How to Regenerate Sulfur-PoisonedSteam Reforming Catalyst

Steam reformers occasionally experience problems of sulfur poi-soning, which, in nearly all cases, result in carbon depositing in thecatalyst bed. Based on several industrial cases, this article de-scribes safe procedures for removing the carbon and reactivatingthe catalyst

Jens HoukenHaldor Topsoe, Inc., Houston, TX 77258

SULFUR POISONING MECHANISM

The poisonous effect of sulfur on steamreforming catalysts is well-known bothfrom industrial experience and fromnumerous theoretical and experimentalstudies.

Interested readers are referred toRostrup-Nielsen (1) where the most up-to-date theories on the mechanism of thepoisoning are discussed with referencesto several other publications on thissubject.

All sulfur compounds reaching thereformer are hydrogenated into HLS whichis strongly chemisorbed to the nickelsurface.

Ni + Ni - S + H, (D

This chemi sorption takes place atsulfur concentrations several orders ofmagnitude lower than the concentrations

(PH2S^

is formed.

) at which bulk nickel sulfide

3Ni + 2H2S (2)

For any catalyst with a given nickelsurface area, it can be shown that for agiven ratio PH $/PH , a certain fraction

9 of the nickel surface area will becovered by chemisorbed sulfur.Rostrup-Nielsen (1) has found the fol-lowing correlation:

0S = 1.45 - 9.53 x 10"5 x T

+ 4 17 x 1 fl x T Y .Ai P /D C*.\T . i / A ±<J A l A -CM r,, c/ rii n ( 3 )

To use this correlation for calcu-lating the distribution of sulfur on thecatalyst in a tubular reformer is com-plex, because of the significant gradi-ents of temperature and hydrogen partialpressure in the tube - both in axial andin radial direction, not to speak aboutthe temperature and concentration gradi-ents inside the catalyst pellets.

If radial and pellet gradients areneglected, however, axial profiles ofsulfur coverage at equilibrium can easilybe calculated and the result is shown inFigure 1 for typical ammonia plantconditions. The decrease of sulfur fromtop towards bottom is the combined effectof the increase of temperature andhydrogen partial pressure. It is evidentthat there is no fixed limit of sulfurbelow which no poisoning will take place.

At first glance, Figure 1 may appearsurprising since the coverage does.notappear to be tremendously different for

19

sulfur concentrations, which we know willaffect the reformer performance and suchwhich are considered safe. It must,however, be remembered that the graphrepresents equilibrium conditions, andthat whereas the equilibrium coverage maybe reached within days or weeks whenfeeding gas with 1 ppm of sulfur, it maylast years with 0.01 ppm of sulfur.During that long period, the catalyst ismost likely to have been shut down andrestarted, at which occasions the cata-lyst is normally steamed, i.e. exposed toconditions where sulfur is removed fromthe catalyst (cf. paragraph 4).

z mFigure 1. Sulfur coverage (&s = sulfur/sulfur ca-pacity) of nickel surface of catalyst in primaryreformer in NH3 plant as function of Z (distancefrom top of catalyst). Equilibrium profiles calcu-lated for different sulfur contents (vol. ppm) innatural gas feed.

In practice, the reformer - whenoperated with a well-functioning desulfu-rization system - is not likely ever toreach the degree of sulfur coverage asdepicted in Figure 1.

Industrial cases of sulfur poisoningare thus normally caused by upsets in thedesulfurization unit rather than the slowaccumulation of sulfur which does takeplace at conditions considered normal andsatisfactory. Such upsets are typicallycaused by changes in the feedstock. Thehigh energy prices have definitely, inthe recent years, led to a diversifica-tion in the feedstock situation, whichhas given unpleasant surprises to manyoperators.

The result of sulfur poisoning is

well-known. Increasing tube skin tempera-tures and methane leakage are the firstsigns, but very often the situationdevelops into hot band formation or"giraffe-necking" as the case may be,followed by increasing pressure drop as aresult of carbon formation in the cata-lyst bed.

As long as the problem is only inthe first phase, and provided that thesulfur "breakthrough is stopped, theoperator will normally let the catalystregain activity by relying on the revers-ibility of the chemisorption reaction(1). This process, however, is slow,because the rate of the diffusion-controlled elution .decreases exponential-ly with time.

It may easily last two to threemonths before a plant has recovered froma sulfur poisoning.

Many operators will, in this situa-tion, operate* the plant at a highersteam/carbon ratio, hoping that this willaccelerate«the removal of sulfur. Thisis, however, hardly so, fcut operation onhigh S/C ratio is, of course, helpful incounteracting the effects of the poison-ing both in terms of methane leakage andpossibly carbon laydown.

If the poisoning has developed intocarbon formation and the related problemsof hot spots and delta p increase, it isoften not acceptable for the operator towait for the slow improvement to beexpected when returning to a normalsulfur level. The alternatives are achangeout of the catalyst or a regenera-tion, which can remove both carbon andsulfur.

CARBON FORMATION AND MORPHOLOGY

In steam reformers, carbon may be formedon the catalyst via different routes,leading to different forms of carbon(Rostrup-Nielsen (!_)).

The three main forms which have beenidentified are:

1. Whisker carbon, formed bycatalytic reactions on thenickel and growing like fila-ments on each nickel crystal.It is typical that the diameter

20

of the filament is the same asthat of the nickel crystal.

2. Pyrolytic carbon is the resultof gas phase non-catalyticpyrolysis of mainly higherhydrocarbons, but in extremecases also of methane and istypically found as dense shaleson the tube wall or completelyencapsulating the catalystparticle.

3. Film of polymers or "gum",encapsulating and thus deacti-vating each nickel crystal.This film is invisible, butdetectable either by dissolutionin some organic solvents or bytemperature programmed reduction(TPR) techniques.

Whisker carbon is the type mostoften encountered in gas based steamreformers, typically being the product offormation of carbon by the methanecracking reaction

CH, C + 2H, (4)

This is the carbon forming reactionin case of operation at too low S/C ratioor on a poisoned catalyst. In bothcases, too little hydrogen is formed tofight the carbon forming equilibrium ofreaction (4). Whisker carbon, when justformed, is very reactive and will oftendisappear if conditions revert to normaland will certainly be removed bysteaming. This is, for example, the caseif, for a short time, the reformer isoperated at too low S/C ratio, upon whichsubsequent operation at a high S/C ratiomay quickly remove carbon just laid downon the catalyst.

In the typical poisoning situation,the carbon is formed slowly as the cata-lyst activity gradually drops off, andwhen the day comes where correctiveaction is required, the carbon may haveremained in the reformer for weeks oreven months. The ageing of the carbonresults in a collapse of the whiskerstructure and the resulting carbon ismuch less reactive. Industrial carbonremoval procedures should be designed forsuch carbon.

CARBON REMOVAL

Traditionally, carbon has been removedfrom steam reforming catalysts bysteaming at high temperature according tothe "water gas" reaction

C + H20 CO + H, (5)

High temperatures are a must as agedcarbon requires at least 650°C (1200°F)for removal by steaming at a reasonablerate. In order to reach this temperaturelevel in the upper part of the tubes,where carbon is most likely to be formed,the outlet temperature would have to bemuch higher, i.e. typically that ofnormal operation.

We find that it is faster and moreefficient to remove the carbon by amixture of steam and air. Air added tothe steam decreases the ignition point ofaged carbon to 450°C (840°F), and removalof carbon is thus possible in the entirefurnace without using extreme tempera-tures. Obviously, the air concentrationshall be kept well controlled in order toavoid too high reaction rates. Concen-trations of air ranging from one to tenmole percent of the steam are used andwill ensure a smooth and quiet carbonburnoff.

SULFUR REMOVAL

As mentioned, operation at normal sulfurlevel, whether at normal conditions or atincreased S/C ratio and/or temperaturelevel will only very slowly regenerate asulfur poisoned catalyst. Steaming thecatalyst, on the other hand, will removequite significant quantities of sulfur inthe form of SO« and HpS. This indicatesa reaction pattern liKe

Ni - S + H20

H2S + 2H20

NiO + H2S (6)

S02 + 3H2 (7)

During the steaming, nickel willreact with steam under formation ofhydrogen

Ni + H20 NiO + H2 (8)

This hydrogen will, according to thereactions (6) and (7)., retard the sulfurremoval. In practice, this means thatthe catalyst must be oxidized completelyto ensure an efficient sulfur removal.For catalysts containing in the order of

21

15 wt percent Ni, we find that completeoxidation is achieved when five to tenpounds of steam per pound of catalysthave passed over the catalyst.

Effect of Alkali

Sulfur removal by steaming is effectiveon most steam reforming catalysts, butone important exception applies. If thecatalyst is promoted with sodium orpotassium, the alkaline oxides react withthe SO« under formation of alkali sul-fates. When such catalysts are laterreduced, the sulfur is released as hLS,which again poisons the nickel, and nonet removal of sulfur has been achieved.Figure 2 shows the result of steamingtests on various sulfur poisoned cata-lysts with and without alkaline.

1.0

0.8

0.6

0.4

0.2

Catalyst

O Ni/Mg A1204O i.i%ca• 0.8%MgD 0.7%K• 0 8%NaV 2.3%K• 4.9%KA Hi/AI203

500 600 700TEMP 'C

800

Figure 2. Regeneration of nickel catalysts withsteam: laboratory-scale test.100 g H2O per g catalyst for three hours.

i = sulfur after reg./'sulfur before reg.

Effect of Steam/Air Treatment

Instead of oxidizing the catalyst insteam alone, one could contemplate to usesteam and air as recommended for carbonremoval .

Experiments, however, show thatthis, to some extent and on all types ofreforming catalysts, will result information of nickel sulfate, which willremain on the catalyst.

A steam plus air treatment is there-

fore - with respect to sulfurremoval - not as effective as a puresteami ng.

Contrary to the alkali sulfatesformed by steaming óf alkali promotedcatalysts, the nickel sulfate formed byexposing the catalysts to air can fairlyeasily be reduced by a hydrogen/steamtreatment at conditions where the nickeloxide is not reduced.

The classical thermodynamics of thereduction of NiO to Ni according to

NiO + H, Ni (9)

tell us that a H20/H2 ratio below approxi-mately 200 would suffice for reduction ofNiO in the relevant temperature region.Due to interaction between the nickel andthe carrier, catalysts need, however,more hydrogen for the reduction.Depending on the actual catalyst carrier,the H20/H2 molar ratio required for thereduction is in practice below the orderof 25 to 100, which is lower than re-quired for the reduction of nickelsulfate.

NiS0 4H NiO 3H20 (10)

It is, therefore, possible, byselecting an appropriate H?0/H? ratio, toreduce the nickel sulfate without forma-tion of metallic nickel.

In consequence of the abovetheoretical considerations, our recom-mended procedures for regeneration ofsteam reforming catalysts are as follows:

1) If sulfur removal alone isrequired, the catalyst issteamed at low pressure and atan outlet temperature in theorder of 650°C (1200°F) withfive to ten pounds of steam perpound of catalyst. Analyses ofS02 and H?S at the reformer exitdetermine the duration of thesteaming. Afterwards, thecatalyst is reactivated as perthe normal procedure.

2) Almost in. all cases of sulfurpoisoning, carbon deposits haveto be removed together with thesulfur. Thus, the proceduremost often required will be moreelaborate.

22

A) Steaming as above.

8) Gradual introduction of airin thé order of five molepercent relative to thesteam.

C) Analyses of CO« at thereformer exit and decreaseof pressure drop willdetermine the end of thesteam/air treatment.

D) While in practice, step A)is sufficient to bring abouta substantial degree ofsulfur removal, experienceshows that some sulfur willoften remain. Thus, ifhydrogen is available, thenext step is Introduction ofH7, aiming at achieving aHgOAL molar ratio of 100.IT appreciable quantities ofH2S are found at the reform-er exit, this treatment is

until H hasto say 50 ppm,

whereafter they hydrogenflow is gradually increasedto the maximum availablefor f#31 catalyst reduc-

E) If no hydrogen is avail-able, the catalyst isreactivated by introductionof gas after completion ofthe steam/air treatment(paragraph C).

We have applied the above regenera-tion procedure on numerous catalystcharges during the last ten years. Theresults have, in all cases, been a verysignificant performance improvement,approaching a full recovery.

It goes without saying that thequestion whether to embark on a regeneration rather than a change of catalyst iscrucial. As usual, the predominantfactor in such an evaluation is theproduction loss during the regenerationand change of catalyst, respectively,whereas the direct costs of the regeneration (fuel and steam mainly) are smallcompared to the cost of a new catalystcharge.

Our experience shows that only in

extreme cases, tne time for a regenera-tion exceeds that of changing catalyst,so given the proven efficiency of theregeneration, economics are much in favorof a regeneration, provided that thestate of the catalyst prior to theincident, which has necessitated eitherchangeout or regeneration, is so that achangeout was not foreseen in the near

; future.

In the remaining part of the paper,five case stories of regeneration arepresented.

CASE STORY 1

The first story deals with a 600 MTPDammonia plant based on naphtha feed. Dueto sulfur breakthrough, in connectionwith a particular batch of high sulfurnaphtha, the reformer developed hot spotsand increasing pressure drop. Even afterthe naphtha quality was brought back tonormal, the increase in pressure dropcontinued, but could be arrested by goingdown in load and increasing the steam tocarbon ratio.

It was decided to take a shutdown,and as other maintenance jobs were to beundertaken at the same time, there wasample time for a thorough regeneration.The plant is well equipped for such anoperation as it has the followingfeatures:

1. Side fired furnace, allowingaxial temperature control.

2. Internal thermocouples in sometubes.

3. Electrically driven recipro-cating compressors.

4. Gas holder with synthesis gasfor start-up.

The above features make it rela-tively easy to comply with the require-ments for a regeneration according to therecommended procedures.

: /

i A graphical display of the regenera-tion is shown in Figures 3 and 4. Itconsisted of the following steps:

1. Steaming. Flow 12 to 15 tons/hcorresponding to 0.7 to 0.9 tonsof steam per ton of catalyst oer

hour. Duration six hours. CO,and S02 leave the furnace indecreasing concentrations.Inlet/outlet temperaturesapproximately 450/620°C.

2. Steam/air treatment at the sametemperature level for 14 hours.Initially CO« and S02 are high,but decrease towards zero.Oxygen breaks through thereformer after three hours andreaches 20 percent after anothersix hours.

3. After a few hours steaming,hydrogen is introduced in minutequantities for reduction ofnickel sulfate to H2S (H20/H2molar ratio decreasing from 150to 50). HpS peaks at ,600 ppmand decreased in the course of13 hours to below 100 ppm.

4. Circulation of hydrogen startedfor reduction of catalyst. Thisbrings down HLS to zero asmetallic nickel is formed.

At this time, the plant was notyet ready for production, so itwas decided to repeat the oxida-tion/reduction cycle in order toremove more sulfur and/or tocheck the efficiency of thefirst cycle. i

5. After one hour steaming, air isagain introduced. Very littleCO, and SO, is detected in thereformer effluent, and aftercomplete breakthrough of air,the oxidation is stopped.

Introduction of hydrogen for thesecond time follows the samepattern, as only little HLS(max. 100 ppm) is found in theeffluent.

7. Circulation of hydrogen isstarted for complete reductionof the catalyst at a H20/Hpmolar ratio of three. Thetemperatura level in thereformer is increased to themaximum permissible to ensurecomplete reduction.

CASE1.RW 1600 MIPO WO PLAKt. TOP

10 45 HOURF

Figures. Case 1, Part 1: 600 MTPD NH3 plant,Topsoe furnace.

CASE1.PARr2

50 55 60 65 70 75 80 HOURS

Figure 4. Case 1, Part 2.

Based on the S0„ and H~S analyses, atotal of two kg of sol fur was accountedfor. More than 95 percent of this sulfurcame off during the first oxidation/reduction cycle, demonstrating the effi-ciency of the regeneration. The quantityof sulfur positively detected correspondsto a little more than 108 ppm relative tothe weight of the catalyst, but beyondany doubt, much sulfur is not detected,as it is dissolved in the sample conden-sate, which was not analyzed.

The performance of the catalystreverted to normal after the regenerationas judged by the delta p, methane leak-age, and disappearance of hot spots.

24

CASE STORY 2

The plant In question is a. large ammoniaplant of MWK design. After a run of twoyears, the reformer developed hot bands,and nine tubes were hot from top tobottom. The temperatures of the hotbands were ranging from 1650 to 1680°F,whereas the maximum temperature outsidethe hot band area was 1580 to 1610°F.

The overall pressure drop hadIncreased only marginally during the twoyears run, and the performance In termsof methane leakage (approach to equilib-rium) was excellent, even better thanthat of the neighbor furnace loaded witha different and only six months oldcatalyst.

As It seemed a shame to dump thiscatalyst, the client decided upon aregeneration, which was performed duringa scheduled shutdown. The regenerationconsisted of a steam/air treatmentaccording to Topsoe's recommended proce-dure and lasted nine hours.

The key parameters are shown inFigure 5.

The regeneration followed prettywell the procedure agreed upon. The onlyproblem to speak of was the difficulty incontrolling the firing at the low flowrates. The relatively small steam flowwas necessary in order to get a reason-ably high percentage of air with theavailable air flow. Had steam and airflows, been higher, the time spent on thedecoking would have been less.

The total quantity of carbon removedas calculated from the CO« analyses andthe air flow was 183 pounas, i.e.0.44 pounds per tube. No doubt, some CCLleft the catalyst before addition of airfand some CO« has escaped detectiondissolved in the condensate, so theactual quantity of carbon removed maywell have been twice the above.Nevertheless, it is surprising, but alsotypical for the hot band formation thateven small quantities of carbon posi-tioned on the catalyst and inner tubesurfaces are able to create this problem.

After the decoking, the plant wassimt down and the delta p was checked onall tubes. The average was ten percenthigher than it was when the catalyst was

loaded two years earlier. Nevertheless,the client decided to change the catalyst'in the nine hot tubes plus In another 18,which had the most, severe hot bandsbefore regeneration.

The catalyst In these 27 tubes cameout free-flowing and with only a slightcoating of carbon on the very top sample.

The reformer was restarted threeweeks after the regeneration and per-formed excellently. At 93% load, the.approach to equilibrium was 13°F, and the'average maximum tube skin temperature haddropped to 1565°F. The hot bands had(disappeared completely.

Nine months, after the regeneration,hot banding appeared again on 47 tubes,and the catalyst was changed to anidentical new charge two months later,

i after a total life of three years.

02.C02.H2MOLE •IORYI 20-

0 1 2 3 4 5 6 7 8 9 HOURS

Figure 5. Case 2: large NH3 plant, MWK furnace.Steam flow: 64,000 Ib/h = 1.7 Ib/lb of catalyst perhour.

CASE STORY 3

This plant is a small Chemico designedammonia plant which experienced a break-through of organic sulfur from theCoMo/ZnO desulfurizer. The reformingcatalyst was, at that time, 16 monthsold, and the approach to equilibriumincreased from about 30°F to 45°F and themaximum tube skin temperature increasedby 25°F. There was no pressure dropincrease.

Two weeks after the breakthrough,the plant was stopped for changeout ofdesulfurization catalyst and the opportu-nity was taken to give the reformingcatalyst a steam/air treatment^ the

25

details of which are shown on Figure 6.

Surprisingly, no sulfur was detecteddownstream the reformer during theoxidation, but equally surprisingly,samples of catalyst taken from some tubeswhich were replaced during the shutdownshowed contamination with arsenic. Twotop samples contained 1650 ppm As and600 ppm As, respectively. Furthermore,the arsenic was concentrated at the verysurface of the rings, and in materialscraped off the surface more thanone percent As was identified.

The source of the As in this plantis most likely the CO« absorption unitwhich is a Vetrocoke design.

In view of these findings, it isquite possible that the deterioration inreforming catalyst performance, which wasthought to be caused by sulfur, was fullyor partly a result of arsenic.

The catalyst fully regained itsoriginal activity level by the regenera-tion and is still (22 months later) inoperation. It should, however, bementioned that since the regeneration,almost half of the tubes and thus alsothe catalyst have been replaced.

MOLEX(DRY)

20

C02MOLEX (DRYIAIR.MOLE %OF STEAM5

C02

EXIT 1300TEMP°f 1250

1200

1150

0 1 2 3 4 5 6 7 8 Hours

Figure 6. Case 3: 300 STPD NH, plant, chemicofurnace.Steam flow = 20,000 Ib/h = 1.8 H2O per Ib ofcatalyst per hour.

CASE STORY 4

This case deals with a large methanolplant with a top-fired reformer where hotbands appeared on a 16 months old cata-lyst after operation on an ethane/propanefeed. It should be mentioned that thisplant operates at a pretty high inlet

temperature (1030 to 1050°F), which makesoperation at the relatively low steam/carbon ratio of 2.8 quite critical on theheavy feed.

As a considerable amount of sulfurwas removed during the regeneration, itis quite possible that this has contrib-uted. This theory is supported by thefact that shortly before operation on theethane/propane feed, the plant receivednatural gas with high contents of sulfur,in the form of HLS, as well as diethyl-disulfide. As tne plant has only zincoxide for purification, a breakthrough ofsulfur - although not identified - isquite likely.

The plant operator, for safetyreasons, decided to blind off all hydro-carbon and hydrogen bearing lines to thereformer before the decoking whilepurging the reformer with nitrogen inpretty cool condition. The decoking thusstarted with a conventional heating up ofthe reformer in steam to 1200°F over aperiod of 12 hours, whereupon air wasadded in a gradually increasing quantity.The pressure level was 70-100 psig.

The steam flow during the decokingwas 35 percent of design flow corre-sponding to 0.7 pounds per pound ofcatalyst per hour. Air was supplied by arented compressor with a maximum capacitycorresponding only to 1.5 mole percent ofthe steam, which was kept on the reformerfor six hours.

Unfortunately, we do not have avail-able details from the steam/airtreatment, but we know that largeconcentrations of S02 and CO« weredetected during the steaming and thesteam/air treatment.

As sulfur seemed to have been con-tributing to the hot bands, it wasdecided to try to remove more sulfurprior to the catalyst reduction by acontrolled reduction at a steam/hydrogenmolar ratio of 100. HpS was not measureduntil 40 minutes later at the reformeroutlet and the result was 192 ppm. TheHpS has thus most likely been much higherjast after adding the. hydrogen. HpStailed off to 12 ppm within two hours,whereupon the catalyst was rereduced inrecirculating hydrogen at asteam/hydrogen ratio of eight to ten.

26

When the plant came back on stream,the bright hot bands had disappeared, buthad left a kind of "shadowy" marks on thetubes. It is possible that the limitedair supply ha,s resulted in an incompleteremoval of coke on the tube walls.

CASE STORY 5

The last case story deals with a case ofmassive and fortunately non-typicalsulfur poisoning in a small hydrogenplant.

The normal feed is à refineryoff-gas containing 45 to 70 percenthydrogen, but the plant is equipped withfacilities to mix in butane as backupfeed. The operation on the two verydifferent feeds has, in the past, givenproblems of steam/carbon ratio control,resulting in coking incidents.

The refinery gas is purified in anami ne wash (not integrated with thehydrogen plant) and by two ZnO vesselsoperated in series. During a failure ofthe ami ne wash and presumably at a timewhen the first ZnO vessel was nearlysaturated, a major breakthrough of H2Soccurred. Only a single figure of 20 ppmis reported. When the operators realizedthe problem, they switched over to thesulfur free butane feed, but as thecatalyst by that time was heavily poi-soned, it was not able to handle thisfeed and coked up (delta p increased tomore than 100 psi).

The butane feed was pulled out andthe catalyst steamed. During the initialpart of the steaming, the condensate wasreturned to the boiler and nitrogen wasrecycled around the reformer. As much as1200 ppm of HgS was found in the circu-lating nitrogen. At this point,Hal dor Topsoe was alerted and it wasrealized that the mode of recirculatingin this situation was not right, as thiswould keep the sulfur enclosed in theunit.

The circulation was stopped, butnitrogen was kept flowing once-throughthe reformer. This is advantageous, asit allows a more accurate estimate of thequantity of sulfur leaving the reformer,by giving a known amount of dry gas inthe reformer effluent.

The steaming conditions were:

Steam flow: 15000 lbs/hrcorresponding to1.5 lbs/lb of catalystper hour

N2 flow (estimated):Temperatures inlet/outletPressure at outlet:Pressure drop:

Duration:

40000 SCFH830/1200°F100 psig100 psi -notdecreasing35 hours

After this steaming, the nitrogenflow was cut and air was introduced andgradually brought up to seven molepercent relative to the steam. Theconditions were:

Steam flow:

Temperatures inlet/outlet:

Pressure at outlet:Pressure drop:

15,000lbs/hr

800/1200-1300°F

85 psig100 psidecreasingto 10 psi

During the steaming, as well asduring the steam/air treatment, largequantities of sulfur left the reformer,as shown on Figure 7. It should be notedthat all analyses are made by Dragertubes. Particularly noteworthy is thevast difference in the carbon removalrate before and after introduction ofair.

H2S.

S02ppm

C02 MOLE % IDR1I

~ ~ 0 5 1 0 Î 5 2 0 25 30 35 40 « 50~HOURS

Figure 7. Case 5: small hydrogen plant.Steam flow : 15,000 Ib/hr = 1.5 Ib per Ib of cata-lyst.

Following the steam/air treatment,the catalyst was again steamed for24 hours (no data available) before

27

hydrogen was introduced for furthersulfur removal.

The conditions were as follows:

Steam flow:H2 flow:

Outlettemperature:

15000-30000 lbs/hr3400-6200 SCFH,corresponding toH£0/H2 = 80-100

1200-1300°FOutlet pressure: 85-100 psig.

Sulfur, in the form of H„S, came offin concentrations varying from 2000 ppmdown to 400 ppm during a period of noless than 90 hours.

After this treatment, refinery gasfeed was introduced, and during the first24 hours, HpS was detected in the efflu-ent in concentrations from 40 ppm down to1 ppm.

If, on the basis of dry flows andmeasured sulfur concentrations, we add uphow much sulfur has definitely left thecatalyst, we find the following:

During 35 hours of steaming: 50 IbsDuring 16 hours of steam/air: 2 IbsDuring 90 hours of partialreduction: 26 Ibs

Total: 78 Ibs.

On top of this could be added thesulfur escaping detection dissolved inthe condensate. Conservatively, there isno doubt that at least 100 pounds ofsulfur have been removed from the cata-lyst, which is indeed a record. As thetotal quantity of reforming catalyst isapproximately 10,000 pounds, the sulfurremoved thus corresponds to one percent

of the catalyst - or 10,000 0pm, which isthe more .familiar unit.

This is more than five times as muchsulfur as could be chemisorbed on thecatalyst and thus indicates formation ofbulk nickel sulfide (Ni2S3) on thecatalyst. According to tne thermo-dynamics, this would require in the «reformer a HpS/H« ratio larger than 10 ,i.e. a sulfur concentration in the sameorder of magnitude (1000 ppm) in thereformer feed gas. Although that high afigure was never detected, it is notimpossible, as the refinery gas upstreamthe failing ami ne scrubber contains manythousands of ppm of sulfur.

As could be expected after such amassive poisoning, the reformer perfor-mance has not been flawless after theregeneration. The tubes were (and stillare) spotted, and the approach to equilib-rium was initially high, but hasgradually decreased to a normal level.

It could be argued whether thedecision to regenerate in such a case ofmassive sulfur poisoning was correct.Many operators would be tempted to changethe catalyst and believe that all is wellafterwards. This is, however, hardly so.In cases as severe as this, sulfur islikely to remain in the feed lines, laterto be released and to poison a new chargeof catalyst. To prevent this happening,a time consuming purification of all thepipework upstream the reformer would havebeen necessary to safeguard againstpoisoning a new charge of catalyst.

LITERATURE CITED

J. R. Rostrup-Nielsen: "CatalyticSteam Reforming", in CATALYSIS,Springer-Verlag, 1984.

Jens Houken

28