8/3/2019 eig16

1/40

Vol. B4 Fixed-Bed Reactors 199

Fixed-Bed ReactorsGERHART EIGENBERGER, Institut flir Chemische Verfahrenstechnik, Universitiit Stuttgart, Stuttgart,Federal Republic of Germany

1.2.

Introduction . . . . . . . . . . . . . . . . . . . . . . . 200Catalyst Forms for Fixed-Bed Reactors 202

2.1. Gas-Catalyst Mass and Heat Transfer 2032.1.1. Random Pac kings . . . . . . . . . . . . . . . . . 2032.1.2. Monolith Structures . . . . . . . . . . . . . . . . 2042.2. Flow and Pressure Loss in Fixed Beds 2052.3. Heat Transport Transverse to the FlowDirection . . . . . . . . . . . . . . . . . . . . . . . . . 2062.4. Comparison and Evaluation of DifferentCatalyst Forms . . . . . . . . . . . . . . . . . . . 2072.4.1. Catalyst Forms for Adiabatic

Operation. . . . . . . . . . . . . . . . . . . . . . . . 2072.4.2. Catalyst Forms for IsothermalOperation . . . . . . . . . . . . . . . . . . . . . 2093. Adiabatic Reaction Control . . . . . 2\04. Reaction Control with Addition or

Removal of Heat in the Reactor . . . . . . 2124.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . 2124.2. Adiabatic Multistage Reactors ~ i t h Interstage Heat Transfer . . . . . . . . . . 214

In additIOn to the standard symbols definedm the front matter and on pp. 6-9 and 121-t 23 of this volume. the following symbols areused:ap specific outer surface area of catalyst.mllm3 packingA flow cross-sectIOnal area, m2c molar concentratton. mol/m 3Co feed concentratton. mol/m 3cpG specific heat of reaction gas, kJ kg - I K - Ic, specific heat of catalyst. kJ kg - I K - ID diffusion coefficient. mllsdp catalyst particle diameter, mdh hydraulic diameter (dh = 4 AIU). m11.2 Ergun coefficients (Eqs. 2.4. 2.5)Gz mass-flow velocity. kgm-ls- IkG gas-catalyst mass-transfer coefficient.m/sL length of catalyst bed. m

Ullmann's Enc)c1opedlaof Industnal Chemistry. Vol B4

4.3. Reactors with Heat ExchangeIntegrated in the Fixed Bed . . . . . . . . . . 2164.3.1. Heat Transfer Media for Fixed-BedReactors . . . . . . . . . . . . . . . . . . . . . . . . . . 2164.3.2. Reactor Designs . . . . . . . . . . . . . . . . . . . 2184.3.3. Influencing the Course of the ReactIOn 220S. Autothermal Reaction Control . . . . . . .. 2235.1. Introduction. . . . . . . . .. . . . . . . . . . . . . 2235.2. Reaction Control ~ i t h Periodic FlowReversal . . . . . . . . . .. . . . . . . . . . . . . 2245.3. Reactor Behavior. . . . . . . .. . . . . . . . . . 2256. Stability, Dynamics, and Control ofIndustrial Fixed-Bed Reactors . . . . . . . . 2286.1. Introduction. . . . . . . . . . . . . .. . . . . . . 2286.2. Parametric Sensitivity . . . . . . . . . . . . . . 2296.3. Multiple Stationary States . . . . . . . .. 2296.4. Migrating Reaction Zones . . . . . . . . . . . 2316.5. Safety Aspects . . . . . . . . . . . . . . . . .. .. 2346.6. Control of Fixed-Bed Reactors . . . . . . 2357. References... . . . . . . . . . . . . . . . . . . 237

NNilQReScShTToUUwt'GVVRII'=ZL':1. w

energy required for coolant circulation.kWNusselt numberheat flow to be transferred. kWReynolds numberSchmidt numberSherwood number (Eq. 2. to )temperature, Kfeed gas temperature. KCircumference. mgas-catalyst heat-transfer coefficient.Wm- 2 K- Iinterparticle velOCity. m/svolume flow rate. m3 /sreactor volume. m 3migration velocity of reaction zone. mlsspace coordmate, mreactor length. m\\all heat-transfer coefficient.Wm- 2 K- 1

(" 1992 VCH Pubh,hers. Inc0-89573-539-3 92 S 5.00 + 50

8/3/2019 eig16

2/40

200 Industrial Reactorsvoid fraction, mJ free space/mJ packingdynamic viscosity, N s m - 2reaction enthalpy, kJ/molpressure drop, N/m2adiabatic temperature change, Kcooling temperature change between inletand outlet, K

L1x molar conversion, mol/molAd r effective axial thermal conductivity,Wm-1K- 1

molecular gas thermal conductivity,Wm-1K-1effective radial thermal conductivity,Wm-1K- 1density, kg/m Jgas density, kgjm3density of catalyst particle, kgjm 3pressure drop coefficient

1. IntroductionCatalytic fixed-bed reactors are the most important type of reactor for the synthesis of large

scale basic chemicals and intermediates. In thesereactors, the reaction takes place in the form ofa heterogeneously catalyzed gas reaction on thesurface of catalysts that are arranged as a socalled fixed bed in the reactor. In addition to thesynthesis of valuable chemicals, fixed-bed reactors have been increasingly used in recent yearsto treat harmful and toxic substances. For example, the reaction chambers used to remove nitrogen oxides from power station flue gases constitute the largest type of fixed-bed reactors asregards reactor volume and throughput, whileautomobile exhaust purification represents byfar the most widely employed application offixed-bed reactors.With regard to application and construction,It is convenient to differentIate between fixedbed reactors for adiabatic operation and thosefor nonadiabatic operation. Since temperaturecontrol is one of the most important methods ofinfluencing a chemical reaction, adiabatic reactors are used only where the heat of reaction issmall, or where there is only one major reactionpathway; in these cases no adverse effects onselectivity or YIeld due to the adiabatic temperature development are expected. The characteristic feature of adiabatic reaction control is thatthe catalyst is present in the form of a uniform

Vol. 84

Heatcarrier

Heatcarrier

Figure 1.1. BaSIC type-; of catalytiC fixed-bed reactorsA) AdiabatiC fixed-bed reactor; B) Multitubular fixed-bedreactor

fixed bed that is surrounded by an outer insulating jacket (Fig. 1.1 A). Adiabatic reactor designsare discussed in Chapter 3.Since the incoming reaction gases in mostcases must be heated to the ignition temperatureof the catalytic reaction, adiabatic reactIOn control is often coupled with heat exchange betweenthe incoming and exiting reaction gas resulting inso-called auto thermal reaction control. ThIS typeof reaction control offers certain specific featuresand development perspectives, which are discussed in Chapter 5.

8/3/2019 eig16

3/40

Vol. B4Reactions with a large heat of reaction as wellas reactions that are extremely temperature-sensitive are carried out in reactors in which indirectheat exchange occurs via a circulating heat transfer medium integrated in the fixed bed.Since in most cases the task of the heat-transfer cycle is to maintain the temperature in thefixed bed within a specific range, this concept isfrequently described as an "isothermal fixed-bedreactor". However, isothermal reaction controldoes not provide optimum selectivity or yield inall cases, and for this reason the concept of heatexchangers integrated in the fixed bed is also be-ing increasingly used to achieve nonisothermaltemperature profiles.The most common arrangement is the multi

tubular fixed-bed reactor, in which the catalyst isarranged in the tubes, and the heat carrier circulates externally around the tubes (Fig. 1.1 B).Fixed-bed reactors with an integrated heat supply or removal are discussed in Chapter 4.Fixed-bed reactors for industrial synthesesare generally operated in a stationary mode (i.e.,under constant operating conditions) over prolonged production runs, and design thereforeconcentrates on achieving an optimum stationary operation. However, the nonstationarydynamic operating mode is also of great importance for industrial operation control. In

Feedl tream

Fixed-Bed Reactors 201particular, fixed-bed reactors with a stronglyexothermic reaction exhibit an, at times, surprising operational behavior which is dIscussed inmore detail in Chapter 6.Within a production plant the reactor mayjustifiably be regarded as the central item of apparatus. However, compared to the remainingparts of the plant for preparing the feed and forseparating and working-up the products, often itis by no means the largest and most cost-intensive component. In many cases the achievableconversion in the reactor is limited for thermodynamic (equilibrium) and kinetic reasons (selectivity). It is then usual to separate the materialdischarged from the reactor into products andunreacted feed components (see Fig. 1.2), whic!1are recycled to the feedstock. This recycling procedure involves costs1) For product separation2) For recycle compression3) For repeated heating and cooling of the circulating matenal to the reaction temperatureand the temperature of the separating device4) Due to loss of product resulting from theneed to remove part of the circulatmg material to limit the amount of mert substances

or byproducts in the recycle stream (bleedstream).

Bleedstream

ProductFigure 1.2. Reaction cycle for syntheSIS reactions With incomplete conversiona) Fixed-bed reactor. b) Feed preheater; c) Exit cooler; d) ReCirculatIOn compressor. e) Separation column

8/3/2019 eig16

4/40

202 Industrial ReactorsTo minimize these costs it is therefore necessary to maximize the conversion in the reactorand avoid as far as possible inert accompanyingsubstances in the reaction mixture. With irreversible reactions (e.g., partial oxidations) thetrend is therefore towards a highly concentrated,approximately stoichiometric feed composition,which may occasionally be in the explosiverange. The resulting problems are discussed inChapters 4 and 6.

2. Catalyst Forms for Fixed-BedReactorsThe heart of a fixed-bed reactor and the siteof the chemical reaction IS the catalyst. The processes taking place on the catalyst may formallybe subdivided into the following separate steps:

I) Diffusion of the reactants from the gas spacethrough the outer gas - particle boundary layer, macropores, and micropores2) Chemisorption on active centers3) Surface reactions4) Desorption of the products5) Back-diffusion of the products into the gasspace

Since most reactions take place with a considerable heat of reaction, a corresponding heattransport is superimposed on the mass transport.The control of the microkinetics, consistingof micropore diffusion, chemisorption, surfacereaction, and desorption, is the task of the catalyst developer and IS not discussed further here.If the catalyst is specified together with its microklnetic properties, then reaction conditions(feedstock concentrations, pressure, temperatureprofile, and residence time) can be found thatlead to optimum yields. The reaction engineermust determine these conditions and ensure thatthey are maintained in an Industrial reactor.In the case of selectivity-sensitive multistepreactions, any deviation from the optimum values Inevitably leads to a decrease in yield. ThiSappIJes to deviations from the uniform residencetime distribution due to flow dispersion and flowbypass phenomena in the fixed bed, as well as todeviations from uniform reaction conditions inthe catalyst as a result of mass-transport resistance in the particles and the outer boundarylayer [2.1].

Vol. B4The influence of mass-transport resistance inthe particles can only be excluded if the criticalreaction rate is substantially lower than the masstransport velocity. This leads on the one hand tothe need for good external mass transfer (i.e. toreasonable flow conditions in the packed bed), aswell as to short diffusion paths in catalyst particles and sufficiently large pores. On the otherhand (in the case of exothermic reactions) thelocal reaction rate must be controlled and limitedby the packed-bed temperature.Temperature control thus plays a predominant role in selective reaction control in general,and in particular in the case ofexothermic multistep reactions. Under nonadiabatic conditions,catalysts must therefore be assembled and arranged in the fixed bed in such a way as to ensuregood heat transport to the heat-transfer medium.A further requirement placed on the catalyst

is a low flow pressure loss. This applies particularly if the reaction conversion in a singlethroughput is low, so that the reaction has to becarried out with a large circulating gas ratio(Fig. 1.2), as well as to off-gas pUrification, inwhich large off-gas streams must be handledwith minimal additional cost.Finally, the catalyst should be available in asufficiently high concentration in order to keepthe construction volume of the reactor low. Thedecisive parameters here are the speCific externalcatalyst surface a p ( = square meters of cata-

CD

Figure 2.1. Usual shapes of monohth catalystsA) Square-channel monohth. B) Parallel-plate monolith.C) Corrugated-plate monohth

8/3/2019 eig16

5/40

Vol. B4Iyst surface per cubic meter reactor volume) forreactions controlled by external mass transfer,the catalyst fraction I - E, where E = cubic meters of free gas space per cubic meter of reactorvolume, describing the void fraction of the fixedbed.The above requirements are to some extentcontradictory, which has led to the propositionof a large number of different catalyst forms andarrangements. However, only a few of the e haveproved really effective in practical operation.Suitable catalyst forms and arrangements include random packings of spheres, solid cylinders, and hollow cylinders, as well a uniformlystructured catalyst packings in the form ofmonoliths with parallel channels, parallelstacked plates, and crossed, corrugated-platepackets (Fig. 2.1).

2.1. Gas-Catalyst Mass and HeatTransfer2.1.1. Random Packings

Industrial fixed-bed reactors a re generallyoperated with a cross-sectio nal loading G;:. I kg gas per square meter of reactor cross section per second. This loading produces a sum-

Boundary layerPellet radius-

Figure 2.2. Temperature a nd concentration profiles ror apartial oxidation reaction in a spherical catalyst pellet

Fixed-Bed Reactors 203ciently strong turbulence in random packing. Asa re ult the external gas - catalyst mass-transferresistance is small compared to the transport resistance in the catalyst pores .However, generally the thermal conductivityof the catalyst matrix is larger than of the gas.This mean that the external gas - catalyst heattransport resistance exceeds the thermal conduction resistance in the catalyst particles. The temperature and concentration conditions estabIi hed in a pherical catalyst are illustrated for apartial oxidation reaction in Figure 2.2. Theconditions can be calculated from the modelequation given in -+ Model Reactors and TheirDesign Equation , pp. 141 - 142 , under the assumptions made there. An essential preconditionis that the catalyst particle i uniformly exposedover its entire urface to a flow with uniformtemperature and concentration. This is , ofcourse, never the case in random packings. Figure 2.3 shows the local mas -transfer distribu-

- ( j ,........ i )-' tr@.

Figure 2.3. Local mas -transrer distribution at the surraceor ind ividual cylindrical or ring-shaped catalyst pellets in afixed-bed packing

8/3/2019 eig16

6/40

204 Industrial Reactorstion, providing an insight into the overall highlyinhomogeneous co nditions. The visualizationtechnique is based on the reaction of traces ofammonia in the gas st ream with a catalyst surface impregnated with manganese chloride so lution . The co nversio n to manganese dioxide is already so fa st under ambient conditions that itpractically depends only on the external mas -transfer resistance of the gas boundary layer.The intensity of the dark coloration is thus directly proportional to the local reaction rate ofthe urface reac tion. At a constant ammoniaconcentration in the gas flow it is thus also proportional to the local ma ss transfer, and if therna s transfer and heat transfer are equivalent,also to the local heat transfer [2.2).

Figure 2.3 shows that the local conditions inrandom packings are much more complex thanas umed in current conventional models. Nevertheless, the models for fixed beds containing alarger number of catalyst particles over the cro ssection can provide reliable information if it isborne in mind that they only describe the meanvalue of a process that varies greatly a regardsdetail. For the same reason , it is inappropriate tocompare model predictions with a few local temperature or concentration measurement . Indeed , a certain degree of averaging is also nece -sary in the measurement procedure (see Section2.2).

The literature contains a number of correlation equations for the mean gas - catalyst masstran fer and heat transfer as a function of gaspropertie , cataly t geometry, and flow conditions [2 .3], [2.4). However, in practice they playonly a minor role for catalyst packings sincedesign and simulation calculations are frequently performed with a model that is qu asi-homogeneous, at least with regard to temperature(-+ Model Reactors and Their Design Equations, pp . 140 - 144 ). The reason for this a re thea bove-mentioned trong local fluctuation s,which make differentiation between the gas temperature an d catalyst temperature difficult.

2.1.2. Monolith tructuresIn contrast to random packings, the externa l

heat transfer and mass transfer in monolith catalysts is much more uniform , but they can alsobecome limiting factors at high reaction rates.This applies in particular to channel-type monoliths with narrow parallel channels. where the

Vol. B 4

A

B

Figure 2.4. Visualization of mass transfer in monolithst ructu res (now from left to right)A) Square duct; 8) Co rrugated monolith 12 .5]

flow i generally laminar under indu trial operating conditions. Example include monolithcatalysts with a square-channel cross section fora utomobi le exhaust purification and for the removal of nitrogen oxides from flue gases .

Figure 2.4 shows results of the visualizationof the local mass transfer by using the ammonia manganese chloride reaction (Section 2.1.1).The marked decrease in coloration in the flowdirection is mainly the result of the increa ingcon umption of ammo ni a in the wall boundarylayer, so that the reactants diffusing to the wallhave to travel an ever increasing distance fromthe flow core (bui ld up of the laminar boundarylayer). The depletion is particularly pronouncedin the corners, since two reaction surfaces meethere. The more acute their enclo ed angle, thegreater is the deplet ion in the corner region andthe sma ller is the contribution of the wall su rfaceTabte I. Asymptotic dimensionless laminar no w heat orma ss- transfer coefficien ts Nu = ,,_ . dh/;'G (constan t wallconditions) and fanning friction factor I for pressure dropfo.p = 2/("Zud:) . I'G fo r ducts of different cross section[2 .6)Geometr) Nu I dh{;; 2.47 13 .33 20 v 3

2.98 14 .23 a

0 3.34 15 .05 3a3.66 16 .00 a0 7.54 24.00 20

8/3/2019 eig16

7/40

Vol. B4 Fixed-Bed Reactors 20510 2

D Re=600 Re=1200C> Re=1800+ Re=2200t 0

..-10 1ti

..

'"'"

10L-__ ~ - L ~ ~ W W U L ~ __ L - ~ ~ ~ ~ ~ ~ ~ __ L - ~ - W ~ 10-4 10-3 10-2 10-1

Z*=Zl/(dh ReSc)_Figure 2.5. BehavIOr of lhe dimensionless heal-transfer and mass-tran,fer coefficients over the dimensIOnless tube lengthNu ='w db/i.". Sh = kG d.ID

to further heat transfer and mass transfer. TheeffiCiency of channel monoliths of equal crosssectional area but different shape thereforedecreases in the sequence: cIrcle. hexagon.rectangle. tnangle. ThiS is illustrated in Table t.which gives the asymptotic dimensionless masstransfer and heat-transfer coefficients for tubesof the above cross sections. As Figure 2.5 shows.the asymptotic limiting value is established aftera flow length roughly corresponding to to timesthe hydraulic channel diameter. The asymptoticlimiting value from Table t can therefore be usedto perform estimation calculations in conventional industnal monolith catalysts with a largelength-to-diameter ratio. For more accurate calculations the radially and aXIally vanable velocity. conCf'ntratJon. and temperature profiles mustalso be taken into account [2.7].In contrast to the monolith types discussed sofar. the flow conditions m corrugated-platemonoliths (Fig. 2. t C) are turbulent under normal industrial velocities. The uniformity of themass-transfer distribution [2.8] depends on geometrical parameters (wave form. amplitude.wavelength. angle of inCidence). The transfercoefficients are considerably higher thanthose of lammarly traversed channel monoliths(Fig. 2.4 B). but the pressure loss IS also high.These structures offer conSIderable advantages

for convective heat transport transverse to theflow direction and for transverse mIXIng. whichare discussed m more detail in the following sections.

2.2. Flow and Pressure Loss inFixed BedsConventional industrial catalyst forms differconsiderably as regards pressure loss. For example. for equal mean dImensIOns and the sameproportIon of empty space. random packmgs

generally have a conSIderably higher pressureloss than monolith structures. and among these.corrugated structures have a hIgher pressure lossthan monoliths with straight parallel channels.Mass transfer and heat transfer are strongly correlated with the pressure loss. For reasons ofenelgy demand. the catalyst form for a givenprocess is chosen to combme the reqUIred masstransport and heat transport with the lowestpressure loss. However. It should be borne inmind that the flow into the fixed-bed reactor(Fig. t. t) is generally achieved by means of afeed pipe and a distnbution hood These musttherefore be constructed so that the fixed bed ortube bundle is uniformly traversed. and the gasreSIdence time m each tube or each flow filament

8/3/2019 eig16

8/40

206 Industrial Reactorsof the fixed bed is the same. This requirement canalways be met simply if the pressure loss in thefixed bed or tube bundle is substantially greaterthan in the entrance hood. In this case it needonly be ensured that the fixed bed is packed souniformly that no flow bypass is formed or, inthe case of a muItitube reactor, the pressure lossis equalized in all the individual tubes of the bundle. Because the pressure loss increases with thesquare of the flow velocity, a uniform distribution of the flow then occurs automatically.

Obtaining a uniform flow distribution is substantially more difficult If the pressure loss of thefixed bed is small. This holds in particular formonolith reactors with straight, laminarly traversed channels (see Section 2.3).

The Hagen - Poiseuille equatIOn is used tocalculate the pressure loss in laminarly traversedmonolith channels(2.1)

The pressure loss of packed tubes can be described either by means of a pressure loss coefficient and the pressure drop equation6p = (. 2or by the Ergun equation:

where for spherical packmgs:

(2.2)

(2.3)

(2.4)

(2.5)

Thus the pressure drop depends very stronglyon the void fraction f. of the packmg. Of thestandard forms for packed catalysts, hollowcylinders ofthm wall thickness (r.:::: 0.6-0.8) aretherefore preferred to spheres (I: :::: 0.37 -0.4) andsolid cylinders (r. :::: 0.35).

The strong dependence of the pressure losson the void fraction underlines the importance ofpacking catalyst beds carefully to avoid bypassflow due to local variations in the packing density. For the same reason the tubes of multitubular reactors for highly exothermic. selectivitysensitive reactions are often filled uniformly by

Vol. 84means of special devices and if necessary individually compensated for pressure loss.

The pressure loss coefficient, (Eq. 2.2) canbe determined for typical packing forms, for example, according to [2.9).

2.3. Heat Transport Transverse to theFlow DirectionWith nonadiabatic reaction control, heat

must be transported perpendicular to the flowdirection through the fixed bed to the heat exchange surfaces. At the usual mass flow rates ofG = 1 kg m - 2 S I , this heat transport takesplace mainly by convection, i.e., the fixed bedmust be constructed so that flow componentstransverse to the main flow direction occur locally. Monolith structures with straight parallelchannels are thus unsuitable for nonadiabatic reaction control.In catalyst packings transverse flow components are automatically established as a result ofthe nonuniform arrangement and the twistedflow around the pellets. Hollow and full cylinders with a length-to-diameter ratio of 1 to 3 areparticularly effective in this respect.Despite the fact that radial heat transport inthe form of twisted flow takes place mamly byconvection. it is formally described by means ofa so-called effective thermal conductivity I.,transverse to the flow direction.

In addition to heat transport through thefixed bed perpendicular to the maIO flow directIOn. the heat transport at the boundary betweenthe fixed bed and heat exchange surface IS alsodecisive for the heat exchange. The latter heattransport is generally described by a wall heattransfer coefficient :xw in which the complex interplay between convective flow at the tube walland conduction transport by contact betweenthe fixed bed and the heat exchange surface ISdescribed 10 overall terms. Heat transport 10packed tubes has been investigated and discussedin detail [2.4). [2.10). However. the correlationsfor :xw and i., given in the literature do not adequately take account of the actual velocity distribution 10 packed tubes. Plug flow was generallyassumed. although the actual velocity profile isnonuniform with a pronounced slip at the wallThis IS due to the fact that the pellets make onlypoint contact With the wall. whereas they overlapand cross over one another inSide the packmg.

8/3/2019 eig16

9/40

Vol. B4

1.0

c: 05u'".t-0'0>

O L - - - - - - - - - - - ~ - - - - - - - - ~ Rad i u s _t 5.0I 4.0> .

3.0OJ>2.0

.; : to,1; O L - - - - - - - - - - - ~ - - - - - - - - - - ~ Rad i u s _Figure 2.6. RadIal d,stribut,on or the vOId rracllon andaXIal no" \eloclt) on a tube packed with spheres

thereby reducing the free volume and hence thevelocity. The conditions for a spherical packingare illustrated schematIcally In FIgure 2.6. Theradially varying empty space distnbution andvelocity distribution must be taken into accountin detailed reactor calculations. as well as in

Fixed-Bed Reactors 207the detenninatJOn of accurate heat-transportparameters [2.12). [2.13).Existing correlation equations for calculatingthe heat-transport parameters were obtainedfrom heating or cooling experiments without reactions and assuming plug flow; they thereforepennit only a semiquantitative evaluation. However. this is adequate for qualitative comparisonof catalyst structures.

2.4. Comparison and Evaluation ofDifferent Catalyst FormsThe choice of a suitable catalyst fonn is always an optimIzation problem that can be com

pletely specified only for a specific process. Eventhen. however. weighting the perfonnance functIon is not easy since. for example. small pressurelosses. unifonn flow through the reactor. andgood mass- and heat-transfer properties generally represent opposing requirements. The following assessment is therefore only a rough guide.

2.4.1. Catalyst Forms for Adiabatic OperationDecisive parameters for adiabatic operationmclude:

1) The active catalyst surface available per unitreactor volume2) The quahty of the mass and heat transfer between the flowing gas and the active catalystsurface3) The flow pressure loss4) The unifonnity of the flow through the reactor and thus the degree of utilization of thefixed bedThe major proportion of the active catalyst

surface IS located In the Intenor of the porouscatalyst structure However. ifit IS assumed that.In the case of sufficIently fast reactIons. the reaction site IS restncted to a thm layer underneaththe external surface. then the actIve catalyst surface area can be taken as proportional to thespecific external catalyst surface area up' If theumfonn distributIon of the flow entering thefixed bed is regarded as a separate problem. thenfor gIven kinetic conditions the evaluation maybe restricted to three parameters. specIfic external surface area up. gas - sohd mass-transfer coefficient kc.. and the flow pressure loss. Theevaluation becomes particularly simple if. wIth a

8/3/2019 eig16

10/40

208 Industrial Reactorssingle dominant reaction, the external gas-solidmass transfer limits the reaction, since in thiscase the specific reaction kinetics have no Innuence. The relationships for a concentration cof a key component A and a simple reactionA + ... -+ products are outlined in [2.11). Assuming mass-transfer limitation. the materialbalance for the key component along the nowpath is

oc["1' -=-k 'Q c c, oz G p (2.6)

If the reactton has a conversion of 99 %, i.e .an outnow concentration of 1% of the innowconcentratIOn, the length of the fixed bed ZL is,by Integration.

For the reqUired reactor volume V. with thecross-sectional area A:

(2.8)

Correlations for kG can be found in [2.3).[2.4). The computation is particularly sImple forstraight, laminarly traversed channels. Theasymptotic limiting value given in Table 1 (Section 2.1) for the square channel

. . h 4gIves. wit ap = -- ,dh . /;12Dkc,' a ::::: - - ,

p d.' /;and

V. . [.V. =O .38 -D-

(2.9)

(2.10)

(2.11)

If Equation (2.1) is used for the pressure loss.then for the laminarly traversed monohth channel. from Equations (2.7) and (2.\0)

12.28 . '1 , 2!'J.p = - - - - . 1'(', . /'D . (2.12)

Vol. 84Using the above equations, reactor variants

for catalytic off-gas purification with a given volume now Vwere calculated in [2.11). The resultis shown in Figure 2.7. The pressure drop of thefixed bed over the required reactor length is plotted; the curve parameter is the hydrauhc diameter dh of the monolith channel or packing body for different empty-tube gas velocitIesv. = E . VG . According to Equation (2.11) the required fixed bed volume V. with square monolith catalysts and a given throughput V dependsonly on dh , whereas the geometric arrangement(small bed cross section A and long bed lengthZL' or large bed cross section and short bed) hasno innuence on V. However, the bed cross-sectional area A for a given gas throughput Ii determines viav= A ' [ 'G ' ' (2.13)the interparticle velocity I 'G' an d according toEquatIOn (2.12) thiS then appears as a quadraticterm in the pressure loss.

In conclusion. a mln[mum catalyst volumewith minimum pressure loss is obtained by usinga very nat bed with a large now cross-sectionalarea A and small hydraulic channel diameter orparticle diameter ci

h This result applies in gener

al to all catalyst forms. The main difficulty withthis arrangement [s the uniform distributIOn ofthe now velocity. Figure 2.7 shows the improvement in pressure loss that can be achieved bychannel monoliths as compared to sphericalpackings. Other catalyst forms such as Raschigrings or corrugated-plate packets lie betweenthese boundary curves.

2500\2000"'... 1500a."" 1000::J1/11/1 500a..

00

.-. Spheres vG=O.5m/s __A-A Monoliths ____ 14_ 12____ a......... 5361

VG=O.5 m/s.-----6 61 3 41 2 3 4 5Reactor length, m_

Figure 2.7. Pre"ure drop mer the lixedbed length for acatal)tlc combu,tlon reactor \\lth gIven throughput and99'0 converSIon J ' ,I functIon of the h ~ d r a u h c catJlystdiameter dh for dlilerent gas cmpt) -tube velOCItIes Ir. I I I

8/3/2019 eig16

11/40

Vol. B42.4.2. Catalyst Forms for IsothermalOperation

In addition to the previously discussed quantities, the lateral heat-transport parameters ofthe fixed bed are further decisive parameters inisothermal reaction control. As shown in Section2.3, heat transport can be characterized by theeffective thennal conductivity perpendicular tothe main flow direction ;., and the wall heattransfer coefficient ~ w . Both quantities arestrongly dependent on the filling or packingfonn. Since transport in the industrially interesting region mainly occurs by convection, theyare approximately proportional to the massthroughput.

For the purposes of overall companson. thecorresponding characteristic parameters are

3000 Dap2500 nnn c1 0001500

;:;..E 1000" ' ~ 500 rI I I I I I

d e

30 300 o a."25 250 mn A,20 100 . 6 p / L

"" 15 150;-E E3 10 3 100- < ~

05 ts 500 b d e

I9

9

Fixed-Bed Reactors 209given in Figure 2.8 for some industrially Important filling and packing fonns in a tube of 50 mmmternal diameter with a mass-flow velocity ofGz = 1 kgm- 2 s- l .The dimensions of the packing bodies werechosen so that their specific external surface areaap is ca. 500 m21m 3. Under these conditions hollow, thin-walled cyhnders have clear advantagesover other packmg fonns. exhibiting the lowestpressure loss and the highest thennal conductivIty. Only as regards wall heat transfer are theyinferior to spheres or cylinders. However. goodwall heat transfer is apparently less deciSIve froma reaction engineering viewpomt than good radial thennal conductivity, since the fonner can becompensated by an appropriate temperatureprofile of the heat-transfer medium. whereas theradial thennal conductivity directly mfluences

09

h

504030 E"-20 E10 ci.

8/3/2019 eig16

12/40

210 Industrial Reactorsthe uniformity of the reaction conditions overthe tube cross section. On the other hand. solidcylinders with a large length-to-diameter ratiohave good heat transport values. but at the costof a very high pressure loss.

Despite its poor heat transport properties. amonolith with straight, parallel channels, such asused for automobile exhaust control. is includedin the companson.Monolith forms have very high speCific surfaces combined with a very low pressure loss.Crossed corrugated structures are considerablymore favorable for isothermal reaction control.They have a very high radial thermal conductivity that IS almost independent of the specific surface area; the latter can be vaned over a widerange by means of the channel dimensions. Thecatalyst of a reactIOn tube can thus be structuredso that wide packings of small specific surfacearea can be used In the region of the main reaction zone, while packlngs of Increasingly narrower structure. I.e .. large specific surface area. areused downstream. In this way a more uniformreaction rate and temperature profile can beachieved over the tube length (see Section 4.3.3).However. it must be remembered that withcrossed corrugated structures, convective radialheat transport occurs only In one plane perpendicular to the main flow direction. In addition.the flow behavior in tubes of circular cross sectIOn is rather nonuniform over the circumference. which is why It is advantageous to arrangeshort packing sections in series. each section being displaced by 90 . The heat transport parameters In Figure 2 8 were determined for structuresarranged in thiS way.

A general problem in the use of monolithstructures In reaction tubes IS Incomplete wallcontact Since individual tubes of multltubularreactors always have a diameter tolerance of ca.1 mm and Interlocking of the structure With thetube wall must be avoided. the bypass of flow atthe wall is even greater than With random packings. Up to now there have been no specific investigations of the magmtude and efTects of thiSphenomenon.

3. Adiabatic Reaction ControlAdiabatiC fixed-bed reactors are the oldest

fixed-bed reactor configuration. In the Simplestcase they consist of a cylindncal jacket in which

Vol. B4

~ : i X l i : : a J i : i l i f " - (a ta ysto.uench

Figure 3.1. Mam deSign concept> for adiabatiC reactorsA) AdiabatiC packed-bed reactor . B) Disk reactor. C) RadIal-now reactor

the catalyst is loosely packed on a screen and istraversed In the axial direction (Fig. 3.1 A). Toavoid catalyst abrasion by partial flutdizatlOn.catalyst packlngs are always traversed from topto bottom. If fixed beds composed of monolithcatalyst sections are used. the flow direction i,arbitrary.

As discussed In Section 2 4.1. the reqUIrement for a low pressure lo,s leads to a fixed bedof large diameter and low height (Fig. 3.1 B).Such an arrangement (disk concept) is used particularly when very short reSidence times. followed by direct quenching of the reactIOn. arerequired. Examples include ammoma oxidationin nitric acid production (-- Nitric ACid. NitrousAcid. and Nitrogen Oxides. A 17. pp 295-325)and oxidative dehydrogenation on silver catalysts (e.g . synthesis of formaldehyde by dehydrogenation ofmethanoL -- Formaldehyde. A II.

8/3/2019 eig16

13/40

Vol. B4pp.624-628). In the first case the "fixed bed"consists of several layers of platinum wiregauzes, and tn the second case. of a porous silverlayer several centimeters in height. The bed diameters can be up to several meters.On account of the difficulties involved withobtaining uniform flow as weIl as for structuralreasons, the disk concept is limited to smaIl catalyst volumes. The radial flow concept (Fig. 3.1 C)is used where large amounts of catalyst are involved. The catalyst packing is accommodated inthe space between two concentric screen rings orperforated plate rings. and is traversed radially.either from the inside to the outside or from theoutside to the inside. ThIS design is particularlysuitable for large catalyst volumes as well as foroperation at elevated pressure, since at moderatereactor diameters the catalyst volume can bevaried over a WIde range by altering the reactorlength, without affecting the flow-through lengthof the packing.A critical feature of packed radial-flow reactors is the shape of the upper bed closure. Asimple horizontal covenng is not practicablesince a gap through which unreacted gas can passis then formed due to the unavoidable settling ofthe packing. The arrangement shown tn Figure3.2 has proved effective since It produces mixedaxial and radIal flow through the bed in theupper bed closure. The required geometricalshape must be determtned by simulation of thelocal two-dimensional flow through the packtng[3.1].The advantages of monohth catalysts withstraight, parallel channels for adiabatic reactorshave already been referred to in Section 2.4.1.Since monolith catalysts are usually producedwith a rectangular cross section. the fixed bed isconstructed by arranging these individual monoliths tn rows in the form of a large "box". Conventional DENOX reactors for removing NO,from power station flue gases are therefore designed as rectangular chambers (FIg. 3.3). Thecatalyst is often arranged in the form of severallayers in senes. the spaces between the individuallayers permitting cross-mixtng, so that the influence of nonuniform flow as weIl as any possiblelocal blockage of the next layer can be compensated to some extent.

Reference is made tn Section 2.2 to the Im-portance of UnIform flow into and throughadiabatic fixed-bed reactors. ThIS is not easyto achieve. particularly with low-pressure-Iossmonolith reactors. and requires a careful deSIgn

Fixed-Bed Reactors 211Perforatedwall

Unperforatedwall

AXial radialgas flowzone

Radialgas flowzone

Figure 3.2. Upper bed clo,ure In a radlal-flO\, redctor [3 IJ

Structurec8Catalystelement

Catalyst module

Flue gas + NH J In

t

Flue gas out

Catalystlayers

Figure 3.3. Reaction chamber for removdl of mtrogen oxIdes from power station flue gas [3 2J

8/3/2019 eig16

14/40

212 Industrial Reactorsof the inflow hood. On account of the low pressure loss, unfavorable flow conditions in the outflow hood may also affect the flow behaviorthrough the catalyst bed.Figure 3.4 shows the velocity distribution mfront of the monolith inlet for an industriallyhoused automobile catalyst [2.11]. Since the flowcannot follow the sudden widening of the inletfunnel, one third of the total cross section is traversed at a velocity that is roughly three times themean velocity. It can be estimated that, with uniform flow through the catalyst, half the catalystvolume would be sufficient for the same meanconversion. Also, bends in the feedpipe can leadto swirl-type components and thus contribute tononuniform flow.

Purely adiabatic fixed-bed reactors are usedmainly for reactions with a small heat of reaction. Such reactions are primarily involved in gaspurification, in which small amounts of interfermg components are converted to noninterferingcompounds. The chambers used to remove NO,from power station flue gases, with a catalyst

,.::,

uo>

15

4 6

Mass flow 204 kg/ho 0 70 kg/h.. .. 35 kg/h

8 10Diameter, em -

Figure 3,4, Velocity dl'tnbutlon In an Industnally houseddutomoblle catalyst [2 11)

Vol. 84volume of more than 1000 m3, are the largestadiabatic reactors, and the exhaust catalysts forinternal combustion engines, with a catalyst volume of ca. 1 L, the smallest. Typical chemicalapplications include the methanation ofCO andCO 2 residues in NH3 synthesis gas, as well as thehydrogenation of small amounts of unsaturatedcompounds in hydrocarbon streams. The lattercase requires accurate monitoring and regulationwhen hydrogen is in excess, in order to preventcomplete methanation due to an uncontrolledrise in temperature, a so-called runaway (seeChap. 6).

4. Reaction Control with Supply orRemoval of Heat in the Reactor4.1. Introduction

In the majority of fixed-bed reactors for industrial synthesis reactions, direct or indIrectsupply or removal of heat in the catalyst bed isutilized to adapt the temperature profile over theflow path as far as possible to the requirementsof an optimal reaction pathway. Here a cleardevelopmental trend can be observed, which isillustrated schematically in Figure 4.1.Development started with the adiabatic reactor (Fig, 4.1 A), which on account of the adiabatic temperature change could only be operated togive a limited conversion. Higher conversionswere achieved at the same mean temperature lev-el when several adiabatic stages were introduced,with intermediate heating or cooling after eachstage. The sImplest form involves injecting hot orcold gas between the stages (Fig. 4.1 8). For aconstant tube diameter. the main disadvantagesof thIS temperature control strategy are crosssectional loading, which increases from stage tostage, and the mixing of hot and cold streams.which IS energetically unfavorable. The composition is changed by injection. which can have apositive or negative effect on the desired reaction.The next development was the replacementof injection cooling by interstage heat exchangers, through which the required or released heatof reactIOn is supplied or removed (FIg. 4.1 C).

The development of reactors in which theheat-exchange surfaces are integrated in thefixed bed to supply or remove the heat of reaction as close as possible to the reactIOn sIteoccurred in parallel WIth the development ofmultistage adiabatic reactors WIth mtermediate

8/3/2019 eig16

15/40

Vol. B4

Feed gasl

-lIIHealcarrierloopII__ J

Feed gas

CDCoolanlloop I

Coolanlloop II

Figure 4.1. Development of fixed-bed reactors

Fixed-Bed Reactors

Inlerslagehealexchangers

213

A) Smgle-bed adiabatic packed-bed reactor; B) Adiabatic reactor with mterstage ga, feed (ICI concept). C) Multlbedadiabatic fixed-bed reactor wllh mterstage heat exchange. D) Multltubular fixed-bed reactor. E) Mulutuhular fixed-bedreactor with two coohng ClrcUlts and nomsothermdl coohng temperdture (the temperature prolilc ,ho ... n I' for d s t r o n g l ~ exothermic reaction)

heating or cooling. The multi tubular fixed-bedreactor (Fig. 4.1 D) constitutes the oldest andstill predominant representative of this class offixed-bed reactors. Here the catalyst packing islocated m the individual tubes of the tube bundle. The heat-transfer medIUm is circulatedaround the tube bundle and through an externalheat exchanger. in which the heat of reaction ISsupplied or removed. Whereas with endothermicreactions. circulatmg gas can be used as heattransfer medium. for strongly exothermiC reactions exclusively liquid or bOiling heat-transfermedia are used. Only In this way can the catalysttemperature (e.g. m the case of partial oxidations) be held m the narrow temperature rangenecessary for selective reaction control.

Initially. the integration of heat exchange mthe fixed bed was utilized to ensure as Isothermala reaction control as possible. which IS why reactors of the type shown m Figure 4 1 D are alsocommonly termed "isothermal reactors". Theyare characterized by reaction tubes of 20 80 mminternal diameter and a carefully deSIgnedflow control of the liqUId heat-transfer medIUm.With largely constant heat-transfer c o n d l t i o n ~ throughout the tube bundle and maxImum temperature changes of the h e a t - t r a n ~ f e r medIum mthe tube bundle of a few degreesThe latest concepb are aimed at establlshmga freely selectable (withm hmib) optimum temperature profile over the tube length. ThIS re-qUires complex heat-transfer medIUm control

8/3/2019 eig16

16/40

214 Industrial Reactorswith several sections and temperature levels (Fig.4.1 E).

The stimulus for the developments outlinedin Figure 4.1 was the need for total raw matenalutilization as regards both mass and energy. ThISinvolves as main criteria the yield of primary endproducts (maxImum), the yield of byproductsthat must be removed and eliminated (minimum), the thermal energy consumptIOn or recovery, and the mechanical energy requirements(gas compression, cIrculation of heat-transfermedium). In addItion to the running costs, whichare determined by the above criteria, the investment costs are decIsive in an Investment decision,and naturally rise sharply with increasing complexity of the reaction cycle.On account of cost degression the projectedplant size and the subsequent degree of utilizatIOn as well as the technological sophistIcationare decIsive in calculating the product price. Several of the fixed-bed reactor vanants illustratedIn Figure 4.1 can. depending on the location ofthe productIon site and the estimated output, beused for the same process. For example, multistage adiabatic reaction systems with intermediate superheated steam feed, multi tubular reactors with circulating gas heating, and strictlyisothermal designs with multi tubular reactorsheated WIth molten salt are currently used for theendothermic synthesis of styrene from ethyl benzene. Also, exothermIC, equilibrium-limited reactions such as methanol synthesis are carried outin multistage adiabatic reactors with interstagecooling as well as in multi tubular reactors. Overall, however, there is a trend towards the morehighly integrated deSIgns, which will accelerateWIth nSlng energy and raw matenal pnces. AdIabatIC multistage designs and reactors WIth heatexchange integrated in the fixed bed are dIS-cussed in the following sections. Autothermal reactIOn control, In whIch the heat of reaction ofmoderately exothermIC reactions is utilized toheat the Incoming feedstock. IS dIscussed separately In Chapter 5.

4.2. Adiabatic Multistage Reactors withInterstage Heat TransferAdtabatlc multistage fixed-bed reactors WIthintermedIate cooling or heating are nowaday,

used partIcularly where the reaction proceeds selectively to gIve a single product but IS limited by

Vol. 84the equilibrium conditions. Intermediate coolingor heating is used to displace the gas temperaturein the directIOn of higher equilibrium conversion. Typical examples include the synthesis ofammonia, sulfur trioxide, and methanol. In theseexothermic reactions the equilibrium conversionto the target product decreases with increasingtemperature. as shown in Figure 4.2 A. For agiven conversion x a temperature can thereforebe found at which the reaction rate, with respectto the target product, becomes a maximum. Thistemperature must be below the equilibrium temperature but not so low that the reaction becomes too slow for kinetic reasons. The pointsobtained in this way can be joined to form amaxImum reaction rate curve (Fig. 4.2 B).

Since, in the case of adiabatic reaction control the temperature increases linearly with theachieved conversion according to the equation,

(4.1)

each adIabatic reaction pathway of an exothermic reaction lies on a straight line of gradient~ T / ~ x (see FIg. 4.2A).A practicable reaction pathway for a multIstage adiabatic reaction can thus be derived fromFigure 4.2 by joining straight-line sections forthe adiabatIC reaction to vertIcal lines for thetemperature reduction due to indirect intermediate cooling (Fig. 4.2C).

The kinetically optimum reactIOn pathwaywith the smallest required catalyst volume resultswhen the trajectory follows, In a large number ofsmall steps. the line of maxImum reactIon rate. Inpractice. the apparatus and equipment expenditure Involved in a large number of stages must beweIghed up against the savings in catalyst. ConventIOnal multIstage reactors for thIS class of reactIOn therefore have three to five stages. FIgure4.3 shows the layout of an ammonia synthesisreactor designed on this baSIS. For structural reasons the heat exchanger is Incorporated betweenthe Inflow and outflow in the lowest part of thepressure casing. The reactIOn gas then flows upward in the annular gap between the pressurecasing and the fixed beds, whereby It is furtherheated and at the same time protects the pressure-beanng structural component, against excessively hIgh fixed-bed temperatures. The threeadtabatlc fixed-beds are traversed from top to

8/3/2019 eig16

17/40

Vol. B4 Fixed-Bed Reactors 215

t I .r::.abatic..... V ~ : ~ ~ t l o n Feed pathtemperature

\ ,,,/" ......... Locus of .... ,optimal ....

reaction rate . ! IConversion. x_ x_ x_

Figure 4.2. EqUlhbnum hmltatlOn of exothermic reactionsA) Equlhbrium conversIOn as a functIOn of temperature. B) Optimum reactIOn rate curve. C) Improvement of conversIOnby mterstage coohng

L . . f ' ' ' - T - - - - - , - -

Condensationcoolers

Stage

Fresh gas N2+3H 2 (+CH 41Clrcu(ating gas

Off-gas

Steam

l l-l"'fl 'I--- WaterSteam

~ 1 - - ~ 4 - - - - W a t e r

FeedEXit

Figure 4.3. Schematic of a multistage reactor for ammOnIa ,ynthesls

bottom. part of the heat of reaction bemg utilized to generate steam In the two mtermediateheat exchangers. To start up the cold reactor. hotgas must be added to the uppermost bed. forexample through an external start-up preheater.

Industnal adiabatic multistage reactors oftendiffer in many details from Figure 4.3. althoughthey are of a comparable baSIC deSign. For example, radially instead of aXIally traversed beds canachieve a smaller pressure loss with a more favorable structural arrangement; heat exchangewith the cold feedstock can be effected by heatexchange surfaces mtegrated in the first catalystbed; or a cold gas quench can be used to achieveadditional temperature regulation.

Modern. adiabatic multistage reactors maythus become so complex that the question anseswhether a multI tubular deSIgn accordmg toFigure 4.1 D or E does not represent the morefavorable alternative. The required heat-exchange surface area in the case where heat ex-change is integrated in the fixed bed IS ~ m a l l e r than In the case of free gas flow on account of thepositive effect of the catalyst packing. Furthermore, It does not involve any additional p r e ~ ~ u r e loss. and the optimum reacllon rate curve(Fig. 4.2 B) can be better approximated by controlhng the coohng temperature profile (see Section 4.3.2.) than by a stepwise temperature reduction

8/3/2019 eig16

18/40

216 Industrial ReactorsOn the other hand, a multistage arrangementmay be considered for structural, operating technology, or kinetic reasons in the following cases:

1) If, in the case of large single-train plants. subdivision Into several individual items of apparatus is necessary for reasons of transportor construction2) If a catalyst must be replaced in individualstages at different times on account of different catalyst compositions and/or aging3) If a gradual addition of a reactant has kineticadvantages compared to the total additionto the feed (here a suitably designed intermediate heat exchanger ensures a uniform distribution and mixing with the reaction gasstream)

4) If the intermediate stages are used to extracta limiting product In the case of equilibriumlimited reactions; an example is the intermediate absorption of SO) before the last stageof the SO) synthesis5) With reaction temperatures above 300 'C intermediate coohng can be performed directlywith boiling water. whereas In a fixed bed ahigh-temperature heat-transfer medium mustbe used as coolant (see Section 4.3.1)

4.3. Reactors with Heat Exchangeintegrated in the Fixed BedThe aim of reaction control with heat ex

change integrated in the fixed bed using acirculating heat-transfer medium is to maintainthe catalyst temperature in a narrow optimumrange under all operating conditions. Withstrongly exothermiC successive reactIOns. such aspartial oxidation, and partial hydrogenations.on account of the danger of a runaway reactIOn(see Section 6.2) this requirement can be metonly If1) The temperature of the heat-transfer medium

IS close to the desired catalyst temperature2) Large heat-exchange surfaces arc availableper unit catalyst volume3) A sufficiently large mass flow velOCity of thereaction gases ensures good heat transport

from the packing to the heat-exchange surfaceWith exothermic equihbnum reactions andendothermiC reactions these requirements are

less stringent since these reactions cannot run-

Vol. 84away, although here too it is beginning to berecognized that a tight and uniform temperaturecontrol over the reactor cross section is advantageous.

4.3.1. Heat Transfer Media for Fixed-BedReactorsThe first of the above requirements presupposes an assortment of heat-transfer media thatcovers the whole temperature range of interestfor gas-phase reactions in fixed-bed reactors. It

IS convenient to distinguish between gaseous.liquid. an d vaporizing heat-transfer media.Gaseous heat-transfer media in the form of hotflue gases are used in the temperature rangeabove 500'C exclUSively to supply heat for endothermic reactions.Conversely. vaporizing heat-transfer mediaare used exclusively to remove heat fromexothermiC reactions. Whereas formerlypetroleum fractions such as kerosene (e.g., Inethylene oxide synthesis) were more widely used.they have now been largely replaced by boihngwater on account of their flammability. lowerheat of vaporization. and the need to producesteam in a downstream condenser/heat exchanger. Depending on the saturated vapor pressure. the temperature range from 100 to 310 C(100 bar) can be covered with boiling water. InthiS range it is the preferred heat-transfer medium for evaporative cooling if an isothermal cooling temperature is reqUired.Locally vanable cooling temperature profilescan be estabhshed most easily with hquid heattransfer media that do not vaporize in the intended operating range. In order to avoid cavitation.pressunzed water should be used only up to ca.220 C; heat-transfer oils cover the temperaturerange up to 300 C. while above thiS temperaturesalt melts are now used exclUSively in reactiontechnology [4.1). Compared to heat-transfer oilsthey have the advantage that they are incombustible and stable. although they have the disadvantage that they solidify at about 200 C (nitrate melts) or 400 C (carbonate melts). Thetemperature ranges of pOSSible heat-transfer media are compared in Figure 4.4. In addit ion to thethermal stabihty. the energy IV per unit amountof transported heat Q required to Circulate aheat-transfer medium IS an Important criterionof choice

8/3/2019 eig16

19/40

Vol. B41000 III800 IIII600 III:;-' 400 I>-- I ..00 0 ..I 3200 !D

Boilingheat transfer

o '"c _.. ....... E0[i

Liquid convectionheat transfer

c: '"o . .1: E

Fixed-Bed Reactors 217

0 '"..'"..... '".3 u:

Gas convectionheat transfer

Figure 4.4. Application ranges for common heat-transfer media

The following equation has been derived forthe evaluation of heat-transfer media withoutphase change [4.11]:Q!!,:rc ( l 7 3 . cpN 0 3 6 - ~ (4.2)

where Q s the heat flow transferred by the heattransfer medium while Its temperature increasesby exactly I1Tc ' N IS the required pump power,and Q, cp ' and '1 are the density, specific heatcapacity, and viscosity of the heat-transfer medium. For liquid heat-transfer media, the equationIS derived from the pressure drop in a turbulentlytraversed, hydraulically smooth pipe. It is practical to solve Equation (4.2) for the pump power'. [Q J2 75N - -11T., (4.3)

Thus the reqUIred power vanes as almost thethird power of Q/11Tc' The component '10 25 ;75. Q2, which depends only on the physicalproperties of the heat-transfer medIUm, is plotted as a functIOn of temperature in Figure 4.5.ThiS shows the exceptional heat-transfer properties of water that result from ItS density and, Inparticular, its high speCific heat. as well as theproperties of gases (air) and vapors (water vapor), which are about seven orders of magnitudeworse. Common heat-transfer oils and salt melts(sodIUm nitrate) lie close together, but about anorder of magnitude above water. The poor performance of molten sodium compared to heattransfer oils and salt melts is noteworthy

To test Equation 4.3 on a more realistic example of the coohng of a multi tubular reactor, aquadratic tube bundle (cross section 1.5 x 1.5 m)with fourfold passage of the coolant was designed by using detailed equations for pressuredrop and heat transfer [4.12] for vanous heattransfer media. The heat release of 1650 kW in1591 (37 x 43) tubes of 30 mm external diameterand 4 m length corresponds to the conditIOns ina partial oxidation reaction. A temperature risebetween feed and outlet of 11 Tc = 5 K was assumed for the heat-transfer medIUm (additionally 150 K for air and water vapor). Table 2 lists asresults the required pump power N, the pressuredrop in the cooling cirCUIt I1p (whereby p = 1 barexit pressure was assumed for water vapor andair), and the excess temperature of the reactorwall above the heat-transfer medIUm temperature I1Tw assuming constant wall temperaturealong the length of the tube. It was shown thatthe results of these more precise deSign e q u a t l o n ~ for the hquid heat-transfer media can be correlated with the equatIOn

/Ii =7800 - ---[ Q 2 73 '1 P11T., c; 73 . (/ (44)

This represents a good confirmation of Equation(4.3).Water is thus the ideal heat-transfer mediumwithin Its temperature range. For higher temperatures molten salts have increasingly replaced

the previously more commonly used heat-transfer Oils. Salt melts cover a larger temperaturerange and have the particular advantage overOils that they are incombustible. The potential

8/3/2019 eig16

20/40

218 Industrial Reactors Vol. B4

o Diphyl THT* BP Transcal N... Waterx Air Water vapor Salt Sodium

......... .. :00.0 .. 400.0T.O(_ 600.0 800.0

Figure 4.5. The matenal-specllic factor F = "Ie; " . 1' for vano us heat-transfer medIa as a functIon of temperatureTable 2_ Results of the example calculatIonHeat-transfermedIum

Water

Producer

Transcal N BPDlphyl TH T BayerSodIUmHT saltWater vaporWater vaporAIrAIr

Pump power N.kW0.060062908624.54131o119x 10"0254x 10 30229 x 10"0456x 10 3

danger of a relatively large amount of hot saltmelt obviously exists. but tS reliably dealt withby experienced reactor construction companies.Special ni trate melts (HITEC) can be used m thetemperature range 200 - 500 C. Gradual decompOSitIOn begms above thiS temperature. whichcan accelerate violently above 600 'C. Access oforgamc components to the melt (mtrate decompOSitIOn) and of water (steam explOSIOn) must beexcluded [4.1). New salt melts. for example.based on carbonates. are being developed for thetemperature range 400 -- 800 C. In this case It isnot so much the thermal stability of the moltensalt but rather the corrosion of the reactor matertals that presents problems.Gases are the only heat transfer media usableover the entIre temperature range. but because oftheIr low density they have an unfavorable heat

Pressure drop Mean tempera- PermItted coohngt.p. bar ture dIfference temperature m-t.T . K crease t.T" K

0.008 07 5 50042 2.32 50054 2.35 50150 0.04 50105 1.45 58759 1.30 50175 1422 15018001 1.08 50341 12.29 150

transport behavior (Fig. 4.5). They are thereforeused exclUSIvely as flue gases to supply heat athIgh temperatures. However. large temperaturedIfferences between the heat-transfer mediumand the reactor wall. with possible adverse effectson the umformity of the heat supply. must thenbe accepted.

4.3.2. Reactor DesignsAs with adiabat ic reactors. a prinCIpal task ofreactor development is to produce uniform reac

tIOn condItIOns over the whole reactor and maintain such conditions durtng the entire operatingtIme. ThIS mvolves the conditIOns m the catalystpacking (residence time. catalyst concentratIOnand actIvity. heat transport) and in the heat-

8/3/2019 eig16

21/40

Vol. B4transfer medium circuit (throughput, temperature, heat transfer). When liqUId or vaporizing heat-transfer media are used, the heat-transfer coefficients are usually one order of magnitude greater than those on the catalyst side.which facilitates the task as regards the heattransfer medium. On the other hand, withstrongly exothermic, selectivity-sensitive reactions (especially partial oxidations), a temperature constancy of the heat-transfer medium ofca. 1DC is often required. This leads to a highenergy requirement for circulation (according toEq. 4.4) and necessitates extremely careful design and control of the heat-transfer medium circuit.Tubular reactors of the type shown in Figure4.1 D have been in use longest and have beenfurthest developed. In the case of a vaporizmgheat-transfer medium an arrangement as shownin Figure 4.6 is generally chosen, in which theliquid heat-transfer medium surrounds the stationary tube bundle. The rising vapor bubblesescape through the ascending pipe into a vapordrum, where they are separated from the liquid.A circulation flow is estabhshed due to the difference in density in the downpipe (pure liquid) andin the reactor jacket, which means that circulatmg pumps are generally not required. The heattransfer medium temperature is regulated VI a thesaturated vapor valve. Specific details of this regulation are discussed in Section 6.6.

If uniform supply and removal of heat-transfer medIUm via annular channels is ensured. thiSarrangement provides reliable isothermal reaction control. Since a steam cushIOn can formunder the upper tube floor, the active catalystlayer should begin only at a deeper level.Apart from the type of multltubular reactorshown in Figure 4.6, other multi tubular reactorsare sometimes used in which the catalyst bed isarranged around the tubes and the heat-transfermt!dium flows through the tubes (Fig. 4.78). Aninterestmg new development has been introduced by Lmde (Fig. 4.7 A): the tube bundle IScomposed of counterwound spirals in which upwardly flowmg water IS evaporated. The tubesrun into a vertical vapor drum located at thereactor head. The tube bundle is connected to acentral downpipe at the bottom so that. as m thearrangement m Figure 4.6, a natural circulationof the evaporating water is established. Advantages in construction and in the heat transferfrom the reaction gas to the tubes of the bundleare claimed for thiS deSign [4.13].

Feed gas;7!.,

I

I

t II

Fixed-Bed Reactors

Saturatedsteam

219

Figure 4.6. Multltubular reactor With bolhng water cool109

CirculatIOn systems with parallel and crossedcocurrent or countercurrent flow of the heattransfer medium (Fig. 4.8) are commonly employed for hquld heat-transfer media. The mainpart of the heat-transfer medIUm is generally circulated with a high-capacIty pump in order toachieve uniform heat-exchange conditions. whilea partial stream IS passed through a heat exchanger for supplying or removing the heat ofreaction. The deSired heat-transfer medIUm temperature is attamed by regulating this partIalstream. WIth exothermic reactions the heat exchanger IS normally a steam generator whichproduces saturated steam at a pressure corresponding to a boihng point of 30-80C belowthe maximum coolmg temperature.A superheater fed with a further partialstream of the heat-transfer medium can if necessary be connected downstream of the steam generator. The arrangement with separate externalUnIts. shown schematically m Figure 4.8 hasstructural and maintenance advantages over acommon arrangement In the tube-free interIor ofthe reactor [4.14].Apparatus constructIOn companies speCializing m these reactors have developed a detailedand comprehenSive know-how as regards flowcontrol of the heat-transfer medIUm [4 15]. ThiSconcerns the UnIform supply and removal of theheat-transfer medIUm. which generally takesplace via external annular channels. as well as the

8/3/2019 eig16

22/40

220 Industrial Reactors

Steam

Recycle

Feedwaler

Figure 4.7. A) Design concept of the Linde isothennal re-actor for methanol synthesis ; B) Cut through the tubebundle surrounded by catalyst pellets (from [4 .16])

flow control within the reactor. Here a distinction is made between para llel flow control andcros ed flow control. Parallel flow control(Fig. 4.8 B) is achieved by two rectifier plateswith narrow bore . On account of the pre surelos through the bores suitably arranged over the

Vol. 8 4reactor radiu , a uniform flow profile is produced over the cross section with uniform heatremoval conditions between the di tributorplates. The advantage of this arrangement i thatthe whole reactor can be equipped with tubes.

However, due to the nonuniform flow conditions in the inflow and outflow region of theheat-transfer medium, only the region betweenthe distributor plates should be filled with activecatalyst. Heat transfer is low because of the parallel flow of the heat transfer medium, and thepre sure loss in the distributor plates does notco ntribute to improving the heat transport. Forthis reason this arrangement is preferably usedfor reac tions with a moderate heat of reac tion .In reactions with a large heat of reaction ,e pecially partial oxidations, tran verse flowcontrol is more widely used , generally in a radially symmetrical arrangement with baffie plates, asillustrated in Figure 4.8 A. Since the heat removal conditions are poorly defined in the region of the flow deflection, with the change froma tran ver e stream to a parallel stream and backto a transverse stream, this region should be freeof tubes. A difficulty arises from the constantchange of the flow cross section in the radialdirection. However, this can be alleviated, forexample, by providing pressure-relief bores inthe baffie plates, increasing in number towardsthe reactor axis. In this way constant heat-transfer conditions shou ld be achieved over the reactor radius. In ummary, ensuring uniform heattransfer conditions as regards the heat-transfermedium thus requires a considerable flow technology know-how. Some recent publications il-lustrate the major differences in the behavior ofdifferent tube sections that ca n arise due to aninadequate design and layout of the heat-trans-fer medium circuit [4 .2)- [4.4) .

4.3.3. Influencing the Course of the ReactionThe cour e of the reaction (i.e., the conversion and se lectivity or yield) can be deci ively

influenced by the arrangements made for controlling the heat-transfer medium. The most obvious, although technically most complex solution, i to a rrange different hea t-transfer mediumcircuits so as to achieve a stepwise ap proximation of an optimum temperature profile. Thepurposeful utiliza tion of the temperat ure changeof the heat-transfer medium flowing through thereactor is, however, technically simpler. and is

8/3/2019 eig16

23/40

Vol. B4

.J

J I Circulationturbine

Fixed-Bed Reactors

Steamgenerator

Steam

Feedwater

221

Figure 4.8. Heat-transfer medIum control In tube-bundle fixed-bed reactorsA) Cross flo\\. B) Parallel flo\\

discussed here in connection with cocurrent orcountercurrent cooling of a fixed-bed reactor involving an exothermIc reaction.

Figure 4.9 shows temperature profiles forthree different ways of controllIng the coolingstream in a partial oxidation reaction. If thecoolant is circulated so fast that its temperaturein the reactor scarcely alters. then its flow direction is irrelevant and a temperature profile witha pronounced temperature maximum becomesestablished; thIS is typical of strongly exothermICreactions (Fig. 4.9 A). If the coolant IS CIrculatedin cocurrent and its velocIty is chosen so that itbecomes noticeably hotter over its path. an almost isothermal temperature behavIor can beachieved (Fig. 4.9 B). ThIS is because the reactivegas at the inlet IS in contact WIth the coldestcoolant and the cooling temperature rises in stepwith the consumption of the reactants. so thatthe reaction rate remains VIrtually constant overa fairly long section [4.5]- [4.7]. This stabilizingeffect of cocurrent cooling has hardly been exploited up to now in industrIal reactors. Thismay be due to the fear that. at the reqUIred lowflow velocity (in the example of Fig. 4.9 B.v, = 0.01 m/s). heat transfer wIll be madequateand natural convectIOn wIll occur in the coolingjacket. However. r, describes the mean coolantvelocity parallel to the tube aXIs. With a crosscocurrent flow of the coolant. the actual flowvelocIty may m fact be substantIally larger. depending on the number of deflectIOns. with theresult that the aforementioned problems do notarIse.

420 - - Reactor temperature--- Cooling temperature- - - Reference temperature's= -0 14m/s380

340 ------------------300

's=+O 01m/s

- - - - - - -

420380 's=-O Olm/s

30006 10 14 18 22 26 3 0

Z I n m ~ Figure 4.9. Influence of the wolanl flo" dIrectIon andflo" \ e l o c l l ~ I. on the reactIon temperature profileA) hothermal. BI ("ocurrcntll

8/3/2019 eig16

24/40

222 Industrial ReactorsCompared to cocurrent flow. countercurrentflow has a markedly destabilizing effect at lowflow velocities (Fig. 4.9 C). Since the incomingreaction mixture in this case is in contact with thewarm coolant outflow. the maximum temperature rises to much higher values. Countercurrent

cooling can even lead to the occurrence of severalstationary states. and in general favors the runaway of a strongly exothermic. irreversible reaction (see SectIon 6.3). In contrast to heat exchange without a reaction. countercurrentcontrol of the heat-transfer medium in reactorsinvolving exothermic reactions should thereforebe chosen only in particular cases.The temperature control of an exothermicequilibTlum reaction can constitute such a case.As Illustrated in Figure 4.2 B. the optimum temperature profile should in this case decrease withIncreasing conversion. i.e . along the tube length.On account of the equilibrium inhibition of thereaction. It is not possible for the reaction torunaway in the front region. Countercurrentflow of the heat-transfer medium is also advantageous for endothermic equilibrium reactions.Figure 4. to shows the calculated temperatureand concentration profiles with different heatingconditions in the synthesis of styrene (dehydrogenatIOn of ethylbenzene). WIth adiabatic andisothermal reaction control. styrene formationdecreases with increasing tube length. whereas Itremains roughly constant with countercurrentcontrol.A significant advantage of nonisothermalcontrol of the heat-transfer medium in cocurrentor countercurrent flow is the saving in circulatIOn energy. since much smaller heat transfermedium streams must be circulated. Overall. thecombination of several heat-transfer mediumcircuits (Fig. 4.1 E) and the purposeful utilization of the temperature change of the heat-transfer medium in the reactor offer a wide range ofpOSSIbilities of establishing optimum temperature profiles for a given reactIon. and of counteracting any changes in actiVIty that occur duringthe operating life of the catalyst by altering thetemperature profiles.A further possibIlity of influencing the courseof the reactIOn is to use catalysts of dIfferentactivities over the reactor length. ParticularlyWIth strongly exothermIC reactions that are liableto runaway. such as partml oxidations. a lessactive catalyst IS occasionally used in the frontpart of the reactor In order to avoid too high amaximum temperature. Figure 4. t t A illustrates

Vol. 84

50 -

- - - - ~ - 1

700

660

620 fu

40

f 30"i 20 __ ----t---i 80 ::..

10 540O ~ ~ - L - - - - ~ - - ~ __o 2 3 4

50 --li-oo----Cata'yst'----oo-li 700

II I, I" EB I 660

" I40

" I f3 0 r - + - - - - - ' ~ , ~ ~ - - ~ T - - - - + - ~ 6 2 0 .....__ _ I_ _ I

10

2 3CD 50

4030

"i 2010

3

~ IIIIII4

4

uo580 0 - '540

700

660

620 Iu

580 : :

540

Figure 4.10. Innuence of the heattng agent now dlrecttonon the temperature profile In styrene syntheSISA) AdIabatIC. B) Isothermal. C) CountercurrentEB = ethylbenLene. St = styrenethe use of two catalysts of dIffering actIvity Inseries. The resulting temperature profiles have atYPIcal double-hump shape [4.8]. [4.9]. This can

8/3/2019 eig16

25/40

Vol. B4

t 50720

""~ 690::J 660OJc .EOJ

11.00.8> ..:; 0.6:;::.;( 0.4

T,= T(=642K

75 150 225 300 375ZL' cm_

100 r,I \I 0 I \I \60

U0 40

200

t 100.8o

~ - ~

Fixed-Bed Reactors

112080u

0 400

223

Reactor length_ Reactor length_ Reactor length_Figure 4.11. Influence of actIVIty profiles on the temperature profile of a strongly exothermic redctlOnA) Catalyst with 66 % actIVIty 10 the front section of the tube and 100% actIVIty 10 the rear section of the tube 14 RI. B) Lmed