1. Introduction
Efficiency in terms of energy and resources is required by the European Green Deal when building and renovating the European building stock [
1] for achieving the European Union’s energy efficiency targets [
2]. To meet the 2050 long-term climate and energy targets, 97% of the residential building stock, which is responsible for around 36% of CO
2 emissions in the European Union [
3], needs to be upgraded to become highly energy efficient and obtain the Energy Performance Certificate (EPC) label A [
4]. This high percentage is due partly to the average age of European residential building stock: more than 80% of residential buildings are over 25 years old (
Table 1) and, therefore, most of them were built without thermal regulation. Construction is considered the sector with the most potential for energy saving. Buildings represent 40% of the energy used in the European Union [
5]. Almost 26% of Europe’s total final energy consumption is used in residential buildings [
6]. Space heating represents a large amount of energy consumption in the housing sector [
7,
8] and, therefore, has a significant capacity for energy saving.
According to the IEA EBC Annex 71-project [
10], measuring the actual energy performances discloses a notable gap between actual and expected energy performances of buildings. The actual thermal behaviour of building envelopes could contribute to this energy performance gap in buildings. Deviations between the predicted and actual behaviour of envelopes are related to aspects of the design and construction stages [
11]. Moreover, assumptions about energy efficiency improvements resulting from building refurbishment are not always met [
12,
13]. Consequently, accurate in situ measurements of the actual U-value of façades are necessary.
The thermal performance of building façades is a fundamental parameter that should be evaluated to obtain an accurate energy diagnosis of buildings [
14,
15,
16,
17,
18,
19,
20,
21]. Successful decision-making during energy renovation processes of existing buildings requires precise characterization of thermal properties of building components [
22,
23]. Classifying the opaque part of the façades is essential to precisely analyse the thermal performance of façades in the housing sector.
In recent years, the thermal behaviour of the materials that compose façades had been studied in depth by several authors such as Laaroussi et al. [
24] measuring the thermal properties of brick materials, Kuman and Suman [
25] or Björk and Enochsson [
26] measuring the properties of thermal insulation materials. However, when analysing the thermal transmittance of walls, assumed U-values have been a meaningful source of error in estimations of energy savings and carbon emissions [
27]. Evidence suggests that assumptions concerning heat loss from a dwelling pre-retrofit and post-retrofit are not correct [
12]. Therefore, accurate on-site measurements are required to provide information on the actual thermal transmittance of façades. To effectively quantify the actual performance of buildings, optimized on-site measurements combined with dynamic data analysis techniques are needed [
28].
Several methods can be used for the in situ measurement of U-value of existing buildings’ façades [
29,
30]. One of the most common is the heat flow meter method, standardised by ISO 9869-1:2014 [
31]. This method obtains the thermal transmittance by measuring the heat flow rate that passes through a wall and the inside and outside environmental temperatures. However, difficulties can arise in on-site measurements of walls U-value in the existing building stock, leading to inaccuracies [
32]. These difficulties can be classified into three groups, according to the IEA EBC Annex 58-project [
33]: errors related to the measurement accuracy, errors related to the analysis of data and errors related to the boundary conditions of the in situ measurement. Difficulties related to the measurement accuracy include factors like reading and calibration of heat flow meters and temperature sensors and had been deeply analysed by authors such as Ficco et al. [
17], Trethowen [
34] and Meng et al. [
35]. Difficulties related to the analysis of data were recently analysed in depth [
32,
36,
37,
38,
39]. Difficulties related to the boundary conditions include factors as an imbalance of the heat flow, edge heat loss and accuracy on the position of sensors, which were highly analysed by Peng and Wu [
32], Meng et al. [
35], Cesaratto et al. [
40], Ahmad et al. [
41] and Guattari et al. [
42]. Delving into boundary monitoring conditions, the factor of contact between the wall surface and the heat flow meter was analysed by simulations but has not been analysed in experimental tests [
40], so as for the present study focuses on this aspect.
The IEA EBC Annex 58-project [
33] and authors such as Cesaratto et al. [
40], Tadeu et al. [
43] and Gori and Elwell [
44] highlight the importance of ensuring good thermal contact between the heat flow meter plate and the wall to be measured to obtain accurate results. However, conducting HFM in situ measurements during the operational stage of dwellings is challenging due to the need for avoiding damage to wall surfaces. In this sense, the usual practice is placing a PVC film to preserve the wall surface [
27,
45].
Considering all the aspects aforementioned, the study aimed to analyse the influence of the non-perfect thermal contact of heat flow meter plates on the accuracy of in situ measurement of the façades’ U-value when a film was applied to avoid damage to the wall surface, as a usual equipment installation procedure during the building’s operational stage. This research provides valuable additional evidence on the accuracy of in situ monitoring of the actual U-value of existing buildings’ façades and will therefore assist practitioners in pre-retrofit diagnosis.
The method used to analyse the influence of the non-perfect thermal contact of heat flow meter plates on the accuracy of in situ measurement of the façades’ U-value is explained in the second section following this introduction. The third section presents the results. Finally, the discussion and conclusions are given in the fourth section.
3. Results
The data acquisition process was conducted following the indications in
Section 2, with a sampling duration of 168 h in Case study 1 (337 readings) and of 144 h in Case study 2 (289 readings). Two measured thermal transmittances and its confidence interval were calculated, for 24-h test cycles using the dynamic method in the two case studies [
54]. One value was calculated using data from the heat flow meter plate installed with a layer of thermal interface material directly on the wall surface and the other value was calculated using data from the heat flow meter installed with a PVC film between the layer of thermal interface material and the wall.
The minimum test duration was checked considering the ISO standard [
31]. The first requirement is that the sampling duration must be an integer of 24 h and at least 72 h (Equation (3)). Consequently, the second and third requirements were verified from the third day onwards. The second requirement for test completion is that the value of thermal transmittance obtained at the end of the sampling duration shall not deviate more than 5% from the value obtained 24 h before (Equation (4)). In accordance with this condition, in Case studies 1 and 2 the test could be ended after 72 h, as the requirement was met for all cycles with both HFM installation methods. The last requirement for ending the test is that the U-value obtained by analysing data from the first 2/3 of the sampling duration shall not deviate more than 5% from the value obtained from the data for the last period of the same length (Equation (5)). According to this condition, in Case study 1 the monitoring process could be ended in 96 h using an HFM installed directly on the wall surface and in 120 h when an HFM was used with a PVC film between the layer of thermal interface material and the wall surface. In Case study 2, the test could be stopped at 72 h when the heat flow meter was installed directly on the wall surface and in 120 h when the heat flow meter was installed with a PVC film between the layer of thermal interface material and the wall surface. The minimum test duration when using an HFM installed directly on the wall surface was found to be shorter than when using an HFM installed with a PVC film between the layer of thermal interface material and the wall surface.
The results of the measured thermal transmittance for the two case studies are shown in
Table 5 and depicted in
Figure 4, where
Um-Dyn ± I95% is the measured U-value using the dynamic method,
HFMNF indicates the use of a heat flow meter plate installed with a layer of thermal interface material directly on the wall surface and
HFMF indicates the use of a heat flow meter installed with a PVC film between the layer of thermal interface material and the wall.
As tests lengthened, the confidence intervals were reduced. These findings are aligned with those analysed in the existing literature, in which the length of the test that was too short led to greater confidence intervals [
14,
37].
Deviations between the two measured thermal transmittance values were calculated following Equation (7), using an HFM installed with a layer of thermal interface material directly on the wall surface and an HFM installed with a PVC film between the layer of thermal interface material and the wall surface. These relative differences between measured U-values using both heat flow meters for the two case studies are depicted in
Figure 5 and summarised in
Table 6.
4. Discussion and Conclusions
This research assesses the influence of the heat flow meter plate non-perfect thermal contact on the accuracy of the in situ measurement of walls’ thermal transmittance using the standardised heat flow meter method. The assessment considers usual practices in the installation of heat flow meters, consisting of the application of a PVC film to avoid damage to the wall in occupied buildings, in two case studies with different theoretical thermal transmittance values. The experimental campaign was designed to isolate measurement errors related to non-perfect thermal contact from those related to the measurement accuracy and the analysis of data. In situ measurements were conducted in the same conditions and equipment, with two heat flow meters: one was carefully installed with a layer of thermal interface material directly on the wall surface and the other was carefully installed by applying a film between the thermal interface material and the wall. Data were analysed in 24-h periods using the dynamic method. Finally, the variability of results was analysed by comparing the deviation between the measured thermal transmittances obtained from the two heat flow meters with the theoretical effect of including a PVC film during the monitoring process.
Test completion results indicate that the minimum duration of experimental campaigns was influenced by the installation of the heat flow meter plate. Generally, when the heat flow meter was installed by applying a PVC film between the thermal interface material and the wall to protect its surface, the minimum duration of the test was longer than when the heat flow meter was directly installed on the wall surface without a PVC film. In Case study 1, the test lasted 24 h more and in Case study 2, the test lasted 48 h more.
The findings show that the use of a PVC film hinders the installation of sensors, worsening the thermal contact between sensor and wall. Relative differences between the measured U-values were found to be greater than the expected by the theoretical calculation and also by the ISO standard quantification in both case studies. Deviations between the two measured thermal transmittance values, using an HFM installed with a layer of thermal interface material directly on the wall surface and an HFM installed with a PVC film between the layer of thermal interface material and the wall surface, were found to be significantly different than those expected resulting from the calculation of the theoretical effect of including a PVC during the monitoring process summarised in
Table 4. In the theoretical estimation, the use of a PVC film between the layer of thermal interface material and the wall surface had an expected deviation of 0.14% in Case study 1 and 0.02% in Case study 2. However, in the experimental tests results from the third cycle onwards showed that for Case study 1, the relative differences were around 24% to 27%, and in Case study 2 around 19% to 21%. On the other hand, random variations caused by a non-perfect thermal contact between the sensors and the surface are quantified by 5% according to ISO 9869-1:2014 [
31]. The experimental results also differ significantly from those quantified by ISO standard [
31]. These considerable differences between the expected differences and the measured ones might be due to collateral effects when installing an HFM with a PVC film between the layer of thermal interface material and the wall surface that hampers a good thermal contact, as Li et al. [
27] suggested. Including a PVC film to protect the wall may complicate the sensor installation process and consequently, the quality of the thermal contact might be hampered. Additionally, the theoretical calculation of including a PVC film on the thermal transmittance does not consider random variations in the thermal contact of the HFM with the surface of the wall tested. Moreover, extending the duration of the test did not seem to reduce the differences between the measured U-values. Therefore, it is highly recommended to avoid the use of a PVC film between the layer of thermal interface material and the wall surface in the in situ measurement of the thermal transmittance of existing buildings’ façades when accurate results are required.
This research could be useful for practitioners when they conduct energy audits. It was found that when HFMs were installed with a PVC film between the layer of thermal interface material and the wall surface, the measured thermal transmittance was around 19 to 27% lower than when HFMs were installed with a layer of thermal interface material directly on the wall surface. Therefore, depending on the installation of heat flow meter plates, technical staff could take into account these possible deviations in the in situ measurement of the façades’ actual thermal transmittance during building’s operational stage and, consequently, propose appropriate energy retrofitting strategies.