2011 John Wiley & Sons A/S

Indoor Air 2011; 21: 442453

wileyonlinelibrary.com/journal/ina
Printed in Singapore. All rights reserved

INDOOR AIR
doi:10.1111/j.1600-0668.2011.00723.x

Commemorating 20 years of Indoor Air

CFD and ventilation research
Abstract There has been a rapid growth of scientic literature on the application
of computational uid dynamics (CFD) in the research of ventilation and
indoor air science. With a 100010,000 times increase in computer hardware
capability in the past 20 years, CFD has become an integral part of scientic
research and engineering development of complex air distribution and ventilation systems in buildings. This review discusses the major and specic challenges
of CFD in terms of turbulence modelling, numerical approximation, and
boundary conditions relevant to building ventilation. We emphasize the growing
need for CFD verication and validation, suggest ongoing needs for analytical
and experimental methods to support the numerical solutions, and discuss the
growing capacity of CFD in opening up new research areas. We suggest that
CFD has not become a replacement for experiment and theoretical analysis in
ventilation research, rather it has become an increasingly important partner.

Y. Li1, P. V. Nielsen2
1

Department of Mechanical Engineering, The University

of Hong Kong, Pokfulam Road, Hong Kong SAR, China,
2
Department of Civil Engineering, Aalborg University,
Aalborg, Denmark
Key words: Computational fluid dynamics; Building
ventilation; Experiment; Theory; Analysis; Validation.
Y. Li
Department of Mechanical Engineering
The University of Hong Kong
Pokfulam Road, Hong Kong SAR
China
Tel.: +852 28592625
Fax: +852 2858 5415
e-mail: liyg@hku.hk
Received for review 29 October 2010. Accepted for
publication 5 May 2011.

Practical Implications

We believe that an eective scientic approach for ventilation studies is still to combine experiments, theory, and
CFD. We argue that CFD verication and validation are becoming more crucial than ever as more complex ventilation problems are solved. It is anticipated that ventilation problems at the city scale will be tackled by CFD in the
next 10 years.

Introduction

Building ventilation, together with source control and

air cleaning, controls indoor air quality. Mechanical
fans and/or natural forces drive ventilation. A major
design objective is to ensure that ventilation air is
delivered from a limited number of supply openings to
every (occupied) part of the room in an eective and
ecient manner. This has become a challenging engineering task as buildings become larger, higher, and
deeper, and the penetration depth of a ventilation
supply air jet is limited. Air distribution is aected by
supply air jets, supply and exhaust location, thermally
induced airows, as well as room geometry and
obstacles.
Air distribution is all about airow and transport of
heat and airborne pollutants. Air distribution is
governed by the conservation principle. It is mathematically described by a set of partial dierential
442

equations, known as the NavierStokes equations.

These can be solved analytically only for simple and
ideal conditions. For complex geometry and/or complex boundary conditions, numerical methods may be
used to solve these equations or solve their modeling
versions, given the initial and boundary conditions, i.e.
computational uid dynamics (CFD). There have been
at least three distinctive or overlapping stages in the
solutions of the governing equations, similar to other
elds involving uid mechanics and heat transfer.
The stage of analytical closed-form or approximate
solutions. Examples include Koestel (1955), Rajaratnam (1976), etc.
The stage of empirical relationships obtained from
measurements in small- and full-scale tests. Examples include Straub et al. (1956) and Mullejans
(1966). Both examples were fully cited in ASHRAE
(2009) and Awbi (2003), respectively.

CFD and ventilation research

The stage of numerical solutions.
During the 1950s to the 1970s, the aerospace
industry and weather/climate prediction were the
primary drivers in early CFD development, but an
engineering application such as building ventilation has
been a main driver in recent CFD development over
the last 30 years. Nielsen (1973) was probably the rst
to apply CFD in ventilation studies. Since 1970s, there
have been considerable developments in both CFD
(faster computers, faster numerical methods, and
improved turbulence modeling) and ventilation (ventilation methods and ventilation control). Indeed, CFD
has now been used in nearly all disciplines where
insights into uid phenomena are needed. CFD simulations with 5 billion cells were carried out for urban
heat island analysis in Tokyo for a 33 km 33 km area
(Ashie and Kono, 2011).
There are a number of reviews on CFD and
ventilation, such as Awbi (1989), Jones and Whittle
(1992), Chow (1996), Nielsen (2004), Zhai (2006) and
Chen (2009). Textbooks on CFD include Ferziger and
Peric (1999), Wesseling (2000) and Nielsen et al.
(2007), on CFD verication and validation: Roache
(1998), on turbulence modeling: Wilcox (2006) and
Sagaut (2001), on building ventilation: Croome
and Roberts (1981), Etheridge and Sandberg (1996)
and Awbi (2003).
This review intends to have a dierent focus from
the existing reviews. We discuss the growing CFD
capability and challenges for ventilation studies, and
suggest continued needs for analytical and experimental methods to support the numerical solutions.
Examples of the contribution of CFD in revealing
new ventilation phenomena as well as its use beyond
building application are also discussed. We do not
intend to review the application of CFD in practical
design of building ventilation, simply as the scientic
literature does not really reect the whole picture of
the eld, and isolated comments will be made in the
text when necessary. We also do not cover the
important topic of comparison of dierent turbulence
models, as it was reviewed, for example, by Chen
(2009).
Two major challenges in CFD

CFD solves the governing equations of uid ow on a

computer and provides both spatial and temporal eld
solutions of variables such as temperature and velocity,
and/or predicts the dispersion of pollutants in a room
or a building, i.e. the computational domain. CFD can
also be used to predict indoor air quality, thermal
comfort, re and smoke spread, and wind ow around
buildings. The eect of gravitational body force can be
signicant in airows in buildings. Owing to relatively
small temperature dierences (with smoke ow as an

exception), the so-called Boussinesq approximation is

commonly used (Tritton, 1988).
In CFD, ventilation parameters such as room/
building geometry, supply and exhaust openings, and
physical variables can be exibly changed. CFD would
allow scientists to test scenarios that were not possible
using experimental methods, and have more in-depth
and in-breadth examination of the physical phenomenon, because of relatively low cost and eciency. CFD
software is now widely available, and complex ventilation problems seem to be easily simulated on a PC in
a researchers oce.
Adapted from Fujii (2005), Table 1 shows a comparison of experimental methods and CFD. Both
methods suer from possible inherent errors. The
major accuracy issue in experimental methods is the
scaling eect, in particular when small-scale models are
used. CFD suers from both turbulence modeling
errors and numerical errors.
CFD has two major challenges. The same applies to
its application in ventilation research. The rst is that
turbulence is still only modeling in CFD.
We illustrate this challenge using two examples.
First, the ow in a ventilated room is generally
assumed to be a fully developed turbulent ow, and
this ow can be handled by most turbulence models.
But in some areas of the room, including the occupied
zone, a low Reynolds number ow can exist at low
room air supply velocity. Figure 1 shows measurements of the maximum velocity in the occupied zone of
a room with mixing ventilation from a wall-mounted
diuser versus the air change rate. The ow is
isothermal (Nielsen, 1992a). It is known from the
similarity principles that any velocity as, for example,
the maximum velocity in the occupied zone is a linear
function of the air change rate (or the supply velocity)
when the ow is a fully developed isothermal turbulent
ow. In Figure 1, this is the case for velocities larger
than 0.25 m/s, but the gure indicates that the ow in

Table 1 Comparison of laboratory experiments and computational fluid dynamics (CFD)

for building ventilation (adapted from Fujii, 2005)
Small-scale or full-scale experiments

CFD

Water tanks,
small-scale models,
or full-scale rooms
Measurement methods
Manufacturing techniques

Computers

Model manufacturing
Data acquisition
Data handling
Scaling effects
(Reynolds number and
Archimedes number)

Numerical algorithms
Programming techniques
(parallel language, etc.)
CAD interface, grid generation
Postprocessing
Visualization
Discretization error, turbulence
modeling error (e.g. because of
low Reynolds number in
the predictions)

443

Li & Nielsen

Fig. 1 The measured maximum velocity (urm, m/s) in the occupied zone of a room versus air change rate (indicating supply
velocity) (n, h)1) in the case of isothermal mixing ventilation.
Proportionality between supply velocity and maximum velocity
in the room indicates a fully developed turbulent ow in the
occupied zone for a supply velocity larger than 0.25 m/s and a
low Reynolds number ow for lower velocities

the occupied zone is a low Reynolds number ow for

velocities below 0.25 m/s (dotted line). The conventional turbulence models may not accurately capture
such important physical phenomena.
Second, the importance of correct turbulence modeling may be seen from predicting a wall jet spread in a
room from a single nozzle in the end wall close to the
ceiling (see Schalin and Nielsen, 2004). The predicted
ow elds are shown in Figure 2, with both the k-
model with the assumption of isotropic turbulence and
the Reynolds stress model (RSM) considering the
anisotropic turbulence.
The wall jet is wrongly predicted by the k- model, as
the wall jet grows at an equal rate parallel to the ceiling
and down into the room. It is known that a threedimensional wall jet grows much faster parallel to the
ceiling than down into the room. In the RSM, with
wall reection terms, a redistribution of normal stresses
takes place near the wall. These dampen the turbulent
uctuations perpendicular to the wall and convert the
energy to uctuations parallel to the wall. The RSM
thus gives a better prediction than the k- model as

compared to the experiments (Schalin and Nielsen,

2004).
The error as a result of the anisotropic issue may
become relatively insignicant (and therefore never
realized) if the room is short and the supply opening
has a large width compared with height (a typical
supply opening). This error is also small if the ow in
the room is nearly two dimensional, for example, if the
supply opening is 20% of the room width or greater,
i.e. when a two-dimensional wall jet in the initial ow
results (data not shown).
Dierent turbulence models for building airows
were evaluated, including Reynolds-averaged modeling
(Chen, 1996; Murakami, 1998; Abdilghanie et al.,
2009) and large eddy simulations (Mochida et al.,
2005; Beghein et al., 2005). Owing to turbulence
modeling diculties, CFD may not be able to reveal
the real physics as in experiments. An increasing
number of large eddy simulations are carried out for
indoor airow analysis, which allows the capture of
some ow physics (Berrouk et al., 2010; Choi and
Edwards, 2008).
The second major challenge in CFD is that the
solutions are still approximate, not exact. Various
numerical phenomena occur, including numerical diffusion and numerical dispersion (Li, 1997), as demonstrated in Figure 3 by the predicted results of a scalar
cone 10 units high after one full counterclockwise
rotation. The maximum Courant number (velocity
times the time step, divided by the grid size) used is
0.09. The grid is 64 64. The numerical diusion is
shown by the reduction in the cone height from
10 units to 8.243 for the QUICK scheme and 5.251
for the second-order upwind scheme (SOU). The
numerical dispersion is shown by the oscillations
behind the cone for QUICK and in front of the cone
for SOU (Li, 1997). Thus, the two commonly used
convection schemes in ventilation studies are shown to
give signicant errors for a very simple problem.
Questions can be asked whether the same numerical
diusion and dispersion exist in the CFD simulations
for complex ventilation problems, and the answer is

(a1)

(a2)

(b1)

(b2)

Fig. 2 Dierences in predicted wall jet spread in a 3D room by the k- model (a1, b1) and the Reynolds stress model (a2, b2), as seen by
the size of the white area where the speed is >3.5 m/s in (a1) and (a2) for the prole near the ceiling, and >3.0 m/s in (b1) and (b2) for
the prole in the mid-vertical plane. The wall jet was issued from the left at the ceiling level (see Schalin and Nielsen, 2004)

444

CFD and ventilation research

Fig. 3 Perspective plots of the predicted proles of a scalar conic shape of 10 units high after one full counterclockwise rotation (Li,
1997)

most likely positive. In theory, the numerical errors

reduce to an acceptable accuracy when the number of
grid points is sucient or the mesh size is suciently
small. The eect of numerical diusion in recirculating
ow is also demonstrated in a Smith and Hutton
scheme by Srensen and Nielsen (2003b). Numerical
accuracy is expected to improve as computer capacity
further increases and better numerical schemes are
developed.
Owing to turbulence modeling errors and numerical
errors, CFD is not only a replacement for the theory
and experimental methods in ventilation analysis but
is also a partner. Obviously, an eective scientic
approach is to combine theory, experiments, and CFD
in a way that exploits the inherent strengths of each
method. This statement may not be true for engineering design of complex realistic problems, in
which data from both theory and experiments are
generally not available, and the required engineering
accuracy for noncritical applications is often not very
high.
Modeling building ventilation specific challenges

Croome and Roberts (1981) asked two questions: (i)

What kind of airow pattern should be in a space
occupied? (ii) How and when can the indoor airow be
predicted? CFD was mostly used to answer the second
question, but it has also increasingly been used to
answer the rst question.
The primary process simulated in CFD ventilation
models is the airow. The primary variables include air
speed, its direction, air temperature, and turbulence
quantities. The buoyancy force can be signicant in
building airows. An important process that aects the
airow in buildings is the transport of heat and
moisture, including thermal radiation (Li et al.,
1993). The related surface condensation is also important for dampness and mold issues in indoor air. An
understanding of the detailed heat transfer through the
building envelope using combined CFD and heat

transfer analysis is also useful for building energy

eciency.
For indoor air quality analysis, detailed emission
characteristics at the material surface (Murakami
et al., 2003; Yang et al., 2004; Mo et al., 2005), and
breathing and coughing may be included. For particulate matter, important processes include particle
deposition and resuspension. An important process
that aects gaseous pollutants and aerosols is indoor
air chemistry.
Two overall ventilation performance indexes are
often used. The air exchange eciency indicates how
eciently the outdoor air is distributed in the room,
while the ventilation eectiveness indicates how eciently the airborne pollutant is removed from the
room. The local mean age of air at a point is dened as
the average time that the air takes to arrive at that
point as it rst enters the room, and the room mean age
of air is the average of the age of air at all points in the
room. The age of air can be measured using tracer gas
techniques. The air exchange eciency and ventilation
eectiveness have been studied using CFD (Peng et al.,
1997).
We next discuss two specic challenges. The rst is in
the description of the supply openings. The specic
challenge in representing supply openings is part of
more general challenges in properly specifying boundary conditions. One major challenge for characterizing
airow in buildings is the central importance of
complex boundary conditions indoors. We now understand rather well how to model thermal and ow
boundary conditions at walls, supply openings, and
natural ventilation openings (Li and Holmberg, 1994;
Fontaine et al., 2005). Eorts made to integrate CFD
with building thermal modeling and multizone airow
modeling (Axley, 2007) have been reviewed by Chen
(2009).
The ow from a diuser in a ventilation system is
determined by very small details in the diuser design.
A numerical prediction method should be able to
handle small details in dimensions of one tenth of a
445

Li & Nielsen
millimeter, as well as dimensions of several meters in
the room. Such a wide range of the geometry necessitates incorporating a large number of cells in the
numerical scheme, which increases the prediction cost
and computing time to a high level.
Various simplications can be suggested. The most
obvious simplied method is to replace the actual
diuser with one of less complicated geometry that
supplies the same momentum of airow to the room.
This may be obtained from a single opening with an
area equivalent to the eective supply area of the
diuser, and such methods were evaluated by Nielsen
(2004).
The box method is a method based on specifying the
wall jet ow (or free jet ow) or measured data close to
the diuser (Nielsen, 1973, 1974). The details of the
ow in the immediate vicinity of the supply opening are
ignored, and the supply jet is described by values along
the surfaces on an imaginary box in front of the
diuser. Two advantages are obtained by using such
boundary conditions. First, it is not required to use a
grid as ne as that needed for full numerical prediction
of the wall jet development close to the opening.
Second, it is possible to make two-dimensional predictions for supply openings that are three-dimensional,
provided that the jets develop into a two-dimensional
wall jet or free jet at a given distance from the opening.
The box method is a possibility because the ow
close to the openings can be considered to be a
parabolic ow in contrast to the general ow in the
room which often will be an elliptic ow. It is therefore
also a possibility to use the full capacity of the
computer to generate the ow around the diuser
and after that, using the capacity to predict the ow in
the room, with the predicted values from the rst run
as boundary conditions in the second set of predictions
(Kondo and Nagasawa, 2002).
The prescribed velocity method has also been successfully developed. The inlet proles are given as boundary conditions at the diuser in the usual way (as
simplied boundary conditions), and they can be
represented by a few grid points only. The velocity

(a)

(b)

proles are given in the inlet volume (box) in front of

the diuser, and all the other variables are calculated in
this volume as well as in the rest of the room. The
velocities are thus prescribed for the box in front of the
diuser as the analytical values obtained for a wall jet
from the diuser, or they are given as measured values
(Gosman et al., 1980; and Nielsen, 1992a,b).
The momentum method is a method where the
momentum and mass ow are decoupled in the CFD
simulation of the diuser, and the initial momentum
and mass ow rate from the diuser are used as the
boundary conditions. Chen and Srebric (2001) give a
detailed discussion of the momentum method.
Continuous development of computational capacity
and speed has undoubtedly made the direct methods
with local grid renements or multigrid solution, a
natural possibility. The diuser in Figure 4 consists,
for example, of 12 small slots which can be adjusted to
dierent ow directions. The diuser is mounted in a
wall below the ceiling. It is a complicated geometry
generating an asymmetrical three-dimensional ow in
the room, and it can only be introduced in the CFD
predictions by making a detailed description of the
boundary conditions corresponding to a partly or fully
resolved diuser.
Comparisons between measurements and predictions
with the grids in Figure 4b,c show that predictions
based on 250,000 cells are sucient to obtain a gridindependent solution (Szczena et al., 2005). Other
diuser designs can be much more demanding to the
computer capacity.
It may be noted that it is always possible to treat a
complicated supply opening such as that shown in
Figure 4 as porous medium. We specify the so-called
intrinsic average velocity and the free opening ratio.
The mass and momentum conservation are automatically satised. In addition, all scalars such as turbulent
kinetic energy are automatically conserved. This total
conservation method has not been widely tested. Such
treatment is similar to the porous model used in tree
canopy layer analysis and recently in the urban
ventilation analysis (Hang and Li, 2010).

(c)

Fig. 4 (a) Air supply diuser with unsymmetrical adjusted nozzles. (b) Representation of the air supply diuser using embedded grid
renement and (c) using unstructured grid

446

CFD and ventilation research

The second specic challenge is in the treatment of
multiple length scales existing in the ventilation problem. Table 2 summarizes a variety of spatial scales of
building ventilation problems that have been simulated. It covers a range of 0.001100 m in building
ventilation and up to10,000 m including city ventilation. For example, CFD has been used to understand
how the supply air jets are generated and how they
develop (e.g. Karimipanah and Awbi, 2002), as well as
how human body plumes evolve (e.g. Gao and Niu,
2006; Craven and Settles, 2006). It is important to
realize that parabolic ows as jets and plumes depend
on a detailed description of the source of the ow
(momentum source scale in Table 2). The ability to
study detailed airows and heat transfer around a
human body allows the detailed examination of thermal comfort of the human body in a room (Gao and
Niu, 2005). Most studies have been at the room scale,
and most of these considered an empty room. Dierent
air distribution strategies are evaluated and compared
(e.g. Lin et al., 2005). Airows between rooms can be
important for overall airow pattern design, such as in
hospitals. CFD has also been useful in analyzing the
eectiveness of open windows, wind catchers, and
chimneys in natural ventilation. CFD has been used
either alone or in combination with the multizone
airow models for the building scale study (Tan and
Glicksman, 2005). Another commonly studied situation is the combined analysis of airow around and
within a building, in particular for natural ventilation
analysis. In the latter situation, the wind ows within a
building cluster can aect the building ventilation. The
neighborhood scale also includes the ventilation of
streets. The ventilation problem can be further

Table 2 Variable scales of building ventilation problems

Problems

Geometric Scale

Study examples

Momentum
source scale

0.0010.1 m
Supply diffusers

Flow element
scale

0.12 m
Coughs and
exhalation puffs
Body plumes

Room scale

220 m
Personalized ventilation
Displacement ventilation,
Mixing ventilation,
Stratum ventilation
20200 m
Multiple rooms
Large enclosures
2002000 m
Multiple buildings
220 km
City ventilation

Nielsen (1992a,b),
Chen and Srebric (2001),
Nielsen et al. (2007),
Kondo and Nagasawa (2002)
Murakami et al. (2000),
Bjrn and Nielsen (2002),
Murakami (2004),
Zhu et al. (2006),
Russo et al. (2009)
Gan (1995), Brohus et al. (2006),
Tian et al. (2008),
Russo et al. (2009)

Building
scale
District
scale
City scale

Lu et al. (1996), Kato et al.

(1995, 1997),
Ji and Cook (2007)
Kato and Huang (2009),
Mirzaei and Haghighat (2010)
Hang and Li (2010)

extended beyond the building scale up to city scale

(Hang et al., 2009a,b; Yang and Li, 2009). Diculties
arise when multiple scales need to be resolved. To some
degree, the diculties in treating supply opening is also
one such multiscale problem.
CFD opens new research areas

CFD has made it possible for a number of new

research directions to be further explored. Examples
include solution multiplicity in building ventilation,
inverse CFD modeling, near-body micro-environment,
and disease transmission, which the authors are familiar with.
Solution multiplicity

Solution multiplicity refers to the existence of more

than one solution of the overall ow patterns in a
room under the same boundary conditions. Dierent
initial conditions can lead to dierent solutions. In
some situations, a solution can be switched to another
when there is a sucient perturbation. Mullejans
(1966) rst experimentally identied the existence of
two solutions in a room with mixed convection. The
same problem was analyzed using CFD by Nielsen
et al. (1979), which is shown in Figure 5. However,
solution multiplicity problems seemed to be forgotten
in the ventilation community until the phenomenon of
multiple solutions was found again in natural ventilation through analytical solutions by Nitta (1996), Li
and Delsante (1998, 2001) and Hunt and Linden
(2005), and for mixing ventilation by Bjerg et al.
(2002). Although the existence of multiple solutions
was not identied by CFD, CFD has provided a
powerful tool to understand solution multiplicity
phenomena in buildings, in addition to experimental
studies (Heiselberg et al., 2004; Li et al., 2006; Yang
et al., 2006).
Inverse CFD modeling

Identication of the airborne pollutant source location(s) and strength can be important in identication
of the index patient in an airborne disease outbreak
and quick determination of the source origin(s) during
intentional release of chemical or biological pollutants
in a building. Examples include Zhang and Chen
(2007) and Liu and Zhai (2008).
Near-body micro-environment

Both thermal comfort and exposure of respiratory

droplets require a full understanding of the near-body
micro-environment. CFD has made it possible to
calculate the detailed airow pattern around a human
body (Gao and Niu, 2005, 2006; Russo and Khalifa,
447

Li & Nielsen
(a)

(b)

Fig. 5 Existence of two solutions (a) and (b) in a room with a heated oor and full width slot, as shown by experimental studies (left
from Mullejans, 1966) and computational uid dynamics (CFD) (right from Nielsen et al., 1979), though the experiments and the CFD
predictions do not cover the same geometrical situation

2010), including near-body chemistry. The movement

of people can also be studied (Choi and Edwards,
2008).
Disease transmission

The importance of building ventilation and airow in

disease transmission and control has been known for a
long time, but it was during and after the 2003 SARS
epidemics that CFD became a modeling tool for
disease transmission in buildings. CFD has to some
degree reproduced the possible transmission routes in
the Amoy Gardens outbreak (Yu et al., 2004), the
PWH 8A Ward outbreak (Li et al., 2005) and in other
outbreaks (e.g. Niu and Tung, 2008). Since 2003, there
have been a great number of studies using CFD to
improve hospital ward ventilation (Chao and Wan,
2006; Noakes et al., 2006; Qian et al., 2006; Lai and
Cheng, 2007 and Bolashikov and Melikov, 2009). An
example of predicted exhaled particle dispersion in a
multibed ward is shown in Figure 6.
CFD verification and validation

The challenges on turbulence modeling and numerical

errors also require serious consideration in CFD
accuracy. The correct governing equations and boundary conditions, and appropriate numerical algorithms
must be carefully selected. This can only be done with a
thorough understanding of the problem before the
simulation takes place. The CFD simulation of a
problem is, at best, as good as the selected governing
equations for the problem. A detailed description of
CFD quality control in indoor air is given by Srensen
and Nielsen (2003b).
448

Fig. 6 Dispersion of the suspended 20-lm particles released

from a patient as predicted by computational uid dynamics in a
six-bed ward. The birth time(s) of the particles are shown in
colors. Details of the simulations can be found in the study by
Qian and Li (2010)

CFD verication deals with the mathematics of the

model, while validation deals with its physics (Oberkampf and Trucano, 2002). In other words, verication
is about equations being solved correctly, while validation is about the correct equations being solved. The
ve most common sources of errors in verication
(Freitas, 2002) include insucient spatial discretization
convergence, insucient temporal discretization convergence, insucient convergence of an iterative
procedure, computer round-o, and computer programming errors.
A number of journals, including the Journal of
Fluids Engineering (Roache et al., 1986), the Journal
of Heat Transfer (ASME, 1994) and the AIAA Journal
(1994), have an editorial policy statement for the
control of numerical accuracy, which includes the need

CFD and ventilation research

to describe the basic features of the method, that the
methods must be at least second-order in space, that
there is a need to assess articial viscosity, and the
phase error in transient calculations, to establish grid
independence, to address the iterative convergence,
and to assess the accuracy of boundary and initial
conditions. The Indoor Air journal produced a guest
editorial statement (Srensen and Nielsen, 2003a) on
ventilation CFD studies with similar requirements for
numerical accuracy, but with additional requirements
for validation. A procedure of calculation and reporting of discretization error estimates in CFD simulation
is available for situations when the experimental data
may or may not be available for comparison (Roache,
1997).
CFD validation involves comparison of the computational solution with experimental results. A number
of validation data sets are available in the literature
(Nielsen, 2010). One needs to quantify or estimate the
error in both the experimental data and CFD results
before comparing the two. One needs to consider the
reliability of the experimental data (e.g. because of
scaling eects), the modeling and matching of the
boundary conditions in the experiment and modeling,
as well as the other operating and physical conditions.
From our own experience, the 3D isothermal mixing
ventilation data set of Restivo (1979) and the 2D
natural convection case of Cheesewright et al. (1986)
are very suitable for new CFD users. The rst data set
is ideal for testing the use of the basic turbulence
models without the buoyancy eect and the boundary
conditions for steady-state problems. The second data
set is ideal for testing the inclusion of the buoyancy
force using the Boussinesq approximation, as well as
the wall boundary conditions for predicting heat ux
at the walls. Readers are also recommended to consider
the benchmarking test cases available at http://www.
cfd-benchmarks.com.
In practice, validation is a process of renement of
selecting the most suitable turbulence models or
making the right decisions about the ow assumptions.
A CFD user routinely makes some important decisions
about the governing equations and boundary conditions before making a prediction. The fundamental
questions before initiating a prediction are the following: (i) Is the ow expected to be laminar or turbulent?
(ii) Is the ow expected to be two dimensional or three
dimensional? (iii) Is the ow expected to have a
symmetry plane? (iv) Is the ow expected to have
one, two, or several solutions? (v) Is the ow expected
to be steady or unsteady? Such decisions will inuence
whether the solutions obtained are the physical ones.
The following example demonstrates the importance of
having a high level of uid mechanical knowledge.
Consider the simple IEA Annex 20 2D test case
(Olmedo and Nielsen, 2010). The assumption of
turbulent, steady, and two-dimensionality will lead

(a)

(b)

(c)

Fig. 7 Predicted path lines using dierent assumptions for the

ow in the simple International Energy Agency (IEA) Annex 20
2D test case (Olmedo and Nielsen, 2010). (a) Using the k-
model and assuming turbulent, steady, and 2D ows; (b) using
the k- model and assuming turbulent, steady, and 3D ows;
and (c) using the k- model and assuming turbulent, unsteady,
and 3D ows. The assumption of turbulent, steady, and threedimensionality will lead to the solution in (b) with a k- model.
It is slightly dierent from (a) in areas below the diuser. We
know that the ow in reality could be three dimensional.
Examining the solution in the third dimension can conrm that
the ow is symmetrical. However, we still do not know whether
the ow has more solutions, although it is unlikely because the
ow is symmetrical around the middle plane. Only the trivial
solution with two unsymmetrical solutions to either side is a
likely situation (see Schalin and Nielsen, 2004)

to the path line solution in Figure 7a with a k-

model. It looks correct, but strictly speaking, we do
not know whether the ow in reality has more
solutions, or whether it is three dimensional, or
perhaps unsteady.
The assumption of turbulent, unsteady, and threedimensionality will lead to the solution in Figure 7c.
The path lines in the middle plane and the velocity
distribution in the lower part of the room show the
steady-state ow similar to the ow found by the
steady-state equations. It is now possible to conclude
that the ow can be predicted with steady-state
equations and even with a set of two-dimensional ow
equations.
Beyond building ventilation

The concept of ventilation can be found in a wide range

of other enclosed or semi-enclosed spaces. In general,
ventilation is to provide exchange of uid into a space
to replace the existing gas/liquid and distribute the new
uid within the space.
Figure 8 shows an incomplete collection of dierent
ventilation problems. The length and time scales are
449

Li & Nielsen
Biomechanical Eng
Not well studied

100 year

Built environment
-well studied

Natural environment
Not well studiedOcean
Sea

1 year

Lake

1 month

Atm bound layer

Mice
burrow

1 day

Owl
burrow
Fox
burrow

1h

1 min

Reservoir

Human
lung

Chicken
shed
Building

City cluster

District
Pool
Green house Street Forest canopy

City

Room

Fish tank

Bird
lung

1s
1 mm

1m

1 km

1000 km

Fig. 8 Length and time scales of ventilation of dierent spaces

approximate. The time scale of ventilation varies from

a few seconds (bird lung) to 1000 years (ocean), while
the length scale ranges from 1 mm to 1000 km. The
following three categories of problems may be found:
Convection dominated problems such as ventilation
of buildings, caves (Christoforou et al., 1996),
streets, swimming pools, greenhouses, sh tanks,
human lungs, and even bird lungs.
Turbulence dominated problems ventilation of a
district, valley (Sivertsen et al., 1983), forest canopy
(Oliver, 1975; Miller et al., 2007), city, city cluster,
atmospheric boundary layer (Rigby et al., 2006),
lake, sea, and even ocean (England, 1995).
Molecular diusion dominated problems these are
mostly the homes of animals such as caves (Kowalczk and Froelich, 2010), burrows (Vogel et al.,
1973; Wilson and Kilgore, 1978; Maclean, 1981),
nests (Ar and Piontkewitz, 1992), fox burrow, owl
burrow, mice burrow (Shams et al., 2005), and termite colony (Turner, 1994). The ventilation ability of
animals for their homes is limited, and we are

changing their environment by making the ground

warmer and the ground level wind-less. Hence, it
may be fair to predict that we will have a new focus
for understanding ventilation conditions for animals, such as in mice burrows (Hansell, 1993; Liow
et al., 2009). Ventilation concerns the health of
animals mammals and non-mammals plants, and
industry processes, those in the ground, in air, and
even in water.
CFD has already been applied to understand greenhouse ventilation design and city ventilation (Hang
et al., 2009a). It may be anticipated that CFD will nd
more applications to other ventilation problems shown
in Figure 8 where it may be dicult to apply experimental methods directly. In the next 10 years, computer speeds are expected to double every 2 years
(Orszag and Staroselsky, 2000; Fujii, 2005), and there
is a potential one to two order-of-magnitude increase
in computer capability. This means that CFD will be
readily applicable to the larger length scale and
multilength-scale problems such as city ventilation.

References
Abdilghanie, A.M., Collins, L.R. and
Caughey, D.A. (2009) Comparison of
turbulence modeling strategies for indoor
flows, J. Fluids Eng., 131, 118.
AIAA (1994) Editorial policy statement on
numerical accuracy and experimental
uncertainty, AIAA J., 32, 3.
Ar, A. and Piontkewitz, Y. (1992) Nest
ventilation explains gas composition in
the nest-chamber of the European beeeater, Resp. Physiol., 87, 407418.
Ashie, Y. and Kono, T. (2011) Urban-scale
CFD analysis in support of a climate-

450

sensitive design for the Tokyo

Bay area, Int. J. Climatol., 31,
174188.
ASHRAE (2009) ASHRAE Handbook. Fundamentals, SI Edition, Chapter 20,
Atlanta, Space Air Diusion.
ASME Editorial Board (1994) Journal of
Heat Transfer editorial policy statement
on numerical accuracy, J. Heat Transfer,
116, 797798.
Awbi, H.B. (1989) Application of computational fluid dynamics in room ventilation,
Build. Environ., 24, 7384.

Awbi, H.B. (2003) Ventilation of Buildings,

2nd edn, London, Spon press.
Axley, J. (2007) Multizone airflow modeling
in buildings: history and theory, HVAC &
R Res., 13, 907928.
Beghein, C., Jiang, Y. and Chen, Q. (2005)
Using large eddy simulation to study
particle motions in a room, Indoor Air,
15, 281290.
Berrouk, A.S., Lai, A.C.K., Cheung, A.C.T.
and Wong, S.L. (2010) Experimental
measurements and large eddy simulation
of expiratory droplet dispersion in a

CFD and ventilation research

mechanically ventilated enclosure with
thermal effects, Build. Environ., 45, 371
379.
Bjerg, B., Svidt, K., Zhang, G., Morsing, S.
and Johnsen, J.O. (2002) Modeling of air
inlets in CFD prediction of airflow in
ventilated animal houses, Comput.
Electron. Agric., 34, 223235.
Bjrn, E. and Nielsen, P.V. (2002) Dispersal
of exhaled air and personal exposure in
displacement ventilated rooms, Indoor
Air, 12, 147164.
Bolashikov, Z.D. and Melikov, A.K. (2009)
Methods for air cleaning and protection
of building occupants from airborne
pathogens, Build. Environ., 44, 1378
1385.
Brohus, H., Balling, K.D. and Jeppesen, D.
(2006) Influence of movements on contaminant transport in an operating room,
Indoor Air, 16, 356372.
Chao, C.Y.H. and Wan, M.P. (2006) A study
of the dispersion of expiratory aerosols in
unidirectional download and ceilingreturn type airflows using a multiphase
approach, Indoor Air, 16, 296
312.
Cheesewright, R., King, K.J. and Ziai, S.
(1986) Experimental data for the validation of computer codes for the prediction
of two-dimensional buoyant cavity flows.
In: Humphrey, J.A.C., Avedisian, C.T.,
Le Tourneau, B.W. and Chen, M.M. (eds)
Signicant Questions in Buoyancy Aected
Enclosure or Cavity Flows, New York,
ASME, 7581.
Chen, Q. (1996) Prediction of room air
motion by Reynolds-Stress models, Build.
Environ., 31, 233244.
Chen, Q. (2009) Ventilation performance
prediction for buildings: a method overview and recent applications, Build.
Environ., 44, 848858.
Chen, Q. and Srebric, J. (2001) Simplified
diffuser boundary conditions for numerical room airflow models. ASHRAE RD1009.
Choi, J.I. and Edwards, J.R. (2008) Large
eddy simulation and zonal modeling of
human-induced contaminant transport,
Indoor Air, 18, 233249.
Chow, W.K. (1996) Application of Computational Fluid Dynamics in building
services engineering, Build. Environ., 31,
425436.
Christoforou, C.S., Salmon, L.G. and Cass,
G.R. (1996) Air exchange within the
Buddhist cave temples at Yungang,
China, Atmos. Environ., 30, 39954006.
Craven, B.A. and Settles, G.S. (2006) A
computational and experimental investigation of the human thermal plume,
J. Fluids Eng., 128, 12511258.
Croome, D.J. and Roberts, B.M. (1981) Air
conditioning and Ventilation of Buildings,
2nd edn, Oxford, Pergamon Press.
England, M.H. (1995) The age of water and
ventilation timescales in a global ocean

model, J. Phys. Oceanogr., 25, 2756

2777.
Etheridge, D. and Sandberg, M. (1996)
Building Ventilation: Theory and
Measurement, Chichester, John Wiley &
Sons.
Ferziger, J.H. and Peric, M. (1999) Computational Methods for Fluid Dynamics,
Berlin, Germany, Springer Verlag.
Fontaine, J.R., Rapp, R., Koskela, H. and
Niemela, R. (2005) Evaluation of air diffuser flow modelling methods experiments and computational fluid dynamics
simulations, Build. Environ., 40, 377
389.
Freitas, C.J. (2002) The issue of numerical
uncertainty, Appl. Math. Model., 26, 237
248.
Fujii, K. (2005) Progress and future prospects of CFD in aerospace-wind tunnel
and beyond, Prog. Aerosp. Sci, 41, 455
470.
Gan, G. (1995) Evaluation of room air
distribution systems using computational
fluid dynamics, Energ. Buildings, 23, 83
93.
Gao, N.P. and Niu, J.L. (2005) CFD study
of the thermal environment around a
human body: a review, Indoor Built
Environ., 14, 516.
Gao, N.P. and Niu, J.L. (2006) Transient
CFD simulation of the respiration process
and inter-person exposure assessment,
Build. Environ., 41, 12141222.
Gosman, A.D., Nielsen, P.V., Restivo, A.
and Whitelaw, J.H. (1980) The flow
properties of rooms with small ventilation
openings, J. Fluids Eng., 102, 316322.
Hang, J. and Li, Y. (2010) Wind conditions
in idealized building clusters: macroscopic
simulations using a porous turbulence
model, Boundary-Layer Meteorol., 136,
129159.
Hang, J., Sandberg, M. and Li, Y. (2009a)
Age of air and air exchange efficiency in
idealized city models, Build. Environ., 44,
17141723.
Hang, J., Sandberg, M. and Li, Y. (2009b)
Effect of urban morphology on wind
condition in idealized city models, Atmos.
Environ., 43, 869878.
Hansell, M.H. (1993) The ecological impact
of animal nests and burrows, Funct. Ecol.,
7, 512.
Heiselberg, P., Li, Y., Andersen, A., Bjerre,
M. and Chen, Z. (2004) Experimental and
CFD evidence of multiple solutions in a
naturally ventilated building, Indoor Air,
14, 4354.
Hunt, G.R. and Linden, P.F. (2005) Displacement and mixing ventilation driven
by opposing wind and buoyancy, J. Fluid
Mech., 527, 2755.
Ji, Y. and Cook, M.J. (2007) Numerical
studies of displacement natural ventilation in multi-storey buildings connected
to an atrium, Build. Serv. Eng. Res.
Technol., 28, 207222.

Jones, P.J. and Whittle, G.E. (1992) Computational fluid dynamics for building air
flow prediction - current status and
capabilities, Build. Environ., 27, 321338.
Karimipanah, T. and Awbi, H.B. (2002)
Theoretical and experimental investigation of impinging jet ventilation and
comparison with wall displacement ventilation, Build. Environ., 37, 13291342.
Kato, S. and Huang, H. (2009) Ventilation
efficiency of void space surrounded by
buildings with wind blowing over built-up
urban area, J. Wind Eng. Ind. Aerod., 97,
358367.
Kato, S., Murakami., S., Shoya, S., Hanyu,
F. and Zeng, J. (1995) CFD analysis of
flow and temperature fields in atrium with
ceiling height of 130 m, ASHRAE Trans.,
101, 11441157.
Kato, S., Murakami, S., Takahashi, T. and
Gyobu, T. (1997) Chained analysis of
wind tunnel test and CFD on cross ventilation of large-scale market building,
J. Wind Eng. Ind. Aerod., 6768, 573587.
Koestel, A. (1955) Paths of horizontally
projected heated and chilled air jets,
ASHRAE Trans., 61, 213232.
Kondo, Y. and Nagasawa, Y. (2002) Modeling of a complex air diffuser for CFD
simulation, Part 2: air diffuser model
based on airflow data obtained by
unstructured CFD simulation Roomvent
2002, Copenhagen, Denmark.
Kowalczk, A.J. and Froelich, P.N. (2010)
Cave air ventilation and CO2 outgassing
by radon-222 modeling: how fast do caves
breathe? Earth Planet. Sci. Lett., 289,
209219.
Lai, A.C.K. and Cheng, Y.C. (2007) Study of
expiratory droplet dispersion and transport using a new Eulerian modeling
approach, Atmos. Environ., 41, 7473
7484.
Li, Y. (1997) Wavenumber-extended highorder upwind-biased finite difference
schemes for convective scalar transport,
J. Comput. Phys., 133, 235255.
Li, Y. and Delsante, A. (1998) On natural
ventilation of a building with two openings. In: Proceedings of the 19th AIVC
Conference: Ventilation Technologies in
Urban Areas, Oslo, Norway, 2830
September, pp. 188196.
Li, Y. and Delsante, A. (2001) Natural ventilation induced by combined wind and
thermal forces, Build. Environ., 36, 5971.
Li, Y. and Holmberg, S. (1994) General flow
and thermal-boundary conditions in indoor air flow simulation, Build. Environ.,
29, 275281.
Li, Y., Sandberg, M. and Fuchs, L. (1993)
Numerical prediction of airflow and heatradiation interaction in a room with displacement ventilation, Energ. Buildings,
20, 2743.
Li, Y., Huang, X., Yu, I.T.S., Wong, T.W.
and Qian, H. (2005) Role of air distribution in SARS transmission during the

451

Li & Nielsen
largest nosocomial outbreak in Hong
Kong, Indoor Air, 15, 8395.
Li, Y., Xu, P., Qian, H., Deng, Q. and Wu, J.
(2006) Flow bifurcation due to opposing
buoyancy in two vertically-connected
open cavities, Int. J. Heat Mass Transf.,
49, 32983312.
Lin, Z., Chow, T.T., Fong, K.F., Wang, Q.
and Li, Y. (2005) Comparison of performances of displacement and mixing ventilations. Part I: thermal comfort, Int. J.
Refrig., 28, 276287.
Liow, L.H., Fortelius, M., Lintulaakso, K.,
Mannila, H. and Stenseth, N.C. (2009)
Lower extinction risk in sleep-or-hide
mammals, Am. Nat., 173, 264272.
Liu, X. and Zhai, Z. (2008) Location identification for indoor instantaneous point
contaminant source by probability-based
inverse Computational Fluid Dynamics
modeling, Indoor Air, 18, 211.
Lu, W., Howarth, A.T., Adam, N. and Riffat, S.B. (1996) Modelling and measurement of airflow and aerosol particle
distribution in a ventilated two-zone
chamber, Build. Environ., 31, 417423.
Maclean, G.S. (1981) Factors influencing the
composition of respiratory gases in
mammal burrows, Comp. Biochem.
Physiol., Part A Mol. Integr. Physiol., 69,
373380.
Miller, S.D., Goulden, M.L. and da Rocha,
H.R. (2007) The effect of canopy gaps on
subcanopy ventilation and scalar fluxes in
a tropical forest, Agr. Forest. Meteorol.,
142, 2534.
Mirzaei, P.A. and Haghighat, F. (2010) A
novel approach to enhance outdoor air
quality: Pedestrian ventilation system,
Build. Environ., 45, 15821593.
Mo, J., Zhang, Y. and Yang, R. (2005)
Novel insight into VOC removal performance of photocatalytic oxidation
reactors, Indoor Air, 15, 291300.
Mochida, A., Yoshino, H., Takeda, T.,
Kakegawa, T. and Miyauchi, S. (2005)
Methods for controlling airflow in and
around a building under cross-ventilation
to improve indoor thermal comfort,
J. Wind Eng. Ind. Aerod., 93, 437
449.
Mullejans, H. (1966) The similarity between
non-isothermal flow and heat transfer in
mechanically ventilated rooms, BSRIA
Trans., 202, Bracknell, UK (Cited in
Awbi, 2003).
Murakami, S. (1998) Overview of turbulence
models applied in CWE-1997, J. Wind
Eng. Ind. Aerod., 74-6, 124.
Murakami, S. (2004) Analysis and design of
micro-climate around the human body
with respiration by CFD, Indoor Air, 14,
144156.
Murakami, S., Ohira, N. and Kato, S. (2000)
CFD analysis of a thermal plume and the
indoor air flow using k- models with
buoyancy effects, Flow, Turbul. Combust.,
63, 113134.

452

Murakami, S., Kato, S., Ito, K. and Zhu, Q.

(2003) Modeling and CFD prediction for
diffusion and adsorption within room
with various adsorption isotherms, Indoor
Air, 13, 2027.
Nielsen, P.V. (1973) Berechnung der Luftbewegung in einem zwangsbelufteten
Raum, Gesundheits-Ingenieur, H10, 94,
299302.
Nielsen, P.V. (1974) Flow in air conditioned
rooms - model experiments and numerical
solutions of the flow equations, PhD
thesis, Technical University of Denmark,
Denmark.
Nielsen, P.V. (1992a) Air distribution systems: room air movement and ventilation
effectiveness, International Symposium
on Room Air Convection and Ventilation
Effectiveness, Tokyo, Japan, ISRACVE.
Nielsen, P.V. (1992b) The description of
supply openings in numerical models for
room air distribution, ASHRAE Trans.,
98, 963971.
Nielsen, P.V. (2004) Computational fluid
dynamics and room air movement, Indoor
Air, 14, 134143.
Nielsen, P.V. (2010) www.cfd-benchmarks.com, IEA Annex 20 2D test case,
last accessed on September 27, 2010.
Nielsen, P.V., Restivo, A. and Whitelaw,
J.H. (1979) Buoyancy-affected flows in
ventilated rooms, Numer. Heat Tr.
A-Appl., 2, 115127.
Nielsen, P.V., Allard, F., Awbi, H.B.,
Davidson, L. and Schalin, A. (2007)
Computational Fluid Dynamics in
Ventilation Design: REHVA Guidebook
No. 10, Rehva, Forssa, Finland.
Nitta, K. (1996) Study on the variety of
theoretical solutions of ventilation network, J. Archit. Plan. Environ. Eng.
(Trans. AIJ), 480, 3138.
Niu, J. and Tung, T.C.W. (2008) On-site
quantification of re-entry ratio of ventilation exhausts in multi-family residential
buildings and implications, Indoor Air, 18,
1226.
Noakes, C.J., Sleigh, P.A., Escombe, A.R.
and Beggs, C.B. (2006) Use of CFD
analysis in modifying a TB ward in Lima,
Peru, Indoor Built Environ., 15, 4147.
Oberkampf, W.L. and Trucano, T.G. (2002)
Verification and validation in computational fluid dynamics, Prog. Aerosp. Sci.,
38, 209272.
Oliver, H.R. (1975) Ventilation in a forest,
Agr. Meteorol., 14, 347355.
Olmedo, I. and Nielsen, P.V. (2010) Analysis
of the IEA 2D test. 2D, 3D, steady or
unsteady airow? DCE Technical
Reports, nr. 106, Aalborg, Aalborg
University, Department of Civil
Engineering.
Orszag, S.A. and Staroselsky, I. (2000) CFD:
progress and problems, Comput. Phys.
Commun., 127, 165171.
Peng, S.H., Holmberg, S. and Davidson, L.
(1997) On the assessment of ventilation

performance with the aid of numerical

simulations, Build. Environ., 32, 497508.
Qian, H. and Li, Y. (2010) Removal of
exhaled particles by ventilation and deposition in a multibed airborne infection
isolation room, Indoor Air, 20, 284297.
Qian, H., Li, Y., Nielsen, P.V., Hyldgaard,
C.E., Wong, T.W. and Chwang, A.T.Y.
(2006) Dispersion of exhaled droplet nuclei in a two-bed hospital ward with three
different ventilation systems, Indoor Air,
16, 111128.
Rajaratnam, N. (1976) Turbulent Jets,
Elsevier, Amsterdam.
Restivo, A. (1979) Turbulent ows in ventilated rooms, PhD thesis, Department of
Mechanical Engineering, Imperial
College, London.
Rigby, M., Timmis, R. and Toumi, R. (2006)
Similarities of boundary layer ventilation
and particulate matter roses, Atmos.
Environ., 40, 51125124.
Roache, P.J. (1997) Quantification of the
uncertainty in computational fluid
dynamics, Annu. Rev. Fluid Mech., 29,
123160.
Roache, P.J. (1998) Verication and Validation in Computational Science and
Engineering, Albuquerque, NM, Hermosa
Publishers.
Roache, P.J., Chia, K. and White, F. (1986)
Editorial policy statement on the control
of numerical accuracy, J. Fluids Eng., 108,
2.
Russo, J.S. and Khalifa, H.E. (2010) CFD
assessment of intake fraction in the
indoor environment, Build. Environ., 45,
19681975.
Russo, J.S., Dang, T.Q. and Khalifa, H.E.
(2009) Computational analysis of
reduced-mixing personal ventilation jets,
Build. Environ., 44, 15591567.
Sagaut, P. (2001) Large Eddy Simulation for
Incompressible Flows: An Introduction,
Berlin, Germany, Springer Verlag.
Schalin, A. and Nielsen, P.V. (2004) Impact
of turbulence anisotropy near walls in
room air flow, Indoor Air, 14, 159168.
Shams, I., Avivi, A. and Nevo, E. (2005)
Oxygen and carbon dioxide fluctuations
in burrows of subterranean blind mole
rats indicate tolerance to hypoxichypercapnic stresses, Comp. Biochem. Physiol.,
Part A Mol. Integr. Physiol., 142, 376
382.
Sivertsen, B., Lamb, B. and Grnskei, K.E.
(1983) A tracer study of pollutant
transport in a deep, Fjord valley, Atmos.
Environ., 17, 19151922.
Srensen, D.N. and Nielsen, P.V. (2003a)
Guest editorial: CFD in Indoor Air,
Indoor Air, 13, 1.
Srensen, D.N. and Nielsen, P.V. (2003b)
Quality control of computational fluid
dynamics in indoor environments, Indoor
Air, 13, 217.
Straub, H.E., Gilman, S.F. and Konzo, S.
(1956) Distribution of air within a room

CFD and ventilation research

for year-around air conditioning Part II,
University of Illinois Engineering Experiment Station Bulletin 435. (Cited in
ASHRAE, 2009).
Szczena, M., Zhang, X., Chacinski, G.,
Malasek, L., Jurelionis, A. and Nielsen,
P.V. (2005) Private communication,
Aalborg, Denmark, Aalborg University.
Tan, G. and Glicksman, L.R. (2005) Application of integrating multi-zone model
with CFD simulation to natural ventilation prediction, Energ. Buildings, 37,
10491057.
Tian, L., Lin, Z., Liu, J. and Wang, Q. (2008)
Numerical study of indoor air quality and
thermal comfort under stratum ventilation, Prog. Comput. Fluid Dyn., 8, 541
548.
Tritton, D.J. (1988) Physical uid dynamics,
Oxford, Clarendon press.
Turner, J.S. (1994) Ventilation and thermal
constancy of a colony of a southern
African termite, J. Arid Environ., 28, 231
248.

Vogel, S., Ellington, C.P. and Kilgore, D.L.

(1973) Wind-induced ventilation of the
burrow of the prairie-dog, Cynomys
ludovicianus, J. Comp. Physiol. A., 85,
114.
Wesseling, P. (2000) Principles of Computational Fluid Dynamics, Berlin, SpringerVerlag.
Wilcox, D.C. (2006) Turbulence Modeling for
CFD, La Canada, CA, USA, DCW
Industries.
Wilson, K.J. and Kilgore, D.L. (1978) The
effects of location and design on the
diffusion of respiratory gases in
mammal burrows, J. Theor. Biol., 71,
73101.
Yang, L. and Li, Y. (2009) City ventilation of
Hong Kong at no-wind conditions,
Atmos. Environ., 43, 31113121.
Yang, X., Srebric, J., Li, X. and He, G.
(2004) Performance of three air distribution systems in VOC removal from an
area source, Build. Environ., 39, 1289
1299.

Yang, L., Xu, P. and Li, Y. (2006) Nonlinear

dynamical analysis and solution multiplicity study of natural ventilation in a
two-zone building: Part B- CFD Simulations, HVAC & R Res., 12, 231256.
Yu, I.T.S., Li, Y.G., Wong, T.W., Tam, W.,
Chan, A., Lee, J.H.W., Leung, D.Y.C.
and Ho, T. (2004) Evidence of airborne
transmission of the Severe Acute Respiratory Syndrome virus, N. Engl. J. Med.,
350, 17311739.
Zhai, Z. (2006) Application of computational
fluid dynamics in building design: aspects
and trends, Indoor Built Environ., 15, 305
313.
Zhang, T. and Chen, Q. (2007) Identification
of contaminant sources in enclosed spaces
by a single sensor, Indoor Air, 17, 439
449.
Zhu, S.W., Kato, S. and Yang, J.H. (2006)
Study on transport characteristics of saliva droplets produced by coughing in a
calm indoor environment, Build. Environ.,
41, 16911702.

453