Você está na página 1de 113

Applications of a Mechanistic Hydrocracking Model

A Dissertation

By

Hemendra Khakhar

Submitted to the Office of Graduate Studies of


Texas A&M University
in partial fulfillment of the requirements for the degree of

MASTER OF SCIENCE

December 2004

Major Subject: Chemical Engineering


Applications of a Mechanistic Hydrocracking Model
A Dissertation

By

Hemendra Khakhar

Submitted to Texas A&M University


in partial fulfillment of the requirements
for the degree of

Master of Science

Approved as to style and content by:

Dr. Gilbert F. Froment Dr. Rayford G. Anthony


(Chair of Committee) (Co-Chair of Committee)

Dr. Robert R. Lucchese Dr. K.R.Hall


(Member) (Head of Department)

December 2004

Major Subject: Chemical Engineering

iv
ABSTRACT

Applications of a Mechanistic Hydrocracking Model

Hemendra Khakhar, B.S. (Chemical Engineering), Gujarat University

Chair of Advisory Committee: Dr. G.F.Froment

Hydrocracking is a key refining process in the commercial production of a variety of

transportation fuels. In the overall petroleum refining process, Hydrocracking is one

of the most important processes which upgrade heavier fractions obtained by

distillation of the crude oil or heavier fractions from other refinery units like cokers

and fluidized catalytic cracker units. Kinetic modeling of Hydrocracking is of

paramount importance to simulate/optimize an existing industrial unit or to design a

new unit.

For a model to be effectively predictive, it must be adequately detailed and

extrapolable to different feedstocks and operating conditions. "Single Event

Kinetics Approach” developed by Prof. Froment has been successfully applied to

these kind of zeolite catalyzed oil refining processes. The complete network of

elementary steps is generated with the help of Boolean matrices and

characterization vectors in accordance with well known carbenium ion chemistry

principles. The reaction network describes chemical transformations on the catalyst

surface by means of elementary steps like (de)protonation, isomerization,

(de)hydrogenation, cracking etc. Rate equations are developed based on Hougen

v
Watson kinetics and pseudo steady state approximation. Transition state theory

and thermodynamic constraints keeps the total number of parameters within

tractable range.

Parameter estimation from experimental data is done by Levenberg Marquardt

algorithm. In contrast to present day available kinetic models, single event rate

parameters are independent of feedstock and reactor configuration due to its

fundamental nature. These rate equations can be plugged into 1-D heterogeneous

model for simulation and optimization of industrial fixed bed reactors. Obtained

kinetic model can also be used for catalyst design, preliminary design of new

hydrocracker units.

vi
To My Parents

vii
ACKNOWLEDGMENTS

First, I would like to express my deepest appreciation to my research advisor Dr.

Gilbert F. Froment for all his sincere support and guidance throughout the course

of this research. I would like to thank Co-chairman Dr. R.G.Anthony for his valuable

advice and financial support.

I am also thankful to Dr. Robert Lucchese for serving on my dissertation

committee. I am grateful to the members of the kinetics, catalysis and reaction

engineering group to their excellent advice and support.

I also feel the deepest amount of gratitude to my all friends here at Texas A&M for

their moral support and enjoyable time. Finally, I cannot adequately expression my

appreciation for my parents and Madhvi for their love, encouragements and

emotional support during my study.

viii
TABLE OF CONTENTS

PAGE
ABSTRACT .............................................................................................................V

ACKNOWLEDGMENTS .......................................................................................VIII

TABLE OF CONTENTS .........................................................................................IX

LIST OF FIGURES .................................................................................................IX

LIST OF TABLES ...................................................................................................XI


I. INTRODUCTION ................................................................................................ 1
1. Hydrocracking process description ........................................................... 4

II. LITERATURE REVIEW ........................................................................................ 8


1. Lumped models......................................................................................... 9
2. Fundamental kinetic models ................................................................... 12

III. SINGLE EVENT KINETICS ................................................................................. 16


1. Theory of SEK......................................................................................... 22
2. Rate expressions .................................................................................... 26

IV. NETWORK GENERATION ................................................................................. 35

V. PARAMETER ESTIMATION AND RESULTS ........................................................... 67

REFERENCES...................................................................................................... 79

APPENDIX A......................................................................................................... 81

APPENDIX B......................................................................................................... 93

VITA .................................................................................................................... 113

ix
LIST OF FIGURES
FIGURE
Page

x
LIST OF TABLES

xi
CHAPTER I

INTRODUCTION

Petroleum refining industries is facing several environmental and market

requirements:

1. The demand for more environmentally acceptable products, low in sulfur,

nitrogen and unsaturated compounds, is increasing

2. The demand for gasoline and middle distillates (kerosene and diesel fuels)

is increasing while the value of heavy fuel oils is declining.

3. The middle distillates to gasoline ratio of products is increasing.

4. A high flexibility is required to respond to changing or seasonal market

demands and changing feeds on a weekly basis (Ward, 1993; Quann and

Jaffe, 1996; Minderhound and Van Veen, 1993).

The oil product demand from 2000 to 2005 in the Organization for Economic

Cooperation and Development and developing countries is shown in Table 1.1. It

has been reported that one of the fastest increases in 3 years from 2002 to 2005

will be in demand for middle distillates (Beck et al, 2001). The petroleum

industries require, therefore a process to convert heavy feedstocks, particularly

heavy gas oils, to higher value middle distillate. They can choose from several

processing options such as hydrocracking, fluid catalytic cracking and coking to

convert heavy feedstocks to middle distillate. However, hdyrocracking is the only

option to produce high quality middle distillate with high hydrogen content.

1
Table 1.1 Outlook for Petroleum demand for Organization for Economic Cooperation and Development (OECD)

and developing countries (Beck, 2001)

Product Demand (1000 b/d) % Change

2000 2001 2002 2003 2004 2005 (2002-2005)

Gasoline 21217 21490 21790 2220 22680 23140 6.2

Middle 24532 24562 24930 25430 25990 26510 6.3

Distillates

Fuel Oil 9958 9852 9990 10200 10390 10570 5.8

Other products 11243 11156 11390 11620 11880 12120 6.4

Total 66950 67060 68100 69470 70940 72340 6.2

2
The history of hydrocracking is found in paper written by Sullivan and Scott

(1983). Thermal hydrocracking is one of the oldest hydrocarbon conversion

processes. An extensive technology for coal conversion was developed in

Germany between 1915 and 1945. Germany needed a secure supply of liquid

fuels derived form coal. The quantity of liquid fuels was important but their quality

was not considered at that time. Therefore, typical reaction conditions were quite

severe: 21-69 MPa and 375-525 0C, giving liquid fuels that were highly aromatic

because of thermodynamic equilibrium.

There was a parallel development in the United States, which was directed

toward the upgrading heavier petroleum fractions. The reaction conditions were

somewhat milder but still severe: 21-31 MPa and >375 0C. After the end of World

War II, thermal hydrocracking became less popular because of the availability of

middle Eastern crude oils and development of new catalytic cracking processes.

During 1950’s, the trend toward automobile engines with high compression

ratios, the shift by the railroads from steam to diesel locomotives, an excess of

refractory cycle stocks produced by catalytic cracking processes, and

introduction of commercial jet aircraft led to increasing demand to convert heavy

fractions into less aromatic liquid fuels.

Once Chevron Research announced a new catalytic hdyrocracking process,

“Isocracking”, in response to this demand in the 1960’s, catalytic hdyrocracking

became a commercial reality and rapidly grew through development of zeolite


based catalyst and large pore catalysts, which permitted operation at moderate

temperatures (200-400 0C) and lower pressures (3-14 MPa). Seven different

hydrocracking processes were offered by the mid-1970’s. Hydrocracking has

grown at a slow pace in the United States since the late 1970’s; however,

worldwide hydrocracker capacity will grow to 3.9 MM barrels per day in 2006.

Hydrocracking Process Description

ISOCRACKING hydrocracker designs include single stage once through (SSOT),

single stage recycle (SSREC) and two stage recycle (TSRE) processes. A two

stage hydrocracker with intermediate distillation represents the most common

process configuration for maximizing middle distillates (Fig.1.1).

The vacuum gas oil is sent to the first stage of the hydrocracker and is severely

hydrotreated. Most of the sulfur and nitrogen compounds are removed from the

oil and many of the aromatics are saturated. In addition, significant conversion to

light products occurs in the first stage. The liquid product from the first stage is

sent to a common fractionation section. To prevent overcracking, lighter products

are removed by distillation. The unconverted oil from the bottom of the

fractionator is routed to the second stage reactor section.

The second reaction stage saturates almost all of the aromatics and cracks the

oil feed to light products. Due to the aromatics saturation, the second stage

4
Fig. 1.1. Process flow diagram for “Isocracking” Process

5
produces excellent quality products. The liquid product from the second stage is

then sent to the common fractionator where light products are distilled. The

second stage operates in a recycle to extinction mode with per-pass-conversions

ranging from 50% to 80%.

The overhead liquid and vapor from the hydrocracker fractionator is further

processed in a light ends recovery unit where fuel gas and liquefied petroleum

gas (LPG) and naphtha are separated. The hydrogen supplied to the reactor

sections of the hydrocracker comes from the steam reformers. The hydrogen is

compressed in stages until it reaches system pressure of reactor sections.

In general, hydrocracking is carried out at 300-450 0C catalyst bed temperature,

8.5-20 MPa pressure, 0.5-2.5 hr-1 liquid hourly space velocity, 505-1685 normal

m3/m3 hydrogen to oil ratio and 200-590 normal m3/m3 hydrogen consumption.

The major drawback of hydrocracking is its high price due to the use of high

pressure. Reactions occurring in hydrocracking are hydrogenation –

dehydrogenations and isomerization – cracking, which is accomplished by

catalyst metal function and acid function respectively. As sulfur and nitrogen

compounds contained in the feedstock have a deleterious effect on

hydrocracking catalysts, the feedstock usually requires hydrotreatment prior to

contact with the hydrocracking catalyst. The operating conditions and properties

of the feedstock have a significant impact on the yield and quality of the products.

For example, an increase in the operating temperature decreases the extent of

6
aromatics hydrogenation due to thermodynamic limitations (Dufresne et al.,

1987). Also, the presence of ammonia decreases the catalytic activity by

poisoning acid sites (Dufresne et al., 1989).

Hydrocracking can be designed with flexibility in the product slate to meet market

demands (Gosselink and Stork, 1997). The particular products of hydrocracking

are oil, petrochemical feedstocks, feedstocks for FCC and thermal crackers, and

basestock for lubricants. The feedstocks for hydrocracking were generally

conventional straight run heavy gas oils. As the market value of heavy oils is

decreasing, the heavy gas oils produced in residuum-upgrading processes such

as coker gas oils are becoming important feedstocks for hydrocracking (Sullivan

et al., 1989). As these heavy gas oils have much more nitrogen and sulfur

compounds and aromatics than conventional straight run heavy gas oils, they are

refractory feedstocks for hydrocracking. Also, the interest in processing

feedstocks from different origins, such as tar sands and coal, is growing because

world reserves of conventional crude oil are dwindling (Kim et al., 1997).

Hydrocracking will continue to play an important role in petroleum industries in

the future. Our goal is to develop a kinetic hydrocracking model for simulation

and optimization of industrial fixed bed reactors. Obtained kinetic model can also

be used for catalyst design, preliminary design of new hydrocracker units.

7
CHAPTER II

LITERATURE SURVEY

Modeling hydrocracking process has become necessary since hydrocracking has

become an important secondary process in refineries. Mathematical models are

needed to predict yields of various products for process design, optimization and

control applications. The feedstocks processed in the petroleum industries

consist of a large number of components. A typical feed for an industrial

hydrocracker unit contains paraffins, isoparaffins, naphthenes, aromatics and

naphtheno aromatic components. These components follow very complicated

reaction pathways with carbenium ion intermediates. Modeling such chemical

processes becomes very complex due to extremely large number of reactions

and difficulties in measuring feed and product compositions.

Kinetic models of hydrocracking process can be categorized as Lumped semi

empirical models and fundamental kinetic models. In early seventies, Weekman

and Nace from Mobil proposed a simple model containing three pseudo

components: Gasoline, Gas oil and light gases with 3 reactions for the catalytic

cracking of gas oil fractions. Jacob et al. (1976) revised this three lump model to

a ten lump model with 17 reactions. The limitation of these lump models is that

rate parameters vary with the type of feed and configuration of reactor. So,

extensive experimentation is required for each component if feed is a complex

mixture of hydrocarbons. Therefore rate parameters increase rapidly with number

8
of lumps. Hence, requirement of a fundamental kinetic model arose, which can

reflect actual mechanism keeping number of rate parameters tractable.

Various types of lump models have been proposed: a model based on a series of

fixed boiling range cuts by Stangeland (1974), the compound type distribution

approach of Krishna (1989), and the Langmuir-Hinshelwood-Hougen-Watson

model by Sundaram et al (1988).

Lumped model of Stangeland (1974)

Stangeland proposed an empirical equation to predict the product distribution of

hydrocracking. One parameter, A, describes the effect of boiling point on the rate

constant, k. Two other parameters, B and C, determine what products will be

generated as each cut cracks. The boiling point distributions of the feedstock and

product are segregated into fixed boiling ranges of 50oF each and the apparent

1st order rate constant k for each component was expressed by a function of the

true boiling point T b (oF) and one cracking rate parameter A as follows:

k (T )  T  A(T 3  T ) ……….(2.1)

Where, T is T b /1000. Experimental data indicate A=0.5. The product distribution

parameter B and butane yield parameter C were also empirically determined to

represent the obtained product distribution.

9
The product distillation curve was predicted with high accuracy using a feedstock

distillation and values of the 3 parameters, A, B and C. However, the parameters

had to be determined experimentally for each feedstock because the information

on boiling point and paraffin content of the feedstock was insufficient to establish

a universal hydrocracking model.

Sundaram Model (1988)

Sundaram et al (1988) analyzed hydrotreating kinetics based on batch reactor

data using a lumped model. The surface reaction was considered as the rate

determining step. Simplified Langmuir-Hinshelwood-Hougen-Watson model were

proposed for the hydroprocessing of quinoline, naphthalene and

dibenzothiophene.

ki ci
ri  n
……….(2.2)
 
1   K j c j 
 
 j 

Where, ki is ith reaction rate constant, ci is concentration of ith species, K j is

adsorption constant for jth species and n is 1 or 2. Then reaction networks for the

hydroprocessing of binary mixtures involving quinoline with naphthalene or

dibenzothiophene were obtained by the superposition of the individual

components based on the competitive adsorption mechanism. The model for the

mixture of quinoline and naphthalene was obtained without any modification for

10
each component. In the case of the model for the mixture of quinoline and

dibenzothiophene, a modification was needed for hydrogenolysis of sulfur

compounds because the sites of hydrogenation and hydrogenolysis of sulfur

compounds are different from each other.

Though the proposed model is still the “ideal hydrocracking” model, their concept

can be applied for the industrial feedstocks with some modifications because the

inhibitors are considered for the hydrocracking of hydrocarbons. Gosselink et al.

(1989) simply used nitrogen for inhibitor content because it was impossible to

determine adsorption constants of all nitrogen-containing compounds in the

industrial feedstocks. The total effect of inhibitors can be written as :

 
n
K l N    K l , j N j ………..(2.3)
j 1

Where, N and N j are total nitrogen and individual nitrogen species respectively.

This relationship is not rigorous, but as demonstrated by Gosselink et al (1989), it

has empirical value for real feedstocks, where the number of components is very

large.

As the reviewed models are adjusted for some particular conventional straight

run heavy gas oil or set of model compounds, it will be impossible to apply them

for the various feedstocks from different origins with significant adjustments and

accuracy.

11
Fundamental kinetic models

Fundamental kinetic models are based on actual chemistry of the hydrocracking

process. A new methodology has been developed by Froment and co-workers

using carbenium ion chemistry principles (Baltanas et al., 1989; Vynckier and

Froment, 1991). Baltanas and Froment (1985) used a computer algorithm to

generate the complete reaction network for hydrocracking of paraffins, taking in

to account all the reactions involving individual molecules. Vynckier and Froment

(1991) extended the single event approach to complex feedstocks and

introduced the concept of lumping coefficients to formulate rate expressions in a

convenient way. Feng et al. (1993) formalized the single event kinetics for

catalytic cracking and estimated single event rate parameters for catalytic

cracking of paraffins on a RE-Y zeolite catalyst.

Svoboda et al. (1995) estimated single event rate parameters from experiments

carried out on hexadecane and validated single event concept for the

hydrocracking of paraffins in a three phase reactor for the Fischer-Tropsch

process. Dewachtere et al. (1999) applied a single event kinetic model in the

simulation of an industrial riser reactor for the catalytic cracking of vacuum oil.

Martens et al. (2000) applied a single event kinetic model for the hydrocracking

of C 8 to C 12 paraffins on Pt/US-Y zeolites. Due to the molecular nature of the

12
approach, it was found that a finite and limited number of kinetic parameters

could describe the hydrocracking of heavy feedstocks.

Liguras and Allen (1989) developed an approach based on carbon centers, in

which, the model components are developed using linear programming. This

methodology was used to describe the reactions in catalytic cracking where it

was found that there were no significant differences in the product distribution for

different sets of pseudo components, as long as the number of pseudo

components is large enough. Klein et al. (1991) considered thousands of model

components and generated reaction networks for asphaltene hydroprocessing

using a Monte Carlo simulation approach.

Quann and Jaffe (1992) proposed a novel approach for modeling called Structure

Oriented Lumping (SOL). The lumping strategy is based on the molecular

structure of the feed and product components.

Structure Oriented Lumping (SOL)

Structure Oriented Lumping (SOL) represented individual molecules as vectors.

The vector elements are referred to as structural increments, which are three

aromatic and six naphthenic ring structures, a –CH 2 - group to specify the carbon

number of alkyl chains of paraffins, bridging between rings, hydrogen deficiency,

13
and heteroatom groups or substitutions in rings and alkyl chains. The vector

representation of molecules is based on the data analyzed by liquid

chromatography coupled with FIMS. As the information on the structure

arrangement of substituent groups or of longer alkyl chain cannot be obtained by

these analyses, all isomers at a specific carbon number were lumped. For

example, three isomers of xylene, which are o-xylene, m-xylene and p-xylene,

are in a group. It was assumed that the isomers of a molecular class at a given

carbon number would exhibit similar properties. The SOL method enabled the

development reaction network with a limited set of reaction rules such as a

reactant selection rule and a product generation rule. The algorithm to estimate

the product properties from the conventional characterization of the feedstock,

such as boiling point, sulfur component, specific gravity etc, is as follows:

1. Translate the characterization data into the model feedstock having

molecular composition which was previously analyzed in detail.

2. Predict the product composition from the feedstock composition referring

to the process chemistry.

3. Translate the product composition into the product properties.

To convert mass spectroscopic data into the vector expression, a set of

conventions, which were based on the general knowledge of straight run streams

derived from conventional crude oils, were used and the reaction networks for

the representative structures were proposed. Some conventions cannot be

14
applied for the streams from residuum upgrading processes such as coker gas

oil. For example, the alkyl groups on ring compounds are assumed to be a long

alkyl chain and an average of one methyl group per ring. However, the alkyl

groups of a coker gas oils are random because they are thermal cracked

asphaltene, alkylated aromatics, polymerized olefins and so on. Therefore, it will

be impossible to represent each group by one structure in the case of coker gas

oils.

The recent shift in research towards the fundamental kinetic models clearly

shows the importance of incorporating the actual chemistry in building high

fidelity models for the hydrocracking process. It is strongly believed that the

single event kinetics can contribute to the understanding of the complexities of

the hydrocracking process, provide economic and operational benefits to the

petroleum refineries, and stimulate further research both in academia and

industry.

15
CHAPTER III

SINGLE EVENT KINETICS

Chemical reactions of Hydrocracking Process

Hydrocracking in the petroleum industry is often carried out in two stages. In the

first stage, sulfur and nitrogen containing compounds are decomposed and

aromatics are hydrogenated. The liquid fraction coming form the first stage is

hydroisomerised and hydrocracked in the second stage.

The hydrocracking of petroleum fractions involves complex chemistry, which

includes hydrogenation-dehydrogenation, isomerization, carbon carbon bond

scission, paring reaction, ring opening, alkylation, disproportionation, and

cyclisaiton reactions. These reactions take place on a bifunctional catalyst

consisting of both cracking (silica /Alumina) and hydrogenation/dehydrogenation

metal components. The cracking catalysts such as SiO 2 -Al 2 O 3 and the zeolites

have been found to have both bronsted and Lewis acid sites (Gates et al., 1979).

The bronsted sites are responsible for proton donation and the Lewis sites for

accepting electron pairs.

The chemistry of hydrocracking is essentially the carbocation chemistry plus the

chemistry of hydrogenation and dehydrogenation. The carbocations can form by

16
several different paths. The protonation of a saturated hydrocarbon in the

presence of a superacid with very high proton donor strength generates a non

classical carbonium ion intermediate. The penta co-ordinated carbon atom loses

a hydrogen molecule to become a carbenium ion. Hydride abstraction from a

saturated hydrocarbon yields a carbenium ion on a Lewis site. Protonation of

olefinic intermediates (on the Bronsted sites) formed by dehydrogenation of the

metal generates a carbenium ion. The reaction path way of the olefins is still

generally accepted for reforming, hydroisomerization, and hydrocracking.

In the primary carbenium ions, the positively charged carbon is connected to two

hydrogen atoms and one alkyl group. The methyl carbenium ion is a primary

cation with three hydrogen atoms connected to the charged carbon. In secondary

carbenium ions, the carbon atom with the positive charge is connected to two

methyl groups and a hydrogen atom. Similarly, in tertiary ions, the charged

carbon atom is connected to three alkyl groups. The energy required for

carbenium ion formation increases with an increase in the number of hydrogen

atoms attached to the carbon atom carrying the positive charge. The stability of

the carbenium ion decreases in the order of increasing ionization energies, E + .

The tertiary carbenium ion is by far the most stable species, and it is the easiest

to form and most prevalent whenever it can be formed.

The most common feed to the hydrocracking is VGO. The VGO feed contains

paraffinic, naphthenic, aromatic and naphtheno-aromatic feed components. The

reactions for paraffins are given in fig 3.1. After physisorption in the zeolite

17
Fig.3.1. Typical Elementary steps for hydrocracking of Paraffins

18
Fig.3.2. Typical Elementary steps for hydrocracking of naphthenes

19
cages of a hydrocracking catalyst, the paraffins are dehydrogenated on the metal

component of the catalyst. The resulting olefins are protonated in the Bronsted

acid sites, yielding the alkane carbenium ions. These cations isomerize through

hydride shifts and methyl shifts. Hydride shifts and methyl shifts preserve the

degree of branching but it is hypothesized that protonated cyclopropane

branching (PCP) steps alter the number of side chains present in the reactant

molecule. Cracking occurs through scission of the carbon-carbon bond in the β-

position with respect to the carbon atom carrying the positive charge. Cracking

requires that the β-carbon with respect to the positively charged carbon be a

tertiary carbon atom. This ensures that the hydrocracking reactions avoid

producing or produce very little amount of methane and ethane in the cracked

products. After deprotonation of the alkane carbenium ions the resulting acyclic

olefins are hydrogenated to produce highly branched paraffin isomers with

reduced chain lengths.

The hydroisomerization and hydrocracking reactions for the naphthenic feed

components are shown in Fig. 3.2. Dehydrogenation of naphthenes on the metal

sites of the catalyst leads to the formation of cyclic mono-olefins, which are

protonated to cyclic carbenium ions. The cyclic carbenium ions isomerize through

hydride shift, methyl shift and PCP branching. Hydride and methyl shifts do not

alter the branching degree in naphthenes. However, methyl shifts alter the

relative positions of the constituents on the ring.

20
Three different kinds of PCP steps are possible in naphthenes. The first one,

known as acyclic PCP branching, which is similar to the PCP step in paraffins,

changes the branching degree in the alkyl side chain attached to the ring. The

second type is known as the intra-ring alkyl shift, which leads to ring contraction

or ring expansion without altering the branching degree of the naphthenic ring.

The third type is known as the cyclic PCP step, which causes the ring contraction

or ring expansion while altering the degree of branching of the naphthenic ring.

There are two kinds of β-scissions possible with the naphthenic carbenium ions.

If the β-scission occurs on the alkyl side chain or if the alkyl side chain is severed

from the ring, then it is known as the exocyclic β-scission. The cleavage of a

carbon-carbon bond that is part of the ring is known as an endocyclic scission.

Exocyclic β-scission produces an olefinic and a cationic species of reduced

carbon number whereas the endocyclic β-scission produces a single species

containing a double bond and a positively charged carbon atom. The alkene or

cyclo-alkene carbenium ions produced by endocyclic β-scission steps may

further undergo reactions similar to that of alkane and cycloalkane carbenium

ions. Compared to β-scission in an aliphatic chain, endocyclic β-scission is

known to proceed at a much slower rate.

21
Theory of Single Event Kinetics

The theoretical background of the single event kinetics approach is shown in

Figure 3.3. The effect of molecular structure on the rate coefficient of an

elementary step can be described by transition state theory and statistical

thermodynamics (Baltanas et al., 1989). A rate coefficient for the transformation

of a reactant in to a product via the activated complex is given by Erying

equation:

k T   S 0     H 0 
k '   b  exp  exp  ……. (3.1)
 h   R   RT 

The influence of the molecular structure on the rate coefficient can be described

in terms of the rotational contribution of the third law entropy S 0 . According to

statistical thermodynamics, the S 0 of a species is determined by several

contributions (Willems and Froment, 1988), due to the different motions of the

species such as translation, vibration and rotation and electronic:

S 0  Strans
0
 S vib
0
 S rot
0
 S ele
0
….(3.2)

The rotational contribution, S rot


0
consists of external and internal contributions,

i.e., S rot
0
 S ext
0
rot  S int rot . The S ext rot and S int rot are composed of two terms: the
0 0 0

22
Fig. 3.3. Theoretical Concept behind Single Event Kinetics (adapted from

Park,1998)

23
‘intrinsic’ value, Ŝ 0 and a term due to symmetry,  , which depends on the

geometry of the molecule.

rot  S ext rot  R ln  ext 


0
S ext ˆ0
……….(3.3)
S 0  Sˆ 0  R ln  
int rot int rot int

When optically active species are involved, the symmetry contribution to the

standard entropy has to be augmented with the contribution due to the mixing of

the different enantiomers. For racemic mixtures, this contribution is given by

R ln(2 n ) , where n is the number of the chiral centers in a species, and the term

2 n , is corresponding to the maximum number of stereoisomers that can exist. On

the basis of equation (3.3) with the effect of racemic mixtures, the rotational

contribution, S rot
0
is given by

  
0
S rot  Sˆrot
0
 R ln ext n int  ………(3.4)
 2 

Where, the intrinsic rotational contribution Ŝ rot , is the sum of Ŝ ext ˆ0


rot and S int rot .
0 0

The expression in the parenthesis in equation (3.4) is called global symmetry

number,  gl . The global symmetry number quantifies all symmetry contributions

of a species.

24
Inserting equation(3.4) in to equation(3.2), the standard entropy of activation

S 0 , which is the S 0 difference between the reactant and the activated complex

is given by

  glr 
S 0  Strans
0 0
 S vib 0
 S rot 0
 S ele  R ln   ……..(3.5)
 
 gl 

The last term of equation(3.5) is the difference in standard entropy between

reactant and activated complex due to symmetry changes. In the Eyring

equation, this contribution can be separated from the entropy of activation.

  glr  k T   Sˆ 0   0

 k '     b  exp  exp  H  ……..(3.6)
   h    
 gl   R   RT 

Where, Sˆ 0  S trans


0 0
 S vib 0
 S rot 0
 S ele

The rate coefficient of the elementary step k ' , can now be written as a multiple of
~
the single event rate coefficient, k (Vynckier and Froment,1991)

~
k '  ne k ………(3.7)

~
Where, the number of single events ne , and the single event rate coefficient k ,

are defined as

25
  glr 
ne    
 
 gl 
………….(3.8)
~  kbT   Sˆ 0    H 0 
&k  
 exp   
 h   exp RT 
 R   

Since the effect of a difference in symmetry, i.e., a difference in structure,

between the reactant and the activated complex has been factored out by

introducing the number of single events, ne , the single event rate coefficient now

truly characterizes the reaction itself at the fundamental level. Comparing this

expression for k ' with the rate coefficient in Arrhenius form indicates that the

activation energy ( Eact ) is corresponding to the enthalpy of activation ( H 0 ) or

potential energy difference between the activated complex and the reactant.

Calculation of number of single events

A simple example can be used to illustrate the need for distinguishing between

elementary steps and single events. Consider the 1,2-methyl shift between 2 –

methyl-3-heptyl carbenium ion and 3-methyl 2-heptyl carbenium ion shown in the

figure 3.2. The activated complex for a methyl shift is commonly regarded as an

alkyl bridged cation where the shifting methyl group is bound to two carbon

atoms via two electrons, three center bond. Both reactants are secondary

carbenium ions, but two methyl groups can shift in the forward reaction, while

only one in the reverse reaction. One might suspect the forward reaction to

26
*
+

* CH 3

*
ne = 4 ne = 2

Fig. 3.4. Calculation of number of single events for an elementary step

27
proceed at twice the rate of the reverse reaction. The global symmetry numbers

for –methyl-3-heptyl carbenium ion, 3-methyl 2-heptyl carbenium ion and

transition state are 33, 33/2 and 33/22 respectively. These global symmetry

numbers correspond to four single events in the forward reaction and two single

events in the reverse reaction. The forward reaction rate constant is twice as

large as the reverse, as expected.

Assumptions concerning single event rate parameters

The number of rate coefficients required to predict the product distribution of

paraffins hydrocracking can be reduced with number of simplifying, but generally

accepted assumptions.

1. Primary carbenium ions and the methyl ions are far less stable than

secondary and tertiary ions and are not retained in the network of

elementary steps.

2. The rate coefficient for isomerization of carbenium ions depend only on

the type of the reactant and the product requiring four single event rate
~ ~ ~ ~
coefficients, kisom ( s; s ), kisom ( s; t ), kisom (t ; s ), kisom (t ; t ) , for each type of

isomerization.

28
3. The rate coefficient of protonation is independent of the olefin involved but

does depend on the type of the resulting carbenium ion (s=secondary,


~ ~
t=tertiary), introducing two single event rate coefficients, k pr ( s ) and k pr (t ) .

4. The cracking rate coefficients depend only on the type of carbenium ions

involved, not on the olefin produced, leading to four single event rate
~ ~ ~ ~
coefficients, k cr ( s; s, O ), k cr ( s; t , O ), kcr (t ; s, O), kcr (t ; t , O) .

Reduction in number of parameters by thermodynamic constraints

While simple assumptions maintain the number of single event rate coefficients

within tractable range for the protonation, isomerization and cracking reactions,

the same cannot be accomplished for the deprotonation rate coefficients. The
~ ~
single event rate coefficients of deprotonation, k De ( s; Oij ) and k De (t ; Oij ) , depend

both on the type of carbenium ion and the character of the olefin Oij . This

condition leaves an overwhelming number of deprotonation coefficients, which

can only be reduced through a thermodynamic analysis. Baltanas et al (1989)

were able to greatly reduce the number of required deprotonation parameters by

employing the following two thermodynamic constraints:

~ ~ ~
k De (m; Oij )  k De (m; Or ) K isom (Or ; Oij ) ……..(3.9)

29
~ ~ ~
kisom (t ; s ) k De (t ; Or ) k pr ( s )
~  ~ ~ ……..(3.10)
kisom ( s; t ) k pr (t ) k De ( s; Or )

In the equation (3.9), Oij is any olefin, Or is a selected reference olefin of the

same carbon number, and m is either s or t. Or is an isomer with the double

bond preferably in a position that produces both a secondary and a tertiary

carbenium ion on protonation. Equations (3.9) and (3.10) are valid for any carbon

number n. This constraint reduces the number of independent single event


~ ~
deprotonation coefficients to two k De ( s; Or ,n ) and k De (t ; Or ,n ) , for each carbon

number.

Equation (3.10), which applies for all the isomerization steps, leads to two more

reductions in the number of parameters. First, concerning isomerization, equation

(3.10) reveals that the rate coefficient for (t ; s ) isomerization is not independent.

This constraint leaves only three parameters for each type of isomerization.
~ ~
Second, concerning deprotonation, the ratio of k De (t ; Or ) to k De ( s; Or ) in equation

(3.10) must be independent of the carbon number n since both the isomerization

and protonation parameters are assumed independent of the carbon number:


~
k De (t ; Or )
~  f ( n) …………..(3.11)
k De ( s; Or )

The final number of independent deprotonation parameters is arrived at through

judicious selection of a homologous series of reference olefins (Froment,1991).

30
The rates of deprotonation of the carbenium ions in the homologous series are

assumed equal. This assumption coupled with equation (3.10), reduces the

number of independent single event deprotonation coefficients to two,


~ ~
k De ( s; Or ) and k De (t ; Or ) , where Or is a reference olefin in the homologous series.

From C 5 onward the deprotonation coefficients for the 2-methyl-2-alkenes are

assumed to be equal, since they have the same structure. For the olefins up to

C 5 , the 1-alkenes are taken as reference olefins and the deprotonation coefficient

is assumed to be independent of the chain length. For C 5 , two references can be

chosen (2-methyl-2-butene and 1-pentene). They are connected through

equation (3.9).

Rate equations for Paraffin Hydrocracking

The kinetic rate equations for observables and surface intermediates can be

formulated in terms of the elementary steps (Baltanas et al.,1989)

First, a paraffin Pi physically adsorbed into the pore of zeolite catalyst.

Pi ( gas )  Pi (ads ) ………(3.12)

Hereafter, the (ads) indication is dropped for ease of notation. The

concentrations of the physisorbed paraffins are determined by a Langmuir

31
isotherm, as was concluded in previous lumped kinetic studies (Steijns and

Froment, 1981)

Csat K Li pi
C Pi  ………(3.13)
1   K Li pi
i

The adsorbed paraffin is dehydrogenated on the metal phase of the bifunctional

catalyst.

Pi  Oij  H 2 ………(3.14)

The olefin Oij is given a second index j because more than one olefin can be

formed from a single paraffin Pi . The (de)hydrogenation step is assumed to be in

quasi equilibrium; therefore the concentration of the olefin is calculated from

thermodynamic equilibrium.

K DH ,ij C Pi
COij  …………..(3.15)
pH 2

Protonation of olefins on Bronsted sites results in carbenium ions :

Oij  H   Rik ………(3.16)

32
Which in turn undergo isomerization and cracking steps:

Rik  Rlo ………(3.17)

Rlo  Rqr  Ouv ………(3.18)

After deprotonation of the carbenium ions and hydrogenation of the produced

olefins the resulting isoparaffins and cracked paraffins can desorb from the

zeolite.

The acid catalyzed steps are assumed to be rate determining, making the

reaction rates of the paraffins identical to the reaction rates of the corresponding

olefins on the acid sites:

RPi   ROij ………(3.19)


j

This can be derived from the pseudo steady state approximation for the olefins.

The rates of formation of the olefins on the acid sites can be expressed in terms

of carbenium ion concentrations as follows.


~ ~ ~
ROij   ne k De (mik ; Oij )C R    ne k pr (mik )CO C H    ne k cr (mlo ; mqr , Oij )C R 
ik ij lo
k k i o

…….(3.20)

33
Where mik is the type of (s or t) of carbenium ion Rik .The concentrations of the

carbenium ions are not directly accessible but can be calculated from the pseudo

steady state approximation for the carbenium ions.

~ ~ ~
RR    ne k pr (mik )CO C H    ne k De (mik ; Oij )C R    ne kisom (mik ; mlo )C R 
ik ij ik ik
k k i o
~ ~ ~
  ne kisom (mlo ; mik )C R   ne kcr (mik ; mlo , Ouv )C R    ne kcr (mlo ; mik , Ouv )C R   0
lo ik lo
i o i o i o

……..(3.21)

and a balance on the Bronsted sites :

Ct  C H    C R  ………(3.22)
ik
i k

The last two equations form a linear system which can be solved to yield the

unknown C R  ‘s and C H  , which in turn can be used to calculated the net rates of
ik

formation of the paraffins from equations (3.19) and (3.20). Similar equations to

those above can be developed for naphthenes and aromatics too.

34
CHAPTER IV

NETWORK GENERATION

The reaction network for the hydrocracking is practically impossible to be

constructed manually because of the overwhelming number of elementary steps

linked in an extremely complicated manner. For this reason, Clymans and

Froment (1984) developed a computer algorithm for the radical reactions

occurring in the thermal cracking of hydrocarbons. This algorithm has been

continuously updated by Prof. Froment’s research group depending on the type

of the catalytic processes.

Numerical representation of species

It is clear that such a large reaction network can only be generated by computer.

Baltanas and Froment first generated the elementary steps by means of Boolean

relation matrices, in combination with a characterization vector further

characterizing the chemical species. Implementing a computer algorithm for the

network generation requires a numerical representation of the species involved n

the elementary steps. Species involved in the elementary steps are represented

by a Boolean relation matrix based on binary representation (0 or 1). The bond

between carbon atoms in a molecule can be readily described by the binary

relation matrix. Note that hydrogen atoms in hydrocarbon are not necessarily to

35
be accounted for in this representation because they can be determined by the

corresponding carbon skeleton.

As a first step for the representation of the olefins or carbenium ions, the carbon

atoms of the species are numbered in an arbitrary sequence. Then the

dimension of the binary relation matrix is given by C N X C N , where C N is the

number of carbon atoms of the given species. The element of the matrix (I,j) is

determined by the existence of a bond between carbon atom I and carbon atom j.

if there is a carbon-carbon bon, a value of “1” is assigned. Otherwise, the

elements are set to “0”.

The constructed matrices based upon this rule are always symmetrical. The

Boolean relation matrix provides general characteristics of the molecule, as well

as the required information to reconstruct the hydrocarbon skeleton of molecule.

The dimension of the matrix provides the number of carbon atoms, while the

number of non-zero elements in a row or column indicates the type of each

carbon atom, i.e., whether it is a primary, secondary, tertiary and quaternary. The

Boolean relation matrices for 2-methyl-3-hexyl and 3-methyl-2-hexyl carbenium

ions is shown in figure 4.1 as an example.

The Boolean relation matrix does not provide all the information on the

hydrocarbon species, however. The two binary relation matrices for the olefin

and equivalent carbenium ion would be identical. This is because the information

on the position of double bond and the positively charged carbon atom can not

36
Fig.4.1 Representation of a methyl shift by means of Boolean relation matrices

and characterization vectors

37
be stored in the binary relation matrix. In addition, several binary relation

matrices can represent the same species because the carbon atoms are

numbered arbitrary.

In order to solve former problem, Vynckier and Froment(1991) introduced a

characterization (auxiliary) vector which represents the component

unequivocally. It consists of 3 rows. The position of positively charge atom is

specified in the first row. The type of carbon atom is given by the second row (1:

primary, 2: secondary, 3: tertiary, 4:quarternary), while the third row indicates the

nature of carbon atom (1: bridge-head aromatic, 2: bridge-head cyclo olefinic, 3:

bridge-head naphthenic, 4: aromatic, 5: cyclo-olefinic, 6: naphthenic, 7: acyclic

olefinic, 8: acyclic paraffinic). Figure 1 also shows the characterization vectors for

2-methyl-3-hexyl and 3-methyl-2-hexyl carbenium ions. The first row of H 1 and

H 2 is 3 and 2 because the positively charged carbon atoms in both ions are 3rd

and 2nd respectively. The type and nature of the carbon atoms are shown in

second and third row of the characterization vectors for both carbenium ions.

The non-uniqueness of the matrix representation can be overcome by

introducing following rules which specify the priority for numbering the various

carbon atoms.

1. The numbering should start at a primary carbon atom.

38
2. The starting atom is chosen so as to obtain the longest possible backbone

with the smallest possible side chain.

3. The lowest possible number is selected for the electron deficient atom.

4. The lowest possible number is given to the carbon atom with the lowest

code describing its nature.

5. The carbon atom of the main chain connected to the highest degree of

substitution is attributed the lowest possible number.

6. The carbon atom with the longest side chain is assigned by the lowest

possible number. These rules make it possible to number the carbon

atoms in the main chain unequivocally. Starting at the tertiary of

quaternary carbon atom with the lowest number, the carbon atoms in the

remaining branched chain are numbered from the main chain towards the

most remote primary carbon atom.

Based upon the definition of the characterization vector and the standardization

rules for numbering the carbon atom, a unique representation of species with

unequivocally defined carbon atoms becomes available for the network

generation program.

39
For species with a large number of carbon atoms the Boolean relation matrices

become very sparse, but they occupy considerable computer memory. The

characterization vector with the standardization rules provides a remarkable

reduction of the computer memory for the storage of the species of the reaction

network. To generate the elementary steps, however, the matrix representation is

required to obtain the necessary information, as will be discussed next.

Generation of the elementary steps

The total number of carbon atoms, the type of each carbon atom, and the

neighboring carbon atoms can be obtained from the Boolean relation matrix. By

means of elementary matrix operations, additional information can also be found.

Implementing reactions in the network generation program is based upon these

operations.

The β-neighbors of each carbon atom can be obtained by squaring the Boolean

relation matrices B.

Bn  B1  B1  I ……….(4.1)

The diagonal elements are irrelevant, since these amounts to double counting

the α-bond, so they are all set to zero. Higher order relations (  -positions, etc.)

can be obtained by

40
Bn  B  Bn 1  Bn  2 , n  3,4..... ….(4.2)

Elimination of a C-C bond in M n , representing β-scission results into two

separate matrices representing the two products of scission.

As shown in Fig.4.1 the isomer 3-methyl-2-hexyl carbenium ion is obtained from

the first by a methyl-shift. The methyl shift changes the structure of the

hydrocarbon chain, and hence the Boolean matrix. The first step is to generate

this type of elementary step is to locate a neighbor of the charge bearing carbon

atom with a methyl substitute. Then, a new bond is created between the methyl

group and the charge bearing carbon atom, and the previous bond of the methyl

group is broken. The positive charge is shifted to the carbon atom which

originally carried the methyl group. In the Boolean relation matrix B 1 , the element

(2,7) and (7,2) are set to zero, whereas in B 2 the element (3,7) and (7,3) are set

to one. The position of the positive charge is adapted in the characterization

vectors from H 1 to H 2 . The location of the tertiary carbon atom is changed also

and stored in the characterization vectors.

The network generation involves all the species, molecules and ions, and their

elementary steps yielding intermediates and products. The number of species in

the network tremendously increases with the number of C-atoms of the

components in the feed. The detailed reaction network results for Paraffins and

naphthenes are shown in the following plots.

41
Reaction Network for Paraffins

160

140

120
Number of Paraffins

100

80

60

40

20

0
3 4 5 6 7 8 9 10 11
Carbon number

Fig. 4.2. Number of Paraffins generated with respect to different Carbon


number of n-Paraffin in the feed

42
700

600

500
Number of Olefins

400

300

200

100

0
3 4 5 6 7 8 9 10 11
Carbon number

Fig. 4.3. Number of Olefins generated with respect to different Carbon


number of n-Paraffin in the feed

43
800

700

600
Number of Carbenium ions

500

400

300

200

100

0
3 4 5 6 7 8 9 10 11
Carbon number

Fig. 4.4. Number of Carbenium ions generated with respect to different


Carbon number of n-Paraffin in the feed

44
Reactions of Paraffins

1600

1400

1200
Protolytic Scissions

1000

800

600

400

200

0
3 4 5 6 7 8 9 10 11
Carbon number

Fig. 4.5. Number of Protolytic Scission reactions with respect to different


Carbon number of n-Paraffin in the feed

45
Reactions of Paraffins

700

600

500
Hydride Transfer

400

300

200

100

0
3 4 5 6 7 8 9 10 11
Carbon number

Fig. 4.6. Number of Hydride transfer reactions with respect to different


Carbon number of n-Paraffin in the feed

46
Reactions ofOlefins

1200

1000

800
Protonations

600

400

200

0
3 4 5 6 7 8 9 10 11
Carbon number

Fig. 4.7. Number of Protonation reactions with respect to different


Carbon number of n-Paraffin in the feed

47
Reactions of Carbenium ions

1200

1000

800
Hydride Shift

600

400

200

0
3 4 5 6 7 8 9 10 11
Carbon number

Fig. 4.8. Number of Hydride shift reactions with respect to different


Carbon number of n-Paraffin in the feed

48
Reactions of Carbenium ions

400

350

300

250
Methyl Shift

200

150

100

50

0
3 4 5 6 7 8 9 10 11
Carbon number

Fig. 4.9. Number of Methyl shift reactions with respect to different


Carbon number of n-Paraffin in the feed

49
Reactions of Carbenium ions

2500

2000
PCP Branchings

1500

1000

500

0
3 4 5 6 7 8 9 10 11
Carbon number

Fig. 4.10. Number of PCP branching reactions with respect to different


Carbon number of n-Paraffin in the feed

50
Reactions of Carbenium ions

800

700

600
Hydride Transfer

500

400

300

200

100

0
3 4 5 6 7 8 9 10 11
Carbon number

Fig. 4.11. Number of Hydride transfer reactions with respect to different


Carbon number of n-Paraffin in the feed

51
Reactions of Carbenium ions

300

250

200
β-scission

150

100

50

0
3 4 5 6 7 8 9 10 11
Carbon number

Fig. 4.12. Number of β-Scission reactions with respect to different


Carbon number of n-Paraffin in the feed

52
Reactions of Carbenium ions

1400

1200

1000
Deprotonations

800

600

400

200

0
3 4 5 6 7 8 9 10 11
Carbon number

Fig. 4.13. Number of Deprotonation reactions with respect to different


Carbon number of n-Paraffin in the feed

53
Reaction Network for Naphthenes

Network of Paraffins

100

90

80

70
Number of Paraffins

60

50

40

30

20

10

0
6 7 8 9 10
Carbon Number

Fig. 4.14. Number of Paraffins generated with respect to different Carbon


number of naphthenes in the feed
Here,

C 6 – n- Cyclohexane

C 7 – n- methyl Cyclohexane

C 8 – n- ethyl Cyclohexane

C 9 – n- propyl Cyclohexane

C 10 – n- butyl Cyclohexane

54
Network of Olefins

800

700

600
Number of Olefins

500

400

300

200

100

0
6 7 8 9 10
Carbon Number
Olefins Cyclic Olefins

Fig. 4.15. Number of olefins generated with respect to different Carbon


number of naphthenes in the feed

55
Network of Cyclic compounds

160

140
Number of Cyclic compounds

120

100

80

60

40

20

0
6 7 8 9 10
Carbon Number
Naphthenes Aromatics

Fig. 4.16. Number of Naphthenes and aromatics generated with respect


to different Carbon number of naphthenes in the feed

56
Network of Carbenium ions

900

800

700
Number of Carbenium ions

600

500

400

300

200

100

0
6 7 8 9 10
Carbon Number
n-Carbenium ions Cyclic carbenium ions Olefinic carbenium ions

Fig. 4.17. Number of carbenium ions generated with respect to different


Carbon number of naphthenes in the feed

57
Reaction of Naphthenes

900

800

700

600
Reactions

500

400

300

200

100

0
6 7 8 9 10
Carbon Number
Exocyclic Protolytic Scissions Endocyclic Protolytic Scissions Hydride Astractions

Fig. 4.18. Number of Naphthene reactions with respect to different


Carbon number of naphthenes in the feed

58
Reactions of cyclic olefins

1400

1200

1000
Protonations

800

600

400

200

0
6 7 8 9 10
Carbon Number

Fig. 4.19. Number of cyclic olefin reactions with respect to different


Carbon number of naphthenes in the feed

59
Reactions of cyclic carbenium ions

1400

1200

1000
Reactions

800

600

400

200

0
6 7 8 9 10
Carbon Number
Hydride shifts Methyl Shifts

Fig. 4.20. Number of cyclic carbenium ion reactions with respect to


different Carbon number of naphthenes in the feed

60
Reactions of cyclic carbenium ions

300

250

200
Alkyl shifts

150

100

50

0
6 7 8 9 10
Carbon Number
Alkyl shifts Intra ring alkyl shifts

Fig. 4.21. Number of Alkyl shift reactions for cyclic carbenium ions with
respect to different Carbon number of naphthenes in the feed

61
Reactions of cyclic carbenium ions

3500

3000

2500
PCP Branchings

2000

1500

1000

500

0
6 7 8 9 10
Carbon Number

Fig. 4.22. Number of PCP branching reactions for cyclic carbenium ions
with respect to different Carbon number of naphthenes in the feed

62
Reactions of cyclic carbenium ions

400

350

300

250
β scissions

200

150

100

50

0
6 7 8 9 10
Carbon Number
Exocyclic β scissions Endocyclic β scissions

Fig. 4.23. Number of β-Scission reactions for cyclic carbenium ions with
respect to different Carbon number of naphthenes in the feed

63
Reactions of cyclic carbenium ions

160

140

120
Aromatizations

100

80

60

40

20

0
6 7 8 9 10
Carbon Number

Fig. 4.24. Number of Aromatization reactions for cyclic carbenium ions


with respect to different Carbon number of naphthenes in the feed

64
Reactions of cyclic carbenium ions

1400

1200

1000
Deprotonations

800

600

400

200

0
6 7 8 9 10
Carbon Number

Fig. 4.25. Number of Deprotonation reactions for cyclic carbenium ions


with respect to different Carbon number of naphthenes in the feed

65
Reactions of cyclic carbenium ions

900

800

700
Hydride donations

600

500

400

300

200

100

0
6 7 8 9 10
Carbon Number

Fig. 4.26. Number of Hydride donation reactions for cyclic carbenium ions
with respect to different Carbon number of naphthenes in the feed

66
CHAPTER V

PARAMETER ESTIMATION AND RESULTS

In the kinetic modeling of Hydrocracking, it had been generally accepted that the

rates of carbenium ion reactions on the acid sites are rate determining. But, that

is not true for whole range of operating conditions. At severe operating conditions

higher temperatures and lower pressures, the product distribution is no longer a

unique function of conversion. The isomerization conversion decreases with

increasing temperature and decreasing pressure. The longer the chain length,

the more important this phenomenon. In such conditions, hydrocracking is

termed as “non ideal” cracking. The term ‘non ideal” has frequently been

associated with skewed curves for the product distributions but these are also

observed when secondary cracking occurs, which is always the case for longer

chain lengths, in particular at higher conversions. In this work “Non ideal” is only

used to express that the rate determining step is not only on the acid sites.

This happens because of a relative decrease in the rate of hydrogenation on

metal sites, compared to the rate of carbenium ion reactions on acid sites. In

other words, the reactions on acid sites are no longer rate determining and to

rates of hydrogenation reactions on metal sites have to be explicitly considered.

Consider a molecular scheme of hdyrocracking (Fig.5.1), in which paraffin A is

hydroisomerized and hydrocracked in to isoparaffin B. The rate expressions

based on Langmuir-Henshelwood-Hougen-Watson kinetics can be derived as

follows.

67
c Al
A( g )  l  Al K1 
p Acl
co1l cH 2l
Al  l  O1l  H 2l K2 
c Al cl
p H 2 cl
H 2l  H 2 ( g )  l K3 
c H 2l
po1 cl
O1l  O1  l K4 
co1l

 
cR1
O1  H  R K5 
po1 cH
1

 
cR2
R1  R K6 
cR1
2


R2  R3  O2
 
po2 cH
R  O2  H K8 
cR3
3

co2l
O2  l  O2l K9 
po2 cl
c H 2l
H 2( g )  l  H 2l K10 
pH 2 cl
cBl cl
O2l  H 2l  Bl  l K11 
co2l cH 2l
p B cl
Bl  B( g )  l K12 
cBl

68
Fig.5.1. Hydrocracking reaction scheme of paraffin A to isoparaffin B

Case – I: RDS is on acid sites (Cracking step)

r  kCR2

k ' C s A K L A p A Ct
 ………..(5.1)
 k k (1  k6 )Cs A K LA C s B K LB 
 p H  DH 5  


2
1   K Li pi 
K H 1   K Li pi  

Where,

k DH  k1k 2 k3 k 4
k H  k9 k10 k11k12
k '  kk5 k6 k DH

Case – II: RDS is on metal sites (Dehydrogenation step)

 po pH 
ki'  p pi  i 2 
 K DH i 
r  
 
2 ………..(5.2)
1   K i' M pi
 K H 2 pH 2  1   K Li pi
M
 

 1   Li i
K p 

Case – III: Most general case (No RDS, neither on acid catalyzed reactions, nor
on metal catalyzed reactions)

In this case, rate expressions are written for all the elementary steps according to

LHHW model and equated rates of all steps. The generalized solution is

mathematically complex and shown in appendix B.

69
Experimental evidence for hydrocracking of n-Hexadecane

The experiments at severe operating conditions (high temperature and lower

pressure) can be modeled if the hydrogenation-dehydrogenation reactions on

metal sites are explicitly considered. The change in isomerization conversion with

total conversion for hydrocracking of n-hexadecane is shown in figure 5.2. and

5.3. As we can see from these plots that at higher temperatures and lower

pressures, the isomerization conversion is significant and it is unique function of

total conversion. This is the case where RDS is on the acid sites. At higher

temperatures and lower pressures, isomerization conversion is rather low. Such

a behavior would be observed when the rate of the carbenium ion reactions on

the acid sites is no longer rate determining.

The reason that ‘non-ideal’ effects are more pronounced for heavier feed, is that

longer chain paraffins are more active on the acid sites, because they undergo

more reactions. (Debrabandere et al., 1997)

The isomer distribution of C 16 isomers in C 16 hydrocracking is shown in fig. 5.4.

For ‘ideal’ hdyrocracking conditions, monobranched isomers are the only primary

products and dibranched isomers are secondary products. While, for the ‘non-

ideal’ conditions, dibranched isomers and cracked products are produced directly

from n-hexadecane through hydrogenolysis.

70
n-Hexadecane Hydrocracking
Catalyst-2

30
25
20
Xiso

15
10
5
0
0 20 40 60 80 100
X

220 0C, 20 bar 200 0C, 10 bar 220 0C, 10 bar


200 0C, 20 bar Poly. (220 0C, 20 bar) Poly. (200 0C, 10 bar)
Poly. (220 0C, 10 bar) Poly. (200 0C, 20 bar)

Fig.5.2 Change in isomerization conversion with respect to change in total


conversion for n-hexadecane hydrocracking for catalyst-2

71
n-Hexadecane Hydrocracking
Catalyst-3

30

25

20
Xiso

15

10

0
0 20 40 60 80 100
X

240 0C, 10 bar 240 0C, 10 bar


Poly. (240 0C, 10 bar) Poly. (240 0C, 10 bar)

Fig.5.3 Change in isomerization conversion with respect to change in total


conversion for n-hexadecane hydrocracking for catalyst-2

72
Isomer distribution for C16 feed

100

80
Mole Fraction of C16

60

40

20

0
0 20 40 60 80 100
Total Conversion (X%)

nC16 MBC16 DBC16

Fig.5.4 Distribution of C16 isomers in n-hexadecane hydrocracking

73
Kinetic modeling of hydrocracking of n-hexadecane (Molecular Scheme)

As discussed earlier, the ‘non ideal’ conditions can be modeled by explicitly

accounting for rates of hydrogenations on metal sites. The reaction scheme for

the hydrocracking of C 16 is shown in fig. 5.5. The monobranched and

multibranched isomers are lumped, since they are in thermodynamic equilibrium

(Froment, 1985).

Derivation of rate expressions

Paraffins are physisorbed in zeolite pores and that can be described by Langmuir

isotherm as shown in chapter 3.

cs K Li pi
cPi  ………..(5.3)
1   K Li pi
i

Where, K Li is the physisorption coefficient for component i.

The rate of dehydrogenation on the metal sites can be expressed , according to

Dumez and Froment (1981) as :

 pO pH 
ki'  p pi  i 2 
 K DH i 
rDH i   '
………..(5.4)
DM

74
nC16 MB C16 mul C16
  
nC16 MB C16 mul C16
  
nC16 MB C16 mul C16   C3
21 22 23 4
C3 C3
C13
5
C13 C13
C4
6
C4 C4
C12
7
C 12 C12
C5
8
C5 C5
C11
9
C11 C11
C6
10
C6 C6
C10
11
C10 C10
C7
12
C7 C7
C9
13
C9 C9
C8
14
C8 C8

75
Here, ki'  k DH i kiM cM2 ,i cs K Li

2
 
 pi   
DM  1   K i
' 'M
 K H 2 pH 2  1   K Li pi 
M
………..(5.5)
 i 1   K Li pi   i 
 i 

The protonation of olefins is so fast compared to other reactions on acid sites

that it can be considered to be in pseudo equilibrium.

Oi  H   Ci

cCi
K Ci  ………..(5.6)
pOi cH

Here, is the chemisorption constant for the component i.

Now, A site balance on acid sites gives ,

 
ct  cH   cCi  cH 1   cOi K Ci  ………..(5.7)
i  i 

Now, we can write the rate expressions for all the carbenium ion reactions on the

acid sites as follows:

 cMBC  
r21  k 21  cnC   16  ………..(5.8)
 16 K 
 21 

76
 cmultiC  
r22  k 22  cMBC   16  ………..(5.9)
 K 22 
 
16

ri  ki cmultiC  ; i  23,24,25,26,27,28 ………..(5.10)


16

Combining equation (8) with equations (3),(6) and (7) that rate of monobranched

carbenium ions is :

 pMBC  
'  16 
k 21 pnC  
 16
 K 21 
'

r21  ………..(5.11)
1   K Li pi   pOi K Ci
i i

Where,

'
k 21  k 21 K c ct cs K L
nC16 nC16

K11 K cnC16 K L 
K 
'
11
nC16
………..(5.12)
Kc KL 
MBC16 MBC16

DA  1   K Li pi   pOi K Ci
i i

Rate equations for the net rate of formation of n-hexadecane paraffins and

olefins can be written as:

 pnC  pH 2 
 k1  pnC16  16 
 K 21' 
RnC16    ………..(5.13)
DM'

77
 pnC  p H 2   pMBC  
k1  pnC16  16  k 21
' 
pnC   16 
 K 21'   16 '
K 21 
RnC       ………..(5.14)
16
DM' DM'

Since the concentrations of the olefins are very low, pseudo steady state can be

assumed for all olefins.

ROi  0 ………..(5.15)

From equations (15), the partial pressures of the paraffins can be expressed as a

function of the partial pressures of the olefins.

Rearranging equation (14), the expression for partial pressure of n-hexadecane

is derived as:

p   k'
 H 2 k '11 DM '  
D '  ………..(5.16)
pn P    p 11 M
C nC  K1 k D  MBC   k K ' D 
16 16  1 A 16  1 11 A 

Similarly expressions for partial pressure of all the normal paraffins are derived

as shown in fig. 5.6. It leads to set of 12 non linear equations. These non linear

equations is combined with non linear continuity equations (for CSTR) to perform

parameter estimation of n-hexadecane hydrocracking in a Berty reactor, which is

a gas phase reactor with complete internal mixing.

78
P   k'
 H 2 k '11 DM '  
D ' 
Pn  PnC   P 
11 M

C 16 K k D MBC ' D
16  1 1 A  kK
16  1 11 A 

P  k' D '
 H2   13 M 
Pn  PnC    P
C 3 K mulC  k4 DA 
3  4 
  16
P  k' D '
 H2   13 M 
Pn  PnC    P
C 13  5  mulC  k5 DA 
K
13   16
P  k' D '
 H2 
Pn  PnC   P  14 M 
C
4 4  6  mulC  k6 DA 
K
  16
P   k' D '
 H 
Pn  PnC  2   P  14 M 
C 12 K mul C  k D 
12  7  16  7 A
 
P   k' D '
 H2   15 M 
Pn  PnC    P
C 5 K mul C  k D 
5  8  16  8 A
 
P  k' D '
 H2   15 M 
Pn  PnC    P
C 11 K mul C  k D 
11  9  16  9 A
 
P   k' D '
 H2   16 M 
Pn  PnC    P
C
6 6 K
 10  mulC  k10 DA 
  16
P   k' D '
 H2   16 M 
Pn  PnC    P
C 10  11  mulC  k11 DA 
K
10   16
P   k' D '
 H2   17 M 
Pn  PnC   P
C 7  12  mulC  k12 DA 
K
7   16
P
H   k' D '
Pn  PnC  2   P  17 M 
C 9 K mul C k D 
9  13  16  13 A
 
P   k'
 H2  D '
Pn  PnC    P  18 M 
C 8 K mul C k D 
8  14  16  14 A
 

Fig. 5.6 Rate equations for n-C 16 hydrocracking

79
The optimal values of the model parameters (Fig.5.6) in a model are determined

by minimizing an objective function with respect to the parameters.

 
 
 

n
f    jk   F j (i )  F j (i )  Fk (i )  Fk (i ) 
k 1 j 1 i 1   


Here, F j (i ) is the observed and F j (i ) is the predicted value of the response

variable for the ith experiment,  is the number of dependent variables. The

weights  jk consist of the elements of the inverse of the error covariance matrix.

A MATLAB code was developed for the Levenberg Marquardt (LM) optimization

algorithm (Appendix A). The LM algorithm is used for minimization because of its

ability to change the distance and direction of every optimization step. The initial

guess for the parameters is made from the n-dodecane hydrocracking parameter

values cited by Debrabandere et al. (1997).

80
APPENDIX A

LEVENBERG MARQUARDT OPTIMISATION ALGORITHM

function [f,p,kvg,iter,corp,covp,covr,stdresid,Z,r2]= ...

leasqr(x,y,pin,F,stol,niter,wt,dp,dFdp,options)

vernum= sscanf(version,'%f');

if vernum(1) >= 4,

global verbose

plotcmd='plot(x(:,1),y,''+'',x(:,1),f); figure(gcf)';

else

plotcmd='plot(x(:,1),y,''+'',x(:,1),f); shg';

end;

if (exist('OCTAVE_VERSION'))

global verbose

plotcmd='plot(x(:,1),y,"+;data;",x(:,1),f,";fit;");';

end;

% x=vec or mat of indep variables, 1 row/observation: x=[x0 x1....xm]

% y=vec of obs values, same no. of rows as x.

% wt=vec(dim=length(x)) of statistical weights. These should be set

% to be proportional to (sqrt of var(y))^-1; (That is, the covariance

% matrix of the data is assumed to be proportional to diagonal with diagonal

81
% equal to (wt.^2)^-1. The constant of proportionality will be estimated.),

% default=ones(length(y),1).

% pin=vector of initial parameters to be adjusted by leasqr.

% dp=fractional incr of p for numerical partials,default= .001*ones(size(pin))

% dp(j)>0 means central differences.

% dp(j)<0 means one-sided differences.

% Note: dp(j)=0 holds p(j) fixed i.e. leasqr wont change initial guess: pin(j)

% F=name of function in quotes,of the form y=f(x,p)

% dFdp=name of partials M-file in quotes default is prt=dfdp(x,f,p,dp,F)

% stol=scalar tolerances on fractional improvement in ss,default stol=.0001

% niter=scalar max no. of iterations, default = 20

% options=matrix of n rows (same number of rows as pin) containing

% column 1: desired fractional precision in parameter estimates.

% Iterations are terminated if change in parameter vector (chg) on two

% consecutive iterations is less than their corresponding elements

% in options(:,1). [ie. all(abs(chg*current parm est) < options(:,1))

% on two consecutive iterations.], default = zeros().

% column 2: maximum fractional step change in parameter vector.

% Fractional change in elements of parameter vector is constrained to be

% at most options(:,2) between sucessive iterations.

% [ie. abs(chg(i))=abs(min([chg(i) options(i,2)*current param estimate])).],

% default = Inf*ones().

82
% OUTPUT VARIABLES

% f=vec function values computed in function func.

% p=vec trial or final parameters. i.e, the solution.

% kvg=scalar: =1 if convergence, =0 otherwise.

% iter=scalar no. of interations used.

% corp= correlation matrix for parameters

% covp= covariance matrix of the parameters

% covr = diag(covariance matrix of the residuals)

% stdresid= standardized residuals

% Z= matrix that defines confidence region

% r2= coefficient of multiple determination

if(exist('verbose')~=1), %If verbose undefined, print nothing

verbose=0; %This will not tell them the results

end;

if (nargin <= 8), dFdp='dfdp'; end;

if (nargin <= 7), dp=.001*(pin*0+1); end; %DT

if (nargin <= 6), wt=ones(length(y),1); end; % SMB modification

if (nargin <= 5), niter=20; end;

if (nargin == 4), stol=.0001; end;

83
y=y(:); wt=wt(:); pin=pin(:); dp=dp(:); %change all vectors to columns

% check data vectors- same length?

m=length(y); n=length(pin); p=pin;[m1,m2]=size(x);

if m1~=m ,error('input(x)/output(y) data must have same number of rows ') ,end;

if (nargin <= 9),

options=[zeros(n,1), Inf*ones(n,1)];

nor = n; noc = 2;

else

[nor, noc]=size(options);

if (nor ~= n),

error('options and parameter matrices must have same number of rows'),

end;

if (noc ~= 2),

options=[options(noc,1), Inf*ones(noc,1)];

end;

end;

pprec=options(:,1);

maxstep=options(:,2);

% set up for iterations

84
f=feval(F,x,p); fbest=f; pbest=p;

r=wt.*(y-f);

sbest=r'*r;

nrm=zeros(n,1);

chgprev=Inf*ones(n,1);

kvg=0;

epsLlast=1;

epstab=[.1, 1, 1e2, 1e4, 1e6];

% do iterations

for iter=1:niter,

pprev=pbest;

prt=feval(dFdp,x,fbest,pprev,dp,F);

r=wt.*(y-fbest);

sprev=sbest;

sgoal=(1-stol)*sprev;

for j=1:n,

if dp(j)==0,

nrm(j)=0;

else

prt(:,j)=wt.*prt(:,j);

nrm(j)=prt(:,j)'*prt(:,j);

85
if nrm(j)>0,

nrm(j)=1/sqrt(nrm(j));

end;

end

prt(:,j)=nrm(j)*prt(:,j);

end;

% above loop could ? be replaced by:

% prt=prt.*wt(:,ones(1,n));

% nrm=dp./sqrt(diag(prt'*prt));

% prt=prt.*nrm(:,ones(1,m))';

[prt,s,v]=svd(prt,0);

s=diag(s);

g=prt'*r;

for jjj=1:length(epstab),

epsL = max(epsLlast*epstab(jjj),1e-7);

se=sqrt((s.*s)+epsL);

gse=g./se;

chg=((v*gse).*nrm);

% check the change constraints and apply as necessary

ochg=chg;

for iii=1:n,

if (maxstep(iii)==Inf), break; end;

chg(iii)=max(chg(iii),-abs(maxstep(iii)*pprev(iii)));

86
chg(iii)=min(chg(iii),abs(maxstep(iii)*pprev(iii)));

end;

if (verbose & any(ochg ~= chg)),

disp(['Change in parameter(s): ', ...

sprintf('%d ',find(ochg ~= chg)), 'were constrained']);

end;

aprec=abs(pprec.*pbest); %---

% ss=scalar sum of squares=sum((wt.*(y-f))^2).

if (any(abs(chg) > 0.1*aprec)),%--- % only worth evaluating function if

p=chg+pprev; % there is some non-miniscule change

f=feval(F,x,p);

r=wt.*(y-f);

ss=r'*r;

if ss<sbest,

pbest=p;

fbest=f;

sbest=ss;

end;

if ss<=sgoal,

break;

end;

end; %---

end;

87
epsLlast = epsL;

if (verbose),

eval(plotcmd);

end;

if ss<eps,

break;

end

aprec=abs(pprec.*pbest);

% [aprec, chg, chgprev]

if (all(abs(chg) < aprec) & all(abs(chgprev) < aprec)),

kvg=1;

if (verbose),

fprintf('Parameter changes converged to specified precision\n');

end;

break;

else

chgprev=chg;

end;

if ss>sgoal,

break;

end;

end;

88
% set return values

p=pbest;

f=fbest;

ss=sbest;

kvg=((sbest>sgoal)|(sbest<=eps)|kvg);

if kvg ~= 1 , disp(' CONVERGENCE NOT ACHIEVED! '), end;

% CALC VARIANCE COV MATRIX AND CORRELATION MATRIX OF

PARAMETERS

% re-evaluate the Jacobian at optimal values

jac=feval(dFdp,x,f,p,dp,F);

msk = dp ~= 0;

n = sum(msk); % reduce n to equal number of estimated parameters

jac = jac(:, msk); % use only fitted parameters

if vernum(1) >= 4,

Q=sparse(1:m,1:m,(0*wt+1)./(wt.^2)); % save memory

Qinv=inv(Q);

else

Q=diag((0*wt+1)./(wt.^2));

Qinv=diag(wt.*wt);

89
end;

resid=y-f; %un-weighted residuals

covr=resid'*Qinv*resid*Q/(m-n); %covariance of residuals

Vy=1/(1-n/m)*covr; % Eq. 7-13-22, Bard %covariance of the data

jtgjinv=inv(jac'*Qinv*jac); %argument of inv may be singular

covp=jtgjinv*jac'*Qinv*Vy*Qinv*jac*jtgjinv; % Eq. 7-5-13, Bard %cov of parm est

d=sqrt(abs(diag(covp)));

corp=covp./(d*d');

covr=diag(covr); % convert returned values to compact storage

stdresid=resid./sqrt(diag(Vy)); % compute then convert for compact storage

Z=((m-n)*jac'*Qinv*jac)/(n*resid'*Qinv*resid);

%%% alt. est. of cov. mat. of parm.:(Delforge, Circulation, 82:1494-1504, 1990

%%disp('Alternate estimate of cov. of param. est.')

%%acovp=resid'*Qinv*resid/(m-n)*jtgjinv

%Calculate R^2 (Ref Draper & Smith p.46)

r=corrcoef(y,f);

if (exist('OCTAVE_VERSION'))

r2=r^2;

90
else

r2=r(1,2).^2;

end

if (verbose),

eval(plotcmd);

disp(' Least Squares Estimates of Parameters')

disp(p')

disp(' Correlation matrix of parameters estimated')

disp(corp)

disp(' Covariance matrix of Residuals' )

disp(covr)

disp(' Correlation Coefficient R^2')

disp(r2)

sprintf(' 95%% conf region: F(0.05)(%.0f,%.0f)>=

delta_pvec''*Z*delta_pvec',n,m-n)

% runs test according to Bard. p 201.

91
n1 = sum((f-y) < 0);

n2 = sum((f-y) > 0);

nrun=sum(abs(diff((f-y)<0)))+1;

if ((n1>10)&(n2>10)), % sufficent data for test?

zed=(nrun-(2*n1*n2/(n1+n2)+1)+0.5)/(2*n1*n2*(2*n1*n2-n1-n2)...

/((n1+n2)^2*(n1+n2-1)));

if (zed < 0),

prob = erfc(-zed/sqrt(2))/2*100;

disp([num2str(prob),'% chance of fewer than ',num2str(nrun),' runs.']);

else,

prob = erfc(zed/sqrt(2))/2*100;

disp([num2str(prob),'% chance of greater than ',num2str(nrun),' runs.']);

end;

end;

end;

92
APPENDIX B

RATE EXPRESSION FOR MOST GENERAL CASE

 2
  k1  k3  k4  k11  k8  ( k13 k1 paph pm k13 )
  
 k3 k4 k13 k12 k14 k15 k22 k19 

r2   
 k2 k14 k15 ( 1k1 pak3 phk4 pmk11 pbk8 pn ) 2
  k1  k3  k4  k11  k8   1 k13  k3 k4 k13 pm  k3 k4 k13 ph  
  
k12 k14 k15 k22 k19   k2 k15  
 
k12 k2 k14

1k1 pak3 phk4 pmk11 pbk8 pn 

  k13 k3 k4 k13 pm k3 k4 k13 ph 
 ct  1   
  k12 k2 k14 k2 k15 
r
 1 k1 pa  k3 ph k4 pm  k11 pb k8 pn

2 ct  
k1 k3 k4 k11 k8  
     ( k13 k1 paph pm k13 ) 
 k12 k14 k15 k22 k19  
 
( 1k1 pak3 phk4 pmk11 pbk8 pn ) 2 

ct2 ( k13 k1 paph pm k13 )
 0
( 1k1 pak3 phk4 pmk11 pbk8 pn ) 2

Solution of Rate equation : ( r )

1
( ( k14 k15 k22 k19 k13 k2 k1 pasqrt( 2 k142 k152 k222 k192 k132 k22 k1 pa k11 pb
2
2 k152 k222 k192 k132 k22 k142 k8 pnk152 k222 k192 k132 k22 k142 k112 pb2
2 k152 k222 k192 k132 k22 k142 k11 pb2 k152 k222 k192 k132 k22 k142 k3 ph
2 k142 k152 k222 k192 k132 k22 k1 pa
k152 k222 k192 k32 k42 k132 pm2 k122 k82 pn2
2 k152 k222 k192 k32 k42 k132 pm2 k122 k8 pn
2 k152 k222 k192 k3 k43 k132 pm3 k12 k2 k14
4 k15 k222 k192 k33 k43 k132 pm2 k122 k14 ph2
4 k15 k222 k192 k32 k43 k132 pm2 k122 k14 ph k8 pn
4 k15 k222 k192 k32 k43 k132 pm2 k122 k14 ph
2 k152 k222 k192 k32 k43 k132 pm3 k122 k1 pa

93
4 k15 k222 k192 k32 k43 k132 pm2 k122 k14 ph k11 pb
4 k152 k222 k192 k3 k42 k13 pm2 k122 k14 k2 k11 pb
2 k152 k222 k192 k32 k43 k132 pm3 k122 k8 pn
8 k15 k222 k192 k32 k122 k132 k2 k1 pa k14 k4 ph k8 pn
2 k152 k222 k192 k33 k43 k132 pm3 k122 ph
2 k152 k222 k192 k3 k43 k13 pm3 k122 k14 k2
4 k15 k222 k192 k33 k122 k132 k2 k1 pa k14 k4 ph2
8 k15 k222 k192 k32 k122 k132 k2 k1 pa k14 k4 ph
8 k15 k222 k192 k32 k122 k132 k2 k1 pa k14 k4 ph k11 pb
k152 k222 k192 k32 k44 k132 pm4 k122
4 k142 k15 k222 k192 k132 k2 k1 pa k3 k4 ph k12 k8 pn
2 k142 k152 k222 k192 k132 k22 k1 pa k4 pm
2 k152 k222 k192 k32 k43 k132 pm3 k122
4 k142 k15 k222 k192 k132 k2 k1 pa k3 k4 ph k12 k11 pb
4 k142 k15 k222 k192 k132 k2 k1 pa k32 k4 ph2 k12
4 k142 k152 k222 k192 k13 k22 k1 pa k12 k4 pm
4 k14 k152 k222 k192 k132 k2 k1 pa k3 k4 pm k12
4 k142 k152 k222 k192 k13 k22 k1 pa k12 k11 pb
4 k14 k152 k222 k192 k132 k2 k1 pa k32 k4 pm k12 ph
2 k14 k152 k222 k192 k132 k2 k12 pa2 k12 k3 k4 pm
4 k142 k15 k222 k192 k132 k2 k1 pa k3 k4 ph k12
k142 k152 k222 k192 k132 k22 k12 pa2
4 k14 k152 k222 k192 k132 k2 k1 pa k3 k42 pm2 k12
4 k14 k152 k222 k192 k132 k2 k1 pa k3 k4 pm k12 k8 pn
2 k142 k222 k192 k33 k42 k132 ph3 k122
4 k15 k142 k222 k192 k122 k2 k42 pm k32 k13 ph2
4 k15 k142 k222 k192 k122 k2 k42 pm k3 k13 ph k8 pn
4 k15 k142 k222 k192 k122 k2 k42 pm k3 k13 ph k11 pb
4 k15 k142 k222 k192 k122 k2 k42 pm k3 k13 ph
4 k152 k14 k222 k192 k122 k2 k42 pm2 k3 k13 k1 pa
k152 k142 k222 k192 k122 k22 k42 pm22 k152 k142 k222 k192 k12 k22 k42 pm2 k13
k152 k222 k192 k132 k22 k142 k42 pm2
4 k15 k142 k222 k192 k122 k2 k11 pb k3 k4 k13 ph k8 pn
4 k152 k142 k222 k192 k12 k22 k11 pb k13 k4 pm
4 k15 k142 k222 k192 k122 k2 k11 pb k32 k4 k13 ph2
4 k152 k14 k222 k192 k122 k2 k11 pb k3 k4 k13 pm k1 pa

94
2 k15 k142 k222 k192 k122 k2 k112 pb2 k3 k4 k13 ph
4 k15 k142 k222 k192 k122 k2 k11 pb k3 k4 k13 ph
2 k152 k142 k222 k192 k122 k22 k11 pb k4 pm
4 k152 k222 k192 k32 k42 k132 pm2 k12 ph k2 k14
2 k15 k222 k192 k34 k42 k132 pm k122 ph3 k14
2 k152 k222 k192 k33 k42 k132 pm2 k122 ph k1 pa
4 k15 k222 k192 k33 k42 k132 pm k122 ph2 k14 k11 pb
k152 k142 k222 k192 k122 k22 k112 pb2
4 k152 k222 k192 k32 k4 k13 pm k122 ph k14 k2 k11 pb
4 k152 k222 k192 k32 k42 k13 pm2 k122 ph k14 k2
k152 k222 k192 k34 k42 k132 pm2 k122 ph2
4 k15 k222 k192 k33 k42 k132 pm k122 k14 ph2
2 k152 k222 k192 k32 k42 k132 pm2 k122 k1 pa
4 k15 k222 k192 k32 k42 k132 pm k122 k14 ph k11 pb
2 k15 k222 k192 k32 k42 k132 pm k122 k14 ph
4 k152 k222 k192 k3 k42 k132 pm2 k12 k2 k14
4 k152 k222 k192 k3 k4 k13 pm k122 k14 k2 k11 pb
4 k152 k222 k192 k3 k42 k13 pm2 k122 k14 k2
k152 k222 k192 k32 k42 k132 pm2 k1222 k152 k222 k192 k33 k42 k132 pm2 k122 ph
4 k15 k132 k2 k222 k192 k33 k122 ph3 pm k14 k4
8 k15 k132 k2 k222 k192 k32 k122 ph2 pm k14 k4
8 k15 k132 k2 k222 k192 k32 k122 ph2 pm k14 k4 k8 pn
8 k15 k132 k2 k222 k192 k32 k122 ph2 pm k14 k4 k11 pb
2 k15 k222 k192 k32 k42 k132 pm k122 k82 pn2 k14 ph
4 k15 k222 k192 k33 k42 k132 pm k122 k8 pn k14 ph2
4 k15 k222 k192 k32 k42 k132 pm k122 k8 pn k14 ph k11 pb
4 k15 k222 k192 k32 k42 k132 pm k122 k8 pn k14 ph
4 k152 k222 k192 k3 k42 k132 pm2 k12 k8 pn k2 k14
2 k152 k222 k192 k32 k42 k132 pm2 k122 k8 pn k1 pa
4 k152 k222 k192 k3 k42 k13 pm2 k122 k8 pn k14 k2
2 k152 k222 k192 k33 k42 k132 pm2 k122 k8 pn ph
4 k152 k222 k192 k3 k4 k13 pm k122 k8 pn k14 k2 k11 pb
8 k15 k222 k192 k32 k122 k132 k2 k12 pa2 k14 k4 ph
8 k15 k222 k192 k32 k122 k132 k2 k1 pa k14 k42 ph pm
4 k142 k152 k222 k192 k13 k22 k1 pa k12 k3 ph
2 k142 k152 k222 k192 k132 k22 k1 pa k3 ph

95
4 k142 k152 k222 k192 k13 k22 k1 pa k12 k8 pn
4 k14 k152 k222 k192 k132 k2 k1 pa k3 k4 pm k12 k11 pb
2 k142 k15 k222 k192 k132 k2 k12 pa2 k12 k3 k4 ph
2 k142 k152 k222 k192 k132 k22 k1 pa k8 pn
4 k142 k15 k222 k192 k132 k2 k1 pa k3 k42 ph k12 pm
k142 k222 k192 k32 k42 k132 ph2 k122 k82 pn2
k152 k222 k192 k132 k22 k142 k82 pn22 k152 k142 k122 k222 k192 k22 k3 ph
4 k142 k152 k222 k192 k13 k22 k1 pa k12
2 k142 k222 k192 k33 k42 k132 ph3 k122 k8 pn
4 k15 k222 k192 k132 k2 k142 k42 pm k3 ph k12 k8 pn
4 k15 k222 k192 k132 k2 k142 k42 pm k32 ph2 k12
2 k142 k222 k192 k32 k42 k132 ph2 k122 k8 pn
4 k142 k222 k192 k3 k42 k132 ph k12 k15 k2 pm
k142 k222 k192 k32 k42 k132 ph2 k122
2 k142 k222 k192 k33 k42 k132 ph3 k122 k11 pb
2 k142 k222 k192 k32 k42 k132 ph2 k122 k11 pb k8 pn
2 k142 k222 k192 k32 k42 k132 ph2 k122 k11 pb
4 k142 k222 k192 k3 k42 k132 ph k12 k11 pb k15 k2 pm
k142 k222 k192 k34 k42 k132 ph4 k122
k142 k222 k192 k32 k42 k132 ph2 k122 k112 pb2
4 k15 k122 k222 k192 k32 k42 k132 pm k1 pa k14 ph
4 k15 k122 k222 k192 k33 k42 k132 pm k1 pa k14 ph2
4 k15 k122 k222 k192 k32 k42 k132 pm k1 pa k14 ph k8 pn
4 k15 k122 k222 k192 k32 k42 k132 pm k1 pa k14 ph k11 pb
2 k142 k152 k222 k192 k13 k22 k12 pa2 k12
4 k152 k222 k192 k13 k22 k142 k12 k3 ph
k152 k122 k222 k192 k32 k42 k132 pm2 k12 pa2
4 k152 k222 k192 k32 k4 k13 pm k122 ph k14 k2 k1 pa
2 k152 k142 k222 k192 k12 k22 k112 pb2 k13
2 k152 k222 k192 k33 k42 k132 pm2 k122 ph k11 pb
2 k152 k222 k192 k33 k4 k132 pm k12 ph2 k14 k2
2 k152 k222 k192 k33 k4 k13 pm k122 ph2 k14 k2
4 k152 k142 k222 k192 k12 k22 k11 pb k13
4 k152 k222 k192 k32 k4 k132 pm k12 ph k2 k14 k11 pb
4 k152 k222 k192 k32 k4 k13 pm k122 k14 k2 ph
4 k152 k222 k192 k3 k4 k13 pm k122 k14 k2 k1 pa

96
8 k15 k222 k192 k3 k42 k132 pm2 k122 k14 k2 ph
2 k152 k222 k192 k32 k42 k132 pm2 k122 k11 pb
4 k152 k222 k192 k32 k4 k132 pm k12 k14 k2 ph
2 k152 k142 k222 k192 k122 k22 k11 pb2 k152 k222 k192 k3 k4 k13 pm k122 k14 k2
4 k152 k222 k192 k3 k4 k132 pm k12 k2 k14 k11 pb
2 k152 k222 k192 k3 k4 k132 pm k12 k2 k14
8 k15 k222 k192 k3 k42 k132 pm k122 k14 k2 k1 pa
8 k15 k132 k2 k222 k192 k32 k122 ph2 pm k14 k4 k1 pa
8 k15 k222 k192 k3 k42 k132 pm2 k122 k8 pn k14 k2 ph
2 k152 k142 k222 k192 k122 k22 k4 pm
8 k15 k132 k2 k222 k192 k32 k122 ph2 pm2 k14 k42
2 k152 k222 k192 k32 k42 k132 pm2 k122 k8 pn k11 pb
4 k152 k222 k192 k32 k4 k13 pm k122 k8 pn k14 k2 ph
4 k152 k222 k192 k3 k4 k13 pm k122 k8 pn k14 k2
4 k152 k222 k192 k3 k4 k13 pm k122 k8 pn k14 k2 k1 pa
4 k152 k142 k222 k192 k12 k22 k4 pm k13
2 k152 k222 k192 k3 k4 k132 pm k12 k82 pn2 k2 k14
2 k152 k222 k192 k3 k4 k13 pm k122 k82 pn2 k14 k2
4 k152 k222 k192 k3 k4 k132 pm k12 k8 pn k2 k14 k11 pb
4 k152 k222 k192 k3 k4 k132 pm k12 k8 pn k2 k14
8 k15 k222 k192 k3 k42 k132 pm k122 k8 pn k14 k2 k1 pa
4 k15 k222 k192 k3 k43 k132 pm3 k122 k14 k2 ph
4 k152 k222 k192 k32 k4 k132 pm k12 k8 pn k14 k2 ph
2 k15 k222 k192 k32 k44 k132 pm3 k122 k14 ph
4 k152 k222 k192 k3 k42 k132 pm2 k12 k2 k14 k11 pb
4 k15 k222 k192 k3 k43 k132 pm2 k122 k14 k2 k1 pa
4 k15 k222 k192 k32 k43 k132 pm2 k122 k14 ph k1 pa
2 k152 k222 k192 k32 k43 k132 pm3 k122 k11 pb
4 k142 k222 k192 k32 k4 k13 ph2 k122 k15 k2
2 k142 k222 k192 k3 k4 k13 ph k122 k15 k2
4 k142 k222 k192 k32 k4 k132 ph2 k12 k15 k2
4 k142 k222 k192 k3 k4 k132 ph k12 k15 k2 k8 pn
4 k142 k222 k192 k3 k4 k13 ph k122 k15 k2 k8 pn
4 k152 k222 k192 k13 k22 k142 k12 k8 pn
2 k14 k222 k192 k32 k42 k132 ph k122 k112 pb2 k15 pm
4 k142 k222 k192 k32 k4 k132 ph2 k12 k11 pb k15 k2

97
2 k142 k222 k192 k32 k43 k132 ph2 k122 pm
4 k142 k222 k192 k3 k4 k132 ph k12 k11 pb k15 k2 k8 pn
2 k142 k222 k192 k3 k4 k132 ph k12 k112 pb2 k15 k2
4 k142 k222 k192 k3 k4 k132 ph k12 k11 pb k15 k2
2 k142 k222 k192 k32 k42 k132 ph2 k122 k11 pb k1 pa
2 k152 k222 k192 k132 k22 k142 k4 pm
8 k15 k122 k222 k192 k3 k42 k132 pm2 k1 pa k14 k2 ph
2 k142 k222 k192 k3 k4 k132 ph k12 k15 k2
2 k152 k122 k222 k192 k3 k4 k13 pm k12 pa2 k14 k2
2 k142 k222 k192 k32 k43 k132 ph2 k122 k11 pb pm
2 k152 k122 k222 k192 k32 k42 k132 pm2 k1 pa k11 pb
2 k15 k122 k222 k192 k32 k42 k132 pm k12 pa2 k14 ph
8 k15 k122 k222 k192 k3 k42 k132 pm k12 pa2 k14 k2
2 k142 k222 k192 k32 k42 k132 ph2 k122 k1 pa
2 k152 k142 k222 k192 k122 k22 k8 pn2 k152 k142 k222 k192 k122 k22 k4 pm k1 pa
2 k152 k142 k222 k192 k122 k22 k4 pm k3 ph
4 k152 k142 k222 k192 k12 k22 k4 pm k13 k3 ph
4 k15 k142 k222 k192 k122 k2 k42 pm k3 k13 ph k1 pa
4 k152 k142 k222 k192 k12 k22 k4 pm k13 k8 pn
2 k15 k142 k222 k192 k122 k2 k43 pm2 k3 k13 ph
2 k152 k142 k222 k192 k122 k22 k4 pm k8 pn
2 k152 k14 k222 k192 k122 k2 k112 pb2 k3 k4 k13 pm
2 k152 k142 k222 k192 k122 k22 k11 pb k3 ph
2 k152 k142 k222 k192 k122 k22 k11 pb k1 pa
2 k152 k142 k222 k192 k122 k22 k11 pb k8 pn
4 k15 k142 k222 k192 k122 k2 k11 pb k3 k4 k13 ph k1 pa
4 k152 k142 k222 k192 k12 k22 k11 pb k13 k8 pn
4 k152 k142 k222 k192 k12 k22 k11 pb k13 k3 ph
8 k14 k222 k192 k42 k122 k132 k2 k1 pa k15 k3 pm k11 pb
4 k142 k222 k192 k3 k4 k13 ph k122 k8 pn k15 k2 k1 pa
2 k142 k222 k192 k3 k4 k132 ph k12 k82 pn2 k15 k2
2 k142 k222 k192 k3 k4 k13 ph k122 k82 pn2 k15 k2
k142 k222 k192 k32 k44 k132 ph2 k122 pm2
2 k142 k222 k192 k32 k42 k132 ph2 k122 k8 pn k1 pa
4 k142 k222 k192 k32 k4 k13 ph2 k122 k15 k2 k1 pa
2 k142 k222 k192 k33 k4 k13 ph3 k122 k15 k2

98
4 k142 k222 k192 k32 k4 k13 ph2 k122 k15 k2 k8 pn
k152 k142 k222 k192 k122 k22 k82 pn2
4 k142 k222 k192 k32 k4 k132 ph2 k12 k15 k2 k8 pn
2 k142 k222 k192 k33 k4 k132 ph3 k12 k15 k2
2 k142 k222 k192 k32 k43 k132 ph2 k122 k8 pn pm
2 k142 k222 k192 k33 k42 k132 ph3 k122 k1 pa
2 k142 k222 k192 k33 k43 k132 ph3 k122 pm
2 k152 k222 k192 k132 k22 k142 k4 pm k3 ph
2 k152 k222 k192 k132 k22 k142 k4 pm k11 pb
2 k15 k222 k192 k132 k2 k142 k43 pm2 k3 ph k12
2 k152 k222 k192 k132 k22 k142 k4 pm k8 pn
4 k142 k222 k192 k3 k4 k13 ph k122 k15 k2 k1 pa
2 k152 k222 k192 k132 k22 k142 k11 pb k3 ph
2 k152 k222 k192 k132 k2 k14 k112 pb2 k3 k4 pm k12
2 k152 k222 k192 k132 k22 k142 k11 pb k8 pn
2 k142 k222 k192 k32 k43 k132 ph2 k122 pm k1 pa
k142 k122 k222 k192 k32 k42 k132 ph2 k12 pa2k152 k222 k192 k132 k22 k142
2 k152 k222 k192 k13 k22 k142 k124 k152 k222 k192 k13 k22 k142 k8 pn k12 k3 ph
2 k152 k222 k192 k132 k22 k142 k8 pn k3 ph
2 k152 k222 k192 k13 k22 k142 k82 pn2 k12
2 k142 k122 k222 k192 k3 k4 k13 ph k12 pa2 k15 k2
2 k152 k142 k122 k222 k192 k22 k1 pak152 k142 k122 k222 k192 k22
2 k152 k142 k222 k192 k122 k22 k8 pn k1 pa
2 k152 k142 k222 k192 k122 k22 k8 pn k3 ph
8 k15 k222 k192 k3 k42 k132 pm2 k122 k11 pb k14 k2 ph
k152 k222 k192 k32 k42 k132 pm2 k122 k112 pb2
2 k152 k142 k222 k192 k13 k22 k32 ph2 k12k152 k142 k222 k192 k132 k22 k32 ph2
4 k3 k4 k132 k14 k15 k122 k222 k192 k2 ph pm
8 k3 k4 k132 k14 k15 k122 k222 k192 k1 pa k2 ph pm
8 k3 k4 k132 k14 k15 k122 k222 k192 k12 pa2 k2
4 k3 k4 k132 k14 k15 k122 k222 k192 k12 pa2 k2 ph pm
4 k3 k4 k132 k14 k15 k122 k222 k192 k112 pb2 k2 k1 pa
4 k3 k4 k132 k14 k15 k122 k222 k192 k112 pb2 k2 ph pm
4 k3 k4 k132 k14 k15 k122 k222 k192 k13 pa3 k2
k152 k142 k122 k222 k192 k22 k12 pa22 k152 k142 k222 k192 k122 k22 k3 ph k1 pa
k152 k142 k222 k192 k122 k22 k32 ph2

99
8 k3 k4 k132 k14 k15 k122 k222 k192 k12 pa2 k8 pn k2
8 k3 k4 k132 k14 k15 k122 k222 k192 k1 pa k8 pn k2 ph pm
8 k3 k4 k132 k14 k15 k122 k222 k192 k8 pn k2 ph pm
8 k3 k4 k132 k14 k15 k122 k222 k192 k8 pn k2 k1 pa
8 k3 k4 k132 k14 k15 k122 k222 k192 k11 pb k8 pn k2 ph pm
8 k3 k4 k132 k14 k15 k122 k222 k192 k11 pb k8 pn k2 k1 pa
4 k3 k4 k132 k14 k15 k122 k222 k192 k82 pn2 k2 ph pm
4 k3 k4 k132 k14 k15 k122 k222 k192 k82 pn2 k2 k1 pa
8 k3 k4 k132 k14 k15 k122 k222 k192 k11 pb k2 k1 pa
8 k3 k4 k132 k14 k15 k122 k222 k192 k11 pb k2 ph pm
4 k3 k4 k132 k14 k15 k122 k222 k192 k2 k1 pa
8 k3 k4 k132 k14 k15 k122 k222 k192 k1 pa k11 pb k2 ph pm
8 k3 k4 k132 k14 k15 k122 k222 k192 k12 pa2 k11 pb k2 )
2 k15 k22 k19 k3 k12 k13 k2 k1 pak15 k22 k19 k3 k42 k13 pm2 k12
k15 k22 k19 k3 k4 k13 pm k12 k8 pn2 k15 k13 k2 k22 k19 k3 k12 ph pm
k15 k22 k19 k3 k4 k13 pm k12k15 k22 k19 k32 k4 k13 pm k12 ph
k15 k14 k22 k19 k12 k2 k11 pbk15 k14 k22 k19 k12 k2 k4 pm
2 k15 k14 k22 k8 k12 k13 k2 k1 pa2 k15 k14 k13 k2 k22 k8 k12 ph pm
k15 k12 k22 k19 k3 k4 k13 pm k1 pak14 k22 k19 k3 k4 k13 ph k12 k11 pb
k14 k22 k19 k3 k4 k13 ph k12k15 k22 k19 k13 k2 k14 k4 pm
k14 k22 k19 k32 k4 k13 ph2 k12k14 k22 k19 k3 k4 k13 ph k12 k8 pn
2 k14 k22 k19 k4 k12 k13 k2 k1 pak14 k22 k19 k3 k42 k13 ph k12 pm
k15 k22 k19 k13 k2 k14 k11 pbk15 k22 k19 k13 k2 k14
k14 k12 k22 k19 k3 k4 k13 ph k1 pak15 k22 k19 k13 k2 k14 k8 pn
k15 k14 k22 k19 k12 k2 k8 pnk15 k14 k12 k22 k19 k2
2 k15 k14 k13 k2 k19 k11 k12 ph pmk15 k14 k22 k19 k13 k2 k3 ph
k15 k22 k19 k3 k4 k13 pm k12 k11 pb2 k15 k14 k19 k11 k12 k13 k2 k1 pa
2 k15 k14 k22 k19 k12 k13 k2 pa2 k15 k14 k22 k19 k13 k2 k1 ph pm
k15 k14 k22 k19 k12 k2 k3 phk15 k14 k12 k22 k19 k2 k1 pa
2 k14 k13 k2 k22 k19 k4 k12 ph pm ) k19 k22 k15 k14 k12 ct ) (
k3 k42 k13 k14 k15 k122 k222 k192 pm k11 pbk13 k2 k13 k142 k152 k222 k192 pa
k222 k192 k4 k122 k142 k2 k15k222 k19 k8 k122 k142 k152 k2 k4 pm
k3 k42 k13 k14 k15 k122 k222 k192 pmk222 k192 k1 k142 k152 k13 k2
k222 k192 k12 k142 k152 k13 k2 pak222 k192 k3 k12 k152 k13 k2 k14

100
k22 k192 k11 k122 k142 k152 k2k22 k192 k112 k122 k142 k152 k2 pb
k3 k4 k13 k14 k15 k122 k222 k192 k112 pb2
k3 k4 k13 k14 k15 k122 k222 k192 k12 pa2
2 k13 k2 k11 k122 k142 k152 k19 k8 k22 k1 pa
2 k3 k4 k13 k14 k15 k122 k222 k192 k1 pak222 k19 k8 k12 k142 k152 k13 k2
k32 k4 k13 k14 k15 k122 k222 k192 phk3 k4 k13 k14 k15 k122 k222 k192
k22 k192 k11 k122 k142 k152 k2 k1 pak222 k192 k32 k122 k152 k4 k13 pm
2 k13 k2 k4 k122 k142 k22 k192 k11 k15 k1 pa
k32 k4 k13 k14 k15 k122 k222 k192 ph k8 pn
2 k3 k4 k13 k14 k15 k122 k222 k192 k11 pbk222 k192 k32 k122 k152 k2 k14 ph
k222 k19 k82 k122 k142 k152 k2 pn2 k13 k2 k3 k122 k15 k222 k192 k4 k14 ph pm
k222 k192 k3 k122 k152 k2 k14k22 k192 k11 k12 k142 k152 k13 k2
k222 k192 k1 k142 k152 k12 k2k222 k192 k42 k122 k142 k3 k13 ph
k222 k192 k4 k12 k142 k13 k2 k15k222 k192 k42 k122 k142 k32 k13 ph2
k22 k192 k11 k12 k142 k152 k13 k2 k1 pak22 k192 k112 k12 k142 k152 k13 k2 pb
k3 k4 k13 k14 k15 k122 k222 k192 k82 pn2k222 k19 k8 k122 k142 k152 k2
k222 k19 k8 k122 k142 k152 k2 k11 pbk222 k19 k82 k12 k142 k152 k13 k2 pn
k222 k192 k32 k122 k152 k42 k13 pm2k222 k19 k8 k122 k142 k152 k2 k1 pa
k222 k192 k43 k122 k142 k3 k13 ph pm
k32 k4 k13 k14 k15 k122 k222 k192 ph k11 pbk222 k19 k8 k122 k142 k152 k2 k3 ph
2 k3 k4 k13 k14 k15 k122 k222 k192 k11 pb k8 pn
k22 k192 k11 k122 k142 k152 k2 k8 pnk32 k4 k13 k14 k15 k122 k222 k192 k1 pa ph
k22 k192 k11 k122 k142 k152 k2 k4 pmk22 k192 k11 k122 k142 k152 k2 k3 ph
k3 k42 k13 k14 k15 k122 k222 k192 k1 pa pm
2 k3 k4 k13 k14 k15 k122 k222 k192 k8 pnk222 k192 k12 k142 k152 k12 k2 pa
k222 k192 k42 k122 k142 k2 k15 pmk3 k42 k13 k14 k15 k122 k222 k192 pm k8 pn
k13 k2 k12 k142 k152 k222 k192 ph pmk13 k2 k32 k122 k152 k222 k192 ph pm
k13 k2 k32 k122 k152 k222 k192 k1 pak22 k192 k11 k12 k142 k152 k13 k2 k3 ph
k22 k192 k112 k122 k14 k152 k3 k4 k13 pm pb
k222 k192 k42 k12 k142 k13 k2 k15 pm
2 k3 k4 k13 k14 k15 k122 k222 k192 k1 pa k8 pn
k222 k192 k4 k122 k142 k2 k15 k8 pnk222 k192 k4 k122 k142 k2 k15 k11 pb
k222 k192 k1 k142 k152 k13 k2 k8 pnk222 k192 k1 k142 k152 k13 k2 k11 pb
k22 k192 k11 k122 k14 k152 k32 k4 k13 pm ph
k22 k192 k112 k122 k142 k15 k3 k4 k13 ph pbk222 k192 k1 k142 k152 k13 k2 k4 pm
k22 k192 k11 k122 k142 k15 k3 k42 k13 ph pmk222 k192 k1 k142 k152 k13 k2 k3 ph

101
k222 k192 k4 k122 k142 k2 k15 k1 pak222 k192 k3 k122 k152 k2 k14 k1 pa
k222 k192 k33 k122 k152 k4 k13 pm phk222 k192 k42 k122 k142 k3 k13 ph k8 pn
k222 k19 k8 k122 k14 k152 k3 k42 k13 pm2k222 k192 k1 k142 k152 k12 k2 k8 pn
k22 k192 k11 k122 k14 k152 k3 k42 k13 pm2k222 k192 k1 k142 k152 k12 k2 k11 pb
k222 k192 k1 k142 k152 k12 k2 k4 pmk222 k192 k1 k142 k152 k12 k2 k3 ph
k222 k192 k32 k12 k152 k13 k2 k14 phk22 k192 k11 k12 k142 k152 k13 k2 k8 pn
2 k3 k4 k13 k14 k15 k122 k222 k192 k1 pa k11 pb
k222 k192 k3 k122 k152 k2 k14 k8 pnk222 k192 k3 k122 k152 k2 k14 k11 pb
k222 k192 k3 k122 k152 k2 k14 k4 pmk13 k2 k112 k122 k142 k152 k192 ph pm
k13 k2 k112 k122 k142 k152 k192 k1 pak22 k192 k11 k12 k142 k152 k13 k2 k4 pm
k222 k19 k8 k122 k14 k152 k32 k4 k13 pm ph
k222 k192 k12 k142 k15 k3 k4 k13 ph k12 pa
k22 k192 k11 k122 k142 k15 k3 k4 k13 ph k8 pn
k222 k192 k1 k14 k152 k3 k42 k13 pm2 k12
k22 k192 k11 k122 k142 k15 k32 k4 k13 ph2
k222 k192 k1 k142 k15 k3 k42 k13 ph k12 pm
k222 k192 k42 k122 k142 k3 k13 ph k1 pak13 k2 k82 k122 k142 k152 k222 ph pm
k13 k2 k82 k122 k142 k152 k222 k1 pak13 k2 k42 k122 k142 k222 k192 ph pm
k13 k2 k42 k122 k142 k222 k192 k1 pak222 k19 k82 k122 k14 k152 k3 k4 k13 pm pn
k222 k192 k4 k122 k142 k2 k15 k3 phk22 k192 k11 k122 k142 k15 k3 k4 k13 ph
k222 k192 k1 k142 k15 k3 k4 k13 ph k12
k222 k19 k8 k122 k14 k152 k3 k4 k13 pm k11 pb
k22 k192 k11 k122 k14 k152 k3 k4 k13 pm
k222 k19 k8 k12 k142 k152 k13 k2 k11 pbk222 k19 k8 k12 k142 k152 k13 k2 k4 pm
k22 k192 k11 k122 k14 k152 k3 k4 k13 pm k1 pa
k222 k19 k8 k12 k142 k152 k13 k2 k3 phk222 k192 k32 k122 k152 k4 k13 pm k8 pn
k222 k192 k1 k142 k15 k3 k4 k13 ph k12 k11 pb
k222 k192 k42 k122 k142 k3 k13 ph k11 pb
k222 k19 k8 k122 k142 k15 k3 k4 k13 ph k1 pa
k222 k192 k1 k142 k15 k32 k4 k13 ph2 k12k222 k192 k3 k12 k152 k13 k2 k14 k8 pn
k222 k192 k4 k12 k142 k13 k2 k15 k8 pn
k222 k192 k1 k142 k15 k3 k4 k13 ph k12 k8 pn
k222 k192 k4 k12 k142 k13 k2 k15 k11 pb
k222 k19 k8 k122 k14 k152 k3 k4 k13 pm k1 pa
k222 k192 k1 k14 k152 k3 k4 k13 pm k12 k8 pn
k222 k192 k4 k12 k142 k13 k2 k15 k3 phk222 k192 k4 k12 k142 k13 k2 k15 k1 pa

102
2 k13 k2 k1 k14 k152 k222 k192 k3 k12 ph pm
k222 k192 k1 k14 k152 k3 k4 k13 pm k12 k11 pb
k222 k192 k12 k14 k152 k3 k4 k13 pm k12 pa
k222 k192 k1 k14 k152 k3 k4 k13 pm k12
k222 k192 k32 k122 k152 k4 k13 pm k11 pbk222 k19 k8 k122 k142 k15 k3 k4 k13 ph
k222 k19 k8 k122 k14 k152 k3 k4 k13 pm
k222 k192 k1 k14 k152 k32 k4 k13 pm k12 ph
k222 k192 k3 k12 k152 k13 k2 k14 k1 pa
2 k13 k2 k3 k122 k152 k222 k19 k8 k14 ph pm
k222 k192 k3 k12 k152 k13 k2 k14 k11 pb
2 k13 k2 k4 k122 k142 k222 k19 k8 k15 k1 pa
2 k13 k2 k3 k122 k152 k222 k19 k8 k14 k1 pa
k222 k192 k3 k12 k152 k13 k2 k14 k4 pm
2 k13 k2 k3 k122 k152 k22 k192 k11 k14 ph pm
2 k13 k2 k3 k122 k152 k22 k192 k11 k14 k1 pa
2 k13 k2 k4 k122 k142 k222 k19 k8 k15 ph pm
2 k13 k2 k3 k122 k15 k222 k192 k4 k14 k1 pa
2 k13 k2 k12 k142 k15 k222 k192 k4 k12 pa
2 k13 k2 k12 k14 k152 k222 k192 k3 k12 pa
2 k13 k2 k1 k142 k152 k222 k19 k8 k12 ph pm
k222 k19 k8 k12 k142 k152 k13 k2 k1 pa
k22 k192 k11 k122 k142 k15 k3 k4 k13 ph k1 pa
2 k13 k2 k12 k142 k152 k222 k19 k8 k12 pa
2 k13 k2 k4 k122 k142 k22 k192 k11 k15 ph pm
k22 k192 k11 k122 k14 k152 k3 k4 k13 pm k8 pn
k222 k19 k82 k122 k142 k15 k3 k4 k13 ph pn
k222 k19 k8 k122 k142 k15 k32 k4 k13 ph2
k222 k19 k8 k122 k142 k15 k3 k42 k13 ph pm
2 k13 k2 k1 k142 k15 k222 k192 k4 k12 ph pm
2 k13 k2 k1 k142 k152 k22 k192 k11 k12 ph pm
k222 k192 k32 k122 k152 k4 k13 pm k1 pa
2 k13 k2 k11 k122 k142 k152 k19 k8 k22 ph pm
2 k13 k2 k12 k142 k152 k22 k192 k11 k12 pa
k222 k19 k8 k122 k142 k15 k3 k4 k13 ph k11 pb )

103
VITAE

Mr. Hemendra Khakhar received his B.S. in Chemical Engineering with summa
cum laude from Gujarat University, India in 1999. Then, he worked for two years
as a process engineer with Indian Petrochemicals Corporation Limited (IPCL) ,
India. He has been a graduate student in the Department of Chemical
Engineering at Texas A&M University since January 2002. He has been working
with Prof. G.F.Froment and Prof. R.G.Anthony in the area of reaction
engineering. His contact details are :

Hemendra Khakhar
413 Nagle St. # 7
College Station,TX-77840
Tel:979-571-2353
E-mail:hmkhakhar@yahoo.com

104

Você também pode gostar