Exxon Chemical Low Energy AmmoniaProcess Start-up Experience

Successful start-up and early operation experience with the new ANC ammo-nia process.

P. A. Ruziska, Exxon Chemical Co., Florham Park, N.J.C. C. Song and R. A. Wilkinson, Esso Chemical Canada, Redwater, Alberta, Canada

and William Unruh, Syncrude Canada Ltd., Fort McMurray, Alberta, Canada

Exxon Chemical's ANC ammonia process is a modern, lowenergy design. The design energy requirement is less than30.2 MJ LHV/kg NH3 (26 x 106 Btu LHV/ST).* Design ca-pacity is 1600 metric tons per day (1764 ST/D). The plant,located at Esso Chemical Canada's Agricultural Chemi-cals Complex near Redwater, Alberta, was commissionedin February of 1983.

The design contains previously demonstrated technol-ogy components, combined in a unique fashion to providean economical, low energy design, with emphasis on pro-cess operability and reliability. Technology employed inthe design includes:

• Exxon steam reforming furnace• Gas turbine driven process air compressor, with gas

turbine exhaust used as reformer furnace combus-tion air.

• Low energy Catacarb process• Molecular sieve syn gas driers• Haldor Topsoe Series 200 radial ammonia converter.• Cryogenic hydrogen recovery from synthesis loop

purge gas.• 10.3 MPa (1500 psi) steam generation and letdown in

synthesis gas compressor turbine driver.• Honeywell PMX/TDC-2000 electronic control

system.

Project Development

Selection of the process technology for this project wasthe responsibility of the Agricultural Chemicals Technol-ogy Division of Exxon Chemical Company. The role of theAgricultural Chemicals Technology Division is to developor obtain the best technology and provide engineering ser-vices for the worldwide Exxon Chemical fertilizer activity.Prior to the initiation of this project, this division was in-volved in a three-year study of the ammonia process toidentify best technology for high energy cost locations.

When the ANC project was approved, a task force wasassembled, headed by Exxon Chemical's Chief Agricul-tural Chemicals Engineer, to develop the process designbasis for this specific project. The task force was com-prised of:

• Process engineers from the Agricultural ChemicalsTechnology Division.

• Engineering specialists in various disciplines (e.g.,machinery, furnace, materials, instrumentation, etc.)

Note: * Electricity converted to joules (Btu's) using factor of 9.5MJ/kWh (9000 Btu LHV/kWh).

from Exxon Chemical's Central EngineeringDivision.

• Personnel from the Esso Chemical Canada site atwhich the plant would be built

• Personnel from other Exxon Chemical fertilizerplants who were able to contribute specific tech-nical and operating experience as needed.

• Personnel from Bechtel, the engineering contractorfor the project, who contributed process engineeringas well as cost estimating services to assist in the de-velopment of the optimum process design condi-tions.

• Haldor Topsoe A/S, who was involved in the synthe-sis loop optimization studies for this project.

In total, this task force numbered in excess of 60 people.The design objective was, from the very beginning, to

give equal importance both to energy efficiency and tooperating reliability. A process design was required whichwould compete favorably with the older, now depreciatedunits through reduced energy consumption. The objectivewas not to develop the lowest energy process, but rather toselect the most economical process for this particular loca-tion. Therefore, any additional investment for energyefficiency had to be justified by the cost of energy saved.

However, the design team was not permitted to considerany commercially unproven process concepts. Reliabilitywas, as indicated, an equally important design criterion. Infact, in an ammonia process, reliability also effects energyconsumption. From a cost point of view, it is the overallyearly energy consumption per ton of ammonia producedthat is important, not just the energy consumed per ton ofammonia when the plant is operating at steady state condi-tions. A process which may be difficult to operate or haveunproven concepts that cause recurring plant outages willconsume a considerable amount of energy during startupsand shutdowns, or during stand-by periods while prob-lems are being corrected (see Figure 1). A plant with a lowservice factor can expect to have a yearly average energyconsumption several million Btu's per ton higher than aplant with' a good service factor.

Not only did this reliability criterion affect the technol-ogy considered for use in this project, but it also led to avery major effort in assuring that the technology selectedwas properly applied in order to minimize potential upsetsand equipment problems. Reliability considerations wereaddressed in the following fashion:

• Operating experience with all technology consid-ered for use in the ANC design was carefully re-viewed. This included experience in Exxon and

22

AIChESURVEY

AVERAGE

60 70 80SERVICE FACTOR. X

90 100

Figure 1. Effect of Service Factor on yearly energy consumption.

other ammonia plants, as well as operating experi-ence in other process applications.

• Risk analyses were conducted to determine appro-priate design features to be incorporated in order toavoid problems or to cope more easily with problemsshould they arise.

• A study of the steam system dynamics was con-ducted, in order to identify stability of key steampressure levels for a variety of typical site upsets.This study led to some design and control systemchanges to improve steam system reliability.

• Extensive pre-startup training of the operating per-sonnel was implemented. The key to this trainingprogram was a newly developed Exxon computer-ized dynamic simulation model of the actual ANCammonia process [1 ]. This model not only enabledthe operators to become familiar with this uniqueammonia process design, but also gave them practicein responding to over one hundred differentoperating upsets that could be encountered in actualpractice.

» Process analysis was conducted on a variety ofoperating modes including start-of-run catalyst,winter vs. summer plant conditions, and variousphases of startup. This information was used to sizecertain equipment for other than normal steady statedesign conditions, as well as to provide data to theoperators on control targets for different points dur-ing the startup sequence.

• Rigorous machinery audits were conducted for eachmajor piece of rotating equipment. This included in-dependent checks of the vendors' designs usingExxon methods, as well as a review of all critical ma-chinery components and auxiliary systems.

« Esso Chemical Canada provided a project manage-ment team that participated in the detailed engineer-ing and procurement phases of the project to ensurethat the owner's specifications were adhered to andthat equipment selection decisions reflected bothenergy efficiency and reliability considerations.

® During the detailed engineering phases, weeklymeetings were held to review the contractor's prog-ress. These reviews involved various owner person-nel representing engineering, operations, mainte-nance, and project functions. Specific sessions wereheld for review of P&I/D's and the model of the plantlayout.

» During construction, Esso Chemical Canada pro-vided a team to oversee the contractor's activity. Thisteam was comprised of personnel who would laterbe assigned to the plant maintenance function.

9 Esson engineering standards were utilized, whereapplicable.

It is estimated that the additional investment resultingfrom reliability considerations has added nearly 5% to theinstalled cost of the ammonia project. This covers featuressuch as:

• Upgraded heat exchanger tube metallurgy.• Installed spares for critical rotating equipment, over

and above normal industry practice.• Provisions for operating the plant without certain

equipment in service.• Redundant level instrumentation on high pressure

steam drum.• Specific vent and isolation valves for improved

operability.• Low temperature shift guard vessel.• Exchanger sizing based on limiting condition (in

some cases occurring at start-of-run, vs. normal end-of-run catalyst activity).

It was not always possible, during the development ofthe project, to prove quantitatively that the added invest-ment for certain reliability features was justified. Instead,ammonia plant operating experience was used as a guidein selecting features intended strictly for improved relia-bility. The justification can be seen from the fact that a ser-vice factor improvement of 1% to 2% (4 to 7 days lessdowntime per year), will provide an acceptable return onthe incremental ANC project investment For improved re-liability. It is now believed based on actual operating ex-perience, that this money has been well spent.

The Startup

The project was completed five months ahead of origi-nal schedule, at a cost of 4.5% below budget. The startupitself was particularly impressive, especially in view of thetiming: the Canadian winter months of January and Febru-ary. Ammonia production was achieved after only 11 daysfrom the introduction of feedgas; and on the 15th day afterfeedgas introduction, on-spec ammonia was delivered tothe storage tank (see Figure 2). The ammonia converterstartup heater was taken offline two days later. This Exxonrecord was in spite of three power interruptions, whichcaused one full outage and several upsets.

The first burner in the primary reformer furnace was litat 7:00 p.m. on January 26,1983, and curing of the reformerfurnace began. Air flow was established through thefeedgas lines, radiant tubes, transfer lines, secondary re-former, and high temperature shift converter. Predeter-mined temperature levels and holding times were ob-served. Front end leaks were also checked out by raisingthe air pressure to 4.1 MPa (600 psig).

Process steam was introduced into the reformer tubes at2:00 a.m., February 5. Feedgas was then introduced a daylater on February 6. While sulfur removal from the hightemperature shift catalyst was proceeding, the synthesisgas compressor turbine was tested without load using 6.9MPa (1000 psi) steam.

Process gas was then introduced through the Catacarbsystem, bypassing the low temperature shift converter,and methanator catalyst reduction was carried out. At thatpoint, at 5:00 a.m. on February 12, there was a total com-plex power failure. Utility boiler instrumentation prob-lems caused the boilers to trip, and steam pressure couldnot be maintained. Feedgas had to be withdrawn. Progresswas delayed 30 hours by this occurrence.

Other problems during the startup period included:• The process air compressor intercoolers became

fouled with foreign material (construction debris) inthe cooling water system. Excessive air tempera-tures were avoided by flushing on the run whenrequired.

• Commissioning of the steam tracing lines wasdifficult due to freeze-ups. Many sections had to bethawed below flow could finally become estab-lished.

• Site glasses blew out on the 10.3 MPa (1500 psi)steam drum, and also on the CO2 absorber level sec-tion. In both cases, the level gauges were manually

23

JANUARY 1983 FEBRUARY 1983

3 | 4 | 5 | 6 i 7I 7 I 8 I 9|to]n|l2|13|l<l|l5| 1

LINE 50«. 600» LINE

CLEANING ""*1500» STEAM LINE

»—• »-•DECREASE ACID CLEAN

S

INITIAL

AIR COMPRESSORSTART

bapobil i

/^TOTH

TEAM BLOW /

o| n| 12) | I8|19| 201 211 221231241 25126127128

BECAME CRITICALHE PROGRESS

AGAINST TARGETS

FRONT READY i" AT 60% RATE

MAINPROCESS

AIR COMP. i GASTURBINE TEST

SYNTHESISSECTION

REFRIGERATIONCOMPRESSION

DISTURBANCES

STA"T I MECHANICAL I 1!

RNERS \ COMPLETION!iSTED _ \ ' I —'

BURNTESTED

METHHTS REDUCTION!

READY OVER

FRONT-ENDT DISTURBANCES PURGEQfl

\ j TO FUEL

0%V TO 75% y TO 85%

POWERBUMP

LEAK TESTFRONT(TO 600«)

\ ST\ IN

STEAM GAS TO \_GAS OUT DUE TOITRODUCED CATACARB POWER BUMP

FURNACE &FRONT DRYOUT t-

LTSREDUCTION

SG TURB.TEST

SYNGASCOMP.

SGTUP,ATMIN

READY LOOPLEAK

CHECK

N2 PURGE

POWERBUMP

t

„IMC /"!GOv7

"3EACTIONEGAN

S/U HEATER

ON

RUN

LEAKFIX ETC

T

PRODUCTION• AT 1070 MT/D

OFF

ON SPEC.AMMONIADELIVERYTO STORAGE

N2 BLOWTHRUP.G.H.

NH3

ABSORBSYSTEMON

POWERBUMPS

NOTES1070 MTD 109S MTD

I 1 I I I I I I I I | I I I I 1 I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I

Figure 2. Ammonia plant start-up.

isolated without incident. It was determined that thesight glass pressure ratings were not adequate.The vent valve upstream of the CO2 absorber failedwide open when the mechanical actuator linkagefailed. Syngas compressor section pressure droppedfrom 2.6 MPa (380 psig) to 1.2 MPa (180 psig). Thecompressor antisurge valves opened and the ventvalve was manually closed, preventing the syngascompressor from going into surge.The boiler feedwater valve to the 10.3 MPa (1500psi) steam drum malfunctioned. Level was main-tained through manual operation of the level valvebypass.Low temperature shift (LTS) converter inlet temper-ature control was sluggish. Wide swings were expe-rienced at load changes. As a result, the operators putthis system on manual control. Boiler feedwaterflow, as shown in Figure 3, is regulated by a three-way valve which divides the total boiler feedwaterflow into two streams: one through the shift section

boiler feedwater preheaters and one through the fur-nace preheat coil. When LTS inlet temperature ishigher than the setpoint, the controller was designedto increase the flow through the boiler feedwaterpreheaters by diverting flow from the furnace pre-heat coil. The preheat coil outlet temperature con-troller was designed to activate the boiler feedwaterbypass valve around the first preheat exchanger lo-cated downstream of the low temperature shiftvessel.

It was found that the response to variations in LTSinlet temperature through adjustments in the boilerfeedwater flow split was too slow to maintain ade-quate control of LTS catalyst temperature, espe-cially when putting on or taking off process steamgenerators which had significant impact on boilerfeedwater flow. The control system was studiedusing the dynamic simulator developed for operatortraining. The modified control system indicated inFigure 4 was found to be very effective on the simu-

MODIFIEDORIGINAL f-VWV-f

PROCESS GAS

Figure 3. LTS temperature control system—original.

PROCESS GAS

HTS EFFLBFW

PREH EATER

Figure 4. LTS temperature control system—modified.

24

lator, and was then implemented on the plant. Thisuses the bypass around the first preheater down-stream of the LTS to adjust boiler feedwater temper-ature entering the LTS inlet boiler feedwaterpreheater. The total boiler feedwater flow split ismaintained constant unless the furnace preheat coiloutlet temperature exceeds a preset maximum, inwhich case more water is directed to the furnace pre-heat coil.

• The refrigeration compressor tripped out severaltimes due to high level in the suction knockoutdrum. It turned out that the liquid level in the pro-cess gas chiller was too close to the internal demisterE ad (Figure 5) resulting in flooding and carryover of

quid ammonia into the downstream knockoutdrum. This problem was solved temporarily byoperating at 10% level on the refrigeration chillerlevel controller. Subsequently, the level controllerwas remounted so that level could be maintained atnormal 50% of instrument range. Although the prob-lem was related to design of the chiller, it was solvedby external means without necessitating a shut-down.

On February 21, 15 days after feedgas introduction, theunit was delivering on-spec ammonia at the rate of 1050metric tons per day (1157 ST/d). However, another powerfailure hit at 5:00 a.m. on February 23 just as the remainingprocess gas vents were being closed. The power failurelasted 20 seconds and affected the steam system, Catacarbsystem, methanator, and converter. The unit was able toride through this upset, thanks to the design philosophywhich provided critical pumps and the control system in-dependent of the imported power supply. Another powerbump occurred at 10:25 a.m. of this same day and again theunit was able to continue operation. In fact, the startupheater was taken offline at 1:00 p.m. ofthat same day.

The fact that the plant continued to operate, safely andwith minimum interruption, through several upsets dur-ing the startup period can be attributed largely to the skillof the operators and to the pre-startup training they re-ceived. This training, involving the Exxon dynamic simu-lator, anticipated such upsets and gave the operators achance to practice their skills on the computer.

The commissioning activities continued subsequent tothat date as sections such as the purge gas recovery unitand the molecular sieve syn gas driers were broughtonline. The performance test was conducted at 100% de-sign feedrate on March 21. Subsequently, the plant has op-erated at over 1815 metric tons per day (2000 ST/d).

Service Factor

Since the initial startup, the service factor for the ANCammonia unit has been excellent. Excluding outage timedue to market limitation in mid-1983, the service factor forthe first 1-1/2 years of operation was 95.5%. In addition,the plant experienced only one shutdown through thefirst 8 months of 1984, accounting for only 3-1/2 hours oflost production. This compares very favorably with the in-dustry average for large tonnage plants, reported in the lastAIChE survey as 87.2% for the period of 1977-1981 [2]. In

DEM1STER PAD

fact, ANC's service factor places it within the top 20% ofthe plants covered in the survey. The ANC ammonia plantservice factor becomes even more impressive when it isconsidered that the period covered is theirs* 1-1/2 yearsof its operation. A comparison of ANC plant performancewith the industry average is shown in Table 1. A detaileddistribution of downtime causes, including a comparisonwith industry averages, is shown in Table 2.

The majority of the failures encountered at ANCoccurred during the first 10 months of operation, and fallinto the "major equipment" classification. Others haveranged from plant-wide power failures to "infant mortal-ity" failures of electronic instrumentation. Some of theproblems experienced have resulted in design changes orspecial precautions being taken to avoid future unit shut-downs. Examples follow:

Ammonia Converter Head Gasket Leakage

The ammonia converter is a Topsoe Series 200 radialflow converter basket, contained in a pressure shell hav-ing an internal diameter of 3.23 m (127 inches). The pres-sure shell, designed by Struthers Wells, has a full diameterclosure. Sealing is accomplished by a double conical gas-ket whose sealing capability is dependent upon the magni-tude of the synthesis loop gas pressure. The gasket ar-rangement consists of a low-chrome steel ring and twoaluminum sheets; one sealing against the shell flange andthe other against the top head. Design calculations pre-dicted complete sealing at synthesis loop pressure of 1.17MPa (170 psig) and above by means of deformation of thegasket ring. However, the converter head gasket leakedwhile at pressures above 1.17 MPa (170 psig) during initialsynthesis loop startup. Once the pressure reached designoperating conditions, the leak stopped. With successivepressure cycles, however, the leakage became worse. Bolttension was increased in an attempt to reduce the amountof leakage. Finally, during a subsequent startup, the leak-age became so extensive that it could not be eliminated byincreasing process pressure. The unit was shutdown andthe converter head was removed.

Extrusion and excessive thinning of the aluminumsheets had occurred due to over-tensioning. Review of thedesign calculations revealed that:

• The thickness of the gasket ring web was rounded upfrom a calculated 36 mm (1.42 inches) to 51 mm (2.0inches). This increased strength reduced the gas-ket's ability to elastically deform and seal at lowpressure.

• The mill test certificate for the ring material showedthat the actual yield stress was 344.7 MPa (50,000psi) compared to the design value of 206.8 MPa(30,000 psi). While the gasket thickness should be aslow as possible to enhance its ability to deform elas-tically at low pressure, it must have a minimumthickness to avoid yield. The higher strength mate-rial used would have permitted a thinner, more elas-

TABLE 1. SERVICE FACTOR* COMPARISON WITH INDUSTRY

ANC Ammonia Unit, RedwaterMarch, 1983 through August, 1984

Large Tonnage Plants—AverageNorth America (1977-1981)Total World (1977-1981)

Best Plant, North America (1977-1981)Largest Scale Plants, Total World (1977-1981)

(1400 to 1650 TPD)AverageBest Plant

95.5%

91.2%87.2%96.9%

87.7%94.1%

Figure 5. Liquid level problem in chiller.*Service Factor is based on total downtime less gas curtail-ment/market downtime.

25

ClassificationI Instrument failures

II Electrical failuresIII Major equipment failures

IV Preventive maintenanceV Other

Controllable downtime

VI Gas curtailment/market

Total downtime

TABLE 2. CLASSIFICATION OF DOWNTIME(Days/Year/Plant)

ANC Ammonia Unit(March 1983 Through Aug. 1984)

0.70.6

10.92.70.8

Large Tonnage PlantsNorth America Total World

(1977-1981) (1977-1981)

15.7

17.0

32.7

1.10.8

15.313.71.2

32.1

12.2

44.3

1.71.6

19.121.82.5

46.7

8.5

55.2

tic gasket, which would have improved its ability toseal at low pressure.

The ammonia unit was successfully restarted with aspare gasket of the same design. The head was set back onthe converter with careful attention paid to prevent over-tensioning of the bolts. Leakage occurred at pressures ap-proaching 5.5 MPa (800 psig), but the gasket sealed effec-tively at higher pressures. Subsequently, the gasket ringhas been redesigned and a replacement fabricated. Thick-ness of the middle portion of the ring has been reduced to29 mm (1.125 inches) to permit greater elastic deformationunder pressure, and the area of the sealing surface hasbeen increased to permit a greater initial bolt load.

Coupling Lock-Up

The synthesis gas compressor train contains two com-pressor casings, located at either end of a 10.3 MPa (1500psi) steam turbine. After nine months of trouble-free oper-ation, increased vibrations at the inboard bearing on thelow pressure case were noticed. The first indication of theproblem was at a vibration level of 0.038 mm (1.5 mils).Within two weeks the vibration had increased to 0.061 mm(2.4 mils). Signature analysis and gap voltage measure-ments indicated that there was no loss of bearing babbit. Acoupling problem was suspected and the compressor wasshutdown.

In the design stage, diaphragm-type couplings werespecified for reliability considerations. However a lubri-cated gear coupling was selected for one location where itwas felt that the axial shaft end expansions would exceedthe capability of conventional diaphragm-type couplings.It was this particular gear coupling that was now indicatingproblems.

On inspection, both the hub and cover teeth showed ma-jor pitting and deep rim markings on the tooth profile. Thespare coupling was installed and the compressor train wassubsequently restarted successfully. Pieces of the failedcoupling were sent to the manufacturer for analysis. Theirconclusion was that the cause was an inadequate numberof oil circulation holes drilled in the coupling, resulting inpoor lubrication and overheating. The design did call forthe correct number of circulation holes, and indeed thespare coupling which was installed did have the correctnumber of oil holes. However, as a contingency action, adiaphragm coupling for this service has now been de-signed and procured.

Steam Generator Tube to Tubesheet Leaks

A 10.3 MPa (1500 psi) steam generator is located imme-diately downstream of the ammonia converter. This ex-changer has a stiffened (thin) tubesheet design, with a

tube sheet thickness of 21 mm. It is a horizontal U-tubebundle with ferrules on the inlet process gas side to protecttube to tubesheet welds against nitriding. Tubes arewelded to the backside of the tubesheet as snown on Fig-ure 6. Both the tubes and tubesheets are fabricated from2-1/4 Cr-1 Mo material, with Inconel overlay on thetubesheet.

In May of 1983, high ammonia content in the 10.3 MPa(1500 psi) steam system was detected, indicating leakageof gas from the synthesis loop into the shell side of thesteam generator. During a unit shutdown, the steam gener-ator was inspected. Six tube to tubesheet welds had failedand the six tubes were plugged. Samples were taken to de-termine the cause of failure. When the unit was restarted, ahydrogen analyzer was installed to monitor for synthesisgas leakage. Hydrogen levels of 180 ppb were detectedand the unit was inspected at the next shutdown opportu-nity in July. One additional leak was found and the tubeplugged. Current hydrogen levels, after an additional yearof operation, are close to 400 ppb. This is indicative of oneor two more leaks, but the leakage is not causing anyoperating difficulty.

Analysis by Exxon of the cracked welds revealed twotypes of failure. One failure was due to hydrogen inducedcracking initiating from the gas side in weld metal withhigh hardness. Subsequent liquid penetrant inspection ofthe outlet ends of the tubes indicated that this was an isola-ted failure. It is concluded that the weld had been repairedafter post weld heat treatment.

The second, more common, type of failure was stress as-sisted oxidation initiating from the waterside at surface ir-regularities on the weld; that is, stress concentrationpoints. Cyclic cracking of the protective magnetic layer onthe waterside permitted the corrosion to promote crackpropagation. The cause of the stress assisted oxidation hasnot been proved by the available evidence but it is be-lieved that the surface irregularities in the weld, combined

INCONEL 60OOVERLAY

TUBESHEET 2 '/, Cr. 1 Ma

TUBES Z '4 Cr. 1 Mo.

Figure 6. Converter effluent steam generator tube-to-tubesheet weld inoriginal unit.

26

with the stress field, led to failure of the protective mag-netic layer.

While there are many similarly designed 10.3 MPa(1500psig) steam boilers operating successfully, including onein this ammonia unit, they are not identical because theprocess pressure is lower than the steam pressure and theyare of the straight tube design. There are a few 10.3 MPa(1500 psig) steam boilers in synthesis gas service, wherethe gas pressure is greater than the steam pressure, andsome of these have had tube-to-tubesheet weld failures.

Based upon the investigation, the history of this steamboiler, and the risk of unreliable operation, a spare steamgenerator has been ordered. The vendor assisted in theanalysis of the cause of the failure and in the developmentof an improved design. The new exchanger will be of thesame basic design except: (1) the tubes will be welded toan Inconel overlay on the face of the tubesheet, (2) at leastthe first weld pass will be welded with the automatic TIGprocess, (3) the tubes will be hydraulically expanded intothe thin tubesheet, (4) insulation will be added to the tubeinlet area of the tubesheet, (5) tube wall thickness will be4.5 mm instead of 3.6 mm, (6) the tube-to-tubesheet weldswill be radiographed, and (7) the base metal tubesheetthickness will be increased from 21 to 24 mm.

CONCLUSIONS

The very successful startup was even more remarkablein view of the fact that this is a large, highly integrated,novel ammonia process design. Industry experience, in-cluding all Exxon Chemical experience, would have pre-dicted many more unanticipated problems or procedural

errors that should have extended the length of time beforefirst ammonia production, or would have reduced its Ser-vice Factor. The fact that these did not occur can be attrib-uted to the following:

• Carefully designed and engineered process and me-chanical systems (the intensive effort on reliabilityin the process design phase).

• Careful selection of machinery, equipment, and ma-terials (owner's participation in the contractor'soffice, the design reviews, and the machineryaudits).

• Extensive involvement of the owner's team (person-nel with fertilizer plant expertise, drawn from the ex-isting plant organization, who participated through-out the project from the development of the basicdesign, through the detailed design, procurement,construction, and startup activities).

• Operator training (classroom as well as simulatortraining) to allow the operators to practice rapid re-sponse to anticipated upset conditions. In fact, thedynamic simulator is continuing to be used for newand refresher operator training, as well as for studiesin improving operating efficiency.

LITERATURE CITED

1. DeMena, H. F., J. A. Litwiller, M. P. Simpson, W. Unruh, "Ni-trogen Complex: Ammonia Plant Operator Training Simula-tor," Ammonia Plant Safety and Related Facilities, Vol. 24,1984, p. 171.

2. Williams, G. P., W. W. Hoehing, "Causes of Ammonia PlantShutdowns: Survey IV," Chemical Engineering Progress, 79No. 3, 11 (1983).

Philip A. Ruziska is Chief Engineer with the Agri-cultural Chemicals Technology Division of ExxonChemical Company in Florham Park, New Jersey.He is responsible for developing and applyingnew agricultural chemicals process technologywithin Exxon Chemical and for licensing. He re-ceived his BS and MS degrees in Chemical Engi-neering from the Massachusetts Institute of Tech-nology and is a member of AIChE.

R. A. Wilkinson graduated from the University ofAlberta with a B.Sc. in Chemical Engineering in1976. Rejoined Esso Chemical (Canada) in June of1976 at the Agricultural Chemicals Complex(ACC) in Redwater, working in process and projectengineering functions. From 1980 throughmid-1983 he acted as an owner's engineer in theAlberta Nitrogen Complex project. He is currentlytechnical supervisor for the nitrogen producingunits at ACC.

Cook C. Song is an Engineering Associate at theEsso Chemical Canada Agricultural ChemicalsComplex near Redwater, Alberta. He is responsi-ble for optimization and retrofit/debottleneckingof both large ammonia plants at this location, plusprovides technical support to plant operations. Hegraduated from Seoul National University in 1959(B.S. in Chemistry) and Hanyang University in1967 (M.S. in Chemical Engineering), both inKorea. He is a licensed professional engineer inthe province of Alberta and a member of AIChE.

William Unruh has worked for the last 28 years inthe agricultural chemicals industry. He has spent10 years with Esso Chemical Canada as Plant Man-ager, Nitrogenous Department. He is employedwith Syncrude Canada Limited as ProductionManager, Extrusion Department. He received aBS Degree in Chemical Engineering from the Uni-versity of Alberta in Edmonton, Alberta.

27

DISCUSSION

QUESTION: Is your simulator a real time simulator?And would this simulator also be suited to make anadvanced control system that enables you to make acloser control and save energy? Secondly, concerningyour waste heat boiler in the synthesis loop: As far as Ihave seen, the one you have in service is a boiler with awelding behind the tubesheets. There are, therefore,no gaps between tubes and tubesheets so some gapcorrosion problems are avoided. You are going backand having again a boiler which is welded on top of thetubesheet, if I have seen it correctly. Don't you feel thatyou would encounter problems in the gaps betweentubesheets and tubes?

P.A. RUZISKA: The dynamic simulator can be op-erated in multiples of real time, but for operatortraining it is used in real time. We have developedimproved or advanced control schemes, tested themon the simulator, and then installed them on the realunit. So, it has been used as a tool for testing anddeveloping improved control strategies that were laterimplemented on the actual unit. The next questionpertained to the synthesis loop waste heat boiler. Therevised design will have a weld at the face of thetubesheet, but the tubes will be hydraulically expandedinto the tubesheet to eliminate crevice for potentialattack.

WALTER GOERS, Goers Assoc.: Is it 26 million Btu(27 GJ) per metric ton or per short ton?

RUZISKA: Twenty-six million Btu (27 GJ) per shortton.

ROGER PARRISH, Parrish Assoc.: Would you care tocomment on the 26 million Btu? Is this a net numberthat includes, say, any electric drives? Does It include,say, refrigerating all the product going to storage?Storage temperature? Are you exporting steam? Doesit include feed-gas compression? Does it includecooling tower—all the usual things?

RUZISKA: The 26 million Btu is a net figure thatincludes production of -28 °F (-33 °C) liquid am-monia, all the energy to operate the associated utilitydrivers, a credit for steam export and a debit forelectricity, which is debited at the rate of 9,000Btu/kWh (13 MJ/MJ).

H.G. ORBONS, Materials and Corrosion ResearchDepartment: Regarding the affair of the tubesheet: Inyour paper it is given that the tubes are from 2% Cr-1Mo steel and you use an overlay of inconel. Did yougive this weld heat treatment?

RUZISKA: Yes.

RON WILKINSON, Esso Chemical Canada: Thosetube to tubesheet welds were heat-treated in thefactory.

28