Thermal sensation and comfort in transient non-uniform thermal environments

Abstract

Most existing thermal comfort models are applicable only to steady-state, uniform thermal environments. This paper presents summary results from 109 human tests that were performed under non-uniform and transient conditions. In these tests, local body areas were independently heated or cooled while the rest of the body was exposed to a warm, neutral or cool environment. Skin temperatures, core temperature, thermal sensation and comfort responses were collected at 1- to 3-min intervals. Based on these tests, we have developed predictive models of local and overall thermal sensation and comfort.

Full text

eScholarship provides open access, scholarly publishing
services to the University of California and delivers a dynamic
research platform to scholars worldwide.
Center for the Built Environment
UC Berkeley
Peer Reviewed
Title:
Thermal Sensation and Comfort in Transient Non-Uniform Thermal Environments
Author:
Zhang, Hui
Huizenga, Charlie
Arens, Edward, Center for the Built Environment, University of California, Berkeley
Wang, Danni
Publication Date:
01-01-2004
Series:
Indoor Environmental Quality (IEQ)
Publication Info:
Indoor Environmental Quality (IEQ), Center for the Built Environment, Center for Environmental
Design Research, UC Berkeley
Permalink:
http://escholarship.org/uc/item/64x0488x
Additional Info:
European Journal of Applied Physiology, Vol. 92, pp. 728-733. Previously presented at the Fifth
International Meeting on Thermal Manikins and Modeling, Strasbourg, September 2003.
Keywords:
Thermal sensation, thermal comfort, model, asymmetry, transient.
Abstract:
Most existing thermal comfort models are applicable only to steady-state, uniform thermal
environments. This paper presents summary results from 109 human subject tests that were
performed under non-uniform and transient conditions. In these tests, local body areas of the
subjects were independently heated or cooled while the rest of the body was exposed to a
warm, neutral or cool environment. Skin temperatures, core temperature, thermal sensation and
comfort responses were collected at one to three minute intervals. Based on these tests, we have
developed predictive models of local and overall thermal sensation and comfort.

Thermal Sensation and Comfort in Transient Non-Uniform Thermal Environments
Hui Zhang, Charlie Huizenga, Edward Arens, Danni Wang
University of California, Berkeley
zhanghui@uclink.berkeley.edu
Abstract
Most existing thermal comfort models are applicable
only to steady-state, uniform thermal environments.
This paper presents summary results from 109 human
subject tests that were performed under non-uniform
and transient conditions. In these tests, local body areas
of the subjects were independently heated or cooled
while the rest of the body was exposed to a warm,
neutral or cool environment. Skin temperatures, core
temperature, thermal sensation and comfort responses
were collected at one to three minute intervals. Based
on these tests, we have developed predictive models of
local and overall thermal sensation and comfort.
Key Words
Thermal sensation, thermal comfort, model, asymmetry,
transient.
1.0 Introduction
The majority of human comfort tests have been done
under steady state conditions in thermally uniform
environments (Nevins et al. 1966; McNall et al. 1967;
Fanger 1970; Rohles and Wallis 1979). Far fewer tests
have been done in transient uniform conditions (Gagge
et al. 1967; Griffiths and McIntyre 1974) and even less
have been done under steady-state conditions in non-
uniform thermal environments (e.g. Wyon et al. 1989;
Bohm et al. 1990). Taniguchi has done a limited
amount of work investigating the effects of cold air on
facial skin temperature during transient conditions in an
automobile (Taniguchi et al. 1992), but no other body
areas were considered.
The two most common thermal sensation models are
Fanger’s PMV (Fanger 1970) model and the TS and
DISC indices obtained from ET* and skin wettedness in
Gagge’s 2-Node model (Gagge et al. 1970). Both of
these models are based on uniform, steady-state test
data and although they work quite well under those
conditions, they have severe limitations under transient
and spatially non-uniform conditions.
There are a few models for non-uniform conditions.
Wyon, Bohm, and others developed Equivalent
Homogenous Temperature (EHT) to characterize non-
uniform environments and defined upper and lower
comfort bounds for each body segment. This approach
is limited to the clothing and metabolic conditions
tested and applies only to steady-state conditions.
Matsunaga proposed a simplified Average Equivalent
Temperature (AET) as a basis for predicting PMV
(Matsunaga et al. 1993). Hagino developed a model
limited to a specific set of test conditions that used a
weighted average of local comfort from the head, upper
arm, thigh, and foot to predict overall thermal sensation
(Hagino and Hara 1992).
Wang and Fiala proposed models for transients in
spatially uniform conditions (Wang 1994; Fiala 2002).
Wang’s model uses a static term from Fanger’s model
and a transient term based on the rate of heat storage in
the skin. Fiala’s model uses skin temperature, skin
temperature rate of change, and core temperature in a
regression based on human subject data from the
literature and from physiological model results. Guan
(Guan et al. 2003) developed a model for transient
environments which uses skin temperature for the static
term and the rate of heat gain for the dynamic term.
Recent work by Frank has shown that skin temperature
and core temperature have equal weighting for
predicting thermal sensation (as opposed to thermal
regulation) in uniform conditions (Frank et al. 1999).
No model exists that can predict thermal sensation in
non-uniform, transient conditions. This paper describes
such a model
.
2.0 Human subject experiment
2.1 Experimental set up
The human subject tests were carried out in the
Controlled Environmental Chamber at UC Berkeley
during January to mid August 2002 (Zhang 2003). To
create transient and non-uniform environments, we put
human subjects in a thermally uniform chamber and
then locally applied cooling or heating air using air-
sleeves attached to individual body segments. Figure 1
shows a subject with a sleeve attached to her back.
Local heating and cooling was applied to 11 separate
body areas: head, face, breath, neck, chest, back, pelvis,
arm, leg, hand, and foot.
A separate set of tests was carried out in an automobile
in a climate-controlled wind tunnel at the Delphi
Harrison facility in Lockport, NY. These tests
simulated conditions found in vehicles during both hot
and cold weather. The
test conditions covered a large
temperature range (–23.3 to 43 °C) with and without
Electronic version available at www.springerlink.com
Published in Eur J Appl Physiol (2004) 92: 728-733 DOI 10.1007/s00421-004-1137-y

Figure 1. An example of the heating/cooling air
sleeve attached to the subject’s back.
solar radiation. During these tests, the human subjects
were allowed to adjust the HVAC settings to their
preference. The results of these tests were mostly used
for model validation.
2.2 Measurements
We measured skin temperature using thermocouples at
28 locations (5-second intervals) and core temperature
(20-second intervals) using an ingestible temperature
device (CorTemp
TM
thermometer pill from HTI
Technologies, Inc.)
with a radio transmitter. We
collected subjective perception of overall and local
thermal sensation and comfort using 9 point analog
scales:
sensation: -4 very cold, -3 cold, -2 cool, -1 slightly
cool, 0 neutral, 1 slightly warm, 2 warm,
3 hot, 4 very hot
comfort: +0 just comfortable to 4 very comfortable;
-0 just uncomfortable to -4 very
uncomfortable
Subjects responded to subjective questions at one to
three minute intervals.
3.0 Observations
3.1 Impact of local thermal sensation on overall
thermal sensation
The influence of local sensation on overall sensation is
different for different body parts. Segments such as the
back and the chest are very dominant at influencing
overall sensation. When these segments were cooled,
overall sensation typically followed local sensation for
the cooled segments. For other segments such as the
hand and the foot, the impact of local sensation is much
less. Figures 2 and 3 shows that although hand skin
temperature and hand sensation changed dramatically
during local cooling the change in overall sensation was
much less than that found during back cooling.
4.0 Models to predict thermal sensation and
comfort
Based on the test results and literature, we developed
four models to predict local and overall sensation and
comfort.
4.1 Local thermal sensation model
We defined a setpoint for each body part as the skin
temperature when the whole body was in a neutral
condition. Figure 4 shows head sensation vs. forehead
skin temperature from 43 tests of steady-state- and
asymmetrical conditions. As forehead skin temperature
became colder (T
forehead
– T
forehead,set
< -2), the head
sensation leveled off, in a way that is effectively
described by a logistic function. On the warm side
(T
forehead
– T
forehead,set
> 0) the sensation did not level off
as clearly because the skin temperature change is small
in this region. Gagge at al. (1967) also found the same
effects for the whole body, on both the cold and warm
sides.
Figure 4 also shows that the head sensation was
modified by the whole-body thermal state. The same
forehead skin temperature felt relatively warmer when
the rest of the body was colder (solid squares: overall
sensation from –3.5 to –2.5, open squares: overall
sensation from –2 to -1) and colder when the rest of the
body was warmer (open triangles: overall sensation
from –0.5 to 0.5, solid triangles: overall sensation form
2 to 3).
g( )
22
24
26
28
30
32
34
36
38
135 140 145 150 155 160 165 170 175 180
elapsed time (minutes)
Skin Temperature (°C)
-4
-3
-2
-1
0
1
2
3
4
Thermal Sensation
back temperature overall sensation back sensation
T
room
= 28°C
T
cooling
= 14°C
cooling applied
cooling removed
Figure 2. Back and overall sensation during a back cooling test
22
24
26
28
30
32
34
36
38
80 85 90 95 100 105 110 115 120 125 130 135
elapsed time (minutes)
Skin Temperature (°C)
-4
-3
-2
-1
0
1
2
3
4
Thermal Sensation
hand temperature overall sensation left hand sensation
T
room
= 28°C
T
cooling
= 14°C
cooling applied
cooling removed
Figure 3. Hand and overall sensation during a hand cooling test
Electronic version available at www.springerlink.com
Published in Eur J Appl Physiol (2004) 92: 728-733 DOI 10.1007/s00421-004-1137-y

Testing local sensations for hand, forehead, and neck,
Hildebrandt also found that for a constant local skin
temperature, the local sensation became warmer when
the overall environment was cooler (Hildebrandt, Engel
et al. 1981).
Our proposed local thermal sensation model
incorporates these three considerations. (1) The local
sensation model is represented by a logistic function of
local skin temperature. As the local skin temperature
gets further away from the local skin temperature set
point, the sensation reaches the sensation scale limits
(+4 and –4). (2) The model is not symmetric for warm
and cool conditions. The slope of the curve is steeper
on the warm side to reflect the smaller range of
tolerable skin temperatures above the setpoint compared
to below the setpoint. (3) Local thermal sensation is not
solely a function of local skin temperature; it is also
influenced by the overall thermal state of the body, as
shown in Figure 4. For a given local skin temperature,
the local sensation is perceived as warmer if the whole
body is colder, and colder if the whole body is warmer.
Our model of local thermal sensation is therefore a
group of contours representing various levels of overall
body thermal state (Figure 5).
In our model, we use mean skin temperature to
represent overall body thermal state.
A transient term, a function of the time derivatives of
skin and core temperatures, is added to the steady-state
model to predict local thermal sensation under transient
conditions.
4.2 Local thermal comfort model
Under steady-state conditions, thermal sensations
farther from neutral are generally perceived as less
comfortable. However, in transient conditions many
researchers have demonstrated that during hyperthermia
or hypothermia, cold or warm stimuli (respectively) to
the hand, forehead, and neck are experienced as very
pleasant (Cabanac 1972; Mower 1976; Attia and Engel
1981). In fact, local comfort under these conditions is
higher than under uniform conditions.
Figure 6 presents results from 30 foot cooling/warming
tests. They compare favorably to findings by the above
researchers. (1) When the whole body was neutral
(gray circles), adding foot cooling reduced foot
comfort. There were no ‘very comfortable’ votes
(above 2) shown in any of our neutral whole body tests.
(2) The ‘very comfortable’ votes occurred when the
whole body was warm or cold and the foot was cooled
or warmed in the opposite direction to relieve
discomfort. When the whole body was warm, foot
cooling was perceived as ‘very comfortable’ (votes
reached 3, triangles on the upper left); when the whole
body was cold, foot warming was perceived as ‘very
comfortable’ (squares on the upper right). The local
sensation where the maximum comfort happened shifted
towards cold or warm based on the body’s thermal
state. (3) As local sensation continued towards very
cold or very warm, the local comfort started to drop
(triangles on the lower left). (4) When the whole body
was cold, adding foot cooling was perceived as
uncomfortable (squares on the lower left). The
discomfort was greater than when the whole body was
neutral (circles) or warm (triangles).
Our model predicts local comfort as a function of local
and overall thermal sensation. As overall thermal
-4
-3
-2
-1
0
1
2
3
4
-15 -10 -5 0 5 10 15
T
skin
-T
setpoint
Local Sensatio
n
whole body warmer
whole body colder
Figure 5. Local sensation model.
-4
-2
0
2
4
-10-8-6-4-2 0 2 4
T
forehead
- T
forehed,set
(°C)
Head Sensation
Overall sensation (-3.5 to -2.5) Overall sensation (-2.0 to -1.0)
Overall sensation (-0.5 to 0.5) Overall sensation (2 to 3)
Figure 4. Head sensation and forehead skin temperature. The
head sensation is influence by the whole-body thermal state.
-4
-3
-2
-1
0
1
2
3
4
-4 -2 0 2 4
Foot Sensation
Foot Comfort
Overall sensation (-3 to -1) Overall sensation (neutral)
Overall sensation (1 to 3)
Figure 6. Foot sensation and comfort. The foot comfort is
influence b
y
the whole-
b
od
y
thermal state.
Electronic version available at www.springerlink.com
Published in Eur J Appl Physiol (2004) 92: 728-733 DOI 10.1007/s00421-004-1137-y

sensation is cooler, a warm local sensation is
increasingly comfortable. Conversely, as the overall
sensation is warmer, a warm local sensation becomes
increasingly uncomfortable. The shifts to cold and
warm are not necessarily equal, so the local thermal
comfort model is asymmetric (Figure 7). We fit this
model to each of the body segments we studied.
4.3 Overall thermal sensation model
Overall thermal sensation is predicted from local
thermal sensations by using a weighted average. In
establishing weights for the different body parts, we
found three effects: (1) As local sensation diverges from
that of the rest of the body (e.g. a cold hand contrasted
to a warm body), the weight becomes larger. This
increase is linear for most body parts. Figure 8 shows
an example for the back (circles are the original test
data, lines are the best linear fits). (2) Certain body
segments dominate the influence on overall sensation
(as shown in Figure 2). These body parts have larger
weights. The differences could be due to segment size
or thermal sensitivity. (3) Segments also differ in their
sensitivity to warm and cold. We observed from our
tests that the head, face, neck, and breathing are more
sensitive to heating than cooling, therefore the weights
for heating are larger than for cooling. Based on the
above three effects, we developed linear models to
calculate weights for all body parts. Figure 9 shows the
linear model for back, face, and hand.
4.4 Overall comfort model
We explored more than a thousand test data points, and
the best model we found for predicting overall comfort
was the following simple rule-based approach:
Rule 1:
Overall comfort is the average of the two
minimum local comfort votes unless Rule 2
applies.
Rule 2:
If the following criteria are met:
• the second lowest local comfort vote
is >–2.5
• the subject has some control over
their thermal environment
• the thermal conditions are transient
then overall comfort is the average of the two
minimum votes and the maximum comfort
vote.
The detailed mathematical descriptions of the four
models are provided in Zhang (2003).
4.5 Model validation
Validation results show that in general the models
predict subjects’ votes very well. We used data from
the Delphi wind tunnel tests to validate our sensation
and comfort models developed from the chamber
studies. Unlike the tests performed in Berkeley, the
Delphi subjects were allowed to adjust their
environment using the vehicle HVAC system.
0
0.2
0.4
0.6
0.8
1
-5 -4 -3 -2 -1 0 1 2 3 4 5
S
local
- S
mean
Weight
non-dominant
segment (hand)
asymmetrically-weighted
segment (face)
dominant segment (back)
Figure 9. Thermal sensation integration model showing three
example segments. Note that a dominant segment like the back has
a much higher weighting factor than a non-dominant segment such
as the hand. The face shows significant asymmetry between warm
and cold sensations.
0
0.2
0.4
0.6
0.8
1
1.2
1.4
-6 -4 -2 0 2 4 6
S
local
- S
mean
Weight
Figure 8. Weight as a function of the difference between local
and the rest body thermal sensation. The solid line is the best
linear fit of the data.
-4
-3
-2
-1
0
1
2
3
4
-4 -3 -2 -1 0 1 2 3 4
Local Thermal Sensation
Local Thermal Comfort
colder overall
warmer overall
neutral overall
Figure 7. Local thermal comfort model
Electronic version available at www.springerlink.com
Published in Eur J Appl Physiol (2004) 92: 728-733 DOI 10.1007/s00421-004-1137-y

Figure 10 presents the validation of the overall
sensation model. It shows the predicted overall
sensation calculated from local sensations vs. actual
overall sensations. The R
2
for the overall sensation
model is 0.95; quite high considering that these data
were not used to develop the model. The standard
deviation of residuals is 0.54.
Figure 11 presents the validation of the overall comfort
model. It shows the predicted vs. actual overall
comfort. The overall R
2
for the comfort model is 0.89;
the standard deviation of the residual is 0.78.
5.0 Conclusion
We have developed new sensation and comfort models
to predict local and overall sensations, and local and
overall comfort in non-uniform transient thermal
environments. The models were proposed based on our
human subject test results and observations of data from
the literature. Our validation work shows that the
models predict sensation and comfort with reasonable
success. Our next step will be to integrate these models
with physiological models (Rugh et al. 2003; Huizenga
et al. 2001). Once integrated, these tools will be very
useful in designing and evaluating thermal
environments.
6.0 References
Attia M, Engel P (1981) Thermal alliesthesial response
in man is independent of skin location stimulated.
Physiology & Behavior 27(3): 439 - 444.
Bohm M., Browen A, Holmer I, Nilsson H, Noren N
(1990) Evaluation of vehicle climate with a thermal
manikin - the relationship between human temperature
experience and local heat loss. Swedish Institute of
Agricultural Engineering.
Cabanac M (1972) Preferred skin temperature as a
function of internal and mean skin temperature. Journal
of Applied Physiology 33(6): 699 - 703.
Fanger PO (1970) Thermal comfort. McGraw-Hill
Book Company.
Fiala D (2002) First principles modeling of thermal
sensation responses in steady state and transient
conditions. ASHRAE Transactions.
Frank SM., Raja SN, Christian FB, Goldstein, DS
(1999) Relative contribution of core and cutaneous
temperatures to thermal comfort and autonomic
responses in humans. J. Appl. Physiol. 86(5): 1588 –
1593.
Gagge AP, Stolwijk JAJ, Hardy JD (1967) Comfort and
thermal sensation and associated physiological
responses at various ambient temperatures.
Environmental Research 1:1-20.
Gagge AP, Stolwijk JAJ, Nishi Y (1970) An effective
temperature scale based on a simple model of human
physiological regulatory response. ASHRAE
Transactions 77(1): 247 - 262.
Griffiths ID, McIntyre DA (1974) Sensitivity to
temporal variations in thermal conditions. Ergonomics
17, No. 4: 499 - 507.
Guan, Y., M. H. Hosni, B. W. Jones, T. P. Gielda
(2003) Investigation of Human Thermal Comfort under
Highly Transient Conditions for Automobile
Applications - Part2: Thermal Sensation Modeling.
ASHRAE
Transactions 109(2).
Hagino M, Hara J (1992) Development of a method for
predicting comfortable airflow in the passenger
compartment. SAE Technical Paper Series 922131: 1 -
10.
Hildebrandt G, Engel P, Attia M (1981)
Temperaturegulation und thermischer komfort.
Zeitschrift fur Physikalische
Medizin 10: 49 - 61.
Huizenga C, Zhang H, Arens E (2001) A model of
human physiology and comfort for assessing complex
thermal environments. Building and Environment 36,
pp 691-699.
Actual overall comfort
Predicted overall comfort
-4 -2 0 2 4
-4 -2 0 2 4
Actual overall comfort
Predicted overall comfort
Actual overall comfort
Predicted overall comfort
-4 -2 0 2 4
-4 -2 0 2 4
Actual overall comfort
Predicted overall comfort
Figure 11. Overall comfort model validation.
Actual overall sensation
Predicted overall sensation
-4 -2 0 2 4
-4 -2 0 2 4
Actual overall sensation
Predicted overall sensation
Actual overall sensation
Predicted overall sensation
-4 -2 0 2 4
-4 -2 0 2 4
Actual overall sensation
Predicted overall sensation
Figure 10. Overall sensation model validation.
Electronic version available at www.springerlink.com
Published in Eur J Appl Physiol (2004) 92: 728-733 DOI 10.1007/s00421-004-1137-y

Matsunaga K, Sudo F, Tanabe S, Madsen TL (1993)
Evaluation and measurement of thermal comfort in the
vehicles with a new thermal Manikin. SAE Paper Series
931958.
McNall PE, Jaax J, Roles FG, Nevins RG, Springer W
(1967) Thermal comfort (thermally neutral) conditions
for three levels of activity. ASHRAE Trans. 73(1): 1.3.1
- 1.3.14.
Mower DM (1976) Perceived intensity of peripheral
thermal stimuli is independent of internal body
temperature. Journal of Comparative and Physiological
Psychology 90(12): 1152 - 1155.
Nevins RG, Rohles FG, Springer W, Feyerherm AM
(1966) Temperature-humidity chart for thermal comfort
of seated persons. ASHRAE Trans. 72: 283 - 291.
Rohles FH, Wallis SB (1979) Comfort criteria for air
conditioned automotive vehicles. SAE Technical Paper
Series
790122.
Rugh J, Farrington R, Bharathan D, Vlahinos A, Burke
R, Huizenga C, Zhang H (2003) Predicting human
thermal comfort in a transient non-uniform thermal
environment. 5th International Meeting on Thermal
Manikins and Modeling, Strasbourg, France.
Taniguchi Y, Aoki H, Fujikake K, Tanaka H, Kitada M
(1992) Study on car air conditioning system controlled
by car occupants' skin temperatures - part 1: research
on a method of quantitative evaluation of car occupants'
thermal sensations by skin temperatures SAE Paper
Series 920169: 13-19.
Wang XL (1994) Thermal comfort and sensation under
transient conditions. Ph.D. Thesis, The Royal Institute
of Technology, pp 136.
Wyon DP, Larsson S, Forsgren B, Lundgren I (1989)
Standard procedures for assessing vehicle climate with
a thermal manikin. SAE Technical Paper Series
890049: 1-11.
Zhang H (2003) Human thermal sensation and comfort
in transient and non-uniform thermal environment.”
Ph.D. thesis, University of California at Berkeley.
Acknowledgments
This work was supported through the National
Renewable Energy Laboratory by U.S. DOE’s Office of
FreedomCAR and Vehicle Technologies (OFCVT).
The authors appreciate the support of NREL project
team members Rom McGuffin, John Rugh and Rob
Farrington. Delphi Harrison contributed in-kind
support to make the wind tunnel testing possible. We
wish to thank Taeyoung Han, Lin-Jie Huang, Greg
Germaine and the volunteer subjects from Delphi
Harrison.
Electronic version available at www.springerlink.com
Published in Eur J Appl Physiol (2004) 92: 728-733 DOI 10.1007/s00421-004-1137-y