Climate resilience of buildings: overheating in buildings: development of framework for overheating risk analysis

Publisher’s version / Version de l'éditeur:

Vous avez des questions? Nous pouvons vous aider. Pour communiquer directement avec un auteur, consultez la première page de la revue dans laquelle son article a été publié afin de trouver ses coordonnées. Si vous n’arrivez Questions? Contact the NRC Publications Archive team at

PublicationsArchive-ArchivesPublications@nrc-cnrc.gc.ca. If you wish to email the authors directly, please see the first page of the publication for their contact information.

https://publications-cnrc.canada.ca/fra/droits

L’accès à ce site Web et l’utilisation de son contenu sont assujettis aux conditions présentées dans le site LISEZ CES CONDITIONS ATTENTIVEMENT AVANT D’UTILISER CE SITE WEB.

Client Report (National Research Council of Canada. Construction); no.

CRBCPI-Y3-R8, 2019-06-10

READ THESE TERMS AND CONDITIONS CAREFULLY BEFORE USING THIS WEBSITE. https://nrc-publications.canada.ca/eng/copyright

NRC Publications Archive Record / Notice des Archives des publications du CNRC :

https://nrc-publications.canada.ca/eng/view/object/?id=f04717ba-4e71-4af9-bc33-cce3d2665160 https://publications-cnrc.canada.ca/fra/voir/objet/?id=f04717ba-4e71-4af9-bc33-cce3d2665160

NRC Publications Archive

Archives des publications du CNRC

For the publisher’s version, please access the DOI link below./ Pour consulter la version de l’éditeur, utilisez le lien DOI ci-dessous.

https://doi.org/10.4224/40001235

Access and use of this website and the material on it are subject to the Terms and Conditions set forth at

Climate resilience of buildings: overheating in buildings: development

of framework for overheating risk analysis

REPORT CRBCPI-Y2-R8

Climate Resilience of Buildings:

Overheating in Buildings

―

Development of

Framework for Overheating Risk Analysis

A. Laouadi, M.A. Lacasse, A. Gaur, M. Bartko, and M. Armstrong

Report: CRBCPI-Y3-R8

10-June-2019

Climate Resilience of Buildings:

Overheating in Buildings — Development of

Framework for Overheating Risk Analysis

Author

A. Laouadi, Sub-task manager

Authors:

Approved

Ms. Marianne Armstrong

Project Manager, CRBCPI Project

Approved

Dr. Phillipe Rizcallah, P. Eng.

Program Leader, Building Regulations for Market Access

Client Report: CRBCPI-Y3-R8

Report Date:

10 June 2019

Contract No:

A1-012020-05

Reference:

29 November 2016

Program:

Building Regulations for Market Access

44 pages Copy no. 1 of 5

This report may not be reproduced in whole or in part without the written consent of the National Research Council Canada and the Client

T

C

Table of Contents ... i 

List of Figures ... ii 

List of Tables ... ii 

Executive Summary ... iii 

About the overheating project ... 1 

Introduction ... 2 

Objectives... 6 

Approach ... 6 

Reference climate data period ... 6 

Metric for heat events ... 7 

Modifications to the SET model ... 12 

Definition and characteristics of heat events ... 13 

Definition of a heat event ... 13 

Definition of heat waves ... 16 

Characteristics of heat waves ... 16 

Generation of RSWY ... 16 

Overheating definition and metric ... 24 

Selection of thresholds for SET ... 26 

Overheating Criteria ... 27  CIBSE’s Method ... 27  DfES’s Method ... 28  PHI’s method ... 29  USGBC’s method ... 29  Proposed method ... 29  Conclusion ... 32  Acknowledgement ... 34  References ... 35 

L

F

Figure 1. Comparison of SET with UTCI and WBGT indices ... 11  Figure 2. Validation of the transient SET model for the skin and rectal temperature predictions ... 13  Figure 3. Temperature and SET during for the extreme heat wave periods obtained using the proposed method for Ottawa, Ontario. ... 19  Figure 4. Temperature and SET during for the extreme heat wave periods obtained using the CIBSE’s method for Ottawa, Ontario. ... 20  Figure 5. Temperature and SET during for the extreme heat wave periods obtained using the proposed method for Montreal, Quebec. ... 21  Figure 6. Temperature and SET during for the extreme heat wave periods obtained using the CIBSE’s method for Montreal, Quebec. ... 21  Figure 7. Temperature and SET during for the extreme heat wave periods obtained using the proposed method for Toronto, Ontario. ... 22  Figure 8. Temperature and SET during for the extreme heat wave periods obtained using the CIBSE’s method for Toronto, Ontario. ... 22  Figure 9. Severity of overheating versus body water loss for healthy subjects ( is the standard deviation) ... 31  Figure 10. Severity of overheating versus body water loss for vulnerable subjects ... 32 

L

T

Table 1. Comparison of the selected indices against the screening criteria ... 8  Table 2. Thermal sensation scale of SET (Parson, 2003) ... 11  Table 3. Reference values of SET for nighttime (sleep) for selected Canadian cities... 15  Table 4. Comparison of extreme years as obtained using the proposed, CIBSE’s and EC’s methods for Ottawa, Ontario. The selected years are highlighted in the bold red color. ... 19  Table 5. Extreme years for selected Canadian cities ... 24  Table 6. Threshold values of SET for un-acclimatized or acclimatized persons per building type ... 27 

E

S

Under the Pan-Canadian Climate Change Framework, Infrastructure Canada has provided funding to the National Research Council (NRC) to deliver the Climate-Resilient Buildings and Core Public Infrastructure (CRBCPI) Project over a five year period. The purpose of the project is to develop decision support tools, including codes, guides and models for the design of new climate resilient buildings (CRB) and core public infrastructure (CPI) and rehabilitation of existing buildings and CPI in key sectors to ensure that the impact of existing and future climate change and extreme weather events are addressed.

This report relates to research sub-project A1-012820-06: Overheating in Buildings under the CRBCPI Project. The goal of sub-project A1-012820-06 is to develop guidelines to address the overheating risk in retrofitted and new buildings as may arise from climate change and extreme heat events. The guidelines are also intended to provide sound evidence for code change requests which will lead to future updates of the National Building Code of Canada on the climate resilient design of buildings.

This report is a deliverable of sub project A1-012820-06. The report addresses the development of a framework to analyse the risk of overheating in buildings from the perspective of comfort and health of occupants through the use of whole building energy simulation. The framework includes five steps: 1. generation of historical climate data to extract various types of extreme heat events at given

locations;

2. development of a heat stress metric to evaluate the effect of heat on the comfort and health of human subjects;

3. definition and characteristics of heat events;

4. generation of reference summer weather years (RSWY) for select Canadian cities; and 5. development of metrics and criteria for overheating events.

Each section of the framework is further described below.

The reference historical climate data were taken from the most recent climate observation databases (1986-2016) available from Environment and Climate Change Canada (ECCC, 2018). The hourly data of each weather variable was collected from all weather stations within a city domain and then averaged to produce a single weather file for each selected city. Any missing data in the climate observations were filled using bias-corrected data from the climate reanalysis database of CFSR (Saha et al., 2010). Selection of an appropriate physiological heat stress metric was determined using unique screening criteria developed in this work. A short list of widely used indices was established, including the Humidex (H), wet-bulb globe temperature index (WBGT), universal thermal climate index (UTCI), the predicted heat strain index (PHS), and the standard effective temperature index (SET). The SET satisfied all the screening criteria and was therefore selected. The SET model was also modified to account for the transient nature of heat events and daily activity levels of subjects, and the physiological response of sleeping persons. The SET index correlated very well with the aforementioned indices under outdoor

exposure with steady state conditions and compared well with third-party experimental results (Lundgern-Kownacki et al., 2017). for body temperatures of human subjects undergoing multi-stage activity levels under constant hot conditions.

The SET index was used to analyze the climate data to define heat events that triggered a specific physiological response in human subjects for daytime and nighttime exposure. From the analysis, three features of heat events were then identified, including: duration, intensity and severity. Severity is obtained as the deviation of the hourly SET values from predefined threshold values, multiplied by the exposure time during the daytime and the preceding nighttime. Duration accounts for the number of days of sustained heat events, and intensity is obtained as the ratio of severity to duration. The daytime threshold value of SET was selected to correspond to the initiation of the sweating mechanism of human subjects (equivalent to a slightly warm sensation) and the nighttime threshold value was selected to correspond to the lower limit of the adaptive thermal comfort temperature. Based on these features, three types of heat events were identified: long, intense and severe.

The extreme weather years were extracted from 31 years of the most recent historical climate data (at the time when the study was being performed) at a given location. Heat events were identified for each year and sorted by their maximum duration, intensity and severity. The return period of extreme heat events was fixed to 15.5 years (the second extreme year out of 31 years). The extreme value frequency distribution functions were used to fit the obtained data, and the extreme years that have return periods close to 15.5 years were selected as RSWY. Three types of RSWY were then generated for 11 Canadian cities. Under some circumstances, one representative RSWY may be sufficient to conduct overheating analysis to minimise parametric computational time.

The proposed method for the generation of RSWY was also compared to the CIBSE’s and EC’s definition of heat events. The CIBSE’s method is based on the weighted cooling degree hour (WCDH) with a single threshold value of temperature to extract extreme years. The CIBSE’s method does not account for other climate variables (e.g., solar radiation, relative humidity, and wind) nor the effect of nighttime exposure for sleeping persons. The EC’s method uses a combination of the daily maximum and minimum

temperatures and Humidex. Comparing the proposed method with the CIBSE’s method for three

Canadian cities (Montreal, Ottawa, and Toronto) revealed that the proposed method yields extreme years that have higher SET values during daytime and nighttime than that using the CIBSE’s method. However, in terms of temperature, the CIBSE’s method may in general yield extreme years with higher daytime temperatures, particularly in humid areas. The CIBSE’s method may therefore under-estimate the health effects of the outdoor environment. For dry or mixed dry and humid areas, the proposed method yields daytime high temperatures that are comparable to the CIBSE’s method. The EC’s method is limited in that it does not yield intense or severe years. The proposed method is therefore suitable for any climate and requires building simulation tools to have the capabilities to model both heat and moisture transfer in buildings since the indoor conditions (temperature and humidity) are dependent on the outdoor conditions

particularly in naturally ventilated buildings. The CIBSE’s method, on the other hand, might not be suitable for mixed or humid climates as the humidity may alter the physiological response of building occupants.

The proposed method for the assessment of overheating risk used the severity index of heat events (SETH), and body water loss criteria to limit the dehydration of building occupants exposed to overheating events. The severity index is expressed as the product of the intensity of heat event and exposure time. Threshold values of SET beyond which overheating events are identified were developed for healthy and vulnerable subjects with or without acclimatization. For healthy subjects, the upper limit of the resilient range of the static thermal comfort where subjects can adapt to restore thermal comfort is used. This corresponds to the range of the slightly warm sensation or slight sweating (skin wettedness = 30%) with a corresponding SET = 30°C for un-acclimatized subjects and 31.2°C for acclimatized subjects. For the vulnerable subjects, the SET value that corresponds to the borderline between comfort and discomfort is used since the vulnerable subjects have limited capacity to adapt to slight discomfort. The corresponding value of SET is 26°C for un-acclimatized subjects and 27.2°C for acclimatized subjects. For subjects in sleeping mode, the threshold SET value was fixed to 26°C (or 27.2°C for acclimatized subjects) to avoid sleep disturbance. The overheating events are identified for the three types of RSWY and the one with highest severity index (SETH) during the summer period is taken to apply the criteria of overheating. To declare that the space is overheated, the overheating index (SETH) should exceed a fixed threshold value. This threshold value is calculated based on the total water loss by sweating (dehydration) of subjects exposed to the overheating events. The allowable maximum dehydration rates for healthy and vulnerable subjects were fixed to 3% and 2% of body weight, respectively. The rehydration rates during the exposure to heat events were fixed to 60% and 40% for the healthy and vulnerable subjects,

respectively. The threshold values of SETH to declare overheating in Canadian residential buildings were calculated and found to be 153 ±18 (°C*h) and 56 ± 27 (°C*h) for healthy and vulnerable subjects, respectively.

Climate Resilience of Buildings:

Overheating in Buildings — Development of Framework for

Overheating Risk Analysis

Authored by:

A. Laouadi, PhD; M. A. Lacasse, PhD.; A. Gaur, PhD; M. Bartko, PhD, and M. Armstrong

A

The goal of the project (A1-012820-06) is to develop guidelines to address the overheating risk in retrofitted and new buildings as may arise from extreme heating events considering the effects of climate change. The guidelines are intended to provide sound evidence for code change requests which will lead to future updates of the National Building Code of Canada for the climate resilience design of buildings. The specific objectives of the project are:

1. To review literature on the effects of climate change and extreme heat events on the overheating risk in buildings under a Canadian context;

2. To evaluate building responses to climate change;

3. To develop and evaluate the effectiveness of selected resilient (adaptation) measures for new and retrofit buildings;

4. To develop guidelines and tools for the overheating risk in buildings;

5. To update the Canadian National Building code to address the climate resilience design of buildings.

The project is composed of five main tasks as follows: Task 1 – Literature review.

Task 2 – Generation of reference climate data for the historical and future projected climates. Task 3 – Building responses to climate change and extreme heat events.

Task 4 – Development of guidelines and tools for overheating risk in buildings. Task 5 – Proposed changes to the National Building Code of Canada (NBC).

I

Overheating in buildings arising from climate change and extreme heat events (EHEs) is a growing health concern in many countries (CIBSE, 2005; CIBSE, 2013; HC, 2011; Saman et al., 2013; Quinn et al., 2014; Eames, 2016). Overheating is the condition of the indoor environment that results in thermal discomfort or heat-related health stress to building occupants. Overheating is found in naturally ventilated buildings, buildings with limited capacity or intermittent use of air conditioning, and buildings that

experience extended periods of power outages or HVAC failure. Continuous exposure to high heat over several days strains the human physiological system and may lead to health issues (such as heat cramps, exhaustion, and stroke) or death, particularly for vulnerable people such as the elderly, the sick and children. Indeed, epidemiological studies have found that the excess mortality data during EHEs are higher than any other natural hazard such as floods and storms (Kenny et al., 2018; ICLR, 2018; Berko et al., 2014; Saman et al., 2013). To reduce the risk of overheating, buildings should be designed and operated to be resilient under such extreme climatic conditions. Current design and retrofit of buildings use reference (average or typical) weather years to conduct building energy simulations. These reference years are, however, not suitable to assess the risk of overheating and therefore new reference summer weather years (RSWY) need to be developed.

The reference summer weather years are used in building simulations to:

1. Assess the overheating risk under extreme weather conditions in terms of the effect of indoor

conditions on the thermal comfort and heat stress of occupants (particularly the vulnerable ones) in all building types such as passive buildings, high performance buildings, and retrofitted buildings;

2. Assess building response and resilience during extended power outages and HVAC failures, which usually occur during extreme heat events (IOM, 2011);

3. Determine the need for mechanical ventilation and cooling of buildings prior to their construction or retrofit;

4. Assess cost-effective measures for new or retrofit buildings to reduce the risk of overheating in buildings;

5. Establish design trade-offs between energy efficiency and thermal resilience of buildings;

6. Assess the compliance of new buildings to the thermal resilience requirement of applicable building codes and energy efficiency standards.

The reference summer weather years (RSWY) should be selected from a time-series of climate data of sufficient length to permit capturing a statistically significant number of occurrences of various types of EHEs of the prevailing local climate. The World Meteorological Organisation (WMO) and climate scientific communities consider 30 years as the minimum period to capture natural climate variabilities (Charron, 2014). Different sets of RSWY should thus be developed for the historical and projected future climates. Historical weather data based on observations or bias-corrected predictions include less uncertainty than

the projected future climate, and should therefore be used to assess the short term (within the 30 year period) response of buildings.

Developing RSWY also requires the use of proper metrics to rank and select representative years having periods of EHEs. Furthermore, the metrics should take into account not only the climate variables, but also the human subjects that are exposed to such EHEs to permit the evaluation of their thermal comfort and heat-related health stress. Extreme heat events (EHEs) are characterised by their intensity, duration, and frequency of occurrence (or return period), and their health outcomes are expected to differ as well (i.e., long and intense EHEs have higher risk to mortality than shorter EHEs; WHO, 2009). For practical applications, there is no worldwide accepted definition of EHEs (Gachon et al., 2016; EPA, 2016; WHO, 2009). The challenge resides in the selection criteria of the threshold values for the intensity of the event. The EHE duration is arbitrarily fixed to one, two or more days. The event frequency is statistically

computed from the cumulative frequency distribution of the intensity variable and its threshold value. The climate and health organisations and scientific communities have used various definitions of EHE intensity, grouped into three categories: climate (or meteorological) based; impact based, or; a combination of both (Vaidyanathan et al., 2017; EPA, 2016; Gachon et al., 2016). The climate based definitions use the absolute or relative values of the maximum, minimum, or average daily temperature, or a combination of the daily temperatures (Laaidi et al., 2012; Medina-Ramon and Schwartz, 2007). The absolute threshold values of temperature are determined from the local historical climatology (NASEM, 2016). The relative threshold values are, however, arbitrarily chosen from either the percentile of exceedance (e.g., 90th to 99th) or deviation from the historical mean value (WHO, 2009).

The impact-based definitions of EHEs are based on the effects of EHEs on the comfort and health of the population. Epidemiological studies on all-cause (non-accidental) mortality data and physiological heat stress metrics are used to establish the thresholds for the EHE intensity. Mortality–based EHE intensity uses the threshold values of the daily maximum, minimum or average temperatures, or their

combinations, that correspond to the onset of the excess mortality above the seasonal average (Gachon et al., 2016; Zeng et al., 2016; Hajat and Kosatky, 2010). Various threshold values are therefore

determined depending on the location, thus delineating the variability of mortality data with the local climate and vulnerability of the population to EHEs. At a community level, the daily average temperature was found to be a better metric than the daily maximum or minimum temperature for predicting mortality events (Guo et al., 2017). Ma et al. (2015) found that the mortality threshold values of the daily (24 hour) average temperature varied from city to city in China with a city-average value of 21°C. Mortality studies in Canada found that the threshold daily maximum temperature is city-dependent and varies from 25 to 28°C (Casati et al., 2013; Goldberg et al., 2011). Similarly, Auger et al. (2015) found that the daily maximum temperature in excess of 29°C was strongly linked to the sudden death syndrome in infants of ages 3 to 12 months in Montreal, Quebec. Mora et al. (2017) found that the global thresholds of the daily average temperature are affected by the daily average relative humidity.

Physiological thresholds for heat stress use the universal steady-state thermal comfort and heat stress metrics to establish the limit at which the exposure to EHEs is perceived as warm, hot, very hot, or becoming dangerous to health (Chindapol et al., 2017; Blazejczyk et al., 2012). Thermal comfort indices include Fanger’s widely used Predicted Mean Vote (PMV) index and the adaptive thermal comfort for naturally ventilated buildings (ASHRAE, 2017). Some of the commonly adopted heat stress indices include the Standard Effective Temperature (ASHRAE, 2017), Perceived Temperature (Jendritzky et al., 2000), Physiological Equivalent Temperature (Höppe, 1999), and Universal Thermal Climate Index (Blazejczyk et al., 2013).

The EHE definitions that combine climate and impact based thresholds (also called bio-meteorological) usually combinesweather variables (e.g. temperature, relative humidity, wind speed, solar radiation) into an empirical index that relates to an individuals’ perceived comfort. Common indices include Humidex (H) (Masterson and Richardson, 1979), Heat Index (HI) (Rothfusz, 1990), and Wet Bulb Globe Temperature (WBGT) (ISO, 2017; NIOSH, 1986). Thresholds of these indices corresponding to selected exposure conditions of a local climate are used to indicate the occurrences of EHEs. These indices have been adopted in: (i) occupational health and safety standards to define heat exposure (WBGT in NIOSH, 1986; ISO, 2017; and CCOHS, 2019); (ii) local jurisdictions and municipalities to produce heat alert and

response plans (Gower et al., 2011; HC, 2012; TPH, 2017), and; (iii) weather service organisations to produce heat alert warnings for the general public (H in EC, 2018; and HI in NOAA, 2018).

Despite the extensive studies on the characterisation of EHE, studies on the development of RSWY are very limited. The Chartered Institution for Building Services Engineers (CIBSE) developed the Design Summer Year (DSY) for overheating analysis of naturally ventilated buildings (CIBSE, 2002). The Design Summer Year represents a warmer summer than the climatic average, and is defined as the third hottest summer within a 21 year climate data set (i.e. equivalent to a return period of 7 years) in terms of the daily average dry bulb temperature from April to September (Levermore and Parkinson, 2006). The CIBSE’s DSY method assumes the indoor operative temperature is equal to the outdoor temperature. In other words, this assumption is equivalent to a conceptual building where occupants are exposed to the

outdoor temperature conditions, but not the effects of solar radiation, relative humidity and wind. The DSY for future climate years are obtained using the morphing methodology that converts the projected monthly averages of the climate variables into hourly quantities, which are then used to adjust the historical DSY data (Belcher et al., 2005; Jentsch et al., 2008). Since its inception, the DSY methodology has gone through a series of modifications due to the fact that the original DSY method was based on a single weather variable (daily average temperature) to classify heat events. Furthermore, studies in some sites found that the DSY under-predicted the overheating risk (Jentsch et al., 2013). In addition, the DSY failed to reproduce some historical extreme heat events such as the 2003 heat wave in Europe (Kershaw et al., 2010). Recently, Eames (2016) has proposed three variants of probabilistic design summer year (pDSY) based on an updated historical climate data set (1984-2013) and the return periods of heat events. Three classifications of heat events were used and analysed for the UK regions to produce these pDSY. The

first classification is based on the Weighted Cooling Degree Hours (WCDH) method, which is a quadratic sum of the operative temperature deviation from the upper temperature of the adaptive thermal comfort, as provided in the BS EN15251 standard (BSI, 2008), over the summer period. The second classification is similar to the first, but uses the Static Weighted Cooling Degree Hours (SWCDH) method in which a static threshold temperature is used instead of the adaptive thermal comfort temperature. The static threshold temperature is calculated as the 93 percentile temperature of the region under consideration. The third classification uses the Threshold Weighted Cooling Degree Hours (TWCDH) method, which combines both WCDH and SWCDH.

The generalized extreme value distribution technique was used to calculate the return period of hot summer events. The three summer design years were established as: pDSY-1 - moderate warm year defined as a year with a SWCDH return period close to 7 years; pDSY-2- intense extreme year chosen as the year with the same duration of heat events as pDSY-1 but with a higher event intensity; and pDSY-3- long extreme year defined as a more intense year than pDSY-1 but with a longer heat event duration. When ordered by the three aforementioned metrics for London (UK), the pDSY-3 and pDSY-2 correspond to the first and second extreme year, respectively. These pDSY have been adopted in CIBSE TM49 (CIBSE, 2014).

More recently, Liu et al. (2016) have reviewed other alternatives to the original definition of DSY, including the near extreme summer reference year (SRY) by Jentsch et al. (2015), and the probabilistic design summer years (pDSY) by Eames (2016) . Liu et al. (2016) have also proposed two new alternatives for the probabilistic hot summer years for the current and future climates: (1) pHSY-1, based on the WCDH, and; (2) pHSY-2, based on the physiological equivalent temperature (PET) of 23°C (upper value of the comfort range). Both pHSY-1 and pHSY-2 weather files were compared with the pDSY files in fourteen UK locations. The study found that both pHSY-1 and pHSY-2 were more robust than pDSY. The pHSY-1 weather file was suggested to be used to study the severity and occurrences of overheating, whereas pHSY- 2 is considered suitable for evaluating thermal discomfort or heat stress.

The forgoing review shows that a universal definition of EHEs, and thus the selection of RSWY for overheating analysis, is a challenging task given that it is difficult to convene the definitions based on climatology and comfort/health in one principal definition. Furthermore, EHEs are transient in nature and their effects on human comfort and health depend on an occupants’ previous (e.g., nighttime conditions which may lead to sleep disturbance) and actual exposure to thermal conditions (Saman et al., 2013; WHO, 2004; NASEM, 2016). In addition, the health outcomes of EHEs are confounded by other climate variables (such as air pollution) and non-climate factors (such as vulnerability of population). However, the effects of heat on human comfort, health, and physiological acclimatization of individuals to the climate conditions are well established, particularly for healthy people. Human comfort and health are affected by absolute temperatures, suggesting that physiological metrics that are based on the transient heat balance of human subjects, and that take into account climate acclimatisation to heat, are likely the

more appropriate approach when establishing universal absolute temperature thresholds for the definition of EHEs and metrics for the RSWY selection for any climate. The absolute threshold temperatures may be determined from laboratory or field studies to cover healthy and vulnerable population categories. The goal of this paper is to address this research gap by developing a methodology to identify EHEs from a long period of historical climate data and select RSWY to analyse the risk of overheating through building simulations. The methodology also applies to RSWY for future climate projections, but the application on projected future climates is not covered in this paper.

O

The specific objectives include:

 To develop methodology to identify reference summer weather data years from a long historical climatic period;

 To develop a novel metric to discern extreme heat events;

 To identify RSWY based on historical weather data for selected Canadian locations.

A

The proposed approach to generate RSWY consists of three steps. The first step is to develop a method to gather the weather data years from a long historical climate data period from which EHEs are

extracted. The second step is to develop a novel metric to characterise all possible types of EHEs that might occur within the reference climate period. The third step is to apply the metric to process and statistically analyse the selected climate data and generate the RSWY for selected Canadian locations. Details of the approach follow in the subsequent sections.

REFERENCE CLIMATE DATA PERIOD

The reference historical climate data spanning 1986-2016 were taken from the observational climate database maintained by Environment and Climate Change Canada (ECCC, 2018). The climate variables include the hourly observations of the global horizontal solar irradiance, total cloud-cover, relative

humidity, air temperature, rainfall, atmospheric pressure, wind speed and direction, and snow-depth. The climate data were collected from all the weather stations within a city domain and then averaged to produce a single weather file for the selected city. The global horizontal solar irradiance was split into its direct beam and diffuse components using the method of Orgill and Hollands (1977). This method was also adopted in the generation of the Canadian Weather for Energy Calculation (CWEC; Morris, 2016). The data gaps in the averaged weather files of cities were filled-in using bias-corrected data from the Climate Forecast System Reanalysis (CFSR) database (Saha et al. 2010). To perform the

bias-correction, multiplicative or additive (depending on the climate variable being corrected) correction factors (CFs) were calculated for each month and hour by comparing the CFSR data with the

METRIC FOR HEAT EVENTS

Heat events as occur for a given climate are time (seasonal) and location dependent. Direct (outdoor) or indirect (indoor) exposure of people to such heat events will induce heat stresses to their

thermoregulatory systems, thus resulting in thermal discomfort (e.g., sweating) and potential health issues (i.e., dehydration, heat cramps, etc.). Such heat events may also be fatal, particularly for vulnerable people such as the elderly, the sick and children. From a climatological perspective, heat events are associated with temperatures warmer than the local seasonal average (NASEM, 2016). However, from the human physiological perspective, fatal heat events may start at temperatures lower than the seasonal average, particularly for vulnerable people (Mora et al., 2017). Furthermore, other climate variables (such as relative humidity, solar radiation, wind) may exacerbate the health effects in humans as occur during heat events. In this regard, the human body acts as both a sensor to detect heat events and actuator responding to the imposed thermal conditions. Therefore, theoretical and empirical physiological heat stress indices are more suitable to depict heat related comfort and the wellbeing of people. To this end many indices having various degrees of complexity and associated limitations have been developed (Blazejczk et al., 2013; Walls et al., 2015; Havenith and Fiala, 2016). One common limitation in the use of these indicies to predict heat events for whole populations is that they have been developed for healthy and young adults in a wakeful (non-dormant) mode, and do not include factors accounting for response of vulnerable populations. The availability of indices for vulnerable people is very limited. In this paper, suitable indices were selected such that they:

1. Are based on a complete heat balance and the thermo-regulatory system of human subjects; 2. Take into account all relevant variables related to the climate and human subjects;

3. Address the transient nature of heat events and changes of activity levels of human subjects throughout the day (such as sleep during nighttime and activity during daytime).

4. Produce objective (not subjective or sensational) indices with an established heat stress scale; 5. Apply to both outdoor and indoor environments so that they can be used to assess heat related

health stresses of people in such environments;

6. Are well validated or adopted in current applicable standards on heat exposure; 7. Are simple to implement in building simulation programs.

Based on literature reviews of existing indices (Walls et al., 2015; Havenith and Fiala, 2016; Holmes et al., 2016), a list of suitable indices was established for each category. A comparison of the different selected indices against the screening criteria is provided in Table 1. Descriptions of the selected indices follow.

Table 1. Comparison of the selected indices against the screening criteria

Criterion SET UTCI PHS WBGT Humidex

Human heat balance Yes Yes yes no no

Thermoregulation controls Yes No no no no

All relevant climate variables Yes Yes yes yes T & RH

Subject variables Yes fixed yes through

adjustments no

Transient Yes No yes no no

Objective scale Yes yes no scale yes yes

Applicable to both outdoor & indoor

environments Yes outdoor yes yes outdoor

Validated & adopted in standards Yes yes yes yes yes

Simple for implementation Yes yes yes yes yes

Humidex — The Humidex is a dimensionless empirical index that combines the air temperature and vapour pressure to indicate the perceived temperature. It was developed by Canadian meteorologists Masterson and Richardson (1979) and is regularly used by the Canadian weather forecast services to produce heat alert warnings for the general public (EC, 2018c), provincial jurisdictions to produce heat alert and response plans (HC, 2012), and occupational health and safety organisations for protecting workers from heat stress (OHCOW, 2019). Humidex values higher than 29 indicate thermal discomfort (EC, 2019c). The Humidex does not explicitly take into account the variables related to human subjects (e.g., clothing, activity level, etc.). Although the index is used for outdoor environment, a comparison of its scale with the SET index reveals that the Humidex is applicable to sedentary human subjects in an indoor environment with still air. For human subjects walking outdoors, the Humidex (with no solar radiation and wind speed effects) corresponding to the same thermal sensation state is several units lower than the one for the indoor environment.

Wet-Bulb Globe Temperature (WBGT) — WBGT is a simplified version of the empirically derived Corrected Effective Temperature (CET) (Parsons, 2003; 2006). It was originally developed for the US Navy for military training in hot environments to avoid heat related injuries (Blazejczyk et al., 2012; Parsons, 2006). Currently, WBGT is the most widely used occupational heat stress index in the world. It has been adopted by the National Institute for Occupational Safety and Health (NIOSH, 1986) and is referenced in the ISO 7243 standard for workplace ergonomics of the thermal environment (ISO, 2017). The WBGT combines weather variables (i.e. air temperature, relative humidity, wind speed, and solar

radiation) in a linear weighting sum of the air, black globe, and naturally ventilated web bulb temperatures. Two equations are produced for healthy and physically fit people wearing reference clothing (i.e., long sleeve cotton shirt and pants with an insulation value of 0.6 clo) and working in either indoor or outdoor environments (ISO, 2017; Parsons, 2006). The American Conference of Government Industrial Hygienists (ACGIH) published the permissible heat exposure Threshold Limit Values (TLVs) of WBGT that correspond to thermal conditions in which the body core temperature does not exceed 38°C for a sustained exposure of up to 8 hours (working day). Some Canadian jurisdictions have adopted these TLVs as occupational exposure limits and others use them as guidelines (CCOHS, 2019). For clothing insulation different than the reference clothing, a clothing adjustment factor is made to the reference values of WBGT. There is, however, no adjustment made to account for gender, body size or height. The preferred calculation of WBGT is based on the direct measurement, using special apparatus and sensors, of the naturally ventilated wet bulb and black globe temperatures (ISO, 2017). Different analytical models to compute WBGT using the weather variables as inputs have also been proposed (ISO, 2017; Kjellstrom et al., 2009), but the most recent model by Liljegren et al. (2008) seems the most accurate, having an accuracy deviation lower than 1°C. Other studies have revealed the limitations of the index to indicate the heat strains under hot exposure. The index generally overestimates heat strains, and is not accurate for low air velocity (< 0.3 m/s) or high humidity levels (Holmer, 2010; d’Ambrosio Alfano et al., 2014; Holmes et al., 2016). Furthermore, the index is expensive and difficult to measure and prone to errors if non-standard measurement equipment is used. It is therefore recommended that the WBGT be used as a conservative index to limit the rise of the body temperature under long steady state heat exposure. Predicted Heat Strain (PHS) — PHS is the ISO 7933 standard model (ISO, 2004) to conduct a detailed heat related stress analysis under long hot exposure as an alternative to the WBGT model of the ISO 7243 standard (ISO, 2017). The PHS uses a non-physiological algorithm to calculate the time limit to reach a body core temperature of 38°C or a body water loss of 3% (or higher with rehydration). The algorithm uses the heat balance of the human body under steady state conditions to calculate the corresponding evaporative heat loss, sweat rate, body core and skin temperatures, and uses those values, combined with their time constants, to calculate the instantaneous values of the predicted evaporative heat loss, sweat rate, body heat storage, core and skin temperatures. As spelled out in the standard, the current PHS model is valid only for constant heat exposure where the body temperatures progress from the neutral condition to steady state values. The PHS takes into account all the relevant environment variables, but is limited to air temperatures between 15°C and 50°C, mean radiant

temperature higher than the air temperature by up to 60°C, activity levels between 0.96 and 4.3 met, and clothing insulation levels between 0.1 and 1 clo. The PHS model was extensively validated by

experimental data from various laboratories and field studies (Malchaire et al., 2001). Given the fact that the PHS does not include any physiological thermo-regulatory controls, it is expected that the model would fail to produce meaningful results under some circumstances out of the validity ranges of the model variables. Indeed, initial testing of the PHS model revealed that the model produces inaccurate values of

body temperatures under multi-stage exposure conditions such as alternating heating and cooling periods. Furthermore, the model is not sufficiently robust to produce steady state results when the initial values of body temperature are different from the neutral values. Laboratory studies confirmed that the PHS model is inaccurate in predicting body temperatures and inappropriate for intermittent exposure and work cycles (Ooka et al., 2010; Havenith and Fiala, 2016; Lundgren-Kownacki et al., 2017).

Universal Thermal Climate Index (UTCI) — UTCI is defined as the equivalent temperature of a reference outdoor environment causing the same physiological response as in the actual environment (Blazejczyk et al., 2013). The reference outdoor conditions are: wind speed of 0.5 m/s measured at 10 m height (corresponding to 0.3 m/s at 1.1 m height), mean radiant temperature (MRT) equal to the air temperature, vapor pressure of 2 kPa (equivalent to 50% relative humidity at 29°C), in which an average person with a clothing level adapted to the outdoor temperature is walking at a speed of 4 km/h,

generating a metabolic rate of 135 W/m2 (2.3 met). The UTCI uses sixth order polynomial regression correlations that link the actual outdoor conditions (air temperature, MRT, vapor pressure, wind speed) to the physiological response of an average person. The correlations were developed using the multi-node bioheat model of Fiala et al. (2012) coupled with a complex clothing model (Havenith et al., 2012) for the outdoor environment. The correlations cover a wide range of activity levels up to heavy exercise and thermal conditions with cold, cool, warm, and hot stresses. The UTCI was designed for weather information services and epidemiological studies (Havenith and Fiala, 2016). Inter-model comparison studies of UTCI with other similar indices found that the UTCI better represents the bio-thermal conditions for humans (Blazejczyk et al., 2012; Havenith and Fiala, 2016).

Standard Effective Temperature (SET) — SET is defined as the temperature of an imaginary indoor environment at 50% relative humidity and MRT equal to the air temperature, in which an imaginary occupant wearing clothing standardized for the activity level has the same heat stress (skin temperature) and strain (skin wettedness) as in the actual environment (ASHRAE, 2017).SET is calculated using the transient two-node bioheat model of Gagge et al. (1986) for indoor exposure. Pickup and de Dear (2000) extended SET to the outdoor environment by including the solar radiation in the definition of MRT. The SET thermal sensation scale (Table 2) is based on the skin temperature and wettedness (Blazejczyk et al., 2012). The SET mode has been successfully used as a comfort and heat stress screening tool with various modifications to its thermo-regulatory control algorithms to suit specific situations such as age, gender, average people versus athletes, transient comfort, and other conditions (Ooka et al., 2010; Havenith and Fiala. 2016; Schweiker et al., 2016; Li et al., 2017). ASHRAE (2017) adopted SET for the calculation of thermal comfort under higher air speeds than the maximum allowable limit in the predicted main vote (PMV) model. Recently, the U.S. Green Building Council has adopted SET for Passive Survivability-Thermal Resilience (space overheating) criteria in buildings to earn LEED points (Overbey, 2018).

Table 2. Thermal sensation scale of SET (Parson, 2003)

SET (°C) Thermal sensation Physiological state

> 37.5 Very hot, very uncomfortable Failure of thermoregulation 34.5 - 37.5 Hot, very unacceptable Profuse sweating

30.0 - 34.5 Warm, uncomfortable, unacceptable Sweating

25.6 - 30.0 Slightly warm, slightly unacceptable Slight sweating, vasodilation 22.2 - 25.6 Comfortable and acceptable Neutrality

17.5 - 22.2 Slightly cool, slightly unacceptable Vasoconstriction 14.5 - 17.5 Cool and unacceptable Slow body cooling 10.0 - 14.5 Cold, very unacceptable Shivering

The transient SET index correlates well with the steady state UTCI and WBGT in Figure 1 for a person walking outdoors under sun shade (incident beam sunlight is set to zero) at a speed of 4 km/h (metabolic rate of 2.3 met) and wearing typical summer clothing (0.5 clo). The calculations used the summer

historical 31-year weather data (May to September) of Ottawa, Ontario, Canada. The figure shows that the SET index correlates well with the UTCI and WBGT indices.

From Table 1, SET satisfies all the screening criteria, and therefore it has been selected as the basis on which to perform the analysis.

Figure 1. Comparison of SET with UTCI and WBGT indices

y = 1.23*x + 0.08 R² = 0.92        ‐5 0 5 10 15 20 25 30 35 40 ‐5 0 5 10 15 20 25 30 35 SE T WBGT y = 0.78*x + 7.60 R² = 0.96 ‐10 ‐5 0 5 10 15 20 25 30 35 40 ‐10 ‐5 0 5 10 15 20 25 30 35 40 SET UTCI

MODIFICATIONS TO THE SET MODEL

The original SET model of ASHRAE (2017) was developed to produce values for steady state conditions (after one hour exposure) when young and healthy adults of an average population are in a wakeful mode. The model also includes default values for the thermoregulatory controls that can be fine-tuned for particular situations. For the purpose of this work, some modifications were brought to the SET model to suite the transient nature of climate heat events. The changes are summarized below:

1. The emissivity and solar absorptivity of typical summer clothing were fixed at 0.97 and 0.66, respectively (Watanabe et al., 2013). The coefficient for vasodilation (CDIL) was fixed at 50

litre/(m2.hr.k) for average people (ASHRAE, 2013). The effective radiation body area and maximum sweat rate were set according to the recent ISO 7933 standard (ISO, 2004).

2. The SET transient model was developed following the approach in Schweiker et al. (2016). The SET value at a given time step is calculated based on the past and present thermal conditions. This calculation continues from day to day.

3. Thermoregulatory controls for a sleeping person were added following the approach in Dongmei et al. (2012). During the sleep mode, the thermoregulatory controls for sweating and vasomotor depend on the sleep stages, namely stages of non-rapid eye movement (NREM or deep sleep) and rapid eye movement (REM). During sleep, a healthy adult experiences four to five cycles of alternating NREM and REM stages. The REM stage constitutes between 20 to 25% of the sleeping time (Colten and Altevogt, 2006; Carskadon and Dement, 2005). Weighed values of the sweating and vasomotor coefficients were computed based on the proportions of the NREM and REM stages. The neutral core and skin temperatures during sleep are also different than those during the wakeful mode. The neutral skin temperature was fixed at 34.6°C (Lin and Deng, 2008a; Dongmei et al., 2012 ), and the core temperature was calculated as an average over a seven-hour sleep period and set equal to 36.41°C.

4. The algorithm of the time limit of exposure of the ISO 7933 standard (ISO, 2004) was added to the transient SET model. The sweating rate and body temperatures were calculated based on the SET bioheat model as compared to that of the PHS model (ISO, 2004) where these factors are simply estimated.

Figure 2 shows a validation of the transient SET model in terms of the average skin and rectal (proxy of core) temperatures as compared with the laboratory measurement, the multi-node Fiala model and the PHS model (ISO, 2004). The measurement and simulation data of the Fiala model were taken from Lundgern-Kownacki et al. (2017). The measurement represents the average of five female and male subjects who had undergone 13 intermittent activities from cycling, stacking, stepping, and resting in an environment of constant air temperature equal to the MRT of 34°C, relative humidity of 60%, and air speed of 0.4 m/s. More details of the experiments may be found in Lundgern-Kownacki et al. (2017). Given the high scatter of the measurement data, the results of the transient SET model show good

representation of the trend of the experimental results (average of female and male data) and are better considering both measurements than that achieved using either the Fiala or the PHS models. The Fiala model underestimated the skin temperature, whereas the PHS model underestimated the rectal

temperature.

Figure 2. Validation of the transient SET model for the skin and rectal temperature predictions

DEFINITION AND CHARACTERISTICS OF HEAT EVENTS

In the context of this paper, a heat event is defined in terms of its effect on the comfort or health of human subjects under its exposure.

DEFINITION OF A HEAT EVENT

A heat event is defined as a meaningful thermal event that occurs in a day over at least a prescribed number of hours and triggers a response of the thermoregulatory system of a human subject under it’s exposure. During a day (24 hours), a human subject takes on various activities in the outdoor and indoor environments. During daytime (defined from wakeup to sunset times), human subjects are assumed walking outdoors under sun shade whilst during nighttime (sunset to wakeup time) they are assumed in a sleeping mode. The magnitude of the heat event of duration (t) is defined as follows:

∑ ∙ ∆

Where SETH (in degree SET*h) is the magnitude of the heat event (zero if negative), t is the calculation time step (h), and SETr is the reference value of SET above the comfort level, which triggers a

physiological response to cool the subject body and/or a subject action to restore thermal comfort. The reference value of SETr may ideally vary with the exposure time according to the activity level of the

human subject throughout a day. In this study, only two daily activities are considered: walking outdoors during daytime and sleeping indoors during nighttime. The SETr should be chosen to correspond to a

end, during daytime, the reference value of SET is chosen to correspond to the initiation of the sweat generation (slightly warm sensation) which corresponds to a Heat Stress Index of 30 (skin wettedness of 30%) according to ASHRAE (2013). Values of skin wettedness greater than 40% would result in health issues, particularly to vulnerable people and therefore should be avoided during exposure to heat events (ASHRAE, 2013). The corresponding reference value of SET is calculated to be 30°C (Table 2) for an average, and un-acclimatized person at the SET standard indoor conditions (sedentary activity at 1 met, summer clothing with 0.6 clo, relative humidity = 50%, and air speed = 0.1 m/s; ASHRAE, 2017). For acclimatized persons, the reference value of SET may be higher, but is not known. Alternatively, the reference value of SET for acclimatized persons is chosen to correspond to the acclimatized value of WBGT of the ISO 7243 standard (ISO, 2017), which is one degree above the un-acclimatized value. According to the profile of SET versus WGBT for outdoor environment (Figure 1), the value of SET for acclimatized persons is 31.2°C. People are assumed not acclimatized during the first month of expected heat events in the summer season, which is set to May for Canadian locations. For other summer months, people are assumed acclimatized to heat events from June to September.

During nighttime when people are assumed asleep, the relationship between the indoor conditions with the outdoor is not known. Therefore, to quantify the direct effect of the outdoor thermal conditions on a subject’s sleep, the reference value of SET for a sleeping person is assumed to correspond to the lower temperature value of the adaptive thermal comfort range for the location under consideration. The rationale behind this selection is that people in indoor environments are more tolerant to heat in naturally ventilated buildings, the indoor conditions are usually higher than the outdoor temperature during

nighttime, and that most people prefer to cover-up and thus endure colder temperatures when sleeping (Lin and Deng, 2008a, b). The corresponding SET values are calculated for subjects in a sleeping mode (0.7 met) with sleepwear insulation value of 1.57 clo (long top and bottom pyjama, sleepers, and

mattress; Lin and Deng, 2008b), air velocity of 0.15 m/s and relative humidity of 50%. Table 3 lists the corresponding calculated values of SET for selected Canadian cities. The average monthly temperatures (Tma) for the period 1981-2010 were taken from Environment Canada’s web site (EC, 2019b). The lower

temperature limits of the adaptive thermal comfort range (TCL) are calculated according to the ASHRAE

Table 3. Reference values of SET for nighttime (sleep) for selected Canadian cities.

City

May Jun. Jul. Aug. Sep.

Tma TCL SET Tma TCL SET Tma TCL SET Tma TCL SET Tma TCL SET

St John's, NF 6.4 16.3 21.1 10.9 17.7 22.2 15.8 19.2 23.4 15.8 16.1 23.5 12.4 18.1 22.6 Charlotte-town, PEI 9.2 17.2 21.8 14.5 18.8 23.1 18.7 20.1 24.2 18.7 18.3 24.1 14.1 18.7 23.0 Halifax, NS 10.1 17.4 22.0 15.2 19.0 23.3 18.8 20.1 24.2 18.8 19.1 24.3 15.5 19.1 23.4 Moncton, NB 10.7 17.6 22.2 16.0 19.3 23.5 19.5 20.3 24.3 19.5 19.0 24.3 14.5 18.8 23.1 Montreal, PQ 12.4 18.1 22.6 17.4 19.7 23.9 19.8 20.4 24.4 19.8 18.7 24.2 14.1 18.7 23.0 Ottawa, ON 13.5 18.5 22.9 18.7 20.1 24.2 21.2 20.9 24.8 21.2 19.9 24.5 15.3 19.0 23.3 Toronto, ON 14.1 18.7 23.1 19.4 20.3 24.3 22.3 21.2 25.1 22.3 21.5 24.9 17.2 19.6 23.8 Winnipeg, MB 11.6 17.9 22.4 17 19.6 23.8 19.7 20.4 24.4 19.7 18.8 24.2 12.7 18.2 22.6 Saskatoon, SK 11.8 18.0 22.5 16.1 19.3 23.5 19 20.2 24.3 19 18.2 24.0 12.0 18.0 22.5 Calgary, AB 9.7 17.3 21.9 13.7 18.5 22.9 16.5 19.4 23.6 15.8 19.2 23.4 11.0 17.7 22.2 Vancouver, BC 12.8 18.3 22.7 15.7 19.2 23.4 18 19.9 24 18 19.9 24.0 14.9 18.9 23.2

The daily magnitude of a heat event as given by equation (1) is further broken down in terms of the nighttime and daytime magnitudes:

∑ ∙ ∆

∑ ∙ ∆

Where:

SETHday : daily magnitude of heat event (degree SET*h);

SETHn : magnitude of heat event for sleeping subjects during the preceding nighttime;

SETHd : magnitude of heat event for active subjects during daytime;

SETsleep : reference value of SET for a sleeping person;

SETawake : reference value of SET a wakeful person.

The start time of the sleeping period is set as the hour immediately after sunset, and the end (wakeup) time is set to 7:00 AM. The daytime period extends from the wakeup time to the sunset hour.

A meaningful heat event is declared if its daytime magnitude (SETHd) is greater than a minimum value

(SETHd,min). This minimum value should be a function of the reference value of SETawake. A high minimum

value may be chosen for low values of SETawake, or vice versa. In this study, this minimum value is fixed at

when subjects are exposed to thermal conditions of one degree SET above the reference value of SETawake(4 hour exposure)

It should be noted that the nighttime magnitude of heat events for the sleeping period (SETHn ) is

accounted for in equation (2) only if the daytime magnitude SETHd is non zero. This stems from the fact

that the excess mortality data from heat events are affected by the exposure conditions during the preceding night (Sheridan and Kalstein, 2004; Loughnan et al., 2010; Saman et al., 2013; Anderson et al, 2013; Kenney et al., 2018). In other words, hot exposure of human subjects during nighttime affects their sleep quality and therefore weakens their thermoregulatory system during hot exposure of daytime (NASEM, 2016; CIBSE, 2015).

DEFINITION OF HEAT WAVES

A heat wave is defined as continuous meaningful heat event occurring over at least two days. Multiple heat waves are declared if they are separated by a recovery period of at least one day. The magnitude of a heat wave is calculated as the summation of the magnitudes of its daily heat events. This is expressed as follows:

∑

CHARACTERISTICS OF HEAT WAVES

Heat waves are characterized by three features: duration, intensity and severity. The duration is

measured in terms of the number of days of sustained heat events. The severity of a heat wave is similar to its magnitude as given by equation (5). Finally, the intensity (degree SET) of a heat wave is calculated as the ratio of severity to duration (expressed in hours).

As per this definition, many types of heat waves during a given summer season may occur. One can distinguish three major types, namely: long, intense and severe. Long heat waves are usually mild, whereas intense heat waves are usually short. Severe heat waves account for both the duration and intensity. Combinations of heat wave types such as long and intense, long and severe, or intense and severe may occur as well. It should, however, be mentioned that any type of heat wave with the same value for severity would result in the same physiological effect, as was found in studies of acclimation to heat (Periard et al., 2015).

G

RSWY

Reference summer weather years are extracted from a 31 year period of climate data, as was previously mentioned, to capture all types of heat waves. The summer period was fixed from May to September. The methodology to develop such weather data includes the following steps:

1. Heat waves for each year are identified and sorted by maximum duration, intensity, and severity. The maximum values are assigned to each year.

2. The return period of heat waves is fixed to 15.5 years, which corresponds to the second extreme year out 31 years.

3. The cumulative frequency distribution of the maximum values of duration, intensity and severity are plotted and fitted with suitable distribution functions such as the generalized extreme value

distribution (GEV), Gumbel distribution, or any other suitable function (Note: There is no single function that satisfies all the maximum value distributions). The extreme years are then chosen if their calculated return periods are close to the fixed value of 15.5 year. In most cases, the extreme years are those ranked second by their maximum values for duration, intensity and severity. If the first ranked years have the same frequency values or are very close, the one with the highest severity value is chosen to represent the second extreme year as the first extreme year is not known. Similarly, if the second ranked extreme years have the same frequency values, the one with the highest severity value is chosen.

It is often preferable to have one representative extreme year instead of three types of extreme years to reduce the amount of effort to conduct parametric studies in whole building simulations. This may be justified by the fact that long extreme years may include some days with high intensity as short intense years. Furthermore, the three types of severe years may be interchangeable. That is, a long extreme year ranked second under duration may be ranked first under intensity or severity, or vice versa. In this regard, a one representative extreme year, if not ranked as the most (first) extreme under either type of heat wave, may be chosen if it combines two features of the heat wave such as long and intense years, long and severe years, or intense and severe years.

The foregoing methodology is applied to generate the RSWY for selected Canadian cities. The methodology is also compared with other methods to identify heat waves and generate RSWY. Of particular interest for comparison is the CIBSE’s method (CIBSE, 2014) to generate Design Summer Years (DSY) and the Environment Canada’s definition of heat waves.

The CIBSE’s method uses the weighted cooling degree hours (WCDH) to characterise heat waves. The magnitude (or severity) of heat waves expressed in WCDH is given by the following equation:

∑ ∑ ,

Where To,i is the operative temperature at hour (i) of the day, and TCU is the upper temperature of the

adaptive thermal comfort for the location under consideration.

The operative temperature is assumed equal to the air temperature, and TCU is calculated using the local

monthly average temperatures according to the ASHRAE-55 standard (ASHRAE, 2017). The duration and intensity of heat waves are computed in a similar way as mentioned before. It should be noted that the threshold temperature TCU is fixed throughout the day (including nighttime) irrespective of the daily

activity levels and clothing of building occupants. Furthermore, the CIBSE method does not take into account any weather variables other than temperature.

Environment Canada (EC) identifies heat waves based on the combination of the daily maximum and minimum temperatures and the daily maximum humidex. These thresholds vary from province to province as outlined in (EC, 2019a). For example, for Ottawa and Toronto, Ontario, heat waves are declared if the daily maximum temperature is equal to or above 31°C and the daily minimum temperature is equal to or above 21°C; or the humidex is equal to or higher than 42. Whereas, for Montreal, Quebec, heat waves are declared if the daily maximum temperature is equal to or above 30°C and humidex is equal to or higher than 40; or the daily maximum temperature is equal to or higher than 40°C. It should be noted that the EC’s definition yields only the duration of the heat wave as it combines two different variables

(temperature and humidex).

Table 4 compares of the first five extreme years as obtained using the proposed method using SET, CIBSE’s and EC’s methods for Ottawa, Ontario. The values of duration, intensity and severity in the table are the maximum values of the heat waves occurring in each year. Under the duration ranking, the proposed method yields several years with the same frequency of occurrence. These years have the closest return period to 15.5 years, and are therefore ranked second (the first extreme year does not exist in this case). The year with the highest severity is then chosen among them, which is 2010 (Jul. 5 to 11; Note the severity values of heat waves in each year other than the maximum values are not shown in the table). Under the intensity and severity rankings, the second ranked years of 2006 (Jul. 31 to Aug. 3) and 1987 (Jul. 8 to 13) are chosen as extreme years, respectively. As per the criteria to select a

representative extreme year, the year of 1987 is chosen as the extreme year for duration, intensity and severity.

The CIBSE’s method yields 2012 (Jul. 11 to 17), 2012 (Jun. 20 to 21) and 2002 (Sep. 7 to 10) as extreme years for duration, intensity and severity, respectively. The year of 2012 is chosen as the representative extreme year. The EC’s method yields the year of 2010 or 1988 as the long extreme years.

Figures 3 and 4 compare the temperatures and SET of the extreme heat waves as obtained by the proposed and CIBSE’s methods for Ottawa, Ontario. The extreme heat waves of the typical meteorological year (TMY) are also plotted in the figures. It is clear that using the TMY weather

underestimates the overheating risk given the fact that the heat waves of TMY are a lot shorter and milder in terms of temperature and SET. The extreme heat waves obtained by the proposed method are

comparable with the CIBSE’s method in terms of the maximum daily temperature, but have higher nighttime temperature and higher SET during both day and night times. It should be noted that the intense heat waves, although shorter than the long or severe heat waves, balance the nighttime and daytime SET. This may result in intense heat waves that have high daytime SET and low nighttime SET, or heat waves with high nighttime SET and moderate or high daytime SET. The CIBSE’s method does not, however, account for nighttime temperature.

Table 4. Comparison of extreme years as obtained using the proposed, CIBSE’s and EC’s methods for Ottawa, Ontario. The selected years are highlighted in the bold red color.

Method  Year Duration, (days) Year Intensity, (_˚C) Year Severity, (_˚C*h) Proposed 2010 7 2002 3.83 2010 437 1987 7 2006 2.89 1987 415 2006 7 1994 2.75 2005 336 2003 7 2010 2.6 2013 280 2005 6 1987 2.47 1988 278 CIBSE 2001 10 2002 11.26 2001 1679 2012 7 2012 10.82 2002 1081 2013 7 2010 9.52 1988 982 1988 6 1995 9.42 2010 979 1989 6 1994 9.06 1994 869 EC 1987 5 - - - - 2010 or 1988 4 - - - -

Figure 3. Temperature and SET during for the extreme heat wave periods obtained using the proposed method for Ottawa, Ontario.

Figure 4. Temperature and SET during for the extreme heat wave periods obtained using the CIBSE’s method for Ottawa, Ontario.

Figures 5 and 6 compare the temperatures and SET of the extreme heat waves as obtained by the proposed and CIBSE’s methods for Montreal, Quebec. Using the proposed method, the extreme years for long, intense and severe heat waves are 2010 (Jul. 5 to 11), 2010 and 1987 (Jul. 8 to 14) respectively. Year 2010 is also the most (first) severe year. The representative extreme year is then 1987.

The CIBSE’s method produces, however, the following years in order 2002 (Aug. 10 to 18), 1994 (Jun. 15 to 18) and 2010 (Jul. 4 to 9). The longest and most severe year is 2001 (Jul. 31 to Aug. 9). The

representative extreme year is chosen to be 2010.

As for the EC’s method, the long extreme year is 2010 (Jul. 5 to 8), and the longest year is 1987. The proposed and EC’s methods produce a more extreme year (2010) for long and intense heat waves than the CIBSE’s method (2002 or 1994). However, the CIBSE method produces a more severe year (2010) than the proposed method (1987). Year 2010 is considered as the most (ranked first) severe year by the proposed method.

Figure 5. Temperature and SET during for the extreme heat wave periods obtained using the proposed method for Montreal, Quebec.

Figure 6. Temperature and SET during for the extreme heat wave periods obtained using the CIBSE’s method for Montreal, Quebec.

Figures 7 and 8 compare the temperatures and SET of the extreme heat waves as obtained by the proposed and CIBSE methods for Toronto, Ontario. Using the proposed method, the extreme years for long, intense and severe heat waves are 1987 (Jul. 7 to 13), 2006 (Jul. 27 to Aug. 3) and 2011

(Jul. 20 to 23) respectively. Year 2006 is also the longest extreme year (ranked first). The representative extreme year is then 1987.

The CIBSE’s method produces the following years in order 2005 (Jul. 9 to 14), 2001 (Aug. 5 to 10) and 2011 (Jul. 20 to 23). The most intense and severe year is 1988 (Jul. 6 to 10). The representative extreme year is chosen to be 2001.

Both the proposed and CIBSE’s methods produce the same intense year (2011). For other heat wave types, the proposed method produces less severe years than the CIBSE’s method in terms of

temperature, but more severe in terms of SET. This is due to the fact that Toronto’s weather is more humid than Ottawa or Montreal. The CIBSE’s method may therefore underestimate the health effects of the outdoor climate.

Figure 7. Temperature and SET during for the extreme heat wave periods obtained using the proposed method for Toronto, Ontario.

Figure 8. Temperature and SET during for the extreme heat wave periods obtained using the CIBSE’s method for Toronto, Ontario.

From the preceding results, it is clear that the proposed method yields extreme years that have higher SET values during daytime and nighttime than the CIBSE’s method. However, in terms of temperature, the CIBSE’s method yields, in general, extreme years with higher daytime temperatures, particularly in

humid areas (e.g., Toronto). This is because the CIBSE’s method screens out heat wave days that have high daytime temperatures without accounting for nighttime (sleep). As per the proposed method, high nighttime temperatures increase the intensity of heat waves and may therefore result in heat wave days with lower daytime temperatures. For locations with dry (e.g., Saskatoon, SK) or mixed dry and humid (e.g., Ottawa, Montreal) climates, the proposed method yields daytime high temperatures that are comparable to the CIBSE’s method, but with higher nighttime temperatures in some cases. The EC’s method is limited in that it does yield intense or severe years.

The proposed method is therefore suitable for any climate and requires building simulation tools to have the capabilities to model both heat and moisture transfer in buildings so the indoor conditions

(temperature and relative humidity) are dependent on the outdoor conditions. The CIBSE’s method, on the other hand, might not be suitable for mixed or humid climates as the humidity can alter the

physiological response of building occupants. Furthermore, the CIBSE’s method does not require building simulation tools to have the capabilities to model the moisture transfer in buildings. In this regard, the health effects of the indoor conditions may be underestimated. It should be added that the CIBSE’s method may be improved to include a different reference temperature for nighttime (sleep) to account for days with high nighttime temperatures that would affect the sleep quality and hence the physiological response of building occupants during the next hot days.

Table 4 lists the extreme years for selected Canadian cities as obtained by the proposed method. All the years encompass heat waves with duration of at least two days, except for the mild weathers of Halifax (NS) and St. John’s (NF) with heat waves of one day duration.