Pyrocracker 1

Priyesh Ranjan Research Article

Pravin Kannan
Ahmed Al Shoaibi Modeling of Ethane Thermal Cracking
Chandrasekar Srinivasakannan
Kinetics in a Pyrocracker
The Petroleum Institute,
Department of Chemical
The kinetics of ethane thermal cracking in industrial pyrocrackers was investi-
Engineering, Abu Dhabi,
gated using Aspen Plus. The radiant coils were modeled as a 1D plug-flow reactor
United Arab Emirates.
in which ethane is subjected to steam cracking. An improved reaction scheme for
ethane cracking was proposed to predict product distribution which was com-
pared with industrial data to validate the model. Further, the model was used to
assess the effects of various process parameters, including coil outlet temperature,
residence time, and steam-to-ethane ratio, on yield and composition of the
cracked gas.

Keywords: Aspen Plus, Ethane thermal cracking, Kinetics, Molecular reaction scheme,
Pyrocracker
Received: October 09, 2011; revised: December 07, 2011; accepted: February 07, 2012
DOI: 10.1002/ceat.201100529

1 Introduction quenched by passing the gas through a set of coolers to arrest

the reaction. The primary goal of this process is to maximize
Typically, pyrolysis or steam cracking is employed for the the yield of olefins and minimize the rate of coke formation
conversion of ethane to ethylene using tubular reactors [1]. by operating the pyrocracker under optimum process condi-
During this cracking process, feedstock ethane is diluted with tions. Reducing the rate of coke formation would ensure a
steam and preheated for a short time in the convection sec- longer duration of the continual run of the furnace, without
tion of the pyrocracker. Steam is used as a diluent to decrease being shut for decoking.
the partial pressure of the cracked gas components and there- Due to the numerous technical and commercial difficulties
by increase the yield of the desired products. The heated mix- involved in the operation of an industrial pyrocracking unit,
ture then enters the radiant section of the furnace that is any investigation focusing on process development has to rely
designed and assembled as a set of coiled tubular reactors, on various simulation packages. The goal of this study is to de-
also termed as radiant coils. Combustion of natural gas in the velop a process model using Aspen Plus that would provide
fireboxes and burners provides the necessary heat flux to the qualitative and quantitative details on the effect of various pro-
radiant coils so as to raise rapidly the reactants flowing inside cess parameters on ethane conversion, and thus serve as a han-
the coils to the desired temperature. The ideal temperature of dy reference to the engineers and operators working with in-
the cracking reaction may vary between 800 K and 1400 K, dustrial units. One of the main components of a process
depending upon the presence or absence of any catalysts. model is the accurate representation of a kinetic reaction
Normally, thermal cracking necessitates carrying out the scheme describing the ethane cracking mechanism. There are
pyrolysis reaction at elevated temperatures for a short span of few reports in literature focusing on simulation of ethane
time so as to reduce the formation of coke. The heat flux to cracking kinetics to predict the product distribution at differ-
the radiant coils is manipulated and controlled such that the ent process conditions [1–4].
coil outlet temperature (a representative of the product The accuracy of predicted product distribution is highly de-
stream temperature) is maintained at the desired value. Inside pendent on the kinetic reaction scheme utilized in the simula-
the radiant coils, the feedstock is cracked to a combination of tion. Different molecular and radical reaction schemes for eth-
various lightweight hydrocarbons of varying proportions. The ane pyrolysis have been proposed and validated for this
cracked gas exiting the radiant coils may undergo secondary purpose. The major fraction of earlier work on ethane pyroly-
reactions resulting in formation of undesired products. sis has been at temperatures until 1173 K, and the mechanisms
Hence, immediately upon exiting, the cracked gas is rapidly that have been proposed are mostly radical mechanisms. Radi-
cal reactions are more significant in predicting the rate of coke
– buildup inside the reactors [2]. Ethylene and aromatics are be-
Correspondence: Dr. P. Kannan (pkannan@pi.ac.ae), The Petroleum lieved to be the precursors of coke formation via secondary re-
Institute, Department of Chemical Engineering, P.O. Box 2533, actions [5]. Although the ethane pyrolysis process proceeds via
Abu Dhabi, United Arab Emirates. free radical mechanisms [6], molecular schemes are more

Chem. Eng. Technol. 2012, 35, No. 00, 1–6 © 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.cet-journal.com
2 P. Kannan et al.

commonly used for design and simulation purposes, mainly and the resultant volume of the PFR were equivalent to the to-
due to its simplicity in computation. One such molecular reac- tal volume of the radiant coils of the industrial reactor.
tion scheme obtained by approximating the free-radical nature One of the most important specifications of the PFR module
of ethane cracking was proposed by Sundaram and Forment is the reactor temperature condition. The molecular reaction
[2] which is considered to be the most accurate and widely scheme proposed by Sundaram and Forment [2] was utilized
adopted. However, as shown later in this report, the model in the base case simulation for assessing the effect of the tem-
proposed by Sundaram and Forment fails to consider certain perature profile. Because of the unavailability of an actual in-
significant components present in the product gas, and thus dustrial temperature profile, simulations were tested against
does not provide satisfactory results when compared with in- two different temperature profiles reported in the literature.
dustrial data. Since the PFR dimensions were different, a nondimensionali-
This study focuses on developing a more comprehensive zation of the temperature and the tube length was performed
ethane thermal cracking model via an improved kinetic molec- using the formula: T* = (T – T0)/(Tmax – T0) and L* = l/L,
ular scheme using Aspen Plus. The model was utilized to ana- where T* is the nondimensionalized temperature, Tmax is the
lyze the effects of various process variables on the overall com- maximum temperature of the reactor, T is the localized tem-
position of product distribution and the yield of ethylene. perature, T0 is the inlet temperature, L is the length of the reac-
tor, L* is nondimensionalized length, and l is the position at
which temperatures were measured.
2 Model Development Fig. 1 presents a comparison of the two estimated tempera-
ture profiles reported by Gao et al. [7] and Gujarathi et al. [3].
The following assumptions have been made in the develop- Simulations were carried out using these two nonlinear pro-
ment of the ethane cracking model: files in addition to an isothermal profile and the results were
(1) Effects due to the bending and other hydrodynamic pa- compared with the available industrial data, see Tab. 1.
rameters were neglected, and the entire set of coils was For all the comparison studies performed in this work, only
modeled as a single plug-flow reactor (PFR). the composition of the major components, including hydro-
(2) No chemical reactions occur prior to the entry of the reac- gen, methane, ethylene, and ethane, were considered. As it can
tant into the radiant section (i.e., before the PFR). be noticed, the temperature profile of Gao et al. yields relative-
(3) Only molecular species are considered in the product list; ly better predictions than the profile reported by Gujarathi et al.
free radicals and their corresponding reactions are not in-
cluded in the simulation.
(4) Since formation of coke is not considered, the reactor core
temperature was assumed to be the same as the reactor
wall temperature.
(5) Steam diluent in the feed is assumed to be inert and reac-
tions involving steam are not considered.
The process in the radiant coils of the pyrocracker could be
well described by means of a PFR model. Aspen’s built-in PFR
module was selected for this purpose. The process conditions
used in the current model are similar to an industrial reactor
from where data has been collected for comparison. Industrial
data was provided by an ethylene production plant that utilizes
ethane as the primary feedstock. The process was carried out
in a furnace, also called a firebox, of about 11 m height, 11 m
length, and 2.5 m width. The firebox houses six sets of radiant Figure 1. Various pyrocracker temperature profiles reported in
coils, with each set consisting of eight inlet tubes, thus literature.
amounting to a total of 48 tubes. The
industrial feedstock was a mixture of
100 % pure ethane and steam fed into Table 1. Comparison of simulation and industrial data for various temperature profile condi-
the reactor at a preheated temperature tions inside the reactor.
of 943 K and at a pressure of 2.2 bar.
Products Product yield [mol.-%]
Under normal conditions, the flow ra-
–1 Simulation Simulation Simulation Industrial data
tio was 0.3 kg h of process steam per
–1 with Gao et al. with Gujarathi et al. with isothermal
1 kg h of ethane. For the base case
simulation, an ethane mass flow rate of temperature profile temperature profile profile
–1
29.6 t h , adhering to a steam-to-eth- H2 31.48 27.60 40.67 36.49
ane ratio (kg kg–1) of 0.3, was em-
ployed. The configuration of PFR in- CH 4 2.27 1.71 3.32 8.42
cluded a tube length of 10.5 m, inner C2H4 29.79 26.26 37.69 33.07
diameter of 0.085 m, and a total of 48
C2H6 34.64 42.99 15.11 18.86
tubes. The dimensions, configuration,

www.cet-journal.com © 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eng. Technol. 2012, 35, No. 00, 1–6
 Pyrocracker 3

in comparison with industrial data. It could Table 2a. Improved molecular reaction scheme for ethane cracking. Kinetic data for re-
be observed from Fig. 1 that in the profile of actions (1), (2), (3), (5), (8), and (10) taken from Sundaram et al. [2] and the rest from
Gao et al. the temperature remains close to Gao et al. [7].
the coil outlet temperature for most parts of
No. Reactions Pre-exponential constant Activation energy
the reactor while in the profile reported by [s–1] or b[cm3mol–1s–1] [J mol–1]
Gujarathi et al. the temperature profile shows
a continuously increasing trend and reaches 1 C2H6 > C2H4 + H2 4.6 · 1013 272 838
the coil outlet temperature (COT) only at the 2 C3H6 > C2H2 + CH4 7.2 · 1012
274 178
exit of the reactor. Hence, based on the above
trend, it can be expected that an isothermal 3 C2H2 + C2H4 → C4H6 1.0 · 1015 b
172 632
temperature profile that represents an aver- 4 C2H4 + C4H6 → C6H6 +2H2 8.3 · 1012 b
144 599
aged value of the nonlinear profile and
5 C3H8 > C3H6 + H2 5.8 · 1010
214 597
equivalent to COT, might achieve results
closer to industrial data. As expected, the iso- 6 C3H8 + C2H4 → C2H6 + C3H6 2.5 · 1016 b
247 107
thermal profile yields relatively better results
7 2C3H6 → 3C2H4 7.3 · 1012 268 487
than the other two nonlinear profiles
(Tab. 1). However, it should be noted that 8 2C2H6 → C3H8 + CH4 3.8 · 1011 273 006
the reactor temperature is expected to vary 9 C4H10 > C4H8 + H2 1.6 · 1012
260 914
along the length of the reactor as it is gov-
erned by the rate of reaction coupled with 10 C2H4 + C2H6 → C3H6 + CH4 7.0 · 1016 b
252 839
the rate of heat transfer from the furnace. In 11 C3H6 + C2H6 → C4H8 + CH4 1.0 · 1017 b
251 082
industries, COT serves as the single reference
to ensure product quality and is maintained
constant by manipulating the heat flux (temperature) from the Table 2b. Reverse reaction rate constants [7].
furnace. Nevertheless, for the current study, a steady-state
model with an isothermal condition seems satisfactory for all No. Reverse reaction rate constant
sensitivity studies. [L mol–1s–1]; E [kJ mol–1]
The components present in the product stream comprises of 1 KR = 8.49 · 108; 10(–136.5/RT)
hydrogen, methane, C2H2, C2H4, C2H6, C3H6, C3H8, and C4+
that includes C6H6, C4H6, and C4H8. However, the scheme 2 KR = 3.81 · 108; 10(–147.2/RT)
proposed by Sundaram and Forment [2] does not account for 5 KR = 9.03 · 105; 10(–93.5/RT)
the formation of these higher-molecular-weight components.
9 KR = 1.78 · 107; 10(–135.1/RT)
Hence, an improved reaction scheme along with the apparent
kinetic parameters taken from different sources has been pro-
posed for pyrolysis of ethane that includes all the components
mentioned above. The scheme involves the original five reac- tion scheme only includes the cracking reactions, but does
tions proposed by Sundaram and Forment along with six not consider the stream reforming reaction, C2H6 + 2 H2O
other additional reactions accounting for the formation of → 5 H2 + 2 CO. If it were to be argued that hydrogen is
benzene via reaction (4), formation of propylene via reaction primarily formed via steam reforming reaction, then it
(6), cracking of ethylene via reaction (7), formation and crack- should have resulted in proportional quantity of CO. How-
ing of n-butane via reaction (9), and formation of butane ever, it should be noted that the composition of CO in the
and methane via reaction (11). The kinetic parameters for re- effluent stream of the industrial reactor was found to be
actions (1), (2), (3), (5), (8), and (10) were taken from only ~ 143 mol ppm–1. Hence, it can be concluded that the
Sundaram and Forment, while the kinetic parameters for all produced hydrogen could only stem from ethane cracking
other reactions were taken from recent
literature (Tab. 2).
Tab. 3 compares the simulated and Table 3. Comparison of simulation and industrial data using an improved reaction scheme
industrial data obtained using Sun- and the Sundaram and Forment reaction scheme [2].
daram’s and Forment’s reaction
Products Product yield [mol.-%]
scheme and the improved scheme
proposed in this work. The following Simulation Simulation Industrial data
process conditions were employed to with Sundaram et al. with improved reaction
obtain the simulation data: ethane reaction scheme scheme
flow rate = 30 t h–1, steam = 10 t h–1, H2 40.67 36.79 36.49
and isothermal condition. As it can
be seen, the improved scheme is able CH4 3.32 3.83 8.42
to predict the product composition C2H4 37.69 34.34 33.07
better due to the additional reactions
C2H6 15.11 22.36 18.86
incorporated in the scheme. The reac-

Chem. Eng. Technol. 2012, 35, No. 00, 1–6 © 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.cet-journal.com
4 P. Kannan et al.

reaction, as rightly indicated in the proposed molecular reac- 3.2 Effect of Feedstock Flow Rate
tion scheme.
In order to study the effect of the flow rate on product compo-
sition, the reactor length was varied instead of altering the feed
3 Sensitivity Analysis flow rate as both variations of the flow rate or the reactor
length would only result in variation of residence times of the
In industrial operation of the pyrocracker, the COT, steam-to- reactants inside the reactor. The reactor length was modified as
hydrocarbon ratio, and feed flow rate/composition can be con- it involves fewer process parameter manipulations in the simu-
sidered as the significant independent process variables. Since lation compared with varying the feedstock flow rate. At a
the current work is specific to an ethane feedstock, the influ- fixed COT of 1123 K, simulations were performed for different
ence of feed composition is not investigated. For the sensitivity reactor lengths ranging from 1 to 15 m. For a feed flow rate of
analysis presented in the following section, the base case model 9.5 m3s–1 and a tube diameter of 0.085 m, the equivalent resi-
assumes an isothermal temperature profile and the improved dence time is between 0.02 to 0.42 s. The results of simulation
molecular reaction scheme, while all the other process condi- runs illustrating the variation of product distribution as a
tions, including flow rates, remain unchanged. function of reactor length are displayed in Fig. 3. It can be no-
ticed that there is a pronounced variation in the product com-
position at shorter reactor length, i.e., at short residence times
3.1 Effects of the Coil Outlet Temperature up to 0.3 s. As the reactor length increases above 10 m, there is
no change in the product composition including that of ethyl-
The COT is defined and measured as the temperature of the ene and hydrogen. Also, at larger residence times, the rate of
product gas exiting the radiant coils upon cracking operation. ethane cracking also seems to decrease and is almost zero
It plays a very important role in determining the composition above 15 m (0.45 s). This is possibly due to the fact that the
of the cracked gas. Simulations were performed at different re- process reaches an equilibrium state which is evidenced by the
actor temperatures ranging from 923 to 1373 K while keeping constant product composition (Fig. 3). The results are very
all the other process parameters same as the base case. The much in agreement with the reported values for residence
variation of product yields obtained through simulations as a times in cracking furnaces. A typical value of mean residence
function of COT is illustrated in Fig. 2. It could be observed time of the reactant mixture in the radiant section is very short
that the ethane conversion level increases with rising tempera- and is in the order of 0.3 s [7]. From Fig. 3 it can also be con-
ture and simultaneously increased levels of ethylene and cluded that hydrogen composition always remains slightly
hydrogen, indicating the dominance of reaction (1) over all higher than that of ethylene, possibly due to the multiple reac-
other reactions. This is in very good agreement with the results tions that lead to the formation of hydrogen.
reported by Masoumi et al. [1] where a maximum ethylene
yield of ca. 33 % was attained at 1153 K. However, above
1173 K, the composition of ethylene and hydrogen in the
product stream seems to decrease due to the increased forma-
tion of methane. The compositional trend of the products,
especially that of ethylene, illustrates that the upper bound
limit for COT must be around 1173 K. At this temperature, the
ethylene yield in the reactor effluent attains a maximum of
39 %. Therefore, the ethane cracking temperature in excess of
1173 K is not reasonable for technological and economic con-
siderations.

Figure 3. Effect of reactor length on product distribution.

3.3 Effect of Steam-to-Ethane Ratio

As mentioned earlier, the purpose of steam dilution is to de-

crease the partial pressure of the product hydrocarbons, there-
by shifting the equilibrium in a direction favorable for the in-
creased yield of products. Simulations were performed at
varying steam-to-ethane ratio ranging from 0.1 to 1 at a fixed
COT of 1123 K. Fig. 4 illustrates the variation of product com-
position as a function of the diluent ratio. It can be observed
that the combined yield of ethylene and hydrogen attains a
Figure 2. Effect of COT on product distribution. maximum when the steam-to-ethane ratio is between 0.3 and

www.cet-journal.com © 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eng. Technol. 2012, 35, No. 00, 1–6
 Pyrocracker 5

relatively better results which compared well with industrial

data. Furthermore, it is conclusive that the simulation results
of the present work using the improved reaction scheme is able
to match well with the industrial data. The effect of several
process parameters, including COT, feed flow rate, and steam-
to-ethane ratio, on the product composition was investigated
using the validated model. Some of the conclusions from the
sensitivity analysis include: (i) The yield of ethylene from the
cracking process reaches a maximum at a coil outlet tempera-
ture of about 1173 K. (ii) Maximum ethane conversion attains
at residence times of ca. 0.3–0.4 s, beyond which it remains un-
affected. (iii) The combined yield of ethylene and hydrogen at-
tains a maximum when the steam-to-ethane ratio is between
0.3 and 0.35.
The capability of the model for predicting the optimum
Figure 4. Effect of steam-to-ethane ratio on product distribution. conditions needed for the cracking unit has been successfully
demonstrated and the results can potentially have a direct im-
0.35. Under conditions of increased steam content (> 0.4 ra- pact on the industrial pyrocracker economics. Furthermore,
tio), the conversion of ethane seems to be drastically affected the modeling and simulation methodology demonstrated in
by steam dilution, indicated by an increased ethane content in this work could be easily interpreted and reproduced for future
the effluent composition. Hence, increasing the steam-to-eth- process evaluation studies.
ane ratio above 0.35 is not favorable for both technical and
economic operations of the plant.
References

4 Conclusions [1] M. E. Masoumi, S. M. Sadrameli, J. Towfighi, A. Niaei,

Energy 2006, 31, 516.
A simplified process model of an industrial ethane cracker was [2] K. M. Sundaram, G. F. Froment, Chem. Eng. Sci. 1977, 32,
developed and simulated using Aspen Plus. The radiant sec- 601.
tion of the pyrocracker, where the major cracking reactions [3] A. M. Gujarathi, D. S. Patle, P. Agarwal, A. L. Karemore, B. V.
take place, has been modeled as series of plug-flow reactors. Babu, CHEMCON-2009, IIChE, Visakhapatnam, December
The model includes an improved molecular reaction scheme 2009.
with ten components and fifteen reactions, which seems to be [4] Z. Belohlav, P. Zamostny, T. Herink, Chem. Eng. Process.
a better representation of the overall ethane cracking kinetics 2003, 42, 461.
than any of the previously reported models. Since the actual [5] R. Zou, Q. Lou, S. Mo, S. Feng, Ind. Eng. Chem. Res. 1993,
temperature data within the radiant coils could not be mea- 32, 843.
sured, various temperature profiles reported in literature were [6] K. J. Laidler, Chemical Kinetics, 2nd ed., McGraw-Hill, New
examined using the developed model. The simulated results York 1965.
were compared with the available industrial data to validate [7] A. Tarafder, B. C. S. Lee, A. K. Ray, G. P. Rangaiah, Ind. Eng.
the model, and it was found that an isothermal profile yielded Chem. Res. 2005, 44, 124.

Chem. Eng. Technol. 2012, 35, No. 00, 1–6 © 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.cet-journal.com
6 P. Kannan et al.

Research Article: A simplified process Modeling of Ethane Thermal Cracking

model of an industrial ethane cracker Kinetics in a Pyrocracker
was developed. The capability of this
model for predicting the optimum P. Ranjan, P. Kannan*, A. Al Shoaibi,
conditions for the cracking unit has C. Srinivasakannan
been successfully demonstrated. The
results potentially have a direct impact Chem. Eng. Technol. 2012, 35 (䊏),
on the industrial pyrocracker economics XXX … XXX
and could serve as a handy reference to
engineers and operators working with DOI: 10.1002/ceat.201100529
industrial units.

www.cet-journal.com © 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eng. Technol. 2012, 35, No. 00, 1–6