Closing the Performance Gap in Building Energy Modelling through Digital Survey methods and Automated Reconstruction

Against the backdrop of increasing global efforts to mitigate the effects of climate change there has been a large focus on the Built Environment. The low level of building stock turnover in the UK, estimated between 1 3% per annum, has reinforced the importance of robust retrofit programmes to meet legislated targets [1,2]; Experts predict that the majority of existing UK buildings will still be in use in 2050 [3]. Residential and commercial buildings account for approximately 20% of energy end use globally with UK industry building services such as space heating and lighting account for between 6-56% of overall building energy use, depending on sector [4]. Building Energy Modelling and Simulation (BEMS) software is used to assess the energy performance of a building based on knowledge of its construction, design, use and location. While design data is readily available for new buildings, existing buildings, that are in need of retrofit, tend to have limited as-built building data. This requires a collection of data through site surveys and manual creation of building models; This is a time consuming and expensive activity. The aim of this research was “Develop a scientific method to remove barriers to urban scale Building Energy Modelling and Simulation (BEMS) using pattern recognition software to extract built forms from large data sets”. This research has developed a process of rapid geometry generation for BEMS applications to substantially improve this workflow. Following an internal site survey, a Point Cloud was produced of a case-study building. This was automatically processed to create recognisable building geometry for BEMS applications that achieved time savings of 85% over traditional methods. It was identified that internal survey methods present limitations to the automated reconstruction process and that existing offerings for UAV mounted survey equipment required high capital expenditure. A low-cost prototype for external scanning underwent initial development and identified areas for further development. The geometry that was reconstructed via internal survey data was simulated in BEMS and compared against measured energy data. The annual energy use was predicted to within 6% of the measured energy data. Limitations to a full reconstruction led to a hybrid approach being conducted. The hybrid approach predicted annual energy use to within 4% of measured data and met industrial validation requirements. The research conducted has demonstrated that improvements to the BEMS workflow can be achieved and in doing so it can contribute to the reduction in emissions from the Built Environment.