Comparison of gas phase mechanisms for
the prediction of coke deposition during thermal cracking of light hydrocarbons
A.Y. Ramírez1, L.O. Almanza2,
J. C. Vivas2, M. Kraft3, A. Molina1
Facultad de Minas, Universidad Nacional de
Colombia - Sede Medellín
2 Instituto Colombiano del Petróleo,
ICP
2 Chemical Engineering and
Biotecnology departemt. Univeristy of Cambridge
Thermal cracking of light hydrocarbons is the
main route for the production of important raw materials for the chemical
industry, such as ethylene and propylene. The current and most used technology
for olefin production involves injection of a mixture of hydrocarbons,
preferably ethane, into a long tubular coil (around 80m long) located in a
furnace with multiple burners that provide the required energy for the highly
endothermical cracking reactions. Steam is added to the hydrocarbon mixture at
a ratio (known as dilution factor) in order to control the reactive flow
temperature and reduce the partial pressure of hydrocarbons,
increasing the forward rate of reaction preferential to light olefins[1, 2] .
An undesirable effect during
thermal cracking is coke deposition on the walls of the tubular reactor. Coke
deposits build with reactor operation time and increase up to a point in which
the reduction in heat transfer across the reactor's wall is so high that
external skin coil temperature needs to be significantly increased to maintain
a constant heat flux to the reactor. Clearly this decreases the thermal
efficiency. This
coke layer leads to a higher pressure drop over the reactor which is a very
undesirable situation because it affects product selectivity. When the pressure drop along the
reactor and the reduction on heat transfer across the reactor's wall are too
high, the furnace operation is interrupted and a decoking operation is
conducted in
which the coke is burned off with a controlled air/steam mixture [3?5]. The operational time before decoking is
of the order of 20 to 90 days, depending on process conditions and load.
Simulation, by a reliable
model, of coke deposition in the thermal cracking furnace for different inlet
conditions is necessary if one wants to understand the effects that changes in
process conditions and raw materials have on process performance. A typical
model for coke deposition includes two independent submodels: one that
considers cracking of steam/hydrocarbons mixtures in gaseous phase and a second
one that predicts coke deposition. Both models have to be integrated in order
to simulate olefins production and the reduction in the diameter during thermal
cracking of light hydrocarbons [1, 6]. This paper deals with the selection of
a proper gaseous phase model that renders the information required to properly
construct a coke deposition model.
In the refereed literature,
the seminal work of Sundaram and Froment is recognized as one of the first
studies on this area. These authors proposed using a global mechanism with 5
and 10 reactions to model the thermal cracking of ethane and ethane/propane
mixtures respectively [3], [4]. After this first approximation, they
proposed a radical
reaction scheme which has
110 reactions and describes the
pyrolysis of ethane, propane, isobutene, ethylene, propylene and n-butane as
well as their mixtures [7] .
Subsequent to Sundaram and
Froment's work, Ranzi and collaborators developed a model which takes into
account elemental reactions for the thermal cracking of light hydrocarbons such
as ethane, propane and propylene as well as heavier feedstocks such as naphtha[8, 9]. The mechanism for light hydrocarbons takes
into account 85 species and 1351 reactions and includes hydrocarbon fuels up to
3 C atoms. The work by Ranzi and collaborators finally lead to the SPYRO code,
currently used in the hydrocarbon industry to predict thermal cracking. This
code is currently licensed by Pyrotec, a divison of Technip [10].
Among the numerous kinetic
mechanisms developed to describe the combustion and pyrolysis of hydrocarbons,
it is important to analyze those which are relevant to pyrolysis of light hydrocarbons,
particularly those developed by and Frenklach et al [11, 12] and Wang et al. [13, 14].
The kinetic mechanism of
Frenklach et al. (ABF mechanism) consists of 99 chemical species and
531reactions [11, 12] . It includes the pyrolysis
and oxidation of C1 and C2 species. One of the most
important things of this mechanism is that it goes until de formation of pyrene
as the higher-weight aromatic compound. This is a good characteristic because pyrene,
as benzene, has been widely used to start the mechanism to describe solid phase
formation of soot and coke.
The kinetic mechanism
described by Wang et al. (USC mechanism) has been changed during the last
decade in more than four occasions. The last published mechanism takes into
account the combustion of H2/CO/C1-C4 and the
description of pyrolysis and combustion at high temperature of normal alkenes
up to n-dodecane. It has as the most complex aromatic compound benzene [13, 14] .
This paper addresses the differences
between all these mechanisms under the light of developing a model to predict coke
deposition as final goal. The predictions with the mechanisms by Sundaram and Froment (global [3, 4] and detailed [7] ), Ranzi et al. [8, 10], Frenklach et al. [11, 12]and Wang et al. [13, 14] were compared with measurements
from an ethane cracker operator.
While the global and detailed mechanisms
by Sundaram and Froment give good prediction of the high-concentration species,
they yield a poor prediction of the low-concentration species and do not
consider key species important to model coke deposition. The ABF mechanism correctly
predicts species concentrations but the highest-molecular weight species it
considers is ethane. In the case of Ranzi's and USC, both mechanisms consider
propane pyrolysis and correctly predict major and minor species. However,
predictions with the USC mechanism are in better agreement with industrial data
and give more insight into some precursors for coke formation. Table 1 shows
the comparison in the prediction of the gaseous species as a ratio between
model predictions and industrial data.
The results suggest that from
the mechanisms available in the open literature, the USC is the one that will
be more suitable to be coupled with a solid-phase model for the prediction of
coke deposition during the pyrolysis of light hydrocarbons.
REFERENCES
[1] G.
P. Froment, B. O. de Steene, P. S. Van Damme, S. Narayanan, and A. G. Goossens,
?Thermal Cracking of Ethane and Ethane-Propane Mixtures,? Industrial &
Engineering Chemistry Process Design and Development, vol. 15, no. 4, pp.
495-504, 1976.
[2] B. L. Crynes,
L. F. Albright, and L.-F. Tan, ?Thermal Cracking,? in Encyclopedia of
Physical Science and Technology (Third Edition), Third Edit., E.-in-C. R.
A. Meyers, Ed. New York: Academic Press, 2003, pp. 613-626.
[3] G. Froment,
K. Sundaram, and P. S. Van Damme, ?Coke deposition in thermal cracking of
ethane,? AIChE journal, vol. 27, no. 6, pp. 946-951, 1981.
[4] K. Sundaram
and G. F. Froment, ?Kinetics of coke deposition in the thermal cracking of
propane,? Chemical Engineering Science, vol. 34, no. 5, pp. 635-644,
1979.
[5] L. F.
Albright and J. C. Marek, ?Coke formation during pyrolysis: roles of residence
time, reactor geometry, and time of operation,? Industrial & Engineering
Chemistry Research, vol. 27, no. 5, pp. 743-751, 1988.
[6] L. F.
Albright and J. C. Marek, ?Mechanistic model for formation of coke in pyrolysis
units producing ethylene,? Industrial & Engineering Chemistry Research,
vol. 27, no. 5, pp. 755-759, 1988.
[7] K. M.
Sundaram and G. F. Froment, ?Modeling of Thermal Cracking Kinetics. 3. Radical
Mechanisms for the Pyrolysis of Simple Paraffins, Olefins, and Their Mixtures,?
Industrial & Engineering Chemistry Fundamentals, vol. 17, no. 3, pp.
174-182, 1978.
[8] M. Dente, G.
Bozzano, T. Faravelli, A. Marongiu, S. Pierucci, and E. Ranzi, ?Kinetic Modelling
of Pyrolysis Processes in Gas and Condensed Phase,? in Chemical Engineering
Kinetics, vol. 32, G. B. Marin, Ed. Academic Press, 2007, pp. 51-166.
[9] M. Dente, S.
Pierucci, E. Ranzi, and G. Bussani, ?new improvements in modeling kinetic
schemes for hydrocarbons pyrolysis reactors,? Chemical Engineering Science,
vol. 47, no. 9?11, pp. 2629-2634, 1992.
[10] M. Dente, E.
Ranzi, and A. G. Goossens, ?Detailed prediction of olefin yields from
hydrocarbon pyrolysis through a fundamental simulation model (SPYRO),? Computers
& Chemical Engineering, vol. 3, no. 1?4, pp. 61-75, 1979.
[11] M. Frenklach,
J. Appel, and H. Bockhorn, ?ABF mechanism.? [Online]. Available:
http://www.me.berkeley.edu/soot/mechanisms/abf.html.
[12] J. Appel, H.
Bockhorn, and M. Frenklach, ?Kinetic modeling of soot formation with detailed
chemistry and physics: laminar premixed flames of C2 hydrocarbons,? Combustion
and Flame, vol. 121, no. 1?2, pp. 122-136, 2000.
[13] H. Wang, X.
You, A. V. Joshi, S. G. Davis, A. Laskin, and F. Egolfopoulos, ?USC Mech
Version II. High-Temperature Combustion Reaction Model of H2/CO/C1-C4
Compounds.? [Online]. Available: http://ignis.usc.edu/USC_Mech_II.htm.
[14] H. Wang and F.
Egolfopoulos, ?JetSurf-A Jet Surrogate Fuel Model.? [Online]. Available:
http://melchior.usc.edu/JetSurF/.