Chemical Engineering and Processing 95 (2015) 276–283

Contents lists available at ScienceDirect

Chemical Engineering and Processing:

Process Intensification
journal homepage: www.elsevier.com/locate/cep

Kinetic study of autothermal reforming of glycerol in a dual layer

monolith catalyst
Yujia Liu* , Adeniyi Lawal
New Jersey Center for MicroChemical Systems, Dept. of Chemical Engineering and Materials Science, Stevens Institute of Technology, Castle Point on Hudson,
Hoboken, NJ 07030, USA

A R T I C L E I N F O A B S T R A C T

Article history: Kinetic study of autothermal reforming of glycerol was performed in a dual layer monolith catalyst to be
Received 19 January 2015 used in the design of chemical reactors to optimize product (H2 and CO) yield and more efficiently adjust
Received in revised form 2 July 2015 the H2/CO ratio. The catalytic partial oxidation catalyst layer is in direct contact with the steam reforming
Accepted 5 July 2015
catalyst layer next to the monolith wall with the two reactions thus occurring in one reactor. Kinetic
Available online 9 July 2015
experiments were carried out to measure the effects of the steam to carbon molar ratio, oxygen to carbon
molar ratio, and temperature on the rate of reaction. Mass and heat transfer resistances were found to be
Keywords:
negligible. Reaction rate expressions were based on proposed reaction mechanisms using the Langmuir-
Biodiesel
Glycerol
Hinshelwood approach. Non-linear regression was performed to obtain kinetic constants and activation
Kinetics energy. The later was found to be about 131 kJ/mol. Mechanistic considerations indicated that glycerol
Reforming underwent non-dissociative adsorption and steam underwent dissociative adsorption. The modeling
Syngas results were successfully validated with experimental data.
Pt Rh/Pt Monolith catalysts ã 2015 Elsevier B.V. All rights reserved.

1. Introduction The monolith reactor features a honeycomb structure comprising

thousands of gas flow passages with sub millimeter and thus
Glycerol is a theoretical byproduct from biodiesel production microchannel dimensions. Such microchannel reactors provide
via transesterification of triglycerides and represents approxi- greatly enhanced mass and heat transfer rates in comparison to
mately 10 wt% of the total product [1,2]. The utilization of glycerol current macroreactor-based technology [5–8]. The associated high
to produce value-added secondary products such as synthesis gas gas hourly space velocity translates into a significant reduction in
would make biodiesel more economically feasible [3,4]. The reactor size thus providing the important basis for process
conversion of glycerol to syngas and then to methanol represents intensification and miniaturization. Further size reduction is
an important niche for byproduct processing that could reduce the provided by the use of a single reactor in contrast to conventional
dependence of biodiesel producers on natural gas, thus reducing ATR processes which use sequential, separate CPO and SR reactors.
fossil fuel needs. To date most studies have used particulates to investigate
In a previous paper,1 we demonstrated that glycerol could be fundamental reaction data, yet a field process will likely be based
successfully converted into synthesis gas by autothermal reform- on a more structured support. For example, Simonetti et al. [4]
ing (ATR) using Pt, Rh/Pt dual layer monolith catalysts. The ATR of have used particulates to study the rate of glycerol conversion to
glycerol into syngas is an exothermic combustion reaction that is syngas mixtures under kinetically controlled reaction conditions
used to drive the endothermic steam reforming reaction, which is over carbon supported platinum and platinum-rhenium catalysts.
essentially an oxidative steam-reforming process under thermally There are very few articles available on kinetic aspects of ATR
neutral conditions. Monolith ceramic or metal supports wash- especially for glycerol, although the performance of the CPO and SR
coated with an active catalyst, allow for low pressure drop and an with methane as the feedstock fuel has been extensively
efficient mechanism for enhanced heat management. The catalytic investigated.
partial oxidation (CPO) catalyst layer is in direct contact with the A literature review has revealed several modeling efforts aimed
steam reforming (SR) catalyst layer which is adjacent to the at the reforming process. Among the most cited is the modeling
monolith wall with the two reactions thus occurring in one reactor. and reactor simulation work by Xu and Froment [9]. Their work
focused on industrial-scale SR of methane over a Ni/MgAl2O4
catalyst; however, the proposed kinetic model structure has
* Corresponding author. influenced almost all subsequent reforming models. In De Groote

http://dx.doi.org/10.1016/j.cep.2015.07.008
0255-2701/ ã 2015 Elsevier B.V. All rights reserved.
 Y. Liu, A. Lawal / Chemical Engineering and Processing 95 (2015) 276–283 277

and Froment [10], the SR kinetic model of Xu and Froment [17] was glycerol and EA values of 60–90 kJ/mol for CO production over Pt
extended to methane-fed CPO and ATR over a nickel catalyst. In the and Pt-Re catalysts.
automotive application, the desire to reduce reactor size has The objective of this study is to investigate the kinetics of ATR of
motivated the investigation of highly active catalysts based on glycerol to be used in the design of chemical reactors to optimize
noble metals. Regarding CPO of methane, Hickman and Schmidt product (H2 and CO) yield and more efficiently adjust the H2/CO
[11] applied a 19-elementary-step model that included adsorption, ratio. Detailed consideration of the reaction mechanisms and
desorption, and surface reaction rates. Application of this model to hence procurement of the most adequate Langmuir-Hinshelwood
platinum and rhodium catalysts provided an activation-energy- kinetic model has been sought based on information from a wider
based explanation of experimentally observed improvements in H2 range of operating conditions.
selectivity when using a rhodium catalyst. Mhadeshwar and
Vlachos [12] developed a thermodynamically consistent C1 2. Material and methods
microkinetic model for CPO and SR of methane and decomposition
of oxygenate on Rh via a hierarchical multiscale methodology. They Experiments were performed for kinetic study in the reactor
employed sensitivity analysis to identify the important parameters system shown in Fig. 1. A BASF dual layer washcoated monolith
which are refined using quantum mechanical density functional catalyst (400 cpsi) consisting of 1–4 wt% Pt in catalytic partial
theory. Maestri et al. [13] investigated SR and dry reforming (DR) of oxidation layer and Rh/Pt in steam reforming layer on a g-alumina-
methane using a microkinetic model. They concluded that based support was used for all catalytic tests; its preparation is
regardless of the co-reactant used, methane consumption is due described elsewhere [17]. The CPO catalyst layer is exposed to the
to pyrolysis and carbon oxidation. The role of the co-reactant feed while the SR catalyst layer is the first coat on the monolith
(either CO2 or H2O) is to provide the main oxidizer. They also found wall. Each layer of the monolith catalyst is between 5 and
that SR and DR always occur with the WGS reaction close to 10 micrometers. The total length of the dual layer monolith catalyst
equilibrium. They concluded that their kinetic analysis was able to varied from 0.75 to 2.5 inches. The monolith catalyst was placed in
correctly predict the most important features found in experi- the middle of the quartz tube (1 inch inner diameter and 36 inches
ments. Donazzi et al. [14] studied CPO of methane over Rh coated long), and the reactor tube was placed in a three-zone furnace
foams using a thermodynamically consistent C1 microkinetic (Applied Test Systems, Inc.). The first zone of the furnace was used
scheme. They confirmed that the consumption of O2 is strictly as the preheating zone and the temperature was set to 350  C. The
governed by mass transfer and that the co-presence of O2 and monolith catalyst was placed in the second zone, which was used
syngas in the bulk of the gas phase is exclusively due to mass as the reaction zone. The monolith outlet temperature (T/C2 in
transfer limitations. Fig. 1) was controlled using an Omega temperature controller, and
In Pacheco et al. [15] an ATR model was developed for the the inlet temperature was monitored using a K-type thermocouple
reforming of liquid hydrocarbons over a Pt-CeO2 catalyst. The (T/C1 in Fig. 1) inserted into the reactor tube and located
model assumes isooctane as a model molecule to represent a immediately upstream of the catalyst bed.
typical C8 naphtha or paraffinic gasoline. Oxidation and reforming The detailed description about this unique ATR system has been
were then proposed to develop a kinetic model. Despite these thoroughly explained in our previous paper [1,18]. In this kinetic
simplifications, the model compares well with experimental data study, ranges of flow rates of glycerol mixture, D.I. water, air and
over a wide range of operating conditions. Adhikari et al. [16] nitrogen gases used were 1–2 ml/min, 0–0.73 ml/min, 547–
presented kinetic parameters in relation to glycerol steam 1250 sccm, and 400–1422 sccm respectively. The purpose of H2/
reforming over Ni/CeO2. They reported activation energy, EA, of N2 mixture was to reduce the monolith catalyst at first use. The
103.4 kJ mol
1 and reaction order of 0.233 using a power-law pure water line was used to adjust the steam to carbon ratio (S/C) in
model that assumes zero-dependency on steam. Using a similar the feed, while the air inlet stream was used to adjust the oxygen to
model, Simonetti et al. [4] obtained reaction order with respect to carbon feed ratio (O2/C). S/C molar ratio and O2/C molar ratio are

Fig. 1. Schematic of experimental set-up [1,19–21].

278 Y. Liu, A. Lawal / Chemical Engineering and Processing 95 (2015) 276–283

defined in the Appendix. The atomizer sprayed the feed in small appearing in the molecule heading the column. Hence, the atomic
droplets which facilitated mixing and vaporization. matrix for our systems is as follows:
The inlet and product gas mixtures were analyzed every 10 min C3H8O3 H2O O2 CO CO2 H2
with an online Agilent gas chromatograph (GC) equipped with two C 3 0 0 1 1 0
columns (Haysep N and Molecular Sieve 5A) and a TCD detector. H 8 2 0 0 0 2
The determination of the CO2 was recognized by column Haysep N, O 3 1 2 1 2 0
while N2, O2, syngas and CH4 were analyzed by the molecular sieve
column.
The liquid mixtures were analyzed with Shimadzu HPLC with The rank of the matrix can be determined to be three. The
Aminex HPX-87H ion Exclusion (300 mm  7.8 mm) column using number of independent reactions is the difference between the
a refractive index detector. With proper operating conditions, number of columns and the rank of the matrix, which in this case is
glycerol autothermal reforming proceeds to equilibrium products: 3.
CO, CO2, CH4, and H2, which can be analyzed by GC mentioned Based on “atomic matrix”, only 3 independent chemical
above, and water. reactions (from Eq. (1) to Eq. (3)) are needed to describe this 6-
component, 3-atomic species system. Other combinations of
3. Results and discussion reactions could be used to describe all possible products; however,
reactions 1 to 3 provide an appropriate set of reactions that can be
3.1. Catalyst activity for kinetic study combined to describe all possible reaction routes.1 ATR is a
combination of steam reforming (Eq. (1)) and partial oxidation
The experiment was conducted for up to 90 h to investigate (Eq. (2)). The possible reactions for glycerol ATR also include water
activity of the dual-layer monolith catalyst. The catalyst length gas shift (Eq. (3)).
used for this purpose was 0.75 inches. Fig. 2 presents the variations Steam reforming:
of conversion or yield as a function of catalyst on stream time at the
temperature of 650  C, S/C molar ratio of 0.4, O2/C molar ratio of C 3 H8 O3 þ 3H2 O$7H2 þ 3CO2 (1)
0.15 and atmospheric pressure. In order to evaluate the catalyst Oxidation:
activity for kinetic study, the conversion and yields of H2, CO, CO2,
CH4 and total gaseous carbon were calculated and listed in C 3 H8 O3 þ 2O2 $4H2 O þ 3CO (2)
Appendix. Water Gas Shift:
The activity of the catalyst drops very rapidly during the first
24 h, after which the drop is gradual. The results shown in Fig. 2 CO þ H2 O$H2 þ CO2 (3)
indicate that the catalyst did not lose its activity for 19 h of on- A set of experiments was conducted under different operating
stream time. Extensive trial tests were carried out to check conditions to determine the kinetic rate expression for the
repeatability and the carbon deposit. Those trial tests show that autothermal reforming of glycerol. The experimental conditions
during the proposed time duration for experiments the repeat- for glycerol autothermal reforming in this study were set to ensure
ability is good and no obvious carbon deposition is seen on normal and reliable operation of the system and catalyst. The
catalysts when removed from the reactor. Because the experiments temperature in the reformer evolved from 550 to 650  C, ensuring
are conducted quite quickly, the effect of the catalyst deactivation high catalyst activity and avoiding reaching a chemical equilibri-
on the reforming performance can be ignored. The kinetic study, um. Six different catalyst lengths were used. For each catalyst
with its specific conditions, was carried out during 70 to 90 h on length, 3 different temperatures, S/C, and O2/C molar ratio
stream, when the rate of deactivation had reduced significantly. experiments were conducted. Results of the typical experiments
are shown in Figs. 3–7 and discussed below. The variation of partial
3.2. Reaction rate determination pressures of steam along monolith catalyst length/weight is shown
in Figs. 3 and 6. The partial pressure of steam increases along
The chemical species involved in this ATR of glycerol are mainly catalyst weight and reaches a peak within the front half of the
C3H8O3, O2, H2O, H2, CO, and CO2, and three different elements (C, monolith catalyst and then decreases in the rear half. Figs. 5 and 7
H, O). The atomic matrix is made up from the molecular formulas show the variation of partial pressures of products along monolith
of the species entering into the chemical reactions. The first catalyst length/weight, with partial pressures of H2 and CO2
column identifies all the elements in the species while the entries increasing significantly while CO decreases along with the
in the columns that follow are the subscripts of each element

60
90
Conversion or Yield (%)

P steam Kpa (@exit)

40
60

30 20

0 0
70 75 80 85 90 95 0 0.01 0.02 0.03
Catalyst on stream time (hr) Cat. Wt. (g)
Conversion H2 CO2 CH4 CO S/C=1.2 S/C=0.8 S/C=0.4

Fig. 2. Dependence of conversion and yield on catalyst on stream time (Temp. = Fig. 3. Dependence of steam partial pressure on S/C molar ratio (Temp. = 650  C,
650  C, atmospheric pressure, S/C = 0.4, O2/C = 0.15, GHSV  36,000 h
1). atmospheric pressure, O2/C = 0.15).
 Y. Liu, A. Lawal / Chemical Engineering and Processing 95 (2015) 276–283 279

P O2 Kpa (@exit) 12 60

P steam Kpa (@exit)

40

20
0
0 0.01 0.02 0.03

Cat. Wt. (g) 0

0 0.01 0.02 0.03
Fig. 4. Dependence of oxygen partial pressure on catalyst weight (Temp. = 650  C,
atmospheric pressure, O2/C = 0.15). Cat. Wt. (g)

O2/C=0.2 O2/C=0.15 O2/C=0.1

Fig. 6. Dependence of partial pressure of steam on O2/C molar ratio (Temp. = 650  C,
atmospheric pressure, S/C = 0.8).

temperature of 650  C, S/C molar ratio of 0.8 and atmospheric

pressure. Accordingly, the variation of partial pressures of products
as a function of O2/C molar ratio along catalyst weight is shown in
the Fig. 7. Outlet steam partial pressure is higher than that of the
inlet until the catalyst weight increased to 0.021 g. Results show
that the oxidation reaction dominates in the beginning of the
reactor, generating steam, while in the rear part of the reactor, the
steam reforming reactions take over and steam is partly consumed.

3.3. Mass and heat transfer limitation

Meaningful kinetic data is only obtainable in the region of

negligible mass and heat transfer limitations. Gas-solid phase

Fig. 5. Dependence of partial pressure of products on S/C molar ratio (Temp. =

650  C, atmospheric pressure, O2/C = 0.15).

increased S/C and O2/C molar ratio. It was observed that at a fixed
reforming temperature, H2 and CO2 increase rapidly with
increasing residence time.
Fig. 3 presents the variations of steam outlet partial pressure as
a function of S/C molar ratio along catalyst weight at the
temperature of 650  C, O2/C molar ratio of 0.15 and atmospheric
pressure. Accordingly, the variation of partial pressures of products
as a function of S/C molar ratio along catalyst weight is shown in
the Fig. 5. For the inlet S/C molar ratio of 1.2, steam outlet partial
pressure is less than that of inlet beyond the catalyst weight of
0.019 g; while for the S/C molar ratio of 0.4, steam outlet partial
pressure is always higher than that of inlet. Additionally, there was
no O2 left in the end of each experiment as shown in Fig. 4, which
indicates oxidation reaction is accomplished completely. Since
combustion reaction is a very fast reaction and there is no O2 left in
the experiment of using catalyst weight of 0.012 g, it can be
concluded that there were only steam reforming and water gas
shift reactions beyond the catalyst weight of 0.012 g.
Fig. 6 shows the variations of steam outlet partial pressure as a Fig. 7. Dependence of product partial pressure on O2/C molar ratio (Temp. = 650  C,
function of O2/C molar ratio along catalyst weight at the atmospheric pressure, S/C = 0.8).
280 Y. Liu, A. Lawal / Chemical Engineering and Processing 95 (2015) 276–283

catalytic reactions involve a series of mass transfer and reaction 700

steps that contribute to the overall observed reaction rate. The step
that has the greatest influence on the overall reaction rate is
considered as the rate controlling or limiting step. From the bulk 675
gas to the catalyst surface, the overall mass transfer resistance is

T ˚C (@exit)
determined by external mass transfer resistance and internal mass
transfer resistance. External mass transfer resistance refers to
650
transport of the reactants to the catalyst surface. Internal mass
transfer resistance refers to the resistance to diffusion of the
reactants in the pores of the catalyst.
625
For the monolithic reactor studied, the external mass transfer
refers to the mass transfer from the gas bulk to the surface of the
catalyst coating. The internal mass transfer refers to the mass
600
transfer from the external surface of the catalyst coating to the
0 0.01 0.02 0.03
interface between the catalyst coating and the support. The dual
layer catalyst contained a coating with a thickness of 10 micron Cat. Wt. (g)
consisting of at least one layer of steam reforming catalyst in
Fig. 8. Dependence of the monolith catalyst exit temperature on the monolith
contact with at least one layer of partial oxidation catalyst.
catalyst weight (Temp. = 650  C, atmospheric pressure, S/C = 0.8, O2/C = 0.15).
The external and internal mass transfer resistance was
computed using criteria presented in Chinkui Cheng’s work [22].
limitation could be neglected. With all the above considerations,
External mass transfer criteria: C RLkc 23 < 0:15 appropriate reaction conditions have been obtained in the reactor
Ab

Internal mass transfer criteria: F ¼ CRL

2 2

As De
<1 where the reaction is controlled by surface kinetics and not
In the above criteria, R is the flow observed reaction rate and L is significantly influenced by mass and heat transfer resistances.
the characteristic length which is four times larger than the
thickness of the washcoat on the flat sides of the channel [23,24]. 3.4. Rate analysis and kinetic modeling
CAb and CAs are molar concentration of glycerol at the gas phase
and washcoat surface respectively. De is the effective diffusion In the mole balances dF dw
G
¼ rG , FG stands for molar flow rate of
coefficient in the catalyst washcoat. kc [19] is the gas-solid mass glycerol; w stands for the catalyst weight; and rG stands for the
Sh Db
transfer coefficient and can be expressed as dh
, where reaction rate of glycerol. rG is actually a function of time or location
relative to the entrance, rG = f(z). The measurable value for the
Sh ¼ 3:53expð0:0298Re dLgh Sc Þ,Sc ¼ rmDb , and Re ¼ rdmh u.
current experimental setup was only the reactant composition at
Substitution of the relevant values into the left-hand side of the
the entrance and the product composition at the exit. It was not
external mass transfer criteria gave values of the order of 10
3,
practical to monitor the rG along the monolith catalyst. In order to
while the Thiele modulus was estimated to be 0.016 based on
integrate over the catalyst length under current experimental
above data which corresponds to F2 of 0.000276. The effectiveness
setup, six different catalyst lengths were used throughout the
factor was close to unity due to its small Thiele modulus. Therefore,
reaction for obtaining rG along the axial direction of monolith
it can be concluded that the reaction rate of glycerol reforming was
catalyst.
not limited by mass transfer within the catalyst washcoat layers.
In the Section 3.2 performance study for reaction rate
Apart from mass transfer limitations, heat transfer limitations
determination, all O2 was consumed within the catalyst length
should also be considered to ensure that the kinetic data obtained
of 0.25 inches, which was equivalent to 0.012 g. Because of
in the experimental reactor was under a kinetic controlled regime.
equipment limitations, the shortest catalyst length can be worked
The novelty and the most important part of the dual-layer catalyst
with was 0.25 inches. It can be concluded that there was no more
was the idea to integrate the exothermic partial oxidation reaction
partial oxidation reaction beyond the catalyst weight of 0.012 g.
and endothermic steam reforming reaction in a micro-size layer by
Steam reforming and water-gas shift reactions are the only two
layer reactor. The heat produced by partial oxidation reaction will
reactions taking place from the catalyst length of 0.25 inches.
be effectively absorbed and used to drive steam reforming reaction.
Based on the existing experimental set up and data, the kinetics of
To ensure that the kinetic data obtained in the experimental
steam reforming of glycerol can be derived. The net rate for the
reactor reflect only kinetic effects, the criterion established by
disappearance of glycerol would be equal to the rate of steam
Mears [25] was used to estimate whether heat transport effects
reforming reaction from the catalyst length of 0.25 inches.
were significantly affecting the experimental results. A criterion for
After eliminating external mass transfer, internal mass diffusion
quantitative analysis of radial heat transport limitations was given
as well as heat transfer limitations, a catalytic reaction in the
by Damköhler number (Da) for heat transfer which is defined by
Langmuir-Hinshelwood model involves a sequence of three steps:
the following equation:

DHð
robs Þð1
eÞRo 2 RT w  1)Adsorption of reactants on catalyst surface

Da ¼ j j < 0:4 (8)
lT w ð1 þ bÞ Ea  2)Reaction of reactants on the catalyst surface
 3)Desorption of the products from the catalyst surface
If the left hand side of the equation is smaller than the right hand
side, the radial temperature gradients would be less than 5%. The
The adsorption of glycerol and steam on the catalyst may be on
exit temperature of the monolith catalyst along the catalyst length
same sites or each may adsorb on different sites. Both single- and
is shown in Fig. 8. It can be seen that the temperature kept constant
dual-site Langmuir-Hinshelwood (LH) mechanisms for the glycerol
as 650  C from the catalyst weight of 0.012 g to 0.026 g. In this case,
steam reforming are considered. The adsorption of OH groups onto
the apparent heat of reaction is approximately 0 since the heat
strong acid sites is possible, as reported before for glycerol
produced by partial oxidation reaction is consumed instantly by
dehydration on acidic catalysts [26,27]. High acidity leads to a
steam reforming in the catalyst layers. Therefore, the equation is
strong glycerol adsorption; [28] hence both non-dissociative and
satisfied and it can be concluded that the radial heat transfer
dissociative glycerol adsorption are possible on the alumina-
 Y. Liu, A. Lawal / Chemical Engineering and Processing 95 (2015) 276–283 281

Table 1 lines in Figs. 9 and 10 indicate theoretical calculations for how

Langmuir-Hinshelwood rate models. much glycerol is needed for input oxygen. The amount of oxygen
No. Model Description fed was known for each experimental run; therefore the amount of
glycerol consumed can be calculated from Eq. (2), the partial
Single-site adsorption of glycerol and steam
oxidation reaction. Therefore, from the catalyst weight of 0.012 g,
1 krxn PG PW Non-dissociative adsorption of both glycerol
there are only steam reforming and water-gas shift reactions. The
ð1 þ KG PG þ KW PW Þ2 and steam
pffiffiffiffiffiffiffi net rate of glycerol will be equaled to the glycerol disappearance
2 krxn PG PW Non-dissociative adsorption of glycerol and rate of the steam reforming reaction. From Figs. 9 and 10, it is
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ð1 þ KG PG þ KW PW Þ2 dissociative adsorption of steam
pffiffiffiffiffiffi
possible to calculate the reaction rate along the axial direction of
3 krxn PG PW Non-dissociative adsorption of steam and monolith catalyst. The reaction rate was calculated by numerically
pffiffiffiffiffiffiffiffiffiffiffiffi
ð1 þ KG PG þ KW PW Þ2 dissociative adsorption of glycerol
differentiating the exponential curves.
pffiffiffiffiffiffiffiffiffiffiffiffiffi
4 krxn PG PW Dissociative adsorption of both glycerol and The rate data obtained were fitted to different rate equations
pffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2
ð1 þ KG PG þ KW PW Þ steam based on Langmuir-Hinshelwood models. In order to estimate the
Dual-site adsorption of glycerol and steam kinetic constants, the individual rate equations were subjected to a
5 krxn PG PW Non-dissociative adsorption of both glycerol nonlinear regression analysis using polymath. The non-linear
ð1 þ KG PG Þ þ ð1 þ KW PW Þ and steam regression analysis was performed to estimate the kinetic
pffiffiffiffiffiffiffi
6 krxn PG PW Non-dissociative adsorption of glycerol and constants such that the difference between the experimental
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ð1 þ KG PG Þ þ ð1 þ KW PW Þ dissociative adsorption of steam and the predicted rate has a minimum value.
pffiffiffiffiffiffi
7 krxn P P
pffiffiffiffiffiffiffiffiffiffiffiffi G W
Non-dissociative adsorption of steam and The regression analysis on the experimental data was
ð1 þ KG PG Þð1 þ KW PW Þ dissociative adsorption of glycerol performed purely on a mathematical basis and therefore did not
pffiffiffiffiffiffiffiffiffiffiffiffiffi
8 krxn P P Dissociative adsorption of both glycerol and account for the thermodynamic significance of the kinetic
pffiffiffiffiffiffiffiffiffiffiffiffi G W pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ð1 þ KG PG Þð1 þ KW PW Þ steam constants. The values for the kinetic constants obtained from
these rate equations had to satisfy certain conditions, which were
derived from the thermodynamic and optimization considerations.
supported catalyst. Different rate equations are derived for each These constraints are:
mechanistic scheme mentioned above assuming each step as the
rate limiting step. These rate equations are listed in Table 1. a Ki,ki > 0
The results shown in the Fig. 9 indicate that the rate of b Ea > 0
disappearance of glycerol increases with increase in steam c R2 (optimization variable) > 95%
concentration; while the rate of disappearance of glycerol d The confidence intervals of the parameter values should be
decreases with increase in oxygen concentration shown in much less than the values themselves.
Fig. 10. All oxygen was consumed at catalyst weight of 0.012 g,
however the oxygen concentration still affected the steam
concentration. It can be explained that the different O2 concentra-
tion results in the formation of the different amount of H2O based
on the partial oxidation reaction shown in equation (6). The red

Fig. 9. Dependence of glycerol outlet molar flow rate on S/C molar ratio. Fig. 10. Dependence of glycerol outlet molar flow rate on O2/C molar ratio.
282 Y. Liu, A. Lawal / Chemical Engineering and Processing 95 (2015) 276–283

Table 2
Reaction parameters estimation from Langmuir-Hinshelwood rate models at temperature of 650  C.

Model krxn KG kPa
1 KW kPa
1 R2

mol min
1 g
1 kPa
(b+g)
krxn PG PW 0.031  0.010 0.240  0.104 0.019  0.008 0.97
ð1 þ K G PG þ K W PW Þ2
pffiffiffiffiffiffiffi
krxn PG PW 0.110  0.0004 0.188  0.003 0.002  4E-5 0.97
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ð1 þ K G PG þ K W PW Þ2
pffiffiffiffiffiffi
krxn PG PW 0.025  3.585E-6 1.366E-11  1.778E-9 0.052  5.974E-6 0.17
pffiffiffiffiffiffiffiffiffiffiffiffi
ð1 þ K G PG þ K W PW Þ2
pffiffiffiffiffiffiffiffiffiffiffiffi
krxn PG PW 0.103  2.89E-6 4.853E-15  3.599E-12 0.008  6.647E-7 0.83
pffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2
ð1 þ K G PG þ K W PW Þ
krxn PG PW 0.040  0.024 0.333  0.151 0.072  0.061 0.97
ð1 þ K G PG Þð1 þ K W PW Þ
pffiffiffiffiffiffiffi
krxn PG PW 0.114  0.001 0.332  0.015 0.013  0.0007 0.97
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ð1 þ K G PG Þð1 þ K W PW Þ
pffiffiffiffiffiffi
krxn P P 49.590  0.001 1.208E-14  8.03E-12 195.295  0.006 0.81
pffiffiffiffiffiffiffiffiffiffiffiffi G W
ð1 þ K G PG Þð1 þ K W PW Þ
pffiffiffiffiffiffiffiffiffiffiffiffi
krxn P P 0.073  1.93E-6 1.537E-15  2.468E-12 0.012  1.616E-6 0.85
pffiffiffiffiffiffiffiffiffiffiffiffi G Wpffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ð1 þ K G PG Þð1 þ K W PW Þ

data it was found that the surface reaction was the rate controlling
Table 3 step assuming non-dissociative adsorption of glycerol and
Kinetic constants for the rate equation. dissociative adsorption of steam. The kinetic constants were
obtained from non-linear regression of the experimental data and
T C 1/T krxn -lnkrxn R2
1/K mol min
1 g
1 kPa
(b+g)
activation energy is 130.73 kJ/mol.
550 0.00121507 0.014  0.000002 4.26869795 0.97
600 0.00114548 0.052  0.0008 2.95651156 0.98
Acknowledgements
650 0.00108342 0.110  0.0004 2.20727491 0.97
The authors would like to thank BASF Catalysts LLC for
providing the catalysts used in this study. Discussions with Dr.
Farrauto, formerly of BASF LLC, and now at Columbia University
Based on the above constraints, the rate data obtained from the were very helpful. Special thanks to Dr. James Manganaro, Dr. John
experiments were best fitted to a rate equation derived for the Adeosun, Dr. Dongying Qian and Dr. Narendra Joshi for their help
surface reaction as a rate controlling step assuming Non- and guidance.
dissociative adsorption of glycerol and dissociative adsorption of
steam.Table 2 Appendix I
The glycerol steam reforming rate data at different temper-
atures (550, 600, 650  C) were also used to estimate the activation 5.1 S/C= Molarf lowrateof steam
molarf lowrateof carbonincrudeglycerol
energy for product formation. The activation energy of steam Molarf lowrateof oxygen
5.2 O2/C= molarf lowrateof
reforming for glycerol was determined from kinetic constants carbonincrudeglycerol

using the Arrhenius equations which are shown in Table 3. Conversion (X) = F in F
F in
out
 100%
Therefore, the activation energy value for glycerol consumption is Fin is the molar flow rate of a reactant into the reactor and Fout is
130.73 kJ/mol. the molar flow rate of the unconverted reactant exiting the reactor.
The yield, YC(i), is the ratio of moles of carbon in product i (CO,
4. Conclusions CO2, CH4) to the number of moles of carbon in the feed, where Ci
indicates the number of carbon atoms in species i, and Fi is the
Kinetic study on autothermal reforming of glycerol has been molar flow rate of species i.
conducted in a dual layer monolith over a range of steam to carbon Ci F1
Yield (YC (i)) = CGlycerol FGlycerol
molar ratios (0.4–1.2), oxygen to carbon molar ratios (0.1–0.2), and The total gaseous carbon yield (YCt) is defined as the sum of
temperatures between 550  C to 650  C. The BASF dual layer atoms of carbon out, in the form of CO, CO2, and CH4, per atoms of
monolith catalyst was used to develop an intrinsic kinetic rate carbon in the feed, which is the best indication of conversion of
equation for the autothermal reforming of glycerol. The reaction carbon in biodiesel byproduct to equilibrium C1 gas products. Ci
rate of glycerol reforming was not limited by mass transfer within indicates the number of carbon atoms in species i which is 1 in CO,
the catalyst washcoat layers. The apparent heat of reaction is CO2, and CH4, and Fi is the molar flow rate of species i.
P
approximately zero since the heat produced by partial oxidation C i F1
Yield (YCt) = CGlycerol F
reaction is consumed instantly by steam reforming in the catalyst Glycerol

layers. Hence, the radial heat transfer limitation could be The yield of product H2 containing no carbon is analogously
neglected. Therefore, autothermal reforming of glycerol in a dual described: YH, defined in Equations 4, where HH2 indicates the
layer monolith catalyst is considered kinetically controlled under number of hydrogen atoms in species H2, and FH2 is the molar flow
these operating conditions. Langmuir-Hinshelwood model was rate of species H2.
HH2 FH2
found to be suitable for rate expressions. Based on experimental Yield (YH (i)) = HGlycerol FGlycerol
 Y. Liu, A. Lawal / Chemical Engineering and Processing 95 (2015) 276–283 283

References [16] S. Adhikari, S.D. Fermando, A. Haryanto, Chem. Eng. Technol. 32 (2009)
541–547.
[1] Y. Liu, R. Farrauto, A. Lawal, Chem. Eng. Sci. 89 (2013) 31–39. [17] Hwang H. S., Robert J. Farrauto. Patent number: 6849572 "Process for
[2] J.M.N. Van Kasteren, A.P. Nisworo, Resour. Conser. Recycl. 50 (2007) 442–458. generating hydrogen-rich gas". 2002.
[3] Y. Fernández, A. Arenillas, M.A. Díez, J.J. Pis, J.A. Menéndez, J. Anal. Appl. [18] Y. Liu, A. Lawal, Energy Technol. 2 (2014) 792–801.
Pyrolysis 84 (2009) 145–150. [19] Liu Y., Lawal, A. Autothermal reforming of glycerol in a dual layer monolith
[4] D.A. Simonetti, E.L. Kunkes, J.A. Dumesic, J. Catal. 247 (2007) 298–306. catalyst. 2012 AIChE Spring Meeting and 8th Global Congress on Process
[5] Y. Liu, A. Lawal, Catalytic cracking of acetic acid in a microchannel reactor Safety, Conference Proceedings, April 3, 2012, Houston, Texas.
Catalytic cracking of acetic acid in a microchannel reactor. Conference [20] Y. Liu, B. Chen, R. Farrauto, A. Lawal, Kinetic study of autothermal reforming of
Proceedings - 2009 AIChE Spring National Meeting and 5th Global Congress on glycerol using dual layer monolith catalyst. 2012 AIChE Annual Meeting,
Process Safety, April 27, 2009, Tampa, Florida. Conference Proceedings, October 31, 2012, Pittsburgh, Pennsylvania.
[6] O.C. Okafor, S. Tadepalli, G. Tampy, A. Lawal, Int. J. Chem. Reactor Eng. 8 (2010) [21] Y. Liu, A. Lawal, Synthesis Gas Production by Autothermal Reforming of Crude
1542–1550. Glycerol using a Dual Layer Monolith Catalyst Texas. 2013 AIChE Spring
[7] D. Qian, A. Lawal, Chem. Eng. Sci. 61 (2006) 7609–7625. Meeting and 9th Global Congress on Process Safety, Conference Proceedings,
[8] S. Tadepalli, R. Halder, A. Lawal, Chem. Eng. Sci. 62 (2007) 2663–2678. May 1, 2013, San Antonio.
[9] J. Xu, G.F. Froment, AIChE J. 35 (1989) 88–96. [22] C.K. Cheng, S.Y. Foo, A.A. Adesina, Catal. Today 178 (2011) 25–33.
[10] A.M. De Groote, G.F. Froment, Appl. Catal. A 138 (1996) 245–264. [23] S.Y. Joshi, M.P. Harold, V. Balakotaiah, Chem. Eng. Sci. 64 (2009) 4976–4991.
[11] D.A. Hickman, L.D. Schmidt, AIChE J. 39 (1993) 1164–1177. [24] R.E. Hayes, S.T. Kolaczkowski, Chem. Eng. Sci. 49 (1994) 3587–3599.
[12] A.B. Mhadeshwar, D.G. Vlachos, J. Phys. Chem. B 109 (2005) 16819–16835. [25] D.E. Mears, Ind. Eng. Chem. Process Des. Dev. 10 (1971) 541–547.
[13] M. Maestri, D.G. Vlachos, A. Beretta, G. Groppi, E. Tronconi, J. Catal. 259 (2008) [26] H. Atia, U.A. Armbruster, J. Catal. 258 (2008) 71–82.
211–222. [27] W. Suprun, M. Lutecki, T. Haber, H. Papp, J. Mol. Catal. A: Chem. 309 (2009)
[14] A. Donazzi, M. Maestri, B.C. Michael, A. Beretta, P. Forzatti, G. Groppi, E. 71–78.
Tronconi, L.D. Schmidt, D.G. Vlachos, J. Catal. 275 (2010) 270–279. [28] Bancquart, S., Vanhove, C., Pouilloux, Y. and Barrault, J., Applied Catalysis, A.
[15] Pacheco M., Jorge S., Kopasz, J. Applied Catalysis, A. 2003, 250, 161-175. 2001, 218, 1-11.