and entropy of wet gases (even if not saturated) could be cal- (1965).

Prausnltz, J. M., “Molecular Thermodynamics of Fluid-Phase Equilibria,”

culated at higher temperatures than those considered here,
using eq 22 and 23.
Literature Cited
Bain, R. W., “NEL Steam Tables 1964,” Her Majesty’s Stationery Office, Edin-
burgh, 1964.
de Santis, R., Breedveld, G. J. F., Prausnitz, J. M., lnd. Eng. Chem., Process Des.
Dev., 13, 374 (1974).
Edwards, T. J., Newman, J., Prausnitz, J. M.,A.6Ch.E. J., 21, 248 (1975).
Eveiein, K. A., Moore, R. G., Heidemann, R. A., Ind. Eng. Chem., Process Des.
Dev., 15, 423 (1976).
Heidernann, R. A.. A.l.Ch.€. J., 20, 847 (1974).
Hilsenrath, J., Beckett, C. W., Benedict, W. S.,Fano, L.. Hoge, H. J., Masi, J. F.,
Nuttall. R. L., Touloukian, Y. S.,Wooiiey, H. W., “Tables of Thermodynamic
and Transport Properties,” Pergamon Press, New York, N.Y., 1960.
Himmelblau, D. M., J. Chem. Eng. Data, 5, 10 (1960).
Kelley, K. K.. US.Bur. Mlnes Bull. 584 (1960).
Luks, K. D., Fitzgibbon, P. D., Banchero, J. T., lnd. Eng. Chem., Process Des.
Dev., 15,326 (1976).
Lyckman. E. W., Eckert, C. A., Prausnitz, J. M., Chem. Eng. Sci., 20, 685
Prentice-Hail, Inc., Engiewood Cliffs. N.J., 1969.
Rigby, M., Prausnitz, J. M., J. Phys. Chem., 72, 330 (1968).
Saddington, A. W., Krase, N. W., J. Am. Chem. Soc., 56, 353 (1934).
StinSon, D. L., Carpenter, H. C., Cegieiski, J. M., J. Pet. Techno/., 645 (June,
1976).
Takenouchi, S., Kennedy, G. C., Am. J. Sci., 262, 1055 (1964).
TCjdheide, K., Franck, E. U., Z. Phys. Chem. (Frankfurt am Main), 37, 387
(1963).
Webster. T. J., Discuss. Faraday SOC., 15, 243 (1953).
Zimmerman, F. J., Chem. Eng., 65, 117 (Aug 25, 1958).
Received for review October 11,1976
Accepted January 18,1977
T h e authors are grateful for financial support given by Zimpro Inc.
o f Rothschild, Wisconsin, Associated Pulp and Paper, Ltd. of De-
vonport, Tasmania, the National Research Council o f Canada, and
the National Science Foundation.

Optimal Temperatures for Ammonia Synthesis Converters

Larry D. Gaines
Systems Research Division, Applied Automation, lnc., Bartlesville, Oklahoma 74004

A steady-state model for a quench-type ammonia converter is developed and used to study the effect of convert-
er temperature profile over a large range of operating conditions. A substantial improvement in converter effi-
ciency is possible by maintaining optimal converter temperatures. The converter efficiency is shown to be pri-
marily a function of fourth bed temperature. The optimal value of this temperature is a function of operating con-
ditions and may be determined from the ammonia effluent concentration and equilibrium concentration. Based
upon these results, a simple method of converter temperature control is proposed.

Introduction opments for quench-type converters. Shah (1967) proposed

Ammonia synthesis, one of the oldest commercial high-
pressure processes, consists of reacting a hydrogen-nitrogen
mixture over a catalyst a t elevated temperatures and pres-
sures. Pressures employed in the reactor range from 120 to 600
atm. The reaction is exothermic and the high temperature in
the reactor is sustained by the heat of reaction through the use
of a feed-effluent interchanger. This process has been the
subject of numerous studies and with the advent of computers,
mathematical models have proven to be an accurate and
convenient way to study ammonia synthesis converters over
their entire operating region. Computers have been widely
accepted in the area of ammonia plant control because of their
abilities to increase operating efficiency and reduce process
upsets. As computer control is extended to all parts of the
process, a basic understanding of process characteristics is
needed.
Several studies (VanHeerdon, 1953; Hay et al., 1963; Brian,
1965; Shah, 1967; Baddour, 1965) have considered the effects
of various pertinent variables upon ammonia production and
converter stability. Annable (1952) fitted data from industrial
reactors using the Temkin-Pyzhev rate expression and cal-
culated ideal temperature and composition profiles. Baddour
et al. (1965) modeled a TVA converter and compared model
results with plant data. They found that the converter was
close to instability a t temperatures favoring maximum am-
monia production. They also found that these temperatures
were a function of process conditions.
Both Shah (1967) and Kjaer (1963) present model devel-
that his theoretical model be used for off-line optimization to
determine quench flows and converter temperatures. Kubec
et al. (1974) developed a model for a radial flow quench-type
converter. They suggest the use of a model which would be
optimized using dynamic programming. Model parameters
would be adjusted using current process data.
The objectives of this study are to develop a model of a
synthesis converter, determine the effects of process variables
upon converter operation, and use the resulting data to ex-
amine practical and easily implemented methods of converter
temperature control. The ammonia plant chosen for this study
is a low-pressure (150 atm) 600 T P D Kellogg plant which uses
two quench-type converters in parallel. This model is used to
determine the “best” converter temperature profile over a
wide range of operating conditions. Process variables that
affect converter efficiency are pressure, catalyst activity, space
velocity, inerts, Hz/Nz ratio, feed temperature, and ammonia
concentration in the converter feed; these variables and their
impact on converter efficiency are amply investigated via
model simulation.
Ammonia Synthesis Converter
The converter feed gas is split into five fractions. The bulk
of the gas enters the top of the converter and flows downward
around the outside of the catalyst section to the exchanger at
the bottom of the pressure shell as shown in Figure 1. The gas
is heated in the exchanger and passes through the riser tube
to the first quench zone where it is mixed with cold feed gas

Ind. Eng. Chem., Process Des. Dev., Vol. 16, No. 3, 1977 381
 Q'JENCI

QUENCH d

QUENCY YLI

Figure 1. Synthesis converter.

I

P
Temperature

Figure 2. Heat produced or consumed.

before entering the first bed. Ammonia is formed according

to eq 1 as the gas flows through the bed.

The reaction is exothermic and the heat released increases the

energy of the reaction mixture. The first bed is followed by
three more quench zones and beds, each containing a larger
quantity of catalyst. The hot gas out of the fourth bed is cooled
on the shell side of the exchanger before leaving the converter.
This transfer of heat is necessary to bring the cold feed gas to
reaction temperature.
Figure 2 shows heat produced by reaction as a function of
converter temperature. The equilibrium concentration of
ammonia decreases as reaction temperature increases and the
heat production curve passes through a maximum as the
temperature is increased. The heat consumed in raising the
cold feed gas to reaction temperature is also shown in Figure

382 Ind. Eng. Chem., Process Des. Dev., Vol. 16, No. 3, 1977
H2
Nz
NH3
Ar
CH4
MF

.
TF

1
2
3
4
5
1.3445
2.3930
1.2907
2.2769
i
i

0.02096
0.05046
0.3415
Bed I1

w2

Quench Zone

113

Bed U 2

H4

Quench Zone

n5 75

Bed d 3

'6

Quench Zone
T6

Bed d 4

I
Figure 3. Block diagram of synthesis converter.
12

13

74

Table I. Constants for Calculation of Activitv Coefficients

i Ai
0.1975
Bi cix 10-4

2 for three different total quench flows and the intersections

of the heat production and heat consumption lines represent
operating points. The normal operating point 0 is the only
Tp

0.0504
0.420
476.87

point of operation considered in this study and the significance

of the two lower points is discussed elsewhere (Brian et al.,
1965; Slack, 1953; Stephens, 1975).
Maximum converter efficiency is obtained when the heat
consumption line passes through the maximum in the heat
production curve, point A. When the two lines become tan-
gent, as shown at point B, converter operation becomes un-
stable. This point is referred to as the "blow-out" point. Both
heat consumption and heat production curves are functions
of operating conditions. To obtain maximum converter effi-
ciency, it is necessary to continuously operate at point A. Since
this point changes with operating conditions, a control system
is needed which is capable of tracking A while ensuring that
the converter does not become unstable.
Model Development
The converter catalyst section consists of an outer pressure
shell and a catalyst basket with cold feed gas flowing in the
annulus between the two (Figure 1).The catalyst basket is
coated with insulation to minimize heat transfer from the hot
catalyst to the cold feed gas. An uninsulated riser tube
transports gas from the heat exchanger section through the
catalyst section to the first bed. Minimal heat transfer to the
riser gas is due to a small heat transfer area and because the
Table 11. Constants for Calculation of Effectiveness Factor
Pressure,
atm bo bi bz bg X lo4 b4 bg X 108 bs
150.0 -17.5391 0.07698 6.90055 -1.08279 -26.4247 4.92765 38.9373
225.0 -8.2126 0.03774 6.19011 -0.53547 -20.8696 2.37914 27.8840
300.0 -4.6757 0.02355 4.68735 0.346331 11.2803 1.54088 10.4663

riser gas has been heated to near reaction temperature. The
catalyst section is nearly adiabatic and an assumption of
uniform radial temperature is made. A uniform velocity profile
in the bed may also be assumed since the bed diameter is much
larger than the catalyst diameter. The assumptions of uniform
radial temperature and velocity reduce the equations to an
easily usable form while retaining the desired accuracy of the
model.
Material Balance. The hydrogen consumed in a differ-
ential element of a catalyst bed is given by
F1 dx = rA dz
where A is the cross-sectional area of the bed in ft2 and x is the
fraction H2 converted. F1 is the flow rate of Hz into the bed
in lb-mol/h. Units are temperature, O R , pressure, atm, and
flow, lb-mol/h unless otherwise stated. Figure 3 is a simplified
diagram of the synthesis converter which shows flows and
temperatures used in the following equations.
The component flows at any point in the bed may be com-
puted from inlet flows by
F1' = Fi(1 - X )
F2' = F2 - F1X/3
F3' = F3 + 2FlX/3
F4' = F4
F5' = F5
where the order of components is hydrogen, nitrogen, am-
monia, argon, and methane. r, the rate of reaction for equation
1 is found from the Temkin (1950) equation
of 0.5 was used in this study to obtain the reverse reaction rate
constant k 2 as outlined by Dyson and Simon (1968), but using
activities as discussed in this paper. Data from Nielsen (1968)
for an industrial catalyst were fitted and the resulting equa-
tion
k2 = 0.146578 x 1015e(-80899/RT) (8)
has units of lb-mol/h f t 3 and T is in OR. R is the universal gas
constant and has a value of 1.987 Btu/lb-mol O R .
The effectiveness factor is derived by Dyson and Simon
(1968) for isothermal spherical catalyst particles. They as-
(2)
sumed the diffusion coefficients were independent of position
in the particle and that Knudsen diffusion does not occur. The
resulting equation
[ = bo + blT + b27 + b3T2 + b472 + b5T3 t be73 (9)
gives the effectiveness factor as a function of 8, temperature
in K, and pressure for a 3 to 1 H2/N2 ratio and 12.7% inerts.
Constants for use with eq 9 are given by Dyson and Simon
(1968) and appear in Table I1 for pressures of 150, 225, and
300 atm. The fractional conversion of nitrogen is 7 and may
be obtained from
7= molar flow of NH~/(molarflow of NH3
+
2 molar flow of Nz) (10)
Catalyst activity, v , deteriorates a t high temperatures due to
changes in catalyst structure. The catalyst is also poisoned by
(3)
oxygen and sulfur compounds. Factors affecting the rate of
deterioration are thoroughly discussed by Nielsen (1968).
Energy Balance. The converter may be divided into three
major sections: catalyst beds, quench zones, and heat ex-

[
r = 3k2t K ,2 a2 (37 v (4)
where K , is the equilibrium constant, k 2 is the rate constant
of the reverse reaction, v is the catalyst activity, is the ef-
fectiveness factor, and al,a2, and a3 are the activities of hy-
drogen, nitrogen, and ammonia respectively. K , is calculated
from the equation of Gillespie and Beattie (1930)
log K , = -2.691122 log T - 5.519265 X 10-jT
+ +
1.848863 X 10-7T2 2001.6/T 2.6899 (5) + of ammonia is dQreaction and is expressed as
where T is in K. Activity coefficients used to calculate the
activities for eq 4 are computed from
where Tis in K and R has the value 0.08206. Constants Ai, Eli,
and Ci are given by Nielsen (1968) and appear in Table I. S ,
is computed as
(7)
where mi refers to the mole fraction of component i in the
mixture and the summation includes all components in the
mixture.
Nielsen (1968) discusses the possible values of a and con-
cludes that a should have a value between 0.5 and 0.75. A value
changer. The energy balance for the converter may be ob-
tained from the energy balance equations for the three sec-
tions by combining them in the proper order. The energy
balance equations for each major section are developed. The
energy balance for a horizontal differential element of catalyst
bed involves four heat terms
dQreaction + dQgas + dQriser + dQannulus = 0 (11)
of which the last two are small as already indicated but are
included in the model. The heat generated by the formation
dQreaction = -?3mRF1 dx (12)
AH, is the heat of reaction corrected for the heat of mixing in
Btu/lb-mol of NH3 formed, and is computed from
AHR = -23840.57 + ( P - 300.0)(1.08 + (P - 300.0)
X (0.01305 + ( P - 300.0) (0.83502 X + ( P - 300.0)
X(0.65934 X + 4.5(1391.0 - T) (13)
This equation was derived from the heat of reaction data
presented by Kazarnovskii (1945) for the formation of a
mixture containing 17.6%NH3 in a 3 to 1Hz/Nz mixture. Most
of the heat liberated by reaction is expended in raising the
temperature of the reaction mixture as expressed by
dQgas= -MC,(T, P ) dT (14)
where M is the moles of reaction gas. CJT, P)is a function of

Ind. Eng. Chem., Process Des. Dev., Vol. 16,No. 3, 1977 383
pressure, temperature, and composition and is calculated from nent out of the zone is equal to the sum of the flows in since
the BWR equation. Calculation of mixture heat capacities, no reaction occurs.
enthalpies, viscosities, and thermal conductivities are ex- The energy balance for the heat exchanger located at the
plained elsewhere in this paper. bottom of the converter may be obtained from the heat
The remaining two heat terms refer to heat transferred from transfer equations. The cold gas flowing from the annulus
the bed to either the gas flowing in the annulus on the outside passes upward through the exchanger tubes into the riser and
of the catalyst basket or the gas in the riser. Heat transferred reacted gas flows countercurrent and is cooled on the shell
to the gas in the annulus is computed by side. The energy balance for the reacted gas in the shell is
dQannuius= k’C’(T - TA)/Gdz = ~ ~ M F C , ( T PF)
A , dTA MpC,(Ts, Pp)dTs = UA’(Ts - TT)dz (25)
(15) where A’ is the heat transfer area in ft2/ft of heat exchanger
since the primary resistance to heat transfer is the catalyst height and Ts and TT are shell side and tube side tempera-
basket insulation. Neglecting all other heat transfer resis- tures, respectively. U ,the overall heat transfer coefficient for
tances reduces computational effort without substantially the heat exchanger, is computed according to eq 17. h, for the
affecting the accuracy of the simulation. For converters heat exchanger is computed by the method of Donohue (1949)
without insulation, heat transfer to the feed gas would be for disk-and-doughnut baffles
considerable and an overall heat transfer coefficient would be
necessary. Since the riser is uninsulated, an overall heat h, = 0.23D,0.6 -)
(DOGe 0.6
N P ~ ~ . ~ ~
transfer coefficient U is considered in calculating this heat
term where De is the equivalent diameter in inches
dQriser = -UC(T - T d d z = f i M p C p ( T ~PF)
, dTR (16) De = 4 (flow arealwetted perimeter) (27)
where C is the outside circumference of the riser. The overall and G, is computed as
heat transfer coefficient is given by G, = WIS, (28)
l l U = llh, + lhi, + Dd(2k In (D,/Di)) + e (17) S, is the geometric mean of the cross-flow and the baffle-hole
h, is calculated for heat transfer inside packed tubes as de- areas, i.e.
scribed by Leva (1948) S, = (cross flow area X baffle-hole area)0.5 (29)
h, = \zr~iWO.~ (18) The energy balance for the gas in the heat exchanger tubes
where W is the mass flow in lb/h. The geometric factor \k is yields
calculated from ~ I M F C ~ ( TPF)dTT
T, = -UA’(Ts - TT)dz (30)
\k = 3.5e(-4.6DdD)(DP/A)O.7/D (19) The individual heat transfer coefficient for the tube side, hi,,
where D, is the catalyst diameter in feet and D is the bed di- is computed according to eq 22.
ameter. Equations 26 and 30 may be solved by numerical integration
The bed heat transfer factor A is a function of reaction gas or analytically as suggested by Shah (1967) if Cp(Ts,Pp) and
thermal conductivity K , and gas viscosity and is given by C,(TT, Pp) are assumed to be constant. The solution may be
obtained by letting
h = K/po‘7 (20)
P = ~ ~ M F C ~ (PF)
TT, (31)
The Colburn equation is used to calculate the heat transfer
coefficient inside the riser and
PF)NPR-~
hi, = 0.023GCp(T~, /~
NRE-O.~D~/D, (21) + = M P C ~ ( T SPP)
, (32)
G is the mass velocity lb-mollh f t 2 and the Prandtl number Equating eq 26 and 30 and integrating
is computed
Ts = Ta + P/$(TT - TRB) (33)
Npr = C ~ ( T RPF)P/K
, (22)
where T8 and TRBresult from evaluating the constant of in-
Di and Do are the inside and outside diameters of the riser, tegration. Substitution of eq 31,32, and 33 into 30 and rear-
respectively, in feet, and the Reynolds number is computed
N R e = DiGIp

The thermal conductivity of the riser material is k , and t, the

(23) xL
ranging
dz = 1 TA

TRB
dTT
(Ta + PI+ (TT- TRB)- TT)
(34)

fouling factor, was assumed to have a value of 0.001.
The energy balance for the quench zones above each bed
is developed assuming the zone to be adiabatic and perfectly
mixed. Equating the energy flow from the previous bed, Figure
3 (or the riser in the case of the first zone), and the energy in
the quench stream to the energy flow into the next bed yields
for the lth quench zone
M is the molar flow from the previous bed or riser and H is the
enthalpy, Btu/lb-mol. The enthalpy of the quench is obtained
by multiplying the fraction of feed used for quench in the lth
zone by the total enthalpy of the feed. Flow rate of a compo-
Integration over the length of the exchanger yields after
rearranging
C,(Ts, P p ) and C,(TT, P p ) are evaluated at average condi-
tions. Tp may be obtained from eq 33 since a t the bottom of
the exchanger TT becomes TAand Ts becomes Tp.
Calculational Procedure
The basic model eq 2,3,11,25,33,and 35 cannot be solved
analytically for component flows and converter temperatures
since they are nonlinear and coupled due to heat transfer
between the feed and reacted gas. The method of solution

384 Ind. Eng. Chem., Process Des. Dev., Vol. 16, No. 3, 1977
Table 111. BWR Constants"
Nitrogen Argon Methane Ammonia Hydrogen
A 0.031 231 9 0.028 835 8 0.049 4 0.103 540 29 0.00810 853 4
A0 0.872 086 0.823 417 1.855 3.789 282 0.182 670 58
b 0.003 235 1 0.002 152 89 0.003 380 04 0.000 719 585 2 0.010 927 08
Bo 0.028 106 6 0.022 285 259 7 0.042 6 0.051 646 121 0.026 812 097
c x 10-6 0.000 547 364 0.000 798 243 7 0.002 545 0.000 157 532 98 0.000 003 553 07
co x 10-6 0.007 813 75 0.013 141 25 0.022 57 0.178 570 9 0.000 099 967 11
a 0.000 070 923 2 0.000 035 588 95 0.000 124 359 0.000 004 652 178 0.000 089 327 7
Y 0.004 5 0.002 338 271 1 0.006 0.019 805 156 0.002 576 974
a Units L/g-mol atm K.

consists of assuming a value of TRTand T1 is then found such Table IV. Base Case for Simulation of Single Converter
that eq 25 is satisfied to within 0.1 O F on TI. At this point,
values of TR,T1, TA,and F exist and an integration through Inlet pressure 140.0 atm
the first bed is performed using a fourth order Runge-Kutta Outlet presure 140.0 atm
Feed temperature 290.0 OF
method. The fraction Hz converted, x is set to zero to start the Hz feed 10243.0 lb-mol/h
integration. Molar flows a t any point in the bed may then be Nz feed 3414.0 lb-molh
found by eq 3. The pressure is assumed to vary linearly with NHs feed 330.0 lb-mol/h
the fraction catalyst traversed Ar feed 727.0 lb-mol/h
CH4 feed 1817.5 lb-mol/h
P = P' + M(B - z)/B (36) Fraction feed through exchanger 0.58
where P' is the pressure at the bottom of the bed, AI' is the Fraction feed for first bed quench 0.03
pressure drop through the bed, B is bed depth, and z is posi- Fraction feed for second bed quench 0.12
Fraction feed for third bed quench 0.12
tion measured from the top. The pressure, temperature, and Fraction feed for fourth bed quench 0.15
mole fractions are used to calculate a rate for eq 2 and a value Outlet temperature of bed 1 897.9 OF
of dx/dz is obtained. Values of dT/dz, dTR/dz, and dTA/dz Outlet temperature of bed 2 891.8 O F
for the element are found from eq 11,15, and 16. New values Outlet temperature of bed 3 880.0 OF
for TA,TR,T, and x result at the end of each integration step. Outlet temperature of bed 4 859.5 OF
Ten steps are normally required to obtain desirable accura- Mole fraction inerts in converter feed 0.154
CY. Hz/Nz ratio 3.0
When the bed integration is completed, compositions are Mole fraction ammonia in feed 0.02
adjusted according to the value of x and the temperature and Mole fraction ammonia in effluent 0.1265
composition in the next quench zone is computed, which leads
to initial values for the next bed integration. This procedure Stiel. Pure component thermal conductivities are computed
is repeated for each bed until the end of the catalyst section as suggested by Bromley and mixture thermal conductivities
is reached. The temperatures of the reacted gas and the cold are computed according to the method of Lindsay and
annulus gas are entered into the exchanger equations along Bromley; the mixture value is then corrected for pressure by
with the estimates of TRBand Tp. Heat exchanger calcula- the equations of Stiel and Thodos.
tions are repeated using the new values of TR and Tp until
convergence on the two temperatures is obtained. Results a n d Discussion
Solution of the heat exchanger equations completes one Model calculations were performed on an Interdata 7/32
pass through the converter. T o check for convergence, the with an average computational time of 10 s for each case.
value of TRBfrom heat exchanger calculations is compared Simulation results showing the effects of synthesis loop
with the value obtained after integration of the last catalyst variables upon converter efficiencies were obtained for the
bed. If they differ by more than 0.25 O F , a new value of TRT feasible operating temperature range for the ammonia con-
is computed verter. The volume of results necessitates a method or orga-
nization so that variable effects may be clearly presented. A
base case for the simulations was established as shown in
and used to start another pass through the converter. Four or Table IV and all simulations result from varying either quench
five passes are usually required to obtain convergence. fractions or quench fractions and one other process variable.
Mixture heat capacities and enthalpies are computed using Thus, the effects of process variables and the temperature
the BWR equation of state. Table I11 contains pure compo- profile are referenced to one set of operating parameters.
nent constants which are used to compute mixture constants The effect of converter temperature upon heat production
to be used in the BWR equation. Mixture constants are is shown in Figure 2. As temperature increases, the concen-
computed according to Reid and Sherwood (1966) except for tration of ammonia in the converter effluent approaches
Bo, which is computed equilibrium. However, the equilibrium concentration de-
creases with increasing temperature and a maximum in the
heat production curve results. Stephens (1975) indicated that
the fourth bed temperature is of primary importance. The
Mixture viscosities and thermal conductivities are calculated effects of bed outlet temperature were determined by holding
by methods suggested by Reid and Sherwood (1966). Pure all but one outlet temperature constant, and those are shown
component viscosities are obtained using the theoritical in Figure 4. The temperatures of the first two beds seem to
treatment of Chapman and Enskog. Low pressure mixture have little effect upon production as long as they are in the
viscosities are then computed by the method of Wilke and acceptable operating range. Ammonia production, however,
corrected for pressure by the equation proposed by Dean and may be affected if the third bed temperature is too high. This

Ind. Eng. Chem., Process Des. Dev., Vol. 16, No. 3, 1977 385
 '930[ 0.15 ~
-+ S i m u l a t i o n
- Equilibrium
Temperature R t Yhrch O t h e r Beds
Were E v a l u a t e d 1 12.0% I n e r t s
2 15.3% l n e r t r
3 18,O%%lnerts
A x = 0.935

1890 / \
0.14 -

,-
r
-
~ 1900
\ E
<
E

o 1940 :. 0.13
+

it
Bed 2

1920
-
r
E
: 1900

0.12 -

-
1900
800 820 840 860 880 900 920 940

0.11
840. 850. 860. 870. 880. 890. 900.
Bed O u t l e t Temperature
Temperature O F

Figure 4. Ammonia production vs. bed outlet temperature. Figure 6. Effect of inerts upon ammonia concentrations.

0.15 - - S~mulation
- Equilibrium
should not be allowed to exceed that of the fourth bed by more
than 20 to 30 O F .
1 130 Atmospheres
2 140 Rtrncspheres The fourth bed temperature is the temperature of greatest
3 150 At-c:pheres
A x * 0.935 consequence, and the effects of other process variables may
be related to this temperature. The process variables con-
sidered are pressure, inerts, H2/N2 ratio, feed temperature,
0.14 - space velocity, catalyst activities, and ammonia concentration
in the feed gas. A series of simulations for each value of the
process variable being studied was made. The fraction of the
E
feed stream to be used as quench in each quench zone was first
4 chosen to give high converter temperatures. The temperature
*
of the fourth bed was then decreased by varying quench flows
: 0.13 and an attempt was made to obtain a declining temperature
profile where possible. The fourth bed temperatures became
-
L L

r very sensitive to the fraction of total flow through the ex-

0.12 - temperature corresponding to point B on Figure 2 is
0.11 I I I I
results from two factors. At high temperatures the reaction
rate in the third bed is decreased because equilibrium condi-
tions are approached. The temperature of the outlet gas must
be greatly reduced to maintain the fourth bed temperature.
Less conversion in the third bed coupled with additional
quench substantially reduces the ammonia concentration in
the gas entering the fourth bed and a decrease in production
occurs. Figure 4 indicates that the third bed temperature
changer as temperature was reduced. The series of simulations
was considered complete when it appeared point A on Figure
2 had been reached. Only trivial solutions exist when the
passed.
Figure 5 illustrates the effect of pressure upon ammonia
equilibrium and converter outlet concentration. Increasing
the pressure from 130 to 150 atm results in an ammonia outlet
concentration increase from 12.3% to 13.7%at 860 O F . The
I 1
effect of increasing synthesis loop pressure upon converter
efficiency is consistent with the Le Chatelier principle since
there is a decrease in the total number of moles of gas as seen
from eq 1.
An increase in loop pressure is normally a result of de-
creasing the purge from the synthesis loop which is accom-
panied by an increase in inerts. Figure 6 shows the effect of
three levels of inerts upon equilibrium and outlet concentra-
tions of ammonia. The advantages of high synthesis loop
pressures are a t least partially offset by increasing inerts
levels.
The H2/N2 ratio was changed while holding the total flow
rate and the flow rates of ammonia, argon, and methane
constant. Although a higher equilibrium concentration was
computed at a 3 to 1H 2 / N 2 ratio than at a 2.5 or 3.5 to 1ratio,

386 Ind. Eng. Chem., Process Des. Dev., Vol. 16, No. 3, 1977
 ,\: -
0.15

0.14
-

-
9
-
1
2
3
A ”
Simulation
Equilibrium

2.5
3.0
to 1 i2/N2
t o 1 H2/N2
Ratio
Ratio
3 .. 5 t o 1 H 2 / h 2 P a t i o
= 0 . 9 3 55
0.15

0.14
\
-
1
2
3
Simulation
Equilibrium

270°F Feed
290’F F e e d
310’F F e e d
A x = 0.935

E E
4 4

1
\
0 0.13 .p 3.1:
i i
2
Y i

-
E
-
L

0.12

I I I I I I 0.11 1 I 1 I I J
850. 860. 870. 880. 590. 900. 843. 850. 860. 870. 880. 890. 900
T e m p e r a t u r e ‘F T e m p e r a t u r e 1r

Figure 7. Effect of H*/Nz ratio upon ammonia concentrations. Figure 8. Effect of feed temperature upon ammonia concentra-

a higher converter effiency was obtained a t a 2.5 to 1 H*/N2

ratio as shown in Figure 7. A lower converter efficiency was
obtained with a 3.5 to 1 H*/Nz ratio. This results from a de-
1 9 s . Base Feed ? a t e
pendence of rate upon component activities and has been 2 100‘ S a r d Feed R a t e
3 110: B a s e F e e d R a t e
noted by Nielsen (1968) and Hays (1964). A X i 3.935
The feed temperature had little effect upon converter ef- \
ficiency as shown in Figure 8, although the lowest feed tem-
perature resulted in slightly higher conversions. It is possible
that the temperatures used covered such a small range,
270-310 O F , that the dependence was not pronounced.
The dependence of converter efficiency upon space velocity
is shown in Figure 9. At lower flow rates, the gas remains in
the catalyst section for a longer time and the outlet concen-
tration approaches equilibrium. Although the outlet con-
centration is increased, the net production is less at the lower
flow rates.
The dependence of converter efficiency upon catalyst ac-
tivity is shown in Figure 10. As would be expected, efficiency
decreases with activity. Since rate data of Nielsen (1968) were
used, 100%activity corresponds to the activity of the catalyst
used in his 1964 rate experiments decreased by the effec-
tiveness factor as expressed by eq 9. Catalyst available today
may have increased activity and therefore an activity of 110%
was considered.
T o determine the effect of ammonia concentration in the
feed gas, runs were made with 2.5% ammonia in the feed and
results are shown in Figure 11. A comparison shows only a 0.11 I 1 I 1 I ,
840 850. 860. 8iC a80 890 93:
slight increase in outlet concentration and a net result of de- Tespe-ature -F
creased conversion. Figure 9. Effect of space velocity upon ammonia concentrations.
Figures 5 through 11 indicate that maximum efficiency
(highest ammonia outlet concentration) occurs with a fourth
bed outlet temperature, Ta, between 850 and 870 O F . This choosing that value which maximizes efficiency. At the “best”
temperature range may be expanded if the simultaneous ef- value of Tg, the converter is close to “blow-out”, point B in
fects of several variables are considered. A control system Figure 2 , and would be very sensitive to changes in converter
designed to control the fourth bed a t a temperature corre- quench flow changes. A trial-and-error determination of the
sponding to maximum efficiency by manipulating the fraction “best” value of Tg would be difficult. If the process is on
feed gas heated in the exchanger should improve converter computer control with an analyzer in the synthesis loop, a
efficiencies. The best set point for Tg may be found by a better method of determining the “best” value of Tg is avail-
trial-and-error controlling on different values of T8 and able. This consists of computing the equilibrium concentration

Ind. Eng. Chem., Process Des. Dev., Vol. 16, No. 3, 1977 387
 0.15

\
-
- sE iqmuui ll iabt ir oi un m
1
2
90% A c t j v i t Y
100% A c t l v i t Y
0'15 1\ -
1
A2
Simulation
Equilibrium

2.0% N H ~i n F e e d
2.5%
X N H J i n Feed
0.935
3 110% A c t i v i t y
A X = 0.935

0.14

E
<

:.c 0.13
\
-r"
Y

0.12

0.11 I I I I I I 0.11 I I I 1 I I I
850. 860. 870. 880. 890. 900. 840. 850. 860. 870. 880. 890. 900.
Temperature O F Temperature O F

Figure 10. Effect of catalyst activity upon ammonia concentra- Figure 11. Effect of ammonia in feed to the converter.
tions.

of ammonia for the converter operating conditions. A function
x may be defined as the optimal or "best" ratio of C"3/
C N H ~The~ ~function
. x would depend upon component flows,
pressure, feed temperature, and other variables affecting
converter efficiency and could be determined from Figures
5 through 11or by other methods. C N His ~easily
~ ~calculated
using eq 5, 6, and 7 and CNH30ut may be estimated as fol-
lows
(39)
where C, refers to feed composition and MR is computed at
an average pressure. C N H ~ ~ ~ J CcalculatedN H ~ ~ ~may be
compared with the desired value x.It is known that CNHsout
approaches G"Hseq at high temperatures. Therefore, if
C"3ouJCNHaeq is greater than x,Ts should be reduced. This
method has the advantage of determining which direction to
change T8 in order to reach the desired efficiency. x may be
set as far from the unstable region, point B in Figure 2, as
desired. As stated earlier, the converter becomes sensitive to
changes as T8 is decreased and the computed value of
C N H ~ ~ ~ J may C N beH ~used
~ ~to calculate a variable gain for
the controller if necessary.
The method of calculation of x may be determined to give
the desired results. The A points in Figures 5 through 11were
determined with x set equal to 0.935. In all cases acceptable
values for T8 resulted, indicating that a constant value of x
may be used.
In order to maintain the desired value x as operating con-
ditions change, the fraction cold feed passing through the
exchanger must be manipulated. As this fraction is changed,
the temperature profile in the converter will change. Figure
4,however, shows that redistribution of quench will be nec-
essary only when one of the first three beds violates an upper
or lower temperature limit. Figure 4 also indicates that the
redistribution of quench will have little affect upon
CNH3oudCNH3eq.
Conclusions
Converter temperature control is normally far down on the
list when considering computer control functions. Plant tests,
however, reveal that good temperature control can increase
production by over 1%at a constant total gas rate to the
plant.
This study shows how converter efficiency may be improved
over current operating practices by the use of computer con-
trol. The equations necessary for computation of the fourth
bed temperature set point are simple and may be linearized
for use with a programmable analog controller although digital
systems are usually employed because of their increased ca-
pability.
The optimum converter operating temperature changes
with inlet conditions and catalyst activity. The figures pre-
sented are for a low-pressure converter (150 atm), but the
trends are consistent with those reported for converters op-
erating at 300 atm (Shah, 1967; Baddour et al., 1965; Slack et
al., 1953).It is felt that the procedure described in this paper
for temperature control could be adapted to most ammonia
converters.
Nomenclature
A = cross-sectional area of catalyst bed
Ai = constantsineq6
A' = heat transfer area per foot of exchanger
ai = activity of component i
B = catalyst bed depth
Bi = constants in eq 6
Boi = BWR constant for component i
Bo, = BWR mixture constant
bi = constantsineq9
C = circumference of riser
C' = circumference of catalyst section
Ci = constantsineq6

388 Ind. Eng. Chem., Process Des. Dev., Vol. 16, No. 3, 1977
(C,) = mixture heat capacity TRB = temperature a t bottom of riser
C N H =~ molar
~ ~ ~concentration of N H 3 out of converter 2's = temperature on shell side of exchanger
CNHseq = equilibrium concentration of NHB a t fourth bed TT = temperature on tube side of exchanger
outlet conditions U = overall heat trassfer coefficient
D = beddiameter W = moss flow rate
Di = inside tube diameter x 5 fraction Hz converted
D e = equivalent diameter z = distance from top of zone
D o = outside tube diameter Greek Symbols
D, = catalyst particle diameter
cy = kinetic parameters
Fi = molar flow of component i
FIi = molar flow of component i within bed P = definedbyeq31
f i = Fraction inlet flow y = activity coefficient
G = massvelocity e = fouling factor
G , = equivalent mass flow rate 7 = conversion based on nitrogen
(H) = molar enthalpy of mixture K = thermal conductivity of gas mixture
L~HR = heat of reaction p = viscosity of gas mixture
= average heat of reaction u = catalyst activity
h, = outside heat transfer coefficient E = effectiveness factor
hi? = tube side heat transfer coefficient x = "best" ratio of C ~ H 3 0 u J C N H 3 e q
i, J = 1refers to H2, 2 refers to Nz, 3 refers to NH3,4 refers A = bed heat transfer factor
to A, 5 refers to CH4 J, = defined by eq 32
K , = equilibrium constant in terms of activities \k = geometric factor
k = thermal conductivity of heat transfer surface d = thickness of insulation
k' = thermal conductivity of catalyst basket insulation
k 2 = rate constant for decomposition of ammonia Literature Cited
L = length of heat exchanger Annabie, D., Chem. Eng. Sci., 1, 145 (1952).
1 = subscript refers to converter zone Baddour, R. F., Brian, P. L. T., Logeais, B. A,, Eymery, J. P., Chem. Eng. Sci., 20,
M = molar flow rate 281 (1965).
M I = molar flow rate in zone 1 Brian, P. L. T., Baddour, R. F., Eymery, J. P., Chem. Eng. Sci., 20, 297
(1965).
M F = molar flow rate of feed Dyson, D. C., Simon, J. M., Ind. Eng. Chem., Fundam., 7,605 (1968).
M p = molar flow rate out of converter Donohue, D. A., Ind. Eng. Chem., 41 (ll), 2499 (1949).
mi = mole fraction of component i Gillespie, L. J., Beattie, J. A., Phys. Rev., 36, 743 (1930).
N p , = Prandtlnumber Hankinson, R. W., Phillips Petroleum Co., Bartiesville, Okla., Letter RWH-1-
76TSD (1976).
N E , = Reynolds number Hay, J. J., Pallai, I. M., Brit. Chem. Eng., 8 (3), 171 (1963).
P = pressure Hays. G. E., Poska, F. L., Stafford, J. D., Chem. Eng. Prog., 60 (l), 61 (1964).
P' = pressure at bottom of bed Heerden, C. van, Ind. Eng. Chem., 45, 1242 (1953).
P = pressure differential across bed Kazarnovskii, Ya. S., Zh. Fiz. Khim., 19, 392 (1945).
Kjaer, J., "Calculation of Ammonia Converters on an Electronic Digital Com-
PF = pressureoffeed puter," Akademisk Forlag, Copenhagen, 1963.
Pp = pressure out of converter Kubec. J., Burianova, J., Burianec, Z., Int. Chem. Eng., 14 (4), 629 (1974).
Q = heatterm Leva, M., Weintraub, M., Grummer, M., Clark, E. L., Ind. Eng. Chem., 40 (4), 747
( 1948).
R = gasconstant Nieisen. A., "An Investigation on Promoted Iron Catalysts for the Synthesis of
r = reaction rate Ammonia," 3rd ed, Jul. Gjellerups, Copenhagen, 1968.
S, = geometric mean of the cross-flow and the baffle-hole Reid, R. C., Sherwood, T. K., "The Properties of Gases and Liquids," 2nd ed,
areas McGraw-Hill, New York, N.Y., 1966.
Shah, M. J., Ind. Eng. Chem., 59 (l), 73 (1967).
S , = term computed from eq 7 Slack, A. V., Allgood, N. Y., Maune, H. E., Chem. Eng. Prog., 49, 393 (1953).
T = temperature Stephens. A. D., Chem. Eng. Sci.. 30, 11 (1975).
~'2 = temperature in lth zone Temkin, M., J. Phys. Chem. (USSR), 24, 1312 (1950).
T A = temperature in annulus
T F = temperature of feed
Tp = temperature out of converter Received far review October 13, 1976
T R = temperature in riser Accepted January 31,1977

Ind. Eng. Chem., Process Des. Dev.. Vol. 16, No. 3, 1977 389