4064 (RP-781)

1997, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in ASHRAE Transactions 1997, Vol 103, Part 2. For personal use only. Additional distribution in either paper or digital form is not permitted without ASHRAEs permission.

Too

Marc E. Fountain, Ph.D. Associate MemberASHRAE ABSTRACT

Charlie Huizenga Member ASHRAE

software for possible inclusion in ASHRAE Standard 5.5 (ASHRAE 1992)o ASHRAEs goals included determining As part of a recent ASHRAE research project (781-RP), whichmodelsto incorporate and providinga user-friendly front thermal sensation prediction tool has been developed. end, a comparative analysis of the models,and informationthat paper introduces the tool, describes the componentthermal allowsa professionalwhois not necessarily involvedin thermal sensation models, and presents examplesof howthe tool can be used in practice. Since the main end product of the HVAC comfort research to apply the modelssuccessfully. The thermal #~dustryis the comfortof occupants indoors, tools for predictsensation prediction tool that resulted fromthe project will be ing occupant thermal response can be an important asset to describedin this paper. designers of indoor climate control systems. The software tool Eight existing physiologically based thermal comfort presented in this paper incorporates several existing models models and five existing non-physiologically based thermal for predicting occupantcomJbrt. comfortmodels wereidentified for possibleinclusion in the software tool (Table 1). A physiologically based thermal comfort INTRODUCTION modelis an algorithm that producesa predicted physiological Many state-of-the-art thermal sensation modelshave been state and predicted thermal comfortvote for a human exposedto developedfiom dozensof research projects conductedduring an indoor environment using certain physical parametersof the the last 35 years. Outof this enormous of work,twothermal body environment(and of the human)as input. Non-physiologically comfort prediction methods--rangers PMV-PPD (ranger basedthermalcomfortmodels statistical fits to data relating are 1970) and Gagges 2-Node(Gagge et al. 1986)--have been comfort indices to the physiological environment. A brief most widely used. These models predict the thermal comfort (including, but not limited to, thermalsensation) of humans. The description of the models is presented below; however, a International Standards Organization (ISO) has adopted the TABLE 1 PMV-PPD model in its thermal comfort standard 7730 (ISO Some Physiologically BasedThermalComfortiViodels 1984), while ASHRAE ET*(one of the indices calculated uses by the 2~Node model)to define the boundaries of the comfort Date Author Description zone in its thermal comfort Standard 55 (ASHRAE 1992). 1964 Wissler 225-node finite elementmodel Both the PMV-PPD 2-Nodemodelssolve heat balance and 1970 ranger PMV steady-state model equations for the human bodyand are generally implemented on Stolwijk 1970 25-node basic heat flow model a computer. Thesemodels in the public domain are availare and 1986" Gagge aL et 2-nodebasic heat flow model able to professionalsby request fromseveral different sources. 1990 de Dearand Ring40-layerfinite differenceskin model However, lack of user-friendly interfaces, lack of informathe tion on how interpret results, andlack of information which to on 1990 Int-Hout Modified PMV modelsto use in different situations wardoff many potential 1992 Jones and Ogawa 2-nodewith transient response users. Tanabe Modified Stolwijk model 1992 ASHRAE recently sponsored a research project (Fountain and Huizenga1995) to prepare thermal sensation prediction * Most recent iteration;many been have released.

Marc Fountain principal, Environmental E. is Analytics,Berkeley, Calif. CharlieHuizenga a researchspecialist at the Centerfbr Envi: is ronmental Design Research, Universityof California, Berkeley.

130

ASHRAE Transactions: Research

completediscussion of the models structure and their strengths and weaknesses beyond scopeof this paperand is available is the elsewhere (Fountain and Huizenga1995). In brief, these physiologicalthermalcomfortmodels haveat their core a statementabout the heat balanceof the human body. Humans gain heat from metabolismand lose heat due to respiration and evapotranspiration. In addition, dependingon the physical environment, they either gain or lose heat by conduction, convection, and radiation. The hypothalamus charged is with regulating heat gain and loss mechanisms maintain the to bodys core temperatureat 37C(98.6F). All the physiological models listed in Table 1 (except for Fangers PMV-PPD and PMV-IH) initial values for physiological constants and use physiologicalvariables and then iterate for a user-specifiedtime period. Eachiteration consists of establishing thermoreceptor signals to the brain, determining physiologicalresponses,calculating heat flows, calculating newcore and skin temperatures, and, finally, calculating the resulting thermoreceptor signals again, usually on a minute-by-minute basis. Fangers PMV-PPD model is also physiologically based, but, instead of iterating changing flowsfor a specific period heat of exposure, the iteration determinesclothing surface temperature, andthe convective transfer coefficientis basedin fixed heat heat flows Theequationuses a steady-state heat balancefor the human body and postulates a link between deviation from the minimum load on heat balance effector mechanisms--e.g, sweating, vasoconstriction, and vasodilation--and thermal comfort vote. The greater the load, the morethe comfort vote deviates from zero or "neutral" sensation. A modifiedform of PMV also considered for inclusion in the software tool. was PMV-IH (Int-Hout 1990) is the Fanger PMV calculation with modification in the wayheat is transferred through the skinclothing systemthat accountsfor the vaporresistance of clothing. Five non-physiologically based modelswere also considered for inclusion in the software: three empirical modelsand two adaptive models. The three empirical models are PD (Fangeret air 1988), or "predicted percent dissatisfied due draft," whichis a fit to data of persons expressing thermal discomfort to drafts; PS(Fountainet al. 1994),which a fit due is to data of comfortable persons choosing velocity levels; and air TS (Rohlesand Nevins1971), whichis a fit to data of thermal sensationas a linear functionof air temperature partial vapor and pressure. Thetwo adaptive modelsinclude variations in outdoor climate for determiningthermal preferencesindoors. Auliciems (Auliciems1983)neutral temperaturemodelfits sensation data based on field investigations of thermal comfort in Australia spanning several climates. Humphreys(Humphreys1978) neutral temperature equationis a fit to morethan 100,000 observations of sensation in climate-controlled and non-climatecontrolled buildings. Equations for PD, PS, and TS are as follows: 0622 PD= 3.413(34- Ta)(v- 005) +0.369vT,~(34 Ta)(v - 0622 005) ASHRAE Transactions: Research (1)

PS = 1.13 T~ - 0.24Top + 2.7~/~- 0.99v and TS = 0.245Ta+ 0.248p - 6.475.

(2)

(3)

The PDequation arises from two studies in which 100 people wereexposedto various combinations air temperature, of air velocity, and turbulence intensity. For each combination of conditions, the people were askedif they felt a draft. ThePS equation arises from a study in which50 people were asked to adjust an air velocity source as they pleased whenexposedto a specific air temperature.PSrepresents the cumulative percent of people choosinga particular air velocity at the temperatures tested. TS is an equation that predicts thermal sensation vote using a linear functionof air temperature partial vaporpresand sure. The adaptive modelsinclude in somewaythe variations in outdoor climate for determining thermal preferences indoors. The equations for the adaptive modelsincluded in the software are Tn = 9.22 + 0.48Ta + O.14Tmmo and Tn = 23.9 + 0.295( Ttmno 22) F(Tmmo 22)]2" e-L 24.,,/~ J METHOD Thefirst task wasto identify the location and status of the thermal comfort modelsdiscussed above. Specifically, does a computer code exist for the model, where is it, and what languageis it in? Asurvey wasthen distributed to HVAC professionals via facsimile. Thesurvey questionedprofessionalsabout basic thermal comfort analysis needs. Based on the survey responses and available computercode, decisions were made about the subset of modelsto include in the software. It wasdecided that several, but not all, of the eight existing physiologically basedthermal sensation models(listed in Table 1) should be incorporated in the software package.It wasalso decided that the "modelselection," or methodfor choosing among these models, should be presented in the documentation via instructions for using the software.Thewidespread availability of veryfast microprocessors itself readily to the task of lends including several models.A60-minute iteration of the Gagge 2Nodemodelthat once took minutes (or longer) of real time process nowtakes less than one-tenth of a second on a fast computer. Asurvey was distributed via facsimile to 60 professional HVAC engineers in the San FranciscoBayarea to assess interest in and experience with thermal sensation models. Basedon the survey results, it was decidedthat to incorporate heat balance models beyond PMV-PPD 2-Node at this time would and provide diminishing returns for most users of the software, addingadditional complexity possibility for error in appliand 131 (5) (4)

cation without providingany truly uniqueinformationuseful to a professional. In addition, it wasdecidedthat all five of the empirical and adaptive models should be implemented to due their simplicity. A complete list of the modelsincludedis given in Table2 and a completelist of the indices computed given is in Table3. USING THE SOFTWARE Utilizing a point-and-click interface (Figure 1), users adjust the input values and the outputs are updated in real time. The right-hand side of the screen controls the input variables, while the left-hand side presents the output from the models. Values can either be entered directly or the up/down arrowscan be used TABLE 2 Modeas |ncluded in Software Model Name Author Year First Introduced Type Heatbalance Heatbalance ...... Heatbalance Empirical Empirical Empirical Adaptive Adaptive

PMV-PPD P.O. Fanger 2-Node Gagge al. et Revised PMV D. Int-Hout Fangeret aL PD PS TS Tn Tn

1970 1970 1990 1988 Fountain al. et 1994 Rohles andNevins1971 MA. Humphreys 1978 A. Auliciems 1983

to scroll to the desired number.When programis loaded, the default values for all inputs are loadedand all tile rnodels are cycled. When input value is changed,all the modelsare run any again. Withtodays fast processorspeeds, there is no time delay between clicking on an arrowto adjust an input and havingall of the output values changein response. Pull-downmenus allow file handling, access to additional modelparameters, adjustmentof units, and interactive program help. A toolbar is "wired"to fiequently used menu options for quick access. Tile modelis written in C++in a waythat will allowintegrationof additionalmodels a latex date if desired. at In addition to numericaloutputs, the program utilizes icons and text that changecolor (or appear and disappear) to indicate whenthe outputs are within cex~ain ranges. Fox" example,the "ASHRAE icon in the upper tight corner turns blue, green, 55" or red whenconditions are below, within, or abovethe ASHRAE 55-92 comfort zone, respectively. When of the input varione ables does not coincide with one of the ASHRAE assump55-92 tions, the icon turns a lighter shadeto indicate that ASHRAE 5592 does not apply. Similarly, when calculated value of PDis the above 15%(ASHRAE 55-92 limit), the words "draft risk" appearnext to the PDcalculation. It shouldbe noted here that it is not sufficient to run this programto showASHRAE Standard 55 compliance it does not incorporateall of the requirements as of ASHRAE 55-92. Three pull-down menusate available at the top of the screen--"File," "Options," and "Help." "File" accesses file input and output functions; "Options"allows control of various control constants,the unit system,and calculationutilities; and "Help" runs the on-line help. Frequently used commands from the pull-down menus linkedto icons in the toolbar (Figure 2). are

TABLE 3 Indices Computed the Software by Index PMV (Predicted meanvote) PPD (Predicted percentdissatisfied) ET* (New effective temperature) SET* (Standardeffective temperature) TSENS (Predictedthermal sensation) DISC (Predictedthermaldiscomfort) PD (Predicted percentdissatisfieddueto draft) PS(Predicted percentsatisfied withthe level of air movement) Tn(Neutral temperaturebasedon mean monthly outdoortemperature) Tn(Neutraltemperature basedon long-termindoorandoutdoor temperature) TS(Predicted thermalsensation) PPD (Calculatedby Int-Houtmodel) PMV-IH (Calculated by Int-Hout model) Fanger(1970) ISO(1984) Gagge al. (1986)as modified Doherty Arens(1988) et by and Gagge al. (1986)as modified Doherty Arens(1988) et by and Gagge al~ (1986)as modified Doherty Arens(1988) et by and Gagge al. (1986)as modified Doherty et by (1988) Fanger al. (1988) et Fountain(1994) Humphreys (1978) Auliciems (1983); de Dearand Ring(1984) Rohlesand Nevins (1971) Int-Hout(1990) Int-Hout(1990) Reference

132

ASHRAE Transactions: Research

Figure 1 Thermalcomfort tnodel user #~te .rface.

Figure 2 Toolbar. From left to right, the icons perform followingfunctions: (1) the select an inputfile, (2) select an outputfile, (3) reada set of input data from input file, (4) write the currentrun to the outputfile, the (5) toggle all inputs and outputs between Fahrenheitand Celsius, (6) select type of humidity input, (7) invoke clo calculator, the access the physiologicalvariables, and (9) use the on-line help facility. The humidity input window (Figure 3) showsthe types humidity input. Userscan select fromrelative humidity,partial vaporpressure, dewpoint, wet bulb, or humidity ratio by clicking on the appropriate radio button. This feature allows easy comparisonof design scenarios that have different control parameters. The"clo calculator" (Figure 4) and the "globe calculator" are modulesthat pop up for specific computations. Clothing insulation is input to the heat balancemodels a single number as (clo) that combines of the different articles of clothingon the all body weighted by insulation value and area of skin surface covered. The clo calculator allows the user to compute clo a value for an ensemble individualarticles by selecting the artiof ASHRAE Transactions: Research

Figure 3 Humidio, specification selection box. cles individually. When of the desired clothing items are all selected, the user can give that specific ensemblea nameand save it to the clothinglibrary. Ensembles the clothinglibrary in are available in the clo selection boxon the mainscreen (Figure 1). The globe calculator allows determinationof MRT based on a measurement of"globe temperature" for a globe thermometer of any size Globe thermometers are a common method for determining MRT indoors, and the globe calculator simplifies the conversionof measured globe temperaturesto MRT field for survey data~ ENGINEERING APPLICATIONS Engineeringapplications for the modelare numerous.The modelcan be used to examinethe relative trade-offs between 133

MyClothes

Figure 4 Clothing calculatop: various physical parameters;for example, increase in relative an humidity 10% predictedto be offset by a drop in air temperof is ature of 0.5F (0.28C) using PMV-IH a measure.This effect as changes only slightly between 50% 90%RH. Similarly, an to increasein temperature one degqee predictedto be offset by of is an increase in air movement 12 fpmbetween77F (25C) and of 78F(25.5C), but this effect is extremelynonlinear--it takes moreair movement fpm) to provide the equivalent cooling (+20 effect between 78F (25.5C)and 79F (26.1 C). The model also be used to determinethe design deadband required to maintain comfortfox groupsof people performing certain activities. For example, with ASHRAE Standard 55 summer clothing (0.5) and office activity that includes walkingabout betweenworkstations with little or no sitting (1.7 met), the temperature whichthe most people will feel comfortableis predicted to be 70~5F(21.3C), the lower limit is 66.3F (19.0C), and recornrnendedupper limit (using PMV) 74.7F (23.7C). is wouldthat temperature range be suitable for people seated quietly at desks?Foxseated, quiet persons,the temperatures shift up to 75.0F (23.8C), 77.7 (25.3C), and 80.4F (26o8C), respectively. Clearly, since the temperatureranges do not overlap, people perfor~ning these different activities cannot be comfortable the samezone. Byfar the mostinteresting appliin 134 cation is to use the prograrnto modelvariousscenariosthat come up during design, Well look at two common situations and see howthe modelcan help inform design decisions. Example 1 Its a cool winter dayin the perimeter zoneof a large office building. Thesetpoint of the zone is 715F(21.9C)and there is someradiant cooling occuning due to cold windows nearby (MRT 68F [20C]). Air movement low and relative humid= is ity is 50%.Apersonsitting at a desk, occasionallygetting up to walkaround(1.1 met), dressedin a businesssuit (0.9 clo) feels slightly cool but comfortableaccordingto PMV. wedecrease If the clothing to underwear,long pants, and a long-sleevedshirt (about0.65)or a full slip, a blouse,anda thick skirt (about0.75), the person becomes very cold using both ASHRAE ISO and criteria. How much wmmer wouldthe ternperature have to be in order to bring these morelightly clothed people into a state of thermal comfort? turns out that we haveto adjust the temperIt ature upward nearly 5F (2.8C) to bring these people up to the lower boundaryof the comfort zone and 7.5F (4.2C) to bring themto a state of thermalneutrality (neither hot nor cold). This exampleshowsthe irnportance of clothing in determining thermal comfort and reveals howsomedress codes can ASHRAE Transactions: Research

cause widespread discomfort. If the zone occupancy will encom- NOMENCLATURE pass significant clothingvariation, as in the example above,there DISC = predicted discomfortvote (scale value) are several options: (1) design a single-setpoint zone control ET* = neweffective temperature (C [F]) systemand assume that people will fight over the thermostatand = vapor pressure (kPa) use fans or heaters to make themselves comfortable,(2) separate p PD = predicted percent dissatisfied due to draft (%) the zoneinto smaller zoneswith the ability to produce least at PMV = predicted meanvote (scale value) five-degree (in the aboveexample) differences between ad.jacent zones, (3) consider a task-conditioningapproachwhereindividPPD = predicted percent dissatisfied (%) uals control the environment their own of workarea and design PS = predictedpercent satisfied withthe level of air a system that can produce at least five-degree (in the above movement (%) example)differences betweenadjacent work areas~ SET* = standard effective temperature(C [F]) = air temperature(C [F]) Ta Example 2--Air Movement Tmmo = mean monthly outdoor temperature (C) Tn = neutral temperature (C) Using the ASHRAE Standard 55-92 winter optimum operTop = operative temperature (C) ative temperatureof 71F(21.6C) and a fairly typical turbulence intensity of 40%,the minimum movement the skin air over TS = thermal sensation vote (scale value) surface is assumed be 20 fpmsimplydue to the thermal plume TSENS = thermal sensation vote (scale value) to of the body. Whathappensto draft risk (PD) whenair velocity = turbulence intensity (%) Tu is increased a little bit, as might occur whena fan-powered v = air velocity (m/s) mixingboxis used to improve circulation rates? Naturally, air diffuser selection has a big impact on whetherdrafts become a REFERENCES problem, if air velocity is increasedto 30 fpm but (barely percepASHRAE. 1992. ANSI/ASHRAEStandard 55-1992, Thertible), the value of PD(predicted draft risk) is over 20%.This mal environmental conditions .for humanoccupancy. exceeds the allowable percentage of 15%in ASHRAE Standard Atlanta: American Society of Heating, Refrigerating 55-92. Does the sameeffect happenin the summer whenhigher and Air-Conditioning Engineers, Inc. space temperatures occur? If we raise the temperatureto the Auliciems, A. 1983. Psycho-physiologicalcriteria for global ASHRAE Standard 55-92 optimum operative temperature of zones of building design Proceedings of the Ninth 76F(24.4C), the draft risk drops to 16%,nearly meetingthe International Society of Biometeorology Conference, standards percentage but not quite. If we assumethat the Stuttgart-Hohenheimo increased air movement a 20%turbulence intensity, we has de Dear, R., and J. Ring. 1990oHuman subjective experience couldraise the air velocity to 35 fpmbut no higher withoutraisof ambient temperature step-changes: Experimental ing the air temperature. results compared to the predictions of a numerical model. Draft for review. On the other hand, in manyparts of the U.So, summer Doherty, T., and E. Arens. 1988. Evaluation of the physiotemperatures indoors routinely exceed the ASHRAE Standard logical bases of thermal comfort models. ASHRAE 55-92 optimum operative temperature of 76F (24.4C). For Transactions 94(1): 1371-1385. space temperatureof 78F (25.5C), the PS modelpredicts that Fanger, P.O. 1970. Thermal comfort. Copenhagen: Danish air movement to 65 fpmmaybe desired by 80%of the occuup Technical Press. pants for cooling. Fanger, P.O., A. Melikov, H. Hanzawa,and J. Ring. 1988. Air turbulence and sensation of draught. Energy and CONCLUSIONS BuiMings12: 21-39. Fountain, M.E., E.A. Arens, RJ. de Dear, F.S. Bauman,and ASHRAE recently sponsored a research project (RP-781) K. Miura. 1994. Locally controlled air movement preto select and prepare a thermal sensation modelfor use by the ferred in warm isothermal environments. ASHRAE profession. Athermal sensation prediction tool has beendevelTransactions 100(2): 937-952. oped as part of this project. This paper introduced the tool, Fountain, M., and C. Huizenga. 1995. A thermal sensation described the tools features, and presented examplesof how model for use by the profession. Final report to the HVAC engineerscan use the tool in practice. Thenext step in the American Society of Heating, Refrigerating and Airdevelopment a design engineering tool is to graft comfort of Conditioning Engineers, Inc. models onto a building energy simulation model (such as Gagge, A.P., J. Stolw~ik, and Y. Nishi. 1971. An effective BLAST) produces interior surface and space temperature that temperature scale based on a simple model of human Establishing this linkage will allow greater feedbackbetween physiological regulatory response. ASHRAE Transacsystemdesign and predicted environmentaleffects. tions 77(1 ): 247-262. ASHRAE Transactions: Research 135

Gagge, A.P., A.P. Fobelets, and L. Berglund. 1986. A standard predictive index of human response to the thermal environment. ASHRAE Transactions 92(2): 709-731. Humphreys, M.A. 1978. Outdoor temperatures and comfort indoors. BREcurrent paper. Garston, Watford, U.K.: Building Research Establishment. Int-Hout, D. 1990. Thermal comfort calculations/A computer model. ASHRAE Transactions 96(1): 840-844. ISO. 1984. Standard 7730, Moderate thermal environments--Determination of the PMV and PPDindices and specification of the conditions for thermal comfort. Geneva:International Organization for Standardization.

Jones, B., and Y. Ogawa. 1992. Transient interaction between the human and the thermal environment. ASHRAE Transactions 98(1): 189-195. Rohles, F.H., Jr., and R.G. Nevins. 1971. The nature of thermal comfort for sedentary man. ASHRAE Transactions 77(1): 239-246 Stolwijk, J. 1970. Mathematicalmodelof thermoregulation. In Physiological and Behavioral Temperature Regulation, Gagge, Stolwijk, and Hardy, eds. Tanabe, S. 1992. A 25-nodemodelof human thermoregulation. Computer code obtained via personal communication. Wissler, E.H. 1964. A mathematical model of the hu~nan thermal system. Bulletin of Mathematical Biophysics 26(147).

136

ASHRAE Transactions: Research