Traditional stone walls in Scotland - validation of RdSAP U-value calculation methodology: research

Appendix B: Discussion of potential factors influencing stone wall U-values

This section discuses some of the factors that may have influenced the in-situ measured U-value results obtained in this study, but that it was not possible to verify, either due to restrictions on the ability to carry out intrusive testing at the properties, or because there was insufficient sample data to identify or verify trends within the data. These may offer potential reasons for variations between the measured U-values and those calculated according to the RdSAP 10 approach.

A number of other published studies reporting on measured U-values of stone walls show similar disparity in results, and even if relevant results from the various studies were considered together for equivalent types of stone, it does not appear that there would necessarily be demonstrable trends against wall thickness that could reliably predict wall U-values for use in RdSAP calculations. ^{[6],[7],[8], [9]}

Uncertainties around wall construction

Actual wall make-ups for the test walls in this study were unknown and it was beyond the scope of the research to carry out intrusive investigations to accurately determine the construction. As stone walls become thicker, it may be expected that rather than being comprised solely of the primary type of stone and mortar, there may be rubble and/or voids within the centre of the wall with facing stone on the internal and external faces, which could influence the overall U-value. While the wall thickness may be measured, in-situ stone wall U-values may diverge from calculated values where the calculations are fundamentally based on the thermal conductivity of assumed types and densities of stone and the percentage of mortar, and where the presence of any voids within the wall are unknown and therefore not fully accounted for.

A generic correction factor is applied to RdSAP stone wall U-value calculations to allow for the presence of an internal wall lining. However, the nature and thickness of the lining can vary in practice (e.g. lath and plaster versus plasterboard), as can the depth of the air void that is introduced by battens or studs that support the lining material. These factors will influence the thermal resistance of the wall, yet this cannot be accurately allowed for without intrusive inspection to confirm the construction.

Assumptions regarding the thermal conductivity of stone

The standard BS EN ISO 10456:2007 presents thermal conductivity values for a range of different stone types, including sandstone and granite at 2.3 W/mK and 2.8 W/mK respectively. However, different sources (quarries) of stone may lead to differences in thermal conductivity. For example, a laboratory study^[10] of physical properties of various sources of Scottish sandstone produced ‘dry’ thermal conductivity λ-values in the range 1.07-1.94 W/mK. The moisture content of the stone can also have a significant impact; the same stone samples produced ‘saturated’ thermal conductivity λ-values in the range 2.41-5.65 W/mK. It follows that subsequent variation in wall U-values can be expected depending on the fundamental characteristics of the stone itself, and on the moisture content of the stone. Indeed, the latter may vary to some extent seasonally within the same wall due to natural cycles of wetting and drying, leading to a varying U-value over time.

Proximity to features that may introduce localised variations in thermal transmittance

Construction features of walls can introduce localised variation in heat flow, often termed ‘thermal bridging’. These include wall corners and adjoining separating walls, window openings and also the presence of timber studs at a wall surface supporting an internal lining. This study attempted to avoid being too close to features such as adjoining walls and window reveals so far as possible, though for some properties it was necessary to install measurement sensors closer to such features than would have been preferred. (In all cases, sensors were placed at least in excess of 300mm away from an adjoining wall or window.) Thermography was used to ensure that measurement plates were not located on areas of the wall with discrepancies in the construction that could significantly influence the heat transfer and give erroneous results, including the presence of studs in wall linings. (A stud-finding device was also utilised.)

When considering the results from individual measurement points on walls in this study, where one measurement plate from a pair was closer to a window reveal than the other, the U-value determined from the plate nearer the window tended to be marginally lower than the plate further away. However, the difference between the values from the respective plates was of the same order as from pairs of plates that were not near to windows, so it is not possible to confidently say that the proximity to windows influenced the overall average results from pairs of heat flux plates in this study.