An approach to modelling envelope airtightness in multi-family social housing in Mediterranean Europe based on the situation in Spain

Depósito de investigación de la Universidad de Sevilla https://idus.us.es/ Esta es la versión aceptada del artículo publicado en: This is an accepted manuscript of a paper published in: Energy and Buildings (2016): 28/01/2025 DOI: https://doi.org/10.1016/j.enbuild.2016.06.074 Copyright: © 2016 Elsevier Ltd. All rights reserved. El acceso a la versión publicada del artículo puede requerir la suscripción de la revista. Access to the published version may require subscription. “This is an Accepted Manuscript of an article published by Elsevier in Energy and Buildings on 2016, available at https://doi.org/10.1016/j.enbuild.2016.06.074” An approach to modelling envelope airtightness in multi-family social housing in Mediterranean Europe based on the situation in Spain Fernández-Agüera, Jesica* ; Domínguez-Amarillo, Samuel; Sendra, Juan José; Suárez, Rafael Universidad de Sevilla *[email protected] Abstract Building hygrothermal performance, indoor air quality and energy consumption depend heavily on envelope airtightness. Over the last three decades, single-family dwellings have been amply studied in this respect by researchers in North Europe, the United States and Canada. However, very few studies have been conducted on airtightness in multi-family housing in warm climates such as Mediterranean Europe. This paper aims to enhance the understanding of building airtightness in early twenty-first-century multi-family buildings in southern Spain. Blower Door tests were conducted in 45 units in seven such buildings. The main airtightness parameter values found are reported and compared to the data for other buildings in southern Europe. The paper includes a statistical analysis of the findings, characterises building types and describes the protocol used to identify and quantify air leakage pathways. The conclusion drawn is that although a rough predictive model can be developed, the results are widely scattered due to the impact of the random component of manual construction, even in buildings with identical construction characteristics and types. The values recorded are nonetheless consistent with the findings for other European surveys. Keywords Airtightness, residential buildings, Blower Door test, air infiltration, southern Europe Abbreviations  V50: air leakage rate at 50 Pa across the building envelope, including the flow rate through joints, fissures and surface pores (m3/h)  Qenv: air flow (m3/h) across the building envelope  Δp: pressure differential (Pa)  n: air flow exponent  Cenv: air flow (or leakage) coefficient.  N: number of dwellings  Z: climate zone  FA: façade area (m2)  S: gross floor area (m2)  V: volume (m3)  WA: window area (m2)  WP: window contour (perimeter) (m)  WIN: winter severity  SUM: summer severity  n50: air change rate at 50 Pa (h-1)  w50: specific leakage area at 50 Pa (m3/(h·m2))  f50: air leakage rate per unit of façade area at 50 Pa (m3/(h·m2))  p50: air leakage rate per unit of window perimeter at 50 Pa (m3/h·m)  U-value: thermal transmission (W/(m2·K))  T: Monthly mean temperature (OC)  TM: Monthly mean of daily maximum temperatures (OC)  Tm: Monthly mean of daily minimum temperatures (OC)  H: Monthly mean relative humidity (%)  I: sunshine hours (h) 1. Introduction At this time, the residential sector accounts for 17 % of Spain’s final energy and 25 % of its electric power consumption. Approximately 48 % of that demand is estimated to be attributable to heating and cooling systems, values that are on the rise in Mediterranean areas [1]. As Spain is one of the largest countries, area-wise, in Mediterranean Europe, the hygrothermal performance, energy consumption and indoor air quality of its housing stock are sufficiently representative to be extrapolated to similar regions in southern Europe. In the wake of the transposition into Spanish Regulation (CTE, 2006) [2] of the European Energy Performance of Buildings Directive (EPBD) and its provisions on thermal envelope requirements, airtightness may be regarded as one of the factors with greatest impact on hygrothermal performance, energy consumption and indoor air quality in modern homes. Airtightness values and solar radiation-induced thermal loads (in both winter and summer) are instrumental in determining the actual energy demand in Mediterranean multi-family buildings. This applies to heating as well as cooling [3, 4, 5] and consequently to building energy strategies and performance labelling. Identification of air leakage pathways is a prerequisite to proposing construction solutions to improve building airtightness in compliance with the EU’s 20-20-20 objectives [6]. Building envelope airtightness has become a standard area of research in the USA [7], where the most comprehensive databases and widest variety of infiltration models are available [8, 9, 10, 11]. In particular, the LBNL air leakage measurements database developed by Chan et al. [12] and the DOE’s Multifamily Envelope Leakage Model [12] are very widely used. Drawing from an extensive review of measurements taken across the country, the former developed a revised model for single-family detached housing. The latter described a model for multifamily buildings which is, however, scantly applicable to southern Europe, where building types, construction techniques and climate vary immensely from the parameters for which the model was developed. Despite these constraints, both can be used a benchmarks for comparison. Intense research is also underway in northern and central Europe [13,14,15], where building airtightness is extensively regulated. In southern Europe, dwelling airtightness and its relationship to the main air inlets has been studied more profusely in France than in any other country. There, multi-family units have been shown to be more airtight than single-family homes and the construction system used has been identified as one of the major factors affecting the results [16, 17]. Legislation on airtightness is lacking in other southern European countries, although insightful papers have been published on conditions in Greece [18], Italy [19] and Portugal [20, 21]. The Italian study covered 20 buildings of different types and in different seasons, while in Portugal the object of research was social housing, namely two multi-family buildings surveyed before and after retrofitting. The mean V50 values found by the latter for an area comparable to the region studied here were 8.9 h-1 prior to and 6.8 h-1 after refurbishment. The most prominent infiltration studies conducted in Spain include research by Meiss and Feijo, who found a mean V50 of 6.26 h-1 [22] for the 13 homes measured and by Tiberio and Branchi, who ran the most extensive survey in the region to date, testing 150 recently built dwellings. Although the data were widely scattered, with a mean V50 of 3.50 h-1, the distribution by low, medium and high airtightness showed that as a rule today’s buildings lie in the medium category [23]. While the data reported in these papers are of interest from the standpoint of characterisation, they refer to north-central Spain, a region with a fairly cold continental climate reminiscent of central European construction techniques geared to airtightness, insulation and effective heating systems. Building conditions there consequently vary widely with respect to the warmer areas of the country. In contrast, a study by Montoya et al. addressing the less directly related issue of protection against air pollution due to toxic gas release, adopted an approach relevant to the present research [24]. Although that study involved no field tests and focused on single-family homes, it proposed a regional prediction model that differed from the Chan model and drew from a database of French tests (CETE Lyon). The authors also proposed a methodology for extrapolating the data based on housing type similarities. Therefore, the studies conducted in southern Europe have yet to deliver a large body of data enough to identify the actual impact of airtightness on the energy efficiency of the residential buildings of the area, and the the link with regional construction systems. This article analyses a comprehensive suite of field data with the aim of enhancing the understanding of housing stock energy performance in an area highly representative of Mediterranean Europe. It focuses on multi-family units built in southern Spain in the first decade of the twenty-first century. The field campaign findings are set out, air leakage pathways are characterised, the scatter observed in envelope performance is discussed and a predictive model is proposed. 2. Context Single-family dwellings account for 64 % of the total European housing stock, although in the southern part of the continent, particularly in Italy and Spain, multi-family buildings prevail [25]. These countries also have a large stock of social housing, most of which adopts multifamily configurations [26] built to minimum construction quality standards [27]. While social housing can be defined in a number of ways, here it is meant in the broadest sense: housing that receives some manner of government subsidy or assistance [28]. Substantial improvements will be needed in these buildings in the years to come, both to improve indoor habitability and to comply with the energy efficiency standards for nearly zero energy buildings laid down in the EPBD [29] and in the EU’s 20-20-20 objectives [30]. This is going to call for a huge effort in Spain and in other Mediterranean countries to improve the existing housing stock by means of effective, optimised energy intervention and refurbishment techniques. One of the main aims of refurbishment policy should be to optimise available public and private resources. Scantly effective retrofitting should be avoided to attain genuine energy savings while reaching the primary aim, the enhancement of indoor comfort. These considerations are applicable to a significant portion of Europe: its entire Mediterranean flank, from southern Portugal to Turkey. The area has a temperate climate, with mild temperatures that fluctuate only narrowly throughout the year. The winters are not very cold (with temperatures dipping below freezing only exceptionally), and although warm, the summers are not torrid. While the climate varies with the lay of the land, it is clearly distinct from conditions in central and northern Europe. The area is classified as Köppen Csa and Csb [31]. Albeit with regional variations, multi-family building architecture and construction are similar across the area [32] and differ widely from the solutions in place in central and northern European countries, especially where social housing is concerned [33]. Many of these solutions are a legacy inherited from the nineteen sixties and seventies residential construction boom [34]. The area’s inhabitants share more than a similar climate, however, for social and cultural mores around housing use and occupancy are likewise analogous. The reinforced concrete frame and continuous concrete slab construction that characterises medium-rise multi-dwelling housing, especially social housing, in Mediterranean cities, results in very airtight inter-storey compartmentation. The outer enclosures typically consist of brick or block (heavyor medium-weight walls), usually with an inner cavity and several layers of plaster on the inside. Lightweight façades are uncommon. As a rule this social buildings has been concibed as Natural Ventilated based on the understanding of the climate as benign, and generally lacks built-in heating or air conditioning systems. Users meet their heating needs with portable electrical radiators that in most cases are energetically inefficient and unable to provide for an even distribution of comfortable temperatures [35]. Nor is this housing fitted with ducted, whole-home cooling systems. , despite the warm summers that characterise the area. As a result, homeowners tend to install room-size split AC-units a posteriori. These buildings were not generally fitted with mechanically controlled ventilation systems until the entry into effect of the Technical Building Code in 2007 [2]. Prior to that date, bathroom and kitchen ventilation was based on the stack effect across static vents. Such extraction systems are often ineffective. Where indoor air quality in Mediterranean housing is not particularly poor it is thanks to the uncontrolled inflow of air through the building envelope (infiltration) or voluntary ventilation (manual operation of windows). Envelope thermal-performance is average in such buildings, with medium-to-low thermal mass and wall thermal transmittance values on the order of 0.5-1.5 W/m2K. Artificial heating and cooling are needed to ensure comfortable indoor temperatures in the winter and summer under normal outdoor conditions. The absence of built-in HVAC is one of the major reasons for the indoor temperature dipersion typically found in this housing [35,36]. 3. Methods Field research to assess building airtightness was conducted between late 2011 and early 2013 in a number of locations in southern Spain (Andalusia) to ensure coverage of the various intraregional climates. Pressurisation and depressurisation were measured with a Blower Door Test System, later supplemented with infrared thermography and smoke tests to identify specific infiltration pathways. 3.1. Sampling The study was conducted on 45 homes in seven open gallery-type buildings in southern Spain. This building type, characterised by outdoor walkways on all floors from which each individual unit is accessed (Figure 1), was chosen for a number of reasons. - It is very commonly found in Mediterranean residential buildings, especially in social housing. - As most of the envelope around each unit separates it directly from the outdoors rather than from indoor staircases or hallways, leakage across elements other than the outer enclosures can be ruled out. - For the above reason, the façade: floor area ratio is larger than in other building types. As a result, the enclosures have a greater impact on airtightness, rendering these homes more vulnerable to air leakage than others. Figure 1: Ground plans of the buildings analysed The buildings chosen were located in five climate zones within the region, with winters ranging from very mild (zone A) to cold (zone C) and summers from warm (zone 3) to very warm (zone 4). The zones were classified by the climate categories set out in the Spanish building code. In this scheme, a given building’s heating/cooling demand is compared to its hypothetical demand in a reference location for which a climate profile has been established based on climate severity defined in terms of degree-days and global solar radiation, as described by de la Flor [37] and the mean climate data shown in Appendix A for each city. Figure 1 shows the ground plans of the buildings analysed and Table 1 lists their location, climate zone and year of construction.The drawings in Figure 1 are shown on a larger scale in Appendix A. Building City Climate zone Year of construction Window type Kind of shutters Number of storeys Ventilation Winter severity (WIN) Summer severity (SUM) CS1 Seville B 4 2004 Hinged Roller blinds 6 Passive vents in bathroom and kitchen CS2 Seville B 4 2006 Sliding Roller blinds 6 Passive vents in bathroom and kitchen CS3 Cordoba B 4 2007 Hinged Roller blinds 6 Passive vents in bathroom and kitchen CS4 Malaga A 3 2007 Sliding Roller blinds 6 Passive vents in bathroom and kitchen CS5 Almeria A 4 2008 Sliding Roller blinds 6 Passive vents in bathroom and kitchen CS6 Granada C 3 2003 Hinged Roller blinds 5/7 Passive vents in bathroom and kitchen CS7 Jaen C 4 2011 Sliding Roller blinds 6 Passive vents in bathroom and kitchen Table 1: Location, climate zone, year of construction and characterisation of the buildings studied. The units chosen in the seven buildings had similar floor areas (Table 2) and construction solutions to enhance comparability and ensure that generally applicable conclusions on performance could be drawn. The sample consisted in reinforced concrete frame, continuous slab buildings with ceramic or cement tile flooring. Only kitchens and bathrooms had continuous suspended ceilings. Vertical envelope consisted in cavity-insulated walls: ceramic-brick wall outside, air chamber with insulation panel, and thin ceramic-brick plastered inner sheeting. Party walls consisted in thin hollowbrick (approximately 15 cm thick) partitions, plastered on both sides. All windows, whether side-hinged or horizontal-sliding, had aluminium frames and double glazing (Table 1) and were fitted with built-in roller blinds (‘monoblock’ system). The cross-section of a typical building solution is shown in Apendix A. The U-values for walls and windows were similar for all the buildings. As these units had no built-in air conditioning or heating, there was no horizontal or vertical ductwork outside the conditioned space. This design minimises interstorey connections, which are confined to drainage and vents in wet rooms, normally embedded in the partitions. With construction systems based on large continuous inter-storey slabs and a paucity of internal service shafts, horizontal air leakage was minimal in these buildings, an advantage reinforced by the outdoor location of staircases and elevators. In open gallery configurations, in turn, dwellings are primarily in contact with the outdoor environment, with scant communication with other building elements as vertical shafts or stairs. The resulting lack of vertical air pathways [38] practically eliminated the stack effect. As all the units studied were one-storey dwellings, the façade would constitute the primary route for potential infiltration, along with theventilation shafts, which could be readily sealed and insulated during the tests. In pursuit of representative sampling, a random, stratified and proportional design was adopted. In all the buildings, at least one of the units tested was located on the lowest storey, one on an intermediate storey and a third under the roof to study the possible impact of horizontal enclosures and the stack effect resulting from differences in height within a given building [39]. Table 2 lists the characteristics of the units in the seven buildings, where Z represents the winter climate, N the number of storeys, S and V the mean floor area and volume, respectively, FA the mean façade area, WA the mean window area and WP the mean window perimeter. The standard deviation (σ) for all the aforementioned variables is also shown. Building Z N S V FA WA WP Mean (m2) σ Mean (m3) σ Mean (m2) σ Mean (m2)  Mean (m) σ CS1 B 7 64.51 6.01 176 16.46 54.29 2.92 8.99 1.87 30.84 3.82 CS2 B 5 70.40 0.00 183 0.00 47.20 0.00 8.20 0.00 28.00 0.00 CS3 B 8 70.00 0.00 185 0.00 46.40 0.00 9.10 0.00 30.30 0.00 CS4 A 4 68.63 21.9 188 60.08 60.60 19.77 12.08 3.33 40.38 10.24 CS5 A 8 68.89 1.72 188 5.79 46.91 0.82 9.00 0.67 29.78 1.44 CS6 C 8 69.88 0.23 175 0.51 45.45 0.63 16.03 0.86 42.78 2.44 CS7 C 5 69.56 0.49 174 1.24 34.06 1.05 8.90 0.00 26.40 0.00 Mean 45 68.80 6.50 181 17.90 47.52 8.50 10.43 3.00 32.71 6.70 Table 2: Characteristics of the buildings selected Although the units in the buildings selected had similar gross floor areas, their mean window areas and perimeters and floor areas differed substantially, favouring statistical analysis. Under these arrangements, the possible effect of each of these three parameters on airtightness could be determined. 3.2. Pressurised fan airtightness measurements The field study consisted in pressurisation/depressurisation tests performed with a Blower Door device supplemented with infrared thermography and smoke tests, a technique generally accepted as valid by the scientific community [40]. The test was conducted with a Minneapolis Blower DoorTM system running Tectite software. Spaces around all manner of openings were sealed with a suitable material. Building envelopes were assessed to Spanish and European standard UNE-EN 13829: 2002, Method B [41]. All the vents, mechanical and otherwise, in bathrooms and kitchens, including smoke extractors in the latter, were sealed and all other outdoor openings (windows) closed. In addition to the Method B trials, a number of other pressurisation/depressurisation tests were conducted, in which the sealing set-up was varied (Figure 9). Alternative tests P5 (bathrooms and kitchens sealed off) and P6 (only bathrooms sealed off) proved to be of particular interest to locate air pathways. The findings were used to compare the performance of the envelope enclosing dry rooms as opposed to the wet rooms that are more prone to uncontrolled cracks and openings around service ductways. Each unit was regarded to constitute a single volume, inasmuch as energy performance was practically even throughout. The test consisted in positioning the fan on the outside door to remove (depressurise) or force (pressurise) air from or into the dwelling to a positive or negative pressure of 50 Pa, and subsequently measuring the airflow at 5 Pa intervals. All estimated. The cumulative frequency curves and boxplots in Figure 6 show the results of the least significant difference (LSD) test, from which the three sub-groups or trends identified for n50, the five for f50 and the four for p50 were derived. Intra-sample dependence and sampling errors were ruled out. Figure 6: Cumulative frequency curves and box and whisker plots for the groups identified in the seven buildings studied 4.3. Dependence on climate zone The dataset did not appear to corroborate the intuitive assumption that as a rule, building construction seeks greater airtightness in colder than in warmer areas [44]. Hence, the initial hypothesis was that the volume-weighted infiltration values (n50) were not dependent upon the climate zone where buildings were located. The explanation may be that all the buildings in the sample were located in southern Spain and despite climate differences from one area to another, shared construction solutions, architecture and usage. Initially, then, location was assumed to have no impact on the prediction of parameter n50 (null hypothesis). Table 7 lists the statistics for n50 by climate zone and Figure 7 reproduces the cumulative frequency curves, likewise by climate zone. Z N umber of units Mean (h-1) Median (h-1) Standard deviation Coefficient of variation Lowest Highest A 12 4.08 4.1 0.51 12.65% 3.20 5.30 B 20 6.59 6.7 1.64 24.93% 3.90 8.70 C 13 5.86 5.3 1.59 27.14% 3.70 8.30 Mean 45 5.71 5.30 1.73 30.36% 3.20 8.70 Table 7: Statistics for n50 (h-1) by climate zone Figure 7: Cumulative frequency curves for n50 (h-1) by climate zone Lastly, a chi-square test was run (with a test value of 65.97) to determine whether the assumption of independence between n50 and climate zone could be rejected. Since the pvalue was greater than 0.01 (0.22), independence between the two variables could not be ruled out at 99 % confidence. In light of those results and the distribution of the values, then, n50 was not necessarily dependent upon climate zone. 4.4. Multiple linear correlation approach to developing a predictive model While acknowledging the stochastic nature of these values due to their association with construction processes, a predictive model could be proposed, subject to a band of uncertainty. This model may prove useful when forecasting expected behaviour and establishing a framework for comparison. Pearson’s bivariate correlation coefficients were found for the geometric parameters associated with the n50 values to ascertain the existence or otherwise of inter-variable dependence. The correlation coefficients ranged from -0.98 to 0.75. For V, they denoted a statistically significant p-value of 0.003, which was somewhat higher for WA and even higher for WP. In light of this possible dependence, which even if present would not prove causality, a series of ANOVAs and linear regression analyses were conducted to determine whether infiltration in the group of dwellings studied could be predicted with a model based on their geometric characteristics, as the initial hypothesis would appear to indicate. A factorial analysis was run on the independent variables in the linear regression model. This procedure, which excludes the least significant variables in the model, reduced the number to the four with the highest statistical significance (Figure 8). All six variables, S, V, FA, WA, WP and Z, were included in the factorial analysis to obtain a better fit. As expected, FA and Z (climatic zone) exhibited low statistical significance and were disregarded. Figure 8: Component+residual plot for n50 and diferent parameters The predictive model proposed can be expressed as shown below (Equation 3), with the coefficients listed in Table 8. 𝑛 = 𝛼 × 𝑆 − 𝛽× 𝑉 − 𝛾 × 𝑊𝐴 + 𝛿 ×𝑊𝑃 Eq.(3) Parameter Coefficient Estimate Standard error S 𝛼 1.019 0.101 V 𝛽 -0.353 0.044 WA 𝛾 -1.763 0.360 WP 𝛿 0.549 0.161 Table 8: Predictive linear regression model: coefficients The model proposed afforded a good approximation, explaining 98.12 % of the variation in the dependent variable (n50). Although the estimated error was significant (0.849 h-1), it provided a more accurate fit than the standard deviation of the measurements (approximately halving the width of the fluctuation band). Barring any prior knowledge of geometric factors, the best prediction was that the air flow in a dwelling would be around 5.721.74 h-1 (Table 5). This model is proposed to predict envelope air-tightness in the social housing stock built in southern Spain, based on a sample of 45 gallery-type buildings. The model was developed using multiple linear regression. 4.5. Typical air leakage pathways The primary air leakage pathways [47] were identified with infrared thermography or smoke during the Blower Door test. These tests were run during the winter with the heating on and the rooms depressurised (ISO 18434-1: 2008) [48]. In the building type studied, the envelopes in all the units were bounded in part by the outdoors and in part by other units. Buildings were tested in detail, sealing off the parts of the unit where air could flow in from the party walls, to determine the percentage of infiltration attributable to the outdoor enclosure (Figure 9). Figure 9: Sealing protocols to locate air leakage pathwaysThe findings for the various types of test (protocols 1 to 9) were analysed to determine the relative weight of each dwelling component in overall air infiltration. The trend identified in this suite of tests reproduced the behaviour observed in the detailed analysis conducted on building CS2, albeit with a significant rate of variation (Figure 10). Infiltration between adjacent units was found to be scantly significant: just 3.6 % of the total. Figure 10: Dwelling airtightness, contribution by component These tests also showed that air leakage took place primarily across the vertical envelope, particularly in dry rooms where most of the windows are located. Nonetheless, plumbing, electrical and ventilation ductways also contributed to infiltration. In contrast, no significant infiltration was found across horizontal elements or party walls, nor were differences observed between under-roof units and other dwellings. As would be expected in façades of the type found on these buildings, the main air pathways were at the interface between windows and opaque enclosures. The wide variability recorded in the test values may have been the result of the manual construction procedures deployed in this type of buildings, inducing differences in dwellings in one and the same building. Hens identified the cause of such variations to be poor workmanship [49]. In this study, the findings for different window types were compared to corroborate that conclusion. No relationship was found between window closure type (hinged or sliding) and infiltration rates, whereas behaviour was observed to vary widely within each type of window. The whisker box in Figure 11 shows the proportion of leakage attributable to the wet (kitchen and bathroom) and dry rooms. Note that over half of the total leakage was associated with the envelope around dry rooms (in the Method A test, i.e., with the dwelling in use). Nonetheless, these values were subject to 17 % variability. The proportion attributable to wet rooms was lower, and higher in bathrooms (from under 10 % to 20 %) than in kitchens (under 7 %). Method B, i.e., considering measurements for the envelope alone (disregarding vents and permanent openings), yielded the following values: 72.69 % for the dry room envelope, 18.85 % for the bathroom envelope and 8.71 % for the kitchen envelope. Figure 11: Proportion of infiltration accounted for by room type 4.6. Comparison to other regions The present findings were compared to the results reported for similar studies on multi-family dwellings conducted in southern Europe. Research focusing on single-family homes or with a small number of cases was disregarded. When the information was provided on the date of construction, the data were broken down for separate analysis of the dwellings built within more or less the same time frame as the ones studied here, in particular units built to legislation deriving from the EPBD. The n50 values are reproduced in Figure 12, which also shows the intervals that define the low, medium and high airtightness categories set out in European and international standard EN-ISO 137900 [50] and the number of tests run to establish the representativeness of the values. Figure 12: Values for European multi-family dwellings The figure shows information on the following datasets: Italya, overall total [19]; Italyb, dwellings built in 2001-2010 [19]; Portugala [20]; Portugalb, refurbished dwellings [21]; Portugalc, non-refurbished dwellings [21]; N-Spaina, dwellings built in 1979-2007 [22]; NSpainb, EPBD dwellings [22]; N-Spainc, mainly EPBD dwellings [23]; S-Spain, present study. The above analysis revealed that the S-Spain n50 values, which lay predominantly in the medium band, were in line with the findings for similar units in the Mediterranean area. These values were lower than in other newer buildings in northern and central Spain, built after the transposition of the EPBD, and lower as well than Italian buildings erected more recently. The lowest n50 values (high airtightness) were found for the oldest buildings (Italya and N-Spaina), while the highest (low airtightness) were recorded in the Portuguese dwellings (Portugala,b,c) even after refurbishment. The scatter observed in the n50 values, while significant, was similar to the findings reported for other Mediterranean regions. The n50 values recorded here were higher than required in northern and central European countries’ design standards, where housing airtightness has long been the object of regulation. The reference n50 value in Germany and Norway is 3 h-1 and in Finland, 4 h-1. While some of the dwellings in this study exhibited values nearly as low as those benchmarks, others, in keeping with results reported by the earlier studies analysed here, doubled the score. 5. Conclusions The air permeability readings at 50 Pa for 45 multi-family units with open galleries built in southern Spain after 2000 averaged 5.72 h-1. This value was higher than energy efficiency standards and recommendations for European buildings, although similar to the results reported for northern Spain and other southern European countries such as Italy and Portugal. The study revealed a wide scatter in the airtightness values recorded and highlighted the difficulties involved in proposing an accurate predictive model. That the scatter for parameter n50 was greater than the mean values is an indication of flaws in building envelope continuity. The values for the entire sample ranged from 3.2 to 8.7 h-1 and in some cases the high value in a given building doubled the low value in the same building. The value of exponent n ranged from 0.56 to 0.61 in all the measurements performed. Values on the order of 0.6 are associated with leakage at interfaces between openings and their opaque surrounds. Despite the uncertainty generated by the artisanal façade construction methods typical of this housing stock in southern Europe, the model developed was able to predict the degree of airtightness in these dwellings with acceptable accuracy. That model may be particularly useful to compare airtightness values for buildings erected in the same time frame. It should be regarded as a regional tool, tailored to southern European construction and architecture that aims to be an alternative to the more universal models developed to date. In addition, the probabilistic model for behaviour proposed here is applicable to the assessment and construction of models for analysing the housing stock. In light of the difficulties described to build a reliable predictive model, no design solutions for buildings and their envelopes that would enhance airtightness can be put forward. Consequently, efforts should focus on construction per se, seeking more assembly-sensitive solutions and reinforcing the control of workmanship in openings and interfaces with the opaque plane. Tests that measure infiltration rates must also be conducted. No infiltration was observed at the interface between the deck slab and vertical members in the dwellings tested, with no perceptible differences between under-roof and intermediate units, nor was any detected at party walls. Therefore, improvements should focus on the joints around façade openings. This study revealed the need to individualise airtightness surveys to be able to identify the main air infiltration pathways across the envelope and hence apply protocols both for construction quality and ascertainment of the condition of buildings with a view to energy refurbishment. Pressurisation techniques have proven to be a cost-effective approach to assessing and improving the energy performance of buildings in general and residential buildings in southern Europe in particular. Given the impact of infiltration on energy demand and consumption in residential buildings, airtightness should be regarded as an instrumental factor for the energy assessment of the existing housing stock and its refurbishment to meet nearly zero energy building (NZEB) aspirations in the southern Mediterranean. Acknowledgements This study was funded by the Spanish Ministry of Economy and Competitiveness under project BIA2012-39020-C02-01 and BES-2013-063097 support. Appendix A Figure A.1: Plan view of CS1. Figure A.2: Plan view of CS2. CS5 8 24.46 7.64 2.76 11.30% 19.80 28.50 8.70 0.04 CS6 8 19.48 11.33 3.37 17.28% 14.60 26.60 12.00 2.02 CS7 5 50.62 11.08 3.33 6.57% 47.80 54.50 6.70 -1.47 Mean 45 33.26 187.38 13.69 41.15% 14.60 55.90 41.30 -1.91 Table B.3: Statistics for p50 (m3/hm) Predictive model R 2 : 0.981 Adjusted R 2 : 0.981 Estimated standard error 0.84 Table 11: Predictive linear regression model: summary References [1] Ministerio de Industria, Turismo y Comercio. PROYECTO SECH-SPAHOUSEC: Análisis del consumo energético del sector residencial en España IDAE, 2011. [2] Real Decreto 314/2006: España, Real Decreto 314/2006, por el que se aprueba el Código Técnico de la Edificación, de 17 de marzo. Boletín Oficial del Estado, 28 de marzo de 2006, n. 74, p. 11816. [3] J.J. Sendra, S. Domínguez, A.L. León, J. Navarro, et al, Proyecto Efficacia: Optimización energética en la vivienda colectiva, Universidad de Sevilla, 2011 [4] S. Domínguez, J.J. Sendra, A.L. León, P.M. Esquivias, Towards Energy Demand Reduction in Social Housing Buildings: Envelope System Optimization Strategies, Energies 5 (2012) 22632287. [5] A.L. León, S. Domínguez, M.A. Campano, C. Ramírez-Balas, Reducing the Energy Demand of Multi-Dwelling Units in a Mediterranean Climate Using Solar Protection Elements Energies 5 (2012) 3398-3424. [6]Directiva 2012/27/UE del Parlamento Europeo y del Consejo, de 25 de octubre de 2012, relativa a la eficiencia energética, por la que se modifican las Directivas 2009/125/CE y 2010/30/UE, y por la que se derogan las Directivas 2004/8/CE y 2006/32/CE [7] M.H. Sherman, W.R. Chan, Building Airtightness: Research and Practice, Ernest Orlando Lawrence Berkeley National Laboratory, 2004. [8] W. R. Chan, W. W. Nazaroff, P. N. Price, M. D. Sohn, A. J. Gadgil, Analyzing a database of residential air leakage in the united states, Atmospheric Environment 39 (2005) 3445-3455. [9] J. McWilliams, M. Jung, Development of a Mathematical Air-Leakage Model from Measured Data, Ernest Orlando Lawrence Berkeley National Laboratory, 2006. [10] M.H. Sherman, D. J. Dickerhoff, Air Tightness of U.S. Dwellings, Ernest Orlando Lawrence Berkeley National Laboratory, 2002. [11] M. H. Sherman, N. Matson, Air Tightness of New U.S. Houses: A Preliminary Report, Department of Energy, Ernest Orlando Lawrence Berkeley National Laboratory, 2002. [12] W.R. Chan, J. Joh, M.H. Sherman, Analysis of air leakage measurements of US houses, Energy and Buildings 66 (2013) 616–625. [13] J. Jokisalo, J. Kurnitski, M. Korpi, T. Kalamees, J. Vinha, Building leakage, infiltration, and energy performance analyses for Finnish detached houses. Building and Environment 44 (2009) 377-387. [14] D. Sinnott, M. Dyer, Airtightness field data for dwellings in Ireland. Building and Environment 51 (2012) 269-275. [15] W. Pan, Relationships between airtightness and its influencing factors of post-2006 newbuild dwellings in the UK, Building and Environment 45 (2010) 2387-2399. [16] R. Carrié, R. Jobert M. Fournier, S. Berthault, Permeabilite´ a` l’air de l’enveloppe des batiments. Generalites et sensibilisation, CETE de Lyon, 2006. [17] F. R. Carrié, B. Rosenthal, August An overview of national trends in envelope and ductwork airtightness, AIVC Information Paper 29 (2008). [18] A. Sfakianaki, K. Pavlou, M. Santamouris, I. Livada, M. Assimakopoulos, P. Mantas, et al., Airtightness measurements of residential houses in Athens, Greece, Building and Environment 43 (2008) 398-405. [19] F.R. d´Ambrosio Alfano, M. DellÍsola, G. Ficco, F. Tassini, Experimental analysis of airtightness in Mediterranean buildings using the fan pressurization method, Building and Environment 53 (2012) 16-25. [20] N. M.M. Ramos, R. M.S.F. Almeida, b, A. Curado, P. F. Pereira, S. Manuel, J. Maia, Airtightness and ventilation in a mild climate country rehabilitated social housing buildings – What users want and what they get, Building and Environment 92 (2015) 97–110. [21] M. Pinto, J. Viegas, V.P. de Freitas, Air permeability measurements of dwellings and building components in Portugal, Building and Environment, 46 (2011) 2480-2489. [22] A. Meiss, J. Feijó-Muñoz, The energy impact of infiltration: a study on buildings located in north central Spain, Energy Efficiency 8 (2015). [23] A. Jimenez Tiberio, P. Branchi, A study of air leakage in residential buildings, New Concepts in Smart Cities: Fostering Public and Private Alliances, (2013) 1-4. [24] M. I. Montoya, E. Pastor, F. Remi Carrie, G. Guyot, E. Planas, Air leakage in Catalan dwellings: Developing an airtightness modeland leakage airflow predictions, Building and Environment 45 (2010) 1458–1469. [25] K. Dol, M. Haffner, Housing statistics in the European Union, Delft University of Technology, 2010. [26] D. Pennestri, The energy and environmental requalification of post-war housing: problematics and innovative solutions for the building envelope, Proceedings of SB13 Prague on Sustainable Building and Refurbishment for Next Generation, 2013. [27] BPIE, Buildings Performance Institute Europe, Europe’s buildings under the microscope, 2011. [28] A. T. Carswell, The Encyclopedia of Housing, Second Edition University of Georgia, 2012. [29] Directive 2010/31/EU of the European Parliament and o the Council of 19 May 2010 on the energy performance of buildings (recast). [30] Europe’s 20-20-20 objectives [31] M. Kottek, J. Grieser, C. Beck, B. Rudolf, and F. Rubel, World Map of the Köppen-Geiger climate classification updated, Meteorol. Z., 15 (2006) 259-263. [32] S. Ferrari, V. Zanotto, Building Energy Performance Assessment in Southern Europe, Editor Springer, 2015. [33] J. Allen, J. Barlow, J. Leal, T. Maloutas, L. Padovani, Housing and Welfare in Southern Europe, Wiley-Blackwell, 2004. [34] D. Matteis et al, Affordable Housing Housing Praxis for Urban Sustainability, Ipertesto Edizioni, Verona Good, 2008. [35] J.J. Sendra, S. Domínguez-Amarillo, P. Bustamante, A.L. León, Energy intervention in the residential sector in the south of Spain: Current challenges, Informes de la Construcción, 65(2013) 457-464. [36] A.L. León, S. Muñoz, J. León, P. Bustamante, Monitorización de Variables Medioambientales y Energéticas en la Construcción de Viviendas Protegidas: Edificio CrosPirotecnia en Sevilla, Informes de la Construcción 62 (2010) 67-82. [37] F. J. S. De la Flor, S. A. Domínguez, J. L. M. Félix, R. G. Falcón, Climatic zoning and its application to Spanish building energy performance regulations, Energy and Buildings 40 (2008) 1984–1990. [38] W.R. Chan, J. Joh, and M. H. Sherman, Analysis of Air Leakage Measurements from Residential Diagnostics Database, Ernest Orlando Lawrence Berkeley National Laboratory, 2012. [39] W. Walther, B. Rosenthal, Airtightness Testing of Large and Multi-family Buildings in an Energy Performance Regulation Contex, Information Paper no P165, EPBD buildingsplatform, 2009. [40] H. E. Feustel, Measurements of air permability in multizone buildings, Energy and Buildings, 14 (1990) 103-116. [41] EN 13829:2002 Thermal performance of buildings - Determination of air permeability of buildings - Fan pressurization method. (ISO 9972:1996, modified) [42] F. R. Carrié, V. Leprince, Uncertainties in building pressurisation tests due to steady wind, Energy and Buildings, 116 (2016) 656–665. [43] D. Etheridge, M. Sandberg, Building ventilation. Theory and measurement, Chichester, England: John Wiley & Sons, 1996. [44] UNE-EN 15242:2007 Ventilation for buildings - Calculation methods for the determination of air flow rates in buildings including infiltration. [45] M. Orme, M. Liddament, A. Wilson, An Analysis and Data Summary of the AIVC's Numerical Database, Technical Note 44, Air Infiltration and Ventilation Centre, Coventry, 1994. [46] AIVC Guide 5, Ventilation Modelling Data Guide, 2002. [47]EN 13187:1998 Thermal Performance Of Buildings - Qualitative Detection Of Thermal Irregularities In Building Envelopes - Infrared Method [48] ISO 18434-1:2008. Condition monitoring and diagnostics of machines. Thermography. Part 1: General procedures. [49] H. Hens, Dwelling air-tightness in a 55 years old estate. In: 9th Nordic Symposium on Building Physics (NSB 2011), Tampere, Finland, 2011. [50] EN-ISO 13790:2008 Energy performance of buildings. Calculation of energy use for space heating and cooling.