Ind. Eng. Chem. Res. 1995,34, 4254-4259
4254
Behavior of Venturi Scrubbers as Chemical Reactors John H. Hills Department of Chemical Engineering, Nottingham University, University Park, Nottingham NG7 2RD, U.K.
Solutions to the equation of absorption with first-order chemical reaction in spherical polar coordinates are presented in graphical form as plots of dimensionless mass absorbed against dimensionless time, with dimensionless rate constant, gas volume, and gas-side mass-transfer coefficient as parameters. An asymptotic solution which combines the effects of finite gas volume and gas-side mass-transfer resistance is derived and its range of validity assessed; a n extension to the case of infinite reaction rate is also proposed. The results are applied to the design of venturi scrubbers and compared with the experimental data of Ravindram and Pyla. Serious discrepancies are apparent, and possible reasons for these are discussed.
Introduction Venturi scrubbers have been used for a long time as devices to remove particulates from gas streams, and there have been many papers published on the modeling of this process. However, the large surface area of the drops also makes them effective gas-liquid reactors, particularly for scrubbing pollutant gases from an inert stream, and this aspect has been much less studied, although a number of papers have been published which give experimental data and attempt to model the process in a variety of ways. The models proposed (Uchida and Wen, 1973; Cooney, 1985; Ravindram and Pyla, 1986) have a number of features in common. They all treat the droplets as rigid spheres which are formed at the throat of the venturi and travel independently along the diffuser without striking each other or the walls. Diffusion and chemical reaction take place within the droplets, and overall mass balances allow conversions of the reactive gas to be calculated. They differ, however, in their degree of complexity, and in their manner of modeling the combined diffusion and reaction effects. Uchida and Wen (1973)used a one-dimensional model of the fluid flow to calculate the motion of the droplet. The gas velocity at any position was calculated from the volumetric flow rate (assumed invariant) and the diffuser cross section at that position, allowing for the volume of liquid drops present. An equation of motion for the droplet was then set up (gravity and drag terms only) and integrated to calculate droplet velocity. Correlations for gas-side heat and mass transfer were selected, based on the droplet-gas relative velocity, and mass and heat balances written in terms of the rate of dissolution of the absorbable gas component in the drop. Having noted that the reactions they were interested in (mainly SO2 into alkaline solutions) were likely to be infinitely fast, the liquid-side mass transfer was modeled by analogy with standard absorbers, multiplying the physical (nonreacting) mass-transfer coefficient, ALP, by an enhancement factor:
which assumes equal diffusivities of the dissolved gas and reactive liquid species. The physical mass-transfer coefficient was estimated from the known solution for unsteady-state absorption into a sphere:
where F is the rate of absorption (kmol s-l) and a the sphere radius. Assuming a zero concentration of absorbing gas in the bulk sphere leads to:
F = 4na2kLPCk
(3)
and ALP is found by eliminating F between (2) and (3). Uchida and Wen solved the complete set of equations to predict pressure profiles and fractional absorption of solute gas and found good agreement between their model and data from the literature. Cooney (1985) followed Uchida and Wen in the calculation of droplet trajectory and external heat and mass transfer but proposed a much more realistic model of the liquid-side transfer and reaction involving dividing the droplet into 50 concentric shells of equal thickness and writing down the flux of each species present into and out of the shell in finite difference form. In each time interval, the amounts of reaction resulting from these fluxes were calculated, and hence the amount of reacting gas dissolving. The model allows for complex equilibria within the liquid phase and could be adapted to allow for finite reaction rates, but the necessary numerical integrations are very time-consuming. He used the model t o fit some unpublished data on the absorption of H2S in ammonia solutions in a pilot scale scrubber. Subsequently, the same model was fitted to laboratory-scale measurements of absorption of SO2 and H2S in various alkalis (Cooney and Olsen, 1987). Ravindram and Pyla (1986) adopted a different approach. They used a much simpler hydrodynamic model, by assuming homogeneous flow in which liquid and gas flow at the same velocity; this may be justified for the very long diffuser used in their experiments (33 in., with an angle of less than 2"). They also neglected any gas-side mass-transfer resistance and assumed isothermal conditions. For the liquid side, they adapted a solution given by Crank (1975) for diffusion in a sphere accompanied by first-order reaction with finite kinetics. Since they were assuming homogeneous flow, they postulated that each droplet travelled surrounded by a finite volume of gas, from which the solute gas was gradually absorbed, which corresponds to one of the cases treated by Crank. They present data collected by themselves on the
0888-5885/95/2634-4254$09.QQ/~0 1995 American Chemical Society
Ind. Eng. Chem. Res., Vol. 34, No. 12, 1995 4255 performance of a laboratory-scale venturi scrubber absorbing COS and SO2 into NaOH solution: their results will be discussed further below. Of the three models, that of Ravindram and Pyla offers the advantage of a physically reasonable representation of the absorption with reaction combined with an analytical solution. This allows a simple graphical representation which would be of general use in the preliminary planning of experiments and interpretation of results. The present work generalizes and extends the work of Ravindram and Pyla to include gas-side mass-transfer resistance and presents appropriate graphical and analytical solutions which demonstrate the effects of the important dimensionless groups.
100
x
10
1 51 n a
EI
$ .-0
1
0.1
E
5 0.01
Theory 0.001
As in previous models, we assume drops are rigid spheres of radius a moving parallel to the axis of the equipment. When created, they have zero content of the gaseous reactant. The model will concentrate on the liquid-side phenomena, deriving extent of reaction as a function of drop age; the equations of Uchida and Wen (1973) can then be used to calculate droplet age and gas-side mass-transfer resistance. Following Ravindram and Pyla (1976), the liquidphase reaction is assumed t o be first order on the gaseous component and zero order on the liquid component. The latter assumption is always true if the liquid component is in large excess, which is the case in much of the published data. We can write the differential mass balance on the gaseous solute within the drop as:
Crank (1975) presents standard solutions to this equation for different boundary conditions a t the drop surface. Ravindram and Pyla (1986) concentrated on one of these boundary conditions, that of a finite gas volume (case b), but it is instructive to look a t other cases also. (a) Surface Concentration Constant: CA = CAO at r = a. This case neglects both gas-phase masstransfer resistance and also depletion of the bulk gas. The solution, in terms of the rate of uptake of gaseous component, F, is (Crank, 1975):
F=
which may be compared with eq 2 for the nonreacting case. It is more convenient to work in terms of the total mass absorbed in time t, Mt, which may be derived from the above by integration, and to make k , t , and Mt dimensionless by writing A = ka2/DA,t = tDA/a2,and X = M t / ( 4 / 3 n a 3 C ~Making ~). these substitutions, we find
0.01
0.1 1 10 Dimensionless time, kt
100
Figure 1. Plot of eq 12. The parameter is a value of A. Dotted lines are a steady-state solution.
It may be noted that X represents the ratio of the amount of gas absorbed to that which would be needed to saturate the liquid in the absence of reaction. As r =, the rate of absorption reaches a steady state (provided the liquid-phase reactant is still present in excess); the solution is analogous t o that for reaction in a porous catalyst pellet; the Thiele modulus, in our nomenclature, is 4 1 3 :
-
In reactor work, dimensionless time is more usually defined as kt, which in our nomenclature is equal to AT, and Figure 1 shows X as a function of kt with A as a parameter. The dotted lines give the asymptotic steadystate solution (eq 71, while the solid lines give the full solution (eq 6). The large enhancement of rate due to the unsteady state a t small times is clearly seen: it is greater for slower reactions (smaller A) where the initial absorption is determined by the diffusion of the gaseous component, rather than its reaction. This unsteady-state enhancement of absorption could be advantageous in a scrubber, but it only occurs at low values of kt, and in a region where the total amount of gas absorbed is less than that needed for physical saturation of the liquid (X< 1). Since the fast reactions used in venturi scrubbers have k values of 1000 s-l or more, while droplet residence times are a few milliseconds, kt will normally be significantly greater than unity, and it can be seen from Figure 1 that the asymptotic solution is valid for values of kt > 10. It may also be noted that droplets in a venturi typically have a diameter of 100 pm, so that a = 5 x m, while liquid-phase diffusion coefficients are typically of order 2 x m2 s-l, so that the term a2/DAis of order unity; thus, A would be expected t o be of order 1000 or more in most practical absorbers. (b) Absorption from a Limited Volume of Gas. This is the solution used by Ravindram and Pyla (1986), who assumed that each droplet is surrounded by an envelope of well-mixed gas which moves with it (i.e., drop and gas velocities are equal) and that only the reactive component within that envelope of volume V is available to the droplet. It can be shown that, in the absence of significant gas-side mass-transfer resistance, this assumption can be relaxed to allow relative motion of droplet and gas without changing the solution, which
4256 Ind. Eng. Chem. Res., Vol. 34,No. 12, 1995
X = 3t[hc o t h ( h ) - 11 +
is given by Crank (1975) as:
7
I1coth(\il) Jn
I1
- cosech2(h) -
1.5
a-
n 2 2exp[-t(A
n=l
where p n and q n are the roots of:
However, Crank assumes a distribution coefficient of unity for component A, so that its equilibrium concentration at the interface in both gas and liquid phases is the same. This is not true in general for real systems, for which we can define a distribution coefficient:
'
equilibrium concn. in the gas phase
= equilibrium concn. in the liquid phase
In this paper, CA will refer to the liquid: the corresponding gas side concentration is thus ~ C A . Making k and t dimensionless as before and writing V, = ?,bV/(4/3~a3), P, = pna2/DA,and Qn = qna, we can generate dimensionless solutions which express X as a function of A t with A and V, as parameters. Vr is equal to the ratio of the amount of A present in the gas envelope to the amount needed to saturate a drop in the absence of reaction. Noting that one of the roots q n of eq 9 is imaginary, we obtain, after some manipulation:
X= V,[l - exp(Pit)l- 1.5[cosech2 Qi - (coth QiYQi1 1.5 1- -[cosech2
c
n=l
-
- coth(Qi)/Qil
Vr V, exp[-t(Q;
where CA represents the interfacial concentration in either phase. If the distribution coefficient is q, eq 13 becomes:
F = 4ZCL2h(qCAo- Wck> = 4ZU21,!Jh(CAo - c,)
(14)
Using the same dimensionless groups as before, with the addition of H = ahqfDA, the solution is:
-
- A;[exp(-t(il
A t ( A +A;) X= 6H2c
(,+A;)2[An2 I
+A;))
- 11
+ H(H - 1)l (15)
+4 1
where An are the roots of (10)
VJ
(12)
(A + n 2 2 ) 2
After checking that eqs 6 and 12 yield the same numerical results, the latter was used to plot the curves of Figure 1because it converges more quickly. Its first term has already been given as the steady-state solution (eq 7)) while the second term represents the accumulated effect of the initial unsteady-state absorption; it is always positive and tends to 1.5/& for large values of A. Figure 2 is a plot of eq 10; it shows the effect of decreasing V, from infinity at three values of A. As would be expected, the major effects are felt at long times, when depletion of the gas envelope becomes severe. It follows from the definitions of X and V, that X N , is equal to the fraction of A absorbed from the gas phase (fractional recovery of A), and since we are assuming a large excess of B, this naturally tends to unity at long times. ( c ) External (Gas-Phase)Mass-Transfer Resistance with Constant Bulk Gas Concentration. Crank (1975) assumes a distribution coefficient of 1and writes:
n=l
Qi
+ n22)1
A
where
and Pi in eq 10 corresponds t o the imaginary root, iQi, of eq 11. Crank points out that, by letting V, tend to infinity, we can obtain an alternative solution to the case of constant surface concentration (case a). The result is:
A,cotA,+H-
1=0
Figure 3 gives the solution of eq 15 for three values of A for comparison with Figures 1and 2. It can be seen that absorption is reduced significantly a t short times, even for quite large values of H, so that gas film resistance cannot be ignored a priori in this case; however, at the larger values of kt typical of venturi scrubbers, the reduction is only significant for small values of H (large gas-side mass-transfer resistance) or large values of A (fast reaction). For these larger values o f t , eq 15 leads to the steady-state solution:
X , = 3Ht
H
dI coth v'I
-1
+ & coth dI - 1
(16)
(d) Gas-Phase Mass-Transfer Resistance and Finite Gas Volume. Crank does not give a solution for this case, which is likely to be important in venturi
Ind. Eng.Chem. Res., Vol. 34,No. 12,1995 4257
I)
c o t h ( 4 ) - 1)
which is a close approximation to the steady-state (first) term of eq 10 (case b). When Izt =- 10, the error is normally less than 5%,even for 1 values as small as 1. Thus, eq 19 has the correct asymptotes both when V, and when H -. This suggests that it is appropriate t o use in cases of rapid (but not infinite) reaction in which both gas-phase resistance and significant depletion of the gas phase may occur. (e)Infinitely Rapid Reaction. As the reaction rate approaches infinity, eq 19 will break down because the assumption of a constant liquid-phase reactant concentration (involved in eq 17) is no longer true. At very high reaction rates, reaction will take place on a spherical surface of radius r I a, with both gaseous reactant A and liquid reactant B diffusing toward this surface, where the concentration of each is zero. This was the case modeled by Uchida and Wen (1973), although their modification t o the classical two-film model used in gas absorption is suspect. The solution in the general case, where r decreases as reactant B is consumed, is very difficult although Hikita et al. (1977) give solutions for the simpler semiinfinite slab geometry (laminar liquid film flow). However, it is quite common to operate the venturi scrubber with a large excess of B, which results in the reaction surface lying at the interface ( r = a ) throughout the absorption and the reaction being totally gas film controlled. The rate is then given by eq 14 with Ch = 0, which may be solved with eq 18 t o give:
--
Dimensionless time = kt Figure 2. Plot of eq 10. The parameter is a value of 1.Lines: (-) v,= m; (-.- ) = 100; (*..) = 10.
v,
v,
n
loo
88
10
n
m
*
E f.-s
1
0.1
C
g
0.01
5 0.001 0.001 0.01 0.1 1 10 100 Dimensionless time M Figure 3. Plot of eq 15. The parameter is a value of 1.Lines: (-) H = -; H = 1000; H = 100;(-.*-) H = 10. (-e-)
(e..)
scrubbers. However, having noted in section a that most real scrubbers will be operating in a region where unsteady-state effects are insignificant, we can derive an approximate, pseudo-steady-state solution, and check its validity against the two limiting cases of infinite gas volume and zero gas-phase resistance. Consider the steady-state absorption rate into a drop when the surface concentration is CA. Using the analogy of the effectiveness factor in catalysis (see eq 7):
rate = F A = 4/3na3kCh(3/A)[4coth(&) - 11 (17) If the distribution coefficient is q, eq 14 applies to the gas-side transfer: FA = 4na2h(qCA-
qch)= 4na2qh(CA - CA) (14)
while a mass balance on the gas envelope surrounding the drop gives: FA
= -V d(qCA)/dt = -4/3n~3V,(dCA/dt) (18)
Assuming eq 17 also holds when C h is varying slowly (the pseudo-steady-state assumption),we can eliminate CK and FAbetween eqs 17,14,and 18 and integrate to give, after some manipulation and substitution of dimensionless groups:
(
X=V,.l-exp-
[
3Ht(.J;i coth .J;i - 1) V,(H .J;icoth .J;i - 1)
+
- -
In eq 19,letting Vr 00, we recover the steady state for case c (eq 16). Letting H 03 gives:
(20)
-
[
X = V , 1 - e x p --
(
3
3
(21)
It is comparatively easy t o check whether the condition C h = 0 is fulfilled in a given case by calculating the maximum rate at which B can be supplied to the interface by diffusion in the liquid and comparing this with the rate at which A is supplied by diffusion in the gas.
Comparison with Experiments The theoretical equations presented above may be compared with the published data of Ravindram and Pyla (1986). They used dilute C02 and SO2 in air as gaseous reactants, with a 0.6 M NaOH solution as the liquidphase reactant, and they present their data in four tables, from which it is possible to back-calculate air and liquid rates for each run and also to correct two typographical errors in their tables. They quote values for DA,K, and a for their system, which allow us to calculate 1,but do not mention gas solubility, which makes it uncertain whether they corrected Crank’s solution for the factor q. However, their data sources give values from which y j may be calculated, and the results of all these calculations are given in Table 1. They checked their assumed constant value of a = 50 pm against a photographic measurement which gave a mean of 46 pm. A graph presented by Azzopardi and Govan (1985)for mean droplet diameter as a function of throat velocity suggests that the assumed diameter is reasonable, although it should decrease slightly with gas rate. Ravindram and Pyla quote Hikita et al. (1977) for the value of k for S02, but these authors refer to S a d (19281,who found that k for SOdNaOH was at least
4258 Ind. Eng. Chem. Res., Vol. 34, No. 12, 1995 Table 1. Parameters for Ravindram and Pyla Data parameter DA/(m2 s-1)
COflaOH 1.868 x 6 x lo3 5 10-5 8030 0.743
kls-l
alm
A.
w
SOflaOH 1.776 x 6 x lo8 5 x 10-5 8.45 x lo8 0.031
Table 2. Ravindram and Pyla Data for C02-NaOH fraction of COz absorbed run
V,
t (s)
1 2 3 4 5 6 7 8 9 10 11 12 13 14
85.4 130.7 170.9 217.6 268.9 301.6 336.8 403.0 336.8 282.6 248.4 215.6 196.7 176.9
0.3548 0.2330 0.1782 0.1400 0.1134 0.1012 0.0907 0.0907 0.0907 0.0907 0.0907 0.0907 0.0907 0.0907
expt
calc
(R&P) 0.7039 0.7585 0.8272 0.9156 0.9459 0.9667 0.9824 0.8662 0.8892 0.9237 0.9467 0.9544 0.9621 0.9736
(R&P) 0.9965 0.9983 0.9933 0.9698 0.9035 0.8497 0.7804 0.7179 0.7811 0.8942 0.8731 0.9069 0.9256 0.9433
calc (this work)
0.5617 0.2983 0.1871 0.1200 0.0804 0.0645 0.0521 0.0438 0.0521 0.0618 0.0700 0.0802 0.0876 0.0969
lo9 m3 kmol-l s-l; most other workers have taken the S O D a O H reaction to be infinitely fast. Using the values in Table 1,with the assumptions of homogeneous flow and zero gas film resistance, we can estimate V, for each run and calculate X /V, (conversion of gaseous reactant) from eq 10 or its steady-state version eq 20. Table 2 presents the results for COS and compares them with the measured values, and also with the calculations of Ravindram and Pyla. Clearly, the absorption predicted by our model is very much less than both the measurements and the original model predictions. Since the two models use the same basic equation, apart from the factor yj, which is close to unity for C02, they should give the same results. The continuity between the infinite series solution (eq 10) and the readily verifiable asymptotic solution (eq 20) gives us confidence in our calculations, and we feel that Ravindram and Pyla must have made a calculation error. However, the big discrepancy between our model and the experimental data needs explaining. It is possible that the experimental data are themselves faulty; for example, if the bubblers used to measure the unreacted C02 were inefficient, the absorption in the venturi would be overestimated. The data also show an increase in conversion with gas rate a t fxed liquid rate (runs 1-7), although both predictions say it should decrease. Inefficient bubblers could also lead to this effect, as the inefficiency would be expected t o increase with gas rate. The model assumes a constant droplet size, but the droplet size will decrease with gas rate, leading to smaller 1,and increased conversion. However, it would need a 4-fold reduction in the droplet size to overcome the effect of the 4-fold reduction in the residence time, and the graph of Azzopardi and Govan (1985) suggests much smaller changes. The model also assumes no collisions of droplets with the wall or with each other; the former effect should reduce conversion by reducing interfacial area, but the latter might cause a significant increase by promoting internal circulation where the model assumes rigid spheres.
The droplet residence time in Table 2 was calculated from the homogeneous model as (volume of venturi)/ (total fluid volumetric flow rate), but we can use the method of Uchida and Wen (1975)to track an individual droplet. The initial droplet velocity is assumed to be the jet velocity in the nozzles, the drop size is taken as 100 pm, and the physical properties are those of air at ambient conditions. The calculations have been made, and they yield times between 75% and 85% of the values calculated by the homogeneous model, which means that this correction would reduce still further the predicted conversions and increase the discrepancies between model and experiment. External mass-transfer resistance was also neglected by Ravindram and Pyla. Uchida and Wen suggest the following formula for its estimation:
The drop-tracking calculations above yield point values of NSh which give an average over the diffuser of between 2.5 and 3.7: the “no-slip” assumption of Ravindram and Pyla would give N R =~ 0, and thus NSh = 2, and we will take this latter as the more pessimistic estimate. Comparing the definitions O f N S h and H
hd NSh=$ G
and
ha
H=q-D A
we see that:
where DG is the gas-phase diffusion coefficient of A. Substituting numerical values leads to H = 2.5 x lo4 for COS and 8.5 x lo2 for SOZ. For COS, the effect of including gas-side resistance and setting H = 2.5 x lo4 rather than H = m can be calculated using the approximate equation (19);for the highest conversion (run 1in Table 2) the calculated conversion falls from 56.19% to 56.06%, which is indeed negligible, and a similar result holds for all the runs. For the SO2 data the present model, neglecting gasside resistance (eq lo), predicts 100.00%conversion (Xl V, = 1.0000 to four places of decimals) in all runs, which is again higher than that predicted by Ravindram and F‘yla using ostensibly the same model and also higher than the experimental results. Using the value of H = 850 calculated above and eq 19 reduces the predicted conversion below 100%in runs 5-12, showing that gasside resistance is not negligible. The greatest effect is in run 8, where predicted conversion falls to 99.46%. The infinite reaction rate equation (21) gives very similar-results, with a conversion in run 8 of 99.37%. The data used by Uchida and Wen (1973) is not readily accessible in the open literature. Moreover, the majority of it is for the SOflaOH system which is expected to be gas-film limited, so the data would do little other than check the correlations used for the gasside mass-transfer coefficient. Work in progress in our laboratory aims to measure absorption in a venturi system for slower reactions, to check the results of this paper. Conclusions Modeling of the absorption with a chemical reaction which takes place in the droplet phase of a venturi
Ind. Eng. Chem. Res., Vol. 34, No. 12, 1995 4269 scrubber suggests that there is an initially rapid absorption, which tends to an asymptotic rate afier a dimensionless time Kt > 10. The rapid reactions normally encountered in real scrubbers (acid gas plus alkali) lie in this asymptotic range, which means that the effect of mass-transfer resistance in the gas phase and depletion of the gaseous reactant on conversion can be readily estimated by an equation such as eq 19. Data with which to check this conclusion are rare, and the major set, presented by Ravindram and Pyla, show large deviations from the model. Work is in progress to collect additional data.
Nomenclature a = droplet radius (m) CA= concentrationof gaseous reactant in liquid (kmol m-3) CA = equilibrium value of CA at surface of drop (kmol m-3) CAO= constant (case a) or initial (other cases) value of C k (kmol m-3) d , = droplet diameter (m) DA = diffusion coefficient of gaseous reactant in liquid phase (m2s-l) DG = diffision coefficient of gaseous reactant in gas phase (m2s-l) E = enhancement factor F = rate of absorption into droplet (kmol s-l) h = mass-transfer coefficient in gas fdm (m 8-l) H = dimensionless mass-transfer coefficient = ahly/DA k = first-order rate constant (s-l) kLp = physical mass-transfer coefficient in liquid (m s-l) Mt = amount absorbed in time t (kmol) N s= ~ Sherwood number in gas film p , qn = roots of eq 9 (s-l and m-2, respectively) P, Qn= dimensionless forms ofp, qn,roots of eq 11 r = radial coordinate within drop (m) t = time (s) V = volume of gas envelope surrounding droplet (m3)
V, = dimensionless volume = v V / ( 4 / 3 ~ a 3 )
X = dimensionless mass absorbed = M t / ( ' 4 / 3 ~ a 3 C ~ ~ ) Greek Letters 1 = dimensionless rate constant = ka2/DA z = dimensionless time = tDA/a2 q j = distribution coefficient
Literature Cited Azzopardi, B. J.; Govan, A. H. Annular Two-Phase Flow in Venturis. European Two-phase Flow Group meeting, Marchwood, U.K., 1985. Cooney, D. 0. Modelling Venturi Scrubber Performance for H2S Removal from Oil-shale Retort Gases Chem. Eng. Commun. 1985, 35, 315. Cooney, D. 0.;Olsen, D. P. Absorption of SO2 and H2S in smallscale Venturi Scrubbers. Chem. Eng. Commun. 1987,51,291306. Crank, J. The Mathematics of Diffusion, 2nd ed.; Clarendon Press: Oxford, 1975; pp 334, 340, 346. Hikita, H.; Asai, S.; Tsuji, J. Absorption of Sulfur Dioxide into Aqueous Sodium Hydroxide and Sodium Sulfite Solutions. AIChE J. 1977,23, 538. Ravindram, M. and Pyla, N. Modelling of a Venturi Scrubber for the Control of Gaseous Pollutants. Ind. Eng. Chem. Process Des. Dev. 1986,25,35-40. Saal, R. N. J. The velocity of ionic Reactions I & 11. Recl. Trav. Chim. 1928,47, 73 and 264. Uchida, S.; Wen, C. Y. Gas Absorption by Alkaline Solutions in a Venturi Scrubber. Ind. Eng. Chem. Process Des. Dev.1973,12, 437-443.
Received for review March 7, 1995 Revised manuscript received July 19, 1995 Accepted July 28, 1995* IE950165U
* Abstract published in Advance ACS Abstracts, October 15, 1995.