Anda di halaman 1dari 6

ARTICLE IN PRESS

Building and Environment 42 (2007) 40004005


www.elsevier.com/locate/buildenv

Thermal comfort evaluation with virtual manikin methods


Hakan O. Nilsson
Department of Technology and Built Environment, Laboratory of Ventilation and Air Quality, University of Gavle, 801 76 Gavle, Sweden
Received 11 January 2006; accepted 21 April 2006

Abstract

Computational uid dynamics has become an important tool in the prediction of thermal comfort in occupied spaces. Despite its
ability to predict temperature and velocity elds, it is more difcult to evaluate the degree of thermal comfort experienced by an
occupant. This article describes the construction of a new numerical thermal manikin, with new comfort evaluation methods based on
data from thermal manikin measurements as well as subjective results from several hundred experiments. The level of thermal comfort is
highly dependent on the local environment. Human beings respond differently to local heat transfer in different parts of their bodies. It is
suggested for that reason that local results from manikins should be presented in new clothing independent comfort zone diagrams. The
research presented in here is intended to be used to evaluate system solutions that provide improved thermal climate in many different
everyday situations, e.g. all types of buildings and vehicles.
r 2006 Elsevier Ltd. All rights reserved.

Keywords: Thermal manikin; Mannequin; Thermal climate assessment; Clothing-independence; Comfort zone diagram

1. Introduction various types and amounts of clothing that manikins use


today, make comparative interpretation of results from
Modern occupants expect thermal indoor environments different manikins/methods very complicated. In order to
of continually increasing quality in many different types of facilitate comparison of results, the methods should be
situations. Thus, it has become necessary to calculate in independent both of the manikin used as well as clothing
advance the effects of different indoor climate solutions worn.
before anything permanent is built. Experimental work
with thermal sensation ratings is expensive, time consum-
ing and difcult to standardise. Simulations with computa- 2. Methods
tional models in virtual environments have been seen as a
new way of evaluating end simulating the inuence of and This method aims at combining different thermal climate
on human beings [13]. However, values from computa- investigation techniques into one evaluation scheme
tional uid dynamics (CFD) models still need to be further (Fig. 1). Results from human experiments and thermal
related to human reactions and veried with measurements manikin measurements are used to develop a methodology
in real environments. It is important to adjust the based on a virtual manikin positioned in a CFD-simulated
evaluation methods used according to clothing worn by environment. The results are presented not only as whole
subjects and manikins. The degree of thermal comfort body inuence, but also with local information on how the
depends greatly on the local environment [47]. Human thermal climate varies over the human body. The devel-
beings usually use different types of clothing and respond opment of virtual models is an efcient complement to
differently to heat transfer from different body areas. The traditional evaluation of the thermal environment.
An effective and clear measure of thermal comfort is the
Tel.: +46702192492; fax: +4626648113. equivalent temperature (teq). This measure comes closer to
E-mail address: hnn@hig.se. the humans experienced temperature, compared to ordin-
URL: http://www.hig.se/hnn. ary operative temperature, with an inclusion of the effects

0360-1323/$ - see front matter r 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.buildenv.2006.04.027
ARTICLE IN PRESS
H.O. Nilsson / Building and Environment 42 (2007) 40004005 4001

Fig. 1. The climate evaluation methods associated in this article: human Fig. 2. MANIKIN3 built with PROSTAR version 3.102.517, PRE/POST
subjective ratings, manikin measurements and computer modelling. processor for STAR-CD (Star CD 3.1B Computational Dynamics
Limited, 2001).

from non-evaporative heat loss from the human body [8,9]. The turbulence model used for MANIKIN3 calculates the
The denition of equivalent temperature can be related turbulent viscosity empirically with
to the whole body as well as local parts of a human being.
mt 0:04ru0 H, (1)
In more than 500 experiments, total and local heat uxes
from full-scale thermal manikins were determined for where mt is the turbulent (eddy) viscosity [Pa s], r the uid
different climatic conditions and compared with thermal density [kg/m3], u0 the characteristic velocity, inlet velocity
sensation votes from subjects exposed to the same [m/s], and H the characteristic length, inlet min length [m].
conditions [10]. The individual votes were averaged for The length scale is a characteristic length; normally the
each condition and reported as a Mean Thermal Vote minium length of the air inlet should be used. In the same
(MTV) [11]. Equivalent temperature values obtained with way, the inlet velocity is used as the characteristic velocity.
different manikins in different test houses should be The empirical constant suitable for different indoor
comparable during the same test conditions. As a airows is a number between 0.038 and 0.040. Although
consequence of methodological and individual variations, this model is found to be sufcient for predicting the total
the use of climate evaluation limit lines is not the optimal characteristic of a turbulent ow, it may not always be
solution. A new diagram should be developed affording suitable for predicting local details. One benet of this
practical support for judgement of complicated local method is that the time used for calculations is much less
thermal climates, this would allow the generation of compared to more complicated models. Furthermore, the
general prole usable with different clothing and manikins use of this model does not need extensive grid renement or
in various environments. the use of special wall functions, two factors that
signicantly speed up the working and iteration process.
2.1. Computational manikin with new surface temperature Consequently the computer power needed to calculate the
regulation airow is less and can quite easily be realised using an
ordinary personal computer.
MANIKIN3 is built inside the computer with a heat ow The STAR-CD CFD code [13] supports continuous
interface to a CFD code (Fig. 2). The virtual manikin is modication boundary conditions during the ow and
built up of cubes and has a similar size, area and number of temperature eld calculations. Functions have been de-
zones as the real MANIKIN2. Both manikins have a free signed that regulate the numerical manikin surface
surface area of 1.6 m2, giving the same constant heat ux to temperature and calculate the equivalent temperature.
the surrounding air. The surface temperature of MANI- These functions are introduced in the CFD calculations
KIN3 is regulated during the iteration process, using the with the results of the ow and temperature elds as input.
adaptive boundary conditions at the rst grid cell. This The subroutines enable the user to calculate new surface
procedure together with a virtual calibration is implemen- temperatures and heat transfer for each zone on the
ted in order to even out the geometrical differences. When manikin. In order to minimise the computational load, the
the general mixing of the air is the main interest, it is surface temperatures and equivalent temperatures are
possible to use a constant turbulent (eddy) viscosity mt calculated in a user subroutine that is programmed to be
instead of the molecular viscosity [12]. This zero-equation called only once in every iteration step. These new Fortran
model uses a constant to express the turbulent viscosity. 77 user subroutines are written so that the surface
This does not require the solution of any additional temperature data and current equivalent temperature
differential equations beyond the NavierStokes equations. data for comfort diagram output are continuously updated
ARTICLE IN PRESS
4002 H.O. Nilsson / Building and Environment 42 (2007) 40004005

for real time monitoring. This method is much faster 2.2. Clothing independent comfort zone diagram evaluation
compared to previously published methods involving two output
or three separate codes for thermoregulation, radiation,
and CFD. No thermoregulation model has to be used In order to make the comfort evaluation clothing
as the routines are based on correlation results from independent, the construction of new comfort zone
extensive studies of subjective responses in different diagrams can be made by inserting any seated total
radiative and convective environments. The conrmed insulation available. Eq. (2) shows how a relationship
hypothesis [4,10] is that human beings do not differentiate between the equivalent temperature level and MTV can be
between various avenues of heat loss, just experience established for each manikin body part. The heat loss
an increased or decreased general heat loss from any zone corresponding to a certain level of comfort, or discomfort,
of the body. in the diagram is consequently considered to be the same.

Table 1
Linear regressions and calculated neutral values for clothing combinations (LS) Light summer clothing (1.0 clo) and (EW) Enhanced winter
clothing (1.9 clo)

Zone a b r2 RT (m2K/W)LS RT (m2K/W)EW IT (clo) LS IT (clo) EW teq (1C) neutral LS teq (1C) neutral EW

Whole body 43.8 13.3 0.97 0.160 0.300 1.03 1.94 21.0 21.4
Scalp 65.5 33.9 0.89 0.199 0.193 1.28 1.25 21.0 21.4
Face 65.5 33.9 0.89 0.199 0.193 1.28 1.25 25.7 17.2
Chest 36.1 20.5 0.95 0.229 0.464 1.48 2.99 25.7 17.2
Up. back 36.1 20.5 0.95 0.229 0.464 1.48 2.99 24.8 15.4
L U arm 43.0 21.1 0.94 0.215 0.432 1.39 2.79 24.8 15.4
R U arm 43.0 21.1 0.94 0.215 0.432 1.39 2.79 28.8 15.4
L L arm 43.0 21.1 0.94 0.122 0.432 0.79 2.79 28.8 15.4
R L arm 43.0 21.1 0.94 0.122 0.432 0.79 2.79 24.1 21.6
L hand 84.9 57.2 0.98 0.117 0.146 0.75 0.94 24.1 21.6
R hand 84.9 57.2 0.98 0.117 0.146 0.75 0.94 28.0 20.4
L thigh 46.7 20.3 0.97 0.128 0.292 0.83 1.88 28.0 20.4
R thigh 46.7 20.3 0.97 0.128 0.292 0.83 1.88 28.0 20.4
L calf 46.7 20.3 0.97 0.128 0.292 0.83 1.88 28.0 20.4
R calf 46.7 20.3 0.97 0.128 0.292 0.83 1.88 28.0 24.0
L foot 46.7 20.3 0.97 0.128 0.215 0.83 1.39 28.0 24.0
R foot 46.7 20.3 0.97 0.128 0.215 0.83 1.39 24.2 19.0
Lo. back 39.5 19.5 0.93 0.247 0.381 1.59 2.46 24.2 19.0
Seat 39.5 19.5 0.93 0.247 0.381 1.59 2.46 21.0 21.4

"Light summer clothing" Comfort Zones "Enhanced winter clothing" Comfort Zones
Whole body Whole body
Scalp Scalp
Face Face
Chest Chest
Up. back Up. back
L U arm L U arm
R U arm R U arm
cold but comfortable

L L arm L L arm
cold but comfortable

R L arm R L arm
hot but comfortable

hot but comfortable

L hand L hand
R hand R hand
L thigh L thigh
R thigh R thigh
L calf L calf
neutral
too cold

too cold
neutral

too hot

too hot

R calf R calf
L foot L foot
R foot R foot
Lo. back Lo. back
Seat Seat

-10 0 10 20 30 40 -10 0 10 20 30 40
Equivalent temperature, teq (C) Equivalent temperature, teq (C)

Fig. 3. Comfort zone diagrams constructed for light summer (LS) and enhanced winter (EW) clothing ensembles. Note that the comfort zones of
acceptance are much narrower for the summer clothing, except for the less sensitive head and hands.
ARTICLE IN PRESS
H.O. Nilsson / Building and Environment 42 (2007) 40004005 4003

The shape of the zones is however changed according to local total insulation value (RT) of the clothing and air
the clothing used: layer together with a MTVzone for the zone borders (1.5
left blue, 0.5 right blue, 0.5 left red, 1.5 right red) in
teq;zone ts  RT a b MTVzone , (2)
Eq. (2). Now teq,zone can be calculated for the four borders
where teq;zone is the equivalent temperature in the zone [1C], of the three shaded comfort zones (bluegreenred) for all
ts the manikin surface temperature (here 34 1C) [1C], RT the zones and the whole body. The result, plotted in a diagram,
total insulation, seated [m2 K/W], a, b the linear regression forms the evaluation background in the clothing indepen-
constants [W/m2], MTVzone the MTV in the zone [n.d.]. dent comfort zone diagram. This is only done once for each
The equation is valid for an interval of seated whole clothing combination, and reects the insulation distribu-
body total insulation (IT) between 0.9 and 1.9 clo. With tion of the clothing used. As could be expected, diagrams
Eq. (2) and the information in Table 1, it is possible for the for clothing with less insulation show an increased
rst time to make a comfort zone diagram (Fig. 3) that sensitivity in most zones, except the normally unclothed
applies to a specic clothing combination used in a given head and hands. The opposite, decreased sensitivity for
situation. Insert corresponding local values of the surface heat loss variations, can be observed for diagrams with
temperature (ts, usually 34 1C), a and b (from Table 1) and increased clothing insulation.

Mixing ventilation office case Displacement ventilation office case


Whole body Whole body
Scalp Scalp
Face Face
Chest Chest
Back U Back U
Arm LU omix meas Arm LU odis meas
Arm RU Arm RU
Arm LL omix cfd Arm LL
cold but comfortable
cold but comfortable

odis cfd
hot but comfortable
hot but comfortable

Arm RL Arm RL
Hand L Hand L
Hand R Hand R
Thigh L Thigh L
Thigh R Thigh R
too cold
neutral

neutral
toocold

Calf L Calf L
too hot
too hot

Calf R Calf R
Foot L Foot L
Foot R Foot R
Back L Back L
Seat Seat

10 15 20 25 30 35 40 10 15 20 25 30 35 40
Equivalent temperature, teq (C) Equivalent temperature, teq (C)

Fig. 4. The virtual and real ofce investigated with MANIKIN3 and MANIKIN2. Below the case with mixing ventilation (omixofce mixing) showing
good agreement between measurements and simulation and the displacement ventilation simulation (odisofce displacement) producing slightly lower
equivalent temperatures compared to the measurements due to an overall lower simulated air temperature. The comfort zone diagrams are drawn on the
basis of the 1.3 clo summer clothing that MANIKIN2 and MANIKIN3 wore.
ARTICLE IN PRESS
4004 H.O. Nilsson / Building and Environment 42 (2007) 40004005

3. Results simulations with MANIKIN3 made good predictions of


how a human will perceive the thermal climate.
3.1. Full-scale measurements vs. CFD simulations The cabin is build with an option to change the glazing used
in order to test the different possibilities to improve the thermal
Results from measurements with the full-scale MANI- climate. Fig. 5 shows few differences between simulation and
KIN2 (meas) have been compared to CFD simulations measurement in the case with clear cabin glass and no sun
with MANIKIN3 (CFD). The heat loss and temperature of (ccleargcabin clear glass). But in the case with solar radiation
the two manikins inuenced the air movements around the and reective glass (cregcabin reective glass) does a lower
body in a realistic way. Thermal interaction with walls, air temperature increase the heat loss from the hands and
ventilation and the seat inuenced each manikins thermal thighs for the measured data not compensated by the radiation
situation. The evaluated ofce room could alternatively be from the windows caused by the sun as in the simulation.
ventilated with mixing (omixofce mixing) or displace-
ment ventilation (odisofce displacement). MANIKIN3 4. Discussion
had no problem simulating the increased insulation of the
seat zones supplied by the virtual ofce chair (Fig. 4). In This work is intended to complement full-scale investi-
relatively homogeneous environments, as in this case, the gations in order to determine human thermal comfort in

Clear glass no sun cabin case Reflective glass and sun cabin case
Whole body Whole body
Scalp Scalp
Face Face
Chest Chest
Back U Back U
Arm LU Arm LU
cold b ut com fortable

cclearg meas
cold b ut com fortable
hot but com fortable

creflg meas
hot but com fortable

Arm RU Arm RU
Arm LL cclearg cfd Arm LL creflg cfd
Arm RL Arm RL
Hand L Hand L
Hand R Hand R
Thigh L Thigh L
Thigh R
too cold

Thigh R
neu tral

too cold
too h ot

neu tral

too h ot

Calf L Calf L
Calf R Calf R
Foot L Foot L
Foot R Foot R
Back L Back L
Seat
Seat
10 15 20 25 30 35 40 10 15 20 25 30 35 40
Equivalent temperature, teq (C)
Equivalent temperature, teq (C)

Fig. 5. This gure shows pictures of the virtual and real cabin environment with MANIKIN3 and MANIKIN2 in driver position. The boxes to the right
of MANIKIN3 are the shielded air temperature probes always connected to the manikin measurements, as seen with MANIKIN2 in the picture to the
right. Below clothing independent comfort zone diagrams with measurements and CFD results with MANIKN3 presenting good agreement between
measurements and simulations for the case with clear cabin glass and no sun (ccleargcabin clear glass). The reective glass with sun (cregcabin
reective glass) produced higher equivalent temperatures at the middle zones in the simulation due to higher air temperatures around the manikin caused
by slightly different inlet and hand positions.
ARTICLE IN PRESS
H.O. Nilsson / Building and Environment 42 (2007) 40004005 4005

different environments. For engineering purposes, the especially in situations with relatively homogeneous
comfort sensation needs to be converted to and climate, like the ofce case, but still have to be further
expressed in measurable, physical quantities. This is developed in order to be reliable in more complicated
facilitated using the concept of clothing independent climate situations like the cabin case.
comfort zone diagrams. The model of clothing indepen- Today, CFD calculation methods have developed
dence assumes that the human being is equally sensitive to further and a growing eld of research is working to
different heat losses independent of the insulation of the establish the methods for simulation of the human thermal
clothing worn. This may not always be the case, especially environment. Taking this into account, there are still too
at the borders of no clothing or heavy clothing and should many unexplained differences in the results within and in
be investigated further. In the future, more data from between simulation methods, pointing out the limitations
validation experiments with subjects and different methods of currently available CFD methods. There is consequently
will contribute to the development of this general evalua- a need for continued validation of CFD results with real
tion criterion. life measurements and benchmark tests.
The numerical methods used need further development.
The computer should become a professional tool for
Acknowledgements
visualising thermal comfort problems, developing new
ventilation strategies and designing new systems. With
The author sincerely appreciates the travel grant from
more research and experience the tools and methods for
the research-funding agency Formas (Swedish Research
simulation of thermal environment conditions can be
Council for Environment, Agricultural Sciences and
continuously improved. Unfortunately too little of the
Spatial Planning) that made it possible to present this
theory behind CFD simulations is available in the public
research at the 10th Indoor Air conference in China 2005. I
domain. Many researchers and companies still use in-house
am also most grateful to the Journal Paper Selection
codes for all or essential parts of their calculations.
Committee for selecting my paper for publication.
A remaining issue that requires more investigation is
how to simulate the correct heat transfer from different
types of wall boundaries to the uid, in order to carry out References
CFD modelling with better accuracy. A lot of research
today uses high-resolution calculations with rened grids. [1] Murakami S, Kato S, Zeng J. Flow and temperature elds around
Research is also needed at the other end of the scale, where human body with various room air distribution. CFD Study on
Computational Thermal Manikin-Part I 1997;103(part 1):315.
the whole cell is substituted for one or a number of wall [2] Brohus H. Personal exposure to contaminant sources in ventilated
functions specially adapted for the environment studied. It rooms. Thesis, Aalborg University, Aalborg, Denmark, 1997.
is important that research concerning the relationships for [3] Tanabe S. Numerical comfort simulator for evaluating thermal
simulation of transient situations continues. The develop- environment. In: Proceedings of the 10th International Conference on
Environmental Ergonomics, Fukuoka, Japan, 2002. p. 43538.
ment of thermal manikins to enable the handling of
[4] Nilsson H, Holmer I, Bohm M, Noren O. Equivalent temperature
thermal transient conditions has started. However, there is and thermal sensationcomparison with subjective responses. In:
still a need for more tests with human subjects in order to Proceedings of Comfort in the Automotive Industry, Bologna, Italy,
investigate the correlation between subjective feelings and ATA vol 1, 1997. p. 15762, 97A3018.
equivalent temperature transients. [5] Huizenga C, Hui Z, Arens E. A model of human physiology and
comfort for assesing complex thermal environments. Building and
Environment 2001(36):6919 /elsevier.comS.
5. Conclusions [6] Bureau C, Kampf H, Taxis-Reischl B, Traebert A, Mayer E, Schwab
R. MARCO BEHRs Method to assess thermal comfort. In: Vehicle
Results from subjective panels and measurements with Thermal Management Systems Conference Proceedings (VTMS 6),
full-scale thermal manikins have been used to develop a SAE Paper No. C599/005/2003, 2003. p. 22333.
[7] Han T, Huang L. A model for relating a thermal comfort scale to
new type of virtual thermal manikin, MANIKIN3, to be
EHT Comfort Index. SAE Paper 2004-01-0919, March 2004.
used in CFD simulations. The surface temperature of the [8] Dufton AF. The equivalent temperature of a warmed room. vol 4,
computational manikin is regulated continuously through 1936. p. 22729.
an iteration process. This procedure together with a new [9] Madsen TL, Olesen BW, Kristensen NK. Comparison between
model for virtual calibration forms the basis of this new operative and equivalent temperature under typical indoor condi-
numerical manikin concept. Results from measurements as tions. vol 90, Part 1, 1984. p. 107790.
[10] Nilsson HO, Holmer I. Comfort climate evaluation with thermal
well as simulations are visualised in new clothing manikin methods and computer simulation models. International
independent comfort zone diagrams, showing how an Journal of Indoor Air Quality and Climate (Munksgaard, Denmark)
average human being would perceive the whole body as vol. 13 2003:2837.
well as local climate. [11] Bedford T. The warmth factor in comfort at work. MRC Industrial
The results shows that the set of equations used in the Health Board Report, vol. 76. London, UK: HMSO; 1936.
[12] Nielsen PV. The selection of turbulence models for prediction of
simulations give good agreement with real life measure- room airow. ASHRAE Transactions 1998;104(part 1B):111927.
ments in the different environments. The use of input data [13] Star CD 3.1B. User guide and methology. Computational Dynamics
from CFD calculations produced reasonable results, Limited, /cd-adapco.comS, Manual, 2001.

Anda mungkin juga menyukai