A Macroscopic Model for Predicting Dust Concentration Distribution in Swine Buildings


Puma, M.C.; Maghirang, R.G.

Indoor and Built Environment 9(3-4): 182-191

2000


KARL
E
Fax
+
41
61
306
12
34
E-Mail
www.
karger.com
©
2001
S.
Karger
AG,
Basel
1420-326X/00/0094-018317.50/0
Accessible
online
at:
www.karger.com/journals/ibe
Indoor+Built
Environment
Original
Paper
Indoor
Built
Environ
2000;9:182-191
Accepted:
May
31,
2000
A
Macroscopic
Model
for
Predicting
Dust
Concentration
Distribution
in
Swine
Buildings
M.C.
Puma
a
R.G.
Maghirangb
a
Department
of
Agricultural
and
Biosystems
Engineering,
Iowa
State
University,
Ames,
Iowa,
bDepartment
of
Biological
and
Agricultural
Engineering,
Kansas
State
University,
Manhattan,
Kans.,
USA
Introduction
There
has
been
limited
research
to
investigate
dust
transport
within
livestock
buildings.
Such
research
is
needed
to
gain
a
better
understanding
of
the
transport
and
removal
of
particulate
contaminants
within
these
facili-
ties.
A
recent
trend
in
the
analysis
of
air
contaminant
transport
in
ventilated
spaces
is
toward
numerical
model-
ling,
which
involves
either
microscopic
or
macroscopic
models.
Microscopic
models
are
used
to
examine
the
details
of
air
and
contaminant
movement
within
an
air-
space.
Macroscopic
models,
on
the
other
hand,
are
em-
ployed
to
estimate
particle
concentration
for
a
microenvi-
ronment.
A
lumped-parameter
model
for
describing
the
dynam-
ics
of
airborne
dust
at
any
location
within
a
ventilated
space
was
developed
and
evaluated
by
Liao
and
Feddes
[1-3].
It
was
based
on
the
population-balance
model,
which
is
a
basic
way
to
describe
the
time-dependent
change
in
properties
of
airborne
dust
undergoing
turbu-
lent
coagulation,
turbulent
diffusive
deposition
and
gravi-
tational
settling
and
while
being
moved
by
ventilation
air-
flow
[1].
In
the
model
the
removal
of
dust
was
represented
by
three
dimensionless
parameters
that
characterized
the
relative
effects
of
turbulent
coagulation,
turbulent
diffu-
Dr.
M.C.
Puma
130
Davidson
Hall
Ames,
IA
50011
(USA)
Tel.
+1
515
292
7967,
Fax
+1
515
294
9973,
E-Mail
Key
Words
Air
quality
Dust
Modelling
Agricultural
buildings
Abstract
A
macroscopic
model
for
predicting
dust
concentration
distribution
in
mechanically
ventilated
swine
rooms
has
been
developed.
The
model
accounts
for
the
effect
of
tur-
bulent
diffusive
deposition,
gravitational
settling,
coagu-
lation
and
ventilation.
Four
particle
diameter
size
ranges
were
chosen
to
represent
the
fine
particles
in
swine
buildings:
0.5-0.9, 0.9-1.6,
1.6-2.8,
and
2.8-5.0
p.m.
Pre-
dicted
results
indicated
that
ventilation
would
be
the
dominant
particle
removal
mechanism
accounting
for
over
90%
of
the
particles
removed;
that
deposition
of
particles
on
surfaces
would
be
small
(2-9%),
and
loss
by
coagulation
negligible
(---.0%).
Additionally,
source
loca-
tion
would
strongly
influence
the
dust
concentration
dis-
tribution
in
the
prototype
swine
room.
Dust
generation
rate
and
presence
or
absence
of
obstructions
in
the
form
of
mock
pigs
would
affect
the
dust
distribution
minimal-
ly.
Temperature
difference
between
supply
and
room
air
(7-9
vs.
14-16°C)
would
not
cause
any
appreciable
dif-
ference
in
dust
distribution.
Copyright
©
2001
S.
Karger
AG,
Basel
sive
deposition
and
gravitational
deposition.
Three
model
cases,
i.e.,
complete
mixing,
displacement
system
and
short
circuiting,
were
studied.
Results
from
the
model
indicated
that
the
displacement
ventilation
system
was
more
effective
than
the
short
circuiting
system
in
remov-
ing
dust.
Predicted
dust
concentrations
compared
well
with
experimental
data.
Nazaroff
and
Cass
[4,
5]
also
developed
a
general
mathematical
model
for
predicting
the
concentrations
and
fate
of
particulate
matter
in
indoor
air
that
considered
the
effects
of
ventilation,
air
cleaning,
deposition,
direct
emission
and
coagulation,
based
on
the
population-balance
model.
The
model
determined
not
only
the
total
concentrations
of
aerosol
material
compo-
nents
but
also
concentrations
of
defined
particle
sizes.
The
authors
simulated
the
patterns
of
particle
size
distri-
bution
in
a
small
ventilated
chamber
and
found
good
agreement
between
predicted
and
experimental
results.
They
also
assessed
the
soiling
rates
of
paintings
in
mu-
seums
due
to
deposition
of
small
particles
and
used
the
model
to
assess
air
quality
in
office
rooms
and
residential
buildings.
Offermann
et
al.
[6]
used
a
macroscopic
model
to
describe
particle
removal
performance
of
air
cleaners
as
a
function
of
particle
diameter
and
the
mean
room
con-
centration.
The
present
study
details
the
implementation
of
a
sim-
plified
macroscopic
model
for
predicting
dust
concentra-
tions,
size
distribution
patterns
and
particle
transport
in
a
simulated
swine
room.
The
relative
contributions
of
grav-
itational
settling,
coagulation
and
ventilation
on
the
re-
moval
of
particulate
contaminants
from
the
swine
room
have
been
determined.
The
effect
of
ventilation
rate,
dust
generation
rate,
temperature
difference
between
supply
and
room
air,
source
location,
and
the
presence
or
ab-
sence
of
obstructions,
in
the
form
of
mock
pigs,
on
dust
concentration
distribution
were
also
evaluated.
Methods
Model
Development
A
macroscopic
model
has
been
developed
for
predicting
dust
con-
centrations
and
particle
size
distribution
in
mechanically
ventilated
swine
rooms
based
on
the
general
mathematical
model
formulated
by
Nazaroff
and
Cass
[4,
5]
and
the
lumped-parameter
modelling
approach
of
Liao
and
Feddes
[1-3].
The
model
will
account
for
the
effects
of
ventilation,
deposition
onto
surfaces,
coagulation
and
direct
emission
of
particles.
The
swine
room
was
represented
as
a
set
of
16
interconnected
chambers
or
control
volumes
each
having
a
well-mixed
core
(fig.
1).
Applying
the
model,
for
each
control
volume,
p
and
particle
size
range,
q,
the
rate
of
change
in
dust
mass
concentration
can
be
written
as:
dC
=
S„-
(LC)„
dt
where
C
pq
=
mass
concentration
of
dust
particles
within
size
range
q
in
control
volume
p
(µg•
m
-3
);
S
pq
=
production
or
emission
rate
of
dust
particles
within
size
range
q
in
control
volume
p
m
-3
s
-1
);
Lp
q
=
loss
or
removal
rate
of
particles
within
size
range
q
in
control
volume
p
(s
-1
);
q
=
1,2...,n
where
n
is
the
total
number
of
particle
size
ranges,
and
p
=
1,2...,m
where
m
is
the
total
number
of
control
vol-
umes.
The
parameter
S
pq
is
considered
to
include
direct
emission
inside
the
ventilated
space
(E),
advective
transport
from
the
ventilation
sys-
tem
or
from
outside
air
(A),
and
coagulation
of
particles
(K).
The
room
was
assumed
windowless,
airtight
and
with
well-insulated
wall
and
ceiling
surfaces
so
that
advection
from
outside
air
could
be
neglected.
The
parameter
L
pq
is
considered
to
include
removal
by
ventilation
(R),
particle
losses
to
surfaces
by
deposition
(D),
and
loss
to
a
larger
size
by
coagulation
(K').
Dust
removal
by
airflow
move-
ment
due
to
ventilation
was
represented
by:
dC
pq
_
(f‘pCxg
.fp.vC
pq)
dt
=
1
V
P
where
f„
=
volumetric
airflow
rate
from
adjacent
control
volumes
or
from
ventilation
system
to
control
volume
p
(m
3
•s-1);
CXq
=
mass
concentration
of
particles
in
size
range
q
in
the
adjacent
control
vol-
umes
or
in
the
ventilation
system
supply
air
(jig
m
-3
);
f
p
=
volumet-
ric
airflow
rate
from
control
volume
p
to
the
adjacent
control
vol-
umes
or
to
the
exhaust
outlet
(m
3
.
s
-
1
);
Cpq
=
mass
concentration
of
particles
in
size
range
q
in
control
volume
p
(.tg
m
-3
),
and
V
p
=
vol-
ume
of
control
volume
p
(m
3
).
The
volumetric
airflow
rates,
f
p
,
and
f„,
were
determined
from
preliminary
experiments
in
which
air
velocities
between
the
control
volumes,
from
the
ventilation
air
supply
diffuser
to
the
adjoining
control
volume,
and
from
an
adjoining
control
volume
to
the
outlet
were
measured.
Then,
taking
small
cross-sectional
areas
between
boundaries
of
the
coplanar
boundaries
between
control
volumes,
cross-flow
ventilation
rates
between
control
volumes
were
estimated
by
multiplying
the
measured
air
velocities
with
the
cross-sectional
areas
of
the
coplanar
boundaries.
Details
of
the
experimental
proce-
dures
and
results
have
been
thoroughly
discussed
elsewhere
[7-9].
The
rate
of
change
of
dust
concentration
due
to
deposition
was
estimated
from
the
equation:
dC„
(
Cpq\
(vdnism)
dt
V
p
I
where
Vd
mq
=
mean
deposition
velocity
of
particles
within
size
range
q
during
deposition
to
the
mth
surface
of
control
volume
p
(m
s
-1
)
and
A
m
=
superficial
area
of
the
mth
surface
of
control
volume
p
(
m
2)
.
The
mean
deposition
velocity,
V
dmq
,
was
estimated
for
three
pos-
sible
airflow
regimes
in
the
chamber:
(1)
natural
convection
driven
by
temperature
differences
between
the
surfaces
(floor,
walls,
ceiling)
and
the
nearby
air;
(2)
homogeneous
turbulence
in
the
core
of
the
room,
and
(3)
forced
laminar
flow
parallel
to
the
surface.
Equations
relating
deposition
velocity
to
particle
size,
surface
characteristics
(orientation,
temperature,
velocity
of
nearby
air),
and
flow
condi-
(1)
(3)
Modelling
Dust
Distribution
in
Swine
Indoor
Built
Environ
2000;9:182-191
183
Buildings
Ventilation
C"
C
‘q
system
Inlet
A
f
ps
C
pq
(Control
volume
p)
K
--II.
4—K'
A
D
Outlet
R
Fig.
1.
Schematic
diagram
of
the
compo-
nents
of
the
dust
concentration
model.
Cpq
=
Concentration
of
dust
particles
within
size
range
q
in
control
volume
p;
Cxq
=
concentra-
tion
of
dust
particles
within
size
range
q
in
adjacent
control
volumes;
E
=
direct
emis-
sion
from
sources
in
ventilated
airspace;
A
=
advective
transport
from
the
ventilation
sys-
tem
or
from
outside
air;
K
=
coagulation
from
smaller
particles;
K'
=
loss
to
a
larger
size
due
to
coagulation;
D
=
deposition
losses
to
surfaces
(floor,
walls,
ceiling);
R
=
removal
by
ventilation;
f
x
,
=
volumetric
air-
flow
rate
from
adjacent
control
volumes
or
from
ventilation
system
to
control
volume
p;
f
ps
=
volumetric
airflow
rate
from
control
volume
p
to
adjacent
control
volumes
or
to
exhaust
outlet;
q
=
1,2,...r
(r
is
total
number
of
sections/particle
size
range);
p
=
1,2,...s
(s
is
the
total
number
of
control
volumes).
tions
were
obtained
from
published
values
[4,
5,
10,
11].
For
exam-
ple,
the
deposition
velocity
for
a
particle
with
diameter
d
p
being
deposited
on
a
horizontal
upward
facing
surface
(e.g.,
floor)
due
to
homogeneous
turbulence
can
be
estimated
by:
V
d
=
V
ep
+
(4)
where
(1<
e
)
1 / 2
)
N,v
a
tan
-1
[8
(L)
1 /2
]
a
and
(6)
[1
-
exp
n
vg
2
(Dic,)
,
)
where
V
et
,
=
deposition
velocity
due
to
combined
effects
of
eddy
dif-
fusion,
Brownian
motion,
advection
and
gravitational
settling
(m
s
-1
);
V
t
=
deposition
velocity
due
to
effect
of
thermophoresis
(m
s
-1
);
Ni
t
=
thermophoresis
parameter
(dimensionless),
=
K(AT/
T
a
);
AT
=
temperature
of
surface,
T
s
,
minus
temperature
of
air
out-
side
boundary
layer,
T
it
,
(K);
K
=
thermophoresis
coefficient
(dimen-
sionless);
v=
kinematic
viscosity
of
air,
(m
2
.
s
--1
);
K
e
=
tur-
bulence
intensity
parameter
(s
-1
);
a
=
thermal
diffusivity
of
air
(m
2
.
s
-1
);
S
=
boundary
layer
thickness
(m),
=
(1.2)(v/KJ
4
/
9
X
s
";
X
s
=
length
of
the
surface
in
the
direction
of
flow
(m);
V,
=
gravitational
settling
velocity
(m
s
-1
),
and
D
=
coefficient
of
Brownian
diffusivity
of
particles
(m
2
.
s
-1
).
The
thermophoresis
coefficient,
K,
depends
on
particle
size
and,
for
particles
larger
than
the
mean
free
path
of
air
molecules,
on
the
ratio
of
thermal
conductivity
of
the
gas
to
that
of
the
particle
(K,/K
p
).
A
formula
for
estimating
K
has
been
given
by
Nazaroff
and
Cass
[4].
For
particles
in
air
with
d
p
3.0
gm
and
K
g
/K
p
in
the
range
of
0.01-
0.50,
K
varies
between
0.10
and
0.60.
A
representative
value
of
K
is
r=1
s=1
r
=
I
13rgiCprC
pq.1
I
2
i
l
kviikx
C
pq[
[
4
13rg
]
C
i
n]
(7)
1
,
1
r
,=q+
I
where
73,.,„=
mean
rate
of
collision
between
dust
particles
in
sections
r
and
s
yielding
particles
in
the
size
range/section
q,
and
sections
r
and
s
were
sections/size
ranges
smaller
than
q
(m
3
•14
-1
.5
-1
);
I3
r
q
=
mean
rate
of
collision
between
dust
particles
in
sections
r
and
q
yielding
particles
in
section
q
or
a
section
larger
than
q
(m
3
.
gg
-1
.
s
-1
);
)
3,
=
mean
rate
of
collision
between
dust
particles
in
section
q
yielding
particles
larger
than
q
(m
3
1.tg
-1
•s
-1
);
C
ps
,
C
pr
,
C
pq
=
mass
concentra-
tions
of
dust
particles
contained
within
control
volume
p
in
size
range/section
s,
r,
q,
respectively
(gg•
m
-3
),
and
u
=
total
number
of
particle
size
ranges/sections.
The
superscripts
1,
2a,
2b,
3,
and
4
in
equation
7
represent
the
possible
types
of
collision
between
particles.
For
a
given
size
range
q,
four
classes
of
collisions
are
possible,
with
the
type
of
collision
distin-
guished
by
the
sizes
of
the
colliding
particles:
(1)
two
particles,
each
from
a
size
range
smaller
than
q
(e.g.,
size
ranges
r
and
s),
collide
to
produce
a
particle
in
size
range
q
(valid
for
1
<q
<
n);
(2)
one
particle
from
a
size
range
smaller
than
q
collides
with
a
particle
in
size
range
q
to
yield
either
a
particle
in
size
range
larger
than
q
or
a
particle
in
size
range
q
(valid
for
1
<q
n);
(3)
two
particles
in
size
range
q
collide
to
yield
a
particle
in
a
size
range
larger
than
q
(valid
for
1
q
<
n),
and
(4)
a
particle
in
q
collides
with
a
particle
in
a
size
range
larger
than
q
(valid
for
1
q
<
n).
Equations
2,
3,
and
7
were
combined
with
the
direct
emission
term
and
cast
into
the
form
of
equation
1.
The
equations
were
solved
numerically
using
the
asymptotic
integration
method
[12].
For
each
(5)
v
ep
=
V
g
0.50,
particularly
for
particles
smaller
than
1
gm
[4].
The
turbulence
intensity
parameter,
lc,
has
the
dimension
of
inverse
time
and
is
obtained
from
a
fit
to
experimental
data
on
the
measured
deposition
flux
to
the
surface
of
the
enclosure.
Effect
of
coagulation
was
treated
in
the
model
using
the
following
equation
[5]:
dC
1
2
[
q
-
(
q
-
I
q
=
-
[
[
1
[3,„]
C
p
,C,
-
-
I
[
[
2
a/3„]
C
m
C,
r
)
-
dt
184
Indoor
Built
Environ
2000;9:182-191
Puma/Maghirang
inlet
2
3
4
outlet
5
-cut
7
3.46
co
13
14
15
16
1.22
1.73
7.11
z
2.44
/
source
location
1
source
location
2
at>
dust
generation
point
(control
volume
P)
/
41—
1.78
—1110
Fig.
2.
Schematic
diagram
of
the
prototype
swine
room
used
in
the
numerical
modeling
(all
units
in
m,
not
drawn
to
scale).
section,
the
initial
indoor
dust
mass
concentration
and
the
hourly
averaged
values
of
outdoor
dust
mass
concentration
were
specified.
Direct
indoor
emissions
at
constant
rates
were
also
specified
as
hour-
ly-averaged
values.
The
program,
written
in
FORTRAN-77,
was
contained
in
10
files
comprising
70
subroutines
and
functions.
It
was
executed
by
first
preparing
an
input
file
that
contained
a
list
of
commands
including
input
data
using
a
text
editor
and
then
compiling
and
linking
the
10
program
files.
Most
of
the
simulations
required
1-2
min
of
CPU
time
on
a
SUNSPARC
station
10
or
20
with
a
SUN
operating
system
ver-
sion
5.5.1.
Assumptions
and
Limitations
of
the
Model
Model
simulation
assumed
the
following:
(1)
the
swine
room
can
be
represented
by
16
control
volumes,
each
with
a
well-mixed
core;
(2)
particles
are
spherical
and
have
equal
densities,
and
(3)
the
parti-
cle
mass
concentration
is
uniformly
distributed
with
respect
to
the
logarithm
of
the
mass
or
diameter
of
the
particle.
Implementation
of
the
Model
The
model
was
implemented
on
a
prototype
room
(7.11
m
wide,
3.46
m
long
and
2.44
m
high)
representing
typical
mechanically
ven-
tilated
swine
nursery
rooms
(fig.
2).
The
room
had
a
rectangular
air
supply
diffuser
on
one
sidewall
and
an
exhaust
opening
on
the
oppo-
site
wall.
The
room
was
divided
into
16
zones
or
control
volumes
with
each
control
volume
measuring
1.78
x
1.73
x
1.22
m.
Four
par-
ticle
size
ranges
were
chosen
to
represent
the
fine
particles:
0.5-0.9,
0.9-1.6,
1.6-2.8,
and
2.8-5.0
gm.
Sixteen
test
cases
involving
combinations
of
two
levels
of
ventila-
tion
rates,
three
airflow
thermal
conditions,
two
dust
generation
rates,
two
source
locations,
and
the
presence
or
absence
of
obstruc-
tions
were
studied
(table
1).
The
airflow
thermal
conditions
were
iso-
thermal,
nonisothermal
(no
heat
load),
and
nonisothermal
(with
heat
load).
A
heat
load
of
1,600
W
representing
the
metabolic
heat
genera-
tion
of
the
pigs
was
provided
for
the
non-isothermal
(with
heat
load)
test
cases.
Isothermal
test
cases
(test
cases
1-4,
11
and
12)
had
a
ventilation
rate
of
0.336
m
3
-
s
-
i;
temperatures
of
supply
and
room
air,
which
Modelling
Dust
Distribution
in
Swine
Indoor
Built
Environ
2000;9:182
-
191
185
Buildings
Table
1.
Description
of
the
16
test
cases
Test
case
No.
Ventila-
tion
rate
m
3
.S
-1
1
0.336
2
0.336
3
0.336
4
0.336
5
0.096
6
0.096
7
0.096
8
0.096
9
0.096
10
0.096
11
0.336
12
0.336
13
0.096
14
0.096
15
0.096
16
0.096
Airflow
thermal
Dust
Obstruction
Source
condition'
generation
location
3
rate
2
isothermal
low
without
1
isothermal
low
with
1
isothermal
high
without
1
isothermal
high
with
1
nonisothermal,
no
heat
load
low
without
1
nonisothermal,
no
heat
load
low
with
1
nonisothermal,
no
heat
load
high
without
1
nonisothermal,
no
heat
load
high
with
1
nonisothermal,
with
heat
load
low
with
1
nonisothermal,
with
heat
load
high
with
1
isothermal
high
without
2
isothermal
high
with
2
nonisothermal,
no
heat
load
high
without
2
nonisothermal,
no
heat
load
high
with
2
nonisothermal,
with
heat
load
low
with
2
nonisothermal,
with
heat
load
high
with
2
I
For
the
isothermal
condition,
supply
(T
o
)
and
room
air
(T,)
temperatures
varied
from
24
to
27°C.
For
the
nonisothermal,
no
heat
load
condition,
temperature
difference
between
supply
and
room
air
was
7-9°C.
For
the
nonisothermal,
with
heat
load
condition,
tempera-
ture
difference
was
14-16
°C.
2
Low
dust
generation
rate
=
201-248
tig•
min
-1
;
high
dust
generation
rate
=
270-335
jig.
min
-
3
1
=
Source
location
near
the
inlet;
2
=
source
location
near
the
exhaust.
were
almost
the
same
for
each
case,
varied
from
24
to
27
°C.
Noniso-
thermal
(no
heat
load)
test
cases
(test
cases
5-8,
13
and
14)
had
a
ventilation
rate
of
0.096
m
3
.
s
-
';
air
temperature
differences
between
supply
and
room
air
ranged
from
7
to
9
°C.
Nonisothermal
(with
heat
load)
test
cases
(test
cases
9,
10,
15
and
16)
had
a
ventilation
rate
of
0.096
m
3
.s
-
I;
air
temperature
differences
varied
from
14
to
16
°C.
The
two
ventilation
rates:
0.336
m
3
.
s
-
'
for
isothermal
test
cases
and
0.096
m
3
.
s
-
I
for
the
non-isothermal
test
cases,
were
close
to
the
rec-
ommended
ventilation
rates
of
0.341
and
0.067
m
3
•5
-
'
for
nursery
pigs
weighing
13.6-34.1
kg
during
mild
and
cold
weather,
respective-
ly
[13].
Dust
generation
rates
were
high
(270-335
tig
min
-
')
or
low
(201-
248
µg•
min
-
').
Two
source
locations
were
considered:
source
loca-
tion
1
was
near
the
supply
diffuser;
source
location
2
was
near
the
exhaust
(fig.
2).
Obstructions
were
represented
by
'mock
pigs',
in
the
form
of
16
galvanized
steel
tubes
(2.80
m
long
and
20.3
cm
in
diame-
ter),
which
were
uniformly
distributed
in
the
test
room.
These
mock
pigs
were
present
for
test
cases
that
had
obstructions
and
removed
for
test
cases
without
obstructions.
Data
Analysis
The
predicted
dust
mass
concentrations
were
normalized
using
the
procedure
suggested
by
Kato
and
Murakami
[14]
to
account
for
differences
in
dust
generation
rates:
C7,(r)
-
[
G,
(t)1
Coq
(t)
Q
where
C,
*
(t)
=
normalized
concentration
of
dust
particles
within
size
range
q
in
control
volume
p
at
time
t
(dimensionless);
C
pq
(t)
=
predicted
concentration
of
dust
particles
within
size
range
q
in
con-
trol
volume
pat
time
t
(1.ig
m
-3
);
C
ki
(t)=
concentration
of
dust
parti-
cles
within
size
range
q
at
the
inlet
at
time
t
(µg•
m
-3
);
G
q
(t)
=
genera-
tion
rate
of
particles
within
size
range
q
at
time
t
(µg•s
-
');
C„
(t)
=
concentration
of
dust
particles
within
size
range
q
at
the
exhaust
at
time
t
(pg.
m
-3
),
and
Q
=
ventilation
rate
(m
3
.
s
-1
).
Model
Validation
To
verify
the
predicted
results,
controlled
laboratory
tests
using
a
full-scale
room
air
distribution
chamber
were
conducted.
Details
of
the
experimental
procedure
and
results
have
been
presented
else-
where
[7-9].
A
summary
of
the
procedures
is
presented
below.
Experiments
were
conducted
in
a
full-scale
test
chamber
similar
to
the
prototype
room
(fig.
2)
for
validating
the
predicted
results.
The
chamber
had
a
cross-flow
jet
ventilation
system
with
air
entering
a
rectangular
air
supply
diffuser
on
one
sidewall
and
exhausted
by
a
variable
speed
fan
mounted
on
the
opposite
sidewall.
A
furnace
filter
was
installed
at
the
air
supply
diffuser
to
filter
out
particles
coming
in
C
pq
(r
)
-
C,,
(t)
(8)
186
Indoor
Built
Environ
2000;9:182
-
191
Puma/Maghirang
from
the
ventilation
system.
For
the
nonisothermal
test
cases,
out-
side
air
was
first
conditioned
through
a
7.44-kW
•h
-
l
air
condi-
tioner.
Cornstarch
powder
served
as
the
test
dust
material
because
corn
constitutes
the
bulk
of
most
swine
rations.
The
material
has
a
wider
range
of
particle
sizes
than
swine
house
dust
but
exhibits
some
of
the
characteristics
(chemical
makeup
and
density)
of
swine
house
dust.
Particle
density
of
the
test
dust
was
determined
to
be
1.6
g
cm
-3
.
To
disperse
the
material
a
dust
generator
originally
developed
at
the
Bureau
of
Standards
was
used.
The
dust
generator
was
placed
outside
the
chamber
because
it
might
have
modified
the
airflow
pattern
if
it
had
been
sited
within
the
chamber.
The
dust
particles
were
emitted
into
the
airstream
through
two
plastic
tubes,
one
on
each
side
of
the
chamber,
with
the
end
of
each
tube
set
at
0.2
m
above
the
floor
(fig.
2).
A
microprocessor-controlled
optical
particle
counter
monitored
the
dust
concentrations.
It
was
placed
outside
the
chamber
and
con-
nected
to
a
manifold
system
to
enable
automatic
measurements
from
16
locations.
Attached
to
the
manifold
was
a
multiple
sampling
port
placed
at
the
center
of
the
room
that
enabled
continuous
measure-
ments
from
the
16
locations
without
interference
with
the
airflow
inside
the
room.
Dust
concentration
was
measured
at
the
center
of
each
control
volume.
Sampling
time
was
15
s
for
each
control
volume
and
sampling
between
any
two
control
volumes
was
15
s.
It
took
8
min
to
measure
the
concentrations
at
all
16
control
volumes
during
each
reading.
Readings
were
at
15-min
interval
starting
from
the
beginning
of
dust
generation
for
1.5
or
2
h
during
which
time
steady-
state
concentrations
in
the
control
volumes
were
attained.
The
inside
surfaces
of
the
chamber
were
cleaned
thoroughly
before
each
test.
The
surfaces
were
wiped
with
wet
mops
and
the
ventilation
system
was
run
for
several
hours
to
remove
airborne
dust.
A
test
was
started
only
when
the
dust
concentrations
in
the
control
volumes
were
about
the
same
as
the
outdoor
concentrations.
Initial
dust
concentrations
inside
the
chamber
were
measured
before
the
start
of
each
dust
generation.
After
dust
generation
was
started,
dust
concentrations
were
read
at
15-min
interval
until
steady-state
con-
centrations
were
attained
(i.e.,
when
concentrations
were
at
levels
±
5%
of
previous
reading).
Measured
dust
concentrations
were
nor-
malized
using
the
same
procedure
used
for
normalizing
predicted
concentrations.
Results
Results
from
representative
test
cases
are
presented
below
to
show
the
performance
of
the
model.
For
all
test
cases,
predicted
concentrations
were
symmetric
about
the
x-z
plane
or
the
longitudinal
(x)
axis
of
the
chambers
so
that
only
half
of
the
chamber
was
considered.
Additional-
ly,
predicted
values
for
the
0.5-
to
0.9-,
0.9-
to
1.6-,
1.6-
to
2.8-1.tm
particle
size
ranges
and
total
dust
(0.5-5.0
iim)
agreed
well
with
measured
values
[8,
9].
Measured
values
for
the
2.8-
to
5.01.1m
particle
sizes
were
higher
than
pre-
dicted
values;
it
was
suspected
that
this
was
due
to
an
error
in
measurement
of
the
concentrations
of
the
2.8-
to
5.0-1.1m
particle
range.
Table
2.
Dust
particle
removal
(percentage
of
total)'
Test
case
No.
Ventilation
Deposition
1
95
5
2
98
2
3
94
6
4
98
2
5
91
9
6
91
9
7
92
8
8
92
8
9
92
8
10
94
6
II
97
3
12
96
4
13
94
6
14
93
7
15
92
8
16
9
3
7
Loss
due
to
coagulation
was
negligible
(--.0%)
for
all
test
cases.
Dust
Removal
Mechanisms
Predicted
values
indicated
that,
for
all
test
cases,
dust
removal
by
ventilation
was
the
dominant
mechanism
(19-98%);
removal
by
deposition
of
particles
on
surfaces
was
2-9%;
removal
by
coagulation
(----,0%)
was
negligible
(table
2).
Effects
of
Ventilation
Rate,
Temperature
Difference,
Dust
Source
Location
and
Obstructions
Isothermal
Case
Normalized
predicted
concentrations
for
the
eight
con-
trol
volumes
(control
volumes
1-8)
for
test
case
1
are
shown
in
figure
3.
Test
case
1
involved
an
isothermal
test
condition,
low
dust
generation
rate,
dust
generation
at
source
location
1
(near
control
volume
5),
and
without
the
mock
pigs
in
the
chamber.
Concentrations
immediately
increased
from
the
initial
values
to
levels
approximating
the
steady-state
values
after
the
first
15
min
of
the
simula-
tion
(fig.
3).
During
this
period,
normalized
values
in-
creased
from
0
to
0.80
for
control
volumes
1-4
and
6-8,
and
from
0
to
1.10
for
control
volume
5.
For
all
control
volumes,
differences
in
the
predicted
concentrations
for
the
four
particle
size
ranges
(0.5-0.9,
0.9-1.6,
1.6-2.8,
Modelling
Dust
Distribution
in
Swine
Indoor
Built
Environ
2000;9:182-191
187
Buildings
2.5
Control
volume
4
III
0.5-0.9
0.9-1.6
1.6-2.8
A
2.8-5.0
2.0_
llll
l
111
2.5
0.5
0.0
Control
volume
3
0.5-0.9
0.9-1.6
1.6-2.8
A
2.8-5.0
IyJ
•••444
I
S
1.11,•••••41
S
1.5
0
1.1
1.0
o
0.5
0.0
2.5
1.1
0.5
0.9
0.9-1.6
1.6-2.8
A
2.8-5.0
1.5
-
0
1.0
E
0.5
-
0.0
a4444
114
444-4
2.5
id
2.0
O
O
1.0
0.5
0.9
0.9-1.6
1.6-2.8
A
2.8-5.0
•••....!
•441.:11
JJs
JP.N.
•••••••••
t
2
0.5'
0.0
Control
volume
I
Control
volume
2
25
0.5-0.9
0.9-1.6
1.6-2.8
A
2.8-5.0
0.5-0.9
0
0.9-1.6
1.6-2.8
A
2.8-5.0
2.0
_
2.5
.2
2.0
1.5
0
1.0
-
0
5
-
••.
,s'
"0::
%---4
.
1.1
14
444.18
4
2
0.5
-
•••••91111
00000
4
0.0
0.0
0
0
15
30
45
Timc
(min)
60
75
90
0
15
30
45
60
75
90
Time
(min)
2.5
2.0
-
1.5
-
1.0
0.5
0.0
0
15
30
45
Time
(min)
60
75
90
0
15
30
45
Time
(min)
60
75
90
Control
volume
5
Control
volume
6
0.5-0.9
0.9-1.6
1.6-2.8
2.8-5.0
0
15
30
45
Time
(min)
Control
volume
7
60
75
90
0
15
30
45
Time
(min)
Control
volume
8
60
75
90
••
.5
Norma
lize
d
concen
tra
t
io
n
2.5
.2
2.0
MI
0.5
0.9
0.9
-1.6
1.6-2.8
A
2.8
5.0
00
0
15
30
45
Time
(min)
60
75
90
0
15
30
45
Time
(min)
60
75
90
Fig.
3.
Variations
of
normalized
predicted
dust
concentrations
(by
particle
size
range)
with
time
(test
case
1).
188
Indoor
Built
Environ
2000;9:182-191
Puma/Maghirang
Norma
lize
d
dus
t
con
ce
n
tra
t
io
n
1.8
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
Norma
lize
d
dus
t
conce
n
tra
t
io
n
1.8
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
in
all
2
3
4
5
6
7
8
Control
Volume
Fig.
4.
Predicted
total
dust
concentrations
for
test
case
1
(isothermal,
low
dust
generation
rate,
without
mock
pigs,
source
location
1).
1
2
3
4
5
6
7
8
Control
Volume
Fig.
5.
Predicted
total
dust
concentrations
for
test
case
11
(iso-
thermal,
high
dust
generation
rate,
without
mock
pigs,
source
loca-
tion
2).
and
2.8-5.0
um)
were
minimal.
Normalized
values
then
increased
gradually
until
the
steady-state
levels
were
reached
in
about
90
min.
Steady-state
normalized
concen-
trations
ranged
from
0.82
to
0.83
for
control
volumes
1-4
and
6-8,
and
1.11
for
control
volume
5
(fig.
4).
Predicted
concentrations
were
higher
for
control
volume
5
than
for
all
the
other
control
volumes.
Similar
trends
were
observed
on
the
predicted
concen-
trations
of
test
cases
2-4.
All
these
test
cases
had
dust
gen-
eration
also
at
source
location
1.
Test
case
2
differed
from
test
case
1
only
in
having
mock
pigs
inside
the
chamber.
Test
cases
3
and
4
were
similar
to
test
cases
1
and
2,
respectively,
but
involved
higher
dust
generation
rates.
The
similarity
in
the
trends
of
predicted
dust
concentra-
tions
between
test
case
1
and
those
of
test
cases
2-4
indi-
cated
that
dust
generation
rate
and
presence
of
the
mock
pigs
did
not
have
much
influence
on
the
dust
concentra-
tion
distribution
in
the
room.
The
mock
pigs
occupied
only
a
small
portion
(--="2.3%)
of
the
control
volumes
so
that
the
airflow
and
dust
transport
between
the
control
volumes
were
not
affected
by
their
presence.
Regarding
the
effect
of
the
dust
generation
rate,
the
actual
values
of
dust
concentrations
were
higher
for
test
cases
with
higher
dust
generation
rates;
however,
normalized
values
were
almost
the
same.
Predicted
values
for
test
cases
involving
source
loca-
tion
2
(test
cases
11
and
12)
also
showed
small
differences
in
the
predicted
values
for
the
four
particle
size
ranges.
However,
the
trend
for
dust
concentrations
among
con-
trol
volumes
was
different
from
test
cases
involving
source
location
1.
The
steady-state
normalized
predicted
total
dust
concentrations
for
test
case
11
(isothermal,
low
dust
generation
rate,
without
mock
pigs,
source
location
2)
is
shown
in
figure
5.
Normalized
concentrations
for
the
upper
control
volumes
(control
volumes
1-4)
were
lower
than
values
for
the
lower
control
volumes
(control
vol-
umes
5-7).
Predicted
concentration
was
higher
for
control
volume
8
(the
dust
source
location)
than
for
all
the
other
control
volumes.
Differences
in
the
predicted
concentra-
tions
between
test
cases
involving
source
location
1
with
those
involving
source
location
2,
as
shown
in
figures
4
and
5,
indicated
that
source
location
exerted
a
strong
influence
on
dust
particle
concentration
distribution.
Nonisothermal
(No
Heat
Load)
Test
Cases
Model
predictions
under
this
condition
are
presented
by
the
results
for
test
case
5
[nonisothermal
(no
heat
load)
condition,
low
dust
generation,
without
the
mock
pigs,
and
dust
source
location
1].
For
this
test
case,
there
were
small
variations
in
the
predicted
values
for
the
eight
con-
trol
volumes.
Likewise,
there
were
small
differences
in
the
predicted
concentrations
for
the
four
particle
size
ranges.
The
steady-state
normalized
predicted
total
dust
concen-
trations
for
control
volumes
2-4
and
6-8
ranged
from
0.98
to
0.99
(fig.
6).
For
control
volume
5,
the
source
loca-
tion
control
volume,
steady-state
normalized
concentra-
tion
was
1.00.
For
control
volume
1,
which
was
above
control
volume
5
but
directly
in
front
of
the
air
supply
Modelling
Dust
Distribution
in
Swine
Indoor
Built
Environ
2000;9:182
-
191
189
Buildings
1
2
3
4
5
Control
Volume
7
8
Norma
lize
d
dus
t
c
oncen
tra
t
ion
1.8
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
11
2
3
4
5
Control
Volume
III
8
Norma
lize
d
dus
t
concen
tra
t
ion
1.8
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
i
ll
I
I I
I
Fig.
6.
Predicted
total
dust
concentrations
for
test
case
5
(nonisother-
Fig.
7.
Predicted
total
dust
concentrations
for
test
case
9
(nonisother-
mal,
low
dust
generation
rate,
without
mock
pigs,
source
location
1).
mal,
low
dust
generation
rate,
with
mock
pigs,
source
location
1).
diffuser,
steady-state
value
was
0.71.
Almost
the
same
variations
were
obtained
for
the
other
test
cases
involving
the
same
thermal
condition
and
source
location,
but
eith-
er
with
higher
dust
generation
and
with
the
presence
of
mock
pigs
in
the
room
(test
cases
6-8).
Again,
this
indi-
cated
a
minimal
effect
of
dust
generation
rate
and
the
presence
or
absence
of
mock
pigs
on
the
dust
concentra-
tion
distribution.
For
test
cases
that
had
source
location
2
(test
cases
13
and
14),
normalized
concentrations
showed
an
increasing
trend
for
the
upper
control
volumes
(from
control
volume
1-4).
For
the
lower
control
volumes,
normalized
concen-
trations
increased
from
control
volume
5-8.
This
differ-
ence
in
predicted
dust
concentration
distribution
with
those
of
test
cases
5-8
was
due
to
the
effects
of
difference
in
source
location.
Non
isothermal
(with
Heat
Load)
Test
Cases
For
these
test
cases,
there
were
also
small
variations
in
the
predicted
concentrations
among
the
eight
control
vol-
umes
and
between
the
four
particle
size
ranges.
The
steady-state
normalized
total
dust
concentrations
for
test
case
9
[nonisothermal
(with
heat
load),
low
dust
genera-
tion
rate,
with
mock
pigs,
and
dust
source
location
1]
are
shown
in
figure
7.
Values
for
control
volumes
2-4
and
6-
8
were
almost
the
same
(0.98).
For
control
volume
5
(source
location),
normalized
value
was
just
slightly
high-
er
(1.00),
while
for
control
volume
1,
it
was
0.68.
The
trends
of
the
predicted
values
for
test
case
10
followed
those
of
test
case
9.
This
trend
was
similar
to
that
of
the
190
Indoor
Built
Environ
2000;9:182-191
nonisothermal
(no
heat
load)
test
cases
(fig.
5),
indicating
that
temperature
difference
between
supply
and
room
air
was
not
enough
to
cause
any
appreciable
difference
in
room
dust
distribution.
Results
of
test
cases
15
and
16,
which
involved
source
location
2,
indicated
increasing
concentrations
for
both
the
upper
and
lower
control
volumes.
This
trend
was
also
similar
to
that
of
the
corresponding
nonisothermal
(no
heat
load)
test
cases.
Summary
and
Conclusions
A
general
macroscopic
model
that
accounted
for
con-
vective
diffusion,
coagulation,
and
gravitational
sedimen-
tation
was
implemented
to
determine
the
dust
concentra-
tion
distribution
in
a
typical
mechanically
ventilated
swine
nursery
room
as
affected
by
ventilation
rate,
tem-
perature
difference
between
supply
and
room
air,
dust
generation
rate,
and
the
presence
or
absence
of
mock
pigs.
The
following
conclusions
were
drawn:
(1)
Ventilation
was
the
dominant
dust
particle
removal
mechanism,
accounting
for
91-98%
of
the
total
dust
removed.
Deposition
of
particles
on
wall,
ceiling
and
floor
surfaces
(2-9
%)
was
small,
and
the
effect
of
coagulation
(.-
-0%)
was
negligible.
(2)
Source
location
had
a
strong
influence
on
dust
con-
centration
distribution.
Dust
concentrations
tended
to
be
higher
at
the
source
location
control
volumes
than
in
all
other
control
volumes.
Puma/Maghirang
(3)
Obstructions
(i.e.,
mock
pigs)
and
dust
generation
rates
had
little
influence
on
dust
distribution.
(4)
The
temperature
difference
between
supply
and
room
air
(7-9
vs.
14-16°C)
was
not
enough
to
cause
any
appreciable
difference
in
dust
distribution.
Acknowledgments
The
contribution
of
Dr.
William
W.
Nazaroff,
Mr.
Steve
Coulson,
and
Dr.
Murali
Narayanan
is
acknowledged.
Support
from
the
Kan-
sas
Agricultural
Experiment
Station
(AES
Contribution
No.
00-236-
J)
is
also
acknowledged.
References
1
Liao
CM,
Feddes
JJR:
Mathematical
analysis
of
a
lumped-parameter
model
for
describing
the
behaviour
of
airborne
dust
in
animal
hous-
ing.
Appl
Math
Modelling,
1990;14:248-257.
2
Liao
CM,
Feddes
JJR:
Modelling
and
analysis
of
airborne
dust
removal
from
a
ventilated
air-
space.
Can
Agric
Eng
1991;33:355-361.
3
Liao
CM,
Feddes
JJR:
A
lumped
parameter
model
for
predicting
airborne
dust
concentra-
tions
in
a
ventilated
airspace.
Trans
Am
Soc
Agric
Engineers
1992;35:1973-1978.
4
Nazaroff
WW,
Cass
GR:
Mathematical
model-
ling
of
chemically
reactive
pollutants
in
indoor
air.
Environ
Sci
Technol
1986;20:924-934.
5
Nazaroff
WW,
Cass
GR:
Mathematical
model-
ling
of
indoor
aerosol
dynamics.
Environ
Sci
and
Technol
1989;23:157-166.
6
Offermann
FJ,
Sextro
RG,
Fisk
WJ,
Grimsrud,
DT,
Nazaroff
WW,
Nero
A,
Rexzan
KL,
Yater
J:
Control
of
respirable
particles
in
indoor
air
with
portable
air
cleaners.
Atmos
Environ
1985;19:1761-1771.
7
Puma
MC:
Modelling
Dust
Concentration
Dis-
tribution
in
Swine
Houses;
PhD
diss
Kansas
State
University,
Manhattan,
Kansas.
1998.
8
Puma
MC.
Maghirang
RG,
Hosni
MH,
Hagen
L:
Modelling
dust
concentration
distribution
in
a
swine
house
under
isothermal
conditions.
Trans
Am
Soc
Agric
Engineers
1999;42:1811-
1821.
9
Puma
MC,
Maghirang
RG,
Hosni
MH,
Hagen
L:
Modelling
dust
concentration
distribution
in
a
swine
house
under
non-isothermal
condi-
tions.
Trans
Am
Soc
Agric
Engineers
1999;42:
1823-1832.
10
Incropera
FP,
DeWitt
DP:
Fundamentals
of
Heat
and
Mass
Transfer.
J.
New
York,
Wiley,
1985.
11
Corner
J,
Pendlebury
ED:
The
coagulation
ad
deposition
of
stirred
aerosol.
Proc
Phys
Soc
1951;B64:645-654.
12
Young
TR,
Boris
JP:
A
numerical
technique
for
solving
stiff
ordinary
differential
equations
associated
with
the
chemical
kinetics
of
reac-
tive-flow
problems.
J
Phys
Chem
1977;81:
2424-2427.
13
Midwest
Plan
Service:
Mechanical
ventilating
systems
for
livestock
housing.
MWPS-32.
Ames,
Midwest
Plan
Service,
1990.
14
Kato
S,
Murakami
S:
New
ventilation
efficien-
cy
scales
based
on
spatial
distribution
of
con-
taminant
concentration
aided
by
numerical
si-
mulation.
Trans
Am
Soc
Heating
Refrigeration
Air
Conditioning
Engineers
1995;94:309-330.
Modelling
Dust
Distribution
in
Swine
Indoor
Built
Environ
2000;9:182-191
191
Buildings