Cl~ernical Cngr,~rrrirrg Sciwzw. Vol. 13, No. 8. pp. lXOlL1806.19X8. Printedin Great Bntain.

m9-2sn9/xx

$3.oo+o.w

Pergamon Pressplc

Simulation

of

Steam

Reformers

for

Methane

M.A.

Soliman,

S.S.E.H.

El-Nashaie*.

AS.

Al-Ubaid

and A. Adris

Chemical Reaction Engineering King Saud University, *Author to whom

Group (CREG), Chemical Engineering Department P-0. Box 800, Riyadh 11421, Saudi Arabia correspondence should be addressed.

Abstract industrial steam A model is developed for reformers for both-top fired and side fired tube model is a onefurnaces. The catalyst model with intradimensional heterogeneous The two particle diffusional resistances. point boundary value differential equations of the catalyst pellets are solved using a modified novel orthogonal collocation technique to effectiveness obtain the factor variation the lenkth of the reactor. The side along equations are algebraic equatfired furnace the top fired furnace equations are twoions, boundary value differential equations point which are solved using the orthogonal collocaA recently developed more tion technique. The model rate expression is used. general industrial against checked performance is The model is used to investsteam reformers. igate the effect of various parameters on the behaviour of the catalyst tubes and the furfactor variation The effectiveness nace. along the length of the catalyst tube is also analysed. Reactor modeling, Keywords: steam Reforming. Digital Simulation, effectiveness factor Introduction Most used today is produced by steam sy*gas reforming. In Saudi Arabia and Egypt this percenta&e is almost 100%. and therefore economic comparison -with other methods is not however based on a plant starting available, in 1987 steam reforming Is up in the U.S.A. the most economic choice (Gaff and Wan& 1987. Table 1). Lfficient steam reformers need to be designed thermodynamic basis rather than on kinetic equilibrium basis, (Rose 1977), and therefore diffusion-reaction kinetic SXp~C2SSiOn and models which are reliable over a wide range of Extensive work dealt parameters are needed. reforming of with kinetics of steam the catalyst, methane over nickel supported and some of the review papers offer a good survey (Rostrup--Nielsen, 1984, of subject the Elnashaie et. al.. Some investigators 1988). obtained effective order for the negative reaction with respect to steam (e.g. Bodrov, Al-Ubaid et. al.. while others 1964, 1987). obtained positive effective order (e-g. DeDeken et. al.. 1982). The more general rate SXpt-SSsiOn which shows this non-monotonic dependence of the rate of reaction upon the steam partial preseure obtained recently by Froment and co-workers (Xu and Froment 1988) is used in this investigation.

The performance of steam reformers is strongly affected by the heat transfer from the furnace to the catalyst in the tubes (Hyman, 1968). The developed model for the whole steam reformer including both the catalyst tubes and the furnace is checked successfully against a number of industrial reformers, with both top and side fired furnaces in the open literature (Singh and Saraf 1979) and in the ammonia and methonal industries in Egypt and Saudi Arabia. The model was then used in a detailed parametric study that elucidates the COmplSX interactions between the different processes taking place in the catalyst tubes, and the complex interaction between the catalyst tube and the furnace. Catalyst Tube Model Development: Reaction Kinetics Used: The kinetic rate expression used in this work is the one developed by (Xu and Froment 1988) with the following equations:

cn4 + Hz0

co

+ 3n2

(I),

co

+ H-JO

cop +

Hp (II), CH,, + 2H20 co2 + 4H2 (III> based upon adsorption-desorption mechanism model consisting of 13 steps, three of them are rate controlling and thus arrived at the following kinetic rate equations: 0.5 PCH4 . PH20 PH2 - PC0 )/DEN*, =1 = kl ( P2.5 Kl H2 PC0 r2 = k2 ( pH2 PCH4 r3 = kg ( P3.5 H2 where, + kH20 DEN = . Pi,, po.5 Hz Kl * PC0 2)/DEN.2 . K2 'H2 and -+ 'CH4 r3 are 'CH4 the (2) * PH20 PC0 - ~)/DEN~ K2

(1)

' + kCOPCO and

+ kU2 rl, r2

PH20/PH2

rates of reactions I. II and III respectively. The rate of disappearance and appearance (formation) of CHq and CO, are given by: rCH4 =rl+r3, (3) rco2 = r2 + r3 For this reaction system the number of moles of the different components can be expressed in terms of number of moles in feed and the and CO2 conversions, which are defined CH4 as: XCH4 = = (FCH4 (n CO2 - nCH4)jFCH4 - FC02)'FCU4 ' XC02 (4)

1801

1802

M.

A.

SOLIMAN

et al.

A2

Detailed analysis of the non-monotonic behaviour of these rate equations is given elsewhere Also all kinetics (Elnashaie et al., 1988). and equilibrium constants and their dependence upon temperature are given elsewhere (Xu and Froment. 19881. The Mathematl&l Model: Single Energy and Momentum Balances: Mass, perf0rmsIIce to be reformer tube is assumed tube in the other representative of snY The material, energy and nxxnentum furnace. balance equations can be written as:

(G StjR

is

the

total

exchange

eree

between

dXcH
4=A.p dll
7

' 'GE4

- rCH4/FCH4

'

d%u 2_

flue gas and reformer tube, and (c '&)R black = ' is that for a black tube surface, A, n . dc . L. P a is a factor accounting for the part of the that falls on radiation the cold plane and -aA / which is observed by the tubes. FR =P where 0 c AR < 0.5a 5 (AR+aAc) =PP Assuming 2% of the net heat release Qn lost to the surroundings and that the heat transfer the by convection is negligible, following heat balance equation csn be written as shown in equation (11) where QG is the sensible heat of the exit flue gas: fL 0 QR de = 0.98 Qn e; (11) ROesler takes the

-A

PB

nco

- fC02/FCH4 "m 4 (rL+r3) + PB

(5)

[PB AH2 nC02(r2+r3) +

AH1

4 &

(Ts-T)]

(6)

bModeling ~of top fired furnaces: model as modified by (Filla 1984) form: d2 EL df2.
Q

heat transfer coefficient U relation calculated by the 1988):

(Kcal/m2.hr.0K) is (Xu and Froment, (7)

al

[6(E1
Er

11 T;)

Et

s,(EL

II,

T:,&
-

*,(<l
st(E2

4) El
~(1 -

4~Ez)

1
+

(12)
Er sr (13)

convective where the a'i (Kcal/m.hr.K) is heat transfer coefficient in the packed bed and is obtained from the correlation of (Leva et. al., 1951). The differential pressure drop Is given by the 1954) and the friction equation of (Fanning, factor is computed using (Hicks 1970) correlation together with the modification of (Singh and Saraf 1979). Catalyst Partifles Equations: The catalyst pellets are assumed to be slabs Material length, a,. with a char&teristic balance equations take the following form:

d2 E2 dL2

u2["t

$) T:,,,)

(4 E2 With ---2 the 1 a2

(1 - 9)
boundary dE dlL

El>1
conditions & dE1 df [(I at Q) 1 a2 * = El dE2 de

1 0, -- dE1 a1 df 'IJ

E2)1

atn-L,-1~=--=[(l - +) El - ;&

er 2-s=

d2% dw2 with

= ai for

10 -5

Q2
ai'l, -1 for at i R CO2 (8)

The differential stream will take d[Tg Gg % - F(T* df

heat balance the form - Tg,J] -28

on

the

flue

gas

RT-rf, Di,e i E en4 and

[E1-WEgl

the

boundary

conditions w-1 Fi,s

w = 0

d -=o pi

.+ti _-

= Fi

The slab is assumed isothermal and external msss and heat transfer resistances are negliAll physico-chemical parameters formugible. taken from (Reid and Sherwod 1966) lae are Also all and Bischoff 1979). and (Froment than methane in feed are hydrocarbons higher assumed to instantaneously crack into CH,,, H2, CO3 and Co. Radiative Heat Transfer Two tvnes ot turnaces firinp: are considered: and top-fired furnaces. side gired furnaces Side side fired furnaces: sModeling of wellare uf3UGZill modeled y fired turnaces stirred enclosure having a mean tema&rature which is different from the exit temperature. The radiative heat transfer equation takes the form (Singh and Saraf, 1979):

(14) where T* is the adiabatic flame temperature, and F is the fraction of the fuel burned along the flue gas Inlet temperature the reformer by T g o is given , _ [El.0]1/4 (15) Tg,o 00 the heat and transfer by radiation to the reformer tubes is given by: - Et) V where E, = CJ T: and V QR - 2 Et st(Er is the free volume of the reformer. Numerical Solution: Differential equations solved using 4th order Runge-Kutta (5,6) are subroutine with automatic step size to ensure The furnace equataccuracy of the solution. ion for side fired becomes a single non-linear algebraic equation in the flue gas temperature which is solved by an IMSL (International Mathematical and Statistical Library) subroutine celled %SPOW based on a discretized Newton method. The differential equations describing the heat transfer in the top fired furnace are discrethe orthogonal collocation method tized by using a cubic polynomial leading to 12 simulnon-linear algebraic equations which taneous are solved by ZSPOW. At every location along the reactor the effectiveness factors have tc~ be calculated and for this purpose a modiffed Using this method collocation method is used. four equations there Is no need to solve

(p 't>R 1 a A, P (L-l)+ Et = 'g 1 (' (a Ac 't)R,black P +

TV

(G Nt)R.black

AR
1 + Et

c,/l(l-~~)
- a 4 Tt,o

FR
t

Eg -

T;

A2

Simulation

of steam reformers

for methane

1803

simultaneously as in the standard two point collocation method. In the present method two equations are solved at the first collocation point followed by the solution of two the second equations in collocation point. Details of the numerical algorithm IS given elsewhere, (Elnashaie et al 1987b). Numerical with indicated a experience this approach saving of 314 the time required to solve the equations three-splines particle using a collocation method developed by accelerated (Xu and Froment 1988) with a comparable accuThe overall computer program is equippracy. ed with the facility to compute the change of the the length of physical properties along the catalyst tube and for the change in feed compositions. The mDde1 developed Results and Discussion: previous in the parts is characterized by including the essential phenomena taking place in the catalytic Steam reforming tubes. The developed model has Model Verifications: been checked anainst a number of industrial reactors with both side fired and top fired furnaces Saudi Arabia and in the in EgyPt. A sample of the results is open literature. both side fired as given in Tables 1, 2 for well as top fired furnaces. The agreement is obviously quite Eood. Effect of -Operaring and Design Variables on the Performance of Steam Reformers: A detailed parametric study using this more general and industrially verified model has been carried A sample of the out (Elnashaie et al 1987b). presented in study performed is parametric this paper. 1. Effect of molar steam to methane ratio: Table 3 shows the results for different values steam to methane ratio which are of molar chosen around the usual industrial values. It is shown that the methane conversion decreases slightly with the increase In steam to methane this decrease in the conversion is not ratio, due to a decrease in the rate of reaction because the system is in the positive reaction order region with respect to steam where the rate of reaction increases with the increase in steam partial pressure, in fact it is due to the decrease in residence time with inThe conversion to crease in steam flow rate. due to the increase in the shift CO2 increases and the methane reaction conversion to CO2 with the increase in steam partial pressure. The increase in methane conversion decreases beC&Use the reactions are the temperatures For the same methane conversion endothermic. the increase of CO2 on the expense of CO causes an increase in the temperature because is less endothermic than that the CO2 reaction Obviously the increase of the heat in for Co. flowrates) decreases the exit (due to higher In Table 3, although the the temperature. converconversion decreases methane and 002 with the increase in steam sion increases feed, the temperature decreases because of the heat increase at exit with the excess steam. Fig. 1 shows the efffectiveness factor profile along the length of the catalyst tubes. The the balance behaviour of n depends upon taking different processes place between The increase in inside the catalyst pellet. the intrinsic rate of reaction tends to cause the to decrease, while the increase in n coefficients diffusion tends to CBUQO 17 to increase and the increase in the equilibrium constants tend to cause n to increase. The final behaviour of n depends upon the

interaction between these length of the tube. The n complex non-monotonic be Y& monotonically nit behaviour decreases. of nCo2 has

factors along the in Fig. lshows a viour, uhile nCH Similar been non-monotolf observed by

(DeDskan et al 1982) also this phenomenon has been observed for ammonia synthesis by Elnashaie et al (1987a). 2. Effect of catalyst activity: Table 4 the effect shows of decreasing the catalyst activity from 1.0 to 0.02. As a result the conversion decreases by about 24% and the catalyst tube temperatures increase by about 5-6%. while the furnace temperature increases by lo-11%. Such relatively limited changes compared with the very large change in catalyst activity is due to the very large diffusional limitation and thus the considerable increase in n with the decrease in catalyst activity as shown in Fig. 2. At this low catalyst activity the non-monotonic behaviour of n is not confined to but it extends nC02 also to nCH,. (Fig. 2). 3. The effect of catalyst particle size: Fig. 2 shows the effect of catalyst particle size on n for catalyt activity = 0.02. It is clear that n increases considerably with the decrease in particle size. It is also noticed that as the particle size increases nCo 2' 'CHq ;pLa;aches each other and the sharpness in the and minima of n profiles decreases. Table 5 shows the effect of pellet size on conversions and temperatures, it is clear that the conversions increases as the pellet sire decreases due to the increase in the n and the the temperatures decreases, while pressure drop increases. 4. Effect of feed temperature: Tbe increase in feed temperature causes considerable increase in the conversion of CR,, as it is clear from the comparison between Table 3 and Table 6. Holowever the increase of CO2 conversion is limited in comparison with that of CR, due to the fact that high temperatures favours the production of CO rather than COP. The high feed temperature cases shown in Table 6 correspond to feed conditions of negative reaction order with respect to steam. The implication of the non-monotonic kinetics of methane steam reforming is discussed in some details elsewhere (Elnashaie et al, 1988). It is noticed that in this region the effect of steam to methane ratio is more pronounced than in the positive effective order cases in Table 3 and therefore the conversion increases with the increase in steam to methane ratio despite the decrease in the residence time. Although the rate of reaction in these cases is much higher than that in Table 3. the effectiveness factor is larger. This is a manifestation that the effect of the increase in diffusion coefficients and equilibrium constants on effectiveness factor is stronger than the effect of the rate of reaction. Acknowledgement: The authors would like to their deep appreciation to Prof. G. express Froment for providing with them his more general kinetic rate equations, and the subroutine for the physical properties and for very stimulating dicussions in Riyadh and Gent during exchange of visits between the two departments. This work is supported by the King Abdulasiz City for Science and Technology grant no. AR-7-19.

1804

M. A. SOLIMAN et al.

A2

Notation &AC #AR

P a?.+ Cp*Cg Ui,e dc.dp

reformer tube cross sectional area. cold plane and refractory surface ar as L::1 re % ractory and tube side surface to furnace free volume half ratio heat capacity of process gas and of flue gas [~cal/~g..~g.~] effective center tubes to and diffusivity center catalyst of component i

x Y

g.%t

ractory slab and tube wall. And solid void fraction process gas and tube metal thermal conductivities viscosity of catalyst ,nCO bed the bulk reacting [Kcal/m.hr.K] gas mixture [Kg/m.hr] density and process of CHA [Kg/m31 dis-

oB.og nCH

gas density effectiveness 2

factor

distance particle

[m'/br] between diameter

dte,dti E1*E2 Ey.Et F,f

[m] external and internal diameters of the reformer tube [ml grey and clear gas component heat flux gas and tube emiesive fired furnaces) fraction reformer [Kcal/m2.hr] powers (in side [Kcal/t&.hr]

o.w

appearance rate and CO2 formation rate stefan-Boltsman constant and dimensionless coordinate of the catalyst pellet References

FH2*FCH4sFC0,

FT

H2, total

of fuel burned along the and friction factor molar flow rate of Fco; FH20S CH4, CO, CO2* H20 in feed[Kmol/hr] molar gas flow and rate mass ve [Kmol/hr] ocities t Kg/hr.m2] i flue of

.G&
AHi KlSK2

process enthalpy

change

reaction for

equilibrium II rate

constant of

[Kcal/mol] reaction I [bar23 > c-1 I, III

klsk3.
k2

coefficients

reactions

respectively rate coefficient adsorption H2

[Kmol.bar/Kgcat.hr] of reaction II [Kmol/K&cat.hr.bar] constants for CK4, CO

kCH4.kC0.kt12 and kH20

[bar-l] adsorption constant for

dissociative Ii20 reformer nate, and pellets number of reacting partial gas bulk

L.fi SEC
"i Fi.Fi.S

t-1
tube heated length, coordiof [ml the characteristic moles of of length

component component

i in

mixture pressure and in

[Kmol/hr] i at the [bar net

catalyst of rate exit of

particle flue heat gas,

QGQrQR

sensible

heat

heat release, by radiation R.Re r1*r2.r3 T.Tg,Ts T r.Tg.0 gas constant, of

transfer [Kcal/hr]

Reynold's I,

rates

reactions of the

number [~calj~m~l.~] II and III res-

pectively temperatures

[Kmol/Kgcat.hr] reacting mixture,

furnace gas and inner tube skin [K] refractory and fuel gas inlet temperature* [K] and and mass inner tube surface [K] [kg/m*.hr], catalyst velocity temperature

outer Ti,o.Tt,i.Tcot temperature Us.V superficial

free volume of reformer [m31 dimensionless distance along the tube, L/L a radiation distribution factor emissivity of the flue gas, refEg,Er.Et,Es

Al-Ubaid. A-S.; Elnashaie, S.S.E.H.; AbbaM.E.E., shar, "The Influence of the 1987, Support on the Effective Order of the Steam Reforming Reactions", Methane Conversion April 27-May 1, Auckland, New SYmP-. Zealand. Bodrov, N-M.; Apel'baum, L-0.; Temkin, M-I., "Kinetics of the Reaction of - Methane 1964, with Steam on the Surface of Nickel", Kinet. Catal., 5, 614. - DeUeken, J-C.; Uevos, E-F.; Froment, G-F., 1982, "Steam Reforming of Natural Gas". Chemical Reaction Engineering ACS Symposium Series 196 (BOSTON), 1982. Elnashaie, S.S.E.H., Al-Ubaid, A-S., Soliand Adris. man, M-A., A.M., "On the 1988. Kinetics and Reactor Modelling for the Steam Reforming of Methane - A Review", J. Ensw Sciences, King Saud Univ.. Vol. 14, 2. Elnashaie. S.S.E.H.; Abbashar, M.E.E.; AlUbaid, A-S., 1987a. "Simulation and Optimization of An Industrial Ammonia Reactor". submitted for publication 1987. Elnashale, S.S.E.H., Al-Ubaid, A-S.. Adris, A.M., and Soliman, M-A., 1987b, "Parametric Investigation of Industrial Steam Reformers in and Saudi Arabia", unpublished Egypt report. Elnashaie, S.S.E.H., Al-Ubaid, A-S., Adris, A.M., and Soliman, M-A.. "On the Non1988, monotonic Behaviour of the Methane steam Reforming Kinetics", submitted for publication. Filla. M., 1984, "An Improved Roesler-type flux method for radiative heat transfer in one-dimensional furnaces", Chem. Eng. Sci., 39, 1. 159-161. Froment. G., and Bischoff, K., 1979, "Chemical Reactor Analysis and Design", Wiley and Sons, 1979. Gaff. S-P.; Wang, S-1.. 1987, "Syngas Production by Reforming". Chem. Eng. Frog.. 38, 8. 46-53. Hicks, E-E., 1970, -Pressure drop in packed bed of spheres", IEC Fund.. K, 500. Hyman, M-H., 1968, "Simulate Methane Reformer Reactions", Hydr.Carb.Froc. 49. 131. Leva, M., Winstraub, M., Grummer, M., Pollchlk, M. and H.H. Starch. "Fluid Flow through Packed and Fluidised Systems". 1951, U.S. Bur. Mines, Bull, 504. "Chemical Reactor Design Rase, W-F.. 1977, for Process Plants", Vol. 2 John Wiley. Reid, R-C., and Sherwood, T-K., "The 1966, Properties of GSSSS Liquids", McGraw and Hill, N.Y. - Rostrup-Nielsen, J-R., 1984, "CatalyticSteam Reforming", In Catalysis Sci. 6 Tech., Vol, 5, p- 1, springer Verlag, Berlin.

A2

Simulation

of steam reformers

for methane

1805

"SimulaSin&h, C.P.P.; Saraf. D.N., 1979, tion of Side Fired Steam-Hydrocarbon Reformers", Ind. Eng. Chem. Proc. Des. Dev.. 18, 1, l-7. and Froment, G-F., "Methane - Ku. J., 1988, Methanation and Water-Gas steam Reforming, Shift on A Ni/MgA120,+ Spine1 Catalyst. IIntrinsic Kinetics, II. Reactor Simulation" (to be published, 1988). "Kinetic Study of Steam Refor- Xu, J., 1986, Methanation". Ph.D. Thesis, and ming RIJKSUNIVERSITEIT, GENT, Belgium.

Side fired furnace, case of Table 1 and Saraf (1979) Table 11. in Singh *Heated length of the reformer tubes = 12 meters, Inside diameter of reformer tube 0.0935 meter, Outside diameter of reformer tube = 0.1379 meter, Number of tubes = 200. Catalyst shaped and size - Raschiy rings (0.016*0.016/0.006 meter), Bulk density of catalyst = 1262.6 kg/m3 *Process gas (natural gas) Composition (volume - 81.5% Cl&. 7% C2hi6, 5.5% C3HB. dry basis) 4.5% C&H,O, 1.5% co2 *Fuel (natural gas) flow rate 7411.5 std. Oxygen volume percent in flue gas 3 &/hr, 2.5%. Number of burners = 584 .Inlet conditions to the catalyst tubes, Process gas flow rate - 13945.5 std m3/hr, Pressure = 34.4 atm. steam flowrate Process gas temperature 3401.515 kmole/hr. 727.4 OK. Methane equivalent at inlet = = 762.84 Kmole/hr Outlet conditions Variable Plant Calculated 2493.6
255.49

Top-fired reformer, Abou-Qir plant data *Total tube length = 12 m, heated length = 10 m, Tube inner diameter = 0.1 m, tube wall thickness = 0.015, Number of tubes = 280, Catalyst used (ICI 5713 nickel oxide) *Process gas (Natural gas) composition (volume dry basis), 95.3% CH,+, 2.8% C2H6, 0.91% CJHe, 0.07% n-butane. 0.045% I-butane, 0.63 CO2, 0.0086% I-pentane, 0.18% N2, and 0.016% hexane oFue1 used (Natural gas) flow rate 8700 m3/h, at temperature 343-K, Pressure 4 bar (same composition as process gas), Combustion air flow rate 11.7*104 m3/hr, at pressure 40 bar. temperature 583K *Number of burners = 120 forced draught burners arranged in the reformer ceiling in 6.7 rows eInlet conditions, Process gas flow rate temperature 723 K. pressure 36.5 26000 m3/b, bar, Steam flow rate 5278 Kmol/hr, temperature 723 K. Pressure 36.5 bar, Methane equivalent at inlet - 1196 Kmol/hr Outlet conditions Table 2 Plant Calculated 4076.72 371.322 1063 33.4

Steam flow rate Kmol/hr Methane flow rate Kmol/hr Process gas temperature, OK Process gas pressure, bar Process gas composition (X volume dry basis) CH4 co CO2
H2 N2

4062.37 364.96 1056 33.7

Steam flow rate Kmol/hr Methane flow rate Kmol/hr Process gas temperature, OK Process gas pressure, bar Process gas composition (X volume dry basis) C% co
co2

2453.3 253.9 1038.55 32.0 10.7 10.17 11.4 67.72

8.6 8.6 11.9 70.6 0.3

8.75 8.45 11.8 70.71 0.29

1042.3 32.02 10.75 9.95 11.45 67.85

H2

1806

M.

A. SOLIMAN et al.

A2

+(bae

3.8)

t-)1.68

+13.9

-4.9

-6.5

-6.3

-1.68

+x7.5

-5.3

-5.8

-5.6

-1.3

+2.02

442.7

+a.7

a-

act-1.0

-39.3

-44

i6.1

l .4

+5.2

-24.0

-27.5

t5.95

+5.*

6.3

Ml.9

+39.24

T6.6

4.m

_____
i

----__-_I

-----_-_---

____

--_-_-__-_____ t3.3 +7.4 +,.1

+2

--_--_-______ +9.2 -2.91 -54.9

Differ3s ur

kasis

-33.7

-36.6

+*.4

+a.,

da.3

-26.6

-30.2

incr-

l 9.2

*XL9

-3.3

-4.5

-4.2

+4.8

+53.,

-3.4

-3.95

-3.7

-1.05

+8.X!

r37.9

eel.4