Application of Printed Circuit Heat Exchanger Technology

within

Heterogeneous Catalytic Reactors

By: A.M.Johnston, W.Levy, S.O.RumboldHeatric division of Meggitt (UK) Ltd

A paper presented at the AIChE Annual Meeting 2001

Copyright © Meggitt (UK) LtdNovember 2001

Introduction

The Printed Circuit Heat Exchanger(PCHE) is an established compact heatexchanger technology, originally inventedas a result of research performed at theUniversity of Sydney in the early 1980’s.Heatric was formed in Australia, in 1985,to commercialise the concept (1), and thefirst applications were in industrialrefrigeration systems. However, it soonbecame apparent that one of the biggestpotential markets for the new technologywas in offshore gas processing, wherespace and weight savings are at a premium.

Following Heatric’s move to the UK in1989, PCHEs rapidly gained acceptance inthe offshore industry, and some of the mostsignificant early projects that used themhave been extensively reported (2)(3)(4).Other applications include LNG, ethyleneoxide, sulphuric acid, naphtha reforming,and caustic soda plants (5). A total of over500 PCHEs have entered service aroundthe world.

PCHE Construction

The compact core of a PCHE isconstructed by chemically milling flowpassages into flat metal plates, and thenstacking and diffusion bonding the platestogether into a single block. The chemicalmilling technique is analogous to that usedfor the manufacture of electronic printedcircuit boards, and this gave rise to the“Printed Circuit” exchanger name.

Diffusion bonding is a high temperaturesolid state joining process that promotesgrain growth across the metal boundaries,resulting in a join exhibiting parent metalstrength and ductility.

If necessary, multiple diffusion bondedblocks may be welded together to formlarger units, before headers and nozzles arewelded on to complete the exchanger.

Alternatively, instead of welded-onheaders, it is also possible to incorporateflow distribution channels within thediffusion bonded block, in a so-called“ported” design.

PCHE Characteristics

PCHEs can be up to 85% smaller andlighter than an equivalent shell-and-tubeexchanger, as a result of the very highsurface area density, high heat transfercoefficient, and potentially near-perfectcounter current flow.

PCHE cores can be designed for extremetemperatures and pressures: units are inoperation at over 500 bar (7000 psi) andtemperatures ranging from cryogenic up toaround 800°C (1472°F). They can bedesigned for very low pressure drops, evenwhen handling highly viscous liquids orhot gas.

Process safety can be enhanced, sincePCHEs are not susceptible to hazards suchas flow induced tube vibration, or tuberupture, and the contained inventory isminimized. A wide range of corrosion-resistant materials may be employed, andthere are no gaskets or braze alloy.

A versatile plate design and manufacturingprocedure permits a wide variety of flowschemes to be used: more than one fluidstream can be included on a single plate; astream can be evenly distributed and mixedwith another stream on an adjacent plate.

New Opportunities for PCHETechnology

The design and fabrication techniquesdeveloped for PCHEs can be applied in arange of new and novel areas. Examplesinclude:

• mixer-reactors,• cooling/heating with heterogeneous

catalysis,• cooling and flow distribution plates for

fuel cells,• cooling of electronics,• reactions using supercritical solvents.

The potential for using a PCHE as a plugflow reactor was recognised by Heatric atan early stage (1). Since then, the mixer /heat exchanger / reactor concept has beenwidely investigated (6)(7), and Heatric arecurrently working with partners tocommercialise such devices: the programremains confidential, but we hope to beable to announce details at a future date.Fuel cell components are also an area ofactive development by Heatric. However,the primary focus of this paper is thetemperature control of heterogeneouscatalytic reactions.

Reactors for Heterogeneous CatalyticProcesses

A wide variety of different reactorconcepts have been used in heterogeneouscatalytic processes, including fixed bed,moving bed, fluid bed, and bubble column.Within each broad category there arefurther variants. For example, a fixed bedmay be arranged as one or more largepacked beds, or may be packed withinmultiple tubes (with external cooling /heating, or with feed quench), or may bepacked within the annulus of a double wallmulti-tubular reactor. Each arrangementhas been devised to address specificprocess requirements – especiallytemperature control. It is this challenge of

temperature control that is of particularinterest to us.

The concept of using PCHEs in reactors isaimed at processes that are either highlyexothermic or endothermic. Further, sincePCHEs are suited to extremes oftemperature and pressure, processesrequiring such extremes are alsopotentially of interest.

Catalysts are commonly supplied on aporous support material such as alumina,inevitably resulting in poor thermalconductivity within the catalyst bed.Consequently, large fixed beds tend tooperate adiabatically. Even in multi-tubular fixed bed reactors, commonly usedfor reactions involving large thermal loads- especially highly exothermic reactions, avery significant temperature gradientbetween the tube wall and the centre of thetube can arise. This poor in-tube heattransfer also frequently results in a highlynon-isothermal axial temperature profile,with a pronounced hot spot close to thetube inlet. Loss of selectivity and productdegradation are typical consequences ofsuch uncontrollable hot spots, whilstthermal runaway and even explosion riskmay occur in certain circumstances.

Such difficulties are sometimes addressedby adoption of some form of fluidised bedreactor but these can bring their owndrawbacks, including high capital cost,catalyst loss (through attrition), possibleflow instability or poor fluidisation, andrisk of line plugging. They are alsosubstantially back-mixed, which may bedetrimental to yield selectivity in certainreaction systems.

Before looking at how we might usePCHEs to improve reactor designs, letsremind ourselves of the primary reactordesign objectives. These are:

• to achieve an optimum temperatureprofile for selectivity andconversion;

• to minimize catalyst volume;• to avoid catalyst overheating (with

consequent irreversible damage);• to minimize pressure drop;• to achieve an optimal trade-off

between capital cost andperformance.

Application of extremely compact andhighly flexible PCHE heat transfertechnology allows us to address each ofthese objectives in new ways. We haveadopted the term Printed Circuit Reactoror PCR to refer to such PCHE reactorapplications.

“In Passage” PCR

The concept of coating a heat transfersurface with catalyst has been quite widelydiscussed: a recent study was presented atthe 4th International Conference on ProcessIntensification for the Chemical Industry(8). However, one of the big challenges ishow to provide sufficient catalyst surfaceat reasonable capital cost. The In Passageor IP PCR offers the desired high surfacedensity at demonstrably competitive cost,over a range of operating conditions farexceeding that offered by any other type ofcompact heat exchanger.

The benefits of such a coated reactor arediscussed in the case study given later inthis paper.

Multiple Adiabatic Bed PCR

Inevitably it will not always be possible tomatch (at least approximately) the requiredcatalyst surface with the required heattransfer surface. In this case the IP PCRrisks simply becoming a rather expensivecatalyst support, and an alternativeapproach is needed.

Where the required catalyst surface area isvery large, and substantially exceeds therequired heat transfer area, a better balancebetween capital cost and performance maybe achieved by approximating the in-passage reactor with a large number ofshallow adiabatic beds, with heat exchangebetween each bed – the Multiple AdiabaticBed or MAB PCR.

Hitherto the feasibility of such an approachhas been constrained by the cost ofsuccessive reactor vessels, heatexchangers, and interconnecting piping.Even in the most sophisticated integratedreactor designs, the number of reaction andheat exchange steps rarely exceeds 3 or 4,since the volume required for conventionalheat exchange within a reactor vesselbecomes increasingly costly withincreasing overall reactor vessel size.However, by making use of the verycompact nature of PCHEs it is possible todevise cost-effective reactor layouts withmany tens of adiabatic beds, withintermediate heating or cooling betweeneach.

A MAB PCR comprising an array ofshallow catalyst beds interposed betweenthin PCHE panels is considered in the casestudy.

Alternatively, a very large number of smalladiabatic beds can be incorporated within asingle PCR block, together with heatexchange zones to adjust the initial feedtemperature, to add or remove heat ofreaction between each catalyst bed, and toadjust the product temperature. Theheating or cooling medium may simply bea utility stream, or it too can undergo aseparate reaction within a separatesequence of catalytic beds. Such anarrangement appears uniquely suited toreactors in applications where space andweight are at a premium – for example fuelreforming for automotive fuel cell systems.

Catalyst Considerations

With the “in-passage” reactor in mind,Heatric is developing and evaluatingtechniques for applying catalyst coatings tothe passages within a PCR, and we believerobust and renewable coatings can beapplied in a cost-effective manner.However, perhaps the biggest challenge isthe availability of suitable catalysts.Substantially improved catalyst activity isneeded if we are to take full advantage ofthe opportunity offered by a coatedpassage reactor. Catalyst life and resistanceto poisoning or deterioration are also ofparamount importance when consideringlayers only a few microns thick.

More encouragingly, as has been reportedelsewhere (9), both catalytic coatings andfinely comminuted catalyst typicallyexhibit significantly improvedeffectiveness, when compared withcommercially available pellets in aconventional packed bed. For a reactorwith multiple small adiabatic bedscontained within a single PCR block, acorrespondingly small catalyst particle sizeis essential to assure good flow distributionand to avoid bypassing.

Reactant Addition and ProcessIntegration

The well-established and proven techniquefor injecting glycol into offshore gascoolers can be extended to admix reactantsinto a process stream on a passage-by-passage basis. Further, it is possible toconveniently achieve staged reactantaddition and mixing.

As previously noted, PCHEs offer theopportunity to combine multiple dutiesinto a single heat exchanger, eithersequentially or in parallel. Together withthe capability for very high thermaleffectiveness, these unique characteristicsprovide opportunities for significantlymore compact and cost effective reactordesigns.

Case Study: Phthalic Anydride

Background

Phthalic anhydride (PA) is predominantlymanufactured by the partial oxidation ofortho-xylene with air. The reaction isstrongly exothermic (∆H = 1285 kJ/kmol),and is carried out in a multi-tubular reactorcooled by circulating molten salt: themolten salt is in turn cooled by raisingsteam in an external boiler. The catalyst isV2O5 on promoted silica gel, and theoperating temperature range is 335°C to415°C.

The following case study, based on a50,000 tonne/year plant, shows clearly thesignificant potential capital cost andperformance advantages achievable byusing PCR technology.

Conventional Reactor

A typical multi-tubular reactor wouldcomprise c. 28,000 tubes, of 3 to 5 mlength, containing a total of 40 to 45 m3

catalyst. Indicative key operatingparameters are:

• o-xylene loading: up to 100 g/Nm3

air• Conversion: 70 to 71 %• Selectivity: 85 %• Catalyst: rings of 5 to 8 mm.

The capital cost of such a reactor is in therange of US$2.5-3.0 million.

Multiple Adiabatic Bed PCR

This concept achieves control of thereaction temperature throughout the reactor

by using multiple adiabatic beds alternatedwith heat exchangers in which the processstream is repeatedly cooled down to theoptimum inlet temperature to thesubsequent bed.

We have assumed the use of a structuredceramic matrix catalyst carrier, in order tominimize bed pressure drop, simplifycatalyst loading and unloading, andmaximize reactor compactness. In thiscase, gas flows along catalyst layers,instead of through the bed as in atraditional fixed-bed reactor: because thegas flows through straight channels thepressure drop is much lower than over apacked bed. Calculations show thatpressure drop through structured catalyst issome 10 times less than through a fixed-bed. This saving can either be takenentirely as an energy credit (keeping thesame superficial velocity as for a packedbed), or can alternatively be used toachieve a more compact, lower cost reactor(by increasing the velocity to give thesame pressure drop as a packed bed).

Key benefits of such a configurationinclude:

• The PCHE panel face area matchesthe catalyst face area, so reactantredistribution problems,recirculating flows and dead spotsare avoided.

• The PCHE and ceramic matrixstructure both comprise passageswith small dimensions – smallenough to quench combustion orexplosion.

• Any ortho-xylene feed loading ispossible, from 44 g/Nm3 to morethan 130 g/Nm3 in total. Indeed webelieve it is quite feasible tooperate entirely in a hydrocarbon-rich regime.

• Loading and unloading of catalystis straightforward, as the wholematrix can be replaced in a veryshort time. In addition, as catalyst

is prepared outside the reactor,there will be no need foradjustment of catalyst load (asfrequently needed in each of the28,000 tubes of a conventionalreactor).

• The cooling medium flow patterncan be readily configured to ensuretight control over catalysttemperature profile, and ifprofitable, ramping the temperatureupwards to compensate for reactiondepletion in the later beds.

• PCHE elements are cost-effectivelymanufactured in stainless steel.

In this case study we have considered aPCR with 50 adiabatic beds and 50 PCHEelements, with ramped temperature profile.Each PCHE element is 4 m high and 1.1 mwide, to match a catalyst bed facial area of4.4 m2.

For 16,000 m2 of structured catalystsurface, the catalyst volume will be 9 m3

and PCHE cores will have a total volumeof 5 m3, so the total PCR volume,(excluding feed and coolant headers) willbe 14 m3. This compares with anequivalent multi-tubular reactor volume ofapproximately 280 m3. This sizecomparison is illustrated in Figure 1. Wehave calculated a conversion of 72%, aselectivity of 78% for a total ortho-xyleneloading of 120 g/Nm3.

The estimated cost of such a PCR,excluding catalyst, is approximatelyUS$2.0 million. The overall operational

performance of this reactor gives potentialsavings of up to US$0.7-0.8 million peryear.

Figure 1: Comparison between MultipleAdiabatic Bed PCR and conventionalmulti-tubular reactor.

Catalyst-Coated IP PCR

A catalyst-coated heat exchanger is clearlyan attractive option for partial oxidationreactions, provided the coating:

• can be applied consistentlythroughout the structure;

• can be applied in a manner whichsustains suitable levels of activityand adherence throughout areasonable design life; andpreferably,

• can be replaced or rejuvenatedcost-effectively.

Catalyst-coated heat exchange surface canclosely control reaction temperature, to anextent not possible with catalyst–packedtubes, as there is no large thermalresistance equivalent to that between thegas, catalyst pellets and the tube wall.Consequently the hot spot designconstraint associated with packed tubes islargely eliminated and the averageoperating temperature of the reaction maybe safely increased, subject only toconstraints imposed by catalyst aging andundesirable side-reactions.

Key benefits of a coated IP PCR include:

• The reactant flow contacting thecatalyst surface has no dead-spots orrecirculating flows, which are knownto compromise PA selectivity.

• The level of turbulence in the reactantflow can be readily adjusted tominimize gas-phase mass transferresistances, which would becomeincreasingly important at higherreaction rates. (We understand, on theother hand, that PA catalyst isrelatively non-porous and porediffusion is therefore unlikely to beimportant.)

• Process pressure drop can be readilyoptimized, trading off compressorpower against reactor cost.

• By skillful application of counter-,cross- and co-flow, the coolingmedium flow pattern can be configuredto ensure tight control over catalysttemperature profiles.

• PCRs are cost-effectivelymanufactured in stainless steel.

• The PCR structure is inherently small-scale, and can be designed to quenchany tendency for explosion. Therefore,there appears to be no fundamentalconstraint on ortho-xylene loading.

• The pre-heaters and post-coolers maybe conveniently configured as PCHEsand close-coupled to the reactor,thereby minimizing piping andstructure costs and avoiding thehandling of explosive mixtures in largepipes.

Higher catalyst operating temperaturescould well permit substantial reductions inrequired catalyst active surface. Thiseffect could potentially be quitepronounced, since it appears the PAreaction rate roughly doubles for every20EC temperature rise. For example, if thepeak temperature at the center of a packedreactor tube were 50EC higher than theaverage temperature throughout thereactor, then conceivably only 25% of the

catalyst area would be required for a givenyield of product in a catalyst-coatedreactor.

For the capacity under consideration, theheat release in the reactor is approximately16MW. Hence, if we consider the same16,000 m2 of catalyst surface as for themultiple adiabatic bed case consideredabove, the heat flux would only average1kW/m2. Clearly there is scope for at leastan order of magnitude increase in heatflux, without overwhelming the heattransfer capabilities of the PCR core.Suitable ramping of the temperature profilecould hold the heat fluxes relativelyconstant throughout.

By making use of the full capabilities ofthe coated PCHE reactor, the capital costmay be brought down to around US$1.5-2.0 million, offering a significant savingover conventional technology.

Other Options

Reaction selectivity is adversely affectedby the over-oxidation of PA to CO/CO2.Yield improvements might therefore beachieved through options such as:

• a cascade of short-bed, low-conversionreactors with intermediate recovery ofPA. PCHEs and PCRs are well suitedto providing the required cascade ofheat exchange surface cost-effectively.

• operation in the hydrocarbon-richregime, perhaps with staged injection

of air. PCRs are able to incorporatesuitable means for intermediatereactant injection with intimate mixing.Enhanced safety would also be abenefit of operation under this regimeif the mixtures were above the upperflammability limit.

Conclusions

PCHEs can be applied in novel reactordesigns in several different ways, toachieve (near) isothermal conditions.Options include catalyst coating, ormultiple adiabatic beds. Further possiblefeatures include staged reactant addition,and sophisticated heat integration.

Both the example tabled herein, and otherongoing design and test activities, indicatethe use of PCHE technology can offersubstantial economic benefits wherever:

• improved temperature control will giveimproved yields, or

• in instances where compact heattransfer can reduce capital cost, or

• wherever space and weight are at apremium.

In specific circumstances, further safetyand utility cost benefits may also berealized.

References

(1) Johnston, Tony, The ChemicalEngineer, December 1986, pp36-38

(2) The Offshore Engineer, May 1991(3) Knott, Terry, The Offshore

Engineer, October 1993(4) Esau, Iain, The Offshore Engineer,

October 1994, pp27-28(5) Samdani, Gulam, et al, Chemical

Engineering, June 1993, pp30-32(6) Jones, D.O., “Process

Intensification of Batch ExothermicReactors”, Health & SafetyExecutive, 1996

(7) Green, A., Johnson, B., Arwyn, J.,Chemical Engineering, December1999, pp66-73

(8) Babovic, M., Gough, A., Leveson,P., Ramshaw, C., Catalytic PlateReactors for Endo- and ExothermicReactions, 4th InternationalConference on ProcessIntensification for the ProcessIndustries, Brugge, September2001

(9) Oroskar, A.R., et al., ProcessIntensification by Miniaturization,Proceedings of MicroTec 2000,Hanover, September 2000.

(10) Froment, G.F., and Bischoff, K. B.,Chemical Reactor Analysis andDesign, John Wiley and Sons,March 1990

(11) Arpentinier, P., Cavani, F., andTrifiro, F., The Technology ofCatalytic Oxidations, EditionsTechnip, March 2001

Filename: Paper 352a illustrated BW.docDirectory: F:\SHARED\PCRTemplate: C:\Program Files\Microsoft

Office\Templates\Normal.dotTitle: Application of Printed Circuit Heat Exchanger

Technology within Heterogeneous Catalytic ReactorsSubject:Author: HeatricKeywords:Comments:Creation Date: 24/01/02 18:09Change Number: 10Last Saved On: 28/01/02 09:59Last Saved By: HeatricTotal Editing Time: 100 MinutesLast Printed On: 01/02/02 08:56As of Last Complete Printing

Number of Pages: 10Number of Words: 3,087 (approx.)Number of Characters: 17,599 (approx.)