I d . Eng. Chem. Process Des. Dev.

1981, 20, 425-433 425

Process Simulation of Ammonia Plant

Chandra P. P. Singh"
Gulf Science d Technology Company, Pittsburgh, Pennsylvania 15230

Deokl N. Saraf
Depadmnt of Chemical Engineering, Indian InstirUte of Technolcgy, Kanpur 2080 16, India

Models of various important process units comprising an ammonia plant, which were validated against extensive
plant data, have been unified in an overall plant simulation. This simulation has been used to check the performance
of two plants, each at two levels of operation, one of which is based on naphthas whereas the other one uses
natural gas as the raw material. The calculated compositions, temperatures, and flow rates at various points along
the process stream, including the recyle around the ammonia synthesis reactor, are in very good agreement with
the corresponding measured values. This shows the generality and accuracy of the process unit models as well
as that of the overall simulation. Effects of errors in the models on calculated rate of ammonia production, pressure,
and concentration of inerts in the ammonia synthesis loop are also illustrated.

Introduction process. The synthesis reaction is expressed stoichiom-

A mathematical description of all the important process trically as
units, taking into consideration all the important physical 0.5Nz + 1.5& = NHB (3)
and chemical processes taking place inside them, is nec-
essary for a successful design and operation of a plant. To evaluate the composition of the process gas a t the
Moreover, when a computer is used for control of a PR inlet, hydrocracking of the higher hydrocarbons in the
chemical process, a mathematical description of the process feed to methane is expressed by the relation
units is necessary to perform such tasks as feed-forward 3CH, + (2 - y)HZO = (1 + y)CH, + (2 - y)CO (4)
and feed-back control, as well as off-line and on-line op-
timization. This paper presents such a mathematical Using appropriate values of the preexponential factors
description of a single-train ammonia plant. Figure 1 and activation energies and considering the effect of
shows a simplified flow diagram of the process. pressure simultaneous with the pore diffusion resistance,
The important processes involved in the production of the rates of reactions in the reformer tubes (reactions 1
ammonia from a hydrocarbon feedstock, such as natural and 2) are expressed as
gas or naphthas, are steam reformation, shift conversion,
and ammonia synthesis reaction. These have been dealt
with in the following sections in the same order. The
processes of removal of COZby absorption and conversion
of CO to methane are of relatively lesser importance and
hence the performance of these units has been assumed
to be ideal.
Steam Reformation
Primary Reformer (PR). The feed to PR (hydro- Heat Transfer. Radiation is the prevailing mode for
carbon-steam mixture) enters the catalyst packed reformer transfer of heat from flue gas and flames to the tube walls.
tubes at the top. Endothermic heat of the reaction as well The total heat transfer is the sum of the transfers from
as heat to increase process gas temperature is supplied by these two sources
the large number of burners distributed uniformly over the 4 = 4- + 9f=t (7)
walls of the reformer. Rate of heat transfer to the process
gas as well as the rates of reactions affects the rate of where qm and qW are the rates of heat transfer to an unit
reformation. The model used in this work (Singh and area on the reformer tube from flue gas and flames, re-
Saraf, 1979a) considers the problem of heat transfer si- spectively
multaneous with the reactions taking place in the catalyst
tubes.
Rate Equations. The hydrocarbons subjected to
steam-reforming range from natural gas to naphthas in the
boiling range up to 220 "C. Higher hydrocarbons in the
feed are assumed to hydrocrack to methane at the inlet
to PR. The following pair of reactions describes the sub-
sequent conversion
CHI + HzO = CO + 3Hz (1)
CO + HzO = COZ + Hz (2)
In addition to the above reactions, the ammonia syn-
thesis reaction is the only important reaction in the overall
0196-4305/81/1120-0425$01.25/0 0 1981 American Chemical Society
420 Ind. Eng. Chem. Process Des. Dev., Vol. 20, No. 3, 1981
diagram for an ammonia plant,
Eg = .Ti; Et = .Tt; Ef= UT:; u is the Stefan-Boltzmann
constant; T,, Tg,and Toare the adiabatic flame, gas, and
outer tube surface temperatures; (?%,)R is the total ex-
change area between flue gas and reformer tubes, and
(Gt)R,black is the total exchange area when tube surfaces
are assumed to be black; nt and nB are the total number
of reformer tubes and burners; tg, tt and tf are emissivities
of gas, tubes and flames;At, AR and Af are the surface areas
of tubes, refractory walls, and assumed spherical surface
of flames;and VF, is the view factor of tubes with respect
to the refractory walls.
The rate of transfer of heat from a unit area of the inner
walls of the reformer tubes to the reacting gas mixture is
obtained from the equation
qin = hin(Tin - T> (12)
where Ti,is the inside tube surface temperature and T is
the temperature of the process gas. The inside heat
transfer coefficient is 0.4 times the value obtained from
Beek's correlation (1962).
Assuming that there is no transfer of heat in the axial
direction at any point inside the reformer, the conduction
of heat through reformer tube is expressed as
2K(To - Tin)
q = (13)
do 1n do/din
and heat transfer to a unit area of the outer surface is
related to that from the inner surface to the gas by
= dinqin/do (14)
where doand dh are outer and inner diameters of reformer
tubes and Toand Ti,are the respective surface tempera-
tures.
It is assumed that the performance of a single reformer
tube is representative of any other tube in the furnace. All
the properties are uniform throughout a general cross
section of the catalyst bed, and axial diffusions of heat and
mass are negligible. With these assumptions the balances
of mass and heat over a differential reformer tube section
of length dz yields the model differential equations.
Pressure drop across the differential section is obtained
from Ergun's correlation (1952). Integration of the model
differential equations along the length of the reformer
yields values of all the variables throughout the reformer.
Adiabatic flame temperature, T,, is evaluated from flow
rate of fuel and oxygen in the flue gas. At each step of
integration, Toand Ti, are evaluated from the iterative
application of eq 7 , 12, 13, and 14. Details of the model
along with the results showing its validity under diverse
conditions of operation are presented elsewhere (Singh and
Saraf, 1979a).
Secondary Reformer (SR).The feed to the secondary
reformer consists of PR exit to which air is mixed. Mixing
of air leads to adiabatic combustion of H2, CO, and CHI
molecules in the gas. The rate as well as equilibrium for
burning (reaction with oxygen) of each of them is different
and therefore for evaluating the exact composition of the
combustion product, all the combustion reactions should
be considered in simultaneous equilibrium (Davis and
Lihou, 1971). However, the rate at which hydrogen burns
is much faster than the other two. Therefore, in this work,
it has been assumed that the extent of combustion of CO
and CH4 is negligible compared to that of H2.Close
agreement was found between the combustion product
temperature calculated on this basis and that calculated
by Davis and Lihou (1971). The amount of air required
is calculated as follows
where P,pH,, PCo,and P c h are the inlet rates of flow
of air,H2,CO, and CH4,respectively, and xaN2and X*O, are
the mole fractions of N2 and O2in air. Amounts of Hz, N2,
and Ar in the combustion product are obtained from the
following component balances
FH2 = PH2- 2x"o;P (16)
FN2 = PdN2 (17)
and
F h = P,,h*~"h (18)
where FH2, FN2, and FA are the rates of flows of H2, N2,
and Ar in the process gas after combustion is complete.
Amounts of the other components in the PR exit remain
unaffected since they do not participate in the reaction
with air.
Temperature of the process gas after combustion and
before it reaches SR catalyst bed is calculated by equating
the s u m of sensible heats in air and PR exit and the heat
from combustion to the sensible heat in combustion
product, by a trial and error procedure.
Rate Equations. As in PR, a nickel catalyst is used in
this reactor also with the only difference that the latter
is capable of standing relatively higher temperatures.
Reactions 1 and 2 describe the conversion. At the SR
temperature conditions, the rate of reaction 1 is so fast that
it could be assumed that methane molecules hydrocrack
instantaneously at the catalyst surface and equilibrium is
attained.
In such situations the reaction is controlled by the rate
at which the reactant diffuses to the catalyst surface from
the bulk gas and the rate of reaction is given by
R1 = Kmam(Cc, - Cac&) (19)
where CcH4and C*chare CHI concentrations in the bulk
gas and at the surface, respectively, K, is the mass transfer
coefficient between bulk gas and solid surface, and a, is
the external surface area per unit mass of the catalyst
pellets. c",, and the catalyst surface temperature, P,are
evaluated from an energy balance on the pellet which can
be expressed as
where C is the total concentration of the process gas. For
the solution of the above correlation, which involves trial
and error calculations, reaction 2 is assumed to be absent.
 Ind. Eng. Chem. Process Des. Dev., Vol. 20, No. 3, I981 427

Otherwise, the rate of reaction 2 is described by eq 6. P u r 9 gas

The assumptions and material, energy and momentum Recycle gas
Mp k mole/ hr

balance equations for this reactor are the same as those
for PR except that no heat is supplied to the process gas
in SR. Integration of the model differential equations,
along the length of SR yields temperature, composition,
and pressure throughout the catalyst bed, in close agree-
ment with the measured values (Singh, 1978).
Shift Conversion
The conversion is expressed by eq 2. It is carried out
in two consecutive adiabatic reactors which use different
catalysts and work under different conditions of temper-
ature. In the first reactor, Le., the high temperature shift
reactor (HT), which uses a chromia promoted iron oxide
catalyst, reaction is carried out ,between 350 and 500 "C.
The low temperature shift reactor (LT) uses a CuO-ZnO
catalyst with various promoting components and operates
in the range of 180-250 O C .
The rate of shift reaction over the shift catalysts (of
FPDIL, India) are given by the general expression
where Rf,A , Pf,Sf,and E f f are factors which account for
the effects of reduction in specific surface area due to
sintering, aging, pressure, H a concentration in process gas,
and diffusional resistances, respectively; u2.and e, are
preexponential factor and intrinsic energy of activation for
reaction 2, and xco and x*co are mole fractions of CO in
the reactor and in equilibrium conditions. Due to low
operating temperature and insignificant presence of HzS,
reduction in rate due to sintering and H2S are negligible
for the LT; i.e., Rf and S, are both equal to 1 for the LT.
Values of u2 and e, and the various factors in the rate
expression 2 are different for HT and LT. Details for
evaluating the various factors numerically along with va-
lidations of the models against extensive plant data are
available in the references (Singh and Saraf, 1977,1980).
Ammonia Synthesis
Equation 3 describes the reaction, and the following
modified Temkin equation (Dyson and Simon, 1968) gives
the intrinsic rate of reaction
B M'k,mOIelhr
C
Figure 2. Recyle scheme of ammonia synthesis.
The outlet from one reactor makes the feed to the sub-
sequent one. Air and steam added at SR and H T inlets,
respectively, are accounted for by mass balance. Tem-
perature changes associated with heat removal in between
any two reactors is not considered; instead the measured
values of temperature after cooling are used in simulation
calculations. The ammonia synthesis reactor involves
recyle of a major part of the product gas. This recyle gas
mixed with makeup synthesis gas, i.e., methanator exit,
is the synthesis reactor feed.
The recyle scheme of the ammonia synthesis reactor is
shown schematically in Figure 2. The flow and compo-
sition of purge, recyle, and feed is obtained from the fol-
lowing component and overall mass balance equations.
The components are numbered for the following equations
as: 1, N,; 2, H,; 3, NH3; 4, CH,; and 5, Ar.
Mass Balance around Synthesis Loop (Figure 2).
Paj" = i@.(xip - ( U ~ ~ . X P-) a3rWr*
j = 1, 2, 4, 5
(24)
Mass Balance around Point A (Figure 2).
P . x j m + iW.x; = Mf-xif j = 1, ..., 5 (25)
Mass Balance around Point D (Figure 2).
Me.x; = (W+ i W ) - x f j = 1, 2, 4,5 (26)
M e . ~ 3=
e (Mp + iW).x3' + MP'* (27)
Component Balances for Various Streams.
5

where kgl is the velocity constant of the reverse reaction Cxi" = 1

j=l
(28)
3; f N 2 , fH2, fm8are the fugacities of nitrogen, hydrogen, and
5
ammonia, respectively, Kq3 is the equilibrium constant of Ex/= 1
reaction 3, and b is a constant. j=l
The velocity constant is expressed as a function of tem-
5
perature in Arrhenius equation form as
CXj" = 1 (30)
1'1
(23) 5
Cxi' = 1
]=1
where u j and egl are the preexponential factor and the
intrinsic energy of activation for reaction 3. and
The effect of mass transfer resistance on the rate of
reaction is accounted for by the effectiveness factor, q, xjP = x; j = 1, ..., 5 (32)
which is evaluated by solving the basic diffusion equation where M is molar flow rate, x j is mole fraction of compo-
for a catalyst of spherical geometry. Details of the model nent j , and superscripts e, f, m, p, prod, and r represent
and its validation against plant data are presented else- the synthesis reactor exit, feed, makeup, purge, product,
where (Singh and Saraf, 197910). and recyle streams, respectively. There are 25 unknowns
Overall Plant Simulation and 22 independent equations (24 through 32). Therefore
There is no recyle of process gas between the point of values of three variables are to be known to obtain the flow
inlet to the PR and the exit from methanator (Figure 1). rates and compositions of all the streams in the synthesis
428 Ind. Eng. Chem. Process Des. Dev., Vol. 20, No. 3, 1981

Table I. Measured and Calculated Data for Naphtha Based Ammonia Plant
process inlet PR case I case I1
number of PR tubes 160 pressure, kg/cm 24.8 24.3
diameter of PR tubes 115.3 mm i.dJ147.7 mm 0.d. temperature, "C 443 440
heated length of PR tubes 11.36 m SR inlet air temperature, "C 231 231
SR catalyst 25 tonsl19.4 m3 H T inlet temperature, "C 377 364
HT catalyst 75 tonsl58.2 m3 L T inlet temperature, "C 202 200
LT catalyst 73 tons/56 m3 synthesis reactor inlet temperature, "C
synthesis catalyst 57.2 tons/20.9 m 3 1st bed 395 389
2nd bed 412 428
process inlet PR case I case I1 3rd bed 406 404
process naphtha flow, kg/h 1 4 100 11900 pressure at synthesis reactor inlet, 228 214
steam flow, kg/h 67200 56000 kg/cm2

Press., t e y ~ , SIG flow,N

location CO CO, H, CH, N, Ar NH, kg/cm2 C ratio m3/h

Case I. Analysis of Process Gas at Different Points in Ammonia Plant

exit PR exptl 9.93 17.43 64.16 8.13 21.90 761 0.927 118 100
calcd 11.21 16.97 64.44 7.38 21.75 770 0.864 118 839
exit SR exptl 13.36 13.06 52.26 0.36 20.63 0.23 21.10 905 0.703 143 100
calcd 13.18 12.74 52.14 0.34 21.34 0.26 21.10 902 0.693 145 980
exit H T exptl 3.21 21.42 55.86 0.32 18.95 0.24 20.50 445 0.530 143 100
calcd 3.08 20.61 56.38 0.31 19.38 0.24 443 0.527 145 980
exit LT exptl 0.31 23.29 57.03 9.32 18.81 0.24 20.00 220 0.480 143 100
calcd 0.33 22.71 57.60 0.30 18.83 0.23 221 0.487 145 980
makeup syn. gas exptl 74.59 0.80 24.32 0.29 72 900
calcd 74.19 0.83 24.68 0.30 74 898
inlet NH, reactor exptl 65.90 6.54 22.03 2.51 3.02 228 248 000
calcd 66.52 6.50 21.78 2.35 2.84 258 805
exit NH, reactor exptl 53.46 7.32 18.12 3.10 18.00 222 216 000
calcd 53.87 7.49 17.50 2.71 18.43 224 741
recycle gas calcd 63.40 8.81 20.60 3.19 4.00 183 907
purge calcd 63.40 8.81 20.60 3.19 4.00 7 052
total amount of ammonia produced, tons/day: plant, 619; calcd, 616
Case 11. Analysis of Process Gas at Different Points in Ammonia Plant
exit PR exptl 12.60 17.20 63.30 6.90 21.60 775 0.821 98 800
calcd 11.32 17.01 64.38 7.29 21.65 772 0.892 98 700
exit SR exptl 14.10 12.80 50.78 0.10 21.90 0.20 21.00 925 0.674 123 900
calcd 13.36 12.42 52.26 0.29 21.42 0.25 20.92 905 0.694 122 037
exit HT exptl 3.10 21.20 55.60 0.10 19.30 0.20 19.90 436 0.550 123 900
calcd 3.11 20.36 56.58 0.25 19.47 0.23 431 0.546 122 037
exit LT exptl 0.30 22.84 57.30 0.28 19.03 0.25 19.30 220 0.510 123 900
calcd 0.32 22.51 57.75 0.24 18.95 0.23 219 0.508 122 037
makeup syn. gas exptl 73.99 0.67 25.01 0.33 63 700
calcd 74.22 0.73 24.75 0.30 62 836
inlet NH, reactor exptl 65.14 6.35 22.32 2.65 3.54 214 226 000
calcd 65.16 6.91 21.81 2.84 3.28 231 393
exit NH, reactor exptl 53.62 6.97 18.33 2.76 18.32 207 197 000
calcd 52.98 7.90 17.77 3.25 18.10 202 353
recycle gas calcd 61.78 9.21 20.72 3.79 4.5 168 557
purge gas calcd 61.78 9.21 20.72 3.79 4.5 4 979
total amount of ammonia produced, tonslday: plant, 543; calcd, 525

loop. The three variables selected for the purpose are
concentration of inerts in the recyle stream, i.e., ( x l + x;),
concentration of ammonia in the recyle stream, i.e., x i , and
the concentration of ammonia at the reactor exit. The
values of the first two variables are taken from the plant
data whereas that of ammonia concentration in the exit
stream is assumed and the flow rate and composition of
the feed stream are calculated. The synthesis reactor
model evaluates the ammonia exit concentration. The
assumed and calculated values are matched by an iterative
procedure to arrive at the final results.
Results and Discussion
Validation of the Simulation. The calculated and
measured values of process and operating variables at
various points in two ammonia plants, each at two levels
of operation, are presented in Tables I and 11. It should
be noted that the measured rates of ammonia production,
in the various cases (bottom lines of the tables), are based
on weight measurements and are independent of the
concentration measurements of ammonia in various
streams of the synthesis loop (Figure 2). Therefore the
rate of ammonia production based on the measured con-
centration of ammonia in the process streams may not
necessarily equal the measured rate of ammonia produc-
tion reported in the bottom lines of Tables I through 111.
Measured flow, temperature, and composition of the
process gas at the exit from PR, SR, HT, LT, and the
synthesis reactor, for all the cases, compare very well with
the corresponding calculated values. For all the four cases,
measured and calculated pressure drops across the PR and
SR are also in very good agreement. Similar agreements
exist for the flow and composition of various streams
around the synthesis loop. The maximum difference be-
tween the calculated and measured rates of ammonia
production are within 4 % . In addition to being small, the
 Ind. Eng. Chem. Process Des. Dev., Vol. 20, No. 3, 1981 429

Table 11. Measured and Calculated Data for Natural Gas Based Ammonia Plant
process inlet PR case I case I1
number of PR tubes 200 temperature, "C 456.4 447.4
diameter of PR tubes 96 mm i.dJ133.8 mm 0.d. air temperature at SR inlet, 460.0 460.0
heated length of PR tubes 12.2 m "C
SR catalyst 14.83 tonsl12.4 m3 steam added at HT inlet, 7200 6000
HT catalyst 53.5 tonsl44.23 m3 kglh
LT catalyst 70.3 tonsI50.2 m3 HT inlet temperature, "C 361 351
synthesis catalyst 52.8 tonsl20.26 m3 LT inlet temperature, "C 198 200
synthesis reactor inlet tem-
process inlet PR case I case I1 perature, "C
process natural gas inlet, N m3/h 13 920 12 014 1st bed 385 406
steam flow, kg/h 61000 55000 2nd bed 433 430
pressure, kg/cmz 35.0 32.3 3rd bed 415 401
pressure at synthesis reactor 235 210
inlet, kglcm
press., t e p p , S/G flow N
location CO CO, H, CH, N, Ar NH, kg/cmz C ratio m3/h

Case I. Analysis of Process Gas at Different Points in Ammonia Plant

exit PR exptl 10.17 12.27 67.22 9.83 31.30 713 0.878 117 100
calcd 9.56 12.56 69.61 8.27 31.56 196 0.898 116 423
exit SR exptl 12.20 9.40 55.58 0.32 22.11 0.38 30.20 962 0.694 144 300
calcd 11.96 9.44 55.73 0.34 22.26 0.21 30.40 957 0.111 144 529
exit HT exptl 2.16 17.62 59.40 0.23 20.12 0.23 29.50 425 0.649 151 600
calcd 1.83 17.52 59.83 0.31 20.31 0.24 424 0.653 153 489
exit LT exptl 0.18 18.93 60.29 0.34 20.01 0.25 28.10 213 0.630 151 600
calcd 0.16 18.87 60.48 0.31 20.03 0.23 208 0.626 153 489
makeup syn. gas exptl 74.55 0.75 24.37 0.32 78 100
calcd 14.39 0.57 24.14 0.30 76 119
inlet NH, reactor exptl 67.03 5.46 22.19 2.56 2.76 235 242 160
calcd 67.25 5.65 21.80 2.95 2.35 244 898
exit NH, reactor exptl 53.90 6.30 17.80 2.90 19.10 228 216 000
calcd 53.38 6.59 17.08 3.44 19.51 209 509
recycle gas calcd 64.31 7.90 20.44 4.10 3.25 168 780
purge gas calcd 64.31 7.90 20.44 4.10 3.25 5 518
total amount of ammonia produced, tonslday: plant, 618; calcd, 641
Case 11. Analysis of Process Gas at Different Points in Ammonia Plant
exit PR exptl 9.86 11.79 69.03 9.32 28.80 193 0.910 103 500
calcd 9.90 12.01 69.59 8.90 29.12 184 0.931 101 098
exit SR exptl 12.40 9.60 55.37 0.33 22.10 0.40 28.00 932 0.135 125 800
calcd 12.52 9.26 55.43 0.33 22.25 0.36 28.30 921 0.148 121 553
exit HT exptl 1.18 18.05 60.02 0.30 20.09 0.36 27.50 415 0.670 134 000
calcd 1.23 18.13 59.80 0.30 20.22 0.32 414 0.667 135 019
exit LT exptl 0.13 18.00 61.60 0.25 19.63 0.39 27.00 209 0.650 134 000
calcd 0.13 19.02 60.24 0.29 19.99 0.31 207 0.649 135 019
makeup syn. gas exptl 74.37 0.49 24.72 0.42
calcd 14.32 0.52 24.18 0.38 65 383
inlet NH, reactor exptl 64.85 5.48 21.54 5.16 2.91 216 228 300
calcd 65.54 5.58 21.91 4.01 2.89 236 198
exit NH, reactor exptl 52.61 6.23 18.21 4.63 18.32 210 204 000
calcd 53.00 6.40 17.76 4.68 18.16 205 619
recycle gas calcd 62.18 7.51 20.82 5.49 4.0 170 815
purge gas calcd 62.18 7.51 20.82 5.49 4.0 4 521
total amount of ammonia produced, tons/day: plant, 548; calcd, 553

Table 111. Compositions and Flow Rates to the Ammonia tion in Table I1 Case I, shows a higher calculated value at
Synthesis Reactor Based on Measured Flow and the HT exit, a lower value at the PR exit and in the
Composition of Makeup Gas make-up gas and almost equals the measured values at all
process gas stream
flow N
other points. As another example, the calculated methane
H, CH, N, Ar NH, m3/h concentration at the PR exit is observed to be less than
the corresponding measured values in three cases (Table
makeup exptl 74.59 0.80 24.32 0.29 72 900 I, Case I and Table 11, Cases I and 111),whereas in the
synthesis remaining case (Table I, Case 11) the reverse is true.
gas Similar observations can be made for concentrations of
inlet NH, exptl 65.90 6.54 22.03 2.51 3.02 248 000
reactor calcd 69.48 6.59 18.71 2.38 2.87 259603 other components and temperatures as well. Thus it can
outlet NH, exptl 53.46 7.32 18.12 3.10 18.00 216 000 be claimed that the overall simulation represents the plant
reactor calcd 57.64 7.52 14.11 2.73 18.00 226332 very well.
total ammonia produced tonslday: plant, 619; calculated, The above discussed results may be expected in view of
601 the fact that models for each of the process units, namely,
PR, SR, HT, LT, and ammonia synthesis reactor, have
differences between measured and calculated values are independently been validated against extensive plant data.
randomly distributed. For example, methane concentra- The ranges for the variables used in the present study are
430 Ind. Eng. Chem. Process Des. Dev., Vol. 20,No. 3, 1981
rather limited, compared to those used in the validations
of unit models (Singh and Saraf, l977,1979a,b, 1980). (It
is worthwhile to mention here that this design, i.e., single
train, of an ammonia plant is self-sufficient in steam and
operating the plant a t relatively lower capacities are im-
practical due to additional steam requirements). However,
there are some limitations to the applicability of the unit
models, based on once-through data, to the recyle systems.
For example, Table 111presents the calculated composition
and flow rates of process gas at the inlet and outlet of the
synthesis reactor based on the measured composition and
flow rate of makeup synthesis gas in Table I, Case I. A
comparison between measured and calculated composition
in Table I11 and that in Table I, Case I clearly shows that
the difference between measured and calculated values
based on the measured makeup gas composition (Table
111)are much higher than the corresponding ones for the
overall simulation (Table I, Case I. In the latter case
(Table I, Case I) the differences are much smaller at
various points in the recyle loop (Figure 2), whereas in the
former case (Table 111)the difference between measured
and calculated nitrogen concentration, at the inlet to and
exit from the synthesis reactor, is around 20%. The dif-
ferences between measured and calculated hydrogen con-
centrations are also much higher in this case (Table 111).
It is important to note here that the differences between
the measured and calculated values of makeup gas rates
and compositions are very small (Table I, Case I) whereas
the corresponding calculated values for the recyle system
show large differences (Table 111). This shows how the
magnitude of an error in the once-through stream is am-
plified in the recyle system which represents the overall
performance of the plant. The above example also shows
the advantage of using the overall process simulation
starting with the input to the plant.
Sensitivity Analysis. In a single train ammonia plant,
the effect of the performance of any of the process units
on the plant performance is not limited to that of the
composition, temperature, and pressure of the exit stream.
For example, the PR can be operated at different
throughputs, and the exit composition, temperature, and
pressure can be maintained constant. However, at a lower
throughput the amount of flue gas generated in the PR
may not be sufficient to obtain enough high-pressuresteam
to run the compressors at the required capacity. This
would make the operation of the plant impractical. Sim-
ilarly, other changes affect the plant performance in var-
ious complex ways. To account for all the effects of change
in the values of a particular variable is outside the scope
of this work, since the models for heat recovery, steam
generation, compressor performance, etc. are not included.
However, this process simulation can be used to illustrate
the effects of poor performance of the process units or the
effects of poor representation of the process units by the
corresponding models.
The effect of poor performance of the reactors upstream
would be reflected in the makeup synthesis gas as reduc-
tion in the rate of generation, higher inerts (CHI + Ar)
concentration and/or H2/N2 ratio different from 3. Con-
versely, a change in composition or flow rate of the syn-
thesis gas reflects on the performance of the process units
upstream. Therefore, the composition and flow rate of
makeup synthesis gas on the performance of the plant has
been selected for study.
Makeup Gas Rate. Figure 3 shows the rate of pro-
duction of ammonia and the flow rate of process gas at the
inlet to the ammonia synthesis reactor, both as functions
of rate of makeup gas flow. The levels of purge rate (as
EE - -1.8
UI
m
L
s
a 0
0.
300- - 1.6
1 1 1 I 1 1 A
a fraction of makeup gas), concentration of ammonia in
the purge or recyle stream, and average pressure in the
system are maintained constant. The H2/N2ratio is 3 and
inlet temperatures to the catalyst beds are the same as that
in Table I, Case I. As shown, the rate of ammonia pro-
duction decreases linearly with the decrease in the rate of
production of the makeup synthesis gas. The rate of flow
of the process gas to the reactor increases with increase
in the makeup gas rate. The rate of increase in process
gas flow, with increase in makeup gas flow, is higher at
higher values of the latter due to decrease in the concen-
tration of ammonia at the synthesis reactor exit. The rate
of production of ammonia increases since the increase in
throughput more than compensates for the decrease in
ammonia concentration in the exit stream.
Inerts (CHI + Ar) in the Makeup Gas. The change
in the rate of ammonia production subject to change in the
concentration of inerts (CHI + Ar) in the makeup gas, and
the corresponding variation in the process gas flow rate,
is shown in Figure 4. The level of inerts in the recyle
stream and pressure is held constant. To maintain con-
stant level of inerts in the recyle gas, the purge rate in-
creases with increase in inerts concentration of the makeup
gas and the production of ammonia decreases. The de-
crease in the rate of ammonia production is linear since
the amount of gas lost in the purge is directly proportional
to the inerts concentration in the makeup gas. The in-
crease in the rate of process gas inlet to the synthesis
reactor is slightly higher at lower inerts Concentration in
the makeup gas. The increase in rate of process gas feed
at lower inerts concentration in makeup gas results from
the counteracting effects of decrease in the purge rate and
increase in the equilibrium concentration of ammonia. The
results in Figure 4 show that the former has a more sig-
nificant effect on the flow rate.
Figure 5 shows the effect of change in the concentration
of inerta in the makeup synthesis gas when the recyle and
purge rates are held constant. For a constant purge rate,
the rate of ammonia production remains constant. How-
ever, the level of inerts in the recyle stream and the average
pressure in the reactor increase. The former increases
 Ind. Eng. Chem. Process Des. Dev., Vol. 20, No. 3, 1981 431

1 0.5 09 1.3

lnerts in makeup gas, % - 1.r 2.1

Figure 4. Effect of inerts in makeup gas on ammonia production

1801
0.1
1
0.5
Inerts in
1
0.9
1
1.3
makeup g a s , % -I
1.7 2.1
I

Figure 5. Effect of inerts in makeup gas on synthesis reactor

pressure and inerts in recyle gas.
and synthesis reactor throughput.

linearly with the inerts concentration in makeup gas

whereas the rate of increase in pressure is higher at higher
inerts concentration.
H2/N2Ratio in the Makeup Gas. The effect of Hz/Nz
ratio in the makeup gas on average pressure in the syn-
thesis reactor is shown in Figure 6. For a constant rate
of makeup gas production, and purge rate, the operating
pressure required to achieve the conversion shows a min-
imum when the H2/N2ratio equals 3. Any change in the
Hz/Nz ratio from 3 increases the average pressure in the
reactor and the rate of increase in pressure also increases
with increase in the difference of H2/N2ratio from 3. The
corresponding change in the concentrations of H2 and Nz
and H2/Nz ratio in the ammonia reactor exit is shown in
Figure 7. As the H2/N2ratio in the makeup synthesis gas
changes from 2.8 to 3.2, the ratio of Hz/Nz in the reactor
exit changes from 1.05 to 20.2. The corresponding change
in mole fractions of hydrogen and nitrogen are from 37.35
to 3.36. It is important to note from the results in Figures
6 and 7 that at H2/N2ratio lower than 3 the decrease in
hydrogen mole fraction has a more significant effect on the
rate of synthesis reaction compared to the effect of increase
in the mole fraction of nitrogen on the same, whereas for
H2/N2 ratio greater than 3 the effect of decrease in ni-
trogen concentration is more important. Thus, the rate
of reaction at a given pressure decreases and the pressure
required to achieve the same extent of conversion increases I
2.8
I
2.9
I
3.0
1
3 .l 2
with any change in H2/N2 ratio from 3. Also, a small H 2 / N 2 in makeup gas __c

change in the H2/N2ratio in the makeup gas results in a
many times greater change in the H2/N2ratio in the re-
actor and all the streams in the recyle loop.
It is clear from the results in Figures 3 through 7 that
the performance of an ammonia plant is very sensitive to
Figure 6. Effect of Hz/N2 ratio on synthesis reactor pressure.
the rate of makeup gas production, and the concentration
of inerts and H2/N2 ratio in the makeup synthesis gas.
Low rate of makeup synthesis gas production, higher inerts
432 Ind. Eng. Chem. Process Des. Dev., Vol. 20, No. 3, 1981

subject to changes in any of the input conditions, and

meaningful optimization of the plant performance can be
obtained.
Nomenclature
A = area, m2
A , = aging factor
a = preexponential factor
a,,, = external surface area per unit mass of catalyst, m2/kg
of catalyst
b = a constant
C = concentration, kg-mol/N m3
C, = average specific heat capacity, kcal/kg K
C,, = specific heat capacity of flue gas, kcal/k%K
C, = concentration of methane, kg-mol/N m
d = diameter, m
E = emissive power, kcal/m3
e = energy of activation, kcal/kg-mol K
e, = apparent energy of activation, kcal/kg-mol K
F = volume flow rate, N m3/h
f = fugacity of component j
ht= heat of ith reaction, kcal/kg-mol
h = heat transfer coefficient, kcal/m2 K h
01 I I I lo K = conductivity, kcal/m2 K h
2.0 2.9 3.0 3.1 3.2
H2/N2 in makeup 98s d Kq = equilibrium constant
k = rate constant
Figure 7. Effect of H2/N2ratio in makeup gas on the H2/N2 ratio It, = mass transfer coefficient, m/h
in the ammonia synthesis loop. M = molar flow rate, kg-mol/h
nB = number of burners
in the makeup gas, or improper H2/N2ratio decrease the n, = number of tubes
rate of ammonia production. The decrease ie the rate of P = pressure, atm
ammonia production can be reduced and in many cases Pf = pressure factor
q = rate of heat transfer per unit area on the outer surface
the negative effects can be compensated by an increase in of reformer tubes, kcal/m2 h
the operating pressure. However, the required increase qftt, q+ = rates of heat transfer to a unit area on reformer
in pressure is limited by the capacity of the compressors tube from flue gas and flames, kcal/m2 h
and availability of steam for running the compressors. In R = rate of reaction, kg-mol/kg of cat. h
general, low makeup gas rate, higher inerts concentration, Rf = reduction factor
or improper H2/N2 ratio would result in reduced pro- R, = universal gas constant
duction of ammonia. Sf = sulfur factor
In view of the sensitivity of the rate of ammonia pro- T = temperature, K
duction to the changes in the makeup synthesis gas pro- VF = view factor
duction, concentration of inerts (CHI + Ar) and H2/N2 3c = mole fraction
a = stoichiometric coefficient
ratio in the makeup gas and the fact that each of the e = emissivity
process units contribute to the rate of generation and
composition of the makeup gas shows that the unit models Subscripts
are very accurate in predicting the performance of the f = flames
process units, and the unified simulation is capable of g = gas
predicting the plant performance. However, to have a i = reaction i
complete picture of the complex interactions between the in = inside wall of reformer tubes
various units and their effects on the overall performance j = component j
o = outside wall of reformer tubes
of an ammonia plant, any futqre work should include R = refractory wall
modules for the recovery of heat at various locations and t = reformer tube
generation of steam from various hot streams and com-
pressors. Superscripts
a = air
Conclusions e = exit synthesis reactor
A process simulation of hydrocarbon based single train f = feed to the synthesis reactor
m = makeup synthesis gas
ammonia plant has been obtained using mathematical o = inlet condition
models for the important process units and considering the p = purge gas
performance of other units to be ideal. The close agree- prod = product stream
ment between measured and calculated values confirms r = recyle gas in the ammonia synthesis loop
the accuracy of the process unit models and the overall s = catalyst surface
simulation in predicting the plant performance. Analysis ' = reverse reaction
based on the simulation shows that the latter is very * = equilibrium condition
sensitive to the flow rate and composition of the makeup
synthesis gas. Literature Cited
Future work on the ammonia plant simulation should Beak, J. Adv. Chem. Eng. 1862. 3, 234.
include models for all the process units as well as those Davis, J.; Lihou, D. A. Chem. Process Eng. 1971. 52, 71.
Dyson, D. C.; Simon, J. M. I d . Eng. Chem. Fundam. 1968, 7, 605.
for heat exchangers and compressors. Such a simulation Ergun, S. Chem. Eng. Prog. 1852. 48, 89.
can be used to study the overall response of the plant, Singh, C. P. P. Ph.D. Thesis, I..T, Kanpur, India, 1978.
 Ind. Eng. Chem. Process Des. Dev. 1981, 20,433-435 433

Singh, C. P. P.; Saraf, D. N. Ind. Eng. Chem. Process Des. Dev. 1977, 16, Singh, C. P. P.; Saraf, D. N. Ind. Eng. Chem. Process Des. Dev. 1980, 10,
313. 393.
Singh, C. P. P.; Saraf, D. N. I d . Eng. Chem. Process Des. Dev. 1979a,
18, 1.
Singh, C. P. P.; Waf, D. N. Ind. Eng. Chem. Process Des. Dev. 1979b, Received f o r review October 12, 1979
18, 364. Accepted March 16,1981

Extraction of Acetic Acid from Dilute Aqueous Solutions with

Trioctylphosphine Oxide

Janvlt Golob," Vlktor Grllc, and Bojan Zadnlk

&Pam" of Chemistry and Chemical Technology, Edvard Kardeu Unlversky, 61000 Ljubljana, Yugoslavia

Extraction characteristics of trioctylphosphine oxide (TOPO) dissolved in commercial kerosene for recovery of acetic
acid from dilute water solutions were studied. Some basic thermodynamic properties such as viscosity, density,
interfacial tension, and primary settling time of the extraction system were measured. Extraction isotherms at 20
OC were obtained for 3, 5 , and 18 wt % of TOPO in kerbsene. Finally, a simulation of countercurrent extraction
of water solution, containing 5 wt % acetic acid, and regeneration of the solvent was performed on laboratory
scale equipment.

Introduction Table I. Properties of TOPO-Kerosene Solutions at 20 "C

The present work deals with the recovery of acetic acid TOPO concn in kerosene, wt 5%
from dilute aqueous solutions. Removal of acetic acid from
waste waters, concentration ranging from 0.5 to 5 wt 90, properties 0 3 5 18
is of increasingly interest due to economic and environ- p0, kg/m3 786.0 787.5 790.0 804.5
mental reasons (Helsel, 1977). For extraction of acetic acid qo, mPas 1.18 1.27 1.34 1.94
Y, mN/m 33.79 24.76 12.15 10.13
on an industrial scale, solvents such as ethyl acetate or
diethyl ketone have been used (Brown, 1963; Ricker et al., t, s 15 25 30 40
1980). Recovery of acetic acid with such a conventional extraction of other components such as formic acid, form-
solvent is not economically attractive as long as the feed aldehyde, methanol, etc., present in acetic acid waste
contains less than 3 wt ?& of acetic acid. Therefore, in- streams. Their presence demands additional separation
vestigations aiming to find solvents that would provide steps for their isolation (Helsel, 1977; Wardell and King,
higher values of distribution coefficients were initiated 1978; Ricker et al., 1980).
(Helsel, 1977; Wardell and King, 1978; Ricker et al., 1979). The aim of the present work is to examine the usefulness
Application of trioctylphosphine oxide (TOPO) as an ef- of TOPO/kerosene solvent. Thermodynamic data of the
fective solvent for extraction of acetic acid from dilute extraction system, such as the solubility of TOPO, dis-
solutions is discussed in a limited number of papers tribution coefficients, viscosity, density, interfacial tension,
(Helsel, 1977; Kohn, 1978; Wardell and King, 1978; Ricker and primary settling time are needed. Simulation of the
et al., 1979). The distribution coefficient varied between extraction/regeneration process on a laboratory scale
3 and 5. In recent years the use of other organophosphorus would also provide useful insight into operation require-
compounds and high molecular weight alkylamines dis- ments.
solved in various diluents (not kerosene) was explored for
the acetic acid (Ricker et al., 1979). Ricker et al. (1979) Extraction Characteristics bf TOPO-in-Kerosene
put forward a comparison between TOPO and amine ex- and Acetic Acid-in-Water Solutions
tractants. Due to the ten times higher price of TOPO The thermodynamic data for extraction of acetic acid
compared with conventional amine solvents (Alamine 336, from dilute aqueous solutions are available for various
Adogen 364), it was concluded that any significant loss of TOPO diluents excluding kerosene and do not cover all
TOPO would have a pronounced negative effect on the relevant property data important for the design of the
economics of an acetic acid recovery process. Another apparatus for extraction and phase separation. Therefore,
disadvantage of TOPO is its very high boiling temperature it was one of the aims of this work to determine experi-
which involves difficult recovery from nonvolatile sub- mentally some most important properties of TOPO-in-
stances (tar) by distillation (Ricker et al., 1980). An ad- kerosene solutions.
vantage introduced by the TOPO/diluent system is that The TOPO used in this study was of technical grade
there is no need for additional thermal or chemical (95% purity), manufactured by Fluka AG, Switzerland.
treatment of the water phase after extraction and coa- Kerosene was commercial grade jet fuel supplied by Petrol
lescence, providing that insoluble diluent is used. Ricker Oil Co., with composition as follows: 16.7 wt % aromatica,
could not detect traces of TOPO in the water phase after 40.8 wt % C8-CI4 n-parafins, and 42.5 wt % naphthenes.
containing 30 wt ?& TOPO in Chevron Solvent No. 25 the solubility tests yielded 21 wt % saturated solution of
(Ricker et al., 1979). Some authors also considered co- TOPO in kerosene at 20 "C. Other physical properties
0196-4305/81/1120-0433$0l.25/0 0 1981 American Chemical Society