8/17/2019 J. 2008 Indoor Air

1/20

Forty years of Fanger s model of thermal comfort:comfort for all?

Introduction

Thermal comfort contributes to overall satisfaction,

well-being and performance. Comfort is an importantparameter in the building design process as modernman spends most of the day indoors. In the 1970s and1980s, the development and usage of energy balancemodels of the human body came within the focus of human biometeorology (Ho p̈pe, 1997). The mostimportant contributor was P.O. Fanger (1934–2006),who created a predictive model for general, or whole-body, thermal comfort during the second half of the1960s from laboratory and climate chamber research.In that period, environmental techniques were improv-ing, wealth increased and workers wanted the bestindoor environment, while at the same time offices were

growing larger (McIntyre, 1984). With his work,Fanger wanted to present a method for use by heatingand air-conditioning engineers to predict, for any type

of activity and clothing, all those combinations of thethermal factors in the environment for which thelargest possible percentage of a given group of peopleexperience thermal comfort (Fanger, 1967). It provideda solution for predicting the optimum temperature fora group in, for instance, an open plan office, whichcould be provided by architects and engineers (McIn-tyre, 1984). This new predicted mean vote (PMV)model has become the internationally accepted modelfor describing the predicted mean thermal perceptionof building occupants.

Fanger (1970) dened PMV as the index thatpredicts, or represents, the mean thermal sensation

Abstract The predicted mean vote (PMV) model of thermal comfort, createdby Fanger in the late 1960s, is used worldwide to assess thermal comfort. Fangerbased his model on college-aged students for use in invariant environmentalconditions in air-conditioned buildings in moderate thermal climate zones.Environmental engineering practice calls for a predictive method that is appli-cable to all types of people in any kind of building in every climate zone. In thispublication, existing support and criticism, as well as modications to the PMVmodel are discussed in light of the requirements by environmental engineeringpractice in the 21st century in order to move from a predicted mean vote to

comfort for all. Improved prediction of thermal comfort can be achievedthrough improving the validity of the PMV model, better specication of themodel s input parameters, and accounting for outdoor thermal conditions andspecial groups. The application range of the PMV model can be enlarged, forinstance, by using the model to assess the effects of the thermal environment onproductivity and behavior, and interactions with other indoor environmentalparameters, and the use of information and communication technologies. Evenwith such modications to thermal comfort evaluation, thermal comfort for allcan only be achieved when occupants have effective control over their ownthermal environment.

J. van HoofFaculty Chair of Demand Driven Care, Faculty of HealthCare, Hogeschool Utrecht, Utrecht, The Netherlands,Faculty of Architecture, Building and Planning,Technische Universiteit Eindhoven, Eindhoven, TheNetherlands

Key words: Thermal comfort; PMV model; Validity;

Adaptive; Personal control.J. van HoofLectoraat Vraaggestuurde ZorgFaculteit GezondheidszorgHogeschool UtrechtBolognalaan 1013584 CJ UtrechtThe NetherlandsTel.: +31 30 2585268Fax: +31 30 2540608e-mail: joost.vanhoof@hu.nl

Received for review 20 May 2007. Accepted forpublication 30 December 2007.

Indoor Air (2008)

Practical ImplicationsThe paper treats the assessment of thermal comfort using the PMV model of Fanger, and deals with the strengths andlimitations of this model. Readers are made familiar to some opportunities for use in the 21st-century informationsociety.

Indoor Air 2008; 18: 182–201www.blackwellpublishing.com/inaPrinted in Singapore. All rights reserved

2008 The AuthorJournal compilation Blackwell Munksgaard 2008

INDOOR AIRdoi:10.1111/j.1600-0668.2007.00516.x

182

8/17/2019 J. 2008 Indoor Air

2/20

vote on a standard scale for a large group of persons forany given combination of the thermal environmentalvariables, activity and clothing levels. PMV is based onFanger s comfort equation (Fanger, 1967). The satisfac-tion of the comfort equation is a condition for optimalthermal comfort of a large group of people, or, whenmost of this group experiences thermal neutrality, and

no local discomfort exists. Based on PMV, the predictedpercentage of dissatised (PPD) can be determined

PPD ¼ 100 95 eð 0:03353 PMV4 0:2179 PMV 2Þ ð1Þ

(Equation 1). Fanger foresaw applications for thecomfort equation in thermostats, the design of lifesupport systems, work units and in relation to mete-orology. For application in environmental engineering,PMV was needed to assess a given room climate interms of deviations from an optimum thermal situa-tion. This became of great importance after the oil

crisis, when the focus of research abandoned thecentral optimum, and began to explore the edges of comfort, searching for how cold or warm it could getbefore getting uncomfortable (McIntyre, 1984). ThePMV model has since been the basis of numerousstudies and experiments worldwide. As with anymodel, it has been subject to criticism and support.

The PMV model applies to healthy adult people andcannot, without corrections, be applied to children,older adults and the disabled. The model has beenglobally applied for almost 40 years throughout allbuilding types, although Fanger was quite clear thathis PMV model was intended for application by theheating, ventilation and air-conditioning (HVAC)industry in the creation of articial climates incontrolled spaces (de Dear and Brager, 2002; Fanger,1970). Fanger s PMV model is adopted by many(inter)national standards and guidelines, for instance,ISO 7730, ASHRAE Standard 55, and CEN CR 1752,for providing an index of thermal comfort, and, evenafter the latest rounds of thermal comfort standardrevisions, is still the official tool to evaluate thermalcomfort, although a new adaptive model developed inthe 1990s by Brager and de Dear was incorporatedalongside the PMV model as an optional method.

Given the passing away of Fanger in September 2006,numerous (eld) validation studies on thermal comfort,the emergence of adaptive thermal comfort models andthe ever growing need and search for improved comfortfor all building occupants, it is time to give consider-ation to Fanger s contribution, or legacy, to the worldthermal comfort research, and to explore what hismodel might bring us in the near future. The central goalof this article is to provide an overview of the develop-ments in thermal comfort practice since the introductionof the PMV model by Fanger in 1970. It explores themeaning and potential of the PMV model by literatureresearch for past and present use, and future application

by, indoor air scientists, environmental engineers, andthe HVAC industry. Fanger s elaborate work in theeld of indoor air quality (Fanger, 2004a, 2006) is notpart of the scope of this paper. The objectives of this publication, which have been chosen in accordancewith questions from engineering practice, are to:(i) summarize criticism regarding the PMV model and

its input parameters, and (ii) the validity of the model,while at the same time, (iii) extending its applicationdomain, and (iv) improving individual comfort.

Literature review

The literature search was divided into several catego-ries: (i) studies regarding the validity of the PMVmodel, (ii) studies regarding the model s input para-meters, (iii) studies regarding the application of themodel in relation to building type and climate, (iv) theuse of the PMV model for special groups, (v) studies

regarding the new applications of the model, and (vi)individualized thermal comfort.Electronic databases of scientic publications that

were searched include ScienceDirect, PubMed, andWeb of Science. Also, selected journals were searchedfor relevant articles: Building and Environment,Energy and Buildings, International Journal of Bio-meteorology, ASHRAE Transactions, Indoor Air,Proceedings of Indoor Air and Healthy Buildings.The university libraries of Eindhoven and Delft werealso browsed for non-digitally available books anddissertations. Articles providing relevant backgroundinformation were also taken into account.

Basis of the PMV model

The human body produces heat, exchanges heat withthe environment, and loses heat by diffusion andevaporation of body liquids. During normal activitiesthese processes result in an average core body temper-ature of approximately 37 C (Prek, 2005). The body stemperature control system tries to maintain thesetemperatures even when thermal disturbances occur.The human body should meet a number of conditionsin order to perceive thermal comfort. According to

Fanger (1970), these requirements for steady-statethermal comfort are: (i) the body is in heat balance,(ii) mean skin temperature and sweat rate, inuencingthis heat balance, are within certain limits, and (iii) nolocal discomfort exists. Local discomfort to be avoidedincludes draughts, radiant asymmetry, or temperaturegradients. Moreover, high frequency of temperatureuctuation should be minimized as well. In order todescribe these physical processes, Fanger derived hiscomfort equation (Fanger, 1967) based on college-agestudents exposed to steady-state conditions in a climatechamber for a 3-h period in winter at sea level(1,013 hPa) while wearing standardized clothing and

Fanger s model of thermal comfort

183

8/17/2019 J. 2008 Indoor Air

3/20

performing standardized activities, for use withintemperate climate zones (Fanger, 1970). Although thecomfort equation may probably be applied in thetropics as well, Fanger (1970) stated this needed furtherresearch. Also, there were no reliable correctionsavailable for extreme climate conditions.

For practical applications, in which subjects do not

feel neutral, an extension of the comfort equation wasneeded. By combining data ( n = 1396) from variousstudies, Fanger expanded his comfort equation into thecurrent PMV model. This thermal sensation indexpredicts the mean thermal sensation vote for a largegroup of persons and indicates the deviation frompresumed optimal thermal comfort, i.e., thermoneu-trality. Results of the PMV model are expressed on the7-point ASHRAE scale of thermal sensation. Thecentral three categories of this scale are labeled slightlycool, neutral, and slightly warm. It is generallyaccepted that a person with a thermal sensation in one

of these three categories considers his environmentacceptable, and that someone voting in one of the fourouter categories is dissatised with his thermal envi-ronment (McIntyre, 1984).

The PMV model includes all the major variablesinuencing thermal sensation and quanties the abso-lute and relative impact of six factors of which airtemperature, mean radiant temperature, air velocityand relative humidity are measured, and activity leveland clothing insulation are estimated using tables.Activity level is measured in terms of metabolic rate, ormet units, and clothing insulation in clo units (Gaggeet al., 1941). To insure a comfortable indoor environ-ment, PMV should be kept 0 with a tolerance of ±0.5scale units (ISO, 2005a). In his dissertation, Fangerstated that the PMV model was derived in laboratorysettings and should therefore be used with care forPMV values below ) 2 and above +2. Especially on thehot side, Fanger foresaw signicant errors.

Criticism of the PMV model

Since the introduction of the PMV model, numerousstudies on thermal comfort in both real-life situationsand in climate chambers have been conducted. The

model s validity and application range were subject tostudy. Many studies have given support to the PMVmodel while others showed discrepancies (Bentonet al., 1990). Criticism involves various aspects, forinstance, the model as a whole, its geographic appli-cation range, application in various types of buildings,and the model s input parameters.

Validation

The process of validation of PMV for everyday userequires theresults of many eld studies, covering indoorenvironmental conditions found in buildings across the

world. An early eld validation by Howell and Kennedy(1979) showed that PMV and PPD provide just a rstapproximation to the prediction of thermal comfort innatural settings. The three middle categories of theASHRAE 7-point scale of thermal sensation seem to benotentirely valid, andmodal comfort is somewhat closerto thecooler-than-normalposition. Croomeet al. (1993)

suggest that the PMV model underestimates thermalimpressions and undervalues the swings of these impres-sions, due to the assumption of steady-state laboratoryconditions in the derivation of the model, an oversim-plicationof metabolic rates, and thesensitivity of PMVto clo-values. When deriving his model, Fanger consid-ered his test persons to be dissatised if they judged thethermal environment as cold, cool, warm, or hot, or outof the range of ) 2 to +2 on the ASHRAE 7-point scaleof thermal sensation. However, data referred to byFangershow that discomfort already arises in somewhatcooler and warmer than neutral conditions, and grad-

ually increases.Results of the PMV model are interpreted as what ahypothetical average person will feel, or as the averageresponse of large group of people experiencing thesame conditions. The model was designed to predictthe latter. Within a large group, optimal thermalconditions are likely to vary between individuals by upto 1.15 K according to Fanger and Langkilde (1975),or up to 1.0 scale unit (on the ASHRAE 7-point scaleof thermal sensation) according to Humphreys andNicol (2002). Fountain et al. (1996) stated that indi-vidual differences are frequently greater than 1.0 scaleunit when people are exposed to the same environment(inter-individual variance). In addition, how a personfeels in the same environment from day-to-day canalso vary on the order of 1.0 scale value (intra-individual variance). This corresponds to approxi-mately 3 K; the full width of the comfort zone ineither summer or winter (Fountain et al., 1996).Therefore, it is impossible to exactly predict thermalcomfort for individuals, and that is the reason thecomfort zone is as wide as it is, and why it isunreasonable to expect all people to be satised withina centrally controlled environment, even when thethermal conditions meet current standards.

A validation by Humphreys and Nicol (2002) basedon a meta-analysis of data from the ASHRAE RP-884database (de Dear et al., 1997), showed that PMV wasa valid index. It predicted thermal sensation within0.11 ± 0.01 scale units of the observed votes. Whenseparate database samples were analyzed, 33 out of atotal of 41 samples showed evidence of bias in PMV,often exceeding 0.25 scale units, and reaching as muchas 1.0 scale unit. Humphreys and Nicol (2002) foundthat the more thermal conditions moved away fromneutral, the larger the bias. PMV is only reliablebetween ) 0.5 and 0.5, unlike the range of validitystated by Fanger and given in ISO 7730

van Hoof

184

8/17/2019 J. 2008 Indoor Air

4/20

() 2.0 < PMV < 2.0). This makes the PMV modelseem unrealistic for application in eld settings.

Researchers have compared the results of the PMVmodel against the votes of subjects (actual mean vote,AMV) on the ASHRAE 7-point scale of thermalsensation. Doherty and Arens (1988) reviewed climatechamber studies and found that there were discrepan-

cies between predicted and actual thermal sensation aslarge as 1.3 scale units. A comparison of the actual andPMVs by Parsons (2002) showed that the PMVspredicted AMVs within 0.5 of a scale value for neutraland warm conditions. However, in slightly cool to coolconditions subjects were between neutral and slightlycool, and hence, warmer than predicted. Parsonsconcluded that changes in thermal comfort responsesin neutral and slightly warm environments are smalland unlikely to be of practical signicance. Ka ḧko n̈en(1991) assessed thermal comfort in 17 enterprises at 129work sites in shops, stores, and offices. The estimated

PMV indicated that the thermal environment was toowarm, and in fact, the calculated PMVs were usuallylower than the estimated ones. In a previous study byKa ḧko n̈en and Ilmarinen (1989) in 13 shops and stores,workers were asked to rate their subjective thermalsensations ( n = 96). No correlation was foundbetween calculated PMV and subjective ratings.

Various researchers found a different relationshipbetween PMV and PPD than the one described byFanger (Figure 1). Araú jo and Araú jo (1999) found aPPD of 47.5% corresponding to a PMV of 0.0 basedon an assessment ( n = 1866 votes, mean clo 0.6,naturally ventilated) carried out in secondary schoolbuildings and a university in Brazil

PPD ¼ 100 52:5 e ð0:03353 PMV4 þ 0:2179 PMV 2Þ ð2Þ

(Equation 2). Korean climate chamber research byYoon et al. (1999) found a minimum PPD of 18%matching with a PMV of ) 0.8

PPD ¼ 11 :37 PMV 2 þ 18:34 PMV þ 24:42 ð3Þ

(Equation 3) ( n = 40 students, mean clo 0.4, summer,mean met 1.2). People involved in the study preferredcooler temperatures. A German study by Mayer (1997)based on approximately 100 subjects found a minimumPPD of 16% instead of 5%, corresponding to a PMVof 0.5 instead of 0.0

PPD ¼ 100 84:3 e½0:01ðPMV 0:4Þ4þ 0:5479 ðPMV 0:4Þ2

ð4Þ(Equation 4). There is a shift in the curve to the warmside and to higher PPD for the cold side. de PaulaXavier and Roberto (2000) produced a thermal sensa-tion index S

S ¼ 0:219t0 þ 0:012 RH 0:547va 5:83 ð5Þ

(Equation 5) similar to PMV, through eld studies in aschool in Brazil (>1200 votes). This thermal sensationindex corresponds to a PPD I

I ¼ 18:94S 2 0:24S þ 25:41 ð6Þ

(Equation 6). If the thermal sensation index is neutral,the PPD is 25.4%. The different outcomes of thesestudies may be due to the building ventilation type(natural ventilation/eld study vs. air-conditioning/climate chamber), and the small number of subjectsthat may have a large inter-individual spread inthermal preferences. The studies mentioned do illus-

trate that the outcomes of the PMV model should beused with caution, as errors may occur when using thislarge-group model for small samples.

Field validations of the PMV model have raiseddiscussions on the validity and reliability of the modelfor use in real-world settings. Large discrepancies werefound between outcomes of the PMV model and

0

10

20

30

40

50

60

70

80

90

100

–3.00 –3.00–2.00 –2.00–1.00 –1.00–2.50 –2.50–1.50 –1.50–0.50 –0.500.00PMV [-]

P P D [ % ]

FangerYoonAraújo & AraújoMayerDe Paula Xavier & Roberto

Fig. 1 Relationship between predicted mean vote (PMV) and predicted percentage of dissatised (PPD) and other thermal sensationindices

Fanger s model of thermal comfort

185

8/17/2019 J. 2008 Indoor Air

5/20

comfort votes of subjects. Also, various researchersfound a different relationship between thermal votesand dissatisfaction than the one described by Fanger,in smaller climate chamber and eld studies.

Thermoneutrality vs. preferred thermal sensation

The PMV model is based on the concept of thermo-neutrality, and PMV expresses how warm or cooloccupants perceive the thermal environment. Thisthermal sensation is a measure of how occupantsperceive a certain thermal condition. Other measuressuch as thermal satisfaction, thermal acceptability,thermal comfort, and thermal preference, also dealwith the appropriateness of a given thermal condition.Values ranging from ) 1 to 1 on the ASHRAE scale of thermal sensation are often considered to reectsatisfaction with a given thermal environment. How-ever, direct measures of thermal satisfaction and

acceptability are not incorporated in the currentPMV model. Thermal neutrality does not necessarilyhave to correspond to the desired or preferred thermalsensation, for instance, when people like it slightlywarm (Humphreys and Hancock, 2007). In addition,Fountain et al. (1996) mentioned two fundamentalassumptions that need further discussion: (i) optimumtemperature corresponds to a neutral thermal sensa-tion, and (ii) the notion of acceptability is associatedwith specic thermal sensations on the ASHRAE7-point scale of thermal sensation. It is suggestedthat thermal sensation cannot be assumed to bethe equivalent of evaluative measures mentionedabove, and that the PMV model requires a critical

interpretation and application (Fountain et al., 1996).These considerations are supported by other studies(Table 1).

Various eld and climate chamber studies showeddifferences between both neutral and preferred tem-peratures, as well as differences between eld andclimate chamber research. In a review by Humphreys

(1994), the observed neutral temperatures in eightclimate chamber studies were between 0.8 K lower and3 K higher than predicted by the PMV model. In acombined eld ( n = 497) and climate chamber(n = 455) study in the UK (season and buildingventilation type unknown), Williams (1995) found thatpeople preferred lower temperatures. Field datashowed a 2.5 K difference between the air temperatureat which people felt most comfortable and thatpredicted using PMV, whereas the climate chamberonly gave a 0.5 K difference compared to PMV. In areview of eld studies in air-conditioned and naturally

ventilated buildings, Brager and de Dear (1998) foundthat the observed and predicted neutral temperatureswere overestimated by as much as 2.1 K (Melbourne,summer), and underestimated by up to 3.4 K (Bang-kok, time of year not mentioned); both extremes innaturally ventilated buildings.

Various studies from Asia on neutral and preferredtemperatures found that these temperatures were oftenhigher than based on the PMV model, and foundconsiderable between-subjects differences (Table 2).For instance, Humphreys (1994) provides an overviewof mainly Asian studies on this matter, and mentionsthat one study on Malays in London and Malaysia hada difference in neutral temperature as large as 3 K.

Table 1 Selection of studies showing that thermal comfort does not only occur around thermal neutrality

Reference Location Setting Time of year Subjects Results

Schil ler (1990) San FranciscoBay Area, USA

Field Winter andsummer 1987

304 subjects(187 females, 117 males)in 10 office buildings(2342 visits)

People generally felt warmer than predicted more often than theyfelt cooler, in both seasons. 22% (winter) and 15% (summer)of people voting +2 and +3 on a 7-point scale of thermalsensation stated to be moderately to very comfortable on a6-point scale of thermal comfort. At the same time, 17% (winter)and 34% (summer) of people voting ) 2 and ) 3 stated to be moderatelyto very comfortable too

Busch (1990, 1992) Bangkok,Thailand

Field Hot seasonand wetseason 1988

Over 1,100 Thai officeworkers in AC andNV buildings

36% of people voting neutral preferred to feel warmer or cooler

Brager et al. (1994) Variouscountries

Review offield studies

Summers andwinters

Office workers Of occupants voting in the outer categories of the thermalsensation scale () 3,) 2,+2,+3), 0 to 50% preferred nochange to their thermal environment. Also, of occupantsvoting in the same outer categories() 3,) 2,+2,+3, other studies reviewed), 3–66% were comfortable

Paciuk and Becker (2002) Haifa, Israel Field Summer 117 AC and NVowner-occupieddwellings

Of residents of NV homes, voting between ) 1 and +1(on the 7-point scale of thermal sensation), about 47%were not comfortable (as rated on a 5-point thermal comfort scale).In AC homes, the majority of dwellers voted to be comfortable (81%),although thermal votes ranged from ) 1 to +3 on the 7-point scale

AC, air-conditioned; NV, naturally ventilated.

van Hoof

186

8/17/2019 J. 2008 Indoor Air

6/20

8/17/2019 J. 2008 Indoor Air

7/20

In a normally distributed population, about 95% of neutral temperatures of subjects will be within therange of four times the standard deviation (Wyon,1994). Wyon (1994) further stated that in the deriva-tion of the PMV model this range was 6.4 K, whereas aFrench study found a 10.4 K range, and an Americanstudy even a 13.6 K range. Wyon (1994) continues by

stating that when individual variation is so large, [ … ]the practical value of estimating group mean neutraltemperatures is very limited. It is more important toprovide the practical means for individuals to adjusttheir own heat loss, than to try to guess what thecollective average of their individually preferred tem-peratures might be.

Differences found between eld and climate chamberresearch are explained by the greater variation inoccupants clothing patterns in real buildings, comparedto the standard uniforms used in laboratory studies. Toohigh estimations by the PMV model might be found in

erroneousvalues formetabolism andclothing insulationduring the derivation of the comfort equation. Fanger stest persons wore the KSU clothing combination, of which the insulation was estimated at 0.6 clo. Accordingto Nishi et al. (1975), this turned out to be only 0.32. Awrong estimation in clothing insulation, however, can-not be the only explanation.

In conclusion, (i) thermal neutrality is not necess-arily ideal for a signicant number of people, and(ii) preferences for non-neutral thermal sensationsare common, very asymmetrically around neutrality,and in several cases are inuenced by season. Also,(iii) thermal sensations outside of the three centralcategories of the ASHRAE 7-point scale of thermalsensation do not necessarily reect discomfort for asubstantial number of persons.

Differences among building types

At present, the PMV model is applied globallythroughout every type of building. However, the modelwas developed from laboratory studies, and the effectsof building type were not considered. In practice,differences in the perception of the thermal environ-ment were found among occupants of naturally

ventilated (also referred to as free-running), fullyair-conditioned and mixed mode or hybrid buildings(de Dear et al., 1997). It was found that for naturallyventilated buildings the indoor temperature regardedas most comfortable increased signicantly in warmerclimatic contexts, and decreased in colder climate zones(de Dear, 2004). This is reected by numerous studies,which showed that the neutral temperature observed inair-conditioned buildings differs from that observed innaturally ventilated buildings in the same climaticcontext (Table 3).

de Dear et al. (1997) showed that occupants of air-conditioned buildings are twice as sensitive to

temperature changes as those of naturally ventilatedbuildings. In contrast, Humphreys and Nicol (2002)found no difference between the accuracy of thepredicted responses in air-conditioned and free-run-ning buildings because of noticeable variations inthermal environments of the buildings. PMV provideslarger accuracy in terms of the average preferences for

air-conditioned buildings due to narrower bandwidthsof the thermal environment. It is wrong to assume thatthe use of PMV is without exception correct when usedfor people in air-conditioned buildings (Humphreysand Nicol, 2002). According to de Dear and Brager(1998), the PMV model is not applicable to naturallyventilated buildings, because it only partly accounts forthermal adaptation to the indoor environment. Or, asstated by de Dear (2004), the PMV model has lostmuch of its predictive ability in naturally ventilatedbuildings. Therefore, a model of adaptive thermalcomfort has been proposed for free-running buildings,

which relates the neutral temperature indoors to themonthly average temperature outdoors (de Dear andBrager, 1998). This model is incorporated into ASH-RAE (2004) Standard 55 as an optional method,applicable in naturally ventilated office buildings forpeople engaged in sedentary activity, when outdoortemperatures are between 10 and 33 C. Above 33 Cthe only predictive tool available is the PMV model,which is unreliable for predicting thermal responses of people in free-running buildings (de Dear et al., 1997;Humphreys and Nicol, 2002). In temperate climatezones, the comfort zones of adaptive models and thePMV model are largely superimposed over one another(van Hoof and Hensen, 2007). An adaptive modelproposed earlier for fully air-conditioned buildings hasnot been included in the ASHRAE Standard.

Fanger and Toftum (2002) see the exclusion of thesix input parameters of PMV that have an impact onthe human heat balance as an obvious weakness of theadaptive model. Although the PMV model is oftenreferred to as a static model, it turns out to be anadaptive model as it accounts for behavioral adjust-ments and fully explains adaptation occurring in air-conditioned buildings (de Dear and Brager, 2002).Fanger (2004b) expressed some thoughts on the

linguistic problems concerning the term

adaptive .‘‘Adaptation should be a process of machines adaptingto human requirements and ergonomics, not theadaptation of humans to technology.’’

Although the PMV model is still applied throughoutevery type of building all across the globe, it was foundthat (i) for naturally ventilated buildings the indoortemperature regarded as most comfortable increasessignicantly in warmer climatic contexts, and decreasesin colder climate zones, and that (ii) the neutraltemperature observed in air-conditioned buildingsdiffers from that observed in naturally ventilatedbuildings in the same climatic context. This led to the

van Hoof

188

8/17/2019 J. 2008 Indoor Air

8/20

8/17/2019 J. 2008 Indoor Air

9/20

development of a model of adaptive thermal comfort,which is now an optional method alongside the PMVmodel for application to naturally ventilated buildings.

Improving the PMV model

Many researchers have tried to nd ways to optimize

the use of the PMV model, for instance, by improvingthe model and enlarging its application domain.Improving the original PMV model is twofold; rst,by correcting or adjusting the model itself, second byintroducing methods to increasing the accuracy of themodel s input parameters.

Adjusting the model

The ability to incorporate expectation into the evalua-tion of thermal comfort is an areawhere the heat balancemodels offer no information. The difference does not lie

in physiology but in expectation (Fountain et al., 1996).Fanger and Toftum (2002) acknowledged some of thelimitations of the PMV model, as Fanger already did inhis 1970 dissertation, and proposed an extension to free-running buildings in warm climates by introducing twocorrection factors. First, factor e should be introducedto account forexpectancy of theoccupantsdependingonprevious thermal experiences. This factor is to bemultiplied with PMV to reach the mean thermalsensation vote of the occupants. Based on professional judgment, e variesbetween 1 and0.5 (1 forbuildingswithHVAC, for buildings without HVAC e is assumed todependon theannualduration of thewarm weather, andif these buildings can be compared with other buildingswithHVACin the region) (Table 4). The secondfactorisa reduced activity level, as it is slowed down uncon-sciously by people feeling warm as a form of adaptation.The new extension predicts the actual votes well, and itacknowledges the importanceof expectationsaccountedfor by the adaptive model, while at the same time notabandoning the current PMV model s input parameterswith direct impact on the human heat balance.

Humphreys and Nicol (2002) provide a statisticalmethod named PMV new , which is illustrative only, andis not recommended for using as a replacement index

T a

b l e

3

( C o n t i n u e d )

R e f e r e n c e

L o c a t i o n

S e t t i n g

T i m e o f y e a r

S u b j e c t s

R e s u l t s

F a t o e t a l

. ( 2 0 0 4 )

, C o n t e

a n d F a t o ( 2 0 0 0 )

B a r i , I t a l y

F i e l d

S u m m e r ( 1 9 9 5 , 1 9 9 9 ) , a n d w i n t e r

( 1 9 9 6 , 2 0 0 0 )

U n i v e r s i t y s t u d e n t s

. S a m p l e s i z e : 4 2 3

i n 1 9 9 5

, 1 0 3 4 i n 1 9 9 6

, 2 5 0 i n 1 9 9 9

,

a n d 1 3 3 i n 2 0 0 0

. B u i l d i n g t y p e

( t w o m o d e s ) : A C i n w i n t e r ,

N V i n s u m m e r

N e u t r a l t e m p e r a t u r e s w e r e 2 4

. 4 C i n s u m m e r 1 9 9 5

, 2 6

. 3 C i n s u m m e r

1 9 9 9

, 2 0

. 7 C i n w i n t e r 1 9 9 6

, a n d 2 0

. 6 C

i n w i n t e r 2 0 0 0

. O c c u p a n t s o f

N V b u i l d i n g s ( s u m m e r ) r e g a r d e d a 3 . 3 K a n d 2 . 1 K b a n d w i d t h t o b e

a c c e p t a b l e c o m p a r e d t o 3 . 6 K i n A C b u i l d i n g s ( w i n t e r )

Y a m t r a i p a t e t a l

. ( 2 0 0 5 )

T h a i l a n d ( C h i a n g M a i

, B a n g k o k &

M a h a s a r a k h a m

, P r a c h u a b k i r i k h a n )

F i e l d

A u g u s t 2 0 0 1

U s e r s o f A C b u i l d i n g s i n p r i v a t e

a n d p u b l i c s e c t o r s ( n =

1 5 2 0 )

T h e n e u t r a l t e m p e r a t u r e o f p e o p l e w i t h a

p o s t - g

r a d u a t e e d u c a t i o n l e v e l

w a s t h e l o w e s t a r o u n d 2 5

. 3 C , w h i l e t h a t o f t h e o t h e r g r o u p s ( g r a d u a t e

a n d s c h o l a r ) w a s h i g h e r a t 2 6

. 0 C . F o r p e o p l e w i t h a i r - c o n d i t i o n i n g

h o m e , t h e d i f f e r e n c e b e t w e e n n e u t r a l t e m p e r a t u r e o f e v e r y e d u c a t i o n

l e v e l i s r a t h e r s m a l l ( 0

. 3 K )

. H o w e v e r

, f o r t h e o t h e r g r o u p ( n o

a i r - c o n d i t i o n i n g ) , t h e d i f f e r e n c e o f 0 . 9 K i s l a r g e r

. P e o p l e w i t h h i g h e r

e d u c a t i o n a l d e g r e e s a r e f o u n d t o p r e f e r l o w e r i n d o o r t e m p e r a t u r e

c o m p a r e d t o t h e l e s s - e

d u c a t e d

Table 4 Expectancy factors in relation to building classification and expectation (Fangerand Toftum, 2002)

ExpectationBuildingclassification

Expectancyfactor e

High Non-HVAC in regions where HVAC buildingsare common. Warm periods occurring brieflyduring summer season

0.9–1.0

Moderate Non-HVAC in regions with some HVAC buildings.Warm summer season

0.7–0.9

Low Non-HVAC in regions with few HVAC buildings.Warm weather during all seasons

0.5–0.7

van Hoof

190

8/17/2019 J. 2008 Indoor Air

10/20

PMV new ¼ 0:8ðPMV DPMV Þ ð7Þ

DPMV ¼ 4:03 þ 0:0949 T 0 þ 0:00584

RH þ 1:201ðMI clÞ þ 0:000838 T 2out ð8Þ

(Equations 7 and 8). They state that the appropriate

method to revise the PMV model would be to revise itspsychophysical and physiological construction, ratherthan to make empirical statistical adjustments. Therevised PMV has a reduced bias against all thecontributing variables and the ranges of validity aregreatly extended. The bias against PMV new itself ismuch less than the PMV, extending the range of application about three-fold, though it is still unsatis-factory in cold environments. Holmé r (2004) statedthat for use in moderately cold and cold environments,the basic comfort criteria may also apply, but thecomplex heat transfer through multilayer clothing

should be considered in a more adequate way. Amodied PMV model was proposed by modication of the sweating criteria and some of the heat transferequations. The suggested criteria provide more realisticand reliable prediction of heat balance and conditionsfor comfort in colder environments.

Gagge et al. (1986) proposed PMV* for any dry orhumid environment by replacing operative temperaturein Fanger s comfort equation by standard effectivetemperature (SET*). Two classic indices of tempera-ture sensation and warm discomfort (PMV and DISC)are combined as generalized index PMV*, which isdirectly related to new effective temperature (ET*).PMV* is responsive to thermal stress caused by heatload, as well as to the physiological heat strain causedby changing humidity of the environment and bychanging vapor pressure permeability properties of clothing worn. Gagge s PMV* model has been used ina limited number of studies, for instance, by Priantoand Depecker (2003). Unfortunately, the modicationis unjust because Gagge and colleagues assumed thatFanger used different comfort criteria for cold andheat, instead of one. Also, Prek (2005) made acomparison between exergy consumption and theexpected level of thermal comfort. Since this analysis

used the two-node model and special attention wasgiven to heat and mass transfer, a new PMV index(PMV*) was used. This model can be used for any dryor humid environment by replacing the operativetemperature in Fanger s comfort equation with rationalET*. According to Prek, this modication representsan improvement as it includes the physiologic heatstrain caused by the relative humidity and vaporpermeability properties of clothing.

The calculation of PMV is complicated as it isnonlinear and iterative. Sherman (1985) proposed amodel based on Fanger s original work to calculatePMV without iteration, by linearizing the radiation

exchange terms, simplifying the convection coefficient,and using dew-point temperature. Sherman indicatedthat these simplications should not affect the precisionof PMV only when the occupants are near the comfortzone. Federspiel (1992) proposed another thermalsensation index V , which is a modication of thePMV index. He simplied the derivation by supposing

that radiative exchange and heat transfer coefficient arelinear, assuming that bodily heat production andclothing insulation are constant, and that the occu-pants are in a thermally neutral condition. V is alinearly parameterized function of the four environ-mental variables and becomes nonlinear when activitylevel or clothing insulation is changed, which is alimitation to the applicability of the model.

Another methodforadjusting thecurrent PMVmodelis through fuzzy logics, for instance,by Feriadi andHien(2003). Hamdi et al. (1999) proposed a fuzzy thermalsensation index based on the original work of Fanger

that canbe used in feedback HVAC control applicationswith online calibration and without requiring anysimplication. The difference between PMV and fuzzyPMV values was found to be lower than 0.05 for anystate of the six input parameters, which indicates highprecision of the proposed fuzzy PMV model.

Although many modications to the original PMVmodel have been proposed and tested to date, none of these developments have yet found widespread appli-cation in environmental engineering practice. The mostpromising application is the extension of the PMVmodel to free-running buildings in warm climates byFanger and Toftum (2002), which incorporates expec-tation into the evaluation of thermal comfort. Engi-neering practice has not yet come up with practicalmethods to use the extension.

The PMV models input parameters

For an appropriate use of the PMV model, theparameters involved in the calculation of PMV needto be within certain limits (Table 5) according to ISO7730. The validity intervals stated in this standardcontribute largely to the biases in PMV (Humphreysand Nicol, 2002). Bandwidths of comfort parameters

matching correct PMV are narrower than that stated inISO 7730. The current intervals are too large, and nearthe extremes there is not sufficient data to conrm thevalidity. However, fortunately the PMV model has abias-free range of conditions (Humphreys and Nicol,2002). In tropical areas air temperatures frequently riseabove 30 C, and air velocities in excess of 1 m/s areseen. In Jakarta, for instance, the maximum average airtemperature is 33 C (Karyono, 2000). In case of naturally ventilated buildings, the PMV model cannotbe used without error according to ISO 7730. Similarsituations are found for architecture in deserts andvernacular buildings in cold climate zones.

Fanger s model of thermal comfort

191

8/17/2019 J. 2008 Indoor Air

11/20

Uncertainty in thermal comfort predictions andmeasurements of physical parameters has also beeninvestigated by de Wit (2001). Error can also be foundwhen instruments are malfunctioning or have not beencalibrated correctly. Early thermal comfort researchmade use of instruments with wider ranges of errorthan instruments used today, particularly instrumentsmeasuring air velocity. Many studies used simple setsof instruments, often measuring on one position in aroom. Uncertainty remains as to whether the indoorclimate measured was representative for the room orbuilding as a whole. ISO 7726 (ISO, 1998) prescribesthat PMV shall be assessed on a height of 0.6 m forsedentary activity. This height is actually indicative, asbody length differs per individual and not all chairs areof the same dimensions. Variations in heights cancontribute to small variations in PMV, which arelargely negligible. In climate chamber settings, physicaland personal parameters can be controlled for moreeasily. In eld setting, all variables tend to show greatervariation. These differences might be the key contri-butor to discrepancies found between PMV and AMV.According to Fanger (1994), it is essential that all fourenvironmental factors are properly measured and that

a careful estimation is made of the activity andclothing, when making a fair comparison. Poor inputdata will provide a poor prediction.

de Dear and Brager (1998) claimed that the accuracyof the PMV model is not as good as that of theadaptive model, as some of its input parameters are notaccurately specied. Activity level and clothing insula-tion are the two main variables that affect theuncertainty of the PMV index (Chamra et al., 2003),and are likely to contribute to differences betweenactual and predicted thermal sensations. Because PMVis much more sensitive to air velocity, metabolism andclothing insulation than for air or radiant temperature,

estimations with some deviation can lead to largedifferences in PMV and thus considerable misinterpre-tations. In eld studies, it is not possible to accuratelydetermine clothing insulation. Therefore, someresearchers specify an average clothing value for thetotal research population, even though there are inter-individual differences, while others make use of elab-orate garment tables for better group averages or forindividual calculations. The calculated clothing insula-tion differs by as much as 20% depending on the sourceof the tables and algorithms commonly used (Brageret al., 1994). Fanger claims this is due to the diversityof methods used to estimate clo-values (de Dear andBrager, 1998). In eld settings, there is always somebody movement increasing air velocity, particularly innon-sedentary activity. Some studies involving thermalmanikins suggest that body movement has a verylimited impact on clothing insulation (Olesen andNielsen, 1984; Olesen et al., 1982). Other studies usinghumans suggested that there are effects of bodymovement (Chang et al., 1988; Vogt et al., 1984). Theclo-value is reduced by increased air movement (Kim-ura and Tanabe, 1993), which can be as much as 25%(Tanabe et al., 1993). Havenith et al. (2002) showed

that effects of body motion and air movement onclothing insulation are so big that they should beaccounted for in comfort models to be physicallyaccurate. However, effects on dry heat exchange aresmall for stationary, light work at low air movement.Also, algorithms for convective heat exchange inprediction models should be reconsidered. Other fac-tors that might contribute to deviations in clo-valuesare body posture, material/fabric and cut, and – oftenomitted – the amount of chair insulation, whichdepends on geometry and material. McCullough et al.(1994) showed that an additional value of 0.15 cloneeds to be added to one s total clothing insulation for

Table 5 Validity intervals for PMV input parameters, taken and adapted from ISO 7730 (ISO, 2005a) and Humphreys and Nicol (2002)

ParameterISO 7730(ISO, 2005a)

Humphreys and Nicol (2002)

PMV freefrom bias if Comment

Clothing insulation 0–2 clo(0–0.310 m2K/W)

0.3 < I cl < 1.2 clo(chair included)

Overestimation of warmth of people in lighter and heavier clothing,serious bias when clothing is heavy. Little information

exists for conditions when I cl < 0.2 cloActivity level 0.8–4 met

(46–232 W/m2 )M < 1.4 met Bias larger with increased activity. At 1.8 met overestimation

sensation of warmth by 1 scale unitHypotheticalheat load

M ÆI cl < 1.2 unitsof metÆclo

Serious bias at 2 units

Air temperature 10–30C Overestimation warmth sensation t o > 27 C. At higher temperaturesbias becomes severe. Upper limit t o approx. 35C in ISO 7730Mean radiant

temperature10–40 C

Vapor pressure orrelative humidity

0–2.7 kPaor 30–70%

RH < 60% Suggested bias becomes important if p a > 2.2 kPa

Air velocity 0–1 m/s v a < 0.2 m/s Overestimation warmth sensation v a > 0.2 m/s. Underestimationcooling effect increased v a

van Hoof

192

8/17/2019 J. 2008 Indoor Air

12/20

the impact of a chair. CEN Report 1752 (CEN, 1998)mentions an additional 0.1–0.3 clo for chairs. A goodexample of the impact of omitting the thermal insula-tion posed by chairs is the aforementioned study byBrager et al. (1994). Data from the Californian officeswere reanalyzed, accounting for an extra chair insula-tion of 0.15 clo. Also, adjustments were made based on

recent clothing insulation tables (an additional 0.1 clo).The results of the new analysis showed a bettercorrespondence between AMV and PMV.

For precise comfort assessments, a precise measureof activity level or metabolic rate is needed. In eldstudies, measuring physiologic parameters can beinvasive and time-consuming tasks. As with clothinginsulation, activity level is often equal for a group as awhole. According to Havenith et al. (2002), the meth-ods for determining metabolic rate provided in ISO8996 (ISO, 2004) do not provide sufficient accuracy toclassify buildings to within 0.3 PMV scale units. In

order to improve metabolic rate estimation based onISO 8996, more data and detail are needed for activitieswith a metabolic rate below 2 met units. According toGoto et al. (2002), the metabolic rate is inuenced by aperson s body mass, blood ow, and tness. Wyon(1975) found that performing mental tasks can increaseactivity levels up to 1.3 met units, while Humphreys(1994) states that the vigor with which certain activitiesare carried out can affect met values. Rowe (2001)conducted a study involving 144 occupants lling outactivity checklists ( n = 1,627). The mean activity levelwas 1.2 met units, with inter-individual and over-timedifferences ranging from 1.0 to 1.9 met.

Errors in clothing insulation and activity level arefound to contribute to the inaccuracy of PMV,particularly in eld studies. As differences betweenAMV and PMV are larger than could reasonably beattributed to these errors, other factors are probablyinvolved as well. As comment on de Dear and Brager(1998), Fanger says: A careful assessment of an‘‘average’’ building occupant s metabolic rate couldbe in error as much as plus or minus 0.3 met units andthat what researchers might rate as a ‘‘typical officetask’’ 1.2 met is in fact a ‘‘siesta’’ 0.9 met in certainparts of the world. In view of this enormous difficulty

in getting the PMV-inputs right, we feel that [ … ] anadaptive model is probably not such a bad thing afterall…

For an appropriate use of the PMV model, theparameters involved in the calculation of PMV need tobe within certain limits, but the bandwidths of comfortparameters are subject to discussion. The quality of predictions and evaluations also depends on theaccuracy of input parameters, especially clothing insu-lation and activity level, and measurements. Wrongestimations of clothing insulation and activity levelmay even be contributing to the inaccuracy of PMV,particularly in eld studies.

Extensions to the application domain

The current PMV model is used for assessing andpredicting thermal comfort in indoor spaces, oftenoffices with presumably stable thermal conditions. Newinsight is gained for application in semi-outdoorenvironments and in transient thermal conditions, as

well as on interaction between various indoor environ-mental parameters and the use of the PMV model fornon-office workers, such as older adults. Moreover, themodel holds interesting promises for other applica-tions, such as in a control strategy for optimizingproductivity and in relation to environmental psycho-logical factors. The latter also play a major role inoffering modes for individualized thermal comfort.

The PMV model for (semi-)outdoor use. Potter and deDear (2000) asked themselves a question: Why doholiday makers deliberately seek out thermal environ-

ments, that would rate ‘‘off the scale’’ if they wereencountered indoors? In their investigation of outdoorscenarios, the researchers found that outdoor neutraltemperatures were about 3 K higher than for indoorspaces. Indoor and outdoor thermal sensations areperceived in a different manner.

Current comfort standards do not clearly addresstransitional spaces, locations where the physical envi-ronment bridges between the interior and exteriorenvironment. Outdoor situations are not includedeither. When designing the PMV model, it was neverintended for use in such spaces. Clothing insulation isgenerally higher in transitional spaces, as people areoften traveling to another destination and dress for theoutdoor weather. Field research from Thailand ontransitional spaces by Jitkhajornwanich et al. (1998)showed that the neutral temperature was 27.1 C whilethe thermal acceptability was between 25.5 and 31.5 C,whereas the expected neutral temperature, based on thePMV model, was 26.7 C. Chun et al. (2004) found thatthe PMV model cannot be used for thermal comfortpredictions in transitional spaces, because of unstableand dynamic physical conditions and met values. Thisis especially true for outdoor pavilions, where PMVoften travels out of the validity range of ) 2 <

PMV < +2. These two studies from Asia suggestthat PMV cannot be used as an evaluative parameter insemi-outdoor spaces.

Thermal sensation outdoors is perceived differentlyfrom that of indoors. This implies that models forindoor thermal comfort are not applicable to theoutdoor settings (Ho p̈pe, 2002). By parameterizationof the complex radiation uxes outdoors, Fanger scomfort equation became a universally acceptablemodel, which is used today under the name of Klima-Michel-Modell (Ho p̈pe, 1997). Various similarefforts have been undertaken by various researchersworldwide.

Fanger s model of thermal comfort

193

8/17/2019 J. 2008 Indoor Air

13/20

8/17/2019 J. 2008 Indoor Air

14/20

based on tests with approximately 1,300 students. Realbuildings involve much larger and diverse samples of real occupants as opposed to college-age subjects (deDear, 2004). In his experiments, Fanger (1970) foundthat women were more sensitive to uctuations inoptimum temperature than men. A comparison of theactual and PMVs by Parsons (2002) showed that for

cool conditions (PMV = ) 2.0), females tended to besignicantly cooler ( P < 0.05) than males, and theresponses of the females were close to the PMVs. Also,Nakano et al. (2002) found differences in neutraltemperature between males and females (Table 2).Fanger (1970) concluded that women are more sensi-tive to deviations than men. Parsons concluded that foridentical clothing and activity, there are few genderdifferences in thermal comfort responses for neutraland slightly warm conditions, although females tend tobe cooler than males in cool conditions.

In a study by Karjalainen (2007), gender differences

in thermal comfort were examined by a quantitativeinterview survey ( n = 3094), and by controlled exper-iments. Females were less satised with room temper-atures, preferred higher room temperatures, and feltboth uncomfortably cold and uncomfortably hot moreoften than males.

Thermal comfort is not only important in worksettings. In our ageing society, the voice of older adultsis growing ever louder and there is a tendency of creating a more tting building stock to accommodatethis group of people. This is also expressed in the desirefor excellent indoor environments, which offer thermalcomfort in compliance with physical needs.

Fanger (1970) validated his model for older adults bycarrying out experiments involving 128 older subjectsto study the inuence of age and aging. Some scientists,most of them from Japan, have since repeated studieson the effects of aging on comfort sensation. Moststudies conrm that in principle older adults do notperceive comfort differently from younger groups.However, older adults have a lower activity level andbasal metabolism, resulting in a preference of higherambient temperature (Havenith, 2001; van Hoof andHensen, 2006). These differences may also result fromdifferences in clothing between the age groups.

Physically disabled persons respond as predicted bythe PMV model, although the range of responses forphysically disabled is much greater than that of healthypeople at PMVs of ) 1.5 and 0 (Webb and Parsons,1998). Multiple sclerosis patients are shown to havelarger actual PPD than based on Fanger s PPD, andthe preferred environment (PMV = 0) is slightlywarmer than 23 C. Further work is needed to qualifywhether preferred environments match that of PMV +1 and PMV ) 1 and to identify whether anyof the factors such as age, duration of disability, andmedication affect the actual mean vote (Webb et al.,1999). According to Haghighat et al. (1999), physically

handicapped persons demonstrate thermoregulatoryabnormalities in the affected portion of their bodies.Measured thermal sensations for physically handi-capped persons were found to have no generaltendencies. In 2002, Parsons stated that there are fewgroup differences between thermal comfort require-ments of people with and without physical disabilities.

However, there is a greater necessity to considerindividual requirements for people with a physicaldisability.

The technical specication ISO/TS 14415 (ISO,2005b) deals with thermal comfort requirements of people with special needs and impairments. However,this technical specication is far from complete, as sofew studies have been conducted in the eld. Moreresearch is needed on the requirements of particularlyvulnerable groups of older adults, such as those withdementia, who are hypothesized to perceive theirthermal environment differently because of atrophy

of their brain tissue, and who are unable to express theneed for cooler or warmer conditions. Similarly, peoplewith psychiatric impairments may respond differentlyto the thermal environment.

A promising new application of the PMV model foruse for older adults is relating (extreme) climaticconditions to increased mortality rates in summer(Luscuere and Borst, 2002). National governments inEurope are stressing the importance of air-conditioningas a protective measure, but the impact on thermalcomfort and preferences remain unclear. In nursinghomes, the PMV model could be used for determiningcomfort of all occupants, mortality/morbidity riskresulting from heat and cold, and even productivitylevel of nursing staff in relation to the thermalconditions.

Besides the aged and aging, more research is neededon infants and children, who have a different meta-bolism and posture, especially in light of studies of theindoor environment in daycare centers and primaryschools in relation to well-being and school produc-tivity.

Productivity and the role of environmental psychology.The application range of the PMV model can be

enlarged, for instance, for the prediction of productiv-ity in relation to the thermal climate. Productivity onthe work oor is affected by indoor environmentalparameters, including noise and thermal discomfort.Kosonen and Tan (2003 and 2004) made a PMV-basedproductivity model to estimate the effects of differentthermal conditions on productivity, as task-relatedperformance is correlated with the human perceptionof thermal environment. The theoretical optimalproductivity occurs when the PMV value is ) 0.21(around 24 C). The normally accepted PMV valueof +0.5 leads to about 12% productivity loss (PL)in thinking and 26% in typing. The mathematical

Fanger s model of thermal comfort

195

8/17/2019 J. 2008 Indoor Air

15/20

expressions of PL for typing and thinking tasks inoffices (boundary constraints: PMV ) 0.21 and PMV+1.28), are given by

PL typing task ¼ 60:543 PMV 6 þ 198:41

PMV 5 183:75 PMV 4 8:1178

PMV3

þ 50:24 PMV

2

þ 32:123

PMV þ 4:8988

ð9Þ

PL thinking ¼ 1:5928 PMV 5 1:5526 PMV 4

10:401 PMV 3 þ 19:226 PMV 2

þ 13:389 PMV þ 1:873

ð10Þ

(Equations 9 and 10). More studies have been carriedout on the inuence of thermal comfort and the indoorenvironment on productivity that promise interestingeld of application for the PMV model in the nearfuture.

The perception of thermal comfort is closely relatedto environmental psychology. The important role of environmental psychology has been discussed earlierby Heijs and Stringer (1987) and de Dear (2004). Forthe integrity of person–environment relationships to bepreserved, thermal comfort research must be con-ducted within natural contexts (de Dear, 2004). In air-conditioned settings, the occupant is regarded as apassive recipient of the thermal stimuli presented by

their environment, with the latter being assessedsubjectively against very specic expectations of whatsuch an environment should be like (de Dear, 2004). Aeld study by Howell and Stramler (1981) on the roleof cognitive psychological variables ( n = 260) foundthat there was a substantial impact of cognitivevariables on thermal sensation judgments. The mostimportant determinant of perceived comfort was,perhaps not surprisingly, perceived temperature. Roh-les (1980) found that adding wood paneling, carpets,comfortable furniture and soft lights, without anychanges in thermal parameters in a climate chamber,made people feel warmer. Even providing information(both true and false) could inuence feelings of warmthand comfort. In winter, people preferred warm tem-peratures to cold temperatures, while in summer theopposite situation was found.

Other psychological factors related to thermal cli-mate are manifold. Anderson et al. (1996) reportedeffects of extreme temperatures on arousal, cognition,and affect. High temperatures increase levels of aggres-sion. Simister and Cooper (2005) reported evidencethat violent crime rates vary from month to month in aseasonal pattern, and this evidence is consistent withthe hypothesis that the seasonal pattern is caused by

temperature. This seasonal pattern might be due to theside effects of stress hormones, which the bodygenerates to cope with thermal stress.

Perhaps, input from environmental psychology canone day be combined with the PMV model, as a meansto quantify the thermal environment, to predict thelevel of arousal and aggression, for instance, in

hospitals, jails, or police offices.

Individual control. It is generally believed that mostcomfort standards specify conditions that providecomfort for everybody. The PMV model is oftenapplied bearing this idea in mind. Unfortunately,fullling the criteria does not mean 100% acceptabilitybecause of individual differences in preferred temper-ature. When the building is later constructed andoccupied, it is sometimes believed that the buildingshould be operated strictly according to [such criteria].This was never the intention. The standard should be a

servant, not a master (Fanger, 1994). Selection of anacceptable PPD often depends on economy andtechnical feasibility (Fanger, 1984, 2001; Olesen et al.,2001). In practice this means quite a number of dissatised persons (if the thermal climate is main-tained in compliance with the international standards,there will be up to 10% dissatised people due towhole-body discomfort), while few seem to be ready tocharacterize the indoor environment as outstanding(Fanger, 2001).

Individual differences in comfort temperature are solarge that it would be a historical mistake to developnew models and standards focusing on guaranteeingthermal comfort for a group as a whole (Wyon, 1994).In air-conditioned environments, the occupants are notexpected to attenuate it, or their responses to it, in anysignicant way. There is no need for prescriptivestandards if individual control over the thermal climateis provided, allowing all occupants to be satised(Fanger, 2001; Fountain et al., 1996), for instance, bybeing able to slightly adjust temperature at the work-place level. This requires systems designed for userinvolvement, both physical (individual temperaturecontrol), or social, where the physical conditions areseen as a natural consequence of the situation, rather

than arbitrarily imposed (McIntyre, 1984).Palonen et al. (1993) recommended that a tempera-ture range of 20–24 C without individual temperaturecontrol is too wide a range from the point of thermalcomfort. Room temperature should stay between 21and 23 C if no control is available. The best solution isan individual temperature control that gives thepossibility to adjust the room temperature at22 ± 2 C. In rooms with conventional mixing of ventilation air, the air should be kept at a moderatelylow temperature that corresponds to the coolesttemperature preferred by any of the occupants(Fanger, 2001). All other occupants should have access

van Hoof

196

8/17/2019 J. 2008 Indoor Air

16/20

to additional personalized local heating to obtain theirown preferred operative temperature. These smallheating ows should be provided by radiation orconduction in order to keep the air for inhalation cool(Fanger, 2001). Individual thermal control by airmovement should be avoided because of the risk of draught (Fanger, 2001). In a study investigating the use

of thermostats, Karjalainen (2007) concluded that eventhough females were more critical of their thermalenvironments, males used thermostats in householdsmore often than females. Would this mean that inoffice settings women are more reluctant to controltheir indoor environment and that men are morewilling to operate technological means? More researchis needed on means to individualize thermal comfort.Having control over the environment is a very effectiveway to limit negative health effects of stress, as one canuse external coping strategies (Vroon, 1990). Providingall kinds of pseudo solutions, for instance, imitation

thermostats, may have a positive effect in the begin-ning, but can have serious counterproductive effectssome time later, as occupants feel not taken seriously(Vroon, 1990). A negative placebo effect ( nocebo effect)can emerge resulting in a further increase in dissatis-faction with the environment (Vroon, 1990). In linewith these statements, attention needs to be paid to thenature of building structures in relation to preferredfast gratication. Changes to the indoor temperaturedepend both on the system of heating and coolingavailable, and the thermal mass of the structure. Inthermal activation of building parts, changes in tem-perature may be slow. Thermostat control may nothave an immediate effect and therefore lead to moredissatisfaction and complaints.

The PMV model can be used for providing a basicthermal environment that meets the preferences of agroup of people. On an individual level, people shouldbe equipped with simple means to control the directthermal environment in order to maximize satisfactionand comfort.

ICT and thermal comfort. The emergence of informationand communication technologies (ICT) in buildingservices engineering will lead to the development of

real-time thermostats that incorporate software basedon the comfort equation and the PMV model. Suchsystems will be at the basis of environmental systemsthat control the indoor climate, and offer means forcontrolling local environments to individual needs andin relation to the thermal conditions in adjacent spaces.Personal user proles, based on a one s preferredtemperature, and one s sensitivity to deviations on thelower and warmer side, may be incorporated in time tocontrol the overall indoor climate. The speed at whichlocal environments can be adapted to these needs will beessential in the success of individualized solutions. Theintroduction and development of these systems will

undoubtedly be closely linked to energy conservationand management. Fanger (2006) stated that compro-mises are needed between energy conservation and bestpossible indoor environments. New developments inthe eld of indoor air quality may provide room forsimultaneous energy savings. According to de Dear andBrager (2002), a more integrative view of optimizing the

indoor environment, energy consumption, and produc-tivity is needed. Based on building automation andadvances in sensor technology, we will see the moni-toring and control of the indoor environment of officesin relation to achieving optimal conditions for maxi-mum productivity. In other types of buildings, otherpsychophysiological parameters may be coupled to thePMV model when controlling the indoor environment.Also, the previously described modied PMV modelsmay be used for computerized control strategies.

The current use of building performance simulationtools providing iso-PMV lines as output are an example

of misuse of ICT, as PMV is not the property of a singlepoint in space, but rather a whole-body sensation. Theincreasing important of ICT solutions means that forcomplex and detailed simulations, new multi-segmentalmodels of human physiology and thermal comfort areincreasingly used, for instance, models by Fiala et al.(2001) and Huizenga et al. (2001). These models arebased on earlier work by Gagge and Stolwijk (seeGagge et al. (1971) for an example of their occasionalcollaborations) from the 1980s, and represent the bodyas a sum of separate body parts. These models candescribe local effects due to non-uniform environmentsand/or non-uniform clothing coverage. Although theyhave found little application outside of the laboratory,the new models are able to deal with real-world thermalcomplexities, and may therefore gain importance.

Conclusions

After almost 40 years of practical experience with thePMV model, the foundation of the model and its use,particularly in air-conditioned buildings, are growingever more solid. In four decades, numerous studieshave been conducted on thermal comfort in both eldsettings and in climate chambers. These studies have

provided valuable insight into the principles on whichthe PMV model is based.Thermal neutrality is not necessarily the ideal ther-

mal condition as preferences for non-neutral thermalsensations are common. At the same time, very lowand very high PMV values do not necessarily reectdiscomfort for a substantial number of persons.

The PMV model is applied throughout every type of building all across the globe and its use is prescribed inthermal comfort standards. However, it was found thatfor naturally ventilated buildings the indoor tempera-ture regarded as most comfortable increases signi-cantly in warmer climatic contexts, and decreases in

Fanger s model of thermal comfort

197

8/17/2019 J. 2008 Indoor Air

17/20

colder climate zones. This important milestone inthermal comfort research led to the development of an adaptive comfort model. The support for the PMVmodel is still larger than that for this adaptive model,which was expressed by the latest round of standardrevisions that only included the adaptive model as anoptional method in very restricted conditions.

Many researchers have tried to improve the PMVmodel, and enlarge its application domain. Improve-ments can be made by correcting or adjusting themodel itself, or by introducing methods to increasingthe accuracy of the model s input parameters.Although many modications to the original PMVmodel have been proposed and tested to date, none of these developments have yet found widespread appli-cation in environmental engineering practice. Theaccuracy of the PMV model and the quality of itspredictions gain in strength when users are able tomore precisely determine the model s input parameters,

particularly clothing and activity levels. This requiresmore sophisticated tables and tools for use in dailypractice, and in case of office settings, should alsoprovide accurate data on the insulation of varioustypes of chairs.

The current PMV model is used for assessing andpredicting thermal comfort in indoor spaces, oftenoffices. The model, however, holds interesting promisesfor other applications that go beyond the originalscope, such as optimizing productivity. Thermal com-fort plays a major role among other indoor environ-mental parameters. The exact interactions betweenthese parameters are not yet fully understood, and

subject of many studies, as are transient thermalconditions. Also, groups other than college-age stu-dents and occupants of office buildings need to bestudied in terms of thermal preferences.

There is a tendency, inuenced by the emancipationof building occupants and energy conservation, to setthe general indoor climate in accordance with require-

ment of the PMV model, while on an individual level,people are equipped with simple means to control thedirect thermal environment. The development of indi-vidualized thermal comfort in work spaces takes placesimultaneously with the computerization of the workoor. In the near future, the thermal environment maybe set in compliance with personal preferences storedin an automated user prole that allows for furtherpersonal ne-tuning, energy conservation, and couplingwith other indoor environmental parameters.

Fanger s PMV model is almost 40 years old and stillused as the number one method for evaluating thermal

comfort. Even the incorporation of an adaptive model inAHSRAE Standard55 did littleharm to the status of thePMV model, as it is still recognized as valid for allbuildings types. The quality of the outcomes of the PMVmodel is as good as that of its input parameters speciedby the model s user. The many input parameters of thecomfort equation make PMV an index with greatpotential for the computerized 21st century in whichoccupants have increased demands regarding the indoorenvironment. In combination with means for individualcontrol, engineers should be able to create indoorenvironments that provide nearly 100% acceptabilityand comfort to all, based on the work of Fanger.

References

Ahmed, Z.N., Howarth, A.T. and Zunde,J.M. (1990) Thermal comfort studies andapplications for Dhaka, Build. Serv. Eng.Res. Technol. , 11 , 105–108.

Alm, O., Witterseh, T., Clausen, G., Toftum,J. and Fanger, P.O. (1999) The impact onhuman perception of simultaneousexposure to thermal load, low-frequencyventilation noise and indoor air pollution.In: Raw, G., Aizlewood, C. and Warren,P. (eds) Proceedings of Indoor Air 99,Vol. 5, Edinburgh, 270–275.

Anderson, C.A., Anderson, K.B. and Deu-ser, W.E. (1996) Examining an affectiveaggression framework: weapon and tem-perature effects on aggressive thoughts,affect, and attitudes, Pers. Soc. Psychol.Bull. , 22 , 366–367.

Araú jo, V.M.D. and Araú jo, E.H.S. (1999)The applicability of ISO 7730 for theassessment of the thermal conditions of users of the buildings in Natal-Brazil. In:Raw, G., Aizlewood, C. and Warren, P.

(eds) Proceedings of Indoor Air 99, Vol. 2,Edinburgh, 148–153.

ASHRAE (2004) Thermal Environmental Conditions for Human Occupancy ,Atlanta, American Society of Heating,Refrigerating and Air-ConditioningEngineers (ANSI/ASHRAE Standard55-2004).

Benton, C.C., Bauman, F.S. and Fountain,M.E. (1990) A field measurement systemfor the study of thermal comfort, ASH-RAE Trans. , 96 , 623–633.

Brager, G.S. and de Dear, R.J. (1998) Ther-mal adaptation in the built environment:a literature review, Energy Build. , 27 , 83– 96.

Brager, G.S., Fountain, M.E., Benton, C.C.,Arens, E.A. and Bauman, F.S. (1994) Acomparison of methods for assessingthermal sensation and acceptability in thefield. In: Oseland, N.A. and Humphreys,M.A. (eds) Proceedings Thermal Comfort,Past, Present and Future , Garston, 17–39.

Busch, J.F. (1990) Thermal responses to theThai office environment, ASHRAE Trans. , 96 , 859–872.

Busch, J.F. (1992) A tale of two populations:thermal comfort in air-conditioned andnaturally ventilated offices in Thailand,Energy Build. , 18 , 235–249.

CEN (1998) Ventilation for Buildings: DesignCriteria for the Indoor Environment ,Brussels, European Committee for Stan-dardization (Technical report CR 1752).

Chamra, L.M., Steele, W.G. and Huynh, K.(2003) The uncertainty associated withthermal comfort, ASHRAE Trans. , 109 ,356–365.

Chan, D.W.T., Burnett, J., de Dear, R.J. andNg, S.C.H. (1998) A large-scale survey of thermal comfort in office premises inHong Kong, ASHRAE Trans. , 104 , 1172– 1180.

Chang, S.K.W., Arens, E. and Gonzalez,R.R. (1988) Determination of the effect of walking on the forced convection heat

van Hoof

198

8/17/2019 J. 2008 Indoor Air

18/20

transfer coefficient using an articulatedmanikin, ASHRAE Trans. , 94 , 71–82.

Chun, C., Kwok, A. and Tamura, A. (2004)Thermal comfort in transitional spaces – basic concepts: literature review and trialmeasurement, Build. Environ. , 39 , 1187– 1192.

Conte, E. and Fato, I. (2000) Thermal com-fort in mixed mode buildings. In: Seppa

¨-

nen, O. and Sa ẗeri, J. (eds) Proceedings of Healthy Buildings 2000 , Vol. 2, Espoo,581–586.

Croome, D.J., Gan, G. and Awbi, H.B.(1993) Thermal comfort and air quality inoffices. In: Jaakkola, J.J.K., Ilmarinen, R.and Seppa n̈en, O. (eds) Proceedings of Indoor Air 93, Vol. 6, Helsinki, 37–42.

de Dear, R. (2004) Thermal comfort inpractice, Indoor Air , 14 , s32–s39.

de Dear, R.J. and Auliciems, A. (1985) Va-lidation of the predicted mean vote modelof thermal comfort in six Australian fieldstudies, ASHRAE Trans. , 91 , 452–468.

de Dear, R.J. and Brager, G.S. (1998) De-veloping an adaptive model of thermalcomfort and preference, ASHRAE Trans. ,104 , 145–167.

de Dear, R.J. and Brager, G.S. (2002) Ther-mal comfort in naturally ventilatedbuildings, revisions to ASHRAE Stan-dard 55, Energy Build. , 34 , 549–561.

de Dear, R.J., Leow, K.G. and Ameen, A.(1991) Thermal comfort in the humidtropics – part I, climate chamber experi-ments on temperature preferences in Sin-gapore, ASHRAE Trans. , 97 , 874–879.

de Dear, R.J., Brager, G.S. and Cooper, D.(1997) Developing an Adaptive Model of

Thermal Comfort and Preference , FinalReport ASHRAE RP- 884, Atlanta,American Society of Heating, Refrigerat-ing and Air-Conditioning Engineers.

Doherty, T.J. and Arens, E. (1988) Evalua-tion of the physiological bases of thermalcomfort models, ASHRAE Trans. , 94 ,1371–1385.

Ealiwa, M.A., Taki, A.H., Howarth, A.T.and Seden, M.R. (1999) Field investiga-tion of thermal comfort in both naturallyand mechanically ventilated buildings inGhadames, Libya. In: Raw, G., Aizle-wood, C. and Warren, P. (eds) Proceed-ings of Indoor Air 99, Vol. 2, Edinburgh,

166–171.Fan, Y., Lang, S. and Xu, W. (1993) Fieldstudy on acceptable thermal conditionsfor residential buildings in transitionzone of China. In: Jaakkola, J.J.K.,Ilmarinen, R. and Seppa n̈en, O. (eds)Proceedings of Indoor Air 93, Vol. 6,Helsinki, 109–114.

Fang, L., Clausen, G. and Fanger, P.O.(2000) Temperature and humidity:important factors for perception of airquality and for ventilation requirements,ASHRAE Trans. , 100 , 503–510.

Fanger, P.O. (1967) Calculation of thermalcomfort: introduction of a basic comfort

equation, ASHRAE Trans. , 73 , III.4.1– III.4.20.

Fanger, P.O. (1970) Thermal Comfort ,Copenhagen, Danish Technical Press.

Fanger, P.O. (1984) The philosophy behind acomfort standard. In: Berglund, B.,Lindvall, T. and Sundell, J. (eds) Pro-ceedings of Indoor Air 84, Vol. 1, Stock-holm, 91–98.

Fanger, P.O. (1994) How to apply modelspredicting thermal sensation and dis-comfort in practice. In: Oseland, N.A.and Humphreys, M.A. (eds) ProceedingsThermal Comfort, Past, Present and Future , Garston, 11–17.

Fanger, P.O. (2001) Human requirements infuture air-conditioned environments, Int.J. Refrigeration , 24 , 148–153.

Fanger, P.O. (2004a) Curriculum vitae,Indoor Air , 14 , s12–s17.

Fanger, P.O. (2004b) Oral communication ,Eindhoven, Technische UniversiteitEindhoven, June 10 2004.

Fanger, P.O. (2006) What is IAQ?, IndoorAir , 16 , 328–334.Fanger, P.O. and Langkilde, G. (1975)

Interindividual differences in ambienttemperatures preferred by seated persons,ASHRAE Trans. , 81 , 140–147.

Fanger, P.O. and Toftum, J. (2002) Exten-sion of the PMV model to non-air-con-ditioned buildings in warm climates,Energy Build. , 34 , 533–536.

Fato, I., Martellotta, F. and Chiancarella, C.(2004) Thermal comfort in the climaticconditions of Southern Italy, ASHRAE Trans. , 110 , 578–593.

Federspiel, C.C. (1992) User-adaptable and

Minimum-Power Thermal Comfort Con-trol , Cambridge MIT (Dissertation).Feriadi, H. and Hien, W.N. (2003) Model-

ling thermal comfort for tropics usingfuzzy logic. In: Hensen, J.L.M. andAugenbroe, G. (eds) Proceedings BuildingSimulation 2003 , Vol. 1, Eindhoven, 323– 330.

Fiala, D., Lomas, K.J. and Stohrer, M.(2001) Computer prediction of humanthermoregulation and temperatureresponses to a wide range of environ-mental conditions, Int. J. Biometeorol. ,45 , 143–159.

Fountain, M.E., Brager, G.S. and de Dear,

R.J. (1996) Expectations of indoor cli-mate control, Energy Build. , 24 , 179–182.Gagge, A.P., Burton, A.C. and Bazett, H.C.

(1941) A practical system of units for thedescription of the heat exchange of manwith his environment, Science , 94 , 428– 430.

Gagge, A.P., Stolwijk, J.A.J. and Nishi, Y.(1971) An effective temperature scalebased on a simple model of human phy-siological regulatory response, ASHRAE Trans. , 77 , 247–262.

Gagge, A.P., Fobelets, A.P. and Berglund,L.G. (1986) A standard predictive indexof human response to the thermal

environment, ASHRAE Trans. , 92 , 709– 731.

Goto, T., Toftum, J., de Dear, R. and Fan-ger, P.O. (2002) Thermal sensation andcomfort with transient metabolic rates,In: Levin, H. (ed.) Proceedings of IndoorAir 02, Vol. 1, Monterey, 1038–1043.

Haghighat, F., Donnini, G., Megri, A.C. andGiorgi, G. (1999) Prediction of thermalperception of disabled persons vs mea-surement. In: Raw, G., Aizlewood, C. andWarren, P. (eds) Proceedings of Indoor Air999, Vol. 2, Edinburgh, 113–118.

Hamdi, M., Lachiver, G. and Michaud, F.(1999) A new predictive thermal sensationindex of human response, Energy Build. ,29 , 167–178.

Havenith, G. (2001) Temperature regulationand technology, Gerontechnology , 1 , 41– 49.

Havenith, G., Holmé r, I. and Parsons, K.(2002) Personal factors in thermal com-fort assessment: clothing properties and

metabolic heat production, Energy Build. ,34 , 581–591.Heidari, S. and Sharples, S. (2002) A com-

parative analysis of short-term and long-term thermal comfort surveys in Iran,Energy Build. , 34 , 607–614.

Heijs, W. and Stringer, P. (1987) Comfort asa property of the dwelling: a conceptualanalysis, Neth. J. Housing Environ. Res. ,2, 331–356.

Hensen, J.L.M. (1990) Literature review onthermal comfort in transient conditions,Build. Environ. , 25 , 309–316.

Holme ŕ, I. (2004) Cold but comfortable?Application of comfort criteria to cold

environments, Indoor Air , 14 , 27–31.van Hoof, J. and Hensen, J.L.M. (2006)Thermal comfort and older adults, Ger-ontechnology , 4 , 223–228.

van Hoof, J. and Hensen, J.L.M. (2007)Quantifying of relevance of adaptivethermal comfort models in moderatethermal climate zones, Build. Environ. , 42,156–170.

Ho p̈pe, P. (1997) Aspects of human bio-meteorology in past, present and future,Int. J. Biometeorol. , 40 , 19–23.

Ho p̈pe, P. (2002) Different aspects of asses-sing indoor and outdoor thermal comfort,Energy Build. , 34 , 661–665.

Howell, W.C. and Kennedy, P.A. (1979)Field validation of the Fanger thermalcomfort model, Hum. Factors , 21 , 229– 239.

Howell, W.C. and Stramler, C.S. (1981) Thecontribution of psychological variables tothe prediction of thermal comfort judg-ments in real world settings, ASHRAE Trans. , 87 , 609–621.

Huizenga, C., Hui, Z. and Arens, E. (2001) Amodel of human physiology and comfortfor assessing complex thermal environ-ments, Build. Environ. , 36 , 691–699.

Humphreys, M.A. (1994) Field studies andclimate chamber experiments in thermal

Fanger s model of thermal comfort

199

8/17/2019 J. 2008 Indoor Air

19/20

comfort research. In: Oseland, N.A. andHumphreys, M.A. (eds) ProceedingsThermal Comfort, Past, Present and Future , Garston, 52–72.

Humphreys, M.A. (2005) Quantifying occu-pant comfort: are combined indices of theindoor environment practicable?, Build.Res. Inf. , 33 , 317–325.

Humphreys, M.A. and Hancock, M. (2007)Do people like to feel neutral ?: exploringthe variation of the desired thermal sen-sation on the ASHRAE scale’’, EnergyBuild. , 39 , 867–874.

Humphreys, M.A. and Nicol, J.F. (2002)The validity of ISO-PMV for predictingcomfort votes in every-day thermalenvironments, Energy Build. , 34, 667– 684.

Ishii, A., Iwamoto, S., Sakai, E., Katayama,T. and Tsutsumi, J. (1993) The effect of humidity reduction on thermal sensa-tions. In: Jaakkola, J.J.K., Ilmarinen, R.and Seppa n̈en, O. (eds) Proceedings of

Indoor Air

93, Vol. 6, Helsinki, 115–120.ISO (1998) Ergonomics of the Thermal Environment – Instruments for MeasuringPhysical Quantities , Geneva, Interna-tional Organization for Standardization(ISO 7726).

ISO (2004) Ergonomics of the Thermal Environment – Determination of MetabolicRate , Geneva, International Organizationfor Standardization (ISO 8996).

ISO (2005a) Ergonomics of the Thermal Environment – Analytical Determinationand Interpretation of Thermal ComfortUsing Calculation of the PMV and PPDIndices and Local Thermal Comfort Cri-

teria , Geneva, International Organizationfor Standardization (ISO 7730).ISO (2005b) Ergonomics of the Thermal

Environment – Application of International Standards to People with Special Requirements , Geneva, InternationalOrganization for Standardization (Tech-nical Specication ISO/TS 14415).

Iwamoto, S. and Yoneki, T. (1999) Indoorthermal environment in transient condi-tion. In: Raw, G., Aizlewood, C. andWarren, P. (eds) Proceedings of Indoor Air99, Vol. 2, Edinburgh, 178–183.

Jitkhajornwanich, K., Pitts, A.C., Malama,A. and Sharples, S. (1998) Thermal com-

fort in transitional spaces in the coolseason of Bangkok, ASHRAE Trans. ,104 , 1181–1193.

Ka ḧko n̈en, E. (1991) Draught, radiant tem-perature asymmetry and air temperature – a comparison between measured andestimated thermal parameters, Indoor Air ,1, 439–447.

Ka ḧko n̈en, E. and Ilmarinen, R. (1989)Assessing indoor thermal climate – a cri-tical discussion. Environ. Int. , 15, 531–535.

Karjalainen, S. (2007) Gender differences inthermal comfort and use of thermostats ineveryday thermal environments, Build.Environ. , 42 , 1594–1603.

Karyono, T.H. (1995) Hig