Академический Документы
Профессиональный Документы
Культура Документы
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/244737216
CITATIONS READS
31 2,469
3 authors:
Aligholi Niaei
University of Tabriz
118 PUBLICATIONS 1,167 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by S.M. Sadrameli on 22 September 2014.
The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document
and are linked to publications on ResearchGate, letting you access and read them immediately.
Journal of Chemical Engineering of Japan, Vol. 35, No. 10, pp. 923-937, 2002 Journal Review
Keywords: Coke Formation, Coke Inhibition, Hydrocarbons Pyrolysis, Cracking Coil, Kinetics
Research has been carried out to understand the mechanisms under which coke formation occurs and
to search for solutions to reduce or eliminate coke deposition. Significant efforts bave been exerted over
the past twenty years in developing coke-inhibiting methods. Coke inhibitors, i.e., chemical additives, or
special coating of metal surfaces which suppress coke formation. Coke inhibitors/surface coating work
by passivating catalytically active metal sites through cbemical bonding interactions, and/or forming a
thin layer to physically isolate the metal sites from coke precursors in a process stream, and/or interfer
ing with those radical reactions leading to coke formation by blocking active radical sites on surfaces.
Tbis review paper investigates the effect of the most important methods such as new chemical additives,
tube material construction, pretreatment of surface on the coke formation of pyrolysis furnaces.
L Coke Precursor
Metal particle on
Metal particle on
* Grain boundaries
~.~.~~.--- \
Metal crystalline
Metal oxides or
5-Carbon Diffusion
I 2-Sitc adsorption '> 6-Nucleution and
I ,"i-Strrface reactions \
carbon forma:ion ..
pyrolysis of hydrocarbon feed. The radical reaction accumulation in the particle causes a pressure buildup
mechanisms of pyrolysis of normal and branched par at the dislocations and the grain boundaries, which may
affins (Murata and Saito, 1973; Dente and Ranzi, 1983; exceed the tensile strength of the metal. The metal
Ranzi etal., 1983. 1997). heavy liquid feeds (Depeyre partic Ie is then Iifted from the surface and carbon
et at., 1989: Dente et al., 2001), naphtha (100 et al., crystallizes at the end of the particle. A growing carbon
200 I), substituted aromatic hydrocarbons (Chen and stem thus develops which carries the crystallite at its
Froment, 1991a, 1991b) have been investigated. top. The precipitation of carbon can give rise to
Recently a complete reaction, rigorous kinetic structural deficiencies in the carbon lattice, thereby
model, for the decomposition of the hydrocarbon feed creating reactive carbon centers along the filament skin.
(ethane to naphtha) has been prepared by Towfighi and Hydrocarbon radicals and molecules from the gas phase
Karimzadeh (1993) and Towfighi et al. (1993, 1998b, are incorporated at these reactive sites, whereby lateral
2002a). This reaction network contains more than 1000 growth of the filaments occurs. In this way a porous
reactions for 91 species such as initiation reactions, layer of interwoven filaments is formed. Carbon
H -abstractions, radical additions, decomposition reac migrating over the metal surface precipitates on the
tions, isomerization reactions, termination reactions, metal surface surrounding the carbon stem. Surface
and Diels-Alder cyclizations. carbon may thus be formed, encapsulating the metal
Many authors however, have developed molecu and preventing further growth.
lar kinetic models for pyrolysis of hydrocarbons. The The properties of the metal are very important in
pyrolysis of ethane, propane, butane (Sundaram and this mechanism. Encapsulation of the metal particles
Fromen!, 1977a, 1977b; Bozga et al., 1988; Towfighi reduces the rate of dehydrogenation of the chemisorbed
et al., 1994b, 1998a; Owtadi and TowfighL 2000), hydrocarbon atoms. At this stage the catalytic activity
naphtha (Kumar and Kunzru, 1985a; Parameswaran et of the metal particle diminishes and both carbon for
al., 1988; Niaei and Towfighi, 1993; Towfighi and mation and CO production slow down (Albright and
Shokuhi, 1994; Towfighi et al., 1994a; Towfighi and McConnel, 1979; Trimm and Turner, 1981; Baker et
Sharfi, 2000; Sadrameli et al., 2000; Sadrameli and al., 1982; Albright and Marek, 1988a, 1988b, 1988c;
Towfighi, 2(00), cyclohexane (Pant and Kunzru, 1997), Francoise et al., 1995).
gas oil (Zou et al., 1993; Depeyre et al., 1989) have 1.2.2 Radical coking mechanism At the gas/coke
been in~estigated and published. interface layer, the polyaromatics are not completely
1.2 Mechanisms of coke formation dehydrogenated. At this surface, hydrogen abstraction
The three mechanisms by which cokes produced reactions by free radicals from the gas phase can oc
in the coil and transfer line exchanger (TLE) of a crack cur. Hydrogen, methyl, and ethyl radicals are the most
ing furnace are reviewed as follows: active species. As a consequence, the concentration of
1.2.1 Catalytic coking mechanism Industrial the active sites at the coke surface becomes a function
cracking coils frequently contain different weight of the gas phase composition. This explains that the
percentage of chromium, nickel and iron. As shown in feedstocks generating more active radicals also yield
Fig. 1, during the startup of a furnace, the reacting gas more coke.
mixture is in contact \\lith the bare reactor walls. As shown in Fig. 2, at the free radical positions
Initially a hydrocarbon molecule is chemisorbed on a on the coke surface, certain gas phase molecules or
metal crystallite on the surface and by a surface reaction radicals (coke precursors) react via an addition mecha
convert to coke. Carbon atoms thus dissolve in and nism. All unsaturated molecules from the gas phase
diffuse through the metal panicle. The carbon are potential precursors. The long aliphatic side chain
~ Beta scission
) CondensatIOn j
/U,
I{~'.~
Dehydrogenates
/ 1 to form coke
~ Coke
Coke
Hydrogen abstnl.ction
. H
. "'~
0~'
~
+ H,
Fig. 3 Growth of coke by condensation mechanism
??
~ ~
vacuum gas oil. Condensation mechanism is generally
Dehydrogenation : i H
.. + , relatively unimportant at temperatures of700C or less.
h h
The exact temperatures depend to some extent on the
Coke
:l:::::a~~=:j
propylene are coke precursors, and ethylene is the more
intense precursor. Froment (1990) has reported the
o w a
U D ~ D ~
coking rate was correlated with respect to the concen Reactor length Iml
tration of a number of products such as C5+ olefins,
Fig, 4 Variation of product yields along the coil
diolefins. butadiene. butene. and aromatics, through a
first order rate equation. Significant information on
mechanisms and relative rates of coking was obtained
by Zimmerman and co-workers from studies using 14 C_ 0.014 ---800 hr
labeled components (Kopinke et al . 1988, 1993a, 0.012 ........... 1700 hr
\993b). Towfighi et at. (2002b) has reported the cok :! 0.01
ing rate was correlated with respect to the concentra
tion of olefins, diolefis and aromatics, through a power
j 0,008
~ 0.006
law rale equation.
~ 0.004
Olefin Research Group of Tarbiat Modares Uni V
0.002
versity achieved the calculation of the temperature and
product distribution in thc reactor with the help of an 0
0
6 12 18 24 30 36 42
accurate rigorous kinetic model for the decomposition Reactor length (m I
of the hydrocarbon feed (Towfighi el al., 2002a). Tak
Fig. 5 Variation of coke thickness along the coil at differ
ing into account the kinetics of coke formation
ent run length
(Towfighi ef at., 2002b). along with the simultaneous
simulation of the reactor and the firebox, provides a
detailed understanding of the behavior and an accu
rate prediction of the cracking furnace run length. The 1360 r------------------------------,
-O.Ohr
coupled furnace and reactor simulation can then be 1310 -800 br
optimized in order to obtain a favorable product dis 1260
tribution or to reduce unwanted side effects. such as ~
-;; 1110
fouling of the heat transfer surfaces by coke deposi S
tion. The test results of simulation in a sample indus i lIoO
trial naphtha cracking furnace as shown in Figs. 4-6, E
~ 1110
which is in good agreement with the industrial data.
The formation of coke in TLX is thought to be IMO~------__--------------------____J
o 10 20 30 40
occured by only two mechanisms. The first consists of Reactor length Iml
surface catalyzed reactions, and the second is the po Fig. 6 Variation of tube skin temperature along the coil at
lymerization of the pyrolysis products through unde different run length
sirable secondary reactions (Lahaye et al., 1977;
Sadrameli and Sharifnia, 1996). In contrast to the for
matitln of coke in cracking coils, only little is known
about the details of the formation of carbon rich prod catalyzed polymerization (Ranzi et al., 1985) or de
ucts which are deposited in the TLE tubes at tempera composition of unsaturated hydrocarbons at the inner
tures between about 350 and 600C. Events taken into surface of the TLE tubes followed by the thermal de
account are a mechanical transfer of coke like parti hydrogenation of the resultant polymers.
cles from the reactor coils into the TLE, a physical
condensation of the high boiling point cracked prod 2. Coke Reduction Techniques
ucts at the TLE walls (Kopinke ef al., 1993c; Batch et
al., 1995; Sadrameli el al .. 1999. 200l), a chemical In this review the effect of several techniques for
condensation of unsaturated cracked products from the reduction of cokc formation is investigated. For in
gas phase to droplets at temperatures above 350C, a stance, the effect of the most important new chemical
chemical interaction between the deposited products additives. alteration of the coil surfaces and the com
and reactive gas phase species, and even a metal position of feed on the coke formation in pyrolysis
furnaces is investigated.
DMDS: dimethyl disulfide: DMS: dimethyl sulfide: DES: diethyl disulfide: MM: methyl mercaptane:
EM: ethyl mercnptane; NBM: normal butyl mereaptane; CS2: carbon disulfide
naphtha feed, its sulfur content is used as a guide. Where C"H", (hydrocarbon) is considered for steam
Reeently contradictorial results were reported on reforming (Rostrup-Nielsen, 1973; Wagner and
the effect of sulfur containing additives on the coking Froment, 1992). Sulfur compounds can alter the rate
rate. Froment (1990) claimed that the CO yield is not of radical reactions through their interference with
a reliable indicator for the coke formation. Velenyi et hydrogen transfer and with termination reactions.
al. (1991) reported an increase of the coking rate on 2.1.2 Phosphorous compounds Phosphorous
Inconel 600 in pyrolysis of ethane with hydrogen containing additives have been shown to be effective
sulfide in the feed. Francoise er al. (1995) observed in inhibiting coking rates. Naberezhnova et al. (1983)
that addition of carbon disulfide results in an impor reported that the organophosphorous compounds were
tant reduction of the CO yield. both in the initial and very effective in controlling the coke formation dur
the asymptotic range. Sulfur compounds from the gas ing pyrolysis of liquid hydrocarbons without affecting
phase are more readily chemisorbed on the metal par the product yields. Boone (1983) of Nulco Company
ticles than water and hydrocarbon molecules. Cover reported the development of proprietary products con
age of metal sites by sulfur then leads to a lower con taining phosphorous with or without sulfur which were
centration of active oxygen atoms at the surface and effective in reducing the coking rate during
hence to a lower catalytic CO production. When S, and hydrocarbon pyrolysis. The additive provided a film
S, are empty sites on the surface of metal catalyst to passivate the metal surface to prevent it from
(cracking coil surfaces contain with high weight per catalyzing the coke formation. The increase in run
cent ofNi and Fe) and the carrier respectively, the fol length varies from 50% to 223%. Moreover, the steam
lowing sequence can be formulated for the adsorption air decoking time decreased by 5-10Ck and cleaning
of steam and formation of active oxygen atom (S,-O): of exchangers was easier because of softer coke
(Towfighi er al., 1999).
The effect of the addition of triethylphosphite
(TEP) on the rate of coke formation was investigated.
The reduction in the rate of coke formation was ini
tially higher and then gradually decreased such that, at
high concentrations of phosphous, the coking rate ap
proached a constant value (Ghosh and Kunzru, 1988).
- + S,- = S,-H,O
H,O -- Such a trend of coking rate with run time is well estab
lished in references (Kumar and Kunzru, 1985b;
Francoise et al., 1995). As can be seen from these re
sults, coking rates were significantly reduced in the
presence of triethylphosphite. This reduction in the
coking rate is comparable to the limited data reported
by Boone (1983). The results discussed above show
that the phosphorous neither takes part in the homoge-
~88;:lC while on steams only and holding for about 10 High temperature ductility
Cost.
1955-1960
1960-1965
1965-1975
1975-1990
1990-Today
Unfortunately, there is no simple assessment of These oxides thermally stable in high temperature,
properties that ensures max imum cost effectiveness for i.e. Crp3' SiO" and Alp) up IOS0, 1100, 11S0C re
the lifetime of a cracking coil. The properties of the spectively. Today the best industrial operating results
reactor surface affect the rate of coke formation. This and maximum effectiveness were fouild by use of Al
is because the physical and chemical properties of the loy 28/48 Wand Alloy 3S/4S. The presence of W im
reactor surface will influence the radical and surface proves carburization resistance; W plus Nb is better
reactions in pyrolysis of hydrocarbons. The formation than W alone. Synergistic additions of W, Nb, Ti and
of a stable, compact, protective oxide film is needed Zr greatly improve carburization resistance (Parks and
to minimize oxidation and carburization of coils. To Schillmoller, 1996).
proviJle adequate oxidation resistance, a minimum of 2.3.2 Effect of polishing By polishing alloy sur
2S% Cr is desirable. Higher Cr imparts a more stable face, there is a change in emmitance of the surface in
oxide film, but introduces instability in the alloy struc heat transfer by radiation. The emittance, which influ
ture. Cr. Ni and Si are effective barriers to carburiza ence radiation and therefore the temperature of the re
tion. High weight percentage are beneficial and provide actor surface, may vary from below 0.1 for a polished
a protective and regenerative oxide film. Oxides of surface up to 0.9 for a plain or a carbonized surface.
silicon, chromium, aluminum, manganese, and others Highly polished surface, with low emittances, oper
formed stable, compact and protective film and reduce ates as much as SO-100C below that of unpolished
catalytic coke by reducing the carbon diffusion rate or surface (Albright and McGill, 1986). Accordingly, re
solubility (Parks and Schillmoller. 1996; Wysiekierski duction of metal surface temperature significantly re
et al., 1999). The order of the stability of these oxides duces the carbon formation for no other reason than
are as below: the slower chemical reaction rate and phase changes
in solution equilibria that occur at these lower tem
Alp, > TiO > TiO~ > Si(\ > Crp3 > SiO > Mnp1 > peratures. Also the rate of carburization of alloys de
Mn 3 0 4 > FeO > FecO) > NiO > Cr01 crease by a factor of O.S for every 40C tube metal
..: 0
71 rms for the unpolished surface to 6 rms for the Ni HP HK Alonized HP
polished. Furthermore, if acti ve metal particles already
Fig. 10 Amount of coke deposition on various substrates
exist on the surface, they will be removed or reduced
in size and number by polishing. Indeed experimental
data showed that significant reduction in coke
deposited during pyrolysis of various feedstocks. The tube did not change.
ratio of coke formed on the unpolished surface to that Based on ED AX analyses, the aluminum content
formed on the polished varied from 5.6 for isobutane was in the range of 42 to 87 wt%.
up to 28.1 for ethane (Crynes and Crynes, 1987). The presence of high percentages of Cr, AI, Nb
One other consideration, polishing, is a change in and Si in the alonized surface.
chemistry of the surface, which is exposed, to the re The alonized tube has not shown any carburiza
acting gas. Not only the reaction chemistry but also tion and intergranular increase of carbon.
the phase behavior relating to carbon solution at the The non alonized coil by contrast had lost approxi
surface will change hy polishing. In reviewing other mately 20% of its wall thickness especially on the in
studies, polishing metah of different chemical com ner surface, because of carburization and subsequent
positions usually results in rates of carbon formation metal dusting.
similar to that of quart!, v. hich is generally considered The non-alonized tube is heavily carburized on
most inert. With lower ~urface temperature (radiation the surface.
effects), less mechanical surface defects, and possibly The increase of carbon along grain boundaries
a change in che111i~try, Ie~, carbon and an inactive car beneath the surface of non-alonized tube.
bon is created under the new conditions, and this pre In other industrial experience, has investigated the
venh or greatly reduce, filamentous carbon formation. effect of surface alonization on HP coils and similar
Therefore, with reduced filamentous carhon, the ac result is obtained for alonized HP surface. Results of
tive metal particles canl10t continue to be generated laboratory work in the pyrolysis system of ethane
and/or be exposed to further catalyze carhon forma (Albright and McGill, 1987) on HP substrate are shown
tion. in Fig. 10, and ethylene pyrolysis (Marek and Albright,
2.3.3 Effect of alonization of surface Albright 1982a) on Incolloy 800 substrate, have indicated the
and Yu (1979) investigated the effect of alonization on reduction in weight of coke by a factor of 6-7. The
coke formation phenomena. They ohtained valuable in fact that alonized coils operate at lower temperatures
formation by com pari ng the morphologies of coke than non-alonized coil is likely caused by at least two
formed on stainless steels by using a scanning elec factors. First, less coke is formed on the alonized
tron microscope (SEM) equipped with EDAX, from surfaces. Second, the thermal conductivity of the coke
different coke precursors in a pilot scale system. Par on the alonized surfaces is greater.
tiCularly noteworthy is the finding that for aluminized 2.3.4 Effect of silica coating Brown et ai. (1982)
(alonized) Incoloy 800 surfaces filamentous coke was found that silica coating process reduced the rate of
not ohserved and iron, nickeL and chromium were not coke formation in pyrolysis coils. Silica coating is pro
detected in the coke (Albright et ai., 1979; Albright duced on alloy steel substrates by the decomposition
and Yu, 1979: Graff and Alhright, 1982; Marek and of an alkoxysilane in a carrier gas. Briefly, an
Albright. 1982a; Albright and McGill, 1987). alkoxysilane (e.g. tetra etheyl orthosilicate) is injected
Results from industrial installations of alonized (I % by weight) into a preheated stearn flow, which
furnace coil composed of Incoloy 802 tubing and non passes into the coating vessel where steel substrates
alonized Incoloy 802 in a Gulf Canada Ltd. ethylene are radiantly heated to 600-850C. Silica films, largely
furnace, indicated that the al uminum rich layer remains amorphous in character, have been deposited with a
on allonized surface after 3 112 years and following wide range of thickness, as thin as 1 micron or less.
results were observed (Albright and McGilL 1987): Metallurgical examination of silica coated alloys
The alonized tube had not any loss of metal from showed no deleterious action on the bulk alloy. Indeed
the surface during the 3 112 years. silica coating has been found to protect the alloy sur
The roughness of the inner surface of the alonized face from oxidative attack in high temperature steam