UC BerkeleyIndoor Environmental Quality (IEQ)

TitleA field measurement system for the study of thermal comfort

Permalinkhttps://escholarship.org/uc/item/9cb6t2wz

AuthorsBenton, C.Bauman, FredFountain, M.

Publication Date1990 Peer reviewed

eScholarship.org Powered by the California Digital LibraryUniversity of California

3372 (RP-462)

A FIELD MEASUREMENT SYSTEM FOR THESTUDY OF THERMAL COMFORT

C.C. Benton F.S. Bauman, P.E.Member ASHRAE

M.E. FountainAssociate Member AS/-/RAE

ABSTRACTThis paper describes the instrumentation and

measurement protoco/ used in ASHRAE research proj-ect RP-462, a fie/d study of environmenta/ conditionsand occupant comfort in 10 office bui/dings /ocatedin the San Francisco Bay region. During this study,we made a tota/ of 2342 visits to 304 participants dur-ing two seasons, co//ecting a furl set of physica/ mea-surements and subjective responses at each visit./nthis paper we describe the design of equipment andtechniques for gathering physica/ measurements withthe detail and accuracy required by both ASHRAEStandard 55-8I (ASHRAE 1981) and/SO Standard7726 (ISO 1985). In addition, the project developed alaptop microcomputer-based thermal assessment sur-vey that collected a substantial subjective data set inmachine-readable form. These components per-formed reliably during nine months of field use, pro-viding a detailed description of the monitoredworkstation environments with a concurrent portrait ofsubjective occupant response. The system is rec-ommended for further use.

INTRODUCTIONThe design of pleasant and efficient working en-

vironments is becoming an increasingly complex task~Our "post-industrial" economy features a major ex-pansion in the white-collar workforce, radical changesin the technology of the office workplace, and growingconcern about the efficiency of office workers. Withthese changes, contemporary designers face newchallenges that will directly affect the design of build-ing comfort systems.

The scale of the office arena is immense; con-sider the following statistics describing the UnitedStates office enterprise (Roark and Dowell 1986): 1985, our office inventory totaled 4 billion ft 2 of interiorspace with a yearly expansion of 250 million fF. Theseoffices provided the work setting for two-thirds of allU.S. employees, who in turn spent a cumulative 100billion hours per year working inside the buildings wedesign. In this vast interior landscape, the quality ofthe physical environment can have profound effectson a worker’s sense of health and well-being, as wellas subtle but significant effects on worker productivity.Current practices in office design raise importantquestions regarding the needs of individual buildingoccupants. For example, does the objective of pro-

viding a constant, uniform interior environment re-spond to the needs of workers with different comfortpreferences, clothing styles, and environmental sen-sitivities? Does today’s application of centralized con-trols for thermal and lighting systems allow workersadequate control to respond to their own hour-to-houror day-to-day environmental preferences? To what ex-tent does the proliferation of desktop computers, act-ing as localized heat sources, contribute to variationsin comfort between adjacent workstations?

ASHRAE provides standards for maintainingcomfortable interior environments and in providing thisservice affects the experiential reality of each year’s100 billion hours of office occupancy. The current com-fort standard, ASHRAE 55-81 (ASHRAE 1981), is pri-marily based on the results of laboratory studies inwhich sedentary people were exposed to uniform ther-mal environments and comfort was equated with neu-trality or the lack of thermal sensation. Whilelaboratory-based studies of comfort allow experimen-tal precision, it is uncertain whether the standards thatthey engendered are optimal for real buildings. For inreal buildings, conditions are both dynamic and fa-miliar; occupants face well-known settings and pursuecustomary tasks° In addition, it is uncertain whetherthe college students often employed in short-term lab-oratory studies are accurate representations of thegeneral office population. There are also numerousfunctional and aesthetic attributes of the office envi-ronment that may influence the worker’s thermal re-sponse differently from that of the laboratory subjecLOne method for assessing the appropriateness of ex-isting laboratory-based comfort standards would beto conduct a field study of building occupants’ re-sponses to their normal office environments.

The instrumentation system presented in this pa-per was developed for ASHRAE research project RP-462 (Schiller et al~ 1988b), a field study of thermalenvironments and occupant comfort in existing officebuildings. We visited 10 San Francisco Bay area officebuildings during the winter and summer seasons of1987, making a total of 2342 visits to 304 participantsand collecting a full range of physical measurementsand occupant responses characterizing the physicalenvironmenL The occupants were volunteers, sur-veyed during their normal work activities. The objec-tives of this study included:

1. The development of a detailed data base on

Charles C. Benton is an Associate Professor of Architecture; Fred Bauman is a Research Specialist with the Building ScienceLaboratory; and Marc Fountain is a Graduate Research Assistant in the Department of Architecture, University of California,Berkeley. 623

the thermal environment and subjective responses ofoccupants in representative Bay Area office buildings,

2. ]he documentation of comfort conditions inthe monitored office environments including their de-gree of compliance with ASHRAE (1981) and ISO(1984).

3. The analysis of the compiled data to identifyrelationships between physical, psychological, anddemographic parameters. We derived comfort param-eters from the monitored data, calculated commonlyused temperature indices, and applied statistical anal-ysis to identify significant correlations between thermalconditions and subjective responses.

4. The development of instrumentation, mea-surement procedures, and occupant survey tech-niques to assess thermal comfort.

Work on the first three objectives has been de-scribed in previous papers (Schiller et al. 1988; Schiller1989). This paper describes the field methods and thesystem developed for collecting comfort-related datain the field. Following a summary review of previousfield studies, we provide a detailed description of thesystems used for field measurements, including spe-cifics on transducers, data acquisition equipment, anddata management techniques~ The paper also in-cludes a description of the system in use and sampledata from the study.

PREVIOUS FIELD STUDIESThermal comfort is defined by ASHRAE (1981)

and ISO (1984) as "that condition of mind which ex-presses satisfaction with the thermal environment."Just as we understand the experimental advantagesfor investigating the effects of physical-envkonmentalfactors in laboratory studies, we must also acknowl-edge the important role played by nonphysical, or psy-chological, parameters in the definition of comfort. Inan effort to provide a practical, real-world context forthe abundance of laboratory-based thermal comfortresearch, several notable field studies have been car-ried out and are briefly discussed below.

Humphreys (1976) provided a worldwide sum-mary of more than 30 field studies performed overmany years. In his paper, Humphreys compared thefield study results with predictions from the PMV heat-balance model (Fanger 1970) based on laboratory re-search. This comparison identified important differ-ences between field and laboratory results, including(1) an increased adaptability of people to variations their thermal environment compared to the model pre-dictions-in fact, a majority of the field study resultsfound that the optimum temperature was lower thanthe laboratory-based preferred temperature--and (2)a dependence of thermal response on acclimatizationto outdoor and indoor temperatures during the pre-ceding weeks, which is not accounted for by the PMVmodel. Several significant and more recent field stud-ies have noted similar differences between field andlaboratory results (Fishman and Pimbert 1978; Howelland Kennedy 1979; Auliciems and Dedear 1984; De-dear and Auliciems 1985; Schiller et al. 1988). In thegreat majority of field studies performed prior to 1984,

it is unusual to find studies that characterize the ther-mal environments with more than a single set of basicsensors positioned at one location near the monitoredworkstation, and in some cases environmental varia-bles are merely assumed or estimated. Typical fieldinstrumentation perrnitted the average building ther-mal conditions to be tested for general compliancewith the ASHRAE and ISO standards but was insuffi-cient to verify all specifications required for thermalcomfort. With regard to the Fishman and Pimbert study(1978), Mclntyre (1980) pointed out that, unlike labo-ratory test subjects, real-world people do not wearstandard clothing and are generally free to adjust theirinsulation level in response to changes in their thermalenvironment. However, during their one-year fieldstudy, Fishman and Pimbert found that their test sub-jects were more (not less) sensitive to uncomfortableconditions compared to laboratory-based predictions.

Just as the level of thermal instrumentation hasvaried from study to study, so has the detail with whichpsychological assessments have been carried ouLThe field work of Howell and Kennedy (1979) and How-ell and Stramler (1981) demonstrated the relative im-portance of psychological vs. physical variables indetermining perceived therrnal comfort. Cena et al.(1986) attribute the differences between measured re-sults and PMV model predictions to a psychologicaladjustment made by their elderly test subjects. Stillother field studies have concluded that the measuredcomfort conditions were in general agreement with therecommendations of ASHRAE (1981) (Gagge Nevins 1976; Lammers et al. 1978). During recentyears, several large office building occupant studieshave virtually dropped all physical measurements andused survey methods alone to address a range ofenvironmental parameters--thermal comfort, air qual-ity, acoustics, visual cornfort, spatial comfort, andfunctional and aesthetic aspects of the space--whichare critically related to occupant comfort, satisfaction,and productivity (Harris 1980; Bril11984; Woods et al.1987; Dillon and Vischer 1987; and Baillie et al. 1987).

The variation of conclusions from previous fieldstudies is due in part to the inherent difficulties of per-forming well-controlled measurements in the field andthe significant variations among the studies in the levelof detail with which both the thermal environment andthe subjective state of mind of the respondent havebeen assessed. Many of the earlier studies avoidedmeasuring air velocity altogether because of the cum-bersome Kata thermometer, the only available low-speed anemometer at that time (Dedear and Auliciems1985). Recent evidence, however, suggests that airmovement (magnitude and variability) is perhaps themost significant of the fundamental comfort parame-ters in its potential influence on occupant response(Purcell and Thorne 1987). Clothing levels and rneta:bolic rates of previous field study respondents havealso been assessed by methods ranging from an es-timated constant value to individual checklists. Themost common psychological assessment methodshave made use of a simple seven-point ASHRAE orBedford scale, while a few more thorough investiga-

624

tions have contained additional survey questionsaimed at determining the effects of alternative psy-chological variables on perceived comfort. In manycases, simplified data constructs have limited the po-tential usefulness of field data in comfort analysis.

ASHRAE (1981) and ISO (1984, 1985) providethe protocols for determining the combination of en-vironmental (temperature, radiation, humidity, and airmovement) and personal (clothing and activity) factorsthat provide conditions for thermal comfort in the builtenvironment. The detail of physical measurements inmany previous field studies is below the requirementsspecified by these standards, requirements necessaryto test for compliance with recommended values foraverage and dynamic conditions, as well as local non-uniformitiesl One of the major objectives of ASHRAEresearch project RP-462 (Schiller et al. 1988) was, fact, to determine whether current comfort standardswere being met in the 10 monitored buildings.

MEASUREMENT SYSTEM DESIGN

In the context of our field study’s objectives andthe measurement specifications of ASHRAE (1981)and ISO (1985), we developed a series of require-ments for our own field measurement system. Our de-sign of a measurement strategy for both physical andsubjective variables was based on the following cri-teria:¯ The system(s) must be capable of collecting con-

current physical data (air temperature, dew pointtemperature, globe temperature, radiant asymme-try, air velocity, and illuminance) from an array oftransducers placed to represent the immediate en-.vironment of our seated subjects: In addition, thesystem should be capable of recording a widerange of subjective responses and questionnaireanswers with direct input from the building occu-pants.

¯ The survey process, including subjective responsesand physical measurements, should be completedin approximately 10 minutes per workstation visit.

¯ The physical measurement transducers and theirinterrogation must meet the ASHRAE (1981) andISO (1985) standards for accuracy and responsetime.

¯ Physical measurements should be made as closeas possible to the exact physical position of thesubject completing the subjective questionnaireand as soon as possible after completion of thequestionnaire.

¯ All physical and subjective data should be collectedin machine-readable form. Compiling data in digitalfiles during the collection process eliminates key-punch errors and expedites daily summary sheetsfor error checking.

¯ The instrumentation package should be mobile andportable with battery power for the system(s) ca-pable of a full day’s operation without recharge.

¯ The data acquisition systems should provide a real-time display of measured values for error-checkingpurposes. These values should be hidden from the

sight of test subjects to avoid bias in their answersto subjective questions.

¯ The data-gathering process should include a con-tinuous temporal record of physical variables in afixed location to monitor transient effects.

¯ The field equipment should be automated to theextent that student assistants with modest trainingcould contribute to the daily data collection effort.

¯ The entire instrumentation package (includingtransducers, data acquisition systems, and com-puters) should not exceed a budget of approxi-mately $25,000.

To meet these criteria, a measurement strategywas developed that combined four separate data ac-quisition systems running simultaneously. The datacollected by each system was time stamped at col-lection and then combined during the data analysisphase to assemble a complete profile of physical andsubjective variables. Detailed measurements of indi-vidual workstations were made with a mobile instru-mentation cart (see Figure 1) that housed threesystems: a laptop computer-based survey system forsubjective assessment, a packaged indoor environ-ment measurement system for the mid-level (0.6 m)physical measurements, and a microdatalogger-based measurement system for all additional physicalmeasurements. The fourth system was a fixed-posi-tion, microdatalogger-based equipment group thatcollected time-series data to provide a temporal re-cord of the building’s interior conditions. The next sec-tion, describing the transducers in detail, is followed

Figure 1 Mobile measurement chart

625

by a complete description of the four measurementsysterns.

TRANSDUCERSASHRAE (1981 ) and ISO (1985) specify minirnum

response time and required accuracy levels for sen-sors used in the rneasurement of environmental vari-ables. Additional project-specific criteria fortransducer selection included reliability, physical in-tegrity, and an overall response time appropriate forthe five-rninute length of workstation measurements.Table 1 provides a description of the primary trans-ducers used in our study (see Figure 2). These includetransducers for the environmental variables listed be-lOW:

Point-in-timemeasurements ateach workstation

Height Above (via mobileFloor instrumentation cart)

Time seriesmeasurements at asinglerepresentativelocation

0.1 rn

air temperatureair velocityglobe temperature

air temperatureair’ velocityglobe temperaturedew-point

temperaturechair surface

0,6 m temperature air temperature

air temperatureair velocity air temperatureglobe temperature air velocityradiant asymmetry globe temperature

1.1 m illuminance illuminance

1.7 m air temperature

Figure 2 Transducer profiles ....

Air TemperatureAir temperature was measured with either a plat-

inum RTD, a thermistor, or a thermocouple dependingon the location of the measurement, In each case, thesensing element was shielded from radiation by a cy-lindrical metal screen. A platinum RTD measured airternperature at the 0.6 m height on the mobile cart,and type-T copper-constantan thermocouples wereused at the 0.1 rn and 1.1 m heights. High-precisionthermistors were used in the stationary system.

Air VelocityThe air velocity transducers, used on the mobile

cart and the stationary system, were omnidirectionalanemometers based on the constant temperature prin-ciple. An elliptical-element anemometer was used forthe 0.6 m height on the mobile cart, while the remaininganemometers featured a spherical sensing element.The anemorneter elements are maintained at a con-stant temperature by an electrical heating element thatcompensates for heat lost to the surrounding air-stream. The amount of current required to keep anelement at a constant temperature is related to theamount of air flowing past the element per unit time(and is easily measured). Both types of sensors wereoptimized for the low air velocities commonly found inindoor environments and featured compensation forambient temperature.

Dew Point TemperatureThe transducers used in this project had rela-

tively fast sampling rates with the exception of the dewpoint sensor, which took a single measurement everytwo minutes. This was considered acceptable, as dewpoint temperature is a relatively stable variable. In thesensor, a small conical mirror is cooled until atrnos-pheric water condenses on it. A light-emitting diode(LED) and a photosensitive transistor (receiver) located above the mirror. When condensation occurs,the receiver detects light scattered by the dew. Thetemperature of the mirror at that time is measured andrecorded as the dew point temperature.

Plane Radiant Temperature and RadiantAsymmetry

Plane radiant temperature is defined as the uni-form surface temperature of an enclosure that pro-duces the same incident radiation on a small planarsurface as in the actual environment. Radiant asym-metry is the difference between the plane radiant tem-peratures of small planes facing opposite directions.The radiant asymmetry probe consisted of two pairsof gold-plated and black-painted elements connectedto thermopiles. Each side of the probe had a gold anda black elernent. The measurement is based onthefact that the gold element exchanges heat primarilyby convection while the black element exchanges heatby both convection and radiation. Thus any voltagegenerated across the thermopiles results from heattransfer by radiation between the black element andthe environment.

626

Illuminance and Chair Surface TemperatureIlluminance was considered an ancillary param-

eter that might be useful in later analysi& Cosine-cor-rected silicon photometers were used to measureilluminance in a horizontal plane. A spring-loaded plat-inum RTD was used to track the surface temperatureof cart surfaces adjacent to the mid-height globe ther-mometer. The comfort standards do not require mea-surement of these variables.

Globe TemperatureThis investigation of human thermal response re-

quired knowledge of the mean radiant temperature ofthe surrounding environmenL Mean radiant tempera-ture can be calculated as a function of air temperature,air velocity, and globe temperature. Globe tempera-ture is measured as the internal temperature of a hol-low sphere (typically painted black or gray) exposedto the environment. This internal temperature indicatesthe balance between heat lost and gained from radia-tion and convection° In principle, globe temperatureis relatively easy to measure; however, the five-minutetime limit at each workstation posed a challengingproblem. We built and tested several alternativeglobes of various materials and sizes to develop asensor with short response time and appropriate ac-curacy. Candidate globes included 5- and 6-in.-di-ameter spherical glass lightbutbs with the filamentremoved and smaller spherical Christmas tree orna-ments. The sensor eventually adopted was con-structed by inserting a type-T thermocouple into atable tennis ball coated with gray paint.

Humphreys (1977) proposed the 38-mm-diam-eter table tennis ball as useful for making indoor mea-surements because of its fast response and heatexchange properties similar to the human body at typ-ical indoor air speeds. Tests were conducted to verifythat 38-mm globes could, in fact, predict mean radianttemperature in 5 minutes as well as a standard 150mm globe could after 15 minute& A smaller globe wasdesirable for the mobile cart in any case, since themeasurement standards required globes at threeheights and three 150-mm globes within a 1~1 m by.5 m plane would pose obstructions for other sensors.

The candidate globes used small lengths of plas-tic pipe to position the thermocouples in the center ofthe globe. To test the accuracy and response timesof the candidate globes, they were suspended on thinwire from a frame mounted on a mobile platform, whichwas rapidly moved from a cold hallway into a hotchamber equipped with radiant and convective heatsources plus a fan. This was done for a range of tem-perature differences, room air velocities, and radiantsource intensities. The equation used for calculatingmean radiant temperature from globe temperaturewas:

[ 6’32D-°4V°5 (Tg - Ta) Tg4]°25 (1

where

D = diameter of the globe (m)e = emissivity of globe~ = Stephan-Boltzmann constant (567 × 108 W/m2K4)Ta = air temperature (K)Tg = globe temperature (K)T, = mean radiant temperature (K)V = air velocity (m/s)

and where the convective heat transfer coefficient(hc = 6.32 D-°4V°5) is taken from ASHRAE (1987).

As expected, the table tennis ball sensor had themost rapid response time. Although the time requiredto reach 90% of final value was 5~8 minutes, slightlylonger than the workstation measurement period, thiswas considered acceptable because in practice thethermal differences between successive workstationswere usually small. After correction for globe diameter,the differences in predicting mean radiant temperature~,sing the 38-mm and larger globes were less than orequal to the accuracy of the thermocouples used.Side-by-side comparisons were made with a com-mercially available "ellipsoid-shaped" comfort sensor(see ASHRAE 1985) to obtain the operative tempeFature comparison shown in the globe temperature sec-tion of Table 1.

CALIBRATIONAlthough many of the transducers featured man-

ufacturers’ specifications that exceeded the comfortstandards’ requirements, a series of calibration mea-surements was performed on all sensors used in thestudy. Air temperature sensors were checked by plac-ing them in a well-insulated lightweight box. The box’sinternal temperature was varied slowly through therange of values anticipated for interior office environ-ments, while readings from the temperature sensorswere compared to a laboratory-grade mercury ther-mometer. Results obtained for the shielded thermo-couples produced an acceptable accuracy rating overthe relatively narrow temperature range found in officebuildings. The omnidirectional air velocity probes werenew from the factory with calibrations within the re-quirements of ASHRAE (1981). Intercomparison of ve-locity readings in the flow of a portable desk fanconfirmed the integrity of the factory calibration be-tween sensors. The dew point sensor, also factory cal-ibrated, was satisfactorily compared to a slingpsychrometer and a thin-film, hygroscopic-salt relativehumidity sensor; however, subsequent comparisonsat the midpoint of the study suggest operational cau-tions concerning the dew point sensor (see the "Troub-leshooting" section). The entire calibration exercisewas conducted before and after the field measure-ments with intermediate calibration measurements onselected sensor& Calibration results are presented inTable 1.

DATA ACQUISITION TECHNIQUESThis section describes the two measurement

units, one mobile and one stationary, developed torecord physical and subjective dat& The mobile sys-tem was used to visit individual workstations, where itrecorded subjective occupant responses to a short

627

TABLE 1Transducer Specifications

SENSOR SENSOR _.. IVlEASURENIENT ACCURACY, RESPONSEQUANTITY DESCRIPTION LOCATION* ASHRAE 55-81 ISO-7726 MANUFACTURER CALIBRATION TIME

Air shielded M: 0.6 m -0 2°CTemperature platinum RTD

shielded S: Q6. 1 1, +02°Cthermistor 1 7 rr

shielded type T M: 01, 1.1 m +0.2°Cthermcouple

Globe type T thermo- M: 0 1, 06, Desired:Temperature couple inside 38 1 lm; -+0 2°C

mm diameter S: 1 1 m (for MRT)table tennis ba(painted grey)

Air Velocity

Humidity

ellir3tical M: 0.6 r~ +0 05 m/somnidirectional over range 0.05constant to 0 5 m/stemperatureanemometer

spherical M: 0 1. 1 1 m; -+005 rn/somnidirectional S: 1 1 m over range 0,05temp, to 0.5 m/scompensatedanemometer

chilled-mirror M: 0 6 m -+0 6°C (fordew point sensor dew point

temp )

Plane Radiant opposing plane M: 1 1 mTemperature radiantAsymmetry temperature

sensors

Surface spring loaded M: 06 m N/ATemperature platinum RTD

Illuminance silicon M: 1.1 m: N/Aphotovoltaic S: 1 1 mphotometer

¯ + 1,0°C

Required: -+0.5°C -+0 2°C over -+0 1°C over 50 sec (90%)Desired: -+0.2°C range range in still air

5 to 40°C 18.7 to 25.1°C

Required: -+ 0 5°C -+0 1°C over + 0.2°C over 5 sec (90%)Desired: -+0 2°C ’ange range

0 to 70°C 20 7 to 28 5°C

Rec~ dired: -+ 0 5°C -+ 1 0°C over -+ 0 1°C over <3 sec (90%)Desired: -+0 2°C ’ange range

0 to 100°C 187 to 25 1°C

Rec~aired: -+2 0°C = 1.0°C over -+QI°C over 2.5 min (63,2%);Desired: -+0 2°C range ranges 187 to 5 8 rain (90%)(for MRT) 0 to 100°C 25 1°C

(for (forthermocou pie) thermocouple);

_+ 1°C (foroperative temp)

Required: -+5% -+0 05 m/s factory 0 2 sec (90%)-+5% -+0.05 m/s over range calibrationDesired: 0 05 to 1 0 m/s checked by-+2% -+0 07 m/s intercomparisonover range 0 05 to1,0 m/s

Required:-+5%+0,05 m/sDesired: -+2%-+0,07 m/sover range005tol0m/s

-+0.15 kPa (for -+Q5°C (for dewwater vapor partial point temp overpress) range:

Ta~,-Top < 10°C)

¯ + 3% -+ 0 02 m/s factory 2 sec (67%);for flow at 90° to calibration 4 1 sec (90%)probe; for other checked byangles: <-+ 10% intercomparison

Required: -+1 0°C -+0,5°C forDesired: -+ 0.50C Tp,-Ta~rl

~< 15°C

N/A -+ 0 5°C overrange 5 to 40°C

N/A -+ 5%

factory 2 minutecalibration measurementchecked with periodslingosychrometer

-+0,4°C over 60 sec (90%)’ange 18 7 to25 1°C (forplane radianttemp.)

-+0 2°C over 7 sec (90%)range 18 7 to251°C

factory instantaneouscalibrationchecked byintercomparison

*M: mobile cart sensor S: stationary sensor

questionnaire and detailed physical measurements tocharacterize the local environment. Each workstationwas visited an average of five times during the week-long period of measurement in each building. The sta-tionary system recorded overall physical trendsthrough the week.

iViOB~LE SYSTENI

Figure 1 shows the cart that carried the mobilemeasuring system A molded fiberglass seat was at-tached to the front of the cart to represent the shieldingeffect of the occupant’s seat. The various sensors used

for physical rneasurements were mounted above andbelow the chair at the 0.1 m, 0.6 m, and 1.1 m levels(representing ankles, mid-body, and head/neck of seated subject). The sensors were protected (at the0.1 m and 1.1 m heights) with a black metal tubingbumper to guard against encounters with office work-ers and furniture. The tubing and sensors were sep-arated by sufficient space to minimize possible effectsof the tubing on the readings. The shelves on the cartbehind the chair contained the remainder of the sys-tem, including signal conditioners, data-recording de-vices, cables, and battery power.

628

As mentioned earlier the measurement cart con-tained three separate data acquisition systems Thefirst, a portable laptop m~crocomputer, was used toadminister the thermal assessment survey to each of-fice worker participating in the study The survey con-sisted of a series of questions and scales addressingthe subject’s response to his or her immediate thermalenvironment. At the time of each workstation visiL thecomputer was placed on the desk and the subjectwas left alone to complete the survey by respondingto a series of questions appearing on the computerscreen (Figures 3a and 4b). Answers (typically yes/no, numerical or positioning of the cursor along ascale) were typed on the keyboard, with results goingdirectly into storage on diskette. To reduce the pos-sibility of typing errors, an opaque plastic cover wasbuilt for the keyboard, exposing only the limited num-ber of keys necessary for answering the question&

The thermal assessment survey used theASHRAE Thermal Sensation and Mclntyre scales asthe primary measures of thermal sensation and com-fort. Data were also collected on a secondary seriesof office work area questions (general comfort, draf-tiness, and brightness using six-point scales) and the

Figure 3a Thermal assessment survey administered onlaptop computer

Figure 3b Cart in position at workstation

appropriateness of 26 affect ratings (adjectives). clothing checklist screen (separate female and maleversions were developed) presented an itemized listof clothing and asked for a rating on a four-point scaleindicating the relative weight of each item. An activitychecklist screen inquired about physical activity, eat-ing, drinking (hot, cold, or caffeinated beverages), andsmoking during the 15 minutes prior to taking the sur-vey. A complete description of the subjective surveymethods employed during the study, including a sepa-rate background information survey, Is given bySchiller et al. (1988a).

The second data acquisition system containedon the mobile cart was a packaged indoor environmentmeasurement system, optimized to measure the fourphysical parameters necessary for the evaluation ofthermal environments: air temperature, humidity, ra-diant temperature asymmetry, and air velocity. Thisspecialized system had the additional advantage thatthe manufacturer’s specifications complied with therequired measurement accuracies of ASHRAE (1981)and ISO (1985). The sensors associated with this sys-tem were positioned to monitor the air temperature,humidity, and air velocity at the &6 m height, the max-imum radiant temperature asymmetry in the horizontaldirection above desk level, and the surface temper-ature of. the chair shell (see Table 1 and Figure 1). set of readings from these sensors was transmittedserially every 10 seconds to a second portable laptopcorn puter, which em ployed custom software to recordthe data on diskette and display current values on itsscreen as a status check for the operator.

Project budget constraints would not allow thepurchase of more than one of the packaged indoorenvironment measurement systems. Therefore, the ad-ditional physical measurement sensors were assem-bled and connected to a general-purpose, battery-operated microdatalogger, and together these formedthe third data acquisition system on the mobile cart.These sensors included thermistors and thermocou-pies for air temperature and globe temperature mea-surements, spherical omnidirectional anemometers forair velocity measurements, and a silicon photometerfor illumination measurements (see Table 1). The mi-crodatalogger controlled the sensor scan rate (onceper second), converted the incoming data to engi-

¯ neering units in real time, and transmitted the data toa cassette tape recorder for final storage.

The use of the multiple data acquisition systemsrequired special procedures to ensure the proper timesequencing of each system over the measurementperiod. At the beginning of each day in the field, allsystem time clocks were synchronized. As all datawere time stamped: at collection during the course ofthe day, this allowed concurrent or sequential datafrom individual systems to be identified and assem-bled into complete data sets during the data analysisphase. In particular, the physical measurements ateach workstation depended on the simultaneous op-eration of two data acquisition systems: the packagedindoor environment system and the microdatalogger-based system. This was accomplished by having the

629

field researcher manually initiate the data collectionprocess for both systerns at the appropriate time. Themeasurement cart could then be left unattended(avoiding undesirable thermal effects of nearby re-search personnel) as the data collection process con-tinued until it was terminated automatically.

The physical data at each workstation were col-lected for a total of five minutes. During this time, allsensors were scanned once per second with the ex-ception of the chilled-mirror dew point sensor, whichproduced a new reading only every two minutes. Thecart was positioned in the vicinity of the subject priorto the measurement period, allowing the transducersto equilibrate in the general area. Following the startof data collection, the first two minutes were used toallow all sensors to achieve final equilibrium with theirsurroundings. Ten-second average data were re-corded for the entire five-minute period along with asingle average value based on the final three-minuteinterval. The three-minute average data served as theprimary values for data analysis and reporting, whilethe ten-second data were retained as a safety rneasurein the event of lost or inconsistent results. For bothlaptop computer-based systems, Custom data man-agement programs were developed to receive thedata inputs, convert them into machine-readable for-mat, and store them for subsequent data analysis.

Both average and sample data from the mobile

cart are shown in Table 2~ The cases presented in thistable were selected to emphasize the utility of havinga high level of detail in certain interior situations. Thedata clearly reveal local environmental asymmetriescaused by solar gain, a personal fan, an open window,and a floor heater.

STATIONARY SYSTEMTernporal variation in each building’s interior ther-

mal environment was monitored throughout the week-long measurement period. The stationary instrumen-tation was rnounted on a large aluminum tripod andplaced in a location representative of the areas beingmonitored (typically an unoccupied workstation). Theenvironmental parameters monitored by the stationarysystem included air temperature at the 0.6 rn, 1.1 m,and 1.7 m heights and globe temperature, air velocity,and illumination at the 1.1 m height (see Table 1). general-purpose microdatalogger, identical to the oneused on the mobile cart, controlled the data collectionproces& All sensors were scanned every 10 seconds,and 10-minute average values were calculated andrecorded. The stationary data provided a continuousrecord of trends in interior conditions that could notbe detected by the roving measurement cart. The datawere used primarily to help diagnose effects observedin the mobile measurements.

TABLE 2Sample Data

Summer Winter

Physical Paramenter Units

average small operable high average small direct highvalues portable window internal values portable sunlight radiantentire fan nearby gains entire floor on desk asymmeteryseason1 season1 heater

Air Temperature 0.1 rn °C 23.16 2&89 22.47 29.45 22.5 32.8 22.1 .......... 23.9Air Temperature 0.6 m °C 23.16 27.58 22.16 29.4 22°64 31 23.2 24Air Temperature 1.1 m °C 23.58 28.34 22.27 29.64 23.14 25,5 23,19 25

GIobe Temperature 0.1 m °C 23.2 27.08 22.5 29.34 22.57 32.2 22.2 24.2GIobeTemperature0.6m °C 23,42 27.71 22.56 29.49 22.9 29.5 23.5 25GIobeTemperature 1.1 m °C 23,65 28.45 22.42 29.62 23.24 25 25 25Radiant Asymmetry2 °C 0.22 0.32 0.28 - 1.68 - 0.08 - 0.1 5.9 11.7Dew Point Temperature °C 15.12 12.2 13.9 14 7.3 6~84 4.39 7.59

Air Velocity 0.1 m m/s 0.08 0.43 0.19 0~93 0.06 0.2 0.04 0~46Air Velocity 0.6 m m/s 0.11 0.28 0.22 1.35 0.04 0.54 0.07 0~13Air Velocity 1.1 m m/s 0.11 0.22 0.3 0.6 0.07 0.15 0.06 0,12

ET* °C 23.53 27.36 22.66 29.13 22.53 27.43 23.21 24.67CIo clo 0.52 0.7 0,54 0.46 0,58 0,55 0,62 0.55Illuminance lux 776 1051 613.4 902.5 913 571.6 11434 3388(horiz. plane)

ASHRAE Vote3 0.28 3 0.5 2 0.18 1 2 1Mclntyre Vote4 0.24 1 0 1 0.1 0 1 1General Comforts 4.38 1 5 2 4.34 5 3 3Ventilation Comforts 3.55 5 5 4 3.5 3 3 3

These averages include data from 1308 workstation visits during the winter season and 1034 during the summerThe radiant asymmetry measurement is the difference between the "A" and the "B" side hence a negative value indicates the "B" side was warmer than

the "A" sideOur convention was to orient the urobe such that the "A" side was aimed toward the window (if any) in the workspace

ASHRAE scale: -3 (too cool) to +3 (too warm)Mclntyre scale: -1 (I want to be watt’net), 0 (I want no change), + 1 (I want to be cooler)General Comfort scale: " (uncomfortable) to 6 (comfortable)Ventilation Comfort scale: 1 (stuffy) to 6 (drafty)

630

FIELD METHODSField researchers (typically one or two per day)

spent a total of one week per season in each monitoredbuilding to set up equipment, collect data, and packup. A shipping crate was constructed to protect themeasurement cart and its instrumentation duringtransit by pickup truck from one building to the next.A large suitcase was modified with foam pads to ac-commodate the stationary measurement equipmentduring transport. These custom-built packages wereextremely important to the smooth operation of the fieldstudy~ They performed adequately for the duration ofthe study, as no catastrophic damage was experi-enced during the many months of work in the field.

Within each building, a secure "homebase" roomwas established in which the measurement cart couldbe parked and its batteries recharged overnighL Themeasurement cart with its fully charged data acqui-sition systems was then capable of collecting datathroughout an 8- to 10-hour working day without furthercharging. A detailed set of instructions was availablefor the various field researchers (a total of eight indi-viduals were used during the project) to follow duringmorning start-up, evening take-down, and trouble-shooting procedures.

In order to study office workers in their normalwork environment, the measurement procedure wasdesigned to minimize disruptions to the subjects. Theprotocol for each workstation visit and approximatelength of time for each task was as follows:

1. Researcher approaches subject and, if con-venient to subject, presents the survey computer (1min [Figure 4a]) while the cart is not visible to thesubject;

2. Subject completes the thermal assessmentsurvey (3 to 10 min [Figure 4b]);

Researcher approaches subjectOccasionally, arrangements forthe timing of the next visit couldbe made at the conclusion of theprevious visit But usually,subiects were approached in arandom order iF the time of thelast visit was not too recentdepending on how busy theyappeared

After finding a willing subject,the researcher initialized thesoftware in the laptop computer,identifying the subject by anumber and then retreated to arespectable distance. While thesubject took the survey, theresearcher scouted possibilitiesfor the next visit andtook noteson anything unusual in thesubject’s clothing, attitude, ~dproximity and/or status ofnearby fans, floor heaters andoperable windows,

When the subject has finishedwith the survey the researcherreturns and requested the subjectto leave his or her desk for fiveminutes, The mobile cart is putinto place and the measurementsequence initiated All sensorsare scanned at least every secondfor five minutes The first twominutes were used to allow thesensors to equilibrate and a singleaverage value is calculated basedon the final three-minute interval

Figure 4 Workstation measurement protocol

3. Subject leaves desk, and field researcher po-sitions measurement cart in place of the subject’s chair(1 min);

4. Thermal measurements are recorded (5 min[Figure 4c]);

5. During survey and measurement periods, re-searcher records additional observations andsketches, takes photographs, and arranges for nextworkstation visit.

In this procedure, it was important for the re-searcher to leave the workstation area during the sur-vey period to avoid disturbing the subject. Theadditional observations and recordings that the re-searcher collected in item 5 above proved to be quitevaluable in interpreting unexpected or inconsistentdata during the subsequent data analysis phase. Thecollected information included (1) sketches of the of-fice layout and cart position (first visit only); (2) tographs of the work area using fisheye and wide-angle lenses (first visit only); (3) location, type, status (on/off) of equipment affecting local thermalconditions (e.g, fans, electric heaters, HVAC diffus-ers, computer equipment, etc.); (4) openable windowand movable shade positions; (5) unusual clothing the subject; (6) unusual subject behavior patterns; and(7) observable thermal conditions (e.go, draft, incidentbeam sunlight, etc.).

DATA FLOW PATHThe daily yield of data from the operation of the

systems was sizable. Data from the indoor environ-ment measurement system, accumulated by its ded-icated laptop microcomputer, were formatted into twoseparate ASCII microcomputer files° The first file con-tained average data from each transducer for 1-sec-ond periods, while the second provided summary dataaveraging the last 3 minutes of each workstation visit.The retrieval of these data, plus the diskette-baseddata from the thermal assessment survey laptop, re-quired only the exchange of diskettes at day’s end.The microdatalogger stored similar physical data fromsensors at O1 m and 1~1 m heights on a proprietarycassette tape storage system. These data were trans-ferred to microcomputer disk files through a specialcassette tape interface at the end of each day. A sim-ilar cassette storage system was used for the station-ary data acquisition system°

Data collected in machine-readable form wereeasy to review for errors on a daily basis. As the day’sdata scrolled by on the screen during transfer, theresearcher scanned each channel to confirm that allsensors were operating properly.

The final step in the data flow path was the as-sembly of all three-minute average data from a build-ing for a given week into a spreadsheet for final errorchecking and analysis. Using the spreadsheet, prob-lems noted during data collection were resolved alongwith a careful rechecking of time synchronization be-tween the data files. The summary data for each build-ing were then transferred to mainframe computers forstatistical analysis. The project’s data spreadsheetsare available from the authors.

631

TROUBLESHOOTINGTwo instrumentation problems are worth noting:

dirt on the dew point sensor mirror and the failure ofboth thermistors on the mobile cart. The dew pointsensor design assumes that light scattered from itsmirror surface is due to condensation. If soiling at themirror surface scatters light instead, the result is aninaccurate reading Measurements were found to beincorrect twice during the course of the project atroughly equal intervals, suggesting a more rigorouscleaning schedule for the mirror surface. The failureof the thermistors occurred when the battery chargeron the cart caused a voltage spike, overloading theircircuitry. This problem was addressed by retrofittingtype-T thermocouples to these locations.

A final observation about cabling: data transfercables on a mobile system of this type must be ableto withstand frequent flexing and other stresses. Atemporary failure of one DAS occurred when a cabledeveloped an intermittent short at one of the plugs,presurnably from recurring movement of the cablewithin the plug body.

CONCLUSIONSIn retrospect, a nurnber of observations arise

concerning the field performance of these data ac-quisition systems and the utility of the detailed datasets they collected. The instrumentation performedwell in a challenging program and provided more than2300 sets of detailed measurements at a performancereliability of approximately 98%. The battery-poweredmobile cart could complete an entire day’s measure-ments on an overnight recharging and proved idealfor roaming the corridors of an office (in one casecovering several miles a day). The laptop microcom-puter proved a convenient device for implementingthe thermal assessment survey by providing both anovel, easy medium for the subjects and machine-readable data for the researchers. The daily collectionand review of field data were remarkably manageabledue to the direct storage of all monitored data in ma-chine-readable form.

The field measurements included all variablesspecified in ASHRAE (1981) and ISO (1985) includingrepetition of measurements at 0.1 m, 0.6 m, and 1.1m above the floor. For many workstation visits thisrepetition was excessive; average data for the 0.6 mheight alone would have been sufficient to character-ize the local environment. However, other workstationsmade it clear that data in this detail are necessary tofully understand the thermal character of workstationssubject to nonuniformities in the surrounding environ-ment. Indeed, many sources of discomfort (e.g.,draffs, cold floors, asymmetric radiant exchanges)represent just this type of nonuniformity as do marlyof the workstation-based efforts at restoring comfort(e.g., fans and space heaters) and the effects of spe-cific workstation geometries. The availability of micro-processor-based measurement equipment lessenstile burden of making detailed measurements. Thevalue of the entire ASHRAE (1981) measurement setseems well worth the modest incrernental cost of ad-

ditional transducers and the slight increase in datamanagement overhead. It is clear from the study thatthe measurement specifications of ASHRAE (1981)can be implemented in the field.

The great majority of our analysis was based onaverage data for the last three minutes of each work-station visiL The more detailed data gathered by themobile cart instrumentation (10-second average data)produced extensive data files that received relativelylittle analysis~ However, these large data sets didprove to be very useful on several occasions, partic-ularly when a portion of the summary data perisheddue to accidental erasure or an unusual transducerreading merited closer inspection.

Although it was scaled 1.4 m by 0.45 m in plan,the cart moved easily through the interior spaces ofthe study. Moving the cart between the test buildingswas more difficult. The cart with well-wrapped cableruns, securely mounted transducers, and a series ofsecured instrument boxes--was not easily disassem-bled. For transport between buildings, the entire cartwas loaded into a custom-built packing crate. Thecrate’s large dimensions, besides being clumsy, re-quired the use of a light pickup truck or van for trans-port~

The coordination of time-stamped data from sev-eral data acquisition systems, although a manageabletask, did require a weekly commitment of time for dataformatting. A desire to simplify management of thefield data sets, plus the appeal of a more transportablesolution, has led to a new approach for the next versionof the mobile cart. This version is based on a collaps-ible wheelchair chassis that carries a single data ac-quisition system for all physical measurements andfeatures readily demountable components for easytransport.

In conclusion, experience with the field data sys-tems of ASHRAE research project RP-462 was posi-tive. We recommend these systems for furtherapplication as practical and useful tools in the studyof thermal comfort.

ACKNOWLEDGiViENTSThe authors gratefully acknowledge the assistance of

ASHRAE, Inc, and the IBM Dace Grant Program for provid-ing funds and computer equipment for this research Prin-cipal investigators Gall Schiller and Edward Arens meritspecial recognition for a well-run project. Eleanor Lee pro-vided solid graphic assistance. Nora Watanabe of the Cen-ter for Environmental Design Research served as an adeptadministrator Professor Kenneth Craik of the Institute of Per-sonality Assessment Research, U.C. Berkeley, and his re-search assistant, Karl Dake, contributed to the developmentof the thermal assessment survey

REFERENCESASHRAE 1981. ASHRAE Standard 55-1981, "Thermall en-

vironmental conditions for human occupancy." Atlanta:American Society of Heating, Refrigerating, and Air-Con-ditioning Engineers, Inc

ASHRAE 1985. ASHRAE handbook--1985 fundamentalsAtlanta: American Society of Heating, Refrigerating, andAir-Conditioning Engineers,

ASHRAE 1987 ASHRAE handbook--1987 HVAC systems

632

and applications, chapter 16 Atlanta: American Society Humphreys, MA 1976. "Field studies of thermal comfortof Heating, Refrigerating, and Air-Conditioning Engi- compared and applied"Bui/ding Services Engineer, Vol.neers, Inc. 44, pp. 5-27.

Auliciems, A, and Dedear, RJ 1984 "Impacts of air cooling Humphreys, M.A 1977 "The optimum diameter of a globeon Europeans in a tropical climate: a case study in Dar- thermometer for use indoors" Annals of Occupationalwin." In Indoor Air Buildings, Ventilation and Thermal Hygiene, Vol. 20, pp 135-140Climate, eds Berglund, B.; Lindvall, G.; and Sundell, J., ISQ 1984~ International Standard 7730, "Moderate thermalpp, 337-342~ Stockholm; Swedish Council for Building environmentsidetermination of the PMV and PPD in-Research. dices and specification of the conditions for thermal com-

Baillie, AP,; Griffiths, I.D,; and Huber, JIW. 1987. "Thermal fort." Geneva: International Standards Organization.comfort assessment: a new approach to comfort criteriain buildings." Department of Psychology, University ofSurrey, Guildford, UK, Final Report ETSU S-1177

Brill, M 1984 Using office design to increase productivi(y.Buffalo, NY: Workplace Design and Productivity, Inc

Cena, K; Spotila, J.R; and Avery, H.W. 1986. "Thermal com-fort of the elderly is affected by clothing, activity, andpsychological adjustment." ASHRAE Transactions, Vol.92, Part 2

Dedear R J., and Auliciems, A 1985 "Validation of the pre-dicted mean vote model of thermal comfort in six Aus-tralian field studies," ASHRAE Transactions, Vol. 91, Part2.

Dillon R., and Vischer, J.C 1987. "Derivation of the tenantquestionnaire survey assessment method: office buildingoccupant survey data analysi&" Public Works Canada,AES/SAG 1-4:87-89

Fanger, P.O 1970 Thermal comfort, Copenhagen: DanishTechnical Press

Fishman, DS. and Pimbert. S.L 1978. "Survey of subjectiveresponses to the thermal environmenl in offices" Pro-ceedings of the Intemational Indoor Climate Symposlum,30 August-1 September, Copenhagen

Gagge, A.P., and Nevins, RG 1976 "Effect of energy con-servation guidelines on comfort, acceptability, andhealth." Final Report of Contract #00-04-51891-00, Fed-eral Energy Administration, Washington,

Harris L., and Associates. 1980. "The Steelcase nationalstudy of office enwronments, no II Comfort and produc-tivity in the office of the 80s." Grand Rapids, MI: Steel-case, nc.

Howell, W,C and Kennedy, P.A 1979. "Field validation ofthe Fanger thermal comfort model "Human Factors, VoL21 No. 2, pp. 229-239.

Howell, W,C. and Stramler, C.S 1981 "The contribution ofpsychological variables to the prediction of thermal com-fort judgements in real world settings." ASHRAE Trans-actions Vol. 87, Part 1.

ISO. 1985. International Standard 7726, "Thermal environ-ments instruments and methods for measuring physicalquantities" Geneva: International Standards Organiza-tion

Lammers, J.T.H.; Berglund, LG.; and Stolwijk, JA J. 1978,"Energy conservation and thermal comfort in a New YorkCity high rise office building." Environmental Manage-ment, Vol 2, No. 2, pp, 113-117

Mclntyre, D 1980 Indoor climate. London, Applied SciencePublishers

Purcell, A.T. and Thorne, R.H. 1987. "The thermal environ-ment and level of arousal, field studies to ascertain theeffects of air-conditioned environments on office work-ers." Architectural Science Review, Vol 30, pp 91-115.

Roark, K S. and Dowall D.E 1986. "The setting for change "’In: The Changing Office Workplace, ~)p. 1-16, eds Black,J.T; Roark, K.S.; and Schwartz, L.S Washington, DC:Urban Land Institute (ULI) and Building Owners and Man-agers Association (BQMA)International

Schiller, G.E. 1990 "A comparison of measured and pre-dicted comfort in office buildings." ASHRAE Transac-tions, Vol. 96, Pad 1

Schiller, G.E.; Arens, EA Bauman. F S.; Benton, C,C.;Fountain, M.; Doherty, T and Craik, K 1988a "A fieldstudy of thermal environments and comfort in office build-ings," Final Report, ASHRAE Research Project RP-462Berkeley: Center for Environmental Design Research,University of California.

Schiller, GE Arens, E.A.; Bauman, RS; Benton, C C:Fountain, VI.; and Doherty, T 1988b "A field study ofthermal environments and comfort in office buildings"ASHRAE Transactions, Vol 94 Part 2.

Woods, J.; Drewry, G.; and Morey, P 1987 "Office workerperceptions of indoor air quality effects on discomfort andperformance" In: Proceedings, 4th International Confer-ence on Indoor Air Quality and Climate, Berlin, 17-21August, pp 464-468 Berlin: Institute for Water, Soil, andAir Hygiene.

633