Anda di halaman 1dari 12

CH-09-049

How to Simplify Computer Simulated


Persons (CSPs) for Modeling
Personal Microenvironments:
Comparison and Case Studies
Wei Yan

Xudong Yang

Student Member ASHRAE

Member ASHRAE

ABSTRACT
Different computer simulated persons (CSPs) have been
developed to represent occupants for indoor airflow simulation using computational fluid dynamics (CFD). Simple CSPs
are preferred in order to avoid very fine grids and long computing time. However, how CSPs with different complexity affect
the accuracy of CFD simulation is not well studied yet. This
paper intends to investigate quantitatively the simulation
accuracy due to CSP simplification. A detailed, human-like
CSP and three different simplification strategies including an
overly simplified rectangular box CSP, a simple CSP using
several cuboids to represent different parts of a person, and an
improved one in which the total heat of different parts (head,
body, arms, legs) matches that of the detailed CSP are
proposed. The simulation results using simplified CSPs are
compared with the benchmark experimental data and simulated results using the detailed CSP in a displacement ventilation case. Results show that all the CSPs, simple or complex,
yield acceptable simulation results for the global field study.
For the micro-environment around the person, there seems no
significant benefit by adopting very complex CSP geometry
either. Instead, attention should be paid to ensure the heat and
pollutant source consistence of the simplified CSP over the
actual human (detailed CSP).
INTRODUCTION
Different computer simulated persons (CSPs) have been
developed to represent occupants for indoor airflow simulation using computational fluid dynamics (CFD). These CSPs
range from very complex, human-like configuration to very
simple, box-like shape. Complicated CSPs usually require
large number of computational grids and long computing

Ming Shan

time. On the other hand, overly simplified CSPs may significantly influence the simulation accuracy of personal microenvironment, especially for non-mixing type ventilation such
as personal or displacement ventilation. A guide to properly
simplify CSPs is urgently necessary for current research.
In the past, different CSPs have been used in room airflow
or indoor air quality models. He and Yang (2005) compared
the contaminant removal by displacement ventilation (DV)
and mixing ventilation based on experiments and CFD simulations. Cubic box was used to represent the persons and
equipment in their study. Zhang et al. (2005) also used cubic
CSPs to evaluate the thermal comfort and indoor air quality in
classrooms, retail shops, and industrial workshops. Hayashi et
al. (2002) adopted a person-like CSP to investigate the effect
of inhalation under DV chamber. The CSP, though more
complex than cuboid box, is still a simplified one. Later, the
same CSP was applied by Zhang et al. (2005) to study the
respiration area caused by unsteady breathing. Besides these
simplified CSPs, very complex, detailed CSPs were also used
in previous research. Bjorn and Nielsen (2002) developed
detailed CSPs to investigate the dispersal of exhaled air in DV
rooms. The simulation results of a seated detailed CSP and a
cuboid CSP were compared by Topp and Nielsen (2002) and
Sideroff and Dang (2005). They concluded that a simple
geometry was sufficient for studying global airflow, but
detailed geometry should be applied to assess the local flow
condition. However, they did not mention to what complexity
the CSP must be in order to accurately assess local flow near
the person. Murakami (2004) developed a standing detailed
CSP to analyze microclimates around the human body. This
CSP was generated according to a real human body, thus it is
supposed to be more accurate than pervious simplified CSPs.

Wei Yan is a doctoral student at the University of Illinois at Urbana-Champaign. Ming Shan is a research assistant and Xudong Yang is the
Chang-Jing Professor in the Department of Building Science, Tsinghua University, Beijing, China.

2009 ASHRAE

473

Gao and Niu (2004) developed a seated, detailed CSP using


laser scanning technology. They applied it to investigate the
performance of a personal ventilation system. A seated
detailed CSP was adopted by Sorensen et al. (2003) to study
the radiative heat transfer at different parts of the CSP. They
divided their CSP into foot, leg, thigh etc. There were totally
10 different parts. The heat transfer coefficient of each part
was estimated based on their simulation result.
Detailed CSPs require fine, and usually unstructured
grids to mesh the geometry accurately. A very simple case
such as a cubic room with a single occupant may need more
than 105-106 grids to ensure the accuracy. And the calculation
time can be up to several days for a single node personal
computer (for example, the detailed CSP case with 400,000
grids running on a P4 2.6GHz PC with 1G memory). Therefore, it is usually too time consuming to use such detailed
CSPs. Simple CSPs are always preferred if they would not
significantly affect the accuracy of simulation. Under what
circumstances such simple CSPs could be used, and how to
simplify the CSPs? These problems have not been well studied yet.
The objective of this paper is to investigate quantitatively
the simulation error due to CSP simplification. Different CSP
simplification strategies are proposed. The simulation results
by applying these strategies are compared with those by using
the detailed CSP in a displacement ventilation case, the latter
being validated by the benchmark experimental data (Kato
and Yang, 2003).
CSP SIMPLIFICATION METHODOLOGY
Both the CSP geometry and heat source distribution may
affect the airflow and temperature field when using simplified
CSPs, especially for the personal microenvironment studies.
Four kinds of CSPs are compared in this paper: (1) Laser
scanned CSP (Detailed CSP); (2) Geometry consistent CSP.
(Simple CSP); (3) Rectangular box CSP (Rectangular CSP),
and (4) Both geometry and heat source distribution consistent
CSP (Improved CSP). These CSPs are illustrated in Figure 1.
The detailed CSP (Figure 1a) is obtained by laser scanning technology based on the real manikin used in the experiment. Its geometry is exactly the same as the manikin used in
the benchmark experiment (Kato and Yang, 2003), and it is
reasonable to assume that the heat flux is uniformly distributed
on the CSP surface.
The simple CSP (Figure 1b) uses several cuboids to represent different parts of a person (head, arms, body, legs etc). It
tries to represent the human body but with simplified way in
terms of the geometry consistency. The heat flux is also
assumed to be uniformly distributed on the CSP surface.
The rectangular CSP (Figure 1c) is the most simplified
manikin by using a single rectangular box to represent the
whole person, and also assuming uniform heat flux distribution on the CSP surface. It does not look like a real person
geometrically, and the heat distribution on different body
parts (thus the thermal plume) is not the same as a real person
474

either. Because this strategy makes the grid generation and


modeling work very easy, it has been widely applied by many
researchers.
The improved CSP has exactly the same geometry as the
simple CSP (Figure 1b), but the heat flux distribution on the
CSP surface is re-calculated so that the total heat of different
parts (head, body, arms, legs) matches that of the detailed CSP.
In other words, the heat flux on the CSP surface is not uniform
any more. The intension is trying to match with the detailed
CSP in terms of both geometry (in a simplified way) and buoyancy (in a consistent way). By doing so, higher simulation
accuracy would be expected than a simple or rectangular CSP.
COMPUTATIONAL FLUID DYNAMICS
METHODOLOGY
Computational Grids
The simulation cases are based on the experiments done
by Kato and Yang (2003) shown in Figure 2. Commercial CFD
software Gambit 2.1 is used to generate the grids shown in
Figure 3. The CSP is put in the center of the chamber. Since the
chamber is symmetrical, only half of the chamber is simulated.
The entire chamber is divided into two computational
domains: (1) The inner domain is rectangular volume
surrounding the geometry of the CSP. The grids in this region
are tetrahedrons; (2) the outer volume is meshed with structured cuboids. The surface connecting these two domains is
defined as grid interface. The total number of grids in the
outer region for all cases is 97362. The numbers of grids in
inner region are 315397, 212374 and 181636 respectively for

Figure 1 Different CSPs for numerical simulation.


(a) Detailed CSP, (b) Simple and Improved CSPs,
(c) Rectangular CSP.
ASHRAE Transactions

the detailed CSP, simple CSP and rectangular CSP. The


improved CSP has the identical grids with the simple CSP. Our
previous studies (Yang et al., 2004; He et al., 2005) show that
such grid distributions are sufficient to reach grid independence status for the displacement ventilation case.
Turbulence Model
Sorensen and Nielsen (2003) and Chen and Srebric
(2002) gave good recommendations on how to control the
quality of CFD for indoor air quality simulation. Proper turbulence model should be selected according to the airflow situ-

ation in the room and the requirement of the simulation. For


room airflow, Chen (1995, 1996) compared the performance
of several commonly used turbulence models and recommended the Re-Normalization Group (RNG) k- model
(Yokhot et al. 1992). This model produced good simulation
results for indoor airflow, temperature, and contaminant distributions in several previous studies (Yuan et al. 1999; Srebric
and Chen 2002; Gao et al 2004; He et al 2005). Encouraged by
these results, RNG k- model together with logarithmic wall
functions (Launder and Spalding 1974) is used in current
study to simulate the airflow field
Boundary Conditions
The benchmark experimental setup (Kato and Yang,
2003) is shown in Figure 2. The dimension of the chamber is
3.0m3.5m2.5m. The air is supplied from the bottom of the
wall with the opening size of 0.4m0.2m. The exhaust opening is located at the top of the opposite wall, and the size is
0.3m0.3m. Detailed parameters and boundary conditions of
this experiment are listed in Table 1.
The boundary condition about thermal radiation for this
case was discussed by Srebric et al. (2008).They recommended that the ratio of convection to radiation (C:R) is 30:70.
Table 1.

Figure 2 Benchmark test cases of displacement ventilation


(Kato and Yang, 2003).

Boundary Conditions in the Experiment

Boundary Condition

Values

Supply inlet (0.4m 0.2m)

Velocity=0.182 m/s
Temperature =21.8C
Turbulence intensity=30%
Scale length =0.1m

CSP (1.471m2 H=1.67m)

Heat source=76W

Exhaust (0.3m 0.3m)

Figure 3 Geometry and mesh in computational domains. (a) Detailed CSP, (b) Simple and improved CSPs,
(c) Rectangular CSP.
ASHRAE Transactions

475

In addition, they suggested another 10W should be added due


to the heat transferred through the wall. According to this
assumption, total heat load of this chamber is 86W. The
convection heat from the CSP is 22.8W. The radiation heat is
53.2W. Heat transferred from the wall is 10W.
COMPARISON METHODOLOGY
The only set of data available from the benchmark test
(Kato and Yang, 2003) is temperature distributions. The data
are used to validate the CFD simulation results. The validated
CFD model is further applied to compare quantitatively the
effect of different CSPs on the distributions of temperature and
pollutant near the CSP. In simulations, the pollutant is treated
as a passive scalar and has no effect on the airflow profile.
Both active and passive pollutant sources are considered
in the simulations. Active source is the pollutant associated
with heat source and in the simulation cases, pollutant that
comes from the CSP surface. Passive source is not associated
with heat release and in the simulation, pollutant that is generated from the floor surface.
Both global environment of the room and the micro-environment around the breathing zone are studied. The global
environment is defined as the location far away from the CSP.
There is no clear boundary for this definition. In current experiment, data are compared at 24 sampling points shown in
Figure 4. The breathing zone is defined as a spherical volume
centered at the mouth of the occupant (radius =0.3m).
Root mean square (RMS) residual is used as the criterion
to analyze simulation error. It is defined as:
2

RMS Residual =

x----N

(1)

where

N
x2

= the number of measured data points.


= variance. It is the sum of the squares of the residuals:
N

x = [ y mi y ei ]

(2)

i=1

ymi =
yei =

the simulated value.


the expected value.

Absolute error is also applied as the criterion to analyze simulation error:


Error = y exp ected y simulated

(3)

Where y is the volume average value:


yi Vi
y = --------------------V

(4)

RESULTS AND DISCUSSION


Verifying Simulation Results by Measured
Temperature Field
In Kato and Yangs experiment, temperatures were
measured along four different poles in the chamber. Each pole
had six sampling points, as shown in Figure 4. L1 and L5 were
close to the wall, while L2 and L4 were near the manikin.
Figure 5 compares the measured and simulated temperatures
at these points. For the global temperature field, the simulation
predicted the trend of displacement ventilation and agreed
well with the experimental data. There is no significant difference among the simulated result using different CSPs. The
most obvious absolute temperature difference between simulation and measured data occurs near the inlet. Because velocity and temperature profiles at the inlet were not given in the
experiment, a uniform velocity and temperature profile was
assumed in simulations. This may affect the simulation accuracy at locations near the inlet (Figure 5a), but the overall
temperature profiles in other regions are well captured.
Comparing with experimental data, the overall RMS residual
of detailed CSP simulation is 0.41 oC, or 0.26 oC if we neglect
the inlet point of (0.2m,0.2m).
In the following analysis, the modeling result of the
detailed CSP will be used as the baseline data for comparison
purpose.
Comparing different CSPs for
Active Pollutant Cases

Figure 4 Arrangement of sampling points (Kato and Yang,


2003).
476

To determine how much the CSP simplification can influence the numerical simulation results, we compared the simulated temperature and concentration fields of the simple,
rectangular, and improved CSPs with the detailed CSP.
Figure 6, Figure 7 and Table 2 present the results. Notice that
the RMS residual results in Table 2 for different CSPs are
compared with those of the detailed CSP case (i.e., used as
baseline data for comparison).
ASHRAE Transactions

Figure 5 Comparison of measured and simulated temperature profiles. (a) Temperature profile at X=0.20m, (b) Temperature
profile at X=1.55m, (c) Temperature profile at X=1.95m, (d) Temperature profile at X=3.30m.

ASHRAE Transactions

477

Figure 6 Simulated iso-temperature contours for four cases. (a) Detailed CSP, (b) Simple CSP, (c) Rectangular CSP,
(d) Improved CSP.

478

ASHRAE Transactions

Figure 7 7 Simulated iso-concentration contours (dimensionless) for four cases (active source). (a) Detailed CSP, (b) Simple
CSP, (c) Rectangular CSP, (d) Improved CSP.

Table 2. RMS at Sampling Points Using


Different CSPs (Active Source)
X=0.2 m X=1.55 m X=1.95 m X=3.3 m

Total

Temperature RMS Residual (C)


Simple

0.064

0.124

0.141

0.139

0.121

Rectangular

0.083

0.247

0.222

0.237

0.208

Improved

0.065

0.093

0.127

0.142

0.111

Concentration RMS Residual (dimensionless)


Simple

0.061

0.054

0.082

0.088

0.073

Rectangular

0.055

0.096

0.124

0.070

0.090

Improved

0.055

0.060

0.164

0.064

0.097

ASHRAE Transactions

For the global environment, different CSPs do not significantly change the simulation results. The maximum temperature RMS is 0.14 oC for the simple CSP model, while this
value is 0.25 oC for the rectangular CSP. The overall temperature RMS residuals using these CSPs are 0.12 oC, 0.21 oC,
and 0.11 oC respectively for the simple, rectangular, and
improved CSPs. Similar results are observed in the concentration field. The overall dimensionless concentration RMS
residuals are 0.073, 0.090 and 0.097 respectively. These residuals are comparable since all of them are less than 10%. It is
notable that the improved CSP does not benefit the global
simulation result much. For example, the overall temperatures
RMS for the simple and improved CSPs are 0.12oC and
0.11oC respectively.
479

However, when it comes to the micro-environment at the


breathing zone, things are different. The RMS residuals
increase near the heat source. For example, the dimensionless
concentration RMS residual at X=0.2m is 0.061 for the simple
CSP and 0.055 for the rectangular CSP. But at X=1.95m, the
RMS residuals are 0.082 and 0.124, respectively. For the
points closer to the CSP, the RMS residuals are larger. Two
reasons may be contributed to this error increasing: (1) The
geometry of the CSP. Since the CSP blocks the airflow at local
position, the velocity field is different with different CSP
geometry. (2) The heat flux of the CSP. In the simulation, total
heat load is consistent in all cases, which means the heat flux
boundary condition is different due to different surface areas.

This may affect the strength of thermal plume produced by the


CSP. Since the active pollutant is associated with heat source,
pollutant concentration distribution will be affected by the
CSP simplification, especially for the micro-environment
close to the heat source.
Figure 8 shows the pollutant concentration contours
around the head. Forty points are selected to evaluate the error
of CSP simplification. They are evenly distributed in the
spherical volume. The temperature RMS residuals of the
simple and rectangular CSPs are 0.32 oC and 0.27 oC, respectively. The residuals of concentration are 0.26 and 0.19,
respectively. It is observed that the error of the simple CSP is
even larger than the rectangular CSP. This is surprising

Figure 8 Simulated iso-concentration contours (dimensionless) at breathing zone for four cases (active source). (a) Detailed
CSP, (b) Simple CSP, (c) Rectangular CSP, (d) Improved CSP
480

ASHRAE Transactions

because people may expect that the simple CSP whose shape
is more like a human would perform better than the rectangular CSP. The larger error of the simple CSP could be
explained by that it does not treat the source distribution well.
It is found that the head of the simple CSP releases more
pollutant than the head of the detailed CSP. Although the
head section of the rectangular CSP is even larger than the
simple CSP, its surface concentration is smaller than the
simple CSP. That is why simulation result of the rectangular
CSP is even better than the simple CSP. To overcome this
problem, the source distribution should be rearranged according to the detailed CSP, that is exactly what has been done in
the improved CSP case. Table 3 lists the source distribution
on different CSPs. Better simulation results should be
expected through carefully arranging the heat and pollutant
source over the geometry of the simplified CSP.
To evaluate the advantage of the improved CSP, the
volume averaged temperature and concentration for different
CSPs are compared. The comparison result is shown in
Table 4. The temperature errors due to the CSP simplification
are 0.44 oC and 0.38 oC respectively for the simple CSP and
rectangular CSP. The concentration errors are 0.44 and 0.36,
respectively. The rectangular CSP performs better than the
simple CSP, and there seems no benefit by adopting humanlike CSP geometry. Instead, attention should be paid to ensure
the heat and pollutant source consistence of the simplified
CSP over the actual human (detailed CSP).

The temperature field is identical to the active cases. The


concentration field is not sensitive to different CSPs. The error
due to CSP simplification is less than 0.05 for global field, and
is 0.03 for breathing zone environment. All are acceptable
Based on the above discussion, the error analysis result is
obtained in Table 7. The reduced rate of computational grids
is defined as the number of reduced grids using simplified
CSPs over the total grid number of the detailed CSP.
It is observed that the improved CSP has acceptable accuracy for both active and passive source cases, and for both
micro and global environment study. Compared to the detailed
CSP case, the improved CSP not only requires 25% less
computational grids, but more importantly, leads to easier
convergence due to less complex geometry and grid distributions.
Table 3.

Source Distribution of Different CSPs


Area
(m2)

Heat Load
Heat Flux
Concentration
(W)
(W/m2)
Detailed CSP

Arm

0.136

7.320

0.003

53.824

Body

0.199

10.711

0.003

53.824

Head

0.056

3.014

0.003

53.824

Leg

0.316

17.008

0.003

53.824

Total

0.706

38.000

Comparing Different CSPs for


Passive Pollutant Cases

Simple CSP

Because the pollutant is treated as a scalar, the location of


pollutant source does not affect the temperature and velocity
field. For passive pollutant case, we assume that the pollutant
is released uniformly from the floor surface. The concentration field of the simple CSP and improved CSP should be similar considering that the CSP heat flux distribution is not
important when heat source and pollutant source are not
coupled. Therefore, only the rectangular and simple CSPs are
compared next.
Figure 9 compares the concentration field of passive
pollutant cases. The concentration field in the breathing zone
is almost uniform for all cases, and CSP has little effect on the
micro-environment. Tables 5 and 6 quantitatively compare the
global field and volume averaged field at the breathing zone.
Table 4.

Arm

0.188

7.641

0.00227

40.642

Body

0.334

13.574

0.00227

40.642

Head

0.112

4.552

0.00227

40.642

Leg

0.301

12.233

0.00227

40.642

Total

0.935

38.000
Improved CSP

Arm

0.188

7.320

0.00217

38.937

Body

0.334

10.711

0.00179

32.069

Head

0.112

3.014

0.00149

26.912

Leg

0.301

17.008

0.00315

56.507

Total

0.935

38.000

Volume-Averaged Temperature and Concentration at Breathing Zone


Using Different CSPs (Active Source)
Error
Detailed

Simple

Rectangular

Improved
Simple

Rectangular

Improved

Temperature (C)

26.51

26.94

26.89

26.66

0.44

0.38

0.16

Concentration

1.005

1.444

1.360

1.076

0.439

0.355

0.071

ASHRAE Transactions

481

Figure 9 Simulated iso-concentration contours (dimensionless) for four cases (passive source). (a) Detailed CSP, (b) Simple
CSP, (c) Rectangular CSP, (d) Improved CSP.

Table 5. Comparison of Temperature and


Concentration RMS at Sampling Points
Using Different CSPs (Passive Source)
X=0.2 m X=1.55 m X=1.95 m X=3.3 m

Table 6. Volume-Averaged Temperature and


Concentration at Breathing Zone
Using Different CSPs (Passive Source)

Total
Detailed

Simple

Rectangular

Error
Simple

Error
Rectangular

Temperature (C)

26.51

26.94

26.86

0.43

0.35

Concentration

0.997

0.994

0.994

0.003

0.003

Temperature RMS Residual (C)


Simple

0.045

0.120

0.131

0.163

0.123

Rectangular

0.083

0.237

0.233

0.236

0.208

Concentration RMS Residual (dimensionless)


Simple

0.014

0.018

0.023

0.036

0.024

Rectangular

0.002

0.063

0.063

0.043

0.049

482

ASHRAE Transactions

Table 7. Comparison of Errors


due to CSP Simplification*
Simple
CSP

Rectangular
CSP

Improved
CSP

Active Pollutant Source


o

Global-T ( C)

0.121

0.208

0.111

Global-C

0.073

0.09

0.097

Micro-T ( C)

0.436

0.384

0.162

Micro-C

0.439

0.355

0.071

Passive Pollutant Source


o

Global-T ( C)

0.123

0.208

Global-C

0.024

0.049

Micro-T ( C)

0.431

0.351

Micro-C

0.003

0.003

Reduced Rate of Computational Grids


%

25.0%

32.4%

25.0%

*The errors of gray color in this table are unacceptable.

CONCLUSION
Based on simulation cases by adopting different CSPs,
the following conclusions can be made:
1.

2.

For the global field study, all the CSPs, simple or


complex, yield acceptable simulation results. There is
no apparent benefit by adopting very complex CSP
geometry.
For the micro-environment around the person, there
seems no significant benefit by adopting human-like CSP
geometry either. Instead, attention should be paid to
ensure the heat and pollutant source consistence of the
simplified CSP over the actual human (detailed CSP).

ACKNOWLEDGMENTS
This study is supported by Chinas Eleventh Five-Year
Scientific Research Support Program, Project No.
2006BAJ02A08. We thank Prof. Shinsuke Kato for providing
experimental data, and Prof. Peter Nielsen for organizing an
interesting ASHRAE seminar on this topic.
REFERENCES
Bjorn, E. and Nielsen, P.V. 2002. Dispersal of exhaled air
and personal exposure in displacement ventilated
rooms, Indoor Air, 12, 147-164.
Chen, Q. 1995. Comparison of different k- models for
indoor airflow computations, Numerical Heat Transfer
Part B, 28, 353-369.
Chen, Q. 1996. Prediction of room air motion by Reynoldsstress models, Building and Environment, 31(3), 233-244.
Chen, Q. and Srebric, J. 2002. "A procedure for verification,
validation, and reporting of indoor environment CFD
analyses," HVAC&R Research, 8(2), 201-216.
ASHRAE Transactions

Gao, N., Niu, J. 2004. CFD study on micro-environment


around human body and personalized ventilation,
Building and Environment, 39, 795-805.
Hayashi, T. 2002. CFD analysis on characteristics of contaminated indoor air ventilation and its application in the evaluation of the effects of contaminant inhalation by a human
occupant Building and Environment, 37, 219-230.
He, G., Yang, X., Srebric, J. 2005 Removal of contaminants
released from room surfaces by displacement and mixing ventilation: modeling and validation, Indoor Air,15,
367-380.
Kato S. and Yang J.H. 2003. Benchmark test for a computer
simulation person: experimental result of displacement
ventilation, Institute of Industrial Science, University of
Tokyo, Japan.
Launder, B.E. and Spalding, D.B. 1974. The numerical
computation of turbulent flows, Computer Methods in
Applied Mechanics and Energy, 3, 269-289.
Murakami, S. 2004. Analysis and design of micro-climate
around the human body with respiration by CFD,
Indoor Air, 14, 144-156
Sideroff, C.N. and Dang, T.Q. 2005. Validation of CFD for
the flow around a computer simulated person in a mixing ventilated room, Proceedings of the 10th International Conference on Indoor Air Quality and Climate,
Beijing, China, 3885-3889.
Sorensen, D.N., Nielsen, P.N. 2003. Quality control of computational fluid dynamics in indoor environments,
Indoor Air, 13, 2-17.
Sorensen, D.N., Voigt, L.K. 2003. Modelling flow and heat
transfer around seated human body by computational
fluid dynamics, Building and Environment, 38, 753-762.
Srebric, J., Vukovic, V., He, G., Yang, X. 2008. CFD
boundary conditions for contaminant dispersion simulations around human occupants in indoor environments,
Building and Environment, 43, 294-303.
Topp, C., Nielsen, P.V., and Sorensen, D.N. 2002 Application
of computer simulated persons in indoor environmental
modeling, ASHRAE Transactions,108(2), 1084-1089.
Yang, X., Srebric, J. Li, X., He G. 2004. Performance of three
air distribution systems in VOC removal from an area
source, Building and Environment, 39(11), 1289-1299.
Yokhot, V., Orzag, S.A., Thangam, S., Gatski, T.B. and Speziale, C.G. 1992. Development of turbulence models
for shear flows by a double expansion technique, Physics Fluids A, 4(7), 1510-1520.
Yuan, X., Chen, Q., Glicksman, L.R. 1999. Performance
evaluation and design guideline for displacement ventilation, ASHRAE Transactions, 105, 3403352.
Zhang, L., Chow, T.T. 2005. Comparison of performances
of displacement and mixing ventilation. Part I: thermal
comfort, International Journal of Refrigeration, 28,
276-287.
Zhang, L., Chow, T.T. 2005. Comparison of performances
of displacement and mixing ventilation. Part II: indoor
air quality, International Journal of Refrigeration, 28,
288-305.
483