Justification for selected research method

The research questions as set out in Chapter 2.9. focus this study on a single building element in in-situ conditions. The use of air-mixing fans in both co-heating tests and tracer-gas studies, preclude the study (or replication) of occupied dwelling conditions. Further, given the difficulty of isolating a single building element in most measuring methods, in-situ heat-flux measurements were considered to be the most suitable research method for the proposed research. Additional reasons in favour of in-situ U-value measurements are listed overleaf:

⁃ In-situ heat-flux measurements have been used extensively for the investigation of the thermal performance of a variety of construction elements, including a limited number of suspended timber ground floors - see Chapter 2.4.2. In this thesis, in-situ heat -flux measurements of floors will add to the other in-situ results, investigate comparison with models and increase knowledge about the thermal performance of floors.

⁃ In-situ heat-flux measurements have been used for some pre/post intervention studies, including for floors - e.g. Currie (2013) observed one point measurement on a floor pre/post insulation and Harris (1997) measured pre/post insulation floor performance in a test-cell - see Chapter 2.4.

Furthermore, a thermographic survey can support in-situ heat-flux measurement placement, while blower door tests will be used as a secondary research method where pre/post

interventions take place to investigate the effect of the floor intervention on the overall dwelling's air leakage - see Chapter 6.5.1. Given the previously described limitations of pressurising the floor void itself as well as the difficulty of mixing tracer gases in confined floor void spaces, the investigation of sub-floor air change rates is considered beyond the scope of this research and is noted for future research purposes.

3.2.6.1. Overcoming limitations

There are however several limitations of in-situ heat-flux measurements in general and more specifically in relation to suspended timber ground floors, including issues raised previously in Section 3.2.1. and Chapter 2.4., such as practical aspects of measuring in occupied houses;

limited floor measurements undertaken so far; measurement resolution issues and

comparability to models and other sources. Other potential issues are resource limitations, inability to capture all heat loss mechanisms and confounding variables. However, with careful research design, the impact of these disadvantages might be minimised. A summary of the main limitations are listed below and overleaf, alongside how these limitations might be controlled for or minimised.

1. Practical issues with measuring floors in occupied dwellings: daily usage of floor surfaces limits where and how many sensors can be placed as well as the monitoring duration. This could be minimised by monitoring in an unoccupied dwelling, a thermal chamber or test-cell, however close replication of a typical floor construction, its junctions and a realistic approach to space-heating would be required to reflect typical construction and occupation - see Section 3.4 and research design in Chapters 4.3 and 5.2.

2. Point measurements on a large surface area and difficulty to compare to models: one point U-value is unlikely to be representative of the total element U-value (ASTM, 2007a) - see Section 3.3.2., Chapter 4.4.2.5, 4.4.3. and 5.3. for further discussion. Undertaking multi-point (i.e. high resolution) measurements will aid the understanding of the spread of floor heat-flow. Doing so will also support the understanding of the applicability of point measurements in pre/post comparisons, and support the testing of point U-value averaging techniques to obtain whole floor U-values to compare to models. Use of infrared thermography can help with sensor placement and with estimation of whole-floor U-values - see Chapter 4.4.2.

3. Not all heat-loss may be captured, such as the impact of air leakage of the floor on dwelling heat loss, which is also excluded in floor U-value models. For pre/post retrofit measures, blower door tests might give an indication of the floor's air leakage and any impacts arising from interventions. Monitoring heat loss with open and sealed airbricks might indicate the impact of floor void ventilation on observed heat-flow - see Chapters 4.4.4. and 5.3.7.

4. Different measurement and analysis conventions and methods exist which could affect U-value determination - this will be further investigated; full discussion in Section 3.3. and in Chapter 4.4.5 and 5.3.3.

5. Time consuming to undertake and limited to the heating season: field studies are limited to the winter heating season, limiting the number of studies that can be undertaken as well as limiting the time-scale of pre/post intervention monitoring - see Section 3.4. Access to an environmental chamber would disconnect

measurements from the winter period, however this is subject to other limitations - see Chapter 4.3.

6. Short-term and seasonally changing variables as confounders in pre/post intervention field studies: changing external environmental conditions and seasonal changes of the ground will affect in-situ U-value estimates; there are potential confounding effects when undertaking pre/post intervention studies. This can be minimised by:

⁃ access to a thermal chamber with a replicated floor construction, such as the Salford Energy House; limitations of a thermal chamber are discussed in Section 3.4. and in Chapter 4.3.

⁃ analysing field data with dynamic methods instead of the more commonly used steady-state analysis techniques. Given that dynamic methods are not well

characterised for suspended ground floors, steady state analysis was applied in this thesis.

⁃ measuring other external variables such as solar radiation, wind-speed as well as void airflow, and ground temperatures and heat-fluxes might give additional useful insights to support understanding of possible confounding variables; see Chapter 6.3.2.

⁃ Access to an unoccupied control house over the same period of the study might be useful; an occupied control house may introduce other variables such as occupant behaviour and different heating patterns, affecting the observed heat-flow - see Chapter 6.

⁃ monitoring heat-flow over a whole year or longer might provide a year-average U-value and other useful insights into the different mechanisms affecting heat transfer; this would also be aligned to current model assumptions. However in practice access to case-study dwellings is normally short-term, making long term longitudinal studies highly unlikely, especially if also monitored at high-resolution.