USF Libraries
USF Digital Collections

Characterization of the esterification reaction in high free-fatty acid oils

MISSING IMAGE

Material Information

Title:
Characterization of the esterification reaction in high free-fatty acid oils
Physical Description:
Book
Language:
English
Creator:
Altic, Lucas
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
ASTM D6751
Brown grease
Ultrasonic
Cavitation
Biodiesel
FFA
Dissertations, Academic -- Mechanical Engineering -- Masters -- USF   ( lcsh )
Genre:
non-fiction   ( marcgt )

Notes

Abstract:
ABSTRACT: Energy and vegetable oil prices have caused many biodiesel producers to turn to waste cooking oils as feedstocks. These oils contain high levels of free fatty acids (FFAs) which make them difficult or impossible to convert to biodiesel by conventional production methods. Esterification is required for ultra-high FFA feedstocks such as Brown Grease. In addition, ultrasonic irradiation has the potential to improve the kinetics of the esterification reaction. 2-level, multi-factor DOE experiments were conducted to characterize the esterification reaction in ultra-high FFA oils as well as determine whether ultrasonic irradiation gives any benefit besides energy input. The study determined that sulfuric acid content had the greatest effect followed by temperature and water content (inhibited reaction). Methanol content had no effect in the range studied. A small interaction term existed between sulfuric acid and temperature. The study also concluded that sonication did not give any additional benefit over energy input.
Thesis:
Thesis (MSME)--University of South Florida, 2010.
Bibliography:
Includes bibliographical references.
System Details:
Mode of access: World Wide Web.
System Details:
System requirements: World Wide Web browser and PDF reader.
Statement of Responsibility:
by Lucas Altic.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains X pages.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
usfldc doi - E14-SFE0004817
usfldc handle - e14.4817
System ID:
SFS0028093:00001


This item is only available as the following downloads:


Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim
leader nam 22 Ka 4500
controlfield tag 007 cr-bnu---uuuuu
008 s2010 flu s 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0004817
035
(OCoLC)
040
FHM
c FHM
049
FHMM
090
XX9999 (Online)
1 100
Altic, Lucas.
0 245
Characterization of the esterification reaction in high free-fatty acid oils
h [electronic resource] /
by Lucas Altic.
260
[Tampa, Fla] :
b University of South Florida,
2010.
500
Title from PDF of title page.
Document formatted into pages; contains X pages.
502
Thesis (MSME)--University of South Florida, 2010.
504
Includes bibliographical references.
516
Text (Electronic thesis) in PDF format.
538
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
3 520
ABSTRACT: Energy and vegetable oil prices have caused many biodiesel producers to turn to waste cooking oils as feedstocks. These oils contain high levels of free fatty acids (FFAs) which make them difficult or impossible to convert to biodiesel by conventional production methods. Esterification is required for ultra-high FFA feedstocks such as Brown Grease. In addition, ultrasonic irradiation has the potential to improve the kinetics of the esterification reaction. 2-level, multi-factor DOE experiments were conducted to characterize the esterification reaction in ultra-high FFA oils as well as determine whether ultrasonic irradiation gives any benefit besides energy input. The study determined that sulfuric acid content had the greatest effect followed by temperature and water content (inhibited reaction). Methanol content had no effect in the range studied. A small interaction term existed between sulfuric acid and temperature. The study also concluded that sonication did not give any additional benefit over energy input.
590
Advisor: Daniel Hess, Ph.D.
653
ASTM D6751
Brown grease
Ultrasonic
Cavitation
Biodiesel
FFA
690
Dissertations, Academic
z USF
x Mechanical Engineering
Masters.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.4817



PAGE 1

Characterization of the Esterification Reaction in High Free Fatty Acid Oils by Lucas Eli Porter Altic A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Mechanical Engineering Department of Mechanical Engineering College of Engineering University of South Florida Major Professors: Daniel Hess, Ph.D. Muhammad Rahman, Ph.D. Stuart Wilkinson, Ph.D. Date of Approval: October 29, 2010 Keywords: ASTM D6751, brown grease, ultrasonic, cavitation, biodiesel, FFA Copyright 2010, Lucas Eli Porter Altic

PAGE 2

DEDICATION To my beautiful wife Lara who gave me the courage to carry on and to my new son Jackson who gave me the will to finish.

PAGE 3

ACKNOWLEDGEMENTS I would like to thank Dr. Daniel Hess for his input, support, and wisdom during the entirety of this process. I would also lik e to thank Dr. Muhammad Rahman for serving as a patient mentor throughout the years and for hi s latest input in this work. I would also like to thank Dr. Stuart Wilkinson for his inspir ation in the field of alternative energy and his honest yet constructive criticism. Al so Mitch Bishop, Mark Tarrien, Laura Blalock, and John Daly for being there from the start. Finally, I would like to thank Sue Britten, Shirley Tervort, and Catherine Burton for guiding me to the finish line.

PAGE 4

i TABLE OF CONTENTS LIST OF TABLES ............................................................................................................. iv LIST OF FIGURES ............................................................................................................ v ABSTRACT ...................................................................................................................... vii 1.0INTRODUCTION ..................................................................................................... 11.1OBJECTIVES ................................................................................................ 41.1.1REACTION CHARACTERIZATION ........................................... 41.1.2ENHANCED REACTION KINE TICS EVALUATION ............... 51.1.3IMPLEMENTATION ..................................................................... 7 2.0LITERATURE REVIEW .......................................................................................... 92.1SCOPE OF SURVEY .................................................................................... 92.2INTRODUCTORY ........................................................................................ 92.3THE BACKGROUND AND HISTORY OF BIODIESEL ......................... 102.4THE BIODIESEL PROCESS ...................................................................... 142.4.1TRANSESTERIFICATION ......................................................... 142.4.2ESTERIFICATION ...................................................................... 172.4.2.1Need for Esterification ................................................... 172.4.3INDUSTRIAL PROCESS ............................................................ 202.4.4ALCOHOLS ................................................................................. 212.4.4.1Methanol ........................................................................ 222.4.4.2Ethanol ........................................................................... 222.5CHARACTERIZING THE ESTERI FICATION REACTION ................... 232.6IMPROVING REACTION KINETICS ...................................................... 262.7SUMMARY OF THE LITERATURE REVIEW ........................................ 27 3.0EXPERIMENTAL ................................................................................................... 293.1SCOPE OF EXPERIMENTS ...................................................................... 293.2DESIGN OF EXPERIMENT ...................................................................... 313.2.1ESTERIFICATION REACTION CHARACTERIZATION ........ 313.2.1.1Justification for the Factorial Design ............................. 333.2.1.2Choice of Levels ............................................................ 343.2.1.3Randomization ............................................................... 353.2.1.4Replicates ....................................................................... 353.2.1.5Consideration of the Fracti onal Factorial Design .......... 403.2.1.6Test for Curvature .......................................................... 413.2.1.7Blocking ......................................................................... 41

PAGE 5

ii 3.2.2ENHANCED REACTION KINE TICS EVALUATION ............. 433.2.2.1Design Choices .............................................................. 463.3EQUIPMENT .............................................................................................. 493.3.1SONICATOR ................................................................................ 493.3.2GLASSWARE .............................................................................. 493.3.3PIPETTING SYSTEM.................................................................. 503.3.4STIR BARS .................................................................................. 513.3.5BALANCE .................................................................................... 513.3.6HEATED STIR PLATE ............................................................... 513.3.7AUTO TITRATOR ....................................................................... 523.3.8POWER METER .......................................................................... 523.3.9KARL FISHER MOISTURE ANALYZER ................................. 523.4MATERIALS ............................................................................................... 533.4.1METHANOL ................................................................................ 533.4.2BROWN GREASE ....................................................................... 533.4.3REAGENT ALCOHOL ................................................................ 533.4.4SULFURIC ACID ........................................................................ 543.4.5DEIONIZED WATER .................................................................. 543.5REACTOR DESIGN ................................................................................... 543.5.1STIRRED REACTOR DESIGN................................................... 543.5.2SONICATION REACTOR DESIGN ........................................... 553.6METHODOLOGY ...................................................................................... 573.6.1EXPERIMENTAL SEQUENCE .................................................. 573.6.2REACTION PARAMETERS ....................................................... 603.6.3MEASUREMENTS ...................................................................... 613.6.3.1Free Fatty Acid Measurements ...................................... 623.6.3.2Energy Input Measurements .......................................... 64 4.0RESULTS AND DISCUSSION .............................................................................. 674.1ESTERIFICATION CHARACTERIZATION ............................................ 674.1.1STATISTICAL RESULTS ........................................................... 684.1.1.1Interaction Effects .......................................................... 684.1.1.2Main Effects ................................................................... 704.1.1.3Covariate Effects ............................................................ 714.1.1.4Blocking Effects............................................................. 724.1.1.5Center Point Effects ....................................................... 724.1.1.6Relative Effects .............................................................. 724.1.1.7Regression Equation ...................................................... 764.1.1.8Model Adequacy Check ................................................. 764.1.1.9Power Analysis .............................................................. 784.1.2PRACTICAL INTERPRETATION ............................................. 814.1.2.1Limitations of the Design............................................... 824.1.2.2Sulfuric Acid Content .................................................... 844.1.2.3Temperature ................................................................... 854.1.2.4Water Content ................................................................ 854.1.2.5Methanol Content........................................................... 86

PAGE 6

iii 4.1.2.6Sulfuric Acid Temperature Interaction .......................... 874.2ENHANCED REACTION KINETICS ....................................................... 88 5.0CONCLUSIONS...................................................................................................... 93 6.0RECOMMENDATIONS ......................................................................................... 95 7.0REFERENCES ........................................................................................................ 96 ABOUT THE AUTHOR ................................................................................... END PAGE

PAGE 7

iv LIST OF TABLES Table 1: Esterification reaction characterization experiment ........................................... 32Table 2: Treatment levels for factorial design .................................................................. 33Table 3: Concentration of treatme nt levels by volume of oil ........................................... 34Table 4: Initial parameters to determine required replicates ............................................ 36Table 5: Initial experimental standard devi ation for esterification characterization ........ 36Table 6: Available factoria l designs (Minitab 16) ............................................................ 40Table 7: Reaction accelerati on evaluation experiment ..................................................... 45Table 8: Reaction acceleration ex periment treatment levels ............................................ 46Table 9: Initial parameters to determine required replicates ............................................ 47Table 10: Initial experiment al standard deviation for sonication experiment .................. 47Table 11: Specification for XL2020 sonicator .................................................................. 49Table 12: Glassware used in experimentation .................................................................. 50Table 13: Fixed reaction conditions for es terification reaction characterization .............. 61Table 14: Fixed reaction conditions for e nhanced reaction kinetics evaluation ............... 61Table 15: Minitab output of the estimated effects ............................................................ 67Table 16: Minitab output of refitted data .......................................................................... 76Table 17: Minitab output of power analysis ..................................................................... 80

PAGE 8

v LIST OF FIGURES Figure 1: Biodiesel publications by year (SciFinder Scholar, 2010) .................................13 Figure 2: Basic methanolysis reaction schematic ..............................................................15 Figure 3: Power and sample size tool in Minitab 16 .........................................................37 Figure 4: Power curve for the esterificati on reaction characteri zation experiment ...........39 Figure 5: Power curve for the sonication experiment ........................................................48 Figure 6: Adjustable pipetter .............................................................................................50 Figure 7: Magnetic stir bar .................................................................................................51 Figure 8: Stirred reactor setup ............................................................................................54 Figure 9: Sonication reactor setup .....................................................................................55 Figure 10: Experime ntal flowchart ....................................................................................60 Figure 11: Tiamo 1.1 workplace GUI ................................................................................62 Figure 12: Tiamo 1.1 titration plot .....................................................................................64 Figure 13: Interaction eff ects plots for the esterification characterization experiment........................................................................................................69 Figure 14: Main effects plot s for the esterification char acterization experiment ..............70 Figure 15: Normal probability plot of the effects of the esterification characterization experiment .............................................................................73 Figure 16: Half-normal probability plot of the effects of the esterification characterization experiment .............................................................................74 Figure 17: Pareto chart of the effects of the esterification characterization experiment........................................................................................................75 Figure 18: Residual plots for the esteri fication characterization experiment ....................77

PAGE 9

vi Figure 19: Power and sample size window ........................................................................79 Figure 20: Power curve for the f our significant factor effects ...........................................80 Figure 21: Esterification reaction rate for two factor combinations ..................................83 Figure 22: Main effects plot for the r eaction kinetics enhancement experiment ...............88 Figure 23: Interaction plot for the reac tion kinetics enhancement experiment .................89 Figure 24: Pareto chart for the reacti on kinetics enhancement experiment .......................90 Figure 25: Power curve for the reaction kinetics enhancement experiment ......................91 Figure 26: Residual plots for the reacti on kinetics enhancement experiment ...................92

PAGE 10

vii Characterization of the Es terification Reaction in High Free Fatty Acid Oils Lucas Eli Porter Altic ABSTRACT Energy and vegetable oil prices have caused many biodiesel producers to turn to waste cooking oils as feedstocks. These oils cont ain high levels of free fatty acids (FFAs) which make them difficult or impossible to convert to biodiesel by conventional production methods. Esterificati on is required for ultra-high FFA feedstocks such as Brown Grease. In addition, ultrasonic irradia tion has the potential to improve the kinetics of the esterification reaction. 2-level, multi-factor DOE experiments were conducted to characterize the esterification reaction in ultra-high FFA oils as well as determine whether ultrasonic irradiation gives any benefit besides energy input. The study determined that sulfuric acid content had th e greatest effect follo wed by temperature and water content (inhibited reaction). Methanol content had no effect in the range studied. A small interaction term existed between sulf uric acid and temperat ure. The study also concluded that sonication did not give any additional benefit over energy input.

PAGE 11

1 1.0 INTRODUCTION As an emerging industry, Biodiesel has blossomed from a little-known, speculative, and exotic alternative energy source into a qua si-mainstream fuel for generally accepted use in diesel vehicles and machinery. Mounti ng political instability in the middle-east, growing environmental concerns, and recent asso ciated hikes in fuel prices have spawned a great deal of public, scientif ic, and capital interest both fo r the fuel and the industry in recent years. Between the years 2001 and 2008, Biodiesel production in the United States increased from 9 million to 678 milli on gallons per year (U.S. Energy Information Administration, 2010). This indicates a majo r boom in a quickly maturing industry. Despite this recent economic, technical, a nd social progress for biodiesel, overwhelming market forces combined with unreliable gove rnment support for the industry has resulted in a dramatic decline in biodiesel production and the near extinction of viable biodiesel companies within the United States over the last two years. Upward pressure within the commodities and energy markets has forced unbridled increases in the price of conventional feedstocks for biodiesel such as soy and rape seed oils. Between the years 2005 and 2008, soybean prices rose from $5 pe r bushel to nearly $11 per bushel on the low side (Wordpress, 2008). En ergy prices also increased in relative proportion. This rapid inflation in biodiesel production costs cau sed the margins for biodiesel sales within the US to shrink. With many companies at breakeven or underwater, the future of biodiesel production hung on the tattered tapestry known as governmental energy policy.

PAGE 12

2 On December 31, 2009, the government subsidy of $1 per gallon of biodiesel production was allowed to lapse. Hope of the pr ogram’s renewal gradually dwindled though 2010 and many biodiesel plants simply closed thei r doors (Daily Times Herald, 2010). By this time, even with regulatory support, biodies el production from conventional feedstocks was, at best, a marginal business and, at worst, economic suicide. As a result, U.S. domestic biodiesel production for the first ha lf of 2010 dropped to levels not seen since 2007 (U.S. Energy Information Administration, 2010). The solution for many was a fundamental shift in the underpinnings of the industry itself: the use of cheaper recycled or second-use oils for biodies el production such as yellow grease and brown grease. With the cost of the lipid stock being nearly 80% of the cost of biodiesel production, any amount saved in this category represented huge potential for economic survival if not prosperity (W ordpress, 2010). As of October 2010, yellow grease prices were around 28 cents per pound wh ile soybean oil prices were approaching 48 cents per pound. Brown grease was lowest at 10 cents per pound (The Jacobsen, 2010). In addition to being cheap, second-use oils are also plentiful. A U.S. EPA study recently found that 1 to 3 billion gallons of waste greases are produced annually (Greer, 2010). This is nearly two to six times the national annual peak production of biodiesel for 2008, so it is evident that recycled greas es have the ability to completely replace conventional feedstocks in volume. As an added benefit, the use of waste greases to make biodiesel is also more environmentally friendly. The EPA concluded that the total lifecycle greenhouse gas emissions from the production and combus tion of biodiesel made from waste greases resulted in an 86 percent reduction over petroleum-derived

PAGE 13

3 diesel. This is in contrast to soy-base d biodiesel yielding only 54% fewer greenhouse gas emissions than petrodiesel (Gre er, 2010). These figures suggest that biodiesel made from waste greases offers a 59% improvement in greenhouse gas emissions over soy-based biodiesel. The problem with the strategy of using waste gr eases to produce biodiesel is that recycled oils such as yellow grease or brown grease require extr a processing steps to produce ASTM quality fuel. The oils themselves are heavily degraded and contain high levels of free fatty acids (FFA) which are the products of thermal degradation and hydrolization during their use in cooking and residence in gr ease traps. Dealing with these free fatty acids has become the primary technological dile mma for the few biodiesel plants still in operation. Esterification of the FFA component was widely adopted as a pre-treatment step to reduce the FFA concentration in th e oil prior to convers ion into biodiesel by conventional means (Tyson, 2002). While several entities have successfully and consistently produced biodiesel from UCO or Yellow Grease (the commodity term for commercially-collected and rendered used cooking oil) the dynamics of this process has remained a mystery for many. It is not widely known, for instance, whether catalys t concentration, moisture content, or methanol quantity has a greater influence on conversion time or reaction completeness. A formal evaluation of the various factors infl uencing the esterificati on reaction is needed as a foundation for optimization. Furthermor e, far fewer operations have had success converting even cheaper, but more heavily de graded, feedstocks such as brown grease

PAGE 14

4 into quality biodiesel. Brown grease (the commodity name for commercially collected and rendered trap grease) and similar oils such as fatty acid distillates contain FFA concentrations in excess of 50% which has resulted in relu ctance for many companies to consider them as viable feedstocks (Tyson, 2002) These latter feedstocks are referred to herein as ultra-high-FFA oils. Some have recently proposed that reaction acceleration techniques such as ultrasonic irradiation can greatly increase the efficiency of conversion of the ultra-high-FFA oils (Hahn, 2009). It is the dual intent of this thesis to both characterize the esterificat ion of ultra-high-FFA oils by examining the influence of the primary process factor s on the reduction of FFA at high concentrations in the oil as well as to evaluate the use of reaction acceleration techniques for applicability to the conversion of low-cost feedstocks into biodiesel. 1.1 OBJECTIVES 1.1.1 REACTION CHARACTERIZATION As mentioned, the primary objec tive of this project was to characterize the esterification reaction of free fatty acids in ultra-high-FFA oils such as brown grease. Several major factors are thought to in fluence this reaction: 1. Relative quantity of methanol 2. Relative quantity of sulfuric acid catalyst 3. Relative quantity of water

PAGE 15

5 4. Temperature of the reaction system 5. Agitation speed. Four of these were selected for evaluation: re lative methanol content, relative sulfuric acid content, relative water content, and temperature. Agitation speed was excluded because its effect was thought to be minor with in the range of standard low-speed mixers and difficult to alter in many industrial setups The primary intent of this study was to determine the relative magnitudes of the effect s for each of the four chosen factors. Future researchers and process engineers ca n use this groundwork to determine on which of the factors to focus optimization efforts and which associated cost categories to dedicate resources. 1.1.2 ENHANCED REACTION KINETICS EVALUATION Biodiesel production from ultrahigh-FFA oils is neither a well understood process nor an efficient one. Conventional production techniqu es require catalyst and alcohol quantities far greater than the theoretical stoichiometric ratio. This results in increased production cost, unnecessary waste streams, impure byprodu cts, and inconsistent product quality. These glaring process issues have incited many researchers to i nvestigate alternative production methods. Most arisi ng technologies are still in developmental stages, with only a fraction having sufficient merit to pa ss into the implementa tion phase (Bournay, 2005). This fraction represents the beginni ng of second-generation biodiesel production. It is one of the objectives of this project to investigate one of these technologies in

PAGE 16

6 particular as it applie s to throughput, quality, and catalysis and several in general as part of a literature review. Initial investigation estab lished sonochemistry as a r eaction technology with a high degree of industrial potential. Its application possessed many characteristics that tailor it to an ideal reaction technology for biodiesel: it decreases the surface to volume ratio of the associated reactants, it shortens reaction times sometimes by orders of magnitude, it increases catalytic efficiency, and it is r eadily suitable for continuous production (Mason 1999). The literature review yielded se veral specific areas of biodies el research which might be improved by the application of sonochemistry. The first is catalysis. As mentioned previously, sonochemistry can increase catal ytic action during a reaction (Mason, 1999). A second area for potential improvement is r eaction length: sonication has been shown to reduce reaction time in transesterification r eactions as well as es terification reactions (Stavarache, 2003). Further inves tigation of the litera ture, however, left the author with a lack of confidence that sonochemistry had a ny additional advantages over other reaction enhancement technologies. All technologies reviewed seemed to have one primary advantage: they had the ability to add far more mechanical energy to the reaction system than conventional stirring t echniques. A simple experime nt was designed to test, on a fundamental level, whether sonication afford ed any benefit over c onventional stirring when the input of thermal and mechanical en ergy using either technique was made the same. With energy input held constant, sonication as an agitation method should show a

PAGE 17

7 demonstrable improvement over conventional s tirring if the former offered any additional benefit beyond the amount of en ergy it is capable of delivering to the reaction system. If no improvement is observed, it can be implicit ly concluded that energy input is the sole benefit of sonochemistry over conventional st irring. Catalyst content was varied along with the agitation method to evaluate whether sonication had an improvement on catalysis at high and lo w levels of catalyst. 1.1.3 IMPLEMENTATION Implementation of this project was conducted in four parts: First a literature review was done to evaluate the state of the art and current level of unde rstanding of the esterification reaction in ultra-high-FFA oils as it applies to biodiesel production as well as to identify viable reaction acceleration technol ogies with suitable merit for experimental evaluation. Next appropriate experimental designs we re chosen using DOE methods for both the esterification reaction characterization and th e enhanced reaction kinetics evaluation. Both experiments were considered to be sc reening experiments because the objective was to determine the relative magn itude of factor effects. Fo r this reason 2-level factorial designs were chosen and are outlined in sections concerning their design and implementation. The experiments were then performed and m easurements were taken according to the experimental design. Stirred reactions were performed with a Fisher Scientific Isotemp magnetic stirrer with maximum rotational ou tput of 1400 RPM. The Isotemp provided

PAGE 18

8 temperature control as well. Sonication r eactions were performe d with a probe-type sonicator (Mason, 1990). The sonicator m odel used was a Misonix XL-2020 with a frequency of 20 kHz and maximum power output of 600 W. Final FFA was the response variable in all experiments and FFA measurements were made by wet titration with an automatic titrator. Finally, statistical methods were applied to the data to identi fy the significant effects and to draw conclusions about their statistical significance and practical applicability to modern biodiesel production. ANOVA and regressi on analysis were applied to the data using the Minitab 16 statistical software suite.

PAGE 19

9 2.0 LITERATURE REVIEW 2.1 SCOPE OF SURVEY The scope of the literature review of the various studies was limited to their application to esterification reactions in ultra-high-FFA or high-FFA oils. Where necessary, nonesterification related works, su ch as those applying to transe sterification, are discussed in order to furnish rudimentary background of the process itself and to familiarize the researcher and the reader with its ba sic application and manifestation. 2.2 INTRODUCTORY In order to fully understand the importanc e of adopting new technologies for the enhanced production of biodiesel fuel, it is first necessary for one to know the background of the fuel itself, the hurdles it has overcome during its journey into mainstream acceptance, and the technologies ava ilable currently or in the near future for use in its production. This chapter serves to bring the reader up to speed on the state of biodiesel production within the United States and abroad. It aims to address the technical problems with the fuel and to hint at possi ble methods for correcting them. After an introduction to biodiesel production and its ma ny technical features, the reader will then be exposed to an array of cutting edge pr ocessing technologies which may or may not have been developed to the point of industr ial implementation. Fina lly, a description of the field of sonochemistry will be presen ted as a means for providing background on the

PAGE 20

10 prevailing technology of this study. With this review, an unde rstanding of the importance of technical improvement of the biodiesel process will be gained and the reader will be better prepared to grasp th e significance of the use of sonication in industrial biodies el production. 2.3 THE BACKGROUND AND HISTORY OF BIODIESEL It was first suggested to use vegetable oil-derived fuels as a means for motive power by the inventor of the diesel engine himsel f, Rudolf Diesel. On April 13, 1912, Diesel proclaimed that, through vegetable oils “…Moti ve power can still be produced from the heat of the sun, always available, even when the natural stores of solid and liquid fuels are completely exhausted” (Pahl, 2005). Th ese prophetic words indicated that Diesel was a visionary, carefully consider ing the instability of the petroleum supply chain well before the common consumer had even imagined that they were fi nite. Indeed, early models of the diesel engine were designed to run on vegetable oils and other alternative fuels, not the low grade petroleum distillate kno wn today as diesel fuel In the course of time, economics favored the use of petrodiese l (then a waste by-product) over higher cost virgin oils such as peanut and hemp (Pahl, 2005). The diesel engine was optimized for the use of petrodiesel and the vision of the engine’s inventor to use clean, renewable fuels was all but forgotten. It was not until nearly 60 years after Diesel’s death that the diesel engine’s ability to utilize vegetable oil-based fuels was re discovered during the oil embargo of 1973. Previous to that however, early work was c onducted at the University of Brussels by G.

PAGE 21

11 Chavanne on the use of ethyl esters of palm oil in a diesel engine. This preliminary research resulted in Belgian patent #422877 (Pet erson, 2006). That patent deals little with the production of methyl ester and focu ses primarily on its use as a transportation fuel (Iowa State University, 2006). Wartime experiments also commenced in these early years of Biofuels research, however intere st was suppressed with the reemergence of cheap petroleum-based fuels in peacetime (Mittelbach, 2004). With the Oil Embargo in full effect during th e Fall of 1973, oil prices and supplies were severely limited (Pahl, 2005). By 1974, the pr ice for a barrel of oil had risen from $3 to over $12 (Pahl, 2005). Renewable energy on ce again moved into the public eye and agriculturally derived fuel s were among the many sour ces investigated. Early experiments suggested that th e diesel engine had been highly optimized for the use of petrodiesel over the years and that the use of straight vegetable oil held the potential for severe engine damage (Pahl, 2005). The two viable options were to either modify the engine, or to modify the fuel in order to at tain compatibility (Mitt elbach, 2004). The first would require a mechanical alteration, the second a chemical one. Modifying the diesel engine for straight vegetable oil use, though po ssible, is somewhat impractical due to the existence of incumbent technology. The enti re world had adopted petroleum-fueled diesel engines and an alterati on to the technology would result in vast infrastructure change accompanied by an extreme resistance to adoption of the fuel for use. A more sensible approach was to modify the fuel to suit the engine. This was the source of a great deal of research and trials in years to come.

PAGE 22

12 Early patents relating to the us e of alkyl esters after the oil embargo began to appear in 1980 (Mittelbach, 2004). In 1981 and 1982, resear chers in South African, Germany, and New Zealand were studying the use of vegetabl e oil esters in diesel engines (Mittelbach, 2004). By 1982, a young chemist and researcher by the name of Martin Mittelbach had begun development of a simplified process fo r producing fatty acid methyl esters under mild conditions and received a patent on th e process soon thereafter. Mittelbach in collaboration with Wrgetter began feasibility testing of rapeseed me thyl ester as diesel fuel at Graz University in 1983 (Mittelbach, 2004). Mittelbach himself even used the early fuel in his own diesel powered vehicle. Most feas ibility testing between 1982 and 1987 were conducted on diesel tractors. It wa s felt that, being an agricultural fuel, the fuel should primarily benefit agricultural pr oducers, namely farmers. Ironically, this vision has yet to come to pass as the major users of biodiesel currently are city and governmental fleets (Pahl, 2005). Mittelbach continued his research thr oughout the 1980s and produced a vast body of work which resulted in the foundation for th e biodiesel industry. The initial process which he developed remains the primary meth od by which biodiesel is produced on an industrial scale. While Mittelbach and Wrgetter were hard at work defining the biodiesel process in Europe, Jon Van Garpen of Idaho University was bringing the Biodi esel movement to America. Charles Peterson of the Colorado Scho ol of Mines was also at the forefront of early biodiesel in the Un ited States (Pahl, 2005).

PAGE 23

13 By 1988, the term “Biodiesel” had made its way into the globa l alternative fuels vocabulary. The term was coined by Wang in a Chinese article on the subject (1988). The number of articles using the term “Biodiesel” to refer to the alkyl esters of vegetable oil increased almost exponentially between 1988 and 2009. Figure 1 shows this trend visually. Publication frequency seems to have begun to accelerate in 1996 and from 2000 on, the magnitude of academic interest in the topic increased rapidly. Figure 1: Biodiesel publications by year (SciFinder Scholar, 2010) A new alternative energy source was born. As of 2009, 8,752 articles on Biodiesel had been referenced in SciFinder alone. Th is does not account for articles published in databases not searched by SciFinder. It al so does not account for ar ticles not using the 0 500 1,000 1,500 2,000 2,500 3,0001988 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 Number of PublicationsYearBiodiesel Publications by Year

PAGE 24

14 phrase “biodiesel” to refer to the fuel. It is ev ident that much scientific interest exists for Biodiesel and associated tec hnologies and this can only l ead to further breakthroughs. 2.4 THE BIODIESEL PROCESS 2.4.1 TRANSESTERIFICATION The chemical reaction by which a lower alcohol reacts with a tri-glyceride (oil) to yield a fatty acid alkyl ester is known as transesterification. The pr ocess was first described by Duffy in 1852 and was referenced by Mittelbach (2004). Transesterification is also more broadly referred to as alcoholysis When the process refers to a specific alcohol, the -ysis suffix is appended to the name of the reacti ng alcohol – for instance, transesterification with methanol is referred to as methanolysis A slightly different reaction (described later), by which glycerol reacts with a fatty acid is known as glycerolysis. This is not explicitly considered as a tr ansesterification reaction beca use glycerol is not a lower alcohol and the reactant is not a triglyceride, however it yi elds diand monoglycerides and even some methyl esters (Tyson, 2002). Transesterification occurs easil y with the lower alcohols such as methanol or ethanol. The process is slow under normal conditions without the presence of a catalyst. Traditionally, an alkaline catal yst such as sodium or pota ssium hydroxide is used to catalyze and accelerate the reaction at standard temperatures and pressures. The catalytic reaction is complicated, however the necessity for a catalyst arises from the relative

PAGE 25

15 insolubility of alcohol in o ils. Catalysts provide a phase-transfer as well as an ionexchange effect which reduces reaction times by many orders of magnitude (Mittelbach, 2004). Transesterification will occur within a reasonable time period without the presence of a catalyst in a process known as supercritical methanolysis. Conditions are typically extreme with temperatures as high as 235 C and pressures in the range of 62 bars (Mittelbach, 2004). If temperatures and pressures are high en ough, methanol becomes fully soluble in oil and the r eaction occurs readily. Thes e conditions are typically not practical for industrial purposes, however in vestigations by Han (2005) using gas-phase co-solvents have been made which have resu lted in much milder reaction conditions. Following is a basic schematic of a methanol ysis reaction. By definition, the catalyst does not participate in th e reaction so is not shown in the schematic. Figure 2: Basic methanolysis reaction schematic Figure 2 is a highly simplified representation of transesterif ication with methanol. In actuality, the reaction oc curs in 3 steps. The triglyceride is first converted into a Glycerin Molecule Fatty Acid Chain Fatty Acid Chain F atty A c i d C h a in Methanol Meth y l Este r Glycerin Molecule Methanol Methanol Meth y l Este r Meth y l Este r Triglyceride

PAGE 26

16 diglyceride, then to a monoglyceride, and finally into a free glycerin or glycerol molecule. In each step, a methyl ester mol ecule is created. The total result is three methyl ester molecules and a glycerol mo lecule. The following equations show the reaction in more detail on a step-by-step basis. CH2 —O—COR1 CH2 —O—COR1 | | CH—O—COR2 + CH3OH <==> CH2 —O—COR2 + R3—COOCH3 (1) | | CH2 —O—COR3 CH2 —OH CH2 —O—COR1 CH2 —O—COR1 | | CH—O—COR2 + CH3OH <==> CH2 —OH + R2—COOCH3 (2) | | CH2 —O—COR3 CH2 —OH CH2 —O—COR1 CH2 — OH | | CH—O—COR2 + CH3OH <==> CH2 — OH + R1—COOCH3 (3) | | CH2 —O—COR3 CH2 — OH Equations (1), (2), and (3) de tail the 3-step methanolysis reaction in which a triglyceride molecule reacts with a methanol molecule to form a diglyceride plus a methyl ester molecule (Equation 1), a monoglyceride plus a methyl ester molecule (Equation 2), and finally a glycerin molecule pl us a methyl ester molecule (E quation 3). The result is 3 methyl ester molecules plus one glycerin mol ecule. The reaction is reversible so excess alcohol is typically used in practice to force the reaction towards ester production (Khan, 2002). The above equation formats were adapted from Mittelbach (2004).

PAGE 27

17 2.4.2 ESTERIFICATION Esterification, as it applies to biodiesel pr oduction, is the chemical reaction by which a fatty acid, typically a free fatty acid in a de graded or second-use o il, reacts with an alcohol to produce an alkyl ester and wa ter. The process differs from the transesterification reaction in that the reacti on is occurring directly between the alcohol and the fatty acid molecule. The intermediate steps of cleaving the fatty acid chains from the glycerin backbone are not present. For th is reason, no glycerin is produced during the esterification reaction. The following formula shows the basic esteri fication reaction with methanol. A fatty acid molecule reacts with a methanol molecule to form a methyl ester plus a water molecule: sulfuric acid R1-COOH + CH3OH R1-COO-CH3 + H2O (4) ffa methanol methyl ester water The above formula was adopted from Desh mane (2006) and represents the basic chemical reaction for all industr ial esterification reactions usi ng methanol as the alcohol. It is the formula for all reac tions performed in the experime ntal section of this thesis. 2.4.2.1 Need for Esterification Conventionally, virgin vegetable oils and hi gh-grade animal fats are the feedstock of choice for biodiesel production due to low levels of impurities, such as free fatty acids and sulfated proteins, which can cause problems with processing and final product quality. Rapeseed alone comprises of roughly 84% of the lipid stocks used for biodiesel

PAGE 28

18 production. By comparison, sunflower and palm oil each represent 13% of the feedstocks with soybean trailing with a 1% sh are. All other feedstocks such as waste fryer oils, animal fats, jatropha, peanut, must ard, etc. make up the remaining 2% (Pahl, 2005). With the economic and regulatory challenges outlined in the introduction, however, this trend is quickly ch anging (Biodiesel Magazine, 2008). Second use oils such as yellow or brown gr ease are thermallyor chemically-degraded waste-oils that primarily cont ain grease collected from rest aurant or industrial grease traps. Most of this oil is spent cooking oil from restaurants that has been thermally degraded by sustained high temperatures. It further degrades when in contact with water in the grease trap through a process know n as hydrolysis (Montefrio, 2010). This degradation produces molecules known as free fatty acids. Fatty acids will chemically react with the typical alkaline catalysts used in base-catalyzed biodiesel reactions to form soap. Two problems result from this: 1. The catalyst is consumed resulting in e ither an increased catalyst requirement and therefore higher chemical costs or an incomplete or failed reaction. 2. The reaction between the fatty acid molecule and catalyst creates soaps which manifest themselves as impurities in the biodiesel and must be washed out (Lotero, 2005).

PAGE 29

19 Free fatty acids are always present in oils, however mass concentrations above 4% will generate more soap than can be dealt with reasonably in a conventional base-catalyzed reaction and will prevent the reaction from going to completion in almost all cases (Tyson, 2002). Brown grease contains fatty acid concentrati ons in excess of 15% with typical values closer to 60%. It is not unusual for heavily degraded br own grease to contain nearly 100% free fatty acid (Tyson, 2002). These fact s clearly imply that conventional methods of biodiesel production will be ineffective with brown grease or other ultra-high-FFA oils as feedstocks. Acid-catalyzed esterification has been de monstrated be an effective method for converting moderately degraded feedstocks such as Yellow Grease and high-FFA animal fats into viable biodiesel. Yellow Grease pr imarily contains spent cooking oil that has not been hydrolyzed. For this reason, yello w grease rarely contains free fatty acid concentrations above 15% making it only moderate ly difficult to convert into biodiesel. Due to the greater FFA concentrations in brown grease, processing requires multiple esterification and dewatering stages as well as additional byproduct separation and purification steps (Tyson, 2002). At present, conversion of ultra-high-FFA o ils into biodiesel remains impractical and costly. Technology in existence today has fa iled to address three primary issues which

PAGE 30

20 collectively contribute to the failure of widespread adoption of brown grease as a principal feedstock for biodiesel: 1. Separation and purification of in-process biodiesel is challenging leading to inefficiencies, yield losses, and increased production costs. 2. Sulfur compounds remain in the finished product which fails to meet the ASTM D6751 specification of < 15 ppm of sulfur. Conventional attempts to distill the product to remove the sulfur are capital in tensive and inefficient. 3. Brown grease processing can lead to wa ste-water discharge that is heavily contaminated and expensive to process/ dispose of. High solid disposal and chemical costs exist for “dry wash” systems. As previously discussed, the es terification reaction of free fatty acids directly into methyl esters is a favored method of pre-treating degraded oils with high free-fatty acid concentrations. Alternative methods invol ve the saponification and then washing, or removal with water, of the resulting soaps. The removal of these soaps translates to a yield loss and can only be economically accomplished with relatively low FFA concentrations in the feedstock. It is not suitable for ultra-high-FFA oils. 2.4.3 INDUSTRIAL PROCESS Industrially, both esterification and transesterification are em ployed in a two-step process to first convert the FFA into alkyl esters and then to conve rt the remaining triglycerides into methyl esters. An outline of the process follows:

PAGE 31

21 1. The high-FFA oil is elevated in temperature and then methanol (Me OH) and sulfuric acid (H2SO4) are added in appropriate quan tity. The reaction is allowed to progress for several hours until the FFA concentration is reduced to an acceptable level for base-catal yzed transesterification. 2. The methanol is decanted which carries the majority of the sulfuric acid and water with it. The methanol is either neutra lized after decanting or neutralization is done prior to decanting with a base su ch as sodium hydroxide or potassium hydroxide. This converts the sulfur ic acid into non-corrosive salts. 3. The esterified oil is then transferred to another reactor where additional methanol and a base catalyst is added to transesterify the remaining triglyceride. 4. Washing and other post-processing steps ar e done to prepare the resulting methyl esters for sale as biodiesel fuel. The gl ycerin byproducts are tr eated to remove the soaps and excess methanol (Zullaikah, 2005). 2.4.4 ALCOHOLS The primary alcohols used for Biodiesel pr oduction in both tran sesterification and esterification reactions are of the lower type s, namely methanol and ethanol (Mittelbach, 2004). Each has distinct adva ntages and disadvantages. Th ey are described below in conjunction with an explanation of other po ssible alcohol species for the production of alkyl esters.

PAGE 32

22 2.4.4.1 Methanol Methanol is by far the most popular alcohol used in industr ial Biodiesel production. The primary reasons for this are due to its low price and high reactivity. In conjunction with alkaline catalysts, practical yi elds greater than 100% are t ypical with 80 of the conversion occurring within the first 5 minutes (Mittel bach, 2004). Post-separ ation of the reacted products occurs at nearly the same rate as the reaction which eliminates process bottlenecks. Other advantages include the fact that methanol has less of an affinity to atmospheric moisture absorption and retenti on and can be obtained in anhydrous form. Moisture removal can be achieve d by simple distillation. Methanol is typically a petroleum-based product although some research has gone into the production of methanol from agricultur al sources (Branson, 2002) from [Mittelbach, 47]. For that reason, it is c onsidered less environmentally friendly than ethanol. Many complain that Biodiesel will not be a true agricultural fuel until ethanol is widely implemented in conventional processing techniques. 2.4.4.2 Ethanol Ethanol is produced from the anaerobic fe rmentation of high-glucose carbohydrates followed by distillation. Also known as ethy l alcohol, ethanol is the alcohol found in alcoholic beverages for human consumption. The carbohydrate stock is typically derived from the germ of grains such as corn and wheat Ethanol itself is used as an alternative fuel in gasoline engines. Et hanol can be blended as high as 10% in most gasoline engines and as high as 100 percent with minor modifica tions or in “flex fuel” vehicles. Though it

PAGE 33

23 is considered a renewable resource, it is far more energy intensive to produce than Biodiesel. Whereas Biodiesel has a positive energy balance as high as 3.5:1 (National Biodiesel Board, 2008), ethanol ha s been reported to have an energy balance as low as 1.2:1 or even a negative balance (Journey to Forever, 2010). The energy balance of a fuel is the ratio between the quantity of fossil energy units co nsumed in the production of the fuel and the number of energy units yielde d by the end use of the fuel. For this reason, ethanol is far more expensive than methanol which is the primary reason for its limited use as a reactant for Biodiesel produc tion. Other advantages of ethanol besides the environmental ones are the comparativel y low toxicity in relation to methanol (Wikipedia, 2010). Ethanol also has an addi tional carbon atom which has been shown to increase the heat and cetane values of ethyl esters (Fillire s et al., 1995). Other disadvantages of ethanol are difficulty separati ng of the ester and glycerol phases, higher reaction temperatures, reaction sensitivity to trace moisture, and lower conversion than methanol (Mittelbach, 2004). 2.5 CHARACTERIZING THE ESTE RIFICATION REACTION The literature review produced no evidence of a formal attempt to characterize the esterification reaction of free fatty acids in ultra-high-FFA oils using methanol as the alcohol and sulfuric acid as th e catalyst. The term “characteriz ation” is used here to refer to a comprehensive comparative study of the primary variables or fa ctors that influence the level of free fatty acid reduction in the su lfuric acid catalyzed esterification reaction. As a product of the various investigations however, some understanding was gained

PAGE 34

24 about the effects of several variables of various unique reaction systems including those in oils containing elevated levels of FFA. In studying the production of methyl esters fr om Nile tilapia (fish ) oil using ultrasonic excitation, Santos et al. identified the meth anol to FFA molar ratio to be the most important factor influencing th e conversion of FFA to methyl esters followed by catalyst content. Response surface methodology was em ployed to determine an optimal operating condition of 9.0 alcohol to oil molar ratio and 2.0% wt/wt catalyst c oncentration at 30 C (Santos, 2010). Cardoso et al. (2008) studied th e effect of oleic acid concen tration, catalyst concentration, and temperature in esterifi cation reactions using SnCl2 catalyst and ethanol as the alcohol. High FFA levels (up to 10%). A ll three factors were determined to have desirable effects on the esteri fication reaction and the study co ncluded that tin chloride was a suitable catalyst for esterification. In addition, the study determined that ethanol in high excess to oleic acid (>120:1 molar ratio) had no discernable effect on the reaction yield or rate at varied levels (Cardoso, 2008). Ngo, et al. (2010) developed a process to manuf acture biodiesel using waste greases with free fatty acid concentrations ranging from 10% to 90% using sustainable methods. The group characterized several catalysts and th e resulting methyl ester products. The importance of this study was that it focused on low-cost feedstocks such as brown and yellow greases (Ngo, 2010).

PAGE 35

25 The 2-step acid esterification / base transest erification of rendered pork lard with acid value of 14.57 mg KOH/g was studied by Dias et al. (2009). The team determined that the prevailing factors for the esterification reaction were sulfuric acid and temperature (Dias, 2009). Moisture or methanol conten t were not explicitly studied nor were interaction effects between factors. Hahn et al. (2009) studied the effects of et hanol to oleic acid molar ratio, catalyst concentration, temperature, alcohol type, a nd free fatty acid type, on esterification reactions catalyzed by sulfuric acid under ultrasonic irradia tion. Optimal conditions were found at ethanol to oleic acid molar ratio of 3:1, catalyst concentration of 3%, and reaction time of 2 hours at 60 C (Hahn, 2009). Liu et al. (2006) studied the effects of wa ter on the esterification of acetic acid using sulfuric acid as the catalyst. They demonstrated that catalytic activity was reduced by increased water content and determined its eff ect to be -0.83. The researchers presumed that the primary mechanism fo r this deactivation was solvat ion of the catalyst by water (Liu, 2006). This gives credence to the pres ent study’s examination of water content in the esterification reaction of ultra-high-FFA oils. Many more such studies exist a nd support several of the postula tes of this thesis; however the literature is lacking a comprehensive study to fully characterize the effects of the primary reaction components (alcohol level, catalyst level, water content, and temperature) of sulfuric acid catalyzed esterification of ultra-high FFA oils.

PAGE 36

26 2.6 IMPROVING REACTION KINETICS Stavarache et al. (2003) first reported that sonochemistry appl ied to the transesterification reaction could dramatically improve reaction time and yield. They concluded that lowfrequency ultrasonic irradiat ion at both 28 and 40 kHz showed dramatic improvements over mechanical stirring (Stavarache, 2003). Others duplicated the effect in following years (Benitez 2004; Fang 2005; Colucci 2005). From 2006 until present, the number of papers on the topic increased dramatically wi th investigations in areas ranging from ultrasonically assisted exotic catalysis (Y ue, 2006) to in-situ tr ansesterification of sunflower oil using ultrasonic excitation (Georgogianni, 2008). In, 2009, Lee et al. used ultrasonic excitation to prepare methyl via esterification of fatty acids. After resolving overheating problems, they determined that reaction time was shortened dramatically 30 minutes for 93% yield (Lee, 2010). No consideration or comparison of mechanical or thermal energy input was made. As mentioned in the previous section, Santos et al. studi ed the effects of ultrasonic excitation on Nile tilapia oil (2010). Co mparisons were made between stirred and ultrasonically agitated reactions however their study did not c onsider energy intensity in comparison to the stirred reactions. As mentioned in the last section, Hahn et al. studied the effects of ethanol to oleic acid molar ratio, catalyst concentra tion, temperature, alcohol type and free fatty acid type, on esterification reactions cataly zed by sulfuric acid under ultr asonic irradiation. Agitated

PAGE 37

27 experiments were done with a stirrer with 35 watt output while s onication experiments were done with an ultrasoni c cleaner outputting 700 watts (Hahn, 2009). Clearly, the improvement of ultrasonic irradiation over st irring was not examined as a function of energy input. Deshmane, et al. studied the e ffects of ultrasonic irradiati on on the esterification of palm fatty acid distillates at 30 C and 40 C with stirring, ultrasonic agitation, and a combination of both. The effects of ultras ound, methanol to distil late ratio, catalyst concentration, and reaction temperature, we re studied. In addition, some kinetic modeling of the reaction was pe rformed. The effect of arti ficial water addition was not studied, nor was the effect of energy input between the two agitation methods. Clearly, though focused and highly valuable re search has been accomplished in the area of ultrasonically assisted esterification reac tions, there has been no attempt to reconcile the difference between stirring and sonicat ion on an energy input basis in either transesterification reactions or esterification reactions in either high or ultra-high-FFA oils. 2.7 SUMMARY OF THE LITERATURE REVIEW Some history and background on both the science and history of biodies el as an industry and as a process has been given. Justifica tion for the use of second-use, high free-fatty acid feedstocks and the associated requirement for acid-catalyzed esterification has been explored. The literature revi ew found a great deal of resear ch in both the characterization

PAGE 38

28 of certain components of the esterification reaction itself as well as sonochemistry as applied to the esterificati on reaction. It failed to find, however, a comprehensive exploration of the sulfuric ac id catalyzed esterification r eaction of ultra-high-FFA oils such as brown grease and fatty acid distilla tes. It also failed to find a dedicated comparison of the effects of agitation by stir ring and by sonication on the basis of energy input. The remainder of this thesis seeks to reco ncile these deficiencies in the current state of the art.

PAGE 39

29 3.0 EXPERIMENTAL From the literature review, it was discovered that the state of unde rstanding about the relative effects between contro llable factors in the esterification reaction of ultra-highFFA oils such as brown grease is lacking bot h in academia and in industry. It is known that these factors all influen ce the reaction, however it is no t known to what degree and to what extent some factors domi nate the others. This information would be valuable to either the scientist or the captain of industry because it is a foundation upon which optimal operating conditions can be built. This foundation will be experimentally established as a pr oduct of this work. In addition, with the many choices of sec ond-generation reaction technologies, little guidance is offered as to whether these reaction technologies ar e worth the capital investment. A second experiment is presente d here to demystify one particular reaction technology, sonication, and present it s sole advantage in the cont ext of its ability to input higher amounts of kinetic energy into the re action matrix than conventional methods. 3.1 SCOPE OF EXPERIMENTS The scope of the first experiment, esterificati on reaction characterization, is to establish a foundation for operational optimization. It is intended to identify the relative magnitudes of the four chosen experimental factors: su lfuric acid content, me thanol content, water

PAGE 40

30 content, and reaction temperature in reac tions involving ultra-hi gh-FFA oils. Brown grease was used in all experiments at an FFA concentration of ~85%. The experiment does not attempt to optimize the examined factor s, nor does it strive to suggest that these are the only important elements. In addition, it evaluates only two levels for each factor and any conclusions drawn can only be applie d to the range bracketed by the high and low level for each. Finally, the studied time range was the first 15 minutes of reaction. Typical esterification reactions in ultra-highFFA oils can run for several hours and it is possible that the relative magn itudes of effects will differ from those found in this work. An assumption was made here that the re lative magnitude of each effect remains proportional throughout the reaction. The scope of the second experiment, enhanced reaction kinetics evaluation, is limited to the evaluation of sonication as a reaction enhancement technique against conventional stirring as a function of energy input. Two factors were vari ed: agitation type (sonication or stirring) and catalyst level. As menti oned in the introduction, energy input was held constant for both agitation methods as a way of filtering out the effect of mechanical and thermal energy input into the system from any other effect. If the agitation-method categorical factor effect wa s found to be significant at constant energy input, then it would suggest that sonication o ffers a benefit over and above its ability simply to add more mechanical energy to the reaction matrix. Interaction between this effect and catalyst treatment level was also examine d. This experiment does not evaluate interactions between any of the other factors from the first experiment, nor does it attempt to optimize the power levels, fre quency, or other sonication parameters for

PAGE 41

31 sonochemistry. Since the power level of the sonication was held constant and only one level evaluated, this experiment does not iden tify any potential interaction effects that might occur at higher power le vels. And as with the firs t experiment, conclusions and observations drawn from the experiment apply only to the regions br acketed by the factor levels. 3.2 DESIGN OF EXPERIMENT The 2k factorial design was chosen for the two primary experimental systems in this body of work. Both systems were completely randomized designs using the default random number generator in Minitab 16. The designs for each system are presented and discussed in the following sections. 3.2.1 ESTERIFICATION REACTION CHARACTERIZATION There are four fundamental fact ors that potentially influence th e rate and completeness of the esterification reaction of free fatty acids in waste oils: 1. Relative quantity of methanol 2. Relative quantity of sulfuric acid catalyst 3. Relative quantity of water 4. Temperature of the reaction system

PAGE 42

32 A potential fifth factor is agita tion speed, however the effect of this factor was thought to be minor and, in many existing industrial desi gns, difficult to alter. For this reason, agitation speed was fixed for the purpose of th is experiment. Table 1 shows the factor levels and run order for the Esterifica tion Reaction Characterization experiment. Table 1: Esterification reacti on characterization experiment Run Order Standard Order Sulfuric Acid Addition Water Addition Methanol Addition Temperature 1 27 + – – + 2 20 – + – – 3 18 – – – – 4 28 – + – + 5 26 – – – + 6 33 + + + + 7 19 + – – – 8 24 – + + – 9 25 + + + – 10 30 – – + + 11 22 – – + – 12 23 + – + – 13 31 + – + + 14 29 + + – + 15 21 + + – – 16 34 0 0 0 0 17 32 – + + + 18 12 + + – + 19 14 + – + + 20 10 + – – + 21 4 + + – – 22 11 – + – + 23 1 – – – – 24 2 + – – – 25 17 0 0 0 0 26 16 + + + + 27 5 – – + – 28 8 + + + – 29 9 – – – + 30 13 – – + + 31 6 + – + – 32 3 – + – – 33 15 – + + + 34 7 – + + –

PAGE 43

33 A plus sign (+) indicates the high factor leve l, a minus sign (-) indicates the low factor level, and a zero (0) indicates the center point level (center points are discussed in future sections). Table 2 shows the corresponding treatment levels associated with the high, low, and center point levels. Table 2: Treatment levels for factorial design + – 0 Sulfuric Acid 500 l 150 l 350 l Water Addition 500 l 0 l 250 l Methanol 100 ml50 ml 75 ml Temperature 60 C 48 C 54 C 3.2.1.1 Justification for the Factorial Design From the literature review, each of the lis ted factors is known to affect reaction efficiency, however the rela tive effects between them are not known. For instance, should an operator bother to increase reaction te mperature or focus on reducing the water content of the lipid stock prior to starting th e reaction? It was desired to determine the relative effects of the main f actors and their interactions to gain a better understanding of where to focus time and capital in the optim ization of real-world industrial systems. Because this experiment is essentially a fact or screening experiment two treatment levels were chosen for each factor in order to simplify the design and minimize the number of required runs. Safeguards to test for curvatur e in the resulting model, which required the introduction of a few intermediate levels called center points, were also taken and this will be discussed in proceeding paragraphs.

PAGE 44

34 Because there were more than one or two f actors to evaluate and because it was desired to examine interactions between factors, a 24 factorial design was chosen for the experiment. The choice of a factorial desi gn provided a relative efficiency of 2.5 compared to an experiment evaluating each independently (Montgomery, 2005). This means that a one-factor-at-a-time experiment would require 2.5 times the number of runs as the 24 factorial design that was chosen in orde r to evaluate the effects and interactions of each factor. 3.2.1.2 Choice of Levels As outlined in Table 2, high, low, and center point levels were chosen for each of the four factors in the factorial design. The volum e of oil used for all reactions for both experiments was 50 ml. The levels of each li quid reactant sulfuric acid, water, and methanol were chosen based on a volumetric pe rcentage of the oil. Table 3 summarizes these percentages. Table 3: Concentration of treat ment levels by volume of oil + – Sulfuric Acid 1% 0.3% Water Addition 1% 0% Methanol 200% 100%

PAGE 45

35 3.2.1.3 Randomization As mentioned previously, the experiment was a completely randomized design. Randomization was achieved using the defau lt generator in Minitab 16. Experiments were run in the exact order of the rand omized output. This randomization can be observed in Table 1. Other experimental elements such as human resource and instrument allocation could not be randomized because, in each case, there was only one instance of each element. Chemicals were dr awn from lots and for this reason, blocking was used as described in a later section. 3.2.1.4 Replicates Determining the number of replicates was (a nd often is) an iterative process. From Montgomery, we recall that a replicate is an independent repeat of a unique factor combination (Montgomery, 2005). From a practical standpoint, replicatio n is required to rule out the possibility that an observed effect is caused by expe rimental error. It is the fundamental technique for determining statisti cal significance. A rep licate is not to be confused with a repeated measurement, or “duplicate” as referred to in this text. Duplication was used in these experime nts to correct for variability caused by instrumentation accuracy as well as iden tify possible botched measurements. In order to estimate the number of required re plicates, it was first necessary to determine the standard deviation between identical factor combinations. Three identical runs were made using the parameters in Table 4:

PAGE 46

36 Table 4: Initial parameters to determine required replicates Brown Grease 50 ml Methanol 50 ml Sulfuric Acid 250 l Water Addition 500 l Reaction Time 10 min. Reaction Temp. 50 C Stir Rate 700 rpm Two repeated measurements (dup licates) were taken after each run. The average value of each run is shown in Table 5 along with the associated standard deviation: Table 5: Initial experimental standard de viation for esterification characterization Sequence Final FFA % Run 1 28.892 Run 2 29.149 Run 3 30.189 S 0.68676 Once the standard deviation of successive identical runs wa s established, it was possible to arrive at an estimate of required replicates or the sample size. Estimated sample size was determined using the “Power and Samp le Size” tool in Minitab 16. An iterative approach was taken. It was desired to k eep the sample size as low as possible to minimize the number of runs for the expe riment. For this reason, it was deemed acceptable to choose the minimum possible number of replicates (2), enter the values for Power, Center Points, and the Standard Devi ation, and calculate the minimum observable effect based on the significan ce level. This effect woul d then be evaluated with a practical eye to determine if it was appropriate for the experiment or if additional

PAGE 47

r e a n F T f a r e P w ( M l i e plicates we r n d the valu e igure 3: Po w T he design h a a ctorial desi g e lative confi ower = 1 w here is th e M ontgomer y i kelihood th a r e needed. s entered as w er and sam p a d 4 factors g n is 4x4 = 1 dence that a e probabilit y y 2005). Fr o a t an observ e The approp r shown: p le size tool as previous l 1 6. A powe r a real effect i y of not reje c o m a practi c e d effect wi l 37 r iate tool fo r in Minitab l y stated an d r value of 0. i s not misse d c ting the nul c al standpoi n l l be identifi e r 2-level fac t 16 d the numbe r 95 was con s d by the mo d l l hypothesi s n t, the powe r e d if it actu a t orial desig n r of corner p s idered suff i d el. s when it is f r value is a m a lly exists. A n s was selec t oints for a f u i cient to giv e (5 f alse m easure of t h A power val u t ed u lle ) h e u e of

PAGE 48

38 0.95 means that there is a 95% chance of iden tifying a real effect and a 5% chance of missing it even though it is real. Blocking was anticipated (discussed in the next section), so 2 blocks were selected in the “Design…” option. A significance level ( ) of 0.05 was entered under “Options…” The significance level is the probability of rejecti ng the null hypothesis if it is true. An of 0.05 indicates that there is a 5% chance that an apparent effect is a product of statistical error and is not a real effect. Both the power level and signi ficance levels chosen for th is design were the default values in Minitab and are common values in many statistical designs. Higher power levels and lower significance le vels would be justified in optimization experiments where high certainty is required, how ever for the purposes and scope of this body of work and for determining the initial sample size, the default values were deemed to be adequate. The tool iteratively determined a maxi mum detectible effect of 0.933 at an level of 0.05 and a power level of 0.95 for a 2-level, 4-fact or, full-factorial design with two blocks, one center point, and a standard deviation of 0.687. The power curve is seen in Figure 17.

PAGE 49

39 Figure 4: Power curve for the esterifica tion reaction characte rization experiment Generally, from a practical standpoint, an eff ect with an impact on FFA percentage of less than about 3 would not be considered significant unless the cost to modify the factor(s) involved was minimal. Exceptions to this would be if the final FFA percentage was close to a spec which would determine the successful sale of product, or if no other methods existed to achieve a desired FFA percenta ge. This is discusse d in more detail in the Results and Discussions section; however it was touched on briefl y here in order to justify the choice of sample size. Since the maximum detectible effect calculated by Minitab was ~1 and an effect less than 3 would generally be considered to be not useful, the initial selection of 2 replicates was considered to be sufficient for the experiment Ultimately, the actual standard deviation 1.0 0.5 0.0 -0.5 -1.0 1.0 0.8 0.6 0.4 0.2 0.0 EffectPower Alpha0.05 StDev0.68676 # Factors4 # Corner Pts16 # Blocks2 # Terms Omitted0 Center PointsYes BlocksYes Terms Included In Model Assumptions 2, 1 Ctr Pts Per Blk Reps,Power Curve for 2-Level Factorial Design

PAGE 50

o v d 3 A i n o r d o t o n r e i m e x 4 t o w T f the experi m alue for one i scussed in t .2.1.5 Co n A fractional f n teractions w r der interac t esign woul d t her main e f n ly with thr e e solution V d m me d iate re j x periment ( M -factor expe o uncertaint y w as the only T able 6: Ava i m ent was fa r of the effec t he Results a n sideration o f actorial des i w ere of no i n t ions: that is d have been s f fect or with e e factor int e d esigns exis j ection of t h M ontgomer y riment; ho w y about the s reasonable p i lable factor i r greater tha n ts, albeit no t a nd Discussi f the Fracti o i gn was brie n terest in thi s interaction s s uitable sinc any two-fa c e ractions. A t only in ce r e possibilit y y 2005). A w ever aliasin g ignificance o p ossibility g i i al d esigns ( 40 n the initial e t to the detri m ons section. o nal Factori a fly conside r s design an d s between t h e no main e f c tor interacti o A brief exam i r tain designs y of using a f resolution I V g between t h o f main-eff e i ven the obj e Minitab 16) e stimate, an d m ent of the a l Design r ed for this e d it was only h e main effe c f fect would h ons, and ali a ination of T a s of 5 factor s f ractional fa c V half-frac t h e 2-factor i n e ct interacti o e ctives of t h d this affect e experiment, e xperiment. desired to e c ts only. A have been a l a sing would able 6 and r e s or more le d c torial desi g t ion design i n teractions w o ns. A fullf h e experime n e d the powe as will be Higher ord e e xamine firs t resolution V l iase d with a have existe d e calling tha t d to the near l g n for this s possible w w ould have l f actorial des i n t. r e r t V a ny d t l y w ith a l ed i gn

PAGE 51

41 3.2.1.6 Test for Curvature As mentioned previously, it was also desired to test for curvature in the design. Because a 2k factorial design assumes linearity by its natu re (predictive intermediate points would fall along a line drawn between the two factor le vels), a test for curvature is valuable if the experimenter suspects non-li nearity in the model or simply wishes to test for its existence (Montgomery, 2005). Center points are added to the model to test for the existence of quadratic effects in the fitted data. These second-order effects will cause a twisting of the plane generated by the interaction terms (Montgomery, 2005). Graphi cally, curvature would be observed if the center points do not lie near th e plane passing through the factor ial points. True center points will be treatment levels that are equidistant from each high and low treatment level. These have the advantage of not aff ecting the standard effects estimates in the 2k design (Montgomery, 2005). To have true cente r points, all factors must be numerical, not categorical, factors. Cent er points can be simulated in experiments that have one or more categorical factors as will be seen in sections dealing with the sonication experiment. For this design, one center point was chosen per block as a simple test for curvature. The center points can be observe d in Table 4 and the center point levels can be seen in Table 5. 3.2.1.7 Blocking A 24 full-factorial design with no blocki ng or center points would yield 24-1 = 16 unique runs or factor combinations. Adding one re plicate would result in 32 runs. Dividing the

PAGE 52

42 experiment into two blocks and adding one cen ter point per block resu lted in a total of 34 unique factor combinations. Blocking was chosen for two primary reasons: 1. Dividing the experiment into two blocks would result in 16 runs per block. At approximately 30 minutes total cycle tim e per run (reaction, sampling, analysis), this fit nicely into an 8-hour day. By completing one block inside of one day, the influence of performing the runs on different days could be adequately controlled for in the statistical analysis. 2. The quantities in each container of the various reagent chemicals used for the reactions and analyses were not sufficient for 34 complete runs. For instance, the container of reagent alcohol used as the solvent in the FFA measurement was 1 gallon or 3.8 liters. Recall that measurements were run in duplicate for a total of 68 total measurements barring any mistakes Each measurement required 100 ml of reagent alcohol for a total requireme nt of 6.8 liters of reagent alcohol. Similarly, about 4 liters of titrant was needed for the measurements, and the volume of each bottle of titrant was 2 liter s. By dividing the experiment into 2 blocks, each block could be r un with 1 batch of chemicals. The end result was that several potential nuisanc e factors could be statistically eliminated from the experimental results. The number of blocks was equal to the number of

PAGE 53

43 replicates, so each replicate occurred in a different block. As a result, no confounding existed with this design. 3.2.2 ENHANCED REACTION KINETICS EVALUATION An additional objective of this study was to examine a reaction kinetics enhancement technology, ultrasonically induced cavitation, fr om a total energy input standpoint. The literature review showed se veral instances of sonochemistry being applied to both transesterification and esteri fication reactions of low and high free fatty acid oils respectively. None, however, attempted a comparative examination of the technique under the lens of energy input. It is the belief of the author that the single advantage of sonochemistry in esterification and transest erification reactions is that the technology possesses the ability to deliver more mechani cal energy to the reacta nts than conventional stirring methods. Some claims of catalys t activation and complex hydroxyl reactions induced by sonication have been made (Mas on, 1990), but the author challenges these claims as having any significant impact on reaction rate improvement. It became an objective of this study to demonstrate that reaction acceleration is a function of energy input only and that any means for inputting an equivalent amount of energy (high-shear mixing for instance) is just as suitable as sonoc hemistry. The author regrets that he must spoil the ending to some degree in order to continue the current discussion because a significant portion of this experimental de sign was based on the findings of the first experiment from the preceding section.

PAGE 54

44 As seen in further detail in the Results and Discussion section, the Esterification Reaction Characterization experiment conc luded that sulfuric acid (cat alyst) addition rate had the single greatest impact on final FFA content af ter 15 minutes of reac tion time. This was followed by temperature and water addition rate (negative magnitude). Methanol content in the range and reaction period evaluated had negligible effect. A primary interaction between temperature and catalyst also existed but was determined to be of no practical significance. Due primarily to experimental logistics, a 22 factorial design was chosen with agitation method (s onication or stirring) at a common energy level and sulfuric acid content as the controllable factors. This choice is justified as follows: 1. A 22 design would give the minimum number of experimental runs while still demonstrating the primary objective. 2. Methanol content could not be practically varied because this would change the volume of the reactant mixture and th erefore influence the energy content required to agitate the mixture. This would make it difficult to fix the energy input for each level. 3. Temperature could not be easily varied between runs because the sonication reaction generated the heat for the reacti on through mechanical excitation as an influence of the cavitation itself. The onl y way to vary the temperature would be to induce additional cooling which would effectively remove energy from the system. This would cause the observer to draw false conclusions about the energy/agitation method relationship.

PAGE 55

45 4. Energy input was not chosen as a variable factor because, in addition to it being difficult to change, due to the range of energy use between the highest and lowest stirring settings, the difference in leve ls would have been so small that a significant effect would have had l ittle chance of being identified. 5. Varied sulfuric acid levels would resu lt in minute volumetric differences between levels and therefore could have potentially impacted th e required energy input to agitate the bulk material just as ch anging methanol volume would, however the order of magnitude between the sulfuric acid volume (l) and the reaction volume (ml) was thought to be negligible and w ithin the range of variability between reaction volumes of different experimental runs. For this reason, and because it was the single most significant effect, sulfur ic acid addition rate was chosen as the second controllable variable for the 2-level, 2-factor design. Table 7: Reaction accelera tion evaluation experiment Run Order Standard Order Agitation Type Sulfuric Acid 1 5 – – 2 9 – 0 3 2 + – 4 1 – – 5 4 + + 6 8 + + 7 3 – + 8 10 + 0 9 6 + – 10 7 – +

PAGE 56

46 Table 8: Reaction acceleration experiment treatment levels + – 0 Agitation Type SonicationStirring N/A Sulfuric Acid 500 l 150 l 325 l 3.2.2.1 Design Choices Since most of the background on the vari ous design choices was covered in the Esterification Reaction Charact erization section, the desi gn choices are covered more concisely in a single section here. A full factorial 22 design required only 2x2 = 4 runs. It was desired to test for curvature, however since one of the fact ors (agitation method) was cat egorical and not numerical, pseudo-center points were added to the model. Pseudo-center points are the center points for the numerical factors at each combination of the categor ical factors (Minitab Help, 2010). In the 22 design, only two numerical factor combinations existed yielding only two pseudo-center points to simulate a single center poin t for the model. A similar series of identical experimental runs as in the Esterification Reaction Characterization was used to establish the required sample size for the Enhanced Reaction Kinetics Evaluation. Reaction paramete rs are summarized in Table 9 and the results of the runs and resulting standard deviation are summarized in Table 10.

PAGE 57

47 Table 9: Initial parameters to determine required replicates Brown Grease 50 ml Methanol 50 ml Sulfuric Acid 325 l Reaction Time 15 min. Reaction Temp. 60 C Table 10: Initial experime ntal standard deviation for sonication experiment Sequence Final FFA % Run 1 26.49537 Run 2 26.41208 Run 3 26.79163 Run 4 27.24879 S 0.378094 The 2-Level Factorial Design Power & Samp le Size tool in Minitab 16 yielded the following power curve for a 22 design with no blocks, a significance level ( ) of 0.05, a power level of 0.95, and a st andard deviation of 0.378094:

PAGE 58

48 Figure 5: Power curve for th e sonication experiment From Figure 18, an effect of 1.31751 can be obs erved with only two replicates. Though higher than the observable effect with the Es terification Characterization Experiment, it still falls within the practical level of 3 referenced earlier. With the addition of the two pseudo-center poi nts, the total number of runs for the enhanced reaction kinetics evaluation experiment is 10. Because of the relatively small number of runs, and referring to the justif ication for blocking in the Esterification Reaction Characterization section, blocking was not used in this design. 1.5 1.0 0.5 0.0 -0.5 -1.0 -1.5 1.0 0.8 0.6 0.4 0.2 0.0 EffectPower Alpha0.05 StDev0.378094 # Factors2 # Corner Pts4 # Blocksnone # Terms Omitted0 Center PointsYes Terms Included In Model Assumptions 2, 1 Ctr Pts Per Blk Reps,Power Curve for 2-Level Factorial Design

PAGE 59

49 3.3 EQUIPMENT 3.3.1 SONICATOR As mentioned previously, th e sonicator used for experimentation was a Misonix XL2020 with a driver frequency of 20 kHz and maximum power output of 600 W. Table 11 shows the basic specifications of the device and related peripherals. Table 11: Specification for XL2020 sonicator Generator Input Voltage 200-260 Vac @ 50/60 Hz Full Load Current 7.5 Amps Fuse Rating 8 Amps (GDB8) Weight 16.5 lbs. (7.4Kg) Dimensions 7.5"x18.5"x11.6" (WxLxH) Output Voltage 1500 V rms (max.) Output Frequency 20 KHz (nom.) Convertor Weight 2 lbs. (0.9 Kg) Dimensions 8" L x 2.5" Dia. Materials Aluminum Standard Horn Weight 0.5 lbs. (0.45Kg) Dimensions 5" L x1.5" Dia. Materials Titanium Alloy 3.3.2 GLASSWARE Standard laboratory glassware was used for measuring and reacting the various chemicals for these experiments. Table 12 shows a lis t of the specific gl assware used and the purpose of each.

PAGE 60

T G 1 1 ( 5 4 5 3 F A c h d s u ( T u T able 12: Gl a G lassware 1 40 ml Beake r 1 00 ml Gradu a ( 1 ml graduati o 5 00 ml beake r 4 oz. glass ja r 5 ml. pipette .3.3 PIPE T igure 6: Ad j A Fisherbran d h annel pipe t i spensing s u u lfuric acid i 0.9 to 0.6 % T hermo Scie n sed in conj u a ssware use d a ted Cylinde r o ns) T TING SY S ustable pip e d ™ Finnpip t ter with an a u lfuric acid a i n the sonic a % of range) a n tific™ Fin n u nction with d in experi m Purpo s Reacti o Measu r alcoho l Interm e vessel f Reacti o Measu r reactio n S TEM e tte r ette II (fish e a djustable v o a nd deionize a tion reactio n nd the preci n tip 1000 pi p the pipetter. 50 m entation s e o n vessel for st r ement of met h l for FFA anal y e diate holding f or reaction cu r o n vessel for s o r ement and dis p n curve deriva t e rsci.com ca t o lume of 20 0 d water in t h n s. The me a s ion was 2. 0 p ette tips (T h t irred reaction s h anol, lipid sto y sis vessel for lipi d r ve derivation o nication react i p ensing of sul f t ion experime n t alog numb e 0 – 1000 l h e stirred re a a surement a c 0 0 to 0.6 l ( h ermo Scie n s ck, and reage n d stock, reacti o i ons f uric acid for n t e r 21-377-8 2 was used fo a ctions and f c curacy wa s ( 0.3 to 0.2% n tific item # 9 n t o n 2 2) singler measuring f or dispensi n s 6.00 to 1. 8 of range). 9 401070) w e and n g 8 0 l e re

PAGE 61

3 F M t h n u r e r e 5 3 A f o d 3 A R t e .3.4 STIR igure 7: Ma g M agnetic sti r h e factorial e u mber 14-5 1 e actants. F o e actions, an 12-127) wa s .3.5 BAL A A digital lab o o r FFA anal y etermine m e .3.6 HEA T A Fisher Sci e R PM was us e e mperature c BARS g netic stir b a r bars were u e xperiments 1 2-126) wit h o r pre-reacti o 8 mm D x 5 s used. A NCE o ratory bala n y sis and for e thanol loss. T ED STIR P e ntific Isote m e d for cond u c ontrol whic h ar u sed to agita t a Fisherbr a h dimension s o n lipid stoc k 0 mm L ma g n ce with an a weighing re P LATE m p magneti c u cting the sti r h means tha t 51 t e the stirre d a nd™ Polyg o s 9.5 mm D k stirring an d g netic stir ba a ccuracy of actor quanti t c stirrer with r re d reactio n t it has heat i d reactions. F o n Stir Bar ( x 25.4 mm L d for the re a ar (fishersci 0.01 g wa s ties before a h maximum r n s. The plat e i ng ability, b For the reac ( fishersci.co m L was used f a ction rate c u com catalo g s used for w e a nd after rea c r otational o u e has unidir e b ut not cooli n tions involv m catalog f or agitatin g u rve derivat i g number 14 e ighing sam p c tion to u tput of 140 0 e ctional n g ability. T ed in g the i on p les 0 T he

PAGE 62

52 Isotemp has a thermocouple temperature probe for the purposes of loop thermal control and temperature monitoring. 3.3.7 AUTO TITRATOR An 809 Titrando potentiometric auto-titrator manufactured by Metrohm of Sweden was used for the FFA measurements. The devi ce was controlled by the Tiamo 1.1 computer interface software. Additional information on the Titrando can be found in the Appendix. Descriptions of the use and setup of th e device are covered in future sections. 3.3.8 POWER METER Power measurements were taking for the reaction kinetics enhancement evaluation experiments using a Watts Up? Pro electricity monitor manufactured by Electronic Educational Devices. Data acquisition was accomplished by a USB connection from the Watts Up? meter to a laptop computer runni ng the Watts Up Real Time version 0.10.7.14 data logging software. 3.3.9 KARL FISHER MOISTURE ANALYZER Moisture analyses were conducted usi ng a Karl-Fisher moisture analyzer.

PAGE 63

53 3.4 MATERIALS 3.4.1 METHANOL The methanol used for the esterificati on reactions in both the esterification characterization experiment and the enhanced reaction kinetics ev aluation experiment was provided by Univar. The lot number was JO09880613 and the product number was 298001. The methanol was analyzed by Karl Fisher in duplicate for an average of 0.1339% moisture. 3.4.2 BROWN GREASE Brown grease was used as the high FFA feedst ock for all esterifica tion reactions and was provided by CHP ByProducts, LLC. The grown grease was tested for FFA using the 809 Titrando automatic titrator in duplicate fo r an average of 85.348% FFA. The moisture content of the brown grease was tested by Ka rl Fisher in duplicate for an average of 0.4956% moisture. 3.4.3 REAGENT ALCOHOL Ethanol Government Formula C was used as the reagent alcohol solvent for the FFA titrations. The reagent alcohol (lot number 8Z040809182, product number 763541) was manufactured by Univar.

PAGE 64

54 3.4.4 SULFURIC ACID Sulfuric acid 98% (manufacturer EMD Chem icals, fishersci.com catalog number 50-947796) was used as the catal yst for the esterification. 3.4.5 DEIONIZED WATER Deionized water was used in the esterifica tion reaction characteri zation experiments to test water content as a factor. The water was produced with a US Filter laboratory deionization system. 3.5 REACTOR DESIGN 3.5.1 STIRRED REACTOR DESIGN The stirred reactor setup is shown in Figure 8. All stirred reactions were conducted using this design. Figure 8: Stirred reactor setup Isotemp Wire Thermocouple Probe Stir Bar Reactants Cling Wrap Magnetic Driver 140 ml Beaker Heater

PAGE 65

55 The stirred reactor design consists of a 140 ml beaker atop a heated stir plate. Reactants (brown grease, methanol, sulfuric acid, and deionized water) are added to the beaker in quantities determined by the experimental design. The beaker is covered with cling wrap which provides a superior barrier for inhibi ting the escape of methanol vapors from the reaction zone. Stirring is accomplished with a 9.5 mm diameter, 25.4 mm long magnetic stir bar driven by a magnetic stirrer within the hot plate. Heat ing is provided by an electric resistance heater within the hotpla te controlled by feedback from a thermocouple probe. The probe is submerged beneath th e liquid level to provide an accurate temperature readout of the reac tant system. A control loop maintains temperature at the desired setpoint. 3.5.2 SONICATION REACTOR DESIGN The design of the sonication reaction vessel underwent severa l incarnations during the course of the current research. A basic sche matic of the final setup is shown in Figure 9. Figure 9: Sonication reactor setup 20kHzGenerator ReactionCell Sonicator Reactants CavitationZone Rubber Coupling

PAGE 66

56 The sonicator is connected to the 20 kHz pow er supply and lowered into the reaction cell at a fixed depth provided by the rubber coupl ing. The reaction cell consists of a 4 oz. glass jar mounted below the sonicator probe. The reaction cell is coupl ed to the sonicator probe with a 1 ” X 1 ” rubber coupling. Th e coupling is secured with pipe clamps to the probe neck as well as to the jar rim. R eactants (high FFA oil, methanol, and sulfuric acid) are added to and contained within th e reaction cell. The coupling inhibits the escape of methanol vapor as well as focuses the sonic energy into the reaction cell. This accomplishes two things: it retards the evapor ation of methanol and prevents sonic energy from reflecting out of the reaction cell. Both provide better experimental control and enhanced repeatability. In addition, the r ubber coupling serves as a suitable spacer to ensure a standard probe depth for all reac tions. A thermocouple wire (not shown in Figure 9) is inserted between the probe neck and the 1 ” end of th e rubber coupling with the junction end submerged beneath the liquid le vel. The thermocouple wire is connected to a digital thermocouple analyzer fo r temperature readout, and monitoring. The basic components of the above setup rema ined the same; however several revisions were made in order to promote stable reacti on conditions. Early experiments resulted in a high degree of inconsistency of performa nce. This was a result of crude and nonrepeatable apparatus setups. Initially, the probe was lowered into a 140 ml glass beaker. The probe was mounted on a stand with clamps that made it difficult to control the probe depth. Parafilm was used to cover the glass beaker and a small hole was punched in the film to allow for probe insertion. Methan ol vapors easily escaped from around the edges of this hole resulting in incomplete and non-repeatable reactions. The rubber coupling

PAGE 67

57 design was eventually implemented and so lved both the methanol escape and sound reflection issues. Initial experiments made no provision for te mperature control and it was soon found that the effects of the initial exothermic energy re lease elevated temperatures near or beyond the boiling point or methanol. Most of the methanol was allowed to escape from the reaction vessel before a good reaction was produ ced. Attempts were made to control the temperature with a water bath, however simplic ity prevailed and the final setup consisted of an electric fan positioned to blow dire ctly on the reaction cell. Cooling was occasionally assisted by squirting a small am ount of anhydrous methanol directly on the cell to provide evaporative cooling. This method of control was manual, however required only occasional attention and was typically only used at the beginning of the reaction to correct for the initial exothermic energy release. Temperature was maintained to within 1 C for the duration of the reaction using this method. 3.6 METHODOLOGY 3.6.1 EXPERIMENTAL SEQUENCE Experiments were run on a laboratory scale using standard laboratory glassware and equipment. The basics steps for the experiments are listed below: 1. Oil and methanol are measured up in th e volumes and order specified by the experimental design and placed into the appropriate reaction vessel.

PAGE 68

58 2. In the stirred reactions, a magnetic stir bar is added. 3. The reaction vessel is weighed and the weight recorded. This step is to monitor methanol loss during the reaction. 4. Agitation is started and, with the stirred reactions, the appropriate temperature set point specified by the experimental desi gn is selected and heating is started. 5. The oil/methanol mixture is allowed to heat to the temperature setpoint. With the sonication reactions, the ultrasonically i nduced cavitation generates the required heat without the need for external heating. 6. Once the temperature set point is reache d, the water is added if required by the design. 7. After the water is added, the sulfuric aci d is added in the volume required by the design and the timer is simultaneously started. The power measurement data acquisition program is also started for th e enhanced reaction kinetics experiment only. 8. The esterification reaction is exothermic and will generate heat. For the stirred reactions, the control loop will reduce the heat input to maintain the temperature setpoint. For the sonication reactions, tr im cooling may be required. The trim cooling process is described in se ction 3.3.2, Sonication Reactor Design. 9. After the 15 minute reaction time period has elapsed, the alarm will sound and two duplicate samples are immediately draw n from the reaction mixture with a 1 ml transfer pipet. The samples are placed into two separate pre-weighed 140 ml beakers.

PAGE 69

59 10. The beakers are re-weighed and their final we ight is recorded. The final weight is subtracted from the initial weight to calculate the weight of the sample. 11. As quickly after the sample is drawn as possible, 100 ml of reagent alcohol is added to the 140 ml beaker containing th e sample. This halts any continued reaction as well as eliminates the evaporat ion of methanol from the sample. The reagent alcohol acts as the solvent for titration analysis. 12. A stir bar is added to the two prepared duplicate samples. 13. The samples are analyzed for FFA using the titration procedure outlined in X using the Metrohm 809 Titrando automatic titrator. 14. The FFA percentages produced are the fr ee fatty acid % by weight of the total reaction system (oil, methanol sulfuric acid, and water) in the ratios specified by the experimental design for that particular run. The weights of the other reactants must be mathematically reduced to l eave only the FFA% of the oil component. This is done by simple arithmetic usi ng a Microsoft Excel™ spreadsheet. 15. The calculated final FFA% of the oil is ente red into the factorial table in Minitab 16 and the effects analyzed using ANOVA a nd regression analysis to determine the statistically significant factor effects.

PAGE 70

60 Figure 10: Experimental flowchart 3.6.2 REACTION PARAMETERS The parameters for the esterification reacti on characterization and the enhanced reaction kinetics evaluation are shown in Tables 13 & 14 below. Continue heating / agitating Start Heater Add Stir Bar Select Stirred Reactor Select Temperature Setpoint Record reaction vessel weight Start Agitation FFA% Select Sonication Reacto r Sonication reaction? Sonication reaction? Temp. reached? Water needed? Add required water vol. Add required H2SO4 vol. Add required Oil & MeOH Heat & agitate for 15 min. Trim cool if necessary Prep & analyze samples for FFA% No Yes Yes No Yes No Yes No Enhanced reaction Kinetics? Start 15 minute timer Start Power (W) data logging No Yes Start

PAGE 71

61 Table 13: Fixed reaction conditions for esterification reacti on characterization Level Brown Grease 50 ml Agitation Rate 700 rpm Table 14: Fixed reaction conditions for enhanced reaction kinetics evaluation Level MeOH Content 50 ml Brown Grease 50 ml Power Level 33 W The variable factors were varied accordi ng to the experimental design for each experiment discussed in section 3.2. 3.6.3 MEASUREMENTS The experimental component of this th esis required two primary categories of measurements to be taken: 1. Free Fatty Acid % for all reactions 2. Rate of energy input for the enha nced reaction kinetics evaluation Other measurements such as weights, temperature, and moisture content were made, however these were secondary to the experime ntal goals and served as support to the laboratory activities. Two software packages we re used in conjunction with the respective instruments for which they were designed. Th ey are detailed in the following sections.

PAGE 72

3 M M t h s c 1 F .6.3.1 Fre e M easuremen t M etrohm 80 9 h e Tiamo 1. 1 c reenshot o f 1. igure 11: Ti e Fatty Acid t s of final fr e 9 Titrando a u 1 computer i f the main g r amo 1.1 wo r Measurem e e e fatty acid u tomatic titr a i nterface so ft r aphical use r r kplace GU I 62 e nts s content in a tor. Stand a ft ware desig n r interface f o I reaction sa m a rd wet titra t n ed specific a o r the Tiam o m ples were w t ion method s a lly for the T o software is w ith the s were adap t T itrando. A shown in F i t ed to i gure

PAGE 73

63 Calculations for the final FFA% were done automatically by the software using the following formula: (‘DET U.EP{2}.VOL’ – ‘DET U. EP{1}.VOL’ – ‘CV.FFAbk’) ‘DET U.CONC’ 28.2 / ‘MV.Sample size’ (6) Where: ‘DET U.EP{2}.VOL’ is the volume of titrant dispensed at the second inflection point ‘DET U.EP{1}.VOL’ is the volume of titrant di spensed at the first inflection point ‘CV.FFAbk’ is the volume of titrant disp ensed with a solvent blank titration ‘DET U.CONC’ is the normality of the titrant 28.2 is the molecular weight of oleic acid divided by 10 and ‘MV.Sample size’ is the user inputted sample weight As the titration runs, the voltage is measur ed with respect to ti me and plotted. Two inflection points will occur: one when the sulfuric acid endpoint is reached, and the other when the FFA endpoint is reached. Equati on 6 calculates the FFA based on these two endpoints. A plot of a typical titration curve with the sulf uric acid and FFA endpoints is shown in Figure 12.

PAGE 74

F E e v F t i 3 T W w a l T a c igure 12: Ti E P1 is the su l v aluated for FA can be d i trations usi n .6.3.2 Ene T he energy i n W atts Up? P r w as measure d l l pieces of e T he baseline c counted fo r amo 1.1 titr a l furic acid e n every FFA d ifferentiate d n g an indica t rgy Input M n put for the E r o meter an d d and recor d e quipment a s power dra w r when tuni n a tion plot n dpoint and analysis. A d from the s u t o r solution a M easurement s E nhanced R d software. T d ed. The ba s s sociated w i was differe n n g the powe r 64 EP2 is the F major adva n u lfuric cont e a s the FFA e s eaction Kin e T he baseline s eline powe r i th either ex p n t for each e r consumpti o F FA endpoi n n tage of thi s e nt. This is i e ndpoint det e e tics experi m e power dra w r draw is de f p erimental s e xperimenta l o n for each e n t. A simila r s titration m e i mpossible w erminant. m ents was m w for each e q f ined as the p etup detaile d l setup and t e xperiment a r curve was e thod is that w ith manual m easured wi t q uipment se t p ower draw n d in section t his value as a s it was onl y the t h the t up n by 3.5. y

PAGE 75

65 desired to measure the power drawn for the reac tions, not that of the equipment itself. In addition, power measurement was started after the reactants, equipment, and glassware was allowed to heat up and reach setpoint. This eliminated the measurement of nonsteady state energy consumption. The author acknowledges that energy losses will exist for each of the experimental setups and that these losses may be different. The losses are defined for each setup as follows: 1. Stirred Reactions a. Radiative and convective thermal losses from the hotplate surface b. Radiative and convective thermal losses from the beaker walls c. Evaporative losses from the methanol vaporization 2. Sonication Reactions a. Sound reflection from the reaction vessel b. Radiative and convective thermal losses from the beaker walls c. Evaporative losses from the methanol vaporization Though the sources of loss are somewhat differe nt between setups, re call that the primary objective is to determine any superiority of ultrasonic irradiation over conventional stirring over and above its ability to deliver more energy to the reaction system. This includes losses, be they for better or for wo rse. In addition, the configurations are thought to approximate those seen in large-sc ale industrial environments. For instance, where thermal losses from the surface of the hotplate might be signifi cant, it is no less so

PAGE 76

66 than losses from the surface of jacketed tank heaters, steam boilers, heat exchangers, etc. seen in industrial processes. It was desired to accommodate these losses in the experiment, and any differences between losse s of the two experimental setups would adequately account for those seen on an industrial scale.

PAGE 77

67 4.0 RESULTS AND DISCUSSION 4.1 ESTERIFICATION CHARACTERIZATION Table 15 shows a table of estimated effect s and coefficients for the esterification characterization experiment from the ANOVA output in Minitab 16. Table 15: Minitab output of the estimated effects ----------------------------------------------------------------------------Estimated Effects and Coefficients for FFA (coded units) Term Effect Coef SE Coef T P Constant 40.62 0.4804 84.56 0.000 Block 1.27 0.4661 2.73 0.015 Sulfuric Acid Content -33.18 -16.59 0.4804 -34.53 0.000 Water Content 6.50 3.25 0.4804 6.77 0.000 Methanol Content 0.69 0.35 0.4804 0.72 0.481 Temperature -13.93 -6.97 0.4804 -14.50 0.000 Sulfuric Acid Content*Water Content 0.47 0.24 0.4804 0.49 0.630 Sulfuric Acid Content* -0.70 -0.35 0.4804 -0.73 0.478 Methanol Content Sulfuric Acid Content*Temperature -2.49 -1.24 0.4804 -2.59 0.020 Water Content*Methanol Content -0.06 -0.03 0.4804 -0.06 0.950 Water Content*Temperature -0.85 -0.43 0.4804 -0.88 0.389 Methanol Content*Temperature -0.51 -0.25 0.4804 -0.53 0.605 Sulfuric Acid Content*Water Content* 0.40 0.20 0.4804 0.42 0.682 Methanol Content Sulfuric Acid Content*Water Content* 1.13 0.56 0.4804 1.17 0.258 Temperature Sulfuric Acid Content* -1.40 -0.70 0.4804 -1.46 0.165 Methanol Content*Temperature Water Content*Methanol Content* 1.22 0.61 0.4804 1.27 0.223 Temperature Sulfuric Acid Content*Water Content* -0.38 -0.19 0.4804 -0.39 0.699 Methanol Content*Temperature Ct Pt -9.77 1.9808 -4.93 0.000 S = 2.71759 PRESS = 742.588 R-Sq = 98.94% R-Sq(pred) = 93.34% R-Sq(adj) = 97.82% ----------------------------------------------------------------------------

PAGE 78

68 An examination of the P-values for each e ffect located along the right-hand column of Table 9 reveals several significant effects. Significance, which is the notion that the observed response is not likely a product of chance as described in section 3.1.2.4, is determined by comparing the P-value to the significance level ( ). Recall from section 3.1.2.4 that the significance level us ed for this experiment is = 0.05. Effects with Pvalues less than are considered to be statistically significant. 4.1.1 STATISTICAL RESULTS 4.1.1.1 Interaction Effects Only one interaction effect, Sulfuric Ac id Content*Temperature, was found to be statistically significant with a P-value of 0.02. The 2-wa y interaction terms can be evaluated visually in Figure 13.

PAGE 79

69 Figure 13: Interaction effects plots for the esterification characterization experiment The interaction effects plots are interpreted by looking at how the lines fall in each plot. Nearly parallel lines indicat e no interaction where non-para llel or intersecting lines suggest interaction. As can be seen, onl y the Sulfuric Acid Content*Temperature interaction term (upper right-hand plot) is appreciably non-paralle l and therefore shows slight interaction. As will be discussed in the Practical Evaluation section, however, this effect is minor and is of no practical significance. 1 0 -1 1 0 -1 1 0 -1 60 40 20 60 40 20 60 40 20Sulfuric Acid Content Water Content Methanol Content Temperature -1Corner 0Center 1Corner Content Acid Sulfuric Point Type -1Corner 0Center 1Corner Content Acid Sulfuric Point Type Content Acid Sulfuric Point Type -1Corner 0Center 1Corner Content Water Point Type -1Corner 0Center 1Corner Content Water Point Type -1Corner 0Center 1Corner Content Methanol Point TypeInteraction Plot for FFAFitted Means

PAGE 80

70 4.1.1.2 Main Effects The main effects Sulfuric Acid Content, Wa ter Content, and Temperature, all of which had P-values of 0.00, were all significant. Figure 14 graphically summarizes these effects. Figure 14: Main effects plot s for the esterification char acterization experiment The coded variables are plotted against the gr and mean. The grand mean is the “mean of all observations across al l factor levels” (Minitab 16 Hel p, “grand mean”) rather than the mean of observations within each factor sett ing. A negative slope in dicates a factor that results in a lower FFA concentration after th e 15 minute reaction time at its highest level and a higher FFA content at its lowest level. A positive slope indicates the inverse. A 1 0 -1 60 50 40 30 20 1 0 -1 1 0 -1 60 50 40 30 20 1 0 -1 Sulfuric Acid ContentMean Water Content Methanol Content Temperature Corner Center Point TypeMain Effects Plot for FFAFitted Means

PAGE 81

71 nearly horizontal line indicates that the f actor has no effect on the system within the studied region. Since the goal is to reduce FFA concentra tion, negative slopes are considered desirable and positive slopes are considered undesirable. Steeper slopes suggest a greater effect on the final FFA concentration. Again, Sulf uric Acid Content and Temperature both accelerate FFA reduction while Water Content retards FFA reduction. Methanol Content was not significant with a Pvalue of 0.481 > 0.05. This came as a surprise as will be testified to in the Practical Interpretation section. 4.1.1.3 Covariate Effects Initially, a covariate, methanol loss was added to the model. During the experiment, a certain amount of methanol evaporated from the reaction chamber. This had the potential to affect the outcome of the design so the lo ss was measured and added as a covariate to the model. Methanol loss was shown to have no statistically signi ficant effect and was subsequently removed and the data refitted without the term to provide a cleaner fit with less statistical noise. From the data, the fact that methanol loss had no statistically significant effect is not surprising since the main effect of methanol content was not significant. Covariates were also added to the Sonication Experiment as will be seen later.

PAGE 82

72 4.1.1.4 Blocking Effects As mentioned previously, blocking was used to rule out influences on the model caused by performing the reaction on different days an d by using different chemical lots during the experiment. This proved to be a good desi gn choice since the Block effect term has a P-value of 0.015 and is therefore considered a significant effect. It is not known whether the effect came from the use of different chem ical lots or from performing two halves of the experiment on different days since the e ffect is a composite of the two variables. 4.1.1.5 Center Point Effects As described in section 3.2.1.6, center points were us ed in the design to test for curvature. As with blocking, this design c hoice proved to be a fortunate one as the model distinctly revealed the center point term to have a si gnificant effect at a P-value of 0.00. The center points are also shown within the main effects plot. Recall that curvature exists when the center point does not lie along the line drawn betw een the two factor levels. Figure 14 clearly reveals the cu rvature in the fitted data. 4.1.1.6 Relative Effects Figure 15 shows a normal proba bility plot of the effects of the esterification characterization experiment. The normal pr obability plot more clearly shows that sulfuric acid content, temperature, water content, and an interaction between temperature and sulfuric acid content have significant effects on the esterification reaction within the

PAGE 83

73 chosen level ranges. Within the chosen ra nge of 50 ml to 100 ml of methanol content (100% to 200% of the volume of oil used), no effect was observed due to methanol. Effects with negative values are considered to be beneficial because the practical objective is to reduce FFA concentration. In versely, effects with positive values are considered to be counteractive to the goal of reducing FFA concentration. Figure 15: Normal probability plot of the e ffects of the esterifi cation characterization experiment Sulfuric acid content had the greatest benefici al effect relative to the other effects at a standardized value of -33.18 followed by te mperature at -13.93 and then the 2-way interaction between temperature and sulfuric acid content at -2.49. Water content had a counteractive effect of 6.50. Sulf uric acid content had, by far, th e greatest effect at nearly twice the next most significant effect. The ha lf-normal probability plot of the effects is 0 -10 -20 -30 -40 99 95 90 80 70 60 50 40 30 20 10 5 1 Standardized EffectPercent ASulfuric Acid Content BWater Content CMethanol Content DTemperature FactorName Not Significant Significant Effect Type AD D B ANormal Plot of the Standardized Effects(response is FFA, Alpha = 0.05)

PAGE 84

74 given in Figure 16. The relative magnitudes of the effects are more easily visualized in the half-normal plot. Again, sulfuric acid co ntent is shown to have the greatest effect accounting for over 95% of the tota l variability of the model. Figure 16: Half-normal probability plot of the effects of the esterification characterization experiment A pareto chart is another useful way to vi ew the fitted data. Figure 17 shows a pareto chart of the standardized effects of the es terification characterizat ion experiment. The error term was calculated from the difference between the two replicates and was used to draw the significance line on the Pareto chart. Factors and their in teractions with an effect of magnitude above the error term of 2.12 as indicated on the chart are considered to be statistically significant while factors or interactions with an effect of magnitude 35 30 25 20 15 10 5 0 98 95 90 85 80 70 60 50 40 30 20 10 0 Absolute Standardized EffectPercent ASulfuric Acid Content BWater Content CMethanol Content DTemperature FactorName Not Significant Significant Effect Type AD D B AHalf Normal Plot of the Standardized Effects(response is FFA, Alpha = 0.05)

PAGE 85

75 below 2.12 are not considered to be statistically significant. This provides a quick, visual way to determine which effects are significant without having to look at the P-values. Figure 17: Pareto chart of the effects of th e esterification character ization experiment Clearly, Figure 17 reveals sulfuric acid cont ent to be twice the ma gnitude of the next greatest effect, temperature. In addition, the interaction between sulf uric acid content and temperature is revealed to be of marginal statistical significance. All other two-, three-, and four-way interactinos are of no statistical significance. BC ABCD ABC AB CD C AC BD ABD BCD ACD AD B D A 35 30 25 20 15 10 5 0 TermStandardized Effect 2.12 ASulfuric Acid Content BWater Content CMethanol Content DTemperature FactorNamePareto Chart of the Standardized Effects(response is FFA, Alpha = 0.05)

PAGE 86

76 4.1.1.7 Regression Equation In order to obtain a cleaner regression equa tion with better fit, the model was refitted excluding the statistically nonsignificant terms. Table 16 shows the Min itab 16 output of the refitted re gression data. Table 16: Minitab output of refitted data -------------------------------------------------------------------------Estimated Effects and Coefficients for FFA (coded units) Term Effect Coef SE Coef T P Constant 40.62 0.4510 90.08 0.000 Block 1.27 0.4375 2.91 0.007 Sulfuric Acid Content -33.18 -16.59 0.4510 -36.78 0.000 Water Content 6.50 3.25 0.4510 7.21 0.000 Temperature -13.93 -6.97 0.4510 -15.44 0.000 Sulfuric Acid Content*Temperature -2.49 -1.24 0.4510 -2.76 0.010 Ct Pt -9.77 1.8594 -5.25 0.000 S = 2.55112 PRESS = 480.486 R-Sq = 98.42% R-Sq(pred) = 95.69% R-Sq(adj) = 98.07% -------------------------------------------------------------------------As can be seen, the standard deviation and R2 terms are improved over the fit that incorporates all of the non-si gnificant terms (Table 14). From the Coef column in Table 15, coefficients for the regr ession equation are listed. 4.1.1.8 Model Adequacy Check Residual plots were used to check the ade quacy of the regression model. Figure 18 shows four diagnostic plots that are usef ul in evaluating whether the regression assumptions apply to the data. Residuals ar e simply the difference between the fitted response data and the observed values. They are indicative of how well the fitted model represents the real-world data.

PAGE 87

77 Figure 18: Residual plots for the esteri fication characterization experiment The normal probability plot in the upper left of the four plot display of Figure 18 shows the residuals falling along a re latively straight line indicating that the data is normally distributed. There is slight curvature at the tails of the plot which indicates only minor skewness in the data suggesting that the mode l is relatively symmetrical. There are no outliers and there is no diverging slope to the plotted residuals which indicates that there are no unidentified variables influencing the design. The versus fits plot in the upper right hand qua drant of the four plot display of Figure 18 shows a plot of residuals vs. the fitted re sponse values. The variance of the residuals seems to be randomly distributed with respec t to final FFA% indicating constant variance in the error term, no missing qua dratic terms, and no outliers. 5.0 2.5 0.0 -2.5 -5.0 99 90 50 10 1 ResidualPercent 60 40 20 4 2 0 -2 -4 Fitted ValueResidual 4 2 0 -2 -4 8 6 4 2 0 ResidualFrequency 30 25 20 15 10 5 1 4 2 0 -2 -4 Observation OrderResidualNormal Probability PlotVersus Fits HistogramVersus OrderResidual Plots for FFA

PAGE 88

78 The histogram in the lower left quadrant of Figure 18 is perfectly bell-shaped and confirms the lack of outliers or skewness in the fitted data. L ong tails or disconnected bars would indicate skewness or outliers respectively and this is not the case with the histogram suggesting that experime ntal technique was satisfactory. The versus order plot in the lower right quadrant of Figure 18 shows a plot of the residuals vs. the order of the experimental ob servations. Patterns in this plot would indicate that the order of the measurements had an influence on the final FFA%. The plot shows relative randomness and therefore indica tes that the order of the runs did not greatly affect the final FFA concentration. The examination of the four residual plots in Figure 18 show a model well suited to ANOVA and regression analysis and gives confidence in the design choices for the factorial design as well as the experimental technique used in the acquisition of the data. 4.1.1.9 Power Analysis As described in section 3.2.1.4, an estimate of the standard deviation of the model was used to determine the sample size at an accep table power level. It is now time to reevaluate the power curve using the actual stan dard deviation of the experimental data to determine whether the initial sample size assumption was correct. From Table 15, the standard deviation of the fitted data of the esterification reaction characterization experiment is 2.71759 vs. the original estimate of 0.687 from Table 5. Since the standard

PAGE 89

d o F F l e s h a eviation dir e f sample siz igure 19: P o igure 19 sh o e vel factoria l h own along power curv e e ctly influe n e for the ex p o wer and sa m o ws the inp u l design. T h with the ot h e for the fo u n ces the sam p p eriment wa s m ple size wi n u ts for the P o h e four signi h er required d u r effects wa 79 p le size, thi s s sufficient t n dow o wer and Sa m f icant effect d ata in their s generated. s raises a co n t o detect the m ple Size t o t s were inpu t respective f n cern that t h desired ch a o ol in Minit a t into the “E f iel d s. The t h e initial cho i a nge. a b 16 for the ffects” field t ool was run ice 2as and

PAGE 90

80 Figure 20: Power curve for the four significant factor effects Figure 20 shows the power curve of the experi ment using the new st andard deviation. The main effects all have close to 100% proba bility of being real effects, however the Sulfuric Acid Content*Temperature (AD) in teraction has only a 68% chance of being a real effect. Table 17 shows the Mi nitab output of the power data. Table 17: Minitab output of power analysis ------------------------------------Center Points Per Total Block Effect Reps Runs Power 1 -33.18 2 34 1.00000 1 6.50 2 34 0.99999 1 -13.93 2 34 1.00000 1 -2.49 2 34 0.68221 ------------------------------------5 0 -5 -10 -15 -20 -25 -30 1.0 0.8 0.6 0.4 0.2 0.0 EffectPower Alpha0.05 StDev2.71759 # Factors4 # Corner Pts16 # Blocks2 # Terms Omitted0 Center PointsYes BlocksYes Terms Included In Model Assumptions 2, 1 Ctr Pts Per Blk Reps,Power Curve for Significant EffectsA D B FactorName D B A Temperature Water Content Sulfuric Acid ContentAD

PAGE 91

81 What this means is that the sample si ze was not large enough to identify the AD interaction with the de sired power of 0.95 at the realized standard deviation of 2.71759. Two possible remedies for this conundrum exist: 1. Rerun the experiment with a larger sample size. 2. Accept the uncertainty of the interaction effect. The second option could only be used if the e ffect was considered to be insignificant by another means. If the effect were of intere st, it would be required to rerun the entire experiment with more replicates or better rep eatibility to reduce the standard deviation of the sample set. In fact, the experiment was salvageable because the interaction term is indeed insignificant from a practial standpoi nt as has been mentioned and as will be discussed in more detail in the next section. 4.1.2 PRACTICAL INTERPRETATION Statistical methods can be very useful for id entifying significant trends from voluminous sets of data. There is, however, often a shar p contrast between stat istical significance and practical significance. The ultimate question th at must be asked when a trend or effect is plucked from the sea of data by a soulless soft ware package is, “do we care?” With that in mind, this section endeavors to examine the results of the analysis with a practical eye and to build a framework upon which the data can be applied to real world processes.

PAGE 92

82 4.1.2.1 Limitations of the Design Without a consideration of the initial experi mental assumptions, an observer could draw false conclusions about the scope of the experi mental results. One must recall that the experimental data is limite d to the first 15 minutes of reaction time. Also, the conclusions only apply to the ranges bracketed by the treatment levels outlined in Figure 5. General conclusions about the data outside of these ranges are in potentia l error. Figure 21 shows a comparison between two differe nt reaction rate curv es at two different factor combinations. Additionally, it shows th e associated moisture increase for one of the two reactions. Figure 21 was attained by running an additional experiment involving two esterification reactions at two different factor combinations. One reaction was run with the highest levels of sulfuric acid and me thanol and the lowest level of water. The other reaction was run at the lo west levels of sulfuric acid and methanol and the highest level of water. Both reactions were run at the highest temper ature level. Please refer to Table 6 for the associated treatment levels. Free fatty acid percentages were measured at various intervals during each reaction until the reactions slowed appreciably. The moisture concentration for the first reacti on described above was also measured.

PAGE 93

83 Figure 21: Esterification reaction ra te for two factor combinations From the curves, it can be observed that the ra te of FFA reduction with time is not linear. An initial steep decline quickly changes slope until negligible reaction occurs with time. A 15 minute reaction period was chosen so that the total factorial experiment could be 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 10 20 30 40 50 60 70 80 90 100 0100200300Percent Moisture (%) Percent FFA (%)Time (min.)Esterification Reaction Rates 1,-1,1,1 -1,1,-1,1 1,-1,1,1 Moisture

PAGE 94

84 achieved within a reasonable timeframe, howe ver the assumption must be made that the relative effects are seen throughout the duration of the reaction. 4.1.2.2 Sulfuric Acid Content Clearly, sulfuric acid content was identified as having the most significant effect on the system response. This may lead a student of this thesis to conclude that the most advantageous thing to do would be to increase the sulfuric acid addition rate in order to accelerate the reaction and reduce the FFA con centration. This is true to a point; however said student should be aware of th e fact that the sulfuric acid must be neutralized in post-processing. After the re action has completed, agitation is stopped and the excess methanol rises to the top of the oi l phase carrying most of the sulfuric acid with it in solution. The methanol/sulfuric ac id solution is then decanted and neutralized prior to recovery of th e methanol for reuse. The neutra lization of the sulfuric acid with sodium hydroxide generates sodium sulf ate salts which can prove operationally problematic in sufficient quantity. If the qua ntity of salts exceeds the capacity for the remaining water to dissolve them, then they wi ll settle out in process vessels, foul heat exchangers, abrade pump seals, and clog piping. This suggests that care must be taken when choosing sulfuric acid addition ra tes in industrial applications. Additionally, it should be remembered that sulf uric acid was shown to have a significant effect within the region studied which was between 150 l and 500 l or 0.3% and 1.0% of the volume of oil used. There is likely a point beyond which adding additional sulfuric acid will result in no further reduction in FFA after a 15 minute reaction period. This

PAGE 95

85 point should be identified with a path of steepest ascent experiment or similar optimization experiment. 4.1.2.3 Temperature The analysis suggests that increasing temperat ure will accelerate the reaction rate within the 15 minute reaction period studie d. Two temperature levels were used for this factor: 48 C and 60 C. Because of the curvature term, it is expected that this relationship increases exponentially between the two levels. A physical limit exists, however, and that is the boiling point of me thanol or 64.7 C at atmosphe ric pressure. As the boiling point is approached, methanol evaporation will increase and then no further temperature increase will be possible as the methanol begins to change into the vapor phase. After the entire volume of methanol has vaporized, temperature increase will once again be possible, however it will be a pointless action since the primary reactant is no longer present. For several reasons this is an undesirable even t and the reaction temperature should be limited to a point below the boi ling point and below which evaporation becomes excessive. 4.1.2.4 Water Content The data suggest that water content has a ne gative impact on reaction rate. One must be aware that the “water content” term is so mewhat of a misnomer because the factor evaluated was simply an addition of a certain vo lume of water. Two levels were studied: 0 l and 500 l. This does not account for the water already present in the methanol

PAGE 96

86 (0.1339%) and the brown grease (0 .4956%). The experiment is measuring the effect of the additional water on the response. Within the studied range of wa ter addition (0% to 1% of th e volume of brown grease) the water was shown to inhibit the FFA reducti on. This was expected and supported by Liu et al. in the literature review (2006). No limit to this rela tionship is anticipated: in other words, reduction of the total system water co ntent down to zero percent will result in the best reaction rate as it relates to water cont ent and further increases in water content will result in further inhibition and poten tial stalling of the FFA conversion. Additionally, water is created as a byproduct of the esterification reaction of free fatty acids into methyl esters. This water gradually slows the reacti on as can be seen in Figure 21. From Figure 21, the associat ed moisture rise with FFA reduction is clearly visible. What this all means is that the operator s hould take great care to reduce initial moisture content when performing esterification r eactions. Obviously, water should never be intentionally added to the reactants! 4.1.2.5 Methanol Content As mentioned earlier, it came as a great surpri se that methanol content was shown to have no significant effect on the e xperimental response. Methanol Content was expected to have a large effect based on pract ical experience. It is likel y that this resulted from a level choice which fell above the point at which the path of steepest ascent is observed.

PAGE 97

87 The path of steepest ascent is used in optim ization experiments to determine the point at which further increase in a trea tment level results in no furt her response from the system. Practical experience shows that increased methanol content will result in increased reaction rates at methanol volumes below 100% of the volume of high FFA oils. From the experimental results, however, it is eviden t that methanol content has little effect within the region between 100% and 200% of the volume of oil, which was the region tested. An optimization study should be c onducted to determine the point at which additional methanol addition resu lts in no further FFA reduction. 4.1.2.6 Sulfuric Acid Temperature Interaction As mentioned, one interaction term was dete rmined by the regression analysis to be statistically significant: Sulfuric Acid*Tempe rature. The magnitude of this term was also shown to be problematic as describe d in section 4.1.1.9 since its power was lower than the 0.95 threshold below which signifi cance is in question. The answer to the previously pondered question, “do we care?” is simply, “no”. The reason for this is twofold: 1. The interaction term is a combination of sulfuric acid and temperature, two factors which would already be maximized within practical constraints based on the analytical conclusions. 2. The effect of the interaction term is ma rginal. As stated earlier in section 3.2.1.4, an effect of less than 3 would generall y be ignored unless th ere was an economic reason to consider it (e.g. borderline spec). The number is rather arbitrary and

PAGE 98

88 would be subject to the demands of the sp ecific situation; however it serves as a good rule of thumb for process planning. The absolute contribution of the interaction effect was 2.49 < 3 and theref ore not generally considered to be practically significant. Given the above evaluation, the hi gh uncertainty associated with the interaction term is of little concern from a pr actical vantage point. 4.2 ENHANCED REACTION KINETICS Figure 22 shows the main effects plots of th e enhanced reaction kinetics experiment. Figure 22: Main effects plot for the r eaction kinetics enhancement experiment Sonication Stirring 55 50 45 40 35 30 25 20 500 325 150 Agitation TypeMean Sulfuric Acid Corner Center Point TypeMain Effects Plot for FFAFitted Means

PAGE 99

89 From the above figure, the magnitude of the main effect between sonication and stirring is relatively minor. The magnitude of the effect between the high and low levels of sulfuric acid concentration is, once again, dramatic. From Figure 23, no evidence of interaction at either high or low leve ls of sulfuric acid content exists. Figure 23: Interaction plot for the re action kinetics enhancement experiment The lines for stirring and sonication and the related center points ne arly overlap which shows a largely identical effect between s onication and stirring at both the high and low levels of sulfuric acid concen tration. No discernable intera ction exists between the two main effects. 500 325 150 60 50 40 30 20 Sulfuric AcidMean StirringCorner StirringCenter SonicationCorner SonicationCenter Agitation TypePoint TypeInteraction Plot for FFAFitted Means

PAGE 100

90 The pareto chart of the standardized effects in Figure 24 shows the sulfuric acid factor effect to be the only significant effect based the effect magnitude threshold of 3.18 calculated from the P-values as described in section 4.1.1.6. Figure 24: Pareto chart for the reacti on kinetics enhancement experiment The power curve in Figure 25 shows that th e model is powerful enough to detect real effects when they exist with the sulfuric ac id effect having a 95% chance of being real. AB A B 30 25 20 15 10 5 0 TermStandardized Effect 3.18 AAgitation Type BSulfuric Acid FactorNamePareto Chart of the Standardized Effects(response is FFA, Alpha = 0.05)

PAGE 101

91 Figure 25: Power curve for the reaction kinetics enhancement experiment The primary conclusion to be drawn from the da ta is that, at a fixe d power (wattage) level of 33 watts of combined mechanical and ther mal energy input, no statistically significant difference exists between the agitation methods of sonication and stirring. This suggests that the observations seen in the literature re view are solely a func tion of the ability of sonicators to administer dramatically incr eased rates of mechanical energy into the reaction system. 5 0 -5 -10 -15 -20 -25 -30 1.0 0.8 0.6 0.4 0.2 0.0 EffectPower Alpha0.05 StDev1.33243 # Factors2 # Corner Pts4 # Blocksnone # Terms Omitted0 Center PointsYes Terms Included In Model Assumptions 2, 1 Ctr Pts Per Blk Reps,Power Curve for 2-Level Factorial Design

PAGE 102

92 Figure 26: Residual plots the reaction kinetics enhancement experiment The residual analysis is shown in Figure 25. While there appears to be some indication of outliers, and evidence of skewness, suggesti ng a less than perfect f it for the regression analysis, these disparities are likely the result of the extremely small number of experimental runs (10). 2 1 0 -1 -2 99 90 50 10 1 ResidualPercent 60 50 40 30 20 1.0 0.5 0.0 -0.5 -1.0 Fitted ValueResidual 1.5 1.0 0.5 0.0 -0.5 -1.0 3 2 1 0 ResidualFrequency 10 9 8 7 6 5 4 3 2 1 1.0 0.5 0.0 -0.5 -1.0 Observation OrderResidualNormal Probability PlotVersus Fits HistogramVersus OrderResidual Plots for FFA

PAGE 103

93 5.0 CONCLUSIONS A study of the esterification reaction in ultr a-high-FFA oils using methanol as the reacting alcohol and sulfuric aci d as the catalyst was presente d herein. It first sought to characterize the major reaction elements and de termine their relative effects on the final FFA% after 15 minutes of reacti on time. In addition, a comparison of a claimed reaction acceleration technique agitati on by ultrasonic irradiation – to conventional stirring was performed at constant energy input to evalua te the potential merits of sonication beyond its ability to input more en ergy into the reaction matrix. A literature review provided direction as well as suggested that additional work in the two areas of study herein was warranted. The experimental section and re sulting analysis conc luded that catalyst concentration (sulfuric acid) and reaction temp erature had the greatest effects on the final FFA content after 15 minutes of reaction time. This was supported by Dias et al. (2009). Water addition was shown to have a retardi ng effect on the final FFA content and was supported by the work of Liu et al. (2006). Methanol was shown to have no significant effect on the final FFA content after 15 minutes of reaction time which is inconsistent with some of the other findings (Santos, 2010) but supported by othe rs (Cardoso, 2008). The author concludes that the acceptance of the null hypothesis in this case is due to the relatively high volumetric ratio of methanol to oil. Effects would likely have been observed within a lower region. Finally, ul trasonic irradiation was shown to have no significant effect after 15 minut es of reaction time when th e stirring and s onication power levels were held constant. This clearly dem onstrates that the sole advantage of sonication

PAGE 104

94 as an agitation method for the esterification of ultra-high-FFA oils such as brown grease is its ability to introduce higher levels of mechanical energy into the reaction system.

PAGE 105

95 6.0 RECOMMENDATIONS Further areas for research include the optimizati on of all of the factor levels identified as being significant in the study. The author also believes that it woul d be beneficial to evaluate the region of oil to methanol volumetri c ratio from 2:1 to 1:1. It is clear that methanol has no effect in the region of ratios from 1:1 to 1:2, however based on some of the findings in the literature, there is indication that meth anol does have an impact on final FFA concentration in ultr a-high-FFA feedstocks such as brown grease and fatty acid distillates. Finally, additi onal study of the sonication react ion as a function of power level is warranted. A larger experiment at multiple levels for power level, methanol content, and sulfuric acid content should be conducted. Optimization of any findings should also be conducted.

PAGE 106

96 7.0 REFERENCES 1. Benitez, F.A. Effects of the use of ultrasonic waves on biodiesel production in alkaline transesterification of bleached tall ow and vegetable oils: Cavitation model. Dissertation. University of Pu erto Rico Mayaguez, 1-169 (2004). 2. Biodiesel Magazine. A Greasy Alternative. http://www.biodieselmagazine.com/article .jsp?article_id=2880, November (2008). 3. Bournay, L., et al. New heterogeneous pr ocess for biodiesel production: A way to improve the quality and value of the crude glycerin produced by biodiesel plants. Catalysis Today, 190-192 (2005). 4. Cardoso, A.L.; Neves, S.C.G.; da Silva, M.J. Esterification of Oleic Acid for Biodiesel Production Catalyzed by SnCl2: A Kinetic Investigation. Energies, 80-92 (2008). 5. Colucci, J.A.; Borrero, E.E.; Alape, F. Biodi esel from an alkaline transesterification reaction of soybean oil using ultrasonic mixing. Journal of the American Oil Chemist’s Society, 525-530 (2005). 6. Daily Times Herald. http://www.carrollspaper.com/main.asp?S ectionID=1&SubSectio nID=1&ArticleID= 9990, May 26 (2010). 7. Deshmane, V.G.; Gogate, P.R.; Pandit, A. B. Ultrasound-Assisted Synthesis of Biodiesel from Palm Fatty Acid Distill ate. Ind. Eng. Chem Res., 7923-7927 (2009). 8. Dias, J.M.; Alvim-Ferraz, M.C.M.; Almeida, M.F. Production of biodiesel from acid waste lard. Bioresource Technology, 6355-6361 (2009). 9. Fang, Y.; Wang, J.; Li, Y.; Ji, J. Study of new method for ultrasonic wave-assisted preparation of biodiesel oil. Huafei Gongye, 40-41, 44 (2005). 10. Georgogianni, K.G.; Kontominas, M.G.; Pomonis P.J.; Avlonitis, D.; Gergis, V. Conventional and in situ tran sesterification of sunflower seed oil for the production of biodiesel. Fuel Processi ng Technology, 503 -509 (2008). 11. Greer, D. Recycling Local Waste Oil and Grease into Biodiesel. BioCycle, 56-58 (2010).

PAGE 107

97 12. Hahn, H.D.; Dong N.T.; Okitsu, K.; Nishimura, R.; Maeda, Y. Biodiesel production by esterification of oleic aci d with short-chain alcohols under ultrasonic irradiation condition. Renewabl e Energy, 780-783 (2009). 13. Han, Hengwen. et al. Preparation of biodies el from soybean oil using supercritical methanol and CO2 as a co-solvent. Process Biochemistry, 3148-3151 (2005). 14. Iowa State University. What is Biodiesel. http://www.me.iastate.edu/biodiesel /Pages/bio6.html, May 3 (2006). 15. Journey to Forever. Is ethanol energy-efficient? http://journeytoforever.org/etha nol_energy.html, October 12 (2010). 16. Khan, A. K. Research Into Biodiesel Kinetics & Catalyst Development. Individual Inquiry. University of Queensland, 7 (2002). 17. Lee, S-B.; Lee, J-D. The effect of ultrason ic energy on esterification of vegetable oil. Kongop Hwahak, 532-535 (2009). 18. Liu, Y.; Lotero, E.; Goodwin J.G. Effect of water on sulfuric acid catalyzed esterification. Journal of Molecula r Catalysis A: Chemical, 132–140 (2006). 19. Lotero, E.; Liu Y.; Lopez D.E.; Suwannakarn K.; Bruce, D.A.; Goodwin, J.G. Synthesis of Biodiesel via Acid Cataly sis. Industrial & Engineering Chemistry Research, 5353–5363 (2005). 20. Mason, T.J. Sonochemistry: The Uses of Ultrasound in Chemistry, 48. The Royal Society of Chemistry (1990). 21. Mason, T. J. Sonochemistry 1, 2, 3. Oxford University Press (1999). 22. Minitab. Help. “pseudo-cen ter points”, October (2010) 23. Mittelbach, Martin; Remschmidt, Claudi a. Biodiesel: The Comprehensive Handbook 1,2,3. Martin Mittelbach (2004). 24. Montefrio, M.J.; Xinwen, T.; Obbard, J.P. Acid catalyzed synthesis of fatty acid methyl esters from waste greases. Applied Energy, 3155-3161 (2010). 25. Montgomery, D. C. Design and Analysis of Experiments 13, 35, 164, 247, 286. John Wiley & Sons, Inc. (2005). 26. National Biodiesel Board. News. http://nbb.grassroots.com/08 Releases/EnergyBalance/, February 6 (2008).

PAGE 108

98 27. Ngo, H.L.; Vanselous, H.; Lin, W. Acid cat alyzed synthesis of fatty acid methyl esters from waste greases. Amer ical Chemical So ciety, CATL-60 (2010). 28. Pahl, Greg. Biodiesel: Growing a Ne w Energy Economy 1,2,3. Chelsea Green Publishing Company (2005). 29. Peterson, C. L. Development of the Biodi esel Industry. Power Point Presentation. http://www.uidaho.edu/bae/biodiesel/ Development_of_the_Biodiesel_I ndustry_ASAE.ppt, May 3 (2006). 30. Santos, F.P.F.; Malveria, J.Q.; Cruz, M.G.A.; Fernandes, F.A.N. Production of biodiesel by ultrasound assi sted esterification of Oreochromis niloticus oil. Fuel, 275 – 279 (2010). 31. SciFinder Scholar. “biodies el,” October 12 (2010). 32. Stavarache, C. et al. Conve rsion of Vegetable Oils to Biodiesel Using Ultrasonic Irradiation. Chemistry letters, 716-717 (2003). 33. The Jacobsen. Renewable Fuels. ht tp://www.thejacobsen.com, October 13 (2010). 34. Tyson, S. K. Biodiesel Technology and Feedstocks. National Renewable Energy Laboratory. http://www.nrbp.or g/pdfs/pub32.pdf, June 19 (2002). 35. U.S. Energy Information Administrati on. Table 10.4 Biodiesel Overview. http://www.eia.doe.gov/emeu/mer/pdf /pages/sec10_8.pdf, October 12 (2010). 36. Van Gerpen, J. H. et al. Comparison of engine performance and emissions for petroleum diesel fuel, yellow grease biodiesel, and soybean oil biodiesel. Transactions of the ASAE, 937-944 (2003). 37. Wang, R. Development of biodiesel fuel. Article in Chinese. Taiyangneng Xuebao, 434-436 (1988). 38. Wikipedia. Methanol. http://en.wikip edia.org/wiki/Methanol, October 12 (2010). 39. Wordpress. Americas. http://www. worldpress.org/Americas/3146.cfm, May 13 (2008). 40. Zullaikah, S., Lai, C.C., Vali, S.R., Ju, Y.H. A two-step acid-catalyzed process for the production of biodiesel from rice bran oil. Bioresource Technology, 1889–1896 (2005).

PAGE 109

ABOUT THE AUTHOR Lucas Altic received his Bach elors in Mechanical Engine ering in the Spring of 2004. Shortly thereafter, he began his career in renewable fu els, specifically biodiesel development and production. He has worked as a process engineer, project manager, plant manager, and independent contractor at biodiesel manufacturing facilities in Florida, Georgia, and South Carolina. His areas of focus are process management and control. He was a speaker at the Siemen ’s Automation Summit, Orlando, FL on the topic of process automation in 2008 and also presen ted at the annual Florida Farm to Fuels summit in August 2009. Upon completion of th is Thesis, Lucas received his Masters degree in Mechanical Engineering from the Un iversity of South Florida. Lucas’ wife, Lara, recently gave birth to their new baby boy, Jackson Joseph Altic.