SWORD - South West Open Research SWORD - South West Open Research Passive Control Strategies for Cooling a Non-Residential Nearly Passive Control Strategies for Cooling a Non-Residential Nearly Zero Energy Office: Simulated Comfort Resilience Now and in the Zero Energy Office: Simulated Comfort Resilience Now and in the Future Future

The average global cooling demand in non-residential buildings is expected to increase by over 275% between now and 2050. Controlled passive cooling is fundamental to successful operational performance of buildings and in mitigating energy that would otherwise be consumed by mechanical systems. The aim of this study was to determine the resilience of different passive cooling control strategies in delivering optimal comfort and energy scenarios in both current and future extreme conditions, for low energy indoor ofﬁce spaces. Simulations were conducted using a calibrated TRNSYS 17 model of a nearly zero energy building. The performance of ten passive cooling control strategies was simulated for climatic conditions in two representative cities, Dublin and Budapest. Each strategy used different combinations of passive cooling systems such as day-time ventilation, night-time ventilation and dynamic solar shad- ing. The effect of static and adaptive indoor temperature set-points and a limit on external relative humidity was also investigated. The thermal comfort performance of each strategy was assessed by using standardised thermal comfort, overheating and overcooling metrics. Findings from the study show that passive control strategies maintained comfortable internal conditions between 57% and 95% of the occu- pied hours, without the need for mechanical cooling. The most resilient strategies were those that combined multiple measures. Passive control strategies were found to be resilient in the medium-term in Dublin, however, the same systems were not able to maintain comfortable conditions in Budapest in 2050. The use of an external relative humidity limit resulted in increases overheating incidences and fail- ure of some overheating criteria. Based on the reductions in mechanical cooling requirements, it was con-cluded that there is regulatory need to consider the use passive control strategies in the design of buildings. (cid:1) 2020 The Authors. Published


Introduction
Between now and 2050 the global average cooling demand in commercial buildings is expected to increase by up to 275% [1]. Increases in cooling demand are likely to lead to a tripling in the energy demand for air-conditioning by 2050 [2]. Cooling systems or techniques that avoid or reduce the need for air-conditioning are therefore important if energy efficiency targets for buildings are to be achieved [3]. Controlled passive cooling systems could play a key role in removing the need for mechanical cooling systems.
A passive cooling system refers to a system that regulates the internal environmental conditions in a building without the need for energy consumption (excluding the energy for actuation or control) [1]. Passive cooling control strategies refer to the combination or application of one or more passive control systems to control the internal conditions in a building. Passively controlling internal conditions can involve heat prevention (e.g. solar heat control), dissipation (into sources like the air, ground or water), or can incorporate some storage or modulation of heat (e.g. thermal mass storage) [4]. Passive cooling strategies have been used in many buildings and their performance have been presented in many studies. Previous work from International Energy Agency, Energy in Buildings and Communities Programme (IEA-EBC) Annexes demonstrated that passive ventilative cooling control strategies [5] or hybrid cooling strategies [6] are viable options for maintaining internal conditions. Measured data from case study buildings in Annex 62 indicated low levels of overheating (<2.6% of the occupied hours greater than 28°C) for buildings with largely naturally driven ventilation systems [5].
Studies focused on controlled passive cooling systems or strategies have often analysed the control of one to two passive cooling systems applied to one building type [7][8][9][10][11][12]. Moeseke et al.
investigated the control of both solar shading and natural ventilation (NV) systems by simulating the energy and comfort performance of a small office building and reported that shading control strategies that used set-points for both internal temperature and external irradiance were found to be more efficient than solar shading strategies that considered only one set-point for either [10]. Karlsen et al. measured and simulated the energy and comfort performance of office buildings in a cold climate using different solar shading control strategies. Their work evaluated the comfort and energy performance with varying levels of complexity in the control of both internal and external shading devices. An optimised control strategy, operating both internal and external shading devices, or, the detailed control of external shading, were found to be best when optimising for both comfort and energy [8]. Fiorentini et al. simulated the effectiveness of various NV control strategies when applied to a house during summer conditions in Sydney, and compared these strategies to a mechanical system for the same building. Day-time NV alone was found to reduce the thermal energy demand of the building by 28.9% when compared to mechanical ventilation (MV). The combined effect of both day-time and night-time ventilation reduced the thermal energy demand of the building by 54.9% when compared to the same MV system [12]. Psomas et al. simulated the control of roof windows in energy renovated dwellings in Copenhagen. This study compared manual and automated control strategies and then assessed the overheating risk using static and dynamic methods. The study found that manual control of roof windows was not sufficient for mitigating overheating risk. However, automated control of roof windows was considered sufficient to mitigate overheating risk [7]. Schulze and Eicker et al. studied the simulated performance of controlled NV in modern buildings located in the moderate climate of Stuttgart in Germany [9]. In this work, it was found that controlled NV can meet the comfort cooling requirements of modern office buildings without the need for MV [9]. Chen et al. simulated varying levels of automation in the control of NV for different climates in China. In this study, varying levels of occupant control were simulated and the performance was evaluated based on energy consumption, indoor thermal comfort and frequency of operation of each theoretical NV system. This study found that automated window control performed best from a thermal comfort perspective and could result in energy savings of between 10% and 80% when compared to an MV system [11]. These examples highlight the benefits of controlled passive cooling systems in current conditions. However, few of these studies investigated the effect of combining multiple passive cooling systems [10], are largely related to control in domestic or generic test buildings [8,10,12,13] and none indicated the performance of controlled passive cooling systems in future conditions.
One of the main risks associated with adopting passive cooling strategies is that their applicability or lifecycle [14] can be limited by future climate change or extreme conditions [15,16]. Lomas and Ji et al. simulated the performance of hospital spaces that used simple single-sided NV and compared them to advanced NV systems and an MV system. In this study, the effect of climate change was investigated by using design summer year (DSY) weather files. It was found that advanced NV outperformed both simple singlesided NV and MV systems from both energy and comfort perspectives. In this study, the overheating risk associated with simple natural ventilation systems were found to be unacceptable. Controlling internal heat gains and external solar gain were seen as important components in reducing this overheating risk [14]. Breesch and Janssens et al. simulated the thermal comfort performance of natural night ventilation for office spaces in Belgium by investigating the uncertainty and sensitivity associated with different NV principles (i.e. single-sided, cross flow and stack) and weather files. Night-time stack ventilation was found to be the least uncertain of all the NV principles when considered under current conditions. Night ventilation systems that were designed using typical weather files were found to be insufficient at regulating thermal comfort under warmer ten-year weather files.
Heat gains (W) found that night ventilation alone was not sufficient for regulating internal conditions. Combining night ventilation with additional measures such as thermal mass increased the likelihood of improved future thermal comfort performance [16]. Pagliano et al. investigated potential options for the energy retrofit of a care centre in Milan. In this study, passive systems designed for current conditions were simulated using future weather files. It was found that passive cooling systems that use multiple measures such as natural ventilation and solar shading, are likely to be able to maintain thermally comfortable conditions for the medium-term. However, it is likely that active cooling will be required in the long-term [15]. These studies highlight the importance of simulating the performance of NV buildings in the future [15] and in extreme conditions [14,16], however, these studies represent a small number of locations and climates, few examine the control of multiple passive systems [14], and few investigate the impact of relative humidity when controlling passive systems [16]. From a regulatory perspective, the energy performance of buildings directive (EPBD) stipulates that all new buildings are now required to be at nearly zero energy building (nZEB) levels [17]. Although nZEB levels of performance can be different depending on the member state [18], there are common characteristics in their design. Annunziata et al. examined the national regulations in various EU member states and provided and overview of existing national regulatory frameworks. This study reported that while different countries used different approaches when designing their regulatory frameworks, commonality existed in that the vast majority of member states prioritised roof, wall and window insulation [19]. This common characteristic in nZEB building design was also supported in other work analysing thermal regulations [20] and main construction features of nZEBs [21]. Papadopoulos et al. presented achievements, perspectives and challenges in thermal building regulations in Europe for the past 40 years. This analysis indicated that the vast majority of 27 EU member states had requirements in their national building codes for various elements of the building fabric including: u-values (100%), thermal bridges (88%), air tightness (81%) and ventilation (77%). However, only a limited number of building codes were found to have requirements for window sizing (23%) and solar protection (8%) [20]. Paoletti et al. presented an overview of the main construction features of nZEBs from the EU IEE ZEBRA2020 project. This work indicated that nZEB design approaches were not influenced significantly by climatic conditions. The vast majority of the buildings that were studied had high levels of insulation (median u-values (W/m 2 /K): wall (0.14), roof (0.12), floor (0.16), window (0.96)) and relied on specific technologies such as, heat pumps (heating mode (32%), cooling mode (75%)) and mechanical ventilation (with heat recovery (84%), without heat recovery (5%)), and only 36% declared passive solutions [21].
Passive cooling control strategies have shown to be effective at reducing energy consumption without compromising on thermal comfort. Studies have highlighted that the control or automation of passive cooling can be a key component to the successful performance of passive cooling strategies [8,9,11,13]. However, Carrilho da Graca and Linden et al. highlighted the lack of adoption of passive cooling systems such as NV systems for use in modern nonresidential buildings. The challenges for NV systems, the reasons for the lack of adoption of NV systems, as well as some of the questions that need addressing for NV research were presented. One of the questions which needed further investigation was what impact climate change would have on the use of NV systems [22]. Understanding the performance of passive cooling strategies in extreme conditions resulting from climate change has also formed a key aspect of the scope of emerging and ongoing international collaborative research projects [23,24]. These projects and others like them could be crucial in ensuring that the new and existing buildings perform consistently in the medium to long-term. The homogenous nature of existing nZEBs (indicated previously) has led to buildings which are, highly insulated [21,25,26], decoupled from climatic conditions [21], mechanically ventilated [21], and to a large extent, actively cooled [21]. There is a significant risk that, by design (which has been heating season dominant), nZEBs are likely to be susceptible to increased overheating events in the future, where, with their existing and typical energy systems, will lead to increased energy consumption for cooling. Understanding the future risks for nZEB buildings is therefore crucial, however, studies which demonstrate the risk of nZEB designs are limited [15].
In order to address the gaps identified in the literature above, this paper presents a simulation based study determining the comfort resilience of various passive cooling control strategies, when applied to an office test-bed nZEB in different locations, for current and future extreme weather conditions. This study investigates combinations of day-time ventilation, night-time ventilation and solar shading as part of an overall passive cooling control strategy. The effect of using adaptive and non-adaptive set-points and an external RH limit are also investigated. The overall aim of this study is to determine the resilience of different passive control strategies at delivering optimal comfort and energy scenarios, in current and future extreme conditions. The methods section of this paper is broken into four main sub-sections: Section 2.1 describes the simulation model used, Section 2.2 describes the operation of passive control strategies, Section 2.3 describes the weather data that is used, and Section 2.4 describes the comfort, climate and energy assessment methods used to evaluate performance. The results section is broken into two main subsections. Section 3.1 presents general climate results when using different weather files, Section 3.2 presents the categorical comfort results for all control strategies, Section 3.3 presents a comparison of each strategy with reference to previous work and Section 3.4 presents the limitations of the work in this paper. Section 4 presents the conclusions of this paper and future work that is required to further develop the work presented.

Simulation method
The simulation method used in this study was a whole building energy simulation method using a multi-zone building model in the simulation software TRNSYS version 17 [27]. A model of the NBERT building (described below) was made using Type 56, which is the standard multi-zone model in TRNSYS, where each room was modelled as a uni-nodal zone. The model (which is mechanistic and physics-based)uses an energy balance of various heat gains (heating, cooling, ventilation, internal and solar gains etc.) to solve for the internal air temperature at each user-defined time-step. The energy balance due to natural ventilation was defined using an airflow network in TRNFLOW. This airflow network linked internal zones to each other, and the connected these zones to the outside using a series of cracks for infiltration, and openings (defined by a discharge co-efficient, the width and height of an opening) for natural ventilation. More information on this method can be found in O' Donovan et al. [28] and the TRNFLOW manual [29].

Natural ventilation system and airflow network
The National Built Energy Retrofit Test-bed (NBERT, messo.cit. ie/nbert) is a 223 m 2 educational test-bed building that is part of the wider Cork Institute of Technology (CIT) main campus in Bishopstown, Co. Cork, Ireland. NBERT is based within the zero2020 building (See Fig. 1), which is a renovated part of the existing CIT main campus building that was originally completed in 1974 and renovated in 2012. The building functions as a live test-bed for a range of research activities in areas such as: energy performance, micro-grid [30], ventilation [31] and thermal comfort applications [32]. The NBERT building is an nZEB which has comparable heating energy consumption to other nZEB retrofit examples in similar climates [33], has similar technologies and construction characteristics in nZEB designs [21], and has similar if not superior envelope characteristics to other case studies with passive or ventilative cooling systems in Europe [5]. A detailed description of building and the calibrated and validated initial model used in this study can be found in O'Donovan et al. [28]. The calibrated model was capable of predicting indoor air temperatures with a RMSE of between 0.27°C and 1.50°C, depending on the season. In this study, the simulated ventilation system differed from the this calibrated model and was simplified by having one large opening per façade that ensured each room had a proportion of net openable area to floor area ratio (POF) of 2.5%, which is reasonably representative of international cases studies [34]. O' Donovan et al. suggested that openings at head height should be avoided in shoulder seasons in order to avoid discomfort [32]. In this study, each opening in each room was placed at a height that was greater than 1.5 m above floor level. Given that the maximum floor to ceiling height of the room was 3.2 m, the maximum height of each opening was considered to be 1.7 m, with one opening defined for each façade (i.e. west or south facing).
This simplification (in combining multiple openings and maximising height) is likely to lead to differences between the ventilation system of the NBERT and the theoretical system proposed here. However, by maximising height this work presents the best case for single-sided and minor differences are expected by combining openings due to the underlying relationships in TRNFLOW. Equally, as all openings are automated, a single opening was considered representative of multiple acting in unison. Table 1 shows the dimensions of each opening created and simulated as part of the airflow network created in TRNFLOW.
The floor to ceiling height was taken as the same as the NBERT test-bed building which is above the minimum floor to ceiling height recommended for buildings in Ireland [35]. A total of four zones were modelled and simulated, however, the results presented in this paper are for are for the Open Plan Office only.

Solar shading system
An external solar shading system was simulated and operated based on the control strategies shown in Section 2.2. The theoretical external solar shading system used in this study (shown in Figure 2) activates a perforated external solar shading device which is 80% opaque when certain conditions are met (20% transmittance assumed). Perforated external shading devices have been described in detail in previous reviews of solar shading devices [36,37]. Perforated shading has been used by many researchers in the past where they remain as a static feature of the building [38][39][40]. A previous study considered the performance of a sliding perforated aluminium panel, however, the panel was static and was not automated or actuation was not tested [41]. In the study presented in this paper, the assumption is that a perforated shading device can be automated for parts of the year when it is necessary. This was done as research has indicated that the optimum perforation percentage can vary depending on the orientation and the season [38]. Automating the perforated shading device was used to activate shading when shading was needed and to retract shading when it was not needed. A perforated shade was selected over a unperforated shade, as a result of the need to allow occupants to have a view to the outside, this would maintain visual comfort levels [40]. Fig. 2 shows the difference in solar transmittance through the buildings glazing system as a result of surface dirt and through the use of the theoretical perforated screen. When the shade was not activated, it was assumed that the layer on the outside of the window (e.g. dirt) results in 39% (i.e. 61% opaque fraction, E d ) of the radiation passing through the first pane of the windows in the building (see Fig. 2). This value was taken from measured percentage of transmitted solar radiation from external values that were representative of values used in the NBERT [28]. When the external shading device was activated, only 8% (i.e. 92% opaque fraction, E s ) of the radiation is passed to the first pane of the windows in the building. Table 2 indicates the heat gains into each zone in the building model. The maximum capacity of the heating system was based on the radiator capacities in the NBERT building. The zone capacitance in Type 56 was set to be five times the zone volume with the guidance from literature suggesting values higher than the air capacitance (typically between 5 and 10 times) when using TRNSYS [42,43] and other similar simulation tools [44,45]. To scale the capacity of the heating system, the percentage of the maximum delivered heat gain to each zone was weather compensated (c wc ) depending on the outside air temperature T e (see Fig. 3). The heating system also followed a schedule for working hours between 08:00 and 18:00 and did not operate during public holidays (c occu ). All gains from the heating system to a zone were taken to be radiative gains. The heat gain into each zone due to the heating system, was dependant on the output of a heating dead-band controller (c set,h ) which typically operated with a set-point of 21°C (±1°C) and varied depending on the control strategies indicated in Section 2.2. Equation (1) describes the heat gain from the heating system into each zone (h h ).

Heat gains, energy systems, and gains from occupants
where, A f describes the floor area of each zone (m 2 ), G h is the heat density taken from Table 2, c occu is the occupancy state of each zone, c wc is the percentage of maximum heating required that is weather-compensated, and c set,h is the controller output of the dead-band controller. For appliances and lighting, a diversity factor for weekdays was applied based on the cellular and landscaped office diversity factors (c div ) shown in Fig. 3. Using ISO 17772-1, gains from appliances and lighting was assumed to be zero on weekends.
The percentage of radiative and convective gains for appliances and lighting was taken from O'Donovan et al. [28]. The heat gains from occupants in the building model were taken to be stochastic and were generated using a stochastic occupancy generator [48]. A  single annual dataset of occupancy values was generated for each location and considered national holidays. These datasets were then post-synchronised to hourly intervals. The appliances and lighting systems in each room were considered to be off during holidays. Equation (2) describes the heat gains due to appliances (h A ) at each time-step into each zone. h where, G A is the appliance density taken from Table 2, and c div describes the percentage of maximum density for appliances and lighting respectively expressed as a diversity factor (see Fig. 3). h Equation (3) describes the heat gains due to lighting (h L ) at each time-step into each zone, G L is the lighting density taken from Table 2.
Occupancy levels during holiday periods was set to zero. No active cooling system was simulated in this study. All simulations took place from April to October and the simulation time-step was one hour. This time of the year was taken as it represented the time of the year where cooling or intermittent cooling would be required. In total 40 simulations were run, 20 for each location, and over 205,000 h were simulated.

Control strategies
Ten control strategies were investigated. Five used adaptive control set-points, and five used non-adaptive set-points. The control strategies were broken into five main types, with varying levels of complexity. The temperature set-points for these control strategies were varied based on air temperature which is typical for building control strategies or the operative air temperature which is typically used in adaptive standards. All strategies were defined using interpretations of previous research in cognate areas [8,13,16,49]. Equation (4) describes the notation used for each control strategy.

A DNSR ð4Þ
where, A denotes an adaptive set-point, D denotes a day-time ventilation strategy is used, N denotes a night-time ventilation strategy is used, S denotes an external solar shading being used, and R denotes an external RH limit is used in the control strategy. For example, the A_DNS control strategy is an adaptive control strategy (A), which uses day-time ventilation (D) night-time ventilation (N) and solar shading (S). Table 3 provides and overview of each control strategy from heating, ventilation and solar shading perspectives. This is then followed by a detailed description of each control strategy (Section 2.2.1 to 2.2.10).

Day-time ventilation (D)
This control strategy was designed to represent the existing ventilation control system in the zero2020 building or that of a typical BMS control system. In this strategy, dead-band control was used in conjunction with limits on the external air temperature for temperatures below 10°C and greater than 23°C, which override the dead-band controller. To further ensure that ventilation was only utilised when it was required, a further constraint was imposed where ventilation was only available/active if the external air temperature was less than the inside air temperature. Finally, this strategy only operated during occupied hours between 08:00 and 18:00. A heating system set-point of 21°C was used T set,dv = 22°C, c occu (t) = 1, 08:00 to 18:00 22:00 to 07:00 where, c Te,ll (t) is the condition that the outside air is greater than a low limit temperature of 10°C, c Te,hl (t) is the condition that the external air is less than a high limit temperature of 23°C, c Ti,Te (t) is the condition that the inside air temperature is greater than the external air temperature, c set,dv (t) is the controller output from a dead-band controller (with a static set-point (T set,dv ) of 22°C) and c occu (t) is the condition that ensures ventilation only during occupied hours. Equation (6) describes the low temperature limit condition. Where, T e is the external air temperature. Equation (7) describes the high temperature condition. Equation (8) describes the inside and outside air temperature condition.

Day-time and night-time ventilation (DN)
In this strategy, the daytime ventilation strategy (D) was used, however, night-time ventilation was implemented outside of occupied hours. Equation (9) where, c set,c,nv (t) is the output of the dead-band controller with a combined day-time and night-time ventilation strategy, c DT,ie (t) is the condition that night-ventilation is limits the operation of the ventilation system at night-time given a difference in temperature between the inside and external temperature, and c occu,nv (t) which is the condition that operates the ventilation system if it is occupied or if night ventilation is to activated (see Equation (10)). Equation (10) describes the condition used to operate the natural ventilation system either for day-time and night-time ventilation modes.
where, c nv (t) is the condition that activates night-time ventilation given certain conditions and during a specific times (see Equation (13)). Equation (11) describes the condition used to limit the system from operating if a difference between the inside and outside air is not greater than 2°C (c DT,ie (t) ).
Equation (12) describes the condition used to select the setpoint for combined night-time and day-time operation of the ventilation system (T set,dv,nv (t) ) T set;dv;nvðtÞ ¼ T set;dv þ ðc nv t ð Þ À 3Þ ð 12Þ where, T set,dv is the day-time ventilation set-point and c nv (t) is the condition that activates night-time ventilation. The natural ventilation set-point (T set,dv ) was 22°C during typical day-time operation. However, if night-ventilation was activated this set-point was reduced to 19°C for hours when the building was considered unoccupied. Under this condition, the cooling system reduced the indoor air temperature to 18°C due to the dead-band of ±1°C. 18°C was selected as it was the minimum suggested by the World Health Organisation [50]. Equation (13) describes the condition which leads to the activation of night-ventilation (c nv (t) ).
where, c Temax (t-24) is the condition that the maximum external temperature from the previous 24-hour period is greater than 20°C, c Timax (t-24) is the condition that the maximum internal temperature from the previous 24-hour period is greater than 23°C and c nvt (t) is the condition that the current time-step is between 22:00 and 07:00. Equation (14) describes the condition used to determine if the maximum external air temperature is greater than the threshold limit of 20°C from the day previous.
where, T emax(t-24) is the maximum external air temperature for the previous day. Equation (15) describes the condition used to determine if the maximum internal air temperature (of the open plan office only) was greater than the threshold limit of 23°C from the day previous.
where, c Temax (t-24) and c Timax (t-24) are the same conditions as described above, and c nvt2 (t) is the inverse of the c nvt (t) schedule.
This inverse schedule was used to change the day-time set-point (after night-ventilation) to a lower temperature during day-time operation.

Day-time ventilation with dynamic solar shading (DS)
This control strategy combined day-time ventilation and dynamic solar shading. The day-time ventilation strategy (D) was used, and the external solar shading system was activated at any point during the day if the external irradiance on a glazed façade of a room was greater than 150 W/m 2 , and maximum room temperature was greater than 23°C. To alter the external shading factor for each glazed façade of each zone, Equation (18) was used, which combines both temperature and solar conditions.
where, E w is the overall external shading factor of each window expressed as a percentage of non-transparent area, E d is the external shading factor of the factory windows including dirt, c T is the condition that the internal air temperature inside (T i ) the open plan office is greater than 23°C, c S is the output from the condition that the outside irradiance on a façade is greater than 150 W/m 2 (G e ), and E s-d is the increase in solar shading factor associated with the external shading device. The shading ratio shown in Equation (18) changed the non-transparent area of each window from 61% to 92% if both conditions are met (see from Section 2.1.2). The maximum value that can be used for this external shading factor is 1.
The control of the dynamic solar shading system indicated here operated independently of the natural ventilation system, however both systems acted on reducing the temperature in each room.

Day-time ventilation with night-time ventilation and dynamic solar shading
The DNS control strategy used the control logic of the natural night ventilation system indicated in the DN control strategy and combined this with the solar shading control strategy indicated in the DS control strategy. Ventilation and solar shading systems were controlled independently of each other.

Day-time ventilation, night-time ventilation, solar shading and external humidity limits (DNSR)
In this control strategy, the same daytime and night-time strategies that were shown in the DNS strategy were used, except the natural ventilation system closed openings when the external RH was greater than 70%. Equation (19) describes the controller output for the DNSR control strategy (CO DNSR (t) ).
where, c He (t) is the condition that the outside relative humidity is greater than 70%. This condition is described in Equation (20). Where, H e is the external relative humidity.

Adaptive day-time ventilation (A_D)
This strategy is similar to the control strategy D, however the set-point in this incidence was adaptive and was based on the comfort temperature, which depended on the externally weighted exponential mean daily temperature (T rm ). For this strategy the internal operative temperature (T o ) was used instead of the internal air temperature (T i ), as the comfort temperature is based on the operative temperature. The strategy controlled within a dead-band around the adaptive set-point that was constantly changing. This strategy attempted to stay within ± 3°C of the comfortable operative temperature. The comfort temperature (T c ) was defined by Equation (21).
where, T rm is the exponential external mean using the 7-day calculation method. In this study, a dead-band of ± 1°C was selected to reduce discomfort if an overshoot in operative temperature occurred. The external temperature limits that were used in all non-adaptive control strategies were also used in this control strategy as is shown in Section 2.2.1.

Day-time and night time ventilation (with adaptive control setpoints) (A_DN)
The principle of this control strategy is similar to that of control scenario DN. The adaptive day-time control system A_D was used and a night-time ventilation strategy DN was also used. However, the internal maximum limit used to trigger night-time ventilation was based on whether the internal operative temperature (T o ) exceeded the comfort temperature (T c ) by 2°C. Equation (22) describes this condition.
where, T omax(t-24) is the maximum operative temperature from 24 h before the current time-step. The set-point for cooling during 22:00 and 07:00 was also based on the operative temperature. Equation (23) below describes how the set-point for adaptive night ventilation (T o,set,v (t) ) was selected at each time-step.
For this control strategy each zone had a night-time cooling setpoint that is 3°C below the comfort temperature. This was selected as the lower limit in EN 15251 is 3°C below the comfort temperature. The dead-band that was operated for night-ventilation was ± 1°C. The heating system operated on the same nonadaptive set-point shown in Section 2.2.2. The low temperature limits of the external air that were used in DN were also used in this control strategy.

Day-time ventilation with solar shading (A_DS)
This control strategy combined the adaptive limits of A_D with the same solar shading system that was described in DS. This solar shading system was not controlled based on the operative temperature.

Day-time ventilation with night-time ventilation and dynamic solar shading (adaptive) (A_DNS)
This control strategy combined the adaptive limits of A_DN with the same solar shading system that was described in A_DS.

Day-time ventilation, night-time ventilation, solar shading and external humidity limits (A_DNSR)
This control strategy combined A_DNS with the humidity limits that are discussed in scenario DNSR.

Weather data
For this study, current and future TMY datasets for both locations were generated using Meteonorm 7.3 [51]. Dublin and Budapest were selected as interpolated cities. Current TMY3 weather files were generated for a radiation period of between 1991 and 2010 and a temperature period of between 2000 and 2009. Extreme weather files were created by using the ten-year extreme temperature model and by selecting the worst case IPCC scenario (A2). Initially, climates were analysed for each location using all of the three weather file types: TMY (Now), ten-year, and tenyear 2050 (2050). Simulations were conducted using the TMY (Now) and ten-year (2050) weather files. Table 4 shows the limits of the key relevant standards used in assessing comfort in buildings. In this study, we attempted to maintain the simulated internal conditions within the limits of EN16798-1. In EN15251 and EN16798-1, the optimal comfort temperature T c is determined by Equation (24) [13]. Using this equation, categories of comfort are typically determined with varying category range limits for those with a normal level of expectation (±3, Category II) and special cases or those with a high level of expectation (±2, Category I). However, in the recent revision of EN15251 (EN16798-1) the low limits were changed which led to an asymmetric range (i.e. +3-4).

Comfort metrics used
In this study, all overheating and overcooling criteria were calculated based on the simulated operative temperature (T o ) exceeding the values shown in Table 5. A lack of overcooling criteria exists in published standards and guidelines, therefore, we adopted the percentage of hours where the simulated operative temperature falls below the limits of Category II in EN 15251 and EN16798-1 as overcooling criteria (OC1, OC2). Assessments of overheating and overcooling were only performed in the Open Plan Office zone.
The most common approaches to estimate overheating risk (that do not use a static value) are the Chartered Institute of Building Services Engineering (CIBSE) document TM52 and the methodologies shown in the adaptive standard of EN 15251 [52,[55][56][57][58][59][60]. These include a weighted exceedance criterion, a weighted exceedance for one-day severity and the final criterion is used to limit severe overheating incidences, where, the difference between T max and T o should not exceed 4°K. Overheating in TM52 is defined by this three criteria, where ''a room or building that fails any two of the three criteria is classed as overheating" [55].

Mechanical cooling potential
No mechanical cooling system was simulated in this study. To estimate the mechanical cooling load that would be required to maintain comfortable conditions (for hours of overheating), internal operative temperature data were analysed post simulations. In this analysis, the number of occupied hours where inside operative air temperature exceeded the overheating criterion OH1 were determined for each control scenario, location and weather file type. In this study, the difference in temperature between the indoor air temperature and the lower dead-band limit of 21°C was calculated when the operative temperature in the open plan office was greater than the upper limit of OH1. The sum of the differences between the lower dead-band limit and the internal air temperature where calculated each hour. Equation (25) describes the degree hour criterion used. The sum of the differences for each hour can be described by the degree hour criterion (C dh ) shown in Equation (25).
where T i (t) is the air temperature in the open plan office and T set,c is the set-point temperature for active cooling.

Analysis of weather files and climates
To assess the potential restrictions of climates for passive cooling control in general, each climate file was analysed with reference to low (T e < 10°C) and high external ambient temperatures (T e > 25°C) as well as low (H e < 30%) and high humidity (H e > 70%). The broad categorisation using static limits was undertaken for an entire year to give an insight into annual and seasonal variation in the general applicability of the climates to typical control parameters. The climate assessment would also allow for an insight into the expected changes in climate. Budapest in Hungary, and Dublin in Ireland, were the two cities, which represented the span of temperature and humidity conditions available in the initial ten cities and covered both KG climate classifications (see Appendix for more information). Budapest has a typical continental climate with cold winters and warm summers with few high humidity incidences. Dublin is a consistently temperate maritime climate with a larger amount of high humidity incidences and with consistent mild temperature conditions. From this section onwards, all results will refer to climate conditions in Budapest (HU) and Dublin (IE).

Analysis of climates based on static external limits for passive control systems
Tables 6 and 7 present summary statistics of external temperature and RH for TMY (Now), ten-year extreme (10-year) and extreme 2050 (2050) weather files in Budapest and Dublin. The change in annual mean temperature between current and extreme future conditions is expected to be 0.8°C in Dublin and 2.0°C in Budapest, with a change in maximum external temperatures of 5.2°C and 6.7°C, respectively. Negligible changes in RH were found between current and extreme future conditions (1-2%). However, Dublin was found to have high humidity levels for the majority of the year. Figs. 4 and 5 present a more detailed view of the climatic limitations when static limits were used for external air temperature and RH. Fig. 4 presents categorisations for external temperature limits. A negligible increase in external temperatures above 23°C was found during occupied hours for Dublin (IE) (0-1%). For Budapest (HU), categorisations suggest that large portions of the summer months will not be suitable for passive cooling. Day-long external temperatures greater than 23°C are reported for current and future conditions which is expected to limit the use of nightcooling. An increase in the percentage of occupied hours where external temperatures are greater than 23°C is expected between current and future conditions (from 24% to 32%). A doubling in the  percentage of unoccupied hours above 23°C is also expected (from 7 to 13%). Fig. 5 presents categorisations for RH. In Dublin, between 57% and 66% of the occupied hours were calculated to be above 70% RH. Based on categorisations the night ventilation potential in Dublin may be severely limited by high humidity. In Dublin, between 92% and 95% of the unoccupied hours were calculated as being above 70% RH. For Budapest, humidity does not appear to have as much of an effect on the use of passive cooling systems. Between 19% and 22% of the occupied hours were calculated as being greater than 70% RH. Between 55% and 59% of the unoccupied hours were calculated as being above 70% RH. Both categorisations for temperature and RH suggest that different challenges will exist in the use of passive cooling systems for Dublin and Budapest. For Dublin, high humidity incidences are likely to limit the use of passive cooling systems like night-ventilation or daytime ventilation. For Budapest, changes in external temperatures during the day and particularly at night-time, are likely to lead to unfavourable conditions for passive cooling systems.

Simulated passive potential
3.2.1. Percentage outside the range Fig. 6 indicates the performance of each simulated passive control strategy. Control strategies that maintained operative temperatures at acceptable levels for greater than 95% of the occupied  hours were deemed satisfactory. In Dublin, all passive control strategies were found be capable of maintaining comfortable operative temperatures for the current conditions. Eight out of the ten passive control strategies were found to be capable of maintaining comfortable operative temperatures in 2050. The strategy that performed the best in Dublin was the adaptive day-time ventilation strategy with dynamic solar shading (A_DS), followed closely by the adaptive day-time ventilation strategy (A_D). However, there were marginal differences between most strategies when used in Dublin. In Budapest, three out of the ten passive control strategies were found to be capable of maintaining comfortable conditions for current conditions (A_DNS, A_DNSR, DNSR). One of the passive control strategies investigated was able to maintain satisfactory internal conditions in 2050 (A_DNS). Overall, passive control strategies were capable of satisfying thermal comfort at the normal level of expectation between 57% and 95% of the occupied hours. Three control strategies that satisfied comfort for greater than 90% of the occupied time in 2050 had night-ventilation and dynamic solar shading as part of their control strategy (DNSR, DNS, A_DNS). In 2050, these passive control strategies increased the percentage of occupied hours where comfortable conditions were maintained by between 33% and 37% when compared to a day-time ventilation strategy (D).

Overheating and overcooling
A negligible amount of overheating was calculated in Dublin irrespective of the metric used for both current and extreme future weather conditions. As a result, details on overheating performance in Dublin have been excluded in this section. Fig. 7 presents the percentage of occupied hours in Budapest where the internal operative temperature exceeds the upper limit of Category II of EN16798-1 (OH1). Strategies that used natural night ventilation or dynamic solar shading were not capable of reducing overheating risk to acceptable levels in current conditions. Control strategies that used a combination of night-ventilation, solar shading, and an external RH limit were capable of maintaining overheating to below 3% of the occupied hours. DS or A_DS strategies reduced overheating hours by between 17% and 19% when compared to a day-time ventilation strategy. DN or A_DN strategies reduced the percentage of occupied overheating hours by between 21% and 22% when compared to day-time ventilation. The reduction in the percentage of occupied overheating hours through the use of a night ventilation and solar shading (DNS, A_DNS, DNSR, A_DNSR) was found to be considerable (between 25% and 28%) when compared to both adaptive and non-adaptive day-time ventilation strategies. None of the passive control strategies were found to be capable of reducing overheating to acceptable levels in an extreme 2050. However, the combined effect of solar shading and night-ventilation reduced the amount of overheating hours by between 29% and 38% when compared to a typical day-time ventilation strategy (D, A_D). When an external humidity limit was included in the control strategy, the amount of overheating hours increased (when compared to the DNS and A_DNS strategies) in current conditions and in 2050. However, the strategies that had a humidity limit reduced the number of overheating hours by 25% to 33% when compared to the overheating hours of day-time ventilation strategies. Overall, the number of overheating hours according to OH1 increased by between 4% and 14% between now and an extreme 2050. The percentage of overheating hours was dependent on the passive control strategy. Fig. 8 indicates the performance of each of the control strategy with respect to the criteria in TM52 for Budapest. Two passive control strategies complied with two of the three criteria in the TM52 guidelines on overheating in current conditions, these were the A_DNS and DNS strategies. The use of an RH limit resulted in an increase in the severity of overheating in any one day (Criterion 2), which resulted in the failure of A_DNSR and DNSR strategies. None of the control strategies were compliant with two out of three of the criteria of TM52 for Budapest in the extreme conditions for an extreme 2050. Fig. 9 shows the overcooling risk based on the acceptable lower limits for EN 15251 and its current revision EN16798-1. From the results shown it is clear that adaptive control strategies can eliminate the overcooling risk, irrespective of the location. If nonadaptive control strategies are used, the greatest overcooling risk is for strategies that have night-ventilation (DNS, DN). For the simulations conducted the difference in the percentage of overcooling observed between both of these standards was between 1.5% and 13.0%, with EN16798-1 always predicting less overcooling than EN15251. Currently, the use of the new limit proposed in EN16798-1 can result in all of the passive control strategies having an overcooling risk of<5% of the occupied hours. This also results in nine out of ten of the control strategies having an overcooling risk for<3% of the occupied hours. Fig. 10 indicates the number of degree hours that would be required to maintain comfortable conditions if the Open Plan Office overheated and went outside of category II of EN16798-1 for Budapest. For Dublin, the calculated number of cooling degree hours  based on simulations was negligible, irrespective of the weather dataset used.

Mechanical cooling potential
Overall, there was a large difference in the cumulative cooling degree hours for each control strategy, and for each location. In Budapest the difference in cooling degree hours between best (DNS) and worst strategies (A_D) was 7539°Ch now and 11224°Ch in 2050. The use of multiple passive cooling systems had the capability to keep the number of degree hours to between 19°Ch and 67°Ch for sensible mechanical cooling in current conditions. In the extreme future, the use of passive cooling strategies with multiple passive systems (DNS, A_DNS) was found to reduce the need for mechanical cooling by 89-91% when compared to a natural day-time ventilation strategy (D). The inclusion of an external RH limit resulted in an increase in degree hours of mechanical cooling. However, both DNSR and A_DNSR control strategies reduced the future need for mechanical cooling by 72-82% when compared to a natural ventilation strategy (D).

Maritime climate (Dublin, IE)
The overall change in external air temperature in individual locations indicated in Section 3.1.1 shows varying levels of increase between Dublin and Budapest. In Dublin, it is indicated that  external temperatures will increase by 0.8°C on average. Simulations indicate a negligible effect on categorical comfort levels with respect to EN16798-1. Overall, an average increase in internal operative temperatures of between 0.1°C and 0.3°C between now and 2050 was indicated, which depended on the strategy that was used (see Fig. 13 in Appendix A.2). This minimal change in internal operative temperature coupled with high levels of categorical comfort would suggest that passive cooling strategies are resilient in the climate presented in Dublin for current conditions and in the medium-term (2050). This suggestion is also supported by the work in Annex 62 which suggested that the Irish climate has a significant potential to cool passively [34]. While limited examples of overheating studies simulated or measured exist for nonresidential buildings in the Irish climate [32], there are examples of measured and simulated comfort performance for low energy dwellings [61,62]. There is also some evidence that indicates that low energy buildings are overheating in current conditions, even in similar northerly climates in the UK [63][64][65]. Most of the UK studies indicate disparities between the north and south of the UK. Simulations would suggest that at more southerly latitudes (i.e. London) overheating is likely to be a major issue in the future, but in more northerly latitudes (i.e. Edinburgh) this may not be as much of an issue [65]. However, there are some examples of overheating in new dwellings in more northerly latitudes of the UK in current conditions [63]. Simulations of Irish dwellings suggests that current practices in design do not lead to high levels of overheating, however, particular design parameters such solar gains (i.e. orientation, shading etc.) and ventilation rates can have a significant impact on results [62]. On the other hand, measurements in passive houses in Ireland suggest that summer overheating is an issue during design and in operation, and that additional passive measures such as passive stack ventilation and solar shading should be considered to avoid summer overheating [61]. The work presented here indicates high levels of comfort performance by using passive cooling strategies if performance is assessed using the adaptive standard for non-residential buildings. The use of a static overheating metric or an adaptive metric adopted in dwellings [62] is likely to increase the hours where overheating is presented [66] for both locations and may be unrepresentative of actual comfort levels in office environments.

Continental climate (Budapest, HU)
The mean change in external temperature in Budapest was projected to be 2.0°C between now and 2050 (see Section 3.1.1). Simulations indicate mean increases from April to October of between 1.0°C and 1.9°C from now to 2050, depending on the passive control strategy used. Over the same period, adaptive set-points were found to increase indoor operative temperatures by 1°C. Passive control strategies were capable of reducing mean indoor operative temperatures by 0.6°C to 2.9°C when compared to a day-time ventilation over the cooling season in current conditions. In 2050, reductions of between 1.1°C and 3.5°C were seen when compared to day-time ventilation. The addition of night cooling to the day time strategy led to reductions of between 2.2°C and 2.5°C, solar shading led to reductions of between 1.4°C and 1.8°C, adding both led to reductions of between 2.9°C and 3.5°C on average over the cooling season in current and future conditions. However, seasonal variations were observed. Fig. 11 indicates the differences between internal operative temperatures for each strategy in current and future conditions during different parts of the cooling season. In April, mean increases of internal operative temperatures of between 0.7°C and 1.1°C were calculated between conditions now and in 2050. In October, mean increases of between 1.8°C and 3.7°C between now and 2050 were calculated. In July, simulated results indicate mean increases in internal operative temperatures of between 3.3°C and 7.0°C from now and 2050 depending on the passive control strategy.
Increases in the shoulder seasons suggest that there may be an additional need for cooling to maintain comfortable conditions. The use of additional measures such as solar shading and night ventilation resulted in moderate reductions in the average internal operative temperatures in both April (0.8°C to 1.1°C) and October (1.0°C to 1.8°C). These additional passive measures reduced internal operative temperatures, however, this did not result in significantly improved comfort conditions as day-time ventilation was considered sufficient for the majority of the time during the same period (see Fig. 15 in Appendix A.2). In summer (July), there already exists a significant risk of overheating when using a daytime ventilation strategy (see Fig. 15 in Appendix A.2). Outside of an increase in temperature due to the adaptive day-time ventilation strategy (A_D) all strategies were capable of reducing mean internal operative temperature, however, adaptive strategies maintained conditions that were slightly 0.5°C warmer on average. This is reflected in Fig. 11 where small differences in trends exist between adaptive and non-adaptive strategies in July. In current conditions, passive control strategies were capable of reducing internal operative temperatures by between 2.3°C and 5.5°C on average when compared to the day-time ventilation strategy. In 2050, passive control strategies were capable of reducing internal operative temperatures by between 4.0°C and 8.1°C on average when compared to the day-time ventilation strategy. The addition of night-ventilation reduced internal operative temperatures by 4.0°C to 5.7°C on average when compared to the day-time strategy. The addition of solar shading reduced internal operative temperatures by 2.3°C to 4.2°C on average when compared to the daytime strategy. A combination of both included reduced internal operative temperatures by 4.0°C to 8.1°C on average when compared to the day-time strategy. The addition of a humidity limit led to reductions (2.8°C to 7.0°C) but not to the same magnitude as the DNS and A_DNS strategies. The magnitude of reductions in internal operative temperatures through the use of passive control strategies is significant for the climate presented in current conditions and future conditions in Budapest. However, passive control strategies are likely to not be capable of maintaining optimal comfort conditions in the summer seasons in the medium-term. While previous studies of simulated comfort performance are limited for Budapest specifically, there other examples of the performance of passive cooling systems in European climates to compare with [9,[14][15][16]. Previous work by Breesch and Janssens et al. indicated that multiple passive measures are required to maintain comfortable internal conditions in the extreme external conditions in Uccle in Belgium [16]. Work by Lomas and Ji et al. suggested that simple single sided natural ventilation would not be robust enough to maintain conditions in current conditions for London [14]. Schuzle et al. reported no need for mechanical cooling through simulations of controlled natural ventilation in Stuttgart. Pagliano et al. highlighted that satisfactory levels of comfort performance for a passive cooled building in Milan and suggested that there may be a longterm need for active cooling but not in the medium-term [15]. The work presented here indicates that individual passive systems may not be successful in maintaining comfortable conditions in Budapest both now and in 2050. Combinations of systems (i.e. solar shading and ventilation) are more comfort resilient in current conditions but are likely to need supplementary active cooling in the medium-term.

Limitations
The work presented in this paper has some limitations that are worth noting. The model used in this paper is based on a calibrated and validated model for one nearly zero energy building (nZEB) that uses a single sided flow regime. The ventilation system described in the paper is a theoretical system that has utilised some simplifying assumptions leading to one opening per façade, and maximising opening height to offer best-case opportunity for single sided natural ventilation. Literature indicates that nZEBs have been shown to be relatively homogenous in their design [21]. However, different types of nZEBs with different applications and internal gains, flow regimes and opening characteristics are likely to lead to different outcomes through simulation. As was indicated previously, the weather files used in this study may not exclusively include some significant effects (i.e. urban heat island). An attempt has been made to include the effect of extreme conditions in 2050. However, as new methods are made for developing future weather files [22] and the effects of global warming different locations are advanced the potential for passive cooling may change. Finally, the work presented here considers combinations of some passive cooling systems (natural ventilation and solar shading). There are a number of systems that were not explored in this paper, other advanced and emerging passive cooling systems and technologies that take advantage of alternative natural sinks (such as the ground or water) should also be considered as a priority before utilising mechanical cooling or air-conditioning. Additionally, the performance of hybrid ventilative cooling systems was not considered, this could expand the cooling potential of locations with minimal energy cost.

Conclusions
The aim of this study was to assess the resilience of different passive control strategies in delivering optimal comfort and energy conditions in current and future extreme conditions. From the results presented here, it is clear that the passive cooling strategies that resulted in optimal comfort and energy scenarios were those that combined multiple controlling interventions. The analysis presented indicates that multiple complimentary passive cooling systems can be more successful in addressing overheating when compared to individual passive systems. However, not all passive systems were required in all locations.
For a maritime climate, many passive cooling control strategies resulted in both a negligible need for sensible mechanical cooling, and high levels of standardised adaptive thermal comfort performance in current and future conditions. Day-time ventilation was found be sufficient for shoulder and cooling seasons, but there is the risk of overcooling using this system. The use of either an adaptive set-point (A_D) or a combination of day-time ventilation and solar shading (A_DS) resulted in a reduction in overcooling without overheating. Given the absence of large field studies assessing overheating in non-residential nZEBs in Ireland, there is need for research which evaluates the measured overheating risk in these types of buildings. Future work should also consider the development of more detailed weather files for Ireland [62]. Weather files for different locations considering the effects of urban heat island, and extreme conditions during the cooling season are required, and more efforts should be made to include simulating the resilience of designs (in extreme or future conditions) as part of demonstrating compliance [64]. Although the adaptive model is likely to be more representative of actual overheating (particularly in NV environments) when compared to static values (which are representative of MV environments), more consideration should be given to the development of new overheating metrics that also consider the effects of humidity in maritime climates. For the continental climate presented, the best control strategies that satisfied both comfort and energy criteria were those that used solar shading and night-time ventilation (DNS, A_DNS). These types of strategies were able to maintain satisfactory comfort levels for over 90% of the occupied hours in current and future weather scenarios. There is a more urgent need to consider the combined effects of natural night ventilation and solar shading as an integral part of building design in continental locations. All passive designs should be tested for resilience with extreme weather files. Passive control strategies with multiple measures have a significant potential to reduce mean internal operative temperatures and reduce the need for active systems. Based on this, there is a regulatory need to consider passive cooling strategies in the design of buildings. The lack of consideration for window sizing, opening sizing and solar protection (particularly in continental climates) could lead to significant increases in energy consumption. External relative humidity limits were found to reduce the use of passive cooling systems and increase overheating incidences. Future work should consider how to expand the cooling potential and passive cool buildings in conditions with high external humidity levels.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.