Академический Документы
Профессиональный Документы
Культура Документы
is the predominant product,
we assumed it is the only product in the particle formation
model. Both particle density and the surface tension are
unknown for carrying out the gas to particle conversion
calculation. Particle density was estimated from data of
(NH
4
)
2
S0
4
(pP = 1.77 gjcm
3
) and NH
4
HS0
3
(pp = 2.03
gfcm
3
) and was varied in the range of 1.8-2.0 gfcm
3
in our
model. The value for surface tension is not available and
was adjusted to fit the experimental results (symbols in
Figure 4). Chemical reaction and particle growth may be
occurring simultaneously. As indicated by (14), the satu-
ration ratio increases due to chemical reaction (till the
reaction reaches equilibrium) and decreases due to nu-
cleation and condensation of the vapors. Limited infor-
mation is available in the literature regarding the chemical
reaction order or the reaction rate for the formation of
(NH
4
)
2
S
2
0
5
Hence an average value of R. was estimated
from our experimental data. The time required to reach
equilibrium, t., is between 0 and 0.4 sin our experiments
(Bai et al., 1991). The average value of Rr' was thus cal-
culated by {Ps,in- P
8
,.)N
8
v/82t.T, where Navis Avogadro's
number. This approach is better than assigning arbitrary
values to reaction rates as done by Kodas et al. (1986).
Three sets of model predictions are plotted in Figure 4.
All three sets are plotted by appropriate choice of param-
eters so that there are best fits with the number concen-
tration, geometric mean particle diameter, and geometric
standard deviation. The solid lines are plotted with the
results of model predictions accounting for simultaneous
chemical reaction and particle growth; the dashed lines are
plots for very fast reaction, or assuming that reaction
equilibrium is attained before particle growth begins. On
assuming very fast chemical reaction and on varying the
surface tension, reasonable agreement of experimental data
(number concentration, geometric mean particle size, and
standard deviation) with model predictions was obtained
for a surface tension of 70 dynfcm, this being close to the
value of H
2
S0
4
(Liu and Levi, 1980). The other dashed
line was plotted by using a higher surface tension of 93
dyn/ em, this being less than the value of NH
4
N0
3
used
in the literature [u = 113-119 dynjcm, Raes et al. (1990)].
The nucleation rate had to be scaled up by a factor of 10
9
for agreement between model predictions and experimental
data. The solid line is plotted using a t. of 0.04 s, and
reasonable agreement is obtained with the experimental
10 "!
10 '1
10
1
'*'
g 10'
:.:>
E
c
10 I I
10.
U
a
10
....
.,
.c
10'
E
::l
z
10.!
c
0
10
f-
1
0.0
A
__ a70 ergs/.em', g/.em:
0
----- aQ93 ergs;tcm!. p02.0 g;tcm,. \,=10 ,
_ a93 ergs/em . p
0
2.0 g/cm, K,6 X 10
---------------
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Average Residence Time (sec)
10 -...------------------,
E'
$
.... 10 _,
E
B
__ a=70 ergs/.cm:. p,t.B g/cm!
1
_____ a93 ergs/.em,. g/./em,. 1,10 ,
_ a-=93 ergs/em, p,2.0 g em. R, ... s X 10
.!2 10 ... ------------------------------
0
" . !
/ i
t:: 10 - I !!
cE '
c:
10 -
:IE
0
;:
., 10 _,
E
0
"
(!)
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Average Residence Time (sec)
2.0 ..,.--------------------,
1.9
1.8
1.7
"
0
'E 1.6
0
-g 1.5
.B
Ul 1.4
0
1.3
E
g1.2
(!)
1.1
c
__ a=70 ergs/.em', p
0
=1.8 g/.em:
0
_____ a93 ergs/em', p,2.0 g;tem,. 1,=10
0
_ a93 ergs/em', p
0
2.0 g/cm, R,=6 X 10
1 .0 +---,----.---,:---r-r--.--r----r:--::T"-:-1
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Average Residence Time (sec)
Figure 4. (A) Measured and simulated results of cup mixing average
total particle concentration as a function of average residence time.
The inlet 80
2
and NH
3
concentrations are 3000 and 6000 ppm, re-
spectively. (B) Measured and simulated results of cup mixing av-
erage geometric mean particle diameter concentration as a function
of average residence time. The inlet 80
2
and NH
3
concentrations
are 3000 and 6000 ppm, respectively. (C) Measured and simulated
results of cup mixing average geometric standard deviation as a
function of average residence time. The inlet 80
2
and NH
3
con-
centrations are 3000 and 6000 ppm, respectively.
data. Increasing t. (slower reaction) does not yield good
agreement with the experiments, and hence the results are
not shown. The two dashed curves indicated that there
is a burst of nucleation and formation of particles, which
then primarily grow by condensation. This is also con-
flrmed by the fl.nal geometric standard deviation tending
to 1. In the case of simultaneous reaction and particle
growth, nucleation is delayed by a fraction of a second
(short delay as reaction is still rapid). The fmal geometric
standard deviation is higher, indicating growth occurs by
both condensation and coagulation. Also, the computed
characteristic time for particle growth (T) is about 0.12 s,
much larger than t, (=0.04 s). On comparing the experi-
mental data to the three sets of model predictions, it can
be assumed that chemical reaction is extremely rapid and
completed before particle growth begins.
Scott et al. (1969) and Vance and Peters (1976b) have
considered the formation of two adducts, NH
3
S0
2
and