Pd-based membrane steam reformers: A simulation study of reactor performance

ARTICLE IN PRESS I N T E R N AT I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y

33 (2008) 3036– 3040

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/he

Pd-based membrane steam reformers: A simulation study of reactor performance Marcello De Falco Chemical Engineering Department, University of Rome ‘‘La Sapienza’’, Via Eudossiana 18, 00184 Rome, Italy

ar t ic l e i n f o

abs tra ct

Article history:

Steam reforming of natural gas is the main process for the production of hydrogen and

Received 7 December 2007

synthesis gas needed in the chemical industry as reactants for the manufacturing of

Received in revised form

important products, mainly ammonia and methanol. Moreover, the development of fuel

27 February 2008

cell technology in many application fields (stationary and automotive) will increase in the

Accepted 4 March 2008

next years the need of industrial quantities of pure hydrogen. This increasing demand is stimulating the development of new technologies for

Keywords: Membrane reactor Hydrogen production Steam reforming

producing large amounts of hydrogen at low cost. The most promising process is based on the integration of ultra-selective membranes in steam reforming reactor, removing continuously hydrogen from reaction environment and preventing equilibrium conditions to be achieved. In this work the integrated membrane methane steam reformer performance is evaluated by a bi-dimensional non-isothermal non-isobaric model and a comparison with the traditional reformer varying the gas mixture residence time and the wall temperature is made. Results obtained attest the remarkable improvement in the steam reforming reactor by integrating the selective membrane: if a long residence time is imposed (50 kgcat s/mol), the methane conversion is more than doubled in respect to the traditional technology. On the other hand, the increase of the wall temperature improves the performance of both the traditional and the membrane reactor. However, selective membranes have a stringent technological threshold (To823 K) which limits the thermal level accepted inside the reactor. & 2008 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights reserved.

1.

which taken together, yield:

Introduction

Methane steam reforming (MSR) is the most important process to produce large amounts of hydrogen for uses in chemical industry or as a fuel. Hydrogen production by this process is based on the following two reactions: CH4 þ H2 O 3 CO þ 3H2

(1)

CO þ H2 O 3 CO2 þ H2

(2)

CH4 þ 2H2 O 3 O2 þ 4H2

(3)

MSR reactions are very fast over Ni-based catalyst, so that equilibrium conditions are quickly reached; however, a significant hydrogen yield is achieved only at high temperatures (850–900 1C), therefore MSR requires a high heat flux between the oven and the reactor and heat transfer plays a leading role in the behaviour of the reactors.

Fax: +39 6 4827453.

E-mail address: [email protected] 0360-3199/$ - see front matter & 2008 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.ijhydene.2008.03.006

ARTICLE IN PRESS I N T E R N AT I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y

The integration of selective membranes in the reaction environment appears to be a promising way to enhance hydrogen yield at lower temperatures, because the selective removal of hydrogen avoids the equilibrium conditions to be achieved and enables greater conversions to be attained. The main benefits in using membrane reformers are:

 A strong reduction of the reaction temperature (from 900

  

to 500 1C). A heat exchanger configuration in place of the furnace can be used. Process efficiency increase. Large methane saving. A less amount of methane has to be burned to supply the process heat duty. More compact process.

Recently several papers have appeared in the literature devoted to the study of the performance of membrane reactors (MR) in comparison with traditional reactors (TR). The Pd-based membranes are usually used, thanks to their very high selectivity towards hydrogen (near 100%). At present, MR technology is not yet applied at industrial scale to steam reforming: only Tokyo Gas has developed a large-scale MR equipment but operating conditions and performance of the device are not published; therefore, problems and advantages of this process cannot be deduced from industrial skills and its feasibility can be assessed only by theoretical simulations. The studies reported in the literature are based on models developed for analysing the experimental data collected in lab-scale equipments but this approach does not allow important effects to be shown since turbulent conditions, typical of industrial applications, are not imposed [1]. Moreover, usually the models proposed are isothermal, and the crucial effects of the temperature profile along the reactor are not considered, or one-dimensional. The aim of this work is the simulation study of the steam reforming MR performance and the comparison with the traditional technology by means of a two-dimensional nonisothermal model based on mass, energy and momentum balances. Operating conditions and geometrical parameters typical for large-scale applications are imposed in order to evaluate the applicability of membrane reformers in the industrial field.

2.

the external annular section a heating fluid is sent to supply the heat duty required by the reactions, in the internal annular section (the reaction zone) the catalytic pellets are packed, in the internal tube (the permeation zone) a sweeping gas is sent to carry out the hydrogen permeated through the membrane. The internal tube is the membrane itself. The reactions considered are (1)–(3), while secondary reactions are neglected. The model is based on mass, energy and momentum balances. The two-dimensional nature of the model allows axial and radial concentrations and temperature profiles to be evaluated and therefore gives much more information about the reactor behaviour than one-dimensional models. Moreover, the author has demonstrated in a previous work [1] that if a one-dimensional model is applied, an overestimation of the hydrogen permeation flux and, consequently, methane and carbon dioxide conversion arises. The size of overestimation is slight but not unimportant (4.6% in terms of hydrogen produced and recovered).

 Mass balances Reaction zone: qðuz  ci Þ dP  L q2 ðuz  ci Þ 1 qðuz  ci Þ ¼  þ  2 qz~ qr~ r~ Pemr  ri;o qr~2  rb  L 

Nm dY H2 H  2p  ro;i ¼ 2 0 dz~ FCH

Fig. 1 – Membrane reactor configuration.

(4)

(5)

4

where Y H2 is the ratio between permeated hydrogen and inlet methane flow rate (F0CH ), ro,i is the inner tube radius and 4 Nm H2 is the hydrogen flux permeating through the membrane, which is calculated by the well-known Sieverts law: Nm H ¼ 2

Reactant mixture

Zj  Rj

where uz is the gas mixture velocity, ci is the mole concentration of component i (i ¼ CH4, H2O, H2 , CO, CO2), z~ and r~ are dimensionless axial and radial coordinates, dp is the particle diameter, L and ri,o are the reactor length and the catalytic bed tube radius, respectively, rb is the packed bed density, Zj and Rj are the effectiveness factor and the intrinsic rate of reaction j [2] and Pemr is the mass effective radial Peclet number, given by [3] for Reynolds number greater than 1000. Permeation zone:

The reactor configuration considered in the following simulations is shown in Fig. 1 and is composed by a triple tube: in

Sweeping flow

3 X

!

j¼1

Mathematical model

Heating fluid

3037

33 (2008) 3036 – 3040



BH 0:5  ðp0:5 H2 ;reac  pH2 ;perm Þ d

(6)

In Eq. (6) BH is the hydrogen permeability, calculated according [4] for Pd–Ag membranes, d is the membrane thickness, pH ;reac and pH ;perm are the hydrogen partial 2 2 pressures in reaction and permeation zone, respectively. The sign  in Eq. (5) means that a counter-current sweeping gas configuration is adopted. Energy balances Reaction zone: ! qTR ler  L q2 TR 1 qTR ¼  þ  qz~ r~ qr~ ðuz  ctot Þ  cp;m  r2i;o qr~ 2 P3 rb  L  i¼1 Zj  ðDHj Þ  Rj (7) þ ðuz  ctot Þ  cp;m

ARTICLE IN PRESS 3038

I N T E R N AT I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y

where ler is the effective radial thermal conductivity of packed bed and gas mixture, considered as a pseudohomogeneous phase; ler is calculated according to [5]. Permeation zone: dTP L ¼   ½U1  2p  ri;i  ðTR  TP Þ FP;tot  cp;perm dz~ þ Nm H  p  ro;i  ðhR;H2  hP;H2 Þ 2



(8)

where U1 is the overall heat transfer coefficient between packed bed and permeation zone. Momentum balance Reaction zone: dP f  G  mg  L ð1  Þ2 ¼  dz~ 3 rg  d2p

(9)

The friction factor f is evaluated by the well-known Ergun equation.

2.1.

Boundary conditions

Boundary conditions of Eqs. (4), (5), (7)–(9) are z~ ¼ 0;

8r~ : in uz  ci ¼ uin z  ci

TR ¼ Tin R PR ¼ Pin R z~ ¼ 1;

(10)

8r~ :

3.

33 (2008) 3036 – 3040

Results and discussion

In the following simulations the operating conditions reported in Table 1 are imposed. Fsweep is the flow rate of the sweeping water vapour sent in the permeation zone: its values are taken equal to those ones of the methane inlet stream. The term S/C is the steam to carbon ratio in the inlet section.

3.1.

The effect of the residence time

First of all, the effect of the gas mixture residence time, i.e. the ratio between catalyst weight and the inlet gas flow rate, is evaluated. In Fig. 2 methane conversions are reported in TR and MR at different residence times and imposing a wall temperature equal to 873 K: it has to be noticed that an increasing residence time has a strong positive effect on the MR performance while the TR is slightly influenced. In fact, the equilibrium conditions threshold is achieved at the reactor outlet in the TR and the slight conversion increase is a consequence of the slight temperature increase of the reactant mixture due to the longer residence time. In the MR the equilibrium conditions are not achieved since the hydrogen is continuously removed and when the residence time is longer the hydrogen permeated is larger, supporting the reaction.

Y H2 ¼ 0 TP ¼ Tin P

(11) Table 1 – Operating conditions and geometric features

r ¼ ri;o ;

8z~ : qðuz  ci Þ ¼0 qr~ qTR ¼ qr ¼ U  ðTw;o  TRjri;o Þ ler  qr~

r ¼ ro;i ;

Tin R (12)

L 12 m

8z~ : qðuz  ci Þ ¼0 qr~ qðu dp z  cH2 Þ ¼ Nm  H2 qr~ Pemr qTR ¼ U1  ðTRjro;i  TP Þ ler  qr~

773 K

Tin P

Pin R

PP

Fsweep

S/C

773 K

10 bar

1 bar

F0CH

3

ri;o

ro;i

e

dp

0.065 m

0.04 m

rb 1016.4 kgcat/m3

0.5

0.011 m

4

(13)

In Eq. (12) the term U is the overall heat transfer coefficient between tubular reactor wall and the packed bed:   1 1 1 þ (14) U¼ amet hW where amet is the thermal conductivity of the steel tube and hW represents the heat transport coefficient of the layer of (gas mixture+packed bed) phase in contact with the tube wall; this layer is considered an ‘‘unmixed zone’’ where heat transport occurs only by molecular mechanism [6]. Moreover, the term qr is the heat flux supplied by the external energy source and Tw,o is the wall temperature, i.e. the temperature at the external surface of the steel tube: in the following simulations the value of Tw,o is an input parameter and, for the sake of simplicity, it is assumed to be constant along the axial reactor coordinate. The qr profile is calculated a posteriori.

Fig. 2 – Methane conversion vs. W/Ftot at Tw,o ¼ 873 K.

ARTICLE IN PRESS I N T E R N AT I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y

In Fig. 3, the effect of the residence time on the percentage of hydrogen recovered, i.e. the ratio between the amount of hydrogen recovered and carried out in the permeation zone and the total hydrogen produced, is shown. It is worth noting that when the gas mixture passes quickly over the catalyst, the permeation through the membrane has a short time to occur and less that a half of the total hydrogen produced is removed. When the residence time is longer, the percentage of hydrogen recovered is much more, up to 85%. Moreover, a longer residence time has a positive effect on the maximum membrane temperature as well. It is known that the Pd-based membranes have a stringent temperature threshold due mainly to loss of adherence between the Pd layer and the support (ceramic or stainless steel): therefore this type of membranes must work at temperature o823 K about, otherwise selectivity properties would fall quickly. At a residence time equal to 10 kgcat/(mol/s), the maximum membrane temperature reached along the axial coordinate is 828.4 vs. 807 K at 50 kgcat/(mol/s). It has to be considered that imposing longer residence time has opposite effects on the temperature inside the reactor:

 the longer residence time leads to a slower feed rate with the consequence that the gas mixture temperature should approach the wall temperature.

33 (2008) 3036 – 3040

3039

quently of the membrane temperature, prevail and the membrane is maintained at a lower thermal level imposing a longer residence time. Therefore, the benefits of the membrane integration in the reformer are more evident when the residence time is long. On the other hand, a long residence time brings about a reduction of the single reformer hydrogen production. In Fig. 4, the amount of hydrogen produced per kg of catalyst at various residence times is depicted. The productivity of unit catalyst mass is reduced by increasing the residence time. Therefore an increase of the size of the reformer is required to obtain the same amount of hydrogen. Consequently, a larger membrane surface is requested at longer residence times: if W/Ftot is equal to 10 kgcat/(mol/s), the hydrogen produced per m2 of membrane is about 120 Nm3/h, while at W/Ftot ¼ 50 kgcat/(mol/s) it is about 40 Nm3/h. In conclusion, in a MR the methane conversion is higher and the hydrogen recovery is promoted at longer residence times, but at the same time the membrane reformer is bulky and more expensive. An economic assessment should be performed in order to evaluate the optimal residence time in terms of reactor performance (i.e. methane required per mole of hydrogen produced) and size.

3.2.

The effect of the wall temperature

On the other hand: the higher methane conversion leads to a reduction of the packed bed temperature thanks to the high endothermicity of the reactions. The slower gas velocity provokes a decrease of the gas Reynolds number and of the heat transport coefficients between the reformer wall and the packed bed and inside the packed bed itself. It is a worth assessment that the effects leading to the reduction of the packed bed temperature, and conse-

In Fig. 5, methane conversions in the TR and MR are reported at different wall temperatures. Increasing Tw,o has a positive effect for both the configurations thanks to the greater heat flux supplied at the reactor. However the membrane temperature threshold is reached at wall temperature 4890 K: therefore, although the methane conversion obtained is higher at a greater wall temperature, the current technological limits prevent to operate at too severe operating conditions.

Fig. 3 – H2 recovered to H2 produced ratio vs. W/Ftot at Tw,o ¼ 873 K.

Fig. 4 – Nm3 of hydrogen produced per unit mass of catalyst at different residence times.







ARTICLE IN PRESS 3040

I N T E R N AT I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y

33 (2008) 3036 – 3040

In conclusion, the advantages in integrating the membrane in the reactor are evident but a R&D effort towards an increase of hydrogen permeability and on upgrading the thermal stability of the membranes is required.

Acknowledgement The author is grateful to Prof. L. Marrelli for his useful suggestions. R E F E R E N C E S

Fig. 5 – Methane conversion vs. Tw,o at W/Ftot ¼ 30 kgcat/(mol/s).

4.

Conclusions

The benefits in integrating a selective Pd-based membrane in the reformer reaction environment have been evaluated in comparison with the traditional technology. The results show that the MR best performance is obtained at long residence time, which; however, brings about an increase of reactor volume and membrane surface. The wall temperature increase has a positive effect as well, although too high temperatures cause the overtaking of membrane technological threshold.

[1] De Falco M, Di Paola L, Marrelli L, Nardella P. Simulation of large-scale membrane reformers by a two-dimensional model. Chem Eng J 2007;128:115–25. [2] Xu J, Froment G. Methane steam reforming, methanation and water–gas shift: I. Intrinsic kinetics. AIChE J 1989;35(1):88–96. [3] Kulkarni BD, Doraiswamy L. Estimation of effective transport properties in packed bed reactors. Catal Rev Sci Eng 1980;22(3):431–83. [4] Shu J, Grandjean BPA, Kaliaguine S. Methane steam reforming in asymmetric Pd- and Pd-Ag/porous SS membrane reactors. Appl Catal A Gen 1994;119:305–25. [5] Elnashaie S, Elshishini S. Modelling, simulation and optimization of industrial fixed bed catalytic reactors. Topics in chemical engineering, vol. 7. London: Gordon and Breach; 1993. [6] Tsotsas E, Schlu¨nder E. Heat transfer in packed beds with fluid flow: remarks on the meaning and the calculation of a heat transfer coefficient at the wall. Chem Eng Sci 1990;45:819–37.