1978: Operating Experience with a Cryogenic Syngas Purifier

Operating Experience with aCryogenic Syngas Purifier

The attraction of the purifier design lies in the aspect of energy conservation.

W.D. VerduijnEsso Chemie, B.V.

Rozenburg, The Netherlands

Esso Chemie B.V. operates a 1,360 metric t/d ammoniaplant in Rozenburg, The Netherlands. The design by C.F.Braun and Co., is of the purifier concept. Syngas compres-sion is preceded by a cryogenic treatment to control thehydrogen-nitrogen ratio and to purify the syngas from mostof the methane and argon. This article discusses the techni-cal experience obtained over 10 years operation with thislarge purifier.

Figure 1 shows a typical over-all flow plan of an ammoniaplant with a cryogenic syngas purifier (coldbox). Some 30%more than the stoichiometric amount of process air is reactedin the secondary reformer. Excess nitrogen, together withmost of the argon and methane, is removed in the coldboxupstream of the synthesis loop. The coldbox itself is preced-ed by driers, removing residual quantities of water, ammoniaand carbon dioxide that could freeze out in the subsequentstep.

The high purity syngas produced by the coldbox requiresonly a slight synloop purge. Ammonia is scrubbed from the

Figure 1. Flow plan for an ammonia plant withcoldbox.

purge gas and recovered. The ammonia-free purge stream re-enters the main flow upstream of the driers. The coldbox alsoproduces a nitrogen stream containing the scrubbed argon,methane and a (small) equilibrium amount of hydrogen.Since this stream is completely dry, it is used to regenerateone drier during the adsorption cycle of the other. The gas issubsequently used as primary reformer fuel.

The coldbox consists of two sets of brazed aluminum platefin heat exchangers, a gas expander, a rectifier column, acondenser and a control (Joule-Thompson) valve. Figure 2shows the operating principle. The feed gas, containing some33% nitrogen, is cooled to a temperature around -140° C(-220° F) in the first heat exchanger section, by heatexchange with the two effluent streams. It then passesthrough the turbo expander to provide the required refrigera-tion. The feed is subsequently cooled to around -175° C(-285° F) in the second section of the heat exchanger andflows then into the rectifier column where the liquid portion(mainly nitrogen) drops into the bottom and the vapor flowsupward.

Liquid from the bottom is tlashed at reduced pressure viathe Joule-Thompson valve; the resulting drop in tempera-ture provides the driving force in the condenser. Reflux fromthis condenser washes methane and argon out of the upflow-ing vapors. As a consequence, the rectifier produces ahydrogen-nitrogen mixture in the ratio 3:1, with only littleresidual argon and methane. This stream and the waste gasstream from the condenser flow upward through both ex-changers and cool the feed.

The coldbox has two controls. The material balance (as in-dicated by hydrogen-nitrogen ratio in syngas product) ismaintaned by controlling the flow of liquid from the bottomof the tower. The heat balance (as indicated by a constantliquid nitrogen level in the bottom of the tower) is maintainedby controlling the speed of the expander. Both controls aremanual at Esso Chemie B. V. The exchanger cores are instal-

130

. W A S T E GAS TO FUEL SYHCAS PRODUCT __

Figure 2. Flow plan for the cryogenic purifier.

led in a perlite-filled box; the tower is installed in a separatebox, filled with perlite as well.

The main attraction of the purifier design lies in the as-pect of energy conservation. Without pretending to be com-plete, the following examples can be mentioned:

Use of excess air in the secondary reformer reduces thefuel requirement in the primary reformer greatly. Althoughair compression requires more energy in this case, the asso-ciated gas turbine driver produces so much hot exhaust air,that no ambient combustion air is required in the reformerfurnace. High purity synthesis gas allows a minimal synlooppurge and consequently a minimum waste of associatedhydrogen.

The resulting energy requirement per ton of produced am-monia is lower than for conventional plants. Although thiswas a positive feature in 1967, when the plant was built, ithas become increasingly attractive after the energy crisis of1973.

Regarding capital costs, the added drier and purifier areoffset by savings in the primary reformer and synthesis loop.Regarding the service factor, the driers and coldbox are addi-tional links in the chain from the reformer to the product-tankand cause their share of interruptions. These are offset, how-ever, by fewer interruptions upstream and downstream,because of less severe conditions and because of protectionof the synthesis section from upstream fluctuations. Shortlywe shall deal with Esso Chemie B.V.'s efforts to minimizeinterruptions attributable to the purifier.

Oil fouling during a tripFigure 3 shows a sketch of the lube/seal oil and seal gas

system of the cryogenic expander. Process gas enters the ex-

pander at a temperature around -140° C (-220° F). It flowsthrough the adjustable inlet vanes and the rotor to the lowercryogenic cores, as shown in Figure 2. Lube oil and seal oilenter the expander in a lower section. The dynamometer,anoil brake that destroys the generated energy, is mounted onthe same shaft.

The oil systems and cold process gas systems are separatedfrom each other by a small seal gas stream; its pressure isslightly higher than either of the two. The seal gas flows par-tially through the upper part of the seal into the cold processgas stream, and partially via the lower seal part into the re-turn line (the so called high-pressure drain).

If one reduces the seal gas pressure sufficiently, the seal-ing effectiveness decreases. If the HP drain pressure is high-er than the process pressure, oil will enter the expander and iscarried away into the process. If the process pressure is thehighest,cold gas will enter the HP drain and, if of sufficientquantity, freeze the oil in the drain ("back freezing"). Backfreezing can be followed in turn by oil fouling, since it blocksthe way out for the oil.

While oil fouling has not been a problem in most otherpurifier plants, it was a major difficulty in Esso ChemieB.V.'s coldbox in the first years of operaton. In fact, theproblem is twofold.

Oil that enters the process freezes out immediately aroundthe adjustable inlet vane system. Finally the actuator is nolonger able to move the vanes, resulting in operation withvanes in a stuck position. That means that one of the cold-box controls falls away, resulting in either the requirement todrain liquid nitrogen from the tower, or to drop to lower feedrates in order to find a new equilibrium condition.

The second problem is that oil which does not freeze out inthe expander moves into the lower cryogenic cores, where itis retained. The resulting drop in heat transfer pulls the tem-peratures at the exchanger's warm-end down. The more

to lowercores

process gas

(DYNAMOMETER)

Figure 3. Seal system for the cryogenic expander.

131

severe the fouling, the bigger the drop in temperature. Tem-perature can drop so low that condensation takes place in theexpander. Since compensation of the degrading exchangerefficiency with pressure drop across the expander is onlypossible up to a certain limit, fouling means often reductionof capacity or even shutdown.

This should make it clear that maintaining positive seal gaspressure drop is of vital importance to a good operating cold-box. Seal gas is therefore dry raw syngas, taken from a pointjust downstream the driers, where the pressure is alwayshigher than below the rotor of the expander.

However, Esso Chemie B. V. has found that this system, atleast in our plant, was sensitive to pressure swings. Especial-ly during frontend trips, the seal gas pressure drops so fastthat positive sealing is not always maintained. It was oftenfound that a previously clean coldbox appeared to be in foul-ed condition during restart-up. A complicating problem wasthat one could only conclude that the box was fouled after therecool-down was in an advanced stage; low temperatures onthe warm-end of the lower exchanger and a decreased orabsent ability to generate a liquid level in the tower were theindicators of fouling. In a substantial number of cases, thecoldbox had to be warmed up again to drain off the oil. Sub-sequent cool-down resulted then in a normal operating cold-box.

It is obvious that this procedure of cool-down/warm-up/cool-down was expensive, both in terms of vented gas and inlost service factor time. In the early years of operation, wehave therefore decided to warm up the coldbox regularlyafter a trip.

Finally, an automatic emergency seal gas cut-in systemwas installed. It consists of a number of full nitrogen bottlesand a pressure differential switch. When the pressure differ-ential between seal gas and process gas drops below a certainvalue, the switch is activated and supplies a powerful streamof nitrogen from the bottles to the seal. At the same time, the"common trouble" alarms in the plant and control room areactivated. In such a case, the responsible operator has 10minutes to take corrective action corresponding to the hold-ing capacity of the installed number of bottles.

The present system cuts in regularly during front-end trips.No more front end trip-induced oil fouling has been noticedsince its commissioning.

Oil fouling during normal operation

Although fouling as a result of a trip was eliminated afterinstallation of the emergency seal gas system, subsequentlysudden fouling was found during stable operation. As usual,temperatures at the warm-end of the lower exchanger drop-ped to values sometimes close to the cold-end temperatures,although it was simultaneously concluded that seal gas/process gas and seal gas/HP drain pressure differentials wereabsolutely positive. At other moments, the HP drain tem-perature dropped sharply. Only with extreme efforts couldthe oil be prevented from freezing up and blocking the drainpipe. To stay in operation, for instance, oil supply tempera-tures to the bearings were raised to extremely high values

(80° C = 175° F). Also the oil flow across the expander wasincreased. Finally the (cryogenic!) expander and high pres-sure drain pipe were steam-jacketed. All these effortsresulted in keeping the HP drain temperature barely above20° C (70° F). Surprisingly, the problem disappeared some-times without a reasonable explanation, and came back againunexpectedly.

Since it was apparent that the problem was caused by back-flow of the cold process gas into the HP drain, the seal instru-mentation system was improved. Pressure differentialrecorders were hooked up between seal gas/process gas andthe seal gas/HP drain to exclude any overlooked excursion.The seal gas pressure measuring point was initially located inthe seal gas supply line; an extra orifice bore was drilledthrough the expander into the seal gas cavity to make surethat the correct seal gas pressure for AP recorders wasmeasured.

The casing drain sensing point (process pressure) wasswitched over regularly to a point just downstream theexpander to exclude possible influence of the pressure mea-surement by frozen oil in the drain. However, after thesemeasures were implemented, data showed no change what-soever from the previous situation. The problem continued.

To exclude an inaccuracy in our assembling of the cryo-genic expander, the supplier was asked to superviseassembly and to adjust the seal clearances to mutually agreedvalues. No abnormalities were found in the machine nor inthe way it was assembled. This virtually excluded problemswith the seal and with the expander itself.

As the situation became worse, seal gas temperature wasincreased from 7 to 70° C (45 to 160° F). For test purposes,the seal gas flow itself was also drastically increased. Con-trary to the expectation, the test did not show the desiredincrease in HP drain temperature. Neither did the seal gas/HPdrain differential pressure rise appreciably. Apparently thevast amount of seal gas almost quantitatively flowed throughthe seal into the process stream as that AP increased appreci-ably.

Based on this finding, we concluded that cold gasbackflow through the seal to the HP drain was not the causeof the problem. As conduction through the shaft was notenough for a plausible explanation, internal bypassing ofcold fluid through the expander was suspected again.

To test that, seal gas was entirely replaced by bottled nitro-gen. Subsequent analysis of the gas flowing through the HPdrain showed no pure nitrogen, but quite som hydrogen. Thecomposition of the HP drain gas showed also relatively highmethane figures.

The conclusion was obvious: internal bypassing of coldgas and cold liquid took place through the expander body intothe HP drain. The presence of liquid can be explained by thelow temperatures on the warm side of the cold exchanger,caused by previous inflow of oil. The gas leaving the expand-er must have been within the condensation area. That back-freezing sometimes stopped must have been caused by asmall raise of the HP drain pressure or a small drop of the pro-cess pressure: the driving force for inflow of cold fluid is thenreduced so much that the effects are hardly disturbing. Note

132

cores

Figure 4. Present instrumentation for the seal gassystem.

waste gas from

condenser

V » L V» L V» L V » L

t t t f

8" DIVISION HEADER( ONLY 1 OUT 2 SHOWN )

SITUATION AT DISPERSED FUDW

f tlondensor

SITUATION AT SEGREGATED FLOW

Figure 5. Waste gas phase maldistribution.

also that in case the HP drain pressure exceeds the processpressure, the leak path would be reversed and oil could flowinto the process.

In cooperation with the vendor of the expander, the leakpath was identified and sealed off. Insufficient sealing forceapparently caused a clearance between heat barrier andjournal bearing support during operation. The resulting HPdrain pressure in that clearance deformed an O-ring, causingHP drain oil and gas to flow via the clearance and O-ringthrough the process gas pressure measurement bore into thespace below the rotor. During flow reversal periods, the O-ring must have stuck in deformed position, thus allowingbackfreezing. A full face gasket was installed between theheat barrier and journal bearing support. Gasket force was in-creased by doubling the amount of bolts. The process pres-sure measurement bore below the rotor was plugged off, anda new measuring tap downstream the expander was takeninto operation.

Extra pressure recorders were placed on the process gasand HP drain taps of the expander. Their values are kept closeto each other with the process gas value slightly higher. Thatwill give a slight and acceptable inflow of cold fluid into theHP drain in case the sealing system fails again duringoperation. Figure 4 shows the present seal gas system instru-mentation.

Since these actions, no more problems of the naturedescribed above have occurred.The loss-of-level phenomenon

After the oil inflow problems at a trip and at continuousoperation were corrected, further fouling of the lower ex-changers could no longer be a problem. A smooth operationwas expected, and, indeed, sometimes realized for consider-able periods. However, from time to time a new problemcame up, expecially at the higher feed rates that were reachedover the years. The phenomenon showed up, again, as a dropin warm-end temperature of the cold exchanger, causing loss

of liquid nitrogen level and the requirement to increase theexpander pressure drop and decrease the feed rate to remainin operation. These symptoms are identical to the ones ob-served during oil fouling and in the past we must have mixedthem up regularly.

The phenomenon strikes suddenly during feed rateincrease at high throughput levels (mostly over 100%design). Dropping below the feed rate where the problemstarted does not restore the old situation: the phenomenonshows hysteresis. Only when feed rate is dropped far belowthe original level can a return to the initial situation be ob-tained. Apparently the coldbox switches at a certain pointfrom one stable condition to another. But only the first stablecondition is the desirable one; the other costs energy and pro-duction.

We believe that both conditions may be caused by,respectively, the absence and the presence of waste gas phasemaldistribution among the parallel cores of the lower ex-changer. To illustrate that, a sketch has been drawn of theheader that supplies waste gas to four cores (Figure 5). EssoChemie B.V.'s coldbox has eight parallel cores; the secondheader is identical to the one as shown in Figure 5. In the onestable situation, the waste gas liquid and gas are dividedequally among all cores. In the second stable situation, mal-distribution has taken place in which the last two cores ofeach header receive mainly liquid and the first two mainlygas. This last situation is the unpreferred one, causing thephenomenon that so much resembles fouling.

Obviously, the duty of the condenser plays an importantrole in this matter. Our calculations show that the measuredduty lies below design, resulting in a higher liquid/vaporratio than design in condenser effluent. The piping systemfrom condenser to cold cores is designed for two-phase flowand would probably have been adequate if the condenser per-formance would have been according to design.

The above analysis of waste gas phase maldistribution is a

133

lalct ttmpcrtture

postulation. Although operators of similar purifier units donot report experience with this phenomenon, our design isnot identical to theirs. Measuring equipment, to be installedin the next turn-around, should give the final proof. Our cal-culations have, however, shown that the predicted tempera-ture drop in the maldistribution case agrees with the one mea-sured in practice. Further, there is support for the postulationin the field of flow regimes.

Corrective action is sought either in increasing thecapacity of the overhead condenser (lowering the liquid por-tion to the cores) or in a better distribution system for liquidand gas upstream of the cores.Liquid nitrogen injection

During periods of oil fouling, the service factor of theentire ammonia plant is in danger. At continued fouling,eventually a point is reached where operation is no longerpossible: Both increase of the expander pressure drop and de-crease of feed rate reach their acceptable limits. In such acase, liquid nitrogen injection can save the plant. Nitrogen,from an onsite liquid nitrogen storage drum, is injecteddirectly through a drain connection into the exit line of theJoule-Thompson valve. Doing so, the level in the tower is re-stored. Consequently, one can go back to the initial plantfeed rate, under a continuous supply of liquid nitrogen. Onecan even supply so much more nitrogen as is required toreduce the expander pressure drop to a desired level.

In general, the injection rate is a matter of economics. EssoChemie B.V. feels it is optimal to go back to a few percentunder the initial feed rate, with simultaneous increase of air/feed ratio and expander pressure drop. Nitrogen injectioncures therefore the service factor aspect and most of theproduction rate aspect of oil fouling and waste gas phase mal-distribution. But a certain production debit and a nitrogen ex-penditure, both appreciable in absolute value, do remain.

Another advantage of nitrogen injection is reducing cool-down time. Normal cool-down time from +10 to -180° C(+50 to -290° F) amounts to roughly 24 hr in our plant; it islimited by the capacity of the expander. In order not to shockthe rectifier tower and to avoid temperature stresses in therectifier condenser, the tower is permitted to cool-down to-100° C (-150° F) using the expander. At that moment, liquidnitrogen is pumped into the inlet line of the tower, where itflashes off. Thus both expander and liquid nitrogen pumpcool down the coldbox further, whereas care is taken not toexceed temperature changes in the tower of more than 50° C(90° F) per hour.

Figure 6 shows that the cool-down time of a completelywarm box can thus be reduced from 24 to 17 hr. That savesseven hr service factor time and seven hr syngas losses percool-down.

Cleaning lower exchangersIf the lower exchangers (the ones downstream the expand-

er) are fouled, the agent is usually oil. Because of our unfor-tunate experience in this field, Esso Chemie B.V. hascompared various cleaning methods. Since oil fouling startswith the cryogenic expander, feed passages of the lower ex-changer are bound to be fouled most. In principle that is true,although we also found oil in the bottom of the tower, the

Figure 6. Cool-Down with liquefied nitrogen.

waste gas side of the condenser and the waste gas passages ofthe exchanger cores. Since good care always was exercisedduring cleaning not to transport any oil from a fouled sectionof the coldbox into a clean one, we do not exclude that at leasta part of the oil transport takes place during normal (cold)operation of the box. Oil in the product passes was neverfound.

After the first incidents of oil fouling, warm-up of thefouled passages with syngas to 10° C (50° F) was enough toremove most of the oil. Depending on the situation, theexchanger cores only, or both the cores and tower werewarmed up.

Warm-up of the cores was executed at the highest possiblevelocity: Temperature changes of 100° C (180° F) per hourwere no exception. The tower was always warmed up muchmore gradually: A temperature increase of 50° C (90° F) perhour was never exceeded.

Principles of syngas blowing for cleaning purposes are: a)downflow to help outflow of oil by gravity effects, and b)section by section treatment to minimize oil transport fromfouled to clean parts of the box. Syngas from the driers isblown via the coldbox inlet valve through the feed passagesof the upper exchangers, the expander bypass, and the feedpassages of the lower exchangers to a blow-off connection ata low point. From there, it is routed via a piping system and amuffler to a safe high point into the atmosphere. As the corewarms up, oil can be drained from the low drain point in thefeed passages. A considerable amount of oil disappears,however, as a mist with the syngas. As the syngas flow in-creases, more oil comes loose. We therefore always maxi-mize our blow-out flow. Feed and waste gas passages aretreated similarly. The waste gas side of the condenser isblown as well.

After a number of successful blow-outs with syngas, cold-box performance showed gradually that not all fouling haddisappeared. This is probably dure to the fact that the lighteroil components tend to be removed, whereas the heavier onestend to stay. Therefore a more rigorous step was introduced:warm-up with natural gas to maximal 120° C (250° F). Allthree passages of the exchanger are fed with 120° C (250° F)warm natural gas via a newly designed inlet system. Gasenters the passages at the top of the upper exchangers andleaves via the piping system at the bottom of the lower ex-changers. In the starting phase, natural gas with a lower tem-perature is used so as not to shock the cores.

This warm-up makes residual oil less viscous. Most of it

134

stays, however, in place. A subsequent blow-out step(section by section) with as much syngas as possible, pushesthe oil to the low point drains. Most of it disappears, again asan oil-mist into atmosphere.

However, natural gas warm-up also did not give sufficientlong-term results. Therefore solvent flushing was applied asa last means, some years ago. Virgin naphtha (b.p. 80-1 IOC)was selected as a solvent over chlorohydrocarbons. At thattime, its fire hazard was accepted over the uncertainty of thecorrosion potential of chlorohydrocarbons. Calculationswere carried out to ensure that the cores and their supportcould take the extra solvent weight. The complete coldbox isflushed, section by section, through a circulation pump.Much attention was paid to the draining step. Subsequently,the box was warmed up to 120° C (250° F) with natural gas toremove the naphtha film and to boil off the liquid in an areathat could not be drained. That is done until naphtha-freenatural gas samples are found and subsequently continued for24 hr.

One-hundred-twenty liters of oil were removed during theflush, most of it from the feed passages of the lower ex-changers, the bottom of the tower, and the waste gas side ofthe condenser. Not withstanding the thorough dry-out proce-dure, trapped naphtha was found during various weeks afterrestart-up. Next time, Esso Chemie B.V. will use inhibited1,1,1,-trichloroethane as a solvent. Hopefully this willreduce the trapped solvent problem because of its lower boil-ing point. Apart from oil removal, the advantage of thesolvent flush has been that new oil fouling situations couldagain be acceptably coped with, by syngas or natural gaswarm up.

Core support-inspectionThe core batteries in the coldbox are welded to aluminum

angle profiles, which in turn, rest on horizontal stainlesssteel support beams. The beams themselves rest on anexcision in the perlite box. During cool-down the coresshrink; the temperature of the angle profiles lags behind.Consequently, the cores shrink faster than the angle profilesand the resulting stress is absorbed by the cores. During rela-tively many, rapid cool-downs, as has been the case at EssoChemie B.V., cracks can develop that may influence the in-tegrity of the cores and of the entire core support construc-tion.

On recommendation of the core manufacturer, we havetherefore inspected the core support construction after nineyears of operation. To that end, perlite was removed from thebox by vacuum trucks; with fluidized transport via thebottom manholes. The perlite was stored in a nearby tent.After the inspection step, it was reloaded; as perlite suffersfrom wear during handling, some 30% fresh additional mate-rial had to be purchased and loaded via the top manholes.

Inspection of the core support system revealed indeedcracks in the suspected locations. However, the extent of thecracks was such that action did not seem necessary. In coop-eration with the supplier, it was decided not to execute anycorrective action. Further outside inspection of cores andtower showed they were in excellent condition.

Rupture disc backblowingAs operation continues over the years, fouling material

tends to accumulate in the inlet passages of the upper ex-changer. Partially those foulants are volatile components,like water vapor and ammonia traces that slip through thedriers and freeze out; they can be easily removed byderiming. An other part is, however, solid material that set-tles in the cores; it is this type of fouling that is discussedbelow.

Each of the driers, located upstream the coldbox, isequipped with a discharge filter. That filter retains solid par-ticles of dessicant material and dessicant bed support mate-rial over the size of three micron (0.000003 m). Consequent-ly, particles smaller than that can pass and accumulate in theupper exchangers. The resulting loss of heat transfer showsup as an increasing pressure drop across the cryogenic ex-pander that maintains the heat balance of the box. (A pres-sure drop in terms of a decreasing free flow area in the coresas a result of fouling has never been demonstrated, although asensitive pressure drop measuring system was used). EssoChemie B.V. has found that the lasting pressure drop in-crease across the expander by solid particle fouling hasamounted to 0.5 - 1.0 kg/cm2 (7-14 lb/in2) per year.

Cores fouled with solid particles can effectively be cleanedduring plant turnarounds by applying the so-called rupturedisc backblow technique. To that end, the cores are firstwarmed up to ambient temperature. The inlet pipe isunbolted and removed. The common inlet flange gets with arupture disc, in our case always a layer of a few gaskets (seeFigure 7). The expander inlet and bypass valves are thenclosed, and the thus isolated inlet passages of the upper ex-changer are pressurized slowly with nitrogen. As the pres-sure rises, gaskets will rupture suddenly and the resultingnitrogen flow tears the solid particles loose from the ex-changer surface and carries them into the atmosphere. Thistechnique was previously applied by other syngas purifieroperators.

Rupture pressure can be set by varying the number andthickness of the gaskets. Esso Chemie B. V. pressurizes to 5or 6 kg/cm2g (70 to 85 lb/in2 gage) and repeats the treat-ment several times. The technique gives very good results. Inthe first five years of operation, for instance, the pressuredrop across the expander rose slowly from 2.5 to 5.5 kg/cm2

Figure 7. Rupture disc backblowing: gasket instal-lation.

135

Figure 8. Rupture disc backblowing: net installation.

(35 to 80 Ib/in2). Rupture disc backblowing did reduce thatfigure to somewhat below 3 kg/cm2 (40 Ib/in2). Now thetechnique is standard practice; it is repeated each one or twoyears.

Several safety precautions are taken during backblowing.Firstly, a number of support beams are welded against thebox behind the open inlet flange. Bolts, in turn, support theflange against these beams. The reaction force of the flange isabsorbed by the box, not by the core system. Secondly, earprotection should be worn, as sound levels during rupturingare considerable for a big flange like ours. Finally, a net is

erected in front of the flange to catch the blown out gasket;without such a net the gasket flies with high velocities overdistances up to 200 m (700 ft) through the plant, which is ahigh potential hazard to all present personnel. Note that gasjet and gasket forces released during rupture can be so high,that scaffolding around the net may bend (Figure 8).

Compensating operating advantages

The purifier does add another link in the chain for ammo-nia production and therefore adds the potential for associatedproduction interruptions. However, the purifier also has theadvantage of compensating for problems that may develop inthe other links. Fluctuations in the secondary reformer airflow rate, primary and secondary reforming methane slip,shift conversion CO slip, and C(>2 removal slip—all arecompensated for in the purifier.

The purifier alows milder conditions in the primary re-former, which leads to less operating problems there. It alsopermits economical operation of the plant when the LTS isseverely deactivated.

Because of the synthesis gas purity, synthesis catalyst lifeis extended. Esso Chemie B. V. has operated 10 years on theoriginal charge of catalyst.

Cryogenic processing of ammonia synthesis gas, in thesense of operating a coldbox as described above, can resultinitially in service factor problems. However, our experi-ence has indicated that these problems can be identified,analyzed, and largely eliminated. #

DISCUSSION

EISUKE WAOA, Asahi Chemical Industry Co., Ltd.:We have a similar type of cold box, though the size issmaller. Fortunately we have not had problems likeyou explained. My question is, first, what type of fil-ter are you using at your outlet of dryer?W.D. VERDUIJN, Esso Chemie, B.V.: We are using aglastex filter that retains particles up to 3 micron.WADÄ: We are using a glass filter of 3 micron, andhave had no pressure drop increase at the cryogenicpurifier. It was very constant during operation, andthere was no deriming or any fouling of exchangers.After shutdown, we open the bypass of oil trap forhigh pressure drain of the expander. Then, the pres-sure of the oil discharge from the expander isdecreased about 10 kg/cm2. After that we dry up theexpander inside with nitrogen. We have had no oilleakage or oil contamination of exchanger.VERDUUN: Yes, I'm glad for you. I'm only speakingfor my plant, and we have experienced this problem.But you are, I guess, referring to operating mistakesor things like that. Of course, we also have our prob-lems with operating mistakes but I haven't menti-oned them. I'm just referring to all our problemswhich have not been caused by operating mistakes.HENNINGS, Georgia Pacific: What has been yourexperience with flammables, like process gas, leak-

ing into the perlite of the box? Have you had muchtrouble with that?VERDUIJN: We have had absolutely no trouble withprocess gas leakage into the perlite. We haveinspected our cores, and we have also inspected ourtower sometime ago, and we found them in excel-lent condition.HENNINGS: Do you monitor for flammables, or doyou keep -a purge on the perlite?VERDUIJN: We keep a purge on the perlite, andfrom time to time we monitor for' what you callflammables.HENNINGS: One other question. Have you had anydifficulties, or what difficulties have you had, withthe flow control from the bottom of the rectifiercolumn? Have you had any level control problems inthat column?VERDUIJN: Not at all. The control works fine.ROY BANKS, Petrocarbon Developments, Inc.: Thissort of problem that you are describing on themolecular sieve dryers which relates to carryover ofthe molecular sieve was a typical problem on manygas processing systems and air separation plants.We have experienced this problem in the early days,but by installing the correct type of filter, it has nowbeen virtually eliminated. The sort of filter operated

136

on Petrocarbon purge gas recovery systems and onmany of our gas processing systems is a full flowtype filter, what is known as stillite, and in our expe-rience, it has solved this type of carryover problem.In some instances we have also installed external fil-ters, but I would like to make a comment that I don'tfeel that this kind of problem is typical of cryogenicsystems today.JOHN LIVINGSTONE, Imperial Chemical Indus-tries: I would just endorse those last remarks. In factwe have not yet had to clean the expansion turbinefor dust, and I wonder if in fact the problem mightnot be better tackled at the other end of the system.It's an intriguing process for cleaning, but I justwonder what is it doing to the seals in the bearingsof the expander? Certainly in our plant on the cry-ogenic unit, the seals on the bearings are very verysensitive, indeed, to the process system pressure.And to have that sort of shock, I shudder to thinkwhat would happen to our turbines. Have you hadany adverse effects as a result of this sort ofcleanup?

VERDUIJN: Yes, we think that we may have somewear on the bearings, and if necessary, we replacethem. Incidentally, we keep our expander and towerblocked during backbiowing, to make sure thatexclusively the cores are exposed to the shockwave.LIVINGSTONE: Yes, I think a good look at this filterup at front end might be of extreme interest to you.Certainly we have very good experience with it.

VERDUUN, W.D.

137