2.3 Post Construction Assessment
2.3.2 Quantifying Whole House Heat Losses
As discussed in Section 2.1.2, heat losses through building fabric can be extensive and have a significant effect on space heating demand. However, the evaluation of such losses is not a simple task, and it has been acknowledged that it is almost impossible to undertake repeatable and reliable in-situ thermal performance testing of constructed buildings at a macro-level (Judkoff et al., 2001).
Short term energy monitoring (STEM) tests are the most commonly used technique, and have been utilised for a number of years in order to assess post-construction building performance (Wouters et al., 2005). The majority of such tests that have been undertaken are related to research projects, as the work involved is time and resource intensive and requires long term access to an empty property (Department for Communities and Local Government (DCLG), 2011a).
The primary technique used in the UK for as-built thermal performance testing is the steady state coheating test (Good Homes Alliance (GHA), 2011a). This methodology was first developed in the USA over 20 years ago (Sonderegger et
al., 1980), but it is only more recently that a semi-standardised protocol has been developed by Leeds Metropolitan University (Wingfield, J. , 2011).
Techniques of this type involve using electrical heaters to maintain a constant internal temperature in a building (Figure 2-15 (Stamp, 2013)), and measuring the power input of the heaters for a number of consecutive days (Department for Communities and Local Government (DCLG), 2011a). Generally, a time period of seven to ten days per test is recommended (Homes and Communities Agency (HCA), 2010). This minimises the effect of fabric heat storage effects and stabilises the variation between measured air and radiant temperatures (Everett et al., 1985).
Figure 2-15 - Principles of Coheating Test Procedure Source: (Stamp, 2013, Web)
The output from a coheating test is a measurement of the as-built HLC of a property. As indicated by Johnston (2013, p. 5), the calculation of the HLC can be undertaken using a rearranged form of the standard heat balance equation (Equations 2-3 and 2.4).
+ . = ( . + 1/3
(HLC) = ( . + 1/3 .
Where:
= Total measured power (W)
= Solar aperture (m2)
= Total south facing solar radiation (W/m2) . = Total fabric heat loss (W/m2)
= Background ventilation rate (h-1)
= Internal volume of the dwelling (m3)
= Temperature difference between the inside and outside of the dwelling (K)
The total measured power value is recorded directly by a power meter attached to each heater and fan, whilst internal temperatures and external temperatures are recorded using thermocouples. The circulation fan is employed in order to encourage mixing of hot and cold air, in order to prevent stratification and maintain an even temperature throughout the test dwelling. Individual thermostats are connected to each heater in order to regulate internal temperature. Solar radiation is measured either directly by a site-based pyranometer, or obtained from a local weather station data. The analytical techniques only require input of the south facing solar radiation values, although, depending on the orientation of the building, this could underestimate the overall impact of solar gains to the east and west elevations.
The raw data from the coheating test provides a measurement of the total heat input from the heaters required to maintain a uniform internal temperature.
However, the effect of solar gains needs to be accounted for, as less power may be required to heat the dwelling on days with high levels of solar radiation (Jenkins et al., 2013).
There are several techniques that are commonly used to assess the effect of solar gains on the whole house HLC. The Siviour Method is a linear regression
Equation 2-4 – Rearranged Heat Balance Equation (Calculation of HLC (W/K) (Johnston, 2013, p. 5) Equation 2-3 – Standard Heat Balance Equation (Johnston, 2013, p. 5)
method, presented by Siviour over twenty years ago. The parameter is plotted against in order to obtain the solar aperture ( in m2), which is represented by the slope of the line. The y-intercept is equal to the total solar corrected HLC (W/K) (Siviour, J., 1981).
A modified version of this methodology, referred to as thermal calibration and shown in Equation 2-5 (Johnston, 2013, p. 6) was developed by Everett (1985) in order to isolate floor heat loss values where this building element was considered to have different thermal properties to the rest of the dwelling (such as a solid concrete slab foundation).
F - 1/3 (HLC) = . .
Where F = Total Ground Floor Heat Loss (W)
When the data is plotted using this technique, the slope of the line represents solar aperture ( in m2), while the y intercept equals the total solar corrected HLC (W/K), excluding ground floor heat losses.
When it is not possible to determine the solar aperture through use of either Siviour or thermal calibration analysis, it may be appropriate to obtain this value through either manual or computer-aided calculation. This requires information relating to total glazing area, and values of solar transmittance, solar access factor, frame factor and average incidence factor. The solar aperture value can be used with the mean solar radiation data to calculate mean solar gains in order to adjust the measured raw power input. The corrected data can be used to calculate the solar corrected HLC through linear regression (Johnston et al., 2013).
Equation 2-5 - Thermal Calibration Equation (Adjusted Calculation of HLC W/K) (Johnston, 2013, p. 6)
Multiple regression analysis provides an alternative to simple linear regression techniques, and can be used to obtain solar aperture data. Solar gains and temperature values are regressed against raw power input in order to obtain the solar aperture, represented by the y intercept in the resultant statistical tables. As in the methods described previously, the solar aperture and solar radiation data is used to calculate solar gains, which are then added to the measured raw power input in order to obtain the total HLC through linear regression (Lowe et al., 2007). Other influential independent factors, such as rainfall and wind velocity, can also be included in the multiple regression model in order to evaluate their effect on the HLC value.
In all of the methods described, the calculated solar aperture ( ) value is applied to the solar radiation data in order to obtain a value in watts for solar gains to the property. The original value of measured power ( ) is then adjusted to reflect the true amount of electrical power required to heat the property, through addition of the solar gains value. This can be of particular importance when undertaking coheating tests in the autumn and spring months, when levels of solar irradiance could potentially be high and may significantly reduce the amount of energy required by heating to maintain a constant internal temperature (Miles Shenton et al., 2010).
In a study of the reliability of coheating test measurements, Bauwens (2012) found that multiple regression provided the most reliable technique in order determine solar aperture. This was due to the ability of the statistical model to allow for experimental error in all of the variables. The work also provided evidence that the HLC generated as a result of the coheating test and multiple regression analysis was reliable, when it is assumed that are independent variables and a zero x/y intercept is used when plotting the data.
However, when setting the intercept at zero, this reflects a situation where there is no power input and no difference between external and internal temperatures. In reality, this relationship is not strictly linear due to factors such as thermal lag in the building fabric and night time heat dissipation/cooling
strategies, which could mean that even when there is no power input, there may be a difference between internal and external temperatures (Bauwens et al., 2012).
An alternative methodology to the coheating test is The Primary and Secondary Terms Analysis and Renormalisation (PSTAR test). This is a slightly different whole house heat loss analytical technique, which utilises the steady state coheating test within its methodology. It was developed by the US National Energy Research Laboratory (formerly the Solar Energy Research Institute) in the 1980’s, and detailed explanation of the process is given in publications by the institution (Subbarao, K. , 1998; Subbarao, K. et al., 1989; Subbarao, K. et al., 1998).
In simple terms, a model is constructed using energy simulation software and information collected relating to the building construction and location/position.
Data relating to building permeability, heat flows through building elements and thermal bridging is obtained through experimentation, followed by a short term heating test and cooling down period. The data collected is used in linear regression techniques to renormalize the building heat flows within the original model developed for the building. Alongside total HLC and solar aperture values, the method also accounts for thermal mass effects (Carrillo et al., 2009).
There are three types of ‘term’ defined within the methodology, as shown in Figure 2-16, which are used to realign the standard heat loss equation parameters into a renormalized form that represents the performance of the building under assessment.
Figure 2-16 – Summary of Key Terms and Equations in PSTAR Methodology (Produced by Author Based on Palmer et al. (2011))
The renormalized parameters (represented by ‘p’) are those that commonly contribute to the lack of agreement between the building model and test results, and are calculated using statistical linear least squares fit analysis (linear regression) (Chun et al., 1997). The test has been shown to be repeatable and predictable and to give reasonably accurate results (Burch et al., 1989).
In a study undertaken to compare the PSTAR and steady state coheating test methodologies, a 30% difference between the HLC values calculated using each technique was noted (Palmer et al., 2011, p. 61). This was partly explained by differences in the state of the property tested due to a year-long interval between the two tests, but also due to the impact of changing thermal conditions that are taken into account in the PSTAR test and not in the coheating test. The research concluded that the coheating test provided a steady state HLC that has characteristics similar to that derived from models such as SAP, and therefore can be used to evaluate design/construction conformity and compliance. The PSTAR test HLC is more comparable to dynamic thermal modelling outputs, as it accounts for the effect of thermal mass.
Whilst it is acknowledged that the PSTAR testing methodology is rigorous and provides a good indication of whole house heat losses, the coheating test protocol has been selected for use in this research. This is largely because the majority of the studies completed in the field of housing performance in the UK have utilised this technique, and so there is more comparative data and analysis available relating to this method. It also provides a HLC value that can be used for comparison with steady state design stage outputs such as those derived from SAP.