Assessing the impact of modelling variability on building performance simulation results

The UK’s goal of transitioning to net zero carbon buildings has led to an increasing focus on the reliability of modelling results for energy consumption. Meanwhile, there is a trend to apply the design-for-performance approach in order to predict and verify the actual operational performance of buildings. The modelling framework proposed by the latest edition of CIBSE TM54 (2022) is a viable route for this approach. However, the modelling uncertainty often results in discrepancies between actual operation and design expectations. This paper will test the uncertainty of building level performance modelling.

The impact of modelling approach variability, modeller variability and inter-model variability on the results will be investigated. In detail, a modelling task will be developed for practitioners based on a simplified school building. Two dynamic simulation approaches (template-level HVAC modelling and detailed component-level HVAC modelling), defined in CIBSE TM54, will be used to predict building energy consumption in two separate building simulation software (IES VE and Design Builder). Five people will be recruited for each of the different modelling approaches under each software to complete the modelling task and submit the results (20 modellers in total).

The input data and outputs from the models of the twenty participants will then be compared to identify the variations in building performance modelling. The impact of the level of detail in modelling HVAC systems on energy prediction will also be discussed. A quantitative analysis will be carried out to explore the underlying causes of the discrepancies and discuss the selection of the optimal modelling approach. Participants will be required to record their modelling time in order to make the trade-off between modelling reliability and time required.

Furthermore, a questionnaire will be developed to obtain feedback from the practitioners to provide a qualitative analysis of existing building performance modelling approaches in the construction industry. Thus, the findings will not only help to investigate the trade-offs between modelling reliability and complexity, but also provide an opportunity to get insightful feedback about the current application of building performance modelling approaches from practitioners’ viewpoint.