1

Oxygen: the Solution for Sulfur Recovery and BTX

By

Jason NormanBOC

Murray Hill, NJ

Stephen Graville and Richard WatsonBOC

Sheffield, UK

ABSTRACT

Oxygen has been in use in sulfur recovery units for more than twenty years. Oxygen allows for flexiblesulfur plant operation for expanded capacity, peak shaving and turndown with minimal capital investment;this is beneficial in assisting an industry under pressure to increase gas capacity for the power generationsector. BTX contaminates, which cause serious problems with sulfur plants, can readily be destroyedwith the use of oxygen. This article reports pilot plant and simulation studies conducted to evaluate burnerdesign features for BTX destruction. The use of oxygen has been evaluated as a cost effective means ofdestroying BTX compared to alternative approaches whilst providing the additional option to increasesulfur plant capacity up to 180% of design.

BACKGROUND

BTX (Benzene, Toluene, and Xylene) are common contaminants in Claus plant feed gases originatingfrom associated and some natural gases. Whilst not difficult to destroy in a conventional combustionprocess, the fact that the associated acid gas often contains much less than 60% H2S means that thetemperature in the Claus furnace is often too low for effective BTX destruction. Failure to remove theBTX, either upstream or in the reaction furnace, leads to carbon and hydrocarbon contamination in thecatalyst beds with subsequent loss of activity, high pressure drop in the first catalytic bed and a need tochange out catalyst on a frequent basis. Bypassing some of the process gas around the furnace toincrease the furnace temperature is not possible since BTX is then passed directly to the catalyst beds.

The conventional way of handling the problem within the reaction furnace, for air-based plants, is topreheat the feed gas (and air) either directly or indirectly. Indirect pre-heat is less efficient thermally andrequires the use of direct-fired heaters, or gas to gas heat exchangers if an external source of heat isavailable. If the heat required for preheat is taken from the Claus unit itself, less steam is available foramine stripping and difficulties can arise during start-up. In either case, preheating is energy intensiveand has a significant effect on the cost and the complexity of the Claus unit. Furthermore, as highlightedin a paper by Chen[1], the heat required to preheat the air and / or acid gas streams tends to increaseoperating costs of the plant. If preheat is to be retrofitted to an existing burner / furnace configuration, theincrease in gas volume to the burner may lead to mixing and pressure drop issues.

Adding natural gas to the Claus feed is more attractive from an efficiency point of view, but this cansignificantly increase the size and cost of the Claus unit. If the added methane is not effectively burnt ittoo can add to problems with carbon deposition. The reducing nature of the furnace also tends to resultin higher CO and COS concentrations due to incomplete combustion of the methane if added directly tothe feed acid gas.

Oxygen enrichment or total replacement of the combustion air with oxygen is an elegant and costeffective solution to the BTX problem and this paper will concentrate on illustrating the benefits andpotential cost savings that can be achieved. It will also describe the programme of test work carried outby BOC on it’s 4 ton per day Claus pilot plant to collect the data which forms the basis of the evaluationwork.

2

BOC SURE™ PROCESSES

BOC began developing oxygen using burners and processes for application in Claus units in the mid80’s. Much of this work was aimed initially at the refining Claus market, but its applicability to gasrecovery Claus was always recognised. BOC offers a complete range of options from simple enrichment,where oxygen is added to the combustion air before it reaches the burner, through burner replacement,which allows higher enrichment levels and mixes oxygen directly with feed gas at the burner tip, to pureoxygen using processes. In the latter case, BOC’s Double Combustion process allows pure oxygen useeven on high combustible content refinery feeds. Three plants now operate successfully with pureoxygen.

Since the feed streams derived from most gas recovery operations (incl. POX) contain less than 60% H2Sby volume (often much less) pure oxygen can be used without process modifications such as DoubleCombustion. The parts of BOC’s SURE™ portfolio most applicable to gas operation therefore are theSURE™ burner design and related technologies that can have a significant effect on BTX destruction.

BOC SURE™ BURNERS Burner development is regarded as a core competency within BOC. Our experience has been built upover many decades and in a wide range of user industries.

The tip-mixed design was chosen because of its safety and flexibility and because it generally has amuch lower pressure drop than the pre-mixed alternative. As the name implies, oxygen (and air in thecase of lower levels of enrichment) does not mix with the acid gas until they leave the burner tip. Keepingthe feed gas, oxygen (and air) apart in this way, facilitates the creation of zones of different temperatureand stoichiometry within the flame. In this way, the flame is used as a chemical reactor and the burner isstaged to promote the beneficial reactions that aid the destruction of unwanted contaminants such asBTX. It should be noted that, within the flame, localised temperatures in excess of 2000oC might exist.This is achieved without any risk to the furnace refractory or any other part of the Claus plant. Since theBTX destruction reactions are kinetically limited, higher localised temperatures result in much greaterdestruction rates.

Burner-furnace matching is of major importance and in order to evaluate this fully, BOC has developed athree-dimensional kinetic computational fluid dynamics (CFD) model of the Claus furnace. This modelhas been fully validated with data from BOC’s 4TPD pilot plant, from commercial installations usingSURE™ technology and from small-scale laminar flow reactor work where applicable.

BOC’s burner and process technology has been in commercial use for approximately 12 years and hasproved highly successful. Burners are designed and produced to exacting standards in order to meet therequirements for oxygen use and to satisfy the design standards of the user industry. The burners havebeen developed to ensure optimum performance no matter what the feed composition may be. Thesedesigns have been obtained after extensive research examining the key parameters associated withgood contaminant destruction and burner / furnace operation within the refinery and gas plant industries.To do this, a purpose built Claus burner development facility was constructed, a brief description of whichfollows.

BOC Burner development facilityThe burner development facility comprises a commercial-scale 4TPD Claus pilot plant with one thermaland one catalytic stage (see Figure 1). Based on the former Courtaulds site in Trafford Park,Manchester, UK, this unit was operated for a period of three years. The plant was equipped withsophisticated in-furnace sampling and gas temperature measuring devices (see Figure 2), some of whichwere developed by BOC specifically for this application.

The plant was able to simulate virtually any feed stream including high ammonia and BTX contaminatedoptions. The programme looked at the performance of all SURETM burner designs and the completerange of oxygen use up to 100%. The work on the pilot facility was supplemented by additional small-

3

scale laminar flow reactor work carried out both within BOC and by external R&D organisations. Furtherinformation relating to the facility is available from Graville et al. [2-5].

Realising that the burner requirements may be different between gas plant and refinery operations, aprogramme of work specifically geared to examine BTX destruction was undertaken. This consisted of arange of operating conditions that included: varying BTX and H2S concentrations in the feed gas to the burner varying levels of oxygen enrichment natural gas addition as an alternative and in combination with oxygen enrichment varying the burner design and modes of operation

Figure 1 BOC SURE™ burner development facility

Figure 2 Gas sampling from within the reaction furnace

4

Full furnace profiles of the different species within the furnace were measured using gas chromatography.Corresponding temperature measurements were taken with in situ thermocouples protruding into thefurnace flow area, refractory thermocouples, sacrificial thermocouples and suction pyrometry. This datawas then used to help develop and validate a kinetic CFD model that is described below. Using thismodel, burner / furnace matching can be performed and process optimisation techniques examined andimplemented.

Claus reaction furnace models

Many models used in the design of Claus units rely upon equilibrium assumptions to represent thechemistry occurring within the reaction furnace. This is adequate if the species being examined are notkinetically limited and one can assume the reactants to be perfectly mixed. In instances where kineticlimitations predominate, such as ammonia and BTX destruction, the equilibrium assumption is erroneousand will often predict far greater destruction levels than actually occurs within the furnace. Some modelstry to overcome this through empirical relationships taken from commercial plant however, due to thelarge number of variables between different plants, this too can give misleading results.

An alternative to relying purely on equilibrium assumptions is to incorporate kinetic expressions for thosespecies whose reactions are deemed to be kinetically limited within the reaction furnace. Unfortunately,kinetics alone are not sufficient to obtain an accurate picture of the reaction furnace chemistry. In manyprocesses, especially combustion, turbulence and mixing play an important role in the overall reactionpathways undertaken. Therefore, both the kinetics and mixing properties need to be combined andsolved simultaneously.

BOC has, over the last twelve years, developed a 3D model of the reaction furnace and associatedchemistry [6]. The model uses a modified computational fluid dynamics code (CFD). Using this code, thechemistry, thermodynamics, fluid flow, turbulence and radiation can be solved simultaneously. Thesolution process is complex and requires relatively powerful computers to obtain a solution in a timelymanner. Having obtained a solution however, a great deal of information can be extracted from themodel. One caveat with this type of model is that the results are only as good as the various sub-modelsused to describe the Claus process. To this end, a great deal of effort has been expended ensuring thatthese are accurate; the sub-models for the furnace chemistry are, by necessity, particularly complex. Asfar as BTX is concerned, those reactions associated with its destruction have appended an alreadycomplex set of chemical reactions used to represent the sulfur chemistry. BTX chemistry relevant to thiswork is discussed below.

BTX chemistry

The reaction pathways associated with the aromatic components within the reaction furnace have beenthe issues of some debate within the industry. Since oxygen is present, the assumption that thehydrocarbons are oxidised to CO2 is often thought to prevail. However, experience shows that invariablythe oxygen requirement is less than stoichiometric for the amount of hydrocarbon present leading toincomplete or partial oxidation. Furthermore, in the flame region of the furnace, the H2, H2S and reactiveflame radicals are competing for the available oxygen. As such, the relatively large aromatic moleculeswith associated large bond energies tend to fair badly in the battle for oxygen resulting in a different set ofreactions occurring to those of hydrocarbon oxidation. An investigation was therefore conductedexamining the possible reaction pathways for these higher hydrocarbon species.

Experiments detailing higher hydrocarbon decomposition reaction rates are well documented [7-10] andvarious reaction pathways have been proposed. Figure 3 illustrates an initial decomposition step for p-xylene derived from Hippler et al.[8]. In this scheme, the initial decomposition of p-xylene is initiatedthrough collision with another molecule (M) within the system. This either leads to a methyl group beingremoved from the ring or, more likely, the removal of an H atom. Once initiated, toluene is eventuallyformed from 4-methylphenyl reactions. In a similar manner, toluene thermally decomposes to benzene[9].

5

CH3

CH3

CH3

CH2

H

CH3

M

+

CH3+

M

H

CH3

CH3

CH2

+H

CH3

CH2

+CH4

CH3H3C

CH3H3C

2

CH3

H3C

CH3

+

CH3

CH2

H, M

CH2C6H4CH2 + H + MM

M

M

Figure 3 Thermal decomposition of p-xylene through 4-methylphenyl to toluene

Examination of the various reaction enthalpies of the dissociation channels for the larger hydrocarbonmolecules, shows that xylene tends to have lower reaction enthalpies than toluene (Table 1). Toluene, inturn, has lower reaction enthalpies than those associated with the relatively stable benzene ring. As aconsequence, once sufficient energy is supplied to decompose xylene, benzene tends to be formedrelatively quickly; in the experimental programme of work undertaken, in furnace sampling rarely showedany species other than benzene downstream of the main flame reaction front.

The deactivation potential varies with different hydrocarbon components; In catalyst deactivationexperiments, Cravier et al.[11] found that xylene and toluene were particularly efficient at deactivatingClaus catalysts with xylene being capable of significant catalyst deactivation within a matter of hours. Incommercial units where catalyst deactivation occurs over a relative long time, the main contributor islikely to be benzene [12]. Where extremely rapid deactivation of the catalyst occurs, toluene and xylenewould be the main contributors however, such events would normally be associated with ‘non-standard’operation.

Parent molecule Reaction products ∆H298/kJmol-1

p-xylene p-methyl-benzyl + H 353.2C7H7+ CH3 423.8

toluene benzyl + H 356.1phenyl + CH3 426.4

benzene phenyl + H 464.2

Table 1 Reaction enthalpies for aromatic hydrocarbon dissociation channels

Using fundamental chemical kinetic packages, the break-up of the large aromatic rings and theassociated kinetics have been investigated. Using this information, sensitivity studies examining the keyreaction pathways were conducted and a series of key reaction steps determined. Before these could beincorporated into the CFD model however, the inclusion and interaction with sulfur chemistry wasrequired. This work has therefore examined and included CS2, S2, H2 and COS chemistry.

COS and CS2 The chemistry associated with CS2 and COS alone is particularly complex as is illustrated in Figure 4. Inthis case, the main source of CS2 comes from reaction of methane with sulfur species in the reactor. In

6

the case of BTX, the methyl radical serves as the main precursor to CS2, this being produced during theinitial decomposition of the larger aromatic structures.

CH4 + 2S2 CS2 + CO2 2COS

H2S

H2O

COS + H2S COS + S2O

SO2

H2O

CO2 + H2S

H2 + 0.5S2CO + H2OCO2 + H2

SO2

CO2 + S2O

CO2

COS + S3

CS2

CO2 + S3

S3

S3

S2

COS + CO + SO2

CO + 0.5S2

CO2

COS + H2O

3S2

Figure 4 Reaction pathways for COS and CS2 chemistry (adapted from Clark et al.[13,14])

In addition to the main stable species that are shown in the above scheme, many flame-generatedradicals are also present and these play a key role in the overall formation and destruction of each stablespecies. Such radicals include H, SH, S, OH, CH, CH2, CH3 and SO. To include the effect of all species,a large kinetic model containing over 250 reactions was used. This large scheme was checked withexperimental work available within the literature and external and internal bench scale experiments. Forcomplexity reasons, it is prohibitive to model all of the reactions present in the fundamental kinetic modelwithin the CFD model. Further studies determining the key reaction pathways were repeated and asimplified model of the reaction chemistry suitable for use within the CFD model was derived.

Using this global kinetic model, a comparison of BTX destruction at two different temperatures is shownin Figure 5. Here, the top graph represents the system at 1000oC whereas the lower represents thesame system at 1100oC. Xylene initially present reacts to form toluene and benzene. The toluene curve(inverted triangles) shows an initial formation peak and subsequent destruction to benzene and otherproducts. Benzene formed is destroyed albeit at a slower rate than toluene. This figure clearly illustratesthat a 100oC change in operating temperature has a marked effect on the rates of BTX destruction;benzene and toluene are still present after 2s residence time. This sensitivity to temperature andresidence time illustrates why Claus furnace models relying on equilibrium assumptions often have errorsin their BTX representation.

7

Figure 5 Comparison of BTX destruction at 1000oC and 1100oC

The reduced reaction scheme used in the simulation above has been incorporated into the CFD model toenable the mixing and fluid dynamics to be coupled with the chemistry. An overview of this is shown inFigure 6.

8

CH3

CH3

CH3

Hy drocarbonPool

S2 CS2

H2O CO + H2

COSH2O, SO2

Sulphur chemistry model

M,H,CHx M,H,CHx

Figure 6 BTX chemistry and tie-in to existing CFD sulfur model

Results from the BOC burner test facility, described earlier, have been used to validate the BTX CFDmodel. The overall technique adopted ensures that the chemical models used are validated at each stepfrom fundamental level through to the global reaction pathways incorporated in the CFD. The nextsection describes some of the CFD validation work whilst illustrating the importance of good burnerdesign and the effect this can have on BTX destruction.

Validation of the CFD model

Figure 7 Illustration of 3D section of reaction furnace showing burner, hot flame region and innerrefractory temperature variation

A full 3D simulation of the BOC burner development facility reaction furnace was performed for a gasstream comprising 40% H2S, 59.3% CO2 and 0.7% xylene firing through a burner using 100% oxygen asthe oxidant. The inside furnace refractory temperature, hot gas zone and burner location are illustrated inFigure 7. For this particular firing configuration, acid gas passes through and around the acid gas burner.Pure oxygen, fed through the central regions of the burner reacts rapidly with the combustible speciesliberating thermal energy. With the BTX chemistry and sulfur sub-models in place, the predictedtemperature, H2S and SO2 variation within the furnace obtained are shown in Figure 8. The figures hererepresent a slice through the middle of the reaction furnace.

9

Figure 8 Predicted contours of temperature, sulfur dioxide and hydrogen sufide

Intense combustion regions result in a relatively short flame with peak temperatures around 1900oC. Inthis figure, the red end of the scale depicts higher temperatures and concentrations. Oxidation of thesulfur species initially liberates SO2 the profiles of which follow maximum temperature relatively closely.The hottest region of the flame consumes virtually all of the H2S, that which is not oxidised tends to bedissociated to H2 and associated species.

Downstream of the main reaction zone, the temperature drops fairly quickly to give bulk gas temperaturesof around 1050oC, which agree well with experimental observations. Using the model, it is possible toconfirm that refractory temperatures throughout the furnace remain below the material temperature limits.This is not as great a concern for gas plant feed streams owing to the low H2S concentrations and highCO2 levels in the feed. In this example, the high temperature region of the flame is shrouded by a stagedfeed gas stream, which ensures the refractory temperature remains well within material design limits.

In Figure 9, xylene, toluene and benzene concentrations are depicted together for the same case. Theinitial break-up of the xylene molecules leads to some formation of toluene part of which subsequentlyforms benzene. As far as kinetics are concerned, the xylene destruction is faster than toluene andbenzene, complete destruction occurring before the first sampling port of the test facility. Benzene takeslonger to be fully removed owing to the stability of the benzene ring. The scale in this figure is notcommon but serves merely to illustrate the high and low concentration regions within the furnace.

In this example, the high temperatures associated with the flame region give very high BTX destructionrates. Even though xylene is depicted to pass around some of the flame, there is sufficient temperature,mixing and time for the key reactions to proceed. Even with a relatively low bulk gas temperature of1050oC, the high levels of feed xylene (7000ppmv) are completely destroyed using pure oxygen whilstensuring refractory temperatures remain well with design limitations.

10

Figure 9 Decomposition of xylene to toluene and subsequent formation and destruction of benzene

Figure 10 illustrates a comparison between predicted and measured benzene concentrations for thiscase. Benzene is illustrated here since it was the only component measured within the furnace at thesampling locations. The burner is located at the left end of the figure. Considering the complexity of theBTX chemistry and sampling and analysis errors, the model gives good agreement with measuredvalues. The lower plot, depicting the predicted values, is slightly conservative with respect to benzenedestruction, which is favourable for design purposes. The figure illustrates that complete BTX destructionis obtained.

Figure 10 Comparison between measured (top) and predicted (bottom) levels of benzene (ppmv)

11

The results from the model described above provide a large amount of data, far more than is presentedhere. Using this CFD model different burner designs and operating conditions have been examined andthe effects and dependencies these have, within the Claus reaction furnace, have been determined. Witha validated model, this can be done without necessarily having to build and test each burner design. Thenext section illustrates an example where the CFD model is used to examine some of the differencesbetween general oxygen enrichment and specific oxygen enrichment through a purpose-built burner.

Oxygen enrichment options

At levels below 28%v/v, oxygen addition to the reaction furnace can be performed through either generalenrichment, lancing or a purpose built burner. Above 28%v/v, oxygen compatibility requires differentmaterials to be used and purpose built oxygen equipment is normally supplied. Oxygen requirement isnormally determined through process modelling of the type discussed previously. Due to the nature ofthis analysis tool, only general enrichment can be considered. Using the CFD model however, differentburner designs and firing configurations can be examined and the best solution determined.

In this example, the same burner is fired in two different modes. The feed stream is the same as theprevious example i.e. 40% H2S, 59.3% CO2 and 0.7% xylene. In this case however, oxygen enrichmentto 28%v/v is utilised as opposed to 100% in the previous example. In dropping from 100% oxygen to28%, the difference is made up with nitrogen from air and as a consequence, the furnace operatingtemperature will drop from the previous example. The two firing options considered are:1. GENERAL: General enrichment of the air fed to burner to a level of 28%v/v oxygen 2. SPECIFIC: Pure oxygen feed to the burner gun with separate air being fed to an air annulus. The

quantity of oxygen used is the same as case 1 thus giving the 28%v/v overall oxygen levels.

Figure 11 Comparison of temperature profiles for the two firing configurations. Top = generalenrichment. Bottom = Pure oxygen addition to same level of overall enrichment

Figure 11 illustrates that, with the SPECIFIC mode of operation, the maximum gas temperature is higherthan the GENERAL case by approximately 300oC. As a consequence of this, the maximum rates of BTXdestruction are higher. Bulk gas temperatures on the other hand, are virtually the same. Figure 12depicts the xylene destruction for the two configurations. The general enrichment case obtainsapproximately 81% destruction of the feed xylene whereas the specific enrichment case achieves 94%destruction. Although in each case, xylene breakthrough would occur, the concentrations are less withthe purpose-built burner. Bulk gas temperatures are ~980oC and one would normally expect some HC

12

breakthrough at these temperatures and feed concentrations. In practise, an increase in oxygenenrichment would be sufficient to destroy all of the BTX.

Through using CFD modelling and pilot plant testing, BOC has been able to design SURE™ burners thatare optimised for BTX destruction. As such, the amount of required oxygen to achieve completedestruction is less than is required for general enrichment. In a similar manner, a purpose built burneroffers the same benefits over oxygen lancing for higher levels of oxygen enrichment.

The decision as to whether to use oxygen or not should not therefore be based solely on the oxygenrequirement predicted by industry standard process simulations since these only account for the generalenrichment option. With correctly design equipment, oxygen usage can be less and a reduction inoperating costs can be gained.

Figure 12 Comparison of xylene concentrations for the two different firing configurations

The economics of oxygen use

Oxygen use within Claus units treating gas plant acid gases offer two main advantages. The first is theability to completely destroy BTX without the need for natural gas addition or preheat. The second is theopportunity to increase the plant throughput due to the removal of diluent nitrogen associated with the airstream. The benefits of this can either be realised through debottlenecking existing plants or reducedplant size for grass roots units. Since the concentrations of H2S are relatively low in gas plants, higherlevels of oxygen enrichment are achievable within the single furnace without furnace overheating. Formaterials reasons, oxygen enrichment above 28% v/v would require dedicated oxygen equipment inthese instances. In cases where the acid gas is reasonably strong, in excess of 70% H2S, alternativetechnologies, such as BOC’s Double Combustion, can be used to control the higher associatedtemperatures.

For grass-roots installations of equivalent throughput, there are capital, space and complexity savings tobe made with a pure oxygen system over an air-based unit. This is mainly as a result of the removal ofair associated nitrogen from the system; an oxygen-based plant treating a 40% H2S feed would beapproximately 55% of the normal size of an equivalent air based unit. In cases where additional fuel gasand air are required for BTX destruction purposes, the size of the air-based unit is necessarily larger andthe oxygen-based unit can be significantly less than 50% of the air-based plant. In addition to the above-mentioned savings, the main air blower would not be required since the air separation plant wouldprovide oxygen already compressed to the required pressure. A small packaged air blower could beused for start-up conditions. The use of oxygen also enables catalyst savings to be made owing to thesmaller sized units. This is of additional benefit in plants using proprietary catalysts.

13

In spite of the on cost of oxygen, the operating costs can also be lower than that of an equivalent air-based plant as shown in Table 2. Three options for achieving the required temperature of 1100oC havebeen evaluated:

preheat acid gas and air up to 500oC add fuel gas into feed to reaction furnace pure oxygen feed

Net requirement Preheat Fuel gas OxygenFuel gas (Heating/RGG/incinerator)

100% 95% 25%

Electrical(Air fans)

100% 123% 16%

Net production Preheat Fuel gas OxygenSteam LP 3.5 Barg(Condensers/coolers)

100% 122% 83%

Steam MP 23 Barg(WHB/reheaters)

100% 98% 65%

Table 2 Operating costs for different process optionsThe fuel gas and power requirements are very much larger for the fuel gas and preheat processes whencompared to oxygen usage. The flip side of the coin is that there is a net increase in steam production forthe preheat and fuel gas processes. However, the value of the steam will be limited by the cost ofproviding boiler feed water (typically BFW costs are a third of the value for steam production) and thedemand for a steam supply on site.

The major operating benefit using pure oxygen is seen in fuel gas (natural gas) savings. Obviously thecost of fuel gas determines the relative savings in operating costs achieved. In order to provide theoxygen for BTX destruction, an air separation unit (ASU) has been assumed as the supply. If the powerrequirements of the ASU to produce the required oxygen can be met with integrated power generation,then the cost of oxygen supply can be reduced by approximately 50% further. In order to do this, anequivalent natural gas requirement of 25% of the preheat case, would be required as a fuel source.

For a grass-roots system, both capital and operating cost savings can lead to oxygen being aneconomically attractive alternative to other technologies. In the instance of retrofitted oxygen supply todestroy BTX, the economics become more dependent upon local gas prices and the local cost of oxygenwhich in turn is dependent upon demand requirements.

Conclusions

• An extensive programme of work has been conducted aimed at improving the understanding of Clausplant operation and the controlling parameters for BTX destruction. A greater understanding of thechemistry associated with BTX and sulfur has been achieved and this work has resulted in a kineticmodel that has been incorporated into a customised CFD package. Through extensive testing over arange of different operating conditions, the model has been validated on a commercial scale pilotfacility.

• In conjunction with pilot plant experiments, modelling work has examined and optimised SURE™burner designs specifically for BTX destruction. The design of burners enables more efficientcontaminant destruction than can be achieved using general enrichment and lancing techniques tothe same level of enrichment. Consequentially, operating costs can be less when SURE™equipment is used.

• Converting a gas plant sulfur recovery unit to oxygen can provide an additional 80% increase incapacity at a marginal capital cost. The added benefits of more reliable operation with goodcontaminant destruction can also be attained.

• In addition, to the capital savings, a significant reduction in energy consumption can be achieved.Virtually all the electrical power cost would be incorporated in the oxygen cost (for over the fencesupply). Additional energy would not be required for pre-heating and the fuel gas supply for the

14

incinerator would be greatly reduced. In these instances, the cost of oxygen can be offset againstthese operating cost savings making oxygen enrichment an attractive alternative to othertechnologies.

References

1. Chen, J.K. ‘Processing Lean Acid Gas in Sulfur Plants’, In proceedings of Brimstone Sulfur RecoveryConference, Canmore, Canada, 2001.

2. Graville, S.R. and Watson, D. ‘BOC Burner Development Technology’, In: Proceedings of the

Brimstone Sulfur Recovery Symposium, Vail, US, 1997. 3. Graville, S.R. and Watson, D. ‘Optimising the use of Oxygen in Claus Plants’, In: Proceedings of the

Brimstone Sulfur Recovery Symposium, Canmore, Canada, 2001. 4. Graville, S.R., Norman, J.S. and Watson, D. ‘Claus Plant Reaction Furnace: Misconceptions’, In:

Proceedings of the Brimstone Sulfur Recovery Symposium, Vail, US, 1998. 5. Graville, S.R., Norman, J.S. and Watson, D. ‘Contaminant destruction using the BOC SURETM

burner’, In: Proceedings of the Brimstone Sulfur Recovery Symposium, Vail, US, 1999. 6. Norman, J. S. and Watson, R. W. ‘ Claus Reaction Furnace Modelling’ Sulfur, pp43-51, August 1999. 7. Kern, R.D., Wu, C.H., Skinner, G.B., Rao, V.S., Keifer, J.H., Towers, J.A. and Mizerka, L.J.

‘Collaborative shock tube studies of benzene pyrolysis’, In: Twentieth Symposium (International) onCombustion, The Combustion Institute, pp789-797, 1984.

8. Hippler, H., Seisel, S. and Troe, J. ‘Pyrolysis of p-xylene and of 4-methylbenzyl radicals’ In: Twenty

Fifth Symposium (International) on Combustion, The Combustion Institute, pp875-882, 1994. 9. Brouwer, L., Muller-Markgraf, W. and Troe, J. ‘Identification of primary reaction products in the

thermal decomposition of aromatic hydrocarbons’ In: Twentieth Symposium (International) onCombustion, The Combustion Institute, pp799-806, 1984.

10. Muller-Markgraf, W. and Troe, J. ‘Shock wave study of benzyl UV absorption spectra: Revised

toluene and benzyle decomposition rates’, In: Twenty-first Symposium (International) on Combustion,The Combustion Institute, pp815-823, 1986.

11. Crevier, P.P, Clark, P.D., Dowling, N. I. and Huang, M. ‘Quantifying the effect of individual aromatic

contaminants on a Claus Catalyst’, In: Saudi Aramco Journal of Technology, pp46-54, 2001. 12. Klint, B. ‘Hydrocarbon destruction in the Claus SRU reaction furnace’, In: Proceedings of the

Laurance Reid Gas Conditioning Conference, Norman, Oklahoma, US, Feb. 2000. 13. Clark, P.D., Dowling, N.I. and Huang, M. ‘Mechanisms of CO and COS formation in the Claus

furnace’, ASRL quarterly bulletin, July-Sept. 1999. 14. Clark, P.D., Dowling, N.I., Huang, M., Cooper, J. and Butlin, G. ‘The chemistry of sulfur recovery by

the Claus process’, ASRL quarterly bulletin, Oct-Dec. 1998.