Modeling of Hydrogen Production from Bio-Oil via Steam Reforming using Matlab

By

Mohd Zharif Bin Imrat

A project dissertation submitted to the

Chemical Engineering Programme
Universiti Teknologi PETRONAS
In partial fulfillment of the requirement for the
Bachelor of Engineering (Hons)
(Chemical Engineering)

January 2009

Universiti Teknologi PETRONAS

Bandar Sen Iskandar
31750 Tronoh
Perak Darul Ridzuan
 CERTIFICATION OF APPROVAL

MODELING OF HYDROGEN PRODUCTION FROM BIO OIL VIA STEAM REFORMING

USING MATLAB

By

Mohd Zharif Bin Imrat

A project dissertation submitted to the

Chemical Engineering Programme
Universiti Teknologi PETRONAS
In partial fulfillment of the requirement for the
BACHELOR OF ENGINEERING (Hons)
(CHEMICAL ENGINEERING)

Approved by,

(Dr. Murni Melati Ahmad)

UNIVERSITI TEKNOLOGI PETRONAS

TRONOH, PERAK
JANUARY 2006
 CERTIFICATION OF ORIGINALITY

This is to certify that I am responsible for the work submitted in this project, that the
original work is my own except as specified in the reference and acknowledgement, and
that the original work contained herein have not been undertaken or done by unspecified
sources or persons.

MOHD ZHARIF BIN IMRAT

 ABSTRACT

Biomass is a product in which is a renewable resource that can be used for production of
hydrogen. The hydrogen has the potential to be a source for alternative fuel required in
the future. The alternative way concept is based on a two-stage process, which are fast
pyrolysis and catalytic steam reforming [17]. There exists a wide range of modeling
approaches to represent hydrogen production from bio oil via steam reforming. The
modeling approach, even it is less expensive and more extendable compared to
experimental procedures, has its own limitation. In this project, the bio oil will be
represented by crude ethanol as there are difficulties of finding the right journals. The
main objective for this project is to screen suitable models to represent hydrogen
production from bio-oil via steam reforming. The models of each process steps in
hydrogen production are compiled by MATLAB. The process steps from steam
reforming to water gas shift reaction, to PSA purification of hydrogen are gathered into
one system so that the integrated results can be determined. In the case of steam
reforming process, Model 1 has the Average Relative Error of 29.1%. Whereas for Model
2 and Power Law, both have the Average Relative Error of 21.1% and 17.2%
respectively. Basically, Model 2 and Power Law are on the active site of the Rate
Determining Step as the Average Relative Error is less than 25 %. For the Water Gas
Shift reaction, it is concluded that the final temperature for the shift reactor for this
system is around 493 K. The conversion of the reactor is 68 % and the GHSV will be
approximately about 6250 h"1. Finally, the H2 purity for this work is 98.9119 %because
the bed only consists ofactivated carbon with the purpose istotrap CO2- It is determined
that the Relative error (%) for this work and Ribeiro ., based on H2 purity is 1.08 %.
 ACKNOWLEDGEMENT

First and foremost, I would like to thank to God Almighty, Allah for giving an
opportunity to perform this Final Year Project in such a great manner. Greatest
appreciation also to my beloved family for being well supportive, which really helps me
to complete this project.

Special thanks also to Dr. Murni Melati, who is a project supervisor for providing
excellent help, guidance, patience and willingness to share out her knowledge and
experience in order for me to complete the project. Endless thanks and appreciation to
friends for their ideas in bringing this project to success. Their kindness and friendliness
will never be forgotten. Their contribution will be remembered and only God can repay
their kindness. Thank you very much.
 TABLE OF CONTENT

CERTIFICATION OFAPPROVAL j

CERTIFICATION OF ORIGINALITY 2

ABSTRACT 3

ACKNOWLEDGEMENT 4

TABLE OFCONTENT. 5

LIST OFFIGURES 7

LIST OFFIGURES 8

CHAPTER 1: INTRODUCTION jo

1.1 Background 10
1.2 Concepts U
1.2.2 Pyrolysis Process jj
1.2.3 Steam Reforming J2
1.2.4 Water GasShift Reaction 72
1.3 Chemical Reaction Analysis 14
1.4 Process Flow Diagram 15
1.5 Problem Statement 17
1.6 Objective 18

CHAPTER 2:LITERATURE REVIEW 19

2.1 Steam Reforming 19

2.2 Water Gas Shift Reaction. 21
2.3 Hydrogen Purification 22

CHAPTER 3:METHODOLOGY. 24

3.1 Overview of Overall Project Methodology 24

3.2 Project Gant Chart 26
3.3 Process Development 26
CHAPTER 4: RESULTSAND DISCUSSION 28

4.1 Steam Reforming of Crude Ethanol 28

4.1. J The calculation ofrate of reaction. 28
4.1.2 The calculation ofComponent Balance 37
4.2 Water Gas Shift Reaction 40
4.2.1 The calculation ofrate ofreaction. 40
4.2.2 The calculation ofComponent Balance 45
4.3 Hydrogen Purification (PSA Purification) 47
4.3.1 The calculation of Component Balance 47

CHAPTER 5: CONCLUSION 53

REFERENCES 54

APPENDICES. 57

Appendix 1: Operating Unit Description 57

Appendlx 2: Project Gant Chart FYP 1 59
Appendix 3: Project Gant Chart FYP 2 60
Appendix4: steam_reforming_model1.m 61
Appendix 5: steam_reforming_model2.m 63
Appendix 6: power_lawjviodel.m 65
Appendix 7: WGS-numec_fit.m 66
 LIST OF FIGURES

Figure 1.1: SimplifiedProcess Flow Diagram 15

Figure 1.2: Processflow diagram ofthe bio-oil reforming 16

Figure 3.1: Block Diagram ofProject Methodology 24

Figure 3.2: Block Diagram Each Step Process in Hydrogen 26

Productionfrom Ethanol

Figure 4.1: Variation ofcrude ethanol conversion with space time 28

Figure 4.2: A comparison ifmeasured andpredicted rates 33

ofreforming crude ethanol within the temperature
Range of593-793 K

Figure 4.3: Water Gas Shift Reaction - CO conversion 41

as afunction of space velocity

Figure 4.4: Graph ofmolarflowrateforH2 along the bed 50

from 0 m to 0.5 mat 80,000s

Figure 4.5: Graph ofmolarflowratefor C02 along the bed 51

from 0m to 0.5mat80,000s
 LIST OF FIGURES

Table 1.1: Crude Ethanol Composition 17

Table 4.1: Value ofslopes (predicted r^) at temperature 30

593,693, 793 K

Table 4.2: Table ofkinetic models 30

Table 4.3: Nomenclaturefor catalytic reforming ofcrude ethanol 31

Table 4.4: Kinetic experimental datafor catalytic reforming 32

ofcrude ethanol

Table 4.5: Fitted values ofkinetic constants ofcrude ethanol 32

reforming tohydrogen based onmodel 1, model 2 and
power law

Table 4.6: The values ofthe measuredrAfor the three models along 33
with thepredicted rA^

Table 4.7: Model 1 average relative error 34

Table 4.8: Model 2 average relative error 34

Table 4.9: Power Model average relative error 35

Table 4JO: CalculatedAveragefor each Temperaturefor 35
Power Law Model

Table 4.11: Conversion, XofCalculatedAveragefor each 36

Temperaturefor Power Law Model

Table 4.12: Value ofslopes (predicted rCo) attemperature 42

428,463, 493K

Table 4.13: Conversion, Xofrcofor each Temperaturefor 44

empirical rate expression derievedfrom the
numericalfitting

Table 4.14: Calculatedmolarflowratesfor H2 andC02 along 50

the bedfrom 0 mto 0.5mat t=80,000 s
 CHAPTER 1: INTRODUCTION

1.1 Background

Due to the rising price offossil fuel in the global market, along with the increasing in
energy consumption and demand, exploration of alternative and renewable resource of
energy has been developed. Together with the depleted source of fossil fuel and the
awareness ofthe environment since 1990s, which is the significant global warming trend,
has lead to the exploration ofa new alternative energy resource such as hydrogen. This
resource is becoming popular alternative in order to replace other energy resources
especially for transportation fuel and power generation, while maintaining the
sustainable, economically viable and eco-friendly source.

Hydrogen production from renewable resources, such as biomass, is gaining attention

since the use of biomass for energy does not increase carbon dioxide emissions. It also
does not contribute to the risk of global climate change, as it is particularly a clean
resource. One of the common processes to produce hydrogen from biomass is via fast
pyrolysis of biomass into bio-oil and steam reforming of bio-oil into hydrogen. The
hydrogen yield from the bio-oil can be further increased via water gas shift reaction. The
combination of fast pyrolysis of biomass followed by steam reforming of the produced
bio-oil has attracted attention of the research community [1-3], as one of the most
promising methods for hydrogen production.

Bio-oil has certain advantages such as higher density than biomass, easy storage and
transportation, easy to be used either as a renewable liquid fuel or for the production of
chemicals. By addition, bio oil can be separated into a hydrophobic lignin derived
fraction and an aqueous fraction (50% ofbio-oil) containing mostly carbohydrate derived
monomeric compounds [4,5]. The aqueous phase ofthe bio-oil consists of20% organics
and 80% water [6-8].

10
 1.2 Concepts

1.2.2 Pyrolysis Process

Pyrolysis is a chemical decomposition of organic materials by heating in the absence of

oxygen or any otherreagents, except steam. It is used in chemical analysis to break down
complex matter into simpler molecules for identification. In industry, it is used to convert
one single chemical to multiple chemicals. For instance, ethylene dichlorideis pyrolysed
to vinyl chloride in order to make PVC. It is also used to convert complex materials to
subtances that are less harmful such as biomass or waste. Extreme pyrolysis can also be
called as carbonization, as it leaves only carbon as the residue.

Fast pyrolysis of biomass feedstocks is required to achieve high yields of liquids. It is

characterized by rapid heating of the biomass particles and a short residence time of
product vapors (0.5 to 2 s). Rapid heating is where the biomass is grounded into fine
particles and the insulating char layer that forms at the surface of the reacting particles
must be removed continuously. Since pyrolysis is endothermic, various methods have
been proposed in order provide heat to thereacting biomass particles:

• Partial combustion of the biomass products through air injection. This will result
in poor-quality products.
• Direct heat transfer with a hot gas, ideally product gas that is reheated and
recycled. The problem is to provide enough heat with reasonable gas flow-rates.
• Indirect heat transfer with exchange surfaces (wall, tubes). It is difficult to achieve
good heat transfer on both sides of the heat exchange surface.
• Direct heat transfer with circulating solids: Solids transfer heat between a burner
and a pyrolysis reactor. This is aneffective but complex technology.

There a lot of technologies that have been proposed for the biomass pyrolysis process.
But, for this project, only two of those technologies will be emphasized. The following
technologies that have been proposed for the biomass pyrolysis process:

11
 • Fixed beds: For the traditional production of charcoal, bulk catalysts are used.
But, the consequence of using this technology is a poor and slow heat transfer will
result in very low liquid yields.
• Fluidized beds: Biomass particles are introduced into a bed of hot sand fluidized
by a gas, which is usually a recirculated product gas. Biomass particles experiense
a rapid heating because of high heat transfer rates from fluidized sand. There is
some ablation by attrition with the sand particles, but it is not as effective as in the
ablative processes. Heat exchanger tubes will provide heat through which hot
combustion gas flows. There will be some dilution of the products, which will
make it difficult to condense.

1.2.3 Steam Reforming

Steam reforming (SR), hydrogen reforming or catalytic oxidation, is a method of

producing hydrogen from hydrocarbons. It isa dominant method for producing hydrogen
in industrial scale.

Steam reforming of natural gas or syngas is the most common method of producing
commercial bulk hydrogen as well as the hydrogen used in the industrial synthesis of
ammonia. It is also the least expensive method. At high temperatures (700 - 1100 °C) and
in the presence of a metal-based catalyst (nickel), steam reacts with methane to yield
carbon monoxide and hydrogen. This steam reforming process is quite different and not
to be confused with catalytic reforming of naphtha, an oil refinery process that also
produces significant amounts of hydrogen along withhigh octane gasoline.

1.2.4 Water Gas Shift Reaction

The water-gas shift reaction (WGS) is a chemical reaction in which carbon monoxide
reacts withwater to form carbon dioxide and hydrogen. The water-gas shift reaction is an
exothermic and also important industrial reaction.

12
The water gas shift reaction is sensitive to temperature, with the tendency to shift towards
reactants as temperature increases due to Le Chatelier's principle. In fuel-rich
hydrocarbon combustion processes, the water gas reaction at equilibrium state is often
employed as a means to provide estimates for molar concentrations of burnt gas
constituents.

The process is used in two stages, stage one a high temperature shift (HTS) at 350 °C
(662 °F) and stage two a low temperature shift (LTS) at 190-210 °C (374-410 °F)'
Standard industrial catalysts for this process are iron oxide promoted with chromium
oxide for the HTS step and copper on a mixed support composed of zinc oxide and
aluminum oxide for the LTS shift step. Attempts to lower the reaction temperature of this
reaction have been done primarily with a catalyst such as Fe304 (magnetite), or other
transition metals and transition metal oxides.

13
1.3 Chemical Reaction Analysis

According to Rioche et al. [15], the steam reforming of bio-oil oxygenated organic
compounds with chemical formula CnHmOk is described as follows:

CMmOk + (n-k) H20 ^ nCO + (n+m/2-k) H2 (1)

The reaction is followed by water gas shift reaction (WGS):

H20 + CO • C02 + H2 (2)

The steam reforming reaction is endothermic (+ve), thus favoring high temperatures
while the carbon monoxide shift reaction is exothermic (-ve). The overall reforming
process giventhat both the reactions above can be represented as follows:

CnHmOk + (2n-k) H20< • nC02 + [(2n+m)/(2-k)J H2 (3)

Due to high temperature ofthe reaction, thermal decomposition or cracking can occur
which causes coke and a mixture of gases such as hydrogen, carbon monoxide, carbon
dioxide, methane, and carbon to be formed. The detail reaction is represented as follows:

CMA A • CxHyOz+ gases (H2, H20, CO, C02, CH4, ....) + coke (4)

Due to the formation of the two undesirable products, carbon monoxide and methane,
which are formed during WGS and methanation reaction, the actual yield of hydrogen is
lower than the stoichiometric maximum.

14
1.4 Process Flow Diagram

The simplified process flow diagram for the purpose of MATLAB studies can be referred
in Figure 1.1 below. Basically, this design comes from the conceptual design of Douglas
J.M [9] and the operating condition of Ahmad et al. [16]. Figure 1.2 on the next page
section is the overall process flow diagram of the bio-oil steam reforming. By designing
the process flow diagram, one can focus and understand the concept of steam reforming
process flow. The feed for this process is bio-oil and water. The details about the
functions of each operating units can be referred in theAppendix 1 section.

According to theFigure 1.2,the three dotted circles are the main operating units thatwill
be involved for the MATLAB feasibility studies. The first unit is the steam reformer, in
which steam-reforming process ofbio oil will take place. The second unit is thewater gas
shift reactor, inwhich water gas shift reaction will occur. Finally, the PSA column, where
the purification of hydrogen will take place.

Feed Hydrogen
->

Steam WGS PSA

Reformer reactor Column

Figure 1.1: SimplifiedProcessFlowDiagram

15
Acetic acid,
Ethylene glycol,
Acetone
Hydrogen rich
Compressor gq*
H20
Pressure
swing
absorption
(j»SA)
Compressor
PSA off gas
Figure 1.2: Processflow diagram ofthe bio-oil reforming [16]
 1.5 Problem Statement

There exists a wide range of modeling approaches to represent hydrogen production from
bio oil via steam reforming. It is known that, modeling approach, even it is less expensive
and more extendable compared to experimental procedures, has its own limitation.
Different model has its different strength and limitation, in which will indicate it is an
accessible model or otherwise. This project will aim to review and to put up models that
can represent each process step in hydrogen production from bio oil in a simplest way.

Due to some difficulties offinding the necessary journals regarding the kinetic modeling
of the bio oil, the bio oil will be represented by crude ethanol. It is assumed that the use
of crude ethanol kinetic modeling somehow replicate the kinetic modeling of bio oil, as
the molecular formula of crude ethanol is C2.j2Hej2Oj.23, which is somewhat similar to
the basic molecular formula of bio oil, Cj.sH4.12Oj.54 obtain from Rioch CE et al. [2]. The
crude ethanol steam reforming general equation canbe represented as:

C2H60^3H20 >. 2C02 + 6H2

Crude Ethanol Components Volume % Mole % on a water free basis

Ethanol 12.0 88.42

Lactic acid 1.0 5.71

Glycerol 1.0 5.87

Maltose 0.001 0.001

Water 86.0 Not applicable

Table LI: Crude Ethanol Composition [10]

However, for the purpose of simplicity, the stoichiometry of ethanol is used to develop
the kinetics, as it is well known.

17
1.6 Objective

1. To screen suitable models to represent hydrogen production from bio-oil via

steam reforming. The bio oil has been represented by ethanol, in which the kinetic
modeling will replicatethe kinetic modeling of bio oil.
a. By referring to the work of Akande etc [10], the modeling approach for
steam reforming for ethanol can be done.
b. The work is found suitable, asthe bio oil has been represented by ethanol.
2. The models of each process step in hydrogen production are compiled by
MATLAB.

a. The process steps from steam reforming to water gas shift reaction, to
PSA purification of hydrogen are gathered into one system so that the
molar balance can be determined.

3. To develop the modeling ways and then used to simulate the sequence of
processes in hydrogen production from bio-oil in MATLAB.
a. To determine the significance of thiswork and then to compare the results
of this work with other previous works.
b. To find out if this work has an advantage compared to the otherworks.

18
 CHAPTER 2: LITERATURE REVIEW

This chapter summarizes the suitable journals or works in order to be used for references
of this project. The journals contain the appropriate information such as theories
equations and also result that will be sufficient to carry out the project. Below are the
summaries of all the suitable journals that are related to the project that managed to be
found. Since there will be three main processes involved in this project, the necessary
journals are crucial so that the system will work.

2.1 Steam Reforming

Journal/ Paper Author
Description
Cinetic Modeling of Abayomi
lydrogen production The kinetic modeling ofH2 production has
Akande
>y the catalytic been performed over 15% Ni-Al203
Ahmed catalyst.
eforming of crude Aboudheir
thanol over a co- The kinetic experiment was done at
Raphael Idem
recipitated Ni-Al203 atmospheric pressure inpacked bed tubular
Ajay Dalai. reactor at temperature 593-793 K.
atalyst in a packed
ed tubular reactor Eley Rideal assumptions where the surface
10] reaction involved an absorbed species and
afree gaseous were used to develop the
reaction mechanism and four models were
proposed.
The model ofdissociation ofadsorbed
crude ethanol as rate determining step was
develop.
J^m0del was rA=(2'08 x10'8e ~
443mT)(NA)/[l +3.83xl07NAf
Absolute deviation between experimental
rates and predicted rates was 6% for this
model.
hanol steam • F. Gallucci
brming in a dense Mathematical model has been formulated
" M.DeFalco for a traditional reactor packed with a Co-
-Ag membrane 1 S. Tosti
ictor: A modeling based catalyst and then applied to a
1 L. Marrelli membrane reactor (MR).
irk. Comparison • A. Basile
th the traditional With MR reactor, high conversion of
item. [22] ethanol and high hydrogen selectivities.

19
 These two works are compared to determine which has the ideal modeling approaches
After afew studies, the work of Akande et al. [10] seems to be the right journal to be
referred to as it has the necessary information in order to develop amodel for steam
reforming ofethanol.

Basically, the work ofAkande et al. [10] is focused more on the hydrogen production by
the catalytic reforming of crude ethanol over aco-precipitated M-Al2Os catalyst in a
packed bed tubular reactor. This method is a tradition to steam reforming process
compared to the work ofGallucci et al. [22], which it uses the more conventional method,
the membrane reactor.

In theory, the performance of the membrane reactor was compared with the traditional
reactor in the journal ofGallucci et al. [22], It indicate, that using the membrane reactor
with the respect to the traditional reactor can increase the ethanol conversion. Moreover,
aCO-iree hydrogen stream can be directly produced in the membrane reactor and 11 can
be directly fed to aproton exchange membrane fuel cell system.

But, the journal of Gallucci et al. [22] does not have complete information on how to
develop the modeling for steam reforming. Thus, it is decided that the work ofGallucci et

al. [22] will be put aside although the used of membrane reactor for steam reforming iis
very practical and feasible in theory.

20
 2.2 Water Gas Shift Reaction.

Journal/ Paper Author Description

Water gas shift Yongtaek The reaction was studied to evaluate
reaction kinetics and Choi existing reaction mechanisms.
reactor modeling for Harvey G. To test various rate expressions and to
fuel cell grade Stenger simulate the performance in a methanol
hydrogen fuel processorfor fuel cell applications.
[11] The reaction was done in a micro reactor-
testing unit using a commercial Sud-
Chemie Cu/2nO/Al2Os catalyst between
120-250 °C with a range of feed rates and
compositions.
Using non-linear least squares
optimization, the parameters in five rate
expressions were fit to the experimental
data.
Numerical integration of a one-dimensional
PFR model was used for this parameter
fitting.
An empirical rate expression, rCo =
kPcoPmo 0 -p) with activation energy of
47.4 kJ/mol was obtained from the
experiment.
Reactor performance was simulated to
determine catalyst loadings required to
achieve specific CO conversions as a
function of temperature and water feed rate
Modeling of C02- Jin Huang WGS reaction is criticalto hydrogen
selective water gas Louei El- purification for fuel cells.
shift membrane Azzami WGS reaction in the traditional fixed bed
reactor for fuel cell W.S. Winston reactor is not efficient.
[18] Using a C02-selective membrane reactor
Ho
shifts the reaction towards the product side,
Which enhances the conversion of CO and
increases the purity of the H2 product at a
high pressure.
Using a one-dimensional non-isothermal
model simulated the simultaneous reaction
and transport process in the countercurrent
WGS membrane reactor.
The effect of several systemparameters
including C02/H2 selectivity, C02
permeability, and sweep-to-feed molar
flow rate ratio were investigated.

21
 The work of Choi et al. [11] and Huang et al. [18] are compared so that the modeling
approach for water gas shift reaction can be developed. It seems that theoretically, the
work ofHuang et al. [18] is much better inenhancing the conversion of CO and increases
the purity ofthe H2 production at high pressure compared to the traditional shift reactor,
which is fixed bed reactor. The journal of Huang et al. [18] also has done some
investigation on the effect ofseveral system parameters including CO/H2 selectivity, C02
permeability, and sweep to feed molar flow rate ratio.

But, the work ofHuang et al. [18] does not have the information regarding the rate of
reaction for CO, whereas for the work ofChoi et al. [11], it has the equation for rate of
reaction for CO, although the use of fixed bed reactor is not very practical in enhancing
the conversion ofCO and increasing the purity ofH2 production. Even though it is not
very efficient, it has the proper modeling equations that is called empirical fitting rate
expression derived from the numerical fitting, which will be accurate and much more
simple.

2.3 Hydrogen Purification

Journal/ Paper Author Description

A parametric study of • AnaM. This work is focused onthe separation of
layered bed PSA for Ribeiro hydrogen from a five-component mixture
hydrogen purification • Carlos A. (H2/CO2/CH4/CO/N2) by pressure swing
[12] Grande absorption.
• FilipeV.S. The mathematical model of this paper is
Lopes applied in the study ofthe behavior of
• Jose M. single column and four column PSA
Loureiro processes with layered activated carbon
• Alirio E. and with an eight-step cycle.
Rodrigues 99.994% purity ofH2 is attained at the end
of the feed step for a process hydrogen
recovery of 51.84% and a productivity of
59.6 molH2 / kg ads /day.
Multicolumn simulation predicts a H2
recovery and purity, 52.11% and 99.995%
respectively.
Activated carbon layer improves both the
purity and recovery of the process.

22
Basically, for the purification ofhydrogen, only journal ofRibeiro et al. [12] is suitable to
develop amodel ofthis process. This work is focused on the separation ofhydrogen from
a five-components mixture, which is H2, C02, CH4, CO, and N2 by a separation
technique namely as pressure swing absorption. This work tells that acomplete model is
able to describe the dynamic behavior of the pressure swing adsorption system. This
work also shows that activated carbon layer will improve both the purity and recovery of
thehydrogen purification process.

The mass balance equation for gas phase of a fixed bed system is applied in order to
determine the amount of hydrogen produced after PSA column. By referring to the table
of boundary condition from Ribeiro et al. [12], it is assumed that co-current
pressurization with feed will determine the amount of hydrogen production for the
purification system.

23
 CHAPTER 3: METHODOLOGY

3.1 Overview ofOverall Project Methodology

Project Statement

Planning

I
Preliminary Literature Research on Background
and Previous Work

Data Gathering

I
Process Synthesis &Conceptual Design

Choose a suitable model for each unit

I No
No Does the model predict with system behavior
well?
Does it match in the exneriment data?

I Yes

ProcessDevelopment

Yes

Result Discussion and

Analysis

Figure 3.1: BlockDiagram ofProjectMethodology

24
 Project Statement

• The FYP begins when the project that shall be chosen will get the supervisor's
approval.

Planning

• Good planning on the project development will help on the direction of the
project to avoid any lost inproject track.

Preliminary Literature on Background and Previous Work Research

• Necessary journals and work will be sufficient in order to understand the basic
concept of the project.

Data Gathering
• After obtaining the right journals and works, it is better to gather all ofthe data so
that any problems regarding the project can easily be referred back to the
information gathered.

Process Synthesis &Conceptual Design

• Building the design ofthe operating units to comprehend the concept ofthe
process involved.

Choose a suitable model for each unit

• Asuitable model is important to aunit so that the model can be checked whether
it converges or otherwise.

Does the model predict with system behavior well? Does it match in the experiment **»<)
' Aproper investigation is needed so that any inaccuracy will be quickly discarded
so that the model will match the experiment data.

ProcessDevelopment
• Ifno problem occurs, then the project will be commenced as usual.

25
 Result Discussion and Analysis
• Finally, the project will be summarized and analysis will be made.

3.2 Project Gant Chart

For this work, it was initially divide into two continuing sessions (FYP 1and FYP 2),
which will be carried out for two semesters. For the FYP1 the project work is more on
emphasizing on the literature modeling research and development. Whereas for the
FYP2, the project work will be more on writing the modeling programming language by
using MATLAB. Basically, the details of the project work for FYP 1and FYP 2 are
illustrated in Appendix 2 and Appendix 3.

3.3 Process Development

Ethanol Steam Water Gas

Reforming Shift Bio oil

Reaction

Figure 3.2: BlockDiagram Each Step Process in Hydrogen Productionfrom Ethanol

Steam Reforming

• The journal of Akande et al. [10] is used to determine the rate of reaction of
ethanol and the work of Koumpouras et al. [21] is used to determine the
component balance for all components inthe reactor reformer.
• Model assumptions
o Steady state and non isothermal (adiabatic) operation
o Perfect gas behavior
o Adsorption kinetics described by linear driving force model (LDF) and
Langmuir isotherm for C02 adsorption equilibrium.
o The general kinetics from Akande et al. [10] is used to determine the
accumulation of hydrogen production.

26
 o C02 is considered as the only absorbate and C02 adsorption is assumed to
take place on the flowing adsorbent particle.
o Ideal plug flow

o Gas and adsorbent particle velocities are assumed constant and equal.

Water Gas Shift Reaction

The work ofChoi et al. [11] and Huang et al. [18] are used to obtain the modeling
approach for the rate of reaction of CO and also the component balance in the
shift reactor.

• Model assumptions
o CO2 and H2 are the only two gasses permeating through the membrane.
o There is no temperature variation in the radial direction inside a hollow
fiber due to its small dimension/
o Membrane permeability is fixed and does not change with temperature
variation.

o The module is adiabatic and operates at steady state.

o There is no axial mixing.
o The pressure drops on both lumen and shell sides are negligible.

PSA Purification System

The work of Ribeiro et al. [12] is used to obtain the mathematical modal of
component balance equation in afixed bed adsorption system.
• Model assumptions
o Ideal gas behavior throughout the column, within the operating conditions
under study, the gases follow nearly the ideal gas law.
o No mass, heat orvelocity gradients inthe radial direction.
o External mass and heat transfer resistances, both expressed with the linear
driving for model (LDF).
o Constant porosity along the bed.
o No temperature gradients inside each particle as the heat transfer in the
solid particles are much higher than in the gas phase.

27
 CHAPTER 4: RESULTS AND DISCUSSION

4.1 Steam Reforming of Crude Ethanol

4.1.1 The calculation of rate of reaction

For the Steam Reforming, the work by Akande et al. [10] is used to obtain the modeling
equations. In this study, crude ethanol conversion (X) was evaluated in terms of ethanol
plus other organics on awater free basis as shown in the following equation:

Crude Ethanol Conversion (X)

= gmol {organics )in - gmol (organics )out
gmol (organics )in

Where organics = ethanol + lactic acid + glycerol + maltose.

Based onthe experimental data ofAkande et al. [10], the results for the variation of crude
ethanol conversion X with ratio of weight of catalyst to crude ethanol flow rate ratio
(W/Fao) at reaction temperatures of593, 693, and 793 Kare presented in Figure 4.1 [10].
Crude Ethanol Conversion with Space Time at
593, 693, 793 K.

1.00E+03 1.50E+03 2,OOE+03 2.50E+03

W/Fao, kg cat s/kg-crude

Figure 4.1: Variation ofcrude ethanol conversion with space time [10]

28
 According to Akande et al. [10], amicro reactor was used to gather the experimental data
and the design equation for the plug flow reactor was therefore applicable for data
analysis. This was used in the differential form.

dX
= rA
d(W/FA0)

To determine the predicted rate ofequation, rA various slopes were taken at each ofthe
three temperatures in order to differentiate with the measured rate of equation, rAof all
three models. The slopes of each ofthe points were calculated manually. Based on Figure
4.1, only three points (1) 7.56 x 102, (2) 1.4 x 103, (3) 2.16 x 103 were taken at each
temperature because in this project, the measured rA are calculated with only 3trials.

For the temperature of 593 K;

0.65-•0.585
A 2300(46.07)--1880(46.07)
-3.36xl0"6

0.54--0.45
rA=
1620(46.07)-1140(46.07)
= 4.096 xlO"6

0.39-0.315
rA
980(46.07)-700(46.07)
= 5.814xl0"6

Thus, by doing the same method offinding the value ofthe slope (predicted rA) to both
temperature, value of each slopes are represented in Table 4.1

29
 ^^^^ Points
1 2 3
Temperatures\^
593 K 5.814 xlO"6 4.096 xlO"6 3.36 xlO"6
693 K 7.96 xlO"6 4.65 x lO-6 2.71 x 10"6

793 K 9.41 x 10"6 5.42x10* 4.34 x!0~6

Table 4.1: Value ofslopes (predicted ra) attemperature 593, 693, 793 K
Based on Akande et al. [10], in order to calculate the value ofrA, four models have to be
chosen that will be the most realistic to the catalytic reforming ofcrude ethanol over Ni-
A1203 catalyst. In the journal, it states that there are four cases as possible rate controlling
mechanisms for the catalytic reforming of crude ethanol. Table 4.2 and Table 4.3 are
obtained from the work of Akande et al. [10].

Table 4.2: Table ofkinetic models [10]

RDS MnJoh Sxi^oJ i.-'ii S|Xhii.n niulur ikw

AUsurpiion •-•.!' >A~

jt'rV- V-1 A'- .V- V-' £' \rl v*
I + + +
A:* Xi
•"h

Tlk' diss.U-MU'.'11 '."i]. ;td-

ApA-^/
r\ =

S'Thcd crudo ethanol

i + KaNa +£L^d +KcXfl
A'B 'B

/ \' - .V-1 7^
,v_ v? -,

Suriacv rvaciinn boKvocn

,A'c.^ *P ,
rA
;iUs'Jt'K\l oxygenated hv-
+
di'ocarhon und %io;im

(.n0-£-/RT ( AA;VB -Vc.Vp>

Surkicv ivadion heiv.eou AcA:h A'p
;idsork\l s hydr*K*urh.>n l+JCAAU +^MD +*:"iV^B\
•ha'a*'b \
fruction spevio ;itxl steam ;VfB

30
 Table 4.3: Nomenclature for catalytic reforming ofcrude ethanol [10]
Ntimt'iK-liiturc

i'A nice of emcte-cihanol conversion, A'. muliir Jlow rate of ^peeic^ fcuiol/s
kmolcrude/kg veil s RDS rate-detenmnmg step
E activation enemy. kJ/kmnl iV weight of eatahsi. kg
h colJisiun frotjuetK-y or pi-e-exjioncntuil A' crude-eiharmI conversion ihoK* or
uwM;inl. ikg eats}"1 fractional conversion
A"; equilibrium ennM;m1 of reartiun slep i AAIK* averago absolute deviation. V;
A'p overall equilibrium constant W;F.w ,pat-e nine, kjn\ii s/kgcrude
R universal cas constant. kJ/kjnot K

The Power Law model is used to fit the experimental data.

rA=k0e-E/RT(NAy

Let C2HtO=A;
H20 = B;
CO2 =C;
H2 -A

It is decided to choose only three models ofrate ofreaction, rA that will be compared
with the predicted rate ofreaction, rA. The models are adsorption ofcrude ethanol (model
1\ the dissociation of adsorbed crude ethanol (model 2\ and also power law model.
Then, the rA is calculated using the models in MATLAB. For the calculation ofmodel 1,
the equations and variables are written in *.m format and then it is saved by the name of
steam reforming_modell.m. The same method is applied to the other two models,
which are model 2, and powerlaw model. Thenames of the files of those two models are
steamreforming model2.m and power law model.m. The MATLAB programs can
be referred inAppendix 4,Appendix 5 and Appendix 6 for more details.

Table 4.4 is the kinetic experimental data, which shows the molar flow for each species
for catalytic reforming ofcrude ethanol [10]. With these data, the calculation ofrA can be

31
done. Table 4.5 is the fitted values of kinetic constant ofcrude ethanol and Table 4.6 is
the summary ofall the measured rA for the three models along with the predicted rA

Table 4.4: Kinetic experimental datafor catalytic reforming ofcrude ethanol [10]
TtIO *VA *B *c % AP
•:km>'ieAi 'klTiokvM tkmukv'M • knfc'irtVi

yty 5.705%E--iW 15"?33IE-07 X.4UM4H — ' » Z4'Jf^5H--:ix 2.45431 E 4- II

5v3 =U3*MK--•;iy > 0"f>3H - 07 K.KI142E- •IS' 2 522K'jE-o* 2 45431E + II
>lj'3 1.M55SE£ - - OK 3.5l4DKE-'i7 *'. 2735&E - •I'j 1*5^2 IE-w 2.4543IE + II
?M\ 5.U2I95E:--{til 1 57S2IE-07 l f/22W - •: IX -."J2S7-SH - -:«N 3.3<V»74E + 15
6*>3 y.w>7iuK--W 24lf4*E-<i7 U>24ttfE- •:ix 2A'32lHH-iiN 3.36674E + 15
m 1 733I&E - - UK .lfth||f>H-ti? L' 1K524K- 0v 2.h2')y3E — 0* 3 3W74E+ 15
?y? 5.II4'>5K~. (M l.(W;44E-<i7 1.IU1X4H- •:w 2 M,s.tiJ2E-'.i,s 4.45742E+ IN
:"••'?• s 52W3E - • 09 ::isy4E-ii7 i.ofOwE- •is; .liMi?iS5H- its 4 45742E+ 1*
N?> 1 (S2475E-•viS J ?0/xiH —ii7 m7|:,im-:_ \W :7;^|f-oH-(iX 4 45742E+ IN

7b*/e 45; Fz'fferf va/wes ofkinetic constants ofcrude ethanol reforming to hydrogen
based on model 1, model 2andpower law [10]
Perimeter NKleJ # f, M^Jof # 2. R.wei law

A'0 HAM x 10- mwv i1^ 3 12 x .1

£ 4 43 \ 11v•-" 4 41 k 1•:.i-'
KA 'S.'r'S v | •l;
(i.d —

(i (i ImI

AY, ('.(! i:«''i

KH -
K0 - -
— i>.43
ADD ^(,f-: 4.5,:;.;

32
 Model 1 Model 2 Power Law Model
Measured rA, kg Measured rA, kg Measured rA, kg Predicted rA, kg mol/kg
Temp, K mol/kg cat s mol/kg cat s mol/kg cat s cat s
593 2.25E-06 3.25E-06 3.64E-06 3.36E-06
593 3.32E-06 4.08E-06 4.31E-06 4.07E-06
593 6.47E-06 5.24E-06 5.74E-06 5.81E-06
693 2.22E-06 3.41E-06 3.92E-06 2.17E-06
693 4.34E-06 5.00E-06 5.23E-06 4.65E-06
693 7.67E-06 6.04E-06 6.67E-06 7.96E-06
793 2.26E-06 3.45E-06 3.95E-06 4.34E-06
793 3.78E-06 4.67E-06 4.92E-06 5.43E-06
793 7.19E-06 5.95E-06 6.49E-06 9.41E-06
Table 4.6: The values ofthe measuredrAfor the three models along with the predicted rA.

Figure 4.3 represents the parity chart of comparison between the measured rates and
predicted rates using the three rate models. The use ofparity chart will be beneficial so
that the comparison can be made significantly by just observing the pattern ofthe chart.
This chart resembles the chart used in Akande etc [10] in order to differentiate and to
compare the measured and predicted rates of rate models.

0.00001

m, 0.000008
(0
o
en

oo 0.000006
E

0.000004

1
"g 0.000002

0.000002 0.000004 0.000006 0.000008 0.00001

measured rA, kg mol/kg cat s

Figure 4.2: Acomparison ifmeasured andpredictedrates ofreforming crude ethanol

within the temperature range of593-793 K[10]

33
A closer look at the chart, one can assume that model 1did not yield a satisfaction result.
This is because the average relative error for model 1is high, which is 29.1% (>25%).
Whereas for model 2 and power law, the average relative error are 21.1% and 17.2%
respectively (<25%).

The calculation of average relative error of all the three models can be referred in Table
4.7,4.8,4.9.

Table 4.7:Model1 average relative error.

Model 1 Model 1
Temp, K Measured rA, kg mol/kg cat s Predicted rA, kg mol/kg cat s Relative error
593 2.25E-06 3.36E-06 4.97E-01
593 3.32E-06 4.07E-06 2.26E-01
593 6.47E-06 5.81E-06 1.02E-01
693 2.22E-06 2.17E-06 2.39E-02
693 4.34E-06 4.65E-06 7.11E-02
693 7.67E-06 7.96E-06 3.75E-02
793 2.26E-06 4.34E-06 9.17E-01
793 3.78E-06 5.43E-06 4.37E-01
793 7.19E-06 9.41E-06 3.08E-01
Average 2.91E-01

Table 4.8:Model 2 average relative error.

Model 2 Model 2
Temp, K Measured rA, kg mol/kg cat s Predicted rA, kg mol/kg cat s Relative error
593 3.25E-06 3.36E-06 3.26E-02
593 4.08E-06 4.07E-06 2.89E-03
593 5.24E-06 5.81E-06 1.08E-01
693 3.41E-06 2.17E-06 3.64E-01
693 5.00E-06 4.65E-06 6.94E-02
693 6.04E-06 7.96E-06 3.18E-01
793 3.45E-06 4.34E-06 2.59E-01
793 4.67E-06 5.43E-06 1.62E-01
793 5.95E-06 9.41E-06 5.81E-01
Average 2.11E-01

34
 Table 4.9: PowerModel average relative error.
Power Law Model Power Law Model
Temp, K Measured rA, kg mol/kg cat s PredictedrA, kg mol/kg cat s Relative error
593 3.64E-06 3.36E-06 7.64E-02
593 4.31E-06 4.07E-06 5.46E-02
593 5.74E-06 5.81E-06 1.27E-02
693 3.92E-06 2.17E-06 4.47E-01
693 5.23E-06 4.65E-06 1.10E-01
693 6.67E-06 7.96E-06 1.93E-01
793 3.95E-06 4.34E-06 9.90E-02
793 4.92E-06 5.43E-06 1.03E-01
793 6.49E-06 9.41E-06 4.50E-01
Average 1.72E-01

With these results, it is concluded that the 2nd model of dissociation of adsorbed crude
ethanol, is on the active site ofthe rate determining steps as the average relative error
seems to be much lesser comparing to the 1st model, which is 21.1%. Besides that, the
power model is also on the active site of the rate determining steps, as the average
relative error is less than 25%, which is 17.2%.

With these, the value ofmeasured rA ofPower Law Model for each temperature 593 K,
693 K and 796 K are gathered in order to find the average. Table 4.10 shows the
calculated average for each temperature for Power Law Model.

Table 4.10: CalculatedAveragefor each TemperatureforPower Law Model

Trials
Average
Temperature^
593 K 3.64E-06 4.31E-06 5.74E-06 4.47E-06

693 K 3.92E-06 5.23E-06 6.67E-06 5.27E-06

793 K 3.95E-06 4.92E-06 6.49E-06 5.12E-06

35
 Then, each average value of rA for each temperature will be manually estimated on
Figure 4.1 by spotting the average value within the appropriate points. Let say, for
average rA of 593 K, which is 4.47E-06 kg mol/kg cat s, the value is somehow located
between point 1and point 2. Thus, the conversion, Xfor rA of4.47E-06 kg mol/kg cat sis
estimated 0.45. The same method is applied to the rest ofrA so that the conversion, Xfor
each rA can be determined. Table 4.11 shows the conversion, Xfor each average rA.

Table 4.11: Conversion, XofCalculatedAveragefor each Temperaturefor Power Law

Model

Temperature, Average rA, Conversion, X

K
kg mol/kg cat s
593 4.47E-06 0.45

693 5.27E-06 0.49

793 5.12E-06 0.6

With the value ofconversion, Xfor each rA obtained, the amount ofproducts produced
after the steam reforming reactor can be determined by using the equation below obtained
from the work of Akande etc [10];

Conversion, X
__ kg Ihr(in) -kg Ihr(out)
kglhr(in)

According to [16], the amount of feed is 1200 kg/hr. The value of conversion, X is 0.6
because it is assumed that 793 K is the final temperature that will exit the reactor. Thus,
by doing the back calculation to get the amount of product after the steam reforming
reactor;

36
 n6_1200kg/hr-kg/hr(out)
UOOkg/hr
720 = 1200 kg/hr - kg/hr (out)
Out - 480 kg/hr (organics)

Where organics - ethanol + lactic acid +glycerol + maltose

4.1.2 The calculation of Component Balance

Based on the work ofKoumpouras et al. [21], the component balance ofgas phase can be
represented as:

dnts 2 / \

Where

ug - superficial velocity
0.46 ms"1
nUg = molar flowrate / of outlet gas mol s"1
niyf - molar flowrate i of feed gas mol s"1
km = mass transfer coefficient
0.15 ms"1
Rc = monolith channel radius, m = I x 10"3 m
8 — 0.5 (dimensionless)
<P = 0.1 (dimensionless)

&ads = 1563 kg m"3

fads = rA (obtained from the measured ofrate of reaction ofPower Law Model)
3.64xl0'6kg

37
 B.C. 1 atz=0, niig=nif
dn. />
-(0.46)-7^-(0.5)(0.1)(1563X3.64xl0~6)-0.15-^-(0.5)«
dz (0.001)
fUo
hf
Simplified

dn
'*f- =-326. l(nif)~ 6.174x10
dz

Integrate both sides

J^L=J-326-1^)-j6.174xl0-
nig =-326.1(^)2 ~6.174x10^ 2+<:

B.C. 2 at z=Z =0.15,

Based on the Huang et al. [18], the ni>s for each components are:
nco,g = 6.35xl0"3mol/s
nHao,s =0.11557 mol/s
nH2,g =0.4134 mol/s
nC02,g =0.09843 mol/s
nCH4,g =1.27xl0"3 mol/s

For CO

6.35xl0"3=-326.l(«co/)(Z)-6.174xl0-4(7) +c
c-6.35xl0-3+326.l(/7ca/)(7) +6.174xl0-4(7)
Fortf20

c=0.11557 +326.l(^A/)(7) +6.174xl0-4(L)

38
 For H2

c=0.11557 +326.l(^2/)(7) +6.174xl0-4(I)

For C02

c=0.09843 +326.l^^Z) +6.174x10^(7)

For CH4

c=0.11557 +326.l(^2/)(7) +6.174xl0^(7)

Thus, the amount offeed ofthe reformer reactor for each component is:

For CO

6.35xl0^3=-326.l(/7co/)(0.15)-6.174xl0-4(0.15)+6.35xl0-3 +326.1^OJP)(0.15)
+ 6.174xl0_4(0.15)
Solve

nco,f =0

The feed that entered the steam reformer reactor contains only ethanol C2H^O. Thus, all
components such as CO, H20, H2, C02, andCH4 are not contained in the feed stream.

So,

For H20, nB2af=0

For#2„ nm,r0
For C02i nC02,r °
For CH4, nCH4/= 0
For C2H60, nC2H60f= nuf= 0.635 mol s"1 (ethanol feed that enters reformer based on
Huang etal. [18])

39
 4.2 Water Gas Shift Reaction

4.2.1 The calculation of rate of reaction

For the water gas shift reaction, the work of Choi et al. [11] is used to obtain the
modeling equations. In this study, a number of rate expressions have been reported and
tested to evaluate the water gas shift reaction rate for various catalyst:

H20 + CO ^ *C02+H2

According to the work of Choi et al.[ll], they focused on the condition of temperature
and pressure as they used a small reactor in an ethanol reformer for a PEM fuel cell.
Basically, in the real industry, WGS reactor operates at a very high temperature and
pressure. With the catalyst of Sud-Chemie Cu/ZnO/Al203, which is acommercial catalyst,
the reaction kinetics and mechanism for the water gas shift reaction have been studied by
them.

In this work, they applied many rate ofreaction for CO, rco, but only the rate of reaction
ofempirical rate expression derieved from the numerical fitting is the most accurate and
simplest among all rate ofequation ofCO, rCo [11].

It is assumed that the reactor is plug flow and isothermal [11]. Although these
assumptions are not entirely correct, they allow observation of the general reactor
behavior.

Figure 4.4 [11] shows for at approximately an equal molar ofH20 and CO mixture, the
effect of space velocity VS conversion of CO (%). The three isotherms in this figure
show that as expected the conversion of the WGS reaction decreases slowly with
increasing space velocity.

40
 CO Conversion with Space Velocity at 428 K, 463 K,
493 K

80 -i
— 70 -

- 60
c
o
.2 50
£ 428
S> 40
> 463
c
o 30 H 493
o
O 20
o
10

6250 12500 17500

Space Velocity (GHSV, 1/hr)

Figure 4.3: Water Gas Shift Reaction -CO conversion as afunction ofspace velocity

To determine the predicted rate ofequation, rco various slopes were taken at each ofthe
three temperatures in order to differentiate with the measured rate of equation, rco of all
three models. The slopes ofeach ofthe points were calculated manually. Based on Figure
4.4, only three points (1) 6250 hr'1, (2) 12500 hr"1, (3) 17500 hr"1 were taken at each
temperature.

For the temperature of 428 K;

25
r, =
6250
-3
-4x10

20
rA =
12500
-3
= 1.6x10

41
 15
rA =
17500

= 8.57xlO"4

Thus, by doing the same method offinding the value ofthe slope (predicted rco) to both
temperature, value of each slopes are represented in Table 4.12

^r
428 K 4x10 1.6 xKT 8.57 xlO"4
^~
463 K 7.2x10 "3.12 xlO-3 1.33 xlO-3"
493 K 0.01104 4.64 xlO"3 2.88 x 10'3

Table 4.12: Value ofslopes (predicted rCo) at temperature 428,463, 493 K

Then, the rco is calculated by using MATLAB software by referring the equation given
in the work of Choi et al. [11], which is the rate of reaction of empirical rate expression
derived from the numerical fitting. The equations and variables are written in *.m format
and then it is then saved by the name of WGS-numec_fit.m. The MATLAB language
can be referred in Appendix 7 for more details.

rC0=2.96xWe~ElRT (P P _ P CQ2rH2
P
\
\iCOrH20 ~ )
K

Where H20 = a

CO = b

C02 = c

H2 = d

42
The activation energy for the empirical model is 47.4 kJ/mol [11], which is consistent
with other values from other literature ofWGS. Equilibrium constant, Ke is calculated by
referring to the equation in Choi et al. [11].

( 4577.8
Ke = exp -4.33
T

Thus, by using MATLAB, the calculated rco are:

For Tj= 428 K; and Ke= 581.6

rco = kxexp(E/(R x Tj)) x(Pbx PJ-(PC xP/Ke})

rco 0.0014

• For Tr* 463 K; and K<r 259.11

rco = kxexp(E/(RxT2)) x(PhxPJ-(PC xP/Ke,)

rco 0.0039

For T3= 493 K; and Ke= 141.96

rco = Ax exp(E/(RxT3)) x(Pbx PJ-(PC xP/Ke3)

rco3 0.0082

43
 Then, each value ofrco for each temperature will be manually estimated on Figure 4.4 by
spotting the average value within the appropriate points. Let say, for the value rco of 428
K, which is 0.0014, the value is somehow located between point 2and point 3. Thus, the
conversion, Xfor rco of0.0014 is estimated 24 %. The same method is applied to the rest
ofrco so that the conversion, Xfor each rco can be determined. Table 4.13 shows the
conversion, Xfor each average rCo-

Temperature, rA, Conversion, X

K
mol gcat"1 h"1
428 0.0014 0.24

463 0.0039 0.45

493 0.0082 0.68

Table 4.13: Conversion, Xofrcofor each Temperature for empirical rate expression
derievedfrom the numericalfitting

Like always, the highest temperature, which is 493 Kwill be selected as it is assumed as
an exit temperature of the shift reactor. The conversion, Xof temperature at 493 K is
0.68. According to Figure 4.4, at 0.68 of X, the space velocity will be approximately
about 6250 h"1.

Based on the journal ofRioche et al. [2], by using the equation ofGHSV, the volumetric
flowrate ofgaseous model compound can bedetermined.

GHSV - Volumetric Flow rate ofGaseous Model Compound/ Volume ofCatalyst

In the work of Choi et al. [11], the volume of catalyst is 22.6 inch3 (360.4 cm3).
Therefore, by doing the back calculation to get the volumetric flow rate ofgaseous model
compound after the shift reactor:

44
 Volumetric Flow rate ofGaseous Model Compound
= 6250 h'1x 360.4 cm3
= 2,252,500 cm3H1
= 2.2525 m3h'J

According the calculation above, it can conclude that the amount of volumetric flow rate
ofgaseous model compound exiting the shift reactor is 2.2525-m3 h"1

4.2.2 The calculation of Component Balance

Based on the work ofHuang et al. [18], the component balance ofgas phase can be
represented as:

—
nfi I* +, nfi Iz+a. xrf&zJ + 1 f3 A
^far = fj
Az Az Az 4Az

Simplify and integrate

^=-<,2A^)~MlinJ(z) +c

Based on the Huang et al. [18], the n^-for each components are:
nco,g =6.35xl0"3mols"1
nmo,g =0.11557 mol s"1
nH2lS =0.4134 mol s"1
nC02,g =0.09843 mol s"1
riN2>s =0
nCH4,g -1.27xl0"3mols"1

And

J= 152400 mol/cm2/s
din - 0.1 cm

45
 B.C. at z=L

nCo,g =0
riH20,g = 0

^-7
nH2,g = 5 x\0" (nt,o) -3.175 x 10"7 mol s"1
nco2lg = 370 x 10-6 (nti0) = 2.3495 x 10"4 mol s"1

Apply B.C. in order to determine the constant, cfor each component

For CO

c=--^(0.1)2(152400)(Z)+^(0.1)(152400)(I,)
ForftO

c=--^(0.1)2(152400)(i.) +^(0.1X152400)(Z)
For H2

c=--^(0.1)2(152400)(i) +^(0.1Xl5240O)(I) +3.175xl0-7

For C02

c=-7^(0.1)2(152400)(I) +^(0.1)(152400)(Z) +2.3495xl0^4

The product produced in the shift reactor contains only H2 and C02 components. Thus,
two components such as CO and H20 are not contained in the product stream, (z =61
cm)

For CO

nco,s=0

For H20

nii2O,s=0

46
 For ft
«H2.s

-^(0.1)2(152400)(61)-^(0.1)(152400)(61) +i,(0.1)(152400)(61)-I^(0.1)2(152400)(61) +3.175x10

3.175 xlO-7 mol s'1

For C02
nco2,s

--^(0.1)2(152400)(61)-^(0.1)(152400)(61) +^(0.1)(152400)(61)-I^(0.1)2(152400)(61) +2.3495xl.

- 2.3495 x!0~4 mol s"1

4.3 Hydrogen Purification (PSA Purification)

4.3.1 The calculation of Component Balance

For the PSA column, literature of Ribeiro et al. [12] is used to model the hydrogen
purification system. Basically, the system from this literature used activated carbon as a
bed for the column with the length of 0.5 m. Based on the work ofRibeiro et al. [12], the
component balance ofgasphase can be represented as:

d( \ dn . (1 - e)akf / —\
dz{ a^dz)

47
 At B.C. 1 which isz^Q

»o >««,,-= *o«i,*-«D„»„,
dz

Then replace the B.C.I into the above equation. The new equation of component balance

kf = 9.28xl0"2 ms"1
u0 = 0.46 ms"1
_ A
ap
V

_1.72xl0"5
6.7xl0"9

= 2567.4 m"1

Rp = 1.17xl0'3m

sp -0.566

nM = 3/5

Dp,H2 = 4.99xl0-6m2/s
Dp,co2 = 279.99 m2/s

Thus, use the new equation in order to determine ngri>

For H2

- j-(0 46rt 17Srin^)-n^^» a-0.38)(2567.4)(9.28xl0-2)r x

dzx ' dt 1+192.46 VbjiJ-v

nf,co2 =3.175 xlO-7 mol s"1

SPQM$P,H2

48
 _2567.4(0.0928)(U7xlQ-3)2
(0.566)(3/5)(4.99xl0-6)
= 192.46 (dimensionless)

Integrate and solve

1.4605x10-'(0 =0.38^ \z) +0.765^ )

At t= 80,000 s(500 cycle number) as stated in literature ofRibeiro et al. [12] (to obtain
the maximum purity ofH2) and z - 0.5 m, thus

ng^2 =0.01 mol s"1

For CO2

-^(04fir7^Qsnn-*).n^^^ (l-0.38)(2567.4)(9.28xlQ-2) , .
^ 7 dt 1+ 279.99 vVcoJ-U
«/co2 -2.3495 xlO-4 mol s"1
&- , 2567.4(0.0928)(l,17xl03)2
(0.566)(3/5)(3.43xl0"6)
- 279.99 (dimensionless)

Integrate and solve

1.08x10^(0 - 0.38^ \z) +0.765^)

At t= 80,000 s and z = 0.5 m, thus

ngC0 -9.05 mol s"1

49
 Since onlyH2 and C02 that are produced from the shift reactor, these two will be the only
products that will flow at the outlet of PSA column stream. Table 4.12 shows the
calculated molar flowrates along the bed ofactivated carbon from 0mto 0.5 mat 80,000
sin order to compare the purity ofhydrogen ofthis project with the work ofRibeiro et al.
[12]. Figure 4.5 and Figure 4.6 show the trends of the molar flowrate for both H2 and
C02 along the bed.

Length Molarflowrate ofH2 Molarflowrate of

(m) (mol s'1) CO;(mol s-1)
0.0 0.02 11.29
0.1 0.01 10.76
0.2 0.01 10.27
0.3 0.01 9.83
0.4 0.01 9.42
0.5 0.01 9.05
Table 4.14: Calculatedmolarflowratesfor H2 andC02 along the bedfrom 0mto 0.5 m
att=80,000s

Molar Flowrate of H2

— 0.02 -

g 0.02 i >
~™ ^- ,_
-~^___
~~# §
2 0.01 -

£ 0.01
IB

| 0.00 - i

c 0.1 0.2 0.3 0.4 0.5 0.6

Mm)

Figure 4.4: Graph ofmolarflowratefor H2 along the bedfrom 0mto 0.5 mat 80,000

50
 Molar Flowrate of CO2

12

5 10
I B
at
1 6
I 4
re

o 2
£
0

0.1 0.2 0.3 0.4 0.5 0.6

Mm)

Figure 4.5: Graph ofmolarflowratefor CO2 along the bedfrom 0mto 0.5 mat 80,000 s

According to these two graphs, it is concluded that the amount of molar flowrates for
both H2 and C02 are decreasing with the increasing length of activated carbon bed (z).
This is because the high diffusivity value makes both the uptake and regeneration more
difficult, which broadens the H2 and C02 along the bed. Although the molar flowrate of
H2 seems to be decreasing with the increasing length ofbed, it is clear that somehow the
flowrate will stay constant.

According to the work ofRibeiro et al. [12], the amount ofcycle number is more than
500 cycle number (>805000 s). This is because with the longer the period, the whole bed
can achieved the stabilization oftemperature, thus make the purification process better.
This process also shows that the C02 does stay in the activated carbon area, which satisfy
the design requisite of this process.

According to the work ofRioche et al. [15], the formula of#2 purity can be represented
as

H2 purity % - molesof H2obtained (at2 =0.5 mat t=80,000s)

(2n + m/2 - k) x moles of carbon in the feed

51
 The feed is ethanol, which C2H(0. Thus, that makes «=2, m=6, and *=1. And also
amount of moles of carbon in the feed is 1.685xl0"3.

0.01
H2 purity % = xlOO
(2(2) +6/2-l)xl.685xl0 -3

-98.9119%

The amount ofH2 purity for this work is 98.9119 %, less than the amount offt purity of
Ribeiro et al. [12], which is 99.9994 %although the cycle number is the same (500 cycles
or 80,000 s). The reason is the work of Ribeiro et al. [12] used double layer beds
consisting of both activated carbon and zeolite. This will make the purification process
more effective and efficient. Whereas for this work, the bed only consist of activated
carbon with the purpose is to trap the carbon dioxide. Thus, by comparing this work and
Ribeiro etal. [12], the relative error (%) can be determined.

98.9119-99.9994
Relative error % xlOO
99.9994

1.08%

52
 CHAPTER 5: CONCLUSION

Basically, in conclusion of this project, the objectives of this project are achieved. But,
there is a weakness for this project. Supposedly, this project should use bio oil as the
main sample of modeling, but due to some difficulties, the bio oil was represented by
ethanol, which will replicate the kinetic modeling of bio oil. By doing this project, the
aim to develop the modeling way to simulate the sequence of processes in hydrogen
production from bio-oil in MATLAB can be achieved significantly.

Based on the results of kinetic modeling of steam reforming, Model 1does not yield a
satisfaction result, as the Average Relative Error is high, which is 29.1%. For Model 2
and Power Law, both have the Average Relative Error of21.1% and 17.2% respectively.
Based on Akande et al. [10], if the Average Relative Error is less than 25 %, then it can
be considered as agood result. Basically, Model 2and Power Law are on the active site
ofthe Rate Determining Step as the Average Relative Error is less than 25 %.

For the Water Gas Shift reaction, it is concluded that the final temperature for the shift
reactor for this system is around 493 K. The conversion ofthe reactor at the temperature
of493 Kis 68 %and thus, the GHSV will be approximately about 6250 h"1. According to
Choi et al. [11] the rate of reaction of empirical rate expression derived from the
numerical fitting is the most accurate and simplest among all rate of equation of CO.
Based on the molar balance calculation, only two products produced in the shift reactor,
which are H2 and CO2 as they will besent toPSA column.

For the PSA purification of hydrogen, H2 and C02 will be the only products that will
flow at the outlet of PSA column stream. Based on the results, the amount of molar
flowrates for both H2 and C02 are decreasing with the increasing length of activated
carbon bed (z). The H2 purity for this work is 98.9119 %because the bed only consists of
activated carbon with the purpose is to trap C02. It is determined that the Relative error
(%) for this work and Ribeiro et al., based on H2 purity is 1.08 %.

53
 REFERENCES

[1] Abedi J, Yeboah YD, RealfFM, McGee D, Howard J, Bota KB. An integrated
approach to hydrogen production from agricultural residues for use in urban
transportation. In: Proceedings of the 2001 DOE hydrogen program review
NREL/CP-570-30535.

[2] Rioch CE, Kulkarni S, Meunier FC, Breen JP5 Burch R. Steam reforming of
model compounds and fast pyrolysis bio-oil on supported noble metal catalysts.
Appl CatalB Environ2005;61:130-9.
[3] Huber GW, Dumesic JA. An overview of aqueous-phase catalytic processes for
production of hydrogen and alkanes in abiorefinery. Catal Today 2006;111(1-
2): 119-32.

[4] Piskorz J, Scott DS, Radlein D. In: Soltes EJ, Milne TA., editors. Pyrolysis oils
from biomass: producing, analyzing and upgrading. ACS Symposium Series 376.
Washington, DC:American Chemical Society; 1988. p. 167-78.
[5] Chernik S, Wang D, Montane D, Chornet E. In: Bridgwater AV, Boocock DGB.,
editors. Developments in thermochemical biomass conversion. London: Blackie
Academic &Professional; 1996. p. 672-86.
[6] Garcia L, French R, Czernik S, Chornet E. Catalytic steam reforming of bio-oils
for the production of hydrogen: effects of catalyst composition. Appl Catal A
General 2000;201:225-39.
[7] Diebold J, Scahill J. In: Soltes EJ, Milne TA., editors. Pyrolysis oil from biomass.
ACS Symposium Series 376. Washington, DC: American Chemical Society.
1998. p. 31.

[8] Evans RJ, Milne TA. Molecular characterization of the pyrolysis of biomass. 1
Fundamental. Energy Fuels 1987; 1(2);123.
[9] Douglas J.M., Conceptual Design of Chemical Process, New York, McGraw Hill
Chemical Engineering Series, 1998.
[10] Kinetic Modeling of Hydrogen production by the catalytic reforming of crude
ethanol over aco-precipitated Ni-Al203 catalyst in apacked bed tubular reactor. -
Abayomi Akande, Ahmed Aboudheir, Raphael Idem, Ajay Dalai.

54
 [11] "Water gas shift reaction kinetics and reactor modeling for fuel cell grade
hydrogen"- Yongtaek Choi, Harvey G. Stenger -Department of Chemical
Engineering, Lehigh University, Bethlehem, PA 18015, USA
[12] "A parametric study of layered bed PSA for hydrogen purification" - Ana M.
Ribeiro, Carlos A. Grande, Filipe V.S. Lopes, Jose M. Loureiro, Alirio E.
Rodrigues Laboratory ofSeparation and Reaction Engineering (LSRE), Associate
Laboratory, Department of Chemical Engineering, Faculty of Engineering,
University ofPorto, Rua Dr. Roberto Frias, s/n, 4200-465 Porto, Portugal
[13] Vagia, CE and Lemonidou, A.A. "Thermodynamic analysis of hydrogen
production via steam reforming of selected components of aqueous bio-oil
fraction", International Journal ofEnergy, 33,212-223,2007.
[14] "Thermodynamic analysis of hydrogen production via steam reforming of
selected components of aqueous bio-oil fraction" - Ekaterini Ch. Vagia, Angeliki
A. Lemonidou - Department of Chemical Engineering, Aristotle University of
Thessaloniki and CERTH/CPERI, P.O. Box 1517, University Campus, GR-54006
Thessaloniki, Greece.

[15] Cyrille Rioche, Shrikant Kulkarni, Frederic C. Meunier , John P. Breen, Robbie
Burch CenTACat, Queen's University Belfast, Belfast, BT9 5AG Northern
Ireland, UK (2005).
[16] "Simulation of Pyrolysis Bio-oil Upgrading Into Hydrogen" - Murni M. Ahmad,
Laveena M. Chugani, Cheng Seong Khor and Suzana Yusup - Chemical
Engineering Department, Universiti Teknologi Petronas, Perak, Malaysia.
[17] "Hydrogen from biomass-production by steam reforming of biomass pyrolysis
oil" - Stefan Czernik, Robert Evans, Richard French - National Renewable
Energy Laboratory, 1617 Cole Boulevard, Golden, CO 80401, United States
[18] "Modeling of C02-selective water gas shift membrane reactor for fuel cell' - Jin
Huang, Louei El-Azzami, and W.S. Winston Ho. Department of Chemical and
Biomolecular Engineering, The Ohio State University, 2041 College Road,
Columbus, OH 43210-1180, USA.
[19] M.-C. Yang, EX. Cussler, Designing hollow-fiber contactors, AIChE J. 32 (1986)
1910-1916.

55
[20] Pure hydrogen from pyrolysis oil using the steam-iron process - M.F. Bleeker,
S.R.A. Kersten, H.J. Veringa. University ofTwente, Department of Science and
Technology, Enschede, The Netherlands
[21] Mathematical modelling of low-temperature hydrogen production with in situ
C02 capture - Georgios C. Koumpouras, Esat Alpay, Frantisek Stepanek
[22] Ethanol steam reforming in adense Pd-Ag membrane reactor: Amodeling work.
Comparison with the traditional system - F Gallucci, MDe Falco, S Tosti, L
Marrelli, A Basile.

56
 APPENDICES

Appendix 1: Operating Unit Description

Mixer

The feed (bio and water) entering the process are mixed before heated up in afurnace.

Furnace

The stream from the mixer is heated up from 41°C to 650°C using afurnace due to the
need ofhigh temperature requirement before it is fed to the steam reformer.

Steam Reformer

The stream from the furnace is fed to fixed bed catalytic reformer that operates at 650°C
and 1bar. The reforming ofthe bio oil will produce CH4, C02, H2, and CO.

Shift Reactor

The stream is cooled to 200°C and compressed to 17.5 bar before entering the shift
reactor. The hydrogen yield is maximized by shift conversion where carbon monoxide
reacts with steamto producecarbon monoxide.

Flash Drum

The gaseous mixture, which is rich in hydrogen content, is cooled to 30°C before it is
separated in flash drum at 16 bar. The steam is condensed to water, which is separated
from the gaseous mixture.

Pressure Swing Absorption (PSA)

The gaseous mixture is purified into hydrogen rich gas stream in PSA. The driving force
for the separation is the difference in impurity partial pressure between the feed and the
tail gas. The feed pressure range is between 15 to 29 bar. The tail gas pressure is chosen
to be as low as possible for the pressure, which is normally around 1to 2 bar. The
operating temperature for PSA is 35°C

57
Coolers

For the 1st cooler, it is used to cool the hot stream from steam reformer from 500°C to
200°C before entering the shift reactor, which operates lower than 200°C. The 2nd cooler
is used to cool the hot stream from the shift reactor 200°C to 30°C before entering the
flash drum. The cooling medium is Molten salt for both ofthese coolers.

58
Appendix 2: Project Gant Chart FYP 1
No. Task/Week 11 2 3 4 5 6 7 8 *) i]l 12 13 14 15
1 Selection of ProjectTopic
Submission of Proposal
Approval of Project Proposal
Selection of Project Title
2 Preliminary Research Work
:|%':
Objective of the Project
InformationGathering
-Bio Oil Components Sfi
X
-Chemical Reaction Analysis
H
3 Submisson of Preliminary Report CO
Ed
4 s
Seminar 1
CO
5 Project Work a
Determining Suitable Journal for Modeling
Conceptual Design
6 Submisson of Progress Report
7 Seminar with Examiner
8 Project Work Continues
9 {Submission of Interim Report Final Draft
10 i3ral Presentation
Appendix 3: Project Gant Chart FYP 2
Task / Week 1 2 3 4 f 6 ' i 9 ii 12 13 U 1* 16 17 18 19 20
Briefing & update on students
progress
Project work commences
Submission ofProgress Report
Submission of Progress Report 2 (Draft
of Final Report) K
B
Poster Exhibition / Pre-EDX/ Progress CO
Reporting
Ed
CO
EDX
a
Submission of Final Report (CD
Softcopy & Softbound)
Final Oral Presentation
Submission of hardbound copies
—
 Appendix 4: steamreforming modell.m

rll=[k*exp(-E/(R*Tl))]*[Nail-
(Ncll-2*Ndll-6)/(Kpl*Nbll-3)]/[l+((Kf^Ncll^Ndll-3)/(Nbll))
11^3)/(Nbll-2)) +((Ke*Ncll*2*Ndll4 /(Nbll*3)J 1 +MKa^Tn11*^
(Kg*Ncll*Nd
k=8.91E+002; J
E=4.03E+003;
R=8.314;
Tl=593;
Nall=5.70596E-009;
Nbll=1.5733lE-007;
Ncll=8.40614E-009;
Ndll=2.40685E-008;
Kpl=2.45431E+011;
Kf=0.0;
Kg=0.0;
Ke=0.0;

rl2=[k*exp(-E/(R*Tl))]*[Nal2-
fNcl2A2*Ndl2-6)/(Kpl*Nbl2-3)I/[l+{(Kf*Ncl2*Ndl2A3)/(Nbl2}) +nKcT*Wnl9*W^
12A3]/(Nbl2-2J)+({Ke*Ncl2-2*Ndl2-6J/(Nbl2-3)n
Nal2=8.43861E-009;
Nbl2-2.0793E-007;
Ncl2=8.811424E-009;
Ndl2=2.52289E-008;

rl3=[k*exp(-E/(R*Tl))]*[Nal3-
(Ncl3-2*Ndl3^6)/(Kpl*Nbl3*3)]/[!+((Kf*Ncl3*Ndl3*3)/(Nbl3))+((Ka*Ncl3*Nd
13^3)/(Nbl3^2)) +({Ke*Ncl3-2*Ndl3-6)/(Nbl3^3})l XJ'' +^K9 Ncl3 Nd
Nal3=1.64558E-008;
Nbl3=3.51408E-007;
Ncl3=9.27356E-009;
Ndl3=2.6552lE-008;

r21^[k*exp(-E/(R*T2))]*[Na21-
(Nc21^2*Nd21-6)/(Kp2*Nb21*3)]/[!+(<Kf*Nc21*Nd21^3)/(Nb21))+((Ka*Nc21*Nd
21A3)/(Nb21-2))+((Ke*Nc21-2*Nd21A6)/(Nb21-3J)l
T2=693; J
J/U^1,) +(<K9 Nc21 Nd
Na21=5.02195E-009;
Nb21=1.5782lE-007;
Nc21=1.0229E-008;
Nd21=2.92878E-008;
Kp2=3.36674E+015;

r22=[k*exp(-E/(R*T2))]*[Na22-
fNc22A2*Wd22A6)/{Kp2*Nb22A3)]/[l+({Kf*Nc22*Nd22-3J/(Nb22))+ffKa*Wn99*HH
22-3)/{Nb22-2))+((Ke-Nc22-2*Nd22-6)/(Nb22^3))l
Na22=9.80704E-009;
?
Nb22=2.41648E-007;

61
Nc22=1.02403E-008;
Nd22=2.93201E-008;

r23=[k*exp(-E/(R*T2))]*[Na23-
(Nc23A2*Nd23A6)/(Kp2*Nb23A3)]/[!+((Kf*Nc23*Nd23A3)/(Nb23))+((Kg*Nc23*Nd
23A3) /(Nb23A2) )+((Ke*Nc23A2*Nd23A6) /(Nb23A3) )]
Na23=1.73306E-008;
Nb23=3.66116E-007;
Nc23=9.18524E-009;
Nd23=2.62993E-008;

r31=[k*exp(-E/(R*T3))]*[Na31-
(Nc31A2*Nd31A6)/(Kp3*Nb31A3)]/[1+((Kf*Nc31*Nd31A3)/(Nb31))+((Kg*Nc31*Nd
31A3)/(Nb31A2))+((Ke*Nc31A2*Nd31A6)/(Nb31A3))]
T3=693;
Na31=5.11495E-009;
Nb31=1.60744E-007;
Nc31=1.04184E-008;
Nd31=2.98302E-008;
Kp3=4.45742E+018;

r32=[k*exp(-E/(R*T3))]*[Na32-
(Nc32A2*Nd32A6)/(Kp3*Nb32A3)]/[1+((Kf*Nc32*Nd32A3)/(Nb32))+((Kg*Nc32*Nd
32A3)/(Nb32A2))+((Ke*Nc32A2*Nd32A6)/(Nb32A3))]
Na32^8.52993E-009;
Nb32=2.21894E-007;
Nc32=1.06204E-008;
Nd32=3.04085E-008;

r33-[k*exp(-E/(R*T3))]*[Na33-
(Nc33A2*Nd33A6)/(Kp3*Nb33A3)]/[1+((Kf*Nc33*Nd33A3)/(Nb33))+((Kg*Nc33*Nd
33A3)/(Nb33A2))+((Ke*Nc33A2*Nd33A6)/(Nb33A3))]
Na33=1.62475E-008;
Nb33=3.50781E-007;
Nc33=9.71516E-009;
Nd33=2.78165E-008;

62
Nc23=9.18524E-009;
Nd23=2.62993E-008;

r31=[k*exp{-E/(R*T3))]*[Na31-
(Nc31A2*Nd31A6)/(Kp3*Nb31A3)]/[1+(Ka*Na31)]A2
T3=693;
Na31-5.11495E-009,
Nb31=1.60744E-007
Nc31=1.04184E-008
Nd31=2.98302E-008
Kp3=4.45742E+018;

r32=[k*exp(-E/(R*T3))]*[Na32-
(Nc32A2*Nd32A6)/(Kp3*Nb32A3)]/[1+(Ka*Na32)]A2
Na32=8.52993E~009
Nb32=2.21894E-007
Nc32=1.06204E-008
Nd32=3.04085E-008

r33=[k*exp(-E/(R*T3))]*[Na33-
(Nc33A2*Nd33A6)/(Kp3*Nb33A3)]/[l+(Ka*Na33
Na33=1.62475E-008
Nb33=3.50781E-007
Nc33=9.71516E-009
Nd33=2.78165E-008

64
Appendix 6: power_law_modeI.m

rll=[k*exp(-E/(R*Tl))]*[Nail]A0.43
k=3.12E-002;
E=4.41E+003;
R=8.314;
Tl=593;
Nall-5.70596E-009;

r!2=[k*exp(-E/(R*Tl))]*[Nal2]A0.43
Nal2=8.4386lE-009;

rl3=[k*exp(-E/(R*Tl))]*[Nal3]A0.43
Nal3=1.64558E-008;

r21=[k*exp(-E/(R*T2))]*[Na21]A0.43
T2=693;
Na21^5.02195E-009;

r22=[k*exp(-E/(R*T2))]*[Na22] A0.43
Na22=9.80704E-009;

r23=[k*exp(-E/(R*T2))]*[Na23]A0.43
Na23=1.73306E-008;

r31=[k*exp{-E/(R*T3))]*[Na31]A0.43
T3=693;
Na31=5.11495E-009;

r32=[k*exp(-E/(R*T3))}*[Na32]A0.43
Na32=8.52993E-009;

r33=[k*exp(-E/(R*T3))]*[Na33]A0.43
Na33=1.62475E-008;

65
 Appendix 7: WGS-numec fitm

k- 2.96E+005;
Pa= 0.58;
Pb^ 0.005;
Pc= 0.015;
Pd= 0.4;
E= -47.4E+003;
R= 8.314;
Tl= 428;
Kel= 581.6;

rC01=k*exp(E/(R*T1))*(Pb*Pa)-(Pc*Pd/Kel

T2= 463;
Ke2= 259.11;

rC02=k*exp{E/(R*T2))*(Pb*Pa)-(Pc*Pd/Ke2

T3= 493;
Ke3= 141.96;

rC03=k*exp(E/(R*T3))*(Pb*Pa)-(Pc*Pd/Ke3

66