lly

02, I

Received in revised form27 April 2015Accepted 21 May 2015

Adaptive thermal comfortClassroomsNatural ventilation

al cf Te

taken during both semesters over the academic year 2013e14. Results of the surveys gave a regressionneutral temperature near 29 C while preferred temperature was found to be 26.8 C. Using student

environment. Analysis of thermal preference and thermal acceptability votes showed a distinct prefer-

ts, higver th

comfort standards will be an essential part of designing thermallycomfortable and energy efcient classrooms. Formulation of suchstandards, that supplement Indian building codes and pave thewaytowards sustainable future development of India, would be

no et al. [7] carriedoms in a couple ofier works, we havefor undergraduatege, no long termfrom Indian class-y was undertaken

To obtain year round data regarding occupant thermal sensationand acceptability for naturally ventilated classrooms in India

To study student adaptive behaviour during classes and makeobservations on how such behaviour helps or hinders thermaladaptation

To compare results from the classroom survey with those fromthe laboratory [8]; subjects in the laboratory had an appreciablyhigher metabolic rate

* Corresponding author.E-mail addresses: asit.mishra@mech.iitkgp.ernet.in (A.K. Mishra), ramg@mech.

Contents lists availab

Building and E

ls

Building and Environment 92 (2015) 396e406iitkgp.ernet.in (M. Ramgopal).growth rate (CAGR) of 11% while student enrolment CAGR was 6%[1]. Taking these numbers as an indication, it is easy to see thatcoming years will witness a dramatic growth in higher educationsector of India. This will mean signicant growth in the sheernumber of classrooms. Conscientious design of classroom thermalenvironment is necessary both because of the high occupant den-sities classrooms have and the adverse impact decient thermalsettings can have on the teaching-learning process. Judicious

and 31 C. Looking at the Indian context, Pellegriout a short duration comfort study for NV classrouniversities located in Kolkata. In two of our earlreported ndings from thermal comfort studylaboratories [8,9]. To the best of our knowledthermal comfort eld study has been reportedrooms. Thus, to address this gap, the current studwith the following aims:higher education institutes in India have seen compounded annual NV classrooms. Comfort zones for these surveys were between 24Hotehumid climateAdaptive opportunities

1. Introduction

With 14.6 million enrolled studenIndia is one of the largest globally. Ohttp://dx.doi.org/10.1016/j.buildenv.2015.05.0240360-1323/ 2015 Elsevier Ltd. All rights reserved.ence amongst occupants for cooler than neutral sensation. Diurnal variation of temperature that wouldbe acceptable to 80% or more occupants was found to be a 4 C wide band. Study of student actionsduring surveys showed that fans were brought into play more often than windows. Variation of clothingshowed strongest correlation with the day's minimum temperature. Overall, observations from the studyshowed broad comfort zones and signicant level of occupant adaptation to the environment of NVclassrooms.

2015 Elsevier Ltd. All rights reserved.

her education system ofe past years, number of

immensely aided by the results of thermal comfort eld studies.Over the past couple of decades, a few studies have been carried

out in naturally ventilated (NV) classrooms located in tropics [2e6].Results of these studies showed that students adapted well to theirKeywords:Available online 27 May 2015 responses to thermal acceptability question, 80% occupant satisfaction was found between 22.1 and31.5 C operative temperature. Over the survey duration, nearly 79% of responses accepted their thermalA thermal comfort eld study of naturaKharagpur, India

Asit Kumar Mishra*, Maddali RamgopalDepartment of Mechanical Engineering, Indian Institute of Technology, Kharagpur 7213

a r t i c l e i n f o

Article history:Received 3 February 2015

a b s t r a c t

To assess occupant thermrooms of Indian Institute o

journal homepage: www.eventilated classrooms in

ndia

omfort, eld studies were carried out in naturally ventilated (NV) class-chnology Kharagpur. The location has a hotehumid climate. Surveys were

le at ScienceDirect

nvironment

evier .com/locate/bui ldenv

dents enrolled while the Spring course had 51 students. Fifteen

g and2. Methodology

Study was conducted in undergraduate classrooms of IndianInstitute of Technology Kharagpur (IIT). Kharagpur, a small town-

Nomenclature

AbbreviationsAC air-conditioningAPD actual percentage dissatisedDBT dry bulb temperatureMRT mean radiant temperatureMTSV mean thermal sensation voteNV naturally ventilatedPMOAT prevailing mean outdoor air temperaturePS percentage of acceptable votes/percentage satisedRH relative humidityRMT running mean temperatureTSV thermal sensation vote (individual's)

Symbolspw partial pressure of water vapour in air, kPata air temperature, Ctc comfort temperature, using Grifths' equation, Ctg globe temperature, Ctmrt mean radiant temperature, Ctop operative temperature, Ctrmt running mean temperature, Cva air velocity, m/sDTc (top tc), C

A.K. Mishra, M. Ramgopal / Buildinship in eastern India, has a tropical climate with dry winters eKoppen climate classication Aw. The region primarily experi-ences three seasons, a hot summerwith some sporadic rain, awarmand humid monsoon and a mild and dry winter. Due to the inu-ence of south-west monsoon, the bulk of rainfall occurs over themonths of July, August, September, and part of June and October.

Surveys were of longitudinal design and followed one courseeach during Autumn 2013 and Spring 2014 respectively. Bothcourses were taught in the same classroom. Spring semester in IIT isfrom January to April while the Autumn semester is from mid-Julyto November. The Autumn semester, unlike the Spring has a mid-semester break of about 10 days. The survey took place on 5 daysduring Autumn of 2013 and 7 days during Spring semester of 2014.

2.1. The classroom and the subjects

The room (room code: CRe310) is on the top oor of a buildingwhose major axis is along EasteWest direction. All the rooms in thetop oor of this building are used as classrooms. The EasteWestorientation of the major axis of the building and the complete topoor being utilized as classrooms is a feature shared by several ofthe neighbouring departmental buildings as well. The building haspillared constructionwith brick andmortar lling. Doors of CRe310open into a corridor while the walls on east and west are internalwalls. Windows are on north faade of the room and are singleglazed with mild steel frames. Windows are 1.8 m wide and 1.5 mhigh, with continuous overhangs about one meter deep. Di-mensions of CRe310 are as follows: ceiling height 3.4 m, length6.7 m, breadth 8.3 m Fig. 1 gives the room layout and an idea onroom furnishings.

The survey followed two courses taken in CRe310 e one during

points were later averaged and used for all further analysis.

Mean radiant temperature (tmrt) was calculated from Equation

(1) [14].

tmrt tg 273

4 1:1 108v0:6aD0:4

tg ta1=4

273 (1)

In the above equation, is surface emissivity for the globe andwas taken as 0.95 while D is diameter of the globe in meters.

Room operative temperature (top) was calculated using the

formula: top hc,tahr,tmrthchr [14]. A constant value of 4.7 W/m2, C

[14] was assumed for the radiative heat transfer coefcient (hr).Convective heat transfer coefcient (hc) was calculated from thecorrelations for persons seated with moving air in ASHRAE Fun-damentals Handbook (Table 6, Chapter 9) [14].

2.2.2. Subjective questionnaireEnglish, being medium of instruction in all courses, was chosenstudents were common to both courses, putting the number ofunique subjects at 67. The student population was a mix of un-dergraduate and graduate students, with age between 19 and 26years. All subjects were Indians and were assumed to be acclima-tized to the local climate. Every effort was made to minimallyinconvenience the classes during which surveys were conducted.To this end, at the beginning of each semester, students were brieyintroduced to the nature and purpose of the surveys. What wasparticularly impressed upon the students was that they should feelfree to provide their forthright responses and that the survey wasnot asking for a mandate regarding usage of air-conditioning in theclassrooms. Towards later survey days during each semester, in abehaviour eerily similar to that observed by Teli et al. [11] in theirstudy with primary school children, some students queried as towhy they had to ll in the same questionnaire repeatedly.

2.2. Collection of survey data

2.2.1. Indoor environmental parametersInformation regarding instruments used for measuring indoor

environmental parameters is given in Table 1. The black globethermometer was made in-house by placing an alcohol ther-mometer at the centre of a 70 mm diameter plastic ball. The ballwas painted black and coated with lamp black. This globe ther-mometer had also been used in our previous studies [8,12].

For the classes during which surveys were conducted, mea-surements were taken during the last 20 min of the class. Paperquestionnaires were given to students for lling up during the lastten minutes. Since the classes were either one or two hours long,the subjects thus had ample time to attain a stable metabolic rate.Measurements were taken at six points around the room for airtemperature (ta), relative humidity (RH), and air velocity (va). Globetemperature (tg) was measured at one central location in the room.Approximate measurement locations are specied in Fig. 1. Allmeasurements were taken at approximately shoulder level of oc-cupants. While measuring air velocity inside the classroom, it wasnoted that little to no wind came in through windows. This may beascribed to the building being surrounded by other buildings andtree lining. Thus the air velocity measured was almost solely due tothe fans. When the fans were not used, no perceptible air velocitywas recorded. From the ta and RH values, partial pressure of watervapour in air (pw) was calculated using the Vaisala HumidityCalculator 3.0 [13]. Values of ta, pw, and va recorded across the sixAutumn and one during Spring. The Autumn course had 31 stu-

Environment 92 (2015) 396e406 397to frame the questionnaire. An attempt was made to keep

a we

A.K. Mishra, M. Ramgopal / Building and Environment 92 (2015) 396e406398Fig. 1. Classroom layout. Points 1e6 are locations where ta, RH, and vquestionnaires brief so as not to take up too much from the classtime. Fig. 2 shows the subjective questionnaire. In addition toquestions regarding how the students perceived their thermalenvironment, a short section regarding clothing was also included.

Table 1Survey instruments.

Instrument and make Measured parameter

Globe thermometer fabricated in-houseLutron LM 8102, Air temperature5-in-1 meter Relative humidity (RH)

Air velocitySound level

Fig. 2. Survey qure measured. GT gives location of globe thermometer during surveys.In this section, the most commonly used clothing ensemblesamongst students were listed out and the responder only had totick mark the appropriate choice. Numbers given along comfortquestionnaire represent the values used for different responses

Range Resolution Accuracy

10 e 110 C 1 C 1 C0e50 C 0.1 C 1 C10e95% 0.1% RH 4% RH0.4e30 m/s 0.1 m/s 3%35e130 dB(A) 0.1 dB 3.5 dB

estionnaire.

during later analysis. Numbers next to the clothing ensemblesrepresent the approximate clo values for each ensemble andinclude contribution from undergarments and foot wear. Thesenumbers have been used only for this illustration and were not partof the original questions.

During winter days, students were asked to write down anyextra winter garment they were wearing next to their clothingensemble. For winter garments, clo values were taken from thework by Nicol et al. [15] in Pakistan: cardigan, 0.22 clo; jacket, 0.30clo; sweater, 0.30 clo; sweat-shirt, 0.34 clo; waistcoat/vest, 0.18 clo.

2.2.3. Outdoor temperature dataDaily maximum and minimum temperature and precipitation

records were taken from the institute in-campus weather station.Daily mean temperature (tm) was taken as average of dailymaximum andminimum temperature. Running mean temperatureof day n (trmt) was calculated based on tm values for the last sevendays, using Equation (2) [16]:

the actual number of responses divided by 20. Including Surveys2012e13 and 2013e14, there were a total of ve days when lessthan 20 responses were collected. All statistical analyses werecarried out with the R statistical computing package [19].

3. Results and discussions

Summary of responses received during Surveys of 2012e13 and2013e14 are presented in Table 2. Responses were deemed to beinvalid under two circumstances:

if a subject found her/his thermal environment unacceptableand yet voted for no change in the thermal preference question

if a subject voted an extreme thermal sensation (3) and againvoted for more of the same sensation in the thermal preferencequestion

Only valid responses were analysed further.

A.K. Mishra, M. Ramgopal / Building and Environment 92 (2015) 396e406 399trmt 1 antm;n1 atm;n2 a2tm;n3/

o(2)

The value of a was taken to be 0.8, same as the value used forcalculating trmt in EN15251.

Prevailing mean outdoor air temperature, the recommendedoutdoor temperature index as per ASHRAE Standard 55 [17], on asurvey day was taken as average of tm from last seven days.

A series of transverse thermal comfort surveys had been carriedout in classrooms near CRe310 for three days during November2012 and four days during March 2013 [12]. Though the surveymethodology followed was similar, these earlier surveys (Surveys2012e13) had a slightly different format of subjective questionnaire(Fig. 3). So, only the thermal sensation votes and thermal accept-ability votes of both surveys were analysed together.

During the course of analysis, whenever day-wise study is donee say for mean thermal sensation vote (MTSV) or acceptabilitylevels e a weighting scheme is used to take into account the dif-ference in number of responses across survey days. Humphreyset al. [18] observed from their meta-analysis of thermal comfortsurvey data that with about 20 observations, the standard error ofestimated comfort temperature was only 0.4 K. Hence, 20 re-sponses was taken as an upper limit for ensuring desirable accuracyof results. A weighting factor of one was used for days withnumber of responses greater than or equal to 20. Days whennumber of responses were less than 20 were assigned a weight ofFig. 3. Survey questionnaire for earlResponses from Surveys 2012e13 were taken to be satised ifthey voted Acceptable to the question on acceptability of tem-perature (Fig. 3). On the other hand, a response from Surveys2013e14 was assumed to be satised if it voted Just Acceptable orbetter to the question on acceptability of thermal environment(Fig. 2).

Maximum and minimum values of some of the important in-door and outdoor conditions during survey are given in Table 3. Thevalues presented consider observations recorded during both sur-veys (2012e13 and 2013e14) but due to the longer length of the2013-14 survey, nal values in Table 3 are all from the later survey'sduration.

3.1. Analysis of thermal sensation votes

A day-wise scatter plot of thermal sensation votes, with somearticial jitter added to distinguish between overlapping votes, isgiven in Fig. 4 (a). Due to variability of individual perception, voteson the thermal sensation scale were quite widely distributed evenon the same day. Average standard deviation of TSV was 0.9, with aminimum value of 0.6 and maximum of 1.2.

3.1.1. Regression neutral temperatureRegression of TSV with respect to operative temperature was

initially done separately for the 2012e13 and 2013-14 studies.Resulting relations are given in Equation (3a) and Equation (3b)ier surveys (Surveys 2012e13).

respectively. Because of the shorter duration and narrower range ofoutdoor conditions of the 2012-13 surveys, Equation (3a) has ahigher slope than Equation (3b). Occupant adaptation was morecompletely expressed during the 2013-14 surveys. Both regressionrelations give a neutral temperature of 29.8 C. This supports ourdecision to analyse TSV values from both surveys together.

TSV 0:36top 10:73; R2 0:35; p

TSV 0:20top 5:96; R2 0:43; p

acceptability (PS) was tted with the corresponding MTSV to give asecond order polynomial, Equation (7).

PS10:34MTSV29:01MTSV89:81; R2adj0:71; p

PS 2:56DT2c 4:54DTc 89:54; R2adj 0:71; p

sity

A.K. Mishra, M. Ramgopal / Building and Environment 92 (2015) 396e406404primarily important in terms of giving occupants more sense ofcontrol and exposing more skin area to direct air movement.

Table 5 gives regression R2 value between daily mean clo valuesand different indices of outdoor temperature. Daily average clovalues had the strongest relation with the day's minimum tem-perature, which would have been recorded during early morning.This nding was similar to that of Schiavon and Lee [34] and sup-ports the idea that subjects' decision regarding their apparel isgreatly affected by early morning temperatures. Next to dailyminimum temperatures, average clo values have the strongestrelation with daily mean temperatures.

3.4. A comparison with ndings from the thermal comfort surveysin undergraduate laboratory

Fig. 8. Student behaviour a) seating denDuring the Spring semester of 2013, thermal comfort surveyswere carried out in an undergraduate laboratory at IIT Kharagpur[8]. The laboratory building is in an annex adjacent to the classroombuildings considered in the current work. During laboratory classes,students had a sustained metabolic rate higher than that duringclasses. At the same time, laboratories have a more exible atmo-sphere as compared to classes. Students can move about the room,adopt relaxed posture, and have discussions amongst themselves.During the surveys on warm days in the laboratory, we hadobserved almost every student taking multiple breaks for drinkingwater. Such behaviour was not observed during classes. A com-parison between the observations and ndings from both surveysare presented in Table 6.

The outdoor and indoor conditions during both surveys weresimilar though a lower top was recorded in the classrooms surveys.While APD range was similar, MTSV range was broader for theclassrooms. This may have to do with the cooler temperaturerecorded during classroom survey. The smaller range of sensations

Table 5Correlations between outdoor indices and mean clo values.

Daily mean RMT Daily min

R2/R2adj 0.70/0.67 0.61/0.57 0.76/0.74may also be because laboratories have a three hour duration ascompared to one or two hour duration of classes. The relations thatRohles and Nevins developed from their study on college age stu-dents [14] indicate that as time of exposure increases, magnitude ofthermal sensation decreases.

Both regression neutral temperature and preferred temperaturewere lower for the laboratory studies and by about an equalmagnitude. This may be ascribed to the higher metabolic rates inlab activity. The mean comfort temperature (tc) calculated fromGrifths equation is also lower for the lab study though not asmuch lower as the regression neutral temperature. The regressionequation slope is smaller for the lab study and this may suggest thatthere is more complete expression of adaptive action in thosestudies. However, it may also be because the laboratory studies

across rows b) window and fan usage.encountered as broad a temperature range as the classroom studiesbut over a single semester instead of two.

More interesting are the comfort zones from the two studies.The lower comfort zone limit from laboratory study carries lessimport as temperatures below 22 C were not encountered. Theupper limits of both comfort zones are nearly equal though (dif-ference of 0.6 C). This would suggest that the highermetabolic ratein laboratories was about effectively negated by availability of moreadaptive opportunities. Breadth of diurnal comfort band was about4 C for both studies. But while the band was symmetric for labstudies, it was biased towards cooler deviations for the classroomstudies. This may again be due to more adaptive opportunities inlaboratory that allow subjects to deal with positive deviationsequally well as negative deviations between top and tc.

Fig. 9 tries to present a summation of all 31 days of thermalcomfort survey (12 from laboratory and 19 from classrooms) interms of temperature and humidity combinations that wereacceptable to students. From 21 to 30 C, conditions recorded werealways acceptable to more than 80% of occupants. Over 30 C,

Daily max PMOAT MMT Last day's mean

0.50/0.45 0.59/0.55 0.56/0.51 0.69/0.66

Regression equation

g andRegression neutral temperaturetc (Grifths' equation)Preferred temperatureComfort zone (from 2nd order polynomial t of top and % satisfaction)Diurnal comfort zone widthTable 6Comparison between studies from laboratory and classroom.

Outdoor daily meantopMTSVAPD

A.K. Mishra, M. Ramgopal / Buildinhigher humidity values, e.g. pw >2:6 kPa, become unacceptable.Above 31 C, conditions are mostly unacceptable except for one ortwo exceptions.

Of the 31 days on which thermal comfort surveys were con-ducted, acceptability of thermal conditions were less than 80% on10 days. Analysis of these 10 days showed that except for an earlymorning (7:30 am) class in January, the other nine dayswere duringApril and March. Of these nine cases, 7 were during afternoon. Thisshows a strong indication that March and April are the problemmonths with the afternoon hours being of particular concern.While this may not be surprising eee as March and April are thewarmest months in Kharagpur during which regular classes takeplace ee this trend does suggest a simple precaution. As IIT has aresidential campus, classes during March and April may bescheduled during early morning/late evening hours so as to limitstudent discomfort. Similarly, for January, which is the coldestmonth, classes may be started later in the day, say 8:30 am insteadof 7:30 am, to avoid the most uncomfortable early morning hours.

4. Conclusion

This study presented results from thermal comfort surveysconducted across an academic year in NV classrooms, in a tropicalregion. A 9 C broad comfort zone is determined while the neutraltemperature found is close to 29 C. These results clearly supportthe use of adaptive comfort standards for NV classrooms in tropics.The students were well acclimatized to the local climate. They

Fig. 9. 80% thermal acceptance spread over operative temperature and humidity.displayed adaptive behaviours like use of fans, clothing adjustment,and window operation. Analysis shows that diurnal temperaturevariation of about 4 Cwill be tolerable to nearly 80% occupants. Butthe results also show a trend of subjects preferring cooler thermalsensation and deviations on the cooler side of neutral being moretolerable. Comparison with the comfort studies done in laboratory,at the same location, shows that both studies have similar upperlimit for their comfort zones. This leads us to believe that greateradaptive opportunities in a laboratory setting e as compared toclassroom e can offset the higher metabolic rates prevalent forlaboratory level activity. Additionally, combined results from bothstudies show that the buildings investigated provided acceptablethermal conditions to occupants over most of the operationalperiod. With some adjustment to scheduling, these classrooms canensure student comfort and are unlikely to need intervention ofmechanical conditioning.

The ndings from this study would be helpful for developingfuturistic comfort standards for NV classrooms in India. As a fastdeveloping nation, number of classrooms in India is also risingquickly. Our results show that with appropriate avenues of adap-tation, comfort in NV classrooms is achievable. Thus adaptivecomfort standards can be the guiding principle for the future ofenergy efcient and comfortable classrooms in Indian context.

Acknowledgements

We are grateful for the cooperation from all subjects of thesethermal comfort studies. We also appreciate the help extended byMr. Ashish Anand and Mr. B. Dinesh Reddy, undergraduate stu-dents, in collecting the data during 2012e13 surveys.

References

[1] Planning Commission Report, Annual Status of Higher Education of States andUTs in India, (Summary Report), Planning Commission. Government of India,New Delhi, India, 2012.

Laboratory Classroom

33e14.5 C 32.5e14.5 C35e22 C 35e20 C1.74e(0.73) 2e(1.8)59%e0% 56.8%e0%TSV 0:184top 4:866 TSV 0:22top 6:36726.4 C 29 C28.2 C 29.5 C24 C 26.7 C19.4e30.9 C 22.1e31.5 C2 3;1:2

Environment 92 (2015) 396e406 405[2] A.G. Kwok, Thermal comfort in tropical classrooms, ASHRAE Tran. 104 (1)(1998) 1031e1050.

[3] N.H. Wong, S.S. Khoo, Thermal comfort in classrooms in the tropics, EnergyBuild. 35 (4) (2003) 337e351.

[4] A.C. Ogbonna, D.J. Harris, Thermal comfort in sub-Saharan Africa: eld studyreport in Jos-Nigeria, Appl. Energy 85 (1) (2008) 1e11.

[5] I. Hussein, M.H.A. Rahman, T. Maria, Field studies on thermal comfort of air-conditioned and non air-conditioned buildings in Malaysia, in: Energy andEnvironment, 2009. ICEE 2009. 3rd International Conference on. IEEE, 2009,pp. 360e368.

[6] C. Ca^ndido, R. de Dear, R. Lamberts, Combined thermal acceptability and airmovement assessments in a hot humid climate, Build. Environ. 46 (2) (2011)379e385.

[7] M. Pellegrino, M. Simonetti, L. Fournier, A eld survey in Calcutta. Architec-tural issues, thermal comfort and adaptive mechanisms in hot humid climates,in: Proceedings of 7th Windsor Conference: The Changing Context of Comfortin an Unpredictable World, NCEUB, Windsor, UK, 2012.

[8] A.K. Mishra, M. Ramgopal, Thermal comfort in undergraduate laboratories e Aeld study in Kharagpur, India, Build. Environ. 71 (2014a) 223e232.

[9] A.K. Mishra, M. Ramgopal, Thermal comfort eld study in undergraduate

laboratories e An analysis of occupant perceptions, Build. Environ. 76 (2014b)62e72.

[10] A.K. Mishra, M. Ramgopal, Field studies on human thermal comforte Anoverview, Build. Environ. 64 (2013) 94e106.

[11] D. Teli, P.A. James, M.F. Jentsch, Thermal comfort in naturally ventilated pri-mary school classrooms, Build. Res. Inf. 41 (3) (2013) 301e316.

[12] A.K. Mishra, M. Ramgopal, Thermal comfort in classrooms in tropics: ananalysis of student preference, in: Proceedings of Conference Efcient, HighPerformance Buildings For Developing Economies, 2014, ASHRAE, April2014c.

[13] Vaisala Humidity Calculator for Windows, Version 3.0, 2013. URL, www.vaisala.com/humiditycalculator.

[14] ASHRAE, 2009 ASHRAE Handbook: Fundamentals, SI ed., American Society ofHeating, Refrigerating and Air-conditioning Engineers, Atlanta, GA, 2009.

[15] J.F. Nicol, I.A. Raja, A. Allaudin, G.N. Jamy, Climatic variations in comfortabletemperatures: the Pakistan projects, Energy Build. 30 (3) (1999) 261e279.

[16] F. Nicol, M. Humphreys, Derivation of the adaptive equations for thermalcomfort in free-running buildings in European standard EN15251, Build. En-viron. 45 (1) (2010) 11e17.

[17] SI/ASHRAE, Standard 55e2013. Thermal Comfort Conditions For Human Oc-cupancy, ASHRAE, Atlanta, 2013.

[18] M. Humphreys, H. Rijal, J. Nicol, Updating the adaptive relation betweenclimate and comfort indoors; new insights and an extended database, Build.Environ. 63 (2013) 40e55.

[19] R Core Team, R: A Language and Environment for Statistical Computing, RFoundation for Statistical Computing, Vienna, Austria, 2014. URL, http://www.R-project.org/.

[20] D.A. Kenny, Statistics for the Social and Behavioral Sciences, Little Brown,Boston, 1987.

[21] F.H. Mallick, Thermal comfort and building design in the tropical climates,Energy Build. 23 (3) (1996) 161e167.

[22] M. Indraganti, Using the adaptive model of thermal comfort for obtainingindoor neutral temperature: ndings from a eld study in Hyderabad, India,Build. Environ. 45 (3) (2010) 519e536.

[23] H.H. Liang, T.P. Lin, R.L. Hwang, Linking occupants' thermal perception and

building thermal performance in naturally ventilated school buildings, Appl.Energy 94 (2012) 355e363.

[24] C.P. Chen, R.L. Hwang, W. Liu, W.M. Shih, S.Y. Chang, The inuence of air-conditioning managerial scheme in hybrid-ventilated classrooms on stu-dents' thermal perception, Indoor Built. Environ. (2014), http://dx.doi.org/10.1177/1420326X14530587.

[25] F. Nicol, Adaptive thermal comfort standards in the hotehumid tropics, En-ergy Build. 36 (7) (2004) 628e637.

[26] H. Zhang, E. Arens, W. Pasut, Air temperature thresholds for indoor comfortand perceived air quality, Build. Res. Inf. 39 (2) (2011) 134e144.

[27] S. Singh, P. Chani, S. Kulkarni, Implication of building energy modeling (BEM)and adaptive model to assess the efciency of multi storied apartments incomposite climate of north India, in: Proceedings of 8th Windsor Conference:Counting the Cost of Comfort in a Changing World, NCEUB, Windsor, UK,2014.

[28] R.L. Hwang, M.J. Cheng, T.P. Lin, M.C. Ho, Thermal perceptions, generaladaptation methods and occupant's idea about the trade-off between thermalcomfort and energy saving in hotehumid regions, Build. Environ. 44 (6)(2009) 1128e1134.

[29] F. Nicol, M. Humphreys, Maximum temperatures in European ofce buildingsto avoid heat discomfort, Sol. Energy 81 (3) (2007) 295e304.

[30] S.P. Corgnati, R. Ansaldi, M. Filippi, Thermal comfort in Italian classroomsunder free running conditions during mid seasons: assessment throughobjective and subjective approaches, Build. Environ. 44 (4) (2009) 785e792.

[31] A.G. Kwok, C. Chun, Thermal comfort in Japanese schools, Sol. Energy 74 (3)(2003) 245e252.

[32] L. Dias Pereira, D. Raimondo, S.P. Corgnati, M. Gameiro da Silva, Assessment ofindoor air quality and thermal comfort in Portuguese secondary classrooms:methodology and results, Build. Environ. 81 (2014) 69e80.

[33] M. Indraganti, R. Ooka, H.B. Rijal, Thermal comfort in ofces in summer:ndings from a eld study under the setsuden conditions in tokyo, japan,Build. Environ. 61 (2013) 114e132.

[34] S. Schiavon, K.H. Lee, Dynamic predictive clothing insulation models based onoutdoor air and indoor operative temperatures, Build. Environ. 59 (2013)250e260.

A.K. Mishra, M. Ramgopal / Building and Environment 92 (2015) 396e406406

A thermal comfort field study of naturally ventilated classrooms in Kharagpur, India1. Introduction2. Methodology2.1. The classroom and the subjects2.2. Collection of survey data2.2.1. Indoor environmental parameters2.2.2. Subjective questionnaire2.2.3. Outdoor temperature data

3. Results and discussions3.1. Analysis of thermal sensation votes3.1.1. Regression neutral temperature3.1.2. Comfort temperature using Griffiths equation3.1.3. Sweating sensation votes

3.2. Acceptability of thermal conditions3.2.1. Thermal comfort votes and thermal acceptance3.2.2. Thermal preference and preferred temperature3.2.3. Diurnal range acceptability for temperature variation

3.3. Observations regarding adaptive actions3.3.1. Use of fans and windows3.3.2. Clothing adaptation

3.4. A comparison with findings from the thermal comfort surveys in undergraduate laboratory

4. ConclusionAcknowledgementsReferences