Energy Performance of Buildings (Scotland) Regulations 2025: stock model research

4 Methodology

The following methodology was developed and implemented to produce the required results to the specification of the Scottish Government’s needs.

4.1 Preparation and ingestion of Scottish housing data

The dataset used for this analysis is based on the Scottish House Condition Survey (SHCS) for the years 2019, 2022, and 2023. This data source was selected as the best available dataset to provide a representative sample of Scottish housing stock. The SHCS is an annual survey which consists of an interview with householders and a physical inspection of the dwelling they occupy, which provides a picture of Scotland's occupied housing stock. It covers all types of households and dwellings across the country: whether owned or rented, flats or houses. The physical data about the dwelling is recorded by surveyors trained to collect detailed information on housing characteristics. This is combined with information about the household collected through the face-to-face interview, covering a range of topics such as household characteristics, tenure, neighbourhood satisfaction, dwelling satisfaction, health status and income. The result is a unique and powerful data set for examining the condition and characteristics of Scotland's housing stock alongside the views and experience of the people living in those dwellings. This ensured a robust foundation for our modelling.

The SHCS data set includes a SAP rating and a SAP rating band, produced using RdSAP 9.93 as part of the physical survey. The available physical inspection data for each dwelling also includes RdSAP compliant energy survey data. Before analysis, this raw SHCS data was processed and mapped to the full RdSAP 9.94 and RdSAP 10.2 formats as required by our bulk stock modelling platform. The SHCS data is aligned with the RdSAP framework. In some instances informed assumptions had to be made, referencing other available data points to ensure the most accurate translation possible.

The SHCS SAP ratings had been calculated using RdSAP 9.93 and did not fully align to the RdSAP 9.94 or RdSAP 10.2 SAP ratings calculated from this adjusted input dataset using Cotality’s BRE-accredited RdSAP engine. To ensure that the SAP ratings and the HEM outputs were comparable, the Cotality-calculated SAP 9.94 ratings were used as the main point of comparison with the HEM HRR results to maintain referential integrity within the research design.

The RdSAP calculation methodology has recently been updated (on 15 June 2025) from version 9.94 to version 10.2. Cotality’s brief was to use the version 9.94 for the main comparisons in this report.

4.2 Energy calculation methodology

The version of HEM (0.34) used for this analysis is still under development by DESNZ, and although the latest version of the Future Homes Standard wrapper (0.25) is designed for calculations on new homes, at present there is no approved HEM methodology for existing homes. RdSAP is the SAP methodology for existing homes, there have been various versions with RdSAP 9.94 having been superseded by RdSAP 10.2 in June 2025.

In this report HEM may be used as a shorthand for HEM in combination with a ‘wrapper’ (for example when referring to the HEM HRR). A wrapper is a separate component that defines the inputs, assumptions and outputs needed for a specific use case, such as producing a Scottish EPC certificate. Wrappers allow the same HEM core calculation engine to be used for many different purposes, without having to reimplement the underlying physics model.

Cotality has an RdSAP engine, accredited by BRE, and built with bulk processing in mind. The project made use of the Scottish EPC wrapper (commit 2248e2d) and HEM (0.34) to run the calculations, in combination with an amended version of RdSAP within the Cotality engine. The HEM input data format has strong commonalities with the SAP input data format, so it was possible, with assumptions, to populate a HEM dataset using the RdSAP dataset. This data development work is described in detail in the next section.

When this development work was completed, Cotality were able to run calculations on the dataset to produce both SAP results and HEM outputs, forming the basis of the analysis presented above and in this report.

4.3 Batch processing using the Home Energy Model

At present there is no HEM methodology for existing homes. We addressed this as follows:

• The Standard Assessment Procedure (SAP) requires a complete "full SAP" dataset to perform calculations. However, for existing dwellings, it is generally not possible for an energy assessor to collect all the necessary information, such as the specific U-values of every building element or precise details of thermal bridging. Consequently, a reduced dataset (RdSAP) is collected. This dataset is then expanded into a "full SAP" dataset by applying the standardised assumptions and conventions set out in the official RdSAP specification applicable at the time of the property survey.
• The SHCS property data (collected for existing homes) did not contain all input variables required to perform a HRR calculation using the Scottish EPC wrapper. However, HEM inputs are very comparable “full SAP” inputs, and as a result it was possible to adapt Cotality’s RdSAP engine to produce HEM inputs. This required some assumptions, including to bridge between the two methodologies, and these are described further in the next section.

As with RdSAP versus SAP, the Scottish EPC wrapper required a number of data fields that were not present in an RdSAP dataset, in the RdSAP methodology for generating “full SAP” data from RdSAP, or the original SHCS data. These included the orientation of the property, nearby and distant shading, and lengths of pipework. We therefore again needed to make a number of assumptions to put together a complete dataset for each property, and these are set out in section 7: Appendix: RdSAP to FHS Assumptions.

Following this integration into our batch processing system it was then possible to calculate both SAP and Scottish EPC wrapper metrics for the approximately 9,000 properties in the SHCS dataset.

Note that there were various bugs and challenges encountered in implementing the wrapper. This was to be expected, given that the versions of HEM and the HEM wrapper that were used were still in the process of development. In particular, there was an unexpected finding that the combined HEM + wrapper package used for the analysis was unable to calculate metrics other than the HRR for less common heating systems and fuels, which had some impact on the results.

The various metrics required that the model be run with different configurations, most notably:

• All ratings required that the model be run for a nominal central Scotland location (‘East Scotland’ climate region). This is a variation from the assumptions used by SAP (which uses East Pennines climate region as a basis for calculations).
• The HRR assumed a standard nominal heating system (on-peak electric heating) to prevent the heating system itself affecting the result, which is intended to reflect the fabric performance without any effects that may be related to the characteristics or performance of the heating system.

4.4 Detailed results

A detailed analysis of the results of the batch processing under HEM is presented in the following sections. In addition, we have provided the raw results in a spreadsheet that accompanies this report.