Energy & Buildings 214 (2020) 109893

Contents lists available at ScienceDirect

Energy & Buildings

journal homepage: www.elsevier.com/locate/enbuild

A review of adaptive thermal comfort research since 1998

R. de Dear a,∗, J. Xiong a, J. Kim a, B. Cao b
a
Indoor Environmental Quality Laboratory, School of Architecture, Design and Planning, The University of Sydney, Australia
b
Department of Building Science, Tsinghua University, Beijing, China

a r t i c l e i n f o a b s t r a c t

Article history: The rapid escalation of cooling demand in buildings set against the backdrop of a global climate emer-
Received 15 December 2019 gency is intensifying research activity on adaptive thermal comfort. In this review of the topic spanning
Revised 19 February 2020
the last 21 years we examine progress or lack thereof, in various research themes. These include adaptive
Accepted 19 February 2020
comfort theory, adaptive comfort practice (standards), contextual effects on adaptive comfort (building
Available online 21 February 2020
typologies), shifting boundaries of the comfort zone and the dynamics of comfort expectations. Lastly we
Keywords: assess the implications of adaptive thermal comfort for cognitive performance of building occupants.
Adaptive © 2020 Elsevier B.V. All rights reserved.
Thermal comfort
Comfort theory
Climate
Standard
Building typology
Cognitive performance

1. Introduction peratures in the summer season (or hot climate zones), or towards
cooler temperatures in winter or cold climate zones [3–5].
Thermal comfort is usually defined as “that condition of mind This paper focuses on the second of these two seemingly op-
that expresses satisfaction with the thermal environment” [1], and posed comfort models - adaptive thermal comfort. Twenty one
the provision of comfort indoors is a fundamental objective of ar- years have passed since Brager and de Dear published their orig-
chitects, engineers, and the allied buildings sector professions. As inal literature review on the subject of ‘thermal adaptation in the
a subject of research, thermal comfort sits at the intersection of built environment’ [6] and a substantial quantum of research on
the physiological, psychological, engineering, and architecture dis- the subject has been published since then. A mid-2019 bibliomet-
ciplines, but it is the last two that have probably contributed most ric analysis [7] using data from the Scopus database indicated that
to the subject since the term ‘thermal comfort’ was coined early over 1200 journal and conference proceedings papers with titles,
last century. Two broad conceptual models have emerged to sci- abstracts, or keywords containing the words ‘adaptive’ AND ‘ther-
entifically understand, and more practically, predict human ther- mal’ AND ‘comfort’ had been published since the first adaptive
mal comfort. The engineering disciplines have contributed a deter- comfort paper by Nicol and Humphreys in 1973 [3]. But research
ministic model analogous to the load estimation concept in HVAC activity on the topic remained muted up until the turn of the cen-
engineering, in which the relationship between building occupant tury, at which point an intensification of research interest has oc-
and their indoor climate is construed as a simple balance between curred, certainly enough to justify a sequel to Brager and de Dears’
the body’s metabolic heat production on the one hand, and the 1998 review [6].
transfer of that heat to the indoor environment through convec- Before beginning the literature review it is instructive to ex-
tive, conductive, radiative and latent heat fluxes, on the other [2]. amine the reasons behind this proliferation of research papers on
The alternative comfort model, labelled ‘adaptive comfort’ is more adaptive comfort in the last 21 years. The root cause is undoubt-
commonly associated with the architecture and allied built envi- edly the climate emergency currently gripping the planet, and the
ronmental disciplines, and posits thermal comfort within a frame- growing realisation of the key role that will be played by the res-
work of contextualized perceptual relativism. Recent past thermal idential and commercial buildings sector in adapting to, and mit-
environmental exposures nudge building occupants’ thermal phys- igating the crisis [8]. Delivering habitable indoor climate through
iology, behaviour, and comfort expectations towards warmer tem- heating and cooling represents the most significant energy end-
use in buildings. With the compounding effects of rising household
∗
incomes, global warming, population growth and urbanisation par-
Corresponding author.
E-mail address: richard.dedear@sydney.edu.au (R. de Dear).
ticularly in hot countries, the use of air conditioning will inevitably

https://doi.org/10.1016/j.enbuild.2020.109893
0378-7788/© 2020 Elsevier B.V. All rights reserved.
2 R. de Dear, J. Xiong and J. Kim et al. / Energy & Buildings 214 (2020) 109893

increase over coming decades, especially in middle-income coun- zones. In their ePMV re-analysis, data from four of the ASHRAE
tries [9]. Since 1990 the annual sales of air conditioning units RP-884 [21] field study cities were included - Bangkok, Brisbane,
increased nearly fourfold to 135 million units, resulting in ap- Athens and Singapore. For each observation of thermal comfort,
proximately 1.6 billion systems currently in use, supplying about metabolic rates registered in the RP-884 database were reduced
11,675 gigawatts (GW) of cooling by 2016 [10]. On current trends by 6.7% for every thermal sensation scale unit that the basic PMV
the energy demand from air conditioning is likely to more than model predicted above neutral. PMV was then recalculated using
triple by 2050, which is roughly equivalent to China’s entire cur- a down-scaled metabolic rate and then adjusted by multiplication
rent electricity demand [10]. Total electricity use for cooling world- with a so-called expectancy factor.
wide was estimated to be 20 0 0 TWh in 2016, representing approx- The a priori logic used to justify this adjustment was that the
imately 10% of all electricity consumption across all sectors world- building occupants, when feeling warm, unconsciously tend to
wide. slow down the intensity of their physical activities, but this spec-
Rising demand for space cooling, particularly at times of peak ulation was not supported by any empirical evidence, as far as we
electricity demand, is putting excessive strain on ageing electric- know. By reduction ad absurdum, a PMV calculation of +3 on the
ity infrastructure in many countries, as well as driving up green- 7-point scale would require a reduction in metabolic rate of 20%;
house gas emissions in places where that infrastructure is based such discrepancy is comparable to the difference between, for ex-
predominantly on fossil fuels. Worldwide carbon dioxide emissions ample, sitting at a desk and reclining on a bed - too large to be
attributable to space cooling have increased threefold since 1990 to explained by unconscious behavioural adaptation to warm expo-
1130 million tonnes in 2016 [10]. Since cooling the indoor climates sure.
of buildings is all about occupants’ thermal comfort, it is not sur- Another ePMV tweak was that expectation affected occupants’
prising that so much research attention has been focussed on re- judgement of a given warm environment. For occupants of a non-
ducing the comfort related energy demand of buildings through a air-conditioned buildings, the expectancy factor ‘e’ was guessti-
better understanding of human thermal adaptability and its more mated to be between 1 and 0.5, depending on the duration of
effective exploitation. warm exposure and whether such buildings could be compared
Another driver for the interest in adaptive thermal comfort with many air conditioned counterparts in the same climate zone.
theory has been the failure of the “one size fits all” approach of A larger e factor signified higher comfort expectations. If the
conventional comfort theory to deliver on the promise of near- weather is warm all year around and there are air-conditioned
universal occupant satisfaction, as revealed in a series of rigorous buildings are scarce in this region, the e may be down-scaled to
field validation studies funded by ASHRAE in the 1980s and 90 s 0.5, but it may be 0.7 if air conditioning is a common occurrence.
[11–14]. Comparable field studies conducted in buildings practic- If warm periods occurred briefly during the summer season in re-
ing adaptive comfort have typically demonstrated higher levels of gions where air-conditioned buildings are common, comfort expec-
occupant comfort and satisfaction (e.g. [15]). tations ‘e’ would be presumed to be 0.9 - 1.0. In the Fanger and
Rather than present an exhaustive review of adaptive thermal Toftum analysis [16], expectancy factor for Brisbane in sub-tropical
comfort literature in the last twenty years this paper is structured Australia was set at 0.9 while for equatorial hot-and-humid Singa-
around salient themes in the last twenty years of publications in pore was arbitrarily lower at 0.7, despite the ubiquity of air con-
this domain. The first theme arises from an Achilles heel of adap- ditioning in the latter location. As will be noted in a subsequent
tive thermal comfort theory; that the main adaptive models are section in this review, differences in comfort expectations are cer-
based on statistical analyses of field data, and as such, essentially tainly a relevant dimension of psychological adaptation to heat, but
“black-boxes.” In the first section of this review we examine the Fanger and Toftums’ [16] attempt to quantify it seems somewhat
main attempts at explanatory or “transparent” adaptive comfort contrived, simply to “correct” discrepant PMV predictions in build-
models. The second section of this review’s foci is the body of ings located in warm-to-hot climate zones. Spatial and temporal
adaptive comfort regulatory documentation that has emerged in quantification of a population’s air conditioning exposure in the
the last 21 years. Those regulatory documents were based predom- metric of AC penetration rate [22] would seem to be a more ra-
inantly on data collected from office buildings, so in the third sec- tional strategy.
tion of this review we examine research directed towards adaptive In a conceptually similar attempt to close the PMV model’s pre-
comfort in other building typologies. The fourth theme of this re- diction ‘scissors’ depicted in Fig. 1, Yao et al. [18] introduced a
view examines the temporal dimension of comfort zones. The final feedback concept to thermal comfort, but it was still contained
section critically reviews another of the adaptive comfort Achilles within the “black box.” The discrepancy between PMV prediction
heels, namely the impact of the warm temperatures found inside and observed comfort is labelled psychological and behavioural ef-
adaptive comfort buildings on the cognitive performance of their fects, and returned to the PMV calculation as a negative ‘‘Adap-
occupants. tive Feedback’’. The adjustment was quantified with an adaptive
coefficient ‘λ’ [18]. In warm or hot environments, the λ coefficient
2. Explanatory adaptive comfort models is positive while in cold conditions it is negative. Higher abso-
lute value of λ indicates larger modification effect on PMV com-
Since the first publication of their adaptive model and pro- ing from psychological and behavioural adaptation. In a field sur-
posal for an ASHRAE Standard 55 adaptive comfort standard in vey in Chongqing University, China, onsite indoor environmental
1998 [4] several research teams have attempted to convert the observations were collected and applied to PMV calculations that
“black box” into an explanatory “transparent” model (e.g. Fanger were linked to simultaneous actual mean votes (ACM). The adap-
et al.’s ePMV [16], Gao et al.’s eSET and aSET [17], Yao et al.’s aPMV tive feedback coefficient λ was then derived from the discrepancy
[18], and Schweiker et al.’s ATHB [19]). All of these approaches, between PMV and actual votes using a least squares method. This
in fact, are attempts to “close the scissors” between the classic curiously circular logic uses its own output (ACM) to modify the
PMV model’s predicted comfort temperatures and those actually input (PMV) to forcibly close the “scissors” between PMV and AMV
observed in field studies, as depicted in Fig. 1 below. in natural ventilated buildings. Moreover, as a design tool and also
Fanger and Toftum [16] proposed the extension of the PMV building operational strategy, the Yao et al. [18] model is impracti-
model (ePMV) by reducing the metabolic heat input parameter cal because it is premised on large-scale onsite questionnaire sur-
and introducing an expectancy factor to make it more applicable vey and environmental monitoring in order to derive the adap-
to non-air-conditioned buildings situated in warm-to- hot climate tive λ coefficient. Circularity of logic notwithstanding, this aPMV
 R. de Dear, J. Xiong and J. Kim et al. / Energy & Buildings 214 (2020) 109893 3

Fig. 1. The “scissor graph” depicts the discrepancy between theoretical predictions (PMV) and actual observations (RP-884) of thermal comfort temperatures in field studies
(modified from [20]). The PMV theoretical predictions were based on observed data for air and mean radiant temperatures, air speed, humidity, clothing insulation and
metabolic rates in actual naturally ventilated buildings where thermal comfort field studies were conducted in the RP-884 project [4].

method forms the basis of the Chinese Evaluation Standard for the unsuited to explaining their outputs or enhancing thermal comfort
indoor thermal environment [23]. To make it useable the standard theory – in essence they remain “black box” models.
is also accompanied by estimates of the adaptive coefficient λ in In summary, none the above-mentioned attempts at “closing
China’s various climates zones [24]. the scissors” in Fig. 1 provide a comprehensive and compelling
Since their publication neither ePMV nor the aPMV has been evidence-based explanation for adaptive comfort models but they
widely accepted and applied by the thermal comfort research com- do make incremental contributions of new knowledge to the do-
munity. In yet another attempt at an explanatory adaptive com- main. It is clear that PMV remains to date the most convincing,
fort model Gao et al. [17] applied the concepts of ePMV and aPMV albeit partial adaptive comfort theory by virtue of its adjustable
across to the alternative index known as Standard Effective Tem- behavioural parameters for clothing, metabolic rate and air move-
perature (SET) [25] because the effects of elevated air velocity ment. However we believe the absence of a rational, evidence-
can be reflected by SET through both convective and evapora- based parameterisation of the psychological adaptive concept of
tive sweating heat exchange mechanisms. The Gao et al. [17] ap- expectation lies behind the persistent discrepancy between PMV
proach predicts thermal sensation (TSENS) based on the SET model predictions and statistically-based adaptive models fitted to com-
and then substitutes it in adjusted PMV models, namely ePMV fort observations, particularly in naturally ventilated buildings lo-
and aPMV, to obtain modified SET models. Therefore these mod- cated in warm to hot climate zones.
ified SET models suffer from the same circularity in logic that we
pointed out for ePMV and aPMV. Gao et al.’s [17] survey was car- 3. Regulatory documents and standards on adaptive comfort
ried out on a university campus in Xi’an which is located in China’s
cold temperature zone where the Li et al. [24] adaptive coefficient The bibliometric analysis of keywords “adaptive comfort” by
λ is estimated between 0.029 to 0.167, differing markedly from the Parkinson et al. [7] discovered the significant role played by com-
λ 0.21 used by Gao et al. [17] to fit their model to Xi’an university fort standards in boosting the visibility of particular adaptive mod-
classroom observations. els. A distinct spike in citations for de Dear and Bragers’ [4] adap-
Schweiker and Wagner [19] presented a highly original frame- tive model results from its inclusion in ASHRAE Standard 55
work for an Adaptive Thermal Heat Balance (ATHB) using cloth- since 2004. And it was EN15251’s [26] inclusion of Nicol and
ing insulation to represent behavioural adaptation and metabolic Humphreys’ [5] that has attracted a high volume of traffic from
rate to quantify both physiological and psychological adaptive pro- the adaptive comfort research community to that paper. There-
cesses. In our estimation this work represents the best attempt to fore it is appropriate that this review paper includes a survey of
date to disentangle the effects of individual adaptive processes op- the plethora of adaptive comfort standards and design guide docu-
erating within the concept of thermal comfort. In summary, ATHB mentation that has evolved in the last two decades. Carlucci et al.
is reflecting adaptations in the two human-related inputs to PMV, [27] conducted a detailed intercomparison of adaptive thermal
namely clothing insulation and metabolic rate but it ignores the comfort models commonly used in built environment standards.
intractable dimension of comfort expectations. Furthermore the Adaptive thermal comfort models were first codified in ASHRAE
plethora of assumptions and complexity of the ATHB model fails standard 55 in 2004 [28], and it has been enhanced in subse-
to significantly improve accuracy of prediction using classic PMV quent revisions. Adaptive comfort models were also included in EN
and adaptive models as the benchmark. 15,251 [26], prEN 16,798–1 [29], GB/T 50785 [23], ISSO 74 [30],
The last couple of years of thermal comfort literature wit- etc., and became the mainstream to reflect the realistic thermal
ness an upsurge in machine learning approaches to the predic- requirements for the occupants of built environments. Table 1 lists
tion of individual comfort responses, including Classification Tree, basic information of different adaptive comfort standards and iter-
Gaussian Process Classification, Gradient Boosting Method, Kernel ations thereof.
Support Vector Machine, Random Forest and Regularized Logis- Since the adaptive model was included in ASHRAE 55 since
tic Regression. Various independant variables have been used in 2004 and many subsequent regulatory documents, it has been
their research designs, including but not limited to skin tempera- widely used for building design and operations. More than 350
tures, subjective comfort questionnaire responses and occupant be- articles are found in Scopus database with keywords containing
haviour. The number of papers in this category is too large for de- “ASHRAE 55” AND titles, abstracts, or keywords containing “adap-
tailed critique here, probably warranting their own focused litera- tive model”. The adaptive comfort standards have been introduced
ture review. Suffice to say that, while machine learning approaches into a large number of case studies conducted worldwide (e.g. [35–
hold promise of increased predictive skill at the resolution of an 40], etc.), in order to analyse the performance of those buildings
individual building (or vehicle) occupant, they are fundamentally and to provide useful feedback to architects and engineers.
4 R. de Dear, J. Xiong and J. Kim et al. / Energy & Buildings 214 (2020) 109893

Table 1
Basic information about various adaptive thermal comfort standards.

Standard/code Scope of application and type Applicable building operation mode Data source Outdoor reference
of building temperature

ASHRAE 55 (2004 and 2010) Predominantly office Occupant-controlled naturally RP-884 Monthly mean outdoor
[28,31] buildings. conditioned spaces (No mechanical temperature
cooling system is installed, and no
heating system is in operation.)
ASHRAE 55 (2013 and 2017) Not specifically mentioned. Occupant-controlled naturally RP-884 Prevailing mean outdoor
[1,32] conditioned spaces (No mechanical temperature (no fewer than
cooling system is installed, and no 7 and no more than 30
heating system is in operation.) sequential days horizon)
EN15251 (2007) [26] and Office buildings, residential Not equipped with mechanical cooling SCATs Running mean external
prEN 16,798–1 (2015) [29] buildings, conference rooms, systems (Mechanical ventilation with temperature (approximate
auditorium, cafeteria, unconditioned air is allowed.) equation with 7-day
restaurants, and classrooms. horizon)
Dutch ISSO 74 (2004) [30] Not specifically mentioned. Alpha space refers to free-running RP-884 Running mean outdoor
situations in summer with operable temperature (approximate
windows. Beta space relies on equation with 3-day
centrally-controlled cooling in summer. horizon)
Dutch ISSO 74 (2014) [33] Not specifically mentioned. Alpha space refers to free-running SCATs Running mean external
situations in summer with operable temperature (approximate
windows. Beta space relies on equation with 7-day
centrally-controlled cooling in summer. horizon)
National Building Code of Not specifically mentioned. Naturally ventilated buildings, Field studies in India 30-day outdoor running
India (2016) [34] mixed-mode buildings, and air mean temperature
conditioned buildings
Chinese GB/T 50,785 (2012) Not specifically mentioned. Free running buildings Field studies in China Running mean outdoor
[23] temperature (7-day horizon)

3.1. ISO Standard 7730 (1984/1994/2005) [41,42] in the preceding hour and their current clothing insulation level.
The occupant’s right-here-right-now subjective evaluations included
PMV/PPD comfort model is the International Standards Organi- thermal sensation, acceptability and preference. For each unit of
sation’s method for evaluation of moderate thermal environment analysis (building/season) the neutral operative temperature was
in ISO 7730. Since PMV/PPD is an explicit heat-balance model derived as the zero-solution of the linear regression equation of
ISO 7730 cannot be regarded as a conventional adaptive standard, thermal sensation vote versus operative temperature. The adap-
but it does include the adaptive behavioural effects of metabolic tive model appearing in the standard was based on a regression
rate, clothing insulation worn, and room air speed. So in this of all building/season units’ neutral operative temperature against
sense it can be seen as a standard incorporating one of the prin- an index of concurrent outdoor temperature. The original de Dear
cipal categories of adaptive thermal comfort response, namely and Brager [4] version used mean monthly outdoor New Effec-
behavioural adjustment of the occupant and/or their indoor tive Temperature (ET∗ ) to capture the effects of humidity as well
environment. as temperature in the climate index, but at the insistence of the
ASHRAE committee charged with drafting Standard 55 [28] this
3.2. ASHRAE Standard 55/RP884 [4,20] was simplified to just mean monthly outdoor air temperature [20].
To make the adaptive model more useable in building energy sim-
The adaptive comfort model embedded in ASHRAE Standard 55 ulation software the 2013 edition of Standard 55 [32] broadened
first appeared in the 2004 edition [28] with explicit restrictions on the options for outdoor climate indices by inclusion of a prevail-
the scope of its applicability. The building should meet the follow- ing mean outdoor temperature – effectively a time-weighted, run-
ing criteria to be eligible to be evaluated by the adaptive comfort ning mean outdoor air temperature which can be derived from a
approach; Typical Meteorological Year (TMY) data file. Eighty and 90% oc-
cupant thermal acceptability criteria for general (i.e. whole-body)
• No mechanical cooling system (e.g., refrigerated air condition- thermal comfort were estimated for each building as the range of
ing, radiant cooling, or desiccant cooling installed, thus preclud- operative temperatures falling between the adaptive model’s pre-
ing mixed-mode ventilation designs). dicted mean thermal sensations of ±0.85 and ±0.5 respectively.
• No heating system is in operation. The 2013 edition of Standard 55 also allowed the upper accept-
• Occupants must have metabolic rates ranging from 1.0 to 1.3 ability adaptive temperature limits to be incremented higher based
met. on an SET-index calculation of air speed correction factors; 1.2,
• Occupants must be free to adapt their clothing to the indoor 1.8 and 2.2 K respectively for air speeds of 0.6, 0.9, and 1.2 m/s
and/or outdoor thermal conditions within a range at least as respectively [32].
wide as 0.5 to 1.0 clo.

ASHRAE Standard 55’s adaptive model was proposed by de Dear 3.3. EN 15251 (and its revision prEN 16798)/SCATs [5,26]
and Brager in 1998 based on field study data. The 21,0 0 0 sets of
raw data observed in 160 mostly office buildings across five dif- In standard EN 15251 and its revision prEN 16798, both PMV
ferent continents in the RP-884 database were sorted, aggregated and adaptive models were included. In the 2007 EN15251, the
and analysed at the resolution of building and season. The obser- adaptive model was based on empirical data from about 1500
vational data included the PMV/PPD model’s main environmen- comfort questionnaire responses recorded in the pan-European
tal parameters - air temperature, globe temperature, mean radi- SCATs (Smart Controls and Thermal Comfort) project described in
ant temperature, air velocity, and relative humidity, as well as es- Nicol and Humphreys [5]. It was aimed to reduce energy use by
timates of occupant-related parameters including metabolic rate air conditioning systems by varying setpoint temperatures in line
 R. de Dear, J. Xiong and J. Kim et al. / Energy & Buildings 214 (2020) 109893 5

with outdoor temperature through the use of an ‘adaptive algo- tionnaires were administrated to occupants once per season at
rithm’. A wide range of physical measurements were made and the same time that contiguous indoor climate measurements were
subjective responses recorded in 26 European offices in France, made by handheld instruments. Due to the non-normal distribu-
Greece Portugal, Sweden and the UK at monthly intervals over ap- tion of temperatures found in the Indian building samples, the
proximately one year. A standard relation between thermal sen- Griffiths method rather than linear regression was used to derive
sation and operative temperature was derived, excluding the ef- thermal neutrality operative temperature for each ‘Building/Season’
fects of day-to-day adaptation and incorporating an adjustment unit of analysis. For ease of further analysis, it was decided to use
for the presence of error in the predictor variable, and then ap- conservative Griffith coefficients of 0.16/K for NV and MM build-
plied to every observation in the data. Due to the smaller SCATS ings and 0.25/K for AC. A 30-day outdoor running mean air tem-
database compared to RP-884, the Griffiths method [43,44] rather perature was adopted as the outdoor climate independent variable
than linear regression was applied to estimate the neutral tem- in the adaptive regression model. The regression lines of neutral
perature with Griffiths Coefficient of G = 0.5, meaning that ther- operative temperature and the upper/lower lines of three classes
mal sensation votes were presumed to change at the rate of one acceptability limit for NV and MM buildings were presented. This
vote (on the 7 point scale) per 2 K change in operative tempera- RP-884 project was used as a precedent research design template,
ture. Running mean outdoor temperature was adopted as the out- notwithstanding the different method to derive neutral tempera-
door climate indicator. Thermal sensations of slightly cool, neutral ture (Griffiths values, but not the constant 0.5) and data aggre-
and slightly warm were deemed comfortable. Thermal acceptabil- gation. The final model agrees well with both ASHRAE 55 s and
ity decreases negligibly within +/- 2 K of the adaptive model’s neu- EN15251’s adaptive models despite some local features. The scope
tral temperature. At +/- 3 K from neutral the risk of discomfort in- of the IMAC model’s application includes both NV and MM build-
creases to about 25%, and increases to about 35% at +/- 4 K. The re- ings.
gression lines of optimal comfort temperature and the upper/lower
lines of three levels acceptability limit were presented in this stan-
3.6. Chinese GB/T 50785 comfort standard for free running buildings
dard and its revision. The 2015 prEN 16798 is a draft revision of
[23,24]
EN15251 with two notable changes: 1 K lowering of the lower limit
of optimal operative temperature, and an extended range of appli-
Nationwide right-here-right-now comfort field studies were con-
cability from 15–30 °C to 10–30 °C in the outdoor running mean
ducted in 14 major cities covering five climatic zones in China
temperature corresponding with lower limit of thermal comfort
as the data source of this standard. The diverse building typolo-
zone.
gies included public (office and educational) buildings as well as
multi-family residential buildings. The field studies covered sum-
3.4. Dutch ISSO 74-2004/2014 [30,33]
mer, winter, spring, and autumn seasons with over 28,0 0 0 sub-
jects were involved in the surveys. A graphical method was used
This standard was established based on adaptive thermal com-
to calculate the neutral temperature, and a running mean tem-
fort theory, but unlike the ASHRAE precedent, it can be applied to
perature was adopted as the outdoor climate indicator. The values
both unconditioned, mixed-mode, and conditioned spaces. There
of Li et al.’s [24] “adaptive coefficient” for different climate zones
were two application scenarios. The term Alpha-space refers to
in China were included in the GB/T 50785 standard. Upper/lower
“free-running situations in summer with operable windows and a
regression lines of two comfort categories were presented in a
non-strict clothing policy for the occupants”, while Beta-spaces are
graphical method. In this standard, both PMV and adaptive mod-
those “which primarily rely on centrally-controlled cooling in sum-
els are included but it no mention was made about the building
mer.” As with ASHRAE 55 the ISSO model was based on data from
types to which each method could be applied. The neutral temper-
the RP-884 database [21]. The ISSO algorithm for thermal neutral-
ature in GB/T 50785 has a discrepancy of 3 K when compared with
ity was the same as that in the RP-884 project as well. An “adapted
other standards [27], which suggests that further validation may be
running mean temperature” was used as the outdoor climate met-
appropriate.
ric. There were three acceptability classes: A: 90% acceptance; B:
80% acceptance; and C: 65% acceptance. The upper/lower regres-
sion lines of three classes for two types of spaces Alpha and Beta 3.7. Extensions in scope for adaptive comfort standards –
were presented in the 2004 version and 2014 version. In the 2014 mixed-mode ventilation
version there were 4 changes from the initial 2004 version. First
it considered specific internal spaces rather than whole buildings. At the time of the first adaptive comfort standard (ASHRAE 55-
Secondly it was based on the smaller, exclusively European SCATs 2004 [28]) there was insufficient data from mixed-mode build-
database rather than the much larger, global RP-884 database of ings to sustain any firm recommendations of how adaptive com-
comfort field studies. Third, temperature requirements were di- fort models worked in such buildings. The cognizant committee
vided into 4 instead of 3 performance classes. And fourth, it had a in charge of the standard, ASHRAE SSPC55, erred on the side of
different outdoor temperature calculation method (7-day outdoor caution and explicitly excluded such buildings from the scope of
temperature horizon in 2014 compared to the 3-day outdoor tem- the adaptive comfort model in Standard 55. Any building that
perature horizon or running mean outdoor temp). had cooling equipment installed was excluded, even if the cool-
ing capabilities were disabled and the building was operating un-
3.5. India model for adaptive comfort 34,45 der natural ventilation mode. Sixteen years later Parkinson et al.
(2020)’s analysis of thermal comfort data contained in the enlarged
The India Model of Adaptive Comfort [45] which forms part of ASHRAE Thermal Comfort Database II [46] provides a compelling
the Building Code of India [34] was based on a sample of 6330 empirical justification for enlarging the scope of applicability to
right-here-right-now thermal comfort questionnaires collected in 16 mixed-mode ventilation buildings. This brings us to a new inter-
buildings located in all of the major climatic divisions of India, pretation of mixed mode buildings as naturally ventilated build-
plus an additional 2002 long-term comfort questionnaires. The re- ings with supplemental cooling capability to temper indoor en-
search design was based directly on the ASHRAE RP-884 project vironmental conditions when external weather overwhelms the
[4] that formed the basis of the adaptive model in ASHRAE Stan- passive capabilities of the building to provide occupant adaptive
dard 55 [1,28,31,32]. Right-here-right-now thermal comfort ques- comfort.
6 R. de Dear, J. Xiong and J. Kim et al. / Energy & Buildings 214 (2020) 109893

4. Differences in adaptive comfort between building typologies indoor conditions falling outside the conventional adaptive com-
fort zone comfortable [52,55]. The Sydney residents’ comfort zone
Since the range and type of adaptive comfort behaviours such width corresponding to 80% acceptability was significantly wider
as clothing adjustments varies considerably between different (about 2 K) than that prescribed in ASHRAE 55 s adaptive comfort
building occupancy types, there seems reasonable grounds for ex- standard. The findings in this Australian study suggested that peo-
pecting different adaptive models in different building types. ple in their homes are more tolerant of greater indoor temperature
variations than those in office settings [52], supporting the hypoth-
4.1. Building typology esis posed earlier in this section. Williamson and Daniel [56] com-
plied a database from residential comfort field studies carried out
An underlying principle of adaptive comfort theory is that over the last several decades in Australia. It includes over 49,0 0 0
building occupants interact with their building to minimise ther- comfort votes from 295 residential buildings spanning across mul-
mal discomfort. Whenever adaptive opportunities exist, occupants tiple climate zones. The slope of the linear adaptive comfort equa-
tend to react in ways to achieve, or at least approximate their tion (which defines the dependence of the neutral temperature
desired conditions [5]. Baker and Standeven [47] emphasised that and outdoor prevailing temperature) was estimated to be 0.60
discomfort generally occurs when the indoor environmental condi- for naturally ventilated (NV) dwellings and 0.43 for mixed-mode
tion exceeds the limits of the adaptive opportunity available to the (MM) dwellings. The regression slope of 0.60 for NV dwellings in
occupant. If there are a number of small adjustments an occupant Williamson and Daniel study is remarkably similar to previously
can make, the cumulative effect can be significant [5,47]. A range reported 0.58 by Nicol [57]. Those regression coefficients derived
of opportunities allowing occupants to adjust or interact with the from residential studies are all significantly greater than that in
environment is central in achieving adaptive thermal comfort in the current office-based adaptive comfort standards (i.e. 0.31 in
buildings. Logically, therefore, the degree to which a building pro- ASHRAE 55 and 0.33 in EN15251). The residential sample’s gradi-
vides adaptive opportunities has significant implications for occu- ent almost twice as steep as the office counterpart signifies the
pants’ thermal adaptive processes. For instance, compared to resi- effectiveness of having a variety of adaptive opportunities.
dential settings where diverse adaptive responses can occur, office Studies carried out in China’s residential context further look
workers are given rather constrained adaptive opportunities pri- into other factors that can play a role in shaping residents’ thermal
marily due to the shared use of space and organisational culture adaptability. Yan et al. [54] collected data from 12 cities in four
in which formal business attire is typically mandated. different climate zones in China. The sample residents’ thermal
Therefore the type of a building, in terms of providing users adaptability varied according to climate region they lived in, with
with adaptive opportunities, can be a major contextual influence the gradient of adaptive equations varying from 0.12 to 0.55. Yan
on occupants’ thermal experience. However the thermal comfort et al.’s [54] results indicate that residents of warmer climates were
guidelines prescribed in the current international standards such more responsive to outdoor temperature variations than those in
as ASHRAE Standard 55 [1] and EN15251 [26] are supposedly ap- colder areas. Liu et al. [58] conducted a similar study but in the
plicable across diverse building typologies. Given that the adap- same climate zone of China across four different seasons. Their
tive comfort guidelines adopted in the current standards were de- data provided empirical evidence indicating that adaptive capac-
rived from field data collected almost exclusively from offices – i.e. ity varies seasonally. According to their analysis, the relationship
ASHRAE RP-884 [21] and SCATS [48], their applicability across di- between the neutral temperature and the prevailing outdoor tem-
verse building typologies needs further verification. A recent anal- perature was much more pronounced in swing seasons (regres-
ysis [49] performed on ASHRAE Global Thermal Comfort Database sion coefficient of 0.73∼0.76 in spring/autumn) compared to win-
2 [46] demonstrated that building typology, categorised as office, ter/summer (0.29∼0.34). Whilst the majority of studies on this
classroom, and multifamily housing, did exert a statistically mean- topic focused on the prevailing external conditions, an interesting
ingful effect on occupant thermal responses when all the relevant study in China’s severe cold northern region of Harbin showed that
environmental covariates were held constant. Other studies also people adapt to not only external conditions but also internal con-
demonstrated that thermal sensitivity of occupants to indoor tem- ditions, and Parkinson et al.’s [7] analysis of the ASHRAE Database
perature fluctuations (quantified as the Griffiths Constant) varied 2 came to the same conclusion. Long-term exposure to overheated
across different building typologies [50,51]. indoor conditions in winter due to centralised district heating sys-
Acknowledging that various building typologies (other than of- tem in Harbin (indoor temperature typically exceeds 24 °C) made
fice buildings) are underrepresented in the literature and the as- residents to adapt to warm indoor conditions in winter [59,60].
sociated international standards, there have been attempts among In other words, the residents’ adaptability to cold outside condi-
the thermal comfort research community to empirically verify tion was undermined after prolonged exposure to warm indoor
whether what is deemed as comfortable in office settings would be environments [61]. The finding has major implications on people’s
also comfortable in other contexts such as homes and schools. The comfort expectations and adaptive capacities, considering the fact
result of our literature review suggests that the current standards that people spend the overwhelming majority of their time in-
do not sufficiently reflect adaptive comfort of occupants in differ- doors.
ent building typologies – especially residential and educational set- Apart from the effect of indoor/outdoor climatic variations, as
tings, which are discussed below. discussed above, living habit or architectural characteristics also
Residential. It is often hypothesised that adaptive comfort the- seem to influence residents’ adaptive comfort. Recent studies in
ory fits better in the residential context than in the office con- China found that residents in traditional housing in rural areas
text because of a higher degree of adaptive opportunities afforded tended to be more tolerant to harsh external conditions, com-
by one’s own home [52]. Not to mention that householders can pared to those who resided in modern housing in urban areas
directly modify their environment, they have a greater degree of [62,63]. The difference between the rural and urban residents was
freedom to adjust their clothing insulation level (clo) than in for- attributed to their clothing habits, daily activity patterns and the
mal spaces like workplaces. Clo-values observed in residential set- typical design characteristics of traditional housing [62,63].
tings are generally wider than that would be normally expected in In residential settings a careful consideration for elderly resi-
the office context (e.g. [53,54]). dents seems necessary, given that the population of older people
Longitudinal field research carried out in Australian residential is increasing rapidly across the globe. Despite that recent climate
settings consistently showed that the householders often found chamber studies showed no effect of age in thermal responses
 R. de Dear, J. Xiong and J. Kim et al. / Energy & Buildings 214 (2020) 109893 7

[64,65], reinforcing Fanger’s earlier findings [2], some field stud- 50 s to 20–24.5 °C post 1981). These historic trends operating in
ies provided empirical evidence that the thermal requirements of parallel indicate that the comfort zone boundary became progres-
the elderly differ from younger people (e.g. [66–68]). Our litera- sively narrower over the last 70 years or so. The authors did not
ture review indicated a systematic discrepancy between what is speculate as to why ASHRAE’s comfort zone became more strin-
regarded as comfortable among the elderly and that suggested in gent through time, therefore at this stage it’s uncertain whether
the comfort standards (e.g. [69–71]). Further research to investigate or not such revisions reflected changing occupant comfort expec-
key aspects like housing design, adaptive behaviours and residents’ tations or the HVAC industry’s ambitions. Nevertheless, some em-
energy concerns is strongly encouraged by researchers as they may pirical evidence from China suggest that long-term thermal expe-
impact on comfort and health of older people, and their ability to riences can raise thermal comfort expectations [83]. An interesting
live independently [64,66,72]. implication from the China study [83] is that it’s easier and quicker
School classrooms. Seemingly diverse adaptive opportunities to enhance one’s thermal expectations but harder to lower them.
are available to students in a typical classroom scenario as class- About half of energy use in buildings is attributed to space
rooms are often equipped with various control measures such as heating and cooling in OECD countries [84]. While the use of en-
operable windows, fans or air conditioners. However, whether or ergy for air conditioning is growing faster than any other energy
not school children actually exercise such adaptive opportunities end uses in buildings [10], and the enormous amount of energy is
is another matter. For example, school children may not actively spent to guarantee the constant indoor conditions irrespective of
adjustment the classroom environment because it would require spatial and temporal variations of outdoor conditions, standards for
permission from their teacher [73]. Although older students (e.g. indoor comfort have become increasingly more stringent over time
secondary/high school students) seem to be more active in ex- [82]. Apart from soaring energy demand associated with the provi-
pressing their desire for thermal adaptations than younger (pri- sion of comfort indoors, a growing concern is that prolonged expo-
mary) students [74], still their indoor environmental conditions sure to tightly-controlled climate may diminish people’s adaptive
in classrooms probably reflect the preferences of their teachers capacity and their tolerance for conditions deviating from what
[75,76]. Moreover, behavioural adjustments they can personally ef- people are accustomed to [85]. Since its inclusion in 2004 version
fect (i.e. activity level and clothing insulation) are also somewhat of ASHRAE Standard 55, a more relaxed (and moving in sync with
constrained in schools because students are often required to wear prevailing outdoor climate) thermal comfort zone promoted by the
uniforms and sit quietly during lessons [77]. Therefore students adaptive model has encouraged more passive and adaptive design
in classroom tend to become passive recipients of environmental strategies. The adaptive comfort concept is central to addressing
conditions, rather than an active agent creating desired conditions, questions of how to enhance adaptive capacity in buildings and
which is central in the adaptive comfort theory. It is uncertain at how to nudge occupant attitudes and behaviours in relation to in-
this stage whether the unique contextual factors in classroom set- door climate.
tings exert influence on comfort status of school children. Accord-
ing to recent review works [78,79], general consensus in the litera- 6. Adaptive comfort and cognitive performance
ture is that (1) students’ desired thermal state is often outside the
comfort zone prescribed in the current standards; and (2) regard- One of the major barriers to full implementation of adaptive
less of stages of their education, students tend to feel comfortable comfort guidelines in the built environment, particularly office and
in cooler conditions than adults, all else being equal. educational buildings, relates to a lingering suspicion that temper-
atures warmer than 22 °C will have deleterious impacts on occu-
5. Shifting comfort zones pant cognitive performance. That suspicion was amplified by Sep-
panen and Fisks’ [86] highly influential meta-analysis of selected
The invention of air-conditioning made it possible to regu- papers in which an “inverted-U” function was statistically fitted
late indoor climate irrespective of outdoor conditions in modern to paired room temperature and relative performance data, show-
buildings. In commercial building sector, for example, maintain- ing a “single temperature optimum” at occurring at 22 °C. Since
ing indoor temperature of around 22 °C that varies little through the costs of human resources in office buildings are universally
time and space has become the norm in the Australian commer- acknowledged to be many times more valuable than the energy
cial buildings sector [80]. But what’s regarded as comfortable can costs maintaining indoor climate near the temperature optimum,
vary from one place (e.g. climate zone) or period (e.g. season) energy conservation in buildings as a way to reduce the emission
to another, and it also seems to be strongly influenced by past of greenhouse gases has been argued to be unaffordable [87].
habits, attitude, and adaptive behaviour, as discussed in our re- The proposition for an optimum performance temperature of
view of residential comfort publications above. For instance, the 22 °C has been highly contentious (e.g. [49,88–91]). Twenty two
contrast in indoor winter thermal preferences between Ameri- degrees falls well below any of the commonly agreed summer
cans and the English in the early 1900s when it was common ‘comfort zones, regardless of which climate zone, which country,
for the latter to complain about overheated rooms in the US, can which comfort model, or which standard one applies. Therefore the
be attributed to differences in space heating methods and indoor implication of the single-temperature hypothesis is that employ-
clothing customs [81]. A more scientific approach by Nicol and ers should deliberately make their office workforces feel cold in
Humphreys [5] demonstrated that the scatter of comfort temper- order to maximize performance. The main empirical evidence ar-
atures inside heated or cooled buildings became much narrower gued by Wyon and Wargocki [91] in support of 22 °C was a small
across a wide range of outdoor temperatures in 1990s, compared Danish laboratory study [92] in which a small sample (12) Dan-
to the 1970s. Yamamoto et al. [82] surveyed historical revisions of ish subjects wearing 0.9 clo (typical winter clothing level indoors)
ASHRAE Standard 55 with the aim understanding how the com- and sitting in a climate chamber and denied any thermally adap-
fort zone changed through time in the most international and tive opportunities such as operable windows for a couple of hours,
widely applied comfort standard. Their literature search went back reported feeling warm at 30 °C compared with the 22 °C condi-
to “Heating Ventilating Air Conditioning Guide 1941 by ASHVE (the tion in which they felt thermally neutral. Under such highly con-
name of ASHRAE pre 1954). Yamamoto et al. [82] indicated that the trived exposure conditions, it is not unsurprising that performance
summer comfort zone shifted down about 1 K cooler (from 23.9– decrements were observed at 30 °C compared with the 22 °C con-
27.2 °C in 1940–50 s to 23–26.1 °C after 1981). In contrast the win- dition. More comprehensive and far-ranging meta-analyses of the
ter comfort zone shifted up about 3 K (from 17.2–21.7 °C in 1940– temperature-performance literature by Zhang et al. [93] could find
8 R. de Dear, J. Xiong and J. Kim et al. / Energy & Buildings 214 (2020) 109893

no credible evidence of any systematic performance decrements cant financial support for this work that could have influenced its
across the broad range of temperatures defined by the adaptive outcome.
comfort zone which led them to support the “extended-U” model We confirm that the manuscript has been read and approved by
that pushes the thresholds of cognitive performance decrements all named authors and that there are no other persons who satis-
to the edges of the adaptive comfort zone. Indeed, the original fied the criteria for authorship but are not listed.
data used to fit Seppanen and Fisks’ [86] highly cited “inverted-U” We further confirm that the order of authors listed in the
model show no evidence of temperature effects between 28 °C and manuscript has been approved by all of us.
30 °C and the so-called 22 °C “optimum” is more likely an artefact We confirm that we have given due consideration to the pro-
of the statistical modelling procedures they chose to apply than tection of intellectual property associated with this work and that
any substantive signal in the empirical data. there are no impediments to publication, including the timing of
publication, with respect to intellectual property. In so doing we
7. Conclusions confirm that we have followed the regulations of our institutions
concerning intellectual property.
This thematic review of the adaptive thermal comfort research We further confirm that any aspect of the work covered in this
literature spanning the 21 years since the Brager and de Dear manuscript that has involved either experimental animals or hu-
[6] literature review of the topic has identified several areas of man subjects has been conducted with the ethical approval of all
progress and points to fruitful foci for future efforts. relevant bodies and that such approvals are acknowledged within
1. None of the published attempts at explaining the discrepancy the manuscript.
between predictions of heat balance comfort models and actual We understand that the Corresponding Author is the sole con-
observations inside adaptive comfort buildings has managed to tact for the Editorial process (including Editorial Manager and di-
fully “close the scissors” (Fig. 1). But there have been incremen- rect communications with the office). He/she is responsible for
tal contributions of new theoretical knowledge to the domain. communicating with the other authors about progress, submis-
We think the absence of an evidence-based parameterisation sions of revisions and final approval of proofs. We confirm that we
of the concept of comfort expectation seems to be the central have provided a current, correct email address which is accessi-
challenge, so future research efforts should be directed there. ble by the Corresponding Author and which has been configured
2. There seems to be broad consistency between the various reg- to accept email from (Richard.Dedear@sydney.edu.au)
ulatory documents and standards on adaptive comfort, which is
not surprising given the general consistency of methods used to Supplementary materials
generate their empirical bases and also their analytical strate-
gies, particularly in the ASHRAE, EN, and ISSO standards. The Supplementary material associated with this article can be
notable outlier seems to be the Chinese GB/T 50785 standard found, in the online version, at doi:10.1016/j.enbuild.2020.109893.
which was developed from a fundamentally different analytical
approach, based on the adaptive PMV model [18].
References
3. Building typology exerts a discernible effect on occupant ther-
mal responses and thermal sensitivity, mainly through varying [1] ANSI/ASHRAE, ANSI/ASHRAE Standard 55 - Thermal Environmental Conditions
adaptive opportunities across different building types and oc- For Human Occupancy, American Society of Heating, Refrigerating and Air–
cupancies. Residents “in the comfort of their own home” seem Conditioning Engineers, Atlanta, GA, USA, 2017.
[2] P.O. Fanger, Thermal comfort, Analysis and Applications in Environmental En-
generally to be more adaptable and tolerant of a wider range of
gineering, Danish Technical Press, Copenhagen, 1970.
indoor thermal exposures than their counterparts in office set- [3] J.F. Nicol, M.A. Humphreys, Thermal comfort as part of a self-regulating sys-
tings. School students tend to feel comfortable in indoor cli- tem, Build. Res. Pr. 1 (1973) 174–179, doi:10.1080/09613217308550237.
[4] R.J. de Dear, G.S. Brager, Developing an adaptive model of thermal comfort and
mates that are consistently cooler than adult thermal neutrali-
preference, ASHRAE Trans. 104 (1998) 1–18.
ties observed in office settings. [5] J.F. Nicol, M.A. Humphreys, Adaptive thermal comfort and sustainable ther-
4. Boundaries of the comfort zone have become progressively nar- mal standards for buildings, Energy Build. 34 (2002) 563–572, doi:10.1016/
rower over the past several decades. Long-term thermal expe- S0378-7788(02)0 0 0 06-3.
[6] G.S. Brager, R.J. De Dear, Thermal adaptation in the built environment: a lit-
riences can raise comfort expectations more readily than they erature review, Energy Build. 27 (1998) 83–96, doi:10.1016/s0378-7788(97)
can lower them. The adaptive comfort concept is central to 0 0 053-4.
addressing questions of how to enhance adaptive capacity in [7] T. Parkinson, R. de Dear, G. Brager, Nudging the adaptive thermal comfort
model, Energy Build. 206 (2020) 109559, doi:10.1016/j.enbuild.2019.109559.
buildings and how to nudge occupant attitudes and behaviours [8] H.Y. Levine, M.D. Ürge-Vorsatz, K. Blok, L. Geng, D. Harvey, S. Lang, G. Lever-
relating to indoor climate. more, A. Mongameli Mehlwana, S. Mirasgedis, A. Novikova, J. Rilling, H. Yoshi-
5. The weight of empirical evidence supports an extended-U noLevine, M.D. Ürge-Vorsatz, K. Blok, L. Geng, D. Harvey, S. Lang, G. Levermore,
A. Mongameli Mehlwan, Residential and commercial buildings, Clim. Chang.
model of temperature-performance effects. It posits a broad 2007 Mitig. 20 (2007) 389–437.
central plateau of temperatures across which there is no dis- [9] L.W. Davis, P.J. Gertler, Contribution of air conditioning adoption to future en-
cernible thermal effect on cognitive performance. We found no ergy use under global warming, Proc. Natl. Acad. Sci. U. S. A. 112 (2015) 5962–
5967, doi:10.1073/pnas.1423558112.
substantive, credible evidence to support the practice of over-
[10] IEA, The future of cooling, Opportunities For Energy Efficient Air Conditioning,
cooling buildings down to 22 °C in order to optimise the per- IEA Publications France, 2018, doi:10.1787/9789264301993-en.
formance of their occupants. The exception that makes the [11] G.E. Schiller, E.A. Arens, F.S. Bauman, C. Benton, M. Fountain, T. Doherty, A field
study of thermal environments and comfort in office buildings, ASHRAR Trans.
rule would be the cold climate zones, in which case the
94 (1988) 280–308.
term overcooling becomes inapplicable. The cognitive perfor- [12] R. de Dear, M. Fountain, Field experiments on occupant comfort and office
mance plateau is bounded by regions of progressive perfor- thermal environments in a hot-humid climate, ASHRAE Trans. 100 (1994)
mance deficits at the acceptability limits of the adaptive com- 457–475.
[13] G. Donnini, J. Molina, C. Martello, D.H.C. Lai, L.H. Kit, C.Y. Chang, M. Laflamme,
fort range. V.H. Nguyen, F. Haghighat, Field study of occupant comfort and office thermal
environments in a cold climate: Final Report on ASHRAE RP-821, 1996 Mon-
Declaration of Competing Interest treal, Quebec, Canada.
[14] K. Cena, R. de Dear, Field study of occupant comfort and office thermal envi-
ronments in a hot, arid climate, ASHRAE Trans. 105 (1999) 204–217.
We wish to confirm that there are no known conflicts of inter- [15] M.P. Deuble, R.J. de Dear, Green occupants for green buildings: the missing
est associated with this publication and there has been no signifi- link? Build. Environ. 56 (2012) 21–27, doi:10.1016/j.buildenv.2012.02.029.
 R. de Dear, J. Xiong and J. Kim et al. / Energy & Buildings 214 (2020) 109893 9

[16] P. Ole Fanger, J. Toftum, Extension of the PMV model to non-air-conditioned [43] M.A. Humphreys, H.B. Rijal, J.F. Nicol, Updating the adaptive relation between
buildings in warm climates, Energy Build. 34 (2002) 533–536, doi:10.1016/ climate and comfort indoors; new insights and an extended database, Build.
S0378-7788(02)0 0 0 03-8. Environ. 63 (2013) 40–55, doi:10.1016/j.buildenv.2013.01.024.
[17] J. Gao, Y. Wang, P. Wargocki, Comparative analysis of modified PMV models [44] I. Griffiths, Thermal comfort studies in buildings with passive solar features,
and set models to predict human thermal sensation in naturally ventilated field studies, Rep. to Comm. Eur. Community. 35 (1990).
buildings, Build. Environ. 92 (2015) 200–208, doi:10.1016/j.buildenv.2015.04. [45] S. Manu, Y. Shukla, R. Rawal, L.E. Thomas, R. de Dear, Field studies of ther-
030. mal comfort across multiple climate zones for the subcontinent: India model
[18] R. Yao, B. Li, J. Liu, A theoretical adaptive model of thermal comfort - Adap- for adaptive comfort (IMAC), Build. Environ. 98 (2016) 55–70, doi:10.1016/j.
tive Predicted mean vote (aPMV), Build. Environ. 44 (2009) 2089–2096, doi:10. buildenv.2015.12.019.
1016/j.buildenv.2009.02.014. [46] V.F. Ličina, T. Cheung, H. Zhang, R. De Dear, T. Parkinson, E. Arens, C. Chun,
[19] M. Schweiker, A. Wagner, A framework for an adaptive thermal heat balance S. Schiavon, M. Luo, G. Brager, Development of the ASHRAE global thermal
model (ATHB), Build. Environ. 94 (2015) 252–262, doi:10.1016/j.buildenv.2015. comfort database II, Build. Environ. 142 (2018) 502–512.
08.018. [47] N. Baker, M. Standeven, Thermal comfort for free-running buildings, Energy
[20] R. De Dear, G.S. Brager, The adaptive model of thermal comfort and energy Build. 23 (1996) 175–182, doi:10.1016/0378- 7788(95)00942- 6.
conservation in the built environment, Int. J. Biometeorol. 45 (2001) 100–108, [48] K.J. McCartney, J. Fergus Nicol, Developing an adaptive control algorithm
doi:10.10 07/s0 0484010 0 093. for Europe, Energy Build. 34 (2002) 623–635, doi:10.1016/S0378-7788(02)
[21] R.J. de Dear, Global database of thermal comfort field experiments, ASHRAE 0 0 013-0.
Trans. 104 (1998) 1141–1152. [49] F. Zhang, R. de Dear, Impacts of demographic, contextual and interaction ef-
[22] M. Isaac, D.P. van Vuuren, Modeling global residential sector energy demand fects on thermal sensation—Evidence from a global database, Build. Environ.
for heating and air conditioning in the context of climate change, Energy Pol- (2019) 162, doi:10.1016/j.buildenv.2019.106286.
icy 37 (2009) 507–521, doi:10.1016/j.enpol.2008.09.051. [50] R.F. Rupp, J. Kim, E. Ghisi, R. de Dear, Thermal sensitivity of occupants in dif-
[23] MOHURD, Evaluation standard for indoor thermal environment in civil build- ferent building typologies: the Griffiths constant is a variable, Energy Build.
ings (GB/T 50785-2012), Ministry of Housing and Urban-Rural Development 200 (2019) 11–20, doi:10.1016/j.enbuild.2019.07.048.
(MOHURD), Beijing, China, 2012. [51] J. Ryu, J. Kim, W. Hong, R. de Dear, Defining the thermal sensitivity (Griffiths
[24] B. Li, R. Yao, Q. Wang, Y. Pan, An introduction to the Chinese evaluation constant) of building occupants in the Korean residential context, Energy Build.
standard for the indoor thermal environment, Energy Build. 82 (2014) 27–36, 208 (2020) 109648, doi:10.1016/j.enbuild.2019.109648.
doi:10.1016/j.enbuild.2014.06.032. [52] R. de Dear, J. Kim, T. Parkinson, Residential adaptive comfort in a humid
[25] A.P. Gagge, A.P. Fobelets, L.G. Berglund, A standard predictive index of hu- subtropical climate—Sydney Australia, Energy Build. 158 (2018) 1296–1305,
man reponse to thermal enviroment, Am. Soc. Heating, Refrig. Air-Conditioning doi:10.1016/j.enbuild.2017.11.028.
Eng. 92 (1986) 709–731 https://www.aivc.org/sites/default/files/airbase_2522. [53] J. Kim, R. de Dear, T. Parkinson, C. Candido, Understanding patterns of adap-
pdf%0Ahttp://oceanrep.geomar.de/42985/. tive comfort behaviour in the Sydney mixed-mode residential context, Energy
[26] CEN Standard EN15251, Indoor environmental input parameters for design and Build. 141 (2017) 274–283, doi:10.1016/j.enbuild.2017.02.061.
assessment of energy performance of buildings addressing indoor air quality, [54] H. Yan, Y. Mao, L. Yang, Thermal adaptive models in the residential buildings
thermal environment, lighting and acoustics, European Committee for Stan- in different climate zones of Eastern China, Energy Build. 141 (2017) 28–38,
dardization, Brussels, Belgium, 2007. doi:10.1016/j.enbuild.2017.02.016.
[27] S. Carlucci, L. Bai, R. de Dear, L. Yang, Review of adaptive thermal com- [55] L. Daniel, T. Williamson, V. Soebarto, D. Chen, Learning from thermal maver-
fort models in built environmental regulatory documents, Build. Environ. 137 icks in Australia: comfort studies in Melbourne and Darwin, Archit. Sci. Rev.
(2018) 73–89, doi:10.1016/j.buildenv.2018.03.053. 58 (2015) 57–66, doi:10.1080/0 0 038628.2014.976537.
[28] ANSI/ASHRAE, ANSI/ASHRAE Standard 55 - thermal environmental Conditions [56] T. Williamson, L. Daniel, Development of a model of adaptive thermal comfort
For Human Occupancy, American Society of Heating, Refrigerating and Air- and preference for housing in Australia, in: 52nd Int. Conf. Archit. Sci. Assoc.
Conditioning Engineers (ASHRAE), Atlanta, GA, USA, 2004. 27 Nov–1 Dec 2018, Melbourne, Aust., 2018, pp. 3–10.
[29] Technical Committee CEN/TC 156, Energy performance of buildings–Part 1: in- [57] F. Nicol, Temperature and adaptive comfort in heated, cooled and free-
door environmental input parameters for design and assessment of energy running dwellings, Build. Res. Inf. 45 (2017) 730–744, doi:10.1080/09613218.
performance of buildings addressing indoor air quality, thermal environment, 2017.1283922.
lighting and acoustics - Module M1–6 (EN 16798-1), (2015). [58] H. Liu, Y. Wu, B. Li, Y. Cheng, R. Yao, Seasonal variation of thermal sensations in
[30] A.C. Van Der Linden, A.C. Boerstra, A.K. Raue, S.R. Kurvers, R.J. De Dear, residential buildings in the hot summer and cold winter zone of China, Energy
Adaptive temperature limits: a new guideline in the Netherlands: a new ap- Build. 140 (2017) 9–18, doi:10.1016/j.enbuild.2017.01.066.
proach for the assessment of building performance with respect to thermal [59] Y. Ji, Z. Wang, Thermal adaptations and logistic regression analysis of ther-
indoor climate, Energy Build. 38 (2006) 8–17, doi:10.1016/j.enbuild.2005.02. mal comfort in severe cold area based on two case studies, Energy Build. 205
008. (2019) 109560, doi:10.1016/j.enbuild.2019.109560.
[31] ANSI/ASHRAE, ANSI/ASHRAE Standard 55 -Thermal Environmental Conditions [60] Z. Wang, Y. Ji, J. Ren, Thermal adaptation in overheated residential buildings
For Human Occupancy, American Society of Heating, Refrigerating and Air– in severe cold area in China, Energy Build. 146 (2017) 322–332, doi:10.1016/j.
Conditioning Engineers, 2010. enbuild.2017.04.053.
[32] ANSI/ASHRAE, ANSI/ASHRAE Standard 55 - thermal environmental Conditions [61] H. Ning, Z. Wang, Y. Ji, Thermal history and adaptation: does a long-term in-
For Human Occupancy, American Society of Heating, Refrigerating and Air- door thermal exposure impact human thermal adaptability? Appl. Energy 183
Conditioning Engineers, Atlanta, GA, USA, 2013. (2016) 22–30, doi:10.1016/j.apenergy.2016.08.157.
[33] A.C. Boerstra, J. Van Hoof, A.M. Van Weele, A new hybrid thermal comfort [62] Y. Xiong, J. Liu, J. Kim, Understanding differences in thermal comfort between
guideline for the Netherlands: background and development, Archit. Sci. Rev. urban and rural residents in hot summer and cold winter climate, Build. Envi-
58 (2015) 24–34, doi:10.1080/0 0 038628.2014.971702. ron. 165 (2019) 106393, doi:10.1016/j.buildenv.2019.106393.
[34] Bureau of Indian Standards, National Building Code, clause 6.2, part 8 building [63] C. Xu, S. Li, X. Zhang, S. Shao, Thermal comfort and thermal adaptive be-
services, section 3, air conditioning, heating and mechanical ventilation, page haviours in traditional dwellings: a case study in Nanjing, China, Build. Env-
19 of volume 2, 2016 Bureau of Indian Standard. iron. 142 (2018) 153–170.
[35] S. Kumar, M.K. Singh, Field investigation on occupant’s thermal comfort and [64] V. Soebarto, H. Zhang, S. Schiavon, A thermal comfort environmental chamber
preferences in naturally ventilated multi-storey hostel buildings over two sea- study of older and younger people, Build. Environ. 155 (2019) 1–14, doi:10.
sons in India, Build. Environ. 163 (2019) 106309. 1016/j.buildenv.2019.03.032.
[36] B. Moujalled, R. Cantin, G. Guarracino, Comparison of thermal comfort al- [65] J. Xiong, T. Ma, Z. Lian, R. de Dear, Perceptual and physiological responses of
gorithms in naturally ventilated office buildings, Energy Build. 40 (2008) elderly subjects to moderate temperatures, Build. Environ. 156 (2019) 117–122,
2215–2223. doi:10.1016/j.buildenv.2019.04.012.
[37] Y. Zhai, A. Honnekeri, M. Pigman, M. Fountain, H. Zhang, X. Zhou, E. Arens, Use [66] F. Tartarini, P. Cooper, R. Fleming, M. Batterham, Indoor air temperature and
of adaptive control and its effects on human comfort in a naturally ventilated agitation of nursing home residents with dementia, Am. J. Alzheimers. Dis.
office in Alameda, California, Energy Build. 203 (2019) 109435. Other Demen. 32 (2017) 272–281, doi:10.1177/1533317517704898.
[38] J. Kim, F. Tartarini, T. Parkinson, P. Cooper, R. de Dear, Thermal comfort in a [67] J. Yang, I. Nam, J.R. Sohn, The influence of seasonal characteristics in elderly
mixed-mode building: are occupants more adaptive? Energy Build. 203 (2019) thermal comfort in Korea, Energy Build. 128 (2016) 583–591, doi:10.1016/j.
109436 %@ 0378–7788, doi:10.1016/j.enbuild.2019.109436. enbuild.2016.07.037.
[39] M. Luo, B. Cao, J. Damiens, B. Lin, Y. Zhu, Evaluating thermal comfort in mixed– [68] L.T. Wong, K.N.K. Fong, K.W. Mui, W.W.Y. Wong, L.W. Lee, A field survey of the
mode buildings: a field study in a subtropical climate, Build. Environ. 88 (2015) expected desirable thermal environment for older people, Indoor Built. Envi-
46–54. ron. 18 (2009) 336–345, doi:10.1177/1420326X09337044.
[40] R.F. Rupp, E. Ghisi, Predicting thermal comfort in office buildings in a Brazilian [69] J. van Hoof, L. Schellen, V. Soebarto, J.K.W. Wong, J.K. Kazak, Ten questions
temperate and humid climate, Energy Build. 144 (2017) 152–166. concerning thermal comfort and ageing, Build. Environ. 120 (2017) 123–133,
[41] ISO, ISO 7730: Ergonomics of the Thermal Environment Analytical Determina- doi:10.1016/j.buildenv.2017.05.008.
tion and Interpretation of Thermal Comfort Using Calculation of the PMV and [70] R.L. Hwang, C.P. Chen, Field study on behaviors and adaptation of elderly peo-
PPD Indices and Local Thermal Comfort Criteria, International Organization for ple and their thermal comfort requirements in residential environments, In-
Standardization, Geneva, 2005, doi:10.1016/j.soildyn.2004.11.005. door Air 20 (2010) 235–245, doi:10.1111/j.1600-0668.2010.00649.x.
[42] B.W. Olesen, International standards and the ergonomics of the thermal [71] Z. Wang, H. Yu, Y. Jiao, Q. Wei, X. Chu, A field study of thermal sensation
environment, Appl. Ergon. 26 (1995) 293–302, doi:10.1016/0 0 03-6870(95) and neutrality in free-running aged-care homes in Shanghai, Energy Build. 158
0 0 033-9. (2018) 1523–1532, doi:10.1016/j.enbuild.2017.11.050.
10 R. de Dear, J. Xiong and J. Kim et al. / Energy & Buildings 214 (2020) 109893

[72] J. van Hoof, H.S.M. Kort, M.S.H. Duijnstee, P.G.S. Rutten, J.L.M. Hensen, The in- [83] M. Luo, R. de Dear, W. Ji, C. Bin, B. Lin, Q. Ouyang, Y. Zhu, The dynamics of
door environment and the integrated design of homes for older people with thermal comfort expectations: the problem, challenge and impication, Build.
dementia, Build. Environ. 45 (2010) 1244–1261, doi:10.1016/j.buildenv.2009.11. Environ. 95 (2016) 322–329, doi:10.1016/j.buildenv.2015.07.015.
008. [84] IEA, Energy Technology Perspectives 2016: Towards Sustainable Urban Energy
[73] N. Bernardi, D.C.C.K. Kowaltowski, Environmental comfort in school buildings: Systems, OECD, 2016, doi:10.1787/energy_tech- 2014- en.
a case study of awareness and participation of users, Environ. Behav. 38 (2006) [85] R. Hitchings, Coping with the immediate experience of climate: regional varia-
155–172, doi:10.1177/0013916505275307. tions and indoor trajectories, Wiley Interdiscip. Rev. Clim. Chang. 2 (2011) 170–
[74] J. Kim, R. de Dear, Thermal comfort expectations and adaptive behavioural 184, doi:10.1002/wcc.106.
characteristics of primary and secondary school students, Build. Environ. 127 [86] O.A. Seppänen, W. Fisk, Some quantitative relations between indoor envi-
(2018) 13–22, doi:10.1016/j.buildenv.2017.10.031. ronmental quality and work performance or health, Hvac &R Res. 12 (2006)
[75] S.M. Kevan, J.D. Howes, Climatic conditions in classrooms, Educ. Rev. 32 (1980) 957–973.
281–292, doi:10.1080/0 01319180 0320305. [87] P. Wargocki, D.P. Wyon, Ten questions concerning thermal and indoor air qual-
[76] D. Teli, M.F. Jentsch, P.A.B. James, Naturally ventilated classrooms: an as- ity effects on the performance of office work and schoolwork, Build. Environ.
sessment of existing comfort models for predicting the thermal sensation 112 (2017) 359–366.
and preference of primary school children, Energy Build. 53 (2012) 166–182, [88] P.A. Hancock, H.C. Ganey, From the inverted-U to the extended-U: the evolu-
doi:10.1016/j.enbuild.2012.06.022. tion of a law of psychology., J. Hum. Perform. Extrem. Environ 7 (1) (2003)
[77] R.L. Hwang, T.P. Lin, C.P. Chen, N.J. Kuo, Investigating the adaptive model of 5–14.
thermal comfort for naturally ventilated school buildings in Taiwan, Int. J. [89] P.A. Hancock, A dynamic model of stress and sustained attention, Hum. Factors
Biometeorol. 53 (2009) 189–200, doi:10.1007/s00484- 008- 0203- 2. 31 (1989) 519–537.
[78] M.K. Singh, R. Ooka, H.B. Rijal, S. Kumar, A. Kumar, S. Mahapatra, Progress in [90] R. de Dear, E.A. Arens, C. Candido, G. Brager, H. Zhang, J. Toftum, Y. Zhu,
thermal comfort studies in classrooms over last 50 years and way forward, S. Kurvers, J. Leyten, C. Sekhar, Indoor temperatures for optimum thermal com-
Energy Build. 188–189 (2019) 149–174, doi:10.1016/j.enbuild.2019.01.051. fort and human performance: reply to the letter by Wyon and Wargocki, In-
[79] Z.S. Zomorodian, M. Tahsildoost, M. Hafezi, Thermal comfort in educational door Air 24 (2014) 554–555.
buildings: a review article, Renew. Sustain. Energy Rev. 59 (2016) 895–906, [91] D.P. Wyon, P. Wargocki, The adaptive thermal comfort model may not always
doi:10.1016/j.rser.2016.01.033. predict thermal effects on performance, Indoor Air 24 (2014) 552–553.
[80] A.C. Roussac, J. Steinfeld, R. De Dear, A preliminary evaluation of two strategies [92] L. Lan, P. Wargocki, D.P. Wyon, Z. Lian, Effects of thermal discomfort in an of-
for raising indoor air temperature setpoints in office buildings, Archit. Sci. Rev. fice on perceived air quality, SBS symptoms, physiological responses, and hu-
54 (2011) 148–156, doi:10.1080/0 0 038628.2011.582390. man performance, Indoor Air 21 (2011) 376–390, doi:10.1111/j.1600-0668.2011.
[81] W.B. Meyer, Why indoor climates change: a case study, Clim. Change 55 (2002) 00714.x.
395–407, doi:10.1023/A:1020586017450. [93] F. Zhang, R. de Dear, P. Hancock, Effects of moderate thermal environments
[82] M. Yamamoto, N. Nishihara, G. Kawaguchi, J. Harigaya, S. Tanabe, Comparison on cognitive performance: a multidisciplinary review, Appl. Energy. 236 (2019)
of the transition of thermal environment in office between Japan and the US, 760–777, doi:10.1016/j.apenergy.2018.12.005.
Clima 2010 - REHVA World Congr., 2010.