelt, Ivan Korolija

Abstract: Urban-scale building energy models capitalise on the increasing accessibility of large-scale urban data sets and allow the rapid evaluation of competing policy options, making them a vital tool for urban responses to the climate emergency. However, the vast number of different inputs required to model a complex urban environment makes it impossible to precisely quantify all inputs and the complex energy flows within models must be simplified to achieve tractable solutions, as a result, the outputs of these models inevitably have a significant range of variation. Without understanding these limits of inference resulting policy advice is inherently defective. Uncertainty Analysis (UA) and Sensitivity Analysis (SA) offer essential tools to determine the limits of inference of a model and explore the factors which have the most effect on the model outputs. Despite a wellestablished body of work applying UA and SA to models of individual buildings, very limited work has been done to apply these tools to urban scale models. This study presents a systematic comparison of three different sensitivity analysis methods for a high resolution, dynamic thermal simulation at the neighbourhood scale. Accuracy, processing time and complexity of application of each method is evaluated to provide guidance which can inform the application of these methods to other urban and large-scale building energy models. The results highlight the importance of considering both model form and input parameter scale when selecting an appropriate method. In this case, the elementary effects method (EER) offers good performance at relatively low simulation cost.
Pages: 102 - 109