... .

corporatton
1:" -------------------,
c
....: - - .
ANALYSIS OF MASS TRANSFER MODELING
AND VAPOR-LIQUID EQUILIBRIUM IN
SCRUBBING RETORT OFF-GAS
Prepared By
MITTELHAUSER CORPORATION
for
U. S. Department of Energy
Laramie Energy Technology Center
P. O. Box 3395, University Station
Laramie, Wyoming 82071
December, 1981
"
.
. -1-:-'-- .-.. ' ,. corporatlon_. ----.......... ----.......... ------------------------------,1-
C, TABLE OF CONTENTS
I
1.0
2.0
3.0
4.0
5.0
6.0
SUMMARY
INTRODUCTION
MASS TRANSFER MODELS
Page No.
1
3
4
3.1 Venturi Scrubber Continuous Contact Model, 4
3.2 Venturi Scrubber Stage Contact Model 12
3.3 Sieve Tray Scrubber Stage Contact 13
3.4 Physical Properties 15
VAPOR-lIQUID EQUILIBRIUM (VlE) IN NH3-H2S-C02-H20
SYSTEM
4.1 VlE Data
4.2 VlE
4.2.1 Van Krevelen
4.2.2 API SWEQ
4.2.3 Institute of Gas Technology
4.2.4 EMNP
4.2.5 Beutier and Renon
4.2.6 ECES
4.3 Process Simulation Software
4.3.1 ChemShare
4.3.2 . SSI Process
4.3.3 OlI Systems
SUPPORTING WORK
5.1 Preliminary Simulation Work
5.2 Determination of Oil Phase Importance
5.3 Determination of Organic Acids and Bases
CONClUS IONS
20
20
23
23
23
24
24
24
25
26
26
26
26
29
29
29
30
33
I'----'" ____
i
tlt1!TTELHAUSER
< - corooration
-4-- ,L'r::::----- -_.-- _. - ._-- .- - ..
. ,

(
.
.... \
1.0 SUMMARY
Previous experiments at the Laramie Energy Technology Center
.- .--.---
indicated that a significant fraction of ammonia (NH3) present in retort
off-gas was removed while operating a venturi scrubbing device for parti-
culate removal. Such results plus the commercial history of using ammonia": __
water solutions for removing hydrogen sulfide (H2S) from coal-gas streams
suggests the possibility of using scrubbers for mass transfer as well as
I
particulate removal. This report addresses mass transfer considerations
related to scrubbing retort off-gas. It examines approaches for calcula-
ting mass transfer in venturi and sieve tray scrubbers and for determining
vapor-liquid equili&rium.
Two different types of mass transfer models are examined-con-
tinuous contact models and stage contact models. Continuous contact
models are derived from the solutions to equations that
mathematically describe material balance, mass transfer rate and equilib-
rium relationships. Typically such models are appropriate where mass
transfer takes place in a continuous manner such as a packed absorber.
Stage contact models treat mass transfer as taking place in discrete
stages such as a tray of a fractionator. Algebrian equations are written
and solved around the stage to account for material, energy and equilibrium
relationships.-
This work concludes that continuous contact models for mass
transfer in a venturi scrubber are not practical. There is little
previous work on such models for venturis. Further the system is relatively
complex due to mass transfer of three species - NH3, H2S and C02 - being
considered. Previous work, usually on packed towers or spray scrubbers,
most often is limited to mass transfer of one species. Also the mathematics
of continuous contact models requires simplifying by means of assumptions
such as linear equilibrium and constant temperature.
Unfortunately these assumptions aren't true in scrubbing retort off-gas.
As a consequence, the most practical type of modeling is a stage contact
model. This type modeling can be used for both venturi and sieve tray
scrubbers. In addition, this type modeling can be implemented using
.MiITTELHAUSER
.. ________________ ---...,
commercially available process simulation packages. These packages are
C computer programs that have been prey; ous ly used for s imu 1 at ing process
operations. They incorporate recent in the prediction of
vapor-liquid equilibrium (VLE) in NH3-H2S-C02-H20 systems.
Significant advances were made in the 1970's relative to
establishing the thermodynamics of electrolytic solutions and predicting
VLE in NH3-H2S-C02-H20 systems. Consequently VLE models are available
and they are described in this report. Several are available as
computer programs or in conjunction with commercial simulation programs.
A couple potential problems warrant research investigation to
determi ne the ir importance. Scrubb i ng retort off -gas wi 11 produce a
hydrocarbon phase in add it i on to the gas and aqueous phases. The presence
of this phase could complicate mass transfer predictions; hopefully it
is of minor. importance and can be neglected. The retort off-gas also
contains polar hydrocarbons that will act as acids and bases in aqueous
solutions. These compounds will tend to fix the soluble NH3 and H2S in
solution. Further research can determine the significance of the hydrocarbon
phase and polar hydrocarbons.
It is recommended that mass transfer calculations be performed
using one of the available simulation packages. Liquid and vapor rates,
compositions and temperatures corresponding to planned field scrubbing
tests should be used in simulation work in order to approximate test
conditions as closely as possible.
MITTELHAUSER _____ _
-' -,1.----------- __ corporation __ ........ _________________ --,
..
.(
2.0 INTRODUCTION
This report addresses mass transfer considerations related to
scrubbing oil shale retort off-gases. Experiments during September, 1980,
at Laramie Energy Technology Center's retort, located at the U.S.
DOE North Site in Laramie, tested a venturi scrubbing device for particulate
removal. Data from these tests indicated that a significant fraction (50
to 75 percent) of ammonia (NH3) present in the retort off-gas was removed in
the venturi scrubber. These results, plus the commercial history of using
ammonia-water solutions for selective removal of hydrogen sulfide (HZS) from
coal-gas streams, suggests the possibility of using scrubbers for mass trans-
fer as well as parliculate removal. Removal of a significant fraction of
the H2S would reduce the costs associated with downstream acid-gas removal
by processes such as Stretford.
Mass transfer in retort off-gas scrubbing systems is complicated.
Gas streams are a mixture of gases, hydrocarbons, water and soluble chemi-
cals. Cooling of these streams produces a three phase mixture--gas, aqueous
and oil. The calculation of vapor-liquid equilibria under such circumstances
is not straight forward. Several of the techniques for calculating vapor-
liquid equilibria and mass transfer have been developed only recently.
This report discusses approaches for calculating mass transfer in
venturi and sieve tray scrubbers and for determining vapor-liquid equi-
librium. Two different mass transfer modeling methods are described. Sources
of VLE data are summarized and VLE models, several of which are quite recent,
are described. Available process simulation software which can be used to
predict removal ofNH3 and HZS is summarized. Supporting engineering and
laboratory work that would assist in better understanding absorption of
NH3 and HZS is discussed.
3
.(MITTELHAUSER
... I.
(
Mass transfer models may be approached on the basis of stage
contact or continuous contact. In one sense the approach parallels the type
of equipment used to bring about mass transfer. For example, trayed
absorbers or fractionators are designed by calculating heat and material
balances around stages that are analogous to the trays in the tower.
Continuous contact equipment such as packed absorbers are designed using
analytical or numerical solutions to differential equations that mathematically
describe material balance, mass transfer rate and equilibrium relationships.
An acceptable approach does not have to parallel the type equipment. Stage
contact calculation procedures, for example, are often used to design packed
towers.
Process conditions and availability of experimental or operating
data can be an important consideration in the selection of a valid, useable
model. A continuous contact approach, for example, is usually dependent
upon process assumptions such as
Constant temperature
Linear vapor-liquid equilibrium relationships.
Further it may require data on mass transfer rates and geometric considerations
that have to be determined experimentally and that are variable with applications
and conditions.
Rigorous stage contact models are not as sensitive although simpli-
fied ones can be. The basic requirements for a good stage contact model are
accurate equilibrium prediction methods and enthalpy data.
3.1 VENTURI SCRUBBER CONTINUOUS CONTACT MODEL
Venturi scrubbers are gas-atomized spray devices. The predominant
mechanism for mass transfer in such devices is absorption by droplets.(l)
4

_+.-_' ___ _ c:;orporation _____________ --_" _--_--_-_____ --,
Almost all of the mass transfer takes place where the gas velocity is highest,
i.e., in the venturi throat and immediately downstream.
Mass transfer in venturi scrubbers is not generally discussed in
the literature. The scrubbers are primarily used for particulate removal.
The most comprehensive treatment of mass transfer is found in a handbook
titled, Wet Scrubber System Study: Volume I, Scrubber Handbook, prepared
for the Environmental Protection Agency. The modeling approach described
here parallels the treatment in this handbook.
The general equation for mass transfer of component i between two
phases is (see nomenclature pages):
(1)
where mass transfer is characterized by a local gas-phase mass-transfer
coefficient kGi' Much of the mass-transfer information in the literature is
expressed in the" form of "overall" coefficients to avoid the problem of
determining interfacial properties. Such a coefficient for the gas-phase
would be written KGi, and the flux would be
A A . *
Ni = KGi (pl - Pi )
(2)
where Pi*, the partial pressure of i at equilibrium with respect to the bulk
liquid phase composition of i, is much easier, at least in theory, to determine.
For example, if component i obeyed Henry's Law,
* !\
Pi + Hixi
(3)
where Xi is the mole fraction of i in the bulk liquid, and is the constant
of proportionality, usually called the Henry's Law Constant.
This simple picture is complicated in the retort off-gas system by
the following considerations:
5
(

Henry's Law in not obeyed .
There may be two additional diffusion films to worry about
because of oil condensation.
1) gas-oil
2) 0 il-water
For the moment, consider the case in which there is no oil film.
is not constant with respect to composition, but if we can -----------
deterwine that the vapor-liquid equilibrium curve in the range of compositions
of inter"est is essentially straight, we can write:
1
(4)
It is vital to note that Hi in this equation is not the,Henry's
Law Constant, but the slope of the equilibrium curve. Also it is essential
to understand that equilibrium curve means the general relation between bulk
liquid and bulk vapor compositions at equilibrium. One could plot:
* Yi
Pi*
fi
versus
versus
versus
* x'
1*
C
1
ai
(mole fraction)
(partial pressure vs concentration)
(fugaCity versus activity)
matter so long as one defines the coefficients and constants of
proportionality in such a manner as to make the interfacial fluxes come out
in the correct units. The important conclusion to be reached is:
It is possible in theory to construct for any component i a vapor-
liquid equilibrium curve, using available models and accurate data
on one of the bulk phases. The slope of this curve in the concen-
tration ranges of interest, when suitably multiplied by appropriate
conversion factors, can be used as the constant Hi in equation
(4). It is not necessary to assume that Henry's Law holds in
order to use equation (4).
To use equation (4), we do need to evaluate kGi, kLi' and Hi' Some
simplification is possible and in the system of interest:
L-.-________ ___ . _________
- 6
LAITTRHAUSER
__ corporation
' l ~
! For NH3 ana H2S, mass-transfer is gas-film controiledl ii, \L
c; and the dissociation reactions in the liquid are so rapid(l
that we can assume that there is no concentration gradient
within the liquid droplets.

For C02, mass-transfer is liquid-film controlled.(2)
First let us consider NH3; gas-film control means that:
(5 )
'"
Reference (1) gives an expression for kGi of:
(
0 .) 0.5
_Gl
t
(6 )
in which t is the "exposure time" between the gas and 1 iquid. This can be
approximated by the ratio of the drop diameter to the relative velocity be-
tween the gas and the drop. If F is defined as the ratio between (1), the
relative velocity of _the drop to that of the gas and (2) the gas velocity
i tse 1f, then
F =
u
r
(7)
llG
and t ~
dd
(8)
FUG
-- -- -----
Therefore '"
,....,
2
(
DGiUGF) 0.5
KGi
"'-
RT
dd
Evaluating DGi will be discussed in a later section.
Referring to equation (2), and setting Pi = YiP and Pi*= Yi*P,
then
l\ /\. *
Ni = KGi
P
(Yi - Yi ) (10)
Now, the molar flux of i that leaves the gas phase can be con-
verted into an overall transfer rate by multiplying both sides by the inter-
facial transfer area, which is assumed to be the total surface area of the
drops, Ad.
7
r!l1 ITT ELHAUSER
, ___ -.-::. __ == corporotion _____________________ --,,-
The droplets in a venturi agglomerate during their passage through
the scrubber. Therefore Ad will change with the distance from the pOint of
atomization. Consider a thin slice of venturi volume at a pOint some distance
away fl"om the point of atomization. -Within this volume, if the molar gas
710wrate is assumed essentially constant, the moles of i transferred from
gas to liquid are
(11 )
where G' is the total gas molar flow rate, gm moles/sec relative
to the flux, G' = GAT, where AT is the total cross-sectional area of the
slice of venturi we are considering.
Then, from equations (10) and (11)
This is the general equation of mass transfer in a venturi. It
incorporates no special assumptions other than that of constant -Gi. Any
,ralid expressions for KGi' Ad, or Yi*' can be inserted to build a new model
as new data are developed.
For the particular case under consideration, if it is assumed that
1 iquid d : ~ o p area varies with distance from the atomization point by the
relation given in Reference 1:
(13 )
equations (9) and (13) can be substituted into equation (12) to get:
,
- G dYi =
(Yi-Yi *)
8
(14 )
MITTELHAUSER
>.--'-- . - corporaUon ........ ______ -_ .. _.--_. -_-_--_-_____ ---'''-'
Now assumptions must be made about:
The variation of F with z. F will always have a limiting initial
value less than 1.0, and that it will decrease with distance
until the drops are traveling at the same velocity as the gas,
at which point F = O.
The variation of AT, the venturi cross-sectional area, with z.
How to express or determine the "average" drop size.
A useful result is obtained if it is assumed that, not only is
there zero liquid-phase resistance, but also that the equilibrium partial
pressure of i is zero over the bulk liquid. The resulting solution to equation
(12) gives the maximum possible degree of mass transfer for a qiven set of
venturi dimensions, rates and initial concentrations.
To obtain this result, assume:
( 15 )
which comes from Reference 1

AT
is constant with z
58,600
0.5 0.45
dd
= d
s
= (6
p
L)
+ 597
I

(I)L(\) 0.5
uR

1.5
10000G (16)
OL
which is an empirical correlation of "average" drop diameter versus
liquid properties and gas-liquid ratio.
Substituting equation (15) into (14) and integrating both sides
(remember, Yi* = 0 by assumption) between z = 0 (Yi = F = Fi) and z = 1
(Yi = F = 0) gives
9
MITTELHAUSER.
, corporatlon ___________ ---' __ ---' ______ ,
- j1I.---- -- .
Yi
y1
1
An expression for dd comes from equatlOn (16); The "average"
value of F can be taken from the following curve taken from Reference 1:

..
o
..
1.0 L/G
o et 'J94G
6 A.M.
___ for
... (or part icle col!cctioll
frol'll FljtUfP 5.3.6""
A
A
2.-0
3.0
VALUES OF F VS. FOR S02 ABSORPTION BY SODIUM
HYDROXIDE SOLUTION
( 17)
Note that L/G is a molar ratio, not a mass ratio; therefore the
molecular weights of both liquid and gas must be known to compute this term
from run data.
The curve of y? Iy for NH3 and H2S versus, say L/G, ca 1 cu 1 ated
from equation (17), should lie below all other curves and below the experimental
data points for the run.
L __ ______ _____________
10
However, even higher mass transfer might be accomplished if a
significant liquid film formed on the walls and added to the available area
for mass transfei. If this happens, the equations must be adjusted since kG
for transfer to a film will be different from that for drops.
MITTELHAUSER .
-'-1',---- . ___ corporotJoo ___________________ t
The presence of condensed oil could affect the efficiency of mass'
transfer if the oil formed a film around the outside of the water droplet.
If, however, the water formed outside the oil droplets, no significant change
in mass transfer would be expected, except for any effect due to changes in
internal liquid circulati?n within the drop and thus to kLi.
~ . _ _ _ _ _ When COZ is absorbed, the liquid-film resistance controls the mass
transfer because of the relatively slow ionization
COz + HZO ! HC03 + H+
which must take place before the combination with dissolved NH3 can occur:
Data on diffusion coefficients in electrolyte solutions are necessary to
evaluate kLi' which must be done to evaluate liquid-film resistances. The
problem is quite complicated especially when mixed electrolyte solutions are
-involved. Ratcliff and Holdcroft (3) reported data on the diffusion of COZ
into electrolyte solutions. The diffusion coefficient was found to decrease
linearly with an increase in salt concentration.
For calcualation purposes DCOZ,L could be computed as if it were
diffusing in pure water and then an empirical correction factor based upon
the Ratcliff and Holdcroft data could be applied.
A
Thus in equation (lZ), for COZ the expression for KG; given by
equation (4) would be used. In integrating equation (lZ) the variation in H
with distance along the z-axis, would have to be recognized; i.e., as Yi
decrease, xi increases, and therefore Hi, which is dependent on the con-
centration not only of i but of all other dissolved species, also changes.
However, it may be possible through construction of a family of Yi vs xi
curves at differing concentrations of NH3 and HZS in the liquid, to establish
* * an average slope of the Yi vs xi curve for COZ, and thus to remove Hi from
the integral.
11
MITTELHAUSER
-:f -__ ____ corporaUon
(
3.2
VENTURI SCRUBBER STAGE CONTACT MODEL
r ~ a s s transfer in a venturi scrubber could be modeled as a stage
contact device wherein the scrubbing liquor and retort off-gas are brought
together in i nt imate contact. The contact is ana 1 ogous to a stage in a
tray tower.
t1ass transfer in a stage contact dev i ce is determi ned by wri t i ng
component material balances and heat balances around each stage. The
material balances must account for the equilibrium relationships that exist
between vapor and liquid. In the typical textbook development, a number of
simplifing assumptions are made in order to reduce the calculations to a
level that can be handled by hand computational methods. The classical
McCabe-Thiele method for analYSis of fractionators is an example.
The problem in scrubbing retort off-gas is not amiable to sim-
plified hand computational methods. Rigorous heat and material balances
must be performed. There are multiple compounds that must be included in
the calculations. Further, soluble gases such as ammonia, carbon dioxide,
and hydrogen sulfide ionize in the liquid phase and the liquid equilibrium
must be accounted for. The equilibrium between the vapor and the liquid is
difficult computationally. Sensible,and latent effects arising from com-
bining a liquid and gas stream at different temperatures must be properly
accounted for. The presence of two liquid phases, oil and aqueous, com-
plicates the calculations and requires that the equilibrium between the two
liquid phases be accounted for.
Nonetheless, the recent development of commercially available
software that is applicable to handling problems of this type makes a stage
contact model viable. Section 4.3 of this report describes this software.
The capabilities of the software described have been used to calculate
absorption of soluble species into aqueous solutions and the subsequent
. ---
stripping of the solutes.
A stage contact model is an equilibrium model. It assumestnat---
everypiace that liquid-vapor contact occurs, the component species in tne
vapor and liquid are in equilibrium of one another. In most stage
12
MITiELHAUSER
I
corporation -- --'".'---" ..
--:- :':._. ------- -- -- - -- - . --------------------'------t
contact devices, such as absorbers or tray towers, equilibrium at each stage
is not usually achieved because there is not sufficient residence time or
contact between phases. The deviation from equilibrium is accounted for by
establishing an efficiency term.
In a highly efficient contact device, such as a venturi, equili-
brium may be approached closely. Absorption of some species such as H2S may
even exceed equilibrium prediction due to relative absorption rates. Be-
cause absorption of NH3 and H2S is fast relative to C02 (due to the slow
ionization of the C02 in the aqueous phase) H2S removal could exceed pro-
jections.
The efficiency term that makes the most sense in the case of a
venturi scrubber is one that is analogous to the Murphree Tray Efficiency.
The term is defined as follows:
EM
y1
-
Yi
o
= 1
i
*
Yi
-
Yi
This efficiency can be evaluated from scrubber test data. yq and y1 are
1 1
taken directly from the test data (after correcting for water vapor that may
or may not be reported in the analysis) .. The equilibrium concentration ii
is determined by running one of the process simulation programs previously
mentioned. The efficiency is calculated for each component.
Other parameters regarding NH3, H2S and C02 removal can be computed
as well and compared to theoretical results predicted by the computer models.
For example, the percentage of each compound actually removed in the scrubber
test can be compared to the amount that the computer program predicts will
be removed.
3.3 SIEVE TRAY SCRUBBER STAGE CONTACT MODEL
Sieve tray scrubbers are stage contact devices and are most appro-
priately modeled as such. The discussion in the previous section regarding
, .,
I.)
MITTELHAUSER
.. -J- _'____corporotioo--'-___________________ --.
venturi scrubber stage contact model applies here too. The approach is
basically the same but the effort is more complici,l.ted in that the sieve tray
scrubber contains multiple stages whereas the venturi is only a single stage
contact.
Again, commercially available software is used to predict scrubber
performance. However, the program used should have the capability to simulate
multiple stage towers whereas a simple flash calculation program will simulate
the venturi scrubber.
The efficiency matter most likely has to be approached differently.
The Murphee efficiency described in Section 3.2 is valid, but its determination
requires gas analysis between stages. The LETC scrubber is not so equipped
and if it were, the analytical requirements would be burdensome.
The common approach on tray towers is to compute an overall effi-
ciency, Eo. The overall efficiency is the ratio of the theoretical stages
required to perform a separation to the actual number of trays in the
scrubber. Theoretical stages are determined from a process simulation using
available software. An overall efficiency is not generally a mathematically
rigorous approach. However, as a practical design matter, it is the most
commonly used efficiency.
An overall efficiency can be mathematically related to a Murphee
efficiency if equilibrium relationships and material balances can be
e x p ~ e s s e d in linear equations and the Murphee efficiency is constant for all
temperatures.
A
log [1 + EGC
HG
- 1)]
't
"
10g(HG) .
r
It is not likely that the applicable restraints will exist when scrubbing
retort off-gas.
- .
MftTELHAUSER
corporation ______________________ --.
3.4 PHYSICAL PROPERTIES
A continuous contact model approach such as described in Section 3.1
requires physical property data. This section gives recommended calculation
methods or data sources for the property data needed. These physical pro-
perties include:
'"'
gas density, PG or PG
gas viscosity,
gas phase diffusivity, DGi
liquid density, Pl
liquid phase diffusivity, Dli
tension, et
Th-e recommended methods are descr i bed in the book, The Pro pert i es
of Gases and Liquids (4). Table 3.0-1 cites the name of the method and the
page in the book where the method is described. The methods are amenable to
hand or computer computation. Alternatively, the commercial process simu-
lation programs described in Section 4.3 can be used to determine several
of these properties. These simulation packages will typically calculate
density and viscosity. Some will estimate surface tension.
15
" I
I
c
Phase
Gas
Gas
Gas
Liquid
Liquid
Liquid
Liquid
TABLE 3.0-1 PHYSICAL PROPERTIES METHODS/SOURCES
Property
Density
Viscosity
Diffusivity
Density----
Viscosity
Diffusivity
Surface
Tension
Method/Source
Use Ideal Gas Law
Use Chapman-Enskog theory
approximation, p.410 of
Reference 4
Use methods described in
Chapter 11, p.544 of
Reference 4
Use field measurement
or .the density for water
Use viscosity of water
or calculate per Table
9-13, p.454 of Reference
4
Use Wilke-Chang method,
p.567 of Reference 4
Use data from Appendix B
of Reference 5
16
Comment
Deviation from ideal at
scrubbing conditions is
negligible
Use Brokaw approximation
of parameters 0ij
Diffusivity in a multi-
component mixture is
difficult. Recommend cal-
culating diffusivity in N2,
the predominant gas phase
constituent
Use water as the solvent
MITTELHAUSER
C
(
~ "
, corporatioo ______________________ ---.
a
A
c
D
d
E
f
F
G
G'
H
k
K
K
1
L
m
M
N
P
P
Q
R
S
r
T
t
u
V
W
w
x
y
z
15
p
]J
Ijl
A
I
*
i
0
NOMENCLATURE
activity
area, cm
2
; also interaction parameter in property equations
mass or molar concentration, gm or gm-mole/cm
3
diffusivity, cm
2
/sec
diameter, cm
efficiency
fugacity, atm
ratio of relative velocity of droglet or particle to gas velocity
gas mass or molar flux, gm/sec/cm
2
or gm-moles/sec/cm
Z
gas mass or molar flow rate
Henry's Law Constant
local mass transfer coefficients
overall mass transfer coefficients
equilibrium ratio between liquid and vapor mole fractions
length, cm
liquid mass or molar flux, gm/sec/cm
2
or gm-mols/sec-cm
2
molality, gm-moles solute/1000 gm solvent
molecular weight gm/gm-mole
molar flux, gm-mol/sec-cm
2
partial pressure, atm
total system pressure, atm
volumetric flow rate, actual cm
3
/sec
gas constant, in cm
3
- atm = 82.06
gm-mole-OK
interaction parameter in viscosity equation
radius, cm
temperature, oK
time, sec
velocity, cm/sec
volume
weight flow, gm/sec
weight fraction
liquid mole fraction
vapor mole fraction
distance, cm
surface tension, dynes/cm
mass or molar density, gms/cm
3
or gm-mol/cm
3
viscosity, gm/cm-sec. (except where noted)
interaction parameter in property equations
Superscripts
molar quantities
value at phase interface
value at equilibrium with respect to bulk property of other phase
(in a two-phase system)
average
inlet
outlet
17
MITTELHAUSER
. corporation _____________________ --,
T
G
L
i
M
a
R
S
d
j
total
gas-phase
1 iquid-phase
Subscripts
pertaining to component "i"
Murphree tray efficiency
overall tray efficiency
relative
Sauter mean
droplet
pertaining to component "j"
18
(
IlAltTELHAUSER
:,' corporation _____________________
i
(1)
(2 )
(3)
(4)
(5 )
REFERENCES FOR SECTION 3
Calvert et al, Wet Scrubber System Study: Volume I, Scrubber
Handbook, Report No. EPA-R2-72-118a (August 1972).
Kohl, A. L. and F. C. Riesenfeld, Gas Purification, 3rd edition,
Gulf Publishing Co., Houston (1979).
Ratcliff, G. A. and J. G. Holdcroft, Trans. Inst. Chern. Eng.
Lond., i,l, 315 (1963).
Reid, R. C.; Prausmitz, J. M. and Sherwood, T.K.; The Properties
of Gases and Liquids, 3rd edition, MCGraw-Hill, New York (1977).
Metcalf & Eddy, Inc., Wasterwater Engineering, 2nd edition,
McGraw-Hill, New York (1979).
19
MITTELHAUSER
+"':--'- corporatioo-=--=--=--=--=-=---=-=-=-________________ ..,
4.0 VAPOR-LIQUID EQUILIBRIUM (VLE) IN SYSTH1
4.1 VLE DATA
Experimental measurements of VLEdata are reported in the following
references.
System Reference
NH3-COZ-HZS-HZO D. L. Cardon'and G. M. Wilson, "Ammonia-Carbon Dioxide-
SOoC t lZOoC Hydrogen Sulfide-Water Vapor-Liquid Study," Publication
________ 0 ____________ of the API, Washington, D.C.
NH3-COZ-HZS-HZO E.H.M. Badger and L. Silver, J. Soc. Chem. Ind.,S7,
_____c-._l.l 0- 1Z (1938).
NH3-HZO
OOC to 60
0
C
NH3-HZO
970C to 147
0
C
NH3-COZ-HZS-H20
20
0
C
NH3-C02-H20
200C to 40
0
C
COZ-
H
20
oOC to 60
0
C
H2
S
-
H
20
800C to lSOoe
NH3-HZS -H20
80
0
e to l20
0
e
NH3-C02-H20
20
0
C to 1000C
NH
r
H
2
0
ODC to 60
D
C
Breitenbach, Bull. Univ. Wis. Eng. Exp. Sta. Ser. 68
(as given in Perry's Chemical Engineers' Handbook,
Fourth ed., 1963).
1. L. Clifford and E. Hunter, J. Phys. Chern., E, 101
(1933) .
1. G. C. Dryden, i!. Soc. Chern. Ind., 66, 59 (1947).
S. Ikendo, Kogyo Kagaku Zasshi, 64, 627 (1961).
Lange's Handbook of Chemistry, Eighth'ed., 1952 (original
source not given).
D. H. Miles and G. M. Wilson, "Vapor-Liquid Equilibrium
Data for Design of Sour Water Strippers," Annual Report
to API for 1974, October 1975.
E. Otsaka, S. Yoshimura, M. Yokabe, S. Inque, Kogyo
Kagaku Zasshi, 63, 1214-1218 (1960).
E. P. Perman, 1. Chern. Soc. London, 83,1168(1903).
ZO
.. MITTELHAUSER
_______ --==_--_---_--_____ ---.
--- -1;'"
----System -- -----
NH3-H20
OOC to 60
0
C
NH3-C02-H20
20
0
C to 40
0
C
---- '--'. __.. _- - --
(Buffered)
80
D
C to 185C
NH., -CO., -H.,O
._._., __ . .. :.t.. .:L ___________ _
700C to 120
0
C
NH3-C02-H20
Phase diagrams
60
0
C to 170
0
C
NH3-H2S-H20
20
0
C to 60
0
C
NH3-C02-H20
NH3-H2S-H20
NH3-C02-H2S-H20
20
0
C to 60
0
C
NH3-H20
114
0
C to 317
0
C
C02-H20
270
0
C to 550
0
C
H2
S
-
H
20
160
0
C to 330
0
C
NH3-H2S-H20
70
0
C to 90
0
C
Reference
Sherwood, Chern., lZ, 745 (1925). (As given in
Perry's Chemi ca 1 Eng i neers I Handbook, Fourth Ed., 1963).
S. Pexton and E. H. M. Badger, i!. Soc. Chem. Ind., 2,
106 (1938).
T. T. C. Shih, B. F. Hrutfiord, K. V. L. N.
Johansen, TAPPI, Technical Assoc. of the Pulp and Paper
Industry, 50, (No. 12) 630-634 (1967)-. - ---
T. Takahashi, Kogyo Kagaku Zasshi, 65, 837-843 (1962).
T.Takahashi, S. Yoshimura, K. Fukii, and E. Otsaka,
Kogyo Kagaku Zasshi, 743-745 (1962).
V. E. Terres, W. Attig, and F. Tscherter, Gas. u.
Wasserfach, 98, 512-516 (1957). - -
D.-W. Van Krevelen, P. J. Hoftijzer, and F. J. Huntjens,
Recuei1 Des Travaux Chimiques Des Pays-Bas, 68, 191-216
(1949). - --
M. E. Jones, i!. Phys. Chem. 67, 1113-1115 (1963).
S. D. Malinin, Geochem. Int., ll, 1060-75 (1975).
T. N. Kozintseva, Geochem. Int., No.4, 750-756 (1964).
V. 1. Oratovskii, A. M. Gamblskii, and N. N. Klimenko,
Appl. Chern. USSR, lZ, 2363-2367 (1964).
21
.. -----------_._-----------------.------ -- .. _---.-
System
Reference
-- --C-----
H
ZS-H2
0
----------H. Gamsjager and P. Schindler, Helv. Chim.
1395-1402 (1969).
25
0
C
r
,
H2S -H20-Sa It
150
0
C to 330
0
C
H2
S
- H20
H2S-CH4-H20
_ 71
0
C to 140
0
C
NH3-C02-H20
----60
o
C to 150
0
C
T. N. Kozintseva, Geokhemiya, 121-134 (1965).
J. L. Vogel, M. S. Thesis, The University of Tulsa, 1971.
G. J. Frohlich, Ph.D. Thesis, Polytechnic Institute of
Brooklyn University, Microfilms No. 60-3497, 1957
More references are given in I. Wichterle, J. Linek, and E. Hala,
Vapor-Liquid Equilibrium Data Bibliography, Elsevier (1973).
22

________ __ ___ -:-- _____ :::::_=_-:-- __ ----------_-_--_-_---_---_______ -:-- __
-( --"4-.2--
\..: --
Until recently there was a serious lack of correlations that pre-
dicted VLE in aqueous systems containing NH3, H2S and C02' Rigorous develop-
ments in the thermodynamics of electrolytic solutions has resulted in improved
approaches. Great interest in coal gasification and oil shale where these
systems must be considered has created incentive to develop improved models.
This section describes VLE models that are currently available.
4.2.1 Van Kreve 1 en
This model is included for historical reasons.(l) Until the develop-
ment of other models described here, the Van Krevelin correlation was the
basis for calculating VLE in sour water systems. Its use is no longer
recommended because of the following deficiencies:
1. Onlyd'ata to 60
0
C were correlated; thus the use of the corre-
lation at sour-water stripper temperatures of 100 to l20
0
C
represented an extrapolation of existing data.
2. The calculation method outlined by Van Krevelen did not allow
for mixtures containing ammonia to hydrogen sulfide ratios
less than 1.5 in the liquid phase.
3. The calculation did not take into account reduced volatilities
of hydrogen sulfide and ammonia at low ppm concentrations due
the ionization constants of the two compounds in water.
4.2.2 API SWEQ
Grant Wilson and co-workers at Brigham Young University developed
a new sour water equilibrium model (SWEQ) during the mid to late 1970's
under a contract sponsored by the API Technical Data Committee.(2) The SWEQ
correlation model is very similar to Van Krevelen's except that some of the
1 imi tat ions have been removed. It is based on new data over wi in
concentration and temperature as well as Van Krevelen's data. _ It considers
23
;1
-.y---- -=-corporation ----
the chemical equilibrium between ionic species of H2S or C02 and undissociated
H2
S
or C02 in the liquid. Van Krevelen assumed that H2S and C02 only exist
in aqueous solutions as ionized species which is virtually true where ammonia
is in excess .
. Both models use Henry's constants to compute NH3, H2S and C02
partial pressures. The SWEQ model improves the correlation of Henry's con-
stants for temperature and composition.
4.2.3
Institute of Gas Technology
In connection with the preparation of a Coal Conversion Systems ______ _
Technical Data Book, _workers at IGT sought to improve the correlation of
partial pressures with concentrations of NH3, C02 and H2S in water.(3) Their
- approach was to analyze data by modification of Van Krevelen's method and to
extend its range by use: of ionization constants, Henry's law correlations
and correlations of activity coefficients as needed. They have extensively
investigated the effect of electrolytes on the solubility or activity of a
gas dissolved in an aqueous solution, commonly called "salting out."
4.2.4 EMNP
Edwards, Maurer, Newman and Prausnitz at the University of California
at Berkeley developed a correlation that calculates equilibrium in the NH3-
H2S-C02-H20 system as well as systems that include sulfur dioxide and hydrogen
This correlation is based on the ionization constants of the
electrolytes, the solubilities of the individual gases in water and salt
solutions, and ionic activity coefficients based on Pitzer's.
The correlation has been incorporated into a computer program called SURFINP.
4.2.5
Beutier and Renon
The procedure developed by Beutier and Renon(6) has the same theoretical
basis as the EMNP correlation. However, it is presently restricted to ternary
systems NH3-C02-H20, NH3-H2S-H20 and NH3-S02-H20. It may be expected that
it is also useful for the complete multisolute system built up with these
compounds.
c
o!
i
corporation __ . __ _
---;. ,------_. _._-_.- --"-- --_._-,---,--- -
4.2.6 ECES
Zemaitis has developed a computer software system called Equilibrium
Composition of Electrolyte Systems (ECES) that is applicable to all multi-
component aqueous electrolyte solutions.(7) Whereas previous VLE models
described here are applicable to. systems containing NH3, C02 and H2S, EeES
was designed to predict vapor-liquid-solid equilibria of multicomponent
aqueous solutions of strong and/or weak electrolytes. The system has been
developed from a basic thermodynamic framework whereas the other models are
based on a Henry's law approach. The thermodynamic framework incorporates
advances made in the area of electrolyte thermodynamics during the early
1970's.
It is not claimed that EeES predicts better results than the other
models described. The other models are directed at predicting equilibria in
specific systems and may be more accurate for those systems. The objective
of EeES is to develop a general purpose system that'is capable of predicting
equilibria in both single and multistage systems over a wide range of conditions.
ECES is available througll software sold byOLI Systems, Inc. The
mode 1 details have not been pub 1 i shed.
25
,;M ITT ELHAUSER
_____ -------------""i
..
c

4.3 PROCESS SIMULATION SOFTWARE

'Process simulation software is available that can predict theoretical
vapor-liquid equilibria and mass transfer in equipment such as venturi scrubbers
and sieve tray towers. While a couple of the simulators have been available
for some time, their capability to handle aqueous electrolytes is recent.
They use the VLE models described in Section 4.2 to perform equilibrium and
mass balances. The best of the process simulators are described below.
4.3.1 ChemShare
ChemShare is a well known computer system used by chemical engineers
to simulate process plant systems. Generally these process systems have
. been hydrocarbon type processes. In 1980 ChemShare incorporated the API
SWEQ model for predicting equilibrium in systems containing NH3, H2S, C02
and H20.
ChemShare is on a royalty use basis.
4.3.2 SST Process
Simulation Sciences Inc. (SST) also provides a process simulation
package. Their current simulator is named PROCESS. In 1980 SSI also in-
corporated the API SWEQ model, giving it much of the same capabilities as
ChemShare in the area of weak electrolyte solutions.
SSI's PROCESS is available on a royalty use basis.
4.3.3 OLI Systems
OlI Systems Inc. offers a process simulation package that is
specifically directed at multicomponent aqueous solutions of electrolytes.
The simulation program is based on the ECES model. Because it is specifically
designed to simulate systems of electrolyte solutions, it has capabilHies
that other simulators do not have.
26

' __ '.___ _ _____ coJPOration ______________ -_" -==_""-_--_--______ ..,
(' OLI's programs are not available on a royalty use basis. The
programs can be purchased or leased. Alternatively, arrangements may be
made with an OLI licensee to have work performed.
27
:-1M ITT ELHAUSER
< corporation --------
-_c_ ---. _______________ ---------=----------.
( 1 )
(2 )
(3 )
(4)
(5 )
(6 )
(7)
REFERENCES FOR SECTION 4
D. W. Van Krevelen, P. J. Hoftijzer and F. J. Huntjens, Recueil Des
Travaux Chimiques Des Pays-Bas, 68, 191-216 (1949).
American Petroleum Institute, "A New Correlation of NH3, C02,
and H2S Volatility Data From Aqueous Sour Water Systems," Publication
955, API, D.C. (March 1978).
Mason, D. M. and R. Kao, "Correlation of Vapor-Liquid Equilibrium
of Aqueous Condensates from Coal Processin9," Thermodynamics of
Aqueous Systems with Industrial Applications, ACS Symposium
Series 133, American Chemical Society, Washington, D.C. (1980).
Edwards, T. J., G. Maurer, J. Newman, and J. M. Prausnitz, "Vapor-
Liquid Equilibria in Multicomponent Aqueous Solutions of Volatile
Weak Electrolytes," A.I.Ch.E. Journal, 24, 966-76 (1978).
Pitzer, K. S., "Thermodynamics of Electrolytes, I," J. Phys. Chern.,
12, 268-277 (1973).
Beutier, D. and H. Renon, "Representation of NH3-
C02-H20 and NH3-S02-H20 Vapor-Liquid Equilibria," Ind. Eng. Chern.
Process Design Dev., ]2, 220-230_ (1978}.
Zemaitis, J. F. Jr., "Predicting Vapor-Liquid-Solid Equilibria in
Multicomponent Aqueous Systems of Electrolytes, II Thermodynamics of
Aqueous Systems with Industrial Applications, ACS Symposium Serico
133, American Chemical Society, Washington, D.C. (1980).
28
i
.,,-_, _____ _-_---_---_--_--_-_____ ---,
5.0 SUPPORTING WORK
Severaljtems of support work are valuable in terms of the
greater understanding that they wi 11 impart toward mass transfer of NH3,
H2Sand C02 in venturi and sieve tray scrubbing systems. This supporting
work is partly engineering in nature and partly laboratory work.
5.1 PRELIMINARY SIMULATION WORK
Using commercial process simulation software such as described
in Section 4.3, the performance of the venturi scrubber or sieve tray
scrubber can be 'predi cted beforehand. It is recommended that fl ash
calculations or absorber calculations be done using one of the software
systems to determine the potential performance with respect to mass
transfer._Ltquid_an_d_vapor rates, compositions and temperatures typical
of planned field scrubbing tests should be used in simulation work in
order to approximate test conditions as closely as possible.
This work would not be expensive. It would indicate the amount
of NH3, H2S, and C02 removal from the gas stream that could be expected.
- -- -
__ It would also indicate typical compositions that would have to be analyzed
during field work. Field test results could be compared against this
simulation work to make preliminary determinations of actual mass transfer
versus predicted. The simulation work should be rerun, however, after
the field tests at the exact field test conditions to accurately compare
actual H2S removal with predicted.
5.2 DETERNINATION OF OIL PHASE IMPORTANCE
Scrubbing of retort off-gas produces a liquid hydrocarbon
phase through condensation of hydrocarbon vapors and collection of liquid
hydrocarqon mist. This hydrocarbon phase presents a complication in the
development of mass transfer models. Theoretically this phase is a
third phase that must be in equilibrium with the other two phases--gas
C and aqueous liquid. Any mass transfer model is going to be complicated
if it has to consider the third phase.
- "--- "-"------ - -
, ,
",,#
) MITTELHAUSER
corporotkm __________________________________________ --.
Just accounting for vapor-liquid equilibrium between three
phases represents a complication. Further, the mechanism of mass transfer
might be complicated. In the presence of two liquid phases, one tends
to be the continuous phase while the other is dispersed.
Review of previous retorting test results and laboratory research
is merited to better understand the importance of a liquid hydrocarbon
phase. Questions that need to be answered are:
How does the volume compare to the phase?
Which phase is continuous?
What is the composition of the hydrocarbon?
Does an emulsion form that would complicate downstream
recovery/rejection of the soluble species?
Hopefully, the hydrocarbon phase is of minor importance. If
so, it could be neglected in formulating mass transfer models. The
importance could be a relative matter. The phase may be so small in
terms of mass or volume that it can be neglected in mass transfer
calculations. However, it might form an emulsion that would make downstream
processing difficult.
5.3 DETERMINATION OF ORGANIC ACIDS AND BASES
Oil shale retorting produces hydrocarbons containing oxygen,
nitrogen and sulfur. These compounds are polar in nature and act as
acids and bases in aqueous solutions. Such compounds absorbed while
scrubbing off-gas can impact mass transfer of NH3, H2S and C02'
The effect of organic acids in the retort off-gas is to increase
ammonia adsorption in the scrubber and reduce ammonia stripping capability
the sour water stripper. The effect on H2S is exactly the opposite.
The concentration of ammonia and organic acids in the stripped process
. condensate increases.
I

30
'". "
; ';, ~ MITTELHAUSER
corporotion _________________ '--___ -,
(
The effect of organic bases in the retort off-gas is to decrease
ammonia absorption and increase ammonia stripping. Hydrogen sulfide
absorption is increased and H2S stripping in the sour water stripper is
reduced. The hydrogen sulfide and organic base concentration in the
stripped process condensate increases.
The combined effect of both organic acids and bases on absorption
and stripping is more difficult to predict. .However, impurity levels in
process condensates are increased.
There are effects that impact the development of mass transfer
models. The presence of compounds present in the scrubbi.ng solutions
resulting from absorption of acids and bases produces a salting out
effect which impacts vapor liquid equilibrium. Also, most of the process
simulation programs described in Section 4.3 do not have the capability
to predict VlE in the presence of these types of compounds. Only the
OlI Systems program has this capability.
Potential organic acids and bases that might be present in
retort off-gas are listed below:
Compound Molecular Weight Boiling Point,OC pK
Phenol 94 196 9.9
Acetic Acid 60 l1S 4.S
Benzoic Acid 122 249 4.2
m-Cresol lOS 203 10.0
Hexanoic Acid 116 205 4.9
p-Toluic Acid 136 275
Propyl Amine 59 49 3.3
Benzyl Amine 107 lS5 4.6
Aniline 93 lS4 9.4
Acetamide 59 222 12.6
Toluidine 107 200 S.9
Diethyl Amine 73 56 2.9
A determination of organic acids and bases in retort off-gas
is required in order to postulate their impact on mass transfer. It is
also relevant to downstream processing. These compounds tend to fix the
31
<
, : A1ITTELHAUSER
'... corporation ______________________ --.,
(0 ..... :
\.
soluble species in solution, thus requiring the addition of caustic or
acid to release the soluble ammonia and hydrogen sulfide. They also
necessitate the use of ion exchange or activated carbon treating t6
remove the organics from the condensate.
32

_______ --==== ___
l
6.0 CONCLUSIONS
Vapor- liquid equilibrium (VLE) models are available that can
adequately predict equilibrium for the NH3-HZS-COZ-HZO system. This
conclusion is predicated on the presence of liquid hydrocarbons and
organic acids or bases not being important factors. Supporting research
can determine the importance of these factors.
The theory exists for developing continuous contact models for
mass transfer in a venturi scrubber. The practical development of such
a model is another matter. There currently is little previous work on
such mOdels for venturis. Further, the system is relatively complex due
to mass transfer of three species--NH3, HZS and COZ--being considered.
Previous work, usually on packed towers or spray scrubbers, most often
is limited to mass transfer of one species. Also, the mathematics of
continuous contact models requires simplifying by means of limiting
assumptions such as linear equilibrium relationships and constant temperature.
These simplifying assumptions aren't true in scrubbing retort off-gas.
As a consequence, the most practical type of modeling is a
stage contact model. This is equilibrium type model and can be
implemented using commercially available process simulation packages.
Actual mass transfer as measured in scrubbing tests is related to
predicted by use of an e!ficiency term.
It is recommended "that mass transfer calculations be performed
using one of the availanle simulation packages. Liquid and vapor rates,
compositions and temperatures corresponding to planned field scrubbing
tests should be used in the simulation work.
33
_________ __ _