+ All Categories
Home > Documents > Burklé-Vitzthum Etal 2011

Burklé-Vitzthum Etal 2011

Date post: 16-Feb-2018
Category:
Upload: geochem1985
View: 220 times
Download: 0 times
Share this document with a friend

of 12

Transcript
  • 7/23/2019 Burkl-Vitzthum Etal 2011

    1/12

    Thermal evolution ofn- and iso-alkanes in oils. Part 1: Pyrolysis model for amixture of 78 alkanes (C1C32) including 13,206 free radical reactions

    V. Burkl-Vitzthum a,, R. Bounaceur a, P.-M. Marquaire a, F. Montel b, L. Fusetti b

    a Laboratory of Reactions and Process Engineering, LRGP CNRS-UPR 3349, Nancy University, ENSIC, BP 20451, 54001 Nancy, Franceb TOTAL Exploration and Production, 64018 Pau Cedex, France

    a r t i c l e i n f o

    Article history:

    Received 3 January 2011

    Received in revised form 11 March 2011

    Accepted 21 March 2011

    Available online 25 March 2011

    a b s t r a c t

    A mechanistic model consisting of 13,206 lumped free radical reactions has been developed to describe

    the thermal evolution of a mixture of 78 alkanes: all n-alkanes from C1 to C32 and 46 branched alkane

    model compounds from C4 to C32. The mixture was meant to represent the major part of the saturated

    fraction of petroleum. The rate constants used are available from the literature. The lumping together

    procedure is described and the model validated on the basis of several experimental results from the lit-

    erature and relating to pure alkanes. The model is also compared to the saturated fraction obtained from

    pyrolysis of Elgin oil at 372 C for up to 1000 h. The cracking global activation energy ofn-C15as well as

    iso-C15is close to 69 kcal/mol in the range 200350 C. The implications of the model for geological res-

    ervoirs will be discussed in a following paper.

    2011 Elsevier Ltd. All rights reserved.

    1. Introduction

    In exploration, the composition of petroleum in deeply buried

    reservoirs (T> 200 C) is of strategic interest. Questions regarding

    the availability of exploitable petroleum reserves in future decades

    can only be answered if the thermal stability of liquid reserves can

    be predicted under geological conditions. The thermal stability of

    petroleum is commonly modelled using kinetic parameters deter-

    mined from laboratory pyrolysis of whole crude oils (e.g. Ungerer

    and Pelet, 1987; Ungerer et al., 1988; Bhar et al., 1992, 1997a,b,

    2008; Schenk et al., 1997; Dieckmann et al., 1998; Lewan and Ru-

    ble, 2002; Lewan et al., 2006; Lehne and Dieckmann, 2007a,b) or of

    pure hydrocarbons and simple mixtures. The experiments typically

    use higher temperatures (300500 C) than those encountered in

    reservoir rocks (80200 C), to compensate for geological time

    (millions of years). Geological timetemperature conditions are

    then applied to a kinetic model derived from experimental results.

    Rate laws for the formation and destruction of hydrocarbons arenot easily obtained from experiments on the cracking of whole oils

    because of the complexity of the chemical composition. This is eas-

    ier when model compounds are used, i.e. pure hydrocarbons and

    simple mixtures representing the reactivity families found in oil.

    Alkanes are usually the most abundant class in non biodegraded

    crude oils (Tissot and Welte, 1984). Pure n-alkanes have been

    extensively studied to derive global kinetic parameters or detailed

    free radical models (e.g. Ford, 1986; Weres et al., 1988; Domin,

    1989; Domin et al., 1990; Song et al., 1994; Jackson et al., 1995;

    Bhar and Vandenbroucke, 1996; Bounaceur et al., 2002b). Except

    for the work ofDomin (1991)at very low conversion, to the best

    of our knowledge, iso-alkane cracking has not been investigated

    extensively. Bounaceur (2001) and Domin et al. (2002) first pro-

    posed a free radical model taking into account a distribution of

    n-alkanes from C1 to n-C30, similar to that found in the saturated

    fraction of crude oils. Only ten individual branched alkanes were

    included in their model as reactants, the branched alkanes being

    formed via n-alkane cracking but not consumed afterwards. It

    should be noted that their mechanism was based on a lumping to-

    gether concept in order to keep the model to a reasonable size.

    The purpose of the present work was to construct a model that

    would be able to take into account the cracking of a complete dis-

    tribution ofn-alkanes, as well as of a distribution of branched al-

    kanes from C1 to C32. A lumping together procedure similar to

    that ofDomin et al. (2002) was applied. Formation, as well as con-

    sumption, radical reactions for every alkane (linear or branched)

    from C1 to C32 were included in the model. After justification ofthe choice of model compounds for branched alkanes, the elabora-

    tion of the lumped free-radical mechanism is detailed and the

    model validated on the basis of literature experimental data. Final-

    ly, the cracking of the distributions ofn- and iso-alkanes character-

    istic of petroleum (Elgin oil, North Sea) is simulated and compared

    with the experimental results. The implications of the model for

    geological reservoirs will be discussed in a following paper.

    2. Branched alkane model compounds

    The structures of the branched alkanes in crude oils are so

    numerous that most are included in the so called unresolved com-

    0146-6380/$ - see front matter 2011 Elsevier Ltd. All rights reserved.doi:10.1016/j.orggeochem.2011.03.017

    Corresponding author. Tel.: +33 383175093; fax: +33 383378120.

    E-mail address: [email protected](V. Burkl-Vitzthum).

    Organic Geochemistry 42 (2011) 439450

    Contents lists available at ScienceDirect

    Organic Geochemistry

    j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / o r g g e o c h e m

    http://dx.doi.org/10.1016/j.orggeochem.2011.03.017mailto:[email protected]://dx.doi.org/10.1016/j.orggeochem.2011.03.017http://www.sciencedirect.com/science/journal/01466380http://www.elsevier.com/locate/orggeochemhttp://www.elsevier.com/locate/orggeochemhttp://www.sciencedirect.com/science/journal/01466380http://dx.doi.org/10.1016/j.orggeochem.2011.03.017mailto:[email protected]://dx.doi.org/10.1016/j.orggeochem.2011.03.017
  • 7/23/2019 Burkl-Vitzthum Etal 2011

    2/12

    plex mixture (UCM). Most of the identified structures are mono-

    methyl alkanes, particularly mid-chain isomers (Jackson et al.,

    1986; Klomp, 1986; Fowler and Douglas, 1987; Hoffmann et al.,

    1987; Kissin, 1987; Summons, 1987; Summons et al., 1988a,b),

    butKissin (1987) suggested that all monomethyl isomers can be

    found in approximately the same amount. In the medium molecu-

    lar weight (MW) range (C10C25), the most remarkable compounds

    belong to the isoprenoid series and they frequently amount to ca.

    1% of a crude oil, with pristane and phytane the most abundant

    (Tissot and Welte, 1984). Components withn-alkyl branches long-

    er than methyl also have been identified by several authors

    (Kurashova et al., 1989; Gough and Rowland, 1990; Killops and

    Al-Juboori, 1990; Gough et al., 1992; Warton et al., 1997).

    Furthermore, it should be noted that in numerous biological

    systems and sedimentary organic matter, some positions for the

    methyl in monomethyl alkanes and their functionalised counter-

    parts are favoured, particularly at C-2 and C-3 (Eglinton and Ham-

    ilton, 1967; Tornabene et al., 1970; Downing, 1976; Kolattukudy,

    1976; Kaneda, 1977; Dowling et al., 1986; Shiea et al., 1991; Gun-

    stone, 1996).

    The goal of this study was to construct an alkane cracking

    mechanism including all the n-alkanes from C1 to n-C32, as well

    as branched alkanes from C4to C32. Obviously, it was not possible

    to take into account all the existing branched alkanes. That is why

    we had to define model compounds, at least one for each chain

    length, on the basis of the previous observations.

    All branched alkanes from C4 to C6 (Fig. 1) are included in the

    mechanism in order to predict the composition of the generated

    gas in detail.

    For odd carbon numbers between C7 and C31, one model com-

    pound per chain length was chosen: the 2-monomethyl alkane.

    This choice enabled us to represent one of the biologically favoured

    methyl positions.

    For even carbon numbers between C8and C32, two model com-

    pounds per chain length were defined. The first represents another

    favoured methyl position, i.e. mid-chain. In addition to a biological

    origin, branched alkanes are formed during alkane cracking viaaddition of alkyl radicals to alkenes. Most are dialkyl alkanes with

    the alkyl groups on neighbouring carbons, which are probably dif-

    ficult to identify in crude oils because of analytical limitations. That

    is why we chose dimethyl alkanes as the third type of model com-

    pound. In order to limit the number of elementary reactions, the

    structures of the dimethyl alkanes, as well as the mid-chain mono-

    methyl alkanes, were chosen to be symmetrical. The three types of

    branched model compounds are summarized inFig. 1.

    Overall, the model comprises 32 n-alkanes and 46 iso-alkanes as

    reactants.

    3. Elaboration of reaction scheme

    It is now widely accepted that elementary free radical reactions

    can adequately describe the thermal transformation of most spe-

    cies found in crude oils (e.g. Ford, 1986; Savage and Klein, 1988;

    Weres et al., 1988; Domin, 1989, 1991; Domin et al., 1990,

    2002; Jackson et al., 1995; Bounaceur et al., 2002a,b; Bhar et al.,

    2002; Burkl-Vitzthum et al., 2004, 2005; Fusetti et al., 2010).The proposed model includes 78 alkanes as reactants, each of

    which can potentially undergo hundreds or thousands of free rad-

    ical reactions. For example, the mechanism of n-C6 pyrolysis, at

    only low conversion, comprises 156 reactions (Domin et al.,

    1990) and includes 53 species (molecules and radicals). A detailed

    model of our mixture at high conversion would thus require mil-

    lions of reactions (the detailed mechanism of each pure alkane

    and the related cross reactions) and the resulting computing time

    would be unmanageable. Some simplification is therefore required,

    which means that reactions and some species must be lumped to-

    gether. The work is inspired by the previous models ofBounaceur

    (2001) and Domin et al. (2002), but the method applied here is

    slightly different. Indeed, the part of the mechanism concerning

    n-alkanes in the model of Domin et al. (2002) was first written

    using an automated procedure and the number of reactions was re-

    duced afterwards by lumping together radicals with the same

    chain length and by lumping together some products as well. In

    contrast, a lumping together procedure was applied immediately

    in this study. All rate constants were derived from Allara and Shaw

    (1980)or from the NIST Kinetics Database.

    3.1. Unimolecular initiation

    At high pressure (several hundred bar) and low or intermediate

    temperature (200400 C), initiation rates are negligible vs. propa-

    gation rates (Bounaceur, 2001). Therefore all radicals formed by

    initiation steps, whatever their structure, can be lumped together

    in a single one and so, for each alkane, there is only one initiation

    step in the model (Fig. 2). Simulations have enabled validation of

    such an assumption.

    For each alkane from C2to C6, an average value of activation en-

    ergy was estimated, and for C7+, the same value was taken, what-

    ever the number of carbons in the molecule. Indeed for long

    chain alkanes, the chain end effect can be neglected. Frequency fac-

    tors are approximately proportional to the number of CC bonds. A

    reference value was set according to the literature and all fre-

    quency factors were then calculated.

    Fig. 1. Model compounds for branched alkanes. Fig. 2. Examples of each type of lumped free radical reaction.

    440 V. Burkl-Vitzthum et al. / Organic Geochemistry 42 (2011) 439450

  • 7/23/2019 Burkl-Vitzthum Etal 2011

    3/12

    3.2. Decomposition via b-scission

    In the mechanism, radicals are lumped together, i.e. the position

    of the single electron is not precisely defined. For example, C3H7represents the primary propyl radical as well as the secondary one.

    However, since it is highly important to determine precisely the

    products formed from each decomposition as well as the corre-

    sponding rate constant, each radical was un-lumped and its

    decomposition precisely analyzed. For example the lumping to-

    gether of the 2-methylbutyl radical (iso-C5) represents four differ-

    ent radicals depending on the position of the single electron. These

    radicals undergo five decompositions (Fig. 3) which are repre-

    sented by three decompositions lumped together in the mecha-

    nism (Fig. 2) because some of them lead to the same products. It

    should be noted that all alkenes with the same number of carbons

    are lumped together in a single class. Moreover, the branched rad-

    icals included in the mechanismhave the same structure as the iso-

    alkane model compounds. Therefore, when a branched radical is

    formed, it is replaced in the mechanism by the radical that has

    the closest structure (example inFig. 4).

    The determination of the rate constants is a critical step: the

    rate of a lumped reaction must be equal to the sum of the rates

    of the reactions for which it stands; this means that, for the previ-

    ous example, where the first, third and fourth decompositions in

    Fig. 3are lumped together in the first decomposition in Fig. 2:

    kaisoC5 radical k1 1ary

    radical k3 3ary

    radical 2 k4

    2ary radical;

    where 1ary stands for primary, 2ary for secondary and 3ary for

    tertiary.

    By implementing the intrinsic distribution of each type of

    radical:

    kaisoC5 radical k1 %1ary isoC5 radical k3 %3

    ary

    isoC5 radical 2 k4 %2ary

    isoC5 radical

    This leads to:ka= k1 %1ary + k3 %3

    ary + 2 k4 %2ary.

    It should be noted that the second and fifth decompositions in

    Fig. 3lead to different products and so cannot be lumped together

    and remain un-lumped in the mechanism (Fig. 2, second and third

    decompositions).

    On the basis of the work ofBounaceur (2001), the intrinsic dis-

    tribution of each radical was estimated (Table 1). Then, depending

    on the type of decomposition, three reference rate constants were

    set (Table 2) and the rate constant of every lumped decomposition

    was calculated independently.

    3.3. H transfer

    H transfers between each alkane and lumped radical are in-

    cluded in the mechanism (Fig. 2). Activation energy depends on

    the class (primary, secondary or tertiary) of the acceptor radical

    and on the class of the carbon bearing the transferred hydrogen.

    Reference values were estimated according to the literature (Table

    3). An average value (1011 cm3 mol1 s1) was taken for the fre-

    quency factor of a single H transfer. In the mechanism, the H trans-

    fers are lumped and therefore: Alumped =P

    Asingle, which implies

    that Alumped is proportional to the number of H transfers the

    lumped step stands for.

    3.4. Addition

    Average values for the activation energy and frequency factors

    were estimated knowing that they are inversely proportional tothe number of carbons in the radical. It should be noted that every

    C33+ radical formed by addition is lumped into a single class, what-

    ever its number of carbons, which is considered as stable in the

    model (Fig. 2).

    3.5. Isomerization

    Obviously, because of the lumpingtogether of radicals, no isom-

    erization reaction is explicitly written into the mechanism, but the

    intrinsic distribution of each radical is implicitly taken into account

    by way of decomposition rate constants.

    3.6. Termination

    Termination steps between all radicals were lumped together

    (Fig. 2), because, like initiation steps, their rates are negligible vs.

    Fig. 3. Detailed decompositions viab-scissionon the example ofiso-C5. For the first

    radical, two different bonds break, leading to different products. The arrow colourcorresponds to the breakage colour. (Forinterpretation of the references to colour in

    this figure legend, the reader is referred to the web version of this article.)

    Fig. 4. Replacement of a formed radical by the closest structure.

    V. Burkl-Vitzthum et al. / Organic Geochemistry 42 (2011) 439450 441

  • 7/23/2019 Burkl-Vitzthum Etal 2011

    4/12

    propagation rates under our conditions. As a consequence, all ter-

    mination steps lead to the same alkane and a reference rate con-stant was used. Our simulations confirmed that the

    concentration of this end member alkane was negligible.

    Neither cyclization nor aromatization reactions were included

    in the mechanism because aromatics are negligible up to 50% con-

    version and the amounts remain low for higher conversions (Song

    et al., 1994; Bhar and Vandenbroucke, 1996) even at high temper-

    ature (450 C). In previous studies (Bounaceur, 2001; Bounaceur

    et al., unpublished data), it was shown that the importance of these

    reactions with regard to alkane cracking decreases when the tem-

    perature decreases, and becomes minor below 300 C, particularly

    at geological temperatures. Indeed, cyclization reactions require

    the formation of alkenyl radicals by H transfer to an alkene, which

    react further by way of intramolecular addition to form a ring.

    There is a competition between these H transfers and the additionof alkyl radical to alkene. The addition activation energy is from 3

    to 5 kcal/mol lower than H transfer activation energy and so, when

    the temperature decreases, the consumption of alkenes by addition

    is favoured to the detriment of the cyclization route. So, cyclization

    and aromatization reactions probably do not affect the n- andiso-

    alkanes distribution to a significant extent. Nevertheless, this

    choice constitutes a limitation of the model.

    Overall, the model comprises 13,206 free radical reactions and

    193 species.

    4. Stoichiometric equations corresponding to n- oriso-alkane

    cracking

    A free-radical mechanism such as the present one is based ontwo generic stoichiometric equations (in moles):

    Cracking equation:

    Alkane Cn ! Alkane-minusCn1 to n2 1-alkene Cn2 to n1

    Alkylation equation:

    Alkane Cn 1-alkene Cn2 to n1 ! Alkane-plusCn2 t o 2n1

    where alkane-minus corresponds to an alkane whose molecularweight (MW) is lower than the reactant and alkane-plus to an al-

    kane whose MW is higher than the reactant.

    Under our conditions [low temperature (200400 C) and high

    pressure (1001000 bar)], addition to alkenes is fast, so both equa-

    tions lead to (in moles):

    2 AlkanesCn ! Alkane-minusCn1 to n2

    Alkane-plusCn2 t o 2n1

    Indeed, the amounts of alkenes are low and become totally neg-

    ligible at low temperature experimentally as well as in simulations,

    as in petroleum (Tissot and Welte, 1984).

    Moreover, in the mechanism, the formation of alkanes-plus is

    due to the addition of alkyl radicals to alkenes. Because primary

    radicals are negligible, alkanes-plus are mainly iso-alkanes.

    So the above analysis leads to:

    In the case ofn-alkanes (in moles):

    2 n-Alkanesn-Cn ! n-Alkane-minusn-Cn1 to n2

    iso-Alkane-plusCn2 t o 2n1

    In the case of iso-alkanes (in moles):

    2 iso-Alkanes n-Cn ! n-or iso Alkane-minusCn1 to n2

    iso-Alkane-plusCn2 t o 2n1

    Then, the produced alkanes react according to the same stoichi-

    ometric equations.

    5. Pyrolysis of a pure n-alkane

    The mechanism was tested on pure n-C15 as an example. The

    model was simulated using the software CHEMKIN II (Kee et al.,

    1989).Fig. 5plots logxi (molar fraction ofn- oriso-alkane Ci) vs.

    the number i of carbons, after 1 week (low conversion) and after

    15 yr (high conversion).

    At low conversion, n-C15 obviously remains predominant. Al-

    kanes-minus are mainly n-alkanes from C2 to n-C13 in almost the

    same proportions. The formation of methyl radicals via decompo-

    sition by b-scission is slightly more difficult than for the other alkyl

    radicals (Table 2). This is why the amount of CH4is lower than the

    amounts of C2C13. Furthermore, the decomposition via b-scission

    of the C15radicals cannot lead to the formation of a C14radical, son-C14cannot be formed primary: its amount is negligible. Alkanes-

    plus are mainly iso-alkanes from C17 to C29. Indeed, alkanes-plus

    are formed by addition of the predominant radicals that are the

    C15radicals, on alkenes comprising between C2and C14(decompo-

    sition via b-scission of C15 radicals). Moreover it should be noted

    thatn-alkanes-minus andiso-alkanes-plus are formed in the same

    proportion, as illustrated by the above stoichiometric equation.

    At high conversion, n-C15does not predominate anymore. Dis-

    tributions ofn-alkanes and ofiso-alkanes are obtained and the evo-

    lution of log(xi) vs.i is almost linear for iP 10. It should be noted

    that methane is much less reactive than the other alkanes and

    therefore it accumulates. The C33+ compounds accumulate but, as

    mentioned above, this is an artifact due to the fact that their con-

    sumption is not taken into account in the mechanism. Further-more, the amount of C2H6 is less than the amount of C3C10

    Table 1

    Estimation of internal distribution of radicals (x is the number of secondary (2ary)

    carbons and y, the number of tertiary (3ary) carbons).

    Iso-CnH2n+2 Primary

    radicals (%)

    Secondary radicals

    (equidistribution)

    Tertiary radicals

    5 6n 67 7.5 2ary% = (100-1ary%)/

    (x+ 2y)

    3ary% = 2 2ary%

    n= 8 7.5 2ary% = (100-1ary%)/

    (x+ 1.5y)

    3ary%=1.5 2ary%

    9 6n 615 5 Ditto 3ary%=1.5 2ary%

    16 6n 622 4 Ditto 3ary%=1.5 2ary%

    23 6n 629 3 Ditto 3ary%=1.5 2ary%

    n> 29 2 Ditto 3ary%=1.5 2ary%

    Table 2

    Decomposition via b-scission: reference rate constants depending on type of CC

    bond.

    Type of CC bond Frequency factor

    (1013 s1)

    Activation energy (cal/

    mol)

    1ary C2ary C 1.5 30,000

    2ary C2ary C 2.5 27,700

    2ary (or 1ary) C3ary

    C

    2.5 27,200

    Table 3

    H transfer: reference activation energy depending on type of radical and type of

    transferred hydrogen.

    Transferred hydrogen radical CH4 CH3 CH2 CH

    CH3 8200 9600 9600

    C2H

    5 14,500 11,600 11,600

    Linear 15,500 12,900 12,200 12,200

    Branched 16,000 13,500 12,200 12,200

    442 V. Burkl-Vitzthum et al. / Organic Geochemistry 42 (2011) 439450

  • 7/23/2019 Burkl-Vitzthum Etal 2011

    5/12

    because it cannot be formed by addition, only by decomposition

    viab-scission.

    Finally it is important to note that, for i > 29, the aforemen-

    tioned trend is not observed anymore. In the following paragraph

    these break points are analyzed in order to understand whether

    they have a chemical meaning or whether they are due to the

    mechanism, which does not take into account the cracking of the

    C33+ fraction.

    6. Effect of number of carbons

    In order to answer the previous question, a reduced mechanism

    was written which comprisesn- andiso-alkanes up to C27. Pyroly-sis ofn-C12is simulated up to the same conversion extent asn-C15previously. The product distributions show the same trends as for

    the full mechanism, but the break points are moved to C24(Fig. 6).

    Therefore, these break points are linked to the maximum number

    of carbons accounted for by the mechanism and have no chemical

    meaning: this is an artifact of the mechanism, more precisely for

    the C33+ compounds which are formed, but do not further react.

    Hence, every conclusion relating to the longest alkanes has to be

    taken with caution.

    7. Experimental validation for pure alkanes as reactants

    7.1. Conversion

    The mechanism was tested on several pure n-alkanes with

    chainlength varying from 6 to 25, submitted to temperatures rang-

    ing from 300 to 450 C, pressures from 120 to 700 bar and conver-

    sions up to 94%. All the data were taken from the literature

    (Domin, 1989; Song et al., 1994; Jackson et al., 1995; Bhar and

    Vandenbroucke, 1996; Burkl-Vitzthum et al., 2004; Lannuzel,

    2006; Lannuzel et al., 2010). The CHEMKIN II software uses the

    perfect gas law to calculate the concentrations; consequently, for

    each condition, the initial concentration was first calculated sepa-

    rately by using the PengRobinson equation of state, in order to

    evaluate the perfect gas pressure that leads to the same initial con-

    centration. This pressure corresponds to the input value.Figs. 79

    plot the conversion vs. time for all data. Agreement between the

    experimental and the simulated conversions is observed for everyn-alkane and under each pressuretemperature-time condition.

    Concerning iso-alkanes, to the best of our knowledge, few

    experimental data are available at high pressure and intermediate

    temperature (300400 C). We only found the results ofDomin

    (1991) relating to 2,4-dimethylpentane pyrolysis at 357 C,

    210 bar and low conversion (up to 5%). Our model does not de-

    scribe this alkane, but the corresponding lumped species is 2-

    methylhexane. By comparing experimental 2,4-dimethylpentane

    conversion to the simulated 2-methylhexane conversion (Fig. 10),

    agreement was obtained.

    7.2. Product distribution

    Conversion is not the only parameter to be checked in order tovalidate the model; product distribution also needs to be assessed.

    Jackson et al. (1995)defined four fractions to describe the product

    distribution (wt%) aftern-C16pyrolysis: C1C4, C5C9, C10C15and

    C16+. For each fraction, the experimental and simulated mass bal-

    ances were represented vs. conversion. Data, at all temperatures

    and pressures are gathered (Fig. 11). The composition of each frac-

    tion is well predicted by the model up to 30% conversion. Agree-

    ment between the experimental and simulated results is

    satisfactory up to 50% conversion. Above 50% discrepancies appear;

    in particular the C16+fraction is overestimated and the C1C9frac-

    tions are underestimated, whereas the C10C15fraction is correctly

    described, as well as conversion (Figs. 79). This can be easily ex-

    plained as the C33+species are lumped together in the mechanism

    and, once formed, do not react further and accumulate at the ex-pense of low MW compounds. This phenomenon only becomes

    important at high conversion, which explains the observed

    discrepancies.

    Table 4compares the simulated results with some data taken

    from Bhar and Vandenbroucke (1996) who defined three fractions

    after n-C25 pyrolysis: C1C4, C7C14 and C14+. As previously, at

    intermediate conversion (37%) the agreement is totally satisfac-

    tory, whereas at high conversion (90%), the light alkanes are under-

    estimated and the heavy ones overestimated.

    8. Apparent activation energy

    Values of apparent cracking activation energy in the case of

    puren-C15andiso-C15 were calculated at several temperatures ofT0 between 200 and 350 C. In fact, the simulations were done at

    Fig. 5. Simulated product distribution obtained from n-C15cracking at 350 C.

    V. Burkl-Vitzthum et al. / Organic Geochemistry 42 (2011) 439450 443

  • 7/23/2019 Burkl-Vitzthum Etal 2011

    6/12

    very short and constant conversion. The initial consumption rate r0was calculated by using CHEMKIN, and by plotting ln r0 vs. 1/RT

    (where R is the perfect gas constant and Tthe temperature in K)

    with T= T0, T0 2, T0 + 2, the slope corresponds to the apparent

    Ea. For both alkanes and whatever the temperature between 200

    and 350 C, the apparent Ea was found to be constant and close

    to 69 kcal/mol.

    This value is consistent forn-alkanes with experimentalEaofn-

    C25 (68.2 kcal/mol) determined by Bhar and Vandenbroucke

    (1996) and used for the n-C14+ fraction byVandenbroucke et al.(1999) in their kinetic modelling of petroleum cracking. But in

    the same model, for the other saturated C14+ fraction, which is

    composed of cyclo- and branched alkanes, Ea is set to 59 kcal/

    mol. Nevertheless, the authors concluded that it may be better to

    group the n- and iso-alkanes on the one hand, and the cycloalkanes

    on the other hand. The conclusion ofVandenbroucke et al. (1999)

    appears to be justified on the basis of this study and of another

    study of cycloalkane cracking (Bounaceur et al., unpublished data)

    that showedEa close to 55 kcal/mol at 200 C.

    9. Saturated fraction of Elgin oil: comparison between

    experimental and simulated results at 372 C

    The mechanism was tested on thermal cracking experimentsinvolving a whole oil sample from the Elgin Field (North Sea,

    UK). This sample has abundant saturated hydrocarbons and negli-

    gible polar compounds (Domin et al., 2002; Vandenbroucke et al.,

    1999), making it well adapted to our model. Pyrolysis experiments

    with the sample were conducted by ELF (TOTAL) in 19921993 but

    the results were not published until now. The temperature was set

    at 372 C, the initial pressure was ca. 400 bar, for several time peri-

    ods up to 1000 h; the experimental conditions are summarized in

    Table 6. Closed cells (20 cm3) were loaded with ca. 5 g Elgin oil and

    added to the same oven at the same time. The C1C10fraction was

    analyzed using GCMS (gas chromatographymass spectrometry)and the C11+ fraction was analyzed using GPC (gel permeation

    chromatography) by applying the method ofSynovec and Yeung

    (1984), which uses toluene and tetrahydrofuran as solvents. The

    product distribution (Fig. 12), e.g. logxi (xi molar fraction of all

    hydrocarbons,n-,iso- and cycloalkanes as well as alkenes and aro-

    matics) vs. the number i of carbons, was determined for several

    time periods up to 1000 h. Nevertheless, it should be noted that

    then- andiso-alkanes represent more than 90% molar of the oil.

    In order to test our mechanism, the molar composition of the

    saturated fraction (Table 5) was deduced from the analysis. The

    n-Ci/iso-Ci ratio is not known fori > 10 because GPC does not sep-

    arate n-alkanes fromiso-alkanes, so it was set to 0.5 according to

    Vandenbroucke et al. (1999). This is an approximation as these

    authors analyzed a different sample from Elgin Field, the composi-tion of which is somewhat different due to inhomogeneity within

    Fig. 6. Simulated product distribution obtained by n-C12 cracking (reduced mechanism) and by n-C15cracking (full mechanism) at 350 C: (a)linear alkanes and (b) branched

    alkanes.

    444 V. Burkl-Vitzthum et al. / Organic Geochemistry 42 (2011) 439450

  • 7/23/2019 Burkl-Vitzthum Etal 2011

    7/12

    the field. The model was simulated using the mixture inTable 5as

    input composition, at 372 C and 400 bar; the product distributions

    are represented inFig. 13. A detailed comparison betweenFigs. 12

    and 13 is complex because Fig. 12 includes all hydrocarbons

    whereasFig. 13only represents n- and iso-alkanes. Nevertheless

    the evolution of the distribution up to 1000 h appears satisfactory:

    experimentally as well as via simulation, the evolution of the light

    compounds is negligible, particularly up to 500 h and for iP 10 the

    distribution follows a straight line whose slope increases with

    time. To better characterize the evolution of the C10+fraction, the

    experimental and simulated average ratios of xi+1/xi with

    15 < i< 25 were calculated for all time periods (Table 7). This ratio

    is perfectly simulated up to about 200 h. However it stabilizes

    experimentally from 200 h, whereas its decrease continues in the

    simulation. The explanation of this apparent experimental stabil-

    ization remains unclear, but it may be because of cycloalkanes that

    crack to alkanes and are not taken into account in the model. Com-

    parisons of the evolution of the experimental and the simulated

    molar fractions for individual alkanes show good agreement

    (Figs. 14 and 15).

    Fig. 7. Comparison between experimental and simulated n-alkanes conversion from 299 to 357 C.

    Fig. 8. Comparison between experimental and simulated n-alkanes conversion from 369 to 400 C.

    V. Burkl-Vitzthum et al. / Organic Geochemistry 42 (2011) 439450 445

  • 7/23/2019 Burkl-Vitzthum Etal 2011

    8/12

    To conclude, all trends are well represented by the model and so

    it is also validated in the case of a mixture as complex as an oil.

    10. Conclusions

    We have proposed the first cracking model ofn- and iso-alkanes

    mixtures that represents their formation as well as their consump-

    tion. It includes all types of free radical reactions that can be rea-

    sonably expected to take place in the pyrolysis of alkanes below

    450 C, except cyclization and aromatization reactions, but these

    reactions are believed to be minor, especially at low temperature.

    The rate constants for elementary processes have been estimatedon the basis of the literature and were not adjusted. The approach

    is not empirical and uses reactions representing what happens at

    the molecular scale.

    Lumping together reaction types and species with similar struc-

    tures and defining model branched alkanes allow limiting the

    model to a reasonable size. The pyrolysis of a mixture containing

    32 n-alkanes and 46iso-alkanes is represented by 13,206 reactions

    involving 193 species. Software like CHEMKIN II is able to solve

    this system easily. The model was compared with several literature

    experimental results with pure alkanes. With respect to conver-

    sion, the agreement is totally satisfactory but product distributions

    are well simulated only up to 50% conversion. Above 50%, gas is

    underestimated and heavy alkanes overestimated because of C33+which, once formed, does not react further in our model. Moreover,

    Fig. 9. Comparison between experimental and simulated n-alkanes conversion at 425 and 450 C.

    Fig. 10. Comparison between experimental and simulated iso-C7

    conversion at 357 C.

    446 V. Burkl-Vitzthum et al. / Organic Geochemistry 42 (2011) 439450

  • 7/23/2019 Burkl-Vitzthum Etal 2011

    9/12

    the model was compared with the pyrolysis results for Elgin oil. In

    this case, the global evolution of the saturated fraction is well rep-

    resented by the model but a quantitative comparison is difficult.

    Indeed, Elgin oil contains cycloalkanes and aromatics (

  • 7/23/2019 Burkl-Vitzthum Etal 2011

    10/12

    Finally, one majorlimitation of themodel is probably theaccumu-

    lationof the C33+ fraction at high conversion. One solution would be,

    at eachcalculation steptime,to extrapolate thealkanedistribution on

    the basis of the middle MWcompound distribution and to compare

    the extrapolation to the simulated results. The difference between

    the extrapolation and the simulation corresponds to the accumula-

    tion which could be equally shared among the C1C32 alkanes for

    the following step time. This method would avoid the accumulation

    of C33+ as well as significant deviation of the C1C32 distribution,but its computational implementation remains unsolved.

    Fig. 12. Experimental product distribution of Elgin saturated fraction after whole oil pyrolysis at 372 C and 400 bar as initial pressure, up to 1000 h.

    Fig. 13. Simulated cracking (372 C, 400 bar) of a linear-branched alkanes mixture in the same proportions as in Elgin oil.

    Table 7

    Comparison between experimental and simulated xCi+1/xCi molar fraction ratio

    (average value from C15to C25).

    Duration (h) Experimental xCi+1/xCi SimulatedxCi+1/xCi

    0 0.85 0.85

    30 0.84 0.84

    100 0.82 0.82

    200 0.78 0.80

    500 0.78 0.74

    1000 0.79 0.65

    448 V. Burkl-Vitzthum et al. / Organic Geochemistry 42 (2011) 439450

  • 7/23/2019 Burkl-Vitzthum Etal 2011

    11/12

    Acknowledgements

    We are grateful for valuable comments and suggestions fromC.C. Walters and an anonymous reviewer.

    Associate EditorR. di Primio

    References

    Allara, D.L., Shaw, R., 1980. A compilation of kinetic parameters for the thermal

    degradation of n-alkane molecules. Journal of Physical Chemistry ReferenceData 9, 523559.

    Bhar, F., Kressmann, S., Rudkiewicz, J.L., Vandenbroucke, M., 1992. Experimental

    simulation in a confined system and kinetic modelling of kerogen and oil

    cracking. Organic Geochemistry 19, 173189.

    Bhar, F., Vandenbroucke, M., 1996. Experimental determination of the rate

    constants of the n-C25 thermal cracking at 120, 400, and 800 bar:

    implications for high-pressure/high-temperature prospects. Energy and Fuels10, 932940.

    Bhar, F., Tang, Y., Liu, J., 1997a. Comparison of rate constants for some molecular

    tracers generated during artificial maturation of kerogens: influence of kerogen

    type. Organic Geochemistry 26, 281287.

    Bhar, F., Vandenbroucke, M., Tang, Y., Marquis, F., Espitali, J., 1997b. Thermal

    cracking of kerogen in open and closed systems: determination of kinetic

    parameters and stoichiometric coefficients for oil and gas generation. Organic

    Geochemistry 26, 321339.

    Bhar, F., Lorant, F., Budzinski, H., Desavis, E., 2002. Thermal stability of

    alkylaromatics in natural systems: kinetics of thermal decomposition of

    dodecylbenzene. Energy and Fuels 16, 831841.

    Bhar, F., Lorant, F., Lewan, M.D., 2008. Role of NSO compounds during primarycracking of a Type II kerogen and a Type III lignite. OrganicGeochemistry 39, 1

    22.

    Bounaceur, R., 2001. Modlisation cintique de lvolution thermique des ptroles

    dans les gisements. PhD Thesis, INPL Nancy, France.

    Bounaceur, R., Scacchi, G., Marquaire, P.-M., Domin, F., Dessort, D., Pradier, B.,

    2002a. Inhibiting effect of tetralin on the pyrolytic decomposition of

    hexadecane. Comparison with toluene. Industrial and Engineering Chemistry

    Research 41, 46894701.

    Bounaceur, R., Warth, V., Marquaire, P.-M., Scacchi, G., Domin, F., Dessort, D.,

    Pradier, B., Brevart, O., 2002b. Modeling of hydrocarbons pyrolysis at low

    temperature. Automatic generation of free radicals mechanisms. Journal of

    Analytical and Applied Pyrolysis 64, 103122.

    Burkl-Vitzthum, V., Michels, R., Scacchi, G., Marquaire, P.-M., Dessort, D., Pradier,

    B., Brevart, O., 2004. Kinetic effect of alkylaromatics on the thermal stability of

    hydrocarbons under geological conditions. Organic Geochemistry 35, 331.

    Burkl-Vitzthum, V., Michels, R., Bounaceur, R., Marquaire, P.-M., Scacchi, G., 2005.

    Experimental study and modeling of the role of hydronaphthalenics on the

    thermal stability of hydrocarbons under laboratory and geological conditions.

    Industrial and Engineering Chemistry Research 44, 89728987.

    Dieckmann, V., Schenk, H.J., Horsfield, B., Welte, D.H., 1998. Kinetics of petroleum

    generation and cracking by programmed-temperature closed-system pyrolysis

    of Toarcian Shales. Fuel 77, 2331.

    Domin, F., 1989. Kinetics of hexane pyrolysis at very high pressures. 1.

    experimental study. Energy and Fuels 3, 8996.

    Domin, F., Marquaire, P.-M., Muller, C., Cme, G.-M., 1990. Kinetics of hexane

    pyrolysis at very high pressures. 2. Computer modeling. Energy and Fuels 4, 2

    10.

    Domin, F., 1991. High pressure pyrolysis ofn-hexane, 2,4-dimethylpentane and 1-phenylbutane. Is pressure an important geochemical parameter? Organic

    Geochemistry 17, 619634.

    Domin, F., Bounaceur, R., Scacchi, G., Marquaire, P.-M., Dessort, D., Pradier, B.,

    Brevart, O., 2002. Up to what temperature is petroleum stable? New insights

    froma 5200 free radical reactions model. Organic Geochemistry 33, 14871499.

    Dowling, N.J.E., Widdel, F., White, D.C., 1986. Phospholipid ester-linked fatty acid

    biomarkers of acetate-oxidising sulphate reducers and other sulphide-forming

    bacteria. Journal of General Microbiology 132, 18151825.

    Downing, D.T., 1976. Mammalian waxes. In: Kolattukudy, P.E. (Ed.), Chemistry andBiochemistry of Natural Waxes. Elsevier, Amsterdam, pp. 1748.

    Eglinton, G., Hamilton, R.J., 1967. Leaf epicuticular waxes. Science 156, 13221325.

    Ford, T.J., 1986. Liquid-phase thermal decomposition of hexadecane: reaction

    mechanisms. Industrial and Engineering Chemistry Fundamentals 25, 240243.

    Fowler, M.G., Douglas, A.G., 1987. Saturated hydrocarbon biomarkers in oils of Late

    Precambrian age from Eastern Siberia. Organic Geochemistry 11, 201213.

    Fusetti, L., Bhar, F., Bounaceur, R., Marquaire, P.-M., Grice, K., Derenne, S., 2010.

    New insights into secondary gas generation from the thermal cracking of oil:

    methylated monoaromatics. A kinetic approach using 1,2,4-trimethylbenzene.

    Part I: a mechanistic kinetic model. Organic Geochemistry 41, 146167.

    Gough, M.A., Rowland, S.J., 1990. Characterisation of unresolved complex mixtures

    of hydrocarbons in petroleum. Nature 344, 648650.

    Gough, M.A., Rhead, M.M., Rowland, S.J., 1992. Biodegradation studies of unresolved

    complex mixtures of hydrocarbons: model UCMhydrocarbons and the aliphatic

    UCM. Organic Geochemistry 18, 1722.

    Gunstone, F.D., 1996. Fatty Acid and Lipid Chemistry. Chapman and Hall, Glasgow.

    Hoffmann, C.F., Foster, C.B., Powell, T.G., Summons, R.E., 1987. Hydrocarbon

    biomarkers from Ordovician sediments and the fossil algaGloeocapsomorpha

    prisca Zalessky 1917. Geochimica et Cosmochimica Acta 51, 26812697.Jackson, M.J., Powell, T.G., Summons, R.E., Sweet, I.P., 1986. Hydrocarbon shows and

    petroleumsourcerocks in sediments as oldas 1.7 109 years. Nature 322, 727

    729.

    Jackson, K.J., Burnham, A.K., Braun, R.L., Knauss, K.G., 1995. Temperature and

    pressure dependence ofn-hexadecane cracking. Organic Geochemistry 23, 941953.

    Kaneda, T., 1977. Fatty acids of the genus Bacillus: an example of branched-chainpreference. Bacteriology Reviews 41, 391418.

    Kee, R.J., Rupley, F.M., Miller, J.A., 1989. Chemkin-II: a Fortran Chemical Kinetics

    Package for the Analysis of Gas-phase Chemical Kinetics. Sandia National

    Laboratories, Livermore, CA.

    Killops, S.D., Al-Juboori, M.A.H.A., 1990. Characterisation of the unresolved complex

    mixture (UCM) in the gas chromatograms of biodegraded petroleums. Organic

    Geochemistry 15, 147160.

    Kissin, Y.V., 1987. Catagenesis and composition of petroleum: origin of n-alkanesand isoalkanes in petroleum crudes. Geochimica et Cosmochimica Acta 51,

    24452457.

    Klomp, U.C., 1986. The chemical structure of a pronounced series of iso-alkanes inSouth Oman crudes. Organic Geochemistry 10, 807814.

    Fig. 14. Comparison between the experimental (in relation to the saturated

    fraction) and the simulated molar fractions (%) of propane.

    Fig. 15. Comparison between the experimental (in relation to the saturated

    fraction) and the simulated molar fractions (%) ofn-octane and C18.

    V. Burkl-Vitzthum et al. / Organic Geochemistry 42 (2011) 439450 449

  • 7/23/2019 Burkl-Vitzthum Etal 2011

    12/12

    Kolattukudy, P.E., 1976. Introduction to natural waxes. In: Kolattukudy, P.E. (Ed.),

    Chemistry and Biochemistry of Natural Waxes. Elsevier, Amsterdam, pp. 115.

    Kurashova, E.K., Musayev, I.A., Smirnov, M.B., Simanyuk, R.N., Mikaya, A.I., Ivanov,

    A.V., Sanin, P.I., 1989. Hydrocarbons of Kharyag crude oil. Petroleum Chemistry

    USSR 29, 206220.

    Lannuzel, F., 2006. Influence des aromatiques sur la stabilit thermique des ptroles

    dans les gisements. PhD Thesis, INPL Nancy, France.

    Lannuzel, F., Bounaceur, R., Michels, R., Scacchi, G., Marquaire, P.-M., 2010.

    Reassessment of the kinetic influence of toluene on n-alkane pyrolysis.Energy and Fuels 24, 38173830.

    Lehne, E., Dieckmann, V., 2007a. Bulk kinetic parameters and structural moieties ofasphaltenes and kerogens from a sulphur-rich source rock sequence and related

    petroleums. Organic Geochemistry 38, 16571679.

    Lehne, E.,Dieckmann,V., 2007b. Thesignificanceof kineticparameters andstructural

    markersin source rockasphaltenes, reservoirasphaltenesand relatedsourcerock

    kerogens, the DuvernayFormation (WCSB). Fuel 86, 887901.

    Lewan, M.D., Ruble, T.E., 2002. Comparison of petroleum generation kinetics by

    isothermal hydrous and nonisothermal open-system pyrolysis. Organic

    Geochemistry 33, 14571475.

    Lewan, M.D., Kotarba, M.J., Curtis, J.B., Wieclaw, D., Kosakowski, P., 2006. Oil-

    generation kinetics for organic facies with Type-II and -IIS kerogen in the

    Menilite Shales of the Polish Carpathians. Geochimica et Cosmochimica Acta 70,

    33513368.

    Savage, P.E., Klein, M.T., 1988. Asphaltene reaction pathways. 4. Pyrolysis of

    tridecylcyclohexane and 2-ethyltetralin. Industrial and Engineering Chemistry

    Research 27, 13481356.

    Schenk, H.J., Di Primio, R., Horsfield, B., 1997. The conversion of oil into gas in

    petroleum reservoirs. Part 1: comparative kinetic investigation of gas

    generation from crude oils of lacustrine, marine and fluviodeltaic origin by

    programmed-temperature closed-system pyrolysis. Organic Geochemistry 26,

    467481.

    Shiea, J., Brassell, S.C., Ward, D.M., 1991. Comparative analysis of extractable lipids

    in hot spring microbial mats and their component photosynthetic bacteria.

    Organic Geochemistry 17, 309319.

    Song, C., Lai, W.C., Schobert, H.H., 1994. Condensed-phase pyrolysis of n-tetradecane at elevated pressures for long duration. Product distribution and

    reaction mechanisms. Industrial and Engineering Chemistry Research 33, 534

    547.

    Summons, R.E., 1987. Branched alkanes from ancient and modern sediments:

    isomer discrimination by GC/MS with multiple reaction monitoring. Organic

    Geochemistry 11, 281289.

    Summons, R.E., Brassell, S.C., Eglinton, G., Evans, E., Horodyski, R.J., Robinson, N.,

    Ward, D.M., 1988a. Distinctive hydrocarbon biomarkers from fossiliferous

    sediment of the Late Proterozoic Walcott Member, Chuar Group, Grand Canyon,

    Arizona. Geochimica et Cosmochimica Acta 52, 26252637.Summons, R.E., Powell, T.G., Boreham, C.J., 1988b. Petroleum geology and

    geochemistry of the Middle Proterozoic McArthur Basin, Northern Australia:

    III composition of extractable hydrocarbons. Geochimica et Cosmochimica Acta

    52, 17471763.

    Synovec, R.E., Yeung, E.S., 1984. Quantitative gel-permeation chromatography

    without standards. Journal of Chromatography A 283, 183190.

    Tissot, B.P., Welte, D.H., 1984. Petroleum Formation and Occurrence, second ed.

    Springer-Verlag, New-York.

    Tornabene, T.G., Morrison, S.J., Kloos, W.E., 1970. Aliphatic hydrocarbon contents of

    various members of the family Micrococcaceae. Lipids 5, 929937.

    Ungerer, P., Pelet, R., 1987. Extrapolation of the kinetics of oil and gas formation

    from laboratory experiments to sedimentary basins. Nature 327, 5254.

    Ungerer, P., Bhar, F., Villalba, M., Heum, O.R., Audibert, A., 1988. Kinetic modelling

    of oil cracking. Organic Geochemistry 13, 857868.

    Vandenbroucke, M., Behar, F., Rudkiewicz, J.L., 1999. Kinetic modelling of petroleum

    formation and cracking: implications from the high pressure/high temperature

    Elgin Field (UK, North Sea). Organic Geochemistry 30, 11051125.

    Warton, B., Alexander, R., Kagi, R.I., 1997. Identification of some single branched

    alkanes in crude oils. Organic Geochemistry 27, 465476.

    Weres, O., Newton, A.S., Tsao, L., 1988. Hydrous pyrolysis of alkanes, alkenes,

    alcohols and ethers. Organic Geochemistry 12, 433444.

    450 V. Burkl-Vitzthum et al. / Organic Geochemistry 42 (2011) 439450


Recommended