Building and Environment 71 (2014) 223e232

Contents lists available at ScienceDirect

Building and Environment

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

Thermal comfort in undergraduate laboratories d A field study in

Kharagpur, India
Asit Kumar Mishra*, Maddali Ramgopal
Department of Mechanical Engineering, Indian Institute of Technology Kharagpur, Kharagpur 721302, West Bengal, India

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

Article history: A thermal comfort field survey is carried out inside a naturally ventilated laboratory in the tropical
Received 3 June 2013 climatic region of India. The building chosen is used for courses in an undergraduate engineering cur-
Received in revised form riculum. We aimed at assessing how the occupants perceive their thermal environment in a free running
6 October 2013
building while carrying out their normally scheduled tasks. A total of 121 acclimatized subjects were
Accepted 9 October 2013
interviewed and 338 responses were collected during the months of spring semester. Survey results
show a strong correlation between indoor comfort conditions and outdoor temperature. Occupants show
Keywords:
adaptability across a comfort zone that is well beyond recommendations of rational models. Overall, 78%
Thermal comfort
Field study
of the responses found their thermal environment to be acceptable. Based on indoor temperature ob-
Adaptive thermal comfort servations, we calculate comfort temperatures for the subjects using Griffiths’ method. The comfort
Tropical climate temperature values are then related with prevailing mean outdoor air temperature to give an adaptive
Laboratory class comfort equation. Predictions from our equation show satisfactory to good agreement with the pre-
dictions from similar equations in comfort standards.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction mind as a tempting and quick fix solution. Projections show India

Activities involved in teaching-learning process are known to be
affected by the immediate thermal environment [1e3]. Thermal
environment’s effect on mental performance is necessarily associ-
ated with occupants’ subjective evaluation of thermal comfort [1,4].
So thermal comfort of occupants in classrooms and other rooms
related to educational facilities is of vital concern.
In India, the demographic eligible for attending institutes of
higher education, i.e. 18e23 year old, makes up 11e12% of the na-
tional population [5]. Actual enrolment is nowhere near 100% of
this age group, yet, at the beginning of the 2010e11 academic year,
India had 14.6 million students enrolled in its 544 universities (or
equivalent institutes) and 31,324 colleges [6]. Add to this the 0.7
million faculty members and that number beats the combined
population of Finland, Ireland, and Norway! Further, data shows
that both the number of institutions and enrolments in them are
growing at more than 5% annually [5].
For providing thermal comfort in all these classrooms spread
across a country having mostly tropical weather, AC can come to
* Corresponding author. Tel.: þ91 8768073875.
E-mail addresses: asit.mishra@mech.iitkgp.ernet.in, writeto.asit@gmail.com
(A.K. Mishra), ramg@mech.iitkgp.ernet.in (M. Ramgopal).
beating both USA and China by 2055 to become the world leader in
energy consumption for AC [7]. But India’s current energy scenario
is hardly reassuring and forebodes a grim future. During the
recession of 2009, while overall world energy demand decreased
by 1%, India’s total final energy consumption increased by about
10% [8]. Also in 2009, India became world’s third largest energy
consumer but with a 26% energy deficit [8]. More than a third of
India’s electricity consumption is in residential and commercial
sectors and the electricity demand of these sectors has been
growing at more than 8% [9]. In both of these sectors, a significant
portion of the energy demand is from buildings while more than
half the buildings India will have in 2030 are yet to be built [10]. All
these statistics and estimations prove a great urgency for energy
saving measures in Indian buildings.
While the National Building Code of India (NBC) [11] is ‘the’
guideline for building construction in India, until 2007, there were
no codes or requirements to be satisfied for energy efficiency in
new buildings. When it comes to AC in classrooms, the NBC pre-
scribes a rather narrow comfort zone of 23e26
C, 50e60% RH. In
2007, the Bureau of Energy Efficiency (a body under Ministry of
Power, Government of India) brought out the Energy Conservation
Building Code (ECBC) for achieving energy efficiency in large
commercial buildings. But ECBC still takes NBC as the reference
standard for both AC requirements and ventilation requirements in

0360-1323/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.buildenv.2013.10.006
224 A.K. Mishra, M. Ramgopal / Building and Environment 71 (2014) 223e232

Nomenclature TSI tropical summer index

TSV thermal sensation vote (individual’s)
TSVS thermal sensation votes in mid three points of scale
Abbreviations WBT wet bulb temperature
AC air-conditioning
ACE adaptive comfort equation Symbols
APD actual percentage dissatisfied pa absolute pressure of water vapour in air
DBT dry bulb temperature ta air temperature
MRT mean radiant temperature tc comfort temperature
MTCV mean thermal comfort vote tg globe temperature
MTSV mean thermal sensation vote tm daily mean temperature
NV naturally ventilated tmm monthly mean temperature
PMOAT prevailing mean outdoor air temperature tmrt mean radiant temperature
PMV predicted mean vote to outdoor temperature index
PPD predicted percentage dissatisfied top operative temperature
PS percentage of acceptable votes/percentage satisfied trmt running mean temperature
RH relative humidity va air velocity
RMT running mean temperature Dtc absolute difference between comfort temperature
TCV thermal comfort vote (individual’s) predictions from different models
TCVS thermal comfort votes in mid three points of scale

NV spaces [12]. For NV buildings, the ECBC User Guide does the study by Pellegrino et al., no other studies have been reported
mention the adaptive model given in ASHRAE Standard 55 as an from Indian classrooms. Considering the importance of comfort
option for determining comfort [12]. requirements in classrooms and the fast growing number of in-
Adaptive thermal comfort standards have shown great promise stitutes for higher education in India, classroom thermal comfort
in reducing energy consumption of buildings, without sacrificing needs serious attention. Our study was aimed at finding the level of
occupant comfort. Several studies done for NV classroom occupants thermal comfort and acceptance amongst students of an under-
in tropical regions (Köppen climate classification type A) have graduate course during their regular semester classes. We also
borne out significant levels of adaptation amongst occupants [13e aimed to ascertain if the adaptive comfort model can be applicable
16]. These studies have come up with comfort zones between 22 at sustained metabolic rates that are slightly over ASHRAE Standard
and 31
C and neutral temperatures in the range of 26.5e29
C. 55-2010 current specification of 1e1.3 met [24]. So we decided to
These findings are an obvious testament to the fact that comfort is survey the conditions during a laboratory class.
not exclusively found within the confines of a conditioned building. Our main objectives for this study were:
Field studies in India also attest to remarkable levels of occupant
adaptation both to the local climate and to NV buildings. In a very e Assessing occupants’ perception of thermal comfort in a free
early comfort study done in Calcutta, Rao reported a comfort running building during their regular classes.
temperature of 25.8
C DBT [17]. Sharma and Ali [18] did a very e Determining the range of temperature found most suitable by
comprehensive study on comfort, developing the sole empirical these occupants.
index of thermal comfort based on Indian subjects and Indian e Determining if adaptive comfort principle can be applied at
conditions d TSI. They found that a TSI value of 27.5
C elicited slightly higher than sedentary metabolic rates.
maximum percentage of comfortable votes. Over the past decade, e Establishing a path and framework for future studies.
there has been a significant rise in the number of thermal comfort
studies being done across India. A summary of some such studies in
presented in Table 1.
2. Study methodology
Pellegrino et al. have also reported a short duration comfort
study done in NV classrooms of two universities in Kolkata [23].
2.1. Location, building, and subjects
This study was over a couple of days and during the summer month
of May. As would be expected from the adaptive hypothesis of
2.1.1. Study location and duration
comfort, the neutral temperature obtained in this survey came out
Kharagpur e located in the eastern Indian state of West Bengal
to be rather high (30.9
C). To the best of our knowledge, apart from
e has a climate classification of Tropical savannah type (Aw) under
the Köppen system. Summers are hot and humid, monsoon months
Table 1 are slightly cooler than summer but more humid, and the winter
Summary of recent Indian studies. months are cool. Maximum outdoor temperature during summer
Location Building type Conditioning Regression neutral
days often crosses 40
C. Temperatures during winter nights have
temperature also been known to fall below 10
C. Historical monthly mean
temperatures over past ten years (collected from the in-campus
Hyderabad [19] Residences NV 29.23
C
Chennai [20] Railway waiting lounge NV 31.93
C weather station manned and operated by the Department of
Hyderabad [21] Offices NV 26.4
C Physics and Meteorology of Indian Institute of Technology (IIT)
AC 26.3
C Kharagpur) show that January is the coolest month while May is
Chennai [21] Offices NV 26.1
C the hottest.
AC 27.5
C
Jaipur [22] Student dormitories NV 30.15
C
The comfort survey was done in a laboratory classroom of IIT
Kharagpur which is a premier institutes of technical education in
 A.K. Mishra, M. Ramgopal / Building and Environment 71 (2014) 223e232 225

India. The Institutes of Technology Act, 1961, declares IIT Kharagpur

to be an institute of national importance. The institute campus is
spread over 2100 green acres and houses nearly 10,000 students
along with around 2000 faculty and supporting staff. The campus
greenery has an ameliorating effect on the harsh summer heat.
In IIT Kharagpur, spring semester starts from the first week of
January and ends in April. Surveys took place on 12 separate days
from January through mid April 2013. Conducting the survey in the
spring semester implies that we can check occupant perception
through both the coldest and the hottest months during which
regular classes take place.

2.1.2. Building
Fig. 1 shows a broad view of the campus as well as a close up of
the building in which the survey took place (as collected from
Google MapsÓ). The figure gives an idea regarding the surrounding
tree line as well as the near by buildings, both of which protect our
building of concern from most of direct solar radiation. The building
has an East-West orientation with windows placed only along the
North and South walls.
The building’s construction is pillared with brick and mortar
filling. Windows are single glazed, have mild-steel frames, and
cover about 23% of total wall area. The windows also have contin-
uous overhangs, roughly 1 m wide, ensuring protection from rain
and direct sunlight. Average thickness of the walls is 40 cm while
ceiling height is slightly over 7 m. Area of occupancy during class
hours is served by 7 ceiling fans with speed controllers. The fans are
suspended using long rods from the ceiling to ensure that they do
not just churn hot air near the ceiling into the comfort zone (Fig. 2).
Overall, the said building has a strong basis in its design to function
under free running conditions.
Fig. 2. An example of a fan suspended much below ceiling level.
2.1.3. Subjects
The survey was designed to follow the schedule of a sixth se- Students had been divided into three groups and each group had
mester laboratory course that is part of an eight semester curricu- to complete their experiments in this particular laboratory over a
lum. A total of 121 students who were registered for the said course period of four weeks. Each group was further divided into four
were surveyed. There were only five female students in this entire subgroups so that in every class, there were roughly 10 students
group. The students were all aged between 19 and 21 years. To performing a particular experiment. After the four weeks of a
maintain this homogeneous age group, it was decided not to survey particular group were complete, the next group would come in to
the instructors. All subjects were Indians, though they came from start their term in the lab. Thus, the course’s schedule in a way
different provinces, and had been staying in Kharagpur for at least helped us to avoid any kind of survey fatigue setting in or the risk of
five semesters. They were thus well acclimatized to the local subjects getting too familiar with the process. As a further measure,
weather. During the survey, all subjects were assumed to be in good we decided to survey only three of the four subgroups present on
health since any ailment gets a student automatically excused from any day. We cyclically altered between subgroups 1-2-3, followed
attending the class. by subgroups 2-3-4 and so on.

Fig. 1. Views of the building (marked by ‘X’) (a) Area of campus around the building (b) Close up view of the laboratory along with surrounding buildings and tree line.
226 A.K. Mishra, M. Ramgopal / Building and Environment 71 (2014) 223e232
Every time a new group of students came in, they were briefly
introduced to the purpose, duration, and process of the survey.
They were assured that this was not some kind of pop quiz and they
should feel free to give honest responses. Since people might often
link survey questionnaires on thermal comfort with chance of
change in conditioning type, we also made it clear to the students
that this was not some kind of referendum on the requirement of
AC in laboratories.
2.2. Data collection
During the field survey, physical measurements of environ-
mental parameters were taken while the paper based subjective
questionnaires were being filled by the occupants. Since we aimed
at recording reactions to typical conditions, every effort was made
not to interfere with the class schedule and thermal environment.
Survey activities around the three subgroups took a total of 35e
40 min.
2.2.1. Objective data
Measurements were taken for dry bulb and wet bulb tempera-
tures, globe temperature, and air velocity. A black globe ther-
mometer was constructed by placing an alcohol thermometer at
the centre of a black painted plastic ball of 70 mm diameter. Dry
and wet bulb temperatures were measured with help of a sling
psychrometer. The globe thermometer had a range 10e110
C and
resolution 1
C. The two thermometers in the psychrometer had a
range 0e120
F, resolution 1
F. All three thermometers were
calibrated against a reference thermometer from ZealÓ
(range 10e50
C, resolution 0.5
C) over the range of 10e40
C. Air
velocity was measured with a digital, vane-type anemometer
(Lutron AM-4201, range 0.1e30.0 m/s).
Students worked in closely clustered groups. So DBT, WBT, and
air velocity were measured at five points around the group and an
average value of these five readings was taken to represent the
thermal condition for a group. Globe temperature readings took
10 min to stabilise and hence we took a single reading of globe
temperature per group. As all subjects were in standing position
during the survey duration, measurements were taken at shoulder
level of the occupants. Measurements were taken within 1 m dis-
tance from the occupants while taking care to keep out of the way
of the students and instructors. Just before starting the survey,
outdoors DBT and WBT in immediate vicinity of the building were
also determined as an arithmetic average of measurements made at
five locations near the building.
Mean radiant temperature (tmrt) was calculated from globe
temperature (tg), air temperature (ta), and air velocity (va) using
Equation (1) [25].


 4 1:1  108 v0:6
a
  1=4
tmrt ¼ tg þ 273 þ  tg  ta  273 (1)
3 D0:4
Here 3 is the emissivity of the globe surface (taken as 0.95) and D is
the globe diameter. Operative temperature (top) was calculated as a
Table 2
Survey questionnaire options.
Thermal sensation Thermal comfort Air velocity sensation
Cold(3) Much too cool(3) Still(1)
Cool(2) Too cool(2) Just right(0)
Slightly cool(1) OK(cool) (1) Breezy(1)
Neutral(0) OK (Just right) (0)
Slightly warm(1) OK (warm) (1)
Warm(2) Too warm(2)
Hot(3) Much too warm(3)
weighted mean of the air temperature and MRT with weighting
factors of convective (hc) and radiative (hr) heat transfer coefficient
respectively. We took a constant value of 4.7 W/m2
C for hr while hc
was calculated from the correlation for standing persons in ASHRAE
Handbook [25].
Values for RH and absolute pressure of water vapour in air (pa)
were obtained from the DBT and WBT values using a humidity
calculator from Vaisala [26]. All statistical analysis were carried out
using the R statistical computing package (version 2.15.2) [27].
2.2.2. Subjective questionnaire
As mentioned in Section 2.1.3, the subjects had been briefed
regarding the nature of the survey and the questionnaires. Any
queries individual students had during the survey were also
resolved. The students were from different regions of India and it
would have been difficult to decide on any one particular regional
language that all students would have been comfortable with. But
the medium of instruction in all their classes is English. Due to their
common comfort level with English, English was determined to be
a suitable compromise. The questionnaire options are given in
Table 2.
Numbers in brackets besides the options are numerical equiv-
alents used for statistical analysis. The questionnaire included the
standard seven point ASHRAE thermal sensation scale, comfort
vote on Bedford scale, thermal preference on the McIntyre scale, a
direct acceptability question, a three point air velocity sensation
question, and a five point humidity sensation question. Occupant
satisfaction percentage was ascertained on basis of acceptability
votes. This involves the inherent assumption that a person voting
acceptable on the survey questionnaire is satisfied with his/her
thermal conditions. Percentage of people voting unacceptable was
taken to be the APD.
As the survey was being conducted in between regular opera-
tions of a class, it was of paramount concern to keep the survey
questionnaire short and compact. Students took 1e5 min to fill up
and return the questionnaire sheet. The humidity and air velocity
sensation questions had fewer options to save time and to not have
the subjects making subtle distinctions in what is a second lan-
guage to them. The options for the Bedford rating scale were the
same as those used by Humphreys [28]. These options had been
used by Humphreys in a study done in Pakistan. Due to the com-
mon heritage of English shared by India and Pakistan, the word ‘OK’
is also very well understood and carries a clear sense in India.
2.2.3. Assessing clothing levels and metabolic rates
Adding an extra section of clothing and activity check-list would
have been time consuming and distracting to the regular class ac-
tivity. Clothing level was assessed using an approach similar to that
used by Morgan and de Dear for a study done “unobtrusively” on
shoppers at a Sydney mall [29]. Students mostly attired in western
style outfits: combinations of shirts/t-shirts, trousers/jeans. Jeans is
a common and frequent apparel amongst students across all sea-
sons and was assumed to have clothing insulation of thick trousers.
Five common ensembles were identified (shown in Table 3) and
Humidity sensation Thermal preference Acceptability
Too dry(2) Cooler(1) Acceptable(0)
Slightly dry(1) No change(0) Not acceptable(1)
Just right(0) Warmer(1)
Slightly humid(1)
Too humid(2)
 A.K. Mishra, M. Ramgopal / Building and Environment 71 (2014) 223e232 227

Table 3
Clothing ensembles.
Ensemble Clo value
1.T-shirt, thick trousers 0.39
2.Half-shirt, thin trousers 0.50
3.Half-shirt, thick trousers 0.59
4.Full-shirt, thin trousers 0.56
5.Full-shirt, thick trouser 0.65
their clo values were assigned to subjects depending on what fit
best. The clo values given in Table 3 take into account insulation of
undergarments and foot wear. Additional winter garments’ clo
values were chosen from those given by Nicol et al. [30] for their
study in Pakistan: sweat-shirt, 0.34 clo; cardigan, 0.22 clo; sweater,
0.30 clo; waistcoat/vest, 0.18 clo; jacket, 0.30 clo.
When girls were not using western outfits, they preferred the
traditional ensemble of salwarekameez. These were treated similar
to ensembles two or four from Table 3 depending on sleeve length
of the kameez. Extra insulation for a dupatta [30] was added when
found in use along with the salwarekameez. The minimum clo
value observed during the survey was 0.39 (ensemble 1) and the
maximum was 0.99 (ensemble 5 with sweat-shirt).
Experiments in the laboratory did not involve any heavy lifting
or strenuous exertion. But since students had to take observa-
tions, change settings on the instruments etc., they stood by their
experimental set-ups or intermittently moved around. We
assumed a metabolic rate of 1.6 met for this activity level, as given
for “Standing, light activity (shopping, laboratory, light industry)” in
ISO 7730-2005 [31]. Students generally walked or cycled from
their hostels to class. So we began our survey 75e90 min after
beginning of class to let the subjects settle down to a stable ac-
tivity level.
2.2.4. Weather data
Temperature data was collected from the Department of Physics
and Meteorology’s weather station. Daily mean temperature was
taken as an arithmetic average of the daily maximum and mini-
mum temperatures. Running mean temperature for day ‘n’ was
calculated from the daily mean temperature data using Equation
(2) [32].
n o
trmt ¼ ð1  aÞ tm;n1 þ atm;n2 þ a2 tm;n3 /
Here, tm,nx is the daily mean temperature for the day ‘x’ days
before day ‘n’. In keeping with EN15251, we take a value of 0.8 for a
and evaluate a seven day running mean.
3. Results and analysis
use of ceiling fans started, air velocity directly under the fans could
rise above 2 m/s and average air velocities were close to 0.5 m/s. But
at some distance from the fans, air velocity rapidly dropped to ‘0’.
Even with multiple windows open, there was very little of wind
flow or cross-ventilation. So even during summer, quite some
portion of the lab had ‘dead zones’ for air velocity. Experimental
set-ups, and hence the students, were always located near fan
positions.
Analysis showed that local thermal conditions of different
groups did not vary much from each other. For example, the
maximum variation operative temperature had between any two
groups for any given day was only 1
C. Hence we decided to
analyse the response data as an average of votes from all the stu-
dents on a particular day and averaged values of the environmental
measurements were used. Further calculations and analysis are
done using these day-average values. Regressions done with the
average values use a weighting factor equivalent to the number of
responses on each day.
Air temperature measurements taken at ankle and shoulder
levels never varied by more than 2
C even on the hottest days. We
believe this is due to the large height of the building. We also
observed that the measured globe temperature exactly matched up
with the calculated operative temperature (rounded up to the
nearest whole number). Predicted mean vote values were calcu-
lated using Fanger’s methodology [33]. Maximum and minimum
values during the survey days for outdoor conditions, indoor con-
ditions, average ensemble clo values, APD values, calculated TSI
values and calculated PMV, PPD values are given in Table 4. In the
same table, we also note maximum and minimum values for the
MTSV on ASHRAE seven point scale and the MTCV on the Bedford
scale. These values are the average votes of all respondents for a
day. Number of respondents on different days varied between 23
and 32.
Maximum values for PMV, PPD, MTCV, and APD occurred on the
same survey day when both top and pa were at their maximum
recorded values. However, MTSV had its highest recorded value on
a different day. We think the higher MTSV value was because it was
the first day in the laboratory for that particular group of students
and as they were not used to the environment, they might have
judged it a bit harshly.
Maximum difference between average values for ta and tmrt is
3
C. We believe such magnitude of variation is due to the large
number of windows present and the value would have been higher
(2)
if not for all the windows being well protected from direct sunlight.
Both tmrt and tg were highly correlated to ta with R2 values of
0.98 and 0.99 respectively. The building’s large thermal mass
ensured that during hot summer days, indoor average DBT could be
as much as 3
C below outdoor DBT even with open doors and
windows.

A total of 342 responses were collected from the subjects. Out of Table 4
these, four responses were classified as ‘invalid’. A certain response Maximum and minimum values for observations.
was deemed invalid if: a subject wanted ‘no change’ in the envi- Parameter Maximum Minimum Parameter Maximum Minimum
ronment, yet found it ‘unacceptable’ or the subject voted an
Meteorological data
extreme thermal sensation but still preferred to have more of the Daily Minimum 25
C 9
C Daily 42
C 20
C
same thermal condition. Results discussed below are for the 338 Maximum
‘valid’ responses. Diurnal Range 18
C 7
C Daily Mean 33
C 14.5
C
RMT 32.8
C 16.9
C
Indoors
3.1. Indoor and outdoor conditions ta 34
C 22
C tmrt 37
C 22
C
top 35
C 22
C pa 3.04 kPa 1.06 kPa
As discussed in Section 2.2.1, we took one set of measurements va 0.6 m/s 0.05 m/s Average clo 0.91 clo 0.44 clo
for every group of students. When ceiling fans were not in use, a TSI 32
C 19
C PMV 3.06 0.50
PPD 99% 10% MTSV 1.74 0.73
0 air velocity was recorded. For such instances, instead of 0 m/s, we MTCV 1.48 0.48 APD 59% 0%
use a nominal value of 0.05 m/s for analysis and calculations. Once
228 A.K. Mishra, M. Ramgopal / Building and Environment 71 (2014) 223e232

3.2. Analysis of votes on ASHRAE and Bedford scales Table 5

Cross-tabulation of TSV and TCV.

Fig. 3 shows a scatter plot of TSV. It is easy to observe from the TSV TCV
figure how widely spread out individual thermal sensations are L3 L2 L1 0 1 2 3
even on the same day. Cross-tabulation of all TCV and TSV is given
L3 0 0 1 0 0 0 0
in Table 5 and Fig. 4 represents a bar graph for the number of votes
L2 0 1 19 3 0 0 0
under different categories for thermal sensation and thermal L1 0 1 41 14 5 0 0
comfort. 0 0 0 14 31 21 0 0
There is match of TSV and TCV 52% of the times. Maximum 1 0 0 4 19 79 7 1
2 0 0 0 1 39 19 0
number of matches occur at voting level ‘1’, followed by ‘1’ and ‘0’.
3 0 0 0 0 3 10 5
Most disagreements for votes on the two scales occur at extreme
sensation values (3,2,2,3) with 30% votes in extreme values of
TSV compared to only 13% in TCV (Fig. 4). The students appear to be
more stringent while voting for thermal sensation than when
amongst the occupants. This was expected since over course of the
voting for thermal comfort at the extreme scale points. But in the
survey, we examined occupant responses both in the hottest and
central three categories (1,0,1), there is a high degree of corre-
coolest month they face during their semesters. The neutral tem-
spondence for votes on the two scales: 96% who voted in the central
perature found in this study is close to that found by Kwok in the
three for TSV also voted in the central three for TCV. There were
study of NV classrooms in Hawaii [13] as well as that found by
only nine instances when the signs of TSV and TCV were different
Ogbonna and Harris in university classrooms in Jos, Nigeria [15].
with all such instances being for votes of 1/1. This similarity in
Though both these study locations have Type A climate, similar to
voting pattern is also observed by Wong and Khoo [14].
Kharagpur, the similarity with results from Kwok’s study is more
A Wilcoxon signed rank test (two-sided, paired) for TSV and TCV
pertinent as it was conducted across both cool and hot seasons, like
(p ¼ 0.24); MTSV and MTCV (p ¼ 0.75) revealed that the votes are
the current study.
not statistically significantly different. So it would seem that using
In view of the similarity of voting pattern for MTSV and MTCV,
TCV or TSV values for analysis should not give us significantly
we also give here the regression equation between MTCV and top.
different results. We further discuss this in the following section.

MTCV ¼ 0:14top  3:72; R2 ¼ 0:87; p < 0:001 (4)

3.3. Regression analysis for neutral temperature
C
Equation (4) yields a neutral temperature of 26.6 and a
comfort zone between 19.4 and 33.7
C. There is a close resem-
We regressed MTSV against three different indices for indoors
blance between comfort zones obtained from regression of MTSV
viz. ta, top, and TSI. Regression correlation came to be strongest with
and MTCV vs top while the two neutral temperatures are practically
TSI (R2 ¼ 0.88) followed by top (R2 ¼ 0.86), and ta (R2 ¼ 0.84). We feel
same. Pellegrino et al. have also observed a similar result [23].
TSI has an unfair advantage in this situation, having been developed
As a check of our decision to take average values of environ-
in Indian conditions and for Indian subjects. So, to allow for com-
mental parameters for the whole building, we also do a regression
parison with other works and in keeping with observations made
analysis between the individual’s TSV and top measured in vicinity
regarding utility of top in certain seminal works on adaptive com-
of his/her group. Equation (5) gives the resulting regression
fort [32,34], we use top as the index for indoor environment.
relation.
Regression equation between top and MTSV is given in Equation (3)

MTSV ¼ 0:18top  4:77; R2 ¼ 0:86; p < 0:001 (3) TSV ¼ 0:184top  4:866; R2 ¼ 0:39; p < 0:001 (5)
This equation yields a neutral temperature of 26.5 and a
C In validation of our approach, Equation (5) yields neutral tem-
comfort zone (based on votes of 1) between 20.9 and 32
C. The perature and comfort zone (26.4
C, 21e31.9
C) near identical to
small slope of the equation implies a high degree of adaptation those found from Equation (3). The individual thermal sensation

Fig. 3. Scatter plot of thermal sensation votes.

 A.K. Mishra, M. Ramgopal / Building and Environment 71 (2014) 223e232 229

Fig. 4. Thermal sensation and thermal comfort votes.

Fig. 5. Thermal sensation votes on ASHRAE seven point scale and their regression lines with operative temperature. (a) Thermal Sensation Votes vs Operative temperature. (b) Mean
Thermal Sensation Vote vs Operative temperature.
230 A.K. Mishra, M. Ramgopal / Building and Environment 71 (2014) 223e232
votes and the MTSV values, along with their corresponding
regression fits, are shown in Fig. 5(a) and Fig. 5(b) respectively.
Since the regression between TSV and top comes with inter-
individual variability of rating same environments differently as
well as all votes having integral values only, R2 value of Equation (5)
is much smaller than that of Equation 3. However, in behavioural
studies, R2 values of even 0.25 would indicate strong correlations
[35]. As Kenny further states, “. Correlations of this size (0.81) are
often between two different measures of the same variable.” [35].
This statement has great import for field studies of thermal com-
fort. As in the current work, often other authors have also found R2
values of the order 0.8 when relating mean votes of a space and
mean indoor temperatures [30,36]. This implies that while inter-
individual perception may vary a lot, average perception of a
large enough group regarding a thermal environment may be taken
as an unbiased estimate of the ambient temperature.
Equation (6) gives the regression relation between PMV values
(calculated using room average conditions) and the corresponding
top.
PMV ¼ 0:18top  3:57; R2 ¼ 0:95; p < 0:001

C,
This correlation yields a neutral temperature of 19.8 i.e.
about 7
C smaller than the neutral temperature obtained from the
occupant votes. Analysing this discrepancy further, we regress
MTSV against PMV values (Equation (7)).
MTSV ¼ 0:996PMV  1:154; R2 ¼ 0:88; p < 0:001
Equation (7) shows that an MTSV of 0 (neutral) would occur for
PMV value of 1.2, indicating the high degree of adaptation occu-
pants have developed to local weather and NV buildings. However,
we must note here that some of the temperature and humidity
values measured in this survey are beyond the standard range for
application of PMV d as specified by ISO 7730 [31].
3.4. Assessment of acceptability
As Brager et al. have observed, different criteria for determining
acceptability levels of occupants are likely to give very different
results [37]. We analyse acceptability levels assuming the following
four types of votes to be accepting of their environment:
e Votes in central three categories of ASHRAE scale.
e Votes in central three categories of Bedford scale.
Fig. 6. Acceptability assessment using different criteria.
e Votes that answer ‘Yes’ to the direct question of acceptability.
e Votes that answer ‘No change’ to the question on thermal
preference.
Fig. 6 shows the total number of students (out of 338) who gave
acceptable votes under the different assumptions. We see that the
first three criteria give similar, though not exactly the same, results.
Votes on the Bedford scale give a higher acceptability than even the
direct acceptability question. This might be due to some subjects
not being able to separate their feelings for overall acceptability of
the class from their perception of thermal comfort alone. The
acceptability pattern we find using different criteria, highest from
Bedford scale to lowest from thermal preference, is similar to that
found by Wong and Khoo in their study of school classrooms in
Singapore [14]. A comparison of acceptability found using the first,
third, and fourth criteria also bears good resemblance to the results
of Kwok from NV classrooms [13], with direct acceptability giving
the highest percentage and thermal preference the least.
Use of the preference question to determine acceptability seems
to be a rather stringent criteria with only 33% finding their envi-
(6) ronment acceptable. As observed by Humphreys et al., votes on the
preference scale may be more about “lifestyle aspirations” than
about acceptability of the current environment [38]. Responses to
the preference question are also affected by thermal experience of
the occupants [39]. Since the study location has a tropical climate, a
majority wanted to feel ‘cooler’ (60%). During the summer months
of March and April, there were zero votes for the ‘warmer’ prefer-
(7)
ence where as even on the coolest day, three subjects wanted to be
‘cooler’.
3.4.1. Acceptable zones of thermal comfort
Using the first three criteria set up for measuring acceptability of
occupants in Section 3.4, we analyse possible zones of thermal
comfort over which students are likely to accept their thermal
environment. We examine the percentage of occupants that
accepted their thermal environment on each day and then a second
order polynomial fit is obtained between percentage of acceptable
votes and average operative temperature on a day. Results are
shown in Fig. 7. From Fig. 7, a sharp decline in acceptability per-
centage at room temperatures beyond 30
C is observed. This is also
reflected in the nature of fitted polynomial curve. Similar trend of a
‘threshold’ has been observed by Zhang et al. from their analysis of
the ASHRAE database for acceptability in summer time in NV and
mixed mode buildings [40]. Unlike Zhang et al., we do not
 A.K. Mishra, M. Ramgopal / Building and Environment 71 (2014) 223e232
Fig. 7. Percentage of acceptable votes vs operative temperature.
encounter a sharply declining acceptability threshold on cool side
as our lowest recorded temperature was 22
C.
Solving the fitted polynomials for an 80% acceptability yields the
following thermal comfort zones:
e Votes in central three categories of ASHRAE scale: 22.7e28.9
C
e Votes in central three categories of Bedford scale: 17.9e32.8
C
e Votes that answer ‘Yes’ to the direct question of acceptability:
19.4e30.9
C
All three of the above zones are broader than those recom-
mended for conditioned spaces in NBC [11]. Assuming direct
acceptability votes as the reliable indicator of an occupant’s overall
comfort level, a comfort zone of 20e31
C may be assumed for the
laboratory building. The comfort zone we find is similar to what
Nicol et al. found from their field studies in Pakistan [30]. This is of
course the range of temperatures occupants can adapt to over
different seasons and not drifts that would be tolerated over a day.
3.5. An adaptive comfort equation
Adaptive comfort equations recommended by both ASHRAE
Standard 55 and EN15251 were developed using database of oc-
cupants engaged in near sedentary activities. A question naturally
arises as to how the comfort temperature (tc) estimates of an ACE
developed from the current observations would compare with
those from the standard ACEs. Using Griffiths’ equation along with
a slope of 0.5/
C d as suggested by Humphreys et al. in a recent
work [41] d we calculate tc on each day of the survey. Following the
recent addenda to ASHRAE Standard 55-2010 [24], we use a seven
day mean of daily mean temperatures as the PMOAT (to).
A regression between PMOAT and tc yields the following ACE:
tc ¼ 0:53to þ 15:23; R2 ¼ 0:93; p < 0:001
Table 6
A comparison with standard adaptive comfort models.
ASHRAE
Comfort equation tc ¼ 0.31to þ 17.38
tc correction for air velocity 1.2
C
Dt c 3.3
C
Dt c for velocity corrected values 2.4
C
231
For the 12 survey days, we compare average values of Dtc be-
tween tc calculated using Equation (8) and using other adaptive
models: Standard 55-2010 [24], EN15251 [32], the work by Hum-
phreys et al. [41], and a recent ACE proposed for South-East Asia by
Nguyen et al. [42]. For outdoor temperature, we use PMOAT based
on a seven day average for ASHRAE equation, RMT for EN15251 and
Humphreys’ model, and current monthly mean temperature (tmm)
for Nguyen’s model.
We also consider values of tc found using corrections suggested
in ASHRAE and EN15251 models for higher air velocities, using
average velocities recorded during the survey for correction. Since
average velocities recorded are always 0.6 m/s, for the ASHRAE
values of tc we include a correction 1.2
C for all days recording an
average velocity greater than 0.3 m/s. Summary results are pre-
sented in Table 6.
An important observation from Table 6 is that though there are
differences in the predicted (tc) values, the differences are not very
stark. The maximum Dt c value found is 3.3
C. For perspective,
ASHRAE recommended 80% comfort zone itself has a width of 7
C.
Considering velocity correction in tc decreases Dt c for both ASHRAE
and EN15251 models. Models of both Nguyen et al. and Humphreys
et al. give reasonably close predictions. We believe this is because
Nguyen et al. gave the model based on meta-analysis of a group of
studies done in Type A climate and due to the similarity of slope
between our and Humphreys’ et al.’s equations. Least deviations are
observed with the EN15251 model after taking into account ve-
locity correction. This shows the necessity for adaptive models to
take into account enhanced air velocities especially in tropical cli-
mates with widespread use of fans. We observe that there are
differences between the predictions from different models, though
not very large ones. Models developed for similar climates have
lesser differences and bringing effect of enhanced air velocity into
consideration also helps reduce difference between model
(8) predictions.
EN15251 Humphreys et al. Nguyen et al.
tc ¼ 0.33trmt þ 18.8 tc ¼ 0.53trmt þ 13.8 tc ¼ 0.341tmm þ 18.83
7  4 þ 50 pffiffiffiffi
n.a. n.a.
10 va C (va>0.1 m/s)
1.3
C 1.2
C 1.2
C
0.5
C n.a. n.a.
232 A.K. Mishra, M. Ramgopal / Building and Environment 71 (2014) 223e232
4. Conclusion
This survey was a first step towards classroom thermal comfort
research in tropical climate of India. Results show a comfort zone
broader than 10
C as well as a high correlation between indoor
comfort temperature and outdoor conditions. Even though the
study was conducted on occupants having sustained metabolic rate
of 1.6 met, adaptive comfort principle was found to be very much
applicable. This would suggest that with further complimentary
research, the metabolic rate limits set for application of adaptive
models in current standards may be broadened.
One thing to note here is that the subjects in this study are not
strangers to AC. Most of their larger lecture theatres and labora-
tories dealing with computation/simulation are air-conditioned. So
the results are not entirely due to low expectations. India, with its
varied climatic zones, economic growth, population size and
number of buildings, and looming energy crisis, could greatly
benefit from development and implementation of an adaptive
comfort standard. Until an ACE tailor made for Indian conditions is
developed, our results indicate that use of EN15251 model (with
velocity correction) could give good results.
Acknowledgements
We appreciate the cooperation of all our subjects. We are
grateful for the cooperation of Mr. Ranendranath Dey and Mr.
Pradip Roychowdhury during the survey and set up of equipments.
We are thankful to Prof. B. K. Mathur of the Department of Physics
and Meteorology for making available to us historical and current
temperature data. We are also grateful for insights on statistical
methods from Prof. P. K. J. Mohapatra, Department of Industrial
Engineering.
References
[1] Auliciems A. Classroom performance as a function of thermal comfort. Int J
Biometeorol 1972;16(3):233e46.
[2] Wyon DP, Andersen I, Lundqvist GR. The effects of moderate heat stress on
mental performance. Scand J Work Environ Health 1979;5(4):352e61.
[3] Mendell MJ, Heath GA. Do indoor pollutants and thermal conditions in schools
influence student performance? A critical review of the literature. Indoor Air
2005;15(1):27e52.
[4] Leyten JL, Kurvers SR, Raue AK. Temperature, thermal sensation and workers’
performance in air-conditioned and free-running environments. Architectural
Sci Rev 2012;56(1):14e21.
[5] UGC, Higher education in India d Issues related to expansion, inclusiveness,
quality and finance, secretary. University Grants Commission: New Delhi.
[6] Annual report 2010e11. Department of School Education & Literacy and
Department of Higher Education, Ministry of Human Resource Development,
Government of India; 2012.
[7] Isaac M, Van Vuuren DP. Modeling global residential sector energy demand for
heating and aireconditioning in the context of climate change. Energy Policy
2009;37(2):507e21.
[8] IEA. Energy balances of non-OECD countries. Paris, France: International En-
ergy Agency; 2011.
[9] CSO. Energy statistics. 19th ed. New Delhi, India: Central Statistics Office.
Ministry of Statistics and Programme Implementation. Government of India;
2012.
[10] Kumar S, Kapoor R, Rawal R, Seth S, Walia A. Developing an energy conser-
vation building code implementation strategy in India; May 2010. Tech. rep.,
ECO-III.
[11] BIS. National building code of India 2005. New Delhi, India: Bureau of Indian
Standards; 2005.
[12] BEE. Energy conservation building code user guide. New Delhi, India: Bureau
of Energy Efficiency; 2009.
[13] Kwok AG. Thermal comfort in tropical classrooms. ASHRAE Trans
1998;104(1):1031e50.
[14] Wong NH, Khoo SS. Thermal comfort in classrooms in the tropics. Energy
Buildings 2003;35(4):337e51.
[15] Ogbonna A, Harris D. Thermal comfort in Sub-Saharan Africa: field study
report in Jos-Nigeria. Appl Energy 2008;85(1):1e11.
[16] Cândido C, de Dear R, Lamberts R. Combined thermal acceptability and air
movement assessments in a hot humid climate. Building Environ 2011;46(2):
379e85.
[17] Rao MN. Comfort range in tropical Calcutta. A preliminary experiment. Indian
J Med Res 1952;40(1):45e52.
[18] Sharma MR, Ali S. Tropical summer index e a study of thermal comfort of
Indian subjects. Building Environ 1986;21(1):11e24.
[19] Indraganti M. Using the adaptive model of thermal comfort for obtaining
indoor neutral temperature: findings from a field study in Hyderabad, India.
Building Environ 2010;45(3):519e36.
[20] Deb C, Ramachandraiah A. Evaluation of thermal comfort in a rail terminal
location in India. Building Environ 2010;45(11):2571e80.
[21] Indraganti M, Ooka R, Rijal HB. Field investigation of comfort temperature in
Indian office buildings: a case of Chennai and Hyderabad. Building Environ
2013;65:195e214.
[22] Dhaka S, Mathur J, Wagnerb A, Agrawal GD, Garg V. Evaluation of thermal
environmental conditions and thermal perception at naturally ventilated
hostels of undergraduate students in composite climate. Building Environ
2013;66:42e53.
[23] Pellegrino M, Simonetti M, Fournier L. A field survey in Calcutta. Architectural
issues, thermal comfort and adaptive mechanisms in hot humid climates. In:
Proceedings of 7th Windsor Conference: the changing context of comfort in
an unpredictable world. Windsor, UK: NCEUB; 2012.
[24] ANSI/ASHRAE, Standard 55e2010. Thermal comfort conditions for human
occupancy. Atlanta: ASHRAE; 2010.
[25] ASHRAE. ASHRAE handbook: fundamentals. SI ed. Atlanta, GA: American So-
ciety of Heating, Refrigerating and Air-conditioning Engineers; 2009.
[26] Vaisala humidity calculator for windows, version 2.2www.vaisala.com/
humiditycalculator; Nov 2009.
[27] R Core Team. R: a language and environment for statistical computing.
Vienna, Austria: R Foundation for Statistical Computing; 2012. http://www.R-
project.org.
[28] Humphreys MA. An adaptive approach to the thermal comfort of office
workers in North West Pakistan. Renew Energy 1994;5(5e8):985e92.
[29] Morgan C, de Dear R. Weather, clothing and thermal adaptation to indoor
climate. Clim Res 2003;24(3):267e84.
[30] Nicol JF, Raja IA, Allaudin A, Jamy GN. Climatic variations in comfortable
temperatures: the Pakistan projects. Energy Buildings 1999;30(3):261e79.
[31] ISO. ISO 7730: Ergonomics of the thermal environment e analytical deter-
mination and interpretation of thermal comfort using calculation of the PMV
and PPD indices and local thermal comfort criteria. 3rd ed.; 2005. Geneva.
[32] Nicol F, Humphreys M. Derivation of the adaptive equations for thermal
comfort in free-running buildings in European standard EN15251. Building
Environ 2010;45(1):11e7.
[33] Fanger PO. Thermal comfort. Analysis and applications in environmental en-
gineering. Copenhagen: Danish Technical Press; 1970.
[34] de Dear RJ, Brager GS. Developing an adaptive model of thermal comfort and
preference. ASHRAE Trans 1998;104(1):145e67.
[35] Kenny DA. Statistics for the social and behavioral sciences. Boston: Little
Brown; 1987.
[36] Rijal HB, Yoshida H, Umemiya N. Seasonal and regional differences in neutral
temperatures in Nepalese traditional vernacular houses. Building Environ
2010;45(12):2743e53.
[37] Brager G, Fountain M, Benton C, Arens E, Bauman F. A comparison of methods
for assessing thermal sensation and acceptability in the field. In: Oseland N,
editor. Proceedings of thermal comfort: past, present and future. Watford, UK:
British Research Establishment; 1993.
[38] Humphreys MA, Nicol JF, Raja IA. Field studies of indoor thermal comfort and
the progress of the adaptive approach. Adv Building Energy Res 2007;1(1):
55e88.
[39] Humphreys MA, Hancock M. Do people like to feel ‘neutral’?: exploring the
variation of the desired thermal sensation on the ASHRAE scale. Energy
Buildings 2007;39(7):867e74.
[40] Zhang H, Arens E, Pasut W. Air temperature thresholds for indoor comfort and
perceived air quality. Building Res Inf 2011;39(2):134e44.
[41] Humphreys MA, Rijal HB, Nicol JF. Updating the adaptive relation between
climate and comfort indoors; new insights and an extended database.
Building Environ 63:40e55.
[42] Nguyen AT, Singh MK, Reiter S. An adaptive thermal comfort model for hot
humid South-East Asia. Building Environ 2012;56(0):291e300.