J. Biomedical Science and Engineering, 2010, 3, 653-663 JBiSE
doi:10.4236/jbise.2010.37089 Published Online July 2010 (http://www.SciRP.org/journal/jbise/).
Published Online July 2010 in SciRes. http://www.scirp.org/journal/jbise
Modelling infection spreading control in a hospital isolation
room
Carla Balocco, Pietro Liò
1Department of Energy Engineering “Sergio Stecco”, Firenze, Italy;
2Computer Laboratory, University of Cambridge, Cambridge, UK.
Received 9 May 2010; revised 21 May 2010; accepted 27 May 2010.
ABSTRACT
This paper investigates the airflow patterns con-
nected to different cough conditions, the effects of
these arrangements on the regions of droplet fallout
and dilution time of virus diffusion of coughed gas.
We focus on some of the physical processes that occur
in a double bed hospital isolation room, investigating
the effect of the ventilation system on the spread of
particles in air. A cough model was carried out and
used for the numerical simulation of virus diffusion
inside an existent isolation room. Transient simula-
tions of air pattern diffusion and air velocity field,
provided by the existing typical HVAC primary air
system designed for infectious patients, were per-
formed using CFD. A multiphysics approach, com-
bined Convection-Conduction, Incompressible Na-
vier-Stokes models on non-isothermal air flow and
Convection-Diffusion, was used. Simulations results
highlighted that the flow field and velocity distribu-
tion induced by the high turbulence air inlet diffuser
combined with the air return diffusers produce wide
recirculation zones near the wall and partial stagna-
tion areas near the ceiling and between the two beds,
but lower particle concentration in the room and
their shorter spreading distance. This type of analysis
is certainly cost effective to identify all the air recir-
culation zones which can harbour lingering patho-
gens.
Keywords: Airborne Diffusion; Hospital Anfection; VAV
System; Transient Simulation; CFD
1. INTRODUCTION
The risk of virus particles dispersal in hospitals mainly
depends on airflow patterns and on airflow directions
changes caused by people’s activity, e.g. moving or
opening doors. The ventilation system of isolation rooms,
operating under closed-door conditions is crucial if the
viruses spread and infection must be contained. The
quality of the hospital environment is provided by an
efficient air-conditioning and ventilating system design,
in controlling temperature, humidity, pressure, and in-
door air quality. The Heating Ventilation Air Condition-
ing systems (HVAC) design is fundamental to maintain
negative pressure within isolation rooms, to protect heal-
th of workers, patients and visitors. This is also neces-
sary to control patient risk from airborne diseases.
In recent years few works have focused on computa-
tional models using fluid dynamics approaches to inves-
tigated airflow patterns and the related spreading of in-
fection in isolation rooms for different ventilation sys-
tems (for example operating under open or closed-door
conditions) [1-4]. The main attention of these papers has
been the evaluation of the effects of negative pressure in
isolation rooms accommodating patients with highly
infectious diseases. Recent studies have highlighted that
an air velocity above 0.2 m s-1 via a doorway effectively
prevents the spread of airborne contaminants out of the
isolation room in the state of door opening [3,5]. HVAC
switch on-off impact on virus and bacteria load has
widely investigated recently [6-10].
Techniques such as aerosol particle tracer sampling
and computational fluid dynamics can be applied to
study the performance of ventilation systems during
coughing episodes [4]. HVAC operating for hospitals,
must establish optimal airflow pattern inside isolation
rooms for infective and in particular for immune-
suppressed patients, such that clean air from the air-
supply vents may carry the air across infectious sources,
and then flow through the exhaust vents completely [2,3].
However, it is not sufficient to provide clean environ-
ments, due to the higher cross-infections risk, useful
guidelines on public health strategies and management
can be obtained studying the risky environments by
CFD-FEM simulation. In particular CFD simulations
based on multiphysics approach can provide a good pre-
dictive efficacy, since the possibility to achieve in hos-

C. Balocco et al. / J. Biomedical Science and Engineering 3 (2010) 653-663 654
pital stays, accurate experimental measurements is lim-
ited and subjected to high costs and difficulties of prac-
tical realization. In this paper, referring to obtained re-
sults in [11], we developed new transient simulations
combining thermo-fluidynamics and diffusion models
with heat transfer analysis (convection and conduction).
Moreover, we defined a particle tracing and diffusion
model taking into account horizontal and vertical com-
ponents of velocity to carry out a coughing function.
Transient simulations were performed to investigate the
air flow patterns, distribution and velocity, and the par-
ticulate dispersion inside an existing typical hospitaliza-
tion room equipped with a HVAC primary air system
designed for infected patients. We analysed three-dimen-
sional models of the room considering different posi-
tions of the patients.
2. PARTICLES TRACING AND
DIFFUSION
A dispersion model of droplets carrying viruses due to
coughing and sneezing was carried out. We believe that
our level of modelling details is an effective trade off
between accurate description of the chemistry physics
process and computational resource. A theoretical gen-
eral framework would consider the mixture of viruses/
bacteria and water as a two-phase fluid; a complete de-
scription would specify virus/bacterial aggregates/
colonies in terms of coupled reaction-diffusion equations
for the bacterial and water.
These equations are general in the sense that the reac-
tion, diffusion, pressure, stresses and external forcing
terms can be chosen differently for different types of mi-
cro-organisms. Further details may take into account di-
fferent shapes of bacteria (round, spirals, ect) and float-
ing behaviour (stroke-averaged swimmer, shakers” (not
self-propelled) and “movers” (self-propelled particles),
“pushers” (most bacteria) and “pullers” (for example
some algae). Coughing and sneezing produce droplets in
size range from less than 1 to up to 2000 m.
Typical diameter of a virus, bacteria and fungal spore
is respectively 0.02  0.3 m, 0.3  10 m and 2  5 m.
Referring to literature we define large droplets those
with diameter larger than 60 m, and small ones if the
diameter is less than 60 m and nucleus droplets when
they are less than 10 m. Winter season provides opti-
mal conditions for airborne influenza spreading. Coughs
and sneezing releases droplets that in cold and low hu-
midity conditions dry out partially the droplets, reducing
their size, so to allows the virus to linger longer in the air
and the nasal passages making virus transmission more
likely. Therefore a multiphysic approach to take into
account for temperature variations (in the room) is of key
importance. Pathogenic bacteria are generally transported
on larger droplets such as skin flakes (13  17 m di-
ameter) or droplets (ejected with peak diameter 20  40
m) coughing/sneezing generates 104  105 droplets. The
potential for fomites to remain in the air is due to their
final fall velocity that varies with the square of their di-
ameter [4]. Finer particles can remain in the air for long
period while larger particles usually sediment to surfaces
in those regions of slowly moving air. Particles of 10 m
which are responsible of the contagious up to 2 meters
distance were taken into account for all the simulations.
A commonly used factor is the Cunningham Slip Cor-
rection (Cc) which is a correction to the drag coefficient
that is used to predict the drag force between a fluid and
a particle moving through this fluid. The drag coefficient
on each particle must be divided by the Cunningham
correction factor, i.e. molecular slip correction that oc-
curs when the size of the particle is of the same magni-
tude as the distance between gas and water molecules.
The particle no longer moves as a continuum in the gas,
but as a particle among discrete gas molecules thereby
reducing the drag force. The correction factor is greater
than 1 which means that the effective particle drag coef-
ficient goes down: this reduction in particle drag is the
“particle slip” [12,13]:
12.34 1.05exp0.39pa
c
pa
D
CD





 

 




 (1)
We considered the free mean path  of a particle, the
average distance the particle travels between collisions
with other moving particles. The following boundary
conditions, that were taken into account, produced a re-
duction of the Cunningham factor towards its lower
boundary of 1: the positive pressure of the isolation
room reduces slightly the mean free path with respect to
atmospheric pressure; the droplets are initially larger and
the evaporation reduces their dimension; the droplets
contain glycoproteins, lipidoproteins and lipoglycids
constituents of the mucous and also viral particles (and
possibly, bacteria); the presence of glycol and lipid-
proteins could increase the local viscosity of the droplet.
The droplet is then subjected to two forces: the force of
gravity and the drag force of air resistance [12].
At low velocity v and low Reynolds number the air
resistance f for a droplet of mass m, is mainly propor-
tional to the viscosity of the medium and to the linear
size of the droplet. This provides the horizontal compo-
nent of motion. The vertical component of sedimentation
velocity is:
2
18
p
ac
vertical
DgC
u


 (2)
The horizontal component of the velocity is provides

C. Balocco et al. / J. Biomedical Science and Engineering 3 (2010) 653-663 655
by f (u) = (b • u) and then:
0
bt
m
horizontal
uve

 (3)
where (b/m) has the dimension of time (in seconds as
unit system used in this study). The horizontal and ver-
tical components of velocity are also important for fo-
mites formation, i.e. contaminated environmental sur-
faces which become secondary sources of infection. For
a rain drop (approximated as a sphere) the coefficient b
has the form b = (D • β). For a spherical drop in air at
atmospheric pressure conditions (101325 Pa) and tem-
perature of 22℃, the value of the coefficient β is 1.6•10-4
Ns m-2 [12,13]. The initial concentration is due to the
liquid and then density is 1; then velocity which repre-
sents the mass transport equation solution is introduced
in the software for CFD simulation based on the FEM
method [14] as initial condition. The diffusion coeffi-
cient of the contaminant is considered to be zero
(1.0•10-10) assuming that the effluent, in the form of
aerosol, is immiscible in the air. The progressive dilution
of the biological effluent concentration is connected to
the mass transport mechanism due to the ventilation
plant. An expression that provides the cough velocity
function (Co), taking into account both the horizontal
and vertical components, was carried out:
 

1
0.07812070.113835exp 15.2864
Co


  (4)
At the initial instant is equal to 28 ms-1. For the case
of mass transport with diffusion we used the following
expression:


2
2
2
HO in
HO
HO
u
Nmm

 
 (5)
assuming for the water molecular mass 18•10-3 kg mol-1.
3. SOLID MODELLING
A full-scale isolation room, present in four hospitals in
Tuscany (Italy) was considered. It represents the most
typical room for hospitalization in Italian hospitals. Solid
model of the room is provided in Figure 1. The ventila-
tion plant is a typical HVAC primary air system de-
signed for immune-depressed hospitalization which pro-
vides 6 vol h-1 of inlet air and 7 vol h-1 of extraction air
[15-18]. The room is ventilated by a commercial turbu-
lence high induction air diffuser located in the centre of
the ceiling and in the middle area between the two beds,
indicated with A (Figure 1) and the exhaust air is ex-
pelled by three return air diffusers located in the ceiling
B, in the door of the toilet C and in the door adjacent to
the corridor D.
To ensure effective control of virus containment in the
A
B
D
C
Figure 1. The 3D model of the isolation room.
isolation room, the door must be closed as often as prac-
tical so that the air flow can be maintained in a stable
state. The scheme of the constant pressures inside the
room was calculated referring to this data and to the air
diffusers dimensions. For the air inlet, the ceiling return
air and the door return air dimensions (transversal sec-
tion), set constant pressure and air flow rates are respec-
tively the following: 0.36 m2, pressure 26 Pa and 480
m3h-1; 0.157 m2, 7 Pa and 420 m3h-1; 0.125 m2, 18 Pa
and 130 m3h-1; 0.125 m2, 18 Pa and 70 m3h-1. The ex-
haust vents of the ventilation system were set at a con-
stant outlet pressure to maintain the negative pressure
within the room as imposed for infectious patients
[15,16,18].
The inlet turbulence high induction air diffuser was
modelled and gradually changed after several attempts
testing the simulation results obtained by [14], to control
the real diffusion and turbulence effect of the inlet air on
the ceiling, at its fixed velocity. The head of the two pa-
tients, lying on two parallel beds, was modelled using a
solid geometry to take into account their different posi-
tions. On the head surface three outflow surfaces were
modelled and used for simulate the inlet surface of the
organic effluent (mouth) related to the different position
of each patient (Figure 2).
Typical positions of the patients and their different
postural conditions (i.e. the first was considered cough-
ing and the second breathing) were modelled: model I
for the “supine position”, model II for the “face to face”
position as while talking each other, and model III for
the “overleaf to overleaf” position. For model simplifi-
cation we considered sneezing and/or coughing once, in
simulation time interval, and its process as pulse. The
two bed-headboard lamps, the safety lamps on the ceil-
ing, the external opaque wall and the french-window
were also modelled. For all the three 3D models the door
was kept closed, such that particles diffusion is influ-

C. Balocco et al. / J. Biomedical Science and Engineering 3 (2010) 653-663 656
Figure 2. Patients head geometry; head detail (right).
enced by ventilation airflow patterns alone.
4. SIMULATIONS
To investigate the temporal dynamics of the ventilation
flow and the particles tracing in the conditions of cough-
ing and breathing of the two patients, time dependent
simulations combining Incompressible Navier-Stokes,
Convection- Diffusion on non-isothermal air flow and
Convection-Conduction models, were performed on the
three 3D models. First extensive transient simulations
were performed employing coarse and normal mesh,
checking its quality to obtain solutions with acceptable
accuracy. The mesh density was selected to combine
solution accuracy with reduction of computational time
needed for convergence: a good quality was obtained by
168000 degrees of freedom with 38000 tetrahedral ele-
ments for a no-structured grid. Numerical integration is
implicit backward Euler method with BDF multistep
which guaranteed the quality and robustness of solution:
this method provided by [14] automatically chooses the
time step on each iteration with checks and controls on
the tolerance imposed to the error and to the stability
region. The stop-convergence criterion was chosen 1*
10-4 with independence of the results to the mesh density
less than 5% compared to the reference mesh. The direct
system solver “UMFPACK” of Unsymmetrical Multi-
Frontal was used. The initial conditions for this transient
computation were obtained by running the simulation
starting from a stationary analysis. These first results
provided 102  103 for Reynolds number which implies a
solution under laminar flow conditions. For the indoor
climatic conditions, a uniform internal air temperature of
22℃ and 50% of relative humidity were assumed as
provided by the plant system and suggested [15,16].
External climatic data were set at the external design
conditions of 0℃ air temperature and 85% relative hu-
midity. Transient simulations were performed consider-
ing the coughing events during 10 seconds and then they
were also carried out for 60 seconds. Input data, sub-
domains settings, boundary conditions and basic equa-
tions used for transient simulations are provided in the
appendix. Due to computational time and complexity,
the optimization of system solver was carried out: the
final computational time for convergence was about 7
hours for each model, using a Processor Pentium 7 Quad
Core with 4 GB Ram.
5. RESULTS
The ventilation and air diffusion pattern inside the room,
at different height, with no coughing and breathing
events the patients, is provided in Figure 3. Noteworthy,
the flow field and velocity distribution induced by the
high turbulence air inlet diffuser combined with the air
return diffusers produce wide recirculation zones near
the wall and small partial stagnation areas near the ceil-
ing and between the two beds. Variable directions of the
air flow due to the high turbulence air inlet diffuser pro-
vide its widespread and homogeneous distribution
(coanda effect) on the ceiling and then its progressive
drop down.
The mean air velocity value of the central diffuser,
calculated using simulation results, is 1.36 ms-1, there-
fore even with reduced surface diffusion, the airflow is
laminar the air flow adherence to the ceiling and its fall
down. In the zone under the safety light at ceiling, the
central inlet air diffuser A and the air return diffuser B, it
can be noted the air velocity increasing. Results about
temperature distribution inside the room provides values
between 22℃ and 26℃ due to the “chimney effect”
and heat transfer near the heat sources (radiant panel,
lamps and patients head; Figure 4). Noteworthy, for
small and low temperature variations the effects on air
distribution and particle diffusion are not negligible in
particular for 60 seconds of simulation time. Thermal
buoyancy produces the “chimney effect” even close to
the head of patients (vertical component of velocity rep-
resented on the surface; Figure 5). The central inlet air
diffuser provides turbulent air distribution near the ceil-
ing and its low and progressive downfall. The flow in
Figure 3. Velocity field at different height; 1 m, 1.5 m, 2.5 m,
2 m from bottom-right side to the upper-right side (clockwise).

C. Balocco et al. / J. Biomedical Science and Engineering 3 (2010) 653-663 657
(a)
(b)
(c)
Figure 4. (a) Temperature field on the wall; (b) slice-detail; (c)
entire room.
the region between the three opposite air return diffusers
interacts with the local unstable flows, resulting in the
generation of an up-flow draft. Particles diffused in the
ambient by patient coughing (10 s simulation time)
mainly remain in the zone of emission sources. The
Figure 5. Velocity field – vertical component “z” (left) and
detail near the bed (right).
analysis of particle tracing and path distribution con-
nected to their concentration proved the formation of a
zone, even though restrained, that should be checked
with a microclimatic and contaminant control between
the two beds and around the ceiling surface. In the area
near the beds the plant guarantees the maximum admis-
sible indoor value of the air velocity and air flow pat-
terns.
The VAV system provides lower particle concentration
and shorter spreading distance of particles (both during
10 s and 60 s of coughing event); this implies that the
transport distance of particle-droplets generated by nor-
mal respiration and coughing of patients is limited.
Concerning the dynamics of transient air flow: in the
all the 3D models near the mouth of the coughing patient
the local air velocity reaches the expected value of 28
ms-1; the transient flow due to the impulse has a length
of about 0.5 seconds. About the effluent concentration:
in model I the effluent concentration in the form of
aerosol achieves the higher values in the time interval of
2-4 seconds and then decreases with time because the

C. Balocco et al. / J. Biomedical Science and Engineering 3 (2010) 653-663 658
continuous dilution provided by the air flow inlet; in the
model II the effluent concentration, achieves the higher
values in the time interval of 3-5 seconds, but during the
transient air flow displacement a local effluent stagna-
tion zone is produced between the two beds and small
recirculation flows are generated near the patients (Fig-
ure 6); in the model III this effluent stagnation reaches
the higher values between 4-5 seconds. For the particle
path and distance: in the model I, during the first 4 sec-
onds particles collide against the headboard and the
head-bed lamp; the trajectories of other particles, in the
successive time instants, provide a particle path toward
the centre of the room; in the model II, during 10 sec-
onds particles go through 1.6-1.8 meters along the or-
thogonal direction to the second patient, but due to the
buoyancy effect, their path takes up in the area between
the two bed; in the model III in about four seconds the
particles go through 1.4 meters along the orthogonal
direction but they do not collide against the wall direct-
ing to the floor at the end of transitory. About particle
tracing: the particle tracing obtained from the three tran-
sient simulations highlights the position during time of
the particles due to cough with a diameter of about
1*10-5 m; the obtained trajectories (Figures 7 and 8) are
referred to those particles which position at the initial
instant overlap with the surface of the mouth of the first
coughing patient and the second breathing. Particles
tracing and diffusion obtained by 60 seconds of transi-
tory are different from those obtained by 10 seconds for
all the three 3D models.
Particles paths are modified by the inertial/gravitational
reduction effect due to thermal buoyancy combined with
the air flow displacement in the room. As a matter of fact,
the vertical component of particles velocity is prevailing.
Comparing results obtained the transient simulations
carried out for 60 s:
1) in the model I at the end of 60 s of transient simula-
tion, particles path after crossing the bed-head board is
oriented to the ceiling due to the thermal buoyancy and
air mixing effects (Figure 9);
2) in the model II a higher number of particles collide
with the second bed and the lying patient. A small num-
ber of particles are driven to the ceiling, up to about 2 m,
by the air flow displacement and the local turbulence
convective flux. This last effect results comparable to
that obtained in model I (Figure 10);
3) in the model III the all particles during 60 s of tran-
sient simulation collide with the opposite wall. Particles
tracing highlights and follows the descending air flow
due to the combined thermal and convective effects (air
mixing and shorting effect). At the end of transitory the
all particles are deposited on the wall (Figure 11).
4. CONCLUSIONS
Results indicate the best conditions for the high induc-
tion air inlet diffuser and the scheme of pressures im-
posed in the room to provide the effective means of con-
trolling flows of virus-loaded droplets. Our findings
stress that the position of the air-supply inlets in the
ceiling and the exhaust vents at the opposite side of the
Figure 6. Particle concentration (mol m-3) during the coughing and breathing of the patients (time from 3 to 6 seconds) - results
obtained by the Model II transient simulation.

C. Balocco et al. / J. Biomedical Science and Engineering 3 (2010) 653-663 659
.
Figure 7. Model I, particle tracing carried out for 10 seconds;
detail near the bed-headboard lamp (right).
room and in the ceiling, provide an up-draft effect and
infection control efficiency. When the air-supply inlet is
located in the middle of the ceiling of the room, and the
exhaust vents are positioned in the wall behind the pa-
tient beds, the coughed particles and gas are contained at
the side of the patient in the region of the exhaust vents.
This can minimize the region of coughed gas diffusion
and droplet fallout. The high induction air inlet diffuser
and the scheme of pressures imposed in the room pro-
vide the effective means of controlling droplet flows
containing viruses. This type of analysis allows to pre-
dict air recirculation zones which can host pathogens.
Indeed our analysis of the particle tracing and path dis-
tribution combined with their concentration, was effec-
tive in identifying examples of such zones between the
two beds, around the ceiling surface and on the surface
of the bed lamps. Current simulation results yield mean-
(a)
(b)
(c)
Figure 8. Particle tracing carried out for 10 seconds. (a) model
I; (b) model II and (c) model III.

C. Balocco et al. / J. Biomedical Science and Engineering 3 (2010) 653-663 660
Figure 9. Model I – particle tracing carried out for 60 seconds.
Figure 10. Model II; particle tracing carried out for 60 sec-
onds.
Figure 11. Model III; particle tracing carried out for 60 sec-
onds.
ingful findings and recommendations for disease control
and careful design and optimization of ventilation in
hospitals for also preventing cross infection processes.
Our work provides the key insights and developments
for hospital management who would like to control the
architecture room (e.g. dimensions, form, arrangement
of the furniture) the position and dimensions of openings
(doors and windows; windows orientation), in particular,
type and location of supply and return air diffusers con-
nected to the design of the HVAC system (variable or
constant air volume) and the better length and shape of
the air-inlet jet.
5. ACKNOWLEDGEMENTS
This work is supported by the EC IST SOCIALNETS project, grant
agreement number 217141, and the International Technology Alliance,
sponsored by the U.S. Army Research Laboratory and the U.K. Minis-
try of Defense. The authors thank Sani Engineering Consulting Office
(Florence, Italy) for provided data on the hospitalization room and
plant system.
REFERENCES
[1] Tang, J., Li, Y., Eames, I., Chan, P. and Ridgway, G.
(2006) Factors involved in the aerosol transmission of
infection and control of ventilation in healthcare prem-
ises. Journal of Hospital Infection, 64(2), 100-114.
[2] Zhao, B., Yang, C.Q., Chen, C., Chao F., Xu D.Y., Sun,
L.C., Wei, G. and Li Y. (2009) How many airborne parti-
cles emitted from a nurse will reach the breathing
zone/body surface of the patient in ISO Class-5 sin-
gle-bed hospital protective environments? A numerical
analysis. Aerosol Science and Technology, 43(10), 990-
1005.
[3] Stanley, N.J., Kuehn, T.H., Kim, S.W., Raynor, P.C., An-
antharaman, S., Ramakrishnan, M.A. and Goyal, S.M.
(2008) Background culturable bacteria aerosol in two
large public buildings using HVAC filters as long term,
passive, high-volume air samplers. Journal of Environ-
mental Monitor, 10(4), 474-481.
[4] Eames, I., Shoabi, D., Klettner, C.A. and Taban, V. (2010)
“Movement of airborne contaminants in a hospital isola-
tion room,” Journal of the Royal Society Interface, 6757-
6766.
[5] Tunga, Y.-C., Hu, S.-C., Tsaia, T.-I. and Changa, I.-L.
(2009) An experimental study on ventilation efficiency of
isolation room. Building and Environment, 44(2), 271-
279.
[6] Walker, J., Hoffman, P., Bennett, A., Vos, M., Thomas, M.
and Tomlinson, N. (2007) Hospital and community ac-
quired infection and the built environment – design and
testing of infection control rooms. Journal of Hospital
Infection, 65(Suppl 2), 43-49.
[7] Soper, M. (2008) Pandemic ready, HVAC systems for
worst-case scenarios. Health Facilities Management,
21(10), 49-52.
[8] Bonetta, S., Bonetta, S., Mosso, S., Sampò, S. and
Carraro, E. (2009) Assessment of microbiological indoor
air quality in an Italian office building equipped with an
HVAC system. Environmental Monitoring and Assess-
ment, 161(1-4), 18-28.
[9] Rygielski, L. and Uden, D. (2007) Creating comfort.
Nine considerations for selecting the right hospital
HVAC system. Health Facilities Management, 20(1),
19-23.
[10] Grinshpun, S.A., Adhikari, A., Honda, T., Kim, K.Y.,
Toivola, M., Rao, K.S. and Reponen, T. (2007) Control
of aerosol contaminants in indoor air: combining the par-
ticle concentration reduction with microbial inactivation.
Environmental Science Technology, 15(41), 606-612.
[11] Balocco, C., Liò, P. and Sani, L. (2010) Simulazione di
un sistema di ventilazione per il controllo degli agenti
eziologici nei reparti infettivi. Un caso reale. Condi-
zionamento dell’ Aria Riscaldamento Refrigerazione,
CdA n.1 febbraio.
[12] Zhao, B., Zhang, Z. and Li, X.T. (2005) Numerical study
of the transport of droplets or particles generated by res-
piratory systems indoors. Building and Environment,
40(8), 1032-1039.

C. Balocco et al. / J. Biomedical Science and Engineering 3 (2010) 653-663 661
[13] Cunningham, E. (1910) On the velocity of steady fall of
spherical particles through fluid medium. Proceedings of
the Royal Society A, 83(561), 94-96.
[14] www.comsol.com
[15] (1995) UNI 10339, Air-conditioning systems for thermal
comfort in buildings. General, classification and require-
ments. Offer, order and supply specifications. Italian
Standard.
[16] (2005) UNI EN 13779, Ventilation for non-residential
buildings - Performance requirements for ventilation and
room conditioning systems. Italian Standard.
[17] ASHARE-Handbook, HVAC-applications, SI, 1995.
[18] (1991) ASHRAE. Health facilities. ASHRAE Handbook
of Applications. S.I. Edition, Atlanta.

C. Balocco et al. / J. Biomedical Science and Engineering 3 (2010) 653-663 662
APPENDIX
The sub-domains settings and boundary conditions, used
for all the time dependent simulations, combining the
incompressible Navier-Stokes, Convection-Conduction
and Convection-Diffusion models on the non-isothermal
air flow, are provided in this appendix.
The equations used by CFD FEM transient simulation
are the following: Sub-domain settings equation used for
Convection-Conduction model:

p
ts p
CT kTQC uT 



  
 (6)
Sub-domain settings equation used for Convection-
Diffusion model:
Air (fluid)

2
22ts
HO DHOR uHO 
  
 (7)
with H20 = concentration
Sub-domain settings equation used for the Incom-
pressible Navier-Stokes model:
 

T
uuupIu uF 
 


 


(8)
with 0u
The boundary settings used are provided for the Con-
vection-Conduction model (1-5), Convection-Diffusion
model (6-8) and then for the Incompressible Navier-
Stokes model (9-14):
1) “insulation” for all the internal walls and doors, the
two beds, ceiling, floor, the two bed-headboard lamps
and the lamps on the ceiling;
2) “heat flux” for the external wall and the french-
window taking into account for the convective coeffi-
cient and the external air temperature a constant value
respectively 25 W m-2 ℃-1 and 0℃;
3) “temperature” for the radiant panel on the ceiling,
at the fixed value of 26℃ as suggested [15];
4) “convective flux” for all the return air diffusers;
5) “temperature” for the inlet air diffuser, assuming
the constant value of the air inlet of 22℃;
6) “insulation symmetry” linked to all the walls and
doors, the two beds, the ceiling, floor and all the lamps;
7) “convective flux” for all the air return diffuser and
also for the air inlet diffuser;
8) for the “supine position” of the patients (Model I),
for the “face to face” position (Model II) and then for the
“overleaf to overleaf” position (Model III): the boundary
condition associated to the mouth of the patient near the
toilet was “convective flux”; the boundary condition
associated to the mouth of the patient near the wall, was
“Inward flux” linked to the “cough function” evaluated
by the Eq.4;
9) “no-slip” for the all walls and doors, the two beds,
the ceiling and floor;
10) “pressure-no viscous stress” with a boundary type
“outlet” for the air return diffuser B, located in the ceil-
ing (Figure 1);
11) “pressure-no viscous stress” with a boundary type
“outlet” for the air return diffuser C, located in the door
of the toilet (Figure 1);
12) “pressure-no viscous stress” with a boundary type
“outlet” for the air return diffuser D, located in the door
adjacent to the corridor (Figure 1);
13) “pressure-no viscous stress” with a boundary type
“inlet” for the air inlet diffuser A, located in the centre of
the ceiling (Figure 1);
14) For model I, model II and model III: the boundary
type and boundary condition associated to the mouth of
the patient near the toilet were respectively “outlet” and
“velocity” with the fixed velocity (normal outflow) of
0.9 ms-1; the boundary type and boundary condition as-
sociated to the mouth of the patient near the wall, were
respectively “inlet” to take into account his breathe and
“velocity” with a velocity value associated to the cough
function.
Input data used were the following:
1) thermal conductivity, density, viscosity conductiv-
ity and specific heat of the air were considered constant;
2) the density, molecular mass (18*10-3 kg mol-1) and
diffusion coefficient of the water (1*10-10 m
2 s -1) were
considered constant;
3) inlet velocity at the initial instant due to the cough
was considered constant (28 ms-1);
4) the inlet/outlet breathe velocity was considered
constant (0.9 ms-1);
5) inlet air temperature form the central air diffuser
was taken at 22°C as provided by the engineering con-
sulting office;
6) mean ambient temperature and relative humidity
were fixed to the values provided by the plant at 22°C
and 50%;
7) sensible and latent loads due to “rest and sat” activ-
ity, taking into account the clothes and the ambient air
temperature of 22°C, are respectively 76W and 26W for
the standard body area of 1.8 m2 from DuBois formula.
Then taking into account the mean body conductivity
and the head volume, the specific dissipation heat and
total power of the head were evaluated;
8) the pressure scheme inside the room was calculated
referring to the HVAC-VAV primary air system and to
the air diffusers dimensions. The exhaust vents of the
ventilation system were set at a constant outlet pressure
to maintain the negative pressure within the room as
imposed for infectious patients;
9) taking into account the thermo-physical properties

C. Balocco et al. / J. Biomedical Science and Engineering 3 (2010) 653-663 663
of the different constituent layers, the equivalent thermal
transmittance of the external wall and french-window
were evaluated and considered constant, (external wall
0.287 W m-2K-1; french-window 1.048 W m-2K-1);
10) specific dissipating heat and power for the two
bed-headboard lamps were calculated taking into ac-
count their lighting parameters (compact fluorescent,
4000 K colour temperature, 58 W absorbed power, 1350
lm lighting flux);
11) safety lamps on the ceiling were considered bar-
rier to the air displacement.
NOMENCLATURE
Latin
B coefficient
C Cunningham factor
Co coughing function
Cp heat capacity at constant pressure (J kg-1 K-1)
D diameter (m)
F air resistance
G acceleration of gravity (m s-2)
M mass (kg)
Mm molecular mass
N mass transport with diffusion (kgs-1)
Q heat source (W m-3)
R thermal resistance (m2 K W-1)
Re Reynolds number
T temperature (℃)
U velocity (m)
Greek
Β coefficient
 mean free path (m)
 dynamic viscosity (Pa s)
 air viscosity (Pa s)
Ρ density (kg m-3)
 time
Footer
C correction
H2O water
In inlet
P pressure
Pa particle