KINETICS

OF COKE DEPOSITION IN THE CRACKING OF PROPANE

K M SUNDARAMandGFFROMENT

THERMAL

Laboratonum

voor Petrochemlsche

Techmek,

hjksumversiteit.

Gent, Belgium

(Recetued 15 July 1978, accepted 18 August 1978)

Abstract-A novel expertmental set up for determmmg the kinetics of coke deposltlon m thermal cracking IS described and Its apphcation to the crackmg of propane IS dlustrated The reactor Itself IS of the completely muted type and provides kmetic mnformation on propane crackmg whch IS m agreement with that obtamed m tubular reactors A sunple reaction scheme aad the correspondmg kmetic equation for the cokmg are fitted to the data, which were determmed by means of an electro-balance techmque The kmetic equations are mserted mto a mathematical model of an mdustnal crackmg cod and the operation of ths reactor IS sunulated over a penod of 700 hr The results are m agreement v&h the scarce mformation avdable

1 INTRODUCFION

crackmg of hydrocarbons IS always accompamed by coke formatlon The coke deposits on the wall and thus hnuts the heat transfer to the reactor, mcreases the pressure drop and even plugs the reactor m extreme cases To bum off the coke the reactor has to be shut down penodlcally The run length of the reactor usually vmes from 10 to 80 days, dependmg upon the feed stock, type of reactor and seventy of operation So far no quantitative treatment of the kmetics of coke deposition m the thermal crackmg of hydrocarbons IS aviulable Gbv~ously, such mformation 1s of extreme Importance for operation, design and optlrmzation of thermal crackmg furnaces
Thermal 2 LlTERATURERFxrEw

Coke may be formed e&her directly from the feed stock or from the products or from both Many components of the feed or the products are capable of yielding coke Hirt and Palmer111 denved a value of 103 kcal/mole for the homogeneous decomposltlon of methane mto coke m a flow reactor between 890 and 1100C Oxley et al [2] pyrolyzed methane and acetylene separately m a fluldlzed bed reactor m the temperature range 85CL1025C and 875-1200C respectively Based on heterogeneous rate expresslons they obtamed activation energies for cokmg from methane and acetylene of respectively 90 4 and 51 0 k&/mole Coke formation 111 the thermal crackmg of acetylene has been studled extensively by Johnson and Anderson[3] They camed out their experunents 111a flow reactor with hehum as dlluent at temperatures rangmg from 500 to 1000C and observed both coke and polymer as products From electron nucroscopy of the coke they concluded that carbon was not formed from the polymer, but that both carbon and polymer were formed by two competmg reactions m the gas phase Gay et al 141 proposed the followmg radrcal mechamsm for the coke formation m shock tube pyrolysis of acetylene
C2H2 C&S C&2 -Cd-I2 -

CsH, -

Kinney and Del Bel[S] reported the pyrolysis of aromatic substances m a flow reactor anti r,eastilea the amount of coke by welghmg the reactor before and after the reacaon They proposed a stepHnse condensation of the aromatic rmgs with the elmunation of hydrogen as a mechamsm of coke formation Wrk et al [6] have reported that coke ISformed from aromatic substances, especially from benzene They concluded that two molecules of benzene condense together to form dlphenyl and two molecules of this component react to gve quadraphenyl and so on until eventually coke ISobtamed Rec&tly Shah et al [7j have reported on the coke formation m the thermal crackmg of n-octane m a tubular reactor between 750 and 800C and space mes up to 1 set The cokmg rate was obtamed by weighing the reactor before and after the reaction They found that the coking rate was high m the mutual penod and observed that the coke formation mamly depends upon the surface to volume ratio of the reactor Froment and BIschoff [8] have shown that, even under Isothermal operation m a tubular reactor, coke IS not deposited unrformly, but accordmg to an ascendmg or descendmg profile dependmg upon the mechamsm of cokmg To develop the reaction scheme for coke formabon, mformation on the coke profIle IS essential Therefore, to detemune the kmehcs, average concentration of coke obtamed by weighmg the reactor IS inadequate Until now only crude methods of measurement, hke welghmg the reactor[5,7] have been used With the avadable analytical techmques the measurement of the coke profile m a tubular reactor IS not possible, unless the reactor were cut mto sections The use of a completely mixed reactor allows to work at pomt con&Qons of partml pressure or converslon To measure the rate of cokmg itself, recourse was taken 111 this work to a techmque apphed by De Pauw and FromentI91 and Dumez and Froment [ IO]to cokmg IIIcatalytic processes, namely welghmg the coke by means of an electrobalance The completely nuxed reactor used m the present work IS too heavy to suspend on the arm of the balance, however Therefore, a small cyhnder was m635

636

SuNDw

and G

FROMENT

serted mto the reactor The cyhnder was of the same materml as the reactor and welghed about 600 mg so that It could be suspended to the arm of the balance and thus allow the coke deposmon to be followed contmuously Knowmg the ratio of surfaces of the cylmder and the wall, the total carbon could then be calculated The reaction chosen to develop and test the procedure was the thermal crackmg of propane The kmebcs of propane crackmg were studled by many workers[l l-151 but no quantltatwe treatment of coke deposltlon m propane crackmg has been reported The dlluent used m this work was mtrogen and not steam, which would partly remove the coke Nitrogen IS completely Inert at the temperatures used and does not alter the results, as follows from a compmson of the results of Buekens and Froment [ 131(Fluent mtrogen) and Van Damme et al [ 111 (dlluent steam) Finally, the coke profile in an industrial reactor 1s simulated
3 E-AL The EQurpMENT AND PROCEDURE

I732
Fig 2 D&ads

of the preheater-reactor assembly 1 Preheater, 2 cybnder, 3 reactor, 4 furnace, 5 msulatlon

equipment had to fulfill a dual purpose to determme the lanetics of the mam reaction and to determme the kmetics of cokmg A schematic representation of the equipment 1s shown m Fig 1 The flow rate of propane IS measured by the cahbrated capillary flow meter A needle valve IS used to mamtam a constant pressure at the outlet of the capillary A snmlar system IS used to meter the flow of mtrogen Propane and mtrogen are preheated separately 111 the preheaters Hl and H2 respectively and then mixed and heated to 400C m the heater H3 The mixed gas IS led mto the preheater, which IS lust above the reactor The preheater and the reactor are deslgned m such a way that perfect mlxmg can be achieved m the reactor The detsuls of the preheater-reactor assembly IS shown 111Fig 2 The wall separating the reactor and the preheater contams 24 small holes dnlled m an angle of 15from either side of the vertJcaI These holes act as nozzles and thus nuxmg JS achieved m the reactor A small cyhnder IS suspended from one arm of the balance and 1s posltloned m the centre of the reactor The reactor IS heated by a manually controlled electnc oven The temperatures m the

preheater and m the reactor are measured by means of thermocouples The product gas passes through the same hole as the wue used to suspend the cyhnder Just above the preheater an outlet for the product gas ISprovided It JS connected to two coolers, one honzontal and one vertical To prevent the reaction products flowmg mto the balance chamber, carbon dloxlde IS contmuously passed mto the balance, m such a way that a posltlve flow ISmamtamed from the chamber to the products exit The cooled gas passes through a cyclone separator A fractlon of the product gas IS wrthdrawn for analysts The
entue reactor-preheater assembly and many parts at hqh temperature are made of Inconel 600 The balance used electroIS an electrobalance (Cahn R G Automatic balance), which provides a contmuous measurement of the weight Two chromatographs are used for on-hne aJ'JdySJS of

the product gas stream The columns are essentially the same as those used by Van Damme et al [ll] For an even better separation of C, and C, fractions 2m of n-octanelporasll C IS added, along with 3 m of phenyl Jsocyanate/porasll C (Durapack) Nitrogen, methane, ethylene, ethane, acetylene, carbon dloxlde, propylene,
propane, butene and buta&ene used as Internal standard are analyzed Nitrogen
JS

TO

GC

Fig 1 Flow sheet of the

equipment

tietIcs
4 E-AL The temperature

of coke deposition

m the thermal crackmg of propane

637

RESULTS

range covered by the experiments extended from 720 to 870C The reactor was isothermal Experiments were camed out at atmosphenc pressure and at a dllutlon factor of approxunately 1 mole mtrogenlmole propane The punty of propane exceeded 99 5% There was no adltion of sulfur Typical conversion vs VJF, curves are shown m Fig 3 The mdlvldual expenments were performed at temperatures shghtly different from those mdlcated m Fig 3, but to allow nevertheless graphIcal representation, they were converted to these temperatures by the equivalent reactor volume concept dlscussed m detti by Van Damme et x -1.
60

al [ll] For the calculation of kmetlc parameters, the actual temperature and space tune were used, however The malor products are hydrogen, methane, ethylene and propylene, but hydrogen was not measured Small amounts of ethane and traces of acetylene, butene and butalene were also detected, dependmg upon the conversion The high molecular weight substances and aromatics were not analyzed The major products dlstrlbutlon ts shown m Fig 4 Of course, by the very nature of the reactor this product dlstrlbutlon 1s dtierent from that obtamed m a plug flow reactor, as can be seen from a comparison wth the data of Van Damme et al [ll]

20

LO

60

so

Vc/Fo

Fig

3 Propane converslon vs space tune

VERSION,%

Fig 4 Product dlsmbutlon

of maJor components

for propane crackmg m a completely

muted reactor

638
5 cHARACS

K M Smm
OF THE REACFOR

and G F F~OMENT The parameters 111eqn (4) were estunated by nummlzmg the residual sum of squares on the converslon, usmg Marquardt routme The strong correlation between the s frequency factor and the actwatlon energy necessitated a reparametnzation accordmg to A = A exp (-EIRT,,,) where T, IS the average of all the temperatures The order was found to be 0 92, the actlvatlon energy 49 04 kcal/mole and the frequency factor 2 098 x 10 Smce the order IS close to unity, first order was unposed for the rest of the work The frequency factor then becomes 4 513 x 10 set- and the activation energy 49 28 kcal/moIe These results are m excellent agreement wth the pilot plant results of Van Damme et al who have obtamed a value of 106 x 10 set- for the frequency factor and a value of 50 83 kcal/mole for the activation energy for first order propane decomposltlon m the total pressure range of 1 2-2 3 atm abs and with dllutlon factors rangmg from 0 2 to 10 kg steam/kg propane The spec&ic rate coefficients obtamed m this work are also compared with the data of Buekens and Froment and Van Damme et al m Fig 6 As can be expected the Ieve of mhlbltlon IS more severe m a mixed vessel than in a tubular reactor Van Damme et al obtamed high conversions at high temperatures and hence the specific rate coefficient 1s close to km of Buekens and Froment, while at low temperatures iow converslons are encountered, wth rate coefficients closer to k. The agreement between rate coefficients determmed m such widely
k

In contrast to the earher workers[ll, 131 the present expenments were camed out m a nuxed reactor To check the amount of mlxmg, residence tie distibution stu&es were camed out With the cooler &sconnected, step signals of carbon &oxide were introduced at the reactor inlet The response curve obtained at the reactor exrt for the lowest total flow rate encountered m the experiments IS shown in Frg 5 Input signals contauung less than 2 5% COZ were used because of the hmltatlon of the IR CO;?-analyzer A blank without the reactorpreheater served to evaluate the dead tune of the mstrument and the lmes The correspondmg response curves are shown 111Ftg 5, together with therr transfer functions Clearly, the reactor may be considered to be of the completely muted type
6 KINETIC ANmYSLS OF PROPANE CRACKING

For a completely nuxed reactor, under steady state condltlons, the contmulty equation for propane may be written Fo=Fo(l-x)+rV or V/F0 = V/Coo0 = x/r (2) (1)

where u0 IS the volumetnc flow rate at the mlet, whtie x and r are the conversion and reaction rate at the exit of the reactor, but also those m the reactor The rate equatron for an nth order propane &sappearance IS r = kCo

I-x 1+ (E - 1)x + 6

(3)

where Q 1s the expansion factor (moles produced/mole of propane cracked) and S IS the ddution factor (moles of NJmole of propane fed) Assummg an expansion factor of 2111, 131 eqn (3) yields (4)
F= l-C/Co *

Fzg 5 Residence tune &stnbution studies of the reactor ~rlth step Input of CO, In N2 Response curves A. expenmentai points wthout reactor, 0, expenmental points urlth reactor, Line 1, F(S) = em9OS/(14769s + 1), L,me 2, F(s) = e-@ tss + S/[(l 4769s + l)(l 3673s + I)]

ot
0 5

15

t/q

Fig 6 Arrhemus hagram for first order overall drsappearance of propane

finetics of coke deposmon 111 thermal crack the different reactors IS remarkable In ad&bon, It IS a rehable basis for the scale-up to be dlustrated m the last section of this paper
7 KINETICS OF COKING

of propane

639

A representative curve of coke vs time and a cokmg rate vs tune curve determined by the electrobalance technique IS shown m Fig 7 It shows that the cokmg rate IS high m the begmnmg, but tends to an asymptotic value This Indicates that uutu&ly surface reactions are takmg place, which dechne after a certam tune Initially the coke formation IScatalyzed by the metal surface, but once the latter IS completely covered a steady state activity IS achieved, probably as a result from the mteraction between gas phase components and already deposited coke precursors The colung 1s therefore considered to be of a heterogeneous* nature A smular phenomenon has been observed recently by Shah et al [q for n-octane crackmg It can be seen that this asymptotic rate IS reached very rapidly, usually wlthm an hour The asymptotic coking rate as a function of converslou 1s shown m Fig 8 The coke yield IS ex*!

tremely small when compared to that of other products 0 142 wt% m the asymptotic range, at 6% propane conversion and 870C Coke may be formed by a parallel mechanism or by a senes mechamsm or by both, m analogy to cokmg m catalpc reactions as shown by Froment and Bischoff [8] and as was recently reviewed by Froment [16] In pyrolysis, the mechamsm IS h&ly complicated due to the large number of rtical reactions only the sunphlied models shown m Table 1 were investigated For the parallel mechamsm of model I c,I-IS- * products C,H, *z coke

the cokmg rate equation may be rotten l-x PI rc = kzCc,m = kz l+G+xRT

(6)

COKE

ON

CYLINDER

mg
I12

X. O6,

I? sza c

, 10,

Fig

7 Rate of coke formatlon and coke content on cyimder as a function of tune

6= 1 md

N2/md

C3H6

CONVERSION .% Fig 8 AsymptoUc cokmg rate vs propane conversion

640

K M SuNDARAhi and G F FROMENT Table 1 Cokmg mechamsms

Parallel

1 C3Hs 2 CsHs -

k1 k*

mechamsm Model I

products coke

Consecutive mechamsms Model II k, 1 C& products 3 CsH6 1 C3I-b4 C2tF41 C3Hs 5 C,Hs 6 &I&kz k1 kr k, ks *6 coke
Model III

products coke Model IV products coke coke

The estnnated parameters of the models are shown in Table 2 Model IV wth both propylene and ethylene as mtermedlates was discarded due to non slgmficantly determmed parameters Smce the actlvahon energy for the cokmg reaction is very high for the parallel mechanism, Model I is not plausible The parameters of Model II and III with respectively propylene or ethylene as coke precursors are posltlve and hlghiy slgmficant I>lscnmmatlon between these models was not possible by statIstIcal tests However, a preference to Model II was gven smce the resldud sum of squares of reactJon rates was smaller for Model II than Model III This was substantiated by expenments with mlxed feeds The asymptotic cokmg rates measured with feeds contammg 80 mole% of propane and 20 mole% of ethylene or propylene are compared m Fig 9 with those obtamed with pure propane This also favours quahtatlvely the conclusion that Model II IS supenor to Model III, but the dlscnmmatlon based on statlstrcal tests faded Further, coke 1s one of the major products m propylene crackrng[l7] The best parameters for Model II were obtamed by consldenng all the data for pure and mixed
feeds They are also shown m Table 2

where r, 1s the coking rate per umt area r, IS obtamed experImentally from the coke deposited on the cylinder, which has a known surface area The frequency factor and the activation energy obtained are shown m Table 2 For the senes mechamsm of Model II, the rate expresslon for the coking reactlon IS

8 CORE PROFILES IN AN INLNJSTRIAL REACTOR

slmulatlon of an mdustnal reactor reqmres the integration of the contmmty equations for each species,
The

Estunatlon of rate coefficients IS extremely sunple m the mixed reactor when compared to case of a completely the case of a tubular reactor For example, If the above senes reactIon 1s studted m a tubular reactor, kl and k3 have to be sunultaneously determined, which requires

the mtegratlon of the contmmty equations for all the m a completely mlxed reactor, species However, measuring the rate of cokmg and the outlet concentration of propylene, which IS also the concentration m the reactor, allows k3 to be determined from eqn (7) only Slmllar series mechamsms wrth ethylene as mtermediate (Model III) and with both propylene and ethylene as intermediates (Model IV) were also tested

2.

20

40

60

.I.

Rg 9 Asymptotic cokmg rate vs propane conversion contammg propylene or ethylene T <82O"C, 6 = 1 mole &/mole hydrocarbon, 0 = C& feed only, 0 = 20 mole% of C&L m C& feed, A = 20 mole% of &I& III C&S feed

Table 2 Parameter estunates for various cokmg models

Reactlon 1 2 3 4 5

Rate coefficient kl
kz ks k4 k?

Frequency factor [g coke/(mole J/I m+ set)]

4 51 x 1 25 x 108x 4 38 x IO? 102 lOI lOI

t
vatue 150 42 92 87

Activation energy
[kcal/mole] 4928 108 70 74 97 6471

7
value 14 3 116 114 87

6 3*

ks k3

9 74 x 10 560x 10-s 5 82 x lOI

18 06 10 1

9046 -3181 73 58

42 -05 12 8

tHomogeneous, umts set- @Zonsldenng both pure and mlxed feeds

Kmetxs

of coke deposlhon

m the thermal crackmg of propane

641

of the energy tlon

equation

and of the pressure

drop equa-

for the bends,

where

and dT 1 = dz 8F,C,, 7rd,* Q(z)ndt + Zr,(-All), 4 C

~ =0051+0

19$

dr=

dpt

-$l/M,,,)+~(~~+Fr) -- 1 Mmpt Pt crG RT

where J represents the species, I reaction number, s,] the stolchlometic coefficients For propane crackmg, the present authors1181 have already proposed a molecular reaction scheme based on an investigation m a pilot cracker The same reaction scheme, shown m Table 3, IS adopted here The characterlsbcs of the mdustrlal reactor are gven in Table 4 The friction factors were computed from Fr = 0 092 Refor the stra&t portion of the tubes */d, and from

Fr = 0 092 Re- Id, + &R,,

The angle of the bend A IS 180 m the present case The heat capacities were calculated by a third degree polynomial m temperature with the coefficients obtamed by Thmh et al 1191 from recent experunental data The viscosity of the nuxture was calculated by Sutherland s method[20] Under clean tube condlhons a contmuous heat flux profile along the reactor shown m Fig 10 was used With this profile an outlet conversion of 90% was obtamed, a value which can be compared with the expenmental value of 90 7% The snnulated gas converslon and pressure profiles under temperature, clean tube con&tions are also shown m Fig 10, along with the mdustnal values Excellent agreement was found between the calculated and mdustrml product dlstnbutions, too The coke deposition alters both the mner tube diameter and the heat transfer To account for this reqmres the local values which are also time dependent Therefore, a suitable cokmg rate equation has to be

Table 3 Reacbon scheme for propane crackmg No

c&g

Reactlon
c2H4 + ec,Hs+Hz

) (set- ort I mole- set-

E (kcal/mole)
5060 5129 5906 55 80 45 50 59 39 6001 65 20 4126 34 32 73 58

a-r.4
+ CsHa

4 692 x 10 5 888 x 10 2 536 x 10?

I 514 x 1423 x 3 794 x 5 553 x 10 lo9 10 lo ?

2C&

2C&

C,& * Cd-Lse
C2I-b G&-c5 + CzHz

GHz + CH, C&II4 Hz +

Cd%

0 5cs + 3CH4

3C2&

4 652 x lOI 1026 x lo ? 6960x 10 5 82 x lOI

Coke Model (II) C& coke

Table 4 Charactenstrcs of an mdustnal reactor for propane crackmg

Length Dmmeter Radms of the bend Length of the bend Thickness of the wall Inlet temperature Inlet pressure (clean) Outlet pressure (clean) Propane Bow rate Dduhon factor Density of the coke Thermal conductivity of the coke = 95m = 0108m = 0178m = 0553m = 0008m = 600C = 30atmabs = 20atmabs = 45 99 kmoles/hr = 0 406 kg steam/kg propane = 16kgfl = 0 00154 kcal/(m secK)

642

K M Sw.4R.w

and G F FROMENT

Fig 10 Conversron,

heat flux, gas temperature and pressure profiles for mdustrlal crackmg of propane under clean tube condtttons Lmes, simulated, pomts, mdustnal values

coupled with (8), (9) and (10)

Y$-

z +Az and pc IS the density of the coke In kg/l The updated dmmeter between z and z + AZ becomes rc d*new- 4 old- 2Atc A step by step apphcatlon of the above formula yields the coke thickness along the reactor at any mstant When the cokmg IS accounted for, a certam pohcy for the operation of the reactor has to be selected Smce the heat flux evolution urlth time 1s not known, the gas temperature protie under clean tube condlttons IS retamed for all tunes The outlet pressure was held constant To achieve this and to mamtam the flow rate, the inlet pressure has to be increased as tune progresses This leads to a two pomt boundary value problem Local coke values after 100,300 and 700 hr are shown m Fig 11 The reaction between coke and steam was not accounted for It IS known from mdustrlal practice to be neghgble under normal operatmg con&tlons As a check this reaction was Included m the reaction model, wtth the rate parameters obtamed by &ede and Hanesmn[21] This reduced the coke thickness after 700 hr by less than OOlcm It is clear from Fig 11 that there is a profile m the coke

act =

Smce the cokmg IS constdered to Involve intermediate polymenc species deposited on the wall its rate IS taken at the internal tube skm temperature Model II with propylene as mtermedlate was adopted for r, Smce the rate of coke deposItion IS low, quasi steady state con&tlons may be assumed Consequently, only the coke thickness IS a function of tune, whale the contmmty equations, the energy equauon and the pressure drop equation do not contam any explicit time dependence The tube diameter d, was updated pentically, however The increase m coke thickness At, m the time Interval At m the reactor between z and AZ for, e g for the heterogeneous case ISgiven by *t c = A exp (-E/RF)cAt pc x IO6

where (I?is the average concentration of propylene, F IS the average coke surface temperature between z and
% REDUCTION 12&/dt 1001 f IN DIAMETER

Ftg 11 Coke profiles rn an mdustnal reactor, Model I1 for cokmg

Kmetxs thtckness

of coke deposItion m the thermal crackmg of propane

643

The h&est amount of coke IS formed at the exit of the reactor, where the surface temperature IS highest At this spot the tube &ameter IS reduced by 18 4% after 700 hr of operatlon The coke layer tluckness IS0 98 cm The coke layer thicknesses obtamed from the simulation are m agreement with observations made III mdustnal crackmg tolls It was found that, when the expenmental cokmg rates were expressed per umt reactor volume, rather than per umt metal surface, the amount of coke predicted by the sunulation at the gas phase temperature, was far too high when compared wzth mdustnal observations The evolution vvlth tune of the maximum tube skm temperature (at the exit), of the mlet pressure and of the exit conversion are shown m Fig 12 The exit conversion drops because of the reduction m space tune, although the mlet pressure increases The evolution of temperature profiles are shown 111 Fig 13 The average rate of mcrease m the maximum tube skm
T,.max

temperature LS1 75C per day Chambers and Potter[22] quote a value of 2 3C per day for mdustnal operations, but the feed stock is not mentioned Even higher values would be obtamed d the mlet pressure were further railsed, so as to mamtam the exit conversion constant The temperature dtierence between the tube skm and the gas at the exit amounts to 50C at the start, but reaches 1OOT after 700 hr The evolution of external tube skm temperature substantiates the conclusion of Mo1[23] who reported that mth naphtha crackmg cods with an mtemal diameter exceedmg 7 5 cm the run length IS hmlted by the tube skm temperature The slmulatlons also clearly show that the temperature has a stronger influence on the location of the maxunum coke thickness than the propylene yield
Acknowledgement-The authors acknowledge the support of the Process Data Group of K T I , Zoetermeer, The Netherlands L,gas 17
atm abs

,;.

)o h,
hrs

Fig 12 MaxImum tube dun temperature,

outIet gas temperature, length

conversIon

and mlet pressure

as a function of run

TUt3: r

ALL

TEMPERATURE

750

25

50

75

100

REACTOR

LENGTH

:I.

Fig 13 Evoluhon

of tube skm temperature

profile as a function of tune Cokmg at mtemal coke surface temperature

K M SUNDARAM and G F
~TA~ON

FROMENT

frequency factor, set- g coke/mole I/I m set , outlet concen~ahon of propane, mole/l coke content, g/m uut.mi concen~tlon of propane, mole/t heat capacity, kc~~mo~ K tube diameter, m actlvatzon energy, kcaljmde fnction factor m&aI molar flow rate of hydrocarbons, molefsec molar ilow rate of Jth component, kmoie/sec total mass flow rate, kg/m2 set specific reaction rate coefficient, mole/I set mean moIecular wezght, k&mole reaction order heat of reactzon, kcaljkmole total pressure, atm abs &et pressure, atm abs heat flux, kcal/m* set reaction rate, mole/l set or kmoI/m3 set rate of cokmg, g/m hr radius of the bend, m gas constant, atm mlkmole K or kcafjmole K Reynolds number coke thickness, m temperature of the flmd, *C or K average surface temperature of the coke, K exit gas temperature, C external tube skm temperature, C voIumetrlc flow rate, 1 jsec reactor volume, I eqmvalent reactor volume, 1 propane conversion selectzvity for propyfene, molelmole propane cracked reactor length coo&mate, m
Greek symbols S ddutlon factor, mole/mole

B expansion factor pc density of the coke, kg/l

REFERENCJS

ill Hut T J and Palmer H B , Carbon 1963 I 6.5 i21 Oxley J 0, Secrest A (3 , VelgelN D and Blocher J M
Jr,AJChE.J 196174% 131 Johnson G L and Anderson R C , Proc 5th Cur&on Cotif, Vol I, p 395 Pergamon Press, Oxford 1962 f41 Gay I D , Kmtmkowsky G B , lMichael J V and N&I H , J C&m Phys 1%5 43 1720 Kmney C R and Del Bet E , Id Engng Chem 1954 44 548 :fi Vuk P S f Chambers L E and Woebcke H N , Adv Chem Ser 1974 131 237 PI Shah Y T. Stuart E B and Sheth K D, Ind Engng them P& Des Dev 197615 518 [gl Froment G F and Btschoff K B . C&emEngns Scr l%t 16 Ig9 r91 De Pauw R and Froment G F , C&m Engng Scr 1975 30
789

flG1 Dumez F J and Froment G F, Ind Engng Chem Proc Des De0 1976 15 291 Cl11 Van Damme P S, Narayanan S and Froment G F, AiChEJ 1975211065 E andAibnghtL F,AIChEJ I121 HemottG E,EckertK r133 BuekensA G and From& G F , fnd Engng ckem Proc Des pw 19687435 El41 Kershenbaum L S and Martm J J , A I Ch EJ 1967 13 148 USI Lrudler K J , Sagert N H and Wopzechowsk! B W , Pr-oc Roy Sot I%2 A270242 2161 Froment G F , Catalyst Deactwatron by Coking, Proc Slxrk Int Congr on Cutulyszs, Vol I, p 10 The Chem Sot , London 1977 [17l Kunum T , Saks T , Soma K and Sasakz Y , Ind Engng Chem Fundls 1970 9 318 [I81 Sundaram K M and Froment G F , Chem Engng Scr 1977 32 601 1191 Thmh T P , hiran J L , Ram&o R S and Kabagume S, Hydrocarbon Processtng Jan 1971 98
I201 Rerd R C and Sherwood T K, The # ropetires of

1972 18 84

Gases

a 5

constant m pressure drop equation friction factor m the bends

and Llqwds, 2nd Edn McGraw Hd, New Yo*k I966 [211 Bede B E and Hanesmn D, Ind Eagng Chem Proc Des Deu 1975 14 70 I221 Chambers L E and Potter W S , Hydrocarbon Processmg Jan 1974 125 [231 Moi A , Hydrocarbon Processrng July 1974 115