Dr Mark Collett

One of the initial steps in globally reducing energy demand involves accurately predicting energy requirements in the building sector and enhancing the thermal performance of buildings. Research indicates that the actual energy performance of building envelopes can differ significantly from initial predictions, often due to inconsistencies in building materials and poor workmanship. As a result, the heat exchange through building components often exceeds designed values, emphasising the importance of evaluating the as-built performance to gain accurate insights into the actual building behaviour. This discrepancy between predicted and the as-built actual energy usage is commonly referred to as the energy performance gap. Addressing this gap necessitates the introduction of performance indicators, such as the Heat Transfer Coefficient (HTC), a metric to quantify the actual as-built energy performance. Over the past 40 years, various methodologies and testing procedures have been developed to measure this indicator. These methods range from dedicated measurements performed on individual level to comprehensive assessments of whole-building performance, utilising in-use measurements in combination with data-driven approaches. Given the complexity and variability in these methodologies, and to reinforce their practical applicability, providing a comprehensive overview of the current state of the art in as-built thermal performance characterisation is of great importance. Therefore, this paper aims to summarise the most up-to-date advancements in this field, highlighting the strengths and limitations of existing methods, and uncovering gaps in robustness, reliability and reproducibility. By doing so, it seeks to provide a deeper understanding of the tools and techniques available to accurately assess the achieved thermal performance of buildings, supporting their future integration in practice.

The performance of building fabric is a critical component in the complex problem of heat decarbonisation. However, a growing body of evidence has shown that there is often a difference between the measured performance of building fabric and its predicted or design performance. This phenomenon is known as the performance gap. Awareness of the performance gap has popularised the concept of measuring fabric performance. This can be characterised by the U-values of individual fabric elements or the HTC (Heat Transfer Coefficient) that describes the whole house fabric performance. HTC measurements are commonly used to evaluate any performance gap. The current standardised method for HTC measurement, coheating, is extremely disruptive as it is required that a home is empty for 15 days or more for steady state conditions to be maintained within the property. As such, coheating can be considered unsuitable for uses outside of research. The QUB method is a dynamic method of HTC and U-value measurement that is completed within a single night. Owing to its short duration the QUB method could potentially be used in mainstream applications such as new build housing and retrofit where a coheating test would not be feasible. This research aims to improve and demonstrate the accuracy (closeness to true value) and precision (dispersion of repeat measurements) of the QUB method. This will identify where the method can be deployed to give informative measurements and its limitations. A method consisting of six field-based case studies was deployed in which repeated QUB measurements were completed and compared to reference HTC and U-value values to determine the accuracy and precision of the measurements. This revealed that variations in test conditions were impacting the dispersion of results and negatively affecting accuracy and precision. These included variability of the temperature ratios in unconditioned spaces, transient thermal mass effects introduced by solar radiation and changes in external temperature, and wind conditions impacting heat transfer. Where observed, this impact is linked to select building characteristics. Consequently, houses with minimal areas of indirect heat loss, insulated building fabric and not of a characteristically high thermal mass often resulted in highest levels of accuracy and precision in QUB measurements. From the results of the case studies, indicative values of accuracy and precision for the QUB HTC measurements were derived based on the building characteristics of the test homes. For homes with characteristics associated with high accuracy and precision the following levels of accuracy and precision are expected: root mean squared error (RMSE) ≤ 15 %, mean bias error (MBE) ≤ |13| %, relative range ≤ 19 % and standard deviation ≤ 7 %. For homes with characteristics associated with low accuracy and precision the equivalent metrics are RMSE ≤ 34 %, MBE ≤ |34| %, relative range ≤ 55 % and standard deviation ≤ 16 %. These values can be used by those conducting QUB HTC measurements to determine the suitability for an application and to provide context on the measurement result. The level of accuracy seen in the QUB U-value measurements was notably lower than that seen in existing work. The reasons for this could not be determined. A novel approach for adjusting HTC measurements for indirect heat loss through unconditioned spaces was proposed. This was done through use of additional temperature and heat flux density measurements and was shown to improve the accuracy and precision of measurements in all applicable instances. However, this practice may, in turn, affect the suitable use cases for the measurement. Future work should consider how these adjustments are communicated in the most understandable way. This study has demonstrated the accuracy and precision of the QUB test in a discrete number of real-world case studies. Whilst limitations of the QUB test are highlighted, its potential to give informative evaluations of building fabric performance are evident. Future work should conduct multiple QUB tests over a duration of close to one year to better understand the impact of changing test conditions. Additionally, the measurements completed in this study could be combined with data from other projects to enable an evaluation of accuracy and precision against a wider range of building characteristics. This will further the understanding gained from this study and give findings that can be generalised across the building stock.

Improved building fabric performance is essential for the decarbonisation of buildings. Evaluating fabric performance is often predicted; but inaccuracies are present within commonly used prediction methodologies. Accurate measurement of building fabric is therefore advantageous when identifying the improvement made through retrofit. The QUB/e method is a practical and effective method of measuring the whole building performance in low-to-medium thermal mass properties. In this paper, a property of high thermal mass was studied for the first time with the QUB/e method. The results identify challenges in undertaking QUB/e measurements in the application of high thermal mass including the impact of stored solar heat contributions resulting in a wider dispersion of measurements. In the case presented; a significant prediction gap is identified when comparing the predicted and measured results. The implications of the prediction gap observed include a change in the regulatory EPC band of the property. Additionally, using performance measurements would avoid overestimations of the reported decarbonisation and annual cost saving benefits of future retrofit works to improve the property at 2.3 Tonnes of equivalent CO2 emissions and £570 respectively.

Measuring the performance of building fabric is increasingly important as stakeholders wish to compare as-built performance with design expectations. When measuring whole house performance (Heat Transfer Coefficient) heat losses through the floor in slab-on-ground type constructions are intractable and introduce uncertainty into measurements. As such efforts are often made to isolate them from measurements. The QUB method is a practical method of measuring whole house building performance. Previous work has shown floor losses can successfully be isolated from measurement through use of heat flux measurements and additional calculation steps. To further test this isolation procedure, three QUB tests were performed on a slab-on-ground Passivhaus dwelling. Whilst the whole house performance measurements agree with the design performance (all results within 11% of the design) the floor losses measured appear unrealistically high. The conditions of the tests, conducted in late summer and in a highly insulated property, are likely causing the heat flux measurements to capture heat being stored in the floor construction rather than heat being lost from the property. Follow up measurements in more preferable conditions are planned which will assist in determining the cause of these observations.

In situ measurement can enable accurate evaluation of a building’s as-built performance. However, when measuring whole house performance, party walls introduce measurement uncertainty. Subsequently, it is common to “adjust” measurements to isolate heat transfer through party walls. This study explores the behaviour and impact of party walls in QUB and coheating measurements of a semi-detached house, presenting empirical evidence on the validity of these measurements where a party wall is present. Two different party wall heat transfer behaviours were observed through heat flux density measurements. Thermal charging is apparent in QUB tests and the initial stages of coheating. After 48 h of coheating, the party wall has become heat saturated and exhibits stable heat transfer. Consequently, using heat flux density measurements to isolate party wall heat transfer in QUB tests, where thermal saturation has not been achieved, can result in misleading inferences. The coheating and QUB measurements without party wall adjustment are in close agreement, irrespective of differing heating patterns in the neighbouring property. The generalisation of these findings is problematic since they describe the impact of the case study-specific built form and the test conditions. Future work to explore the impact of built form and test conditions is needed.

Forecasting retrofit performance often relies on building energy models, which can be inaccurate due to differences between predicted and actual performance. This introduces uncertainty to energy and carbon savings estimates. Research suggests that heat loss measurements improve model accuracy and provide more robust evaluations of retrofit effectiveness. Several whole-house heat loss measurement methods have undergone field trials for validation. However, the assessment of these measurements is limited to the specific field trial test conditions. Simulated measurements in a virtual environment could complement field trials by exploring conditions unattainable in real-world settings, increasing confidence in the measurements. This study replicated dynamic whole-house heat loss measurements from field trials in a calibrated energy model using local weather data. Differences of 7% (pre-retrofit) and 26% (post-retrofit) were observed between simulations and field trial results. The differences could be associated with the modelled heat dynamics not reflecting the true thermal behaviour of the house recorded in the field trials with a particular focus on thermal bridging heat loss. This study has shown that for simulations to be used in validating measurements, further work is needed to determine if the dynamic thermal behaviour of buildings can be replicated in simulations.

Thermal and hygrothermal simulations are undertaken to estimate energy performance, condensation risks, the potential for moisture accumulation, and timber rot. These simulations use default book values to estimate the material properties of solid brick walls. This report investigates the variability of brick properties found in solid walled homes in the UK and compares these to the default book values. It also explores how varying material property inputs in models affects thermal performance and moisture risk in solid walled homes.

Surveys and air tests were performed at 160 solid and cavity walled homes in Northern England, which had a mix of insulated and uninsulated walls. Blower door tests and Pulse tests were compared and used to quantify the airtightness of the homes. An evaluation of how building characteristics affected the results was performed, and common leakage pathways were identified. Data was also collected on the condition of the homes, potential barriers to external wall insulation (EWI) retrofit, as well as perceptions of occupants.

19BA is a mid-terraced pre 1900 solid walled home where airtightness improvements and room-in-roof retrofits have been installed. Building performance testing has been undertaken to collect data on the performance and risks of these improvements, and to evaluate the accuracy of modelled predictions on the retrofit performance and risk.

07LT and 09LT are two of fourteen case study homes retrofitted in the DEEP project. The case studies have been used to identify the performance of, and risks associated with, retrofitting solid walled homes. The data have also been used to evaluate the accuracy of the modelled predictions of the retrofit performance and risk.

08OL is one of fourteen case study homes being retrofitted in the DEEP project. The case studies are used to identify the performance of, and risks associated with, retrofitting homes without conventional cavities. The data from the case studies are used to evaluate the accuracy of modelled predictions of retrofit performance and risk.

27BG is one of fourteen solid walled DEEP case study homes. In this home building performance tests were undertaken to investigate the success and risk of retrofitting suspended timber floors and how the results compare to predictions.

Retrofitting solid walled homes is one of the greatest challenges for the UK in achieving its net zero ambitions. Solid walled homes have unique features, that require special consideration. They are among the least efficient in the UK, and their occupants are more likely to be in fuel poverty. They are also at elevated risk of surface condensation, excessive cold in winter and overheating in summer. Retrofitting these homes is a cornerstone of UK policy to tackle fuel poverty and to facilitate the delivery of decarbonised electrified heat into homes. However, installing solid wall insulation is costly and poses more risks of unintended consequences than any other retrofit. Previous projects investigating solid wall insulation have identified major failures when retrofits are installed in a ‘piecemeal’ way i.e., they did not consider how the retrofit measure affects risks of damp, inadequate ventilation, and overheating in homes. This led to the adoption of the whole house approach in new technical standards for retrofit installers (PAS 20351) to ensure that all risks of retrofit measures were always considered, even if only one measure was being installed at a time. Industry is beginning to adapt to these standards, but more research is needed to explore the benefits of adopting the whole house approach, and more guidance is needed to support retrofits in solid walled homes. Insights from this project explain how solid walled homes can be retrofitted more safely and effectively.

The DEEP case study retrofits provide compelling evidence on how a whole house approach to retrofit can reduce heat loss, surface condensation risk and overheating risks in solid walled homes. From the data collected, specific guidance is produced outlining how to install retrofits in solid walled homes more safely and effectively. Recommendations are provided on how to make measurements and modelling predictions of the technical performance of retrofits more accurate. The findings can inform evidence-led decisions at multiple levels to ensure retrofits in solid walled homes are safe and effective.

17BG was one of fifteen case study homes retrofitted in the DEEP project. The case studies were used to identify the performance of, and risks associated with, retrofitting solid walled homes. The data from the case studies was also used to evaluate the accuracy of modelled predictions around retrofit performance and risk.

56TR is one of fifteen homes being retrofitted in the DEEP project. The case studies are being used to identify the performance of, and risks associated with, retrofitting solid walled homes as well as to evaluate the accuracy of retrofit models.

01BA is one of fourteen case study homes retrofitted in the DEEP project. The case studies identify the performance of, and risks associated with, retrofitting solid walled homes. A retrofit was undertaken in stages, reflecting a piecemeal approach to retrofit, followed by undertaking activities that would be required for a whole house approach as a final stage. The data from the case studies is also being used to evaluate modelled predictions of retrofit performance and risk.

55AD and 57AD, are a pair of identical semi-detached homes, and are two of fourteen DEEP case study homes in which the comparison between a whole house and piecemeal approach to retrofit was evaluated.

00CS is one of fifteen case study homes retrofitted in the DEEP project. The case studies were used to identify the performance of, and risks associated with, retrofitting solid walled homes. The data from the case studies was used to evaluate the accuracy of modelled predictions around retrofit performance and risk.

04KG is one of fourteen case study homes being retrofitted in the DEEP project. The case studies are being used to understand the performance of, and risks associated with, retrofitting solid walled homes. The data from the case studies is also being used to evaluate modelled predictions of retrofit performance and risk.

52NP and 54NP are two of fourteen case study homes retrofitted in the DEEP project. The case studies were used to identify the performance of, and risks associated with, retrofitting solid walled homes. The data from the case studies were also used to evaluate modelled predictions of retrofit performance and risk.

This report describes the common data collection and analysis methods used in the DEEP retrofit case studies. These are generically described to avoid repetition in the individual case study reports.

The York Passivhaus is a 3-bed home in York, North Yorkshire, that achieved Passivhaus certification on completion in 2015. The project aim is to evaluate the building fabric and system performance of the home seven years post-completion against design targets and initial performance tests. Areas of interest are energy consumption, ventilation and air quality, thermal comfort, airtightness and building fabric. Looking at these in turn, fuel bills were used to explore how gas and electricity consumption had changed since occupation in 2016. Gas use was higher during the first year postcompletion in 2016 but has steadily declined since. Electricity use has remained relatively constant. The annual energy consumption in 2023 was 2467kWh for gas (20kWh/m2/year) and 1652kWh (13kWh/m2/year) for electricity, which is between 60 and 74 per cent less for gas and between 9 and 39 per cent less for electricity than the average UK house. The mechanical ventilation heat recovery (MVHR) system was not balanced when flow rate test results were compared against commissioning figures, as extract air flow rates were higher than intake air flow rates. This meant that the system no longer satisfied Passivhaus requirements. Air quality was monitored inside and outside of the home over 12 months. For CO2, a high level of IAQ was recorded, with an average of less than 872 ppm. CO2 levels dropped when the MVHR filters were changed coupled with the onset of warmer weather. Higher noise levels associated with the MVHR system ceased following a service. Higher levels of particulate matter (PM) were recorded at the front of the house, close to a car parking area. Three peak periods were examined to see how particulates generated externally or internally rose and fell over time. Spikes in internal PM levels were generally due to cooking or use of the woodburning stove and dissipated quickly. Elevated PM level patterns recorded outside were often mirrored inside but at a much lower level. Twenty internal sensors monitored temperature and humidity levels. Temperatures remained constant above 15°C throughout winter with all sensors staying within a 3-4°C range, indicating a low level of thermal variation across the home. However, internal temperatures were quite low – usually under 20°C, despite the space heating system defaulting to set points of 24°C during the day and 15°C at night during the winter months. This suggests that the space heating system was undersized for the current occupancy level, as design calculations were based on higher occupancy assumptions. It was assumed at the design stage that the wood-burning stove would meet 30 per cent of the home’s heating demand when during the monitoring period it was rarely used. During warmer weather, higher temperatures were recorded across the two southwest facing first-floor bedrooms. There was no evidence of overheating when the home was occupied during warmer weather. In general, the house is still extremely airtight with a mean permeability of 0.86 m3/(h.m2) @50Pa. However, this is a significant increase in air leakage in relative (rather than absolute) terms since certification was carried out in October 2015, where a mean permeability of 0.39m3/(h.m2) @ 50Pa was recorded. The little air leakage detected appears to come from window seals at casements, the boiler flue, plus some air movement behind plasterboard in the upstairs rooflights, and at wall-to-ceiling, or wall-to wall-junctions. The air leakage area has increased only slightly – from around 73cm2 to 104cm2. Therefore, after seven years the home now satisfies EnerPHit rather than Passivhaus airtightness requirements. A QUB test was used to measure fabric performance. First, a design-stage heat transfer coefficient (HTC) for the home was calculated, which was 69.5 W/K and then tested against. Three tests were done in the summer/autumn of 2022 and two in the winter of 2023. The average measurement was 76.3 W/K. This is a low HTC but 10 % greater than the designstage performance calculation. Overall, as a seven-year-old Passivhaus, the home’s performance is still exceptional compared to current-day new-build homes. Some performance aspects have deteriorated since completion, such as the airtightness and MVHR performance, which could be associated with wear and tear. It is not possible to compare changes to air quality, thermal comfort and HTC, as they were not monitored post-completion. The only area of note is thermal comfort in winter depending on the temperature sought by occupants, as the space heating system is not designed for the current occupancy level and could be considered on the cool side of comfortable.

Whole house heat loss or heat transfer coefficient (HTC) measurements are rarely undertaken to validate the performance of retrofits installed in homes. This means policy, certification and householders must rely on predictions made by energy models. Multiple domestic energy models exist, with varying underlying rules and input requirements. This means predictions made by different models may not always agree. However, few studies have compared the predictions from these models with each other, and with measured whole house heat losses for a home before and after a retrofit. This paper compares the HTC of a three bed, semi-detached, solid-walled home measured via the coheating test, with the HTCs predicted by the Reduced Data Standard Assessment Procedure (RdSAP), Building Research Establishment Domestic Energy Model (BREDEM), Dynamic Simulation Modelling (DSM) and the Passive House Planning Package (PHPP). The results show that most predicted HTCs from the models are not similar to the measured HTC, and there is a large variation between the different modelled HTCs. The paper explores why these differences occur and reflects on how to improve the accuracy and consistency of domestic energy models.