Review of co-heating test methodologies (NF54)
The co-heating test has been developed to assess the amount of heat that leaks through the thermal envelope of a building – the ground floor, walls, roofs and windows, etc. It is currently the only test available for assessing whole buildings and providing a measure of as-built performance. This report shows results from a programme of work designed to evaluate the reproducibility of the co-heating test. Tests were carried out on a single house by the seven separate test organisations which partnered on this project. Results varied considerably, suggesting that more work was needed to ensure that the test was an accurate predictor of as-build performance. The work considered some effects which may be contributing to the variation in test result and how these might be controlled.
Summary of content
The co-heating test is an experimental method of determining a building’s ‘as-built’ heat loss coefficient (HLC), a parameter which is calculated by plotting the daily heat input against the daily difference in temperature between the inside and outside of the building. The test potentially allows deviations from the design performance to be identified by comparing the HLC, derived using the co-heating procedure, with the HLC determined through the Standard Assessment Procedure (SAP) using the building design parameter values.
In practice the reliability and practicality of the co-heating test method has been questioned due to the long test duration and uncertainty in the HLC. For example, a test duration is typically two weeks and during that time no access to the building interior is permitted. External weather parameters including temperature, wind and in particular solar radiation are major confounders. For these reasons it is recommended that the current approach to co-heating testing is at present unsuitable for large scale application across the construction industry.
This report describes co-heating tests undertaken by test teams from BRE and six project partners including BSRIA, Stroma Technology and four universities in one of BRE’s test houses, with a second identical test house used as a control. In addition BRE also undertook some further co-heating tests with various forms of window solar shading.
The test teams based their co-heating tests on a methodology previously developed by Leeds Metropolitan University. Although that guidance did not cover analysis and derivation of the HLC, all of the test teams used a form of regression analysis including a Siviour plot which is a method of graphically analysing the house heat balance to determine the whole house HLC. The test teams also made measurements in order to derive the solar aperture.
It was clear from the tests carried out that the external weather conditions, particularly solar radiation, represented a major confounder and had a major impact on the accuracy and repeatability of the co-heating test by making it difficult to achieve true steady state. The maximum uncertainty in the results from the co-heating tests and the SAP equivalent HLC was 17%.
A major factor in determining repeatability and accuracy was the spread in external temperature and solar radiation during the co-heating tests. In order to derive the solar aperture accurately it was necessary to obtain a large spread or range in external temperature and solar radiation values. This means that shortening the test duration very likely reduces its accuracy. It also follows that a long spell of consistent weather conditions (temperature and sunshine) with a small range or variation may have the effect of reducing accuracy and repeatability.
Additional co-heating tests with window shading appear to have reduced the uncertainty such that the measured values were within -3.8% of the SAP HLC. However, with just three tests carried out (each one undertaken with a different type of shading) this result cannot be considered to be conclusive and should be tested with other types of building and at other times of year. However, if it is subsequently shown that physical window shading is effective elsewhere then it would make the co-heating test a more accurate and therefore useful method of determining the as-built HLC. It is therefore recommended that the effect of external shading on reducing the uncertainties should be investigated further by trialling in other types of houses, other building types and under different weather conditions.
It has also been suggested that the analysis of co-heating test data for overnight hours only, might reduce the uncertainty caused by solar heat gains and might also reduce the test duration. The use of window shading may also contribute to the reduction in uncertainty from night data analysis by reducing the daytime solar heat gains stored in the building and therefore the effect on the night time thermal balance. More research and testing in buildings with a range of thermal mass would be necessary to prove the effectiveness of such measures, and also to determine the optimal day beginning and end times.