8/7/2019 10_2_PITTSBURGH_03-66_0088

1/12

SYMPOSIUM ON PYROLYSIS REACTIONS OF FOSSIL FUELSPRESENTED BEFORE THE DMSION OF PETROLEUM CHEMISTRY, INC.

AMERICAN CHEMICAL SOCIETYPITTSBURGH MEETING, MARCH 23-26, 1966

KINETICS OF BUTADIENE PRODUCTION - DEHYDROGENATION OFMIXED n-BUTANE + BUTENES AT 1/6-ATMOSPHERE PRESSUREEXPERIMENTAL RESULTS AND EMPIRICAL CORRELATIONS

BYR. E. Sattler and E. W. Pitzer

Phillips Petroleum CompanyBartle sville, Oklahoma

I. INTRODUCTIONIn connection with Phillips continuing in te rest in butadiene production, data we re de sired

for the conversion of n-butene to butadiene in a single s tage and at le ss than atmosphericpressure . Earlier experimental data were limited to the conversion of pure n-butane to butenes+ butadiene. For pr es en t considerations of this proce ss, data were needed for the conversionof mixed n-butane + butenes (steady-state feed). In addition, data we re desired for use in studiesof n-butane and butenes dehydrogenat ion kinet ics.

To accomplish thes e objectives, steady-state feed mix ture s wer e dehydrogenated with1/70-inch particles of chromia-alumina catalysts at 125 mm. mercury absolute pre ssu re andover the temperature range of 1050 to 1250F. Produc t distributions and reaction rat es werecalculated from the exp erimen tal resu lts.

II. EXPERIMENTAL WORKA. Apparatus and Equipment

The equipment used fo r the study of low pr es su re dehydrogenation reaction ra te s isschematically outlined in Figure 1. The total volume of th e syst em was 8-9 cc and the reacto rcontained 2.2 cc. The schematic description of the furnace and reactor used for 1/7O-inchcatalyst particles is shown in Figure 2.instantaneously between hydrocarbon and other ga s feeds. Electri cal ly heated furnaces wereused to preheat the gases, to heat the catalyst and to produce a "heat seal" a t the react or outletto minimize hea t lost through the exit of the furnace. The preheat and cat alyst furnaces werecontrolled with Wheelco Amplitrol tem peratu re c on tro llers which operated in conjunction withchromelalumel thermocouples located above the catalyst bed (for prehea t furnace) and on theexternal w a l l of the cata lyst zone (for the catalyst zone furnace). The heat-seal furnace w a sfound to give bet ter heat insulation if used in connection with Var iac control only. Ceramicorifice gas flowmeters and Ideal needle valves we re used to control the gas flows. Pre ssu recontrol was manual and absolute pr es su re was m easured at the reactor inlet and outlet withmer cur y manometers. The total effluent w a s collected and mixed with a Toepler pump. Samplesof the total mixture w er e analyzed by ga s chromatography.B. Materials and Reagents

Catalyst was p repared by impregnating alumina granules with aqueous chromic acidsolution, drying, calcining at 100O0F, and prepar ing a 40-50 mesh fract ion by sieving. Thecataly st was analyzed and found to contain 29.52 weight pe r cent CrgO3 and .12 weight pe r centNa2O on d r y basis. Surface are a by B. E. T. method was found to b e 70 square meters per gram.resea rch gr ade ethylene and propylene from Phillips Chemical Company, (2) prepurified nitrogenand cylinder hydrogen fr om National G a s Company, (3)and building service air which contained0.0006 grams of C02 per liter and approximately 0.0010 gr am s H20 pe r lite r.C. Proceduresthe ca talyst was not in use it w as usually heated with air going over the catalyst (and in someinstan ces N2) at a space rat e of 1000. Approximately 2.0 cc of l/7O-inch part icle s wer e tested.

Three-way solenoid valves were used to switch

Gases used for th e test s were: (1) pure gr ade normal butane, pure gra de butene-2,

A standard procedure was adopted for testing the ca talyst to insu re reproducibility. When

C-88

8/7/2019 10_2_PITTSBURGH_03-66_0088

2/12

A te st began by treat ing the ca talyst with H2 at atmospheric pre ssu re for 10 minutes at a spacera te equivalent to the hydrocarbon space ra te to be used but not going below 1000 spac e rat e.The system w a s evacuated, checked for leaks for a period of one minute. Immediately after theleak checking period the three-way solenoid on the re ac tor outlet was activated so there was nopossibility of hydrocarbon in the by-pass pres surin g back to the catal yst zone. Nitrogen wasthen turned on over the catalys t at a space ra te equal to the hydrocarbon space ra te and a momentla ter the hydrocarbon was turned on and allowed to exit through the by-pass. When the pres su rein the react or built up to the d esir ed pres su re of 125 mm Hg absolute (or to atmospheric as in thecas e of standard butane runs) the re ac tor outlet solenoid was deactivated so as to commutenitrogen and hydrocarbon below the reac to r and control was maintained with one ra te valve. Thetime for thi s operation was kept to a minimum and usually was accomplished in 2-3 minutes.After a ll flows and pressures,appea red to be steady, a switch controlling the two three-waysolenoids above the reactor was activated so that the ga s flows were instantaneously revers ed(i .e. hydrocarbon sta rted going over the cata lyst and nitrogen through the by-pass). An electricclock was connected with th is switch s o the time measure ments w er e begun simultaneously withthe introduction of hydrocarbon. A t the same time the switch controlling the solenoids above thereac tor was activated, a second switch was activated energizing the solenoid below the r eac tor .Thus the effluent was conducted into the s ample collection system and the pre ss ur e in thereactor w as manually contro lled by a second rat e valve. During the dehydrogenation per iod ofthree minutes, the by-pass and the reacto r were maintained at the sa me operating pr es su re tominimize possible leaks through the re act or outlet solenoid. A t the end of the dehydrogenationperiod, the three-way solenoid on the reactor outlet was deactivated and nitrogen was allowed tosweep out any re sidual hydrocarbon a t the same sp ace ra te a s that used for hydrocarbon.Flushing w a s conducted long enough to give time f or all hydrocarbon to be swept out of thereactor and into the sample collection system. A t space r at es of 100, flushing was continuedfo r about 7 minutes--the tim e being measured accurately. After a ll hydrocarbon was removedfrom the rea ctor the sample was "closed in" and the p re ss ur e was allowed to build up in Ngto atmospheric. The sample was removed after the temperature , pr es su re and volumes wererecorded. The sample was replaced with columns of CaSO4 and Ascari te (asbestos impregnatedwith KOH) connected in se ri es. Nitrogen was turned off and service air was started at 1000space rate to remo ve hydrogen and polymer (as water and carbon dioxide) fo r about one hour.

Temperature measurements were made at six different locations. These a r e shownschematically in Figure 2 . At two of the six locations double thermocouples were used--one forcontrolling and the other for reco rding tempera tures . Thermocouples were located:(1)Externally between the r eactor and the core midway down the preheate r to indicate howefficiently the gase s were heating to reaction te mperatures. (2 ) Internally at the preheat outletor about 1/4-inch above the cata lys t bed and inside the thermocouple well. Here also is a secondthermocouple used to control the temperature of the preheat furnace. (3 ) Internally at 1/4-inchbelow the catalyst bed and inside of the thermocouple well. (4) Externally between the catalystrea cto r and heating core l/4-inch above the catalys t bed and on the s ame horizontal axis a s (2)above. This thermocouple usually gave lower tem peratu re readings than (2 ) above because atthi s point the junction of the narrow pre heat furnace and the broader rea cto r furnace allowedsome heat to be lost. Eventually this thermocouple showed signs of radical behavior by givinglower temp eratu re readings. However, it w a s not replaced because of the danger of dilocatingother thermocouples in the pro ces s.heating core and midway between the upper and lower ca talyst zone. Two hermocouples werelocated at this position; one for control of the ca talyst zone furnance and the other for tempera-tu re recordings. ( 6 ) Externally between cata lyst rea cto r wall and cataly st heating zone coreabout 3/4-inch below the catalyst zone and at the upper edge of the hea t seal reactor. Tempera-tures were adjusted so as to sim ula te adiabatic conditions. Before going on hydrocarbon,temperature s at positions 2, 3, 5 desc ribed above wer e maintained within ? 5F of the de sir edreaction temp eratu re. Tem pera ture readings were taken at one minute inte rvals and thetemperature drop recorded.average tempera tures over the total dehydrogenation period. In actual prac tice per fect adiabaticconditions were not possible because some heat was supplied through the walls as the endothermicreaction took place.temp eratu re employing a molecular si eve column for hydrogen, nitrogen, oxygen, methane andcarbon monoxide, and a hexamethyl phosphoramine column for the remaining components. Themaximum deviation for these analyses was around 5% of each component which w a s present inthe amount of 10% or les s. For components pre sent in large r amounts, the maximum deviationw a s ? 1-98. Typical deviations a r e usually found to be around +1%of each component.

(5) Externally between the catalyst reac tor zone and the

Tables showing temperature recordings therefore rep rese nt the

Analysis of hydrocarbon sam ples were made by chromatographic procedures at room

c-89

8/7/2019 10_2_PITTSBURGH_03-66_0088

3/12

TABLE IRESULTS OF REPRODUCIB IL ITY STUDIES

3.oo3.155.202.593.821.652.50

33.930.431.627.739.037.4

31.5131.30

Melean 3.u 35.0 30.63Std.Det. 21.13 (236%of mean) 2 3.6 (+l@ of mean) 2 1.06 (+3.5$ of mean)

(a)(b) Conditions: ll00 F, 1/6 atmosphere absolute pressure,Convwrsion mole of butane destroyed per 100 mole of feed.3 minute dehydrogenation period; feed: the mixtareobtained mer 20 per cent of t he butane is convertedin an in t eg ra l run using a ateady s t a t e butane-butenemixtm.

(e ) Conditiom: 1100 F, 250 GBV, feed; 60% butanebatenee (stead y-state) ; 1/6 of an atmosphere pressur e;3 minute dehydro&enation period.

(a ) With n-butane feed o-(b) Vlth d x e d n-butane + butene feed on l y .(e ) Volumes of feed (S.T.P.)/mlume catalyst/hotw

c-90

8/7/2019 10_2_PITTSBURGH_03-66_0088

4/12

In,practic e dual analyses we re run on each s ample and if the deviations we re norm al they we reaveraged and these were used for computations.

Catalyst deposits were removed after each runby oxidation with a ir and me asured byabsorbing the water formed in a column of Cas04 and absorbing the C 0 2 formed on Ascarite.

Conversion, yield, selectivity and reac tion ra te data wer e calculated with the IBM 7090computer.

Precisio n studies were made to determine the reproduc ibility of resul ts with the equipmentand techniques that we re employed. Table I presents precision data for differential and integralstudies by the original techniques along with data after the equipment and techniques w e r emodified in an attempt to improve reproducibility.

The result s of this study show that the prec ision of integral runs is th ree times betterthan that fo r differential run. This is due to the problem of m easuring smal l changes in reactionproducts resulting from the low conversions requir ed for differential status. After modifyingthe procedures and equipment the precision was increased three-fold to a significantly goodfigure e 1 . 0 6 standard deviation or 3 . 5% of mean). Prec isio n data were not obtained for thedifferentialruns using the improved techniques, but it is believed that a similar three-foldimprovement would be found.D. Ranges of Experimental Conditions and Data

Table I1 presents the rang es of experimental conditions used in this study. All possibl ecombinations of these variables were not used; only pure n-butane was used as feed at oneatm osphere and it only at a temperature of 1100F. Sufficient levels of space velocity we re usedat mo st temp erature levels to establish a butane conv ersi odspac e velocity profile. The resultingexperimental data and calculated result s a r e quite extensive and a r e not reproduced here .

Since one objective of the mixed feed runs was to obtain data representativ e of operationof a plant operating on "steady-state" feed, the feed composition was changed with th e tempera-tu re level. Feed composition vari ed from approximately 62 mole % n-butane, 38 mole %butene-2 at 1050F to 73 mole % n-butane at 1250F.

III. CORRELATION OF DATAAs a result of using "steady-state" feed s, net butene conversion in all runs was zer o plusThe complicating fea tur e of th e corre lat ion of the dehydrogenation data was th e gradual

Data taken over a short period of ca taly st life, particu-or minus a few per cent. There fore, corre lat ion of butene conversion was not attempted.decline in catalyst activity with time.lar ly early in the catalyst life, indicated that conversion could be well cor rela ted with spa cevelocity by an equation of the form:

x = fractional conversion, mols converted per mol of feed.fa ,b = empirica l constants.

= reciproca l space velocity, ft3 catalyst-hr per mol feed.

Reciprocal space velocity in te rm s of mols of feed rather than volume of feed wa s chosen as theindependent variable to give the proper units fo r the ra te of reaction calculated by taking thederivative of x with respect to f .

The normal butane dehydrogenation data obtained at atmospheric pr es su re and 1100F.we re used to determine the effect of catalyst age on conversion. A number of equation fo rm swer e trie d; the only acceptable fit of the data was obtained with an equation of the for m:

fa + c ed + bf=where x = fractional conversion, m ols co nve rted mol feed.f = reciproc al spac e velocity, f t3 catalyst-Hr/mol feed.e = age of cataly st, days.

a ,b , c , d = empiric al constants.A computer curve fit of the data gave the following values for the constants in equation (2):

a = 0.13707b = 1 .26470c = 4 . 1 8770 xd = 2 . 8 4670

A comparison of the exper imenta l and calcula ted onvers ions is shown in Figu re 3.

c-91

8/7/2019 10_2_PITTSBURGH_03-66_0088

5/12

VACUUM

\

SAM?

CERAMIC C ER A MI CO R I F I C E THREE-WAY O l l F l C E

RATERATE FLOWMETER 50LE11110S I 'LOWMETER

B U T M E , B U TE NE S.MIXED BLENDS

VACUUM

tTOEPLER VACUUMPUMP

F I G U R E 1A S S E M B L Y OF C A T A L Y S T T E s T l N G E O U I P M E N T

ELEMENTI OLRECTLY.Y*T E X I T

TUBE.

Ie

ON5 IN MILLIMETERS VOLUMES

THERMOCOUPLE INLET

CATALYST INLET1 - 3 MM IO 2. 2 cc

6.f c

1D I M E N S I O N S SAME- A 5 ABOVEWERMOCOUPLE INLETTHERMOCOUPLE LOCATIONS

F I G U R E 2R E A CT O R A N D F U R N A C E A S S E M B L Y

c-92

8/7/2019 10_2_PITTSBURGH_03-66_0088

6/12

There a r e a number of ways of express ing the activity of one cata lys t rela tive to that ofFrom the standpoint of r eac tion kinetics work, thesecond (or of the first at a different age).

pref erre d method (1)is to express the activity as the ratio of space velociti es neces sary toobtain the same conversion with the two catalysts.desired to convert all the data available to the ir equivalent with a new (zer o age) cat alyst. Fora z ero age catalyst equation (2) reduce s to:

To obtain consistent corre lations , i t was

f 0x = a + bfowhere fo = reciprocal space velocity with zero age catalyst.To find an expression for the equivalent space velocity with a zero age cataly st, equations (2)and (2a) a re equated: f (2b)- 0a + bfo a + cod + bfSolving this expression for fo in ter ms of f and @ gives:

The quantity ( a +aced ) is, in effect, the catalyst activity factor. Equation (3) w a s used tocorrect all the experimental space velocities for th e mixed butane-butene feed data to theirequivalents at ze ro catalyst age. In so doing it w a s tacitly assu med that the constants a , c, andd we re &functions of temperatur e and pre ssure , or, if they we re , that the effect of tempera-tur e and pre ssure w ere cancelled out by the form of the activity factor relation. It should beemphasized that this relation for ac tivity is stri ctly emp irica l and mus t used with anyother se t of data or calculations without experimental confirmation.

The mixed butane-butene feed data we re corre late d a s follows:1. The experimental butane conversions, XI, w here co rrel ated a s a function of the

equivalent reciproc al spa ce velocities, fo, calculated as outlined above in the form:x1 =f,a + bfo

A separat e correlation was obtained at each te mpe ratu re leve l except 1200'F. A t 1200F. onlythree data points were available, all at essentially the same space velocity, so that a validcor rela tion could not be obtained.

2. The logarithms of the values of a and b obtained at each temperat ure level wereplotted against recipro cal absolute tempe rature, Figu re 4.appeared that the 11CD"F points were out of line. These points wer e ignored and the remainingthre e points on each plot, rep rese ntin g data at 1050, 1150, and 1250"F., were f i t to an equationof the form:

From inspection of th ese plots , it

In a (or In b) = CY + ,9 (1. x l o 5 ) (5)TEvaluation of the constan ts gave the following result s:In a = -15.97115 + 0.25232 $ x l o 5 ) (6 )

(7)n b = -1.82226 + 0.04281 (1 x l o 5 )Twhere T = OR, F + 460.

As a test to justify ignoring the 1100F. points , equation (6) was used to calculate 5 a t 1200F.This value of a and equation (4) were then used to calculate a value of b from each of the th re edata points available at 1200F. The three values of b we re then averaged. This average valueof b wa s identical with the value calculated by equation (7) for 1200F. It was concluded thatignoring the 1100F. points was justified sinc e the 1200F. data had not been used to developequations (6 ) and (7 ) .

To check the overall correlation, equations (6) and (7) were used in conjunction withequation (4) to calculate a butane conversion value for each experimental point.a plot of observed versus calculated butane conversion and Figure 6 shows the experimentaldata plotted as conversion versu s equivalent recipro cal space velocity; constant tempera tureslines a r e shown calculated from equations (4), (6), and (7).predicting butane conversion is 0.02429, very close to the stand ard deviation of re plic ateexperimental determinations.

Figure 5 shows

The standard e rr or of estimate for

c-93

8/7/2019 10_2_PITTSBURGH_03-66_0088

7/12

c-94

8/7/2019 10_2_PITTSBURGH_03-66_0088

8/12

A similar correlation procedure was used to cor rela te the data on butadieneappearance. A t each temperatu re leve l (except 1200F. ), the data were f it to:

x2 = ___a + bfowhere x2 = butadiene appearance, mols per mol of feed. To be consistent with the re sult fo rbutane disappearance, the 1100F. data we re ignored, and the a and b values fo r 1050, 1150,1250F. fi t as functions of tempera ture. The resu lts were:

In a = -13.03794 + 0.19267 (1 l o 5 )In b = -4.48793 + 0.09814(Il o 5 )

(sa)(7a)

TT

Figure 7 shows plo ts of equations ( sa ) and (7a) with the data points. A check calculation'a t1200F. confirmed the validity of ignoring the 1100F. data . Fig ure 8 shows a plot of observedve rs us calculated butadiene appearance and Figure 9 shows the experimental data plotted asappearance versus equivalent reciproc al space velocity, with constant temp erature lines calcu-lated from equations (4a), (sa ) and (7a). The standard e rr or of predicting butadiene appearanceis 0.01414.velocity with res pec t to reciproc al sp ace velocity, an expression is obtained for the rate . Thus,fo r butane disappearance the ra te i s:

By differentiating the equation expressing conversion as a function of reciprocal space

a(a + bfo)2r l =

and for butadiene appearance the rate is:a

(a + bfo)2-2 =where r l = mols butane converted/ft3 catalyst-hr.

r2 = mols butadiene produced/ft3 cat alys t-hr .f0 = equivalent reciproc al space velocity for zer o age catalyst, ft3catalyst-hr/mol feed.a and b = constants calculated from equations (6) and (7 ) for butane or (sa)

and (7a) fo r butadiene.

IV . DISCUSSIONIn prepara tion for this study, the tw o w e l l known methods of obtaining react ion r a te data

were compared, viz. , differential experiments (using various feed mix ture s which rep res en tthe total product at various locations in the rea ctor , and a relatively short residen ce time togive a small conversion), and integral experiments (using a feed mixture which re pres ent s thesteady-state feed to the re ac to r, and various re sidenc e times to give various conversion levels ).Based on this comparison, the int egra l experimental method was chosen because it gave mo repre ci se result s than wer e given by differential experiments. This resulted from the relativelylarg e changes in feed composition which occ urred in the integral experiments compared to th erelatively sma ll changes in differential experiments.

In this work to obtain data for us e in stud ies of n-butane and butenes dehydrogenationkinetics, temperatures above those normally used with chromia-alumina catalysts wereemployed. This caused a reduction in catalyst activity and requ ired the use of a catalystactivi ty index. The value of thi s index wa s establ ished at frequent inte rva ls by testing theca talys t fo r the dehydrogenation of steady-state feed at 1100F. This index has been used incorr ela tion s of these data; however, it may not be entirely sat isfacto ry according to recentte sts . These test s with pure n-butane and pure butenes showed that aging o r heat-treating thechromia-alumina catalyst reduced its activity for n-butane dehydrogenation but did not affect itsactivity for butenes dehydrogenation.they we re obtained before the experimental procedure and equipment wer e entirely sa tisf actory.Some of the rea sons f o r the unreliability of the data were: (1)flow rat es in the catalyst zonevar ied widely during each te st due tn the relatively large volume of the gas handling system,and (2) the product sampling procedure did not insure collection of the total product from a tes t.These problems affected only the 1100F. da ta for the steady-state feed.

Data for the 1100F. profile are questionable because

c-95

8/7/2019 10_2_PITTSBURGH_03-66_0088

9/12

FIGURE 5CALC ULAT EDVS OBS ERVE D BUTANE CONVERSIONM l X E D BUTME-BUTENE FEED. 12 5 L(M HC1 0 5 0 V . I I O O F . 1150F . l P 0 0 F M D 1 2 5 0 F

FIGURE 6BUTANE CONVERSION

VSEOUIVALENT RECIPROCAL SPACE VELOCITY

C-96

8/7/2019 10_2_PITTSBURGH_03-66_0088

10/12

The correl ation s developed ap pear to fit the data within the experimental accuracy.Figure 3 shows a slight trend toward a breakdown of the butane conversion corr ela tio n at hi&conversion levels.It must be emphasized that the correlations a r e empirica l and ca re must be exercised intheir us e. Within the range of operating varia bles covered the corre latio ns are adequate fo rpro ces s design work. Their most effective use, however, may be to use them to smooth th eexperimental data pr io r to evaluation of reaction mechanism type equations. The us e of thecorre latio ns in this manner will give an internally consistent s et of values for ra te s to evaluatethe Constants in reaction mechanism equations.

V. CONCLUSIONSFrom the results of this study it is concluded that the conversion of n-butane and th e

appearance of butadiene in the dehydrogenation of n-butane can be satisf actor ily corre lat ed asfunctions of recip roca l space velocity using a two constant hyperbolic equation. The twoconstants of t his equation can in turn be satisf actor ily correlated as functions of tem per atu re byan Arrhenius type equation.

Within the range of the var iable s studi es the r esult ing equations a r e suitable for proc essdesign work. The equations should als o be useful i n smoothing the experimental data to obtainra te s f or a mechanistic type study of the reaction kinetics.

V I. LITERATURE CITED(1) Hougen, 0. A ., and Watson, K. M . , "Chemical Pro ce ss Principles, Part Three,Kinetics and Catalysis", p. 936. John Wiley and Sons, Inc., New York, 1949.

c-97

8/7/2019 10_2_PITTSBURGH_03-66_0088

11/12

1

i

c-98

8/7/2019 10_2_PITTSBURGH_03-66_0088

12/12

I