i Developing an Adaptive Model of Thermal Comfort and Preference FINAL REPORT ASHRAE RP- 884 March 1997 Richard de Dear , Gail Brager ` , Donna Cooper Macquarie Research Ltd., Macquarie University, Sydney, NSW 2109 AUSTRALIA ` Center for Environmental Design Research, University of California, Berkeley, CA 94720 USA “Results of Cooperative Research between the American Society of Heating, Refrigerating and Air Conditioning Engineers, Inc., and Macquarie Research, Ltd.”
312
Embed
Developing an Adaptive Model of Thermal Comfort …...i Developing an Adaptive Model of Thermal Comfort and Preference FINAL REPORT ASHRAE RP - 884 March 1997 Richard de Dear À, Gail
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
i
Developing an Adaptive Model of Thermal Comfort and Preference
FINAL REPORT
ASHRAE RP- 884
March 1997
Richard de DearÀ, Gail BragerÁ, Donna CooperÀ
À Macquarie Research Ltd., Macquarie University, Sydney, NSW 2109 AUSTRALIA
Á Center for Environmental Design Research, University of California,
Berkeley, CA 94720 USA
“Results of Cooperative Research between the American Society of Heating, Refrigerating
and Air Conditioning Engineers, Inc., and Macquarie Research, Ltd.”
ii
iii
i
TABLE OF CONTENTS iii
ACKNOWLEDGMENTS vii
EXECUTIVE SUMMARY ix
CHAPTER 1 - INTRODUCTION & BACKGROUND 1
1.1. Introduction 1
1.2. Defining the adaptive process 3 1.2.1. The dialectic of contemporary thermal comfort theory 3 1.2.2. The “adaptive” hypothesis 4
1.3. A conceptual model of adaptation -- feedback loops 6 1.3.1. Behavioral feedback - adjustment 8 1.3.2. Physiological feedback -- acclimatization 10 1.3.3. Psychological feedback -- habituation and expectation 12
1.4. Literature review 13 1.4.1. Climate chamber evidence for adaptation to climate 13 1.4.2. Field evidence for adaptation 15
1.4.2.1. The earlier field evidence for adaptation 16 1.4.2.2. Analysis of neutral temperatures using recent field experiments 18 1.4.2.3. Evidence for behavioral adaptation - personal/environmental adjustment 22 1.4.2.4. Evidence for psychological adaptation - expectation and context 23
1.5. Implications for RP-884 26 1.5.1. Lessons from static heat balance models 26 1.5.2. Time scales of thermal adaptation 29
1.6. Aims 31
CHAPTER 2 - METHODS 33
2.1. Overview of the RP-884 approach 33
2.2. Establishing the database for RP-884 36 2.2.1. Sourcing the raw data 36 2.2.2. Ratings of raw data submitted to RP-884 40
2.3. Raw data standardisation 41 2.3.1. Creation of a standard data template 41 2.3.2. Consistent mean radiant temperatures within the database. 42 2.3.3. Consistent comfort index calculations within the database 42 2.3.4. Predicted draft risk index (PD) 43 2.3.5. Clothing insulation in the ASHRAE RP-884 database 44
2.3.5.1. Discrepancies between field estimation methods for clo. 45 2.3.5.2. The chair insulation effect 49
2.4. Developing an index for perceived thermal control 49
iv
2.5. Thermal acceptability issues within the RP-884 database 51 2.5.1. Developing a proxy variable for thermal acceptability based on thermal
sensation votes. 51 2.5.2. Rating buildings in terms of their compliance with ASHRAE Standard 55
acceptable indoor climate guidelines 52
2.6. Outdoor meteorological/climatological data for the data base 52 2.6.1. Appending outdoor weather observations to each row of data 52 2.6.2. Climate classification applied to RP-884 raw data 53
2.7. Subdivision of the standardized field experiments 54
2.8. The meta-analysis 54 2.8.1. The unit of analysis for the RP-884 meta-analysis 54 2.8.2. Meta-file’s structure and coding conventions 55 2.8.3. General assumptions within the statistical meta-analysis 55 2.8.4. Statistical treatments on the various subjective thermal ratings 56 2.8.5. Preferred temperatures 59
2.9. The RP-884 database in the public domain and disseminated via the world wide web 60
2.10. Summary of the methods used in RP-884 64
CHAPTER 3 - BASIC RESULTS 67
3.1. Interactions with indoor climate 67 3.1.1. Thermal sensation 67
3.1.1.1. Dependence of thermal sensation on indoor operative temperature 68 3.1.1.2. Dependence of thermal sensation on indoor ET 69 3.1.1.3. Dependence of thermal sensation on PMV 70 3.1.1.4. Dependence of thermal sensation on indoor SET 71
3.1.2. Thermal neutrality 72 3.1.2.1. Neutral operative temperatures (neut_top) 72 3.1.2.2. Neutral effective temperatures (neut_et) 74 3.1.2.3. Neutral predicted mean votes (neut_pmv) 74 3.1.2.4. Predicted neutralities with the PMV heat balance model 75 3.1.2.5. Neutral standard effective temperatures (neut_set) 77
3.1.3. Thermal acceptability and indoor climate 78 3.1.3.1. Relationship between direct and inferred thermal acceptability 78 3.1.3.2. Directly determined thermal acceptability 80 3.1.3.3. Thermal acceptability inferred from thermal sensation 83 3.1.3.4. Thermal sensitivity and the range of thermally acceptable temperatures. 84
3.1.4. Thermal preferences and indoor climate 89 3.1.5. Comparisons between neutral and preferred temperatures indoors. 91 3.1.6. Behavioural adjustments to indoor climate 93
3.2. Interactions with outdoor weather and climate 102 3.2.1. Thermal neutrality and outdoor climate 102
3.2.1.1. Seasonal comparisons 103 3.2.1.2. Dependence of observed neutrality on outdoor climate 104 3.2.1.3. Analysis of predicted neutralities with respect to mean outdoor temperature 106
3.2.2. Thermal acceptability and outdoor climate 108 3.2.3. Thermal preference and outdoor climate 110 3.2.4. Behavioral responses to outdoor climate 113
3.2.4.1. Indoor clothing and outdoor climate 114 3.2.4.2. Metabolic rate indoors related to outdoor climate 115 3.2.4.3. Indoor air speeds in relation to outdoor climate 116
3.3. Influence of building characteristics on thermal comfort 118 3.3.1. HVAC versus natural ventilation 118
3.3.1.1. Thermal sensation and sensitivity in HVAC versus naturally ventilated buildings 119
3.3.1.2. Thermal acceptability in HVAC versus naturally ventilated buildings 121 3.3.1.3. Thermal preferences in HVAC versus naturally ventilated buildings. 122
3.3.2. Personal environmental control 124 3.3.3. Building occupancy types - offices, residential and industrial 127
3.4. Summary of basic results 130 3.4.1. Summary of thermal sensation, acceptability and preference 131 3.4.2. Summary of thermal sensitivity and behavioral thermoregulation 133 3.4.3. Summary of the effects of outdoor climate on thermal perception indoors 134 3.4.4. Summary of the effects of contextual factors and perceived control 135
CHAPTER 4 - TOWARDS ADAPTIVE MODELS 139
4.1. The semantics of thermal comfort 139
4.2. Comparison of RP-884 models with earlier adaptive model publications 141
4.3. Comparison of RP-884 models with the PMV “static model” 145 4.3.1. Comparisons within the centrally conditioned building sample 146 4.3.2. Comparisons within the naturally ventilated building sample 150
4.4. Adaptive models for acceptable ranges of indoor temperatures 152
CHAPTER 5 - VARIABLE TEMPERATURE STANDARDS 155
5.1. A variable temperature standard for application in buildings with centrally controlled HVAC 155
5.1.1. Purpose 155 5.1.2. Scope 156 5.1.3. Definitions 156 5.1.4. Conditions for an acceptable thermal environment. 161
5.2. A variable temperature standard for application in naturaly ventilated buildings 165
vi
5.2.1. Purpose 165 5.2.2. Scope 165 5.2.3. Definitions 166 5.2.4. Conditions for an acceptable thermal environment. 168
BIBLIOGRAPHY 171
APPENDIX A - THERMAL SENSATION AND NEUTRALITY FOR EACH BUILDING IN THE RP-884 DATABASE 185
APPENDIX B - PREFERRED TEMPERATURE FOR EACH BUILDING IN THE RP-884 DATABASE 227
APPENDIX C - SUMMARY OF THE ORIGINAL FIELD EXPERIMENTS COMPRISING THE ASHRAE RP-884 DATABASE 235
C.1. Project Title - ASHRAE TC 2.1 sponsored RP-702 236
C.2. Project Title - Thermal comfort studies in modern industrial buildings. 239
C.3. Project Title - Doctoral dissertation. From comfort to kilowatts: An integrated assessment of electricity conservation in Thailand’s commercial sector. 242
C.4. Project Title - The CSAA, Antioch (1995) component of the Advanced Customer Technology Test (ACT2) project. 245
C.5. Project Title - Higher PMV causes higher energy consumption in air- conditioned buildings: a case study in Jakarta, Indonesia. 248
C.6. Project Title - Montreal ASHRAE RP-821. 250
C.7. Project Title - Richard de Dear’s PhD research project in Australia. 253
C.8. Project Title - A field study of thermal comfort using questionnaire software. 256
C.9. Project Title - “Thermal comfort in Pakistan.” 258
C.10. Project Title - Comfort criteria for passively cooled buildings. a PASCOOL task. 262
C.11. Project Title - Developing indoor temperatures for naturally ventilated buildings. 264
C.12. Project Title - Mixed mode climate control: some hands-on experience. 267
C.13. Project Title - ASHRAE sponsored RP-462. San Francisco area. 269
C.14. Project Title - A field investigation of thermal comfort environmental satisfaction and perceived control levels in UK office buildings, University of Liverpool. 272
C.15. Project Title - Thermal comfort in the humid tropics: field experiments in air conditioned and naturally ventilated buildings in Singapore. 275
C.16. Project Title - The Steelcase building. Grand Rapids Michigan, US 277
C.17. Project Title - Sunset building: a study of occupant thermal comfort in support of PG&E’s Advanced Customer Technology Test (ACT2) for maximum energy efficiency 279
C.18. Project Title - The Verifone building, a component of the Advanced Customer Technology Test (ACT2) Project. 282
APPENDIX D - CLIMATE CLASSIFICATION 285
APPENDIX E - CODEBOOK FOR RAW DATA IN RP-884 DATABASE 287
vii
APPENDIX F - CODEBOOK FOR THE RP-884 META-ANALYSIS 291
APPENDIX G - AULICIEMS’ ADAPTIVE MODEL DATABASE 295
ACKNOWLEDGMENTS
The successful completion of this project depended very heavily on the willingness of field
researchers to make available their raw data for re-analysis and incorporation into the RP-
884 database. In particular, we would like to thank the following contributors:
Dr. Jill Brown, formerly of University of Wales, Cardiff; Dr. John Busch Jr. Lawrence
Berkeley Labs., California; Prof. Cris Benton, CEDR, University of California at Berkeley;
Dr. Tri Karyono, Agency for the Assessment and Application of Technology (BPPT),
Jakarta, Indonesia (formerly of the Department of Architecture, University of Sheffield, UK);
Dr. Giovanna Donnini, formerly of Auger, Donnini and Nguyen Inc, Montreal, Canada; Dr.
Guy Newsham, Institute for Research in Construction, National Research Council of
Canada, Ottawa; Fergus Nicol, School of Architecture, Oxford-Brookes University, UK.;
Iftikhar Raja, School of Architecture, Oxford-Brookes University, UK; Prof. Nick Baker, The
Martin Centre for Architecture and Urban Studies, University of Cambridge, UK; David
Rowe, Dept. of Architectural and Design Science, University of Sydney, Australia; Dr Ruth
Williams, The Building Services Research and Information Association, UK (formerly
Liverpool University, UK); Fred Bauman, CEDR, University of California at Berkeley.
RP-884 also depended on weather and climate data resources. Such data was required for
the relevant sites and periods covered by field experiments within the database. Apart from
resources available on the WWW and various CD-ROM publications, the following
organisations provided data. The Australian Bureau of Meteorology’s National Climate
Centre supplied meteorological data for the Melbourne, Brisbane and Darwin field
experiments. The Oxford University Radcliffe Observatory supplied meteorological
observations for some of the UK experiments. Macquarie University’s Meteorological Site
supplied observations for the Sydney field data. The US National Climate Data Center
(NCDC) supplied meteorological data for the Californian experiments. Meteorological
viii
observations for Grand Rapids were supplied by the Michigan State Climatologist.
Bangkok meteorological data were supplied by the Royal Thai Meteorological Department.
Special thanks are also due to Andris Auliciems of the University of Queensland, Fergus
Nicol of Oxford-Brookes University and Michael Humphreys of Oxford University for their
pioneering work in the area of adaptive models and also for their encouragement at various
stages during the ASHRAE RP-884 project.
ix
EXECUTIVE SUMMARY
One of the more contentious theoretical issues in the applied research area of thermal
comfort has been the dialectic between “adaptive” and “static” models. Apart from having
disparate methodological bases (the former laboratory-experimental, the latter field-based),
the two approaches have yielded starkly differing prescriptions for how the indoor climate of
buildings should be managed. These prescriptions carry implications for the types of
permissible building designs, the means by which their thermal environments are controlled,
and the amounts of energy they consume in the production of habitable indoor climates.
Static models have led to indoor climate standards that have been universally applied
across all building types, are characterised by minimal recognition of outdoor climatic
context, and are contributing to an increased reliance on mechanical cooling. In contrast,
proponents of adaptive models have advocated variable indoor temperature standards that
more fully exercise the adaptive capabilities of building occupants. This approach
potentially leads to more responsive environmental control algorithms, enhanced levels of
occupant comfort, reduced energy consumption, and the encouragement of climatically
responsive building design.
Despite these apparent differences, our review of the research literature emerging from both
approaches indicated that this seemingly irreconcilable split was primarily the result of
narrow definitions of the term “thermal adaptation”, and that there were opportunities to
bridge some of the gap between the hypotheses. We suggest that human thermal
adaptation is comprised of three distinct yet interrelated processes - behavioral,
physiological, and psychological. The adoption of this tripartite definition goes some way
towards reconciling the static and adaptive approaches and the indoor climate standards
derived from them.
This project’s principal objective was the proposal of a variable temperature standard based
on the adaptive approach. Where it differs from earlier attempts is in the quality control
applied throughout its adaptive modelling method. About 21,000 sets of raw thermal
comfort data from 160 buildings were collected from most of the thermal comfort field
research groups around the world who are currently active. Data selection criteria
x
emphasized precision of indoor climatic instruments, while data assimilation involved a
variety of questionnaire standardization processes. For example, each one of the over
21,000 building subjects’ clothing thermal insulation estimates was transformed into an
equivalent clo value using consistent procedures specified in ASHRAE Standard 55-1992.
The thermal effects of chairs for seated subjects was also included. For each set of raw
data, outdoor meteorological and climatological data were appended to the RP-884
database. All indoor and outdoor thermal indices were recalculated using a standard
Figure 1.7: Thermal comfort experiments in the field: Observed and predicted neutralities in relation to outdoor climate
Also depicted in Figure 1.7 are some results from six ASHRAE-sponsored Class I field
experiments in climate-controlled buildings across a variety of climatic contexts. Two
experiments are from San Francisco (Schiller et al 1988a; Schiller 1990). Another two
Code Location & season Climate-controlled or Free Running
Author
P Montreal-Winter CC Donnini et al (1996) A San Francisco-winter CC Schiller et al (1988a) O Montreal-Summer CC Donnini et al (1996) B San Francisco-summer CC Schiller et al (1988a) M Townsville-Dry CC de Dear + Fountain (1994) C Melbourne-summer FR de Dear + Auliciems (1985) D Melbourne-summer CC de Dear + Auliciems (1985) E Brisbane-summer CC de Dear + Auliciems (1985) F Brisbane-summer FR de Dear + Auliciems (1985) G Darwin - Dry CC de Dear + Auliciems (1985) N Townsville-Wet CC de Dear + Fountain (1994) H Singapore FR de Dear et al (1991) I Singapore CC de Dear et al (1991) J Bangkok FR Busch (1990) K Darwin-Wet CC de Dear + Auliciems (1985) L Bangkok CC Busch (1990)
ASHRAE RP-884 Final Report
Introduction & Background page MRL Australia 21
are from tropical Townsville (de Dear and Fountain 1994), and another pair from
Montreal (Donnini et al, 1996).
Plotted along with the San Francisco observed neutralities are some predictions from
Auliciems' (1983) adaptive model as well as Fanger's PMV (heat balance) model, after
the effect of chair insulation (0.15 clo) has been added to Schiller et al's published
clothing insulation estimates. Clearly in both seasons, the adaptive model comes very
close to observation, but so too does the static heat balance model. This general
pattern of consistency between neutralities observed in air-conditioned buildings and
PMV predictions also extends to the more recent ASHRAE-sponsored studies in office
buildings located in a hot-humid climate (de Dear and Fountain 1994) and cold climate
(Donnini et al, 1996).
Busch's (1990) field experiments in office buildings in tropical Bangkok have also been
included in Figure 1.7. Both climate controlled (air-conditioned) and free running
buildings were studied, so a diverse range of thermal environments was covered by the
sample size of 1146. For the climate controlled buildings, neutrality was established at
24.5°C (code L in Figure 1.7), within a degree of the PMV prediction based on Busch's
mean clo value of 0.56 plus some chair insulation (0.15 clo). In Bangkok's free running
buildings, Busch observed a neutrality of 28.5°C (code J in Figure 1.7), which appears
to be over three degrees (K) warmer than predicted by Fanger's PMV. Auliciems'
(1983) adaptive model, on the other hand, came within half a degree of the observed
result. Busch suggested that the lighter clothing and higher local wind explain most of
the disparity between observed thermal neutralities in the naturally ventilated and air-
conditioned buildings, implying that behavioral adjustments were playing a strong
adaptive role. But there are clearly other factors at play, as well. Noting that clothing
and air velocity are used as input parameters to the heat balance models, the fact that
PMV still underestimates neutrality suggests that occupants were influenced by other
modes of adaptation unaccounted for by the heat balance inputs. In particular, PMV’s
underestimation of thermal neutrality more significantly in the free running building
sample than in the climate controlled building sample suggests that context and
adaptive opportunity can influence expectations and thermal response to the indoor
environment.
ASHRAE RP-884 Final Report
Introduction & Background page MRL Australia 22
Another example of this is found in a more recent field experiment, in which de Dear et
al (1991c) examined climate controlled office buildings and free running residential
apartment blocks in equatorial Singapore. As seen in Figure 1.7 (code I), the observed
neutrality of 24.2°C in the air conditioned buildings was accurately predicted by both the
adaptive and heat balance models after 0.15 clo chair insulation was added to clothing
estimates. But as with Busch's Bangkok experiment, the 28.5°C neutrality observed in
Singapore's naturally ventilated apartment buildings (code H) was most closely
approximated by the adaptive model with a prediction of 27.2°C.
1.4.2.3. Evidence for behavioral adaptation - personal/environmental adjustment
There have been a few studies that examined direct evidence of exercised control, or
adjustment. One of the earlier studies that looked closely at clothing patterns was by
Fishman and Pimbert (1982), who studied 26 subjects in a UK office building for an
entire year. The estimated clo values of the Watson House sample had a strong linear
dependence on outdoor weather and season, especially in the case of women subjects,
with a regression gradient of -0.02 clo units per degree of outdoor mean weekly
temperature. This supports the hypothesis that the statistical dependence of indoor
neutrality on outdoor climate may, in part, be due to behavioral adjustments that directly
affect the heat balance, rather than acclimatization or habituation.
This hypothesis is also supported by the work of Humphreys (1994b) and Nicol et al
(1994), in which a study of naturally ventilated buildings in North West Pakistan
concluded that the office workers were comfortable across a wide range of seasonal
temperatures (neutralities varying between 15.7°C in winter, and 26.4°C in summer).
They also concluded that 1~B of the seasonal changes in comfort temperature could
be attributed to the flexibility in the traditional Pakistani clothing worn.
Personal behavioral adjustments over time were looked at in an exploratory study by
Nicol and Raja (1996) in the UK. They found that clothing changes were more strongly
dependent on the succession of outdoor temperatures that occurred prior to the
measurement, compared to the instantaneous or daily mean outdoor temperature, or the
instantaneous indoor temperature. This suggests the importance of time-series
measurements in future field studies designed to evaluate the effect of behavioral
ASHRAE RP-884 Final Report
Introduction & Background page MRL Australia 23
adaptation on thermal comfort. Posture is another example of behavioral adaptation,
and they found a correlation with temperature such that posture would change to
increase the effective body surface area available for dry and latent heat exchange as it
got warmer.
In addition to adjusting to the environment, one can directly manipulate the environment
itself. Baker and Standeven (1994) used hourly questionnaires to ask whether subjects
had made adjustments to their clothing or to furniture, doors, windows, shades, fans or
any other part of the building to improve their comfort. Results indicated extensive
occupant-environment interaction - for 23 subjects in 7 buildings, over a total of 864
hours - there were a total of 273 adjustments to controls or other environmental aspects
of the room, and 62 adjustments to clothing.
The extent to which adjustments actually improve thermal comfort is as important as the
frequency with which they’re made. Benton and Brager (1994) conducted a field
experiment of thermal comfort in a centrally-conditioned office building in California,
before and after energy-efficiency retrofit measures were installed. Adaptive opportunity
was addressed by a series of questions on the availability, use, and effectiveness of
coping mechanisms that either altered the physical environment or personal variables.
While modification mechanisms were infrequently cited, when exercised, they
consistently received high ratings for effectiveness. Behavioral mechanisms received
the highest number of citations, and clothing adjustments in particular were given a
relatively high effectiveness rating.
1.4.2.4. Evidence for psychological adaptation - expectation and context
While there is limited field data providing direct evidence for the effects of psychological
adaptation on thermal comfort, the previous analysis of Figure 1.7 suggests that it can
be implied through comparing comfort responses in different contexts. Paciuk (1990)
provided a more direct analysis of the distinction between available control (adaptive
opportunity), exercised control (behavioral adjustment) and perceived control (related to
the psychological dimension and expectation). She found that, in addition to the
traditional list of thermal inputs to the heat balance models, perceived degree of control
was one of the strongest predictors of thermal comfort in office buildings, and had a
ASHRAE RP-884 Final Report
Introduction & Background page MRL Australia 24
significant impact in shaping both thermal comfort and satisfaction. This finding was
also supported by the work of Williams (1995), in her study in office buildings in the
Northwest of England. The subjects in this study expressed higher levels of satisfaction
when they perceived themselves to have more control over their environment.
Increasing levels of both perceived and available control have implications for the
design of buildings, including their mechanical systems and interior layouts. A good
example is shown in Figure 1.8, which comes from the English researchers Leaman and
Bordass (1993). They administered a standardized indoor environmental quality
questionnaire to thousands of office workers across the UK, and found a strong negative
relationship between perceived control and occupant density in the workplace.
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
1 2-4 5-9 10-29 30+
Number of Occupants in Office
Per
ceiv
ed C
on
tro
l R
atin
g
ventilation
heating
Figure 1.8: Relationship between the number of workers sharing an office and perceived level of control over room heating and ventilation systems. (Leaman & Bordass 1993).
This relationship also has implications for air-conditioned vs. naturally-ventilated
buildings. Naturally ventilated buildings typically consist of small offices with single
occupants or small groups of who are usually within reach of an operable window. This
is clearly not the case in most modern air-conditioned office buildings which are
ASHRAE RP-884 Final Report
Introduction & Background page MRL Australia 25
characterized by deep-space or open-plan floor layouts with dozens if not hundreds of
employees being required to share the same space. The effects of this may be evident
in Figure 1.7, where the naturally ventilated buildings had thermal neutralities
significantly different from the predictions of heat-balance (static) models such as PMV.
These same buildings probably had occupants who perceived a higher degree of
personal environmental control by comparison to their counterparts in centrally air-
conditioned office buildings. The poor predictive capabilities of PMV in naturally
ventilated buildings suggests that adaptive processes other than behavioral adjustment
(which would be accounted for in the heat balance models) must be occurring.
Expectation seems the most likely explanation, since expectation has all but been
eliminated by the climate-chamber method of comfort research. Within the adaptive
hypothesis, such buildings would be expected by their occupants to provide variable
indoor temperatures, and therefore be judged less critically than centrally air-conditioned
buildings. The RP-884 data analysis will pay careful attention to the distinction between
thermal perception in air-conditioned vs. naturally ventilated buildings.
Although naturally ventilated buildings might generally offer a higher level of adaptive
opportunity than air-conditioned buildings, they could still differ in the actual degree of
occupant control they offer. Rowe (1995a) looked at studies in 1) air conditioned
buildings, 2) naturally ventilated buildings, and 3) naturally ventilated buildings with
supplementary on-demand cooling and heating equipment. He found a significantly
higher level of satisfaction in the naturally ventilated buildings with additional
supplementary control. This led to the conclusion that people have a wider tolerance of
variations in indoor thermal conditions if they can exert some control over them, and that
a considerably higher level of satisfaction will be reached if occupants have means of
controlling the upper and lower temperature limits. In Fishman and Pimberts’ (1982)
year-long study in a UK office building, seven of the 26 subjects worked in air-
conditioned areas. The rest were in naturally ventilated offices. While the sample size
was not large, there was still a difference in the thermal responses of these two groups
as temperatures rose above 24°C. People in the air-conditioned offices began voting
much higher on the thermal sensation scale than their colleagues in the naturally
ASHRAE RP-884 Final Report
Introduction & Background page MRL Australia 26
ventilated work areas, suggesting that they were less tolerant of higher temperatures
and expected homogeneity in their thermal environment.
Several other researchers support this hypothesis regarding occupant expectations and
their effects on thermal perception. In a study conducted by Black and Milroy (1966) in
both air-conditioned and non-air-conditioned office buildings in London, occupants in
the air-conditioned buildings expressed more complaints about temperature
fluctuations, even though the free running building experienced much greater variability.
The occupants were basing their evaluations on the benchmark of their own
preconceptions of what air-conditioning should achieve, rather than on what it actually
provided. In effect, this suggests that increasing levels of sophistication in
environmental control systems and building services are on a treadmill of attempting to
satisfy ever-increasing occupant expectations (de Dear and Auliciems 1986). Another
study by Rohles et al (1977) found that Michigan subjects were more tolerant of high
indoor summer temperatures (32°C ET*) than Texan subjects. Since other heat balance
variables such as clothing or activity could not account for the difference, it was
speculated that the Texans took summer air-conditioning for granted and came to
expect or even demand cool temperatures, therefore becoming more critical of warmer
indoor conditions than their northern counterparts.
1.5. Implications for RP-884
1.5.1. Lessons from static heat balance models
We believe that the split between “adaptive” and “static” heat balance models, or
schools of thought, is not as irreconcilable as the protagonists have suggested. As
mentioned previously, the terms "static” and “constancy" have given rise to a mistaken
idea that models such as PMV and 2-node, plus the thermal comfort standards based
on them, prescribe a single, constant temperature for thermal comfort the world over.
But the PMV and 2-node models do, in fact, predict comfort temperatures moving in the
direction of prevailing outdoor climate -- as seen in the offset of winter and summer
comfort zones in the last few revisions to ASHRAE’s Standard 55. So the static model
of comfort is in reality an “adaptive” model in its own right -- the fundamental distinction
between the static and adaptive models is their underlying basis or postulated cause for
ASHRAE RP-884 Final Report
Introduction & Background page MRL Australia 27
the shift in comfort temperatures. The former permits only behavioral adjustments
(personal/technological) to heat balance variables such as clothing or air velocity,
whereas the original adaptive models were premised on changing physiological (i.e.
acclimatization) and psychological (i.e. expectations/habituation) setpoints. While this
may seem to be a fine distinction, failure to appreciate it has, in the opinion of the
authors, been responsible for unnecessary controversy between the two sides of this
debate. An important contribution of the RP-884 adaptive model will be to go beyond
the “black-box” approaches of the earlier adaptive models, so that we can better
understanding the underlying processes of adaptive comfort.
Understanding the challenges of applying laboratory-based static models in the field can
provide guidance on issues to consider when developing a new adaptive model that
combines the best of both static and adaptive theories of thermal comfort. One place to
start is to learn from some of the explanations that have been offered for the
discrepancies between predicted and observed thermal sensations in real buildings:
1. Estimating insulation of clothing garments or ensembles. Brager et al. (1994) have
demonstrated the significance of the clothing insulation estimation method on the
actual clo value obtained. The ensemble insulation value differs by as much as 20%
depending on whether one uses the tables and algorithms in the older or newer
versions of ASHRAE Standard 55 (1981, 1992), or ISO 7730 (1994). It will therefore
be important that rigorous statistical correction factors are used to create consistent
ensemble clo values across the RP-884 database.
2. Accounting for the chair insulation. The tendency for PMV to overestimate thermal
neutralities has been reported in several field studies (Schiller 1990), prompting
Fanger and Wyon (1990) to suggest that the method of estimating clothing insulation
might be systematically flawed by omission of the thermal effect that chairs have on
their occupants. McCullough and Olesen (1994) responded by examining the effects
of upholstered office furniture on the total thermal insulation of a heated manikin, and
found that a typical office chair adds approximately 0.15 clo to the value that one gets
by simple addition of individual garment values, as described in ASHRAE Standard
55-92 (ASHRAE 1992) or ISO-7730 (ISO 1994). Even if the original researchers
ASHRAE RP-884 Final Report
Introduction & Background page MRL Australia 28
supplying raw data to the RP-884 database omitted the effect of chair insulation, it
will be included as part of the RP-884 analysis.
3. Non-uniformities of physical measurements. If field studies take spot-
measurements of general ambient thermal parameters that are separated from the
occupant’s location in space and/or time, then they might not be representative of
what the occupant is actually experiencing at all (Baker 1993). This becomes
particularly important in rooms with transient or spatially non-uniform thermal
conditions, which are more likely to be the case in passive, or naturally ventilated
buildings, or any situations where workers have high levels of personal or
environmental control available to them. An analysis of adaptive comfort would best
be served by using data taken close to the occupant’s location, and at the same time
as the thermal questionnaire. This will be carefully considered when selecting data
for inclusion in the RP-884 database.
4. Behavioral adjustments and perceived control. People adapt to the environment by
adjusting their clothing or activity, modifying their posture or moving to another part of
the room, opening/closing windows, operating fans or other environmental controls.
But why would this cause a discrepancy between the observed and predicted
conditions? In theory, static heat balance models account for clothing, activity, and
thermal environmental parameters, and should therefore, be able to factor the
consequences of the behavioral adjustments into their equations. Probably the most
likely impact of thermal adjustments is the perception of control -- psychologists are
quick to point out that adverse or noxious stimuli are less irritating if the subject
perceives she/he has control over them (Paciuk 1990, Veitch and Arkkelin 1995,
Kaplan and Kaplan 1982). Issues of behavioral adjustment and perceived control will
be given a high priority in the RP-884 analysis, as this represents a potentially
significant feedback loop between discomfort and purposive behavioral
thermoregulation.
5. Thermal sensation, preference, and acceptability. Existing thermal comfort
standards provide guidelines for “thermal acceptability”, while the static heat balance
models on which they’re based only predict “thermal sensation”. As a result, the
ASHRAE RP-884 Final Report
Introduction & Background page MRL Australia 29
traditional approach has been to indirectly associate specific thermal sensations with
“acceptability”, and to assume that thermal “preference” is synonymous with thermal
“neutrality”. RP-884 will strive to include field experiments in its database that directly
asked about sensation, acceptability and preference, so these assumptions can be
tested.
1.5.2. Time scales of thermal adaptation
Since each class of adaptive response depends on repeated exposure to a given
regime of thermal conditions, the questions of duration of exposure and lag in response
seem relevant to adjustment, acclimatization and habituation adaptive processes. A
review of the literature in this area will reveal, in part, which mechanisms are likely to
play the most significant role in thermal response to the indoor environment and,
therefore, which should receive the greatest attention in the RP-884 analysis.
The significance of the temporal dimension of thermal adaptation is realized when one
considers applications of adaptive models to control algorithms for HVAC systems.
Auliciems was the first to propose such an adaptive algorithm (Auliciems 1986) which
he referred to as a thermobile (as opposed to a thermostat). It was premised on the
adaptive model described in equation 1.4. The question of how long the averaging
period for the algorithm’s temperature inputs should be was left open but, as an initial
guess, Auliciems proposed that the running means, one for both indoor and outdoor
temperatures, should comprise hourly observations across the preceding fortnight.
More recently, Humphreys and Nicol (1995) proposed a similar adaptive algorithm for
UK office temperatures. The gist of his proposed guideline is that a weighted, running
mean of the preceding week’s outdoor temperature is combined with current outdoor
temperature in a ratio of 3:7, thereby reflecting the overriding importance of today’s
weather on clothing decisions and behavior. Humphreys proposed that this outdoor
temperature index be used to specify the target indoor temperature.
Adjustment. Thermal adjustment and behavioral adaptation operate across several
time scales. Cutaneous thermoreceptors provide almost instantaneous neural
information about sudden changes in the thermal environment. For example, as
experienced, when crossing the indoor/outdoor threshold, thus enabling clothing
ASHRAE RP-884 Final Report
Introduction & Background page MRL Australia 30
adjustments and other behavioral adaptations to be effected well in advance of any
significant alteration in the body’s heat balance. As for other behavioral adaptations,
very little research has been published on adaptive time lags. A notable exception is a
study by Humphreys (1979) on clothing adjustments at the seasonal and synoptic
weather time-scales. He was able to statistically relate clothing insulation levels on any
given day to an exponentially weighted moving average of outdoor temperatures on the
days leading up to, and including, the day in question. It was suggested that the half-life
for daytime clothing regulation was of the order of 20 hours.
Acclimatization. The literature on acclimatization reviewed earlier indicates that the
physiological adaptations to heat exposure begin on the first day of exposure and
progress rapidly to full development by the third or fourth day, providing the heat
exposures are sufficiently severe to elevate core temperatures (Bean and Eichna, 1943;
Fox, 1974). This has been achieved experimentally with daily work-in-heat regimes or
hyperthermic suits. Passive exposures to heat in the course of normal day-to-day
acclimatization cannot be expected to induce acclimatization responses as quickly nor
as thoroughly, although Wyndham (1970) reports that passive exposures to the normal
course of the seasons in South Africa induced definite signs of at least partial
acclimatization. The time-scales of interest for office workers, therefore, may be of the
order of weeks to months.
Habituation and expectations. Unfortunately this literature review was unable to find
reference to any research on the time-scales of psychological adaptive responses,
probably for the simple reason that no researchers have previously attempted to
disentangle psychological from other thermal adaptive processes. However, anecdotal
evidence suggest that building occupants become accustomed to levels of warmth
prevailing within buildings on time scales of weeks to months. These scales translate
into synoptic and seasonal processes operating in the outdoor atmospheric
environment.
To summarize, the adaptive processes are operating on time scales ranging from
seasonal, through synoptic to diurnal. Critics of the adaptive approach at various
symposia or seminars have repeatedly asked the question: “... how long must your
ASHRAE RP-884 Final Report
Introduction & Background page MRL Australia 31
people suffer in sub-optimal indoor climates before they become adapted?” Ignoring
the emotive language in this question, we feel its answer, if there is one, depends on
which of the adaptive processes is being relied upon. The consensus within what little
has been written on the temporal dimension of adaptation is that meteorological
conditions on the day in question, and to a lesser extent, the preceding week or two,
exert an overriding influence on thermal adaptation in general, and clothing
thermoregulation in particular. This has important implications for future field
experimental protocols. While traditional research designs tend to look at responses at
a given moment, experiments that intend to evaluate adaptive mechanisms need to take
measurements over extended periods of time. Available evidence reviewed in this
paper indicates that, in climate chamber experiments at least, the slower physiological
adaptive process of acclimatization appears not to be relevant to this question of
thermal neutrality and its fluctuations from day-to-day, week-to-week and season-to-
season. As a result, the RP-884 data analysis and model development will focus more
heavily on the adaptive mechanisms of adjustment, and habituation/expectation. This
also suggests the need for field experiments in which data were rigorously obtained,
including accurate measurements of air movement
1.6. Aims
The specific objectives of RP-884 can now be listed:
1. Elaborate and define adaptive processes in the context of indoor climatic
perception.
2. Develop an internally consistent and quality controlled database of thermal comfort
field experimental data from a variety of buildings and climates across the world. To
then make this database as widely available to other thermal comfort researchers
as possible.
3. Examine the semantics of thermal sensation, acceptability and preference scales
within the context of an adaptive model of thermal comfort.
4. Develop statistical models of thermal comfort based explicitly on the various
processes of adaptation, including adjustment, acclimatization and habituation.
ASHRAE RP-884 Final Report
Introduction & Background page MRL Australia 32
5. Explore the influence of contextual and non-thermal factors on thermal perception
indoors. This investigation will include (but not be restricted to) season, building
purpose (residential, office) and climatic setting, on thermal perception. This will
inevitably include comparisons with the thermal comfort predictions of heat-balance
models such as PMV/PPD.
6. Proposing a variable temperature standard that, in time, might eventually
supplement and/or modify ASHRAE Standard 55.
ASHRAE RP-884 Final Report
Methods page MRL Australia 33
CHAPTER 2 - METHODS
2.1. Overview of the RP-884 approach
In view of the vital role played by perceptual and cognitive factors in the adaptive
hypothesis, a consensus emerging from the literature is that observational data to test
the hypothesis must be come from field rather than climate-chamber research. The
reductionist, laboratory approach to comfort research runs the risk of stripping away
those very aspects of thermal perception that are the focus of the adaptive hypothesis
(McIntyre, 1982). The approach in RP-884 has, therefore, been to focus on research
conducted in “real” buildings, occupied by “real” subjects going about their normal day-
to-day activities rather than paid college-age subjects sitting in the highly contrived and
controlled setting of the climate chamber.
In order to identify and disentangle various adaptive processes from the data, it became
apparent in the research design stages of RP-884 that the field data needed to be of a
high standard. The database underpinning RP-884’s adaptive models comprised field
experiments where the standard of measurements, both physical and subjective, was as
close as possible to laboratory-grade, and comprehensive enough to enable heat-
balance indices (static model) to be calculated. Where possible, the RP-884 database
comprised field experiments rather than field studies. Furthermore, the database
needed to be built up from the raw data files generated by the original researchers
instead of their processed or published findings. This approach allowed a variety of
quality controls to be applied and enhanced the internal consistency of the entire
database.
Considerable effort and resources from RP-884 and numerous field researchers around
the world have been dedicated to the assembly of this database of thermal comfort field
experiments. It therefore seems highly likely that the database will have numerous
applications well beyond the scope and lifetime of RP-884. Therefore a decision was
made to provide global and unrestricted access via the World Wide Web (WWW).
ASHRAE RP-884 Final Report
Methods page MRL Australia 34
The ultimate application of any thermal comfort model, adaptive or otherwise, is to
predict the response of a given group of human subjects to a given set of input
parameters (temperatures, humidity, air speeds etc). Typically this means either the
occupants of an extant building, or the hypothetical occupants of a yet-to-be-built
structure. Since RP-884 adaptive models are to be applied at the level of single
buildings, the meta-analysis used to derive the models should be conducted at the same
unit of analysis -- that of the single building. Therefore the 21,000 rows of raw data in the
RP-884 database were subsequently sorted, aggregated and analyzed at the building
level. Figure 2.1 is a schematic depiction of the database process, and how it evolved
into the adaptive model meta-analysis. The remainder of this chapter describes the
detailed steps underpinning this schematic flow chart.
ASHRAE RP-884 Final Report
Methods page MRL Australia 35
Figure 2.1: Schematic depiction of the RP-884 database process and its evolution into the adaptive model meta-analysis.
ASHRAE RP-884 Final Report
Methods page MRL Australia 36
2.2. Establishing the database for RP-884
The RP-884 database is the project’s fundamental research resource. This section
describes where the raw data came from, how they were quality controlled, and what
processes of data assimilation were developed to ensure internal consistency within the
database.
2.2.1. Sourcing the raw data
The literature review in Chapter 1 uncovered numerous thermal field studies and
experiments. Combined with the authors’ and ASHRAE TC 2.1’s knowledge of
researchers currently or recently active in this area, we compiled a mailing list. An initial
fax was broadcast to dozens of researchers around the world requesting information
about field methods and soliciting contributions to the database (see Figure 2.3a and
Figure 2.3b). On the basis of the returns to that questionnaire, a list of the contributors
and their field methods was collated. Figure 2.2 depicts the geographic locations of the
contributors to the RP-884 database. Data came from four continents and a broad
spectrum of climatic zones.
Figure 2.2: Geographic origins of the raw data contributions to RP-884 world database of thermal comfort field research
ASHRAE RP-884 Final Report
Methods page MRL Australia 37
Table 2.1: Sources of raw data for the RP-884 world database of thermal comfort Researcher File
No. Experiment Location Building
Type Research Design
Sample Size
No. of Blgds
Jill Brown (U of Wales - UK) 1 South Wales, UK (summer) HVAC cross-sectional 80 4
Jill Brown (U of Wales - UK) 2 South Wales, UK (winter) HVAC cross-sectional 38 4
(Fountain and Huizenga, 1996). Daily averages for air temperature, relative humidity
and effective temperature were also calculated.
2.3.2. Consistent mean radiant temperatures within the database.
Mean radiant temperature was recalculated from each row of data using the ASHRAE HoF
formula (1993), based on raw globe and air temperatures plus air speed.
t (t 273)1.10 10 V
D(t t ) 273
r g4
8 0.6
0.4 g a
14
= + −
−+
•
•ε
where ε is emissivity (0.95 for a black globe), D is globe diameter (0.04 m for “ping-pong”), V is air speed in m s-1, ta is air temperature in oC, tg is globe thermometer’s temperature in oC. N.B. the globe thermometer has a lagged response and requires about 10 to 15 minutes to equilibrate. Larger diameter globes have longer lags.
2.3.3. Consistent comfort index calculations within the database
With models as complex as PMV and SET*, it is to be expected that several different
algorithms and implementations exist in engineering and research circles around the
world. ASHRAE TC 2.1 has recently acknowledged this potential source of “noise” in
comfort research and engineering applications, and has sought to standardize the
Figure 3.2: Dependence of neutrality predicted by the PMV heat-balance model (PREDNEUT) on mean indoor temperatures (TOP).
The clear dependence of predneut on mean indoor top in Figure 3.2 suggests that the other
heat balance variables that change in response to indoor temperature -- such as clothing
insulation and air speeds, were driving the PMV predicted neutralities. The correlation
appears to be strongest in the case of the naturally ventilated buildings (r=0.84). In the case
ASHRAE RP-884 Final Report
Basic Results page MRL Australia 77
of the HVAC and mixed mode buildings in Figure 3.2, the seven outlying buildings below the
fitted regression line were from the Brown (1992/3) study in industrial settings where
metabolic rates were elevated well above those encountered in the remaining office and
residential buildings, causing the predicted neutrality to be depressed by as much as 10°C
below the trendline. These outliers account for the reduction in correlation coefficients.
3.1.2.5. Neutral standard effective temperatures (neut_set)
Solution of the regression equations for the “neutral” sensation in relation to SET was
performed building-by-building. Table 3.8 below summarises the neut_set findings from
160 buildings.
Table 3.8: Summary of the neutral standard effective temperatures (neut_set) (°C).
centrally heated/air-conditioned buildings
naturally ventilated buildings
mixed-mode buildings
number of buildings in summer sample*
24 out of 73
(6 missing values)
25 out of 31
(2 missing values)
1 out of 1
(1 missing value)
mean neut_set (±stdev) in the summer sample
24.5
(±1.51)
24.1
(±2.85)
23.8 (±0)
number of buildings in winter sample*
8 out of 30
(2 missing values)
2 out of 11
(1 missing value)
2 out of 2
(no missing values)
mean neut_set (±stdev) in the winter sample
25.2
(±3.51)
31.3
(±1.89)
19.6
(±6.73)
* only results from buildings with statistically significant regression models used
Less than one third of the centrally heated/air-conditioned buildings in the sample yielded
significant regression models, probably because of the restricted range of thermal
environmental conditions in such buildings. Naturally ventilated buildings, with their greater
internal climatic variety produced a significant regression equation against the SET index in
the majority of cases. In those buildings sampled during the summer season, there was an
average neutral SET in the 24~24.5°C region and the difference of 0.4 K between the
ASHRAE RP-884 Final Report
Basic Results page MRL Australia 78
means of naturally ventilated and centrally heated/air-conditioned buildings was not
statistically significant (T = 0.61, df = 47, p > 0.5). The anomalously high average neutral
SET of 31.3°C observed in naturally ventilated buildings in winter was based on only two
buildings. The seasonal difference in neutrality for HVAC buildings was not statistically
significant (T = 0.8, df = 30, p > 0.2).
3.1.3. Thermal acceptability and indoor climate
Thermal acceptability was directly assessed in a small subset of studies in the RP-884
database, but it could be inferred from thermal sensation votes, which were recorded in all
studies. This section analyzes both direct and inferred versions of thermal acceptability
response.
3.1.3.1. Relationship between direct and inferred thermal acceptability
Testing the assumption that a thermal sensation vote falling in the interval -1.5<ASH<+1.5
equated with thermal acceptability was possible by comparing frequencies of both direct
and indirect thermally acceptable votes within each building. Each building’s frequency of
directly assessed acceptable votes was coded as f_tsa_2 and the frequency of acceptable
thermal sensations was coded as prxy_tsa in the meta-analysis.
The result have be depicted in Figure 3.3. The graphs represent weighted regression
models of prxy_tsa versus f_tsa_2. Each point in the graphs represents a specific building
in the database. The solid line plotted through the data points represents the expected
relationship (gradient 1:1) whereas the dotted line represents the line of best fit (with model
equation and statistics annotated on each graph).
ASHRAE RP-884 Final Report
Basic Results page MRL Australia 79
Townsville Australia RP-702 (tropical dry season), HVAC buildings.
prxy_tsa = 0.55 * tsa + 38.849
R2 = 0.4685
60
70
80
90
100
60 70 80 90 100
TSA (% aceptable)
Pro
xy T
SA
(%
"sa
tisfa
ctor
y" A
SH
vot
es)
Townsville Australia RP-702(tropical wet season), HVAC buildings.
prxy_tsa = 0.4288 * tsa + 46.302
R2 = 0.2038
60
70
80
90
100
60 70 80 90 100
TSA (% acceptable)
Pro
xy T
SA
(%
"sa
tisfa
ctor
y" A
SH
vot
es)
Montreal Canada RP-821 (summer), HVAC buildings.
prxy_tsa = 0.5007 * tsa + 32.536
R2 = 0.4752
50
60
70
80
90
100
50 60 70 80 90 100
TSA (% acceptable)
Pro
xy T
SA
(%
"sa
tisfa
ctor
y" A
SH
vot
es)
Montreal Canada RP-821 (winter), HVAC buildings.
prxy_tsa = 1.3448 * tsa - 35.093
R2 = 0.7155
50
60
70
80
90
100
50 60 70 80 90 100
TSA (% acceptable)
Pro
xy T
SA
(%
"sa
tisfa
ctor
y" A
SH
vot
es)
Sydney Australia (summer and winter) HVAC and Mixed buildings.
prxy_tsa = -0.2419 * tsa + 92.27
R2 = 0.9872
60
70
80
90
100
60 70 80 90 100
TSA (% acceptable)
Pro
xy T
SA
(%
"sa
tisfa
ctor
y" A
SH
vot
es)
proxy acceptability versus actual acceptability for all available buildings
30
40
50
60
70
80
90
100
30 40 50 60 70 80 90 100
TSA (% acceptable)
Pro
xy T
SA
(% "
satis
fact
ory"
AS
H v
otes
)
prxy_tsa = 30.74 + 0.61 * tsa
R2 = 0.48, p = 0.0001
Figure 3.3: Comparison of directly determined and inferred thermal acceptability. Each data point represents an individual building from the RP-884 database.
Figure 3.3 indicates that the strength of association between direct and inferred thermal
acceptability varied considerably across field experiments. For the pooled analysis of all
seven experiments’ buildings (top left panel in Figure 3.3), about half of the variance in
acceptable thermal sensations (prxy_tsa) could be accounted for by the direct thermal
acceptability ratings. Discounting Rowe’s (1996) Sydney experiment due to its small
sample size (three building data points), the highest correlation was found in Donnini’s
(1996) Montreal winter experiment (r = 0.85). For the remaining experiments, the
correlations can be described as moderate. In five out of the seven individual project graphs
ASHRAE RP-884 Final Report
Basic Results page MRL Australia 80
depicted in Figure 3.3, the gradient of the observed dependence of inferred acceptability on
directly stated acceptability was significantly lower than the unity we expected. Subjects
apparently were voting that thermal sensations outside the central three categories of the
ASHRAE 7-pt scale were still acceptable. Expressing that a different way, buildings that
had a high rating of thermal acceptability on the direct scale (>80%), typically scored lower
acceptability ratings on the basis of percentage of thermal sensations falling within the
As noted in the preceding section, “acceptable” TSA votes (“At the present time, is this
thermal environment acceptable to you or not?”) were tallied for each building and
expressed as a percentage of all responses in the building. This percentage was coded as
f_tsa_2 for each building in the RP-884 meta-analysis.
Ignoring the upper and lower humidity boundaries of the summer and winter comfort zones
depicted in ASHRAE Standard 55-92, the percentage of indoor climate measurements
within each RP-884 database building complying with the relevant summer or winter
ASHRAE comfort zone ET boundaries was coded as ASH55_92 in the meta-analysis. A
simple assessment of the practical utility of the ASHRAE comfort zones can be performed
by comparing these acceptability levels predicted from indoor climatic measurements
(ASH55_92) with the corresponding thermal acceptability ratings for each building. These
comparisons have been performed in Figures 3.4 and 3.5 respectively -- each data point in
the graphs represents a single building.
ASHRAE RP-884 Final Report
Basic Results page MRL Australia 81
Thermal Acceptability (all buildings)
0
10
20
30
40
50
60
70
80
90
100
0 10 20 30 40 50 60 70 80 90 100
% indoor climates falling within ASHRAE 55-92 comfort zones
TS
A (
% a
ccep
tabl
e)
tsa = 75.05 + 0.03 * ASH55_92
R2 = 0.01, p = 0.5258
Figure 3.4: Relationship between direct thermal acceptability ratings of buildings (f_tsa_2) and buildings’ compliance with ASHRAE Standard 55-1992 comfort zone ET boundaries (ASH55_92).
Thermal Acceptability (all buildings)
0
10
20
30
40
50
60
70
80
90
100
0 10 20 30 40 50 60 70 80 90 100
% indoor climates falling within ASHRAE 55-92 comfort zones
Pro
xy T
SA
(%
vot
es -
1.5
< a
shra
e <
1.5
)
prxy_tsa = 70.96 + 0.16*ASH55_92
R2 = 0.14, p = 0.0001
Figure 3.5: Relationship between acceptability thermal sensation ratings of buildings (prxy-tsa) and buildings’ compliance with ASHRAE Standard 55-1992 comfort zone ET boundaries (ASH55_92).
Regardless of which thermal acceptability measure was adopted, Figures 3.4 and 3.5
indicate that compliance with the ET prescriptions of ASHRAE Standard 55-1992 had
little or no bearing on the buildings’ acceptability ratings by occupants. This is indicated
ASHRAE RP-884 Final Report
Basic Results page MRL Australia 82
clearly by the complete lack of statistical significance in the regression models plotted on the
graphs in Figures 3.4 and 3.5. A logical extension of this null result is that most of the
buildings which had very low levels of compliance with ASHRAE 55-92 (say, ASH55_92 <
30%) still had occupant ratings of thermal acceptability better than 60 to 70%
TSA versus TOP for all buildings
0
20
40
60
80
100
18 20 22 24 26 28 30 32 34
mean indoor operative temperature (oC)
TSA
(% a
ccep
tabl
e)
tsa = -287.56 + 27.75 * top - 0.52 * top2
R2 = 0.08, p = 0.0883
TSA versus ET for all buildings
0
20
40
60
80
100
18 20 22 24 26 28 30 32 34
mean indoor effective temperature (oC)
TSA
(% a
ccep
tabl
e)
tsa = -253.80 + 25.64 * et - 0.49 * et 2
R2 = 0.06, p = 0.1554
TSA versus SET for all buildings
0
20
40
60
80
100
18 20 22 24 26 28 30
mean indoor standard effective temperature (oC)
TSA
(% a
ccep
tabl
e)
tsa = -724.32 + 61.02 * set - 1.16 * set 2
R 2 = 0.22, p = 0.0005
TSA versus PMV for all buildings
0
20
40
60
80
100
-3 -2 -1 0 1 2 3
mean indoor predicted mean vote
TSA
(% a
ccep
tabl
e)
tsa = 77.23 + 16.00 * pmv - 11.63 * pmv2
R2 = 0.18, p = 0.0030
Figure 3.6: Dependence of direct thermal acceptability ratings on mean thermal index values. Each data point represents a building.
Figure 3.6 shows the percentage of occupants within each building voting “acceptable”
(f_tsa_2) as a function of the mean indoor climatic index values recorded for each building.
The indices selected for this analysis covered the spectrum from relatively simple operative
temperature up to fully developed heat balance indices such as PMV and SET. The
expected relationship between percentage satisfied and indoor warmth is hyperbolic,
peaking around the database’s mean neutrality or preferred temperature. Unfortunately the
majority of buildings available for the analysis were clustered within a fairly narrow band of
indoor temperatures, centred on 23°C, and so the data are not well suited to regression
analysis. As a result, the weighted 2nd order polynomial models fitted to the TOP and ET
ASHRAE RP-884 Final Report
Basic Results page MRL Australia 83
indices in Figure 3.6 were not statistically significant. While the models fitted for the more
sophisticated heat balance models such as PMV and SET did achieve statistical
significance, the explained variance was about 20% in both cases.
The underlying concept of Fanger’s Predicted Percentage Dissatisfied index (1970) is
simply that as mean indoor climatic conditions depart from the optimum (assumed to be
PMV=0), the percentage of persons experiencing unacceptable thermal sensations
increases. Despite the obvious lack of normality in statistical distributions for TSA % across
the RP-884 building database, the 2nd order polynomial regression equation fitted to mean
building PMV values in Figure 3.6 is of the hyperbolic form suggested by the PPD concept.
The fact that the PMV index produced a statistically significant relationship (R2= 0.18) where
the simpler indices of TOP and ET failed suggests that the inclusion of other heat balance
factors such as air speed, metabolic rate and clothing actually does what it’s supposed to do
-- improve predictions. The same interpretation can be applied to the SET index, since it
too incorporates the full array of heat balance variables, and as seen in Figure 3.6, its 2nd
order polynomial model was also statistically significant (p=0.0005).
3.1.3.3. Thermal acceptability inferred from thermal sensation.
Fanger’s Predicted Percentage Dissatisfied (PPD) model is premised on the assumption
that a thermal sensation vote within the central three categories of the ASHRAE 7-point
scale (slightly cool + neutral + slightly warm) is acceptable and satisfactory. Since it is
derived from the core thermal response item of ASH, this proxy for thermal acceptability was
obtained for every respondent in the cumulative ASHRAE RP-884 database (n>21,000) and
coded as prxy_tsa.
ASHRAE RP-884 Final Report
Basic Results page MRL Australia 84
Proxy TSA versus TOP for all buildings
0
20
40
60
80
100
10 15 20 25 30 35
mean indoor operative temperature (oC)
Pro
xy T
SA
(% a
ccep
tabl
e)
prxy_tsa = -176.48 +20.56 * top - 0.41 * top2
R2 = 0.49, p = 0.0001
Proxy TSA versus ET for all buildings
0
20
40
60
80
100
10 15 20 25 30 35
mean indoor effective temperature (oC)
Pro
xy T
SA
(% a
ccep
tabl
e)
prxy_tsa = -171.19 + 20.21 * et - 0.40 * et 2
R2 = 0.51, p = 0.0001
Proxy TSA versus SET for all buildings
0
20
40
60
80
100
14 16 18 20 22 24 26 28 30 32 34
mean indoor standard effective temperature (oC)
Pro
xy T
SA
(% a
ccep
tabl
e)
prxy_tsa = -679.20 + 56.70 * set - 1.05 * set 2
R 2 = 0.41, p = 0.0001
Proxy TSA versus PMV for all buildings
0
20
40
60
80
100
-3 -2 -1 0 1 2 3
mean indoor predicted mean vote
Pro
xy T
SA
(% a
ccep
tabl
e)
prxy_tsa = 83.72 + 10.75 * pmv - 10.52 * pmv2
R2 = 0.42, p = 0.0001
Figure 3.7: Dependence of building acceptability ratings (derived from thermal sensation) on mean thermal index values. Each data point represents a building.
As noted in the preceding section, the majority of buildings in the RP-884 database were
clustered within a narrow range of mean indoor temperatures, severely limiting the scope for
regression analyses. However in Figure 3.7, because of the larger sample size compared
with the preceding section, all four indoor climatic indices showed statistically significant
relationships with this proxy building thermal acceptability index.
3.1.3.4. Thermal sensitivity and the range of thermally acceptable temperatures.
Given the relatively weak correlations for thermal acceptability in the preceding sections, the
use of the associated regression models to define acceptable ranges of thermal indices
would be not very reliable. A more feasible alternative, based on Fanger’s Predicted
Percentage Dissatisfied (PPD) concept (1970), can be applied to this question of
acceptable ranges. As noted earlier, PPD is a function of mean thermal sensation (PMV in
Fanger’s terminology) and a PMV of ±0.85 is assumed to correspond with 80%
acceptability. Logically therefore, assuming that actual thermal sensation votes (ASH) are
distributed around their mean with a similar variance as predicted votes are (PMV/PPD),
the values of a particular indoor thermal index (e.g. TOP, ET, PMV or SET) corresponding
ASHRAE RP-884 Final Report
Basic Results page MRL Australia 85
with mean ASH votes of ±0.85 can be interpreted as the limits of acceptable thermal
environments (for 80% acceptability). This derivation of acceptable ranges was
operationalized by solving the ASH linear regression models (Appendix A) that we defined
for each of the main indoor thermal indices (TOP, ET, PMV, SET) for each of the buildings
in the RP-884 database, using ASH=-0.85 and again using ASH=+0.85. Subtraction of the
index value, say TOP, corresponding with -0.85 from the corresponding +0.85 value defines
the width of 80% acceptable TOP values for that particular building and the variable thus
defined was codenamed RANG_TOP in the meta-analysis.
Note that acceptable temperature ranges using these techniques were only feasible in
buildings whose ASH regression models (Appendix A) achieved statistical significance at
the 95% confidence level.
Table 3.9: Range of acceptable operative temperatures (Kelvin).
centrally heated/air-conditioned
buildings
naturally ventilated buildings
mixed-mode buildings
number of buildings 108 (3 missing values)
41 (4 missing values)
4 (no missing values)
number of buildings with regression models achieving 95% significance*
62
(57% of total)
33
(75% of total)
3
(75% of total)
80% acceptability criterion (RANG_TOP) Mean (±stdev)
4.1
(±1.91)
6.9
(±2.79)
4.5
(±1.24) 90% acceptability criterion (RANTOP10) Mean (±stdev)
2.4
(±1.12)
4.9
(±3.27)
2.7
(±0.73)
* Based on those thermal sensation models in Appendix A (y=a + b*TOP) achieving 95% statistical significance or better
The 80% acceptable range of operative temperatures was, on average, 6.9 K wide in
naturally ventilated buildings, which was about 70% wider than in centrally heated/air-
conditioned buildings. This difference was statistically significant (T = 5.69, df = 93,
p<0.001). The acceptable range of operative temperatures for mixed mode buildings was,
ASHRAE RP-884 Final Report
Basic Results page MRL Australia 86
on average, between that of HVAC and NV buildings, but the small number of cases
precludes any statistical tests.
Also included in Table 3.9 are the acceptable operative temperature ranges for a more
stringent criterion of 90% acceptability (labelled RANTOP10 in the meta-anlysis). These
were derived from each building’s thermal sensation v operative temperature regression
equation, but instead of solving for neutrality ± 0.85 sensation units (as was the case for the
80% criterion used for RANG_TOP), we applied the PPD=10% assumption, namely
neutrality ± 0.5 sensation units. The acceptable ranges in Table 3.9 reduced from 4.1 K for
80% acceptability in HVAC buildings to 2.4 K using the 90% acceptability criterion (not
dissimilar to the prescriptive ranges found in ASHRAE Standard 55-92 for the same
acceptability criterion for general thermal comfort, excluding local discomforts). However,
the average 90% acceptability range observed for RP-884’s naturally ventilated sample was
twice as wide as observed in the HVAC sample in Table 3.9 (and also prescribed in
Figure 3.9: Dependence of the acceptable range of operative temperatures (TOP) within buildings on their standard deviation of operative temperature indoors
The adaptive hypothesis emphasises the effects of expectation on thermal acceptability. If a
particular building’s indoor climate is characterized by large variations in temperature, both
temporally and spatially, the adaptive hypothesis predicts a corresponding widening in the
range of indoor temperatures considered acceptable by its occupants. Figure 3.9 depicts
the linear relationship between the range of acceptable operative temperatures and the
standard deviation of indoor operative temperature. The model was statistically significant
with a correlation coefficient r = +0.59, and the regression equation indicates that the
acceptable range (-0.85< ASH < +0.85) increases by about two degrees for a single degree
increase in standard deviation of operative temperature. So, in tightly controlled HVAC
ASHRAE RP-884 Final Report
Basic Results page MRL Australia 89
buildings depicted in Figure 3.9 where we find relatively small standard deviations of
operative temperature, there is a clear trend for the gradient of ASH v TOP regression
models to increase (see Appendix A). Consequently the range of acceptable temperatures
appears to be much greater in naturally ventilated buildings where thermal variability is the
norm compared to HVAC buildings.
3.1.4. Thermal preferences and indoor climate
One hundred and sixteen of the 160 buildings in ASHRAE RP-884’s database assessed
thermal preferences with a questionnaire item along these lines:
“At this point in time, would you prefer to feel warmer, cooler, or no change?”
Probit regression analysis (Finney, 1971; Ballantyne, 1977) rather than linear regression has
been separately applied to the votes for warmer and cooler conditions for each building.
Preferred temperature (of whatever index) was defined as that value of the independent
variable (thermal index) corresponding to the intersection of the “want cooler” and “want
warmer” probit models (see Appendix B). Table 3.10 below summarises the main statistics
for preferred operative temperatures for the 116 buildings in which the questionnaire item
was available.
Table 3.10: Summary of the preferred operative temperatures (preftemp) (°C).
centrally heated/air-conditioned buildings
naturally ventilated buildings
mixed-mode buildings
number of buildings in summer sample*
62 (17 missing values)
24 (9 missing values)
1 (1 missing value)
mean preftemp (±stdev) in the summer sample
23.1 (±1.26)
24.3 (±2.13)
24 (±0)
number of buildings in winter sample*
22 (10 missing values)
6 (6 missing values)
1 (1 missing value)
mean preftemp (±stdev) in the winter sample
22.9 (±1.19)
23.1 (±1.61)
21.7 (±0)
* results not based on statistically significant regression models.
ASHRAE RP-884 Final Report
Basic Results page MRL Australia 90
The results in Table 3.10 indicate a fairly constant temperature preference of about 23°C in
centrally controlled HVAC buildings. Winter temperature preferences in naturally ventilated
buildings were on average about a degree cooler than summer preferences, but this failed
to meet statistical significance, due to the small sizes and large standard deviations. The
summer temperature preferences in HVAC buildings were on average about one degree
cooler than those in naturally ventilated buildings and because of the more substantial
sample sizes in this season, the difference was significant (T = 3.23, df = 84, p < 0.002).
The trivial difference in temperature preferences between HVAC and NV buildings in winter
at less than a third of a degree was statistically insignificant (T = 0.34, df = 26, p > 0.5).
Figure 3.10 Thermal preferences as a function of mean indoor thermal index values (TOP, ET, PMV, SET). Each data point represents a single building.
Figure 3.10 indicates that the operative temperature preferred by building occupants was
moderately correlated with mean levels of warmth prevailing within their buildings at the time
of the field survey. The strength of correlation was reasonably consistent at about r=+0.55
across all four indoor climatic indices (TOP, ET, PMV and SET).
ASHRAE RP-884 Final Report
Basic Results page MRL Australia 91
3.1.5. Comparisons between neutral and preferred temperatures indoors.
With various aspects of perceived indoor climates being assessed with different
questionnaire items, there is a possibility that the indoor temperatures defined as optimal
for a particular building and climatic context may in fact vary, depending on whether one is
talking in terms of thermal sensation (neutrality), thermal acceptability (satisfaction) or
thermal preference (preferred temperatures). Indeed, some authors (McIntyre, 1978; de
Dear, 1991c) have suggested that at least some of the statistical dependence of neutrality
on prevailing outdoor climates observed by the pioneers of adaptive models (Auliciems and
Humphreys) may in fact be due to a semantic artefact in the ASHRAE (or Bedford) 7-pt
scale of thermal sensation. Persons living in cold climates may in fact describe their
preferred thermal environment with words like “warm and cosy” while for persons in hot
climates, words like “cool and fresh” may connote their thermal ideal.
The RP-884 database contains 55 buildings in which both thermal sensations (ASH) and
thermal preferences were registered, and so each of these buildings had both a neutrality
and a preferred temperature available in the meta-analysis. A new variable called “semantic
discrepancy” (discrep) was calculated as neutrality minus preferred temperature and
expressed in degrees (°C).
Table 3.11: Statistics for the semantic discrepancy (discrep) between observed neutrality (neut_top) and observed temperature preference (preftemp) (°C).
centrally heated/air-conditioned buildings
naturally ventilated buildings
mixed-mode buildings
number of buildings in summer sample*
43 out of 62
(17 missing values)
23 out of 24
(9 missing values)
1 out of 1
(1 missing value)
mean discrep (±stdev) in the summer sample
0.7
(±0.78)
0.2
(±1.38)
-0.14 (±0)
number of buildings in winter sample*
13 out of 22
(10 missing values)
6 out of 6
(no missing values)
1 out of 1
(1 missing values)
mean discrep (±stdev) in the winter sample
0.0
(±0.45)
0.3
(±1.00)
-0.7 (±0)
* only results from buildings with statistically significant regression models (neut_top) and probit analyses (preftemp) were used to define discrep
ASHRAE RP-884 Final Report
Basic Results page MRL Australia 92
Table 3.11 indicates that in both seasons and in HVAC and NV buildings alike, the average
semantic discrepancy between neutrality and preference was greater than or equal to zero
degrees. Although the seasonal difference between mean DISCREP in NV buildings was
negligible (T = 0.15, df = 27, p > 0.5), the seasonal difference in HVAC buildings was
statistically significant (T = 2.93, df = 54, p < 0.01). Neither the summer nor the winter
differences between DISCREP in HVAC and NV buildings were significant (T = 1.96, df =
64, p > 0.05 and T = 0.73, df = 17, p > 0.2 respectively).
Figure 3.11 below was designed to test the hypothesis that warm environments promote
positive semantic discrepancies between thermal sensations and preferences, while cool
environments promote negative discrepancies. The seemingly random distribution of data
points in the graph and statistically insignificant correlation and regression in the “all
buildings” panel of Figure 3.11 suggest that mean indoor climatic warmth (top) appears to
exert no systematic influence on the semantics of thermal sensation scales.
Pursuing the semantic artefact hypothesis a little further, the database was disaggregated
into HVAC and naturally ventilated buildings. The lower panels of Figure 3.11 indicates
again that, for the naturally ventilated buildings at least, the mean levels of warmth indoors
had no systematic effect on DISCREP. However, there was a positive, albeit modest,
relationship between the DISCREP variable and mean indoor operative temperature in
HVAC buildings. The gradient on that regression model indicates that, on average,
neutrality inside a centrally air-conditioned building becomes elevated above preferred
temperature by about one degree for every two degrees the mean indoor operative
temperature increases between 21 and 26°C. That is, persons living and/or working in
generally warm centrally air-conditioned buildings seem to be describing their preferred
indoor climate with terms like “slightly cool” while persons in generally cool centrally air-
conditioned buildings seem inclined to describe their preferred indoor climate as “slightly
warm.”
ASHRAE RP-884 Final Report
Basic Results page MRL Australia 93
All Buildings
-4
-3
-2
-1
0
1
2
3
18 20 22 24 26 28 30 32 34
mean indoor operative temperature (oC)
dis
cre
p (o
C)
discrep = 0.12 + 0.01 * top
R2 = 0.002, p = 0.7034
Central HVAC and Mixed Mode buildings
-4
-3
-2
-1
0
1
2
3
18 20 22 24 26 28 30 32 34
mean indoor operative temperature (oC)
dis
cre
p (
o C)
discrep = -12.12 + 0.54 * top
R2 0.25, p = 0.0001
Naturally Ventilated Buildings
-4
-3
-2
-1
0
1
2
3
18 20 22 24 26 28 30 32 34
mean indoor operative temperature (oC)
dis
cre
p (
oC
)
discrep = 0.08 + 0.01 * top
R2 = 0.001, p = 0.8450
Figure 3.11: Dependence of discrep on mean indoor operative temperatures
3.1.6. Behavioural adjustments to indoor climate
As noted in the introductory chapter to this monograph, behavioral thermoregulation involves
a variety of purposive actions that modify the heat and mass exchanges that define the
body’s heat balance with its thermal environment. The most obvious behavioural response
for which we have quantitative data in the RP-884 database is that of clothing insulation.
The other “personal” or behavioral parameter governing the human body’s heat balance for
which we have quantitative estimates in the RP-884 database is metabolic heat. Thirdly,
indoor air speeds which were measured throughout the RP-884 database, is another
parameter over which building occupants exert some behavioral control, either by
ASHRAE RP-884 Final Report
Basic Results page MRL Australia 94
opening/closing windows, or turning on/off fans and similar devices. The following sections
examine these data and their relationships with various indices of indoor climate.
3.1.6.1. Thermal insulation adjustments indoors
Clothing insulation and also the incremental insulation of the chairs upon which the subjects
were sitting at the time of their questionnaire response were converted into clo units
according to the ASHRAE Standard 55 1992 methods. Table 3.12 summarises the main
statistics.
Table 3.12: Statistics for the thermal insulation variable (clothes + furniture) (clo).
centrally heated/air-conditioned buildings
naturally ventilated buildings
mixed-mode buildings
number of buildings in summer sample*
79
33
2
mean INSUL (±stdev) in the summer sample
0.70
(± 0.077)
0.66
(± 0.125)
0.71
(± 0.008) number of buildings in winter sample*
32
12
2
mean INSUL (±stdev) in the winter sample
0.92 (± 0.126)
0.93 (± 0.331)
0.83 (± 0.259)
Table 3.12 indicates significant seasonal differences in thermal insulation, with average
winter values exceeding 0.9 clo and average summer values around 0.7 clo (T=11.2,
df=109, p<0.001 for HVAC buildings; T=4.0, df=43, p<0.001 for NV buildings). While
seasonal differences were significant, the trivial differences between HVAC and NV sample
means failed to reach statistical significance in either season. However, building mean
insulation values showed greater variability in the naturally ventilated building sample
compared to the HVAC sample.
Figure 3.12 indicates a statistically significant relationship between the mean level of
thermal insulation worn inside a building and its mean indoor temperature. The
scattergrams in Figure 3.12 suggest that an exponential decay model might fit better than
the straight line printed in the graphs. However, due to the particular weighting
ASHRAE RP-884 Final Report
Basic Results page MRL Australia 95
Central HVAC and Mixed Mode Buildings
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
10 15 20 25 30 35mean indoor operative temperature (oC)
cha
ir +
clo
thin
g (
clo
) insul = 1.73 - 0.04 * top
R2 = 0.18, p = 0.0001
Naturally Ventilated buildings
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
10 15 20 25 30 35
mean indoor operative temperature (oC)
clo
thin
g +
cha
ir (
clo
)
insul = 2.08 - 0.05 * top
R2 = 0.66, p = 0.0001
All Buildings
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
10 15 20 25 30 35
mean indoor operative temperature (oC)
clo
thin
g +
ch
air
(cl
o)
insul = 1.87 - 0.04 * top
R2 = 0.51, p = 0.0001
Figure 3.12: Thermal insulation inside buildings (mean ± stdev) as a function of mean indoor operative temperatures
factors (i.e. sample sizes) applying to each point (building) in the graphs, the R2 statistic was
greatest for the simple linear fits shown in Figure 3.12. The correlation can be described as
“moderate” in the case of the “all buildings” graph of Figure 3.12. The lower panels indicate
that a much stronger relationship in the naturally ventilated buildings. This finding is possibly
due to the greater range of temperatures (independent variable) encountered in the naturally
ventilated building sample compared to the central HVAC/mixed mode building sample.
ASHRAE RP-884 Final Report
Basic Results page MRL Australia 96
The error bars either side of the plotted points in Figure 3.12 represent ± one standard
deviation around the within-building mean. All three panels of Figure 3.12 indicate a general
tendency for the standard deviation bars to contract in towards the mean, i.e. the variability
of clothing insulation to decrease, as indoor temperature increased. This possibly reflects a
diminution of degrees of freedom to adjust clothing as the number of individual garments
being worn reduced towards the socially acceptable minimum dress standards.
Figure 3.21: Thermal acceptability and outdoor climate. TSA represents directly assessed thermal acceptability levels within each building and PRXY_TSA represents thermal acceptability inferred from thermal sensation votes.
A building’s thermal acceptability rating, as inferred from thermal sensation votes, is logically
related to the gradient of that building’s thermal sensation regression models with respect to
indoor temperature (Appendix A). The range of acceptable operative temperatures for each
building was defined as rang_top in the RP-884 meta-analysis and presented earlier in
Section 3.1.3.4 simply by solving the mean thermal sensation versus mean indoor operative
temperature regression model (Appendix A) for mean ASHRAE thermal sensation votes of
±0.85. These values were chosen on the basis of Fanger’s PMV/PPD model (Fanger,
1970) which suggests they correspond to 80% acceptability levels (PPD=20%).
The complete lack of any statistical relationship between the range of acceptable indoor
operative temperatures (rang_top) and mean daily outdoor effective temperature (dayavet)
Figure 3.30: Regression analysis between thermal sensitivity and mean perceived control index (pcc_ag)
The adaptive model predicts that occupants of buildings in which there is a high level of
perceived control over thermal conditions will be less critical of indoor climatic conditions
than those in tightly regulated environments. Translating this hypothesis to the RP-884 meta-
analysis, Figure 3.30 plots each building’s thermal sensitivity statistic (dependence of
thermal sensation votes on either operative or standard effective temperature indices) in
relation to the building’s perceived control index score. The failure to reach statistical
significance in both the operative temperature and standard effective temperature index
graphs of Figure 3.30 lends no support to the adaptive hypothesis.
ASHRAE RP-884 Final Report
Basic Results page MRL Australia 126
Another extension of this perceived control hypothesis predicts that buildings with high
degrees of occupant control would score higher ratings on thermal acceptability than those
with low levels of control. Figure 3.31 fails to support this hypothesis since there is a
complete absence of any relationship between buildings’ direct thermal acceptability ratings
and their perceived control index score.
All Buildings
0
10
20
30
40
50
60
70
80
90
100
0 1 2 3 4 5 6 7
mean pcc_ag
TS
A (
% a
ccep
tabl
e)
tsa = 82.16 - 0.90 * pcc_ag
R2 = 0.03, p = 0.1904
Figure 3.31: Regression analysis between direct thermal acceptability rating of buildings (f_tsa_2) and their mean level of perceived control (pcc_ag)
Another corollary of the adaptive thermal control hypothesis is that occupants of buildings in
which there is high thermal controllability should be less likely to request a change of
temperature when presented with the thermal preference questionnaire item (MCI). Testing
this prediction with the RP-884 database can be done by tallying the percentage of each
building’s occupant sample who voted for either warmer or cooler temperatures (100 -
F_MCI_2). The thermal control hypothesis predicts that this percentage should decrease in
buildings where the degree of perceived control increases, but as seen in Figure 3.32, the
RP-884 database offers no empirical support for this hypothesis.
ASHRAE RP-884 Final Report
Basic Results page MRL Australia 127
All Buildings
0
20
40
60
80
100
0 1 2 3 4 5 6 7mean pcc_ag
% w
antin
g ch
ange
(1
00 -
f_m
ci_2
)
(100 - f_mci_2) = 47.33 + 2.02 * pcc_ag
R2 = 0.05, p = 0.0237
Figure 3.32: Regression analysis between the percentage of building occupants requesting a change in temperature (100- f_mci_2) and the mean level of perceived control (pcc_ag) for the building.
3.3.3. Building occupancy types - offices, residential and industrial
Another corollary of the adaptive hypothesis is that the thermal perception of a particular set
of thermal environmental factors is determined, in part, by the physics of the body’s heat
balance, but also by the functional context of the building setting. That is, perception of a
given state of body heat balance may differ, depending on the setting, because the
occupants’ expectations are context specific and as such, not directly transferable from, say,
the office setting to residential. In order to explore these issues in the RP-884 database,
building function was classified within the database using the information supplied by the
original researchers (Appendix C). A simple three-fold classification consisted of 1)
residential, 2) office, and 3) industrial.
Table 3.18 presents the summary statistics for each of the main thermal environmental
parameters across all three functional classes of building in the RP-884 database.
Obviously the overwhelming majority of buildings in the database were offices and so the
analyses and conclusions developed in earlier sections of this chapter apply primarily to this
ASHRAE RP-884 Final Report
Basic Results page MRL Australia 128
class of building. However, the small sample of residential buildings summarized in Table
3.18 permit some comparisons to be drawn with office buildings. Firstly, the percentage of
physical measurements of indoor climates actually meeting the ET* recommendations of
ASHRAE Standard 55-1992 was remarkably low for the 16 residential buildings in the
sample, ranging from an average of 6% in summer to 21% in winter. These low compliance
levels mainly resulted from the high mean indoor summer temperature of 30°C and low
indoor temperature means of 19°C in winter. Table 3.18 also indicates that mean indoor air
speeds were generally higher in residential buildings compared with office and industrial
settings, and they also showed a much larger seasonal variation in the residential cases.
While mean metabolic rate estimates indoors remained relatively constant across
residential and office settings at about 1.2 met units, they were noticeably higher in the small
number of industrial buildings included in the RP-884 sample, with means ranging between
2 and 2.5 met units. The seasonal swing in mean building occupant thermal insulation levels
was relatively small in the case of office and industrial buildings, amounting to less than 0.2
clo units. However, there was a much larger seasonal adjustment of insulation means
across the residential buildings in the sample, suggesting that clothing adjustment
represents a more powerful adaptive response in the home than in the workplace.
ASHRAE RP-884 Final Report
Basic Results page MRL Australia 129
Table 3.18: Summary of the thermal environmental conditions in three classes of building included in the RP-884 database.
residential summer winter
offices summer winter
industrial summer winter
number of buildings in sample$
12
4
98
38
4
4
mean (±stdev) * operative temp (°C)
30.2 ±0.932
18.8 ±4.86
24.3 ±2.07
22.6 ±0.74
22.6 ±1.40
20.4 ±1.64
mean (±stdev) relative humidity (%)
43.9 ±17.0
45.9 ±13.3
53.2 ±9.1
32.8 ±10.5
48.9 ±3.4
43.7 ±8.3
mean (±stdev) % compliance with ASHRAE Standard 55@
6.4 ±7.4
20.9 ±13.0
55.8 ±30.6
86.5 ±16.6
0 ±0
0 ±0
mean (±stdev) air speed (m s-1)
0.31 ±0.10
0.15 ±0.04
0.13 ±0.07
0.08 ±0.03
0.06 ±0.00
0.06 ±0.002
mean (±stdev) insulation (clothes+chair) (clo)
0.58 ±0.16
1.34 ±0.16
0.70 ±0.075
0.89 ±0.16
0.66 ±0.06
0.82 ±0.08
mean (±stdev) metabolic rate (met)
1.20 ±0.08
1.12 ±0.04
1.20 ±0.10
1.17 ±0.05
2.54 ±0.08
2.14 ±0.573
* Mean and stdev figures quoted are averages across the n buildings in the table’s cell $ Note that in some studies large samples across several buildings have been treated as one building due to
the way the data were originally supplied to the RP-884 database @ Percentage of physical observations within each building falling between ASHRAE Standard 55-92 ET*
limits for the relevant season
Having summarized the physical environmental and behavioral factors in three classes of
building in Table 3.18, the main task of Table 3.19 is to summarize the subjective thermal
responses to those indoor climatic conditions, again for residential, office and industrial
settings. It appears that the samples of residential building occupants were, on average,
less than half as sensitive to indoor temperature as the office building samples, since the
gradient of their thermal sensation votes with respect to indoor operative temperature was
about one vote per every 3~5 K change in temperature. In comparison the statistic from the
sample of office buildings was closer to one sensation unit to every two degrees. Another
noteworthy comparison between building function in Table 3.19 concerns acceptability
ratings of buildings. Despite the very low level of ASHRAE Standard 55 compliance in the
residential buildings in the database (Table 3.18), their acceptability ratings, at least in
summer, were not appreciably lower than those registered in office buildings where the
Standard 55 compliance levels were a good deal higher. Even in winter the acceptability
ratings in residential sample buildings dropped only about 10% below the office buildings’
average rating, whereas the ASHRAE Standard compliance levels dropped by over 60%
ASHRAE RP-884 Final Report
Basic Results page MRL Australia 130
from office to residential settings. The implication of these comparisons is clearly that
contextual factors have a strong bearing on how a given set of indoor thermal environmental
parameters will be perceived by the occupants.
Table 3.19: Summary of the subjective thermal responses across the three classes of building included in the RP-884 database.
residential summer winter
offices summer winter
industrial summer winter
number of buildings in sample$
11
3
66
21
1
0
mean ±stdev thermal sensitivity (i.e. regression gradient (sensation vote/deg K top)
* Mean and stdev figures quoted are averages across the n buildings in the table’s cell $ Note that in some studies large samples across several buildings have been treated as one building due to
the way the data were originally supplied to the RP-884 database. Also, sample size is based only on statistically significant regression models, except where otherwise indicated (i.e. n=...)
n.a. “not applicable”
3.4. Summary of basic results
This chapter has presented a complex array of findings, exploring different thermal indices,
different dimensions of subjective comfort, the effects of different seasons and climates,
different modes of indoor climate control, and different patterns of building occupancy. This
final section summarizes and interprets the key findings in relation to the adaptive
hypothesis of thermal perception. This synthesis provides the starting point for developing
more complex adaptive models in the next chapter.
3.4.1. Summary of thermal sensation, acceptability and preference
Subjective thermal comfort research has been unfortunately complicated over the last thirty
or forty years with the adoption of several different constructs of thermal perception. This
ASHRAE RP-884 Final Report
Basic Results page MRL Australia 131
chapter dealt with three of these -- thermal sensation, thermal acceptability and thermal
preference. Sensation appears to be the most universally used version of questionnaire
scale, and certainly the most ubiquitous within the RP-884 database. Consequently the
more complex and abstract statistical analyses in this project were necessarily confined to
this expression of thermal comfort. However, there was a useable quantity of data on
thermal acceptability and preference within database as well, permitting several
observations to be made about the semantic similarities and differences between all three
constructs and the implications for practical applications.
A very clear observation that emerges from the RP-884 analyses of direct assessments of
thermal acceptability is that building occupants’ responses to direct questions such as this:
“Is the thermal environment in this building at the moment acceptable to you or not?”
bear virtually no relationship to the objective, physical conditions prevailing within the
building at the time of the questionnaire. Evaluations of RP-884 database buildings’ indoor
climatic quality in terms of its compliance with the relevant summer or winter temperature
prescriptions of ASHRAE Standard 55 were completely dissociated from the direct
acceptability ratings of those same buildings by their occupants. We therefore regard
questionnaire items on direct thermal acceptability as being too ambiguous and vague to be
of any practical value in thermal comfort research or practice.
While direct ratings of thermal acceptability for indoor climates may not be particularly useful,
there remains a practical need for information about the range of temperatures which can be
regarded as acceptable for a given building in a specific climatic context. Accepting
Fanger’s (1970) assumption that a mean sensation vote of ±0.85 corresponds with 80%
thermal acceptability (or ±0.50 corresponds with 90%), it was possible in Section 3.1.3.4 to
extract from the database ranges of acceptable temperatures within each of the sample
buildings. ASHRAE Standard 55 suggests operative temperature ranges between 3K and
3.5 K. The RP-884 database, on the other hand, indicated that only a 2.5 K range was
acceptable, on average, within HVAC buildings. In NV buildings, however, the 90%
acceptable range extended significantly further, with an average of 5 K. This stretched to 7
K for the less stringent 80% acceptability criterion.
ASHRAE RP-884 Final Report
Basic Results page MRL Australia 132
The actual range of acceptable operative temperatures within any particular building was
found to depend to a large extent on the degree of indoor climatic variability measured within
that building (r=+0.66). This relationship suggests that if, through prior experience,
occupants of a building come to expect considerable thermal variability, the range of
temperatures regarded as acceptable will extend accordingly.
Compared to direct thermal acceptability ratings, thermal sensation rating scales showed a
much more consistent pattern of association with indoor thermal environmental indices.
Correlations within each building in the database were statistically significant in the majority
of cases in Appendix A, and this permitted the derivation of thermal neutralities wherever the
sample of building occupants was large enough. Thermal neutrality is defined as that value
of a thermal index (TOP, ET, SET or PMV) corresponding with a mean thermal sensation
rating of “neutral” by the building’s sample of occupants. Assuming that neutrality is
synonymous with the “optimum thermal condition” for a particular building, it should be more
useful than direct acceptability ratings as a basis for application and practice.
The temperatures which building occupants felt to be neutral were broadly similar in both
HVAC and NV buildings, coming in at about 24°C in summer and 22.5°C in winter (TOP or
ET). These figures approximate the centre of ASHRAE Standard 55’s summer and winter
comfort zones. Neutrality defined in terms of the fully developed heat balance index such as
SET also fell within the same range. Thermal neutrality depended on mean temperatures
within both HVAC and NV buildings, but the rate of change of neutrality with respect to mean
building operative temperature was twice as steep in NV buildings as it was in HVAC
buildings. This finding suggests that occupants of NV buildings were twice as adaptable in
terms of making themselves feel neutral than their counterparts in HVAC buildings.
Fanger’s PMV model seemed to be reasonably accurate at predicting building neutralities
across the whole sample of buildings, with an average prediction error less than half a
degree (compared to observed neutralities). However, the standard deviation of the
prediction error between buildings was quite high at 3.8 K. This suggests that, while the
model worked well across a large sample of buildings, its predictions within any single
building could be significantly wrong. Assuming that the quality of input data in the RP-884
database is of a uniformly high standard (Class 1 and II studies only), the explanations for the
ASHRAE RP-884 Final Report
Basic Results page MRL Australia 133
PMV model’s building-specific prediction errors must lie in non-thermal factors beyond the
human heat balance. For example, the average prediction error in NV buildings was nearly
a full degree, and its between-building standard deviation exceeded 5 K, suggesting that
contextual factors degraded the model’s predictive powers.
Preferred temperatures (as distinct from neutral or acceptable temperatures) could only be
derived within only a subset of the RP-884 building sample. Preferences were found to
occur within the 21~27°C range in most buildings. The average semantic discrepancy
between neutral and preferred temperatures in buildings was generally within half a degree.
While the sign and magnitude of the semantic discrepancy was unrelated to mean warmth
within naturally ventilated buildings, Section 3.1.5 reported a significant tendency for
neutrality to diverge away from preferred temperature within the sample of HVAC buildings
(r=+0.50) as mean temperatures within buildings departed from 22.5°C.
3.4.2. Summary of thermal sensitivity and behavioural thermoregulation
The linear dependence of thermal sensation votes and indoor climate showed a complex
pattern of differences between HVAC and NV buildings, and also between the various
indices of indoor climate (Section 3.1.1). Using the simpler indices such as TOP and ET,
we found that persons in centrally controlled HVAC buildings were, on average, more than
twice as sensitive to changes in temperature as their counterparts in naturally ventilated
buildings. However, this heightened sensitivity diminished when the more complex heat-
balance indices of warmth such as PMV and SET were used, with a sensation category
having a fairly constant temperature width of about four degrees. One interpretation is that
occupants of naturally ventilated buildings behaviorally regulate their heat balance with
clothing and air speed adjustments such that they can accommodate wide variations in
temperature indoors without adverse impacts on thermal sensation -- that is, they are
actively thermoregulating their sensations. In contrast, occupants of centrally heated and air-
conditioned buildings seem less adaptive behaviorally, and as a result their thermal
sensations appear more sensitive to excursions of indoor temperature away from average,
expected set-points.
This interpretation is further supported by the clear relationships between behavioral factors
(clothing and air speed) and indoor temperature (Section 3.1.6). While the seasonal mean
ASHRAE RP-884 Final Report
Basic Results page MRL Australia 134
insulation (clothes plus chair) levels were broadly similar in NV and HVAC buildings, there
was considerably greater variability within and between the NV buildings. Also, there was a
much higher correlation between insulation and indoor temperature in NV buildings (r= -
0.81) compared to HVAC buildings (r= -0.42). The data for indoor air speeds lends
additional support to this interpretation, with the summer average being twice as high in NV
as in HVAC buildings, and the variability within NV buildings also being greater.
Furthermore, the closer correlation between indoor air speed and indoor temperature in NV
buildings (r=+0.73) compared with HVAC buildings (r=+0.58) reinforces the conclusion that
occupants of naturally ventilated buildings were behaviorally more active in thermoregulating
their thermal sensations than were their counterparts in HVAC buildings.
3.4.3. Summary of the effects of outdoor climate on thermal perception indoors
The temperatures found to be neutral within both HVAC and NV buildings varied, depending
on season, with significantly warmer neutralities (defined in terms of operative temperature)
occurring in summer compared to winter. These seasonal differences became less
consistent as the thermal index used to define neutrality increased in complexity (PMV and
SET), but this may simply result from the climatologically inaccurate definition of “summer”
and “winter” applied throughout the database.
Parameterizing outdoor climate simply as the mean of daily maximum and minimum
effective temperatures (in shade) provided a more rational basis for exploring these effects
in Section 3.2.1. Thermal neutrality within buildings was found to correlate positively
(r=+0.65) with mean outdoor ET. While the strength of correlation was roughly comparable
between HVAC and NV buildings, the slope of the linear relationship was not -- indoor
neutrality was about twice as responsive to outdoor temperature in naturally ventilated
buildings compared to air-conditioned. This difference suggests that much of the
adaptability observed in free-running buildings, described earlier as being driven by
expectations of warmth indoors, may in fact be driven by outdoor climate. Obviously indoor
and outdoor temperatures are highly correlated in naturally ventilated buildings (r=+0.91,
compared to r=+0.53 in HVAC buildings), so the temptation to include both in a multiple
regression model of thermal neutrality must be resisted if the stability of regression
coefficients is to be maintained.
ASHRAE RP-884 Final Report
Basic Results page MRL Australia 135
As for explaining why thermal comfort adaptability might be related to outdoor climate, the
role of behavioral adjustments was the first place to look (Section 3.2.4). In particular, mean
clothing insulation worn inside buildings, both HVAC and NV, was found to correlate
negatively with mean warmth in the outdoor environment (r=-0.63). Mean air speeds inside
buildings were also found to be, correlated with outdoor warmth, but much more so in the
case of NV buildings (r=+0.80 compared to r=+0.44 in HVAC). It was also clear that the
range of mean air speeds found within naturally ventilated buildings (often exceeding
0.4~0.5 m/s) was much wider than in HVAC buildings where it rarely exceeding the 0.2 m/s
mandated in ASHRAE Standard 55 (1992).
The combined effect of these behavioral thermoregulatory processes and their relationships
with outdoor climate were examined in 3.2.1.4 where building neutralities predicted by the
heat-balance index PMV were regressed on mean outdoor ET. The simplistic description of
the PMV index as a “static” model throughout much of the adaptive comfort literature
(reviewed in Chapter 1) was clearly not supported in this analysis, because observed
regression equations were statistically significant and positive in both HVAC and NV
samples.
3.4.4. Summary of the effects of contextual factors and perceived control
The RP-884 index of perceived thermal control comprised a check-list of specific adaptive
opportunities and their relative efficacy which we applied to each of the buildings in the
database. The index clearly differentiated mixed-mode buildings from naturally ventilated
buildings as affording their occupants the greatest degree of thermal control, largely due to
their provision of both thermostats and operable windows. Naturally ventilated buildings
came up second in average control index rankings, while the centrally-controlled HVAC
buildings scored worst on the index. Despite the ability of the index to differentiate the three
type of building in the RP-884 database, we found it had no correlation with thermal
acceptability, sensitivity or preferences.
Rather than interpreting this as a categorical negation of the role of perceived control in
thermal perception, we think there are at least two alternative explanations:
ASHRAE RP-884 Final Report
Basic Results page MRL Australia 136
1. The validity of the index itself is flawed. The perceived control scale was constructed very
simplistically, mainly due to the nature of the raw data supplied to the RP-884 database.
2. Alternatively, the effects of perceived control may not have such a simple and direct
relationship with thermal perception. The constructs of perceived control and adaptive
opportunity within buildings may in fact exert more complex effects on thermal perception,
and as a result, be statistically significant once other dimensions of indoor and outdoor
climate have been taken into account. The possibility of complex, interactive effects of
the pcc_ag index on thermal perception will be explored further in Chapter 4.
The functional classification of RP-884 sample buildings into office, residential and industrial
uncovered sharp differences in the basic indoor thermal environmental parameters such as
air speed and temperature. The three classes of building were also clearly differentiated in
terms of their compliance with effective temperature index limits within ASHRAE Standard
55. For example, a majority of the observations made inside office buildings, regardless of
whether they were air-conditioned or not, complied with the ASHRAE Standard 55’s ET*
limits, whereas the typical residential or industrial building scored less than 20% compliance
with the standard. We also observed distinct differences in the degree of behavioral
thermoregulatory adjustment made by residential building occupants compared to office
workers. For example, seasonal clothing insulation contrasts were sharper in the residential
as opposed to office setting.
Despite these obvious differences in physical and behavioral features of indoor climate for
office and residential buildings in the sample, we couldn’t discern sharp differences in
occupant evaluations of the buildings’ indoor climatic quality. Despite their relatively poor
performance on the objective physical indoor climatic criteria, occupants’ thermal
acceptability ratings for residential buildings were comparable to those within office
buildings. The strength of this contextual effect on subjective response is no doubt part of
the explanation for the lack of any statistical correlation between thermal acceptability
responses and indoor or outdoor climatic indices, as noted in Section 3.4.1 of this chapter.
ASHRAE RP-884 Final Report
Basic Results page MRL Australia 137
ASHRAE RP-884 Final Report
__________________________________________________________________________________ Towards Adaptive Models page MRL Australia 139
CHAPTER 4 -- TOWARDS ADAPTIVE MODELS
The preceding chapter presented field evidence for the effects of indoor and outdoor
climatic factors on the way building occupants perceive the environments provided by their
buildings. Contextual factors such as whether the building is residential or work-place, and
how much adaptive opportunity it affords were also investigated. The aim of this chapter is
to develop these themes into adaptive models of thermal comfort, relating them back to
previous research on the topic, as reviewed in Chapter 1. These adaptive models will form
the basis of a variable temperature standard for indoor climate to be proposed in the next
chapter.
The reader should note that we have tried to make the terminology of equations in this and
subsequent chapters more descriptive than the nomenclature of earlier chapters.
4.1. The semantics of thermal comfort
Chapter 3 demonstrated that the indoor temperature regarded by building occupants as
“neutral” did not always coincide with that which they rated as most “acceptable” or
“preferable.” Evidence was presented for a “semantic artefact” which caused neutrality
(derived from thermal sensation) to be displaced to the right of preference (warmer) in hot
climates, and to the left of preference (cooler) in cold climates. In other words, in hot
climates people preferred a thermal sensation slightly cooler than neutral, while in colder
climates they preferred to feel slightly warmer than neutral.
Earlier researchers have found similar semantic effects -- de Dear et al. (1991a) recorded a
group-mean thermal sensation of -0.33 for their Singaporean climate chamber subjects
while they were seated in their self-determined preferred temperature (i.e. they preferred to
feel cooler than neutral). This is the equivalent of one whole degree (K) in semantic offset for
the clothing, metabolic rate and air speed in question. Oseland (1994a,b) also reported
semantic discrepancies between preference and thermal sensations, finding that they were
stronger in their winter study where subjects decidedly preferred thermal sensations that
were slightly warmer than neutral. The extent to which culture and climate affect people’s
ASHRAE RP-884 Final Report
__________________________________________________________________________________ Towards Adaptive Models page MRL Australia 140
thermal preferences, and the semantics they use to describe them, has also been discussed
at length by McIntyre (1978a, 1978b, 1982).
A test of this semantic artefact hypothesis within the RP-884 database appeared in Figure
3.24 as a set of graphs relating the discrepancy (discrep) between neutrality and preference
within each building to the mean level of effective temperature in outdoor climate. There we
found a significant linear correlation of r=+0.62 between discrep and mean outdoor effective
temperature (dayavet) within the HVAC (and a few mixed-mode) buildings in the database.
The linear equation indicated that neutrality and preference coincided only in those HVAC
(and mixed-mode) buildings located in climates where the mean outdoor effective
temperature was 13.6°C.
semantic effect = -0.95 + 0.07 * mean outdoor ET* for HVAC buildings eq. 4.1
In climates warmer than this, indoor preference became progressively cooler than neutrality,
while in regions where mean outdoor effective temperature fell below 13.6°C, preferred
temperature was warmer than neutral temperature. Clearly this semantic effect needs to be
accounted for when we develop variable temperature standards for HVAC buildings in the
next chapter.
In contrast to the situation just described for HVAC (and a few mixed mode) buildings, there
was no empirical evidence in the RP-884 database for a semantic effect within naturally
ventilated buildings (Figure 3.24). Exactly why people use words like “slightly warm” or
“slightly cool” differently in different types of buildings remains unclear at this stage.
Whatever the interpretation, it seems reasonable to develop a variable temperature
standard for naturally ventilated buildings exclusively on the basis of thermal neutrality, as
derived from rating scales such as the ASHRAE and Bedford 7-pt scales, ignoring the
semantic offset altogether.
The implications of the semantic effect on the HVAC building adaptive model can be
depicted graphically in Figure 4.1. There the “adaptive model” represents the thermal
neutrality function with respect to outdoor temperature from chapter three minus the
semantic effect just discussed.
ASHRAE RP-884 Final Report
__________________________________________________________________________________ Towards Adaptive Models page MRL Australia 141
buildings with centralized HVAC
-5
0
5
10
15
20
25
30
-5 0 5 10 15 20 25 30 35
mean daily outdoor effective temperature (oC)
com
fort
tem
per
atu
re (
oC
)
-5
0
5
10
15
20
25
30
sem
anti
c ef
fect
(K)
neutrality
semantic effect
adaptive modelincluding semantics
Figure 4.1: An adaptive model for HVAC buildings that accounts for the semantic offset between neutrality and preference.
4.2. Comparison of RP-884 models with earlier adaptive model publications
Indoor Climate: As noted in Chapter 1 (literature review), Humphreys (1975, 1978; 1981)
published various statistical models of the adaptive dependence of indoor thermal neutrality
on mean indoor and outdoor temperature. His statistical analysis of thirty six Class III studies
from various countries around the world revealed a clear dependence of thermal neutralities
(roughly equivalent to neut_top in RP-884 nomenclature) on the mean levels of air or globe
temperature (roughly equivalent to top in RP-884) recorded within the buildings (Humphreys,
Figure 4.3: Comparison of the RP-884 adaptive model (based on observed neutralities corrected for semantic effects) and the “static” model (based on PMV predictions) for HVAC buildings.
It is interesting to note that this graph so closely matches predictions of PMV with
observations in real HVAC buildings, whereas so many of the earlier thermal comfort field
research papers which we discussed in Chapter 1’s literature review indicated quite the
opposite. Indeed, some of those anomalous papers were from authors who contributed their
raw data to this project’s database. Therefore our success at bringing PMV predictions
into line with observations in HVAC buildings most probably can be attributed to the quality
controls and precautions we took when assembling the RP-884 database, which
transformed, to some extent, the raw data used in the authors’ original analyses. Among the
more important of these were probably:
• setting minimum standards on instrumentation and protocols for data going into the RP-
884 database,
• conversion of all clo estimates throughout the entire database to a single standard
(ASHRAE 55-92),
• inclusion of the thermal insulation effects of the chairs used by subjects (McCullough and
Olesen, 1994),
• recalculation of thermal indices from raw data throughout the entire database with a
consistent software tool (Fountain and Huizenga, 1995),
ASHRAE RP-884 Final Report
__________________________________________________________________________________ Towards Adaptive Models page MRL Australia 148
• application of a consistent set of statistical techniques to all raw data instead of relying on
different author’s approaches to thermal neutrality, preference and other statistically
derived parameters,
• conducting the meta-analysis at the appropriate scale of statistical aggregation, namely
the individual building.
We are therefore led to the conclusion that Fanger’s PMV model is, in reality, an “adaptive”
model which is suitable for application as it was initially proposed back in 1970 by Fanger
himself; as an engineering guide in HVAC buildings the world over1. The main sticking point
with its application in predictive mode before a building is constructed, or occupied, is that it
is unusual to have detailed observations on mean clo values or air speeds within a building
at the design stage. The practical solution here is to seek further guidance from the RP-884
database. We know from Figure 3.25 that the thermal insulation (in clo units, clothes plus
chair) applicable to PMV calculations is highly correlated with mean outdoor effective
Figure 3.27 indicates that mean room air speeds (m s-1) within HVAC buildings are also
correlated with mean outdoor effective temperature:
mean room air speed = 0.08 * e+0.014*(mean outdoor ET*) (r= +0.44) eq 4.7
so it seems not unreasonable to anticipate the unknown inputs to PMV simply from a
knowledge of the outdoor weather/climate conditions for the site in question.
An even simpler approach is to directly predict PMV-based neutrality using the linear
regression model depicted in Figure 3.20. In effect this amounts to predicting the aggregate
effects of climate on clothing insulation and room air speeds within HVAC buildings.
1 In the introductory chapter to his book entitled “Thermal Comfort - Analysis and Applications in Environmental Engineering” which introduced the PMV model, Fanger was quite clear that the book, and by implication, the PMV model at its core, were intended for application by the HVAC industry in the creation of “artificial climates” in “controlled spaces.” The generalisation of the PMV model to all spaces intended for human occupancy, HVAC or NV, was a much later development that we disagree with.
ASHRAE RP-884 Final Report
__________________________________________________________________________________ Towards Adaptive Models page MRL Australia 149
Therefore, the adaptive model for HVAC buildings (noted here as “PMV, adaptive”) reduces
to this simple linear equation (from Figure 3.20):
comfort temperature in HVAC = 22.6 + 0.04 * mean outdoor ET* (r= +0.50) eq 4.8
So, it would appear that the occupants of buildings with centralized HVAC systems have
become adapted to the temperatures that they encounter within their buildings -- generally
within the narrow 22~24°C range. Of course this begs the question of whether or not it is
possible to extend the range of comfort adaptation by deliberately letting indoor HVAC
setpoints more closely track outdoor weather and climatic conditions? We concede that a
purposely designed intervention field experiment on a “real” building, would be the most
appropriate way to test this hypothesis. However, we can draw comparisons with the
naturally ventilated RP-884 sample where the rate of change in thermal insulation with
respect to variations in outdoor climate (Figure 3.25) was significantly greater than in
centrally conditioned buildings. Across the 5⇒ 30°C range of mean outdoor effective
temperatures in Figure 3.25, building occupants’ mean insulation (including chair effects)
varied by about 0.3 clo units in the HVAC sample, whereas the clothing response was more
than double this in naturally ventilated buildings across the same outdoor temperatures. In
short, naturally ventilated building occupants appear to be prepared to take on greater
personal responsibility for maintaining their thermal comfort, when required to. Whether they
would be prepared to do likewise if required in HVAC buildings remains a moot point
deserving further research.
The same line of reasoning can be applied to indoor air velocities in HVAC buildings. We
noted they were confined to very low levels (virtually still air, at <0.2 m s-1) within the RP-884
sample of HVAC buildings, almost regardless of outdoor climate (see Figure 3.27). This
stood in marked contrast to the naturally ventilated sample where within-building mean
velocities went up to 0.4 m s-1 for outdoor mean effective temperatures of about 30°C.
These velocities could possibly be feasible inside centrally conditioned buildings, perhaps
with supplementary air movement in the occupied zone provided by local fans or other
means for individual thermal control. Present-day HVAC building occupants appear
adapted to conditioned, still-air conditions, but they may be willing to more actively regulate
ASHRAE RP-884 Final Report
__________________________________________________________________________________ Towards Adaptive Models page MRL Australia 150
convective and latent heat losses if the behavioral opportunities to do so were made
available to them and HVAC set-points provided the stimulus (eg allowing for warmer indoor
temperatures during summer).
As noted above, new research is required to establish just how much thermoregulatory
responsibility occupants of HVAC buildings may be prepared to accept. Future field
experimentation may suggest that simply predicting PMV-based neutralities from a
knowledge of mean outdoor temperature, as in eq. 4.8 above, is inappropriate for the
purpose of establishing HVAC set-points. Instead, it may well be more appropriate to first
estimate likely indoor clothing insulation levels and air velocities from equations resembling
those established for the naturally ventilated sample in Figures 3.25 and 3.27 (instead of
the HVAC models presented in eqs. 4.6 and 4.7 above), and then iteratively solving the
PMV model for “neutral” operative temperature. Clearly further field experimentation on
these questions of thermal adaptation in HVAC buildings is required.
4.3.2. Comparisons within the naturally ventilated building sample
Figure 4.4 repeats the “adaptive” versus “static” comparisons for the naturally ventilated
buildings within the RP-884 database. One important departure from the method just
applied to HVAC buildings, however, is the omission of the semantic effect, as discussed in
Section 4.1. This is because we were unable to discern any systematic relationship
between the preferred and neutral temperatures for the naturally ventilated buildings
analysed in Figure 3.24.
The remarkable agreement found between PMV and adaptive models in the HVAC building
sample clearly breaks down in the context of naturally ventilated buildings where the adaptive
model shows a gradient almost twice as steep as the heat-balance PMV model’s. This
divergence tested positive using the Kleinbaum et al. technique (1988) (T=2.43, df=80,
p<0.05). It therefore appears as if behavioral adjustments to body heat balance (i.e.
biophysical effects) account for only about half of the climatic dependence of comfort
temperatures within naturally ventilated buildings. In effect, the PMV model has been
demonstrated to function as a partially adaptive model of thermal comfort in naturally
ventilated buildings.
ASHRAE RP-884 Final Report
__________________________________________________________________________________ Towards Adaptive Models page MRL Australia 151
However, there still remains the other half of the adaptive effect to be explained. Having
partialled out the effects of behavioral adaptations, we’re left with the physiological
(acclimatization) and psychological (habituation) hypotheses discussed in Chapter 1. There
we noted that effects of acclimatization were not in evidence during climate chamber
experiments on moderate heat/cold stress exposures, so it is not surprising that they failed
to reappear in the field settings analyzed in RP-884. Therefore, by a process of elimination,
we are left with psychological adaptation (i.e. expectation and habituation) as the most
likely explanation for the divergence between field observations and heat-balance (PMV)
Figure 4.4: Comparison of the RP-884 adaptive model (based on observed neutralities in Figure 3.18) and the “static” model (based on PMV predictions) applied to naturally ventilated buildings.
One might wonder why the laboratory-based PMV heat balance model works so well in RP-
884’s HVAC buildings but not so for the NV buildings? Perhaps we can regard the former
as being quite comparable to the climate chamber setting? In both climate chambers and
HVAC buildings the thermal environment is entirely regulated by processes outside the
person-environment feedback loop discussed in Chapter 1. Naturally ventilated buildings,
on the other hand, are much more “interactive,” with adaptive feedback loops being closed
at both behavioral and psychological levels.
ASHRAE RP-884 Final Report
__________________________________________________________________________________ Towards Adaptive Models page MRL Australia 152
4.4. Adaptive models for acceptable ranges of indoor temperatures
Preceding sections defined some simple adaptive models for predicting optimal comfort
temperatures indoors, but overlooked the question of what sort of temperature
inhomogeneity might be acceptable. We saw in Section 3.1.3.4, particularly in Figure 3.9, a
direct correlation between the range of acceptable operative temperatures within each
building and its internal temperature variability (standard deviation). The relationship
reached statistical significance only for the naturally ventilated buildings in the RP-884
sample where the following linear regression model achieved a correlation coefficient of r=
+0.51:
range of acceptable temperatures = 4.2 + 1.65 * (stdev of indoor temperature) eq 4.9
The failure of a similar model to reach significance within centrally conditioned buildings
reinforces the fundamental difference between the HVAC and natural ventilation contexts
discussed at length throughout in this report. In the naturally ventilated setting, it appears as
if building occupants extend their range of thermal acceptability to accommodate the range
of thermal variation expected within their buildings.
We propose the simple model in eq 4.9 as the adaptive approach to prediction of 80%
acceptable ranges within naturally ventilated buildings. But for many applications it simply
will not be feasible to anticipate the standard deviation of indoor operative temperatures for
a building that is either yet to be built or not fully monitored for any significant length of time.
Therefore a more practical alternative for prescribing acceptable indoor temperature ranges
may be to rely on the RP-884 observations, as described in Table 3.9. In HVAC buildings
the general comfort 80% acceptability criterion corresponded, on average, to a range of two
degrees (K) either side of the optimal comfort temperature. Tightening the acceptability
criterion from 80% to just 90% in RP-884’s HVAC building sample meant a narrowing of the
acceptable range to ± 1.2 K. In either case, the corresponding 80% and 90% ranges
observed in the naturally ventilated RP-884 sample were significantly wider, at ± 3.5 K and ±
2.5 K respectively.
ASHRAE RP-884 Final Report
__________________________________________________________________________________ Towards Adaptive Models page MRL Australia 153
If these acceptable ranges are going to be applied to adaptive models which predict
optimum indoor temperatures on the basis of outdoor climatic conditions, we need to
address the possibility that the acceptable ranges themselves are also dependent on
outdoor climate. For example, one might speculate that the acceptable range diminishes as
outdoor climate becomes hotter for the simple reason that indoor clothing insulation levels
also decrease with increasing outdoor temperature (see Section 3.2.4.1). A statistical
regression test of this possibility was performed by fitting a regression model to the
dependence of acceptable indoor temperature ranges on outdoor effective temperature,
and the results are reported in Table 4.1. It can be assumed that, if the regression model
turns out to have a statistically insignificant gradient term, the subsample’s mean acceptable
range (as described in Table 3.9) can legitimately be applied across all climate zones. As
seen below in Table 4.1, none of the acceptable range models achieved statistical
significance at the 95% confidence level, regardless of building type nor acceptability level.
Therefore, the variable temperature standards to be proposed in the next chapter can be
based on an optimal temperature predicted from outdoor climate, plus or minus a constant
acceptable temperature range for the building type in question, which does not vary with
climate.
ASHRAE RP-884 Final Report
__________________________________________________________________________________ Towards Adaptive Models page MRL Australia 154
Table 4.1: Assessment of the dependence of acceptable indoor temperature ranges on outdoor effective temperature.
centrally heated/air-conditioned
buildings
naturally ventilated buildings
number of buildings 108 (3 missing values)
41 (4 missing values)
number of buildings with thermal sensation regression models achieving 95% significance*
63
(58% of total)
33
(75% of total) Mean range of indoor temperatures based on the 80% acceptability criterion (K)
4.1
6.9
Regression model for the dependence of the 80% acceptable temperature ranges on outdoor effective temperature Statistical T-test for the regression gradient Statistical significance of T-test
y=3.08 + 0.05*x
1.81 p>0.05
y=6.28 + 0.03*x
0.36 p>0.10
Mean range of indoor temperatures based on the 90% acceptability criterion (K)
2.4
4.9
Regression Model for the dependence of the 90% acceptable temperature ranges on outdoor effective temperature Statistical T-test for the regression gradient Statistical significance of T-test
y=1.81 + 0.03*x
1.81 p>0.05
y=3.70 + 0.02*x
0.36 p>0.10
* Based on those thermal sensation (ASH) models in Appendix A (y=a + b*TOP) achieving 95% statistical significance or better
The next chapter in this report will summarize this chapter’s adaptive models into a pair of
variable temperature standards - one for application in HVAC buildings and another for
application in the naturally ventilated context.
ASHRAE RP-884 Final Report
Variable Temperature Standard page 155 MRL Australia
CHAPTER 5 - VARIABLE TEMPERATURE STANDARDS
The last remaining task for ASHRAE RP-884 is to propose variable temperature thermal comfort
standards. The statistical analyses and adaptive models in Chapters 3 and 4 were presented separately
for buildings with and without centrally controlled HVAC systems. It seems logical, therefore, to partition
this chapter’s variable temperature standards along the same lines. This distinction between centrally-
controlled HVAC buildings in which individual occupants have little or no control over their immediate
thermal environment, and naturally ventilated buildings in which occupants at least have control over
windows, is a unique feature of the ASHRAE RP-884 project. All thermal comfort standards to date
(see Chapter 1), both extant and proposed, regardless of whether they were based on so-called “static”
or “adaptive” models, have been promulgated as universally applicable across all types of building.
By not differentiating their contexts for application, earlier comfort standards are, in effect, extrapolating
from relationships established in centrally controlled HVAC settings to naturally ventilated contexts, and
vice versa. In contrast, a fundamental tenet of RP-884 has been that the indoor climates found in HVAC
and naturally ventilated buildings are not only quantitatively different, but also qualitatively different, and
as such, they require separate comfort standards.
The reader is requested to regard the two standards in this Chapter as self-contained documents. There
is, therefore, some duplication of definitions and related material across the two standards.
5.1. A variable temperature standard for application in buildings with centrally controlled
HVAC
5.1.1. Purpose
To specify the combinations of indoor space environment and personal factors that will produce thermal
environmental conditions acceptable to a majority of the occupants within centrally heated and air-
conditioned spaces.
ASHRAE RP-884 Final Report
Variable Temperature Standard page 156 MRL Australia
5.1.2. Scope
• The environmental factors addressed are temperature, thermal radiation, humidity, and air speed; the
personal factors are those of activity and clothing.
• It is intended that all of the criteria in this standard be applied together, since comfort in the space
environment is complex and responds to the interaction of all of the factors that are addressed.
• This standard applies to general thermal comfort conditions and excludes local discomforts such as
draft, vertical thermal stratification, and radiant asymmetry.
• This standard specifies thermal environmental conditions acceptable for healthy people at atmospheric
pressure equivalent to altitudes up to 3000 m in indoor spaces designed for human occupancy for
periods not less than 15 minutes.
• This standard does not address such non-thermal environmental factors as air quality, acoustics, and
illumination; nor other physical, chemical or biological space contaminants which may affect comfort or
health.
• This standard is intended for use in design of HVAC-systems, design of buildings, evaluation of
existing thermal environments, building ratings or labelling, and testing of HVAC system performance.
• The standard applies exclusively to indoor environments with HVAC systems over which the
occupants have no control. The occupants of such buildings are presumed to have no option to
open/close windows.
5.1.3. Definitions
adaptive model: A linear regression model that relates indoor design temperatures or acceptable
temperature ranges to outdoor meteorological or climatological parameters. Note that the range of
applicable outdoor climates should be restricted to that appearing on the X-axis of the adaptive model’s
graph (i.e. they should not be extrapolated beyond the range of the regression models’ X-variable).
adaptive opportunity: Buildings provide their occupants with varying degrees of adaptive opportunity or
scope to adjust the internal environment (and themselves) to achieve thermal comfort. Sealed, centrally
air-conditioned office buildings with open-plan floor layouts provide minimal adaptive opportunity, while
ASHRAE RP-884 Final Report
Variable Temperature Standard page 157 MRL Australia
naturally ventilated buildings with operable windows and ceiling fans within small single- or dual-occupant
offices typically afford high degrees of adaptive opportunity.
clo: a unit used to express the thermal insulation provided by garments and clothing ensembles, where 1
clo = 0.155 m2 K/W.
comfort, thermal: that condition of mind which expresses satisfaction with the thermal environment; it
requires subjective evaluation. Optimum thermal comfort is assumed to correspond with a thermal
preference vote of “want no change”
environment, thermal: the characteristics of the environment which affect a person’s heat loss.
environment, acceptable thermal: an environment which at least 80% of the occupants would find
thermally acceptable.
humidity, relative (rh): the ratio of the mole fraction of water vapor present in the air to the mole fraction
of water vapor present in saturated air at the same temperature and barometric pressure; alternatively, it
equals the ratio of the partial pressure (or density) of the water vapor in the air to the saturation pressure
(or density) of water vapor at the same temperature.
insulation, chair: incremental thermal insulation of chairs used by building occupants. The typical office
chair’s clo value is ~0.15 clo units. This effect needs to be included in overall thermal insulation estimates
for the PMV model to yield accurate results.
insulation, clothing (Icl): the resistance to sensible heat transfer provided by a clothing ensemble (i.e.,
more than one garment). It is described as the intrinsic insulation from the skin to the clothing surface,
not including the resistance provided by the air layer around the clothed body; it is usually expressed in
clo units. Clothing worn by people indoors is modified to a great extent by the season and outside
weather conditions. During the summer months, typical clothing in commercial establishments consists of
lightweight dresses, lightweight trousers, short or long sleeved shirts and blouses and occasionally a suit
jacket or sweater. These ensembles have clothing insulation values (Icl) ranging from 0.35 to 0.6 clo.
During the winter season, people wear garments constructed of thicker, heavier (ie. warmer) fabrics and
often add more garment layers to an ensemble. A typical indoor winter ensemble would have an Icl value
ranging from 0.8 to 1.2 clo. Where the outside temperature range does not vary a great deal from season
ASHRAE RP-884 Final Report
Variable Temperature Standard page 158 MRL Australia
to season, people do not change the types of garments they wear year round as much as people who
experience extreme hot and cold climates. The (Icl) provided by clothing ensembles can be estimated by
summing the garment Iclu values as described in ASHRAE Standard 55-92 (1992).
insulation, garment (Iclu): the increased resistance to sensible heat transfer obtained from adding an
individual garment over the nude body. It is the effective increase in overall insulation attributable to the
garment and is usually expressed in clo units.
mean air speed (velocity): arithmetic mean of instantaneous air speed measurements within the occupied
zone, integrated over a period of not less than three minutes (m s-1).
mean monthly (or daily) outdoor effective temperature: Arithmetic average of 6am outdoor ET*
(assumed minimum), and 3pm outdoor ET* (assumed maximum) for a calendar month or particular day.
metabolic rate (met): rate of energy production of the body. Metabolism, which varies with activity, is
expressed in met units in this standard. One met is defined as 58.2 Wm-2 which is equal to the energy
produced per unit surface area of a seated person at rest. The surface area of an average person is about
1.8 m2. In today’s society, most people are occupied with light, primarily a sedentary activity level
corresponding to 1 to 1.6 met. Metabolic activity should be assessed for a period between 30 and 60
minutes before any thermal assessment is made. For more detailed values see ASHRAE Standard 55-
1992, ISO 7730, ISO 8996 or the ASHRAE Handbook of Fundamentals (1993).
neutrality, thermal: the indoor thermal index value (usually operative temperature) corresponding with a
maximum number of building occupants voting “neutral” on the thermal sensation scale.
preference, thermal: a conscious desire for change in one’s thermal state, commonly graded into the
categories, 1“want cooler,” 2 “want no change,” 3“want warmer”; it requires subjective evaluation.
Preferred temperature, i.e. that corresponding with a maximum number of “2” votes, does not necessarily
correspond with thermal neutrality.
PMV: Predicted Mean Vote is a thermal index derived from the heat-balance model of thermal comfort
developed by Fanger (1970). PMV predicts the mean thermal sensation of a large group of subjects
experiencing a thermal environment specified in terms of mean air and radiant temperatures, air speed,
humidity, thermal insulation and metabolic rate.
ASHRAE RP-884 Final Report
Variable Temperature Standard page 159 MRL Australia
PMV, analytic: Predicted Mean Vote index calculated analytically from mean measurements or estimates
of the six primary comfort parameters: mean air and radiant temperatures, mean air speed, humidity,
clothing (+ chair) thermal insulation and metabolic rate.
PMV, adaptive: the RP-884 adaptive regression model that predicts optimum thermal comfort
temperature (thermal sensation corrected for semantics). The name “adaptive PMV” is used for the
model because it predicts essentially the same optimum operative temperature answer as the analytic
PMV approach, but uses mean outdoor effective temperature as the only input instead of the usual four
inputs (clo, met, rh and v) required by the analytic PMV method.
sensation, thermal: a conscious feeling commonly graded into the categories, -3 cold, -2 cool, -1 slightly
cool, 0 neutral, +1 slightly warm, +2 warm, and +3 hot; it requires subjective evaluation. An individual’s
ideal thermal comfort does not necessarily correspond with a thermal sensation vote of “neutral” (zero).
summer: operationally defined as the cooling season; climatologically defined for the purposes of this
standard as having a mean daily outdoor effective temperature of 25oC.
temperature, air (ta): the dry-bulb temperature of the air surrounding the occupant.
temperature, dew point (tdp): [or ambient water vapor pressure (Pa)], the temperature at which moist air
becomes saturated (100% relative humidity) with water vapor (Psdp = Pa) when cooled at constant
pressure.
temperature, mean radiant (tr): the uniform surface temperature of an imaginary black enclosure in which
an occupant would exchange the same amount of radiant heat as in the actual nonuniform space.
temperature, operative (to): the uniform temperature of an imaginary black enclosure in which an
occupant would exchange the same amount of heat by radiation plus convection as in the actual non-
uniform environment. Operative temperature is numerically the average of the air temperature (ta) and
mean radiant temperature (tr), weighted by their respective heat transfer coefficients (hc and hr):
th t h t
h ho
c a r r
c r
= ++
( )( )
which typically equates to the arithmetic average of mean air and radiant temperatures.
ASHRAE RP-884 Final Report
Variable Temperature Standard page 160 MRL Australia
temperature, effective (ET*): the operative temperature (to) of an enclosure at 50% relative humidity
which would cause the same sensible plus latent heat exchange from a person as would the actual
environment.
temperature, optimum operative: the operative temperature that satisfies the greatest possible number of
people at a given clothing and activity level. Due to the semantic offset between preferred and neutral
temperatures, optimum operative temperature in this standard does not necessarily correspond exactly
with thermal neutrality (i.e. optimum temperature is neutrality after correction for semantic offset).
temperature, thermodynamic wet bulb: (also called the Adiabatic Saturation Temperature), that
temperature at which water, by evaporating into air, can bring the air to saturation adiabatically at the
same temperature. The wet bulb temperature measured with an appropriate psychrometer can approach
the thermodynamic wet bulb temperature.
winter: operationally defined as the heating season; climatologically, for the purposes of this standard a
typical winter condition is assumed to have a mean daily outdoor effective temperature of 0oC.
zone, occupied: the region normally occupied by people within a space, generally considered to be
between the floor and 1.8 m above the floor and more than 0.6 m from walls or fixed air conditioning
equipment.
5.1.4. Conditions for an acceptable thermal environment.
The conditions for an acceptable thermal environment shall be based on one of the following three
techniques, listed in descending order of preference:
• the analytic PMV method, as described in ISO 7730 (1994) , if mean clothing and metabolic rates
are known in advance, or
• the adaptive PMV method in which indoor optimum operative temperature is predicted from a
knowledge of outdoor effective temperature using RP-884 regression models, or
• the prescriptive method in which summer and/or winter comfort zones for either 90% or 80% thermal
acceptability levels are selected from the RP-884 psychrometric charts.
ASHRAE RP-884 Final Report
Variable Temperature Standard page 161 MRL Australia
5.1.4.1 Analytic PMV Method
See the detailed procedures for estimation of the optimum temperature for a group of building occupants
described in ISO 7730 (1994). The only departure from the methods described there is the inclusion of
the incremental thermal insulation of the chair into the seated occupants’ overall thermal insulation.
Optimum operative temperature may be predicted by inputting measured or estimated values of insulation
(clothing + chair), metabolic rate, relative humidity, air speed and solving for the unknown operative
temperature by setting PMV = zero. Note that the actual group mean thermal sensation expressed by
building occupants under the optimum operative temperature predicted by this method may not
necessarily equal zero (“neutral”). This is due to the semantic offset between group thermal neutrality and
preference. Therefore PMV equal to zero may correspond with a non-zero mean thermal sensation for
the group of building occupants in question, but they will still be in their optimum operative temperature.
5.1.4.2. Adaptive PMV method
In HVAC situations where the mean thermal insulation (clothing and chairs) and mean air speed cannot be
observed or accurately anticipated, the adaptive PMV method may be applied. Weather data in the form
of mean outdoor effective temperature for the relevant time of year is required. In the absence of current
meteorological observation, published
mean climatological data for the relevant month from the nearest or most relevant weather station may
suffice.
ASHRAE RP-884 Final Report
Variable Temperature Standard page 162 MRL Australia
Figure 5.1: The adaptive PMV comfort zone’s optimum and limits for an 80% acceptability level in HVAC premises.
18
20
22
24
26
28
-5 0 5 10 15 20 25 30 35
mean daily outdoor effective temperature (oC)
com
fort
tem
pera
ture
(oC
)
comfort temp. in HVAC = 22.6 + 0.04 * outdoor ET*
90% acceptability upper limit
90% acceptability lower limit
Figure 5.2: The adaptive PMV comfort zone’s optimum and limits for an 90% acceptability level in HVAC premises.
ASHRAE RP-884 Final Report
Variable Temperature Standard page 163 MRL Australia
5.1.4.3. Prescriptive method
Where outdoor meteorological or climatological data are unavailable, the RP-884 prescriptive method
may be used to define acceptable ranges of temperatures. The prescriptions are designed to provide
environments in which minimum levels of thermal acceptability (based on general thermal comfort) can be
selected as either 90% or 80%.
0
5
10
15
15 20 25 30
OPERATIVE TEMPERATURE (oC)
HU
MID
ITY
MIX
ING
RA
TIO
(g
/kg
)
30% rh
19o C Wet Bulb
18o C Wet Bulb
100% rh
70% rh60% rh
50% rh
Winter
Summer
24.7 ET*21.3 ET*
0
5
10
15
15 20 25 30
OPERATIVE TEMPERATURE (oC)
HU
MID
ITY
MIX
ING
RA
TIO
(g
/kg
)
30% rh
19 o C Wet Bulb18 o C Wet Bulb
100% rh
70% rh60% rh
50% rh
Winter
Summer
25.5 ET*20.5 ET*
Figure 5.3: Psychrometric charts showing summer and winter comfort zone prescriptions for 90% acceptability (left panel) and 80% acceptability (right panel)
Operative Temperature. The operative temperature range between which, theoretically, no more than
20% of occupants during light, primarily sedentary activity (� 1.2. met), assuming they wear the same
level of clothing insulation, will find the environment thermally unacceptable is given in Table 5.1. The
acceptable range of operative temperatures and humidities for winter and summer is further defined on the
psychrometric chart of Figure 5.3. The comfort zones are:
a) Winter: to = 20.5oC to 24.5oC at 50% rh for 80% acceptability level.
to = 21.3oC to 23.7oC at 50% rh for 90% acceptability level.
The slanting side boundaries of the winter zones in Figure 5.3 are defined in terms of effective
temperature (ET*) lines and are loci of constant thermal sensations.
b) Summer: to = 21.5oC to 25.5oC at 50% rh for 80% acceptability level.
to = 22.3oC to 24.7oC at 50% rh for 90% acceptability level.
The slanting side boundaries of the summer zones in Figure 5.3 are defined in terms of effective
temperature (ET*) lines and are loci of constant thermal sensations.
ASHRAE RP-884 Final Report
Variable Temperature Standard page 164 MRL Australia
The winter and summer comfort zones overlap in the 22oC to 23oC range. In this region people in
summer dress would tend to approach slightly cool sensation while those in winter clothing would be near
the slightly warm sensation. In reality, the boundaries of each zone are not as sharp as depicted in Figure
5.3 due to inter-individual clothing and activity differences.
Table 5.1: Optimum and acceptable ranges of operative temperature for persons engaged in light, primarily sedentary activity (� 1.2 mets) at 50% relative humidity and mean air speed � 0.15 ms-1. For use in buildings with central HVAC systems.
Description of Icl Operative Temperature Season typical thermal insulation clo optimum
temperature range (90% accept.)
range (80% accept.)
Winter
heavy slacks, long sleeve shirt, sweater and office chair
1.05
22.5 oC
21.3 - 23.7 oC
20.5 - 24.5 oC
Summer
light slacks, short sleeve shirt and office chair
0.65
23.5 oC
22.3 - 24.7 oC
21.5 - 25.5 oC
For infants, certain elderly persons, and individuals who are physically disabled, the lower limits of Table 5.1 should be avoided.
ASHRAE RP-884 Final Report
Variable Temperature Standard page 165 MRL Australia
5.2. A variable temperature standard for application in naturally ventilated buildings
5.2.1. Purpose
To specify the thermal environmental conditions that will be acceptable to a majority of the occupants
within naturally ventilated spaces.
5.2.2 Scope
• The environmental factors addressed are temperature, thermal radiation, humidity.
• It is intended that all of the criteria in this standard be applied together, since comfort in the space
environment is complex and responds to the interaction of all of the factors that are addressed.
• This standard applies to general thermal comfort conditions and excludes local discomforts such as
draft, vertical thermal stratification, and radiant asymmetry.
• This standard specifies thermal environmental conditions acceptable for healthy people at atmospheric
pressure equivalent to altitudes up to 3000 m in indoor spaces designed for human occupancy for
periods not less than 15 minutes.
• This standard does not address such non-thermal environmental factors as air quality, acoustics, and
illumination; nor other physical, chemical or biological space contaminants which may affect comfort or
health.
• This standard is intended for use in design of naturally ventilated buildings and evaluation of existing
thermal environments within such buildings.
• The standard applies exclusively to indoor environments without centralised HVAC systems. Such
buildings are presumed to have operable windows which the occupants have some degree of control
over. They may have some form of heating installed, but it would be controlled by the building
occupants, either individually or in small groups.
• The standard cannot be used to decide when and where to install centralised air-conditioning. While it
may provide useful information in relation to such decisions, the standard cannot be regarded as the
ASHRAE RP-884 Final Report
Variable Temperature Standard page 166 MRL Australia
sole criterion. For example, the adaptive opportunity afforded the occupants of naturally ventilated
buildings should also be borne in mind.
5.2.3. Definitions
adaptive model: A linear regression model that relates indoor design temperatures or acceptable
temperature ranges to outdoor meteorological or climatological parameters. Note that the range of
applicable outdoor climates should be restricted to that appearing on the X-axis of the adaptive model’s
graph (i.e. they should not be extrapolated beyond the range of the regression models’ X-variable).
adaptive opportunity: Buildings provide their occupants with varying degrees of adaptive opportunity or
scope to adjust the internal environment (and themselves) to achieve thermal comfort. Sealed, centrally
air-conditioned office buildings with open-plan floor layouts provide minimal adaptive opportunity, while
naturally ventilated buildings with operable windows and ceiling fans within small single- or dual-occupant
offices typically afford high degrees of adaptive opportunity.
comfort, thermal: that condition of mind which expresses satisfaction with the thermal environment; it
requires subjective evaluation. Optimum thermal comfort is assumed to correspond with a thermal
preference vote of “want no change”
environment, thermal: the characteristics of the environment which affect a person’s heat loss.
environment, acceptable thermal: an environment which at least 80% of the occupants would find
thermally acceptable.
humidity, relative (rh): the ratio of the mole fraction of water vapor present in the air to the mole fraction
of water vapor present in saturated air at the same temperature and barometric pressure; alternatively, it
equals the ratio of the partial pressure (or density) of the water vapor in the air to the saturation pressure
(or density) of water vapor at the same temperature.
mean monthly (or daily) outdoor effective temperature: Arithmetic average of 6am outdoor ET*
(assumed minimum), and 3pm outdoor ET* (assumed maximum) for a calendar month or particular day.
ASHRAE RP-884 Final Report
Variable Temperature Standard page 167 MRL Australia
naturally ventilated: Those premises in which a centralised heating, ventilation and air-conditioning
systems are absent and windows are operable. Some form of heating may be present, but it would
normally be under the control of building occupants, either individually or in small groups.
neutrality, thermal: the indoor thermal index value (usually operative temperature) corresponding with a
maximum number of building occupants voting “neutral” on the thermal sensation scale.
preference, thermal: a conscious desire for change in one’s thermal state, commonly graded into the
categories, 1“want cooler,” 2 “want no change,” 3“want warmer”; it requires subjective evaluation.
Preferred temperature, i.e. that corresponding with a maximum number of “2” votes, corresponds
reasonably well with thermal neutrality in naturally ventilated buildings.
sensation, thermal: a conscious feeling commonly graded into the categories, -3 cold, -2 cool, -1 slightly
cool, 0 neutral, +1 slightly warm, +2 warm, and +3 hot; it requires subjective evaluation. Optimum
thermal comfort corresponds reasonably well with a thermal sensation vote of “neutral” in naturally
ventilated buildings.
temperature, air (ta): the dry-bulb temperature of the air surrounding the occupant.
temperature, mean radiant (tr): the uniform surface temperature of an imaginary black enclosure in which
an occupant would exchange the same amount of radiant heat as in the actual nonuniform space.
ASHRAE RP-884 Final Report
Variable Temperature Standard page 168 MRL Australia
temperature, operative (to): the uniform temperature of an imaginary black enclosure in which an
occupant would exchange the same amount of heat by radiation plus convection as in the actual non-
uniform environment. Operative temperature is numerically the average of the air temperature (ta) and
mean radiant temperature (tr), weighted by their respective heat transfer coefficients (hc and hr):
th t h t
h ho
c a r r
c r
= ++
( )( )
which typically equates to the arithmetic average of mean air and radiant temperatures
temperature, effective (ET*): the operative temperature (to) of an enclosure at 50% relative humidity
which would cause the same sensible plus latent heat exchange from a person as would the actual
environment.
temperature, optimum operative: the operative temperature that satisfies the greatest possible number of
people at a given clothing and activity level. Optimum operative temperature in this standard corresponds
reasonably well with both thermal neutrality and preferred temperature.
zone, occupied: the region normally occupied by people within a space, generally considered to be
between the floor and 1.8 m above the floor and more than 0.6 m from walls or fixed air conditioning
equipment.
5.2.4. Conditions for an acceptable thermal environment.
The conditions for an acceptable thermal environment shall be based exclusively on the adaptive model
(linear regression) approach. The PMV/PPD model is inapplicable to naturally ventilated premises
because it only partially accounts for processes of thermal adaptation to indoor climate. The prescription
of summer and winter comfort zones is inappropriate for this standard because the steep gradient on the
naturally ventilated adaptive model would render climatological definitions of universal “summer” and
“winter” conditions misleading.
The adaptive models in this section can be applied where weather data in the form of mean outdoor
effective temperature for the relevant time of year are available. These need to be calculated from basic
outdoor air temperature maxima (3 pm) and minima
ASHRAE RP-884 Final Report
Variable Temperature Standard page 169 MRL Australia
(6 am), along with coincident humidity. In the absence of current meteorological observations, published
mean climatological data for the relevant month from the nearest weather station may suffice.
Indoor climatic instrumentation was recorded by a portable datalogger. Relative humidity
and air temperature were monitored by a Hanna Instruments probe. This consisted of a
polished aluminium sheath 19mm in diameter, containing in its ventilated tip a humidity
sensor (solid-state hygrometer) and a thermistor. The instrumentation measured air
temperature, globe temperature and humidity. The globe thermometer had a 38mm
diameter ping pong ball with appropriate emissivity attached over the sensor. All variables
were measured at subjects’ waist height.
Questionnaire
The questionnaire addressed conditions at time of physical measurements. Time lapse
between instrument measurements and questionnaire response was never more than 10
minutes. Comfort was rated using the 7-pt semantic differential based on Bedford. Thermal
preference was rated on a want to be warmer/cooler descriptive scale and thermal
acceptability questions were not considered. Other thermal environmental parameters
included were air movement, draft and skin moisture. Metabolic activity was based on a
descriptive scale and noted at the time the questionnaire was being carried out. Total
ASHRAE RP-884 Final Report
Appendix C page 261 MRL Australia
clothing ensemble insulation experienced by the subject was estimated using the ISO 7730
checklist and work of McCullough (eg 1985) and others.
Outdoor meteorological data
Daily outdoor maximum and minimum temperatures were obtained for a number of the
centres from the Pakistan Meteorological office for July and December 1993 and January
1994. Where temperatures were not provided they were replaced with climatological data
(monthly means) from the International Station Meteorological and Climate Summary Vol.2
CDROM (ISMCS, 1992). All outdoor humidities were also obtained from this source and
had to be derived from mean dewpoint temperature and mean temperature minima and
maxima.
RP-884 standardization assumptions
The Bedford 7-point thermal comfort scale was mapped directly to the ASHRAE 7-point
thermal sensation scale for RP-884 purposes. The data was presented as subjects in
individual houses, with studies conducted in summer and winter, so the project was of
longitudinal research design. For the purpose of this study all houses in the same city were
considered to be identical buildings, thus it was assumed there was a number of subjects
from one building for each city and the subjects were independent between both the summer
and winter studies. Some indices in the original data set had to be re-defined to conform to
RP-884 standards. Clo was estimated by ISO 7730 (1984) and corrected to the ASHRAE
55-92 Standard using the regression models developed within RP-884. The activity variable
in the original data set was used such that if activity was <= 4 then 0.15 clo was added to the
total clothing ensemble to form another variable (insul), that accounted for the additional
insulation provided by a chair for subjects that were seated. Velocity measurements in the
raw data file indicated a systematic bias that was time-dependent. The original data in all
summer files was found to be less affected and so original data were used. In the winter
files, values >1.5 m/s were replaced with an average. The Multan, Winter field experiment
was omitted from the RP-884 database.
ASHRAE RP-884 Final Report
Appendix C page 262 MRL Australia
C.10. Project Title - Comfort criteria for passively cooled buildings. a PASCOOL
task.
Project researchers and class of investigation
N. Baker and M. Standeven, The Martin Centre for Architecture and Urban Studies,
University of Cambridge, UK. This is a CLASS-2 field experiment.
Project file names in the RP-884 database
This project is disseminated as file number 27 (summer - NV) in the RP-884 database.
Project publications
Baker, N and M. Standeven. (1995) “A Behavioural Approach to Thermal Comfort
Assessment in Naturally Ventilated Buildings”. Proceedings from CIBSE National
Conference, Ch 76-84.
Baker, N. and M. Standeven. (1994) Comfort criteria for passively cooled buildings. A
PASCOOL task. Renewable Energy. V 5. n 5-8 Aug 1994. p 977-984.
Project location, climate and season
This field experiment was carried out in Athens, Greece for the summer season.
Athens has a Mediterranean climate.
Sample buildings
Building Code (blcode)
Sample Size (n)
Climate Controls (bldgtype)
Floor Area
Occupancy Pattern
1 409 NV residential building 2 276 NV residential building 3 443 NV residential building 4 176 NV residential building 5 187 NV residential building 6 135 NV residential building
Instruments
Indoor room climate instrumentation included: a thermistor to measure air temperature, an
omnidirectional hot-wire sensor to measure air speed, a solid-state hygrometer to measure
humidity and a globe thermometer with 38mm diameter ping pong ball to measure globe
temperature.
ASHRAE RP-884 Final Report
Appendix C page 263 MRL Australia
Local climate instrumentation consisted of: a calibrated sensor array comprising air
temperature thermistor, omnidirectional thermistor anemometer and two hemispherical
globe thermometers, mounted on a headset similar to that of a walkman. Data was logged
on a portable logger allowing complete thermal histories to be recorded for the day,
including time when the subject was away from the room.
The local data (headsets) were attached ot questionnaire responses in the RP-884
database file for this PASCOOL project. In cases where local data were unsuitable or
unavailable, room data were substituted.
Questionnaire
The questionnaire addressed the conditions at the time physical measurements were being
taken. Sensation was rated on the ASHRAE 7-pt scale. Questions of thermal acceptability
and thermal preference where both considered and metabolic ratings were taken. Clothing
insulation was estimated using the ISO 7730 checklist. Adaptive behaviour questions
regarding changes in clothing and adjustment to controls such as opening or closing shades,
blinds or windows and relocations within the room were recorded.
Outdoor meteorological data
Outdoor Meteorological air temperature data was recorded simultaneously with indoor
measurement made. For the purposes of RP-884 outdoor temperatures at 600 hrs and
1500 hrs were extracted. Humidities at 600 hrs and 1500 hrs were obtained from the
International Station Meteorological and Climate Summary (ISMCS, 1992) CDROM.
RP-884 standardization assumptions
This project was of longitudinal research design, but for the purposes of RP-884 subjects
were assumed to be independent (ie. cross-sectional). Clothing insulation was estimated
using the ISO 7730 (1984) Standard, it was therefore necessary to adjust clo to conform to
the ASHRAE 55-92 Standard. Also where the metabolic rate was <= 2 met it was assumed
the subject was seated and so 0.15 clo was added to the total clothing ensemble in these
cases to account for the insulation provided by a chair. The 5-pt variable PRF_VOTE in the
original data was re-coded to our 3-pt McIntrye (MCI) scale. Where air velocity was missing
0.1 m/s was temporarily inserted for the software based index calculation and then removed
from the database.
ASHRAE RP-884 Final Report
Appendix C page 264 MRL Australia
ASHRAE RP-884 Final Report
Appendix C page 265 MRL Australia
C.11. Project Title - Developing indoor temperatures for naturally ventilated
buildings.
Project researchers and class of investigation
I. A. Raja, J. F. Nicol and M. A. Humphreys (Oxford-Brookes University, UK). This is a
CLASS-3 investigation.
Project publications
Nicol, J. F., M. A. Humphreys and I. A. Raja (1995). “Developing Indoor Temperatures for
Naturally Ventilated Buildings”. Proceeding for CIBSE National Conference.
Also see the Full Report.
Project file names in the RP-884 database
This project is disseminated as file number 28 (summer - NV) in the RP-884 database.
Project location, climate and season
The project is located in Oxford, South Britain about 63m above sea level and is situated at
51o 46’ North and 1o 16’ West. The climate of Oxford is typical of the low lying part of the
English midlands and is also influenced by its proximity to the Atlantic. Oxford experiences
one of the warmer maxima in the surrounding area with a mean maximum temperature of
21.7°C in July. The mean minima of 1.3°C in January and February reflects weather similar
to that of the midlands and south-east. This field experiment was completed in the summer
months of August and September and comes under the climate classification of west coast
marine.
Sample buildings
Building Code (blcode)
Sample Size (n)
Climate Controls
(bldgtype)
Floor Area
Occupancy Pattern
1 496 NV School of Architecture plus Biological and Molecular Sciences
2 334 NV Headington Hill Hall.
3 47 NV Tonge building.
ASHRAE RP-884 Final Report
Appendix C page 266 MRL Australia
Instruments
Air temperature was measured using a thermistor. An adapted thermistor probe with a
38mm diameter ping pong ball of suitable emissivity attached, was used to measure globe
temperature. Air speed was registered using an omnidirectional sensor and a solid-state
hygrometer was used to measure humidity. All measurements were taken at waist
(generally desk) height.
Questionnaire
A comfort rating on the 7pt Bedford scale was addressed in the questionnaire as well as
thermal preference. Thermal acceptability was not recorded. Metabolic ratings were taken
at the time the questionnaire was being answered, but covered the 15 minute period before
the questionnaire was completed. Clothing insulation estimates were based on the ISO
7730 checklist with the insulation effects of chair included in the total clothing ensemble of
the subject. Questions of adaptive behaviour and perceived control on a subjects thermal
environment were addressed. Specifically, whether doors, window and curtains or blinds
could be opened and closed as well as the influence of fans and heater that could be
switched on or off.
Outdoor meteorological data
Outdoor Meteorological data was obtained for every 0.25 hours from the Oxford University
Radcliffe Observatory by the original researchers. From this, air temperatures and relative
humidities at 600 hours and 1500 hours were extracted for the purposes of RP-884.
RP-884 standardization assumptions
The research design of this project was longitudinal, but for RP-884 purposes all subjects
were assumed to be independent (ie. cross-sectional). The Bedford 7-point thermal comfort
scale was mapped directly to the ASHRAE 7-point thermal sensation scale for RP-884
purposes. Clothing insulation, estimated using the ISO 7730 (1984) Standard was
corrected to the ASHRAE 55-92 Standard via regression models developed within RP-884.
Allowance for the insulation provided by a chair was incorporated into the total clothing
ensemble by the original researchers only when the subjects reported themselves as seated.
This provided the RP-884 insul variable. To obtain clothing insulation (clo) in isolation, 0.15
clo was subtracted. All rows with missing air temperature were deleted, but where velocity
ASHRAE RP-884 Final Report
Appendix C page 267 MRL Australia
was missing, 0.1 m/s was temporarily substituted and where indoor relative humidity and
metabolic rate were missing, 50% and 1 met respectively were temporarily substituted for
the purposes of index calculations and then removed from the database.
ASHRAE RP-884 Final Report
Appendix C page 268 MRL Australia
C.12. Project Title - Mixed mode climate control: some hands-on experience.
Project researchers and class of investigation
David Rowe. Department of Architectural and Design Science, Sydney University, Australia.
This is a CLASS-2 investigation
Project file names in the RP-884 database
This project is disseminated as file numbers 29 (Summer - Mixed Mode), 30 (winter - Mixed
Mode) and 31 (winter - HVAC) in the RP-884 database.
Project publications
Nothing published yet.
Project location, climate and season
The field experiment was conducted in Sydney, the capital of the state of New South Wales
in Australia. Sydney’s climate is humid and sub-tropical. The project conducted in both
summer and winter seasons.
Sample buildings
Building Code (blcode)
Sample Size (n)
and season
Climate Controls (bldgtype)
Floor Area
Occupancy Pattern
1 137 - summer 170 - winter
Mixed (hybrid) university offices
2 83 - winter HVAC administration offices
Instruments
RTD devices were used to measure air temperature. No globe temperatures were
measured but mean radiant temperature was provided based on the average of six
orthogonal plane radiant temperatures, areally weighted for the projection area factors of the
human body. Air speed was assessed using an omnidirectional sensor and included
turbulence intensity measurements (> 10Hz). A chilled-mirror dewpoint sensor was used to
measure humidity. All measurements were taken at a single height.
Questionnaire
ASHRAE RP-884 Final Report
Appendix C page 269 MRL Australia
The questionnaire for this project was based directly on that used for the ASHRAE RP-702
Hot Humid Field Experiment in Townsville Australia (see above for de Dear et al., 1994).
Thermal sensation rated on the 7-pt ASHRAE scale was recorded at the time physical
measurements were being taken, along with the other items on the questionnaire that follow.
Thermal acceptability and thermal preference was addressed. Metabolic ratings at the time
of and one hour before the questionnaire were recorded. The total clothing ensemble
insulation was estimated using the ASHRAE 55-92 checklist. Other thermal environmental
parameters considered include air movement.
Outdoor meteorological data
Outdoor Meteorological Data consisting of air temperature and relative humidity at 600
hours and 1500 hours was obtained for this field experiment from Macquarie University’s
Meteorological site, Sydney, Australia.
RP-884 standardization assumptions
The research design for this project was longitudinal and for the purpose of RP-884 all
subjects were assumed to be independent (ie. cross-sectional). Clothing insulation was
estimated from ASHRAE 55-92 checklists so no alterations were necessary apart from the
addition of 0.15 clo to account for the insulation effects of a chair in creating our insul
variable. Throughout the field experiment where mean radiant temperature was not provided
air temperature was entered as a substitute.
ASHRAE RP-884 Final Report
Appendix C page 270 MRL Australia
C.13. Project Title - ASHRAE sponsored RP-462. San Francisco area.
Project researchers and class of investigation
Gail Schiller, Edward Arens, Fred Bauman, Charles Benton, Marc Fountain and Tammy
Doherty (CEDR at University of California, Berkeley). This is a CLASS-1 field experiment
Project file names in the RP-884 database
This project is disseminated as file numbers 32 (summer - HVAC), 33 (summer - NV), 34
(winter - HVAC) and 35 (winter - NV) in the RP-884 database.
Project publications
Schiller, G. E., E. Arens, F. Bauman, C. Benton, M Fountain and T. Doherty. (1988) A Field
Study of Thermal Environments and Comfort in Office Buildings: Final Report--ASHRAE
462. (CEDR:UC Berkeley).
Schiller, G. E. (1990) A comparison of measured and predicted comfort in office buildings.
ASHRAE Transactions, 96(1).
Project location, climate and season
RP-462 was conducted over five locations within the San Francisco Bay area including
Berkeley, San Ramon, Palo Alto, San Francisco and walnut Creek. All five cities are within a
Mediterranean climate zone, but all have different local climates due to their location around
the San Francisco Bay area. San Francisco is located right on the coast, but also very close
to the Bay. Palo Alto is situated further from the coast close to southern end of the Bay and
behind the Santa Cruz Mountains. Berkeley is located across the Bay from the Golden Gate
and Walnut Creek is further inland almost directly east of Berkeley. San Ramon is a similar
but shorter distance from the Bay as Walnut Creek, but instead it is almost directly east of
San Francisco. The field experiments were conducted across both summer and winter
seasons.
ASHRAE RP-884 Final Report
Appendix C page 271 MRL Australia
Sample buildings
Location Buildg Code
(blcode)
Sample Size
(n) and Season*
Climate Controls
(bldgtype)
Floor Area
Occupancy Pattern
Berkeley 1 122 - S 121 - W
NV 236,600 ft2. crowed open plan offices.
San Ramon 2 119 - S 123 - W
HVAC - thermal ice storage and evap. ponds.
2,000,000 ft2. office building
Palo Alto 3 92 - S 101 - W
HVAC (multizone HVAC with EMS)
187,000 ft2. mostly private offices.
San Francisco
4 108 - S 134 - W
HVAC - heat pump mech. system.
191,000 ft2. open plan with private balconies on perimeter.
San Francisco
5 115 - S 132 - W
roof-mounted HV unit, no mech. a.c.,
54,000 ft2. open plan converted factory.
San Francisco
6 123 - S 136 - W
NV 90,000 ft2. open plan and private offices.
San Francisco
7 107 - S 122 - W
HVAC - thermal ice storage, VAV perimeter reheat.
265,000 ft2. open plan and private offices.
San Francisco
8 117 - S 147 - W
HVAC 634,000 ft2. large open plan.
Walnut Creek
9 23 - S 145 - W
HVAC 316,400 ft2. open plan and private offices.
Walnut Creek
10 107 - S 146 - W
HVAC 368,000 ft2. open plan with partitions and private offices.
* S = summer, W = winter in the sample size and season column.
Instruments
Air temperature, air velocity, humidity, and globe temperatures were measured using a
mobile cart at the heights indicated below, with the exception of the one stationary
observation point. Air temperature was measured with a shielded platinum RTD at 0.6m
and shielded type T thermocouples at 0.1m, 0.6m and 1.1m were used. Air velocity was
measured by an elliptical omnidirectional constant temperature anemometer at 0.6m and
spherical omnidirectional temperature compensated anemometer at 0.1m and 1.1m.
Humidity was measured by a chilled-mirror dew point sensor at 0.6m. Globe temperatures
were measured by a type T thermocouple inside a 38 mm diameter table tennis ball (painted
ASHRAE RP-884 Final Report
Appendix C page 272 MRL Australia
grey) at heights of 0.1m, 0.6m and 1.1m on the mobile cart and at 1.1m in the stationary set
up. Other variables measured not of relevance to RP-884 include radiant temperature
asymmetry, surface temperature and illumination.
Questionnaire
Questionnaire responses were collected at the time physical measurements were being
taken. The ASHRAE 7-pt scale was used to determine thermal sensation. The McIntyre
scale was used to assess thermal preference. Thermal acceptability was not addressed.
Metabolic rating and clothing insulation estimates were based on checklists in ASHRAE
Standard 55-81 (1981). The background section of the survey (not necessarily completed
when physical measurements were being made) covered general descriptions of office work
areas; degree of satisfaction with components of their work environment; personal and
comparative comfort and personal subject related information.
Outdoor meteorological data
Outdoor Meteorological air temperature minima and maxima were purchased from the US
National Climate Data Center (NCDC) for sites considered of similar climatic situations to
the study locations. Where a suitable site could not be requisitioned, climatological data was
extracted from the International Station Meteorological and Climate Summary (ISMCS,
1992) CDROM. All climatological humidity data were also obtained from ISMCS (1992).
RP-884 standardization assumptions
RP-884 is the fourth ASHRAE sponsored project in the series RP-462, RP-702 and RP-
821. A lot of the assumptions and standards of RP-462 project have formed the basis for
the later projects including RP-884, thus limited standardisation has been necessary here.
Clothing insulation was converted from ASHRAE 55-81 to the 55-92 standard. 0.15 clo was
added to the total clothing ensemble for the insulation effects of a chair to create our insul
variable. The research design of this project was part longitudinal and part cross-sectional,
but for RP-884 purposes all subjects were assumed to be independent.
ASHRAE RP-884 Final Report
Appendix C page 273 MRL Australia
C.14. Project Title - A field investigation of thermal comfort environmental
satisfaction and perceived control levels in UK office buildings, University of
Liverpool.
Project file names in the RP-884 database
This project is disseminated as file numbers 38 (summer -NV), 39 (winter - NV) and 40
(winter - Mixed Mode) in the RP-884 database.
Project researchers and class of investigation
Ruth N. Williams (The Building Services Research and Information Association, Berkshire,
UK). This is a CLASS-2 investigation
Project publications
Williams, R. N. (1995). A field investigation of thermal comfort environmental satisfaction
and perceived control levels in UK office buildings. Healthy Buildings. Vol. 3 pp. 1181-1186.
Williams, R (1996) “Predicting environmental dissatisfaction in UK offices,
“CIBSE/ASHRAE Joint National Conference, Harrogate UK, VII., pp.167-178.
Project location, climate and season
This project was conducted across three towns/cities in the UK, including Liverpool, St
Helens and Chester. All three come under the west coast marine climate classification. The
study was carried out in summer and winter months.
Instruments
Air temperature was measured using thermistors and an omnidirectional hot bead sensor to
measure air speed. A Envirlog supplied sensor (type unknown) was used to measure
humidity and by attaching 38mm diameter ping pong balls globe temperature was also
measured. Air Speed and humidity were measured at waist height. Air temperature and
globe temperature were measured at all three heights (ankle, waist and head), but provided
to the RP-884 database as a single average.
ASHRAE RP-884 Final Report
Appendix C page 274 MRL Australia
Sample buildings
Location Building Code
(blcode)
Sample Size (n)
and season
Climate Controls
(bldgtype)
Floor Area
Occupancy Pattern
Liverpool 1 19 - summer NV office buildings A&B
St Helens 2 8 - summer NV LC
St Helens 3 140 - summer 31 - winter
NV WH
St Helens 4 121 - winter Mixed (hybrid)
NWB
Chester 5 44 - winter NV CCH
Chester 6 31 - winter NV COM
Chester 7 67 - winter NV ANN
Liverpool 8 36 - winter NV SEN
Questionnaire
The questionnaire addressed both conditions at the time of physical measurements and
typical overall conditions. Thermal sensation was rated using a 7-pt ASHRAE scale.
Thermal comfort was rated using the 7-pt Bedford scale. Thermal acceptability was
addressed but not thermal preference. Metabolic rating was dealt with by asking if the
subject was sitting or standing during most of their work time, from which an estimate was
derived. Clothing insulation estimates were based on the ISO 7730 (1994) checklist with
corrections for the insulation from a chair included. Adaptive behaviour questions of the
subjects perceived control on temperature, humidity, freshness, smell, appearance, lighting,
noise and layout within their working environment was noted.
Outdoor meteorological data
Outdoor Climatological air temperature data at 600 hours and 1500 hours was obtained
from Weather (the journal, for site - Ringway). Relative humidity at 600 hours and 1500 hours
was obtained from the International Station Meteorological and Climate Summary (site -
Liverpool) CDROM.
RP-884 standardization assumptions
The research design of this study was cross-sectional which satisfies the assumption of
independence between subjects for RP-884. Coding conventions for some variables was
ASHRAE RP-884 Final Report
Appendix C page 275 MRL Australia
altered to conform to RP-884 definitions. Clothing insulation estimated using ISO 7730
(1984) checklists, was corrected to follow the ASHRAE 55-92 Standard. The sex (gender) of
subjects was not indicated in the study so an average of the adjusted clo to the ASHRAE 55-
92 Standard for males and female was used in all cases. 0.15 clo was then subtracted from
this corrected clothing plus chair insulation to create our clo variable.
ASHRAE RP-884 Final Report
Appendix C page 276 MRL Australia
C.15. Project Title - Thermal comfort in the humid tropics: field experiments in air
conditioned and naturally ventilated buildings in Singapore.
Project researchers and class of investigation
R. J. de Dear, K. G. Leow and S. C. Foo (National University of Singapore). This is a
CLASS-2 field experiment.
Project file names in the RP-884 database
This project is disseminated as file numbers 41 (summer - HVAC) and 42 (summer -NV) in
the RP-884 database.
Project publications
de Dear, R. J., Leow, K. G. and S. C. Foo (1991) “Thermal comfort in the humid tropics:
Field experiments in air conditioned and naturally ventilated buildings in Singapore”.
International Journal of Biometeorology, Vol. 34, pp. 259-265.
de Dear, R.J., Leow, K. G. and A. Ameen (1991) “Thermal comfort in the equatorial climatic
zone -- Part II: Climate chamber experiments on thermal acceptability in Singapore”.
ASHRAE Transactions, Vol. 97(1), pp. 880-886.
Project location, climate and season
The field experiments were conducted in both summer and winter seasons in Singapore
which is a wet equatorial climate.
Sample buildings
Building Code (blcode)
Sample Size (n) Climate Controls (bldgtype)
Floor Area
Occupancy Pattern
1 333 HVAC office building 2 583 NV residential building
Instruments
A hot-wire sensor was used to measure air speed. Relative humidity was measured using
an aspirated psychrometer and mercury-in-glass thermometers were used to measure air
and globe temperature. For globe temperature a 0.15m copper sphere was used.
ASHRAE RP-884 Final Report
Appendix C page 277 MRL Australia
Questionnaire
Thermal sensation was rated on the ASHRAE 7-pt scale. Thermal acceptability and thermal
preference was not addressed. Metabolic ratings were taken and clothing insulation was
estimated using the ISO7730 1984 standard. Questions of adaptive behaviour were not
considered.
Outdoor meteorological data
Outdoor Climatological air temperature and relative humidity data at 600 hours and 1500
hours was obtained from the International Station Meteorological and Climate Summary
CDROM (ISMCS, 1992) for Paya Lebar, the closest site.
RP-884 standardization assumptions
The research design was cross-sectional which satisfied the assumptions for RP-884, that
all subjects were independent. Clothing insulation estimated using the ISO7730 1984
standard was corrected to the ASHRAE55 1992 standard. 0.15 clo was added to the total
clothing ensemble insulation for the insulation effects of a chair forming a separate variable
in RP-884.
ASHRAE RP-884 Final Report
Appendix C page 278 MRL Australia
C.16. Project Title - The Steelcase building. Grand Rapids Michigan, US
Project researchers and class of investigation
F. Bauman et al. (CEDR at the University of California at Berkeley).
This is a CLASS-1 field experiment.
Project file names in the RP-884 database
This project is disseminated as file number 43 (winter - HVAC) in the RP-884 database.
Project publications
Project location, climate and season
This project was conducted in winter in Grand Rapids, Michigan. Grand Rapids has a
continental location in the Great Lakes region of North America and has a humid mid-
latitude climate.
Sample buildings
Building Code (blcode)
Sample Size (n) Climate Controls (bldgtype)
Floor Area
Occupancy Pattern
1 85 HVAC office building
Instruments
The Grand Rapids, Michigan field experiment was not part of the Advanced Customer
Technology Test (ACT2) study but was carried out in an identical format. A cart was set up
with all sensors attached in desired positions of 0.1m, 0.6m and 1.1m. The sensors chosen
were selected to meet the response time and accuracy requirements of ASHRAE Standard
55-81 and ISO Standard 7730 for thermal assessment. YSI series 700 probes with vinyl-
coated tips were used to measure air temperature. Globe temperature was measured by
attaching a 38 mm diameter table tennis ball on the temperature sensors. The balls were
painted grey for correct emissivity. Air velocity was measured by Dantec 54R10
anemometers, which are omnidirectional fully temperature-compensated sensors. Dewpoint
temperature was measured by a General Eastern DEW-10 chilled mirror dewpoint
transducer. All parameters were measured at all three heights except dewpoint temperature
which was only measured at 0.6m. Radiant asymmetry and illuminance where also recorded
but were not essential to the purpose of RP-884.
ASHRAE RP-884 Final Report
Appendix C page 279 MRL Australia
Questionnaire
The questionnaire consisted of an on-line questionnaire, which addressed conditions at the
time physical measurements were being taken and a background questionnaire. The latter
covered subject details such as, health and emotional characteristics, office description,
work area and job satisfaction, environmental sensitivity, plus personal comfort, satisfaction
and perceived control. In the on-line section thermal sensation was rated on the 7-pt
ASHRAE scale. Thermal preference was assessed on a descriptive 3-pt scale. Thermal
acceptability was not rated. Metabolic rate was estimated based on a checklist referring to
the subjects activity in the 15 minutes before completing the on-line questionnaire, using
tables in the ASHRAE Handbook of Fundamentals (HOF, 1985). Clo estimates were based
on responses to the clothing item checklist provided in the on-line questionnaire from the
ASHRAE Standard 55-81 method.
Outdoor meteorological data
Outdoor Meteorological data files are for Grand Rapids, MI, USA for the period January to
February 1992 were bought from the State Climatologist for Michigan by RP-884. The files
supplied had 24 hourly Temperatures (F) and Relative Humidity (%) for the 60 day period
required, from which air temperatures and relative humidities at 600 hrs and 1500 hrs were
extracted.
RP-884 standardization assumptions
The detailed methods and protocol used in ASHRAE RP-462 (and extended to the
ASHRAE RP-702 project described above) were carried out in full for the ACT2 Project.
Since RP-884 itself is based primarily on RP-702 and subsequently on RP-462 little
standardisation was necessary. However, clothing was based on the ASHRAE 55-81
method, and so required conversion into equivalent ASHRAE 55-92 values. 0.15 clo was
then added for chair insulation. The research design of this field experiment was part
longitudinal and part cross-sectional, but for the purposes of RP-884, independence
between subjects was assumed.
ASHRAE RP-884 Final Report
Appendix C page 280 MRL Australia
C.17. Project Title - Sunset Building: a study of occupant thermal comfort in
support of PG&E’s Advanced Customer Technology Test (ACT2) for maximum
energy efficiency
Project researchers and class of investigation
Charles C. Benton and Gail S. Brager (CEDR at University of California at Berkeley). This
is a CLASS-1 investigation.
Project file names in the RP-884 database
This project is disseminated as file numbers 44 (summer - HVAC) and 45 (winter - HVAC) in
the RP-884 database.
Project publications
Benton, C. C. and Brager, G. S. (1994) Sunset Building: Final Report; A study of occupant
thermal comfort in support of PG&E’s advanced customer technology test (ACT2) for
Maximum Energy Efficiency, CEDR.
Benton, C. C. and Brager, G. S. Advanced Customer Technology Test (ACT2) Verifone
Progress Report (CEDR UC Berkeley).
Project location, climate and season
San Ramon is one of 3 location, in which 2 of the 4 components of the ACT2 project were
carried out. San Ramon falls within a Mediterranean climate zone, but experiences local
climatic effects due its location. San Ramon is inland east of San Francisco Bay and almost
directly east of the city of San Francisco. The field experiments were conducted across the
summer and winter months.
Sample buildings
Building Code (blcode)
Sample Size (n) Climate Controls (bldgtype)
Floor Area
Occupancy Pattern
1 152 HVAC office building 2 133 HVAC office building 3 96 HVAC office building
Instruments
ASHRAE RP-884 Final Report
Appendix C page 281 MRL Australia
A cart was set up with all sensors attached in desired positions of 0.1m, 0.6m and 1.1m.
The sensors chosen were selected to meet the response time and accuracy requirements of
ASHRAE Standard 55-81 and ISO Standard 7730 for thermal assessment. YSI series 700
probes with vinyl-coated tips were used to measure air temperature. Globe temperature
was measured by attaching a 38 mm diameter table tennis ball on the temperature sensors.
The balls were painted grey for correct emissivity. Air velocity was measured by Dantec
54R10 anemometers, which are omnidirectional fully temperature-compensated sensors.
Dewpoint temperature was measured by a General Eastern DEW-10 chilled mirror
dewpoint transducer. All parameters were measured at all three heights except dewpoint
temperature which was only measured at 0.6m. Radiant asymmetry and illuminance where
also recorded but were not essential to the purpose of RP-884.
Questionnaire
The questionnaire consisted of an on-line, laptop-computer based questionnaire, which
addressed conditions at the time physical measurements were being taken and a
background questionnaire. The latter covered subject details such as health and emotional
characteristics, office description, work area and job satisfaction, environmental sensitivity,
plus personal comfort, satisfaction and perceived control. In the on-line section thermal
sensation was rated on the 7-pt ASHRAE scale. Thermal preference was assessed on a
descriptive 3-pt scale. Thermal acceptability was not rated. Metabolic rate was estimated
based on a checklist referring to the subjects activity in the 15 minutes before completing the
on-line questionnaire, using tables in the ASHRAE Handbook of Fundamentals (HOF,
1985). Clo estimates were based on responses to the clothing item checklist provided in the
on-line questionnaire from the ASHRAE Standard 55-81 method.
Outdoor meteorological data
Outdoor Meteorological air temperature data was obtained by request to the National
Climate Data Center (NCDC) for San Ramon and humidity was obtained from the
International Station Meteorological Climate Summary CDROM for the closest available site
(Stockton). From this data air temperatures and relative humidities at 600 hrs and 1500 hrs
were extracted for RP-884 purposes.
ASHRAE RP-884 Final Report
Appendix C page 282 MRL Australia
RP-884 standardization assumptions
This project was conducted based on the format of RP-462 (RP-702). Since RP-884 itself is
based primarily on RP-702 and subsequently on RP-462 little standardisation was
necessary. However, clothing was based on the ASHRAE 55-81 method, and so required
conversion into equivalent ASHRAE 55-92 values. 0.15 clo was then added for chair
insulation. The research design of this project was longitudinal, but for RP-884 purposes all
subjects were assumed to be independent (ie. cross-sectional).
ASHRAE RP-884 Final Report
Appendix C page 283 MRL Australia
C.18. Project Title - The Verifone building, a component of the Advanced Customer
Technology Test (ACT2) project.
Project researchers and class of investigation
Charles C. Benton and Gail S. Brager (CEDR at University of California at Berkeley). This
is a CLASS-1 field experiment.
Project file names in the RP-884 database
This project is disseminated as file number 46 (winter - HVAC) in the RP-884 database.
Project publications
Benton, C. C. and Brager, G. S. Advanced Customer Technology Test (ACT2) Verifone
Progress Report (CEDR UC Berkeley)
Project location, climate and season
This field experiment was conducted in winter in Auburn, California and is one of the
components of the ACT2 project. Auburn has a Mediterranean bordering on high altitude
climate and is located inland and to the north east of San Francisco.
Sample buildings
Building Code (blcode)
Sample Size (n) Climate Controls (bldgtype)
Floor Area
Occupancy Pattern
1 128 HVAC office building
Instruments
A cart was set up with all sensors attached in desired positions of 0.1m, 0.6m and 1.1m.
The sensors chosen were selected to meet the response time and accuracy requirements of
ASHRAE Standard 55-81 and ISO Standard 7730 for thermal assessment. YSI series 700
probes with vinyl-coated tips were used to measure air temperature. Globe temperature
was measured by attaching a 38 mm diameter table tennis ball on the temperature sensors.
The balls were painted grey for correct emissivity. Air velocity was measured by Dantec
54R10 anemometers, which are omnidirectional fully temperature-compensated sensors.
Dewpoint temperature was measured by a General Eastern DEW-10 chilled mirror
dewpoint transducer. All parameters were measured at all three heights except dewpoint
ASHRAE RP-884 Final Report
Appendix C page 284 MRL Australia
temperature which was only measured at 0.6m. Radiant asymmetry and illuminance were
also recorded, but not essential to the purpose of RP-884.
Questionnaire
The questionnaire consisted of an on-line questionnaire, which addressed conditions at the
time physical measurements were being taken and a background questionnaire. The latter
covered subject details such as, health and emotional characteristics, office description,
work area and job satisfaction, environmental sensitivity, plus personal comfort, satisfaction
and perceived control. In the on-line section thermal sensation was rated on the 7-pt
ASHRAE scale. Thermal preference was assessed on a descriptive 3-pt scale. Thermal
acceptability was not rated. Metabolic rate was estimated based on a checklist referring to
the subjects activity in the 15 minutes before completing the on-line questionnaire, using
tables in the ASHRAE Handbook of Fundamentals (HOF, 1985). Clo estimates were based
on responses to the clothing item checklist provided in the on-line questionnaire from the
ASHRAE Standard 55-81 method.
Outdoor meteorological data
An error in dates requesting outdoor air temperature data for Auburn from the National
Climate Data Center (NCDC) resulted in the use of climatological data for both air
temperature and relative humidity at 600 hours and 1500 hours. The data was obtained
from the International Station Meteorological Climate Summary CDROM (ISMCS, 1992) for
the closest available site, Sacramento.
RP-884 standardization assumptions
The detailed methods and protocol used in ASHRAE RP-462 (and extended to the
ASHRAE RP-702 project described above) were carried out in full for the ACT2 Project.
Since RP-884 itself is based primarily on RP-702 and subsequently on RP-462 little
standardisation was necessary. However, clothing was based on the ASHRAE 55-81
method, and so required conversion into equivalent ASHRAE 55-92 values. 0.15 clo was
then added for chair insulation. The research design of this project was longitudinal, so for
RP-884 purposes all subjects were assumed to be independent (ie. cross-sectional).
ASHRAE RP-884 Final Report
Appendix D page 285 MRL Australia
APPENDIX D - CLIMATE CLASSIFICATION
ASHRAE RP-884 Final Report
Appendix D page 286 MRL Australia
ASHRAE RP-884 Final Report
Appendix D page 287 MRL Australia
Figure D.1: The climate classification used throughout the RP-884 database.
ASHRAE RP-884 Final Report
Appendix E page 287 MRL Australia
APPENDIX E - CODEBOOK FOR RAW DATA IN RP-884 DATABASE
ASHRAE RP-884 Final Report
Appendix E page 288 MRL Australia
RP-884 Variable Coding Conventions variable's
Type of data code name Description of variable and units
Basic blcode building ID code Identifiers sub subject number
age subject's age [years] sex subject's gender [0=male, 1=female] year year
day julian date (jan 1=1, dec 31=365)
time time thermal ash ASHRAE Thermal Sensation Scale [-3, +3] questionnaire prxy_tsa Thermal acceptability defined as -1.5<=ASH<=+1.5
3=want warmer] vent air movement acceptability [6(very acc), 1(very unacc)] avm air movement preference [3(more), 2(no change),
1(less)] comf General thermal comfort right now [1=very uncomf,
6=very comf] act10 metabolic activity in last 10 minutes [met] act20 metabolic activity between 20 and 10 minutes ago [met] act30 metabolic activity between 30 and 20 minutes ago [met] act60 metabolic activity between 60 and 30 minutes ago [met] met average metabolic rate of subject [met] clo ensemble clothing insulation [clo] upholst insulation of the subject's chair [clo] insul clothing plus upholstery insulation [clo]
Indoor Climate ta_h air temperature at 1.1m above floor [oC] Physical Obs ta_m air temperature at 0.6m above floor [oC]
ta_l air temperature at 0.1m above floor [oC] dewpt dewpoint temperature [oC] prta_b plane radiant asymmetry temperature [oC] tg_h globe temperature at 1.1m above floor [oC] tg_m globe temperature at 0.6m above floor [oC] tg_l globe temperature at 0.1m above floor [oC] vel_h air speed 1.1m [m/s] vel_m air speed 0.6m [m/s] vel_l air speed 0.1m [m/s] turb_h turbulence intensity at 1.1m above floor [frac] turb_m turbulence intensity at 0.6m above floor [frac] turb_l turbulence intensity at 0.1m above floor [frac]
Continue Table.
ASHRAE RP-884 Final Report
Appendix E page 289 MRL Australia
variable's Type of data code name Description of variable and units
calculated taav average of three heights' air temperature [oC] indices trav average of three heights' mean radiant temperature [oC]
top average of TAAV and TRAV (operative temperature) [oC]
velav average of three heights' air speed [m/s] velmax maximum of three heights' air speeds [m/s] tuav average of three heights' turbulence [frac] pa vapor pressure [kPa] rh relative humidity [%] et new effective temperature index et* [oC] set new standard effective temperature index set* [oC] tsens two-node tsens index [-1.5, +2.0] disc two-node disc index [-4, +4] pmv Predicted Mean Vote, Fanger's Model [-3, +3] ppd Predicted Percentage Dissatisfied, Fanger's Model
[frac] pd_h Percent Dissatisfied due to Draft at 1.1m height, Fanger
et al [frac] pd_m Percent Dissatisfied due to Draft at 0.6m height, Fanger
et al [frac] pd_l Percent Dissatisfied due to Draft at 0.1m height, Fanger
et al [frac] pd_max Percent Dissatisfied due to Draft, max of all 3 heights,
Fanger et al [frac]
personal PCC perceived control over thermal environ [1=no control, 5=complete control]
environmental PCC_AG aggregate perceived control from PCEC1...PCEC7 control PCS how satisfied are you with PCC [1=very dissat, 6=very
sat] PCEC1 can you open/close windows? [1=yes, 0=no] PCEC2 can you open/close external doors [1=yes, 0=no] PCEC3 can you open/close internal doors [1=yes, 0=no] PCEC4 can you adjust thermostats [1=yes, 0=no] PCEC5 can you adjust curtains/blinds [1=yes, 0=no] PCEC6 can you adjust local heaters [1=yes, 0=no] PCEC7 can you adjust local fans [1=yes, 0=no]
Do you exercise any PCED1 windows [1=na, 2=never, 3=rarely, 4=sometimes, 5=often, 6=always]
of these options? PCED2 external door [1=na, 2=never, 3=rarely, 4=sometimes, 5=often, 6=always]
PCED3 internal door [1=na, 2=never, 3=rarely, 4=sometimes, 5=often, 6=always]
PCED6 local heater [1=na, 2=never, 3=rarely, 4=sometimes, 5=often, 6=always]
PCED7 local fan [1=na, 2=never, 3=rarely, 4=sometimes, 5=often, 6=always]
ASHRAE RP-884 Final Report
Appendix E page 290 MRL Australia
Continue Table. variable's
Type of data code name Description of variable and units
Outdoor Meteorol day15_ta outdoor 3pm (max) air temp on day of survey [oC] Observations day06_ta outdoor 6am (min) air temp on day of survey [oC]
dayav_ta outdoor average of min/max air temp on day of survey [oC]
day15_rh outdoor 3pm (min) rel humid on day of survey [%] day06_rh outdoor 6am (max) rel humid on day of survey [%] dayav_rh outdoor average min/max rel humid on day of survey [%] day15_et outdoor 3pm ET* on day of survey (Ta and rh at time of
daymx_ta) [oC] day06_et outdoor ET* on day of survey (Ta and rh at time of
daymn_ta) [oC] dayav_et outdoor average of min/max ET* on day of survey [oC]
ASHRAE RP-884 Final Report
Appendix F page 291 MRL Australia
APPENDIX F - CODEBOOK FOR THE RP-884 META ANALYSIS
ASHRAE RP-884 Final Report
Appendix F page 292 MRL Australia
Meta Analysis Codebook Variable Description
authors investigators of the study season season of study country country study carried out in
city city where the study was done seasnum 1 = summer (cooling season), 2 = winter (heating season) dataclas grade of research methods and resulting data (1st, 2nd, 3rd) bldgtype type of building, ie. 1 = climate controlled (HVAC), 2 = free running (NV) and 3
= mixed blcode individual building code
n sample size per building m_taav mean taav s_taav standard deviation taav m_trav mean trav s_trav standard deviation trav m_top mean top s_top standard deviation top
m_velav mean velav s_velav standard deviation velav m_rh mean relative humidity s_rh standard deviation relative humidity m_et mean et* s_et standard deviation of et*
ASH55_92 % indoor climatic obs falling within the relevant ASHRAE 55-92 comfort zone predneut the predicted neutral operative temperature given conditions of vel, rh, insul and
met. deltneut neut_top minus predneut preftemp defined in terms of operative temperature by probit analysis of MCI discrep neut_top minus preftemp m_set mean set* s_set standard deviation set*
m_pmv mean pmv s_pmv standard deviation pmv m_ppd mean ppd s_ppd standard deviation ppd
mpd_max mean pd_max (pd_max being the largest PD of the three heights measured) spd_max standard deviation pd_max m_ash mean ashrae thermal sensation vote s_ash standard deviation ashrae thermal sensation vote m_met mean metabolic rate (met) s_met standard deviation metabolic rate (met)
m_insul mean value of the summed clothing and chair insulation (clo) s_insul standard deviation of the summed clothing and chair insulation (clo) m_clo mean clothing insulation (clo) s_clo standard deviation of clothing insulation (clo)
mpcc_ag mean pcc_ag (pcc_ag is the index of perceived control) spcc_ag standard deviation pcc_ag (pcc_ag is the index of perceived control)
ASHRAE RP-884 Final Report
Appendix F page 293 MRL Australia
Continue Table Variable Description
mday15ta mean day15_ta (daily maximum outdoor air temperature degC) sday15ta standard deviation day15_ta (daily maximum outdoor air temperature degC) mday06ta mean day06_ta (daily minimum outdoor temperature degC) sday06ta standard deviation day06_ta (daily minimum outdoor temperature degC) mdayavta mean dayav_ta (mean of daily min and max temperatures degC) sdayavta standard deviation dayav_ta (mean of daily min and max temperatures degC) mday15et mean day15_et (ET* at time of max outdoor air temperature degC) sday15et standard deviation day15_et mday06et mean day06_et sday06et standard deviation day06_et mdayavet mean dayav_et sdayavet standard deviation dayav_et f_mci_2 % frequency when mci = 2 (no change) f_tsa_2 % frequency when tsa = 2 (acceptable) fprxysat % frequency when prxy_tsa = 2 (-1.5<ASH<+1.5) assumed acceptable grad_top gradient of the regression model mean_ash verses dose_top (ashtop)
p_top p value of the regression model testing gradient coefficient = 0 neut_top neutrality for dose_top (mean_ash = 0 in regression model ashtop) rang_top the acceptibility range for dose_top (mean_ash = 1.5 - mean_ash = -1.5 in regr
model ashtop). grad_et gradient of the regression model mean_ash verses dose_et (ashet)
p_et p value of the regression model testing gradient coefficient = 0 neut_et neutrality for dose_et (mean_ash = 0 in regression model ashtop) rang_et the acceptibility range for dose_et (mean_ash = 1.5 - mean_ash = -1.5 in regr model
ashtop). grad_set gradient of the regression model mean_ash verses dose_set (ashset)
p_set p value of the regression model testing gradient coefficient = 0 neut_set neutrality for dose_set (mean_ash = 0 in regression model ashset) rang_set the acceptibility range for dose_set (mean_ash = 1.5 - mean_ash = -1.5 in regr
model ashset) grad_pmv gradient of the regression model mean_ash verses dose_pmv (ashpmv)
p_pmv p value of the regression model testing gradient coefficient = 0 neut_pmv neutrality for dose_pmv (mean_ash = 0 in regression model ashpmv) rang_pmv the acceptibility range for dose_pmv (mean_ash = 1.5 - mean_ash = -1.5 in regr
22.8 19.8 22.7 Melbourne A Ballantyne 20.7 9.5 21.3 Melbourne A Ballantyne 20.4 11.1 20.5 Melbourne A Auliciems 1977
15.2 23.9 Sydney A Hindmarsh 13.3 22.3 Sydney A Hindmarsh 21.6 24.2 Sydney P Hindmarsh 19.4 21.4 Sydney P Hindmarsh 12.4 21 Sydney A Wong 21.3 23 Sydney A Wong
19.5 12.8 20.6 Adelaide A Auliciems 22.6 17.1 23.1 Brisbane P Auliciems 19.6 14.7 21.9 Perth A Auliciems 22.4 8.3 21.3 Armidale A Auliciems
28.1 26.2 Darwin P Macpherson 28.1 27.6 Darwin P Macpherson 28.9 26.2 Weipa P Wyndham
28.3 27.8 25.4 Pt Moresby P Ballantyne 25.9 25.4 Pt Moresby P Ballantyne
26.9 25 Pt Moresby P Ballantyne 26.9 27.2 Pt Moresby P Ballantyne
27.8 27 27.5 Honiara P Woolard 28.2 28.9 26.1 Singapore P Ellis 28.6 27 26.1 Singapore P Ellis 28.8 27 27.3 Singapore P Webb 33.4 33.5 30.1 New Delhi P Nicol 30.3 26.4 26.1 Calcutta P Rao 35.9 33.9 31.2 Baghdad P Nicol 28.8 24.8 25.8 Rio de Janeiro P Sa 24.7 21.3 24.6 Rio de Janeiro P Sa
22 22.5 Toronto A Tasker 22.8 23.9 New York A Gagge 21.5 23.6 Minneapolis A Newton
23.5 12.5 24.4 Portland A Pepler 23.6 12.5 22.1 Portland A Pepler 21.1 3.5 19.8 Swedish Towns A SIB 24.1 10.2 19 Swedish Towns A SIB 23.9 0.6 21.5 Zurich A Wanner 23.5 18.3 23.1 Zurich A Wanner 22.7 2.2 20.9 Zurich A Grandjean 23.2 17.8 21.3 Zurich/Basel/Bern A/P Grandjean 18.1 4.7 18.4 London A Bedford 18.8 6.7 19.2 London A Black 19 17 22.2 London A Black
17.2 5.2 17.5 London A Fox 20.5 10.6 22.4 London A Wyon 19.7 4.7 18.9 London A Angus 21.4 15.9 19.4 London P Hickish 21.1 17.9 21.3 Garston A Humphreys 21.4 3.8 19.9 Garston A Humphreys 21.4 7.7 19.7 Garston A Humphreys 21.4 10.8 19.3 Garston A Humphreys 21.4 14.4 20 Garston P Humphreys 21.4 16.4 20.2 Garston P Humphreys