Thermal bridge study questions default retrofit heat-loss values

Research led by the University of Salford indicates that generic thermal-bridge allowances may misrepresent predicted heat loss after domestic insulation retrofits. Scientists modelled a single 1930s solid-wall semi-detached house and compared junction-specific heat transfer with simplified default assumptions across external and internal wall insulation options and both staged and whole-house improvement programmes.

Salford publicised the findings on 12 October, following publication of the peer-reviewed study in Energy Science & Engineering on 9 September. The research, undertaken with Leeds Beckett University, examined the junctions where walls meet roofs and floors, and where windows intersect with the surrounding fabric. Insulating the principal building elements does not necessarily eliminate these additional pathways for heat flow.

A thermal bridge occurs where heat can flow more readily through a part of the envelope than through the surrounding insulated fabric. Junctions are especially important because a nominal wall insulation specification does not necessarily describe what happens where that insulation meets a floor slab, roof construction or window opening. Differences in material conductivity, geometry and insulation continuity can leave a concentrated route for heat to pass from the interior to the outside.

Conventional calculations can represent those junctions using linear thermal transmittance values, commonly described by the Greek letter psi. These values indicate the additional heat loss associated with a junction per metre of its length for a given temperature difference. Summed across the building’s junctions, they contribute to the overall heat-transfer estimate alongside heat passing through the main areas of walls, roofs, floors and other elements.

The study compared default thermal-bridge assumptions with junction-specific values produced by finite-element modelling. That approach uses a numerical representation of heat flow through the relevant construction geometry and material layers. It is more detailed than assigning one generic coefficient to every junction, but obtaining accurate model inputs can be demanding in older homes where materials, dimensions and construction details are not always documented.

Researchers considered external wall insulation and internal wall insulation, building regulations and EnerPHit targets, and staged or whole-house retrofit approaches. They modelled a single representative 1930s dwelling rather than measuring completed retrofit works across a sample of properties. The results therefore describe the performance of the selected building and scenarios, not an observed national average for housing retrofits.

Under the modelled whole-house approaches, the building’s overall heat transfer coefficient fell by up to 63% with external wall insulation and up to 61% with internal wall insulation, relative to its uninsulated baseline. These are reductions in calculated whole-building transmission heat loss, not measured energy bill savings and not the proportion of thermal bridges eliminated. Actual dwelling performance also depends on ventilation, airtightness, heating operation, workmanship and occupants’ use of the property, none of which can be inferred from those percentages alone.

Seven additional junction detailing measures, modelled in line with PAS 2035, reduced the remaining thermal-bridging contribution to 17.5 watts per kelvin, representing 16% of total transmission in the external wall insulation scenario. For the internal insulation scenario, the modelled contribution was 12.7 watts per kelvin, or 11%. Both figures relate to the selected building and specified whole-house construction details, rather than a representative average for the UK’s older housing stock.

The influence of thermal bridges changes as conventional heat loss through insulated surfaces is reduced. A poorly treated junction may then account for a greater share of the remaining transfer through the envelope, even though the overall building has become more efficient. Cold interior surfaces at such locations can also elevate the risk of surface condensation when internal moisture conditions permit it, creating a building fabric and occupant health concern beyond energy demand alone.

The research focuses particularly on the Reduced Data Standard Assessment Procedure, or RdSAP, which uses simplified assumptions to calculate energy performance for existing dwellings. Default figures make broad assessments practicable without requiring a numerical model of every floor-to-wall or window-to-wall connection. However, the approach may not reflect the construction of a particular property or the differences introduced when insulation has been added in stages over several years.

A database of junction values that differentiates construction type, junction geometry and retrofit detailing could allow assessors to choose inputs more representative than one general default. The researchers recommend this approach for typical building forms and PAS 2035-compliant scenarios, recognising that bespoke finite-element modelling for every dwelling would add cost and require fabric information that is often unavailable. Any such database would still need adequate underlying evidence and consistent assumptions about the proposed installation.

Funded by the Department for Energy Security and Net Zero, the study adds quantitative detail to the limitations of simplified retrofit calculations. It neither introduces a revised statutory assessment method nor establishes a measured failure rate for completed installations. The peer-reviewed results concern one modelled dwelling; applying them more widely would require values that reflect the range of construction assemblies, junction details and retrofit sequences encountered across the existing housing stock.