Purge Gas Purification and Recovery in

Ammonia Plants
Discussion of design considerations, construction, and operating experience of a
hydrogen recovery unit. It reviews the benefits and operating considerations of running
two ammonia plants through a single purification unit.
R. Harmon
Agrico Chemical Co.
Verdigris, OK
W.H. Isalski
Petrocarbon Developments Ltd.
Manchester, England

Cryogenic recovery of hydrogen from ammonia
plant purge gas is becoming more attractive
as costs of feedstock and energy continue to
skyrocket.
The following reviews the construction
and start-up of a Hydrogen Recovery Unit at
Agrico Chemical's Verdigris Plant, with input
from the designer, Petrocarbon Developments
Ltd. The unit .recovers hydrogen from two
1150 STPD Kellogg ammonia plant purge streams.
PROCESS DESCRIPTION
Hydrogen recycle gas is produced from the
purge gas in two main stages, a pre-treatment
section and the cryogenic section. An overall
process schematic is shown in Figure 1.
Pre-treatment section
Raw purge gas enters the unit through a
control valve, reducing the pressure to 1000
psig (6895kPa). The gas then passes through a
pre-heater and is admitted to the base of an
ammonia absorption column. Ammonia is removed
from the gas by absorption in a lean aqueous
ammonia solution.
Rich aqueous solution is withdrawn from
the base of the absorber, pre-heated and
passed into the regeneration column. Heat for
the regeneration is provided by a reboiler
using medium pressure steam. Hot lean solution
is drawn off the reboiler thru a level con-
0149-3701/81/4401 $02.00 © 1981 AlChE
39
troll er, then cooled and recycled to the
absorber. Ammonia product from the regene-
ration column overheads is condensed and used
for reflux. Excess ammonia is combined with
the inert purge from the regeneration pres-
sure controller and sub-cooled in a chiller
by Agrico, then sent to cold product storage.
Feed gas leaving the absorber then enters
one of two molecular sieve adsorbers where
residual ammonia together with water from the
absorption system is removed. The adsorbers
operate on a fixed time sequence so that one
is on line and the other undergoes regene-
ration.
Low temperature section
The dry, ammonia free purge gas from the
adsorbers enters the cold box where the gas
is cooled in an exchanger using hydrogen
product and fuel gas streams for refrigera-
tion. Supplementary ammonia refrigeration is
used on start-ups and can be discontinued
during normal operation. A vapor/liquid
phase separation in a feed gas separator is
used to produce a hydrogen rich vapor phase
and a liquid phase containing a large pro-
portion of methane and argon with some nitro-
gen and a small amount of dissolved hydrogen.
The overall rejection of the i nerts (argon
and methane) is therefore high. Typically,
90% of the argon and methane in the feed is
condensed and subsequently rejected to fuel
gas. This is the main objective of the plant.
 PRETREATMENT
ANHYDROUS
AMMONIA
X—^""X COOLING WATER
AMMONIA 55O PSIG STEAM
PURGE CAS
DISTILLATION
COLUMN
Figure 1. Simplified flow diagram of hydrogen recovery unit.
Depending on the operating conditions and the
feed gas composition, the hydrogen product
should contain 87 to 90% hydrogen, with inerts
between 2.2% and 1%, see Figure 2.
90
-89
88
87
86
85
1.0 2.0
% AR+CH« IN H, PRODUCT
Figure 2. Variation of H2 product composition and
inerts content in H2 product.
To assist re-evaporation of liquids which
provides cooling to the feed gas, a quantity
of hydrogen product is injected down stream of
the separator level control valve. This quan-
tity is variable with rate, feed gas composi-
tion and other factors, but must be kept to a
minimum to obtain highest hydrogen recovery.
The hydrogen product stream is warmed to
approximately ambient temperature in heat ex-
change with the incoming feed gas and then re-
turns to the synthesis gas compressors in the
ammonia plants. The liquid phase from the
separator is expanded to fuel gas pressure
FINAL CLEAN-UP CRYOGENIC SECTION
FUEL GAS TO
REFORMER BURNERS
MOLECULAR ELECTRIC
SIEVE ADSORBERS HEATER EXCHANGER
along with the injected hydrogen and then
evaporated and reheated to approximately
ambient temperature by the incoming feed gas.
This fuel gas stream is split after
leaving the cold box to provide a sidestream
which is used for regeneration of the mole
sieve. The regeneration gas is recombined
with the fuel product and is returned to the
ammonia plants for use as fuel in the primary
reformer arch burners.
IMPORTANT DESIGN CONSIDERATIONS
The main component of the cold box is an
aluminum exchanger constructed of a series
of finned plates which are dip-brazed to-
gether. This arrangement allows the counter
flow of multiple streams through the ex-
changer. The design specifications and con-
3.0 struction of this exchanger are probably the
most critical aspects to be considered if
satisfactory performance of the unit is to
be obtained. Consideration must be given to
the actual flow rate and percentage of inerts
in the ammonia plant purge gas. The impor-
tance of this can be shown by listing the
gases available for producing Joule-Thompson
refrigeration in order of significance.
Methane - The component of purge gas
that will condense at
higher temperatures than
the others. On separation,
expansion and re-evapo-
ration it will be the best
aid to/the Joule-Thompson
effect.
40
 Nitrogen - Although it gives a lower
Joule-Thompson effect than
methane, it is available
at higher percentages than
methane and argon, there-
fore its refrigeration
effect is also signifi-
cant.
Argon - The volume percentage and
condensing temperature are
lower than methane. Part
of the volume present as
gas will produce some re-
frigeration.
Hydrogen - Present mainly in vapor
phase at these tempera-
tures. A small amount will
be dissolved in the liquid.
The hydrogen supplies
least amount of Joule-
Thompson refrigeration.
In fact, high concentra-
tions of hydrogen can be
detrimental to the overall
refrigeration balance.
Therefore, the feed gas composition is of
importance, in that higher percentages of
methane, argon and nitrogen favor refrigera-
tion balance. The cold box exchanger heat
transfer balance, should therefore be calcu-
lated using actual operating data from the
ammonia plant covering the whole range of
purge gas compositions and flowrates.
Operating difficulties can develop in
obtaining and controlling the low temperatures
in the cold box unless a correct range of in-
erts and flow range of the ammonia purge gas
is specified during the design of a unit.
For example, the material balance data
supplied by Kellogg for Agrico's ammonia
units indicates a purge gas methane content
of 11.4%. But, this data is usually based
on "worst case" conditions, such as deterior-
ating catalyst activity in the primary re-
former or shift converters,-which will tend
to increase methane. Another factor that can
affect the synthesis loop purge is the hydro-
gen-nitrogen ratio. This ratio is commonly
held at a lower value than the 3:1 design,
which will increase argon and decrease methane
percentages. The actual methane in the purge
analysis at the Verdigris plant averages 9.5%.
Keeping in mind that methane provides the most
Joule-Thompson refrigeration, one can under-
stand the error that can be made if the 11.4%
figure is used for design.
41
CONSTRUCTION AND PRE-COMMISSIONING
Several items are worth emphasizing here that
may be of help to anyone considering the ad-
dition of a Hydrogen Recovery Unit.
The most important factor is when and how
to effect the process tie-ins.
The tie-ins necessary to incorporate the
Hydrogen Recovery Unit to one or more existing
ammonia plants can best be accomplished during
a plant turnaround. A brief description of
these follows:
1. A tee and gate valve were installed
upstream of the loop purge flow
control valve. This allowed com-
pletion of the tie-in after turn-
around by addition of a second con-
trol valve. This valve works on
split range control with the existing
control valve. With this arrangement
good control of the loop purge flow
is maintained when switching to hyd-
rogen recovery feed.
2. A valve was installed to return the
fuel gas product to the primary re-
former fuel system. This tie-in was
made to the existing purge to fuel
line.
3. Tie-ins were made for refrigerant
supply and return, also for ammonia
product return.
4. The product hydrogen return tie-in
was made between the intercoolers
for the second stage suction to the
synthesis gas compressors, (116-C
and 129-C). This allows cooling of
the return hydrogen to normal suction
temperature.
5. All utility tie-ins were also made.
These include medium and low pressure
steam, nitrogen, instrument air, de-
mineralized water, return "condensate
and cooling water.
START-UP OF THE UNIT
Start-up of the unit is accomplished using
the feed from one ammonia plant. The product
hydrogen and fuel are returned to the same
plant from which the feed is supplied. In-
troducing the feed from the second ammonia
plant is done by slowly unblocking the split
range purge flow control valve. With the unit
being used to increase production, the opera-
tors simply increase the air flow to the must be increased to maintain cold end temp-
ammonia plant secondary reformers to control eratures. This results in decreased hydro-
the loop ratios. gen recovery. However, this problem has
During the initial start-up of the unit, only developed at ambient temperatures above
several problems occurred which have subse- 105T (314°K).
quently been remedied.
1. A leak developed in the cold box.
This was detected by sampling of the
cold box nitrogen purge. The cause
was later found to be a leaking
bonnet gasket on the separator out-
let block valve.
2. The design cold end temperatures were Table 1. Range of gas compositions during initial
impossible to obtain, even with hyd- operation of unit. H2 Recovery = 90 + 92%
rogen injection to the fuel gas at
maximum. This has been corrected FEED GAS HVOROGEN PRODUCT FUEL GAS
after discussions between Petrocarbon COMPONENT
Composition - Mole % Composition - Mole % Composition - Mole %

and Stewart-Warner, the exchanger Hydrogen 62.9 - 63.7 86.9 - 87.8 14.2 - 15.3

manufacturer concluded that although Nitrogen

Argon
20.69
5.33
- 21.2
- 5.44
9.9
1.3
-
-
10,3
1.5
41.2
13.0
-
-
43.4
13.3

the exchanger was adequately surfaced Methane

Ammonia
10.19
1.5
- 10.78
- 2.5
0.91 - 1.05 29.2 - 31.2

some maldistribution could occur into Temperature °F -5 60 60

the fuel gas passages at the cold end Pressure psig 1050 92S-950 45-47

of the exchanger. A specially de- Flowrale ISCFHI Design:

signed baffle plate distribution
system was installed to disperse the
liquid and vapor phases of the fuel
stream across the exchanger cold end.
This modification was performed by
Stewart-Warner, the manufacturer.
3. Cavitation in the lean aqua circu-
lation pumps during hot weather and
numerous packing failures was cor-
rected by increasing the size of the
suction piping.
PLANT PERFORMANCE
A range of gas flow compositions encountered
during the plant performance run are shown
in Table 1.
The optimum conditions for increased
ammonia production are attained by allowing
the hydrogen purity to vary between 87-88%,
with an inert level of 1.6-2.0%. This can
be accomplished with no hydrogen injection to
fuel, resulting in hydrogen recovery of 94-95%
and maximum recovery of nitrogen.
Normal:
720,000
680,000
448.700
444,200
Approx.
Appro».
253,300
218,800
The product hydrogen and fuel Flows are
returned equally to the ammonia plants, each
plant using 50%. The unit can also operate
at reduced rate with one ammonia plant.
With the ammonia plants operating at
110-115% production rate, the feed flow to
the Hydrogen Recovery Unit averages 584,000
SCFH (16,53Qm3/Hr). From this, ammonia re-
covered in the water wash section is approxi-
mately six tons per day. The additional
production from recycled hydrogen averages 70-
80 tons per day. Table 2 illustrates a typi-
cal mass balance with the unit operating at
maximum hydrogen recovery.
The onstream time for the unit has been
80% since the start-up in January of this
year. The down time was a result of repair-
ing the leak in the cold box and to install
the baffles discussed under start-up of the
unit.

At the initial start-up of the unit,
ambient temperatures in this area were aver-
aging 20°F (266°K). The unit has since
operated in ambient temperatures to 110 F
(316K). A noticeable effect in plant per-
formance occurs with higher ambient tempera-
tures, and the hydrogen injection to fuel
When comparing the benefits of having
one unit to recover hydrogen from two ammonia
plants versus a separate recovery unit for
each plant, savings in construction costs
would be the primary factor. Agrico has in-
stalled recovery units on our ammonia plants
at Donaldsonville, Louisiana and Blytheville,

42
Arkansas. The Verdigris unit was installed
for 900,000 dollars less than the combined
cost of the other two single train plants.
Operator time which amounts to approximately
two hours per day is another benefit. With
separate units, it could require two operators
at two hours per day. Maintenance cost should
also be about half for a single unit compared
to two units.
DISCUSSION
GLEN COMBS, IMC: In the ammonia plants, there
are slight traces of helium present. How does this
unit go for this helium and do you need separate
purge on your single train to remove the helium? Do
you see any influence on the reformer firing or the
synthesis loop pressure or inerts when you depres-
surize and repressurize the ammonia adsorbers
from service to regeneration?
HARMON: We don't have any problem with helium
in the gas. The adsorption system works automa-
tically on time sequence; the valves open and close
automatically when the unit that has been regen-
erated with the fuel gas stream is pressuring up to
system pressure. We see no change other than a
slight increase in the indication on the feed-gas flow
to the unit. The loop purge flow controllers in the
ammonia units are on automatic and this has pre-
sented no problem. On depressurization to the fuel
gas header we put in redundant pressure switches
to make sure that we don't have any upsets in the
fuel gas system in the reformers and we see no
problem. We have several alarms for high pressure
and low pressure on the fuel gas system in the
ammonia plants; the high pressure alarm goes on as
the pressure rises four or five pounds. Normally, the
fuel gas returning from the hydrogen recovery unit
runs about 45 to 55 pounds.
COMBS: Have you considered using the ammonia
refrigerant stream shown on the process diagram in
continuous service to enable you to lessen the
amount of high-pressure hydrogen product injected
to fuel to maintain cold box temperature? The
design of the unit calls for that ammonia exchanger
to be in service at all times.
HARMON: At one time, during a shutdown in the
ammonia plant the operators neglected to block in
water to our product ammonia. The supply for this
exchanger was the first thing on the line, so the
exchanger was filled with water. The hydrogen
recovery unit was still in service with the number
one ammonia plant. The exchanger froze up. So, we
ran for several months with it frozen until we shut
down to repair the valve on the separator outlet.
During the hot summer, we have left the refrigeration
43
Table 2. Current cold box mass balance with unit
operating at highest hydrogen recovery. H2 Recovery =
94.6%
FEED GAS HYDROGEN PRODUCT FUEL GAS
COMPONENT
Composition - Mote B* Composition- Mole e. Composition - Mole ".
Hydrogen
Nitrogen
63.03
21.37
86.71
10.69
10.82
44.21
Argon 5.54
10.06
1.63
0.97
14.18
30.79
Methane
Temperature I'FI -5 60 60
Pressure Ipsig]
Flowrate ISCFHI
1.050 925-950
401.600
55
182.400
584.000
in service but we don't have to inject the hydrogen
for additional cooling except as I mentioned when
the temperature is very high outside.
B.F. RYAN, NET, Ireland: We are presently evalu-
ating hydrogen recovery. Could I ask a question
relating to another interesting area which was not
covered by the two papers presented and that
concerns relatively new area of prism separators?
Could we hear from somebody who has that process
in operation and what their experience has been?
REPRESENTATIVE, Monsanto Chemical Co.: I'll
offer a comment on that and more or less update on
a paper we presented last year. Our commercial
prism unit has now been in operation for about a
year. It still operates successfully. The hydrogen
recovery is about 87 to 88%. The energy saving in
the plant, which is the way we operate, about 0.6 to
0.7 million Btu/ton.
G. ASHTON, Petrocarbon Development, Ltd.: I'd
like to comment on the last question, comparison
between the prism units and the cryogenic routes.
Our comparison shows that the cryogenic process
has higher hydrogen recovery, higher nitrogen
recovery and a lower inert recycle to the synthesis
loop. The Agrico unit which Bob has discussed is
currently working with 94% hydrogen recovery
against about 87% which was just mentioned for the
prism separator.
WERNER, BASF, Germany: Yes, in oxidation, do
you mean wet corrosion? In our paper, I mentioned
some problems we had with storage of construction
materials. The distribution trays and pall rings in the
absorber and regenerator are made of carbon steel
and they were left out in the weather and had
presented quite a problem because the system was
dirty during start-up due to rust formation. The
problem of getting wet the stellte material in your
molecular sieves. What's that background? The
stellte material that is used in the adsorbers can be
ruined if it gets wet. When we were installing the
stellte, it started raining. We didn't ruin the stellte,
but we did have to discontinue installation and
cover everything for our protection of the material.