Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Chemical Engineering Science 60 (2005) 4119 – 4129

www.elsevier.com/locate/ces

Kinetics of n-dodecane dehydrogenation on promoted platinum catalyst


G. Padmavathia , K.K. Chaudhuria,∗ , D. Rajeshwerb , G. Sreenivasa Raob , K.R. Krishnamurthyb ,
P.C. Trivedic , K.K. Hathic , N. Subramanyamc
a Chemical Engineering Division, Research Centre, Indian Petrochemicals Corporation Limited, Baroda-391 346, India
b Catalysis Division, Research Centre, Indian Petrochemicals Corporation Limited, Baroda-391 346, India
c Chemical Engineering Department, Faculty of Technology, MS University of Baroda, Baroda-390 002, India

Received 1 April 2004; received in revised form 4 January 2005; accepted 21 January 2005
Available online 15 April 2005

Abstract
Kinetics of n-dodecane dehydrogenation over a promoted Pt–Sn–/Al2 O3 catalyst between 733 and 763 K were determined. The reaction
scheme includes three consecutive dehydrogenation reactions and also cracking of olefins to light paraffins. Various rate models based
on Langmuir–Hinshelwood–Hougen–Watson (LHHW) and Eley Rideal mechanisms were derived for the three dehydrogenation reactions
and subjected to model discrimination and parameter estimation by using Box’s complex optimisation method. Of the 12 different models
tested, model based on LHHW mechanism with surface reaction as rate-determining step fitted the experimental data well. The estimated
rate constants of the best model are thermodynamically sound and statistically consistent.
䉷 2005 Elsevier Ltd. All rights reserved.

Keywords: Kinetics; Dehydrogenation; n-dodecane; Surface reaction; Mono olefin

1. Introduction Literature on kinetics of higher paraffin dehydrogenation


is very limited. Studies by Krylova et al. (1980a) was based
Selective dehydrogenation of higher paraffins (C10 –C14 ) on H2 –D2 exchange and dehydrogenation of n-decane and
to the corresponding mono olefins on promoted Pt/alumina decene for Pt/Al2 O3 and other promoted systems. Step-
catalyst is an industrially important process and a key step in wise dehydrogenation involving desorption of olefins and
the manufacture of bio-degradable detergents. The reaction di-olefins as the rate determining step was proposed by
is carried out in a radial flow fixed bed reactor at higher tem- Krylova et al. (1980b). However no mechanistic models
perature (723–773 K) and low pressure (200–300 kPa) envi- were tested. Initially the rate expressions were derived for
ronment, with hydrogen as diluent. The catalyst employed is Pt–Li–W/Al2 O3 catalyst and later extended to Pt–Sn/Al2 O3
highly selective to mono olefins by major dehydrogenation catalyst as well (Sadykhova et al., 1981). It has been es-
reaction. However, small amounts of di-olefins and aromat- tablished that relative to mono metallic Pt/Al2 O3 catalyst,
ics are formed by subsequent secondary dehydrogenation bi metallic Pt–Sn/Al2 O3 promotes desorption of olefins
reactions. All these products can lead to the formation of and hence may follow a different mechanism (Sadykhova
light paraffins by cracking. Isomerization reaction may also et al., 1984). Kinetic study of coke formation in dehydro-
be possible. Studies on kinetics help to develop the reaction genation of n-decane on Pt–Al2 O3 catalyst was reported by
mechanism and establish reaction network of the process Mart’yanaova et al. (1982). George (1991) studied the kinet-
and kinetic model for simulation and design of commercial ics of n-dodecane dehydrogenation on Pt–Sn/Al2 O3 catalyst
reactors. The kinetics studies also help to evaluate catalyst but their study was limited to lower conversions and ig-
efficiency and attempt improvements thereof. nored the secondary dehydrogenation reactions. Taskar and
∗ Corresponding author. Tel.: +91 265 3062255. Riggs (1997) modelled a detailed kinetic scheme involv-
E-mail address: kumar.chaudhari@ipcl.co.in (K.K. Chaudhuri). ing 35 pseudo components connected by a network of 36

0009-2509/$ - see front matter 䉷 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ces.2005.01.039
4120 G. Padmavathi et al. / Chemical Engineering Science 60 (2005) 4119 – 4129

Fig. 1. Schematic diagram of the experimental set-up.

reactions covering dehydrogenation of naphthenes, isomer- Langmuir–Hinshelwood–Hougen–Watson and Eley Rideal


ization of paraffins, hydro cracking of single branched and mechanisms have been derived for the three dehydrogena-
multiple branched paraffins, dehydrocyclization reactions in tion reactions and subjected to model discrimination and
the C5 –C10 range using Langmuir–Hinshelwood type reac- parameter estimation.
tion rate expressions. Basrur et al. (1998) studied dehydro-
genation of n-decane on promoted Pt–alumina catalyst us-
ing Box–Wilson experimental design and derived quadratic 2. Experimental
response surface correlations for predicting paraffin conver-
sion and olefins selectivity for optimisation of process con- 2.1. Experimental set-up
ditions. Zaera (2002) provided an overview of the advances
in the understanding of the reaction mechanisms of hydro- The experimental set-up is a fully automated system sup-
carbon reforming processes. Emphasis was on the knowl- plied by M/s Xytel, India. The schematic diagram of the ex-
edge developed by using model single crystal metals and perimental set-up used is shown in Fig. 1. Major sections in
modern surface analytical techniques. Liu et al. (2002) de- the unit are: (1) Hydrogen flow control module (2) Liquid
veloped a lumped reaction network and corresponding ki- pumping system (3) Preheater and mixer for hydrogen and
netics model for n-heptane reforming over an unsulfided n-dodecane (4) Reactor (5) Condenser and gas–liquid sepa-
Pt–Re/Al2 O3 . They coupled the kinetics model with cata- rator (6) Gas recirculation module (7) Adequate flow meters
lyst deactivation kinetics which provided a powerful tool to account material balance. The reactor is a down flow of
for a priori predictions of the evolution of catalyst coke homogenised hydrogen and n-dodecane vapour fixed bed re-
profiles and product compositions as a function of time actor. Re-circulation of hydrogen is by gas compressor. The
on stream. Pacheco et al. (2003) studied the kinetics of inner diameter of the SS-321 reactor tube is 1.4 cm. Space
iso-octane reforming on ceria-oxide catalyst impregnated above and below the catalyst in the reactor was filled with
with platinum and derived the reaction kinetics and reaction inert alumina to ensure proper distribution of fluid flow. The
scheme using LHHW formulation to account for the ad- total bed height including catalyst and inert alumina with an
sorption of products and reactants on the active sites of the average particle size of 0.5 mm measures to 3.6 cm. The ther-
catalyst. mocouples for temperature control were placed between the
This paper gives the results of kinetics of n-dodecane outer surface of the inner tube and inner surface of the fur-
dehydrogenation on a commercial catalyst based on pro- nace. The thermocouple for bed temperature indication was
moted Pt–Sn/Al2 O3 . While the earlier studies (George, inserted at the centre of the catalyst bed. One thermocou-
1991; Basrur et al., 1998) were carried out in once through ple was inserted at the end of the bed to measure the outlet
mode, the present study was in hydrogen gas recycle mode, temperature. The wall effects, channelling, and nonunifor-
closer to industrial operation. Various rate models based on mity of the catalyst dispersion are common problems in
G. Padmavathi et al. / Chemical Engineering Science 60 (2005) 4119 – 4129 4121

laboratory reactors. The channelling and heat transfer ef- 2.4. Experimental procedure
fects at reactor wall can be neglected as the radial aspect
ratio (bed diameter to catalyst particle diameter) is > 15 The reactor is first heated in a hydrogen flow of 15 l/min
(Chu and Ng, 1989). The axial aspect ratio i.e. the ratio at about 1 K/min to 423 K to desorb the adsorbed moisture
of catalyst bed length to catalyst particle diameter is > 30 if any and then to 753 K at the rate of 10 K/min to enable
and hence the dispersion effects can be neglected (Carberry, reduction of platinum for 4 h. Fresh n-dodecane from stor-
1976). The catalyst bed dilution with inert particles of the age tank is pumped by HPLC grade high precision metering
same size allowed uniform dispersion of the catalyst parti- pump without any pulsation in to the vaporizer (pre heater).
cles. Accurate mass flow controllers measured amounts of The feed is combined with recycle hydrogen from recycle
hydrogen in to and out of system. Feed to reactor and prod- gas compressor discharge. The desired hydrogen to hydro-
uct from reactor are measured accurately up to second dec- carbon ratio is maintained before the combined mixture is
imal by electronic balances. sent to the pre heater, which is maintained at 573 K. The
vapourizer is designed to mix n-dodecane vapours and hy-
drogen homogeneously. The pre-heater outlet stream enters
2.2. Materials
the reactor from the top and electrical heaters are used to
achieve the desired reaction temperature. The reactor outlet
’ The n-dodecane feed was supplied by M/s. S.D. Fine
stream is cooled in a condenser and separated in to gas and
Chemicals, India and its purity is 98.63 wt% and aromat-
liquid products in gas–liquid separator. The gas from the
ics content is 0.122%. M/s Inox Air Products Ltd supplied
top of the gas–liquid separator is directed to the suction of
the hydrogen gas with the trade name INOX-II and its pu-
the recycle gas compressor. The condensed liquid product
rity is 99.99%. The diluent, alpha-alumina was supplied by
is collected at definite intervals of time.
M/s Hoechst Cerametec, inertness of diluents is checked by
External mass transfer resistance can be minimized by
extent of isomerization activity.
operating at a high velocity of gases and the pore diffusion
resistance can be minimized by proper sizing of the catalyst.
2.3. Preparation of catalyst Absence of external mass transfer resistance was ensured
by carrying out experiments at different mass flow rates and
The dehydrogenation catalyst prepared according to US keeping W/F ratio constant. The conversion at two differ-
Pat. 5677260 by Dongara et al. (1997) consists of highly ent liquid feed flow rate having same W/F ratio remains
porous spheroidal gamma alumina support containing 0.4 almost constant which indicates that the mass transfer effect
wt% of platinum, 0.5 wt% of tin, 0.4wt% of indium, 0.2wt% on conversion in the given range of feed flow rate is neg-
of iron, 0.6wt% of lithium and less than 0.1wt% of chlo- ligible. The feed flow rate used in all experimental runs is
ride. Analysis of the final catalyst sample has been car- in the range of 10–50 ml/h. Preliminary runs were carried
ried out by inductive coupled plasma (ICP) technique. Pre- out to find out the particle size that allows neglecting in-
formed alumina support consists of pure alumina with  as ternal transport limitation. Experiments were carried out by
dominating phase, BET surface area of 160 m2 /g and bulk varying particle size and keeping W/F ratio constant. The
density is 0.27–0.33 g/cm3 . Surface area measurements are results did not show any significant deviation in conversion,
carried out using Sorptomatic-1900 unit by M/s Carloerba, which indicates the absence of intra particle diffusion.
Italy that works on nitrogen adsorption at liquid nitrogen Kinetic runs were carried out at different weight hourly
temperature. space velocity, WHSV, H2 /HC ratio at three different tem-
On the pre-formed alumina support, as the first step, Li peratures of 733, 748 and 763 K. The WHSV is varied in the
was impregnated using an aqueous solution of LiNO3 and range of 25–150. The H2 /HC ratio is varied in the range of
incipient technique, dried at 120 ◦ C for 12 h and calcined 1–9. The liquid and gas products were analysed using Ag-
at 600 ◦ C for 4 h in dry air. In a subsequent second step, ilent 6890 GC. Innowax-5 column of 50 m length, 0.2 mm
Pt, Sn, Fe and In were impregnated in a solution contain- diameter and 0.4 m film thickness and FID detector were
ing H2 PtCl6 , SnCl2 , Fe[NO3 ]3, In[NO3 ]3 and 10 wt% HCl. used for liquid product analysis. The gases were analysed
The above conditions were favourable for the formation of using PLOT alumina column of 50 m length and 0.2 mm
a complex between Pt and Sn and also ensure uniform dis- diameter capillary column and FID detector. To ascertain
tribution of active metal. The impregnated support was once the composition, a representative sample was subjected to
again dried at 120 ◦ C for 12 h and calcined at 550 ◦ C. In GC–MS analysis as per the method published (Bhatt et al.,
order to reduce the chloride content to 0.1wt%, which is 1993). The five groups identified in the order of column re-
responsible for side reactions, the sample is subjected to tention times are (1) C5 –C11 light ends from cracking re-
a de-halogenation step using 5% aqueous ammonia extrac- actions, (2) n-dodecane (used as the reference peak), (3)
tion technique. Resultant sample is further calcined in dry mono olefins (five positional isomers are separated keeping
air at 500 ◦ C and reduced in hydrogen at 480 ◦ C for 4 h un- end application in view), (4) conjugated di-olefins and non
der moisture free conditions at controlled temperature incre- conjugated di-olefins are grouped as dienes (5) aromatics
ments of 3 ◦ C/min rate. (primarily di-substituted types).
4122 G. Padmavathi et al. / Chemical Engineering Science 60 (2005) 4119 – 4129

17 3 Secondary dehydrogenation
16 Dines formation:
Conversion & yields, wt%

2.5
15
2 k2

Yields, wt%
14 C12 H24 ↔ C12 H22 + H2 . (2)
k−2
13 1.5
12 Aromatics formation:
1
11
0.5 k3
10 C12 H22 → C12 H18 + 2H2 . (3)
9 0
0 2 4 6 8 10 Olefins cracking
Time on stream, h
k4
Conversion Aromatics C12 H24 + 2H2 → < C11 hydrocarbons. (4)
Olefin Light paraffins
Dienes Hydrogen
Five possible reaction schemes, shown in Table 1, were de-
rived for the three dehydrogenation reactions. For each of
Fig. 2. Effect of time on stream.
these mechanisms several rate equations were derived de-
pending upon the assumed rate determining step. Twelve
possible rate equations were retained. The rate equations
3. Experimental results
are presented in Table 2. In Mechanism A, LHHW formal-
ism was used for formulation of kinetics of solid catalysed
All the experiments were carried out in the absence of cat-
reversible dehydrogenation reaction. It was reported that
alyst deactivation. The conversion and selectivity of various
though the assumptions made below may not always be ful-
products are presented as a function of time in Fig. 2. The
filled completely it is accepted that this approach is the most
results do not show any deactivation during the run time as
suitable and reliable way of rationalizing observed catalytic
can be observed from the Fig. 2. Each experiment is carried
rate data (Froment and Hosten, 1981). The reaction scheme
out with fresh catalyst to ensure same efficacy of the cata-
corresponding to the three major dehydrogenation reactions
lyst. The data considered in the present investigation for es-
have steps for adsorption, surface reaction and desorption
timation of kinetic parameters corresponds to 1st hour data.
as shown in Table 1. The symbol L denotes the active site
The dehydrogenation catalyst deactivates with time. Hence
of the catalyst surface. In step 1 molecular adsorption of n-
to get the kinetics corresponding to start of run conditions
dodecane takes place on catalyst active site. In the second
the first hour data was considered when the catalyst was
step the adsorbed n-dodecane reacts with another active site
fresh and free from deactivation. The material balance for all
to form C12 H24 -L and H2 -L. In third step dodecene is des-
the runs is in the range of 98.0–99.9%. The reproducibility
orbed. The olefin is readsorbed (step 4) and diene is formed
of the experimental results is around 99.5%.
in stages 5 and 6. Dehydrogenation of diene takes place in
a similar way. All the reactions are assumed to have a step
that is slower than the other steps, i.e. rate determining step.
4. Kinetic analysis
For example, if adsorption is considered as rate determining,
then step 1, step 5 and step 9 in Table 1 are rate determin-
4.1. Reaction scheme
ing steps. The rate equations were derived in the same way
similar to a single reaction. Rate equations were written for
Dodecene is the desired product in n-dodecane dehydro-
the rate determining step of each of the three reactions in
genation. The liquid product of the reaction is about 99%
terms of the concentrations of the adsorbed species. These
and rest is gas. The hydrogen purity of gas is 99.8 wt% and
concentrations were then eliminated by means of the Lang-
the balance is light paraffins. The amount of light paraffins
muir equilibrium relations. This gives rate equations in terms
in the liquid product is around 2–2.5%, which are formed
of accessible gas-phase partial pressures and contains a de-
from cracking. All the products of n-dodecane dehydrogena-
nominator resulting from the adsorption of reacting species.
tion such as, dodecene, dodecadiene, etc. and n-dodecane
Since all the three reactions are assumed to take place on
itself can crack to form light paraffins. In was assumed that
the same active sites, the three rate equations have the same
the light paraffins are formed by olefins cracking. There is
denominator.
a possibility of isomerization but was ignored in the present
The mechanism B is dissociative dehydrogenation and
study. In accordance with the product distribution and the
surface recombination of hydrogen. The final rate expres-
above observations the system can be described by the fol-
sions resulted will be same as mechanism A if it is assumed
lowing reactions:
that the surface contains adsorbed paraffins, olefins, dienes,
Dehydrogenation
aromatics and hydrogen.
Olefins formation:
The mechanism C was dissociative adsorption of n-
k1 dodecane with abstraction of one hydrogen atom in first
C12 H26 ↔ C12 H24 + H2 . (1)
k−1 stage and abstraction of another hydrogen atom in second
G. Padmavathi et al. / Chemical Engineering Science 60 (2005) 4119 – 4129 4123

Table 1 stage to form C12 H24 -L after which desorption of the olefin
Reaction schemes for n-dodecane dehydrogenation takes place in third stage. Final rate expression was derived
A. LHHW model: molecular dehydrogenation assuming that the adsorbed paraffins, olefins, dienes, and
1. C12 H26 + L ↔ C12 H26 L trines and dissociatively adsorbed hydrogen are present on
2. C12 H26 L + L ↔ C12 H24 L + H2 L the surface and the concentrations of other surface compo-
3. C12 H24 L ↔ C12 H24 + L nents are negligible.
4. C12 H24 + L ↔ C12 H24 L
5. C12 H24 L + L ↔ C12 H22 L + H2 L
Mechanism D is Eley Rideal mechanism in which one
6. C12 H22 L ↔ C12 H22 + L adsorbed species reacts with another species in the gas phase.
7. C12 H22 + L ↔ C12 H22 L Mechanism E was proposed by Biloen et al. (1977) (14)
8. C12 H22 L + L ↔ C12 H18 L + H2 + H2 L which involves: (i) cleavage of the C–H bond in dissociative
9. C12 H18 L ↔ C12 H18 + L adsorption to form an adsorbed alkyl group (ii) cleavage of
10. H2 L ↔ H2 + L
the second C–H bond resulting in - or -bonded olefin and
B. Atomic dehydrogenation and surface recombination of hydrogen (iii) desorption of the olefin.
1. C12 H26 + L ↔ C12 H26 L
2. C12 H26 L + L ↔ C12 H25 L + HL
3. C12 H25 L + L ↔ C12 H24 L + HL 5. Parameter estimation and model discrimination
4. C12 H24 L ↔ C12 H24 + L
5. C12 H24 + L ↔ C12 H24 L
6. C12 H24 L + L ↔ C12 H23 L + HL Since the reactor was operated in integral mode, parame-
7. C12 H23 L + L ↔ C12 H22 L + HL ter estimation was based on the minimization of the sum of
8. C12 H22 L ↔ C12 H22 + L the residual squares of the conversion and selectivity of the
9. C12 H22 + L ↔ C12 H22 L products by Box’s complex optimisation method. The cal-
10. C12 H22 L + L ↔ C12 H21 L + HL
culated conversion and selectivity of the products were ob-
11. C12 H21 L + L ↔ C12 H18 L + H2 + HL
12. C12 H18 L ↔ C12 H18 + L tained by numerical integration of the set of ordinary differ-
13. 2HL ↔ H2 L + L ential material balance equations of reference components.
14. H2 L ↔ H2 + L
5.1. Material balance
C. Stepwise dehydrogenation (Kipperman)
1. C12 H26 + 2L ↔ C12 H25 L + HL
2. C12 H25 L + L ↔ C12 H24 L + HL The rate of formation of each component can be expressed
3. C12 H24 L ↔ C12 H24 + L as
4. C12 H24 + 2L ↔ C12 H23 L + HL Rate of disappearance of paraffin (n-dodecane):
5. C12 H23 L + L ↔ C12 H22 L + HL
6. C12 H22 L ↔ C12 H22 + L dXP
7. C12 H22 + 2L ↔ C12 H21 L + HL = −r1 . (5)
d
8. C12 H21 L + L ↔ C12 H20 L + HL
9. C12 H20 L ↔ C12 H18 + H2 + L Rate of formation of olefins (dodecene):
10. 2HL ↔ H2 + 2L
dXO
= r1 − r2 − r4 . (6)
D. Eley Rideal mechanism d
1. C12 H26 + L ↔ C12 H26 L
2. C12 H26 L ↔ C12 H24 L + H2 Rate of formation of diene (dodecadine):
3. C12 H24 L ↔ C12 H24 + L
dXD
4. C12 H24 + L ↔ C12 H24 L = r2 − r3 . (7)
5. C12 H24 L ↔ C12 H22 L + H2 d
6. C12 H22 L ↔ C12 H22 + L
Rate of formation of aromatics:
7. C12 H22 + L ↔ C12 H22 L
8. C12 H22 L ↔ C12 H18 L + 2H2 dXA
9. C12 H18 L ↔ C12 H18 + L = r3 − r4 . (8)
d
E. Rate of formation of light paraffins (< C11 ):
1. H2 + 2L ↔ 2HL
2. HL + C12 H26 ↔ C12 H25 L + H2 dXLP
= r4 . (9)
3. C12 H25 L ↔ C12 H24 HL d
4. C12 H24 HL ↔ C12 H24 + HL
5. HL + C12 H24 ↔ C12 H23 L + H2 Rate of formation of H2 :
6. C12 H23 L ↔ C12 H22 HL
7. C12 H22 HL ↔ C12 H22 + HL
dXH2
= r1 + r2 + 2r3 − 2r4 , (10)
8. HL + C12 H22 ↔ C12 H21 L + H2 d
9. C12 H21 L ↔ C12 H20 HL
10. C12 H20 HL ↔ C12 H18 + H2 + HL
where  = W/F = g cat h/mol paraffin feed; ri (i = 1–4) is
11. HL + HL ↔ H2 + 2L rate of ith reaction. Reaction rate is expressed in terms of
partial pressure of the components. The moles of different
4124 G. Padmavathi et al. / Chemical Engineering Science 60 (2005) 4119 – 4129

Table 2
Rate equations for various rate models

Mechanism A
r1 = k1 (pP − pO pH /KE1 )/DEN (adsorption rds)
r2 = k2 (pO − pD pH /KE2 )/DEN
r3 = k3 pD /DEN
DEN = (1 + KP pO pH /KE1 + KO pD pH /KE2 + KD pA pH 2 + K p + p /K )
A A H H
r1 = k1 KP (pP − pO pH /KE1 )/DEN (surface reaction rds)
r2 = k1 KO (pO − pD pH /KE2 )/DEN
r3 = k3 KD PD /DEN
DEN = (1 + KP Pp + KO pO + KD pD + KA pA + KH pH )2
r1 = k1 KE1 (pP /pH − pO /KE1 )/DEN (desorption rds)
r2 = k2 KE2 (pO /pH − pD /KE2 )/DEN
r3 = k3 KD pD /DEN
DEN = (1 + KP pP + KE1 pP /(KO pH ) + KE2 pO /(KD pH ) + pD /(KA pH 2 ) + p /K )
H H
Mechanism C
r1 = k1 (pP − pO pH /KE1 )/DEN (adsorption rds)
r2 = k2 (pO − pD pH /KE2 )/DEN
r3 = k3 pD /DEN
1.5 /k8k9√K10 + √p /K )2
DEN = (1 + pO √ pH /K2 K3√ K10 + pD √ pH /K5 K6√ K10 + pA pH H 10
r1 = k1 (pP − pO pH /KE1 )/DEN (surface reaction rds)
r2 = k1 (pO − pD pH /KE2 )/DEN
r3 = k3 PD /DEN
√ √
DEN = pH(1 + pO /K3 + pD /K6 + pHpA /K9 + pH /K10 )2
r1 = k1 KE1 (pP − pO pH /KE1 )/DEN (desorption rds)
r2 = k2 KE2 (pO − pD pH /KE2 )/DEN
r3 = k3 KD pD /DEN
DEN = (1 + K4K5pO + K7K8pD + K10−1.5 pH 1.5 )
Mechanism D
r1 = k1 (pP − pO pH /KE1 )/DEN (adsorption rds)
r2 = k2 (pO − pD pH /KE2 )/DEN
r3 = k3 pD /DEN
DEN = (1 + KP pO pH /KE1 + KO pD pH /KE2 + KD pA pH 2 +K p )
A A
r1 = k1 KP (pP − pO pH /KE1 )/DEN (surface reaction rds)
r2 = k2 KO (pO − pD pH /KE2 )/DEN
r3 = k3 KD pD /DEN
DEN = (1 + KP pP + KO pO + KD pD + KA pA )2
r1 = k1 KE1 (pP − pO pH /KE1 )/DEN (desorption rds)
r2 = k2 KE2 (pO − pD pH /KE2 )/DEN
r3 = k3 KA pD /DEN
DEN = 1 + KP pP + KE1 pP /(KO pH ) + KE2 pO /(KD pH ) + KD pD /pH 2)
Mechanism E
r1 = k1 (pP − pO pH /KE1 )/DEN (adsorption rds)
r2 = k2 (pO − pD pH /KE2 )/DEN
r3 = k3 pD /DEN

DEN = (1 + K3 pO + K5 pD + K7 pA pH + K1 /pH )
Power law
r1 = k1 (pP − pO pH m1 /K )
E1
r2 = k2 (pO − pD pH /KE2 )
r3 = k 3 p D
r 4 = k 4 pO
r4 is same for all the models
KEi = ki /k−i

components present at any point when conversion of paraffin where T M= total moles (feed rate + recycle gas flow
is Xp ; moles of Paraffin 1−XP ; olefins = XO ; diene = XD ; rate).
aromatics = XA ; light paraffins = XLP ; hydrogen = XH2 . The above ordinary differential equations were integrated
Partial pressure of each species can be expressed as numerically by the fourth order Runge–Kutta Method for a
given set of kinetic parameters and operating conditions to
pP = (1 − XP )pt /T M; pO = XO pt /T M; obtain concentration of each component at the end of reactor.
pD = XD pt /T M; pA = XA pt /T M; The boundary conditions are

pH2 = XH2 pt /T M, (11) At  = 0; Xi = 0; i = 1 − n (n = no. of component).


G. Padmavathi et al. / Chemical Engineering Science 60 (2005) 4119 – 4129 4125

Table 3
Results of objective function for various rate models

Type of model Rate determining step Objective Fc Ft


function

1. LHHW model Table 1 A Adsorption 7.87 2.096 43573


2. LHHW model Table 1 A Surface reaction 13.01 2.096 29350
3. LHHW model Table 1 A Desorption 16.11 2.096 28864
4. Rate model Table 1 C Adsorption 15.42 2.096 14973
5. Rate model Table 1 C Surface reaction 14.59 2.096 17893
6. Rate model Table 1 C Desorption 18.16 2.096 14449
7. Eley Rideal Table 1 D Adsorption 8.87 2.166 34924
8. Eley Rideal Table 1 D Surface reaction 14.67 2.166 28841
9. Eley Rideal model Table 1 D Desorption 15.24 2.166 28817
10. Rate model Table 1 E Adsorption 13.47 2.096 19401
11. Power-law model (m1 = 1.2) 23.80 2.557 25630
12. Power-law model at (m1 = 1; T = 733 K) 16.71 2.557 64497

The kinetic parameters were estimated by minimization of 200

the sum of the squares of the residuals of calculated and


r0, gmol/h/gcat
150
experimental values of conversion and selectivity of all the 733 K
100
components by using Box complex optimization method. 748 K

The objective function  is 50


763 K


= (Xn,m cal − Xn,m exp )2 , (12) 0
190 210 230 250 270 290
n m
Total pressure, KPa

when Xn,m cal is the predicted selectivity of species m in


Fig. 3. Initial reaction rate vs. total pressure.
the nth experiment and Xn,m exp. is experimental selectivity
value. The objective function values of all the rate models
tested are presented in Table 3.
 value was retained for each mechanism. This retained
The conditions used to discriminate between the above
LHHW models 1 and 2 with adsorption and surface reaction
described various models are: (i) if negative values were
as rate determining step. Finally both these LHHW mod-
obtained for one or several parameters, the model was re-
els with adsorption and surface reaction as rate determining
jected (ii) objective function value should be the least of all
step were considered for further processing with complete
the models (iii) when the objective function of two models
data from further experimentation that allows to check the
are same, then the model with less number of parameters is
parametric sensitivity of the model. The final objective func-
considered. Another criteria for model discrimination is (iv)
tion value of the rate model with surface reaction as rate
an F -test (Sotelo et al., 1993) based on the ratio, Ft
determining step obtained from Eq. (12) is 296 as compared

Xi pr 2 /k to that of 738, for rate model with adsorption as rate de-
Ft =   , (13) termining step, thereby eliminating the LHHW model with
(Xipr − Xi exp )2 /(nm − k)
adsorption as rate determining step.
where k is number of parameters in the model, n is number To verify the observations further, method of initial rates
of experimental points, m is number of species. A given is used. Initial rate is reaction rate at extremely low con-
regression is considered to be meaningful when Ft is larger version of a feed containing no reaction products. The ini-
than Fc (k, (nm−k), ). The  value considered in the present tial reaction rates, r0 , were calculated at different total pres-
work is 0.05. Ft and Fc values are also presented in Table 3. sures for three different temperatures and the variation of r0
Model discrimination is done by rejecting models based with total pressure as a function of temperature is shown in
on their objective function, , values with minimum number Fig. 3. The trend from the Fig. 3 clearly indicates that sur-
of experimental data. Rate models 11 and 12 based on power face reaction is rate determining step at all the three temper-
law in Table 3 were rejected because of their high  values. atures (Froment and Bischoff, 1979).
Then the models with maximum  value in each mechanism The rate of reaction was calculated from the W/F vs.
of A, C & D were rejected. This, eliminated models with conversion data. The conversion of each reaction was calcu-
desorption as rate determining step (models 3, 6, 9) in each lated by material balance. The data were fitted in to a smooth
mechanism. logarithmic curve of the type X = m ∗ ln(W/F ) + C. Then
Of the remaining models, with adsorption and surface re- the equation is differentiated dX/d(W/F ) to obtain rate as
action as rate determining step, the model with minimum a function of W/F . The rate data indicate that the rate of
4126 G. Padmavathi et al. / Chemical Engineering Science 60 (2005) 4119 – 4129

Fig. 5. Arhenius plots of kinetic parameters.

Fig. 4. Rate vs. conversion of paraffins.

formation of aromatics is proportional to the concentration


of dienes and does not depend on the concentration of olefins
and paraffins which also supports the step wise conversion
in the order paraffin–olefin–diene–aromatics. Fig. 4 shows
the plot of rate of disappearance of n-dodecane and building
up of the desired product olefin. The rate decreased linearly
with conversion at the three temperatures. As per Tashkirova
and Kiperman (1970) the reaction is inhibited by at least one
of its products, i.e. by olefins, dienes, or aromatics, if the
rate vs. conversion curve is concave. The results show that
the desorption of the products may not be controlling the
rate in the present case which was confirmed from objective Fig. 6. Adsorption constants vs. temperature.
function values.
Based on the observations drawn from rate data and anal-
ysis of objective function values as discussed above, rate The Arhenius plot of the rate constants is given in Fig. 5.
model based on LHHW mechanism with surface reaction as Adsorption equilibrium constants decreased with increase
rate determining step was retained as final model for the pro- in temperature as it should. Hence the estimates of activa-
cess. The rate expressions according to the retained model is tion energies have to be statistically positive and those for
adsorption enthalpies negative (Froment and Hosten, 1981).
r1 = k1 KP (pP − pO pH /KE1 )/(1 + KP pP The plot of adsorption constants vs. 1/T is given in Fig. 6.
+ KO pO + KD pD + KA pA + KH pH )2 , (14) The activation energy and pre-exponential factor values of
the retained model for all the reactions are in the same range
r2 = k2 KO (pO − pD pH /KE2 )/(1 + KP pP with that reported for n-decane dehydrogenation (Krylova
+ KO pO + KD pD + KA pA + KH pH )2 , (15) et al., 1980a,b).
The values of the kinetic parameters along with standard
r3 = k3 KD pD /(1 + KP pP + KO pO + KD pD
error of kinetic coefficients are given below:
+ KA pA + KH pH )2 , (16)
k1 = 3.25 × 1015 exp(−78.0 ± 0.2/RT ) mol Pa2 /h/g,
r4 = k4 pO , (17)
k−1 = 3.65 × 103 exp(29.5 ± 0.0001/RT ) mol Pa/h/g,
where KP , KO , KH , KD , KA are adsorption equilibrium
constants and are related to the equilibrium constants of the k2 = 8.60 × 1015 exp(−81.9 ± 0.021/RT ) mol Pa2 /h/g,
steps of the reaction in the following way:
k−2 = 1.28 × 103 exp(38.8 ± 0.22/RT ) mol Pa/h/g,
KP = K1 ; KO = 1/K3 ; KD = 1/K6 ;
k3 = 2.49 × 1018 exp(−111.2 ± 0.12/RT ) mol Pa2 /h/g,
KA = 1/K9 ; KH = 1/K10 . (18)
k4 = 3.55 × 1013 exp(−288.4 ± 3.2/RT ) mol Pa−1 /h/g,
The rate coefficients ki obeyed the Arhenius temperature
dependence KP = 2.78 × 102 exp(25.4 ± 0.061/RT ) Pa,

ki = A exp(−E/RT ). (19) KO = 3.80 exp(61.3 ± 0.34/RT ) Pa,


G. Padmavathi et al. / Chemical Engineering Science 60 (2005) 4119 – 4129 4127

Fig. 9. Effect of H2 /HC ratio on conversion and selectivity of products.


Fig. 7. Effect of W/F on conversion.

100 50

Conversion, %
80 Olefin 40
Selectivity, %
60 30
T= 733 K
40 20
Conversion
20 10
0 Diene 0
0 5 10 15 20
Run number ____ : Pred
Symbol: Expt.

100 50

Conversion, %
80 Olefins 40
Selectivity, %

60 30
T = 748K
40 Conversion 20

20 10
Dienes
0 0
0 5 10 15 20
Fig. 8. Effect of W/F on selectivity of olefins. Run number ______ Pred
Symbol: Expt.

KH = 3.04 × 102 exp(25.2 ± 0.1/RT ) Pa, 100 50


T = 763 K Olefins

Conversion, %
80 40
Selectivity, %

KD = 7.6 exp(78.3 ± 0.48/RT ) Pa,


60 Conversion 30
KA = 8.2 exp(78.0 ± 3.06/RT ) Pa. 40 20

20 Dienes 10

0 0
6. Effect of operating parameters 0 5 10 15 20
Run number ______ Pred
Figs. 7 and 8 show the effect of W/F on conversion and Symbol: Expt.
selectivity of mono olefins at three different temperatures. At
Fig. 10. Comparison of predicted conversion and selectivity with experi-
constant W/F paraffin conversion increased with increase in
mental values.
reaction temperature. The selectivity of mono olefin and di-
ene decreased with increase in temperature. The decrease in
selectivity of mono olefins with increase in conversion, sup- perature. The effect of H2 /HC ratio on paraffin conversion
ports the stepwise mechanism of the paraffin dehydrogena- and selectivity of the products is shown in Fig. 9. The paraf-
tion reaction (Kiperman, 1975, 1977). The loss of olefins fin conversion decreased with increase in H2 /HC ratio. The
selectivity is due to secondary dehydrogenation to diene and decreased H2 /HC ratio at a constant pressure lowers the hy-
then to aromatics and parallel formation of light paraffins drogen partial pressure, which favors higher conversion.
by olefin cracking and finally to coke. The selectivity of Fig. 10 shows the comparison of model predicted con-
aromatics and light paraffins increases with increase in tem- version and selectivity of olefins and dienes with that of
4128 G. Padmavathi et al. / Chemical Engineering Science 60 (2005) 4119 – 4129

experimental values at three temperatures for different runs D diene


with different operating conditions, which included the ef- A aromatics
fect of WHSV, H2 /HC ratio and temperature. The agreement LE light ends
between the predicted and experimental values is good as
shown in the Fig. 10.
References

Basrur, A.G., Rajeshwer, D., Gokak, D.T., Rao, G.S., Krishna Murthy,
7. Conclusions
K.R, 1998. Recent advances in basic and applied aspects of industrial
catalysis. Modelling of n-decane Dehydrogenation Using Box–Wilson
Kinetics of n-dodecane dehydrogenation on promoted Experimental Design 113, 809–814.
Pt–Sn/Al2 O3 catalyst was investigated corresponding to Bhatt, B.D., Ali, S., Prasad, J.V., 1993. GC/GC-MS studies on the
start of run conditions. The results support the stepwise determination of the position and geometry of the double bond
mechanism of the reaction. Twelve sets of rate models based in straight chain olefins by derivatization technique. Journal of
on Power law, LHHW and Eley Rideal mechanisms for all Chromatographic Science 31, 113–119.
Biloen, P., Dautzenberg, F.M., Sachtler, W.M.H., 1977. Catalytic
the three dehydrogenation reactions were derived and sub- dehydrogenation of propane over platinum and platinum–gold alloys.
jected to parameter estimation and model discrimination. Journal of Catalysis 50, 77–86.
The kinetic parameters were estimated using Box optimiza- Carberry, J.J., 1976. Chemical and Catalytic Reaction Engineering.
tion method. Rate model based on LHHW mechanism with McGraw-Hill, New York, pp. 157.
surface reaction fitted the data well. The kinetic parameters Chu, C.F., Ng, K.M., 1989. Flow in packed tubes with a small tube to
particle diameter ratio. A.I.Ch.E. Journal 35, 148–163.
of the selected model are statistically significant and follow
Dongara, et al., 1997. Catalyst Composite for Dehydrogenation of Paraffins
Arhenius law. A good agreement between the model pre- to Mono-Olefins and Method for the Preparation Thereof. October 14,
dicted values with those of experimental is observed with U.S. Patent No. 5,677,260.
derived kinetic parameters. Froment, G.F., Bischoff, K.B., 1979. Chemical Reactor Analysis and
Design. Kinetics of Heterogeneous Catalytic Reactions. Wiley, New
York, pp. 110–111 (Chapter 2).
Froment, G.F., Hosten, L.H., 1981. Catalysis science and technology.
Notation In: Anderson, R., Boudart, M. (Eds.), Catalytic Kinetics Modelling.
Springer, Berlin, Heidelberg, New York, pp. 97–170 (Chapter 3).
George, N., 1991. Studies on Alumina Supported Noble Metal Catalysts.
A pre-exponential factor
Ph.D. Thesis, submitted to M.S. University, Vadodara, India.
E activation energy, kJ/mol Kiperman, S.L., 1975. Methods for Testing Catalyst Activity. Inst. Katal,
F molar flow rate, mol/h Sib. Otd. Akad. Nauk SSSR, Novosibirsk., 4, 7.
H2 /HC ratio hydrogen to hydrocarbon Kiperman, S.L., 1977. Kinetika i Kataliz 81, 18.
ratio, mol/mol Krylova, T.L., Nekrasov, N.V., Gudkov, B.S., Gurevich, V.R., Kiperman,
S.L., 1980a. Mechanism and Kinetics of Dehydrogenation of Higher
KP , KO , KD , KA , KH adsorption equilibrium
n-Paraffins on Promoted Platinum Catalysts I. Hydrogen/Deuterium
constants for P , O, D, A, H2 , Pa Isotope Exchange in n-decane. (Translated from Kinetika i Kataliz,
Ki equilibrium constant for step i. 1980, 21, 1482–1486).
ki reaction rate constant, i = 1 to 3, Krylova, T.L., Nekrasov, N.V., Gaidal, N.A., Gurevich, V.R., Kiperman,
mol Pa2 /h/gcat S.L., 1980b. Mechanism and Kinetics of Dehydrogenation of
Higher n-Paraffins on Promoted Platinum Catalysts II. Kinetics
ki reaction rate constant,
of Dehydrogenation of n-decane on a Platinum–Alumina Catalyst
i = 4, mol Pa−1 /h/gcat (Translated from Kinetika i Kataliz 21, 1487–1493).
k−i reaction rate constant, Liu, K., Fung, S.C., Ho, T.C., Rumschitzki, D.S., 2002. Heptane reforming
i = 1–2, mol Pa/h/gcat over Pt–Re/Al2 O3 : reaction network, kinetics, and apparent selective
pi partial pressure of component i, Pa catalyst deactivation. Journal of Catalysis 206, 188–201.
Mart’yanaova, S.K., Gaidai, N.A., Kostyukovskii, M.M., Kiperman, S.L.,
pt total pressure, Pa
Shaskhin, D.P., 1982. Mechanism and Kinetics of Dehydrogenation
ri rate of reaction, mol/(gcat h) of Higher n-Paraffins on promoted Platinum Catalysts 1V. Kinetic
TM total moles present in the Study of Coke Formation in Dehydrogenation of n-decane on
reaction mixture Platinum–Alumina Catalyst (Translated from Kinetika i Kataliz 23,
W weight of catalyst, g 907–912).
Pacheco, M., Sira, J., Kopasz, J., 2003. Reaction kinetics and reactor
W/F space time, gcat h/mol
modelling for fuel processing of liquid hydrocarbons to produce
Xij exp . conversion of component i in hydrogen: isooctane reforming. Applied Catalysis A. General 250,
j th experiment, % 161–175.
Xij cal. calculated conversion of component Sadykhova, Z.A., Nekrasov, N.V., Gurevich, V.R., Kiperman, S.L., 1981.
i for j th experiment, % Mechanism and Kinetics of Dehydrogenation of Higher n-Paraffins on
Promoted Platinum Catalysts III. Study of the Kinetics and Mechanism
Subscript letters of the Dehydrogenation of n-decane on a Platinum–Tin Catalyst
(Translated from Kinetika i Kataliz 22, 396–402).
P paraffin Sadykhova, Z.A., Nekrasov, N.V., Gurevich, V.R., Kiperman, S.L., 1984.
O olefin Mechanism and Kinetics of Dehydrogenation of Higher n-Paraffins
G. Padmavathi et al. / Chemical Engineering Science 60 (2005) 4119 – 4129 4129

on promoted Platinum Catalysts V. Kinetics of Dehydrogenation of Tashkirova, S.G., Kiperman, S.L., 1970. Analysis of kinetic equations.
n-Undecane and n-dodecane on a Platinum–Tin Catalyst (Translated Kinetika i Kataliz 11, 631–637.
from Kinetika i Kataliz 25, 593–597). Taskar, U., Riggs, J.B., 1997. Modelling and optimization of a semi
Sotelo, J.L., Uguina, M.A., Valverde, J.L., Serrano, D.P., 1993. Kinetics of regenerative catalytic naphtha reformer. A.I.Ch.E. Journal 43, 740–753.
toluene alkylation with methanol over Mg-modified ZSM-5. Industrial Zaera, F., 2002. Selectivity in hydrocarbon catalytic reforming: a surface
Engineering Chemistry Research 32, 2548–2554. chemistry perspective. Applied Catalysis A. General 229, 75–91.

You might also like