- Email: [email protected]
Kinetic studies of the partial oxidation of isooctane for hydrogen production over a nickel–alumina catalyst
Chemical Engineering Science 61 (2006) 5912 – 5918 www.elsevier.com/locate/ces
Shorter Communication
Kinetic studies of the partial oxidation of isooctane for hydrogen production over a nickel–alumina catalyst Hussam H. Ibrahim, Raphael O. Idem ∗ Process Systems Engineering Laboratory, Faculty of Engineering, University of Regina, 3737 Wascana Parkway, Regina, Canada S4S 0A2 Received 25 January 2006; received in revised form 7 April 2006; accepted 7 April 2006 Available online 28 April 2006
Abstract The kinetics of the catalytic partial oxidation of isooctane for hydrogen (H2 ) production over a stable Ni/-Al2 O3 catalyst was investigated o ) in the at atmospheric pressure in the temperature range of 863–913 K, ratio of weight of catalyst to the molar feed rate of isooctane (W/FiC8 −1 range of 7.09.30.89 kg mol , and molar feed ratio O2 /i-C8 H18 of 4.0 in a 12.7 mm diameter Inconel micro-reactor housed in an electrically controlled furnace. The developed rate models were based on the Langmuir–Hinshelwood–Hougen–Watson (LHHW) and Eley–Rideal (ER) formulations. Out of the 18 models developed, 10 were eliminated due to poor predictive efficiency. A LHHW mechanism requiring the dissociative adsorption of isooctane and molecular adsorption of oxygen on a single site was the most likely pathway for the partial oxidation of isooctane. The reaction order of 1.5 indicates a strong coverage of nickel by isooctane. 䉷 2006 Elsevier Ltd. All rights reserved. Keywords: Isooctane; Gasoline; Partial oxidation; Nickel-based catalyst; Kinetics
1. Introduction The use of catalytic converters to control vehicle exhaust emissions is well known (Heck and Farrauto, 2002). However, increasingly stringent legislation has led to consideration of alternative means of reducing emissions. Of these measures, the use of fuel cells appears to be very promising and has seen remarkable progress in the past decade because of an increasing need to improve energy efficiency as well as to address environmental concerns (Krumpelt et al., 2003). The polymer electrolyte membrane fuel cell (PEMFC) is a potential candidate for direct electricity production from hydrogen for transportation applications and also for distributed and portable power generation. Also, hydrogen is projected to be one of the primary energy sources in the 21st century (Sabacchi et al., 2002). Its combustion and oxidation is pollution free and supplies sufficient energy for transportation and other applications. However, the absence of a viable hydrogen storage option and a hydrogen marketing infrastructure, at least in the near term, ∗ Corresponding author. Tel.: +1 306 585 4470; fax: +1 306 585 4855.
E-mail address: [email protected] (R.O. Idem). 0009-2509/$ - see front matter 䉷 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.ces.2006.04.015
necessitate the search for an appropriate fuel (Ahmed et al., 1999). Petroleum-derived fuels such as gasoline are attractive sources of hydrogen because of their existing production, distribution, and retailing infrastructure. Also, gasoline has a much higher energy density and larger hydrogen content compared to methanol and ethanol (Zhu et al., 2004; Moon et al., 2004), which are also considered as good sources for hydrogen. The conversion of hydrocarbons to hydrogen can be carried out using one of three major techniques: steam reforming, partial oxidation and autothermal reforming. The major by-product CO2 can be captured and sequestered prior to hydrogen use. Partial oxidation is chosen for this study for the reasons that: (i) the reaction is exothermic, making it much more energy-efficient than steam reforming; (e.g. C8 H18 + 4(O2 + 3.76N2 )8CO + 9H2 + 15.04N2 ; o =−6.60×105 J/mol, (ii) a smaller reformer can be used H298 to achieve a high conversion of the hydrocarbon selectively in favor of the production of H2 at short contact times; and (iii) the partial oxidation setup is more compact and mechanically simpler than the steam reforming, since no additional heating is required (Li et al., 2000). The development of an efficient partial oxidation reformer for the production of hydrogen from
H.H. Ibrahim, R.O. Idem / Chemical Engineering Science 61 (2006) 5912 – 5918
liquid hydrocarbon fuels such as gasoline is very challenging. It requires the development of a stable catalyst for the process. It also requires detailed thermodynamic and kinetic information about the reaction and such information is presently scant (Springmann et al., 2002). In aprevious study (Ibrahim et al., 2006), we obtained a stable performance for a Ni/-Al2 O3 catalyst for the production of hydrogen by the partial oxidation of isooctane. Thus, the objective of the present study is to develop a mechanistic based kinetic model to describe the catalytic partial oxidation of isooctane to produce hydrogen using this stable Ni/Al2 O3 catalyst. The developed models are based essentially on the LHHW and ER hypotheses and these were cross-examined with actual experimental kinetic data. The results are presented in this communication. 2. Theory Initially, an empirical, irreversible fixed feed molar ratio power law rate model was developed as shown in Eq. (1). rA = k0 e(−E/RT ) CAm .
3.1.1. Experimental apparatus and analysis An Inconel fixed bed reactor (i.d. = 12.7 mm) housed in a furnace with a single heating zone was used for catalyst performance evaluation. Liquid isooctane was introduced by a syringe pump (kd Scientific-200) while the gas flows were metered and regulated by an Aalborg digital flow controller (DFC-26). The air to isooctane molar ratio was 12.0. The catalyst bed temperature was measured by means of a sliding thermocouple dipped inside the catalyst bed. The diluent used in the catalyst bed was -Al2 O3 having the same particle size as the catalyst (150 m). Pure -Al2 O3 (150 m) was used in the preheating zone and the section after the catalyst bed. The total catalyst bed height was 70 mm. The exit product from the reactor was separated into permanent gases and liquid condensate by passing first through a condenser and then a gas/liquid separator. The gases were analyzed with an on-line gas chromatograph (HP-6890, Agilent Technologies) equipped with a TCD using Haysep and Molsieve columns (Alltech Associates) for complete gas product separation. The unreacted isooctane was analyzed with a GC-MS (HP-6890/5073, HP) using a 30 m GS-GasPro column (J&W Scientific).
(1)
Then, different mechanistic models based on LHHW and ER mechanisms were proposed. The overall reaction considered in the development of all models is given in Eq. (2). C8 H18 + 4O2 ←→ 8CO + 9H2
5913
o H298 = −660 kJ mol−1 .
(2) The discrimination between rival models was based on predictive efficiency (i.e., the average absolute deviation (AAD%) between predicted and experimental results) and the similarity in activation energy values between the model and the power law model. Other supporting criteria such as residue analysis and chemical common sense were also used where appropriate. The development of these two classical mechanisms was based on the assumptions that: (1) the rate-determining step (RDS) could be controlled by the surface reaction step, adsorption step or desorption step; (2) there is the presence of uniformly energetic adsorption sites; and (3) there is monolayer coverage. The only differing assumption between LHHW and ER mechanisms was the use of a dual site mechanism for the reaction in the former case and a single site mechanism in the latter case. Sixteen LHHW and two ER models were proposed and tested. All the rate models, and the assumptions used in their derivation are given in Table 1(a–c). 3. Experimental 3.1. Catalyst selection The catalyst used in this work was a 3% Ni/-Al2 O3 catalyst prepared by the precipitation method. This catalyst showed good activity and stability in terms of isooctane conversion and hydrogen selectivity. Details of preparation, characterization and performance evaluation of this catalyst are given elsewhere (Ibrahim et al., 2006).
4. Kinetic studies Experimental kinetic data were collected at atmospheric pressure; temperatures of 863, 883, 903 and 913 K; and o of 255.6, 198.0, 162.0 and 136.8 kg-cat s kg-iC−1 . In W/FiC8 8 order to approach plug flow conditions, and minimize backmixing and channeling, certain operating criteria as prescribed by Froment and Bischoff (Froment and Bischoff, 1990) were used. These were that the ratio of catalyst bed length to catalyst particle diameter (L/Dp ) was 350 and the ratio of the inside diameter of reactor to particle diameter (D/Dp ) was 63.5. Catalyst particle size of 150 m and feed flow rate of 390 mL min−1 were used to eliminate pore and external diffusional limitations, respectively. The pressure drop along the reactor was calculated for the particle size used and found to be insignificant (< 5 × 103 Pa) at the highest reaction temperature. 5. Results and discussion 5.1. Heat and mass transport limitations It is well known that intrinsic kinetic data can only be obtained in the absence of heat and mass transport resistances. Theoretical criteria were used to determine whether there were any effects of interparticle and intraparticle heat and mass transport limitations on the rate of reaction at 913 K. 5.1.1. Heat transport effects Internal pore heat transfer resistance was estimated using the Prater analysis given by Tmax,particle =
Deff (CAs − CAc )(−Hr ) , keff
(3)
5914
H.H. Ibrahim, R.O. Idem / Chemical Engineering Science 61 (2006) 5912 – 5918
Table 1 Rate models based on LHHW and ER mechanisms Model # / type
Assumptions
Rate model
Eq. #
(a) M2/LHHW
Dissociative adsorption C8 H18 (2C4 ) and O2
of
both
k0 e−(E/RT ) CA
rA = 1 + KA
(CA0 − CA )17/2 2 CA
C8 H18 dissociative adsorption assumed to be the RDS M3/LHHW
Dissociative adsorption C8 H18 (2C4 ) and O2
of
both
4 k0 e−(E/RT ) CA
rA =
1 + (KA CA )1/2 +
(CA0 − CA )11/2
Dissociative adsorption C8 H18 (2C4 ) and O2
of
both
Dissociative adsorption C8 H18 (2C4 ) and O2
1 + KA
of
both
8
4
k0 e−(E/RT ) CA
rA =
Desorption of H2 and CO C8 H18 dissociative adsorption assumed to be the RDS M5/LHHW
3
(KB CA )1/2
O2 dissociative adsorption assumed to be the RDS M4/LHHW
2
+ (KB CA )1/2
(CA0 − CA )17/2 2 CA
5 k0 e−(E/RT ) CA
rA =
2
5
+ (KB CA )1/2 + KC (CA0 − CA )
17
6
1 + (KA CA )1/2 + KB (CA0 − CA )
Desorption of H2 and CO Surface reaction assumed to be the RDS 5 CA
k0 e−(E/RT )
M6/LHHW
Dissociative adsorption C8 H18 (2C4 ) and O2
of
both
(CA0 − CA )9 8 1/2 5/8 1 + KA CA + KB CA (CA0 − CA )−9/8 + KC (CA0 − CA )
rA =
Desorption of H2 and CO CO desorption assumed to be the RDS M7/LHHW
7
5 k0 e−(E/RT ) CA
Non-dissociative adsorption of both C8 H18 and O2 Desorption ofH2 and CO Surface reaction assumed to be the RDS
rA =
Dissociative adsorption C8 H18 (C5 , C3 ) and O2
rA = ⎛
8
[1 + KA CA + KB CA0 ]17
(b) M8/LHHW
of
both
⎜1 + ⎝
Dissociative adsorption C8 H18 (C5 , C3 ) and O2
of
both
rA = ⎛
Dissociative adsorption C8 H18 (C5 , C3 ) and O2
5/2
of
both
Desorption of H2 and CO Surface reaction (C3 + O) assumed to be the RDS
+
KB (CA0 − CA )6 3/2
5/2
KB (CA0 − CA )12 3/2
CA +KD (CA0 − CA )7
rA = ⎛
⎞2
9
⎟ ⎠
5 k0 e−(E/RT ) CA
⎜ K A CA + ⎝
Desorption of H2 and CO Surface reaction (C5 + O) assumed to be the RDS M10/LHHW
KA (CA0 − CA )11
CA CA +(KC CA )1/2 + KD (CA0 − CA )
Desorption of H2 and CO C8 H18 dissociative adsorption assumed to be the RDS M9/LHHW
k0 e−(E/RT ) CA
⎞11 + (1 + (KC CA )1/2 )(CA0 − CA )6 ⎟ ⎠
5 k0 e−(E/RT ) CA
⎞6
KB (CA0 − CA )22 7/2 + (1 + (KC CA )1/2 )(CA0 − CA )11 ⎟ ⎜ K A CA + 5/2 ⎝ ⎠ CA +KD (CA0 − CA )12
10
11
H.H. Ibrahim, R.O. Idem / Chemical Engineering Science 61 (2006) 5912 – 5918
5915
Table 1(Contd.) Model # / type
Assumptions
Rate model
Dissociative adsorption of C8 H18 (2C4 ) with O2 in the gas phase
rA =
Eq. #
(c) M11/ER
5 k0 e−(E/RT ) CA
17
12
1 + (KA CA )1/2 + KB (CA0 − CA )
Desorption of H2 and CO Surface reaction assumed to be the RDS M12/LHHW
Dissociative adsorption of both C8 H18 O2 in the gas phase
k0 e−(E/RT ) CA 2 1 + (KA CA )1/2
rA =
13
Surface reaction assumed to be the RDS 2 k0 e−(E/RT ) CA
M13/LHHW
Molecular adsorption of both C8 H18 O2 in the gas phase Surface reaction assumed to be the RDS
rA =
M14/LHHW
Dissociative adsorption of C8 H18 and molecular O2 in the gas phase
rA =
14
[1 + KA CA ]2
k0 e−(E/RT ) CA
3/2
1 + (KA CA )1/2 + KB CA
2
15
Surface reaction assumed to be the RDS k0 e−(E/RT ) CA rA = 1 + (KA CA )1/2 3/2
M15/ER
Dissociative adsorption of C8 H18 with O2 in the gas phase Surface reaction assumed to be the RDS
M16/ER
Non-dissociative adsorption of C8 H18 with O2 in the gas phase Surface reaction assumed to be the RDS
rA =
M17/LHHW
Dual site dissociative adsorption of both C8 H18 and O2 Surface reaction assumed to be the RDS
rA =
M18/LHHW
Dual site non-dissociative adsorption of both C8 H18 and O2 Surface reaction assumed to be the RDS
rA =
2 k0 e−(E/RT ) CA
17
[1 + KA CA ]
k0 e−(E/RT ) CA √ √ (1 + KA CA ) + (1 + KB CA )
where Tmax,particle is the upper limit to temperature variation between pellet center and its surface, Hr is the heat of reaction, and CAs & CAc are the concentrations at the pellet surface and center, respectively, Deff is the effective mass diffusivity obtained from Deff = (DAB /) (Fogler, 1999) where DAB is the bulk diffusivity of component A in B (i.e., isooctane in air), which in turn, is estimated using Fuller–Schettler–Giddings equation (Perry and Green, 1997). The value for DAB was obtained at the maximum temperature of 913 K as 0.466 cm2 /s. The effective diffusivity Deff was estimated to be 0.583 cm2 /s. is the void fraction (estimated as the ratio of the volume occupied by voids to the total bed volume = 0.5), is the tortuosity factor taken as 0.4 (Fogler, 1999), keff is the effective thermal conductivity obtained using the correlation keff /kc = 5.5 + 0.05NRe (Walas, 1990) for packed bed tubular reactors. kc is the mass transfer coefficient calculated by equating two J factors −0.4069 given by the following correlations JD = (0.4548/)NRe 2/3 and JD =(kc/)NSc (Geankoplis, 2003) where is the superficial velocity and NSc is the Schmidt number; NSc =(/DAB ), NRe is the Reynolds number; NRe = dp /(1 − ) where dp is the particle diameter. The effective thermal conductivity keff was calculated to be 1.19 × 104 kJ/(hr m K). A value of 0.03 K was obtained for Tmax,particle which shows that the pellet more or less had a uniform temperature.
16
2 k0 e−(E/RT ) CA
[(1 + KA CA ) + (1 + KB CA )]
18
19
The heat transfer limitations across the gas film was determined using Eq. (4) Tmax,film =
L(−rA,obs )(−Hr ) , h
(4)
where Tmax,film is the upper limit to temperature difference between the gas bulk and the pellet surface, L is the characteristic length, rA,obs is the observed rate of reaction, h is heat transfer coefficient (estimated from the correlation JH = 2/3 JD = (h/Cp )NP r where JH is the heat transfer J factor, NP r is Prandtl number; NP r = Cp /, Cp is the heat capacity, is the thermal conductivity). The heat transfer coefficient h was determined to be 4.77 kJ/(m2 K s). A value of 0.02 K was obtained for Tmax,film . These results confirm the absence of heat transfer limitations externally and internally. These led us to assume isothermal operation conditions during the reaction (Levenspiel, 1999). 5.1.2. Mass transport effects The internal pore mass transfer resistance was calculated using Weisz–Prater criterion as given by Eq. (5). Cwp,ipd =
−rA,obs c Rc2 , Deff CAs
(5)
5916
H.H. Ibrahim, R.O. Idem / Chemical Engineering Science 61 (2006) 5912 – 5918
where Cwp,ipd the Weisz–Prater criterion for internal pore diffusion, c the pellet density, Rc catalyst radius. The estimated value for Cwp,ipd was 2.31×10−4 . This value is much less than 1. Thus, this result indicates that the concentration on the catalyst surface is more or less the same as the concentration within
its pores. This result comes as a consequence of the absence of internal pore diffusion limitations (Fogler, 1999). To determine whether film mass transfer resistance has any effect on the rate of reaction, the ratio of observed rate to the rate if film resistance controls was examined. Eq. (6) illustrates this criterion. observed rate −rA,obs dp = , rate if film resistance controls CAb kc 6
45 T=590 isooctane conversion (mol%)
40
where dp is the catalyst particle diameter. The estimated value for the ratio in Eq. (6) was 3.435 × 10−15 . The result indicates that the observed rate is very much lower than the limiting film mass transfer rate. Thus, the resistance to film mass transfer certainly should not influence the rate of reaction (Levenspiel, 1999).
T=610 T=630
35
(6)
T=640
30 25
5.2. Catalyst evaluation and experimental rates 20
The catalysts were evaluated in terms of isooctane conversion. Material balance calculations based on carbon balance were performed for a selected number of runs. The overall recovery was > 93% in all cases. Fig. 1 gives the relationship o ratio for temperabetween isooctane conversion and W/FiC8 tures in the range of 863–913 K. The experimental rates of reaction were obtained using the differential method of analysis o data. Thus, the experimental rates for the conversions-W/FiC8 of reaction of isooctane were taken as the slopes or tangents
15 10 0.030
0.038
0.046
0.054
0.062
0.070
0.078
W/FA° (kg-cat kg-iC8-1 h) Fig. 1. Isooctane conversion as a function of contact-time at 863, 883, 903 and 913 K.
Table 2 Estimate of the values of the parameters of the models Parameter
Power law
M1
M2
M3
M4
M5
M6
M7
M8
k0 (kg-cat −1 s−1 )
4.33 × 1018
No convergence
No convergence
1.23 × 1023
No convergence
No convergence
No convergence
1.25 × 1023
E (J mol−1 ) m KA KB KC KD
No convergence
2.99 × 105 0.82 – – – –
3.43 × 105 – 5.85 × 10−4 1.51 × 107 4.37 × 107 –
3.48 × 105 – 5.85 × 10−4 2.51 × 107 7.17 × 106 1.96 × 107
Parameter
M9
M10
M11
M12
M13
M14
M15
M16
k0 (kg-cat −1 s−1 ) E (J mol−1 ) M KA KB KC KD
No convergence
No convergence
No convergence
5.24 × 1020 3.04 × 105 – 2.51 × 107 – – –
No convergence
1.25 × 1023 2.81 × 105 – 5.85 × 10−4 2.51 × 107 – –
1.25 × 1023 2.87 × 105 – 1.59 × 10−24 – – –
1.25 × 1023 2.08 × 105 – 2.94 × 108 – – –
Parameter
M17
M18
k0 (kg-cat −1 s−1 ) E (J mol−1 ) M KA KB KC KD
1.25 × 1023 3.40 × 105 – 2.94 × 108 2.51 × 107 – –
1.25 × 1023 2.02 × 105 – 2.39 × 108 7.44 × 107 – –
H.H. Ibrahim, R.O. Idem / Chemical Engineering Science 61 (2006) 5912 – 5918
2.0E-05
Also, Table 2 shows that M14 had an activation energy of 2.81 × 105 J mol−1 , which is very close to the value obtained for the power law model (2.99 × 105 J mol−1 ) implying no change in the mechanism between M14 and the power law model.
Rpre_PLM [AAD%=6.75]
Predicted rate (kmol kg-1 s-1)
Rpre_M4 [AAD%=9.80]
1.6E-05
5917
Rpre_M12 [AAD%=9.27] Rpre_M14 [AAD%=6.38] Rpre_M15 [[AAD%=1022]
1.2E-05
6. Conclusions 8.0E-06
4.0E-06
0.0E+00 0.0E+00 4.0E-06
8.0E-06
1.2E-05
1.6E-05
2.0E-05
Observed rate (kmol kg-1 s-1) Fig. 2. Parity plot for rates of partial oxidation of isooctane—models vs. experimental rates.
to the curves in Fig. 1 at desired W/FA0 values for the reaction temperatures. To verify that the dominant reaction step is the same at high and low temperatures, the data in Fig. 1 were further examined by treating the two regions separately. The results for the high temperature region (903–913 and 883–913 K) and the low temperature region (863–883 and 863–903 K) gave similar activation energies which were also identical to the global value (i.e., Ea = 2.8 × 105 J mol−1 ) obtained over the entire temperature domain. This result indicates that there is no change in the controlling mechanism for the two temperature regions. 5.3. Estimation of the values of the parameters of the kinetic models The estimation of the values of the parameters was based on the minimization algorithm, which essentially involves a combination of Gauss–Newton and Levenberg–Marquardt methods, and employed a non-linear regression software, NLREG. Table 2 gives the estimates of the values of the parameters for the partial oxidation of isooctane for all the final eight models in addition to the power law model. 5.4. Model validation The validation of the developed models was done through the measurement of the percentage absolute average deviation (AAD%) for each model from the actual experimental observations. Out of the models that converged, those with AAD of less than 15% were deemed acceptable and were short-listed. Fig. 2 depicts how well the models fitted the experimental data. The empirical power law model showed a good fit for the experimental (AAD = 6.75%). Among the mechanistic models, the LHHW model (M14) exhibited the best data fit with an AAD of 6.40%. As shown in Fig. 2, this model exhibited a better fit than the power law model.
The work investigated the kinetics of the partial oxidation of isooctane over a stable Ni/-Al2 O3 catalyst. The chemical reaction was modeled using rate models developed from the LHHW and ER mechanisms. Out of the eighteen models proposed, eight converged but only four exhibited acceptable AAD% and comparable activation energy as compared to the power law model. A LHHW mechanism requiring the dissociative adsorption of isooctane and molecular adsorption of oxygen on a single site appeared to be the most likely pathway for the partial oxidation of isooctane. The reaction order of 1.5 for this model indicates a strong coverage of nickel by isooctane. The activation energy estimated from the LHHW model was 2.81 × 105 J mol−1 . Acknowledgements The authors would like to thank the AUTO21-Network of Centers of Excellence (NCE), Canada for their financial support.
References Ahmed, S., Kumar, R., Krumpelt, M., 1999. Fuel processing for fuel cell power systems. Fuel Cells Bulletin 12, 4–7. Fogler, H.S., 1999. Elements of Chemical Reaction Engineering, third ed. Prentice-Hall, Englewood Cliffs, NJ. Froment, G.F., Bischoff, K.B., 1990. Chemical Reactor Analysis and Design, second ed. Wiley, New York. Geankoplis, C.H., 2003. Transport Processes and Separation Process Principles, fourth ed. Prentice-Hall, NJ. Heck, R.M., Farrauto, R.J., 2002. Catalytic Air Pollution Technology: Commercial Technology, second ed. Wiley, New York. Ibrahim, H.H., Kumar, P., Idem, R.O., 2006. Preparation, characterization and performance evaluation of nickel-based catalysts for the production of hydrogen by the catalytic partial oxidation of isooctane. Applied Catalysis, 2006, submitted for publication. Krumpelt, M., Krause, T.R., Carter, J.D., Kopasz, J.P., Ahmed, S., 2003. Fuel processing for fuel cell systems in transportation and portable power applications. Catalysis Today 77, 3–16. Levenspiel, O., 1999. Chemical Reaction Engineering, thrid ed. Wiley, New York. Li, C., Yu, C., Sahen, S., 2000. Role of the surface state of Ni/Al2 O3 in partial oxidation of CH4 . Catalysis Letters 67, 139–145. Moon, D.J., Ryu, J.W., Lee, S.D., Lee, B.G., Ahn, B.S., 2004. Ni-based catalyst for partial oxidation reforming of isooctane. Applied Catalysis A: General 272, 53–60. Perry, R.H., Green, D.W., 1997. Perry’s Chemical Engineers’ Handbook, seventh ed. McGraw-Hill, New York. Sabacchi, M.G., Saveliev, A.V., Fridman, A.A., Kennedy, L.A., Ahmed, S., Krause, T., 2002. Experimental assessment of a combined plasma/catalytic
5918
H.H. Ibrahim, R.O. Idem / Chemical Engineering Science 61 (2006) 5912 – 5918
system for hydrogen production via partial oxidation of hydrocarbon fuels. International Journal of Hydrogen Energy 27, 635–642. Springmann, S., Friedrich, G., Himmen, M., Sommer, M., Eigenberger, G., 2002. Isothermal kinetic measurements for hydrogen for hydrogen production from hydrocarbon fuels using a novel kinetic reactor concept. Applied Catalysis A: General 235, 101–111.
Walas, S.M., 1990. Chemical Process Equipment—Selection and Design. Butterworth-Heinemann, Newton, MA. Zhu, W, Xiong, G, Han, W, Yang, W., 2004. Catalytic partial oxidation of gasoline to syngas in a dense membrane reactor. Catalysis Today 93–95, 257–261.
Shorter Communication
Kinetic studies of the partial oxidation of isooctane for hydrogen production over a nickel–alumina catalyst Hussam H. Ibrahim, Raphael O. Idem ∗ Process Systems Engineering Laboratory, Faculty of Engineering, University of Regina, 3737 Wascana Parkway, Regina, Canada S4S 0A2 Received 25 January 2006; received in revised form 7 April 2006; accepted 7 April 2006 Available online 28 April 2006
Abstract The kinetics of the catalytic partial oxidation of isooctane for hydrogen (H2 ) production over a stable Ni/-Al2 O3 catalyst was investigated o ) in the at atmospheric pressure in the temperature range of 863–913 K, ratio of weight of catalyst to the molar feed rate of isooctane (W/FiC8 −1 range of 7.09.30.89 kg mol , and molar feed ratio O2 /i-C8 H18 of 4.0 in a 12.7 mm diameter Inconel micro-reactor housed in an electrically controlled furnace. The developed rate models were based on the Langmuir–Hinshelwood–Hougen–Watson (LHHW) and Eley–Rideal (ER) formulations. Out of the 18 models developed, 10 were eliminated due to poor predictive efficiency. A LHHW mechanism requiring the dissociative adsorption of isooctane and molecular adsorption of oxygen on a single site was the most likely pathway for the partial oxidation of isooctane. The reaction order of 1.5 indicates a strong coverage of nickel by isooctane. 䉷 2006 Elsevier Ltd. All rights reserved. Keywords: Isooctane; Gasoline; Partial oxidation; Nickel-based catalyst; Kinetics
1. Introduction The use of catalytic converters to control vehicle exhaust emissions is well known (Heck and Farrauto, 2002). However, increasingly stringent legislation has led to consideration of alternative means of reducing emissions. Of these measures, the use of fuel cells appears to be very promising and has seen remarkable progress in the past decade because of an increasing need to improve energy efficiency as well as to address environmental concerns (Krumpelt et al., 2003). The polymer electrolyte membrane fuel cell (PEMFC) is a potential candidate for direct electricity production from hydrogen for transportation applications and also for distributed and portable power generation. Also, hydrogen is projected to be one of the primary energy sources in the 21st century (Sabacchi et al., 2002). Its combustion and oxidation is pollution free and supplies sufficient energy for transportation and other applications. However, the absence of a viable hydrogen storage option and a hydrogen marketing infrastructure, at least in the near term, ∗ Corresponding author. Tel.: +1 306 585 4470; fax: +1 306 585 4855.
E-mail address: [email protected] (R.O. Idem). 0009-2509/$ - see front matter 䉷 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.ces.2006.04.015
necessitate the search for an appropriate fuel (Ahmed et al., 1999). Petroleum-derived fuels such as gasoline are attractive sources of hydrogen because of their existing production, distribution, and retailing infrastructure. Also, gasoline has a much higher energy density and larger hydrogen content compared to methanol and ethanol (Zhu et al., 2004; Moon et al., 2004), which are also considered as good sources for hydrogen. The conversion of hydrocarbons to hydrogen can be carried out using one of three major techniques: steam reforming, partial oxidation and autothermal reforming. The major by-product CO2 can be captured and sequestered prior to hydrogen use. Partial oxidation is chosen for this study for the reasons that: (i) the reaction is exothermic, making it much more energy-efficient than steam reforming; (e.g. C8 H18 + 4(O2 + 3.76N2 )8CO + 9H2 + 15.04N2 ; o =−6.60×105 J/mol, (ii) a smaller reformer can be used H298 to achieve a high conversion of the hydrocarbon selectively in favor of the production of H2 at short contact times; and (iii) the partial oxidation setup is more compact and mechanically simpler than the steam reforming, since no additional heating is required (Li et al., 2000). The development of an efficient partial oxidation reformer for the production of hydrogen from
H.H. Ibrahim, R.O. Idem / Chemical Engineering Science 61 (2006) 5912 – 5918
liquid hydrocarbon fuels such as gasoline is very challenging. It requires the development of a stable catalyst for the process. It also requires detailed thermodynamic and kinetic information about the reaction and such information is presently scant (Springmann et al., 2002). In aprevious study (Ibrahim et al., 2006), we obtained a stable performance for a Ni/-Al2 O3 catalyst for the production of hydrogen by the partial oxidation of isooctane. Thus, the objective of the present study is to develop a mechanistic based kinetic model to describe the catalytic partial oxidation of isooctane to produce hydrogen using this stable Ni/Al2 O3 catalyst. The developed models are based essentially on the LHHW and ER hypotheses and these were cross-examined with actual experimental kinetic data. The results are presented in this communication. 2. Theory Initially, an empirical, irreversible fixed feed molar ratio power law rate model was developed as shown in Eq. (1). rA = k0 e(−E/RT ) CAm .
3.1.1. Experimental apparatus and analysis An Inconel fixed bed reactor (i.d. = 12.7 mm) housed in a furnace with a single heating zone was used for catalyst performance evaluation. Liquid isooctane was introduced by a syringe pump (kd Scientific-200) while the gas flows were metered and regulated by an Aalborg digital flow controller (DFC-26). The air to isooctane molar ratio was 12.0. The catalyst bed temperature was measured by means of a sliding thermocouple dipped inside the catalyst bed. The diluent used in the catalyst bed was -Al2 O3 having the same particle size as the catalyst (150 m). Pure -Al2 O3 (150 m) was used in the preheating zone and the section after the catalyst bed. The total catalyst bed height was 70 mm. The exit product from the reactor was separated into permanent gases and liquid condensate by passing first through a condenser and then a gas/liquid separator. The gases were analyzed with an on-line gas chromatograph (HP-6890, Agilent Technologies) equipped with a TCD using Haysep and Molsieve columns (Alltech Associates) for complete gas product separation. The unreacted isooctane was analyzed with a GC-MS (HP-6890/5073, HP) using a 30 m GS-GasPro column (J&W Scientific).
(1)
Then, different mechanistic models based on LHHW and ER mechanisms were proposed. The overall reaction considered in the development of all models is given in Eq. (2). C8 H18 + 4O2 ←→ 8CO + 9H2
5913
o H298 = −660 kJ mol−1 .
(2) The discrimination between rival models was based on predictive efficiency (i.e., the average absolute deviation (AAD%) between predicted and experimental results) and the similarity in activation energy values between the model and the power law model. Other supporting criteria such as residue analysis and chemical common sense were also used where appropriate. The development of these two classical mechanisms was based on the assumptions that: (1) the rate-determining step (RDS) could be controlled by the surface reaction step, adsorption step or desorption step; (2) there is the presence of uniformly energetic adsorption sites; and (3) there is monolayer coverage. The only differing assumption between LHHW and ER mechanisms was the use of a dual site mechanism for the reaction in the former case and a single site mechanism in the latter case. Sixteen LHHW and two ER models were proposed and tested. All the rate models, and the assumptions used in their derivation are given in Table 1(a–c). 3. Experimental 3.1. Catalyst selection The catalyst used in this work was a 3% Ni/-Al2 O3 catalyst prepared by the precipitation method. This catalyst showed good activity and stability in terms of isooctane conversion and hydrogen selectivity. Details of preparation, characterization and performance evaluation of this catalyst are given elsewhere (Ibrahim et al., 2006).
4. Kinetic studies Experimental kinetic data were collected at atmospheric pressure; temperatures of 863, 883, 903 and 913 K; and o of 255.6, 198.0, 162.0 and 136.8 kg-cat s kg-iC−1 . In W/FiC8 8 order to approach plug flow conditions, and minimize backmixing and channeling, certain operating criteria as prescribed by Froment and Bischoff (Froment and Bischoff, 1990) were used. These were that the ratio of catalyst bed length to catalyst particle diameter (L/Dp ) was 350 and the ratio of the inside diameter of reactor to particle diameter (D/Dp ) was 63.5. Catalyst particle size of 150 m and feed flow rate of 390 mL min−1 were used to eliminate pore and external diffusional limitations, respectively. The pressure drop along the reactor was calculated for the particle size used and found to be insignificant (< 5 × 103 Pa) at the highest reaction temperature. 5. Results and discussion 5.1. Heat and mass transport limitations It is well known that intrinsic kinetic data can only be obtained in the absence of heat and mass transport resistances. Theoretical criteria were used to determine whether there were any effects of interparticle and intraparticle heat and mass transport limitations on the rate of reaction at 913 K. 5.1.1. Heat transport effects Internal pore heat transfer resistance was estimated using the Prater analysis given by Tmax,particle =
Deff (CAs − CAc )(−Hr ) , keff
(3)
5914
H.H. Ibrahim, R.O. Idem / Chemical Engineering Science 61 (2006) 5912 – 5918
Table 1 Rate models based on LHHW and ER mechanisms Model # / type
Assumptions
Rate model
Eq. #
(a) M2/LHHW
Dissociative adsorption C8 H18 (2C4 ) and O2
of
both
k0 e−(E/RT ) CA
rA = 1 + KA
(CA0 − CA )17/2 2 CA
C8 H18 dissociative adsorption assumed to be the RDS M3/LHHW
Dissociative adsorption C8 H18 (2C4 ) and O2
of
both
4 k0 e−(E/RT ) CA
rA =
1 + (KA CA )1/2 +
(CA0 − CA )11/2
Dissociative adsorption C8 H18 (2C4 ) and O2
of
both
Dissociative adsorption C8 H18 (2C4 ) and O2
1 + KA
of
both
8
4
k0 e−(E/RT ) CA
rA =
Desorption of H2 and CO C8 H18 dissociative adsorption assumed to be the RDS M5/LHHW
3
(KB CA )1/2
O2 dissociative adsorption assumed to be the RDS M4/LHHW
2
+ (KB CA )1/2
(CA0 − CA )17/2 2 CA
5 k0 e−(E/RT ) CA
rA =
2
5
+ (KB CA )1/2 + KC (CA0 − CA )
17
6
1 + (KA CA )1/2 + KB (CA0 − CA )
Desorption of H2 and CO Surface reaction assumed to be the RDS 5 CA
k0 e−(E/RT )
M6/LHHW
Dissociative adsorption C8 H18 (2C4 ) and O2
of
both
(CA0 − CA )9 8 1/2 5/8 1 + KA CA + KB CA (CA0 − CA )−9/8 + KC (CA0 − CA )
rA =
Desorption of H2 and CO CO desorption assumed to be the RDS M7/LHHW
7
5 k0 e−(E/RT ) CA
Non-dissociative adsorption of both C8 H18 and O2 Desorption ofH2 and CO Surface reaction assumed to be the RDS
rA =
Dissociative adsorption C8 H18 (C5 , C3 ) and O2
rA = ⎛
8
[1 + KA CA + KB CA0 ]17
(b) M8/LHHW
of
both
⎜1 + ⎝
Dissociative adsorption C8 H18 (C5 , C3 ) and O2
of
both
rA = ⎛
Dissociative adsorption C8 H18 (C5 , C3 ) and O2
5/2
of
both
Desorption of H2 and CO Surface reaction (C3 + O) assumed to be the RDS
+
KB (CA0 − CA )6 3/2
5/2
KB (CA0 − CA )12 3/2
CA +KD (CA0 − CA )7
rA = ⎛
⎞2
9
⎟ ⎠
5 k0 e−(E/RT ) CA
⎜ K A CA + ⎝
Desorption of H2 and CO Surface reaction (C5 + O) assumed to be the RDS M10/LHHW
KA (CA0 − CA )11
CA CA +(KC CA )1/2 + KD (CA0 − CA )
Desorption of H2 and CO C8 H18 dissociative adsorption assumed to be the RDS M9/LHHW
k0 e−(E/RT ) CA
⎞11 + (1 + (KC CA )1/2 )(CA0 − CA )6 ⎟ ⎠
5 k0 e−(E/RT ) CA
⎞6
KB (CA0 − CA )22 7/2 + (1 + (KC CA )1/2 )(CA0 − CA )11 ⎟ ⎜ K A CA + 5/2 ⎝ ⎠ CA +KD (CA0 − CA )12
10
11
H.H. Ibrahim, R.O. Idem / Chemical Engineering Science 61 (2006) 5912 – 5918
5915
Table 1(Contd.) Model # / type
Assumptions
Rate model
Dissociative adsorption of C8 H18 (2C4 ) with O2 in the gas phase
rA =
Eq. #
(c) M11/ER
5 k0 e−(E/RT ) CA
17
12
1 + (KA CA )1/2 + KB (CA0 − CA )
Desorption of H2 and CO Surface reaction assumed to be the RDS M12/LHHW
Dissociative adsorption of both C8 H18 O2 in the gas phase
k0 e−(E/RT ) CA 2 1 + (KA CA )1/2
rA =
13
Surface reaction assumed to be the RDS 2 k0 e−(E/RT ) CA
M13/LHHW
Molecular adsorption of both C8 H18 O2 in the gas phase Surface reaction assumed to be the RDS
rA =
M14/LHHW
Dissociative adsorption of C8 H18 and molecular O2 in the gas phase
rA =
14
[1 + KA CA ]2
k0 e−(E/RT ) CA
3/2
1 + (KA CA )1/2 + KB CA
2
15
Surface reaction assumed to be the RDS k0 e−(E/RT ) CA rA = 1 + (KA CA )1/2 3/2
M15/ER
Dissociative adsorption of C8 H18 with O2 in the gas phase Surface reaction assumed to be the RDS
M16/ER
Non-dissociative adsorption of C8 H18 with O2 in the gas phase Surface reaction assumed to be the RDS
rA =
M17/LHHW
Dual site dissociative adsorption of both C8 H18 and O2 Surface reaction assumed to be the RDS
rA =
M18/LHHW
Dual site non-dissociative adsorption of both C8 H18 and O2 Surface reaction assumed to be the RDS
rA =
2 k0 e−(E/RT ) CA
17
[1 + KA CA ]
k0 e−(E/RT ) CA √ √ (1 + KA CA ) + (1 + KB CA )
where Tmax,particle is the upper limit to temperature variation between pellet center and its surface, Hr is the heat of reaction, and CAs & CAc are the concentrations at the pellet surface and center, respectively, Deff is the effective mass diffusivity obtained from Deff = (DAB /) (Fogler, 1999) where DAB is the bulk diffusivity of component A in B (i.e., isooctane in air), which in turn, is estimated using Fuller–Schettler–Giddings equation (Perry and Green, 1997). The value for DAB was obtained at the maximum temperature of 913 K as 0.466 cm2 /s. The effective diffusivity Deff was estimated to be 0.583 cm2 /s. is the void fraction (estimated as the ratio of the volume occupied by voids to the total bed volume = 0.5), is the tortuosity factor taken as 0.4 (Fogler, 1999), keff is the effective thermal conductivity obtained using the correlation keff /kc = 5.5 + 0.05NRe (Walas, 1990) for packed bed tubular reactors. kc is the mass transfer coefficient calculated by equating two J factors −0.4069 given by the following correlations JD = (0.4548/)NRe 2/3 and JD =(kc/)NSc (Geankoplis, 2003) where is the superficial velocity and NSc is the Schmidt number; NSc =(/DAB ), NRe is the Reynolds number; NRe = dp /(1 − ) where dp is the particle diameter. The effective thermal conductivity keff was calculated to be 1.19 × 104 kJ/(hr m K). A value of 0.03 K was obtained for Tmax,particle which shows that the pellet more or less had a uniform temperature.
16
2 k0 e−(E/RT ) CA
[(1 + KA CA ) + (1 + KB CA )]
18
19
The heat transfer limitations across the gas film was determined using Eq. (4) Tmax,film =
L(−rA,obs )(−Hr ) , h
(4)
where Tmax,film is the upper limit to temperature difference between the gas bulk and the pellet surface, L is the characteristic length, rA,obs is the observed rate of reaction, h is heat transfer coefficient (estimated from the correlation JH = 2/3 JD = (h/Cp )NP r where JH is the heat transfer J factor, NP r is Prandtl number; NP r = Cp /, Cp is the heat capacity, is the thermal conductivity). The heat transfer coefficient h was determined to be 4.77 kJ/(m2 K s). A value of 0.02 K was obtained for Tmax,film . These results confirm the absence of heat transfer limitations externally and internally. These led us to assume isothermal operation conditions during the reaction (Levenspiel, 1999). 5.1.2. Mass transport effects The internal pore mass transfer resistance was calculated using Weisz–Prater criterion as given by Eq. (5). Cwp,ipd =
−rA,obs c Rc2 , Deff CAs
(5)
5916
H.H. Ibrahim, R.O. Idem / Chemical Engineering Science 61 (2006) 5912 – 5918
where Cwp,ipd the Weisz–Prater criterion for internal pore diffusion, c the pellet density, Rc catalyst radius. The estimated value for Cwp,ipd was 2.31×10−4 . This value is much less than 1. Thus, this result indicates that the concentration on the catalyst surface is more or less the same as the concentration within
its pores. This result comes as a consequence of the absence of internal pore diffusion limitations (Fogler, 1999). To determine whether film mass transfer resistance has any effect on the rate of reaction, the ratio of observed rate to the rate if film resistance controls was examined. Eq. (6) illustrates this criterion. observed rate −rA,obs dp = , rate if film resistance controls CAb kc 6
45 T=590 isooctane conversion (mol%)
40
where dp is the catalyst particle diameter. The estimated value for the ratio in Eq. (6) was 3.435 × 10−15 . The result indicates that the observed rate is very much lower than the limiting film mass transfer rate. Thus, the resistance to film mass transfer certainly should not influence the rate of reaction (Levenspiel, 1999).
T=610 T=630
35
(6)
T=640
30 25
5.2. Catalyst evaluation and experimental rates 20
The catalysts were evaluated in terms of isooctane conversion. Material balance calculations based on carbon balance were performed for a selected number of runs. The overall recovery was > 93% in all cases. Fig. 1 gives the relationship o ratio for temperabetween isooctane conversion and W/FiC8 tures in the range of 863–913 K. The experimental rates of reaction were obtained using the differential method of analysis o data. Thus, the experimental rates for the conversions-W/FiC8 of reaction of isooctane were taken as the slopes or tangents
15 10 0.030
0.038
0.046
0.054
0.062
0.070
0.078
W/FA° (kg-cat kg-iC8-1 h) Fig. 1. Isooctane conversion as a function of contact-time at 863, 883, 903 and 913 K.
Table 2 Estimate of the values of the parameters of the models Parameter
Power law
M1
M2
M3
M4
M5
M6
M7
M8
k0 (kg-cat −1 s−1 )
4.33 × 1018
No convergence
No convergence
1.23 × 1023
No convergence
No convergence
No convergence
1.25 × 1023
E (J mol−1 ) m KA KB KC KD
No convergence
2.99 × 105 0.82 – – – –
3.43 × 105 – 5.85 × 10−4 1.51 × 107 4.37 × 107 –
3.48 × 105 – 5.85 × 10−4 2.51 × 107 7.17 × 106 1.96 × 107
Parameter
M9
M10
M11
M12
M13
M14
M15
M16
k0 (kg-cat −1 s−1 ) E (J mol−1 ) M KA KB KC KD
No convergence
No convergence
No convergence
5.24 × 1020 3.04 × 105 – 2.51 × 107 – – –
No convergence
1.25 × 1023 2.81 × 105 – 5.85 × 10−4 2.51 × 107 – –
1.25 × 1023 2.87 × 105 – 1.59 × 10−24 – – –
1.25 × 1023 2.08 × 105 – 2.94 × 108 – – –
Parameter
M17
M18
k0 (kg-cat −1 s−1 ) E (J mol−1 ) M KA KB KC KD
1.25 × 1023 3.40 × 105 – 2.94 × 108 2.51 × 107 – –
1.25 × 1023 2.02 × 105 – 2.39 × 108 7.44 × 107 – –
H.H. Ibrahim, R.O. Idem / Chemical Engineering Science 61 (2006) 5912 – 5918
2.0E-05
Also, Table 2 shows that M14 had an activation energy of 2.81 × 105 J mol−1 , which is very close to the value obtained for the power law model (2.99 × 105 J mol−1 ) implying no change in the mechanism between M14 and the power law model.
Rpre_PLM [AAD%=6.75]
Predicted rate (kmol kg-1 s-1)
Rpre_M4 [AAD%=9.80]
1.6E-05
5917
Rpre_M12 [AAD%=9.27] Rpre_M14 [AAD%=6.38] Rpre_M15 [[AAD%=1022]
1.2E-05
6. Conclusions 8.0E-06
4.0E-06
0.0E+00 0.0E+00 4.0E-06
8.0E-06
1.2E-05
1.6E-05
2.0E-05
Observed rate (kmol kg-1 s-1) Fig. 2. Parity plot for rates of partial oxidation of isooctane—models vs. experimental rates.
to the curves in Fig. 1 at desired W/FA0 values for the reaction temperatures. To verify that the dominant reaction step is the same at high and low temperatures, the data in Fig. 1 were further examined by treating the two regions separately. The results for the high temperature region (903–913 and 883–913 K) and the low temperature region (863–883 and 863–903 K) gave similar activation energies which were also identical to the global value (i.e., Ea = 2.8 × 105 J mol−1 ) obtained over the entire temperature domain. This result indicates that there is no change in the controlling mechanism for the two temperature regions. 5.3. Estimation of the values of the parameters of the kinetic models The estimation of the values of the parameters was based on the minimization algorithm, which essentially involves a combination of Gauss–Newton and Levenberg–Marquardt methods, and employed a non-linear regression software, NLREG. Table 2 gives the estimates of the values of the parameters for the partial oxidation of isooctane for all the final eight models in addition to the power law model. 5.4. Model validation The validation of the developed models was done through the measurement of the percentage absolute average deviation (AAD%) for each model from the actual experimental observations. Out of the models that converged, those with AAD of less than 15% were deemed acceptable and were short-listed. Fig. 2 depicts how well the models fitted the experimental data. The empirical power law model showed a good fit for the experimental (AAD = 6.75%). Among the mechanistic models, the LHHW model (M14) exhibited the best data fit with an AAD of 6.40%. As shown in Fig. 2, this model exhibited a better fit than the power law model.
The work investigated the kinetics of the partial oxidation of isooctane over a stable Ni/-Al2 O3 catalyst. The chemical reaction was modeled using rate models developed from the LHHW and ER mechanisms. Out of the eighteen models proposed, eight converged but only four exhibited acceptable AAD% and comparable activation energy as compared to the power law model. A LHHW mechanism requiring the dissociative adsorption of isooctane and molecular adsorption of oxygen on a single site appeared to be the most likely pathway for the partial oxidation of isooctane. The reaction order of 1.5 for this model indicates a strong coverage of nickel by isooctane. The activation energy estimated from the LHHW model was 2.81 × 105 J mol−1 . Acknowledgements The authors would like to thank the AUTO21-Network of Centers of Excellence (NCE), Canada for their financial support.
References Ahmed, S., Kumar, R., Krumpelt, M., 1999. Fuel processing for fuel cell power systems. Fuel Cells Bulletin 12, 4–7. Fogler, H.S., 1999. Elements of Chemical Reaction Engineering, third ed. Prentice-Hall, Englewood Cliffs, NJ. Froment, G.F., Bischoff, K.B., 1990. Chemical Reactor Analysis and Design, second ed. Wiley, New York. Geankoplis, C.H., 2003. Transport Processes and Separation Process Principles, fourth ed. Prentice-Hall, NJ. Heck, R.M., Farrauto, R.J., 2002. Catalytic Air Pollution Technology: Commercial Technology, second ed. Wiley, New York. Ibrahim, H.H., Kumar, P., Idem, R.O., 2006. Preparation, characterization and performance evaluation of nickel-based catalysts for the production of hydrogen by the catalytic partial oxidation of isooctane. Applied Catalysis, 2006, submitted for publication. Krumpelt, M., Krause, T.R., Carter, J.D., Kopasz, J.P., Ahmed, S., 2003. Fuel processing for fuel cell systems in transportation and portable power applications. Catalysis Today 77, 3–16. Levenspiel, O., 1999. Chemical Reaction Engineering, thrid ed. Wiley, New York. Li, C., Yu, C., Sahen, S., 2000. Role of the surface state of Ni/Al2 O3 in partial oxidation of CH4 . Catalysis Letters 67, 139–145. Moon, D.J., Ryu, J.W., Lee, S.D., Lee, B.G., Ahn, B.S., 2004. Ni-based catalyst for partial oxidation reforming of isooctane. Applied Catalysis A: General 272, 53–60. Perry, R.H., Green, D.W., 1997. Perry’s Chemical Engineers’ Handbook, seventh ed. McGraw-Hill, New York. Sabacchi, M.G., Saveliev, A.V., Fridman, A.A., Kennedy, L.A., Ahmed, S., Krause, T., 2002. Experimental assessment of a combined plasma/catalytic
5918
H.H. Ibrahim, R.O. Idem / Chemical Engineering Science 61 (2006) 5912 – 5918
system for hydrogen production via partial oxidation of hydrocarbon fuels. International Journal of Hydrogen Energy 27, 635–642. Springmann, S., Friedrich, G., Himmen, M., Sommer, M., Eigenberger, G., 2002. Isothermal kinetic measurements for hydrogen for hydrogen production from hydrocarbon fuels using a novel kinetic reactor concept. Applied Catalysis A: General 235, 101–111.
Walas, S.M., 1990. Chemical Process Equipment—Selection and Design. Butterworth-Heinemann, Newton, MA. Zhu, W, Xiong, G, Han, W, Yang, W., 2004. Catalytic partial oxidation of gasoline to syngas in a dense membrane reactor. Catalysis Today 93–95, 257–261.