Vous êtes sur la page 1sur 230

The Pennsylvania State University

The Graduate School


Department of Chemical Engineering

A SOLID CATALYST METHOD FOR BIODIESEL PRODUCTION

A Dissertation in
Chemical Engineering
by
Dheeban Chakravarthi Kannan

Submitted in Partial Fulfillment


of the Requirements
for the Degree of

Doctor of Philosophy

August 2009

The dissertation of Dheeban Chakravarthi Kannan was reviewed and approved* by the
following:

Jack V. Matson
Professor of Environmental Engineering
Dissertation Adviser

Themis Matsoukas
Associate Professor of Chemical Engineering
Chair of Committee

Joseph M. Perez
Senior Research Scientist, Department of Chemical Engineering

Wallis A. Lloyd
Adjunct Professor of Chemical Engineering

Brian A. Dempsey
Professor of Environmental Engineering

Thomas P. Hettmansperger
Professor of Statistics

Andrew Zydney
Professor of Chemical Engineering
Head of the Department of Chemical Engineering

*Signatures are on file in the Graduate School

ABSTRACT

Biodiesel has considerable production potential as a renewable source of energy. The


conventional processes use soluble alkali catalysts that contaminate the biodiesel and
glycerol products, and present separation problems. An efficient and clean process is
crucial for large scale commercial production. Solid catalysts have the potential to
eliminate these problems.

A method has been developed to produce biodiesel using a solid catalyst. The reaction is
carried out at high temperature and pressure conditions (260 C, 70 atm). The high
temperature is not a problem since the solid catalyst is part of a continuous process in
which heat energy can be recovered. The reaction time is short (15 minutes) compared to
that of the conventional processes (~ 100 minutes).

Promising catalysts were identified from batch tests; and MnO was found to be the most
effective catalyst from the lab-scale packed-bed reactor tests. The reaction conditions
allow for this mild solid base which can have a long life. The conventional process
conditions (60~70 C, 70 atm) need specially prepared strong bases that lose their activity
quickly. The kinetics of the MnO-catalyzed reaction has been studied. External mass
transfer limited effects have been identified. MnO was not the strongest of the bases
tested, but it was the most effective. The reasons have been evaluated. MnO has a long
catalyst life (more than 500 hours). Also, the amphoteric nature of MnO aids in the
conversion of free fatty acids. This means used vegetable oil can be converted to
biodiesel efficiently. MnO can sustain high water content in the feedstock without
iii

diminishing efficiency. The method has been found to be favourable in terms of low
pressure drop.

Lastly, the MnO solid catalyst process has been compared with other processes and was
found to be favourable in terms of reaction rate, catalyst life, pressure, temperature, side
reactions and maintenance.

iv

TABLE OF CONTENTS

ACKNOWLEDGEMENTS... ix
Chapter 1 Introduction............................................................... 1
Hypothesis.. 3
Chapter 2 Background....................... 5
Vegetable oil composition......... 5
Problems of direct vegetable oil usage in diesel engines.. 6
Methods of modifying vegetable oil for diesel engines..... 7
Fuel consumption statistics........ 7
Oil yields........ 9
Biodiesel energy balance.......... 10
The scope for biodiesel production.................................. 10
Biodiesel and diesel Performance, handling and usage comparisons... 12
Conventional Process............... 14
Drawbacks of the conventional process..... 18
Solid catalysts.......................... 20
Solid bases...................... 20
Solid acids....................... 22
Solid catalysts for biodiesel reaction...................... 24
Chapter 3 Approach to Method Development......................... 31
Chapter 4 Initial Tests and Catalyst-Screening........................ 36
Batch reactor set-up and experimental procedure............ 36
Product analysis....................... 37
Solid base salts......................... 37
Metal oxide bases..................... 38
Free fatty acid conversion........ 43
Activity sustenance of the catalysts..... 43

Effect of molar ratio................. 44


Effect of cosolvent................... 44
Ultrasonic cavitation without catalyst............................. 46
Chapter 5 Lab-Scale Packed-Bed Reactor.............................. 47
Experimental set-up............ 47
Experiments............................. 54
Results......... 59
Data validation........ 62
Chapter 6 Reaction Kinetics........... 63
External mass transfer............. 63
Internal mass transfer.............. 66
Order of reaction and rate constant................................. 67
Effect of temperature on reaction rate............................ 74
Surface area and activity......... 82
BET surface area........... 82
Cumulative pore area over pore size range.......... 84
Comparative acidity/basicity scale........................ 86
Thermogravimetric analysis of temperature-programmed CO2 desorption.. 88
Rate constant vs surface area per unit volume...... 93
Proposed Reaction Mechanism....................................... 94
Chapter 7 Studies for Commercialization............................. 100
Activity sustenance tests....... 100
Optimal molar ratio................... 104
Effect of pressure.................. 107
Free fatty acid conversion..... 112
Effect of water.............................. 116
Equilibrium conversion........ 120
Process pressure-drop........... 124
High temperature considerations.................................. 129

vi

Alcohol side reaction.... 129


Vegetable oil side reaction................................... 135
Chapter 8 Comparison of MnO Catalyst Method with Other Methods 137
Reaction comparison............. 137
Supercritical methanol process............................ 138
Tateno/Goto method.... 138
Mcgyan process........... 143
Two-step supercritical process............................ 144
Propane cosolvent process................................... 144
Di Serio catalysts and calcined hydrotalcites...... 145
Other methods.............. 147
Comparison with calcium oxide... 149
Activation energy......... 150
Process design comparison with the conventional process... 150
Chapter 9 Conclusion.................... 157
Future work....... 160
Abbreviations, Notations and Symbols................................. 164
References..................................... 166
Appendices.................................... 181
Appendix A Thin Layer Chromatography...................... 181
Appendix B Gas Chromatography.............................. 182
Appendix C Water and Saponification....................... 189
Appendix D Titrimetry....... 190
Appendix E LHSV and WHSV.................................. 191
Appendix F ZSM-5............. 191
Appendix G Cetane number... 192
Appendix H Capital Cost Estimation of 34 Million Gallon per Year Biodiesel

vii

Plant.. 193

Appendix I Process Design of a Mobile Biodiesel Production Unit.. 199

viii

ACKNOWLEDGEMENTS

My time of doctoral research has been a dream, and I shall cherish that forever. There
are many wonderful people who made it what it was. This has been a different project
from other doctoral projects in that the basic idea behind the research did not exist in
advance, nor were there any pre-allocated funds. I had basically approached my adviser
Dr. Jack Matson that I wanted to do a project of environmental significance. We looked
at a variety of ideas and settled on the biodiesel project at the end of the first semester,
fall 2004. We took the whole spring 2004 semester for literature review. There was not a
single experimental data collected in the first two semesters. He had that kind of
confidence in me and that kind of interest in environmental issues. I never knew the
pressure of finances, and we did not have any outside funding for the research. That is
how well he took care of me and the project. The freedom he gave me and the
confidence he had in me are what that essentially made up this research. On looking
back, that kind of blind confidence in the project to be shared by two people until we
identified MnO as a promising catalyst 20 months into the project, feels remarkable. Dr.
Matson is a special adviser, and a friend for life.

I cannot thank enough Drs. Larry Duda and Ronald Danner, my masters thesis advisers.
Since this was a project decided upon by ourselves and was started from scratch, we had
no related experimental and analytical equipments to begin with. Dr. Duda and Dr.
Danner offered me one of their lab rooms and their equipments to carry out my research.
They have been my thorough well-wishers. Dr. Duda, in fact, was the one who gave me

ix

the idea that I could find an adviser outside the Department if I dont find a project of my
environmental interest inside the Department and still make it a chemical engineering
project.

I thank my committee members, Drs. Joe Perez, Wallis Lloyd, Brian Dempsey, Themis
Matsoukas and Tom Hettmansperger. They have not been just committee members, but
they have been closely involved in pursuing the project itself. They have offered me
suggestions, guided me on the right path when needed, and have taken special interest in
the project itself. Drs. Perez and Lloyd helped me set up experimental equipments and
provided me technical resources. Dr. Dempsey has always been there for me if I needed
suggestions whether it was on the Ph.D. project or on my algae-oil comprehensive
project. He has always asked the right questions to help me move forward, and was
always willing to spend any amount of time needed. Dr. Matsoukas has been a favourite
professor of mine for a long time, and has always been nice to me in offering suggestions
and inputs for the project. Dr. Hettmansperger, the outside member of the committee,
showed keen interest in the project be it the algae-oil comprehensive project or the main
biodiesel project, more than just supervising the committee procedures. He was always
interested to spend time on discussing and becoming familiar with the project.

I have to acknowledge my friend, Shaun Pardi. Shaun was the one who introduced me to
biodiesel. Some time in 2003, Shaun talked to me about biodiesel casually as it was a
passion of his, and that impressed me straight away. When I started working on the
project in fall 2004, Shaun spent a lot of time discussing about the project and giving me

guidance on experimental and analytical procedures (I did not have any experience on
conducting experiments then). He has inspired me throughout the research with his
passion for biodiesel.

I thank the Penn State Institutes of Energy and Environment for supporting me with my
tuition fee during my dissertation years. I extend my sincere acknowledgements to them
for their support.

My warm acknowledgements go to my friends at the Center for the Study of Polymeric


systems, Adam Jones, Marc Russel, Ida Balashova, Lourdes Serna and Roman Galdamez,
who were part of the research group of Drs. Duda and Danner. They have been of great
help to me in setting up my gas chromatograph, building my batch and packed-bed
reactors, and guiding me with lab supplies.

I have warm regards to my friend, Kevin Gombotz. Kevin has been in-charge of the
commercialization aspect of the project through Matson Biofuels, Inc. We have worked
closely in attempting to make this project a reality in the outside world. I am full of
appreciation of his efforts in making it happen.

I extend my sincere acknowledgements to Magda Salama of Penn States Material


Research Institute. She took lot of initiatives in conducting the CO2 desorption analysis
of the catalysts. After many initial hurdles, she figured out ways to conduct the analysis

xi

since the existing set up could do only ammonia desorption analysis of acids. She had
spent lot of time in making the procedure work.

I take this opportunity to recognize the efforts my parents have put into my studies since
my childhood. They have made lot of sacrifices to enable me pursue my academic
studies. I would like all my academic and professional credits go to them. It is only fair
that all my work would go under the name Kannan, which is actually my fathers name as
there is no concept of last name in my native place.

I would like to acknowledge my close friend Reenu, who has always been there for me,
has been very understanding of me, and has been a great source of encouragement and
confidence. Balasubramanian is another close friend of mine whose support I value. I
would also like to acknowledge my friend, Vamsi Salaka, who has always been kind to
me, and cared for me throughout my years in US.

Sachin Tendulkar, my favourite cricket player, and Kimi Raikkonen, my favourite racing
driver, have been a great source of inspiration to me. I would like to acknowledge the
way they have positively influenced me in real life beyond sports. They have a role in
my research efforts.

xii

Chapter 1

Introduction

The idea of fuel from a biosource is fascinating, and a sustainable one at that is
compelling. Biodiesel fits the bill. Biodiesel can be produced from high-yield oil crops
from dispensable land and used oil. Its a clean renewable fuel and locally produced.
Biodiesel has a high energy output ratio. It has considerable production capacity and
supply demand. Exploring biodiesel production methods is thus an attractive and worthy
proposition.

Biodiesel is a mixture of fatty acid alkyl esters produced from biosource. Normally this
mixture of methyl or ethyl esters is produced from a transesterification reaction involving
triglyceride esters (vegetable oil) and alcohol (methanol or ethanol) with glycerol as the
by-product [1]. The produced methyl and ethyl esters are similar to petroleum diesel in
structure and properties as fuel which makes it suitable for direct use in present day diesel
engines [1].

There are various reasons to consider biodiesel seriously. One is its importance in terms
of energy. The worlds fossil fuel reserves are rapidly being depleted. This means that
we have switch over to another viable option or combination of options soon. Biodiesel
is a promising option, and could be a part of the solution. Another reason is the
environmental concern. Fossil fuel carbon dioxide emissions pose global warming
problems, a real threat on a planet like earth, where the life system and ecology are

delicately tied to the make-up of atmospheric gases. Carbon dioxide is one of the main
greenhouse gases. Burning fossil fuels brings out the carbon buried underneath the
earths surface and cause carbon dioxide to accumulate. The carbon dioxide emitted
from burning biodiesel comes from the carbon fixed from the carbon dioxide in the
atmosphere by plants through photosynthesis, so biodiesel is carbon-neutral. The push
for renewable energy is greater than ever. Biodiesel assumes significance in this sense.

Depending on foreign resources for energy jeopardizes a nations economic


independence and security. Biodiesel could help agriculture flourish with crops like
jatropha that grow on arid land, and could even create new second generation
technologies such as algae oil. This may mean new opportunities, businesses and better
economy. Setting progressive production goals would establish and refine infrastructure,
and inspire business to better technologies. Local economy, particularly, has a potential
to benefit.

Among the many renewable alternative energy options available currently, biodiesel
could enable a smooth transition to a new fuel since the diesel engines of the present day
need no modification to burn biodiesel. In fact, when Rudolf Diesel invented the diesel
engine he designed it to burn peanut oil. With the USDA reporting that biodiesel cost are
coming as close as about 50~70 cents [2, 3] to gasoline prices, and with the gasoline
prices ever increasing, the time to embrace biodiesel may be very near. Add to this the
fact that diesel gives about 50 % more mileage than gasoline. Large scale production of
biodiesel could be the next step. Many more diesel-powered cars could be seen in the

future, and the diesel vehicles are not being considered polluting anymore. Also, used
vegetable oil could be used, which means value from wastes.

The method of biodiesel production though needs some rethinking. For such a green
product, the current process of production is rather polluting and cumbersome. Biodiesel
is currently produced using soluble alkali catalysts [1]. This leads to separation hassles
after the reaction. Dissolved catalyst and soap in the products require repeated waterwashing, generating about 8 gallons of waste water per gallon of biodiesel produced [4].
Separation issues of the conventional process may pose a problem for the large scale
commercial production of biodiesel. This led to contemplate on alternative methods of
production in the hope of a simpler and more efficient process. Solid catalysts are the
preferred route for successful large scale production of products requiring catalysis.
Development of solid catalyst processes for biodiesel is an emerging technique and there
is lot of scope for work in this area. A product holding such a huge potential warrants
extensive work, and that paved the road for this dissertation.

Hypothesis:
The basic hypothesis of the research is: A solid catalyst would be better than the
homogeneous catalysts for the production of biodiesel. A solid catalyst would avoid the
separation problems, and eliminate the various additional process units currently
required. Even if high temperatures are required, heat energy can be recovered since it
would be a continuous process, and the reaction times would be short. Efficiency of
biodiesel conversion would be better, because (i) feedstock would not be wasted as soap,

(ii) catalyst would not be wasted, (iii) the glycerol by-product would be pure and yield
good value, and (iv) wastage of water resource and pollution would be avoided.

Chapter 2

Background

When Rudolf Diesel originally invented the diesel engine, he tested it with vegetable oil
[5]. Then cheap petroleum became available, and diesel fuel fractionated from it was
used in the diesel engine. The diesel engine has evolved over the years.

Vegetable oil composition:


Oils and fats from natural biological sources are normally various mixtures of fatty acid
glycerol esters [6]. The fatty acid components vary in carbon chain length and number of
double bonds. Examples of these natural oils and fats are castor oil, soybean oil,
sunflower oil, rapeseed oil, jatropha oil, corn oil, palm oil, coconut oil, beef tallow,

Fatty acid %*

Soybean

Palm

Lard

Tallow

Coconut

Lauric (12:0)

0.1

0.1

0.1

0.1

46.5

Myristic (14:0)

0.1

1.0

1.4

2.8

19.2

Palmitic (16:0)

10.2

42.8

23.6

23.3

9.8

Stearic (18:0)

3.7

4.5

14.2

19.4

3.0

Oleic (18:1)

22.8

40.5

44.2

42.4

46.9

Linoleic (18:2)

53.7

10.1

10.7

2.9

2.2

Linolenic (18:3)

8.6

0.2

0.4

0.9

0.0

*Number of carbons and double bonds in parantheses


Table 2.1. Fatty acid composition of common oils [7]

chicken fat and algae oil. Fatty acid compositions vary with different sources. See Table
2.1. Apart from the glycerol esters, these natural oils and fats can also contain free fatty
acids (FFA) as such (not being glycerol esters), and their content is low (but that can also
vary with different sources).

Problems of direct vegetable oil usage in diesel engines:


As mentioned earlier, the diesel engine has evolved over the years for petroleum diesel
combustion. Using vegetable oil directly causes problems with the present diesel
engines. The ideal diesel fuel molecules are non-branched hydrocarbon molecules with
carbon numbers ranging from 12 to 18 whereas vegetable oil molecules are triglycerides
with carbon numbers that are three times greater [8]. The viscosity of vegetable oil is
several times higher than that of diesel oil. It has been reported that the high viscosity of
vegetable oil, 35~45 centistokes, as opposed to 4.0 centistokes for diesel oil at 40C,
leads to problems in pumping and atomization in the injection system of a diesel engine
[8]. It also leads to high carbon residue responsible for heavy smoke emissions from the
engine. The low volatility of vegetable oil (because of high molecular weight) makes it
stick to the injector or cylinder walls [9]. It then polymerizes forming a depositional film
that continues to trap fuel and interfere with combustion. That leads to more deposit
formation, carbonization of injector tips, ring sticking, and lubricating oil dilution and
degradation. The accompanying inefficient mixing with air contributes to incomplete
combustion. The combination of high viscosity and low volatility of vegetable oils also
causes poor cold engine start-up, misfire, and ignition delay.

Methods of modifying vegetable oil for diesel engines:


There have been four primary methods to lower vegetable oils viscosity and derivatize it
to be compatible with the modern diesel engines [1, 8]:
1. Direct use and blending: This involves blending vegetable oil with diesel fuel in
varying degrees. This is generally considered not satisfactory for use in diesel
engines.
2. Thermal cracking (pyrolysis): Vegetable oil, fats and soap can be processed
similar to thermal cracking of petroleum oil. This is not considered to be
efficient.
3. Micro-emulsion: This involves forming micro-emulsions of vegetable oil with
solvents such as methanol, ethanol and1-butanol. It addresses the high viscosity
problem to some extent and has been reported to meet the requirements of No. 2
diesel sometimes, but not always.
4. Transesterification: This is the preferred method to break down vegetable oil
efficiently to biodiesel for use in diesel engines.

Fuel consumption statistics:


Transportation amounts to 29 % of the total energy consumption of US [10].
Annual US transportation gasoline consumption is 142 billion gallons [11].
Annual US transportation diesel consumption is 53 billion gallons [12].
Annual world transportation gasoline consumption is 324 billion gallons [11].
Annual world transportation diesel consumption is 181 billion gallons [13].
The statistics listed are based on the year 2008 and are representative of the recent trend.

Biodiesel can also be used as heating oil (normally referred to as No. 2 fuel oil in US) in
residential and commercial buildings. Annual US heating oil consumption is 17 billion
gallons [12]. Combining the transportation diesel consumption and the heating oil
consumption in US makes a total of 70 billion gallons per year.

Biodiesel can either be used directly or as blends with petroleum diesel, like B5 and B20.
B5, for example, means that biodiesel is blended with petroleum diesel at 5 %. Biodiesel
is also exported to Europe from US.

Million Gallons per Year .

2000

1600
US Production
EU Production
1200

800

400

0
2002

2003

2004

2005

2006

2007

Figure 2.1. Biodiesel production of US and European Union [14, 15]

Figure 2.1 shows the biodiesel production of US and European Union over the past
several years. Even though US is larger than Europe in geographical area, Europe has

produced more biodiesel because of government initiatives and wider usage of the more
efficient diesel cars.

Oil yields:
The yields of various oil sources are given in Table 2.2. Crops have varying degrees of
yield. Palm trees cannot be grown at all geographic locations where soybeans can be
grown. Soybean has lower yield compared to rapeseed, but its fertilizer requirements are
considerably less than that of rapeseed since it can fix nitrogen. Jatropha produces
inedible oil, and can be grown in arid land that may not be suitable for other crops and
would require minimal irrigation requirements. The Caribbean, Africa, India, Pakistan
and the Philippines are some of the best places to grow jatropha economically.

Oil

Yield (gallons/acre/year)

Corn

15

Soybean

48

Sunflower

102

Rapeseed

127

Jatropha

202

Palm

635

Micro-algae (actual)

1850

(theoretical)

15000

Table 2.2. Oil sources and yields [16, 17]

Biodiesel energy balance:


The energy output ratio of biodiesel has been estimated to be 3.2:1 [18]. This is the
amount of fuel energy per unit of energy used to produce it. The energy output ratio of
biodiesel is high compared to the energy balance of bioethanol at 1.05:1 to 1.25:1 [19,
20].

The scope for biodiesel production:


Food vs fuel is an important and delicate issue to bear in mind in the attempt to pursue
biofuels. Apart from food considerations, the creation of biofuels can defeat the original
purpose if it leads to deforestation to farm for biomass. Also it is unlikely that sources
like soybean and rapeseed can supply for the current transportation needs of the entire
world. Even planting the entire geographical area of US with soybean crops would only
equate to 2 % of its transportation fuel needs. Crops for biofuels have to be grown to a
potential for what it is. In some locations of a country, such crops can add value to
agriculture and the local economy. Some countries can afford to grow crops for biofuels
to a certain extent without compromising food supply. It has to be remembered that a
farmer is also interested in earning money just like the oil companies and the food
enterprises. Plants like jatropha can be grown in arid lands which are not suitable for
other vegetation, and definitely add considerable value as fuel. Its yield is among the
highest. India has recently decided to use biodiesel from jatropha oil in its trains [21].

10

Considering the aforementioned points, annual global biodiesel production potential has
been estimated to be about 13 billon gallons [22]. Annual US biodiesel production
potential has been estimated to be 850 million gallons [22].

Used cooking oil could be used as well to produce biodiesel. In US the quantity of used
oil from restaurants and chicken fat and other animal tallow from the meat industry
amounts to about 4 billion gallons per year [23, 24]. These are cheap feedstock that can
be recycled to produce biodiesel, meaning fuel from waste. The cost of raw materials
(oils and fats) amounts to 75 % of biodiesel cost [9], so these cheap feedstock are
important. The 4 billion gallons of waste oil generated per year in US, if converted to
biodiesel, could amount to about 8 % of diesel fuel used for transportation in US.
Overall, biodiesel can never replace the total transportation fuel consumption, but it still
has considerable potential to be converted to fuel effectively in several circumstances as
mentioned earlier.

Micro-algae promises yields up to 1000 times higher, through second-generation


technologies [16]. Recently algae have been researched to produce oil. The US
Department of Energy originally investigated algae for fuel in the 1970s through its
Aquatic Species Program. Research efforts have begun again recently incorporating new
approaches. Algae-oil technology is one of the few alternative energy technologies that
qualify by available resources to supply for entire transportation needs theoretically.
Even conservative estimates indicate that the oil needs of the entire US could be supplied
by an area as small as a part of the Sonora desert in the southwestern US (15,000 square

11

miles, 13 % of the total US geographical area) [25, 26]. The attractions to this method
are the considerable reduction in area and other resources such as water and nutrients
needed for oil cultivation, and the considerable reduction in the by-products associated
with oil production. Oil content in such algae amounts to 50~90 % of their dry weight
(eg.: Botryococcus braunii: 86 %, Nannochloropsis sp.: 68 %, Schizochytrium sp.: 77 %,
Neochloris oleoabundans: 54 %) [27, 28]. It is algal species such as these that were
responsible for the present day petroleum deposits formed millions of years ago. So, it is
only logical that algae are considered to be a prominent energy source for the future.

The current US and global biodiesel production potential numbers mentioned earlier is
still huge. The recyclable oil and animal fat number is also considerable. Algae-oil is
also a promising technology that theoretically qualifies to supply for the energy demands
of the entire world, albeit as a second-generation technology in the future. So it is
imperative to develop an efficient process for large scale commercial production of
biodiesel.

Biodiesel and diesel Performance, handling and usage comparisons:


Biodiesel has poor cold-weather properties compared to diesel. It has a higher cloud
point than diesel (soy biodiesel: 2 C; No. 2 diesel: -23 C) and starts to gel at a higher
temperature than diesel in cold weather [14]. This problem can be resolved by blending
with diesel or kerosene. B20 biodiesel-diesel blend has the same cloud point as diesel [8,
14]. The cloud point is higher for biodiesel made from chicken fat as it contains longer
fatty acid chains. If biodiesel were to become the norm in future, its not difficult to

12

imagine that this problem would be solved with an anti-gelling agent, or a radiator heattransfer with a petroleum diesel start or a battery-heated biodiesel start.

Biodiesel improves lubricity. Even low-level blends of biodiesel such as 1% or 2% can


improve lubricity of diesel fuels, and this may be particularly important for ultra-lowsulfur diesel, as these fuels can have poor lubricating properties. Biodiesel has 8 % less
energy content than petroleum diesel [29]. B100 may soften and degrade some hoses and
gaskets that are made of buna N, nitrile or natural rubber, but some systems already have
materials like VitonTM that are resistant to this degradation [29]. Biodiesel has a better
cetane number (48-65) than petroleum diesel (40-55), meaning better engine performance
(better combustion quality during compression ignition and better cold start properties).
See Appendix G for explanation of cetane number.

Comparing the tail-pipe emissions of biodiesel and petroleum diesel, biodiesel has less
particulate matter emissions (50 %), less carbon monoxide emissions (50 %) and less
unburned or partially burned hydrocarbons (70 %) than petroleum diesel [29, 30]. But
nitrogen oxide emissions of biodiesel are higher (15 %) than that of petroleum diesel due
to high in-cylinder combustion temperatures [29, 30]. But modern catalytic converters
that reduce nitrogen oxides back to nitrogen and oxygen can minimize the problem in the
future.

13

Conventional Process:
Basic reaction:
R-COOCH2
|
catalyst
R-COOCH + 3CH3OH
3R-COOCH3
|
R-COOCH2
triglyceride
(vegetable oil)

methanol

methyl esters
(biodiesel)

HOCH2
|
HOCH
|
HOCH2
glycerol

where R denotes a hydrocarbon chain of carbon number 12~18.


The reaction associated with biodiesel production is a transesterification reaction [1], and
has been known for a long time (about 100 years). This is a reaction that converts one
form of ester to another. Vegetable oils are triglyceride esters (esters of 3 molecules of
fatty acids with one molecule of glycerol). They react with monohydroxy alcohols like
methanol and ethanol, producing corresponding esters. Glycerol is the by-product.

The triglyceride is converted stepwise to diglyceride, monoglyceride and finally to


glycerol as illustrated below [31, 32]. A mole of ester is liberated in each of the three
steps as follows.
Step1:
R-COOCH2
|
R-COOCH + CH3OH
|
R-COOCH2

R-COOCH3

HOCH2
|
R-COOCH
|
R-COOCH2

HOCH2
|
HOCH
|
R-COOCH2

Step 2:
HOCH2
|
R-COOCH + CH3OH
|
R-COOCH2

R-COOCH3

14

Step 3:
HOCH2
|
HOCH + CH3OH
|
R-COOCH2

R-COOCH3

HOCH2
|
HOCH
|
HOCH2

The key to this reaction is the production of alkoxide ions (methoxide (CH3O-),
ethoxide(C2H5O-), etc.). The conventional process involves production of these alkoxide
ions using soluble (homogeneous) base catalysts like sodium hydroxide (NaOH) and
potassium hydroxide (KOH). While using such base catalysts, the reaction mechanism is
a nucleophilic substitution reaction, as illustrated below.
CH3OH + NaOH
CH3ONa

CH3ONa + H2O
CH3O-- + Na+

Figure 2.2. Homogeneous base catalyzed reaction mechanism of transesterification

15

Methanol reacts with the base catalyst, sodium hydroxide, to form sodium methoxide [1].
The negatively charged methoxide ion from sodium methoxide attacks the carbonyl
carbon atom, which is slightly positively charged due to polarization of double bond
electrons toward the more electronegative oxygen atom, and forms a tetrahedral structure.
This tetrahedral structure is unstable, and the carbon atom would much rather revert back
to a carbonyl double bond with the oxygen atom, and give up the electrons to one of the
oxygen atoms of two alkoxides (methoxide and glycoxide (of diglyceride)). The electrons
go back to the diglyceride glycoxide, as methoxide is a much stronger nucleophile. This
results in the formation of a methyl ester biodiesel molecule.

In the conventional reaction, the temperature is 60~70 C, and it takes about an hour and
half for the reaction to proceed to completion [1]. The alcohol and the alkali metal
hydroxide catalyst are first mixed before the reaction in a separate unit to form the metal
alkoxide. The alcohol and the alkoxide catalyst are then mixed with oil in the reactor.
The reactor is a batch reactor with stirrer, or a continuous or semi-continuous stirred tank
reactor. A condenser is used to condense the alcohol continuously. The boiling point of
the alcohol is in the range of the reaction temperature. Oil and alcohol are immiscible;
they are constantly stirred, and the reaction takes place at the interface of two phases. The
reaction is described by second order reaction kinetics [32]. The stoichiometric ratio of
triglyceride and alcohol for the reaction is 3:1, but 6:1 molar ratio is the optimal molar
ratio used to push equilibrium to one side for maximum biodiesel conversion (97~98 %)
[32]. 0.5 wt % of catalyst is considered optimal for maximum activity.

16

Both methanol and ethanol are considered to have comparable reactivities. But methanol
is considered to be a slightly better reactant that ethanol because (i) methanol is more
acidic than ethanol in giving up the proton to form the alkoxide nucleophile, because of
its shorter chain length; and (ii) ethanol can have stearic effect due to its bigger size.
Biodiesel can also be made using branched-chain alcohols like isopropyl and 2-butyl
alcohol, which reduces the crystallization temperature of biodiesel [33].

Used cooking oil, an important source for biodiesel, has significant free fatty acid (FFA)
content, up to 15 % by weight. Virgin vegetable oils also contain FFAs, up to 5 %, in
some cases. The conventional process needs a separate soluble acid catalyzed step
preceding the soluble base catalyzed unit to process such feedstock. Otherwise, the
significant free acid content would be wasted as soap, presenting a difficult separation of
the biodiesel from the product mixture [1]. Sulphuric acid is the most commonly used
acid catalyst, although hydrochloric acid is also used [34]. This acid catalyzed process
requires higher molar ratios, 15:1 to 30:1 [1, 34]. Transesterification of triglycerides
takes ~ 69 hours. At 180 kPa and 50:1 molar ratio, it takes 4 hours [35]. But the
esterification of FFA is fast, taking only 1 hour [36]. The reaction mechanism is different
to that of the base catalyzed reaction, and this enables esterification of FFA as well as
transesterification of triglycerides. The key step of the reaction is the protonation of the
carbonyl oxygen, which makes the carbonyl carbon more electrophilic. This carbon
attracts the alcohol directly as a result, instead of requiring a stronger nucleophile like
methoxide as seen in the base catalyzed mechanism. The reaction mechanism is
illustrated in Figure 2.3.

17

Figure 2.3. Homogeneous acid catalyzed reaction mechanism of transesterification

The three conversion steps indicated in Figure 2.3 are explained as follows: (1)
protonation of the carbonyl group by the acid catalyst; (2) nucleophilic attack of the
alcohol, forming a tetrahedral intermediate; (3) proton migration and breakdown of the
intermediate. The sequence is repeated twice [34].

Drawbacks of the conventional process:


The main drawback with the conventional process is the use of soluble catalysts (NaOH,
KOH) that end up contaminating the biodiesel and glycerol by-product. In the case of
sodium hydroxide, the sodium methoxide produced is dissolved in the final product
mixture, mostly in the glycerol phase and partly in the biodiesel phase. Also, a small
fraction of the triglycerides are always wasted as soap due to the water produced by

18

sodium methoxide reaction. See Appendix C. Presence of water (as in used oil and even
virgin oil sometimes) results in soap formation. It is not easy to separate the sodium
methoxide and the soap from the product mixture. They have to be separated by repeated
water washing steps. The catalyst has to be wasted ultimately, as regeneration is very
expensive and fresh catalyst has to be constantly used. Add to this the waste disposal
problem of the catalyst. For every gallon of biodiesel produced, about 8 gallons of waste
water is generated [4]. The generated waste water is usually dumped into the municipal
drains in US. Biodiesel producers mention that water is one of the most expensive
resources in their plants. Water remaining in the biodiesel is then usually removed by
additional heating. The conventional process ends up having numerous additional
processing units before the produced biodiesel goes through the final polishing step.

Glycerol, the by-product of the transesterification reaction, is only about 80 % pure, as it


contaminated with catalyst, soap and water. It is either sold for a low value to glycerol
refineries or just disposed of. Contaminated glycerol requires vacuum distillation for
purification. Pure glycerol has good value for various applications (food, medicine,
pharmaceuticals, personal care, cosmetics and organic synthesis).

The separation issues of the conventional process are illustrated by catalyst and soap
contamination, numerous water washing steps, waste water generation and water
removal, disposal of the toxic alkali catalyst, the various additional processing units like
catalyst-alcohol pre-mixing unit and FFA pretreatment, and contaminated glycerol
purification. This makes the conventional process cumbersome and tedious. A simple,

19

neat, efficient and robust process is needed for large scale commercial production of
biodiesel.

Solid catalysts:
A solid catalyst process eliminates the separation issues associated with the conventional
process. A solid catalyst would not dissolve in the reactant mixture, so it would not need
to be separated from the product stream.

For the transesterification reaction, base catalysts are always preferred to acid catalysts
wherever possible because of their higher activity. While looking for solid catalysts for
this reaction, the research and development of solid bases in general is lot less compared
to that of solid acids [37, 38].

Solid bases:
Solid bases catalyze reactions by donating electrons. Certain reactions can be used as test
reactions to evaluate the basicity of solid catalysts. Isomerization of alkenes like 2,3dimethylbut-1-ene and 1-butene and decomposition of 2-methyl-3-butyn-2-ol are
examples of test reactions [39, 38]. Solid bases have been tested for various organic
reactions to evaluate their basicity [37, 38]. Reactions include double bond
isomerization, hydrogenation, amination, aldol addition, Michael addition, methylation
and Knoevenagel condensation. Some of the effective solid bases listed are [37, 38, 39]:
1. Alkali earth metal oxides
CaO, MgO

20

2. Rare earth oxides


3. Alkali ion-exchanged/added zeolites
RbX, CsX CsNaX, CsNaY/Cs-loaded CsX, Cs-loaded ZSM-5, Europiumloaded KY
4. Alkali metals on supports like alumina and silica
5. Clay minerals
Hydrotalcite, Chrystotile, Sepiolite
6. Alkali metal compounds supported on alumina
KNO3/Al2O3, K2CO3/Al2O3, KOH/Al2O3, CsOH/Al2O3, KF/Al2O3
7. Lanthanide imide and nitride on zeolite
8. Metals supported from liquid NH3
KNH2/Al2O3, RbNH2/Al2O3, K(NH3)/Al2O3, Na(NH3)/Al2O3, Eu(NH3)/Al2O3,
Yb(NH3)/Al2O3, Eu(NH3)/KY
9. Oxynitrides
Silicon oxynitrides, Aluminophosphate oxynitrides

The surfaces of materials listed above are covered with carbon dioxide, water and
oxygen. Hattori illustrated the generation of basic sites in magnesium oxide from a
magnesium hydroxide precursor [38]. While evacuating the hydroxide at about 700 K,
water and carbon dioxide begin to evolve. As the pretreatment temperature is increased,
the molecules covering the surfaces are successively desorbed according to the strength
of the interaction with the surface sites. The basic sites that appear at higher temperatures
are stronger. Rearrangement of surface and bulk atoms also occurs during pretreatment,

21

which may change the number and nature of the surface basic sites. Mg-O ion pairs of
lower co-ordination numbers exist at corners and edges and the ones at multiple-fold
corners are even stronger [40]. Highest-temperature pretreatment is required to reveal
such ion pairs, and they tend to rearrange easily as well, resulting in optimal pretreatment
temperature range. The various changes together result in different pretreatment
temperatures showing maximum activity for different reactions. This also means that, in
general, even without pretreatment considerations, basic sites of different strength are
required for different reactions, as in different catalysts with different basic strength.
This was evident in the way the catalysts listed above varied in their activity for different
reactions. For example, K2CO3/Al2O3 was the best catalyst for the decomposition of 2methyl-3-butyn-2-ol, but it was much less active than KNO3/Al2O3 for the isomerization
of 2,3-dimethylbut-1-ene (4 % vs 49 % conversion in 30 minutes) [39].

Solid acids:
Solid acids catalyze reaction either by donating protons or abstracting electrons. The
various solid acids are [41]:
1. Zeolites
ZSM-5 (See Appendix F), Mordenite, Y-Zeolite, US-Y
2. Oxides, phosphates
SiO2-Al2O3, Al2O3-BF3, SO42-(Fe, Mn)/ZrO2, SrHPO4, FePO4, Li3PO4,
phosphoric acid, SAPO-11 (silico-alumino-phosphate), SAPO-34
3. Ion-exchange resins
Amberlyst, Nafion

22

4. Clays
Kaolinite, Montmorillonite, pillared clays

Solid acids have been widely used in petroleum industry for applications like cracking
and reforming, and have been studied extensively. Aromatization, isomerization (xylene,
n-hexane), synthesis of ethers like MTBE and TAME, Friedel-Crafts acylation of
aromatics, hydrogenolysis for propane production and alkylation of benzene are various
other reactions where solid acids are used.

Solid acids are so well-developed that numerous solid acids are even stronger than
sulphuric acid. They are called solid superacids. A superacid is defined as an acid that
exhibits an acid strength higher than the acid strength of 100 % sulphuric acid (H0 = 11.93) [42]. The acidity of highly acidic media can be determined by using appropriate
Hammett indicators. The Hammett acidity function, H0, has been extensively used as a
measure of acidity [43]. The acid strength of a solid is defined as the ability of the
surface to convert an adsorbed neutral base into its conjugate acid [42].
H0 = pKa + log [B]/[BH+]
where [B] and [BH+] are the concentrations of the neutral base (Hammett indicator) and
its conjugate acid, and pKa is pKBH+. Examples of solid superacids are Ti(SO4)2/ZrO2 (H0
= -14.5), SO42-/ZrO2 (H0 < -16) and SO4 2/TiO2-La2O3 (-16.04< H0 < -14.52) [43, 44,
45]. Examples of Hammett indicators are m-nitrochlorobenzene (pKa = 13.16), 2,4dinitrofluorobenzene (pKa = -14.52), and 1,3,5-trinitrobenzene (pKa = -16.04) [32c].

23

Solid catalysts for biodiesel reaction:


Solid catalysts have been tested for heterogeneous catalysis of the transesterification
reaction. Peterson and Scarrah tested bases like MgO, Al2O3, CaO-Al2O3, K2CO3,
Na2CO3, activated carbon, Fe2O3, CH3ONa supported on SiO2, NaAlO2, zinc, tin and
lead; and acids like ZnO-SiO2 and Dowex 2-X8 anion exchange resin in the hydroxide
form [47]. Compared to the conventional homogeneous sodium methoxide catalyzed
reaction, only K2CO3 gave reasonable conversion, and that was found to have dissolved
into the reactant mixture. Sodium methoxide loaded onto silica also dissolved into the
solution. Other catalysts showed poor or no conversion. ZnO also dissolved into the
solution. The Dowex anion exchange resin was tested at 200 C (1000 psi/68 atm) for 12
hours and at 91 C (135 psi/9.2 atm); biodiesel yield was poor and the products were
mostly thermally cracked.

Leclercq et al. tested solid bases like cesium-exchanged sodium zeolites (NaCsX),
hydrotalcites (Mg6Al2(CO3)(OH)164(H2O)), Barium hydroxide (Ba(OH)2) and MgO [48].
The best result of the zeolites tested took 22 hours for 99 % biodiesel conversion.
Hydrotalcites and MgO took 22 hours to show 67 % and 69 % biodiesel conversion.
Ba(OH)2, obtained from dehydration of the octahydrate form by calcination overnight at
473 K under reduced pressure, showed 79 % biodiesel conversion in 1 hour. They
concluded that strong basic properties are required to perform this reaction and that
cation-exchanged zeolites are not really appropriate. Alkali earth metal oxides (CaO,
BaO) and alkali earth metal hydroxides (Ca(OH)2, Ba(OH)2) are not completely insoluble
in water and alcohol; they have some solubility. They have potential as solid catalysts if

24

they are attached to supports preventing dissolution. The oxides are always hydrated and
have to be heated to high temperatures (1000 C) for complete dehydration. The
hydroxides are known to be soluble in polar solvents. Ba(OH)2 is reported to likely have
homogeneous mechanism [49]. It is known to have useful solubility in water and
alcohols [50]. The data of Mazzochia et al confirm that Ba(OH)2 is not a completely
heterogeneous catalyst [51]. When the product obtained after reaction is not washed
several times with distilled water, the resulting biodiesel and glycerine contain 0.06 %
and 0.25 % of barium, respectively.

Suppes et al tested zeolites like NaX, NaY faujasites, potassium-exchanged NaX (KX),
cesium-exchanged NaX (CsX), ETS-10 zeolite (microporous titanosilicate), K-ETS10,
Cs-ETS10 and NaX occluded with excess sodium (1-4 NaO/NaX), palladium, nickel,
cast-iron, stainless steel, zinc oxide and zinc carbonate [49]. At 60 C, after 24 hours,
ETS-10 showed 80 % biodiesel conversion, 3 NaO/NaX showed 82 % and KX showed
10 %. At 120 C, after 24 hours, their conversions were 95 %, 93 % and 23 %. At
150C, after 24 hours, their conversions were 96 %, 94 % and 92 %. Calculations based
on the rate constant of ETS-10 showed 80 % biodiesel conversion after 1 hour at 120 C.
At 120 C, zinc oxide, nickel, palladium, zinc carbonate, stainless steel and no catalyst
corresponded to 80 % conversion after ~ 15, 20, 30, 100, 280, 330, 400 and 10,000 hours.
The base form of ETS-10 showed alkali methoxide leaching resulting in homogeneous
catalysis [52]. Zinc oxide and zinc carbonate were believed to have a significant
homogeneous component to their activity, like Ba(OH)2. FFA content higher than 25 %
quenched the catalyst activity, indicating a reaction mechanism similar to that observed

25

with soluble alkali catalyst. Their data indicate that FFA can inhibit solid catalysts
relying on highly basic sites.

Gryglewicz tested CaO, MgO, Ca(OH)2, Ba(OH)2 and calcium methoxide (Ca(CH3O)2)
for biodiesel reaction [53]. The reaction was carried out at the boiling point of methanol
(65 C) with constant reflux condensation. Though Ba(OH)2 showed good activity, it
dissolved into methanol. It was reported that CaO and Ca(OH)2 react with methanol
forming calcium methoxide on the surface. These methoxide ions catalyze the reaction
similar to methoxide ions formed by NaOH, but they have low activity. Ca(OH)2 was
reported not to have catalyzed the reaction. MgO was found not to be an active catalyst.
CaO was found to be 3~4 times slower than NaOH, giving 80 % biodiesel conversion in
1.5 hours and 90 % conversion in 2.5 hours. Solubilities of alkali earth metal catalysts in
methanol were listed as: MgO (0.13 %), Ca(OH)2 (0.01 %), CaO (0.035 %), Ca(CH3O)2
(0.04 %), Ba(OH)2 (1.17 %). Mechanism of active species formation suggests that the
catalysts have a tendency to dissolve. The expensive pure calcium methoxide was
reported to be strongly basic (80 % conversion in 1 hour), forming a colloidal suspension
in methanol. Colloidal suspension, though not a solution, may still pose separation
problems.

Gryglewicz published some interesting results on ultrasonic-aided reaction [53].


Ultrasonication only helped the reactions that involved soluble and suspended catalysts
(NaOH, Ba(OH)2 and Ca(CH3O)2), and it proved to be detrimental for heterogeneous
catalysis (CaO) by preventing the adsorption of reactants on the catalyst surface.

26

Gryglewicz used tetrahydrofuran as a cosolvent (10 %) with CaO catalyst and achieved
80 % biodiesel conversion in ~1 hour and 95 % conversion in 2 hours. It has to be
remembered that vegetable oil and alcohol are already an immiscible two-phase system,
and introducing a solid catalyst would make it a three-phase system. A cosolvent can
help address the reaction process weakened by a combination of mass-transfer limitation
and weaker activity of a heterogeneous catalyst. But alcohol and biodiesel cannot be
separated after the reaction by simple phase-separated gravity settling anymore, the
whole reaction mixture has to be heated to the boiling point of cosolvent.

Dehydrated pure CaO needs calcinations above 1000 C, otherwise hydrated water would
lead to saponification. CaO reacts with water and forms Ca(OH)2 which has a solubility
of 0.19 % with water. CaO loaded on a support may lose some of its activity due to its
physical bond with the support. It has been reported that even if the CaO can be reused
for several runs without significant deactivation, dissolution of CaO does occur [54]. The
catalytic reaction is considered to be the result of the contribution of both heterogeneous
and homogeneous catalysis for the formation of leached active species.

Schuchardt et al. tested alkylguanidines supported on polystyrenes [55]. They showed


good activity for the first run, but started to lose activity gradually for the subsequent runs
as the catalysts leached from the polymers. They leached out completely after 9 runs.

Immobilized lipases from microbial organisms have also been tested as catalysts for
biodiesel production [56, 57, 58]. Lipases like those from Candida rugosa, Pseudomonas

27

cepacia, Pseudomonas fluorescens, Thermomyces lanuginosa and Rhizomucor miehei


have been tested. Reaction usually takes a long time 6~24 hours. The ones that show
good conversions in about 6 hours tend to lose activity in 10 runs. They can be used only
in 30~50 C temperature range. Lipase catalysis is comparatively quite expensive and is
not considered to be viable for commercial production.

Kim et al. used Na/NaOH/-Al2O3, a solid superbase, for the tranesterification of


vegetable oil to biodiesel [59]. The catalyst was originally developed by Suzukamo et al
and has a Hammett basicity value (H_) greater than 37 [60]. This is the strongest
superbase known. Superbase, in concept, is similar to superacid explained earlier.
NaOH and -Al2O3 were first mixed together to form sodium aluminate (NaAlO2) at 320
C, stirred for 3 hours, then sodium metal was added and stirring was continued for
another hour. Sodium is considered to have been ionized and dispersed into the defect
sites of -Al2O3. The electrons donated by sodium leads to superbasic sites of increased
electron pair donating capability in surface oxygen [59]. The catalyst when employed for
biodiesel reaction, gave an equilibrium conversion of only 75 %. The equilibrium
conversion was reached in 1 hour and the conversion stayed there even after 8 hours. But
when they used n-hexane as a cosolvent in 5:1 vegetable oil:hexane molar ratio, and
adjusted the methanol:vegetable oil molar ratio from the conventional 6:1 to 9:1, at 1 wt
% catalyst, 94 % equilibrium conversion was reached in 2 hours. There was no report on
catalyst reusability.

28

In spite of previously mentioned research efforts, there has not been successful
commercial production of biodiesel using solid bases which reflects the drawbacks
related to activity, maintenance and reusability.

Solid acids like sulphuric acid impregnated aluminosilicates, sulphuric acid impregnated
with montmorillonite KSF aluminophosphate, hydrous tin oxide, Amberlyst-15,
Envirocat EPZG, natural kaolinite clay, B2O3/ZrO2, sulphated SnO2, zeolites, Nafion,
MCM-41 supported heteropolyacids, mesoporous silica modified with sulphonic group,
sulphated zirnoia (SO4/ ZrO2) and sulphated tin oxide (SO42-/ZrO2) have been tested for
biodiesel reaction [34]. Catalysts usually require high temperatures, around 180 C, for
better reaction rates (except the ones like Amberlyst-15 and Nafion which are not stable
at high temperatures). Tungstated zirconia-alumina, sulphated zirconia-alumina and
sulphated tin oxide were reported to take 20 hours to give conversions of 50 %, 25 % and
10 % at 200 C, 90 %, 55 % and 17 % at 250 C and 94 %, 77 % and 70 % at 300 C for
soybean transesterification [61]. For the esterification of n-octanoic acid (a
representation of FFA), tungstated zirconia-alumina, sulphated zirconia-alumina and
sulphated tin oxide gave conversions of 94 %, 99 % and 100 % at 175 C [61]. FFA
esterification is a faster reaction than triglyceride transesterification [62]. Furuta et al.
showed sulphated tin oxide (SO42-/ZrO2) gave close to complete conversion of n-octanoic
acid at 120 C after 20 hours and sulphated zirconia gave close to complete conversion of
n-octanoic acid at 160 C after 20 hours [63]. Lopez et al. tested titania zirconia (TiZ),
sulfated zirconia (SZ) and tungstated zirconia (WZ) on glyceryl trioctanoate (tricaprylin)
and oleic acid for biodiesel ester production with ethanol [62]. For oleic acid

29

esterification, it took 9 hours at 120 C to reach close to complete conversion, and for
tricaprylin, it took about 8 hours at 120 C to reach 23 % biodiesel conversion. Leaching
of sulphate groups has been a problem and Lopez et al. also report this. What used to be
a 61 % caprylic acid conversion after a 3-hour cycle decreased to 36 % after 5 cycles
with sulphated zirconia and it could not reactivated, whereas tungstated zirconia which
only showed 21 % conversion initially could be reactivated to its original activity after
the 5-cycle activity loss. Reusability and reactivation are important requirements for
commercial production. Acid catalysts are susceptible to dehydration etherification of
alcohol even at temperatures as low as 100 C and the catalysts tested by Furuta et al [63]
also showed good activity for methanol etherification. French Petroleum Industry (IFP)
and Axens, IFP Group Technologies developed a biodiesel process using zinc aluminate
solid acid catalyst at high temperature and high pressure that had the overall process time
estimated to be about 8 hours [64]. The catalyst gets deactivated by water and the water
content in the reaction medium must be less than 1000 ppm [65]. No data has been
reported on the lifetime of the catalyst [66]. Even the strongest solid acids take long
reaction times at high temperatures, and, deactivation and reactivation have also been a
problem as is the case normally with the specially prepared catalysts.

30

Chapter 3

Approach to Method Development

The approach taken that led to the method developed in this research, leading from the
problems identified in the conventional process, is narrated in this chapter. The thoughtprocess that led to the developed solid catalyst method and the hypotheses behind the
approach have been outlined.

The separation problems associated with the conventional process using homogeneous
catalysts have been explained. The scope for a solid catalyst process to eliminate those
problems and improve the production method has been mentioned. For the
transesterification reaction, base catalysts are always preferred to acid catalysts wherever
possible because of their higher activity. While looking for solid catalysts for this
reaction, the research and development in solid bases in general is lot less compared to
that of solid acids [37, 38].

In fact, normal solid bases are not effective for the conventional reaction conditions
(60~70 C, methanol/ethanol reflux condition), and a solid superbase has to be used [59].
With the solid superbase catalyst come the maintenance issues: water vapour and carbon
dioxide adsorb readily on the active sites on exposure to air; the catalyst loses activity
and requires frequent regeneration; the catalyst gets leached into the reaction mixture; the
catalyst is susceptible to poisoning by glycerol [37]. With the reaction being run in
continuous stirred tank reactors (CSTRs) or batch reactors, the exposure of the solid

31

superbase catalyst to air is inevitable. As such, the strong solid superbases lose their
activity in few runs and the frequent regeneration in addition to the complex preparation
steps involved make them high-maintenance products. The solid superbase still needs a
cosolvent to mix the immiscible reaction mixture. This means heating the whole reaction
mixture to above the boiling point of the cosolvent after the reaction to separate biodiesel.
These are the problems in attempting to use solid catalysts under the conventional
reaction conditions.

Saka and Kusdiana developed a method to produce biodiesel under supercritical


methanol conditions at a temperature of 350 C and 500 atmospheres without using any
catalyst [67, 68]. They used a 40:1 methanol:oil molar ratio. The reaction time was only
4 minutes, but the pressure is too high for industrial applications. They mention that they
need such high pressure to extract the full potential of supercritical regime. The critical
point of methanol is 240 C, 80 atm; and that of ethanol is 247 C, 67 atm. The critical
pressures of these alcohols themselves (65~80 atm) are quite low compared to the 500
atmospheres they employed. But they showed that the reaction rate constant drops down
considerably at lower pressures in their rate constant vs temperature Arrhenius plot.
They reason that the supercritical region may not be stable or strong enough until the
higher temperatures and pressures.

Saka and Kusdiana propose that the higher molar ratios (40:1) provide a better medium of
interaction needed for the reaction to take place at its full potential at the supercritical
region. They report that the rate constant increases with higher alcohol ratios until

32

saturating at 40:1. They also proposed the reaction follows pseudo first order kinetics
based on vegetable oil concentration and confirmed it with kinetic modeling. But thermal
breakdowns are considered an issue at such high temperatures.

Cao et al tried to reduce the pressure by using propane as a cosolvent [69]. The pressure
in their method was 130 atmospheres and the temperature was 280 C. The reaction time
was 10 minutes. The pressure was still high, twice the critical pressure of ethanol, and
had scope for improvement.

Critical region of the alcohol is desired for the miscibility of the alcohol and the oil
phases. Alcohol would act as a reactant as well as a cosolvent. In the present work, it
was desired to see if the reaction could be achieved to completion at critical or nearcritical conditions employing a solid catalyst. The reason is that critical pressure is 67
atmospheres while using ethanol, which is half the pressure of the Cao et als process. It
was desired to see if mild solid base catalysts with no high-maintenance issues could be
effective in this region with reasonably less reaction time (15~20 minutes). Though the
temperature might be high (240~260 C), it should not be a problem for a continuous
process. Very effective heat exchangers are available for industrial applications to make
high temperature continuous processes perfectly feasible. Moreover, a continuous reactor
of fixed-bed type is preferable over the batch reactors and the CSTRs that are normally
employed in the present conventional process. A continuous reactor is better than batch
reactor for large scale production. A CSTR operates at exit concentration, so the reaction

33

rate would be slower. So a CSTR can be seen as not an efficient reactor compared to a
fixed-bed reactor.

Suppes et al had tried calcium carbonate, a weak solid base, in a packed-bed reactor [70].
The reaction temperature of their process was 260 C. They reported that the alcohol was
maintained in liquid state by pressure, and this means a pressure around 70 atmospheres.
But they reported that they had to use 50 % finished product biodiesel as a cosolvent to
get yields greater than 98 %, otherwise the yield was poor. The residence time for their
reaction was 18 minutes, which effectively means a reaction time of 36 minutes,
considering the 50 % finished product being fed back to the reactor as cosolvent. The
work of Suppes et al is planned to be used as a standard to compare the results of the
present research. They did not mention any intention to exploit the critical properties of
the alcohol and the resulting miscibility though their reaction condition happens to be
near-critical.

Mild bases, unlike custom-made high-activity catalysts that have to be regenerated


frequently [37, 38], could make for a good catalyst in a packed-bed reactor. These mild
bases can be expected to be naturally occurring, and be robust and stable as a result.
They could prove to be useful if they provide that just enough improvement in activity at
the critical point of alcohol and sustain their activity for a long time.

To sum it up, solid catalysts are desired to overcome the separation issues associated with
the conventional process employing soluble catalysts. Solid acids are too weak to

34

catalyze the reaction. With solid bases, while attempting to use them under conventional
reaction conditions (methanol reflux), problems such as requirement of a solid superbase
instead of a normal solid base, frequent loss of activity and requirement of a cosolvent are
faced. This led to the consideration of a high temperature process, which if continuous
would facilitate heat recovery. A high temperature process would enable the use of solid
base catalysts that are milder and could sustain their activity much better. High
temperature processes require extremely high pressures if a catalyst is not used. It was
desired to see if solid base catalysts could be effective and bring down the pressure to an
industrially manageable range. If the milder solid bases could sustain their activity in
these reaction conditions and result in reasonably short reaction times, there is scope for
an efficient solid catalyst process for biodiesel production.

35

Chapter 4

Initial Tests and Catalyst-Screening

Though the intended design is ultimately a continuous process, the initial part of method
development, testing and screening the various catalysts, was conducted in batch
experiments. A wide variety of catalysts were tested based on their properties in a batch
reactor.

Batch reactor set-up and experimental procedure:


The batch reactor used was a

inch stainless steel cap and hex plug arrangement with

an internal volume of 3 ml. This could withstand temperatures up to 400 C and


pressures up to 500 atmospheres. Pressure was generated by transferring appropriate
amount of reactants and heating them up to the temperature desired. The reaction
temperature commonly adopted was 260 C. After transferring the reactants and the
catalyst, the cap and plug were tightened with Teflon lining in-between. The reactor was
then placed in an aluminum heating block that can heat up to 400 C. It took about 25
minutes (preheat time) for the reactor to reach the reaction temperature set in the heating
block. The reaction time was counted from then on, and after the desired reaction time,
the reactor was quenched in cold water and opened for product analysis. The reaction
time adopted was 18 minutes as this seemed to be a short enough reaction time for a high
temperature process, and could help in the comparison with Suppes et als calcium
carbonate process [70] as a standard. The reactor could be tumbled as needed to keep the
catalyst well-mixed.

36

Soybean oil was used as the vegetable oil reactant throughout this research. The soybean
oil used was of the commercial Wesson brand. Ethanol was used as the alcohol reactant
throughout this research, as it is renewable keeping in line with the biodiesel, and nontoxic for safety reasons in lab use.

40:1 molar ratio of ethanol and oil was adopted as suggested by Saka and Kusdiana [67].
No cosolvent was added unless otherwise mentioned. The catalyst was added well in
excess, 20% by weight, as opposed to the usual 1~5 % adopted for batch mode. This was
done to make sure that the reaction rate was saturated with respect to the catalyst amount
or surface area while evaluating a wide range of catalysts.

Product analysis:
For quick analysis of various experimental samples, a simple thin layer chromatographic
(TLC) procedure was adopted. See Appendix A. This procedure allows for a quick test
to screen numerous catalysts at a time. This is a qualitative and semi-quantitative
analysis. Samples corresponding to catalysts showing promise were tested in a Varian
3600 gas chromatography (GC) instrument. See Appendix B. This is a clear qualitative
and quantitative tool, and was planned to be used for the rest of the research. The
procedure was adopted from the ASTM procedure to analyze biodiesel composition [71].

Solid base salts:


It was desired to try calcium carbonate (CaCO3) first to start off from the work of Suppes
et al [70] and see what the results were. No cosolvent was used, unlike the work of

37

Suppes et al. 20 % by weight of calcium carbonate (particle size 5 microns) was added.
The product mixture was first analyzed by TLC. The TLC results showed considerable
conversion of vegetable oil to ethyl esters, but not complete conversion. Later when
analyzed by GC, it corresponded to a biodiesel conversion of 80 %. Basicity of the
anionic part (eg.: CO32- in CaCO3) of a salt is based on the charge per anion, number of
oxygen atoms and electronegativity of the central atom (eg.: C in CO32-) [72]. It was also
desired to see if the size and the electronegativity of the cationic part (eg.: Ca2+ in
CaCO3) of salts had any influence. As a result, a variety of salts, barium carbonate
(BaCO3), strontium carbonate (SrCO3), barium sulfate (BaSO4), barium chromate
(BaCrO4), calcium phosphate (Ca3(PO4)2), calcium hydroxyl phosphate (Ca5(PO4)3OH)
and zirconium silicate (ZrSiO4) were tested. None of these salts performed any better
than calcium carbonate as seen from TLC, and lot of them fared poorer than calcium
carbonate.

Metal oxide bases:


Having seen that the salts were not giving any improvement, it was decided to test some
metal oxides that are known to be basic in nature, but not too strong to have maintenance
issues. The first one to be tested was calcium oxide (CaO), as it is one of the prominent
oxides that is considered not to dissolve in alcohol, and was tried in various applications
for its basicity. Once the reactor was opened after the reaction, a large amount of soap
flakes were found. Though calcium oxide is termed insoluble in polar solvents, it is
actually sparingly soluble, not completely insoluble. Moreover, water is innately present
(5~10 % by weight) and cannot be removed until heated above 1000 C. A metal oxide

38

soluble in alcohol, and comprising water, would react with oil and form soap, calcium
salt of fatty acids.

It was thought that one way to tackle this problem was to try calcium oxide in
combination with another component that could act as a support for calcium oxide
molecules and keep them intact without dissolving into the alcohol. Calcium aluminum
oxide (CaO-Al2O3) and calcium titanate (CaO-TiO2) were considered as a result. Also,
similar supported alkali earth metal oxide combinationsbarium titanate (BaO-TiO2)
and magnesium aluminum oxide (MgO-Al2O3)were considered. With pure calcium
oxide proving problematic, it was also desired to try other metal oxides apart from trying
supported alkali earth metal oxides. There could be many metals across the periodic
table that may form insoluble metal oxide bases, and their basicity might vary across the
periodic table [73]. What could be a good base for one reaction may not be good for
another reaction [38]. Reaction with solid base catalysts involves adsorption of reactant
molecules onto the solid catalyst surface before reaction and desorption of the product
molecule after reaction to facilitate reaction with the next reactant molecule. A base that
is too strong in adsorbing a reactant molecule may be too strong to let the product
molecule go, and may not initiate reaction with the next reactant molecule; as a result, it
may not be effective. A base that is good at letting the product molecule leave during
desorption may not be effective in adsorbing the reactant molecule originally. Normally,
when the metals or metal oxides are tested as solid catalysts, across the periodic table
they show an optimal region where the corresponding metals or metal oxides are most
effective for a particular reaction [73]. So, metal oxides were selected corresponding to

39

various metals across the periodic table. Zinc oxide (ZnO), copper (II) oxide (CuO),
nickel oxide (NiO), manganese oxide (MnO) and titanium (II) oxide (TiO) were tried.
All the aforementioned catalysts after calcium oxideCaO-Al2O3, CaO-TiO2, MgOAl2O3, BaO-TiO2, ZnO, CuO, NiO, MnO and TiOwere tested as catalysts for the
biodiesel reaction at 260 C. Of these catalysts, MgO-Al2O3, BaO-TiO2, ZnO and NiO
did not show much improvement over calcium carbonate as seen from TLC. However,
CaO-Al2O3, CaO-TiO2, CuO, MnO and TiO showed almost complete conversion of
vegetable oil as seen from TLC. Using TLC to screen the catalysts, the catalysts that
seemed impressive from the TLC analysis were analyzed using GC. All these catalysts
gave biodiesel conversions of 92~97 % as shown in the GC results in Figure 4.1.

Since supported calcium oxide showed good results, it was desired to investigate the
effect of the supports. It was desired to see if the supports themselves, aluminum oxide
and titanium (IV) oxide, gave good conversion. The results of aluminum oxide
tested as -Al2O3 and titanium (IV) oxide (TiO2) were not any better than calcium
carbonate as seen from TLC and did not seem to be the key in contributing a conversion
around 95 % corresponding to CaO support combinations. In fact, titanium (IV) oxide
(TiO2) showed a conversion of only 82 % when tested by GC. See Figure 4.1. The
conversions of TiO2 and CaCO3, ~ 80 %, may seem close to 95 % in terms of numbers,
but in the later stages of the reaction, when the concentration of reactants is lower, the
reaction rate is slower. By equating the conversion numbers to pseudo first order
kinetics, the reaction time to reach 95 % conversion is twice as long as that to reach 80 %
conversion.

40

Biodiesel conversion (%) .

100
260 C, 70 atm

95

18 min

90

40:1 Ethanol:Oil Molar Ratio

85
80
75
70
65
60
TiO

CaOAl2O3

MnO CaTiO3 CuO CaCO3 TiO2

no
catalyst

Figure 4.1. Performance of the catalysts in batch tests

To study the behavior of the catalysts confirmed earlier to be effective from GC results, it
was desired to see how they fare when the temperature is decreased. As a result,
performance of these catalysts was tested at various temperatures from 150 C to 260 C.
The results are shown in Figure 4.2. The results of calcium carbonate are also shown in
Figure 4.2 for comparison. Along with these catalysts, titanium (IV) oxide (TiO2) was
also tested at these temperatures and the results are shown in Figure 4.2. The bottommost curve corresponds to the runs without catalyst. Of all the catalysts, TiO looks to be
the most effective, giving conversions of about 97 % at 260 C. TiO gave 95 %
conversion even at 20 C lower, at 240 C. 95% conversion itself is good enough as
commercial conversion. This could enable design of a commercial continuous reactor at
240 C. Without a catalyst, the conversion was about 60 % at 260 C, whereas the
promising catalysts showed conversions around 45~50 %, even at 150 C. Apart from

41

these, there were not significant differences in the trend of the catalysts at lower
temperatures.

Biodiesel conversion (%) .

100
90

Ca Al oxide

80

MnO

70

TiO

60

CaTiO3

50

CuO

40

CaCO3

30

TiO2

20

no catalyst

10
125

150

175

200

225

250

275

Temperature (Celsius)
Figure 4.2. Performance of the catalysts with temperature in the batch tests

The proposed reaction mechanism is: The hydrogen in the hydroxyl group of ethanol
gets attached to the electron-pair donating sites of the metal oxides (illustrated in chapter
6). As this hydrogen atom gets adsorbed to the solid catalyst, the remaining part of the
alcohol, the ethoxide ion, is capable of attacking the partially positively charged carbon
atom of the carbonyl bond of the triglyceride ester, just like the methoxide ion attack
illustrated earlier (in chapter 2) in the basic reaction mechanism.

TiO looks to be the most promising of the identified catalysts, but other identified
catalysts have to be tested in the continuous packed-bed reactor to get a conclusive idea.
42

It may help in identifying the optimal catalyst based on cost, activity, catalyst life, ease of
regeneration and availability considerations.

Free fatty acid conversion:


Conversion of FFA to esters is important in the production of biodiesel. Used vegetable
oil is a significant source for biodiesel and has high free acid content. Virgin vegetable
oil can also contain reasonable free acid content in some cases. The reaction was carried
out with oleic acid and ethanol in the batch reactor using two of the identified catalysts,
TiO and MnO. TLC showed very good conversions. When tested with GC, conversions
over 93 % were observed. Under the reaction conditions, proposed conversion of free
fatty acids is good even without a catalyst [74]. Metal oxides can be amphoteric in
nature, having both acidic and basic properties. It looks like the metal oxides identified
are amphoteric in nature as well. Base catalysts do not convert free acids, and normally
waste them as soap. Acid catalysts aid FFA conversion. So, by employing such metal
oxides, conversion of free acids also can be expected. This is not the case with the
conventional process reaction at 60~70 C, and they require an additional acid catalyzed
processing unit. Considering this, the pursued process may be a better way to produce
biodiesel.

Activity sustenance of the catalysts:


Catalyst life is crucial for a successful commercial process. TiO was reused for three
successive runs in the batch tests. The biodiesel conversion for the three runs read 94 %,

43

94 % and 95 %. The catalysts have to be tested in the continuous reactor for over 100
hours to reach any conclusion on the life of catalysts.

Effect of molar ratio:


The aforementioned batch tests were conducted at a molar ratio of 40:1 ethanol:vegetable
oil. During the initial batch tests of calcium carbonate, various molar ratios of 40:1, 30:1,
20:1, 10:1 and 6:1 were tested. During those runs, there was no significant change in the
conversion from 40:1 to 20:1 as seen from TLC, while the 10:1 and 6:1 ratios showed a
decrease in conversions. Effect of molar ratio should be studied with the continuous
reactor and the optimal molar-ratio has to found. More energy and reaction space
associated with excess alcohol has to be optimized with the better medium of interaction.
Ethanol has to be well in excess of the stoichiometric ratio since it is a critical fluid
medium as well as a reactant. While using the alcohol as a critical fluid reactant, higher
molar ratios are seen to have a direct effect on the reaction rate [67, 68].

Effect of Cosolvent:
From the initial tests, having seen the conversions decrease for the promising catalysts
with temperature, one reason was thought to be that decrease in temperature lowers the
kinetics of the reaction. However, there is another possibility. Lowering the temperature
below 240 C means going reasonably below the critical point of ethanol. This could
mean that ethanol and oil become two different phases and may not be miscible as well as
near-critical or supercritical ethanol is miscible with oil. With two different phases, mass
transfer limitation could also be a reason for the lower conversions while lowering the

44

temperature. As a result, it was desired to see how the use of a cosolvent would effect the
reaction. Kim et al had reported that hexane proved to be the best cosolvent for biodiesel
transesterification reaction at conventional reaction conditions [59]. So hexane was used
as the cosolvent. Based on the cosolvent molar ratios employed in other biodiesel
processes [59, 69], it was intended to test three molar ratios, 40:1:2, 40:1:1.4 and 40:1:0.8
(ethanol:oil:cosolvent), using TiO as the catalyst. The results are shown in Figure 4.3. A
point corresponding to a molar ratio of 40:1:4 with the TiO catalyst is also shown. As
shown in Figure 4.3, the use of a cosolvent does not seem to have any significant effect in
improving the reaction conversion. 40:1:4 run itself means that the cosolvent makes up
about 10% of the reactant mixture, and anything more than that is not worth the energy
spent on temperature and pressure as cosolvent composition.

Biodiesel conversion (%) .

100

no catalyst/ no cosolvent

90

no catalyst 40:1:2

80
70

no catalyst 40:1:1.4

60

TiO 40:1:2

50

TiO 40:1:1.4

40

TiO 40:1:0.8

30

TiO 40:1:4

20

TiO - no cosolvent

10
125

150

175
200
225
Temperature (Celsius)

250

275

Figure 4.3. Effect of cosolvent in batch tests

45

Two series of runs employing cosolvent without any catalyst corresponding to molar
ratios 40:1:2 and 40:1:1.4 were also conducted and the results are shown in Figure 4.3.
Again, the use of cosolvent does not seem to give any crucial advantage to the reaction.
It was decided that the cosolvent does not seem to be effective, at least as seen from the
batch runs.

Ultrasonic cavitation without catalyst:


Prior to testing solid catalysts, it was also desired to see if an ultrasonic sonicator could
be used without any catalyst to test if ultrasonic cavitation alone can effect the biodiesel
reaction, which would mean a much simpler process. Ultrasonic cavitation produces
localized high pressure compartments. The cavitation creates the much higher pressures
and temperatures as required by the no-catalyst supercritical alcohol process, without any
pressure vessel. The reaction mixture consisted of 10 ml of methanol and 10 ml of
soybean oil added to a 40 ml vial. The sonicator used was a Sonics Ultrasonic Processor,
Model GEX130. Run time was 15 minutes. The power input was varied from 32.5 W to
105 W. On analysis, it was found that there was no conversion of vegetable oil to
biodiesel. It was concluded that the cavitations produced were too transient for any
appreciable reaction to take place, and that the temperature and pressure of the cavitations
may not be controllable. So it was decided to go ahead with the testing of solid catalysts.

46

Chapter 5

Lab-Scale Packed-Bed Reactor

Having identified potential catalysts by batch mode, the next step naturally is to develop
a continuous process design. This is a high pressure process employing the identified
catalysts in a packed bed. The lab scale packed-bed reactor intended is a test of
feasibility for an effective commercial reactor and would provide the details needed for
the commercial reactor design. This would serve as a bridge between the initial
chemistry tests and the final commercial production, which is classic chemical
engineering. Testing catalyst life and commercial reactor flow modeling that gives an
idea of pressure drop, viscosity and flow regime would be part of these lab-scale packed
bed reactor studies.

Experimental set-up:
Figure 5.1 illustrates the experimental set-up of the lab-scale packed-bed reactor. A
Perkin Elmer Series 200 HPLC pump was used to feed the reactants to the reactor. It has
a back-pressure limit of 6000 psi. The pump can feed up to 4 different components.
However, since oil and ethanol have very different viscosities, the pump had to be
calibrated for the desired composition of reactants. It can pump reactants at a constant
flow rate and has a flow rate range of 0 to 10 ml/min.

The feed line from the pump was connected to the reactor which was housed in an oven.
1
16

inch 316 stainless steel tubings from McMaster-Carr were used for feed lines.

47

Quencher
Oven

Thermocouple

Pressure Gauge

Packed-Bed Reactor

Back-Pressure
Regulator

Thermocouple
Pre-Heat Coil

HPLC Pump

Oil

Alcohol

Figure 5.1. Lab-scale packed-bed reactor set-up Schematic diagram


48

Swagelok compression fittings were used for connections. The reactor was typically
made from inch or inch 316 stainless steel pipes from McMaster-Carr. 2 m frits
were fitted on each end of the reactor to prevent the solid catalyst particles from escaping
the reactor. Various solid catalysts (conforming to reasonable reactor diameter:particle
size ratio [75, 76]) were packed into the reactor by tapped-packing mode [77].

The oven used was part of a spare Hewlett-Packard gas chromatograph that can be heated
up to 400 C. Enough feed line length was introduced into the oven before the reactor as
preheat coil to facilitate the reactant mixture reach the reaction temperature before
entering the reactor. A Type-K thermocouple from Omega Engineering was used to
measure the temperature of the reactant mixture just before entering the reactor. The
thermocouple was connected to a thermometer (Omegaette HH308 Type-K) from Omega
Engineering that read the temperature. Glass wool was used to insulate the various line
openings in the oven.

The feed line from the reactor was passed through a water bath quencher that cooled the
product mixture to room temperature. A similar thermocouple was attached to the feed
line after the quencher to measure the temperature. A glycerin filled pressure gauge from
McMaster-Carr was then attached to the feed line. Then the back-pressure regulator was
attached to the feed line. The back-pressure regulator was purchased from Alltech
Chromatography. The back-pressure regulator has a ball bearing pressing against the
inlet feed line. The reactant mixture inside the feed line must have enough force to press
against the ball bearing to pass out. This force relates to the pressure inside the feed line

49

until this point and this pressure was indicated by the pressure gauge. The pressure
regulator has a valve that can be adjusted for setting various pressures. The pump keeps
on pumping the reactant mixture into the feed line until the set pressure is reached. Thus
it relates to back-pressure; the pressure being built in the feed line behind the regulator by
the resistance offered by the regulators ball bearing. The regulator uses plastic finger
ferrules for the in-flow and out-flow line connections. The product was collected
thereafter to be used for analysis.

The regulator keeps the pressure in a narrow range. For the usual experimental runs with
ethanol at 260 C, the head pressure shown in the pump monitor can vary from 1040 to
1100 psi.

The feed line just before the outlet pressure gauge was connected directly to the pump
outlet, and the pressure readings on the pump monitor and the pressure gauge matched
well.

Residence times were usually varied by changing the flow rates. The selected flow rate
ranges were made sure that there was no external mass transfer limitation (explained in
Chapter 6). Reactors of different sizes were also used to get residence times of different
ranges. Sometimes, specifically the reactor length was increased to get longer residence
times.

50

Samples were collected for analysis after the reactor had run for at least six residence
time cycles or 45 minutes (whichever was longer) to make sure that the reactor reached
steady state. Usually two to four samples were collected for the data points listed in a
study. In some cases, more samples were collected. Experiments were also completely
re-run at a later time and experiments were also run in new reactors, regularly, to check
for various degrees of repeatability. The time gap between the two subsequent samples
collected for a series of run was usually around 30 minutes, but it also varied from 10
minutes to 2 hours in some cases.

The lab-scale packed-bed reactor is illustrated by the real photographs of various sections
of the experimental set-up. See Figures 5.2 through 5.5.

Mass balance tests were done to make sure there was no loss of mass from the point
where the reactants were drawn into the pump to the point where the sample was
collected at the outlet. The change in the mass of the reactants in the reactant chambers
during a particular time (mass-in) and the mass of the product mixture collected during
the same time (mass-out) were measured and compared. The mass-in and mass-out
matched each other closely. The masses were compared for different flow rates, different
temperatures (200 C and 260 C), different feeds (ethanol alone and 40:1 ethanol:oil)
and with and without an MnO catalyst. The masses matched the same way as those of
runs that were not passed through the reactor and collected directly.

51

Oven

Feed-line

Reactor segments

Figure 5.2. Photo: Packed-bed reactor segments inside oven

Ethanol

Vegetable oil

Figure 5.3. Photo: HPLC pump for the reactor set-up


52

Figure 5.4. Photo: Reactor GC oven

Thermocouple

Temperature
Reading

Water Bath
Back-Pressure
Regulator

Pressure
Gauge

Product
Collection

Figure 5.5. Photo: Reactor set-up following oven


53

Experiments:
Catalysts were tested in the packed-bed reactor, following up from the initial batch tests.
The various catalysts tested were titanium (II) oxide (TiO), calcium titanate (CaTiO3),
manganese (II) oxide (MnO), calcium carbonate (CaCO3), nickel (II) oxide (NiO),
vanadium (II) oxide (VO), iron (II) oxide (FeO), zinc oxide (ZnO) and cobalt oxide
(CoO). It was found that most of the catalysts had surface area in a similar range as
explained in chapter 6.

Commercial Wesson soybean oil and ethanol were the reactants, and the oil:ethanol
molar ratio was 40:1. Reactor run times (the cumulative time period for which the
reactor has been running) corresponding to the following experiments were from 0 to 15
hours. The following are the experiments conducted, the results of which are discussed
in the subsequent section.

Titanium (II) oxide TiO:


TiO used was a +50 mesh size sample purchased from Sigma-Aldrich. 0.18 inch ID (
inch OD) and 30 cm long reactor was used. Reactor volume was 4.93 ml. Void fraction
of the packed-bed reactor was 40.9 %. Void volume of the reactor was 2.015 ml. Flow
rates were varied to get various residence times.

Another reactor, 10 mm ID ( inch OD) and 25 cm long, was also used. Reactor volume
was 19.64 ml. Void fraction of the reactor was 39.33 %. Void volume of the reactor was
7.73 ml.

54

Calcium titanate (CaTiO3):


CaTiO3 used was the same sample used in the batch tests and was purchased from Alfa
Aesar. However, that sample was granulated to 35-20 mesh size by J.M. Huber, Inc. and
was loaded into the reactor. The granules were found to be stable when checked after the
runs opening the reactor. 10 mm ID ( inch OD) and 25 cm long reactor was used.
Reactor volume was 19.64 ml. Void fraction of the packed-bed reactor was 52.3 %. Void
volume of the reactor was 10.28 ml. Flow rates were varied to get various residence
times.

Another reactor, 0.18 inch ID ( inch OD) and 15 cm long, was also used. Reactor
volume was 2.46 ml. Void fraction of the reactor was 49.77 %. Void volume of the
reactor was 1.226 ml.

Manganese (II) oxide (MnO):


MnO used was of 200-60 mesh size, the same sample used in the batch tests purchased
from Alfa Aesar. 0.334 inch ID ( inch OD) and 30 cm long reactor was used. Reactor
volume was 16.96 ml. Void fraction of the packed-bed reactor was 61.2 %. Void volume
of the reactor was 10.38 ml. Flow rates were varied to get various residence times.

Another reactor, 0.18 inch ID ( inch OD) and 15 cm long, was also used. Reactor
volume was 2.46 ml. Void fraction of the reactor was 60 %. Void volume of the reactor
was 1.48 ml.

55

The above MnO sample was also used in granulated form, 35-20 mesh size. The sample
was granulated by J.M. Huber, Inc. 10 mm ID ( inch OD) and 25 cm long reactor was
used for the granulated sample. Reactor volume was 19.64 ml. The granules were found
to be stable when checked after the runs opening the reactor. Void fraction of the
packed-bed reactor was 53.8 %. Void volume of the reactor was 10.56 ml. Flow rates
were varied to get various residence times.

Calcium carbonate (CaCO3):


The sample used was the commercial brand, Fre Flow, provided by Iowa Limestone
Company, and was of 20-10 mesh size. This was the same sample used by Suppes et al
in their calcium carbonate catalyst studies of biodiesel reaction [70]. 10 mm ID ( inch
OD) and 25 cm long reactor was used. Reactor volume was 19.64 ml. Void fraction of
the packed-bed reactor was 36.55 %. Void volume of the reactor was 7.18 ml. Flow rates
were varied to get various residence times.

Vanadium oxide (VO):


VO used was the same sample used in the batch tests purchased from Alfa Aesar. The
particle size was in the range of the other catalysts tested. 0.18 inch ID ( inch OD) and
19.55 cm long reactor was used. Reactor volume was 3.21 ml. Void fraction of the
packed-bed reactor was 46.55 %. Void volume of the reactor was 1.49 ml. Only one flow
rate, 0.3 ml/min, was adopted to get the conversion at the optimal residence time
(explained in chapter 7) of 5 minutes.

56

Iron (II) oxide (FeO):


FeO was purchased from Aldrich. The particle size was in the range of the other
catalysts tested. 0.18 inch ID ( inch OD) and 21.9 cm long reactor was used. Reactor
volume was 3.6 ml. Void fraction of the packed-bed reactor was 38.06 %. Void volume
of the reactor was 1.37 ml. Only one flow rate of 0.27 ml/min was adopted to get the
conversion at the optimal residence time (explained in chapter 7) of 5 minutes.

Nickel oxide (NiO):


NiO used was the same sample used in the batch tests purchased from Alfa Aesar. The
particle size was in the range of the other catalysts tested. 0.18 inch ID ( inch OD) and
24.6 cm long reactor was used. Reactor volume was 4.04 ml. Void fraction of the
packed-bed reactor was 71.09 %. Void volume of the reactor was 2.87 ml. Only one flow
rate of 0.57 ml/min was adopted to get the conversion at the optimal residence time
(explained in chapter 7) of 5 minutes.

Cobalt oxide (CoO):


CoO used was the same sample used in the batch tests purchased from Alfa Aesar. The
particle size was in the range of the other catalysts tested. 0.18 inch ID ( inch OD) and
17.1 cm long reactor was used. Reactor volume was 2.81 ml. Void fraction of the
packed-bed reactor was 66.5 %. Void volume of the reactor was 1.87 ml. Only one flow
rate of 0.37 ml/min was adopted to get the conversion at the optimal residence time
(explained in chapter 7) of 5 minutes.

57

Zinc oxide (ZnO):


ZnO used was the same sample used in the batch tests purchased from Alfa Aesar. The
particle size was in the range of the other catalysts tested. 0.18 inch ID ( inch OD) and
15 cm long reactor was used. Reactor volume was 2.46 ml. Void fraction of the packedbed reactor was 60.63 %. Void volume of the reactor was 1.49 ml. Only one flow rate of
0.29 ml/min was adopted to get the conversion at the optimal residence time (explained
in chapter 7) of 5 minutes.

Magnesium silicate (MgSiO3):


The sample used was the commercial brand, Florisil, purchased from Sigma-Aldrich and
was of 200-100 mesh size. 0.334 inch ID ( inch OD) and 30 cm long reactor was used.
Reactor volume was 16.96 ml. Void fraction of the packed-bed reactor was 76.2 %. Void
volume of the reactor was 12.93 ml. Flow rates were varied to get various residence
times.

No catalyst:
Empty reactor runs without catalyst were also performed. 0.18 inch ID ( inch OD) and
15 cm long reactor was used. Reactor volume was 2.49 ml.

Suitable experiments could not be carried out with copper (II) oxide (CuO) and calcium
aluminum oxide (CaO-Al2O3). The two CuO samples evaluated were dusty and there
was visible leaching in the product stream for an extended period of time (more than 2
hours). 13 wt % CuO loaded on alumina resulted in a smoky product stream which is

58

discussed in chapter 7. The granulation facility was not available for calcium aluminum
oxide, so it was not pursued. Aluminum oxide was seen as a support and only calcium
oxide was seen as the catalyst. So it was decided that calcium titanium oxide (calcium
titanate) was similar enough to evaluate the activity of calcium oxide.

Results:
The results are shown in Figure 5.6. There is a clear demarcation in the performance of
catalysts identified in the batch tests and the other catalysts. Among those catalysts
identified earlier in the batch test, MnO clearly proves to be the best. The three catalysts
identified earlier to be better (MnO, TiO and CaTiO3), showed conversions greater than
95 % in 18 minutes as confirmed by the excess catalyst used in the batch tests. One starts
to see the finer differences between these catalysts in the shorter residence times. MnO
conversions are significantly better compared to CaTiO3 at shorter residence times, even
though MnOs packing fraction and surface area (explained in chapter 6) were less than
that of CaTiO3. The results of granular 35-20 mesh MnO were also similar to that of
200-60 mesh MnO. Even though TiO conversion reaches around 95 % at 18 minute
mark, its conversion curve falls even below CaTiO3 at shorter residence times. MnO had
shown better conversions at lower temperatures in the batch tests also. That is another
indication that MnO has the best activity for the reaction.

Of the two catalysts (CaCO3 and TiO2) used as standards in the batch tests, CaCO3 was
tested in the packed-bed reactor. Only CaCO3 was tested because both of them showed
similar conversions with excess catalyst during the batch tests. Those similar results

59

correspond to lower conversion ranges that are not close to equilibrium conversions in
the higher range, and can be seen as a range that shows differences in activity well, unlike
those of CaTiO3 and MnO around 95 % conversions. CaCO3 requires 4~5 times longer
residence times than MnO to reach the same conversion (~80 %).

100
90

Biodiesel conversion (%) .

80
70
60
50
40
30
20
40:1 Ethanol:Oil Molar Ratio
10

260C, Pressure ~70 atm

0
0

10

15

Residence time (min)


TiO

CaTiO3

MnO (200-60 mesh)

CaCO3

no catalyst (125 atm) - Reference [68]

FeO

VO

CoO

ZnO

NiO

No catalyst - this work


Figure 5.6. Lab-scale packed-bed reactor results

60

20

25

The catalyst-free supercritical alcohol process of Saka and Kusdiana [68] was also
compared at 260 C. The pressure corresponding to 260 C in the supercritical alcohol
process was 125 atmospheres whereas the pressure in the lab-scale packed-bed reactor
was about 70 atmospheres. Conversion is around 90 % in 10 minutes for MnO, whereas
it is around 30 % in 10 minutes for the supercritical alcohol process without catalyst.
Considering the reaction kinetics (explained in chapter 6), the catalyst-free reaction
reported by Saka and Kusdiana is 7 times slower than the MnO catalyzed reaction.
Reaction was also performed on an empty reactor in the lab-scale reactor set-up keeping
the pressure around 70 atmospheres. Conversion was around 25 % in 12 minutes. This
corresponds to a reaction that is 11 times slower than that of MnO.

FeO and some of the other metal oxide catalysts not seen to be sufficiently effective in
the batch tests were also tested. Only FeO showed an appreciable conversion, 40 %, in 5
minutes. Other catalystsNiO, VO, ZnO and CoOshowed poor conversions,
comparable to CaCO3 and no-catalyst reactions. They thus confirmed their poor reaction
rates observed in the batch tests.

The Florisil MgSiO3 reactor clogged an hour after it was started. The reactor was opened
for examination, and the catalyst inside seemed to have fused together. It was not
possible to run the reactor anymore.

Other alkali earth metal oxides are considered to be similar to CaO in their catalytic
properties, and CaO is the popular catalyst to use as solid base from the alkali earth metal

61

oxides. So CaO can be considered to be a representative of that class of oxides. Since


the metal oxide is the important part compared to the support, calcium titanate can be
considered as a representative of catalysts like calcium silicate, calcium aluminum oxide
and magnesium silicate.

Data validation:
For a single data point, three or four samples were collected in most cases, and two
samples in some cases, throughout the research. The results of the individual samples
were averaged and checked if the range of individual sample results falls within 5 % error
range i.e. +/- 2.5 % from the average value. Only the runs conforming to this were
documented as data points, and the non-conforming runs were rerun. The experiments,
particularly the initial stages of new procedures, were also checked for repeatability
randomly. This data validation was conducted throughout the research.

To sum it up, manganese (II) oxide clearly looks to be the best catalyst from the
aforementioned tests. It was not quite clear from the batch tests. MnO samples tested did
not have higher surface area than the other catalysts (explained in Chapter 6). MnO was
also much cheaper compared to TiO. The 500 g sample purchased from Alfa Aesar and
Sigma-Aldrich cost only around $22, whereas TiO cost around $50 for a 25 g sample.
MnO was among the cheapest catalysts tested. It was also sturdier in structure compared
to most of the other catalysts. MnO is also one of the catalysts that are available in nature
in +2 oxidation state, unlike TiO. Hence, it was decided to use MnO as the first choice
for the various studies to follow.

62

Chapter 6

Reaction Kinetics

Reaction kinetic studies are important to design the chemical reactor. Size of the reactor
can be determined only from the reaction rate. Modeling the reaction with a consistent
kinetic model and validating the rate constant of the reaction by Arrhenius plot enable
proper design of reactor over a range of temperature pressure. Otherwise one would have
to bank only on a handful of data that may have been collected at certain reaction times
and particular temperatures. In order to study the kinetics of the reaction, one must
understand the various facets of the reaction regime. These facets include the flow
regime inside the reactor, external and internal mass transfer, order of the reaction,
catalyst surface area, effect of temperature, phase-separation susceptibility and critical
phase effectiveness to arrive at the understanding of the catalytic mechanism and the
activity of the solid catalyst for the reaction.

External mass transfer:


External mass transfer is the transfer of mass from the bulk phase to the surface of
catalyst particles. If the external mass transfer rate is slower than the reaction rate in the
catalyst particle, the overall rate of reaction is dictated by the external mass transfer
limitation. It does not matter how superior the solid catalyst is in catalyzing the reaction.
It is only as fast as the external mass transfer rate. To effectively exploit the activity of
the solid catalyst, the external mass transfer rate must not be less than the reaction rate.
In adopting various flow rates, reactor dimensions and catalyst particle sizes in the lab-

63

scale reactor studies, it is important to make sure there is no external mass transfer
limitation.

One way to examine external mass transfer limitation is to increase the flow rate
(superficial velocity) for the same residence time and measure the conversion. If the
conversion is the same for the increased flow rate, then there is no external mass transfer
limitation at the original flow rate. This method acts as a confirmation of no external
mass transfer limitation for the whole range between the lower and higher flow rates
tested [78].

An example of the method is as follows. The MnO catalyst with particle size range, 20060 mesh, was loaded onto two stainless reactors with the same internal diameter, 0.18
inch, the same way. One reactor was 22 cm long and the other was 88 cm long. A flow
rate of 0.43 ml/min was adopted for the former reactor and 1.72 ml/min was adopted for
the latter reactor, keeping the ethanol:soybean oil molar ratio constant at 36.2:1. The
residence time for both the reactors was the same, 5 minutes. When the product samples
corresponding to the two flow rates were analyzed, the biodiesel conversion was found to
be 75 % in both cases. This result indicated that there was no external mass transfer
limitation for the flow rate range 0.43 ml/min to 1.72 ml/min.

The superficial velocity is the actual relevant parameter behind the flow rate in the above
method. Superficial velocity refers to the velocity of the feed stream through the void
spaces in the catalyst bed. It is calculated by dividing the flow rate by the cross-sectional

64

area and the void fraction of the bed. If it is not divided in combination with the void
fraction of the bed, the calculated velocity would not be the true representation of flow
inside the bed. If the conversion increases from a lower superficial velocity to higher
superficial velocity for the same residence time, it means that there is increased mixing
and increased contact with catalyst at the higher velocity with the likelihood of a
diffusion-limited boundary layer at the catalyst surface at the lower velocity.
Superficial velocity =

Volumetric flow rate


Cross sec tional area x Void fraction of bed

Thus, for the given particle size and catalyst, the aforementioned flow rate range 0.43 ~
1.72 ml/min corresponds to superficial velocity range 72.7 x 10-3 cm/s to 290.9 x 10-3
cm/s, and no external mass transfer limitation was experienced.

For the 200-60 mesh (74~250 microns) MnO catalyst used extensively throughout the
research, flow rates ranging from 0.10 ml/min to 2.0 ml/min were observed to have no
external mass transfer limitation for the 0.18 inch ID reactor. This flow rate range
corresponds to the superficial velocity range 16.9 x 10-3 ~ 336.6 x 10-3 cm/s for any
reactor using this type and size of catalyst. The tests conducted with various other
catalysts in the research were run so that there was no external mass transfer limitation.

Also, the various catalysts were packed such that the packing conformed to a minimum
D/dp ratio. D refers to the diameter of the packed-bed and dp refers to the diameter of
catalyst particle. Small D/dp ratios, 1~2, would result in much of the fluid flow near the
wall of the reactor, resulting in inefficient contact with the catalyst (short-circuiting).
Packed-bed experiments were run at D/dp ratios that were reported to be reasonable [75,

65

76]. Common D/dp ratios associated with the packed-bed runs in the research were
around 10~70, with a minimum of 6 in some experiments. The aforementioned test of
using a longer packed bed by increasing the superficial velocity for the same residence
time, was also used to test for short-circuiting and reasonable D/dp ratios. Thus, the
external mass transfer limitation was eliminated for the purpose modeling kinetics of the
chemical reaction itself.

Internal mass transfer:


Internal mass transfer relates to the rate of mass transport inside the catalyst particles. In
a porous catalyst, the rate of diffusion of components from the catalyst surface to active
sites inside the pores can be slower than the rate of chemical reaction. As a result, all the
active sites present in the catalyst may not be exploited for the maximum reaction rate.
The ratio of observed rate constant to the intrinsic rate constant (maximum rate constant
accounting for all the internal active sites) is called the effectiveness factor. Commercial
reactors are normally not designed for a catalyst effectiveness factor of 1, as production
rate is the overriding factor [79, 78]. The pellets can be loaded with lower catalyst per
unit volume so that all the catalyst is utilized in the reaction. This would mean a lower
rate constant and lower production rate as a result. It is desirable to increase the catalyst
loading to increase the rate constant even if it means a fraction of the catalyst may not be
utilized. Most of the commercial reactors are designed for an effectiveness factor of
0.4~0.8.

66

In a packed-bed reactor, the rate constant of the reaction for a given catalyst is defined by
the surface area of the catalyst per unit volume of the reactor [78]. The rate constant
increases linearly with the surface area per unit volume and ultimately reaches a
saturation value.

For a given residence time, all the surface area inside a porous catalyst may not be
exploited. Part of the catalyst would be redundant as the time taken for diffusion into
deeper pores would be slower for the given residence time. If the residence time is
longer, surface area in deeper, previously unused pores could be used. In essence, the
chemical reaction itself can be studied for any effectiveness factor. Various fractions of
the catalyst can just be considered a non-factor for various conditions. The fraction of the
catalyst that did not contribute to conversion for a given residence time due to slower
diffusion time could still be part of the apparent conversion through longer residence time
paths. Thus, the basic kinetics of the chemical reaction remains the same and can
invariably be studied by consistent modeling with a definitive rate constant.

Order of reaction and rate constant:


Having clarified external mass transfer and internal mass transfer issues, the kinetics of
the heterogeneously catalyzed chemical reaction can be defined. Saka and Kusdiana
proposed through their supercritical alcohol process that excess alcohol as in 40:1 molar
ratio of alcohol:oil (as opposed to the 6:1 ratio employed in conventional process)
provided a much better rate constant for the reaction, and that reaction follows pseudo
first order kinetics based only on the concentration of vegetable oil [67, 68]. The alcohol

67

acts both as reactant and cosolvent in this regime. The effectiveness of the excess alcohol
can be explained by a reaction environment where the large triglyceride molecule is
surrounded all over by the alcohol molecules, and the alcohol is present readily at the
carbonyl group reaction site. The coverage of alcohol molecules all around the oil
molecule, while simultaneously facilitating their ready presence at the carbonyl reaction
site for reaction, makes alcohol a cosolvent as well as a reactant. The situation is
predicated by one reactant, the oil molecule, being extremely larger than the other
reactant, the alcohol molecule. It is this presence of excess alcohol that makes the
reaction rate depend solely on the concentration of vegetable oil, making the reaction
pseudo-first order with respect to oil concentration. The reaction in conventional process
with 6:1 molar ratio is second order.

In the present research, the reaction is catalyzed by a solid catalyst. As opposed to the
catalyst-free supercritical methanol process, the reactants are not freely surrounded just
by themselves. They have to react with each other around a catalytic site. Would the
reaction regime microcosm explained earlier be altered by any surface phenomena of the
catalyst? It has to be seen if the pseudo-first order kinetic model holds good for the
reaction catalyzed by a solid catalyst. The reaction regime is near-critical and not
supercritical. Would that alter the assumed kinetic model? Saka and Kusdiana claimed
in their no-catalyst reaction that the near-critical reaction regime that was not
supercritical resulted in significantly less effective reaction. All the above questions can
be answered while checking the presumed kinetic model with the experimental data.

68

As mentioned in Chapter 2, the reaction of triglycerides and monohydroxy alcohol to


form the biodiesel fatty acid esters and glycerol is a series of reversible reactions. The
triglyceride first reacts with ethanol to form diglyceride and a biodiesel ethyl ester. The
diglyceride then reacts with ethanol to form monoglyceride and another biodiesel ester.
The monoglyceride then reacts with ethanol to finally form glycerol and a third biodiesel
ester. The overall reaction is the formation of 3 moles of biodiesel esters and 1 mole of
glycerol from 1 mole of triglyceride and 3 moles of ethanol.
R-COOCH2
|
R-COOCH + C2H5OH
|
R-COOCH2
HOCH2
|
R-COOCH + C2H5OH
|
R-COOCH2
HOCH2
|
HOCH + C2H5OH
|
R-COOCH2

R-COOC2H5

HOCH2
|
+ R-COOCH
|
R-COOCH2

R-COOC2H5

HOCH2
|
+
HOCH
|
R-COOCH2

R-COOC2H5

HOCH2
|
HOCH
|
HOCH2

Even though glycerides at various stages (tri, di and mono) react with ethanol to form
biodiesel ethyl esters, they all can be seen to react equivalently with ethanol. The three
glyceride chains of a triglyceride can be seen as individual reactants reacting with ethanol
to form ethyl esters, in modeling the kinetics of the reaction. Saka and Kusdiana also
modeled satisfactorily the catalyst-free supercritical reaction this way [68]. The reaction

69

mixture can be seen as glyceride chains of various glyceride forms and ethanol. A
triglyceride would have three unconverted glyceride chains, a diglyceride would have
two unconverted glyceride chains, and a monoglyceride would have one unconverted
glyceride chain.

Let these glyceride chains be referred to as g. The reaction rate expression can be
written as
Reaction rate = -

dC g
dt

(1)

where Cg is the concentration of the triglyceride chains (moles/liter),


t is the reaction time (min).
Since it is a first order reaction with respect to glyceride chain concentration and not
dependent on the concentration of excess ethanol at 40:1 molar ratio, the reaction rate can
be written as
-

dC g
dt

= kCg

(2)

= -kdt

(3)

where k is the rate constant (min-1).


Eq. (2) can be rearranged as

dC g
Cg
On integrating eq. (3), the result is

ln Cg = -kt + c
where c is the integration constant.
At time t = 0, Cg = Cg0, where Cg0 is the initial concentration of glyceride chains.

70

(4)

Incorporating this into eq. (4), the result is


ln Cg0 = -k(0) + c
c = ln Cg0

(5)

Substituting this in eq. (4), the result is


ln Cg = -kt + ln Cg0
ln Cg - ln Cg0 = -kt
ln

Cg
C g0
Cg
C g0

(6)
(7)

= -kt

(8)

= e-kt

(9)

It would be much easier to relate to the extent of reaction in terms of fraction of original
reactants (glyceride chains) that were converted to biodiesel ester. Lets denote this
fractional conversion to biodiesel esters as X.
X =

C g0 C g
C g0

= 1-

Cg
C g0

(10)

Incorporating eq. (9) into eq. (10), the result is


X = 1 - e-kt

(11)

This is the rate expression that can be used to calculate the biodiesel conversion for a
given residence time for a first order reaction.

The experiments can be run to confirm the aforementioned pseudo first order kinetics of
the heterogeneously catalyzed reaction. If the experimental conversion data plotted
against reaction time matches with eq. (11), a precise value of rate constant, k, would be
seen confirming the first order reaction. While testing the kinetic model of any reaction

71

for a reasonable degree of order, the first order is checked first for regression fit. So, one
does not need to be concerned about a wrong reaction order matching accidentally.

Experimental details:
Three separate inch OD, 0.18 inch ID reactor segments, 30 cm, 29 cm and 29 cm long
respectively, loaded with 200-60 mesh MnO, were used in various combinations (series
and individual) to get residence times from 0 to 15 minutes. Total reactor volume was
14.45 ml. Void volume was 8.38 ml, meaning 58.01 % void in the packed bed. 40:1
molar ratio of ethanol:oil was adopted. The cumulative reactor run time corresponding to
these runs were from 48 to 82 hours.

The biodiesel conversion was plotted against residence time, as shown in Figure 6.1. A
plot of 1-X vs t, corresponding specifically to the first order kinetic models eq. (11), is
also shown in Figure 6.1. 1-X was fitted exponentially with t with an R2 value of 0.9945.
This confirms that the reaction catalyzed by the solid catalyst is a first order reaction with
respect to oil concentration. Even though only low conversion ranges are normally used
to evaluate the kinetic model, here it can be seen that conversions up to 88 % fit the
model curve, after which it falls off slightly, approaching the equilibrium conversion of
97 %. The rate constant, k, as given by the exponential fit of the model, is 0.283 min-1.

The first order rate expression with a fixed rate constant value matches with the
experimental data. This proves that the reaction follows first order kinetics. The
assumption of modeling individual glyceride chains of mono-, di- and triglycerides as

72

Biodiesel conversion, X (%) .

100
90
80
70
60
50

260C, 70 atm

40

40:1 Ethanol:Oil Molar Ratio

30
20

200-60 mesh MnO

10
0
0

12

15

Residence time, t (min)

0.8

1-X

0.6

0.4
-0.2829 t

1-X = e
2

R = 0 .9 9 4 5

0.2

0
0

12

15

Residence time, t (min)


Figure 6.1. Kinetic modeling of the MnO catalyzed biodiesel reaction

73

equivalent individual reactant molecules is also seen to be valid. The mechanism of the
reactants reacting through a heterogeneous catalytic site is proven to be consistent with
the first order reaction model. The kinetic model does not seem to have been affected by
any surface phenomena of the solid catalyst either. The near-critical regime, which is
unlike the supercritical regime that was confirmed to be stable by Saka and Kusdiana
[68], still conforms to first order reaction kinetics.

Effect of temperature on reaction rate:


It was intended to study what happens to the solid-catalyzed reaction at lower
temperatures. Would it follow the Arrhenius plot? The Arrhenius plot defines the
dependence of the rate constant on temperature through a straight line plot of the
logarithmic rate constant against the inverse of absolute temperature. It validates reaction
rate data collected for the reaction of interest. It gives the quantitative basis of the
relationship between the activation energy and the rate at which a reaction proceeds.
Would there be a decrease in the rate of reaction due to phase separation at lower
temperatures? This would help identify the temperature range where the reaction should
be conducted in order to be not compromised by phase separation. Reaction at lower
temperatures was studied in the batch tests as well. However, it was not as straightforward as doing it in the packed-bed reactor, because the batch reactor needed a certain
initial preheat time that is much desired in calculating the rate constant. At lower
temperature where phase separation happens, this preheat time is a significant part of
phase separated mixture before the reaction mixture becomes miscible as more biodiesel
esters are produced. Studying the reaction at lower temperatures would give an idea of

74

the degree to which the reaction time changes compared to the reaction time at 260 C.
Would there be any unusual change in reaction rate near the critical point of the alcohol?
The Arrhenius plot would reveal whether the reaction regime around 260 C is enhanced
in rate constant because of the critical regime, or whether it is a normal extension of a
miscible regime from whatever temperature miscibility originated. The classic Arrhenius
plot of logarithmic rate constant against inverse of absolute temperature, where a straight
line is usually anticipated, would reveal various details about the reaction, the reaction
regime and the catalyst.

Experimental details:
200-60 mesh MnO catalyst loaded onto a stainless steel reactor (1/4 inch O.D, 0.18 inch
I.D., 22 cm long), the same reactor used for activity sustenance tests discussed in Chapter
7, was used. Reactor volume was 3.61 ml. Considering MnO particle density of 5.445
g/cc and the observed packing fraction of 40.19 %, void volume of the reactor was
calculated to be 2.16 ml. Flow rate was set at 0.43 ml/min. This corresponds to a
residence time of 5 minutes, which was found to be optimal to detect changes in rate
constant, as explained in Chapter 7. An alcohol:oil molar ratio of 36.2:1 was adopted.
The reaction was run at the following temperatures: 260 C, 240 C, 220 C, 200 C, 175
C, 150 C and 100 C. The runs at various temperatures were also collected for two
different pressures, one was a constant pressure of 1070 psi (73 atm), and the other was
the saturation vapour pressure corresponding to each temperature. Saturation vapour
pressure is the minimum pressure required at any particular temperature to maintain a
component in its liquid state. The pressure was varied to corresponding saturation vapour

75

pressure to see if reaction at lower temperatures would allow lower operating pressure,
and it has been discussed as Effect of pressure in a separate section in Chapter 7. The
runs with constant 1070 psi (73 atm) at all the temperatures were conducted to study
more accurately the effect of temperature on reaction rate. Samples were collected at
various time intervals and analyzed in the gas chromatograph.

The conversions at various temperatures for 5 minute residence time are shown in Figure
6.2. The conversion at 100 C is for 20 minute residence time. The conversions were
used to calculate the respective rate constants. The logarithmic rate constants were
plotted against the inverse of temperature to obtain the Arrhenius plot as shown in Figure
6.3. The conversions and rate constants corresponding to saturation vapour pressure are
also shown in both the figures. A linear trend line corresponding to the rate constants
(constant 1070 psi) from 260 C and 175 C is also shown. The rate constants from 260
C to 175 C followed a well-defined linear plot, as illustrated by the linear trend line.
This shows that the reaction catalyzed by the MnO catalyst follows the Arrhenius plot in
this particular temperature range.

However, as the temperature is lowered, the rate constants fall below the trend line, as
illustrated at 150 C and 100 C. This could be due to phase separation of oil and ethanol
phases at lower temperatures, as a reactant mixture that is not well-mixed would have its
reaction rate constant compromised. Tang et al reported vapour-liquid equilibrium data
at these temperatures and pressures [80]. They showed that for methanol:oil molar ratio
~ 40:1, the phase separation starts from 175 C and increases at lower temperatures.

76

Ethanol has higher solubility in triglycerides than methanol. So ethanol can be expected
to have a slightly lower phase separation temperature than methanol. This could be the
reason why the rate constants at 150 C and 100 C fell below the linear trend line
corresponding to the rate constants at higher temperatures. If the alcohol and oil are
phase separated, the reaction rate would be affected. Thus, it can be clearly seen that at
lower temperatures where the reaction mixture is phase separated, the reaction rate is
negatively affected.

Biodiesel conversion (%) .

80
70

1070 psi

60

Saturation vapour
pressure

50

MnO (200-60 mesh),


Residence time = 5 min

40
30
20
10
20 min

0
50

100

150

200

250

300

Temperature (C)
Figure 6.2. Biodiesel conversions at various temperatures using 200-60 mesh MnO

77

0.0017
0.0
-1.0

0.0019

0.0021

0.0023

0.0025

0.0027

260 C
240 C
220 C

-2.0

36:1 Molar ratio


200 C

ln k

-3.0

175 C

-4.0

150 C

-5.0
-6.0
-7.0

k at 1070 psi
100 C

k at saturation vapour pressure


-8.0
1/T (1/K)

Figure 6.3. Arrhenius plot of the transesterification reaction using 200-60 mesh MnO
solid catalyst

The rate constant expression corresponding to the linear Arrhenius plot is

1
ln k = -6398 + 10.718
T
where T is the absolute temperature (Kelvin),
k is the rate constant (min-1).

78

Compared to 260 C, the reaction is 4.5 times slower at 200 C and 10 times slower at
175 C. To reach 95 % biodiesel conversion, the rate constant corresponds to 11 minutes
at 260 C, whereas it corresponds to 50 minutes at 200 C and 106 minutes at 175 C.
The reaction times at 200 C and 175 C are too long for insulation heat loss of the
process, considering that 260 C is not very far off from 200 C. The higher
temperatures allow for more efficient heat recovery in heat exchangers. Moreover, lower
temperatures mean larger reactors and more catalyst. From the aforementioned reaction
rate comparisons at different temperatures, the reactor at 175 C should be 10 times as
large as the reactor at 260 C. Also, the pressure drop along the reactor would be higher
as the temperature is lowered. It has been explained in Chapter 7 that the reaction at 260
C has extremely low pressure drop. But the pressure drop, as confirmed from the labscale reactor, was much higher at lower temperatures such as 100 C and 150 C. This
almost negates advantage of the low pressure facilitated by the saturation vapor pressure
of alcohol at lower temperatures. It is normally preferred to be in the higher range in
high temperature reactions and have shorter reaction times.

Tang et al showed that the phase separation temperature increases as the pressure is
lowered [80]. At 60 atmospheres, the phase separation temperature of 50:1 molar ratio of
methanol:oil is listed as 200 C, as opposed to 175 C, as mentioned earlier for 80
atmospheres at 40:1 molar ratio. This could be a reason why, at 175 C and 150 C, the
rate constants corresponding to lower saturation pressures (20 atm and 12 atm) are lower
than the rate constants corresponding to 1070 psi (73 atm). They also showed that the
phase separation temperature is lowered as the molar ratio is decreased.

79

Another point of interest was whether there was any spike in rate constant as the critical
temperature of ethanol is approached. As shown in Figure 6.3, the Arrhenius plot had a
constant slope from 175 C to 260 C. There was no sudden increase in rate constant as
the temperature approached 240 or 260 C (critical temperature of ethanol: 247 C). This
suggests that there was no special enhancement in the reaction regime with respect to rate
constant, due to a supercritical reaction mixture. It seems that the reaction regime around
260 C, 70 atm is just an extension of the miscible phase region that started at 175 C.

In a collaborative project funded by USDA Small Business Innovation Research (SBIR)


program, Greg Austic of Piedmont Biofuels carried out a variety of experiments using
methanol with the same MnO catalyst method [81]. (Methanol, though toxic, is much
cheaper than ethanol). He carried out experiments at higher temperatures, 280 C and
300 C, also. The conversions at those higher temperatures matched with the
aforementioned rate constant expression. This is a further confirmation of the rate
constant expression.

The Arrhenius equation is


k = k0 e

Ea
RT

It can also be expressed as

E
ln k = ln k0 - a
RT
where k is the rate constant (min-1),
k0 is the prefactor (min-1),

80

Ea is the activation energy (J mol1),


R is the gas constant (8.314 J K1 mol1),
T is the absolute temperature (K).
Activation energy is the energy necessary for a chemical reaction to happen. It represents
the energy barrier that must be overcome to transform the reactants in one energy state
into the products in a different energy state. The rate constant, k, represents the number
of collisions which result in a reaction per second. The prefactor, k0, represents the total
number of collisions (leading to a reaction or not) per second. The exponential term,
e E a / RT , represents the probability that any given collision will result in a reaction. By
using the linear Arrhenius plot shown in Figure 6.3, the values of prefactor and activation
energy were calculated and are reported below. The y-intercept of the Arrhenius plot
gives the value of the logarithmic prefactor, and the slope represents the value of -Ea/R,
from which activation energy can be calculated.
k0 = 45,142 min1
Ea = 53.19 kJ mol1

81

Surface area and activity:


BET surface area:
The BET surface areas of the catalysts identified from batch tests were measured and are
listed below.
MnO (200-60 mesh) : 1.2639 m2/g
CaTiO3 (35-20 mesh) : 5.2383 m2/g
CaTiO3 fine

: 5.1817 m2/g

CaO-Al2O3

: 1.4338 m2/g

MnO (35-20 mesh)

: 1.6167 m2/g

TiO (+50 mesh)

: 1.5839 m2/g

CuO

: 13.1700 m2/g

The BET surface area measurements were done in the Materials Characterization
Laboratory at Penn States Material Research Institute by Magda Salama.

Of all these identified catalysts, MnO has the lowest surface area. Yet, MnO proved to
be the best catalyst from the packed-bed tests. This is an indication of the superiority of
its activity for the studied biodiesel reaction. CaTiO3 and CuO catalysts have 5 times and
13 times more surface area than MnO catalysts. CaO-Al2O3 and TiO have 13 % and 25
% more surface area than 200-60 mesh MnO. 35-20 mesh MnO that was granulated from
200-60 mesh sample ground to a finer sample by a ball mill shows 28 % more surface
area.

82

The surface areas of all the identified catalysts are lower than the typical high surface
area (50~100 m2/g) of catalysts that are specially prepared. This suggests that the surface
areas reported above are primarily external surface area. MnO has an internal crystal
lattice structure of the Halite NaCl rock salt, where the cations and anions are
octahedrally coordinated [82]. See Figure 6.4. The lattice structure means that neither
the gas used in the BET analysis nor the reactants are small enough to enter the interstitial
atomic space. This means that the surface area calculated cannot correspond to the
interstitial atomic surface area found in zeolites. So the measured surface area has to
correspond to external (non-interstitial) pores such as cracks, mesopores and macropores
that may be associated with the manufacturing process.

Figure 6.4. Rock salt crystal structure with octahedral geometry [83]

CaO and TiO also have rock salt crystal structure [84]. So their surface areas also
correspond to non-interstitial external pores. CuO has a monoclinic mS8 crystal structure

83

that has slightly more interstitial space [84, 85], but it is still not large enough for either
the BET analysis gas molecule or the reactant molecules to get inside.

Cumulative pore area over pore size range:


Figure 6.5 shows the cumulative pore area of the catalysts tested over the entire range of
pore size. The 200-60 mesh MnO widely employed in the present research has the lowest
surface area over the entire range of the pore size. Not only does it have the least overall
surface area, but it has the least surface area in all the intermittent pore size ranges. So,
irrespective of the mesoporous and macroporous nature of the catalysts involved, the
200-60 mesh MnO is found to have the least surface area and the best activity. This is
evidence in support of the superiority of the activity of MnO for this biodiesel reaction.
MnO has also been tested for activity over a much longer time, up to 500 hours, while the
other catalysts run times were in the range 10~40 hours.

Though MnO is a base, CaO and TiO are stronger bases than MnO. This takes us back to
the points mentioned in Chapters 2 and 4 regarding what an effective catalyst is. The
strongest base is not necessarily the best catalyst. A solid catalyst that is too strong of a
base for the reaction may hold up the product molecule from desorbing after the reaction.
A solid base that is too weak may not adsorb the reactant to begin with. So there has to
be an optimal strength of the basic sites that is the best for a given reaction. MnO has the
most optimal strength of the catalytic site required to catalyze this biodiesel reaction.

84

13

12

11

MnO (200-60 mesh)


CaTiO3 (35-20 mesh)

10

CaTiO3 fine
CaO-Al2O3

CuO
MnO (35-20 mesh)

Pore area (m/g)

TiO (+50 mesh)

0
10

100

1000
Pore diameter ()

Figure 6.5. Cumulative pore area of catalysts over pore size range
85

10000

Comparative acidity/basicity scale:


Jeong et al listed and compared the acidity of various metal oxides [86]. The comparison
was based on relative acidity, meaning that it can be inferred for relative basicity as well.
They devised a comparative scale considering the partial charge of the metal ion, the
formal oxidation state of the metal in the compound, and the Sandersons
electronegativity of the metal. The values following in parentheses are the relative
acidities, so the lower the value, higher is the basicity. CaO (1.10) and MgO (1.27) are
listed to be more basic than MnO (1.40). TiO (0.98) is also more basic than MnO. In
fact, TiO is more basic than CaO and MgO. CuO (1.52) is less basic than MnO (1.40).
For some of the metal oxides of interest, the scale rates the acidity as shown in Figure
6.6.

NiO, CoO, CuO, ZnO and Al2O3 are less basic than MnO. TiO, CaO and MgO are more
basic than MnO. So this supports the earlier assertion that MnO is not the most basic
catalyst, but it seems to have the perfect activity for the reaction being studied. CuO and
TiO have better activity than the other metal oxides tested and lie in between MnO and
themselves. It may be because there are more than one optimal region for activity, and
MnO has the most optimal region of the various optimal regions. It is also possible that
some metal oxides like VO that are close to TiO may indeed have the requisite activity,
but the tested sample may have been inactive for some reason. So there may be some
uncertainties with some oxides. But, there are strong hints overall that MnO has the most
optimal activity.

86

1.6

2.4

ZnO

CuO
NiO

1.5

CoO

2.3

3.2

4.0

3.1

3.9

Al2O3
TiO2

MnO
FeO

1.4

Relative acidity.

1.3

2.2

3.0

3.8

2.1

2.9

3.7

MgO

MnO2

1.2

1.1

CaO

1.0

2.0

2.8

3.6

1.9

2.7

3.5

1.8

2.6

1.7

2.5

3.3

1.6

2.4

3.2

3.4

ZrO2

TiO
VO
BaO

0.9
La2O3

0.8
0

Figure 6.6. Relative acidities of metal oxides

87

Thermogravimetric analysis of temperature-programmed CO2 desorption:


To understand more about the basic strength of the catalyst sites and the related activity
for the reaction, thermogravimetric analysis (TGA) of temperature-programmed CO2
desorption was conducted on samples of interest. Temperature programmed desorption
of CO2 that was adsorbed on solid bases gives an idea of the strength of the basic sites.
CO2 is a common probe to test the basicity of a solid catalyst. Samples are first placed in
CO2 atmosphere to adsorb the CO2. The amount of CO2 adsorbed per unit mass of the
catalyst is a measure of the number of basic sites on the catalyst surface per unit mass.
But it does not give any information on the strength of various sites. The samples are
heated to higher temperatures with programmed temperature ramp to desorb CO2. CO2
desorption is tracked by monitoring the change due to the loss of CO2 mass
gravimetrically. The basic sites that are stronger need to be heated to higher temperatures
to desorb the CO2 molecules adsorbed onto them. So, the higher the desorption
temperature, stronger is the basicity of the site. Loss of mass plotted against temperature
also provides the number of sites of particular strength through the corresponding area
under the curve. The analysis was carried out in the Materials Characterization
Laboratory at Penn States Material Research Institute by Magda Salama. The analytic
procedure followed by Klepel and Hunger was adopted [87].

Figure 6.7 shows the temperature-programmed CO2 desorption profile. The catalyst
samples were placed under 1 atm CO pressure for adsorption. They were then heated up
to 1000 C to desorb the adsorbed CO2. The loss of mass was measured continuously.
Mass of the samples from the initial room temperature was plotted against the

88

corresponding temperature in Figure 6.7. The analyzed catalysts are listed below in the
order of CO2 % desorbed (indicated in parentheses) from each of them:
MnO (0.018 %) < FeO (0.046 %) < NiO (0.153 %) < CaTiO3 fine (0.432 %) < CaTiO3
(35-20 mesh) (0.575 %) < TiO (0.553 %) < CaO-Al2O3 (1.162 %)

The MnO catalyst adsorbed the least amount of CO2 to desorb, and CaO-Al2O3 had
adsorbed the most amount of CO2 to desorb. CaO-Al2O3 shows a steep desorption profile
up to 300 C. Thereafter, the profile becomes more moderate. From 700 C, the profile
becomes flat, with no CO2 mass loss after that. The basic sites that desorbed CO2 before
300 C are weaker compared to the sites that desorbed CO2 in the 300~700 C range.
MnO desorbs CO2 only till 200 C. Both the CaTiO3 samples show similar profiles, and
the masses of CO2 desorbed were close. However, they continue to desorb CO2 after 700
C, even if the amount desorbed is less compared to the earlier ranges. These results
suggest that CaTiO3 samples have basic sites that are even stronger than those of CaOAl2O3. CaTiO3, thus, has stronger basic sites than CaO-Al2O3, but has less number of
sites than CaO-Al2O3.

After MnO, FeO has desorbed the least amount of CO2 and the desorption continues only
till 450 C. FeO looks the closest to MnO among the desorption profiles and it is also
close to MnO in the relative acidity scale plotted by Jeong et al [86]. But, as seen from
the experimental results, the difference in the activity for transesterification reaction is
still considerable.

89

100.1

100.0

99.9

99.8

99.7

Weight (%)

99.6

99.5

99.4

99.3

99.2

MnO (200-60 mesh)


CaO-Al2O3

99.1

CaTiO3 fine
CaTiO3 (35-20 mesh)

99.0

NiO
FeO

98.9

TiO (+50 mesh)


98.8
0

100

200

300

400 500 600 700


Temperature (Celsius)

800

900

1000

Figure 6.7. Thermogravimetric analysis of temperature-programmed CO2 desorption

90

The amount of CO2 desorbed by the TiO (+50 mesh) sample lies in between those of
CaTiO3 and CaO-Al2O3, and the desorption profile has a steady slope. However, the
desorption stops after 550 C. This means that the range of site-strength is less than that
of CaO-Al2O3.

NiO is at the lower end when it comes to the amount of CO2 desorbed. The amount of
CO2 desorbed by NiO is greater only than those of MnO and FeO, and it is considerably
less than that of CaTiO3, the next highest. But it continues to desorb CO2 even after 1000
C. This suggests that NiO has a higher range of catalyst site-strength, even though the
amount of CO2 desorbed is much less than that of TiO and CaO-Al2O3.

MnO has shown the best activity of all the tested catalysts for the transesterification
reaction. MnO was found to have the lowest surface area, and this was reiterated by
MnO adsorbing the least amount of CO2. Also, its basic sites were not the strongest.
However, other catalysts that are weaker bases than MnO, such as ZnO, CoO, Al2O3,
CaCO3 and TiO2, did not show good activity for the reaction either. This suggests that
the reaction has the best activity when the basic strength is optimum. The strongest base
is not the best catalyst for the reaction in this reaction regime. As discussed in Chapter 4,
a catalyst that is too strong may not desorb the product, and a catalyst that is too weak
may not adsorb the reactant. A solid catalyst needs to have the optimal strength for the
given reaction. Thus, it can be concluded that MnO shows the best activity for the
reaction because it has the most optimal basic strength.

91

It can also be inferred that the reaction has more than one optimal point of activity for the
reaction, although MnO has been the most optimal. The amount of CO2 adsorbed by TiO
was in between those of CaO-Al2O3 and CaTiO3, and the strength of the catalytic sites
was also in between those of CaO-Al2O3 and CaTiO3. The surface area of TiO was less
than that of CaO-Al2O3 and more than that of CaTiO3. This suggests the existence of
optimal activity in other ranges. Also, as seen earlier from the order of basicity/acidity in
Figure 6.6, TiO and CaO show good activity, unlike La2O3 and VO, which have higher
basicity, and MgO and FeO, which have lower basicity. MnO shows good activity,
unlike MgO and FeO, which have higher basicity, and NiO and CoO, which have lower
basicity. Similarly, the basicity of CuO is lower than NiO and CoO, and higher than ZnO
and Al2O3.

Another point of observation from Figure 6.7 is the effect of higher basic sites on the
activity of lower basic optimal sites. It can be seen from Figure 6.7 that even though
CaO-Al2O3, CaTiO3, FeO and NiO have sites of higher basic strength than MnO, they
also have sites with the same basic strength as MnO. But those sites with the same
strength did not result in the same activity as that of MnO. The sites of higher basic
strength could have a negative effect on the activity shown by the sites of optimal
strength. The reactants still continue to adsorb on to the sites of higher strength while
being adsorbed on the sites of optimal strength. For example, if the concentration of
reactants is 1 mmol/l and the fraction of higher basic sites is four times that of optimal
sites, the effective concentration for the optimal sites could be reduced by up to 0.25

92

mmol/l and result in a lower reaction rate. Thus, the sites of higher basic strength likely
have a negative effect on the optimal activity of the overall reaction rate.

Rate constant vs surface area per unit volume:


MnO was ground to finer particles before being agglomerated to a larger 35-20 mesh size
by J.M. Huber personnel. This finer sample was also tested in a packed-bed reactor. The
catalyst was run for 40 hours before collecting data to allow for any initial spike in
activity to settle down. The surface area of this finer MnO sample was 1.62 m2/g,
compared to the 1.26 m2/g surface area of the normally used 200-65 mesh MnO. When
packed into a inch O.D., 0.18 inch I.D. reactor, the packing fraction corresponding to
the finer MnO sample was 0.55. The packing fraction of 200-65 mesh MnO used in a
similar reactor normally was 0.40. The rate constant of a packed-bed reactor depends on
the surface area per unit volume of the reactor.

Mass of catalyst per

Surface area per


Surface area per

=
x

unit volume of reactor


unit volume of reactor
unit mass of catalyst
The surface area per unit volume of the finer MnO sample is higher than that of 200-65
mesh MnO.

The rate constant is known to increase linearly with the surface area of a catalyst per unit
volume, and the slope decreases and gets saturated with the catalyst later [88, 89, 90].
After certain degree, the concentration (packing density) of the catalyst becomes so high
that it becomes saturated with respect to the concentration of the reactants. The rate
constant observed for the finer MnO sample was 0.354 min-1, and this has been the
maximum value observed in the present research while using MnO. The rate constant
93

corresponding to the 200-65 mesh MnO was 0.277 min-1. The rate constant without
catalyst is 0.0244 min-1. The fine MnO sample corresponded to 1.77 times the surface
area per unit volume of 200-65 mesh MnO. However, the rate constant of the fine MnO
sample corresponded to only 1.30 times the rate constant 200-65 mesh MnO after
negating the rate constant without catalyst. The rate constant increased beyond the rate
constant of 200-65 mesh MnO, but the increase was not in proportion with the increase in
surface area per unit reactor volume. It was 26 % below the expected linear proportional
increase. So any effort to increase the rate constant has to consider the additional
maintenance issues (frequent loss of activity and regeneration) of the highly custommade catalyst.

Proposed reaction mechanism:


The reaction mechanism is important to understand the catalytic properties, to design and
evaluate the catalyst particles and the type of loading, and to evaluate different reactants.
While using homogeneous base catalysts such as sodium hydroxide, the reaction
mechanism is as illustrated below. Such a base catalyzed mechanism, whether it is
homogeneous or heterogeneous, is a nucleophilic substitution reaction.

CH3OH + NaOH  CH3ONa + H2O


CH3ONa

CH3O-- + Na+

94

Methanol reacts with the base catalyst, sodium hydroxide to form sodium methoxide [1].
The negatively charged methoxide ion from sodium methoxide attacks the carbon atom
double bonded to oxygen and slightly positively charged due to polarization of double
bond electrons toward the more electronegative oxygen atom, and forms a tetrahedral
structure. This tetrahedral structure is unstable, and the carbon atom would much rather
revert back to a carbonyl double bond with the oxygen atom and give up the electrons to
one of the oxygen atoms of two alkoxides (methoxide and glycoxide (of diglyceride)).
The electrons go back to the diglyceride glycoxide as methoxide is a much stronger
nucleophile. This results in the formation of a methyl ester biodiesel molecule.

Here the methoxide ion is a nucleophile. During the reaction, the methoxide nucleophile
substitutes the glycoxide (of diglyceride in the above example, or monglyceride and

95

glycerol in the subsequent steps), converting a glyceride fatty acid ester to methyl ester.
Hence, the name of the reaction mechanism is nucleophilic substitution reaction.

Solid acid and solid base catalysts can either by Lewis or Bronsted type. Lewis acids or
bases catalyze by abstracting or donating electrons. Bronsted acids or bases catalyze by
donating or abstracting protons. In the following example, BF3 is a Lewis acid.
BF3 + F BF4
Zirconia is another example of a Lewis acid. HCl is an example a Bronsted acid. OH
and SO42 are examples of Bronsted bases. MnO and other catalysts identified in the
present research, such as TiO and CaTiO3, are Lewis bases. MnO catalyzes by donating
electrons through its oxygen atom, which is negatively polarized.

The heterogeneous catalyst mechanism is illustrated in Figure 6.8. R denotes long alkyl
chain. Let us consider methanol as the reacting alcohol for the purpose of easy notation,
and the reaction mechanism is similar for ethanol. The oxygen part of manganese oxide
is electron rich and likes to donate electrons. Methanol is a weak acid and has a tendency
to give up its proton. The basic site of MnO abstracts the proton from methanol by
donating its electrons. Much like how the electron cloud was polarized toward the
oxygen atom in MnO, the electrons are polarized toward the oxygen atom in methanol
now, as the proton has been strongly abstracted by the solid catalyst. This means that the
oxygen atom in methanol is looking to attack any electropositive site in the vicinity.
Through this oxygen atom, the methanol molecule behaves as a methoxide nucleophile,
like the free methoxide ion in the homogeneously catalyzed mechanism illustrated earlier.

96

When a triglyceride molecule approaches the methanol molecule adsorbed onto the solid
catalyst, the nucleophilic methoxide of methanol attacks the partially positively charged
carbon atom of the carbonyl bond of the triglyceride. The carbon atom then forms an
intermediate structure, as shown in Figure 6.8, giving up one of its laterally bonded
electron pair back to the oxygen atom of the original carbonyl bond. This intermediate
structure is unstable and it wants to revert back to a normal structure. It tends to form the
carbonyl bond back with the same oxygen atom and give up the pair of electrons of one
of the bonds it has with oxygen atoms that form the bridges to methyl group and glycerol
backbone. The electropositive carbon would rather give up the glycoxide (glyceride
chain) than the methoxide, since methoxide is a stronger nucleophile than glycoxide.
Thus, by forming the carbonyl bond back and by giving up the glyceride chain, the
electropositive carbon makes a biodiesel methyl ester. The negatively polarized oxygen
of the methoxide group had lost its hold on the proton adsorbed on the catalytic site
earlier, when it bonded with the carbonyl carbon of the triglyceride. The glycoxide ion
just set free by the carbonyl carbon attaches to the proton in the place of the earlier
methoxide group. The biodiesel methyl ester molecule has migrated away by now. The
glycoxide group that is attached to the proton is now a diglyceride adsorbed to the
catalyst. A new methanol molecule then replaces the diglyceride molecule and the cycle
continues.

97

MnO
-

CH3
|
O|
H

MnO
-

O
||
H2COCR
|
RC OCH
|
||
O H2COCR
||
O
-

O
||
H2COCR
|
RCOCH
||
|
O H2COCR
||
O

CH3O

CH3OH

MnO

CH3
|
O|
H

MnO

O
||
H2COCR
|
RCOCH
|
|
O H2COCR
||
O

RCOCH3
||
O

MnO

CH3O|
H

MnO

O
||
H2COCR
|
RCOCH
|
||
O H2COCR
||
O
+

O
||
H2COCR
|
HOCH
|
H2COCR
||
O

O
||
H2COCR
|
RCOCH
|
|
O H2COCR
||
O
CH3O

CH3
|
O|
H

MnO

CH3OH

Electron donating
Lewis base site

Solid catalyst

O
||
H2COCR
|
OCH
|
H2COCR
||
O

CH3O

MnO

RCOCH3
||
O
O
||
H2COCR
|
OCH
|
H2COCR
||
O

MnO
-

Figure 6.8. Heterogeneous catalytic reaction mechanism of transesterification

MnO

CH3OC=O
|
R

MnO

RCOCH3
||
O
O
||
H2COCR
|
OCH
|
H2COCR
||
O

Thus, the kinetics of the chemical reaction catalyzed heterogeneously by MnO catalyst
was modeled, after making sure it was not affected by external and internal mass transfer
limitations. The reaction followed pseudo first order kinetics. The reaction was
validated by the Arrhenius plot. Decrease in the rate of the reaction due to phase
separation was inferred from the experiments conducted at lower temperatures. The
superior effectiveness of MnO over other catalysts has been substantiated from studies
relating surface area to activity. A reaction mechanism of the heterogeneously catalyzed
transesterification reaction has been proposed.

99

Chapter 7

Studies for Commercialization

This chapter focuses on studies that have commercial implications. The features of the
method that are important to establish the commercialization and scale-up feasibility have
been discussed. Studies optimizing process parameters for commercial production have
been performed. Robustness of the method to handle a variety of feedstock has been
evaluated. Considerations on optimal reaction temperature and pressure have been
discussed.

Activity sustenance tests:


Identifying solid catalysts that show promising activity in short term tests is only a start.
The commercial success of the catalyst would be dictated by how long the catalyst can
remain active. A commercial packed-bed reactor requiring catalyst change every day or
week would make the process non-viable. Catalyst life would have commercial
implications more than any other feature of the method. The relevant questions are: How
long can the catalyst remain active? Is it possible to regenerate its activity if it loses
activity? If yes, how often would it need regeneration and how easy would this be?

Tests were conducted to study the longevity of the catalyst activity. Performance of the
catalyst was checked for deterioration over time. MnO catalyst was packed in a reactor
and a residence time of 5 minute was chosen. This residence time was chosen because
any change in catalyst activity is most pronounced around this residence time. See

100

Figure 7.1. At longer residence times, the conversion lines of various rate constants
become closer. At shorter residence times, the conversion lines of various rate constants
just start to branch out.

Difference in conversion (%) .

30
k1-k2
k1-k3

25

k1-k4
20
15

k1 = 0.125 min-1
k2 = 0.175 min-1

10

k3 = 0.225 min-1
k4 = 0.275 min-1

5
0
0

5
10
Residence time (min)

15

Figure 7.1. Differences in conversion with change in rate constant

Experimental details:
200-60 mesh MnO purchased from Alfa Aesar was loaded onto a stainless steel reactor
(1/4 inch O.D, 0.18 inch I.D., 22 cm long). Reactor volume was 3.61 ml. Considering
MnO particle density of 5.445 g/cc, the observed packing fraction was calculated to be
40.19 %. Void volume of the reactor was 2.16 ml. Flow rate was set at 0.43 ml/min, as
this corresponds to a residence time of 5.026 min. Commercial Wesson soybean oil and
ethanol were the reactants. Alcohol:oil molar ratio of 36.2:1 (69:31 volume ratio) was

101

adopted. Samples were collected at various time intervals and analyzed in the gas
chromatograph.

Typically catalyst activity sustenance tests are run up to 100 hours [70]. Figure 7.2
shows that the catalyst activity remains stable for over 200 hours. The same reactor was
used for various other studies and proved to have sustained the activity for 500 hours.
Since the residence time adopted was 5 minutes, a run of 1 hour would equate to 12
cycles. So 200 and 500 hour runs correspond to 2400 and 6000 cycles. These are
promising numbers for commercial applicability.

Figure 7.2 shows that activity of the catalyst shows a spike initially, and it stabilizes after
10 hours. The initial spike is regarded to be typical of solid catalysts. Since the sample
used was established earlier to have its surface area mainly externally, the explanation for
the initial spike could be one of the following: (1) Some of the surface area of sharper
nature was rounded by erosion, with possible leaching of some fresh catalyst. (2) Some of
the active sites on the defect lines and corners (probably even stronger sites) of the fresh
catalyst restructured shortly after being used and that activity was lost. (3) Strength of the
fresh active sites was stabilized by being thoroughly adsorbed by reaction components.

It was noted during longer runs around 300~400 hours that periodic flushing with 100%
alcohol tended to pick up any little drop in activity over a period of time. This suggests
that alcohol flushing helps overcome any clogging/by-product adsorption over time (eg.:

102

oil-related by-products due to side reactions like FFA formation, saponification and
polymerization, and glycerol adsorption).

Biodiesel conversion (%) .

100
90
80
70
60

260C, 70 atm

50
40

MnO (200-60 mesh)

30
20

Residence time = 5 min

10
0
0

20

40

60

80

100

120

140

160

180

200

Catalyst run time (hour)

Figure 7.2. Activity sustenance of MnO solid catalyst

During the batch tests, TiO catalyst was reused for two subsequent runs and the results
showed no decrease in activity. However, the results of TiO loaded onto a packed-bed
reactor were inconclusive. The catalyst retained its activity for the first 5 hours. After 5
hours, TiO catalyst from one of the reactors showed significant decrease in activity (82 %
conversion in 20 minutes, compared to 94 % in 17 minutes earlier), whereas the catalyst
from another reactor showed only marginal decrease in activity (91 % conversion in 19
minutes). Greg Austic in the USDA-SBIR collaborative project reported a conversion of
93 % in 10 minutes after 30 hours of run time and 87 % in 10 minutes after 60 hours,
employing methanol [81]. TiO results were inconclusive and need more investigation to

103

understand its activity sustenance. But MnO showed better activity and better activity
sustenance, and it was decided to go ahead with MnO for further tests.

Optimal molar ratio:


For the conventional process employing soluble alkali catalyst, 6:1 molar ratio is
considered to be the optimal molar ratio. The reaction is a second order reaction.
However, Saka and Kusdiana and other researchers have reported that 40:1 molar ratio is
the optimal molar ratio for supercritical and near-critical reaction conditions [67, 68, 91].
Under these conditions, the reaction is a pseudo first order reaction and the excess
alcohol is considered to have a direct effect on the rate constant of the reaction. 40:1
alcohol:oil molar ratio is reported to be the ratio where rate constant of the reaction
reaches saturation. The researchers have reported an increase in rate constant with molar
ratio until 40:1. In a high temperature, high pressure commercial reactor, reaction space,
reactor size, heat loss and catalyst amount are at a premium. With a lower molar ratio,
the rate constant might be lower, but it would be possible to process more reactants in a
given reaction space. That is, biodiesel productivity might be higher even though the rate
constant would be lower. In simple terms, it is a matter of balancing the better medium
of interaction provided by the excess alcohol with reaction space. In a commercial
reactor, this is reflected by Liquid Hourly Space Velocity (LHSV) and Weight Hourly
Space Velocity (WHSV) in the case of solid catalyst process. See Appendix E for LHSV
and WHSV definitions. WHSV gives the number that matters for commercial
production: mass of biodiesel produced per hour per unit mass of the catalyst.

104

Experimental details:
The same reactor prepared for activity sustenance tests was continued to be used for this
series of experiments. Reaction temperature was 260 C. Flow rate of 0.43 ml/min was
maintained for 5 minute residence time. Ethanol:oil molar ratio of 36.2:1 was
maintained. Reactor run time corresponding to these experiments is from 470 to 500
hours.

The top of Figure 7.3 shows biodiesel conversion after 5 min for various molar ratios,
and the bottom of the figure shows the corresponding rate constants. The conversion
keeps on increasing progressively from 5:1 to 30:1, after which it starts to settle down.
This matches well with the literature reports that 40:1 molar ratio gives the best
conversion. The trend is reflected in the rate constant graph as well. For a commercial
reactor giving 95% yield, WHSV calculated for the rate constants of various molar ratios
is also given in the lower graph. WHSV increases along with rate constant up to 15:1
molar ratio, after which it stops climbing, and starts to fall down from 30:1 molar ratio.
WHSV corresponding to 40:1 molar ratio is similar to that of 12:1. An optimal molar
ratio, based on WHSV, seems to lie around 20:1 to 30:1.

Molar ratio may still have an effect on equilibrium conversion. If that is the case, it may
be required to introduce excess alcohol just for the final part of the reactor to push the
equilibrium. By this way, most of the conversion would have been performed with lower
molar ratio, saving reaction space. This was to be investigated later.

105

80
Biodiesel Conversion (%) .

70
60
50
260C, 70 atm

40
30

MnO (200-60 mesh)

20
Residence time = 5 min

10
0
0

10

20

30

40

50

0.3

10

0.24

0.18

6
4

Rate constant

0.12

WHSV
0.06

0
0

10

20

30

40

50

Ethanol:vegetable oil molar ratio

Figure7.3. Effect of molar ratio on MnO solid catalyzed biodiesel reaction

106

WHSV

Rate constant, k (1/min) .

Ethanol:vegetable oil molar ratio

Effect of pressure:
Researchers that worked on supercritical alcohol process have mentioned that high
pressure (500 atmospheres) is crucial to get fast reaction times at high temperatures such
as 380 C. It was not clear, however, whether high pressure is as crucial around nearcritical temperature (247 C for ethanol). One of the main reasons to conduct the reaction
around the critical temperature of alcohol is to have a miscible reactant mixture of
vegetable oil and alcohol. Pressures around critical pressure of alcohol (70 atmospheres
for ethanol) would render the reactant mixture miscible. It was desired to see what effect
additional pressure has in this regime with solid catalyst. Pressure has crucial cost
implications in a commercial process.

Along the same lines, is it really important to even be at the critical pressure? One of the
reasons it is desired to have pressures above critical pressure around critical temperature
(or saturation vapour pressure at any temperature) is to maintain the alcohol in liquid
phase. It is generally considered that a liquid phase reaction is more effective than a
vapour phase reaction (vapour phase alcohol with liquid phase oil). Would the vapor
phase alcohol reaction be effective enough? That is, even if the rate constant is lower for
vapour phase reaction, would it still be comparable considering that there is no pressure
requirement?

How does the rate constant of the solid catalyst reaction vary with pressure? To
understand this, a series of experiments were performed with an MnO packed-bed reactor

107

at various pressures. Again, a 5 minute residence time was chosen, since it is the optimal
time range to note decrease in catalyst activity.

Experimental details:
The reactor prepared for activity sustenance tests was continued to be used for this series
of experiments. Reaction temperature was 260 C. Flow rate of 0.43 ml/min was
maintained for the 5 minute residence time. Ethanol:oil molar ratio of 36.2:1 was
maintained. Outlet pressure was set at atmospheric pressure, 400 psi, 750 psi, 1100 psi
and 1450 psi. Reactor run time corresponding to these experiments is from 500 to 510
hours.

Figure 7.4 shows biodiesel conversion and corresponding rate constant at various
pressures. The transesterification reaction does take place even at atmospheric pressure,
albeit four times slower than that at 70 atmospheres. As the pressure is increased, the rate
constant of the reaction increases linearly. But this linear increase happens only up to the
critical pressure (or the saturation vapor pressure for lower temperatures). This is the
minimum pressure needed to keep the reactant mixture in liquid phase instead of vapor
phase. Any increase in pressure after this, does not show the same increase in rate
constant as seen in the vapor phase region. The rate constant almost remains constant for
any pressure increase after the critical pressure. This indicates that we only need a
pressure high enough to keep the reactants in liquid phase, and any excess pressure does
not make much difference. This finding differs from the findings of catalyst-free
supercritical alcohol research that showed significant dependence on pressures much

108

Biodiesel conversion (%)


..

80
70
60
50
40

260C

30

MnO (200-60 mesh)

20

Residence time = 5 min

10
0
0

250

500

750

1000

1250

1500

1250

1500

Outlet pressure (psi)

Rate constant, k (1/min).

0.3
0.24
0.18
0.12
0.06
0
0

250

500

750

1000

Outlet pressure (psi)

Figure 7.4. Effect of pressure on MnO solid catalyzed biodiesel reaction

higher than 70 atmospheres. This may suggest that the solid catalyst employed accounts
for the potential for increase in activity related to higher pressures as well. The results
also indicate that vapor-phase reaction would have a considerable compromise on the rate

109

constant. The reaction at atmospheric pressure and half the critical pressure is 4 times
and 2 times slower than the reaction at critical pressure.

The negligible effect of pressure increase after critical pressure (or saturation vapor
pressure) is also seen in the lower temperature reactions discussed in Chapter 6. When
the reaction was conducted at lower temperatures, two outlet pressures were tested; one
keeping the 70 atmosphere pressure (~ 1070 psi) constant for all the lower temperatures,
and the other which is the saturation vapor pressure corresponding to each temperature
(the pressure above which a component becomes a liquid). Outlet pressures set to
saturation vapor pressures for the various lower temperatures listed in Figure 7.5 are
given in Table 7.1.

Temperature

Saturation vapor pressure

Outlet pressure

(C)

(psi)

(psi)

260

980

1070

240

877

900

220

625

650

200

431

450

175

257

300

150

142

175

Table 7.1. Outlet pressures set for reaction at lower temperatures

110

Figure 7.5 shows the biodiesel conversion at various temperatures for 5 minute residence
time and Figure 7.6 shows the Arrhenius plot of logarithmic rate constant vs the inverse
of absolute temperature. The figures indicate that there is almost no difference between
1050 psi and the respective lower saturation vapor pressures, at least for the certain
miscible regions above 200 C. Below 200 C the difference is still very marginal and
Tang et al [80] show miscibility can improve with increase in pressure in immiscible
regime. So it can be concluded that pressure increase above the saturation vapor pressure
does not give any appreciable improvement for the MnO solid catalyzed reaction.

80

Biodiesel conversion (%) .

70

1070 psi
Saturation vapour pressure

60
50

MnO (200-60 mesh),


Residence time = 5 min

40
30
20
10
20 min

0
50

100

150

200

250

300

Temperature (C)
Figure 7.5. Biodiesel conversions at various temperatures using 200-60 mesh MnO

111

0.0017
0.0

0.0019

0.0021

0.0023

0.0025

0.0027

260 C

-1.0

240 C
220 C

-2.0

200 C

ln k

-3.0

36:1 Molar ratio


175 C

-4.0

150 C

-5.0
-6.0
-7.0

k at 1070 psi
100 C

k at saturation vapour pressure


-8.0
1/T (1/K)

Figure 7.6. Arrhenius plot of the transesterification reaction using 200-60 mesh MnO

Free fatty acid conversion:


Conversion of free fatty acids (FFA) is crucial in exploiting cheap feedstock like used oil
and animal tallow. Such cheap feedstock contain ~15 % FFA by weight. This amounts
to a significant portion of revenue in a commercial biodiesel facility.

It was seen earlier in the batch tests that TiO converted 93 % of a pure FFA sample (oleic
acid) to biodiesel esters in the same reaction time adopted for triglycerides. It was

112

decided to do the same evaluation in the packed-bed reactor loaded with 200-60 mesh
MnO. The hypothesis is that the metal oxides identified during the batch tests that
catalyze FFA conversion, are amphoteric in nature. They act as mild solid base while
being surrounded by triglycerides, and as solid acid while being surrounded by free fatty
acids. These oxides are seen to be bases while catalyzing the transesterification of
triglycerides, but as acids while converting FFA as only acids can catalyze the
esterification of FFA.

Experimental details:
200-60 mesh MnO purchased from Alfa Aesar was loaded onto a inch OD, 8.48 mm
ID, 30 cm long stainless steel reactor. Reactor volume was 16.96 ml. Considering MnO
particle density of 5.445 g/cc, the observed packing fraction was calculated to be 38.8 %.
Void volume of the reactor was 10.38 ml. Flow rate was set at 1.38 ml/min to correspond
to a residence time of 7.5 minutes. Ethanol:oil molar ratio of 36.2:1 (69:31 volume ratio)
was adopted. Samples were collected at various time intervals and analyzed. Reactor run
time corresponding to these runs is 32 to 41 hours.

97 % of oleic acid was found to be converted to ethyl oleate biodiesel esters in 7.5
minutes. This means that esterification of oleic acid is twice as fast as the
transesterification of triglycerides. Not only does MnO catalyze esterification of FFA,
the reaction rate of esterification is even faster than transesterification. The analysis was
done by titrimetry (See Appendix D), and the results were also confirmed by gas
chromatography. Lopez et al showed that while using solid acids esterification of free

113

fatty acids is ~ 20 times faster than transesterification of triglycerides [62]. Minami and
Sakas data indicate that esterification of oleic acid is ~ 10 times faster than
transesterification of triglycerides without catalyst at high temperatures [74]. There have
been enough hints that esterification is a faster reaction than transesterification. 97 %
conversion of oleic acid in 7.5 minutes compared to the same triglyceride conversion in
twice the time proves that FFA esterification is faster than triglyceride transesterification
while using MnO as catalyst.

The metal oxide should act as a solid acid to catalyze FFA esterification. The electrons
polarized toward the oxygen atom (partially negative) of the metal oxide donate
themselves acting as base during transesterification. During FFA esterification, the metal
atom (partially positive) deficient of electrons polarized away from it, abstracts electrons
similar to the way the proton abstracts electrons as explained in the homogeneous acid
catalytic mechanism in Figure 2.3 in Chapter 2, and thus acts as a solid acid. Any metal
oxide can be theoretically viewed as amphoteric, the metal atom part can abstract
electrons and the oxygen atom part can donate electrons, but the degree differs. In the
case of MnO and TiO, it appears that the metal oxide tends to donate electrons while
being in an environment surrounded by triglycerides and tends to abstract electrons while
being in an environment surrounded by free acids. It appears that these metal oxides
have the optimal degree of amphotericity to do both, whereas, calcium oxide forms soap
with free fatty acids.

114

There was occasional clogging of the reactor lines while using oleic acid. This could be
because oleic acid has a higher gelling point than triglycerides. Brownish suspension was
seen sometimes. It was seen even before entering the reactor coming out of the pump.
Formation of some amount of soap cannot be ruled out as free acids have a tendency to
form soap with metal oxides that have a degree of basicity. Since the brownish
suspension was seen even before entering the reactor, it could also be due to physical
reaction or oxidation that has no effect from catalyst. Vegetable oils that undergo
oxidation decomposition during storage show similar brownish suspension.

In the USDA-SBIR collaborative project, Greg Austic carried out experiments using
soybean oil mixed with 15 % FFA to mimic the real world cheap feedstock [81].
Biodiesel conversions in the range of 98.65 ~ 99.16 % was observed consistently for a
reactor run time of up to 85 hours. Residence time was 10 minutes and a molar ratio of
30:1 was used. 172 ppm of Mn soap (2 ppm of Mn) was found in the biodiesel layer and
6036 ppm of Mn soap (527 ppm of Mn) was found in glycerin layer. This amounts to a
catalyst leaching of 0.009 % by mass in the output product and means that it would take
about 20 days for 1 % of packed catalyst to leach. This could be improved further with
future research. So it is clear that even though there is some soap formation, the rate is
very low (within ASTM limit) and consistent high conversions have been obtained for a
bed test time up to 85 hours. The experiments carried out by Austic also prove that the
reaction is significantly quicker with FFA. Greg Austic carried out experiments with the
real world yellow grease feedstock also, and the aforementioned results were confirmed.

115

The by-product while using oleic acid is water. The effect of water in biodiesel reaction
involving triglycerides is discussed in the following section.

Effect of water:
It was intended to see the effect of water in the biodiesel reaction mixture. Water can be
present in cheap feedstock like used oil. Water is also formed as by-product from the
reaction of FFA present in cheap feedstock. FFA may be present up to 15 % by weight in
cheap feedstock. Ethanol with 5 % water is considerably cheaper than pure ethanol if
intended to be used as a cheap reactant. To study the presence of water in the reaction
mixture as in the aforementioned scenarios, ethanol mixed with water at various mass
percentages up to 12 % was used for the reaction. The tests were carried out starting
from 30:1 ethanol:oil molar ratio as this was seen to be in the optimal range for
commercial process. This means that the tests were started from 30:1 ethanol:oil molar
ratio as the basis, and ethanol was substituted with ethanol-water mixture with mass
content varying up to 12 %. 30:1 ethanol:oil molar ratio equates to 1.58:1 mass ratio.
Feedstock with 15 % FFA would correspond to ~ 0.4 % water by weight in the reaction
mixture at full conversion. Therefore, ethanol with 12 % water by weight, which
corresponds ~ 7 % water by weight in the overall reaction mixture, covers a range that is
more than what could be expected in the real world reaction scenario.

Experimental details:
A inch OD, 4.57 mm ID, 31 cm long reactor loaded with 200-60 mesh MnO was used.
Reactor volume was 5.09 ml. Considering MnO particle density of 5.445 g/cc, the

116

observed packing fraction was calculated to be 42.3 %. Void volume of the reactor was
2.93 ml. Flow rate was set at 0.59 ml/min to correspond to a residence time of 5 minutes
which was found to be the most optimal to sense any change in the rate constant.
Ethanol:oil molar ratio of 30:1 was adopted to start with. Then experiments were
continued replacing the pure ethanol with ethanol-water mixtures with varying water
contents of 2.54 %, 6.12 % and 11.53 % (by weight). Samples were collected at various
time intervals and analyzed in the gas chromatograph. The samples were also analyzed
for any FFA formed due to hydrolysis by the presence of water by titration. Reactor run
time corresponding to these runs is 49 hours to 68 hours.

Figure 7.7 shows markedly increased conversion with water content initially compared to
that of reaction mixture devoid of water. The conversion increases up to 6 % water in
ethanol (3.6 % water in reaction mixture) after which it starts to decrease. It starts to fall
below the non-water conversion after 9 % water in ethanol (5.4 % water in reaction
mixture) and continues to decrease after that. The conversion has also been related to
rate constant and shown below in Figure 7.7. Compared to the rate constant
corresponding to no water content at 30:1 molar ratio, ethanol with 1.29 % water by
weight (0.78 % water in reaction mixture) shows 12 % increase in rate constant and
ethanol with 6 % water by weight (3.6 % water in reaction mixture) shows 29 % increase
in rate constant. This is a considerable improvement in rate constant. The rate constant
starts to decrease after that, falling below the original mark without water for ethanol
with 9 % water (5.4 % water in reaction mixture) and lowering by 12 % compared to the
original mark for ethanol with 12 % water (7 % water in reaction mixture). The free acid

117

content corresponding to different water content is also shown in Figure 7.7. The free
acid content is seen to be increasing with water content.

90
80

(%)

70
60

Free acid content

50

Biodiesel conversion

40

Residence time = 5 min


200-60 mesh MnO

30
20
10
0
0

Ethanol:Oil
30:1 Molar ratio
1.58:1 Mass ratio

12

12

Water wt. % in ethanol

0.35

Rate constant, k (1/min) .

0.3
0.25
0.2
0.15
0.1
0.05
0
0

Water wt. % in ethanol


Figure 7.7. Effect of water on MnO solid catalyzed biodiesel reaction

118

The increase in rate constant covers a range that is well above what could be encountered
in the reaction involving cheap feedstock.

The primary concern behind doing this test is to see if there is any negative effect on the
reaction due to water. But we ended up seeing significant increase in the rate of reaction
for a large range of water content, which for all practical purposes is more than any water
content that would be encountered in the real world scenario. So, not only does water not
affect the reaction negatively, it actually enhances the reaction significantly. We also
notice appreciable FFA content in the reaction mixture. This suggests that water has
caused hydrolysis of triglycerides to form FFA. This is a good hint that the reaction is
proceeding through an alternate pathway, hydrolysis of triglycerides forming FFA,
followed by esterification of FFA forming biodiesel esters. It was shown in the previous
section that esterification of FFA is twice as fast as transesterification of triglycerides.
All of these suggest that the hydrolysis of triglycerides is also faster than the alcoholysis
of triglycerides during transesterification, which makes the alternate pathway faster.

Normal pathway Transesterification by alcoholysis:


R-COOCH2
|
R-COOCH + C2H5OH
|
R-COOCH2

R-COOC2H5

119

HOCH2
|
+ R-COOCH
|
R-COOCH2

Alternate pathway Hydrolysis followed by esterification:


R-COOCH2
|
R-COOCH +
|
R-COOCH2
R-COOH

H2O

C2H5OH

HOCH2
|
R-COOCH
|
R-COOCH2

R-COOH +

R-COOC2H5

H2O

Lopez et al also mention that water helps in speeding up the reaction through
esterification of FFA produced by hydrolysis of triglycerides while using solid acids [62].
However, when the water content exceeds 6 % in ethanol, the concentration of water has
become too high, compromising the concentration of ethanol as a reactant and begins to
hinder the biodiesel ester formation. But this water content is too high to be encountered
in real world scenario.

Equilibrium conversion:
It was desired to see if the reaction can be pushed all the way to completion to meet the
ASTM specification on residual glycerin. Glycerin is termed to be of two types: free
glycerin and bound glycerin. Free glycerin refers to the glycerol formed after all the
three fatty acid chains of the triglyceride have reacted. Bound glycerin refers to the
glyceride backbone of mono-, di- and triglycerides. Bound glycerin includes the mass of
the glyceride backbone only, not the whole molecular weight of the long molecule. Total
glycerin is the sum of free and bound glycerin. ASTM specifications mention a limit of
0.24 % by weight for total glycerin and 0.02 % by weight for free glycerin [71]. There is
no need to push the reaction up to these limits (~ 99.15 %) in the reactor itself. A

120

separate adsorption or polishing unit at the end of the process can be used to make the
biodiesel meet the ASTM specifications after achieving a conversion of 95 ~ 98 % in the
reactor. However, if the reaction can be pushed to the ASTM limits in the reactor itself,
it would mean an even more efficient process that avoids the final adsorption unit and the
related processing energy. Free glycerin content has been found to be not a problem from
the lab tests, as the natural liquid-liquid equilibrium between the biodiesel and the
glycerol layer leaves only ~ 0.01 % glycerol by weight in the biodiesel layer.

It was decided to conduct some runs at longer residence times to see if the reaction can be
pushed all the way to reach ASTM specifications in the reactor itself.

Experimental details:
The three inch OD, 0.18 inch ID reactor segments used in the kinetic modeling in
Chapter 6 (30 cm, 29 cm and 29 cm long, and loaded with 200-60 mesh MnO) were used
in combination to get longer residence times. Total reactor volume was 14.45 ml. Void
volume was 8.38 ml, meaning 58.01 % void in the packed bed. 40:1 molar ratio of
ethanol:oil was adopted. The reactor run time corresponding to these runs was from 48 to
82 hours.

Figure 7.8 shows that the biodiesel conversion reaches an equilibrium in the 13 to 21
minute range for 40:1 molar ratio of ethanol:oil. The biodiesel conversion hovers around
95 ~ 97 % in this range and remains constant in that range showing that the reaction has
reached equilibrium. It was clear that the reaction was not going to be pushed to ASTM

121

requirement beyond this. Since equilibrium conversion is dependent on the respective


molar ratio of reactants and products, a higher molar ratio of alcohol may increase the
equilibrium conversion. 40:1 molar ratio may have been found to be better for the rate
constant for the reaction, but it need not be the optimal one for equilibrium conversion.
As a result, it was decided to test a higher molar ratio of 85:1. If the forward reaction
corresponding to 40:1 molar ratio had been resisted toward further progress by glycerol,
the reaction corresponding to 80:1 molar ratio would have only half the concentration of
glycerol that was there for 40:1 molar ratio at the same biodiesel conversion. The oil is
also a reactant, so it also has an influence on equilibrium conversion, and its
concentration would be halved as well. So a dramatic change cannot be expected.
Nevertheless, glycerol is seen to have more influence on equilibrium conversion, as
evidenced with better conversion from two-stage reactors with glycerol removal after the
first stage [64], and an improvement, however small, can be hoped for. Flashing the final
stage of the reactor with excess ethanol would be better than using a two-stage reactor
that would require twice the amount of heat exchangers and the associated heat losses.
Flashing just the final stage of the reactor with excess ethanol would only slightly
increase the size of the reactor, but could keep everything else constant, since the excess
ethanol can be flash-vaporized immediately after the reactor. The main problem with
using twice the ethanol would have been the increase in the size of the reactor, but this is
avoided by using the excess ethanol just at the final stage of the reaction.

So a run was conducted with 85:1 ethanol:oil molar ratio corresponding to a residence
time of 18 minutes. A conversion of 98.55 % biodiesel was observed. This is a marked

122

increase compared to the 95 ~ 97 % conversion seen for 40:1 molar ratio, and is very
close to the 99.15 % ASTM requirement. This corresponds to a bound glycerin content
of 0.36 % compared to the ASTM requirement, 0.22 %. It is quite possible that this
conversion was reached even before 18 minutes, as the equilibrium conversion seen
earlier for the 40:1 molar ratio was reached much earlier than 18 minutes, around 13

100
90

Biodiesel conversion (%) .

80
70
60
40:1 molar ratio

50

85:1 molar ratio


40

ASTM limit

30
200-60 mesh MnO
20
260 C, 73 atm
10
0
0

10

15

20

25

time (min)

Figure 7.8. Equilibrium conversion of ethanol-soybean oil transesterification

123

minutes. Greg Austic reported that methanol resulted in an equilibrium conversion


corresponding to 0.246 % bound glycerin (ASTM requirement: 0.22~0.24 %) for 30:1
molar ratio [81]. This is probably due to the better reactivity of methanol. Since a
conversion so close to the ASTM requirement has been obtained with methanol for a
lower 30:1 molar ratio, there is a good chance of meeting the ASTM requirement with a
final stage flash of higher molar ratio in the reactor itself. It can be concluded that the
reaction can be pushed all the way to the ASTM specification with a higher molar ratio
flash for both the alcohols, with methanol being better.

The ASTM limit on acid value is 0.5 relating to FFA. While using methanol at 30:1
molar ratio, Greg Austic observed an acid value of 0.6 at equilibrium for soybean oil that
was not mixed with any FFA and an acid value of 1.0 at equilibrium for soybean oil with
original FFA content up to 6 % [81]. Final stage flash with higher alcohol molar ratio
can lower the acid value also to ASTM requirement.

Process pressure-drop:
Pressure drop in a packed-bed reactor can be calculated using the Ergun equation [92].
The Ergun equation can be written as

150 (1 ) 2 v
1.75 (1 ) v 2
P
=
+
L
3 d 2p
3 dp
P = the pressure drop,

L = the height of the bed,


= the fluid viscosity,

= the void fraction of the bed,


124

v = the superficial velocity of the bed,


d p = the particle diameter,
= the density of the fluid.

The Ergun equation gives the pressure drop along the length of the bed for a given fluid
velocity. It also shows that the pressure drop depends on fluid viscosity, packing size and
fluid density.

Pressure drop is an important factor to consider in a packed-bed reactor for commercial


operation. Surprisingly, low pressure drop has been observed for the researched MnO
solid catalyst method.

The lab reactor of following parameters can be considered for the pressure drop scenario
in the lab-scale runs.
L = 93 cm,

= 0.6,
v = 0.0068 m/s,
d p = 70 microns,
= 451 kg/m3 (calculation was based on the expanded volume inside the reactor, as
done for the batch reactor).
The Andrade equation can be used to estimate the viscosity at a desired temperature
given the viscosities at two temperatures [93]. The Andrade equation is given as

125


1 1
ln 2 = C
1
T2 T1
where 2 and 1 are the kinematic viscosities at temperatures T2 and T1, and C is the
characteristic constant for a given component that can be calculated given the viscosities
at two different temperatures.

Ethanol viscosity data for a range of temperatures and pressures has been reported in the
literature [94, 95]. Ethanol viscosity, which is 1.087 mPa s at 298.15 K and 1 atm,
lowers significantly to 0.04 mPa s at 533.15 K and 100 atm. Vegetable oil viscosity can
be calculated similarly from the viscosity and density data reported in the literature [96,
97, 98]. Vegetable oil viscosity, which is ~ 80 mPa s at 298.15 K, lowers to 0.46 mPa s
at 533.15 K. So the viscosities lower dramatically at the reaction temperature, 260 C.
By incorporating the above values in the Ergun equation, the pressure drop was
calculated to be ~ 5 psi.

This matches well with the pressure readings observed in the reactor set up. The pressure
drop corresponding to the 93 cm reactor length was found to be 30~40 psi. (The pressure
reading fluctuates with 10 psi being the minimal difference of output sensitivity.) Most
of this pressure drop is caused by the six 2 m frits along the reactor length (two each for
the three segments making up the reactor length). A blank run through a stainless steel
tube fitting with just the six frits showed a pressure drop of 30 psi. This means that the
pressure drop reading corresponding only to the catalyst packing along the 93 cm length
of the reactor set-up is 0 to 10 psi. Thus, it matches well with the calculated pressure

126

drop of ~ 5 psi, as the pressure drop reading cannot be indicated anymore accurately in
the reactor set-up. The aforementioned pressure readings of the reactor set-up are
representative of the experimental runs in general.

Let us consider a scaled-up commercial reactor with the following parameters for a
similar pressure drop calculation.
Capacity: 35 million gallons/year (Reasonable capacity for capital cost)
Let us assume that the plant runs 330 days a year.
Hourly biodiesel production rate: 4,419 gallons/hour
Reactant mixture molar ratio: 30:1
Total feedstream rate (including alcohol): 12,581 gallons/hour
Let us assume a packing fraction of 0.5 for the packed-bed reactor.
Volume of the reactor: 23.81 m3
Let us consider two reactor set-ups, one where the reactor unit is modeled as a single
reactor and the other where the reactor unit is modeled as five parallel reactors.
Reactor dimensions if volume is modeled for a single reactor (L/D ratio = 15):
Diameter: 1.26 m; Height: 18.96 m
Reactor dimensions if volume is modeled for five multiple parallel reactors (L/D ratio =
15):
Diameter: 0.74 m; Height: 11.09 m
Superficial velocity (single reactor): 36.9 x 10-3 m/s
Superficial velocity (5 parallel reactors): 21.6 x 10-3 m/s

127

The corresponding Ergun equation parameters for the scaled-up 35 million gallons/year
plant reactor (similar to those listed for the lab-scale reactor) are listed below.
L = 18.96 m for single reactor set-up, 11.09 m for 5-parallel reactor set-up,

= 0.5,
v = 0.0369 m/s for single reactor set-up, 0.0216 m/s for 5-parallel reactor set-up,
d p = 0.5 mm,
= 460 kg/m3 (Calculation was based on the expanded volume inside the reactor).
The following is the pressure drop calculated using Ergun equation.
P = 52 psi (3.55 atm) for single reactor set-up, 14.4 psi (1 atm) for five parallel reactor

set-up.
These pressure drops are easily manageable for a commercial process.

Sometimes the packing fraction can be higher, and a packing fraction of 0.65 is generally
considered the upper limit. The packing fraction of 0.65 equates to a void fraction, =
0.35, in the above calculation. The pressure drop is calculated to be 187 psi (12.7 atm)
across the corresponding reactor length of 21.36 m for single reactor set-up, and 55 psi
(3.7 atm) across the corresponding reactor length of 12.49 m for five parallel reactor setup. These pressure drops are still very manageable.

The pressure drops corresponding to the various scenarios of the process have been found
to be very low. This is mainly due to the elevated temperature of the reaction. Low
pressure drop is a positive result for commercial packed-bed reactor design.

128

High temperature considerations:


Alcohol side reaction:
Alcohols are known to dehydrate at high temperatures to form ethers (intermolecular
dehydration) and alkenes (intramolecular dehydration). Acid catalysts particularly
catalyze dehydration. Sulphuric acid is known to cause etherification dehydration even
from 100 C. In fact, etherification of glycerol has been willingly pursued to get value
out of glycerol. But any attempt of in situ glycerol etherification faced a problem that
reactant alcohols, methanol and ethanol, would also undergo etherification [34]. Many of
the solid acids mentioned for biodiesel production are good catalysts for alcohol
etherification. Khandan et al reported that catalysts such as ZSM-5, Y-zeolites,
mordenites, silica and alumina showed conversions in the range of 82~96 % at 250 C in
3 hours for methanol [99]. Di Cosimo et al [100] and Kim et al [101] also reported that
alumina is an active catalyst for alcohol etherification due to the presence of surface acid
sites.

Base catalysts also can catalyze alcohol etherification, but their activities are
comparatively much lower and usually require higher temperatures [102]. Solid bases
have even lower activity than homogeneous bases for etherification. The data from Di
Cosimo et al show that etherification activity lowers with increase in basicity of catalysts
[100]. The reaction rate of MgO for ethanol etherification is ~ 500 times slower than that
of alumina at 300 C.

129

It was intended to check for etherification in the lab-scale packed-bed reactor set-up.
Pure ethanol as well as soybean oil-ethanol reactant mixture was sent through reactors
with and without MnO catalyst at various temperatures from 200 C to 400 C. There
was visible bubble formation in the product stream for temperatures above 300 C with
and without MnO catalyst. At 350 C, the product stream became squirty with and
without MnO catalyst. At 400 C, the product stream became smoky and violent, and the
squirts became loud.

The samples were analyzed in gas chromatograph and compared with the original pure
ethanol sample as standard. The same GC column used for biodiesel composition
analysis was used, but the temperature was programmed differently. The GC oven
temperature was programmed to hold the column at 35 C for 1 minute and then ramp to
160 C at 15 C/min in 9 minutes (total time = 10 minutes). The injector and detector
temperatures were maintained at the same temperatures of biodiesel composition analysis
at 88 C and 380 C, as mentioned in Appendix B. Samples passed through empty
reactor and MnO catalyst showed peaks identical to the standard pure ethanol up to 300
C. Above 300 C, new peaks were observed in both the cases. Figure 7.9 shows the GC
chromatogram of the pure ethanol sample used as reactant in the research. Figure 7.10
shows the GC chromatogram of the ethanol reagent passed through the MnO catalyst at
260 C for the corresponding reaction time. There is no difference with the original pure
ethanol, and the ethanol exiting the reactor at 260 C matches exactly with the
chromatogram of the pure ethanol. Figure 7.11 shows the GC chromatogram of the
ethanol reagent passed through the MnO catalyst at 400 C for the corresponding reaction

130

time. This chromatogram shows at least two new peaks at shorter residence times and
various small blunt peaks later. The first new peak matches exactly with diethyl ether.
Figure 7.12 shows the chromatogram of pure diethyl ether injected as a standard. It
looked quite convincing that diethyl ether was being formed from etherification of
ethanol at temperatures above 300 C. The second new peak in Figure 7.11 suggests the
presence of a higher molecular weight compound formed by the further association of
molecules. This places a constraint on how high the reaction temperature can be and the
situation can only be worse with solid acids like zirconia as they aid alcohol etherification
at much lower temperatures. Also, samples collected at 350 C and 400 C looked
notably yellowish, further suggesting the presence of higher molecular weight

Signal

compounds like aromatics being formed by dehydration.

Time (min)

Figure 7.9. Chromatogram of pure ethanol

131

Signal

Time (min)

Signal

Figure 7.10. Chromatogram of ethanol passed through MnO at 260 C

Time (min)

Figure 7.11. Chromatogram of ethanol passed through MnO at 400 C

132

Signal

Time (min)

Figure 7.12. Chromatogram of diethyl ether

Mass balance tests were conducted comparing the mass input to the reactor and the mass
output from the reactor for a given time. Samples corresponding to 400 C reaction
temperature showed up to 4 % mass loss immediately on collection at the product outlet,
unlike the samples corresponding to 260 C and 200 C.

However, this may not be all the mass that was lost, as some of the gas products may still
be dissolved in the collected sample, and the ether formed may remain a liquid at room
temperature. So some samples were tested for rate of evaporation after collection.
Samples that have more volatile components like ether would show a faster rate of
evaporation. Three samples were chosen, 100 % ethanol run through reactor without
catalyst run at 260 C for 11.28 minutes, 100 % ethanol run through reactor without

133

catalyst at 400 C and pure ethanol straight off the shelf without passing through reactor.
The third sample of pure ethanol was chosen so that the earlier two samples could be
compared with the rate of evaporation of pure ethanol. Figure 7.13 shows the mass lost
due to varying rates of evaporation for the three samples. It has to be remembered that
the 400 C sample has already lost 4 % mass before starting to weigh the mass at various

14
Original mass
12

Evaporated mass (g).

10

0
420

720

1155

time (min)
no catalyst, 100% ethanol, 260 deg C, 11.28 min
no catalyst, 100% ethanol, 400 deg C, 11.28 min
ethanol not passed through reactor
Figure 7.13. Rate of evaporation of samples analyzed for alcohol side reaction

134

times. The 260 C sample and pure ethanol standard sample were fairly similar, whereas
the 400 C sample showed increased loss of mass. Near the end, the 400 C sample has
lost about 94 % of its original mass, whereas the off-the-shelf pure ethanol and the 260
C sample had lost 85 % and 87 % of their original mass. This is a strong indicator of the
presence of more volatile components such as ether in the 400 C sample. Another set of
runs (not shown in Figure 7.8) showed 73 % loss of mass for the 400 C sample near the
end, as opposed to 66 % and 64 % losses for off-the-shelf pure ethanol and the 260 C
sample.

Vegetable oil side reaction:


The possible side reactions associated with vegetable oil at high temperatures (greater
than 300 C) are cracking and polymerization.

Vegetable oil tends to undergo cracking at higher temperatures. Vegetable oil starts
showing the first signs of cracking at 280 C, starting with the formation of free acids
[103]. That is when Gibbs free energy for the reaction crosses zero and makes the
reaction possible. Demirbas showed that cracking becomes pronounced around 280 C
[104]. Sadrameli and Green show that at 300 C, conversion of vegetable oil for cracking
could be as high as 55 % (38 % liquids, 17 % gases) in the time range of biodiesel
reaction time [105]; and at 350 C and 400 C, the cracking conversions could reach 94
% and 98 % [106]. Type of catalysts can have an effect, and some result in more
conversion than the others, but the potential is still significant. Calcium oxide results in

135

44 % cracking conversion, while -alumina and empty reactor runs result in 99.4 % and
98.4 % conversions at 400 C [107].

Polymerization is another potential side reaction. Polymerization becomes significant


from 300 C [108, 109]. Polymerization depends on the number of double bonds in the
long chain, meaning that linseed oil would polymerize twice as fast as soybean oil. Still
polymerization of soybean oil can be substantial. A conversion of 75 % in 15 hours has
been reported, which even by a conservative estimate would mean at least 1 %
conversion in the biodiesel reaction time. That may be enough to pose problems in a
packed-bed reactor.

136

Chapter 8

Comparison of MnO Catalyst Method with Other Methods

In this chapter, the MnO solid catalyst method has been compared to other biodiesel
methods. First, the comparison has been done in terms of reaction speed and reaction
conditions. The method has then been compared to the conventional biodiesel process in
terms of overall process design and capital and processing cost estimates.

Reaction comparison:
The reaction rates of various methods have been compared to that of the MnO solid
catalyst method. In addition to comparing the reaction rates, other differences in methods
such as pressure, reactor type (packed-bed, CSTR), related side reactions, catalyst life
and reusability, conversion of FFA and reactant molar ratios have also been discussed.

Figure 8.1 shows the logarithmic rate constants of various reaction methods plotted
against absolute temperature inverse. This is a good visual plot for comparison because it
provides the same multiplicative scale of difference over the entire plot area. That is, the
difference in rate constants 0.03 and 0.012 are shown in the same scale as that of rate
constants 0.3 and 0.12. Unlike the logarithmic rate constant plot, an absolute rate
constant plot would have diminished the difference in the former set of rate constants
compared to the later set if both the sets of rate constants were plotted in a single graph.
The rate constants differ by the same factor in both sets and are shown by the same scale
even at different temperatures in Figure 8.1. Figure 8.2 and Figure 8.3 show

137

representative examples of various method comparisons. The figures show reaction


times for a given conversion. Figure 8.2 has a reaction time scale that corresponds to
temperatures less than 300 C. Figure 8.3 has a reaction time scale that corresponds to
temperatures more than 300 C.

Supercritical methanol process:


The supercritical methanol process [67, 68] developed by Saka and Kusdiana requires the
supercritical conditions to be well developed for the rate constant plot that lies on a low
plateau up to 300 C (240 atm) to spike up (See Figure 8.1). The authors report that
pressures as high as 500 atm are required. Even in the temperature range where the rate
constant has risen up, the reaction is ~ 80 % slower (See Figures 8.1 and 8.3). In the low
plateau, the reaction is 400~700 % slower than the MnO catalyzed reaction (See Figures
8.1 and 8.2). It was discussed earlier how MnO solid catalyst overcomes the high
pressure requirement of the no-catalyst reaction in addition to increased reaction rate.
While using MnO, only the minimum pressure that maintains the reactants in liquid state,
~70 atm, is required. Molar ratio corresponding to both the methods is ~ 40:1.

It was also been discussed earlier how temperatures higher than 300 C are not desirable,
due to the side reactions related to alcohol and oil.

Tateno/Goto method:
Tateno and Sasaki [110] used solid catalysts such as sodium carbonate, calcium oxide,
calcium hydroxide, magnesium oxide and a mixture of nickel oxides in the temperature

138

range 200 ~ 400 C, to lower the extremely high pressures required in the supercritical
methanol process. Molar ratio in their method was 40:1. The pressure was above the
critical pressure of methanol, 80 atm. The results are shown in Figures 8.1, 8.2 and 8.3.
The tests were conducted in batch reactor and involved a preheat time of 12~15 minutes.
The conversion during the preheat time can be significant. The preheat time was taken
into account conservatively while calculating the overall reaction time and rate constant.
The preheat temperature vs time profile was reported. The data reported at lower
temperatures was used to calculate the conversions at lower preheating temperature and
conservative interpolation was used to calculate the conversion corresponding to any
intermittent temperature.

At 250 C, the reaction time was 105 % longer compared to that of MnO. Tateno and
Sasaki mention that higher temperatures are preferred for the reaction. It was discussed
in the earlier chapters how temperatures above 300 C cause undesirable side reactions
related to alcohol and oil. At higher temperatures, the reaction tends to be even slower
than that catalyzed by MnO (164 % longer reaction time at 300 C, 356 % longer reaction
time at 350 C). These results reflect the much better effectiveness of MnO as catalyst.

Goto and Sasaki report a similar method using MnO2 and MoO3 as catalysts [111]. The
reaction catalyzed by these catalysts was even slower. The reaction time was 715 %
longer at 200 C and 474 % longer at 300 C.

139

450C 400C 350C 300C

250C

200C

150C

100C

101 min-1

100 min-1

ln k

-2

10-1 min-1

-4
10-2 min-1

-6
10-3 min-1

-8
0.0010

0.0015

0.0020

0.0025

0.0030

1/T (1/K)
No catalyst supercritical [68]
Tateno method [110]
Goto method [111]
Mcgyan process [112]
WZA (1atm) [61]
Calcined hydrotalcite - fresh
MnO maximum observed
Lacome catalysts [119]
Fe-Zn cyanide [121]

MnO
Goto method - No catalyst [111]
MnO Arrhenius plot
Propane cosolvent [69]
Di Serio catalysts [113]
Calcined hydrotalcite - used
Tungstated zirconia [62]
ZnO, zinc aluminate [120]
ZnO-Ln2O3 [122]

Figure 8.1. Comparison of reaction rates of various biodiesel methods

140

Method
Supercritical
methanol
process [67,
68]

Details
Comparison
200 C, 70 atm no 75 % conversion: 115.5 min 404 % more
catalyst
MnO: 22.9 min reaction time
270 C, 120 atm no 88 % conversion: 50.5 min
catalyst
MnO: 6.1 min

723 % more
reaction time

Goto work
[111]

200 C, P>85 atm


MnO2

55 % conversion: 107.5 min


MnO: 13.2 min

715 % more
reaction time

300 C, P>85 atm


MoO3

90 % conversion: 20.6 min


MnO: 3.6 min

474 % more
reaction time

Tateno work
[110]

250 C, P>85 atm


Na2CO3

95 % conversion: 27.9 min


MnO: 13.6 min

105 % more
reaction time

Two-step
supercritical
process [74]
Propane
cosolvent [69]

270 C, 100 atm


88 % conversion: 60.0 min
(hydrolysis), 200
MnO: 6.1 min
atm (esterification)
280 C, 128 atm
98 % conversion: 12.0 min
no catalyst
MnO: 9.1 min

879 % more
reaction time

270 C, 128 atm


no catalyst

85 % conversion: 12.2 min


MnO: 5.5 min

119 % more
reaction time

0.9

180 C, P>25atm
Di Serio
catalysts [113] Titanium-silicate
12:1 molar ratio
120 C, P>6 atm
Calcined
fresh catalyst
hydrotalcite

62 % conversion: 60.0 min


MnO: 29.0 min

106% more
reaction time

0.6

30 % conversion: 44.2 min


MnO: 92.3 min

52 % less
reaction time

0.3

120 C, P>6 atm


used catalyst

30 % conversion: 120.0 min


MnO: 92.3 min

30 % more
reaction time

0.3

94 % conversion: 60.0 min


MnO: 18.6 min

222 % more
reaction time

Zinc
235 C, P>75 atm
aluminate[120]

31 % more
reaction time

20

40

time (min)
60

100

0.8

0.9

0.6

0.9

1.0

0.9

1.0

Comparative method
MnO

0.9

Figure 8.2. Comparison of biodiesel methods (T 300 C)

141

80

120

Method

Details

Comparison

Supercritical
methanol
process [67,
68]

300 C, 140 atm


no catalyst

88 % conversion: 4.98 min


MnO: 3.31 min

50 % more
reaction time

385 C, 650 atm


no catalyst

88 % conversion: 1.42 min


MnO: 0.78 min

81 % more
reaction time

Tateno work
[110]

300 C, P>Pc
Na2CO3

90 % conversion: 9.49 min


MnO: 3.59 min

164 % more
reaction time

350 C, P>Pc
Na2CO3

90 % conversion: 6.69 min


MnO: 1.47 min

356 % more
reaction time

309 C, 218 atm


BMZ

16 % conversion: 0.39 min


MnO: 0.22 min

76 % more
reaction time

405 C, 218 atm


BMZ

87 % conversion: 0.39 min


MnO: 0.57 min

32 % less
reaction time

Propane
300 C, 128 atm
cosolvent [69]

98 % conversion: 10.87 min


MnO: 6.11 min

78 % more
reaction time

Mcgyan
process [112]

10

88.0 %

88.0 %

90.0 %

90.0 %

15.5 %

Comparative method
MnO

87.3 %

98.0 %

Figure 8.3. Comparison of biodiesel methods (T 300 C)

142

time (min)
6

12

Mcgyan process:
McNeff et al developed a process dubbed the Mcgyan process [112]. They employed
zirconia-based catalysts in packed-bed reactor under supercritical conditions (33:1 molar
ratio). Unmodified zirconia, zirconia modified with phosphoric acid (PMZ) and zirconia
modified with sodium hydroxide base (BMZ) were tested. They specifically prefer
higher temperatures in the range 350 ~ 465 C. The high temperature side reaction
involved with alcohol becomes even more pronounced while using acid catalysts like
zirconia. They report that more than 90 % of glycerol formed was decomposed to
gaseous products (dimethylether (42%), carbon monoxide (17%), methane (17%) and
carbon dioxide (16%)). If glycerol undergoes etherification, the reactant methanol has an
even better tendency to undergo etherification, as discussed in Chapter 7. The reaction at
309 C is 76 % longer than the MnO catalyzed reaction. The reaction time is 32 %
shorter at 405 C and 73 % shorter at 465 C. Etherification side reaction of alcohol was
found to be more pronounced above 350 C as discussed in Chapter 7. The 32 % faster
Mcgyan reaction at 405 C is still close enough to that of the widely tested 200-65 mesh
MnO, and is on par with the rate observed with the finer MnO sample mentioned in
Chapter 6. The pressure of the Mcgyan process was considerably higher than that of the
MnO solid catalyst method, 218 atm as compared to 72 atm.

The ln k vs 1/T plot in Figure 8.1 shows that Mcgyan process rate constant is lower than
that of MnO for temperatures up to 350 C, after which it becomes higher.

143

The zirconia catalyst life span was 66 hours compared to MnO that was active even after
500 hours.

Two-step supercritical process:


Minami and Saka proposed a two-step supercritical process (no catalyst) [74] to get
around the extremely high pressure (500 atm) required in the original supercritical
process [67]. They proposed hydrolysis of triglycerides followed by esterification.
Hydrolysis produces FFA and glycerol. FFA is converted to biodiesel esters in the next
esterification step. For a reaction temperature of 270 C, the pressures for hydrolysis and
esterification steps are 100 and 200 atm. They used 10 % FFA during hydrolysis since
FFA was seen to act as a catalyst improving the reaction time. FFA was seen to act as an
autocatalyst in the esterification step. The esterification step was twice as fast as the
hydrolysis step and it was already discussed in earlier Chapters that esterification is a
much faster reaction than transesterification. For 88 % conversion, the total reaction time
of the two-step supercritical process is 60 minutes (40 minutes for hydrolysis and 20
minutes for esterification). Compared to the reaction time of MnO, 6.13 minutes, the
reaction time of the two-step process is 878 % longer.

Propane cosolvent process:


Cao et al [69] reported the process using propane as cosolvent to ease the extreme
conditions proposed in the original supercritical methanol process of Saka and Kusdiana
[67]. They report 280 C, 128 atm, 24:1 ~ 33:1 methanol: oil molar ratio and 0.05:1
propane:methanol cosolvent ratio as the optimal conditions. As shown in Figure 8.3, the

144

process appears to bring the rate constant values up to the supercritical effectiveness
earlier than the supercritical methanol process. The supercritical methanol process shows
the spike at 300 C, whereas propane enables an earlier spike at 280 C. As with the
batch methods mentioned earlier, there is a preheating time during which the reaction
may have occurred. Using their own lower temperature data by conservative estimates,
the reaction time at 280 C for 98 % conversion is 12 minutes which corresponds to the
reported 10 minute reaction time after taking into account the conversion during
preheating. The MnO solid catalyst method corresponds to 9 minute reaction time at 280
C. This means that the propane cosolvent process has 31 % longer reaction time. The
reaction pressure while using cosolvent is still reasonably high, 78 % higher than that of
MnO. The propane cosolvent has only one optimal point, 280 C, at which the
aforementioned comparison was made. At higher and lower temperatures, the reaction
rate falls off even more.

Di Serio catalysts and calcined hydrotalcites:


Di Serio et al tested solid acids like vanadyl phosphate, metal-substituted vanadyl
phosphates (Al, Ga, Fe and Cr) and titanium silicate (7.47 % by weight TiO2 on SiO2);
and solid bases like MgO and hydrotalcites [113, 114]. Hydrotalcites, mixtures of
magnesium and aluminum oxides, were custom prepared with optimal composition and
calcined before use. Both custom-made and commercially available samples were tested
in the case of MgO. The tests were conducted in stirred batch reactors at 180 C and at
saturation vapour pressure of methanol (~25 atm) with a molar ratio of 12:1. Calcined
hydrotalcite showed slightly better activity than MnO when a fresh sample was used.

145

MgOs activity was marginally less than that of MnO. The solid acids were at least two
times slower than MnO. The bases were affected by FFA in reaction mixtures containing
FFA. The acids, though less active, converted FFA. Among the solid acids, though the
phosphates showed more activity, TS was the only stable catalyst that retained its activity
after the first run. Among the solid bases, the commercial MgO sample retained it
activity after the first run. No data was reported on the reusability of custom-made MgO
and calcined hydrotalcite.

Liu et al also tested similar hydrotalcites, but at 120 C and 30:1 molar ratio [115]. They
tested catalyst reusability. Reused catalysts showed considerable decrease in activity.
Fresh hydrotalcite showed 52 % faster reaction rate compared to MnO, but reused
hydrotalcite showed 30 % slower reaction rate compared to MnO.

All these catalysts were tested for a batch or CSTR reactor design. The catalyst lifetime
would be expected to be shorter compared to packed-bed catalysts in a CSTR, and a
CSTR reactor is less efficient than a packed-bed reactor since the reaction takes place at
the exit concentration of reactants. A batch reactor is not suitable for high temperature
reaction to recover heat. Xie et al used similar highly active calcined hydrotalcite at
methanol reflux conditions (~65 C) in batch reactors [116]. At 15:1 molar ratio and with
7.5 % catalyst by weight, it took 9 hours for a conversion of 66 %.

Oku et al showed that, in a hydrotalcite catalyzed run at 150 C that gave 77 %


conversion after 24hours, high concentration of Mg and Al ions were detected in the

146

products (Mg 17,800 ppm; Al 6,900 ppm) [117]. They report that the problems of
catalyst leaching need more in-depth study to confirm the possibility of using MgO and
related hydrotalcites as industrial catalysts [118].

Other methods:
Demirbas conducted batch tests using CaO as catalyst in the temperature range 192 ~ 252
C at 240 atmospheres [91]. From the batch and packed-bed tests of the present research,
MnO was shown to be a better catalyst compared to other good catalysts such as TiO and
CaO. Where CaO tested by Demirbas showed 64.7 % conversion after 25.7 minutes,
MnO data correspond to 21.8 minutes for the same conversion.

Solid acids tend to have lower activity than solid bases. Lopez et al tested various
modified zirconia catalysts (mentioned in Chapter 2) [62]. Tungstated zirconia, the only
catalyst that could be reactivated after the five-cycle activity loss, showed 23 %
tricaprylin conversion in 8 hours at 120 C.

Lacome et al tested zirconia, titanium and antimony supported on alumina in batch


reactors [119]. The reaction was at least four times slower than that of MnO at 200 C.
Stern et al tested ZnO and zinc aluminate in batch reactors in the temperature range
170~240 C [120]. The zinc aluminate tested by French Petroleum Industry was
mentioned in Chapter 2 to be deactivated if water content is more than 1000 ppm.

147

Double metal Fe-Zn cyanides are hydrophobic Lewis acid catalysts that can be used in
the presence of water [121]. Using 3 % by weight catalysts in batch reactor, at 170 C
and 15:1 molar ratio, ~96 % conversion was obtained after 8 hours.

Tungstated zirconia-alumina (WZA) catalyst employed by Furuta et al at atmospheric


pressure and at 40:1 molar ratio gave 50 %, 90 % and 94 % conversions after 20 hours at
200 C, 250 C and 300 C [61]. Acid catalysts are susceptible to dehydration
etherification of alcohol even at temperatures as low as 100 C, and the catalysts tested
by Furuta et al [63] also showed good activity for methanol etherification.

Yan et al tested mixed oxides of zinc and lanthanum (ZnO-Ln2O3) in batch tests [122].
Compared to MnO, the reaction rate was 60 % slower at 210 C, 45 % slower at 200 C
and 200 % slower at 180 C. The catalyst has both acid and base sites. It showed good
conversion of FFA and was tolerant to water content. No data on catalyst reusability was
reported, however.

Microwaves have been reported to enhance the transesterification rate [123, 124].
Microwaves are claimed to impart heat energy more effectively by directly transferring
energy to the molecules instead of the usual conductive and convective heat transfer. In
this way, microwaves are believed to selectively transfer energy to the polar alcohol
molecules and selectively energize the interaction with catalysts. There have been
reports of specially prepared solid bases losing activity after the first run under methanol
reflux conditions while employing microwaves as well [125]. With microwave

148

irradiation, sodium silicate is reported to catalyze the transesterification of oils to


complete conversion in 1 hour at 120 C with 20:1 molar ratio [123]. However, no data
on catalyst reusability was reported. Microwave irradiation is seen as an enhancement of
catalytic reaction, so it can aid the catalytic reaction of MnO as well. MnO has already
been seen as a top notch catalyst in terms of activity as well as activity sustenance, and
microwave-irradiated enhancement could make it even better.

A number of specially prepared heterogeneous catalysts have been reported to leach or


lose activity in a few runs [118]. MnO results discussed were from a packed-bed reactor,
whereas the results of most of the other methods were from batch processes. There is
always some back-mixing in a packed-bed reactor, and this lowers the efficiency of the
reaction at least to some degree. So the already apparent superior activity of MnO could
be actually even better.

Comparison with calcium oxide:


Calcium oxide is one of the most widely investigated heterogeneous catalysts for
biodiesel production. Calcium oxide is one of the representative catalysts tested by
Tateno and Sasaki [110]. One of their illustrative runs at 300 C was done with calcium
oxide. The comparison of results discussed earlier showed that calcium oxide took 164
% longer reaction time than MnO. This illustrates the superiority of MnO over calcium
oxide. Moreover, calcium oxide does not covert free fatty acids, whereas MnO has been
seen to convert free fatty acids due to its amphoteric nature. MnO corresponded to 18 %
faster reaction time compared to calcium oxide at one of the experimental conditions

149

adopted by Demeirbas as mentioned earlier [91]. It was also seen from the present work
itself that calcium oxide in support combination with TiO2 (as CaTiO3) did not fare as
well as MnO.

Activation energy:
Activation energy represents the energy barrier that must be overcome for a reaction to
happen, and catalysts help by lowering this energy barrier. Activation energy using the
MnO solid catalyst was calculated to be 53.19 kJ mol1. Activation energy using ZnOLn2O3 solid catalyst, one of the catalysts that have closer comparison with MnO, was
90.9 kJ mol1 [122]. The activation energy using NaOH catalyst in the conventional
homogeneously catalyzed process was reported to be 19.9 kJ mol1 [126]. The activation
energy of heterogeneous catalysts is expected to be higher than that of homogeneous
catalysts. But, among the heterogeneous catalysts, the activation energy using MnO is
almost half of the activation energy using ZnO-Ln2O3. This represents a significant
reduction in energy barrier of the reaction.

Process design comparison with the conventional process:


In this section, the overall process design of the solid catalyst method will be discussed in
comparison to the conventional biodiesel process, which uses homogeneous catalysts.
The main advantage of a solid catalyst process over the conventional process is the ease
of separation of products. As mentioned earlier, eight units of waste water are generated
for every unit of biodiesel in removing the dissolved catalyst and soap in a conventional
biodiesel plant.

150

The high temperature of the MnO solid catalyst method is not a problem, since it is a
continuous process in which the heat energy can be recovered using heat exchangers.
Excess methanol can be easily recycled by flash vaporization. In conventional process,
the reaction temperature is 60 C, and the scope for heat recovery is either none or
minimal.

Table 8.1 lists the advantages of the MnO solid catalyst method over the conventional
process. Figure 8.4 illustrates how the biodiesel production process can be simplified by
avoiding the various additional processing units required in the conventional process.
The following are the units that would be avoided in the MnO solid catalyst process:
1. Alcohol-catalyst premixing unit:
While using sodium hydroxide, the catalyst has to be mixed with alcohol before
entering the reactor. Mixing alcohol and sodium hydroxide prior produces sodium
methoxide, the real active catalyst species, and water. While using a solid catalyst
like MnO in a packed-bed, this unit is avoided.
2. Free fatty acid pretreatment unit:
The conventional process needs a separate acid catalyzed unit to convert FFA to
biodiesel esters. This unit precedes the main alkali catalyzed transesterification
reactor. Cheap feedstock such as used oil and even virgin oil in many cases contain
considerable FFA content. If a separate FFA pretreatment unit is not employed, the
FFA would be wasted as soap in the main alkali catalyzed reactor. In the MnO solid
catalyst method, MnO was seen to catalyze the esterification of FFA as well. Thus the
solid catalyst process eliminates the need for a separate FFA pretreatment unit.

151

3. Water-washing units:
After the reaction, the biodiesel product of the conventional process contains the
dissolved catalyst, soap, any acid used to neutralize the alkali catalyst or used in FFA
pretreatment and the related water formed. These impurities require numerous waterwashing units to be removed from biodiesel to meet the ASTM requirements.
Generating 8 gallons of waste water per gallon of biodiesel produced indicates the
intensity of water-washing required. Water requirement is one of the main processing
expenses for a conventional biodiesel plant. In most cases, the waste water is dumped
to the municipal sewage. This represents a significant source of pollution, in addition
to the tedious separation hassles of product purification. The MnO solid catalyst
process would not require any water-washing since the catalyst is not dissolved in the
reaction mixture.
4. Product-drying units:
Since there is water-washing involved, there is a drying unit to remove the excess
water from water-washing steps to meet the ASTM allowed limit for water content.
This is done by methods such as trickling biodiesel through superheated walls.
5. Glycerol purification:
The glycerol by-product from conventional biodiesel plants is even more impure than
biodiesel. The concentration of dissolved catalyst, soap, any acid used and related
water would be even more in the glycerol phase since it is polar. The glycerol
produced from the conventional process is only 80 % pure, and it has no or little
value. Pure glycerol has good market value. The biodiesel producers have to either

152

sell the impure glycerol cheap to outside glycerol refineries or install their own
vacuum distillation units if they want value for glycerol.

Conventional Process

MnO Solid Catalyst Method

Process Type

Batch

Continuous

Catalyst Life

Very Short (consumed)

Long

Conversion Time

~90 minutes

10~20 minutes

Impurities (ASTM)

Catalyst, soap, water, glycerol,

Glycerol, glycerides

glycerides
FFA Pretreatment

Yes

No

Catalyst Mixing Unit

Yes

No

Distillation of Glycerol

Yes

No

Washing & Wastewater

Yes

No

Product Drying

Yes

No

Table 8.1. Comparison of conventional process and MnO solid catalyst method

153

Alcohol
(Methanol)
Mixing Tank

Catalyst
(NaOH)

Transesterfication
Reaction

Methanol
Recovery

Product
Separation

Various
Oil Feedstock

Free Fatty Acid


Pretreatment

Water
Wash

Product
Drying

Adsorption
Polishing

Glycerol
Distillation

High Quality
Glycerol

Waste
Water

Figure 8.4. Process simplification of the MnO solid catalyst method

154

High Quality
Biodiesel

The capital cost of a typical conventional plant can be approximated to $1 per gallon of
its annual biodiesel capacity, if the plant is large enough (i.e. above 30 million gallons per
year). The aforementioned units represent ~30 % of total capital cost as seen from the
method developed by Haas et al [127] to calculate biodiesel production cost.

Preliminary cost analysis studies have been carried out based on the MnO solid catalyst
process. The first one was carried out in 2007 when certain processing details were not
finalized. But it was still good enough to compare capital costs. See Appendix H. The
second one was carried out in 2008 for a mobile production unit of capacity 50,000
gallons per year. See Appendix I. The first study was conducted for a 34 million gallon
per year plant. So it is a better scale for capital cost comparison. As seen from the
analysis, the capital cost was 53 % lower than a conventional biodiesel plant. This was
mainly due to the riddance of various additional processing units, as mentioned earlier.

The second analysis, based on a smaller scale of 50,000 gallons per year mobile unit,
shows a higher capital cost, as expected (because of its smaller scale). But the processing
cost (excluding capital cost) per gallon, even at this smaller scale, decreases by 28 to 59
% compared to a conventional plant (28 % for virgin oil; 59 % for used oil) by the Singh
model. Compared to the Haas model, it is not clear by how much the processing cost is
lowered. On a larger scale, because of the simplicity of the MnO solid catalyst process,
the labor costs would be expected to be lower than that of the conventional process. The
glycerol by-product would have good value as well. These are two unknowns to compare
with the Haas model. So, the processing cost can be expected to be lower by this model

155

as well. The total processing cost for a larger scale plant, by incorporating the capital
cost as depreciation value, turns out to be 51 to 71 % lower by Singhs model. By Haas
model, the total processing cost decreases by at least 18 % this way. For a larger plant,
the labor costs of the much simpler MnO process can be assumed to be at the most the
labor costs of the process being compared to. For a larger plant, the reduction in the
capital cost of the first analysis can be assumed.

156

Chapter 9

Conclusion

1. Biodiesel production method was pursued because of its importance in terms of


environment and energy. Biodiesel has considerable production potential. The current
method of biodiesel production has numerous separation issues due to the use of soluble
catalysts. It was intended to develop a solid catalyst method that would be neat and
efficient.

2. For the biodiesel transesterification reaction, base catalysts are always preferred to
acid catalysts wherever possible because of their higher activity. But, normal solid bases
are not effective for the conventional reaction conditions (60~70 C, methanol/ethanol
reflux condition). Solid superbases that are custom-made have to be used for the
conventional reaction conditions. Such highly basic catalysts result in maintenance
issues such as frequent loss of activity, water vapour/carbon dioxide adsorption from
ambient air, and leaching into reaction mixture. So the reaction at higher temperatures
was considered.

3. High temperature biodiesel reaction without catalyst required extremely high


pressures that were not suitable for industrial production. It was desired to test if mild
solid base catalysts with no high-maintenance issues could be effective at high
temperature with reasonably short reaction time. High temperature would not be a
problem since a continuous process would facilitate heat recovery.

157

4. Various known solid bases and previously untested solid bases such as transition
metal oxides were tested for the biodiesel reaction with an idea of what the optimal
activity is, based on surface chemistry principles. A reaction temperature of 260 C and a
pressure of 70 atmospheres were adopted for the investigated solid catalyst reaction.
TiO, MnO, CaO-Al2O3, CaO-TiO2, and CuO were identified as promising catalysts from
batch tests.

5. The identified catalysts were then tested in a lab-scale packed-bed reactor. MnO was
found to be the most effective of the identified catalysts. MnO gave ~95 % conversion in
13 minutes.

6. The kinetics of the chemical reaction catalyzed heterogeneously by MnO catalyst was
modeled, after making sure there was no interference from external and internal mass
transfer limitations. The reaction followed pseudo first order kinetics. The reaction was
validated by the Arrhenius plot.

7. Reaction rate deviated below the Arrhenius plot at temperatures below 175 C. The
oil-ethanol reactant mixture tends to phase separate below 175 C. It was seen how the
phase separation of the reactant mixture lowers the reaction rate at lower temperatures.

8. A reaction mechanism of the heterogeneously catalyzed transesterification reaction


has been proposed.

158

9. MnO was found to have the lowest surface area of the tested catalysts. Literature
reports and temperature-programmed CO2 desorption tests indicated that MnO was not
the strongest of the solid bases tested. Yet, MnO was found to be the most effective
catalyst.

10. The MnO catalyst sustained its activity for more than 500 hours. This is an
important result for the commercial viability of the solid catalyst method.

11. Alcohol:oil molar ratio of 40:1 was found to result in the maximum reaction rate.
However, 20:1~30:1 was found to be the range of optimal molar ratio for maximum
productivity.

12. The MnO catalyzed reaction required the saturation vapour pressure at 260 C as the
minimum pressure required for a good reaction rate. However, pressures higher than that
were not required for better reaction rate, unlike the catalyst-free supercritical methanol
process.

13. MnO was also found to catalyze the esterification of free fatty acids. This suggested
that MnO is amphoteric. 97 % conversion was obtained in 7.5 minutes. This would help
to recycle cheap feedstock such as used cooking oil and animal tallow to biodiesel.

14. The MnO catalyst also showed good resistance to the presence of water. FFA
esterification and resistance to water facilitates the MnO solid catalyst method to exploit

159

cheap feedstock such as used oil and animal tallow for biodiesel production. In fact, the
water content of the reaction associated with the real world cheap feedstock corresponds
to a range that enhances the reaction rate.

15. The MnO solid catalyst method was modeled to calculate the pressure drop of
commercial-scale reactor. The method corresponds to low pressure drops. This is a
positive result for commercial design.

16. Alcohol was evaluated to be prone to an undesirable side reaction, etherification, at


temperatures higher than 300 C.

17. The MnO solid catalyst method was found to compare favourably with other
researched methods in terms of reaction rate, catalyst life, pressure, temperature, side
reaction and maintenance.

18. The advantages of the MnO solid catalyst method over the conventional process have
been illustrated. The potential of the MnO solid catalyst method to be better than the
conventional process in terms of production cost has been discussed.

Future work:
A scaled-up reactor is the next step to do further studies in designing the commercial
production system. The possibility of eliminating the final polishing step has to be
evaluated further following up from the higher molar ratio tests to push equilibrium

160

conversion. The limits of the process to handle the levels of impurities in various grades
of cheap feedstock, used oil, brown grease and animal tallow, has to be evaluated.

The temperature range of the reaction can be taken advantage of in the attempts to meet
the ASTM limits with respect to total glycerin, free glycerin, acid value and soap content.
It has been seen that temperatures from 175 C to 325 C are favourable for the reaction.
The ASTM limits of the aforementioned components is related to equilibrium
conversions (FFA esterification, triglyceride transesterification), reverse reactions
(hydrolysis of biodiesel esters by water formed during FFA esterification), undesirable
side reactions (soap formation, alcohol dehydration) and poisoning of catalytic sites.
These reactions can be manipulated by changing the reaction temperature. Some of these
reactions tend to be less pronounced at lower temperatures, and the process can be
designed with multiple temperature-stage reactors. It was mentioned in Chapter 8 how
the equilibrium of the reaction can be pushed further either by flashing with excess
alcohol in the final stage of the reactor or by designing a two-stage reactor after stripping
the products in between the two stages. Such optimization of the reaction process is one
of the future works that need to be conducted.

Catalyst particle size, related surface area per unit volume and catalyst-support
considerations for commercial reactor have to be assessed in a scaled-up reactor. It was
seen in Chapter 6 that the surface area per unit volume higher than that of the normally
used reactor in the study showed an increase in rate constant in a lower proportion.
However, the potential to increase the rate constant is still there. As demonstrated in the

161

pressure drop calculations, the method has been seen to have low pressure drops, and
particle sizes as low as 0.5~1 mm could be employed. More than the surface area per
unit mass of the catalyst, its the surface area per unit volume of the reactor that
determines the rate constant of the reaction. The packing fraction could be increased
from the 40 % seen in the lab reactor to 65 % in a commercial reactor. This is equivalent
to a reaction rate that is twice as fast. So there is scope to load the catalysts as such in a
simple fashion.

Other options are to load the catalyst at a higher surface area onto a support, or to alter
the catalyst manufacturing mechanism, making it mesoporous. Internal mass transfer
may not be a problem because similar catalysts are already being used in the petroleum
industry in reactions involving molecules as long as triglycerides. However, the
maintenance issues that come with specially prepared catalysts should be evaluated. A
simpler catalyst with less maintenance issues (even if it means lower rate constant), may
be preferred over a catalyst with lot of maintenance issues. This would be a useful study
in a scaled-up reactor.

Other possibilities to enhance the catalyzed reaction, such as microwave irradiation, may
also be evaluated. Studies checking that the production process meets all the ASTM
standards have to be performed.

162

Most of the essential studies of the method have already been performed through the labscale reactor. However, the remaining details of a commercial reactor have to be studied
through a scaled-up pilot reactor.

163

Abbreviations, Notations and Symbols

Abbreviations:
ASTM American Society for Testing and Materials
B100 100 % Biodiesel
B20 20 % biodiesel, 80 % petroleum diesel
BET Brunauer, Emmett, Teller
BMZ Base Modified Zirconia
CSTR Continuous Stirred Tank Reactor
FFA Free Fatty Acid
GC Gas Chromatography
HPLC High Performance Liquid Chromatography
ID Internal Diameter
LHSV Liquid Hourly Space Velocity
MTBE Methyl Tertiary-Butyl Ether
OD Outer Diameter
SBIR Small Business Innovation Research
TAME Tertiary-Amyl Methyl Ether
TGA Thermogravimetric Analysis
TLC Thin Layer Chromatography
USDA United States Department of Agriculture
WHSV Weight Hourly Space Velocity
WZA Tungstated zirconia-alumina

164

Notations and symbols:


dp

- Diameter of particle

- Diameter of bed

Ea

- Activation energy

- Rate constant

k0

- Arrhenius equation prefactor

- Pressure

Pc

- Critical pressure

- Gas constant

- Absolute temperature (K)

165

References

[1]

Ma, F. and Hanna, M.A., Biodiesel production: a review, Bioresour.


Technology, 70 (1999) 1-15.

[2]

National Renewable Energy Laboratory


(http://www.co.snohomish.wa.us/documents/County_Services/FocusOnFarming/
USDABiodieselppt.pdf)

[3]

Personal communication, Fats, Oils and Animal Coproducts Research Unit,


Agricultural Research Service, United States Department of Agriculture.

[4]

Gregory Austic, Personal communication, Piedmont Biofuels, Pittsboro, NC.

[5]

Shay, E.G., Diesel fuel from vegetable oils: status and opportunities, Biomass
and Bioenergy, 4 (1993) 227-242.

[6]

Sonntag, N.O.V., Structure and composition of fats and oils. Bailey's industrial oil
and fat products, vol. 1, 4th edition, ed. Swern, D., John Wiley and Sons, New
York, 1979, p. 1.

[7]

Kincs, F.R., Meat fat formulation, J. Am. Oil Chem. Soc., 62 (1985) 815-818.

[8]

Agarwal, A.K. and Das, L.M., Biodiesel Development and Characterization for
Use as a Fuel in Compression Ignition Engines, Trans. ASME, J. Eng. Gas
Turbines Power, 123 (2001) 440-447.

[9]

Krawczyk, T., Biodiesel: alternative fuel makes inroads but hurdles remain,
Inform, 7 (1996) 801814.

166

[10]

Petroleum Products Consumption, Energy Information Sheets Index, Energy


Information Administration, U.S. Department of Energy.
(http://www.eia.doe.gov/neic/infosheets/petroleumproductsconsumption.html)

[11]

International Petroleum (Oil) Consumption, Energy Information Administration,


U.S. Department of Energy.
(http://www.eia.doe.gov/emeu/international/oilconsumption.html)

[12]

Annual Energy Outlook, Energy Information Administration, U.S. Department of


Energy. (http://www.eia.doe.gov/oiaf/aeo/aeoref_tab.html)

[13]

Transportation: Diesel oil consumption, Energy and Resources, EarthTrends


(World Resources Institute). (http://earthtrends.wri.org/text/energyresources/variable-817.html)

[14]

National Biodiesel Board

[15]

European Biodiesel Board

[16]

Pure Energy Systems Wiki. Directory: Biodiesel for Algae Oil


(http://peswiki.com/index.php/Directory:Biodiesel_from_Algae_Oil)

[17]

Fitzgerald, M., India's Big Plans for Biodiesel, Technology Review


(Massachusetts Institute of Technology), December 27, 2006.
(http://www.technologyreview.com/Energy/17940/)

[18]

Sheehan, J., Camobreco, V., Duffield, J., Graboski, M. and Shapouri, H., Life
Cycle Inventory of Biodiesel and Petroleum Diesel for Use in an Urban Bus,
Final Report, National Renewable Energy Laboratory, NREL/SR-580-24089 UC
Category 1503, May, 1998. (http://www.nrel.gov/docs/legosti/fy98/24089.pdf)

167

[19]

Shapouri, H., Duffield, J.A. and Graboski, M.S., Estimating the Net Energy
Balance of Corn Ethanol, Agricultural Economics Report No. (AER721) 24 pp,
July 1995. (www.ers.usda.gov/publications/aer721/)

[20]

Hill, J., Nelson, E., Tilman, D., Polasky, S. and Douglas, T., Environmental,
economic, and energetic costs and benefits of biodiesel and ethanol biofuels,
Proc. Nat. Acad. Sci. U.S.A., 103 (2006) 11206-11210.

[21]

Railways to plant jatropha for captive use, The Hindu Business Line, October
19, 2007.
(http://www.thehindubusinessline.com/2005/10/20/stories/2005102002021100.ht
m)

[22]

Johnston, M., and Holloway, T., A global comparison of national biodiesel


production potentials, Environ. Sci. Technol., 41 (2007) 79677973.

[23]

DC Biodiesel Resource Guide, The Chesapeake Chapter of the Urban and


Regional Information Systems Association (CCURISA).
(http://www.ccurisa.org/?q=node/90)

[24]

2007 Commodity Articles, Fundamentals, Commodity Research Bureau.


(http://www.crbtrader.com/fund/articles/tallow.asp)

[25]

Brune, D.E., Yen, H.W., Schwartz, G., Benemann, J.R., Massingill, M.J., Van
Olst, J.C. and Carlberg, J.A., The Controlled Eutrophication process: Using
Microalgae for CO2 Utilization and Agircultural Fertilizer Recycling at the
Salton Sea, California", Second Annual Conference on Carbon Sequestration
(2003), National Energy Technology Laboratory.
(http://www.netl.doe.gov/publications/proceedings/03/carbon-seq/PDFs/034.pdf)

168

[26]

The UNH Biodiesel Group, University of New Hampshire.


(http://www.unh.edu/p2/biodiesel/about_index.html)

[27]

A Look Back at the U.S. Department of Energys Aquatic Species Program:


Biodiesel from Algae, July 1998, Report Number NREL/TP-580-24190,
National Renewable Energy Laboratory.
(http://www.nrel.gov/docs/legosti/fy98/24190.pdf)

[28]

Chisti, Y., Biodiesel from microalgae, Biotechnology Advances, 25 (2007) 294306.

[29]

Biodiesel Handling and Use Guidelines, DOE/GO-102006-2358, Third Edition


September 2006, U.S. Department of Energy.
(http://www.nrel.gov/vehiclesandfuels/npbf/pdfs/40555.pdf)

[30]

A Comprehensive Analysis of Biodiesel Impacts on Exhaust Emissions, October


2002 Draft Technical Report, EPA420-P-02-001, U.S. Environmental Protection
Agency. (www.epa.gov/OMS/models/biodsl.htm).

[31]

Schwab, A.W., Bagby, M.O., Freedman, B., Preparation and properties of diesel
fuels from vegetable oils, Fuel, 66 (1987) 1372-1378.

[32]

Freedman, B., Butterfield, R.O., Pryde, E.H., Transesterification kinetics of


soybean oil, J. Am. Oil Chem. Soc., 63 (1986) 1375-1380.

[33]

Lee, I., Johnson, L.A. and Hammond, E.G., Use of branched-chain esters to
reduce the crystallization temperature of biodiesel, J. Am. Oil Chem. Soc., 72
(1995) 1155-1160.

169

[34]

Lotero, E., Liu, Y., Lopez, D.E., Suwannakarn, K., Bruce, D.A. and Goodwin Jr.,
J.G., Synthesis of Biodiesel via Acid Catalysis, Ind. Eng. Chem. Res., 44 (2005)
5353-5363.

[35]

Zhang, Y., Dube, M.A., McLean, D.D. and Kates, M., Biodiesel production from
waste cooking oil: 1. Process design and technological assessment, Bioresour.
Technol., 89 (2003) 1-16.

[36]

Chongkhong, S., Tongurai, C., Chetpattananondh, P. and Bunyakan, C.,


"Environmental, economic, and energetic costs and benefits of biodiesel and
ethanol biofuels, Biomass Bioenergy, 31 (2007) 563-568.

[37]

Ono, Y. and Baba, T., Selective reactions over solid base catalysts, Catal.
Today, 38 (1997) 321-337.

[38]

Hattori, H., Solid base catalysts: generation of basic sites and application to
organic synthesis, Appl. Catal., A, 222 (2001) 247-259.

[39]

Handa, H., Fu, Y., Baba, T. and Ono, Y., Characterization of strong solid bases
by test reactions, Catal. Lett., A, 59 (1999) 195-200.

[40]

Coluccia, S. and Tench, A.J., 7th Proc. Int. Congr. Catal, (1980) p.1160.

[41]

Tanabe, K. and Hlderich, W.F., Industrial application of solid acid-base


catalysts, Appl. Catal., A, 181 (1999) 399-434.

[42]

Sohn, J.R., Recent advances in solid superacids, J. Ind. Eng. Chem., 10 (2004)
1-15.

[43]

Hammett, L.P. and Deyrup, A.J., A series of simple basic indicators. I. The
acidity functions of mixtures of sulfuric and perchloric acids with water, J. Am.
Chem. Soc., 54 (1932) 2721-2739.

170

[44]

Matsuhashi, H., Miyazaki, H., Kawamura, Y., Nakamura, H. and Arata., K.,
Preparation of a Solid Superacid of Sulfated Tin Oxide with Acidity Higher
Than That of Sulfated Zirconia and Its Applications to Aldol Condensation and
Benzoylation, Chem. Mater., 13 (2001) 3038-3042.

[45]

Yang, S., Bai, A. and Sun, J., Preparation of SO4 2/TiO2-La2O3 solid superacid
and its catalytic activities in acetalation and ketalation, J. Zhejiang Univ. Sci. B,
7 (2006) 553-558.

[46]

Hino, M. and Arata, K., Synthesis of Solid Superacid Catalyst with Acid
Strength of H0 16.04, J. Chem. Soc., Chem. Commun., (1980) 851-852.

[47]

Peterson, G.R. and Scarrah, W.P., Rapeseed Oil Transesterification by


Heterogeneous Catalysis, J. Am. Oil Chem. Soc., 61 (1984) 1593-1597.

[48]

Leclercq, E., Finiels, A. and Moreau, C., Transesterification of Rapeseed Oil in


the Presence of Basic Zeolites and Related Solid Catalysts, J. Am. Oil Chem.
Soc., 78 (2001) 1161-1165.

[49]

Suppes, G.J., Dasari, M.A., Doskocil, E.J., Mankidy, P.J. and Goff, M.J.,
Transesterification of soybean oil with zeolite and metal catalysts, Appl. Catal.,
A, 257 (2004) 213-223.

[50]

J.P. Cosman, P.D. Bettge, R.J. Elwell, R.A. and Plepys, Process for purifying
polyethers, US Patent 6,376,625 (2002).

[51]

Mazzocchia, C., Modica, G., Nannicini, R. and Kaddouri, A., Fatty acid methyl
esters synthesis from triglycerides over heterogeneous catalysts in the presence of
microwaves, C. R. Chim., 7 (2004) 601-605.

171

[52]

Lotero, E., Goodwin Jr., J.G., Bruce, D.A., Suwannakarn, K., Liu, Y., Lopez,
D.E, ed. Spivey, J., The Catalysis of Biodiesel Synthesis in Catalysis, Royal
Society of Chemistry, London, 2006.

[53]

Gryglewicz, S., Rapeseed oil methyl esters preparation using heterogeneous


catalysts, Bioresour. Technol., 70 (1999) 249-253.

[54]

Lopez Granados, M., Zafra Poves, M.D., Martin Alonso, D., Mariscal, R., Cabello
Galisteo, F., Moreno Tost, R., Santamaria, J. and Fierro, J.L.G., Biodiesel from
sunflower oil by using activated calcium oxide, App. Cat. B, 73 (2007) 317326.

[55]

Schuchardt, U., Vargas, R.M. and Gelbard, G., Transesterification of soybean oil
catalyzed by alkylguanidines heterogenized on different substituted polystyrenes,
J. Mol. Catal. A: Chem., 109 (1996) 37-44.

[56]

Soumanou, M.M. and Bornscheuer, U.T., Improvement in lipase-catalyzed


synthesis of fatty acid methyl esters from sunflower oil, Enzyme Microb.
Technol., 33 (2003) 97-103.

[57]

Du, W., Xu, Y. and Liu, D., Lipase-catalysed transesterification of soya bean oil
for biodiesel production during continuous batch operation, Biotechnol. Appl.
Biochem., 38 (2003) 103-106.

[58]

Shieh, C.-J., Liao, H.-F. and Lee, C.-C., Optimization of lipase-catalyzed


biodiesel by response surface methodology, Bioresour. Technol., 88 (2003) 103106.

[59]

Kim, H.J., Kang, B.S., Kim, B.J., Park, Y.M., Kim, D.K., Lee, J.S. and Lee, K.Y.,
Transesterification of vegetable oil to biodiesel using heterogeneous base
catalyst, Catal. Today, 93-95 (2004) 315-320.

172

[60]

Suzukamo, G, Fukao, M. and Minobe, M., Preparation of New Solid Superbase


and Its Catalytic Activity, Chem. Lett., (1987) 585-588.

[61]

Furuta, S., Matsuhashi, H. and Arata, K., Biodiesel fuel production with solid
superacid catalysis in fixed bed reactor under atmospheric pressure, Catal.
Commun., 5 (2004) 721-723.

[62]

Lopez, D.E., Goodwin Jr., J.G., Bruce, D.A. and Furuta, S., Esterification and
transesterification using modified-zirconia catalysts, Appl. Catal., A, 339 (2008)
76-83.

[63]

Furuta, S., Matsuhashi, H. and Arata, K., Catalytic action of sulfated tin oxide
for etherification and esterification in comparison with sulfated zirconia, Appl.
Catal., A, 269 (2004) 187-191.

[64]

Bournay, L, Casanave, D., Delfort, B., Hillion, G. and Chodorge, J.A., New
heterogeneous process for biodiesel production: A way to improve the quality and
the value of the crude glycerin produced by biodiesel plants, Catal. Today, 106
(2005) 190-192.

[65]

Bournay, L., Hillion, G., Boucot, P., Chodorge, J., Bronner, C. and Forestiere, A.,
Process for producing alkyl esters from a vegetable or animal oil and an aliphatic
monoalcohol, U.S. Patent 6,878,837 (2005).

[66]

Hillion, G. and LePennec, D., Process for the alcoholysis of acid oils of
vegetable or animal origin, U.S. Patent 7,420,073 (2008).

[67]

Saka, S. and Kusdiana, D., Biodiesel fuel from rapeseed oil as prepared in
supercritical methanol, Fuel, 80 (2001) 225-231.

173

[68]

Saka, S. and Kusdiana, D., Kinetics of transesterification in rapeseed oil to


biodiesel fuel as treated in supercritical methanol, Fuel, 80 (2001) 693-698.

[69]

Cao, W., Han, H. and Zhang, J., Preparation of biodiesel from soybean oil using
supercritical methanol and co-solvent, Fuel, 84 (2005) 347-351.

[70]

Suppes, G.J., Bockwinkel, K., Lucas, S., Botts, J.B., Mason, M.H. and Heppert,
J.H., Calcium Carbonate Catalyzed Alcoholysis of Fats and Oils, J. Am. Oil
Chem. Soc., 78 (2001) 139-145.

[71]

Test Method for Determination of Free and Total Glycerin in B-100 Biodiesel
Methyl Esters by Gas Chromatography, Designation: D 6584-00, ASTM
International.

[72]

Wulfsberg, G., Principles of Descriptive Chemistry, Brooks/Cole Publishing:


Monterey CA, 1987, p.37.

[73]

Somorjai, G.A., Introduction to Surface Chemistry and Catalysis, John Wiley &
Sons, Inc. 1994, p. 452.

[74]

Minami, E. and Saka, S., Kinetics of hydrolysis and methyl esterification for
biodiesel production in two-step supercritical methanol process, Fuel, 85 (2006)
2479-2483.

[75]

Benenati, R.F. and Brosilow, C.B., Void Fraction Distribution in Beds of


Spheres, AlChE J., 8 (1962) 359-361.

[76]

Mueller, G.E., Radial void fraction distributions in randomly packed tied beds of
uniformly sized spheres in cylindrical containers, Powder Technol., 72 (1992)
269-275.

174

[77]

Hoffmann, A.C. and Finkers, H.J., A relation for the void fraction of randomly
packed particle beds, Powder Technol., 82 (1995) 197-203.

[78]

Doraiswamy, L.K., Organic Synthesis Engineering, 1st edition, Oxford University


Press US, 2001, p. 171.

[79]

Lee, S., Methanol Synthesis Technology, CRC Press: Boca Raton, Florida, 1990,
p. 54.

[80]

Tang, Z., Du. Z., Min, E., Gao, L., Jiang, T. and Han, B., Phase equilibria of
methanoltriolein system at elevated temperature and pressure, Fluid Phase
Equilib., 239 (2006) 8-11.

[81]

Austic, G., Gombotz, K., Kannan, D.C. and Matson, J.V., USDA Small Business
Innovation Program Phase I Project Report, July 2008-April 2009.

[82]

Greenwood, N.; Earnshaw, A., Chemistry of the Elements, 2nd Edition, 1997
Oxford: Butterworth-Heinemann.

[83]

Sodium Chloride, Wikipedia. (http://en.wikipedia.org/wiki/Sodium_chloride)

[84]

WebElements Periodic Table. (http://www.webelements.com/compounds)

[85]

Forsyth, J.B. and Hull, S., The effect of hydrostatic pressure on the ambient
temperature structure of CuO, J. Phys.: Condens. Matter, 3 (1991) 5257-5261.

[86]

Jeong, N.C., Lee, J.S., Tae, E.L., Lee, Y.J. and Yoon, K.B., Acidity Scale for
Metal Oxides and Sandersons Electronegativities of Lanthanide Elements,
Angew Chem. Int. Ed., 47 (2008) 10128-10132.

[87]

Klepel, O. and Hunger, B., Temperature-programmed desorption of carbon


dioxide on alkali-metal cation-exchanged faujasite type zeolites, J. Therm. Anal.
Calorim., 80 (2005) 201-206.

175

[88]

Csontos, G., Heil, B. and L. Mark, Hydroformylation of olefins with Rhodium


carbonyls as catalysts. IV. Mechanism of the reaction, Ann. N.Y. Acad. Sci., 239
(1974) 47-54.

[89]

Turberfield, A.J., Mitchell, J.C., Yurke, B., Mills Jr., A.P., Blakey, M.I. and
Simmel, F.C., DNA Fuel for Free-Running Nanomachines, Phys. Rev. Lett., 90
(2003) 118102-1-118102-4.

[90]

Chen, J., Chen, L. and Cheng, W., Kinetics of glycolysis of polyethylene


terephthalate with zinc catalyst, Polym. Int., 48 (1999) 885-888.

[91]

Demirbas, A., Biodiesel from sunflower oil in supercritical methanol with


calcium oxide, Energy Convers. Manage., 48 (2007) 937-941.

[92]

Perrys Chemical Engineers Handbook, 7th Edition, 1997 (Ed: R.H. Perry, D.W.
Green, and J.O. Maloney), McGraw-Hill, New York.

[93]

Andrade, E.N.D., A theory of the viscosity of liquids - Part I, Philos. Mag., 17


(1934) 497-511.

[94]

Zberg-Mikkelsen, C.K., Baylaucq, A., Watson, G. and Boned, C., HighPressure Viscosity Measurements for the Ethanol + Toluene Binary System, Int.
J. Thermophys., 26 (2005) 1289-1302.

[95]

Tanaka, Y., Matsuda, Y., Fujiwara, H., Kubota, H. and Makita, T., Viscosity of
(Water + Alcohol) Mixtures Under High Pressure, Int. J. Thermophys., 8 (1987)
147-163.

[96]

Abramovi, H. and Klofutar, C., The temperature dependence of dynamic


viscosity for some vegetable oils, Acta Chim. Slov., 45 (1998) 69-77.

176

[97]

Davis, J.P., Dean, L.O., Faircloth, W.H. and Sanders, T.H., Physical and
Chemical Characterizations of Normal and High-Oleic Oils from Nine
Commercial Cultivars of Peanut, J. Am. Oil Chem. Soc., 85 (2008) 235-243.

[98]

Esko Industries, Ltd., Tools, Advanced viscosity calculator.


(http://www.eskoindustries.com/viscadv.php)

[99]

Khandan, N., Kazemeini, M. and Aghaziarati, M.,"Determining an optimum


catalyst for liquid-phase dehydration of methanol to dimethyl ether", Appl. Catal.,
A, 349 (2008) 6-12.

[100] Di Cosimo, J.I., Dez, V.K., Xu, M., Iglesia, E. and Apestegua, C.R., "Structure
and Surface and Catalytic Properties of Mg-Al Basic Oxides", J. Catal., 178
(1998) 499-510.
[101] Kim, S., Lee, Y., Bae, J.W., Potdar, H.S. and Jun, K.," Synthesis and
characterization of a highly active alumina catalyst for methanol dehydration to
dimethyl ether", Appl. Catal., A, 348 (2008) 113-120.
[102] Barrault, J., Pouilloux, Y., Clacens, J.M., Vanhove, C. and Bancquart, S.,
Catalysis and fine chemistry, Catal. Today, 75 (2001) 177-181.
[103] Zhenyi, C., Xing, J. Shuyuan, L. and Li, L., Thermodynamics Calculation of the
Pyrolysis of Vegetable Oils, Energy Sources, 26 (2004) 849-856.
[104] Demirbas, A. and Kara, H., New Options for Conversion of Vegetable Oils to
Alternative Fuels, Energy Sources Part A, 28 (2006) 619-626.
[105] Sadrameli, S.M. and Green, A.E.S., Systematics of renewable olefins from
thermal cracking of canola oil, J. Anal. Appl. Pyrolysis, 78 (2007) 445-451.

177

[106] Idem, R.O., Katikaneni, S.P.R. and Bakhshi, N.N., Thermal Cracking of Canola
Oil: Reaction Products in the Presence and Absence of Steam, Energy Fuels, 10
(1996) 1150-1162.
[107] Idem, R.O., Katikaneni, S.P.R. and Bakhshi, N.N., Catalytic conversion of
canola oil to fuels and chemicals: roles of catalyst acidity, basicity and shape
selectivity on product distribution, Fuel Process. Technol., 51 (1997) 101- 125.
[108] Bernstein, M., Heat Polymerization of Nonconjugated Vegetable Oils, J.
Polym. Sci., 1 (1946) 495-528.
[109] Rhoades, W.F. and Da Valle, A.J., Heat Polymerization of Safflower Oil, J.
Am. Oil Chem. Soc., 28 (1951) 466-468.
[110] Tateno, T. and Sasaki, T., Process for producing fatty acid esters and fuels
comprising fatty acid ester, US Patent 6,818,026 (2004).
[111] Goto, F. and Sasaki, T., Method for producing fatty acid ester and fuel
containing fatty acid ester, Japanese Patent Publication number 2003-055299
(2003).
[112] McNeff, C.V., McNeff, L.C., Yan, B., Nowlan, D.T., Rasmussen, M., Gyberg,
A.E., Krohn, B.J., Fedie, R.L., Hoye, T.R, A continuous catalytic system for
biodiesel production, Appl. Catal., A, 343 (2008) 39-48.
[113] Di Serio, M., Cozzolino, M., Giordano, M., Tesser, R., Patrono, P. and
Santacesaria, E., From Homogeneous to Heterogeneous Catalysts in Biodiesel
Production, Ind. Eng. Chem. Res., 46 (2007) 6379-6384.

178

[114] Di Serio, M., Ledda, M., Cozzolino, M., Minutillo, G., Tesser, R. and
Santacesaria, E., Transesterification of Soybean Oil to Biodiesel by Using
Heterogeneous Basic Catalysts, Ind. Eng. Chem. Res., 45 (2006) 3009-3014.
[115] Liu, Y., Lotero, E., Goodwin Jr., J.G., Mo, X., Transesterification of poultry fat
with methanol using MgAl hydrotalcite derived catalysts, Appl. Catal., A, 331
(2007) 138148.
[116] Perin, G., lvaro, G., Westphal, E., Viana, L.H., Jacob, R.G., Lenardo, E.J. and
DOca, M.G.M., Transesterification of castor oil assisted by microwave
irradiation, Fuel, 87 (2008) 2838-2841.
[117] Xie, W., Peng, H. and Chen, L., Calcined Mg-Al hydrotalcites as solid base
catalysts for methanolysis of soybean oil, J. Mol. Catal. A: Chem., 246 (2006)
24-32.
[118] Di Serio, M., Tesser, R., Pengmei, L. and Santacesaria, E., Heterogeneous
Catalysts for Biodiesel Production, Energy Fuels, 22 (2008) 207217.
[119] Lacome, T., Hillion, G., Delfort, B., Revel, R., Leporq, S. and Acakpo, G.,
Process for transesterification of vegetable or animal oils using heterogeneous
catalysts based on titanium, zirconium or antimony and aluminum, U.S. Patent
Application No. US 2005/0266139, Dec 1, 2005.
[120] Stern, R., Hillion, G., Rouxel, J. and Leporq, J., Process for the production of
esters from vegetable oils or animal oils alcohols, U.S. Patent 5,908,946 (1999).
[121] Sreeprasanth, P. S., Srivastava, R., Srinivas, D. and Ratnasamy, P.,
Hydrophobic, solid acid catalysts for production of biofuels and lubricants,
Appl. Catal., A, 314 (2006) 148159.

179

[122] Yan, S., Salley, S.O. and Simon Ng, K.Y., Simultaneous transesterification and
esterification of unrefined or waste oils over ZnO-La2O3 catalysts, Appl. Catal.,
A, 353 (2009) 203212.
[123] Portnoff, M.A., Purta, D.A., Nasta, M.A., Zhang, J. and Pourarian, F., Methods
for producing biodiesel, PCT No. WO2006/002087, Jan 5, 2006.
[124] Refaat, A.A. and El Sheltaway, S.T., Time Factor in Microwave-enhanced
Biodiesel Production, WSEAS Transactions on Environment and Development, 4
(2008) 279288.
[125] Perin, G., lvaro, G., Westphal, E., Viana, L.H., Jacob, R.G., Lenardo, E.J. and
DOca, M.G.M., Transesterification of castor oil assisted by microwave
irradiation, Fuel, 87 (2008) 2838-2841.
[126] Noureddini, H. and Zhu, D., Kinetics of Transesterification of Soybean Oil, J.
Am. Oil Chem. Soc., 74 (1997) 14571463.
[127] Haas, M.J., McAloon, A.J., Yee, W.C. and Foglia, T.A., A process model to
estimate biodiesel production costs, Bioresour. Technology, 97 (2006) 671678.
[128] Skipski, V.P., Good, J.J., Barclay, M. and Reggio, R.B., Quantitative analysis of
simple lipid classes by thin-layer chromatography, Biochim. Biophys. Acta, 152
(1968) 10.
[129] Jork, H., Funk, W., Fischer, W. and Wimmer, H., Thin-layer chromatography:
reagents and detection methods (English Translation of DnnschichtChromatographie.), New York, NY (USA): VCH, 1990, p. 405.
[130] Zeolites and Ordered Mesoporous Materials: Progress and Prospects, Ed: J. ejka
and H. van Bekkum, Elsevier, Amsterdam, Stud. Surf. Sci. Catal., 157 (2005).

180

Appendices

Appendix A
Thin Layer Chromatography (TLC):
Thin Layer Chromatography is based on the principle that components differ in the rate at
which they elute a porous plate placed on a suitable solvent mixture. A spot of the
sample to be tested was placed on Merck silica gel 40 plate. It was then placed in a
solvent chamber with solvents hexane and ether in 5:1 ratio by volume and covered.
Solvent elutes through the plate and the components present in the spot elute according to
their characteristics (size, solvent affinity, functional group). The plate was placed in the
solvent chamber until the solvent front reaches the top and was removed from the
chamber once the solvent front reached the top. The plate was then air dried and dipped
in a detecting reagent to see the various elution fronts of the components in the originally
placed spot [128]. The detecting reagent is 50 mg of Rhodamine 6G in 100 ml of ethanol
[128, 129]. The plate was dipped into the detecting reagent for a second and air dried for
few minutes until the various elution fronts of the components start to appear. Each
component has its own characteristic retention factor. The original vegetable oil, final
reacted ethyl esters and intermediate mono and diglycerides can be identified by their
retention factor, if they are present. The retention factor of each component can be
known by testing with the pure component spots. This is a qualitative and a semiquantitative analysis test. It is semi quantitative, in the sense, that relatively bigger and
stronger spots mean more concentration of a particular component in the original spot,

181

though it would not give accurate results and the size or intensity may not be correlated to
exact proportionality.

Figure A.1. Photo: Typical TLC analytic plates

Appendix B
Gas Chromatography (GC):
Principle:
Gas Chromatography is based on the principle that components, when carried by a carrier
gas through a chromatography column, differ in retention times. The difference in
retention time is due to difference in molecular size and the interaction of the components
with the column. A longer molecule tends to take more time than a shorter molecule to
elute. A polar component tends to have a shorter retention time than a comparable non182

polar compound if the column coating is non-polar. The column can also be temperature
programmed to analyze a range of components.

Instrument set-up:
The gas chromatography analysis procedure was adopted from the ASTM method D
6584 to analyze biodiesel composition [71]. The gas chromatograph used was a Varian
3600. The gas chromatography column used was an open tubular column with a 5%
phenyl polydimethlylslioxane bonded and crosslinked phase internal coating with 0.32
mm internal diameter and 0.1 m film thickness. A computer capable of providing real
time graphic and digital presentation of the chromatographic data was used with the
software corresponding to the Varian 3600 gas chromatograph. Peak areas and the
retention times were measured by the computer and the measurement settings can be
changed as necessary.

Detector type used was a flame ionization detector. The detector temperature was set at
380 C. Carrier gas used was Helium at a flow rate of 3 ml/minute. The ASTM
procedure [71] corresponds to a cool-on column injector. A cool-on column injector
allows for the sample to be injected directly into the column that extends to the injector
and tracks and follows the programmed temperature of the GC oven. The sample
contains high boiling components that may otherwise condense and be retained on the
glass lining of a normal injector. Since the cool-on column injector is expensive, an
alternate way of injecting the components was adopted. The sample was injected directly
into the column inside the oven (instead of an external injector) using a syringe with 15

183

cm long needle. A suitable glass lining that is narrow enough to compactly house the GC
column that extends into the normal injector was used. Since the sample is injected
directly into the oven, there is no problem of the high boiling components left over
outside the oven.

The following table shows the temperature program adopted for the chromatographic run.
Column Temperature Program
Initial temperature

50C

Hold 1 minute

Rate 1

15C/minute to 180 C

Rate 2

7C/minute to 230 C

Rate 3

30C/minute to 380 C
Table A.1. GC column temperature program

184

Hold 10 minutes

Figure A.1. Photo: GC oven and chromatographic column

Figure A.2. Photo: Real-time graphics of GC analysis


185

Figure A.3. Photo: GC front view

Standardization, calibration and analysis:


The biodiesel conversion of the reaction was measured by calculating the content of the
reactant triglycerides and intermediate products, diglycerides and monoglycerides, as
prescribed in the ASTM method D 6584 [71].

Glycerol and pure oleic acid derivatives were used for standardization. Oleic acid
derivatives can be related to quantify the similar vegetable oil components (stearic,
palmitic, linoleic, linolenic acid derivatives). Standards consist of the pure components
dissolved in pyridine. Two internal standards, tricaprin and 1,2,4-butanetriol, were used
to quantify fatty acid derivatives and glycerol respectively. Standardization samples with
different concentrations of the pure components were prepared, injected into the GC,
186

peaks were identified by the pure components and the corresponding peak areas were
calibrated using the internal standards. The calibration was then used in the quantitative
analysis of the samples that need to be analyzed.

The samples, standards as well as product mixtures from the reactor, were measured and
added to glass vials. The internal standards were then added. The samples were then
silyated with N-methyl-N-trimethylsilyltrifluora acetamide (MSFTA). After 15~20
minutes, the solvent heptane was added. The size of the sample injected into the column
was 1 microliter.

Figure A.4 and Figure A.5 show representative chromatograms of samples analyzed for
biodiesel composition. The samples correspond to experiments run on 200-60 mesh
MnO packed in the reactor at 260 C and 70 atm. Figure A.4 represents a sample
collected during activity sustenance tests at a residence time of 5 minutes and had 75 %
conversion. Figure A.5 represents a sample from the reaction kinetics studies that had a
residence time of 13.6 minutes and 95 % conversion.

187

Heptane
solvent

Biodiesel esters

Monoglycerides

Tricaprin internal standard

Signal (volts)

Triglycerides

Diglycerides

Time (minutes)

Figure A.4. Chromatogram of MnO catalyzed sample (260 C, 70 atm, 5 min)

188

Biodiesel esters

Signal (volts)

Tricaprin internal standard

Heptane
solvent

Monoglycerides
Triglycerides

Diglycerides

Time (minutes)

Figure A.5. Chromatogram of MnO catalyzed sample (260 C, 70 atm, 13.6 min)

Appendix C
Water and Saponification:
Water present in reaction mixture would form FFA through hydrolysis of triglycerides.
The presence of a metal, either as ion or metal oxide, can neutralize FFA forming soap.
This is the mechanism by which the well known saponification of triglycerides by
aqueous NaOH happens. Just like sodium, it can happen with any reactive metal. In the
case of CaO and water, water can ionize calcium methoxide (formed by reaction of CaO

189

and methanol) just like it ionizes sodium methoxide to form soap in conventional base
catalyzed process. Calcium ions thus formed or CaO itself can react with FFA formed by
triglyceride hydrolysis in a neutralization reaction forming soap.

Appendix D
Titrimetry:
Fatty acid concentrations can be determined by titrimetry. The following method can be
used to determine fatty acid concentrations equal to or higher than.1 milliMolar.

Reagents:
Solvent mixture: 95% ethanol/diethyl ether, 1/1, v/v,
0.1 M KOH in ethanol accurately standardized with 0.1 M HCl,
Indicator: 1 % phenolphthalein in 95% ethanol.

Procedure:
0.1 to 10 g of oil, fatty acid or reaction product mixture (according to the expected acid
value) was weighed in glass vial and dissolved in at least 50 ml of the solvent mixture.
5 drops of indicator was added to the glass vial mixture. It was then titrated, with
shaking, with the KOH solution (in a 25 ml burette graduated in 0.1 ml) to the end point
of the indicator, the pink color persisting for at least 10 s.

The free fatty acid content (% by wt) is calculated by the formula:


M KOH x n proton x

V
1
x MWacid x 100
x
1000 m

190

where
M KOH is the molarity of KOH solution,
n proton is the number of protons exchanged in a reaction involving KOH and the acid (1 in
this case),
V is the volume of KOH solution used (ml),
m is the mass in g of the sample,.
MWacid is the molecular weight of the acid measured (282.46 in this case, for oleic acid).

Appendix E
LHSV and WHSV:
Liquid Hourly Space Velocity (LHSV) is the volume of liquid processed in one hour
through the unit volume of reactor.

Weight Hourly Space Velocity (WHSV) is the weight of product (biodiesel in this case)
produced in one hour by passing through the unit weight of catalyst. It is a measure of
productivity. This hourly production rate per unit weight of the catalyst is the number the
commercial production facilities are most concerned of.

Appendix F
ZSM-5:
ZSM-5 is a zeolite composed of pentasil units linked together by oxygen bridges to form
pentasil chains. Each pentasil unit consists of eight five-membered rings, with Al or Si as
vertices and oxygen bonded between the vertices. The pentasil chains together form a

191

corrugated sheet with 10-ring holes which has Al or Si vertices like the rings of pentasil
unit. Several corrugated sheets are connected together by oxygen bridges with straight
10-ring channels running parallel and sinusoidal 10-ring channels running perpendicular
to the corrugated sheets [130]. The pore size of the channel (~ 5.5 ) makes them shapeselective molecular sieves.

Appendix G
Cetane number:
Cetane number of a fuel is defined as the percentage by volume of normal cetane in a
mixture of normal cetane and alpha-methyl naphthalene which has the same ignition
characteristics (ignition delay) as the test fuel when combustion is carried out in a
standard engine under specified operating conditions.

Cetane number is actually a measure of a fuel's ignition delay; the time period between
the start of injection and start of combustion (ignition) of the fuel. In a particular diesel
engine, higher cetane fuels will have shorter ignition delay periods than lower cetane
fuels.

192

Cetane is an un-branched open chain alkane molecule that ignites very easily under
compression, so it was assigned a cetane number of 100, while alpha-methyl napthalene
was assigned a cetane number of 0. All other hydrocarbons in diesel fuel are indexed to
cetane as to how well they ignite under compression. The cetane number therefore
measures how quickly the fuel starts to burn (auto-ignites) under diesel engine
conditions. Since there are hundreds of components in diesel fuel, with each having a
different cetane quality, the overall cetane number of the diesel is the average cetane
quality of all the components. There is very little actual cetane in diesel fuel.

Appendix H
Capital Cost Estimation of 34 Million Gallon per Year Biodiesel Plant:
Chris Torres, employed by Matson Biofuels, carried out a capital cost estimate of the
MnO solid catalyst method, dubbed the Green Biodiesel process, in the summer of 2007.
The annual production capacity of the plant estimated for cost was 34 million gallons.
Though some of the processing details (process heat (steam)) were not finalized, the
analysis was good enough for capital estimate. The main details are traced and listed
below.

The design was done with process parameters such as molar ratio, reaction rate and
packing fraction optimized for the 34 million gallon per year commercial production.
Table A.2 shows the overall equipment list. The table also shows equipments related to
water washing, since at that time it was felt water washing may be needed at least once.
Table A.3 shows the detailed capital estimate of the individual equipments. As part of

193

the overall design, the annual utility cost was also calculated and it was also accounted
for in the overall capital cost. Table A.4 shows the overall capital cost of the 34 million
gallon per year plant. The total capital cost is estimated to be $ 16.1 million. This is 58
% lower than the capital cost of a conventional plant. Table A.5 shows the summary of
the overall design estimates. It includes estimates for the scenario of glycerol yielding
maximum value as pure product.

194

Process Segment

Component Name

Total Direct Cost


($)

Catalyst Preparation
Catalyst Preparation
Catalyst Preparation

Cat Prep Tank


MeOH To Cat Prep
Cat Prep Pump Out

Reactor Area
Reactor Area
Reactor Area
Reactor Area
Reactor Area

Reactor
Oil To Reactor
Heating Loop Pump
Reactor Heater
Crude Output Pump

Flash
Flash
Flash
Flash
Flash
Flash

Equipment Cost
($)

729,900.00
33,300.00
53,200.00

480,200.00
5,400.00
10,200.00

1,600,000.00
41,400.00
93,900.00
173,400.00
94,900.00

457,600.00
7,400.00
14,600.00
40,000.00
15,400.00

Methanol Flash
Condenser
Liquid Output Pump
Flash Heater
Liq MeOH Holdup
Liq MeOH Output Pump

329,000.00
105,700.00
45,200.00
156,300.00
110,700.00
28,600.00

142,200.00
36,000.00
7,700.00
58,700.00
25,000.00
5,700.00

Primary Separations
Primary Separations
Primary Separations

Separation Tank
Glycerol Pump
Crude Pump

345,500.00
24,500.00
34,100.00

185,400.00
4,300.00
5,300.00

Water Wash
Water Wash
Water Wash

Water Pump In
Water Wash Tank
Crude Pump Out

28,100.00
444,600.00
32,900.00

5,000.00
264,700.00
5,700.00

MtOH/H2O Distillation
MtOH/H2O Distillation
MtOH/H2O Distillation
MtOH/H2O Distillation
MtOH/H2O Distillation
MtOH/H2O Distillation
MtOH/H2O Distillation
MtOH/H2O Distillation

Distillation Column
Condenser
Reboiler
Reflux Pump
Reflux Drum
Water Recycle Pump
MeOH Recycle Pump
Wash Water Pump In

276,700.00
85,800.00
63,900.00
41,800.00
72,500.00
28,400.00
22,700.00
29,100.00

106,300.00
22,700.00
20,500.00
19,400.00
13,900.00
5,000.00
4,300.00
6,100.00

Product Drying
Product Drying
Product Drying
Product Drying
Product Drying
Product Drying

Flash Dryer
Dryer Heater
Crude Pump Out
H2O Condenser
Product Cooler
Liq H2O Holdup Tank

105,500.00
71,700.00
33,600.00
69,400.00
69,700.00
60,100.00

27,500.00
18,000.00
6,300.00
35,800.00
17,100.00
8,300.00

Table A.2. Equipment list.

195

Direct
Installed
cost
M$

Allocated Costs
Indirects
Eng/Pro
Rack/Sewers
M$
M$

T otal
Contractor
Cost
M$

Catalyst Pre paration


Catalyst Prep T ank
MeOH T o Cat Prep
Catalyst Prep Pump

2
2
2

730
33
53
816

369
17
27
413

74
3
5
83

1173
54
85
1312

Reactor Unit
Reactor
Oil T o Reactor
Heating Loop Pump
Reactor Heater
Crude Ouptup Pump

2
2
2
2
2

1600
41
94
173
95
2004

809
21
47
88
48
1013

162
4
9
18
10
203

2571
67
151
279
153
3220

Me O H Flash Syste m
Flash T ower
Condenser
Liquid Output Pump
Flash Heater
Liq MeOH Holdup
Liq MeOH Output Pump

1
1
1
1
1
1

329
106
45
156
111
29
776

166
53
23
79
56
14
392

33
11
5
16
11
3
78

529
170
73
251
178
46
1246

Primary Separation
Water Pump In
Water Wash T ank
Crude Pump Out

1
1
1

346
25
34
404

175
12
17
204

35
2
3
41

555
39
55
649

Wate r Wash
Separation T ank
Glycerol Pump
Crude BioD Pump

1
1
1

28
445
33
506

14
225
17
256

3
45
3
51

45
714
53
812

Distillation System
Distillation Column
Condenser
Reboiler
Reflux Pump
Reflux Drum
Water Recycle Pump
MeOH Recycle Pump
Wash Water Pump In

1
1
1
1
1
1
1
1

277
86
64
42
73
28
23
29
621

140
43
32
21
37
14
11
15
314

28
9
6
4
7
3
2
3
63

445
138
103
67
117
46
36
47
998

Drying System
Flash Dryer
Dryer Heater
Crude Pump Out
H2O Condenser
Product Cooler
Liq H2O Holdup T ank
Liq H2O Pump Out
Vaccum Ejector

1
1
1
1
1
1
1
1

106
72
34
69
70
60
23

53
36
17
35
35
30
12
Not Yet Sized
219
2812

11
7
3
7
7
6
2

170
115
54
112
112
97
37

44
562

696
8934

Total IBL

433
5559

Table A.3. Detailed capital estimate.

196

Base Case: 34 MMGal/Yr Crude Output; 1 Train Model


Basis
Startup in 2007
33MMGal/Yr Crude Output
1 Process Trains
IBL

Rev 0

05/13/2007
Base Case
TCC, M$
1,312
3,220
1,246
649
812
998
696

Catalyst Preparation
Reactor System
Flash System
Primary Separation
Water Wash
Distillation System
Drying System
TCC IBL

7,240
500
1,140
1,099
100
2,838

OBL Site Development


Utilities
Raw Matl/Prod Storage
Environmental
TCC OBL
Total Contractor Cost, M$
Team Costs
Other Costs
PreOp & S/U (of TCC)

10,078
8%
4%
7%

of TCC
of TCC
of TCC

Current Point Estimate, M$


Escalation
Project Contingency

806
403
655
11,943

13%
20%

Escalation
Years
4
3%
of CPE+Esc

Total Plant Cost, M$

1,499
2,688

16,130

Table A.4. Total capital cost.

197

Biodiesel Production (34 million gallons per year)


Base Case
PG

Capacity, MM lb/yr
Capital, MM$

31356 lb/hr
January 2007 Startup

Units

unit/lb

/unit

Input & Summary Sheet


250.85
16.13

/lb Prod

Raw Materials
Oil
Methanol
NaOH
Water

lb
lb
lb
lb

31855 1.0159
16.7
6647 0.2120
15.6
40 0.0013
18.9
31513 1.0050
0.0
Total Raw Materials

ByProducts
Glycerol

lb

32193 1.0267
18.0 -18.48
Total ByProducts -18.48

Utilities
Fuel
MMBTU
-81 -0.0026
1000.0
Estimation of Cost Recovery Potential
if Glycerol is Used as a Fuel - Not added into Utilities
Refrigeration
MMBTU
0 0.0000
1800.0
LPS
MMBTU
34 0.0011
1400.0
Power
kw
122 0.0039
8.0
CW
MMBTU
13 0.0004
50.0
Waste Water
gal
811 0.0259
10.0
Total Utilities
Fixed Costs
$250M per man-shift

Labor, m/shift

250000.0

Overhead (1.5 x labor)

16.93
3.30
0.02
0.00
20.26

Economic Assumptions
Discount Rate
Income Tax Rate
Working Capital, % of Revenues
Inflation/Escalation
SG&A, % of Sales
hours/year

12%
35%
10%
2%
1%
8000

FYI Reference -> Product Price =

$ 0.58 /Gal

-2.58

0.00
1.50
0.03
0.02
0.26
1.81
0.40
0.60

Rep & maint.

% Capital = 1.5%

0.10

Property tax
Insurance

% Capital = 2.5%
% Capital = 0.5%
Fixed Costs

0.16
0.03
2.50

#
1

TOTAL CASH COSTS


SG&A
Return on Capital

CFPO, yrs =

6.09
0.08
1.50

4.27

0
0
1

NPW ($MM)

Required Product Price for 15.0% ATROR

7.67

ATROR, %

15.00

2.2

Table A.5. Summary of 34 million gallon/year plant design.

198

Appendix I
Process Design of a Mobile Biodiesel Production Unit:
Anita Lin, employed by Matson Biofuels, carried out an analysis of the MnO solid
catalyst method, dubbed the Green Biodiesel process, for a mobile biodiesel production
unit. The mobile production unit designed was to collect used oil and convert it to
biodiesel. The capacity was 50,000 gallons per year. The following is her report on the
design analysis of the unit.

Mobile Green Biodiesel Process and Analysis


Introduction
The term biodiesel is used to describe alternative diesel fuel sources from renewable,
organic sources. Typically biodiesel is composed of molecules of fatty acid methyl esters
derived from triacylglyeride molecules through transesterification. These long
hydrocarbon chains are similar to the hydrocarbons in conventional petroleum diesel, and
can be used in the same manner for fuel.

Description of conventional process


The conventional biodiesel commercial process has several definitive steps, those being
purification, esterification, mixing, transesterification, alcohol recovery, separation and
purification.

The first step, purification, is necessary for the reason that excess moisture, free fatty acid
chains, water and other impurities typically promote saponification in the later base
catalyzed transesterification. This is highly undesirable, as the soap not only clogs the

199

machinery but wastes reactants. The specific type of purification depends on the state of
the waste oil. Several different purification steps may have to be undertaken before the
oil is ready for the reactor. For example, if the oil is wet, drying is required. If there is
intolerable amount of gum, protein or particulate, those must be removed. After
purification, if there is a high percentage of free fatty acid, the oil must undergo acid
catalyzed esterification to remove them. This is often the case with waste oil, which may
be 30-40% FFA by mass of the oil. After esterification, the oil proceeds to the reactor to
undergo base catalyzed transesterification with catalyst and excess alcohol. Here, the
catalyst will abstract a hydrogen molecule from the alcohol and the alcohol will then
attack the nucelophilic carbonyl atom of the triacylglyceride. In this way the fatty acid
chain will be separated from the glycerol backbone. After the reactor, there is an alcohol
recovery unit, typically a distillation column or flash drum, which recovers excess
alcohol and recycles it back to the reactor. After methanol recovery the biodiesel and
glycerol can undergo separation and purification or polishing steps.

There are several disadvantages to this process; mainly that it is not flexible and is
sensitive to changes in feed stocks. In addition, the process requires many purification
and polishing steps, and has toxic components. A more robust, simplified system is
suggested in the form of the Mobile Green Biodiesel process.

Description of the mobile green biodiesel process


The mobile green biodiesel process (MGB) is derived from the research conducted by
Dheeban Kannan, whereby the conventional liquid catalyst is replaced with a recoverable
transition metal oxide solid catalyst at supercritical temperature and pressures.

200

Unlike the convention biodiesel process, the MGB uses a solid, nontoxic, recoverable
catalyst, is accommodating to a wide range of virgin oil and waste feedstocks and can
handle much greater moisture content (up to 5%). It is thus a simpler, more robust system
than the conventional process.

The Mobile aspect was developed with the problem of constant waste oil supply in mind.
A constant supply is difficult to maintain, especially in todays competitive market. A
biodiesel production system which is able to go to the waste source may be the solution
to this problem. Therefore the MGB plant is envisioned as being mounted on skids on a
pickup truck. Since the system has to be accommodated on a truck, the process was
designed with the space limitations that a conventional truck would be able to manage (2
to 3 meter long, 1.5 meters wide). It is estimated that the annual capacity of the plant
would be 50,000 gal/yr with the uptime being 4869 hrs/yr.

For the MGB process, several unit operations are necessary, these being the following.
1. A reactor to perform transesterfication and heat the reactants
2. A operation by which to recover excess alcohol
3. A operation by which to remove excess water
4. A operation by which to separate biodiesel and glycerin
5. A system to keep fluids moving and lastly
6. A system to recover heat

Therefore the green mobile biodiesel process is thus; after pretreatment (not shown) a
reactor (1) with solid catalyst produces biodiesel and glycerin from methanol/ethanol and

201

waste oil. After the reaction, the respective components need to be separated. To separate
the alcohol and water from the biodiesel and glycerin, a flash evaporator (2) is used. Then
water is removed from the alcohol by the use of a zeolite molecular sieve (3). The sieve
requires fairly constant regeneration; therefore 2 sieves are used in parallel, one of which
is in operation at any given time. The sieve that is not in operation will undergo
regeneration with hot dry air. The alcohol flowing from sieve is recycled to the reactor.
To separate the biodiesel and glycerin, a settling tank (4) is used.

The total capital costs for the plant are estimated at $63,439.60. Total operating costs per
year are estimated at $83,502.80 and the total manufacturing to produce one gallon of
biodiesel is $1.80/gal.

8
10
D

3.1

C
11

5
4.1
7

13

14

A Mixer
B Plug Flow Reactor
C Flash/Heat Exchanger
D - Condenser
E Heat Exchanger
F Settling tank
G Molecular Sieve
1, 2, 4 Pumps
3 Air blower
*The Red stream carries feedstock to be heated.
**The Green stream carries air that is heated to regenerate the adsorption bed.

Diagram 1 MGB Process Diagram

202

12

Equipment
Heat Exchanger 1
Heat Exchanger 2
Pump 1
Pump 2
Air Blower
Pump 4
Reactor
Adsorbent bed
settling tank

0.831844
5.677719
0.092169
0.080078
1.998123
0.065997
0.032859
0.037401
1.652629

unit
ft^2
ft^2
hp
hp
hp
hp
m3
m3
m3

Table A.2. Summary of Equipment Sizing


Capacity
Hours operating/year
Biodiesel mass flow
Glycerin mass flow
Biodiesel volume flow
Glycerin volume flow
Biodiesel yearly flow
Glycerin yearly flow

5000
34.2084
3.5417
10.06129
0.645118
50306.47
3225.592

hr/yr
kg/hr
kg/hr
gal/hr
gal/hr
gal/yr
gal/yr

Table A.2. Summary of Capacity


Utilities
Cost of utilities
Utilities steady state
utilities unsteady state

0.09
0.046753
0.439968

$/kwH
$/gal
$/gal

Table A.3. Summary of utilities cost

Derivation of the MGB process


Reactor
Firstly, the reactor was considered. Unlike the conventional process, where components
in the reactor are in one physical phase, MGB has two phases; the liquid reactants and the
solid catalyst. Because of these two phases, it is necessary that a reactor be chosen that
allows the catalyst to stay within the reactor while the reactants move pass it. With these
requirements, a fluidized bed or a plug flow reactor are logical. The plug flow reactor
was chosen based on the simplicity of the design and availability of experimental data.

203

Experiments conducted by Mr. Kannan and Greg Austic of Piedmont Biodiesel indicate
that in a plug flow reactor with MnO catalyst at 0.6 void fraction, a conversion of over
95% can be achieved in approximately 15 min. A fluidized bed reactor may reach even
greater conversion because it often achieves greater mass transfer. However, at this time
in the design process, there is no experimental data to support this.

With a 15 min residence time, the total reactor volume was calculated 0.0328 m3. With a
length to diameter ratio (L/D) of 20, the length is 2.56 m and the diameter is 0.127 m
(5.06 in). These volumes and lengths are of acceptable size to fit on a truck.

Volumetric flow rate (Q) = mass flow rate (M)/ density ()


t = 0.25 hr
L/D = 20
Void fraction = 0.6
Qin=Min/ in
=(70.6 kg/hr)/(898.8 kg/m3)
=0.0786 m3/hr
Qout=Mout/ out
=(70.6 kg/hr)/(892.3 kg/m3)
=0.0791 m3/hr
QAvg=Min + Mout/2
= 0.0788 m3/hr
Volume (V) = QAvg * t
= 0.0788 m3/hr * 0.25 hr
= 0.0197 m3
Vw/catalyst = V/Void fraction
= (0.0593 m3/hr)/0.6
= 0.0328 m3
Diameter (D) = [(V*4)/(20*)]^(1/3)
= [(0.0328 m3 * 4)/ (200*)]^(1/3)
= 0.127 m
Length (L) = 2.56 m

204

Flash
Next the methanol recovery system was considered. Because an excess of alcohol (25:1
molar ratio) is used to drive the reaction to completion, it is desired to recover that
excess. Alcohol and biodiesel are miscible but have different normal boiling points (64
C and 338 C respectively). Therefore for this process a distillation column or a flash
drum are viable options. Adsorption is a possibility, however it would be too difficult
with the level of alcohol (25:1 molar ratio), so it can be eliminated from consideration.
Distillation and flash utilize the same principle of flashing off one component, with a
flash drum being one stage and a distillation column being many. It was found through
modeling in Intelligen SuperPro that a flash drum was sufficient to perform the
separation. Additionally, at the temperatures and pressures present, it should be viable to
flash off >95% of the alcohol.

Temperature loss in the flash calculations


P = 65 atm
Joule Thompson coefficient () = ((1/)/Cp) * (T-1)
= ((1/)/Cp) * (T-1)
=((1/(1000 * 892.8 kg/m3))/(1.8125 J/g-k)) * ((0.0116 1/ C * 240 c) -1)
=1.103 * 10-6 k/Pa
T = * P
= (1.103 * 10-6 k/Pa) * (65 atm) * (101325 Pa/atm)
= 7.26 C
T = 240 7.26
= 232 C

205

Adsorption bed
In the production of biodiesel there are several sources of water or water generation. For
instance, waste oil and lower quality alcohols inherently have water in them (1-5%).
Also, in the reaction of free fatty acid chains to methyl esters, 1 mole of water is
produced for every mole of methyl ester produced. However water is generally
undesirable in the process for the reason that even a small fraction of water will
encourage saponification. However, the MGB, can handle greater quantities of water and
a maximum of 5% water is permissible without negative effects occurring.

In either process is it best to remove water and prevent build up. To remove water from a
fluid process, there are several options, from distillation/flash/steam to
stripping/desiccants. It was decided to use a desiccant, Zeolite 3A. Zeolite is an alumina
silica compound which forms regular pores in its molecular structure. This option
requires no utilities, few moving parts, and is economical, with Zeolite prices around $4/
kg.

With a run of 3 hours, the adsorption bed has a volume of 0.037 m3. With an L/D ratio of
3 the length is 0.75 m and the diameter is 0.25 m. This is an acceptable size to fit on a
truck.

Absorption bed sizing calculations


Mass reactants = Mass flow rate (M) * time(t)
= 1.7 kg/hr * 3 hr
= 5.26 kg
Mass desiccant (Md) = Total Mass/ Mass/mass desiccant)

206

= 5.26 kg/(0.22 kg/kg desiccant)


= 23.9 kg desiccant
Volume (V) = Md/density
= 23.9 kg desiccant/0.64 kg/L
= 0.037 m3
Diameter (D) = [(V*4)/(3*)]^(1/3)
= [(0.037 m3 * 4)/ (3*)]^(1/3)
= 0.25 m
Length (L) = L/D * D
= 3 * 0.25 m
= 0.75 m

Settling tank
Downstream of the reactor, it is necessary to separation biodiesel and glycerin. To
achieve this, a centrifuge or a setting tank was considered. An analysis of passive settling
time was necessary to determine which unit operation would be preferable. The settling
velocity of biodiesel was found to be 0.611 m/s, which was more than adequate enough
for separation. Therefore, the settling tank was chosen. Not only does this reduce the
complexity and utilities, it can also be used as a temporary storage tank.
Setting time calculations
Dynamic viscosity(vd) = 0.00188 m2/s
R = 0.0035 in
Settling velocity (v) = (2 * g/9) * ((glcerol - biodiesel)/vd) * (R2)
= (2 * g/9) * ((1450 kg/m3 880 kg/m3)/0.00188 m2/s) * (0.000962)
= 0.611m/s
Settling time(lights) = chord height * v
Settling time (heavy) = (D chord height) * v
Volume= (Vbiodiesl * t) + (Vglycerol * t)/void fraction
= (1.39 + 0.088)/0.1 = 1.65 m3

207

Diameter (D) = [(V*4)/(3*)]^(1/3)


= [(1.65 m3 * 4)/ (3*)]^(1/3)
= 0.66 m
Length = 3 * D
= 1.99 m
Vglycerol/ V = 0.0532
From charts; Chord length/D = 0.604
Chord height/D = 0.1015
Chord height = Chord height/D * D
= 0.1015 * 0.66 m
= 0.067 m
Settling time(heavy) = chord height * v
= 0.067 m * 0.611m/s
= 0.11 s
Settling time (lights) = (D chord height) * v
= (0.66 m 0.067 m) * 0.611m/s
= 0.975 s

Heat Recovery
The current process uses supercritical temperatures and pressures, where the heat source
is the reactor, which brings the reactants from 25 c to 240c in unsteady state. It is
desired to recover as much of the heat generated as possible, so that no additional heater
will be required.

In the process there are two feed streams and two product streams. The product streams
and intermediate streams can be used to heat the two feed streams. In this fashion a
system where the most efficient heat usage can be implemented. There were several
configurations which are possible (see below diagrams 2-5). The energy balance was
calculated for each process, then the efficiency of the heat exchangers maximized by

208

conducting sensitivity analysis. Afterwards, for each of the configurations the utilities
cost per gal of product was calculated. Based on this analysis, configuration 3, which
minimized costs, was chosen.

Configuration 3 has 2 heat exchange units, one which is the condenser of the flash tank,
and the other which is a shell and tube heat exchanger between streams 5 and 6. Both
heat exchange units were modeled as adiabatic heat exchangers with an overall heat
exchanger coefficient of 300 W/m2K. The condenser requires 0.8 ft2 and the heat
exchanger requires 5.2 ft2 of tubing.

Diagram 2 Proposed heat exchange network 1

Diagram 3 Proposed heat exchange network 2

209

8
10

3.1

11

12

5
4.1
7

13

14

Diagram 4 Proposed heat exchange network 3


8
10

D
4

3.1

11

12

4.1
13

6
7
G

14

Diagram 5 Proposed heat exchange network 4

Utilities Cost ($)


Money lost ($)
Temp in stream 6 (c)
Temp in stream 3 (c)
HEX Area 1
HEX Area 2
HEX Area 3

Configuration 1
all 3
0.087
0.013
63
157.5
0.679
0.132
0.507

Configuration 2
before
0.14
0.012
58
123
3.14
0.276
0.396

Configuration 3
1 and 3
0.032
0.007
31
n/a
n/a
0.137
2.8

Table A.6. Comparison of 4 heat exchange networks


Heat Balance Calculation
Q1 = M * Cp1 * T1
Q2 = M * Cp2 * T2
Q1=Q2
M * Cp2 * T2 = M * Cp1 * T1
T2 = (M * Cp1 * T1)/ (M * Cp2)
T2F = [11.7 g/s * 2.32 J/g-k * (200-25 c)]/ [12.28 g/s * 2.14 J/g-k] + T2I
= 52.1 c

210

Configuration 4
1 and 2
0.044
0.027
n/a
n/a
1.28
0.151
n/a

Pump system
Lastly, the flow through the process must be steady and continuous. To achieve this, a
pumping system is used to push fluids to overcome resistances like piping, unit
operations and friction by generating pressure differences. The type of pump to use is
determined by several factors, for instance, temperatures handled type of fluid, desired
fluid velocity and final pressure. Based on these factors, positive displacement pumps
were chosen.

Resistance and Power Calculation


Q = M/
= (212 kg/hr)/ (898 kg/m3)
= 0.236
Head = f * (L/D) * ( * v2/ 2g)
= 0.023 * (10/0.0508) * (898.8 * 2.13 m/s/2g)
= 0.0068 atm

f= from moody chart


v= calculated with viscosity blending number
VBNA = 14.534 * ln(ln(v + 0.8)) + 10.975
= 14.534 * ln(ln(0.735 cSt + 0.8)) + 10.975
= -1.24
VBNblend = [XA * VBNA] + [XB * VBNB] .
= -1.24 + 0.136
= -1.11

v= exp^[e(VBNblend 10.975)/(14.534)] - 0.8


= exp^[e(-1.11 10.975)/(14.534)] - 0.8
= 0.757 cSt
Utilities calculations
KW = (Head or resistance*Q)/359900
= (5216332.9 pa*0.0474 m3/hr)/ 359900

211

= 0.0687 KW
Fudge Factor = 0.54
0.0687 KW/0.54 = 0.12 KW
Cost = 0.09$/KWh * KW
= 0.09$/KWh * 0.12 KW
= $0.011

Comparison to Conventional Process


The MGB 50,000 gal/yr plant is estimated to have a capital cost of $73,439.60, including
the price of the pickup truck for transportation. With 15 years of depreciation and a final
scrap value of $10,000, yearly depreciation is $3,562.64. Total Operating costs come to
$83,502.89, of which 75% is the feedstock. Total processing costs per gallon of biodiesel
is $0.41/gal, which is better than or comparable to existing processes on waste oils or
animal fats ($0.81, Singh, $0.42 and $0.34, Bender). See Table A.7.

There are several reasons why the MGB process is able to achieve this competitive
processing cost. Firstly, because of the solid catalyst, there are no extra chemicals in use.
Conventional plants like Singh (A) have extra chemicals costing an additional $0.17/gal
to manufacturing cost. Secondly, there are fewer unit operations and less moving parts
than conventional plants. Specifically, there is no pre-esterification, no downstream
purification or water washing, no safety hazards to remove toxic chemicals. This not only
means that the capital costs are reduced, but also that maintenance and repair costs are
lessened. Thirdly, while large plants can often save on labor costs, smaller plants are
often not able to. However while the MGB is a small plant, labor costs are comparable to
larger plants (0.23 $/gal) because the simplicity of the process requires less labor.

212

Fourthly, the process can accommodate many types of feed stocks without adding or
changing existing unit operations. A conventional process which has been designed to
process virgin feed stocks must have additional units added to the process to process
waste feed stocks.

With the current price of diesel oil being about ~$4.40, the profitability of biodiesel
produced with the MGB system is certainly feasible.

Operation Costs - Feedstock


Operations costs - Labor

$8,706.55 , 9%

Operation costs - Utilities

$3,562.64 , 4%
$2,116.70 , 2%

depreciation

$10,000.00 , 10%

Taxes, insurance, interest, maintainance,


repairs, contingency, administration, safety

$71,386.19 , 75%

Diagram 6 Cost of plant per year

213

Comparison
MGB

Haas*

Singh(A)**

Singh(D)***

57.08%
50,000

90.14%
10,000,000

??
20,000,000

??
20,000,000

$/gal
$
1.39

$/gal
$
1.89
$
3.65

$/gal
$
1.86
$
3.65

$/gal
$
1.39

$
$
$
$
$
$

$/gal
0.19
0.04
0.24
(0.07)
0.41

$
$
$
$
$
$

$/gal
0.05
0.04
0.14
(0.13)
0.11

$
$
$
$
$
$

$/gal
0.08
0.17
0.06
0.23
(0.07)
0.47

$
$
$
$
$
$

$/gal
0.15
0.21
0.13
0.39
(0.07)
0.81

$/gal
1.80

$
$

$/gal
2.00
3.76

$
$

$/gal
2.33
4.12

$/gal
2.20

Capacity
Uptime/yr
Capacity (gal/yr)
Feedstock costs
Feedstock
(adjusted for todays prices)
Processing costs
Labor
Chemicals
Utilities
Fixed charges and plant overheads
Co-product Credit (Glycerol)
Subtotal - Processing costs
Total Manufacturing costs
(adjusted for todays prices)

*Haas molded a virgin oil processing plant with a pre-esterification and biodiesel and glycerin purification components
in the process
**Singh (A) is a virgin oil processing plant
*** Singh (D) is a waste oil processing plant that differs from case A because of the additional of degumming, acid
esterification and biodiesel distillation units.

Table A.7. Cost Comparison of MGB process to other processes.

214

The other relevant tables are listed below.


number
feedstock
Methanol
Waste Oil
labor
Operator

utilities
Electric - steady state
Electric - unsteady state

cost

Total

8466.8
50115.9

gal/yr
gal/yr

$
$

people

$ 10,000.00

46918.7
3081.1

gal/yr
gal/yr

$
$

1.58
1.12

$
$

13,377.52
56,129.79

Subtotal

$
$

10,000.00
10,000.00

Subtotal

$
$
$

1,548.32
523.78
2,072.10

0.03
0.17

total operating costs


$
81,579.41

Table A.8. Annual operating cost of the 50,000 gallon mobile production unit.

215

Number

Cost

Installation Insulation

Multiplication factors

Pumps
Pump 1
Pump 2
Pump 3
Pump 4

1
1
1
1

$ 750.00
$ 750.00
$ 1,000.00
$ 750.00

$
750.00
$
750.00
$ 1,000.00
$
750.00

$
$
$
$

150.00
150.00
200.00
150.00

$
$
$
$

150.00
150.00
200.00
150.00

$
$
$
$

150.00
150.00
200.00
150.00

$
$
$
$
Subtotal $

Heat exchangers
Hex 1
Hex 2

1
1

$ 1,048.00
$ 1,048.00

$ 1,048.00
$ 1,048.00

$
$

209.60
209.60

$ 209.60
$ 209.60

$
$

209.60
209.60

$ 2,724.80
$ 2,724.80
Subtotal $ 5,449.60

Reactor
Plug Flow reactor
catalyst

1
6

$ 4,500.00
$
50.00

$ 4,500.00
$
50.00

$
$

900.00 $ 900.00
$
-

$
$

900.00
-

$ 11,700.00
$ 600.00
Subtotal $ 12,300.00

Flash
Flash drum

$ 2,500.00

$ 2,500.00

500.00

$ 500.00

500.00

$ 6,500.00
Subtotal $ 6,500.00

Adsorbent
molecular sieve
zeolite

2
120

$
$

750.00 $
4.50 $

750.00
-

$
$

150.00 $ 150.00
$
-

$
$

150.00
-

$ 3,900.00
$ 540.00
Subtotal $ 4,440.00

Storage
settling tank
misc tanks

1
0

$
$

500.00 $
$

500.00
-

$
$

100.00 $ 100.00
$
-

$
$

100.00
-

$ 1,300.00
$
Subtotal $ 1,300.00

Site
truck
trailer
motor

1
0
0

$ 20,000.00 $ 5,000.00
$
$
$
$
-

$
$
$

$
$
$

Total
unit
costs
$ 33,650.50

Total
insulation
costs
$ 2,719.20

Total
installation
costs
$ 18,646.00

0.2

$
$
$

Piping

Instrumentation

0.2

0.2

Total
piping
costs
$ 2,719.20

Total

$ 25,000.00
$
$
Subtotal $ 25,000.00
Total
Total
instrumentation
capital
costs
costs
$
2,719.20
$ 63,439.60

Table A.9. Total capital cost of the 50,000 gallons per year mobile unit.

216

1,950.00
1,950.00
2,600.00
1,950.00
8,450.00

Sources
3A Molecular Sieve. 2006. Molecular Sieve Study. July 2008.
<http://www.molecularsieve.org/>
3A Zeolite Molecular Sieve. 22 August 2008. Biofuel system. August 2008.
<http://www.biofuelsystems.com/shop/product_info.php?products_id=297>
Bender, Martin. Economic feasibility review for community-scale farmer cooperatives
for biodiesel. Bioresource Technology 70 (1999): 81-87.
Biodiesel Case Studies & Examples. Oilgae.com. August 2008.
<http://www.oilgae.com/energy/sou/ae/re/be/bd/cs/cs.html>
Biodiesel Fact Sheets Specification for Biodiesel (B100). Biodiesel.org. Aug 2008.
<http://www.biodiesel.org/pdf_files/fuelfactsheets/BDSpec.PDF>
The Chemical Engineers Resource Page.19 August 2008. Cheresources.com. July 2008.
<http://www.cheresources.com/>.
Chicago Board of Trade Soybean Oil. 20 August 2008. Chicago Board of
Trade.<http://www.cbot.com/>.
Haas, Michael J, et al. "A process model to estimate biodiesel production costs."
Bioresource Technology 96 (2006): 671-678.
ITT Standard. SSCF - stainless steel shell & tube heat exchanger 2008. ITT Standard.
July 2008. < http://www.ittstandard.com/>. Path: Shell and Tube; SSCF.
Methanol Price. 31 July 2008. Methanex Aug 2008.
<http://www.methanex.com/products/methanolprice.html>
National Tank Outlet Online Store. 2008. National Tank Outlet. July 2008.
<http://www.ntotank.com/index.html>
Nelik, Lev. Pumps. Kirk-Othmer Encyclopedia of Chemical Technology. 5th ed. Ed.
Kirk-Othmer. Vol. 21. New York: Wiley, John & Son, Incorporated, 2001.
Perry, Robert H. and Don W. Green. Perrys Chemical Engineers Handbook. 7th ed. New
York: McGraw Hill, 1997. 10-23, 16-9.

217

Vita
DHEEBAN CHAKRAVARTHI KANNAN
EDUCATION
Ph. D. in Chemical Engineering, The Pennsylvania State University, University Park, PA August 2009.
Dissertation Title: A Solid Catalyst Method for Biodiesel Production
M.S. in Chemical Engineering, The Pennsylvania State University, University Park, PA December 2004.
Thesis Title: The UNIFAC Model and Free-Volume Effect
B. Tech. in Chemical Engineering, Anna University, Chennai, India July 2002.
EXPERIENCE
Senior Research Engineer
OriginOil, Inc, Los Angeles, CA.

April 2009 Present

Graduate Research Assistant


The Pennsylvania State University, University Park, PA.

June 2004 April 2009

Graduate Teaching Assistant


The Pennsylvania State University, University Park, PA.

August 2004 May 2005

Graduate Research Assistant


The Pennsylvania State University, University Park, PA.

August 2002 July 2004

PUBLICATIONS
1. Kannan, D.C., Duda, J.L. and Danner, R.P., A free-volume term based on the van der Waals partition
function for the UNIFAC model, Fluid Phase Equilibria, 228-229 (2005) 321-328.
2. Kannan, D.C., Duda, J.L. and Danner, R.P., Application of UNIFAC-vdW-FV model to water-PEO
systems, Fluid Phase Equilibria, 237 (2005) 86-88.
3. Kannan, D.C. and Matson, J.V., A novel solid catalyst method for biodiesel production Method
development and Reaction kinetics, (in preparation, 2009).
4. Kannan, D.C. and Matson, J.V., A novel solid catalyst method for biodiesel production Studies for
commercialization and Method comparison, (in preparation, 2009).
PATENTS
Matson, J.V. and Kannan, D.C., Green Biodiesel, U.S. Patent No. 7,563,915, July 21, 2009.

Vous aimerez peut-être aussi