1. Introduction
Given the contrasted climate evolution towards an increase in global temperatures, numerous studies and organizations agree on the need to warn society about the risks faced. This global warming, directly related to greenhouse gas emissions, is attributed to a large extent to the building sector, considered one of the major energy consumers worldwide. In the European Union specifically, over 25% of final end-use energy was associated to the household category in 2015 [
1].
The Intergovernmental Panel on Climate Change (IPCC) [
2] periodically publishes a series of reports compiling and synthesizing the most significant studies on climate change to date. In the Special Report on Emissions Scenarios [
3], the IPCC presents a set of future scenarios—A1, A2, B1 and B2—in accordance with social and economic variables and with different predictions for the evolution of CO
2 concentration. A2 is the most unfavourable scenario, with the highest CO
2 concentration predicted for the year 2100, and is therefore considered a priority in the research of the impact and consequences of climate change.
Different authors evaluated the influence of the global increase in temperature on the energy performance of buildings [
4,
5]. It is worth highlighting the research by Santamouris [
6], who, based on recent studies conducted in different regions of the world, developed a model which predicts an increase of up to 35% in global cooling consumption in the housing sector for 2050, mainly due to climate change, the increase in population, and income growth. Focusing on the Mediterranean climate, Kapsomenakis et al. [
7] evaluated the impact of the outdoor temperature increase over the last 40 years on the energy consumption of an office building in Greece, concluding a cooling demand increase of around 5 kWh/m
2 per decade. Suárez et al. [
8] estimated that the buildings built in southern Spain since 2006—in accordance with the limitations set by the current Spanish energy saving regulation [
9]—will double their cooling demand under the 2050 climate change scenario.
In addition to the increase in energy consumption, climate change will worsen indoor thermal conditions in the dwellings inhabited by the social groups which are economically more vulnerable, as they cannot afford to use active air conditioning systems [
10,
11]. The global increase in outdoor temperatures will lead to indoor overheating conditions which must be addressed in order to eliminate the risk of energy poverty [
12].
This situation poses a series of challenges for the building sector [
13,
14]. On the one hand, the design of new buildings should ensure the minimum polluting emissions from construction sites, maximum energy efficiency and a near-zero energy consumption, but what will happen to existing buildings? Their environmental performance will worsen and thermal conditioning needs will increase, meaning that they are in urgent need of retrofitting in order to be able to tolerate the future climatic conditions [
15].
In southern Europe, a large percentage of dwellings, between 63% [
16] and 76% [
17], predate the first regulations limiting energy demand in buildings (1976–1979). As a result, there is a relevant part of the residential stock with obsolete energy performance, and its population will suffer the most due to the negative effects of climate change if investment is not made in retrofitting [
18]. Numerous studies defend the need to characterize case studies before proposing energy retrofitting measures [
19,
20] in order to be able to calibrate simulation models and to avoid the ‘performance gap’ between real and predicted consumptions [
21,
22]. This characterization must take into account the long-term climate perspective to ensure that retrofitting projects do not become obsolete as soon as they are completed.
This environmental or energy characterization can be carried out individually or through building stock models. When evaluating a set of buildings—such as a building category or entire housing stock—there are two methods: top-down and bottom-up [
23]. A high percentage of the models evaluating the energy performance of large building stocks makes use of top-down techniques, based mainly on historical and statistical data. These models have been used in order to characterize the energy consumption of residential stocks in northern [
24,
25] and southern Europe [
26,
27], attempting to identify key challenges to energy saving. Given that the results of these models are largely based on previous data, these techniques are less convenient when evaluating the building performance under climate projections.
In the last few decades, bottom-up engineering techniques have been the most commonly used when a detailed computation of the energy performance of buildings is required. These techniques are based on data collection from study samples to be extrapolated at the regional or national level using simulation models [
28]. The simulation tools most frequently employed by the scientific community are EnergyPlus [
29], ESP-r [
30] and TRNSYS [
31], which ensure highly accurate results. This is the case of the model developed by Wang et al. [
32] for the evaluation of retrofitting measures, in terms of energy demand, for residential districts in Switzerland and that designed by Penna et al. [
33] for different Italian cities. Based on the results of these building stock models, surrogate models can be developed, which drastically reduce computational times [
34,
35].
Nowadays, most simulations use typical weather files, which are available in the databases frequently used by the scientific community and are usually obsolete. These weather data do not match the real situation of housing stock in the coming years as a result of global warming, a phenomenon which will be further worsened in cities due to urban heat island effects, according to the available literature [
10,
36]. Therefore, this work aims to quantify the impact of future climate change scenarios on the thermal behaviour of a building category. In order to do this, a building stock model will be developed using the EnergyPlus [
29] simulation tool, characterizing the environmental behaviour of an entire building category under the typical climate conditions of the city of Seville and those of its corresponding climate change scenario for 2050.
2. Materials and Methods
The starting point for the methodology of this work is a building stock model previously developed for the environmental characterization of the entire building category: linear multi-family social housing developed in southern Spain between the 1950s and the 1980s. The development and calibration of this model through in-situ measurements is described in depth in a previous publication [
37].
This research applied the SLABE (Simulation-based Large-scale uncertainty/sensitivity Analysis of Building Energy performance) method defined by Mauro et al. [
38], providing a reliable environmental evaluation model for any building belonging to the category studied. This model focuses its evaluation on the percentage of discomfort hours, since this building category is generally characterized by summer and winter natural ventilation, i.e., free-running indoor temperatures, with a very sporadic use of mechanical local cooling or heating systems. In order to be able to contrast with in-situ measurements, this initial model made use of outdoor climatic conditions referenced for the monitoring period (years 2014 and 2015).
A climate data projection for 2050 will be developed for application in the model simulation in order to assess the influence of global warming on the thermal behaviour of this building category. This comparison of the results obtained for the typical and 2050 climatic conditions is the main originality of the proposed methodology. In addition, a sensitivity analysis will be performed to ascertain the parameters with most influence on thermal discomfort.
Finally, the simulation models within the building category with the best and worst annual thermal performance will be determined to develop a detailed evaluation of their characteristic parameters and environmental behaviour. The influence of climate change in terms of energy demand will also be estimated for these case studies.
Figure 1 summarizes the process followed throughout this study.
2.1. Building Category Characterization: The SLABE Method
The simulation tool EnergyPlus [
29] was chosen to assess the level of thermal comfort in the building category as it allows a detailed assessment of the indoor temperature and works with text-based format inputs and outputs, allowing communication with mathematical tools for processing the results. A calibrated model representing a specific building belonging to the category was used as a starting point. The peripheral and surrounding buildings were also defined (geometry, material and transmittance) in this model, interacting with the case study in terms of shading, reflection and longwave emission. In addition, the roughness of the terrain typical of suburban locations was considered for the calculation of local wind speed.
Based on this initial model, the input data are replaced by parameters. For the purpose of switching from an individual case study model to a building stock one, a variability range and probability distribution (uniform or normal) should be attributed to these parameters. Latin Hypercube Sampling is then applied to these variables within a Monte Carlo framework, developing a specific number of case studies. This method ensures the uniformity and representativeness of the study sample [
39]. In this case, the sample size was set as N = 750 cases, a value previously considered optimum for this building stock model in order to ensure the best performance in the case of future development of an Artificial Neural Network based on these simulation results [
37]. This sample size gives rise to a ratio between N and the number of characteristic parameters over 25, which is much higher than the ratio of between 2 and 5 recommended in the available bibliography commonly studying a specific case study [
40]. Conraud sets a minimum N of
5 x number of inputs x number of outputs [
41]. Given that the analysis of a building category involves higher ranges of variability in the characteristic parameters, a higher sample size was set to guarantee the significance of the results.
In this work, the building category represents a stock of buildings with the same geometric and constructive typology, similar user profiles (e.g., a limited use of HVAC, restricted economic conditions since it is social housing, etc.) and the same climate conditions. It should not include different building typologies, since this makes the ranges of variability of the characteristic parameters excessively broad and compromises the model’s reliability.
With regards to the extensive research carried out here, the simulation of the N cases in EnergyPlus is automatically commanded by MATLAB [
42] and the thermal comfort prediction is obtained for the study sample in terms of percentage of discomfort hours (DH). Discomfort hours are defined as the occupied hours with indoor temperatures outside the comfort acceptability range, according to the equations established in
Section 2.2. The DH provided is a global value for the whole building, obtained from the calculation of the average DH value for each dwelling.
In addition, a sensitivity analysis (SA) is developed to determine which parameters have the most influence on DH. A global method was selected for the SA, evaluating the Standardized Rank Regression Coefficients (SRRCs) [
38]. According to the available literature [
43,
44], this approach is the most appropriate for the type of relations between inputs and outputs in this study: non-linear but monotonic. The SRRC ranges from -1 to 1, with a positive value representing input and output parameters changing with the same sign, and the opposite for a negative one.
2.2. Adaptive Comfort Standards
For the purposes of the evaluation of thermal comfort, most of the international standards are based on Fanger studies [
45]. They assess thermal comfort according to two indices: the Predicted Mean Vote (PMV) and the Predicted Percentage of Dissatisfied (PPD). Standard EN ISO 7730 [
46], based on this methodology, is one of the most widely used by the scientific community. However, when the case study is a naturally ventilated building, the use of adaptive standards is considered more suitable for thermal comfort assessment than standards based on the PMV index [
47,
48,
49]. The main reason is that in residential buildings, users are completely free to vary the amount of clothing they have on and to open the windows to improve thermal sensation. As a result, these variables are very difficult to fix.
According to previous research [
50,
51], the specific case study of social housing in southern Spain requires a revision of the suitability of applying the adaptive thermal comfort standards most commonly accepted by the scientific community. These studies conclude that, while the two most extensively used adaptive thermal comfort standards—ASHRAE Standard 55 [
52], ISO-EN-15251 [
53] and its revision prEN 16798-1:2015 [
54]—function properly for the winter climatic conditions in southern Spain, this is not the case with indoor overheating conditions in summer. In fact, with the maximum outdoor temperatures considered in both standards, 33.5 °C and 30 °C, respectively, indoor temperatures above 31 °C are included in the comfort range. In this specific study and based on the conclusions of the analysis carried out, thermal comfort levels are assessed as follows:
From December to February (winter period), the adaptive thermal comfort standard defined in ISO-EN-15251 was applied (optimum comfort temperature (
Tco) according to Equation (1)). This equation is only suited to naturally ventilated buildings, with low metabolic rate activities. The acceptability range defined is a temperature interval of ± 4 °C, associated to building category III (corresponding to PPD < 15%), which is defined only for existing buildings.
where:
TeR: running mean dry bulb outdoor temperature for today (Equation 2)
where:
Ted-1: daily mean dry bulb outdoor temperature for previous day;
TeR-1: running mean dry bulb outdoor temperature for previous day;
α: a constant between 0 and 1. Use of 0.8 is recommended.
From March to November (summer, spring and autumn periods), the alternative adaptive thermal comfort standard defined by Barbadilla-Martín et al. for the Mediterranean climate was applied (
Tco calculated following equation 3) [
55]. This standard is only applicable to ‘Mixed Mode’ buildings, i.e., naturally ventilated buildings which also have an air conditioning system in occasional use. For this standard, the acceptability range defined is a temperature interval of ± 3.5 °C, corresponding to a predicted percentage of dissatisfied (PPD) under 20%. In Equation (3),
TeR is the running mean dry bulb outdoor temperature for today (see also Equation (2)).
2.3. Environmental Assessment for Climatic Conditions of the Year 2050
In order to evaluate the influence of a climate change scenario on the environmental performance of a whole building category, climate data are modified in the initial model simulation, taking into account the climate change scenario projection for the year 2050.
The climate data projection for 2050 was carried out using CCWorldWeatherGen software, developed by the University of Southampton [
56]. The operation of this tool is based on the ‘morphing’ method for climate change conversion of weather data defined by Belcher et al. [
57]. The original climate data file from the Seville Airport meteorological station, provided by the EnergyPlus database [
58], was converted into one that takes into account future conditions resulting from a climate change scenario [
59]. In this specific case study, the IPCC Third Assessment Report model summary data of the Hadley Centre Coupled Model version 3 (HadCM3) for the A2 scenario was used [
60]. It is the most adverse scenario, regarding the 2050 prediction of CO
2 particles per million.
The simulation of the environmental behaviour of the building category follows the same methodology described in
Section 2.1, but replaces the original climate data file with that developed by the CCWorldWeatherGen tool for 2050. The comparison between the results of both simulation models —current climate vs. 2050 projection—allows the influence of the temperature increase in the climate change scenario to be assessed for an entire residential building category in the Mediterranean climate.
2.4. Environmental and Energy Assessment of the ‘Best’ and the ‘Worst’ Case Study
Once environmental behaviour is evaluated in terms of discomfort hours for the entire building category, the cases with the best, that is, the lowest DH, and worst thermal behaviour or highest DH throughout the year are determined. For these two case studies, the hourly evolution of the temperature for a dwelling located on an intermediate floor was analysed on a typical winter (January 16) and summer day (July 27), taking into consideration typical climatic conditions and those projected for the year 2050. This establishes a range of results in order to quantify the influence of global warming on the indoor temperature of this building category in southern Spain.
Finally, the annual energy demand for heating and cooling is estimated to calculate the requirements of both case studies in order to ensure indoor comfort conditions under the climate change scenario. These results are also compared with the values obtained for the typical climatic conditions.
5. Discussion
The results obtained in this work and detailed in the previous sections predict a generalized worsening of indoor thermal comfort conditions for the climate change scenario. Although these results are in line with the conclusions from similar studies, they are even more worrying given the energy vulnerability of the building category evaluated and severe summer climatic conditions in the location selected.
Similar research carried out in the Mediterranean climate, such as that by Kapsomenakis et al. [
7] on an office building in Greece or that by Pierangioli et al. [
67] on an apartment building in Italy, established an increase in cooling demand for 2050 of 25 and almost 10 kWh/m
2 respectively. In the Greek case study, the increase is between 15 and 29%, while in the Italian case study, it is around 30%. Cellura et al. [
68] evaluated the impact of climate change in office buildings, designed following current thermal regulations in different cities of southern Europe. In Valencia (eastern Spain), a rise in cooling demand of up to 30 kWh/m
2, 220% in relative terms, is estimated for the year 2035. Nevertheless, as this study focuses on a case study with very low thermal performance, that increase is estimated to range from 25 to more than 41 kWh/m
2. This translates into several relative increments of current cooling demand, up to 250%. The energy assessment of detached houses designed according to current Spanish regulations by Pérez-Andreu et al. [
69] in eastern Spain and by Suárez et al. [
8] in southern Spain concluded that the cooling demand will have more than doubled in 2050, with a predicted increase in the average monthly outdoor temperature of between 3 and 4 °C. In the case study from eastern Spain, cooling demand will increase from 15 kWh/m
2 to 40 kWh/m
2, while in the case study in southern Spain it will go from 20 to 40 kWh/m
2. In the building category under here, it is estimated that this demand will even triple, rising from 10 to 35 kWh/m
2 for the ‘best’ case, with a predicted increase in average monthly outdoor temperature of over 4 °C.
There is a lack of research quantifying the impact of climate change in the Mediterranean climate in terms of thermal comfort. In this study, it is estimated that the percentage of discomfort hours during the summer period in the building category studied will increase by an average of 36.6% in 2050, while Barbosa et al. [
70], analysing a similar building in Lisbon, concluded that this increase will be under 17%. It should be taken into account that Barbosa‘s work applied the ISO-EN-15251 adaptive thermal comfort standard [
53], which is less restrictive than that used in this study [
55]. Both studies concluded that operational parameters have the highest influence on thermal comfort.
In this work, the impact of climate change has been quantified both in terms of thermal comfort and energy demand. In the residential stock evaluated, the average percentage of discomfort hours will go from 17 to 54% in 2050, i.e., it will triple. The estimated cooling demand for the ‘best’ case will increase from 10 to 35 kWh/m
2 in 2050, more than triple, while for the ‘worst’ case it will go from 51 to 92 kWh/m
2, almost double. In this case study, the results obtained in terms of comfort are much more representative of the real behaviour of the dwellings, given that there is almost no use of HVAC systems due to socio-economic reasons. However, the evaluation of the global warming impact in terms of energy demand showcases the needs of this case study to ensure full thermal comfort. Regarding the assessment of climate change impact, most studies focus on the quantification of heating and cooling demand [
7,
67,
69]. The main contribution of the proposed methodology is the ability to quantify the impact of climate change, in terms of adaptive thermal comfort, in an entire building category representing more than 40% of post-war housing stock from southern Spain, rather than focusing the study on a specific building as is frequently found in the available literature [
70,
71,
72]. Despite some references with similar aims in northern Europe [
73], published research on this topic is very limited in the south, where climate change will be most noticeable.
Mostly as a result of the inclusion of the perspective of a future climate scenario, this becomes a very useful tool for characterizing the environmental performance of the housing stock before proposing retrofitting plans which are urgently required to avoid energy poverty within this building category. It makes no sense to retrofit considering current building performance without looking to the future, given that the intervention would immediately become obsolete.
6. Strengths and Limitations
The reliability of the results obtained in this work is supported by the careful simulation process carried out using the EnergyPlus simulation tool, which is widely validated by the scientific community [
35,
74,
75]. This process takes a model calibrated thanks to in-situ measurements as a starting point [
37], resulting in a yearly value of NMBE under 3% and a CV(RMSE) value of 7%, far below the limits established in ASHRAE Guideline 14-2014 [
76]. In addition, the ‘morphing’ method applied for projecting the weather file for the year 2050, described in
Section 2.3. [
57], is one of the most widely used and best regarded by the scientific community [
59,
77].
However, the method applied also has some limitations, which are discussed in this section. The building stock model developed in this work demonstrates the great potential of broadening the sample size, controlling modelling and computation effort [
38]. However, this requires the simplification or generalization of some modelling parameters which greatly affect the thermal and energy behaviour of buildings. This is the case of user profiles, which are specific to each case study, with different occupation schedules and preferences for HVAC use [
22]. The ranges of variability proposed in this work for the operational parameters are based on user profiles defined through in-situ measurements in a limited sample [
50,
51], but these should be improved in the future through a statistical study on a larger sample representing all types of families and socioeconomic conditions. In addition, given that operational parameters were found to be the most relevant for thermal comfort in this case study, new variables such as the use of solar shading devices or light and equipment loads should be considered.
This paper focuses on the energy and environmental characterization of a particular building category, limited to post-war linear geometric typology in Seville. However, it should be feasible to complete this characterization by extending the methods applied in this work to other building categories, including all significative typologies, such as H-shape and high-rise [
61], and climate zones of the Mediterranean area. Future work will aim to address this.
Another possibility for future research is the incorporation of this prediction model as a plugin into Building Information Modelling (BIM) tools. It should read the building properties, replacing the characteristic parameters with the corresponding values in order to easily obtain the IDF file for the simulation in EnergyPlus, which will in turn provide the thermal comfort assessment.
The characterization of the energy and environmental behaviour of the housing stock under the future climate change scenario is a great starting point for proposing retrofitting measures. However, the building stock model developed in this work should be improved by including the evaluation of the energy performance of proven retrofitting measures [
78]. For future research, the authors are currently working on the optimized evaluation of different passive retrofitting strategies recognized by the scientific community as the most effective for the Mediterranean climate [
8,
15,
36]. These include night-time ventilation, window shading devices and the adoption of reflective coatings for the building shell, in order to improve thermal comfort conditions under a climate change scenario.
7. Conclusions
This study evaluates the influence of global warming on the environmental performance of an entire building category, by comparing the results obtained under typical climatic conditions and under an A2 climate change scenario for 2050. It was evaluated in terms of comfort, checking the increase of the percentage of hours outside the comfort band in 2050; that of outdoor and indoor air temperatures, quantified in degrees; and that of energy demand, quantifying the annual cooling requirement (kWh/m2). This contributes to filling the gap in research assessing the effects of global warming on adaptive thermal comfort under the severe summer climate conditions of the Mediterranean arch. Furthermore, there is an added value provided by the development of simulation models representing an important part of the residential stock, instead of a particular case study.
An increase in average monthly outdoor temperatures of about 1.5 °C in winter and 4.0 °C in summer is expected by the year 2050 in Seville. This will logically have a major impact on thermal comfort, indoor temperatures and energy demand.
Based on the characterization of the thermal comfort in post-war linear-type multi-family social housing in southern Spain, it can be determined that the increase in outdoor temperatures predicted for 2050 will reduce the percentage of discomfort hours in winter but give rise to a major increase in summer. There will be an average increase of 36.6% in discomfort hours in summer for the building category.
The results from the sensitivity analysis demonstrate the major influence of user behaviour on thermal comfort in this case study, which will increase with climate change. This should encourage energy policies requiring the characterization of user profiles prior to the proposal of retrofitting measures and their evaluation under climate conditions for 2050. Due to global warming, passive strategies such as natural ventilation will be essential in reducing indoor overheating conditions expected in the future.
The environmental evaluation of the cases with the lowest and highest percentage of discomfort hours throughout the year on a typical winter and summer day gives rise to a large increase in outdoor temperatures in a climate change scenario, up to 2 °C in winter and 5.5 °C in summer, reaching temperatures of 47.5 °C. Therefore, a large increase in indoor temperatures is also predicted for the summer, up to 5 °C, with a maximum daytime value of 42 °C as the ‘worst’ case, and a minimum value of 29 °C during the night as the ‘best’ case. These values are clearly far from adaptive thermal comfort conditions. The estimated cooling demand for these case studies will also notably increase under the climate change scenario, reaching up to three times the results obtained for typical climatic conditions.
It is essential to take the severe repercussions of the climate change scenario in southern Spain into account when planning future energy retrofit policies badly needed for this housing stock. The next steps in this research will aim to propose retrofitting measures based on the results of this characterization for a 2050 climate projection, as well as on the sensitivity analysis, in combination with an economic assessment.