See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/258139973

An Air Distribution Index for Assessing the Thermal Comfort and Air Quality
in Uniform and Nonuniform Thermal Environments

Article in Indoor and Built Environment · August 2013

DOI: 10.1177/1420326X12451186

CITATIONS READS

29 4,196

4 authors, including:

H.B Awbi Ehab Foda

University of Reading 15 PUBLICATIONS 357 CITATIONS
220 PUBLICATIONS 5,888 CITATIONS
SEE PROFILE
SEE PROFILE

Kai Sirén
Aalto University
112 PUBLICATIONS 4,638 CITATIONS

SEE PROFILE

All content following this page was uploaded by Ehab Foda on 29 December 2015.

The user has requested enhancement of the downloaded file.

 Original Paper
Indoor and Built Accepted: May 17, 2012
Environment Indoor Built Environ 2013;22;4:618–639

An Air Distribution Index for

Assessing the Thermal
Comfort and Air Quality
in Uniform and Nonuniform
Thermal Environments
I. Almesria,b H. B. Awbia E. Fodac K. Sirénc
a
School of Construction Management and Engineering, University of Reading, Reading, UK
b
Department of Mechanical Engineering, College of Technological Studies, Kuwait
c
Department of Energy Technology, Aalto University, Espoo, Finland

Key Words during their exposure to the MV and DV systems. The

Air distribution index E Room air distribution E results from CFD predictions were compared with
Ventilation systems E Local thermal comfort E Virtual measurements in the test chamber. The predicted
thermal manikin E Computational fluid dynamics (ADI)New parameters, such as the thermal comfort, the
ventilation effectiveness for both heat and contaminant
removals as well as the local mean age of air, were
Abstract compared with measured values in the chamber and
This study introduces a new Air Distribution Index found to be in good agreement. The results demon-
(ADI)New to assess the ventilation performance in strate that (ADI)New is a useful index for evaluating the
uniform and nonuniform thermal environments. The performance of a ventilation system under uniform
index comprises parameters for assessing the indoor thermal environment and can also be applied to
thermal comfort and air quality in occupied spaces. The nonuniform environment.
thermal comfort assessment was carried out using a
virtual thermal manikin that is adjusted by a model of
human thermoregulation and coupled with a psycho-
logical comfort model. The virtual manikin was used in Introduction
a computational fluid dynamics (CFD) code to simulate
tests in an environmental chamber that were largely The main purpose of a ventilation system is to provide
conducted under thermally neutral conditions with thermal comfort to occupants as well as improve the
mixing ventilation (MV) or displacement ventilation indoor air quality by removing pollutants generated in
(DV) systems. Eight human subjects were used in the rooms. In the past, these two parameters (i.e. thermal
study that included measurement of their skin tem- comfort and air quality) have been examined separately
perature, local and overall thermal sensation votes in evaluating an air distribution system performance.

ß The Author(s), 2012. Reprints and permissions:

http://www.sagepub.co.uk/journalsPermissions.nav H. B. Awbi,
DOI: 10.1177/1420326X12451186 School of Construction Management and Engineering, University of Reading,
Accessible online at http://ibe.sagepub.com Reading, UK.
Figures 1–8 and 10–17 appear in colour online E-Mail [email protected]
Downloaded from ibe.sagepub.com by guest on December 28, 2015
Many studies have investigated the performance of the local comfort under different conditions obtained from
ventilation systems for providing thermal comfort [1–3] the literature. Compared with subjective data, the results
and others have focused on the ability of a ventilation showed satisfactory predictions and were nearly in good
system in removing indoor pollutants [4–7]. However, a agreement with predictions using the equivalent tempera-
ventilation system may be effective in providing good air ture (teq) approach of Nilsson’s model [17].
quality but could be less effective in providing thermal In this study, local subjective votes (based on the
comfort and energy performance and vice versa. Bedford scale) from 45 different indoor conditions as
Therefore, it is beneficial to develop a method to assess given in two studies [17,18] were correlated with the
the ventilation performance in providing both thermal predicted skin temperatures using the MS-Pierce model
comfort and air quality. Awbi and Gan [8] and Awbi [9] under the same test conditions. This was used to develop a
developed the concept of the ventilation parameter (VP), local comfort model that was adapted from the UCB
which was later renamed the air distribution index (ADI) comfort model. The prediction of the overall thermal
[10,11]. The ADI represents a holistic approach for comfort was based on a weighted average of the local
assessing the thermal comfort, indoor air quality and comfort predictions. A virtual manikin was constructed
energy performance of an air distribution system. Since the and used in the computational fluid dynamics (CFD) code
thermal comfort part of the ADI was based on Fanger’s VORTEX 4.0 [19] to predict the room environment. The
PMV model [12], the ADI is applicable to and can be virtual manikin simulated the occupant’s presence in the
effective for evaluating the performance of a ventilation room and was thermally adjusted using the MS-Pierce
system that creates a uniform environment but it could be thermoregulation model. This simulation approach for the
less accurate for evaluating systems that create nonuni- prediction of the (ADI)New was then tested against
form thermal environments. However, in practice the measured data for two cases in an environmental test
majority of indoor environments are seldom uniform. chamber served by mixing and displacement ventilation
Hence, this study proposes a new air distribution index systems.
(ADI)New which can be applicable for the assessment of
ventilation systems that provide either uniform or nonuni-
form thermal environments.
The concept of local thermal comfort that corresponds Methods
to body segments underlies the thermal prediction for the
whole body and represents the current state-of-the-art in New Air Distribution Index (ADI)New
the prediction of the thermal comfort of individuals. The The proposed air distribution index (ADI)New combines
prediction of thermal comfort that is based on local effects the thermal comfort and air quality numbers as repre-
accounts for the nonuniformity in the thermal environ- sented by Equation (1) [20,21]:
ments. Generally, the prediction of local thermal comfort      
jSj n
is based on the local skin temperatures as stipulated in the ðADIÞNew ¼ 1   "t þ  "c ð1Þ
3 p
concept of the equivalent (homogeneous) temperature [13] |fflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflffl} |fflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflffl}
NT:C: NA:Q:
and the use of a comfort model such as the University of
California, Berkeley (UCB) model [14]. Therefore, the where NT.C. is the thermal comfort number, NA.Q. is the air
multisegmental physiological (thermoregulation) models quality number, jSj is the absolute value of the average
which can predict local skin temperatures along with other overall thermal sensation over the exposure time, "t is the
physiological variables have gained more importance with ventilation effectiveness for heat removal, n is the room
the local thermal comfort concept. Such models of human time constant, p is the local mean age of air and "c is the
thermoregulation can be used to predict the local skin ventilation effectiveness for contaminant removal. These
temperature, hence evaluating the local thermal comfort of parameters are calculated as shown by Equation (2):
individuals. This methodology has not yet been adopted
To  Ti C  Ci 1
by any of the international standards and guidelines and "t ¼ , "c ¼ o , n ¼ ð2Þ
Tm  Ti Cm  Ci ACH
needs further validations.
In an earlier study, the authors compared predictive where To , Ti and Tm are the temperature at the outlet, at
methods of the human skin temperature and thermal the inlet and the mean value in the occupied zone
sensations [15]. The multisegmental (MS) Pierce model respectively; Co , Ci and Cm are the contaminant (CO2)
[16] was coupled with the UCB comfort model to predict concentrations at the same locations and ACH is the room

A New Room Air Distribution Index Indoor Built Environ 2013;22:618–639 619

Downloaded from ibe.sagepub.com by guest on December 28, 2015

air change rate per hour. The local mean age of air is conditions and produced an average absolute skin
calculated using Equation (3): temperature deviation in the range of 0.3–0.8 K [15]. The
Z1 model combines useful features such as simplicity along
1
p ¼ Cp ðtÞdt ð3Þ with a good accuracy in estimating the local skin
Cð0Þ 0
temperatures.
where C(0) is the initial concentration of a tracer gas and In this study, the MS-Pierce model was used in the
Cp is the gas concentration at a certain point in the room
construction of the comfort model and for the regulation
(e.g. breathing zone).
of the virtual manikin in the CFD simulations.
The logic beyond developing the (ADI)New in the
manner shown in Equation (1) is that when the occupant’s
thermal sensation is neutral (i.e. jSj ¼ 0), which is the ideal Thermal Comfort Model
thermal condition, NT.C. reaches its maximum value and Model Construction
when jSj reaches its extreme values (i.e. 3 or þ3), NT.C. The overall thermal comfort is estimated for the
reaches its minimum value (zero). Also a high value of "t calculation of (ADI)New using a weighted average of the
implies that the ventilation system is efficient in removing local thermal comfort perceived by the different body
heat from the occupied zone. n , p and "c are important segments. The local comfort is predicted using a model
factors that are incorporated for evaluating the air quality that was adapted from the UCB local sensation model
number in (ADI)New and consequently assessing the air [14]. However, for the purpose of this research, the UCB
distribution performance in an enclosed space. A high
model’s proposed term for the impact of the overall
value of n and "c and a low value of p mean that the
sensation on body segments was not included for
ventilation system is good in removing contaminants as
simplicity. Hence, the adapted model accounts only for
well as providing fresh air to the occupied zone. Therefore,
the local skin temperature’s deviation from its set-point
the (ADI)New presented in the form shown in Equation (1)
value. In addition, a standard 7-point scale (Bedford scale)
could be a useful tool for evaluating both thermal comfort
was used instead of the extended 9-point scale that is used
(based on the local comfort concept) and air quality
with the UCB model (Figure 1). With this scale, when the
whether the thermal environment is uniform or nonuni-
local skin temperature deviates much from its set-point,
form. Although (ADI)New does not represent an absolute
the local comfort approaches its extreme values (i.e. 3,
measure of the performance of an air distribution system,
þ3). The Bedford scale was used in earlier studies to asses
the concept does however provide useful information on
both thermal sensation and comfort and may be seen in
how a system is performing in terms of thermal comfort
that sense as an inclusive good option to use for (ADI)New.
and indoor air quality provision. In principle, a high value
The proposed local thermal comfort model has the
of (ADI)New would mean good performance but in
following form as represented by Equation (4) [14]:
addition NT.C. and NA.Q. should have nearly equal values !
to ensure that the system provides thermal comfort and 2
LTC ¼ 3   1
indoor air quality equally. 1 þ EXP C  ðTskin, local  Tskin, local, set Þ
ð4Þ
The Multisegmental (MS) Pierce Model
The MS-Pierce model [16] was developed on the basis of where LTC is the local thermal comfort (based on Bedford
the original 2-node Pierce model [22] using: measured data 7-point scale); C is a coefficient with a value between 0 and
in neutral condition to adjust the local skin set-points and 1 that depends on the different body parts; Tskin,local is
to calculate the local core set-points (using a line search the local skin temperature (8C); and Tskin,local,set is its set-
method) that allows the model to predict the skin point (8C).
temperatures in the neutral condition with high accuracy; The regression coefficient C is obtained for each body
a modified calculation procedure for the convective heat segment from the correlation of the skin temperature
transfer coefficients; and adjustment to the heat transfer deviation from its set-point against the subjective local
term from core to skin using a common blood temperature comfort votes under defined conditions. In this study, this
along with the local core temperatures. The model was carried out using subjective votes (Bedford scale) from
predictability was verified for steady-state and dynamic two other studies [17,18] along with the predicted skin
conditions using measured data at uniform neutral, cold temperature’s deviation by the MS-Pierce model using the
and warm as well as several different asymmetric thermal test conditions from the same studies.

620 Indoor Built Environ 2013;22:618–639 Almesri et al.

Downloaded from ibe.sagepub.com by guest on December 28, 2015

Fig. 1. Bedford 7-point scale and the extended 9-point scale (UCB model).

Table 1. Local thermal comfort model coefficients and correlation with actual votes
Segment Regression coefficient, C Coefficient of Actual vs Predicted
determination (R2)

T50
T 4 0

Head 0.14 0.85 0.46 3

2

Predicted vote
1
0
-1
-2
-3
-3 -2 -1 0 1 2 3
Actual vote

U. Back 0.47 0.55 0.75 3

2
Predicted vote

1
0
-1
-2
-3
-3 -2 -1 0 1 2 3
Actual vote

Chest 0.4 1 0.79 3

2
Predicted vote

1
0
-1
-2
-3
-3 -2 -1 0 1 2 3
Actual vote

Continued

A New Room Air Distribution Index Indoor Built Environ 2013;22:618–639 621

Downloaded from ibe.sagepub.com by guest on December 28, 2015

Table 1. Continued
Segment Regression coefficient, C Coefficient of Actual vs Predicted
determination (R2)

T50
T 4 0

U. Arm 0.58 0.59 0.75 3

2

Predicted vote
1
0
-1
-2
-3
-3 -2 -1 0 1 2 3
Actual vote

L. Arm 0.72 0.55 0.69 3

2

Predicted vote
1
0
-1
-2
-3
-3 -2 -1 0 1 2 3
Actual vote

Hand 0.1 0.4 0.56 3

2

Predicted vote
1
0
-1
-2
-3
-3 -2 -1 0 1 2 3
Actual vote

Thigh 0.1 0.49 0.88 3

2
Predicted vote

1
0
-1
-2
-3
-3 -2 -1 0 1 2 3
Actual vote

L. Leg 0.1 0.83 0.75 3

2
Predicted vote

1
0
-1
-2
-3
-3 -2 -1 0 1 2 3
Actual vote

Foot 0.29 0.85 0.8 3

2
Predicted vote

1
0
-1
-2
-3
-3 -2 -1 0 1 2 3
Actual vote

622 Indoor Built Environ 2013;22:618–639 Almesri et al.

Downloaded from ibe.sagepub.com by guest on December 28, 2015

Table 1. Continued
Segment Regression coefficient, C Coefficient of Actual vs Predicted
determination (R2)

T50
T 4 0

L. Back 0.65 0.61 0.86 3

2

Predicted vote
1
0
-1
-2
-3
-3 -2 -1 0 1 2 3
Actual vote

Abdomen 1 1 0.86 3
2

Predicted vote
1
0
-1
-2
-3
-3 -2 -1 0 1 2 3
Actual vote

Overall NA NA 0.75 3
2

Predicted vote
1
0
-1
-2
-3
-3 -2 -1 0 1 2 3
Actual vote

Nilsson [17] carried out experiments for 30 different performed under different room air temperatures, differ-
climatic conditions using two thermal manikins to meas- ent temperature gradients between ankle and head, and
ure the heat flux for individual body segments and the different clothing ensembles. In total, 60 male and female
measured local heat flux was used to calculate the subjects participated in these tests (30 for each condition)
segmental equivalent temperature. The equivalent tem- in which they voted their thermal comfort perception on
peratures obtained under these conditions were correlated the Bedford scale.
with subjective votes from human subjects’ tests under the The environmental parameters (i.e. air temperature, air
same test conditions. The subjects (20 subjects for each velocity, radiant temperature and relative humidity) and
condition) reported their local and overall thermal the segmental clothing insulation values from the afore-
sensations on the Bedford scale. The individual votes mentioned studies [17,18] were entered into the MS-Pierce
were averaged for each test condition and reported as a thermoregulation model [16] to predict the local skin
mean thermal vote (MTV). temperatures. The predicted local skin temperatures were
Cheong et al. [18] conducted human subjects’ experi- correlated with the actual local comfort votes to obtain the
ments under 15 different test conditions. This was carried coefficient C for the different body segments. The
out in a climatic chamber to investigate the local and coefficient C for each body segment was obtained for
overall thermal sensation and comfort in environments two cases: when the skin temperature deviation
served with displacement ventilation. The tests were
Tð¼ Tskin,local  Tskin,local,set Þ50 and when
T 4 0.

A New Room Air Distribution Index Indoor Built Environ 2013;22:618–639 623

Downloaded from ibe.sagepub.com by guest on December 28, 2015

 Case 1 conditions (Foda and Sirén 2011)
Actual vote Predicted vote
1

0
Thermal vote

–1

–2

–3
Head
UR.Back
UL.Back
R.Chest
UR.Arm
LR.Arm
R.Hand
L.Chest
UL.Arm
LL.Arm
L.Hand
R.Thigh
R.Leg
R.Foot
L.Thigh
L.Leg
L.Foot
LR.Back
LL.Back
Abdomen
Overall
Fig. 2. Comparison between predicted votes and actual votes under Case 1 (slightly cool conditions).

Table 1 shows the regression coefficients obtained, plots of The test duration was 1 h during which the subjects
the predicted votes against the actual votes and their reported their segmental and overall thermal votes twice.
correlation coefficients (R2). The votes were on the scale given by Nilsson [17] in the
The overall thermal comfort (OTC) is then estimated comfort zone diagram which is similar to the Bedford scale
from the local thermal comfort (LTC) using the 7-point (Figure 1). Figures 2 and 3 show the predicted and actual
weighted average [23] as represented by Equation (5): votes under the conditions given in the two cases. As can
OTC ¼ 0:07  LTChead þ 0:35  LTCtrunk be seen, the model’s predictability based on the 7-point
weighting was, in general, very good for most body
þ 0:14  LTCarm þ 0:05  LTChand
segments and for the whole body. However, the model
þ 0:19  LTCthigh þ 0:13  LTCleg þ 0:07  LTCfoot predictability was slightly lower for the feet in Case 1
ð5Þ (slightly cool condition) and for the hands in Case 2
In Equation (5), the trunk local thermal comfort (slightly warm condition).
represents the average local comfort of the chest, back
and abdomen, whereas the local comfort of the extremities Laboratory Measurements
represent the average of the right and left segments. This
Experimental Setup
approach was used to account for thermal asymmetries in
The experiments were conducted in the University of
the space.
Reading’s environmental test chamber (Figure 4 (a)). The
working compartment of the chamber has dimensions
Model Validity 2.78 m  2.78 m  2.3 m ceiling height. Two ventilation
The model developed is applicable for the prediction of systems were tested using the chamber; mixing ventilation
thermal comfort in the physiological steady-state con- (MV) and displacement ventilation (DV). The mixing
dition under uniform or nonuniform environments. Prior ventilation supply diffuser (0.4 m width and 0.01 m height)
to the integration with the (ADI)New calculations, the is located at the front wall (the wall facing the occupant) at
model was verified against subjective data in a recent study a distance of 0.17 m below the ceiling. The air jet is
by Foda and Sirén [24]. In that study, 17 male subjects directed towards the ceiling using a 458 deflector
participated in the assessment of two thermally asym- (Figure 4(c)). The DV diffuser has a semi-cylindrical
metric conditions (Case 1 and Case 2). The two cases shape with a radius of 0.25 m and a height of 0.20 m. The
respectively represented slightly cool and slightly warm DV diffuser is mounted at floor level and located in the
conditions. The subjects had normal office work with their middle of the front wall (Figure 5). The test conditions
own portable computers during the whole test period. were the same for both ventilation systems with an air

624 Indoor Built Environ 2013;22:618–639 Almesri et al.

Downloaded from ibe.sagepub.com by guest on December 28, 2015

 Case 2 Conditions (Foda and Sirén 2011)
Actual vote Predicted vote
3

2
Thermal vote

–1
Head
UR.Back
UL.Back
R.Chest
UR.Arm
LR.Arm
R.Hand
L.Chest
UL.Arm
LL.Arm
L.Hand
R.Thigh
R.Leg
R.Foot
L.Thigh
L.Leg
L.Foot
LR.Back
LL.Back
Abdomen
Overall
Fig. 3. Comparison between predicted votes and actual votes under Case 2 (slightly warm conditions).

Fig. 4. (a) The environmental test chamber, (b) flow rate control panel and (c) the 458 deflector.

supply flow rate of 15 Ls1, air supply temperature of a Dantec air flow analyser (Model 54N10) with omnidir-
188C, RH  40 % and total room load of 21.2 Wm2 of ectional anemometers (accuracy 10% within the lower
floor area. range of the instrument) and CO2 concentrations were
The physical indoor environmental parameters as well measured using a 12-channel gas analyser type Brüel &
as the CO2 concentrations were measured during the Kjaer 1302 (accuracy approximately  12%). Air tempera-
experiments. The air temperatures were measured using tures and velocities were measured at the inlet and outlet
Platinum Resistance Thermometer (PRT) sensors with an and at the heights of 0.1, 1.1 and 1.8 m for a number of
accuracy of ( 0.15 K), air velocities were measured using locations in the occupied zone (Figure 5). For the stand

A New Room Air Distribution Index Indoor Built Environ 2013;22:618–639 625

Downloaded from ibe.sagepub.com by guest on December 28, 2015

 Locations of the stands:
Stand 1: X =0.6 m, Z =1.4 m
Stand 2: X = 1.37 m, Z = 0.62 m
Stand 3: X = 2.34 m, Z =1.4 m
Stand 4: X =1.2 m, Z = 2.18 m
Stand 5: X =1.2 m, Z =1.38 m

Fig. 5. The locations of temperature, velocity and CO2 measuring points in the test chamber.

located near the occupant (i.e. stand no. 5), the air activated and the measurements of SF6 concentration
temperature was measured at four different heights (0.1, decay commenced.
0.6, 1.1 and 1.6 m) while the air velocity was measured at
the heights of (0.1, 1.1 and 1.6 m). The CO2 concentrations Subjects
were measured at the inlet and outlet and at the heights of Eight college-aged and healthy subjects (4 male and 4
1.1 and 1.8 m for many locations in the occupied zone as female) participated in the experiments. All subjects
shown in Figure 5. The local mean age of air was measured participated in the two tests (with mixing ventilation and
at the breathing zone at a point located 20 cm from an displacement ventilation) on two separate days. The
occupant’s nose using a standard tracer gas decay subjects arrived at the test place 30 mins prior start of
technique with SF6 (sulphur hexafluoride) as a tracer the experiment to allow sufficient time for completion of
gas. Before injecting SF6, the ventilation system was shut the consent form and to take his/her physical meas-
down and the tracer gas was then injected in the chamber urements (height, weight and body fat). A summary of the
from an SF6 cylinder and a mixing fan was kept running physical measurements of the subjects is shown in Table 2.
for 5 mins to ensure a good mixing of SF6 in the working After entering the chamber, the participants were allowed
compartment. After that, the ventilation system was to acclimatize to the thermal environment created by the

626 Indoor Built Environ 2013;22:618–639 Almesri et al.

Downloaded from ibe.sagepub.com by guest on December 28, 2015

Table 2. Anthropometric measurements of the subjects and the standard deviation
Gender No. of subjects Age (year) Height (m) Weight (kg) Body fat (%)

Male 4 35  (5.48) 1.73  (0.06) 73.28  (14.21) 23.03  (12)

Female 4 30.25  (5.91) 1.65  (0.098) 67.6  (12.54) 28.68  (6.57)
Male & female 8 32.63  (5.85) 1.69  (0.085) 70.44  (12.77) 25.85  (9.45)

The values within brackets represent the standard deviation.

Fig. 6. Different numerical grids used for: (a) DV system (b) MV system.

A New Room Air Distribution Index Indoor Built Environ 2013;22:618–639 627

Downloaded from ibe.sagepub.com by guest on December 28, 2015

Fig. 6. Continued.

ventilation systems for 30 mins before commencing the based on the 7-point ASHRAE scale. All participants
tests. In the main tests, the subjects sat at a desk facing wore an ensemble of typical office clothing consisting
the diffuser (Figure 5) and performed sedentary of long sleeve shirt, cotton undershirt (T-shirt),
activities such as reading or using a laptop and were trousers, pants and athletic socks. Female subjects
not allowed to move inside the chamber. The subjects wore bra instead of underwear cotton T-shirt. The
were exposed to the thermal environment for 2 h and estimated overall clo value for the male clothing
had been asked to fill in a questionnaire every 15 mins ensemble was 0.72 while the estimated one for the
to rate their local and overall thermal sensation levels female clothing ensemble was 0.64.

628 Indoor Built Environ 2013;22:618–639 Almesri et al.

Downloaded from ibe.sagepub.com by guest on December 28, 2015

 Air Velocity (DV System)
0.160
Measured
0.140 Predicted (DV grid 1)

Predicted (DV grid 2)

0.120
Predicted (DV grid 3)
Air Velocity (m/s)
0.100

0.080

0.060

0.040

0.020

0.000
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Velocity Sensors

Air Velocity (MV System)

0.160
Measured
0.140 Predicted (MV grid 1)

Predicted (MV grid 2)

0.120
Predicted (MV grid 3)
Air Velocity (m/s)

0.100

0.080

0.060

0.040

0.020

0.000
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Velocity Sensors
Fig. 7. The measured and predicted velocities using different numerical grids for the DV and MV systems.

CFD Simulations transfer, concentration and mean age of air distribution in

CFD technique has progressed during the last few indoor environments. The code uses the standard k  "
decades and has been proven to be a powerful and efficient and the Renormalization Group (RNG) k  " turbulence
tool for simulating air flow and contaminant dispersion in models and has been developed for ventilation research,
indoor environments [10]. The numerical calculations to which may be more suitable for ventilation simulation
predict the airflow properties in the chamber were carried than other general-purpose codes.
out using the CFD code VORTEX 4.0 [19]. This program For the CFD simulations, three different numerical
has been developed for the simulation of airflow, heat grids have been tried for both the DV and MV systems;

A New Room Air Distribution Index Indoor Built Environ 2013;22:618–639 629

Downloaded from ibe.sagepub.com by guest on December 28, 2015

 Mass residual error
140
DV grid 1
120
MV grid 1
100

Mass residual error 80

60

40

20

0
1

119

237

355

473

591

709

827

945

1063

1181

1299

1417

1535

1653

1771

1889
Iteration No.
Fig. 8. Mass residual errors for DV grid 1 and MV grid 1.

DV grid 1, DV grid 2, DV grid 3, MV grid 1, MV grid 2

and MV grid 3. All these grids are 3-D structured
Cartesian grids with 80*80*80 points (in x, y and z
direction) for DV grid 1, DV grid 2, MV grid 1 and MV
grid 2, 66*67*63 points for DV grid 3 and 62*63*57 points
for MV grid 3. Although DV grid 2 and MV grid 2 have
the same grid points as DV grid 1 and MV grid 1 (i.e.
80*80*80), they have a different configuration (Figure 6).
When comparing the predicted air velocities using the
grids mentioned earlier with the measured ones (Figure 7)
it is quite obvious that DV grid 1 and MV grid 1 provide
better results than the other grids. Also, the solutions
converged better when using DV grid 1 and MV grid 1
(Figure 8). Therefore, DV grid 1 and MV grid 1 have been
Fig. 9. The simulated DV inlet diffuser used in VORTEX code.
used in the current study to simulate the air flow pattern in
the environmental test chamber that was created by the
two ventilation systems.
The equations for the momentum, energy, concentra-
tion, kinetic energy and turbulent dissipation rate are the experiments. It is worth mentioning that the calculated
discretised using a HYBRID scheme. The pressure– flow rate using the simulated MV inlet diffuser is exactly
velocity coupling algorithm SIMPLE was used to solve the same as the measured one. For the DV inlet diffuser,
the continuity equation and the RNG k  " model was since it is not possible to use a curved grid in the VORTEX
used to represent the turbulent behaviour of the flow code to represent the semi-cylindrical shape, the DV inlet
within the chamber. The near wall nodes were located was simulated using a rectangular opening. To allow for
5 mm from the wall surfaces as this is the optimum the radial air flow distribution for this diffuser, the
distance to use as recommended by Awbi [25]. rectangular opening was divided into three sections as
The geometry of the MV inlet used in the CFD shown in Figure 9. For the middle section, the direction of
simulations is a simple rectangular opening which has a the air flow is perpendicular to the inlet surface while for
width of 0.4 m and a height of 0.01 m. This geometry is the left and the right sections, the air flow is inclined by an
exactly the same as the MV inlet slot used in angle of 458 to the central part of the diffuser.

630 Indoor Built Environ 2013;22:618–639 Almesri et al.

Downloaded from ibe.sagepub.com by guest on December 28, 2015

 area (m2). The body surface area is calculated using
Equation (7) [26]:
A ¼ 0:203  Ht0:725  Wt0:425 ð7Þ
where Ht is the body height (m) and Wt is the body weight
(kg). In this study, the average height and weight for the
occupants was 1.69 m and 70.44 kg, respectively. The
estimated CO2 production per person used in the CFD
simulations was 0.00464 Ls1 based on assumed meta-
bolic rate for the sedentary activity (i.e. 1.1
met ¼ 64 Wm2).
The data for the boundary conditions that were based
on laboratory measurements and used in the simulations
Fig. 10. The virtual manikin used in the CFD simulations.
related to the temperatures of the chamber’s six surfaces;
inlet air velocity, temperature, CO2 concentration and
turbulence intensity; appliance’s heat flux; and the light
Table 3. The surface area of the individual fixture’s surface temperature. The boundary conditions
segments of the virtual manikin for the virtual manikin were predicted using the MS-Pierce
Body part Surface area (m2)
model and were assigned to each body segment. This
included the segmental heat fluxes and the clothed body
Head 0.1596 temperatures. In order to predict these quantities in the
Chest 0.2328
concurrent simulations with the shortest time possible,
Back 0.2388
Pelvis 0.2460 the MS-Pierce thermoregulation model was fed with the
Upper arm (2 segments) 0.1671 measured environmental physical parameters at different
Lower arm (2 segments) 0.1298 heights (i.e. air temperature, air velocity, radiant tempera-
Hand (2 segments) 0.1045
Thigh (2 segments) 0.3757 ture, relative humidity) together with the personal data
Leg (2 segments) 0.2590 (i.e. clothing insulation and metabolic rate).
Foot (2 segments) 0.1293
Whole body 2.0426

Results and Discussion

Thermal Comfort
The geometry of the simulated DV diffuser, which has The predicted local comfort votes from the simulations
been adjusted to give the same flow rate as the measured along with the actual subjective votes are shown in Figures
one, has a width of 0.5 m and a height of 0.09 m. 11 and 12 for the two cases of the DV and MV systems
The occupant’s presence in the room was simulated respectively. The figures show the predicted and actual
using a virtual thermal manikin (Figure 10). The virtual votes for the head, chest, back, pelvis and the right-side
manikin has a cubical shape and formed, as much as body limbs as well as for the whole body. The differences
possible, to have the same size and area as those between the actual and predicted votes under the test
corresponding to the thermal manikin ‘‘Therminator’’ conditions by the DV and the MV systems were not
[24]. The total surface area of the clothed manikin is significant for most body segments. The actual votes were
2.043 m2. Table 3 shows the surface area of the individual the average from all subjects’ votes (4 male and 4 female).
segments of the virtual manikin. The rate of CO2 The test conditions in the chamber for both ventilation
production in humans used in the CFD simulations was methods produced almost neutral conditions (Tm ¼ 25.58C
estimated using Equation (6) [10]: and 25.38C for DV and MV, respectively). Therefore, the
difference in the average overall sensation votes for the
G ¼ 4  105  M  A ð6Þ two test conditions was not significant and was nearly
consistent with the planned condition (i.e. neutral).
where G is the CO2 production per person (Ls1), M is the However, in the DV tests the subjects felt slightly cooler
metabolic rate (Wm2) and A is the body surface especially for the lower extremities when compared with

A New Room Air Distribution Index Indoor Built Environ 2013;22:618–639 631

Downloaded from ibe.sagepub.com by guest on December 28, 2015

 DV system
Subjective votes Predicted located in the middle of the chamber where the manikin is
3 seated. There was no sign of draft observed around the
manikin or in the occupied zone for both ventilation
Local comfort

2
1
0
systems. The only high air velocity region was near the
–1 ceiling for the case of mixing ventilation. This region of
–2 high air velocity was generated by the air jet supply from
–3
the wall diffuser which then decreased as the flow
Head

Chest

Back

Pelvis

R.U.arm

R.L.arm

R. hand

R. thigh

R. leg

R. foot

Overall
progressed towards the opposite wall. However, to observe
Fig. 11. Predicted and actual votes for the exposure to the DV the plumes around the manikin, the contour plot scale was
system. selected to be from 0 to 0.2 ms1.
Figure 14 shows the air temperature contours at planes
of heights of 0.1, 0.6 and 1.1 m. It is clear from Figure 14
that the air temperatures for the DV system at these levels
were in general lower than those for the MV system. The
MV system
predicted air temperatures at these levels (0.1, 0.6 and
Subjective votes Predicted
1.1 m) and 0.1 m away from the corresponding body
3
2 segments (i.e. ankle, abdomen and head) were 22.4, 27.2
Local comfort

1 and 26.88C respectively for the DV system and 25.1, 28.6

0
and 278C respectively for the MV system. The thermal
–1
–2 plumes from heat sources (i.e. manikin and laptop) are
–3 shown in Figure 15. As can be seen, the lower air
Head

Chest

Back

Pelvis

R.U.arm

R.L.arm

R. hand

R. thigh

R. leg

R. foot

Overall

temperatures were near the inlet diffusers which then

increased gradually until they reached maximum values
Fig. 12. Predicted and actual votes for the exposure to the MV around the heat sources.
system.
Figure 16 shows the CO2 concentration in (ppm) at a
plane located in the middle of the chamber. The CO2
concentration in the occupied zone was more stratified and
showed lower values for the DV system case compared
the MV tests and this consequently slightly affected their with the MV system. Clearly this suggests that the DV
overall sensations. This may be related to the temperature system provided better air quality in the occupied zone
stratification in the chamber where the lower parts of the than in the MV system. The maximum CO2 concentration
subjects’ bodies were subjected to lower temperatures. for both systems was located at the breathing zone which
The predicted votes were in good agreement with the is the source of the CO2 production.
actual votes for most body parts under the two test A comparison between the predicted and measured air
conditions (DV and MV systems). The maximum devi- temperatures, velocities and CO2 concentrations at differ-
ation was 51 on the comfort scale at the pelvis ent locations in the occupied zone is shown in Figure 17.
segment for the exposure with the DV system and at The air temperature, velocity and CO2 sensors are located
the foot segment for the exposure with the MV system. on the five stands shown in Figure 5.
The subjective votes in these tests were based on As can be seen from Figure 17, the predicted and
the ASHRAE 7-point scale while the comfort model measured air velocities for DV system were generally in
was based on the Bedford 7-point scale. However, good agreement with some minor discrepancy especially
the difference between the two scales is believed to at the measuring point nos 1, 2, 4 and 11. In contrast,
be very minor in the range close to the neutral sensations the discrepancy between the predicted and measured air
(i.e. from 1 to 1) which is perceived under these two velocities is noticeable for the MV system at some
exposures. measuring points. However, for the locations where the
measured air velocity is lower than 0.1 ms1, the
Room Physical Parameters discrepancy could be due to the velocity sensors’ large
The predicted air velocity, temperature and CO2 uncertainty for velocities below 0.1 ms1. Whilst for the
concentration profiles are presented in Figures 13–16. locations where the measured air velocity is greater
Figure 13 shows the air velocity profile in a vertical plane than 0.1 ms1 (e.g. measuring point nos 1, 4 and 13

632 Indoor Built Environ 2013;22:618–639 Almesri et al.

Downloaded from ibe.sagepub.com by guest on December 28, 2015

Fig. 13. Velocity contour plots at a plane located in the middle of the chamber (a) DV system (b) MV system.

for the MV system, which are located a distance of the range from 0.13% to 11.09% for the DV system and
0.1 m from the floor surface), the discrepancy could be from 0.78% to 10.4% for the MV system.
attributed to disturbance of the air flow at these In general, the minor discrepancies between the
locations of the test chamber resulting from the dense predicted and measured quantities may be related to the
sensors’ wiring close to the floor level. simplifications in the modelling of the real enclosure and/
The predicted and measured air temperatures were in or due to measurement errors.
good agreement. The difference between the predicted and
measured temperatures was in a range from 0.08 to Assessment of the Performance of Ventilation Systems
0.97 K for the DV system and from 0.03 to 0.8 K for the Using (ADI)New
MV system. While most of the predicted CO2 concentra- The measured mean air temperature in the occupied
tions were overestimated, the percentage difference zone, which was used to calculate "t , represents an average
between measured and predicted concentration was in of the measurements from 15 temperature sensors

A New Room Air Distribution Index Indoor Built Environ 2013;22:618–639 633

Downloaded from ibe.sagepub.com by guest on December 28, 2015

Fig. 14. Temperature contour at different heights (0.1, 0.6 and 1.1 m) for both the DV and MV systems.

634 Indoor Built Environ 2013;22:618–639 Almesri et al.

Downloaded from ibe.sagepub.com by guest on December 28, 2015

Fig. 15. Thermal plumes from heat sources for the DV and MV systems.

distributed in the occupied zone for the last 1 h of the this is related to its better performance in heat removal
recorded data. Similarly, the measured mean CO2 from the occupied zone (represented by "t ) compared
concentration, which was used to calculate "c , represents with the MV system. Moreover, for the DV system
an average of the measurements by seven CO2 sampler case, the air flow was supplied directly to the occupied
points in the occupied zone for the last 1 h of the recorded zone close to the floor level which was then entrained
data. by plumes rising from the heat sources in the chamber
Although the occupants’ thermal sensation from the due to buoyancy. This air flow mechanism of the DV
subjective votes (jSj) for the MV system was slightly system makes it more efficient in removing contami-
better than that for the DV system, the DV system nants (represented by "c ) from the occupied zone as
produced better thermal comfort number (NT.C.) and well as delivering fresh air to the breathing zone in less

A New Room Air Distribution Index Indoor Built Environ 2013;22:618–639 635

Downloaded from ibe.sagepub.com by guest on December 28, 2015

Fig. 16. CO2 concentration in (ppm) at a plane located in the middle of the chamber (a) DV system, (b) MV system.

time than that for the MV system. As a result, the DV systems along with the other parameters that were based
system showed better air quality number (NA.Q.) on the predicted and measured (or subjective) quantities.
compared with the MV system and consequently As can be seen from Table 4, the differences between the
better (higher) (ADI)New value as shown in Table 4. predicted and measured quantities used for the calculation
The differences between the DV and MV systems found of (ADI)New were not significant. Consequently, the
in this study are in general agreement with those found differences between the (ADI)New values based on the
in previous studies [2,27,28]. CFD predictions and those based on the measured (or
Table 4 shows the thermal comfort and air quality subjective) quantities were minor. Based on this compar-
numbers used in calculating (ADI)New for both ventilation ison, therefore, it may be concluded that the simulation

636 Indoor Built Environ 2013;22:618–639 Almesri et al.

Downloaded from ibe.sagepub.com by guest on December 28, 2015

 Air Velocity (DV system) Air Velocity (MV system)
0.160 0.16
Air Velocity (m/s)

Air Velocity (m/s)

0.120 0.12

0.080 0.08

0.040 0.04

0.000 0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Velocity Sensors Velocity Sensors

Air Temperature (DV system) Air Temperature (MV system)

29 29

Air Temperature (°C)

Air Temperature (°C)

27 27

25 25

23 23

21 21

19 19
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Temperature sensors Temperature Sensors

CO2Concentration (MV system)

CO2Concentration (DV system)
1000 1000
CO2conc. (ppm)

900 900
CO2conc. (ppm)

800 800

700 700

600 600

500 500
1 2 3 4 5 6 7 1 2 3 4 5 6 7
CO2Sampling point
CO2Sampling point

Fig. 17. The predicted and measured air velocities, temperatures and CO2 concentrations for the DV and MV systems.

Table 4. Comparison between measured and calculated air distribution index (ADI)New.
System jS j "t NT:C: n (h) p (h) "c NA:Q: ðADIÞNew

MV Measured/Subjective j0:14  0:51j 0.95 0.91 0.33 0.57 1.04 0.60 1.51
Predicted j0:31j 0.96 0.86 0.33 0.59 0.91 0.51 1.37
DV Measured/Subjective j0:43  0:38j 1.13 0.97 0.33 0.49 1.10 0.74 1.71
Predicted j0:07j 1.14 1.11 0.33 0.66 1.10 0.55 1.66

The values after () sign represents the standard deviation.

The predicted |S| is calculated using the proposed overall thermal comfort model (OTS) (Equation (5)).

A New Room Air Distribution Index Indoor Built Environ 2013;22:618–639 637

Downloaded from ibe.sagepub.com by guest on December 28, 2015

tool used in this study provides a sufficiently accurate The CFD simulations were carried out to calculate the
prediction for the assessment of ventilation systems. parameters used in calculating the value of (ADI)New, such
as the ventilation effectiveness for heat removal ("t ), the
ventilation effectiveness for contaminant removal ("c ) and
Conclusions the local mean age of air (p ). In general, the predicted
parameters were in good agreement with the measured
This study introduced a new air distribution index ones. Since the experimental measurements are quite
(ADI)New to assess the indoor thermal comfort and air expensive and time consuming, the implementation of
quality that is applicable for uniform and nonuniform CFD with a thermoregulation thermal comfort model is
thermal environments. The overall thermal comfort faster and more cost-effective and can be considered as a
assessment used in calculating (ADI)New, was based on powerful tool for assessing the performance of ventilation
the local thermal comfort concept. This was carried out systems. The initial validation of this combination of
using a weighted average of the local thermal comfort simulation tools (i.e. CFD and thermoregulation thermal
perceived by the different body segments. The local comfort models) is promising. However, further valida-
thermal comfort was predicted using a model that was tions are needed before extensive application of this
adapted for use with the MS-Pierce thermoregulation simulation tool could be made. Although the analysis
model and based on the structure of the UCB comfort was based on a relatively small test chamber, the new
model. This index was used in the present study to index would also be applicable to larger spaces represent-
compare the performance of two well-known ventilation ing actual buildings where variations in indoor environ-
systems, MV and DV systems, with both CFD simulations ment parameters could be larger than what was
and laboratory-based measurements. Based on the pre-
experienced in this chamber study. For such large indoor
dicted and the measured data, the DV system showed
spaces with many occupants, (ADI)New should be calcu-
better thermal comfort and air quality numbers compared
lated at different locations.
with the MV system and consequently better (higher)
(ADI)New value.

References

1 Gan G: Numerical investigation of local 8 Awbi HB, Gan G: Evaluation of the overall 16 Foda E, Sirén K: A new approach using Pierce
thermal discomfort in offices with performance of room air distribution: in two-node model for different body parts: Int J
displacement ventilation: Energ Build 1995; 23: Proceedings of Indoor Air’93, The 6th Biometeorol 2011; 55(4):519–532.
73–81. International Conference on Indoor Air 17 Nilsson H: Comfort climate evaluation with
2 Lin Z, Chow TT, Fong KF, Wang Q, Li Y: Quality and Climate, 4–8 July 1993, Helsinki, thermal manikin methods and computer simu-
Comparison of performances of displacement Finland; Vol 5, pp. 283–288. lation models. PhD thesis, University of Gävle,
and mixing ventilations. Part I: Thermal com- 9 Awbi HB: Energy efficient room air distri- Sweden, 2004.
fort: Int J Refrig 2005; 28: 276–287. bution: Renew Energ 1998; 15: 293–299. 18 Cheong KWD, Yu WJ, Sekhar SC, Tham
3 Pan CS, Chiang HC, Yen MC, Wang CC: 10 Awbi HB: Ventilation of Buildings. KW, Kosonen R: Local thermal sensation and
Thermal comfort and energy saving of a London, Spon Press, Taylor & Francis comfort study in a field environment chamber
personalized PFCU air-conditioning system: Group, 2003. served by displacement ventilation system in
Energ Build 2005; 37: 443–449. 11 Karimipanah T, Awbi HB, Sandberg M, the tropics: Build Environ 2007; 42: 525–533.
4 Yang X, Srebric J, Li X, He G: Performance of Blomqvist C: Investigation of air quality, 19 Awbi HB: VORTEX: A computer code for
three air distribution systems in VOC removal comfort parameters and effectiveness for two airflow, heat transfer and concentration in
from an area source: Build Environ 2004; 39: floor-level air supply systems in classrooms: enclosure. Version 4D-RNG, Reading, UK,
1289–1299. Build Environ 2007; 42: 647–655. 2005.
5 Cheong KWD, Yu WJ, Tham KW, Sekhar SC, 12 Fanger PO: Thermal Comfort. Copenhagen, 20 Awbi HB: Energy efficient ventilation for
Kosonen R: A study of perceived air quality and Danish Technical Press, 1970. retrofit buildings: in Proceedings of
sick building syndrome in a field environment 13 Wyon DP, Larsson S, Forsgren B, Lundgren I: 48thAiCARR International Conference on
chamber served by displacement ventilation Standard procedures for assessing vehicle Energy Performance of Existing
system in the tropics: Build Environ 2006; 41: climate with a thermal manikin: SAE Paper Buildings, 22–23 September 2011, Baveno,
1530–1539. 890049, 1989. Italy, pp. 23–46.
6 Dascalaki EG, Lagoudi A, Balaras CA, Gaglia 14 Zhang H: Human thermal sensation and 21 Almesri I, Awbi HB: Evaluation of air distri-
AG: Air quality in hospital operating rooms: comfort in transient and non-uniform thermal bution systems using a new air distribution
Build Environ 2008; 43: 1945–1952. environments. PhD thesis, The University of index: in Proceedings of Indoor Air 2011, 12th
7 Cho Y, Awbi HB, Karimipanah T: Theoretical California, Berkeley, CA, 2003. International Conference on Indoor Air
and experimental investigation of wall confluent 15 Foda E, Almesri I, Awbi HB, Sirén K: Models Quality & Climate, 5–10 June 2011, Austin,
jets ventilation and comparison with wall of human thermoregulation and the prediction Texas, USA, Paper a1004-5.
displacement ventilation: Build Environ 2008; of local and overall thermal sensations: Build 22 Gagge AP, Fobelets AP, Berglund LG: A
43: 1091–1100. Environ 2011; 46: 2023–2032. standard predictive index of human response

638 Indoor Built Environ 2013;22:618–639 Almesri et al.

Downloaded from ibe.sagepub.com by guest on December 28, 2015

 to the thermal environment: ASHRAE Trans 25 Awbi HB: Calculations of convective heat ment and mixing ventilations. Part II: Indoor
1986; 92(2): 709–731. transfer coefficients of room surfaces air quality: Int J Refrig 2005; 28: 288–305.
23 Hardy JD, DuBois EF: The technique of for natural convection: Energ Build 28 Cho Y, Awbi HB, Karimipanah T: A compar-
measuring radiation and convection: J Nutr 1998;28:219–27. ison between four different ventilation systems:
1938; 15(5): 461–475. 26 Dubois D, Dubois EF: A formula to estimate in Proceedings of Roomvent 2002, the 8th
24 Foda E, Sirén K: A thermal manikin with a the approximate surface area if height and International Conference on Air Distribution
human thermoregulatory control – weight be known: Arch Intern Med 1916; in Rooms, 8–11 September 2002, Copenhagen,
Implementation and validation: Int J 17:863–871. Denmark, pp 181–184.
Biometeorol 2011. DOI: 10.1007/s00484-011- 27 Lin Z, Chow TT, Fong KF, Tsang CF, Wang
0506-6. Q: Comparison of performances of displace-

A New Room Air Distribution Index Indoor Built Environ 2013;22:618–639 639

Downloaded from ibe.sagepub.com by guest on December 28, 2015

View publication stats