Energy-Saving Modifications inAmmonia Plants

At the 600 t/d Luling plant, the program included a Ljungstrom airpreheater to improve primary reformer efficiency, revised surfacecondenser, and hydrogen recovery from the synloop purge gas, usingseparation technology developed by Monsanto.

D. L. Maclean, Monsanto, Research Triangle Park, N.C., C. E. Prince, Monsanto, Luling, La., and Y. C. Chae,Monsanto, St. Louis, Mo.

2.59(2.37)

e/j«z«3

0.«0(0.38)

0.20(0.19.

0.10(8.69)

O

1959 19SO 1970 1980

Figure 1. Rise in natural gas cost.

0360-7275/80/3760-0098 $01.00 e 1980 AIChE

Natural gas costs have skyrocketed in the last few years(Figure 1). In less than 10 years, the cost has risen morethan ten-fold, while from 1950-1970, the price of naturalgas only doubled. The rapid escalation of natural gas costsin the U.S. is expected to continue, and in the face of suchrising costs, ammonia producers have been forced to evalu-ate energy-saving measures.

Monsanto operates three ammonia plants at Luling,Louisiana. One of these, built in 1954 by Chemico Construc-tion Company, has been upgraded from 300 (272 metric) t/dto 500 (454 metric) t/d. The other two, rated at 600 (545metric) t/d and 1150 (1,043 metric) t/d were built by M. W.Kellogg in 1965 and 1976, respectively.

Energy-savings programs have been initiated in all of theammonia plants at Luling. This article will discuss designand operating experience using these modifications in the600 (545 metric) t/d plant. Energy-saving modificationsinclude a Ljungstrom air preheater to improve heat effi-ciency of the primary reformer, revised surface condenser,and hydrogen recovery from the synthesis loop purge gas,using gas separation technology developed by Monsanto.

Furnace combustion air preheat

The addition of combustion air preheat to the 600 (544metric) t/d Kellogg ammonia plant was made with theexpectation of saving JX84K Btu/ton (977 M J/metric t).Actual savings of 0.86K Btu/ton (1,000 MJ/metric t) arebeing realized. The project cost $3 million.

The recovery of additional heat from the primaryreformer flue gas system was accomplished by adding aLjungstrom rotary type regenerative heat exchanger.Shown in Figure 2, this heat exchanger has reduced the fluegas temperature 210°F (99°C) to an outlet temperature of320°F (160°C). To support the exchanger, we added aforced draft fan (Figure 3), and the existing induced draftfan was replaced with one of higher capacity. The conver-sion to forced preheated combustion air also requiredadding new burners to the primary reformer arch andauxiliary boiler burners.

The air preheater is driven at 3 rev/min by a 5 hp (3.7kW) electric motor. A backup air driven motor is provided.

Figure 2.Combustion

airpreheater

andduct

system.

To maintain clean heat transfer surfaces during possiblefuture oil firing, there is both soot blowing and waterwashing capabilities.

A 450 hp (336 kW) electric motor was installed to drivethe new forced draft fan. Although no backup fan driverwas provided, we added emergency air doors and dualelectric power feeds to increase reliability. The induceddraft fan, however, has both a primary and a secondarysteam turbine driver and is independent of the electricalpower supply. A Marland disengaging clutch was installedbetween each turbine drive and the induced draft fan, sothat if the main turbine trips, the standby turbine can startup automatically.

The air duct system is designed to provide uniform

pressure to all burners by supplying air from a header ateach end of the burner rows, shown in Figure 4. There areflow control dampers for each row and for each burner.Figures 5 and 6 show individual air ducts coming from theheader, and one duct connected to an air box. The system isperforming well, and the firing across the furnace is veryeven. The new reform arch and auxiliary boiler burners aredesigned for burning natural gas and/or vaporized oil, andwill accept a steam automized liquid oil gun.

Surface condenser

The four gas compressors in the ammonia plant are steamturbine driven. Each compressor has a condensing-type

Figure 3.Combustion air

forcesdraftfan.

E L E C T R I C MOTOR

Figure 4.Furnace

combustionair preheat.

P R I M A R Y REFORMER A U X I L I A R YBOILER

ELECTRIC A I RMOTOR MOTOR

AIRP R E R E A T E R

INDUCEDDRAFTFAN

steam driver that exhausts steam into the surface condens-er. Vacuum losses of 4 to 5 in. Hg (13.5 to 16.9 kPa) duringthe hot months had been observed as a result of operatingthe plant above design and the pluggage of 5% of the tubesdue to leaks.

An energy audit in 1976 indicated that a reduction indriver steam could be achieved by upgrading the perform-ance of the surface condenser. The rising cost of energymade this project attractive.

Several alternatives to upgrade the surface condenser

Figure 5. Individual air ducts andheader.

were considered. They were: retubing the existing condens-er, adding a parallel condenser, and complete replacement.Because of space limitations and the desire to increaseperformance for rates above design, we chose completereplacement. A 20% increase in surface area was achievedby adding 520 tubes in a 14 inch (0.36 m) vertical segmentat the horizontal center line of the condenser. A sketch ofthis is shown in Figure 7. The new condenser fit the same

Figure 6. Air duct to air box connec-tion.

3-36"D (0.91M)

72"D (1.83RS)

ORIGINAL

2-48"D (1.22M)

1-42-0 (1.07M)

14" (0.36M)

72"D (1.83M)

REPLACEMENT

Figure 7.140-C surface condenser.

Non-permeote gas outlet

Fiber bundle plug

Hollow fibei

Separators, 4" to 8" di-ameter by 10" to 20*

long, length, diameterand number of separa-

tors determined by pro-cess need.

Permeate gas outlet

A Figure 9. Prism separator.

mounting platform and has the same inlet and outlet nozzlelocations.

To remove the additional head load without shifting it toa downstream heat exchanger, extra cooling tower watersupply and return pipes were paralleled into the existingpiping at the surface condenser. The new condenser andassociated piping are shown in Figure 8. Other supportingequipment for this project included new uprated coolingwater pumps and motors. Also the exhaust steam lines andentry domes were revised to reduce pressure drop from thesteam drivers and provide more uniform distribution intothe surface condenser.

Current operation of this new system is slightly betterthan design. Nominal vacuum values in the summer monthsof 27 in. Hg (91.2 kPa) versus 26 in. Hg (87.8 kPa) design arebeing achieved. Recovery of the 4 to 5 in. Hg (13.5 to 16.9kPa) of vacuum loss represents a savings of 0.95K Btu/ton

Figure 8. Surface condenser andpiping.

Figure 10.

Interim

test

unit.

90

oo

85- •

..

10 20 30 10 20 30 10 20APRIL MAY JUNE

30-H-f-

10 20JULY

Figure 11. Interim unit product H2 purity.

Figure 12. Prism separator system,front view.

Figure 13. Prism separator system,rear view.

jnn n

i — 4jNJHJ'>n i l IWATER V V V

l™™— """""̂ ^ __B-M ^gj BANK p

X WATERSCRUBBER

PURGEGAS^v

"̂̂ """̂ ^W

1 1H', e,SOLUTION

r

i>

r

Ï

r

>i

r

»i

i

»1

M«

mSNT" SEPARATORS 1

-*.

r

i

r

t

T

r

iT

i

»I

••••Hi RECYCLE TO w

J 2ND STAGE SUCTION "̂SYN GAS COMPRESSOR

1 FUEL1 GAS TO w

NOX ABATEMENT OR ™PRIMARY REFORMER

H. RECYCLE TO v

2ND BANK 1ST STAGE SUCTION ^PRISM™ SEPARATORS SYN GAS COMPRESSOR

Figure 14. Prism separator system flowsheet, 600 (545 metric) t/d ammoniaplant at Luling.

(1.093 MJ/metric t). The yearly average savings is expectedto be 0.8K Btu/ton (930 M J/metric t). The cost of thisproject was $800. K.

Hydrogen recovery from purge gas

Monsanto has developed proprietary hydrogen recoverysystems for effective and economical separations of hydro-gen from process or purge gas streams, including purgestreams from ammomia plants (1). The new hydrogenrecovery system, containing Prism separators, utilizes theprinciple of selective gas permeation through membranes toseparate hydrogen from other gases.

The membranes are in the form of hollow fibers. The

design of the separator resembles a shell and tube heatexchanger. A bundle of hollow fibers is sealed on one endand embedded in a tube sheet at the other. The entirebundle is encased in a shell with suitable connections(Figure 9).

The hollow fiber allows the selective transport of variousgases across its wall. Certain gases, for example hydrogen,have high permeation rates while others, such as methane,argon, and nitrogen, have lower rates. This allows for aseparation between hydrogen and the slower gases. Pressur-ized feed gas enters the separator on the shell side and flowsaxially along the fiber wall. The hydrogen rich permeate gasflows through the fiber bores and leaves the separator viathe tube sheet at reduced pressure. The non-permeate gases

having high inert concentrations exit the separator from theshell side, at essentially the same pressure as the enteringfeed gas.

Prism separators have been operating successfully formore than two years in Monsanto's petrochemical installa-tions. Our first commercial scale operation was for adjustingthe ratio of carbon monoxide to hydrogen in an oxo alcoholsyngas stream at Texas City, Texas. In another unit, 99%purity hydrogen was recovered from a purge stream in ahydrogénation operation at the Pensacola, Fla. plant.

Monsanto's high pressure separators are capable of oper-ating at pressures up to 2,150 psi (14,824 kPa) and providehigh hydrogen permeation rates. This makes the applica-tion of hydrogen recovery from an ammonia synthesis purgegas stream economically viable.

Initial field tests using Prism separators were undertakenin a reciprocating compressor ammonia plant at Luling, in1978. The purge gas pressure was reduced from 5,000 psi(34,457 kPa) to 2,000 psi (13,790 kPa) before it was fed tothe separators. The unit ran successfully for the test dura-tion of over four months. A similar pilot demonstration wasconducted at Mississippi Chemical Company's 1,000 t/dKellogg ammonia plant at Yazoo City, Miss, in early 1979.This test ran continuously for over 100 days, and demon-strated hydrogen recovery and purity as high as 90%, withno indication of performance decline.

Test results

An interim scale test unit was installed in the 600 (545metric) t/d Kellogg plant at Luling incorporating twocommercial-size Prism separators (4 in. diameter and 10 ft.long), as shown in Figure 10. The unit treated about 20% ofthe purge gas produced in the plant, 0.58K scf/day (0.016Km3/d). The permeate, or product stream, of 89% hydrogenwas recycled to the second stage suction of the syngascompressor at 1,000 psi (6,895 kPa) pressure. The non-permeate, or reject stream, was sent to the primaryreformer fuel system. Figure 11 illustrates the results of theinterim unit operation, using two commercial-size separa-tors. Essentially, no decline in separator performance wasobserved over the three-month test period.

These successful demonstrations of the Prism separatorsystem in several ammonia units led to the commercial scaleinstallation in the 600 (545 metric) t/d Kellogg ammoniaplant at Luling. It started up in September, 1979.JThe unitwas designed to treat the available purge gas of 2.6K scf/day(0.072K m3/d), butjt could easily handle purge rate up to3.2K scf/day (0.089k m3/d) with minor modifications.

In order to minimize field installation time and expenses,the Prism system (incorporating 12 commercial-size separa-tors) was delivered to the construction site as a pre-fabricated, skid-mounted package. Site preparation wassimple, requiring a small plot size (40' x 10') and necessaryproduct and utility lines. Total on-site construction timewas less than one month, and tie-ins required minimumdowntime in the ammonia plant. Figures 12 and 13 show thecomplete separator system operating as a part of the ammo-nia plant. The skids provide a compact package with easyaccess to the equipment.

The Prism separator system flowsheet is shown in Figure14. The high pressure purge from the 600 (545 metric) t/dKellogg ammonia plant contains upward of 2% ammonia.The ammonia in the purge stream is recovered by waterscrubbing. In our Luling facility, the aqua ammonia solu-tion is sent to an existing ammonia recovery unit, not shownin the flowsheet.

The scrubbed purge gas containing about 60% hydrogenis fed to the first bank of the separators at 2,000 psi (13,790kPa). As the shell (non-permeate) gases pass through theseries of separators, they are depleted of hydrogen. Thehydrogen accumulates on the bore (permeate) side of thehollow fiber separators.

In the first bank of separators, 90% purity hydrogen isrecovered on the bore side and returned to the second stagesuction of the syngas compressor at 1,000 psi (6,845 kPa).The hydrogen partial pressure on the bore side is main-tained at 900 psi (6,205 kPa) at the bore exit from the firstbank of separators.

Since the hydrogen permeation is driven by the hydrogenpartial pressure difference, the rate of permeation decreaseswhen the hydrogen partial pressure on the shell sideapproaches the hydrogen partial pressure of the bore. In oursystem, a substantial amount of the hydrogen is recoveredat the higher bore pressure for return to the plant—anenergy saving (2).

Additional hydrogen is recovered from the residue streamby lowering the hydrogen partial pressure at the bore side ofthe second bank of the separators. The second bank returnsthe recovered hydrogen to the first stage suction of thesyngas compressor at 350 psi (2,413 kPa). The hydrogencontent of the final purge, or reject, stream is reduced toapproximately 20% hydrogen. In our Luling plant, thereject stream is sent to the nitric acid plant for NOXabatement, but it could be also returned to the primaryreformer as fuel.

The average composition of both recovered hydrogenstreams is 89% hydrogen, 6% nitrogen and 5% inerts. Thetypical reject stream contains 20% hydrogen, 42% nitrogenand 38% inerts. Shortly after the startup, the Prism separa-tor system in the Luling plant provided recovered hydrogenstreams having an average hydrogen purity of 89%, withhydrogen recovery of^ 86%. At this level of performance,energy savings of 0.5K Btu/t (0.5815 MJ/metric t) of ammo-nia are indicated. If an ammonia plant has incremental aircompressor and syngas compressor capacities, the recov-ered hydrogen can be used for approximately 4% additionalproduction or 40 (36 metric) t/d in a standard 1,000 (907metric) t/d plant. The Prism separator system has provento be an economical, reliable, and maintenance-free hydro-gen recovery unit for ammonia purge gas streams. #

Literature cited1. Perry, E., U.S. Patent 4,172,885, Monsanto Co.2. Graham, T. E. and D. L. Maclean, U.S. Patent 4,180,552, Monsanto Co.

D. I. MacLean, engineering manager of the Sepa-rations Business Group of Monsanto is responsiblefor the design of PRISM separator systems. Theauthor of several patents and papers in separationprocesses, polymer properties and polymer process-ing, he has a B.S.ChJE. from the Univ. of Cincinnatiand earned his M.S.Ch.E. and Ph.D. degrees fromthe Univ. of Tennessee.

C. E. Prince, a principal engineering specialist atMonsanto's Luling, La., facility earned hisB.S.Ch.E. and M.S.Ch.E. degrees at the Univ. ofArkansas. Current responsibilities include processevaluations and process computer control.

Y. C. Chae, commercial development manager ofMonsanto's Separation Business Group, is responsi-ble for worldwide commercialization of PRISMseparator systems in ammonia and petrochemicalindustries. He holds a B.S.Ch.E. from South DakotaSchool of Mines & Technology and a Ph.0. inchemical engineering from Case Institute of Tech-nology, At Monsanto, he has also been involved inseveral new product developments including plasti-cizers, flame retardants, and reinforced plastics.

DISCUSSION

UNIDENTIFIED: Why is it necessary in the process to scrub theammonia out before you separate the hydrogen? Or is itnecessary?

D.LMacLEAN: We recommend scrubbing out the ammonia toinsure proper functioning of the unit. We scrub it to an order ofmagnitude or so lower than it needs to be. This also makespossible about 5 ton/day extra output from a 1,000-ton/day plantor an 1,150-ton/day plant.

SÄUE QUESTIONER: Is it necessary to do this? Can't you leavethe ammonia in it?

MacLEAN: No, we do not recommend that. The performancedeteriorates. That's one of the reasons we have a scrubberinstalled, along with all the associated controls and trips. Thetwo-fold purpose of the scrubber is to recover the ammonia and toincrease the life of the unit. Pretreatment of some sort is typical ofall hydrogen recovery systems.

UNIDENTIFIED, Petrocarbon Ltd: My comments relate to thepressure relief arrangement on this unit. It seems to be quitesimilar to what one might expect on a heat exchanger, whereobviously one can find an accepted method for calculating thesize of the relief system. How do you allow for this in anexchanger of the type of configuration using thousands of tinytubes, for example?

MacLEAN: The vessel itself is rated at the full pressure of theunit. For a coded vessel we also use relief valves in the system, toprotect from overpresses just as on any vessel.

UNIDENTIFIED: Do you then rate the fuel system for the highpressures as well?

MacLEAN: We have a relief, so we don't go above that on the fuelsystem.

UNIDENTIFIED: So my question really was - do you assume thenthat there is a possible rupture of all the tubes all at once?

MacLEAN: We always make that assumption, and we alwaysassume the worst possible case. I'm sure that would happen ifthe unit were for hydrogen recovery. It is always wisest to assumethe worst possible case, and to design for safety.

UNIDENTIFIED: How does the unit behave on the turndown? Ipresume that the permeability is quite sensitive to velocitythrough the unit?

MacLEAN: It's not very sensitive to velocity. With the unit atPensacola we had about a factor of three on the change in flowrates. We had about 99% purity from that unit.

UNIDENTIFIED: The recovery stays the same?

MacLEAN: The recovery changes slightly. The unit at Luting hada somewhat lower rate when we started up. Now we are runningthe whole purge and just the shell only for Nox abatement. Wevaried that flow rate by a factor of almost two. There are somechanges in recovery; the lower the flow rate, the higher therecovery.

UNIDENTIFIED: Why do you insulate the unit? Is there a thermaleffect in the system which is required?

MacLEAN: It can run without insulation. We insulate because therate coefficients are temperature dependent. The rate across thefiber is somewhat temperature dependent, so we insulate theseparator just to achieve consistent performance.