Indoor Summer Thermal Comfort in a Changing Climate: The Case of a Nearly Zero Energy House in Wallonia (Belgium) †

: While the potential impact of climate change mitigation measures is well documented in building sciences literature, there are only relatively sparse studies focusing on the efﬁciency of adaptation strategies. This paper aims to contribute to this topic by evaluating the extent to which the design of a typical nearly Zero Energy Buildling (nZEB) house in Wallonia (Belgium), and its current operation, could provide summer thermal comfort in a changing climate. Based on calibrated whole building energy simulations, and on the integration of future climate data directly derived from a high-resolution climate model, this study evaluates the potential evolution of overheating risks in the living room and in the main bedroom of the house. Discussing the compliance with existing overheating criteria, the study shows that the passive strategies currently deployed in the house might not be sufﬁcient to guarantee summer thermal comfort especially in the bedroom, and that other strategies might be necessary in the future to limit the use of active cooling systems and curb their environmental impacts. This study concludes that considering the potential of these strategies to guarantee summer thermal comfort in a changing climate should be a priority for the design of nZEB houses (and their related policies) also in temperate oceanic climates.


Introduction
"Warming of the climate system is unequivocal and the effects of anthropogenic greenhouse gas emissions, together with those of other anthropogenic drivers, are extremely likely to have been the dominant cause of the observed warming since the mid-20th century", the Intergovernmental Panel on Climate Change (IPCC) writes in its Fifth Assessment Report (AR5) in 2014 [1]. In this context, a drastic reduction of greenhouse gas emissions is needed to mitigate climate change effects. For the building sector, Wallonia (southern part of Belgium), like many other European regions, aims to reach a carbon neutral building stock by 2050 [2]. This implies achieving nearly Zero Energy Buildings (nZEB) standards for all buildings, including existing ones [3]. According to the definition provided in the European Directive 2010/31/EU on the energy performance of buildings [4], nZEB implies that "the very low amount of energy required is covered to a very significant extent by energy from renewable sources".
In a region such as Wallonia, characterized by a temperate oceanic climate with predominant heating energy use (more than 70% of the energy used in residential buildings [4]), nZEB performance requirements aim first to drastically reduce heating energy needs of buildings. Practically, measures to increase the insulation and airtightness of the thermal limited [9] -be sufficient to avoid the generalisation of active cooling systems and their 45 induced energy consumption [10] and environmental impacts [11] in a country with no 46 such tradition? 47 While the impact of climate change mitigation measures is now well studied, re-48 searches focusing on the efficiency of these adaptation strategies are only at their infancy 49 in the building sector for a number of reasons: slow knowledge transfer from climate mod-50 elling to building engineering science, stochastics behaviour and uncertainties associated 51 with future climates that induce large datasets and specific method to deal with, compu-52 tational time needed, etc. [10]. These analyses are however crucial to define proper build-53 ing energy standards that will effectively reduce energy demands without compromising 54 thermal comfort. This paper aims to contribute to answer this question. Based on a real 55 case study, it evaluates the extent to which the design of a typical nZEB house in Wallonia 56 and its current operation could guarantee summer thermal comfort in a changing climate. 57 58 2.1. Case study 59 The case study is a real detached single-family house. The house is located in a sub-60 urban area of Namur (Erpent, southern part of Belgium, 50° North). In Wallonia, detached 61 houses represent the most frequent type (39%) of single-family housing (approximately 62 80% of the housing stock [12]). 63 The house, constructed in 2011, is composed of 2 levels of around 83m² (net floor 64 area) each and is occupied by 2 adults and 3 children (1 baby). Regarding its thermal per-65 formances, it complies with current requirements for nZEB in Wallonia [13] relying on the 66 definition provided in the European Directive 2010/31/EU on the energy performance of 67 buildings [14]. Main thermal characteristics of the envelope of the buildings analysed are 68 presented in Table 1.The main characteristics and performances of the house have been 69 measured or monitored during one complete year (2016): air temperatures, electricity con-70 sumption (internal gains), heating demand, airflows induced by mechanical ventilation, 71 efficiency of the heat recovery system, U values of walls, airtightness. Some weather data 72 have also been monitored on site (dry bulb temperature and relative humidity) and com-73 pleted with data from the Royal Meteorological Institute of Belgium (RMI) (radiation, 74 cloud cover, etc.). The characteristics of the house presented in Table 1 derive thus from 75 on-site verifications, interviews with the occupants (e.g., about windows opening behav-76 iours) and calculations from the Walloon EPBD (Energy Performance of Buildings) [13]. 77 In the frame of this study two main rooms were analysed : the living room (Z1) and the 78 main bedroom (Z7). Their spatial characteristics are illustrated in Figure 1.   [20]. These requirements imply a primary energy use of a maximum 85 kWh/m 2 ·year (mainly including heating, domestic hot water consumption as well as on site electricity production, here supplied by photovoltaics cells covering the roof). The nZEB standard also requires compliance with a simplified overheating risk indicator based on a static thermal calculation under the current climate [14].

Method
Following an initial measurement of its main characteristics, the performance of the house has been monitored during one full year (2016), including: air temperatures, electricity use (internal gains), heating demands, airflows induced by mechanical ventilation, efficiency of the heat recovery system, U values of the walls and airtightness. Some weather data have also been monitored on site (dry bulb temperature and relative humidity) and complemented by data from the Royal Meteorological Institute of Belgium (RMI) (radiation, cloud cover, etc.). The characteristics of the house presented in Table 1 derive, thus, from on-site verifications, interviews with the occupants (e.g., about windows opening behaviours) and calculations from the Walloon EPBD (Energy Performance of Buildings) tool [19]. In the framework of this study, two main rooms were analysed: the living room (Z1) and the main bedroom (Z7). A description of their spatial characteristics is included in Table 1  * 1 On-site verification/measurements; 2 Questionnaire-based interviews; 3 Belgian EPBD calcu tion; 4 Literature. ** n.a.: not applicable.
In terms of thermal performances, the house complies with the current requiremen for nZEB in Wallonia [19] relying on the definition provided in the European Directi 2010/31/EU on the energy performance of buildings [20]. These requirements imply a p mary energy use of a maximum 85 kWh/m 2 ·year (mainly including heating, domestic h water consumption as well as on site electricity production, here supplied by photovolt ics cells covering the roof). The nZEB standard also requires compliance with a simplifi overheating risk indicator based on a static thermal calculation under the current clima [14].
Following an initial measurement of its main characteristics, the performance of t house has been monitored during one full year (2016), including: air temperatures, ele tricity use (internal gains), heating demands, airflows induced by mechanical ventilatio efficiency of the heat recovery system, U values of the walls and airtightness. Som weather data have also been monitored on site (dry bulb temperature and relative humi ity) and complemented by data from the Royal Meteorological Institute of Belgium (RM (radiation, cloud cover, etc.). The characteristics of the house presented in Table 1 deriv thus, from on-site verifications, interviews with the occupants (e.g., about windows ope ing behaviours) and calculations from the Walloon EPBD (Energy Performance of Buil ings) tool [19]. In the framework of this study, two main rooms were analysed: the livin room (Z1) and the main bedroom (Z7). A description of their spatial characteristics is i cluded in Table 1   Based on the behaviours reported by the occupants, the closure of the white intern curtain to protect from solar radiation and the opening of windows to dissipate the he in the living room have been integrated in the thermal model (except during the perio of absence of the occupants). According to the occupants, no such strategies were reg larly applied in the main bedroom and, therefore, they have not been integrated for th room in the thermal model. Globally, no (severe) overheating problems have been mea ured during the year 2016. The house complied with all the summer thermal comfort c teria (see Section 2.4)defined by the Chartered Institution of Building Services Enginee (CIBSE) (UK), even if the occupants perceived the bedroom temperature in summer Based on the behaviours reported by the occupants, the closure of the white internal curtain to protect from solar radiation and the opening of windows to dissipate the heat in the living room have been integrated in the thermal model (except during the periods of absence of the occupants). According to the occupants, no such strategies were regularly applied in the main bedroom and, therefore, they have not been integrated for this room in the thermal model. Globally, no (severe) overheating problems have been measured during the year 2016. The house complied with all the summer thermal comfort criteria (see Section 2.4) defined by the Chartered Institution of Building Services Engineers (CIBSE) (UK), even if the occupants perceived the bedroom temperature in summer as slightly too hot. The methodologies implemented for interviews and data collection are described in more detail in a previous publication [21].
The house is located in a suburban area of Namur (the capital city of Wallonia), in the city of Erpent in the southern part of Belgium (50 • North) (see Figure 3) and is exposed to a temperate oceanic climate (Cfb zone according to the Köppen-Geiger climate classification system [22]). Section 2.3 presents an overview of the current and potential future characteristics of this climate. described in more detail in a previous publication [21].
The house is located in a suburban area of Namur (the capital city of Wallonia the city of Erpent in the southern part of Belgium (50° North) (see Figure 3) and is exp to a temperate oceanic climate (Cfb zone according to the Köppen-Geiger climate cl fication system [22]). Section 2.3 presents an overview of the current and potential fu characteristics of this climate.

Calibrated Thermal Model
A whole building energy simulation of the house has been carried out ( Figure 1 ing the TRNsys18 simulation engine [24]. In order to ensure a realistic evaluation o indoor summer thermal conditions, a calibration process of the model was perform The calibration consisted in two main steps. First, measured values (thermal proper weather data, etc.) (cf. Table 1) were introduced in the model (=base model). Then, a iation of the entries data within the range of their estimated instrumental uncertain (obtained from manufacturers) was performed in order to minimise the deviation of ulated temperatures (hourly values) from measured values according to the widely u (see for example [25]) indicators defined by ASHRAE guideline 14 [26]. These indica are based on the Coefficient of Variation of the Root Mean-Square Error (CV-RMSE) the Mean Bias Error (MBE). This last calibration step was performed using an optimisa algorithm (Particle Swarm Optimization (PSO) [27]) and has involved more than 900 ulations. Table 2 presents the compliance of the final model with ASHRAE indicators. last column of Table 1 illustrates in detail the related entries data.   [28] (for an overvie climatic historical data, see Figure 5).

Calibrated Thermal Model
A whole building energy simulation of the house has been carried out (Figure 1) using the TRNsys18 simulation engine [24]. In order to ensure a realistic evaluation of the indoor summer thermal conditions, a calibration process of the model was performed. The calibration consisted in two main steps. First, measured values (thermal properties, weather data, etc.) (cf. Table 1) were introduced in the model (=base model). Then, a variation of the entries data within the range of their estimated instrumental uncertainties (obtained from manufacturers) was performed in order to minimise the deviation of simulated temperatures (hourly values) from measured values according to the widely used (see for example [25]) indicators defined by ASHRAE guideline 14 [26]. These indicators are based on the Coefficient of Variation of the Root Mean-Square Error (CV-RMSE) and the Mean Bias Error (MBE). This last calibration step was performed using an optimisation algorithm (Particle Swarm Optimization (PSO) [27]) and has involved more than 900 simulations. Table 2 presents the compliance of the final model with ASHRAE indicators. The last column of Table 1 illustrates in detail the related entries data.  Figure 4 illustrates the capability of the calibrated model to estimate the time exceeding 25 • C in the living room and in the main bedroom compared to the values that were measured in 2016. The year 2016 can be considered broadly in line with the average of the last 30 years, both in terms of solar radiation and of temperatures [28] (for an overview of climatic historical data, see Figure 5).

Climate Change Model
Data for future and historical (here intended as similar to current) climatic conditions (temperature, humidity, radiation, wind speed and direction, etc.) have been directly derived from high-resolution climate modelling. These models, dynamically downscaled from Global Climate Models (GCMs), consider local climate variations and anomalies and allow a detailed temporal and geographical resolution (horizontal grid spacing <4 km). Compared to the classical use of morphed weather data [30], this method allows for accounting the variability in occurrence and intensity of heatwaves phenomena throughout the years. Today, these data based on the ALARO model are available in Belgium for different representative concentration pathways (RCP) [29]. For the purpose of this paper, two future climates scenarios-Representative Concentration Pathway (RCP) 2.6 and 8.5 -for the period 2071-2100 were considered and compared to historical data  (also derived from the same climate model) for the city of Uccle (the main meteorological station of Belgium, located in the southern part of Brussels, about 50 km from Namur). These two RCPs scenarios were chosen since they represent opposite (and extreme) perspectives with respect to the deployment of mitigation strategies and, consequently, the severity of induced climate change: a stringent mitigation scenario (RCP2.6) that would keep global warming likely below 2 °C above pre-industrial temperatures, which is in line with the objectives of the Paris agreement [31], and a very high Green House Gas (GHG) emissions scenario (RCP 8.5), leading to an increase of global mean temperature comprised between 2.6 °C to 4.8 °C (95% confidence interval) by the end of the 21st century

Climate Change Model
Data for future and historical (here intended as similar to current) climatic conditions (temperature, humidity, radiation, wind speed and direction, etc.) have been directly derived from high-resolution climate modelling. These models, dynamically downscaled from Global Climate Models (GCMs), consider local climate variations and anomalies and allow a detailed temporal and geographical resolution (horizontal grid spacing <4 km). Compared to the classical use of morphed weather data [30], this method allows for accounting the variability in occurrence and intensity of heatwaves phenomena throughout the years. Today, these data based on the ALARO model are available in Belgium for different representative concentration pathways (RCP) [29]. For the purpose of this paper, two future climates scenarios-Representative Concentration Pathway (RCP) 2.6 and 8.5 -for the period 2071-2100 were considered and compared to historical data  (also derived from the same climate model) for the city of Uccle (the main meteorological station of Belgium, located in the southern part of Brussels, about 50 km from Namur). These two RCPs scenarios were chosen since they represent opposite (and extreme) perspectives with respect to the deployment of mitigation strategies and, consequently, the severity of induced climate change: a stringent mitigation scenario (RCP2.6) that would keep global warming likely below 2 °C above pre-industrial temperatures, which is in line with the objectives of the Paris agreement [31], and a very high Green House Gas (GHG) emissions scenario (RCP 8.5), leading to an increase of global mean temperature comprised between 2.6 °C to 4.8 °C (95% confidence interval) by the end of the 21st century

Climate Change Model
Data for future and historical (here intended as similar to current) climatic conditions (temperature, humidity, radiation, wind speed and direction, etc.) have been directly derived from high-resolution climate modelling. These models, dynamically downscaled from Global Climate Models (GCMs), consider local climate variations and anomalies and allow a detailed temporal and geographical resolution (horizontal grid spacing <4 km). Compared to the classical use of morphed weather data [30], this method allows for accounting the variability in occurrence and intensity of heatwaves phenomena throughout the years. Today, these data based on the ALARO model are available in Belgium for different representative concentration pathways (RCP) [29]. For the purpose of this paper, two future climates scenarios-Representative Concentration Pathway (RCP) 2.6 and 8.5for the period 2071-2100 were considered and compared to historical data (1976-2005) (also derived from the same climate model) for the city of Uccle (the main meteorological station of Belgium, located in the southern part of Brussels, about 50 km from Namur). These two RCPs scenarios were chosen since they represent opposite (and extreme) perspectives with respect to the deployment of mitigation strategies and, consequently, the severity of induced climate change: a stringent mitigation scenario (RCP2.6) that would keep global warming likely below 2 • C above pre-industrial temperatures, which is in line with the objectives of the Paris agreement [31], and a very high Green House Gas (GHG) emissions scenario (RCP 8.5), leading to an increase of global mean temperature comprised between 2.6 • C to 4.8 • C (95% confidence interval) by the end of the 21st century (2081-2100) relative to 1986-2005 [1]. Figure 5 shows the monthly mean outdoor temperature evolution according to these scenarios compared to the historical one (yearly global horizontal radiation remains stable-around 1040 kWh/m 2 ·year-under these different scenarios).
As mentioned in the introduction, there is wide consensus that these scenarios will impact the frequency and the duration of heat waves [1]. In Belgium, heatwaves are defined as a period of minimum 5 consecutive days with air temperature reaching at least 25 • C and more than 30 • C for at least 3 days [32]. Figure 6 presents the evolution of the number of heat waves per year as well as the number of days reaching 25 • C and 30 • C according to RCP 2.6 and 8.5 for the end of the century (2071-2100) compared to the historical period  for Uccle. A standard boxplot representation has been used to illustrate the distribution (min, 25th percentile, median, 75th percentile, max) of the encountered values for the considered period. Median values have also been specified to increase the readability of these graphs. These sets of data, covering a period of 30 years each, were directly introduced in the TRNsys simulation software with an hourly resolution (using an IWEC format [33]) in order to provide a detailed overview of their implications on the building indoor climate.  Figure 5 shows the monthly mean outdoor temperature evolution according to these scenarios compared to the historical one (yearly global horizontal radiation remains stable-around 1040 kWh/m 2 ·year-under these different scenarios). As mentioned in the introduction, there is wide consensus that these scenarios will impact the frequency and the duration of heat waves [1]. In Belgium, heatwaves are defined as a period of minimum 5 consecutive days with air temperature reaching at least 25 °C and more than 30 °C for at least 3 days [32]. Figure 6 presents the evolution of the number of heat waves per year as well as the number of days reaching 25 °C and 30 °C according to RCP 2.6 and 8.5 for the end of the century (2071-2100) compared to the historical period  for Uccle. A standard boxplot representation has been used to illustrate the distribution (min, 25th percentile, median, 75th percentile, max) of the encountered values for the considered period. Median values have also been specified to increase the readability of these graphs. These sets of data, covering a period of 30 years each, were directly introduced in the TRNsys simulation software with an hourly resolution (using an IWEC format [33]) in order to provide a detailed overview of their implications on the building indoor climate.

Summer Thermal Comfort Evaluation
The evaluation of summer thermal comfort has been conducted according to CIBSE Technical Memoranda TM52 [34] and TM59 [35], as they offer a comprehensive and widely-used method for overheating risk analysis in buildings (see for example [36]). The first document (TM52) presents three criteria based on the frequency, duration and intensity of the deviations of operative temperature from the EN15251 comfort categories [37]: Adaptive hours of exceedance (Tmax + 1); Daily weighted exceedance (We) and Upper limit temperature (Tupp) [34]. In the context of climate change scenarios, these criteria-based on the theory of adaptive comfort-seem particularly appropriate as they already imply a physiological, behavioural and psychological adaptive capacity of humans [38,39]. In homes, CIBSE only imposes compliance with the first criterion (Adaptive hours of exceedance) for living rooms, kitchens, and bedrooms [35]. Nevertheless, all these criteria have been considered in this study with respect to the limit of the comfort Category II [37], recommended for new buildings and major refurbishments in the absence of very sensitive and

Summer Thermal Comfort Evaluation
The evaluation of summer thermal comfort has been conducted according to CIBSE Technical Memoranda TM52 [34] and TM59 [35], as they offer a comprehensive and widelyused method for overheating risk analysis in buildings (see for example [36]). The first document (TM52) presents three criteria based on the frequency, duration and intensity of the deviations of operative temperature from the EN15251 comfort categories [37]: Adaptive hours of exceedance (T max + 1); Daily weighted exceedance (W e ) and Upper limit temperature (T upp ) [34]. In the context of climate change scenarios, these criteria-based on the theory of adaptive comfort-seem particularly appropriate as they already imply a physiological, behavioural and psychological adaptive capacity of humans [38,39]. In homes, CIBSE only imposes compliance with the first criterion (Adaptive hours of exceedance) for living rooms, kitchens, and bedrooms [35]. Nevertheless, all these criteria have been considered in this study with respect to the limit of the comfort Category II [37], recommended for new buildings and major refurbishments in the absence of very sensitive and fragile persons [40]. For bedrooms, a criterion based on an absolute threshold of 26 • C was also considered as suggested by TM59 [35]. This static criterion is based on the fact that thermal adaptation is considerably reduced during sleep [41]. Table 3 illustrates in detail the application of these criteria. In accordance with TM59, occupied hours were considered to be from 7 am to 10 pm for the living room and from 10 pm to 7 am for the bedroom [35].  (1976-2005), RCP2.6 (2071-2100) and RCP8.5 (2071-2100) scenarios resulting from the application of the criteria presented in Table 3 for the living room ( Figure 7) and for the main bedroom (Figure 8) of the house. fragile persons [40]. For bedrooms, a criterion based on an absolute threshold of 26 °C was also considered as suggested by TM59 [35]. This static criterion is based on the fact that thermal adaptation is considerably reduced during sleep [41]. Table 3 illustrates in detail the application of these criteria. In accordance with TM59, occupied hours were considered to be from 7 am to 10 pm for the living room and from 10 pm to 7 am for the bedroom [35].   Table 3 for the living room ( Figure 7) and for the main bedroom (Figure 8) of the house.  Table 3 for the living room.  Table 3 for the living room.  Table 3 for the bedroom.

Results
For the living room, Figure 7 shows that the Adaptive hours of exceedance criteria (not more than 3% of occupied hours where Top -Tmax > 1 K) is met under the three climatic scenarios. The Daily weighted exceedance criterion is met for 75% of years between 1976 and 2005, but this result could drop under 50% of the years between 2071-2100 (38% under RCP8.5 scenario). Moreover, this criterion is not met for more than 2 days per year for half of the years between 2071 and 2100 under the RCP8.5, while it is only exceeded once per year in 25% of the years between 1976 and 2005. The Upper limit temperature criteria is only exceeded under the RCP8.5 scenario (three times and up to 6 K). For the main bedroom, Figure 8 shows that the Static hours of exceedance criterion (max 1% of occupied hours per year where Top > 26 °C ) is only exceeded 25% of the years during the historical period. The non-compliance with this criterion could, however, become systematic: 25 years are exceeding this criterion over the period 2071-2100 under the RCP8.5 scenario. On the contrary, the Adaptive hours of exceedance criterion (not more than 3% of occupied hours where Top − Tmax > 1 K) is exceeded only one year out of four under this scenario (while it is met under the other scenarios). The median value of the number of days exceeding the Daily weighted exceedance reaches 4 between 2071-2100 (under RCP8.5 scenario), while it is 0 for the historical period. The Upper limit temperature criterion is not frequently exceeded but it reaches higher frequencies per year under the RCP8.5 scenario (up to 39 times per year).

Discussion
As shown in Figure 6, days over 25 °C and heat waves will become quite common in the future in Belgium: around 27 days reaching 25 °C and one heatwave (up to four) on  Table 3 for the bedroom.
For the living room, Figure 7 shows that the Adaptive hours of exceedance criteria (not more than 3% of occupied hours where T op − T max > 1 K) is met under the three climatic scenarios. The Daily weighted exceedance criterion is met for 75% of years between 1976 and 2005, but this result could drop under 50% of the years between 2071-2100 (38% under RCP8.5 scenario). Moreover, this criterion is not met for more than 2 days per year for half of the years between 2071 and 2100 under the RCP8.5, while it is only exceeded once per year in 25% of the years between 1976 and 2005. The Upper limit temperature criteria is only exceeded under the RCP8.5 scenario (three times and up to 6 K). For the main bedroom, Figure 8 shows that the Static hours of exceedance criterion (max 1% of occupied hours per year where T op > 26 • C) is only exceeded 25% of the years during the historical period. The non-compliance with this criterion could, however, become systematic: 25 years are exceeding this criterion over the period 2071-2100 under the RCP8.5 scenario. On the contrary, the Adaptive hours of exceedance criterion (not more than 3% of occupied hours where T op − T max > 1 K) is exceeded only one year out of four under this scenario (while it is met under the other scenarios). The median value of the number of days exceeding the Daily weighted exceedance reaches 4 between 2071-2100 (under RCP8.5 scenario), while it is 0 for the historical period. The Upper limit temperature criterion is not frequently exceeded but it reaches higher frequencies per year under the RCP8.5 scenario (up to 39 times per year).

Discussion
As shown in Figure 6, days over 25 • C and heat waves will become quite common in the future in Belgium: around 27 days reaching 25 • C and one heatwave (up to four) on average per year should be excepted under the RCP8.5 scenario. Conversely, these phenomena were very exceptional in the historical data (only one year with heatwaves and a mean value of less than 11 days per year reaching 25 • C).
The analysis of the historical period shows that the implementation of passive strategies in the building (e.g., internal shading devices and natural cooling by windows opening only practiced in the living room) generally allows compliance with the CIBSE overheating criteria. It should also be remembered here that the configuration of the building analysed for this study (east oriented, limited glazing area, suburban area, etc.) allowed to reduce the occurrence of overheating phenomena.
The evolution of the climate could have, however, a clear impact on future indoor climate conditions, as shown in Figures 7 and 8. The severity of this evolution depends on the criterion and on the studied room. In this way, the results confirm the observation made in a previous study [42].
More specifically, the CIBSE Daily weighted exceedance criterion appears to be difficult to meet in the future (more than 50% of the years during the period 2071-2100 under RCP2.6 do not comply with this criterion for both the living room and the bedroom). This shows an increasing difficulty in dissipating the heat accumulated inside the building, probably linked to higher external temperatures. The bedroom suffers particularly from this phenomenon (median value of 4 days per year where the Daily weighted exceedance criterion is reached under the RCP8.5 scenario, versus 2 for the living room). The Static hours of exceedance (T op > 26 • C) criterion in the bedroom appears to be the most difficult to comply with in the future (criterion exceeded more than 50% of the years during the period 2071-2100 under the RCP2.6 scenario, and more than 75% of the years under the RCP8.5).
The Adaptive hours of exceedance and Upper limit temperature criteria are, however, rarely exceeded. These CIBSE criteria based on adaptive comfort appear, thus, easier to meet than benchmarks based on a fixed static value. It has been already shown that these criteria, based on a significant deviation from the comfort limits set by EN15251, are not very restrictive for this type of building and are not always representative of the number of complaints reported by the occupants [21]. It is, therefore, crucial to properly (re)consider (this also probably depending on the room type) the applicability and the limits of adaptive thermal comfort principles and strategies in residential buildings (EN15251 was originally established from data collected in offices) [43,44].
More generally, the passive strategies currently deployed (internal solar shading and natural cooling by windows opening in the living room only) might not be sufficient in the future. The widespread use of window curtains (which could be replaced by external solar shadings) and natural cooling by windows opening in all rooms (including bedrooms) could certainly contribute to improve the summer thermal comfort of the house, and probably limit the need of active cooling systems in the future. In this context, the potential decrease of (night)ventilative cooling efficiency due to higher (night)temperatures induced by climate evolution (as illustrated by Figure 5) should be explored in detail.
Further research focusing on the potential of such strategies in a global warming context is, therefore, needed. A more thorough understanding appears crucial to limit the potential global increase in cooling energy consumption [16] and to properly define policies and standards aiming at a better resilience of the building stock to heat waves [45].

Conclusions
Based on a whole building energy simulation, this paper investigated the potential evolution of summer thermal comfort in a typical nZEB house under different climate evolution scenarios (RCP2.6 and RCP8.5).
The study, based on compliance with CIBSE TM52 and TM59 summer thermal comfort criteria analyses, has shown that the passive strategies currently used in the selected case study (internal solar shading and intensive natural ventilation in the living room only) will not be sufficient in the future (especially in the bedroom) and that other strategies will be necessary to guarantee summer thermal comfort.
It has also been noticed that CIBSE criteria based on the theory of adaptive thermal comfort are much less restrictive in a context of changing climate than criteria based on a fixed static value. It is, therefore, crucial to verify the applicability of adaptive thermal comfort defined in EN15251 to the residential sector and its limits in the context of a changing climate.
More generally, the paper shows that considering summer thermal comfort, and its possible evolution, should be a priority for the design of nZEB houses (and related standards and policies) to avoid the generalised use of active cooling systems, with their induced environmental impacts, in temperate oceanic climates. Therefore, the full potential of passive design strategies to ensure summer thermal comfort under future climate scenarios have to be analysed more in depth.