COMPARISON OF METHODS FOR THE PURIFICATION OF BIODIESEL A THESIS PRESENTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE WITH A MAJOR IN BIOLOGICAL AND AGRICULTURAL ENGINEERING IN COLLEGE OF GRADUATE STUDIES UNIVERSITY OF IDAHO BY JACOB WALL AUGUST 2009 MAJOR PROFESSOR: DR. JON VAN GERPEN ii AUTHORIZATION TO SUBMIT THESIS This thesis of Jacob Wall submitted for the degree of Master of Science with a major in Biological and Agricultural Engineering and titled “COMPARISON OF METHODS FOR THE PURIFICATION OF BIODIESEL” has been reviewed in final form. Permission, as indicated by the signatures and dates given below, is now granted to submit final copies to the College of Graduate Studies for approval. Major Professor ______________________ _________________ Dr. Jon Van Gerpen Date Committee Members ______________________ _________________ Dr. B. Brian He Date ______________________ _________________ Dr. W. Daniel Edwards Date Department Administrator ______________________ _________________ Dr. Jon Van Gerpen Date Discipline’s College Dean ______________________ _________________ Dr. Don Blackketter Date Final approval and acceptance by College of Graduate Studies _______________________ _________________ Dr. Margrit von Braun Date iii ABSTRACT Biodiesel is a proven alternative to petroleum diesel fuel. Once produced, biodiesel contains several impurities such as soap and glycerin. The free fatty acids in the oil react with the sodium or potassium catalyst to form soaps. After the biodiesel and by product glycerin are separated, trace amounts of glycerin remain in the biodiesel. Soap and glycerin impurities in the biodiesel can lead to engine wear and fuel storage problems. Traditionally, soap and glycerin are removed from the biodiesel by water washing. Water washing has several disadvantages such as producing large amounts of waste water that requires treatment and causing plant operational problems such as emulsion. Recently, several alternative “waterless” purification procedures have been used, such as ion exchange resins and Magnesol, a brand- named adsorbent. The objective of this study was to compare these two “waterless” purification methods by their mode of operation, effectiveness, life, and its ease of use. Experiments were developed in order to compare each purification method. Ion exchange resins were tested in packed bed columns and Magnesol was tested by mixing and filtering. During this research it was found that soap and glycerin can be removed by a combination of four modes. These modes are filtration, physical adsorption, ion exchange and soap removal by glycerin affinity. It was found that the four mode theory was supported by the experimental data. iv It was also found that ion exchange resins can reduce soap levels from 1200 ppm to below 50 ppm for about 550 BV of processed biodiesel. Ion exchange resin resins can reduce glycerin levels from 0.08% to below 0.02% for about 200 BV of processed biodiesel. Magnesol can reduce soap levels from 1000 ppm to below 50 ppm and glycerin levels from 0.08% to below 0.02% at a Magnesol concentration of 1%. v ACKNOWLEDGEMENTS I thank my major professor, Dr. Jon Van Gerpen for the opportunity to investigate a practical problem in the biodiesel production process. I would like to thank my committee members Dr. Brian He and Dr. W. Daniel Edwards for their guidance and support. I would like to thank Dr. Joseph C. Thompson for his ideas and support. I also want to mention his help with the development of equipment and procedure to make this project possible. I would like to thank my lab assistants Sweta Khanal and Sohana Khanal for their help preparing for and conducting the experiments. I would like to thank Jim Sabzali and Vivek Naik from the Thermax Company for their visits and technical and financial support. I, especially, want to thank my wife Christina Wall for her patience and support. vi Table of Contents Authorization to submit thesis………………………………………….ii Abstract…..………………………………………………………………….iii Acknowledgements…………………………………………….………….v List of figures……………………………………………………………...viii List of tables………………………………………………………………...x Abbreviations……………………………………………………………….xi Chapter 1. Introduction…...………………………………………...1 Chapter 2. Literature review……..…………………………………4 2.1 Water wash………………………………………………………………….5 2.2 Filtration……………………………………………………………………..7 2.3 Adsorbents…………………………………………………………………..8 2.4 Ion Exchange……………………………………………………………...12 Chapter 3. Theory...………………………………………………...14 3.1 Four Mode Theory………………………………………………………...14 3.2 Purification Processes……………………………………………………15 Chapter 4. Experimental methods and equipment………..….21 4.1 Packed Bed Purification………………………………………………….21 4.2 Adsorbent Purification…………………………………………………….22 4.3 Feedstock Preparation……………………………………………………22 4.4 Purification Media…………………………………………………………23 4.5 Measurement Techniques……………………………………………….24 4.6 Experimental Procedures………………………………………………...26 vii Chapter 5. Results and discussions……...…………………….26 5.1 Four Mode Theory Validation……………………………………………29 5.1.1 Filtration Mode…………………………………………………….29 5.1.2 Ion Exchange Mode………………………………………………32 5.2 Ion Exchange Resin Exhaustion………………………………………...34 5.2.1 Soap Only………………………………………………………….34 5.2.2 Glycerin Only………………………………………………………36 5.2.3 Soap and Glycerin Combined……………………………………38 5.2.4 Macroporous Resin Test…………………………………………38 5.2.5 Parallel Gel Resin Test…………………………………………...41 5.3 Ion Exchange Resin Operating Parameters……………………………44 5.4 Adsorbent Media Effectiveness………………………………………….47 5.5 Sulfur……………………………………………………………………….50 Chapter 6. Conclusions…..……………………………………….52 6.1 Conclusions………………………………………………………………..52 6.2 Recommendations for Future Work………………………………...…..54 References….……………………………………………………………...56 Appendices..………………………………………………………………58 A. Saw Dust Tests……………………………………………………………58 B. Soap and Glycerin Addition Calculations………………………………60 C. Autotitrator Procedure…………………………………………………….61 D. GC Procedure……………………………………………………………..62 E. Sample Titrator Calibration………………………………………………63 F. Sample GC Calibration…………………………………………………...64 G. Ion Exchange Resin Bed Start-up Procedure………………………….65 H. Methanol Wash Procedure………………………………………………66 viii LIST OF FIGURES Figure 1.1 Saponification reaction………………………………………………………..1 Figure 2.1 Diagram of ceramic membrane filtration for purification of biodiesel…….8 Figure 3.1 Physical adsorption mechanics……………………………………………..17 Figure 3.2 Ion exchange mechanics……………………………………………………18 Figure 3.3 Manufacture of a strong acid cation exchange resin……………………..19 Figure 3.4 Structure of a strong acid cation exchange resin…………………………20 Figure 4.1 Ion exchange test setup……………………………………………………..22 Figure 4.2 Adsorbent purification equipment…………………………………………..23 Figure 4.3 Autotitrator…………………………………………………………………….26 Figure 4.4 Gas Chromagraph……………………………………………………………26 Figure 5.1 Effect of filter porosity on soap removal……………………………………30 Figure 5.2 Methanol removal from biodiesel containing sodium and potassium………………………………………………………………...31 Figure 5.3 Effect of biodiesel soap level on acid value……………………………….33 Figure 5.4 Statistical analysis of experimental data…………………………………..34 Figure 5.5 Effect of resin age on soap removal………………………………………..36 Figure 5.6 Glycerin removal with no soap present…………………………………….37 Figure 5.7 Soap removal from biodiesel containing soap and glycerin……………..39 Figure 5.8 Comparison of gel resins for soap removal……………………………….41 Figure 5.9 Comparison of gel resins for glycerin removal……………………………43 Figure 5.10 Effect of flow rate and aspect ratio on soap removal…………………...46 Figure 5.11 Effect of flow rate on glycerin removal……………………………………46 ix Figure 5.12 Soap removal by adsorption and filtration………………………………..48 Figure 5.13 Glycerin removal by adsorption and filtration……………………………49 Figure 5.14 Reduction in acid value by processing with Magnesol………………….50 x LIST OF TABLES Table 2.1 Effect of impurities on biodiesel and engines………………………………..4 Table 2.2 Results of purification with a ceramic membrane…………………………...8 Table 2.3 Monoglyceride, bound glycerin content from two step reaction…………...9 Table 2.4 Effect of Magnesol purification on biodiesel properties…………………...10 Table 4.1 Comparison of ion exchange resins product data…………………………24 Table 5.1 Effect of purification media on biodiesel sulfur level………………………51 xi ABBERVIATIONS The following abbreviations were used throughout this thesis: FFA: Free fatty acid GC: Gas Chromatography PPM: Parts per million BV: Bed volume 1 CHAPTER 1. INTRODUCTION Biodiesel is a well known and proven alternative to petroleum diesel fuel. It can be used in most diesel engines with few modifications. Biodiesel has been shown to reduce engine exhaust emissions leading to less pollution of the environment. (McCormick et al., 1998) Biodiesel is typically produced using a transesterification reaction. Plant oils or animal fats are mixed with alcohol and a base catalyst, such as sodium hydroxide, and heated until a separation occurs between the biodiesel and the glycerin. The majority of the glycerin is then removed from the solution, by phase separation. After the transesterfication reaction has been completed and the glycerin has been separated and removed, the raw biodiesel still contains a small percentage of free glycerin and soap. Soap is the product of saponification reactions that occur in parallel to tranesterification. Figure 1.1 depicts the saponification reaction. Figure 1.1 Saponification reaction (Bruice, 2004). Soap is generally formed because of free fatty acids reacting with the catalyst. If the oil contains free fatty acids, these will be converted to soap by the 2 saponification reaction. The water contained in the reactants of transesterification causes an increase in free fatty acids by hydrolysis of the oil and therefore more soap will be created. (Ma et al., 1999) Traditionally, the soap and trace amounts of glycerin are removed by washing with water. This involves mixing water with the biodiesel, agitating them gently, and then allowing them to separate. The soap and glycerin are extracted into the water phase and removed when the water is separated from the solution. The use of water to wash the biodiesel feedstock causes many problems. • Water washing produces a large amount of waste water that must be treated. • Water washing introduces water to the fuel that can cause it to degrade. • After water washing, the fuel must be dried. This causes an increased energy cost. • Water washing with a high soap level biodiesel can lead to emulsions that can cause significant yield loss and other plant operational problems. • Water washing can require a great deal of time for drying, multiple washes and water-biodiesel separation. Prior to use, wash water must be deionized to remove metal ions, such as calcium and magnesium that could be transferred to the fuel causing the fuel to be out of specification for those compounds. (Bryan, 2008) As an alternative to water washing, several methods are available to remove soap and glycerin from biodiesel. These impurities can be removed by passing the fuel through a bed of ion exchange resin. Other methods for removing impurities 3 involve the use of adsorbent compounds such as magnesium silicate (Berrios et al., 2008; Kucek et al., 2007). The object of this research is to compare ion exchange resins to adsorbents for the purification of biodiesel. Each method will be compared by its mode of operation, effectiveness, life, and ease of use. The operating parameters and recommendations for use will also be presented. 4 CHAPTER 2. Literature Review Various purification methods have been proposed in the literature to purify biodiesel. Researchers have used water washing, filtration with a membrane, adsorbents and ion exchange resins to remove impurities such as soap and glycerin (Berrios et al., 2008; Predojevic et al., 2008; He et al., 2006; Wang et al., 2009; Kucek et al., 2007; Yoris et al., 2008). Silica gel, magnesium silicate and ion exchange resin are the common materials used in these studies. This chapter discusses the results of these studies in purifying biodiesel. Crude biodiesel contains impurities that can cause problems in its use and storage. The impurities contained in crude biodiesel include free fatty acids, water, methanol, glycerides, free glycerin, and metal compounds such as soap and catalyst. Table 2.1 shows how the impurities affect the biodiesel and engines. Table 2.1 Effect of impurities on biodiesel and engines (Berrios et al., 2008). Impurity Effect Free Fatty Acid (FFA) Corrosion, low oxidative stability Water Formation of FFA, corrosion, bacterial growth (filter blockage) Methanol Low density and viscosity, low flash point, corrosion Glycerides High viscosity, injector deposits, crystallization Metals(soap, catalyst) Injector deposits, high sulfated ash (filter blockage), abrasive engine deposits Free Glycerin Settling problems, Increases in aldehydes and acrolein exhaust emissions 5 2.1 Water Wash Washing the crude biodiesel with water is the traditional way to remove impurities. There are several variations to water washing, including washing with deionised water, washing with acidified water, and using a membrane to prevent emulsions with water washing. Studies by Van Gerpen et al. (2003, 2005) have shown that adding acid to the wash water aids in removing soaps. Also, multiple washes are required to achieve satisfactory soap removal. Some cases required as many as six individual washes (Van Gerpen et al., 2003). Berrios et al. (2008) tested water washing with deionised water and acidified water and found that acidified water performed better in removing soaps. However, glycerin removal required longer wash times with acidified water compared to deionised water. Their results showed that increased wash water temperatures improved soap removal but glycerin removal was not significantly improved by temperatures elevated above ambient. Predojevic et al. (2008) compared water washing with hot distilled water to a 5% phosphoric acid solution. In this study, the effect of wash method on the biodiesel density, kinematic viscosity, acid value, iodine number, water content, saponification number, cetane index and yield loss were evaluated. It was found that there was no significant change in density, kinematic viscosity, iodine number, water content, saponification number, cetane index between a distilled water wash and a phosphoric acid wash. It was found that the phosphoric acid wash decreased the acid value to a greater degree, and the phosphoric acid wash produced higher yields than the distilled water wash. 6 He et al. (2006) conducted experiments using several methods of water washing. These wash methods included distilled water, HCl acid wash, solvent extraction, and hollow fiber membrane extraction. This study compared yield loss for each method but did not give information about impurity removal. The distilled water wash involved adding a 1:1 volume ratio of water to biodiesel. The solution was agitated in a water bath oscillator for 20 min, allowed to phase separate, and then the water was removed. This process was repeated three times. The solution was then dried over sodium sulfate. It was found that the highest biodiesel yield was achieved at a water temperature of 50°C and an agitation rate of 125 rpm. The HCl wash was performed by adding a 1:1 volume ratio of HCl (pH =1) to biodiesel and agitated and separated similar to the distilled water wash. The biodiesel was then washed with distilled water two additional times and then the solution was dried over sodium sulfate. It was found that the addition of acid led to reduced biodiesel yield loss compared to a distilled water wash at the same temperature. The acid value increased due to the addition of acid. The solvent extraction was performed by adding petroleum ether to biodiesel at a 1:1 volume ratio. The solution was then water washed three times. Then, the solvent was removed and the solution was dried over sodium sulfate. Addition of solvent led to a yield loss due to emulsion formation at the biodiesel/water interface. Evaporation of the solvent also lead to yield loss. The hollow fiber membrane extraction involved using a 1 m long by 1 mm diameter hollow fiber membrane. Two membranes were tested, a hydrophilic 7 polysulfone membrane and a hydrophobic polyacrylonitrile membrane. Each membrane was placed in a beaker filled with distilled water and biodiesel was pumped through it. The purified biodiesel was then dried over sodium sulfate. It was found that when using membrane separation, the mass transfer occurred without phase mixing with each other. The hydrophilic polysulfone membrane provided the lowest yield lost of the study, 8.1% weight, whereas washing with 20°C distilled water provided the highest yield loss of 15.2% weight. (He et al., 2006) 2.2 Filtration Wang et al. (2009) found that soaps and glycerin could be removed by filtration using a membrane. This study involved passing raw biodiesel through a ceramic membrane after methanol removal. To test the effectiveness of the ceramic membrane, filtration pore sizes of 0.1 µm, 0.2 µm, and 0.6 µm were evaluated. The membrane was configured in a tube shape with an outside diameter of 26 mm and a length of 250 mm. The inside of the tube contained 19 three mm channels giving a total filtration area of 0.045 m 2 . Figure 2.1 shows the setup for filtration using the ceramic membrane tubes. To begin the test, methanol at 30 °C was passed through the membrane to provide initial cleaning. Biodiesel was then passed through the membrane at 60 °C. After 3 minutes, a sample was collected and analyzed for potassium, sodium, calcium, magnesium, and free glycerin. These results were compared for each membrane pore size used. The effects of temperature and transmembrane pressure were also tested at the 0.1 µm pore size. 8 Figure 2.1 Diagram of ceramic membrane filtration for purification of biodiesel (Wang et al., 2009). Table 2.2 shows the results comparing the three pore sizes and water washing for the crude biodiesel. From this table it can be seen that as the pore size decreases the effectiveness of removal of potassium and free glycerin were enhanced significantly. The potassium was in the form of soap and catalyst. Table 2.2 Results of purification with a ceramic membrane (Wang et al., 2009). 2.3 Adsorbents Solid adsorbents have shown success in the removal of impurities from biodiesel. Magnesium silicate and silica gel were the materials used. Kucek et al. Metals Content (mg/kg) Potassium Sodium Calcium Magnesium Free Glycerin (%wt) Crude Biodiesel 160 8.98 1.45 0.33 0.261 Pore: 0.6 µm 4.25 0.68 0.7 0.25 0.0276 Pore: 0.2 µm 2.2 0.88 0.55 0.26 0.0257 Pore: 0.1 µm 1.7 1.36 0.95 0.15 0.0152 Water Wash 2.46 1.41 0.64 0.18 0.0179 9 (2007) conducted a study on analyzing the use of the commercial magnesium silicate product, Magnesol. In this study biodiesel was produced in a two stage reaction. The first stage involved producing biodiesel with ethanol and KOH or NaOH. After the first stage, the solution was allowed to separate and the glycerin was drained off and the ethanol was removed. Next, 2% by weight of Magnesol was added to the biodiesel and stirred continuously for 20 min at 65°C. After the Magnesol was filtered from the biodiesel, a second reaction was performed and the biodiesel was water-washed. Using this procedure it was found that the use of Magnesol provided a significant reduction in monoglycerides and bound glycerin compared to a two stage reaction with a water wash between the stages. Use of Magnesol provided no significant yield loss. All other biodiesel impurities were below the Brazilian biodiesel standard. Table 2.3 shows the reduction in monoglyceride and bond glycerin obtained from this study. Table 2.3 Monoglyceride (MAG), bonded glycerin content from a two step reaction (Kucek et al., 2007). Berrios et al. (2008) evaluated the use of Magnesol for the purification of biodiesel. Biodiesel fuels from two different sources were compared. The Magnesol was mixed with the biodiesel fuels for 30 min and removed by filtration. The biodiesel fuels were sampled after 10, 20 and 30 min of agitation to determine MAG (wt%) Bonded Glycerin (wt%) NaOH 1.65 0.42 KOH 1.89 0.48 NaOH-Magneso 0.97 0.25 KOH-Magnesol 1.19 0.3 10 the effect of residence time on purification effectiveness. It was found that soap and glycerin were removed satisfactorily by the 10 min case. Four experiments were conducted using 0.25, 0.5, 0.75 and 1 % by weight of Magnesol in the biodiesel. Each of the samples was evaluated for methanol content, free glycerol, soap, acid value and oxidative stability. Table 2.4 shows the averaged performance of Magnesol purification on the properties of raw biodiesel. It can be seen that the free glycerin, soap, and methanol content are reduced significantly. Table 2.4 Effect of Magnesol purification on biodiesel properties (Berrios et al., 2008). Predojevic et al. (2008) evaluated the use of silica gel in a column with bed dimensions of 2 cm by 2 cm. To dry the biodiesel, a 1 cm layer of sodium sulfate was placed above the silica gel. This study analyzed the biodiesel density, kinematic viscosity, acid value, iodine number, water content, saponification number, and cetane index and yield loss. It was found that there was no significant change in density, kinematic viscosity, iodine number, water content, saponification number or cetane index due to purification by silica gel. This study compared the results from purification with silica gel to that of water washing. It was found that silica gel produced yields of about 94% compared to 90% from water washing with Raw Fuel Magnesol Treated Biodiesel 1 Biodiesel 2 Biodiesel 1 Biodiesel 2 EN 14214 Maximum Methyl Ester Content (%(m/m)) 95.3 90.1 Free Glycerol (%(m/m)) 0.11 0.1 0.03 0.05 0.02 Monoglyceride ((%(m/m)) 0.46 0.73 0.41 0.71 0.8 Diglyceride ((%(m/m)) 0.21 1.21 0.21 1.19 0.2 Triglyceride ((%(m/m)) 0.43 5.69 0.35 5.68 0.2 Methanol ((%(m/m)) 1.4 1.32 0.51 0.8 0.2 Soap (ppm) 909 1240 250 340 Water (mg/kg) 1050 1050 1050 1050 500 Oxidative Stabilty (h) 1.8 3.3 1.7 3.3 6 Acid Value (mg KOH/g sample) 0.18 0.24 0.11 0.2 0.5 11 distilled water. Silica gel reduced the acid value by greater than 90% compared to a 75% reduction by distilled water washing. Two studies were conducted by Yoris et al. (2007, 2008) to evaluate the effectiveness of silica gel for the removal of glycerin from biodiesel. In the first study, a test solution was prepared by adding glycerin to clean biodiesel with no methanol. A column of 0.25 inch inside diameter and 30 cm in length was loaded with silica gel. Biodiesel containing glycerin levels of 0.1 - 0.2% was passed through the bed at a space velocity of 3-11 cm 3 /min. Tests were conducted to regenerate the bed by washing with methanol. For this test, a breakthrough curve was generated. From this curve it was found that the maximum adsorption capacity was 0.139 grams glycerin per grams silica gel. The breakthrough time, which is the time needed to reach the ASTM standard for glycerin, was found to exceed 1 hour at a space velocity of 14 cm 3 /min. It was also found that the glycerin adsorption capacity could be restored by flushing the bed with 4 bed volumes of methanol. In the second study, test solutions were created by adding glycerin to clean biodiesel and also solutions prepared containing water, methanol, monoglycerides and soaps in addition to glycerin. Sodium stearate and Glyceryl monosterate were used to simulate soap and monoglycerides. After the test solution was prepared, 3 grams of silica gel were added to the solution. Stirring was performed at 25°C for 2 hours. The silica gel was then allowed to settle and the liquid phase was sampled for analysis. 12 The samples were analyzed to determine the effect of soaps, water, methanol and monoglycerides on the adsorption of glycerin. It was found that soaps and water did not affect the adsorption of glycerin. Methanol and monoglycerides were shown to reduce the adsorption capacity of the silica gel for glycerin. 2.4 Ion Exchange Berrios et al. (2008) tested the use of ion exchange resins on the purification of biodiesel. The ion exchange resins were tested their ability effect methanol, glycerin, soap, acid value and oxidative stability parameters of the biodiesel. Two raw biodiesel fuels were used for testing, as was shown in Table 2.4. Berrios et al. (2008) evaluated Ion exchange resins from Purolite and Rohm and Haas to determine their effectiveness. Two ion exchange columns were tested in parallel. The 3 cm diameter resin columns were loaded with 80 grams of dry resin and flow rates of 0.25 L/h or 2.5 bed volumes per hour were used. From this test it was found that ion exchange resin has little effect on methanol concentration after the bed has reached equilibrium. Purification with the ion exchange resin caused the acid value of the purified biodiesel to increase slightly. Oxidative stability did not seem to be affected. Soap and glycerin were reduced significantly. The two resins were found to perform similarly and the purification capacity was found to be 500 L biodiesel /kg resin for the biodiesel #1 and 720 L biodiesel/kg resin for the biodiesel #2. In table 2.4, it can be seen that biodiesel fuels #1 and # 2 had about the same glycerin content but biodiesel fuel #2 13 had about 300 ppm more soap. This difference in initial soap level is the reason why the resin had longer life processing biodiesel fuel #1. 14 CHAPTER 3. Theory The production of biodiesel produces byproducts that can have a negative impact on the biodiesel’s end use; these byproducts are mainly soap and free glycerin. This chapter discusses theory behind the removal of these impurities. 3.1 Four Mode Theory In the course of the research conducted for this project, the authors proposed that the removal of trace amounts of soap and free glycerin from biodiesel involves four different operating modes. The various media and methods of purification take advantage of at least one of these modes. The four modes of purification are: 1. Filtration: Filtration is the removal of impurities that are insoluble in biodiesel by a mechanical action. This mechanical action can take the form of surface filtration or depth filtration. In most cases, both types of filtration are involved. Filtration can be used to remove both insoluble soap and glycerin. Filtration is most effective when biodiesel methanol levels are low. 2. Adsorption: Adsorption is the removal of soluble impurities by a chemical action. The pores on the adsorbent are charged and attract the impurities. The impurities become chemically bonded to the charged surface of the adsorbent. Adsorption can be used to remove both soluble soap and glycerin. The soap and glycerin adsorption capacity of ion exchange resins can be regenerated using a methanol wash. 3. Ion Exchange: Ion exchange is the removal of impurities by exchanging an ion from the ion exchange material, for the metal portion of the impurity 15 contained in the raw solution. Ion exchange is the chemical breakdown of the impurity. Ion exchange can be used to remove soap from biodiesel by exchanging the hydrogen ions from the ion exchange material for the sodium or potassium ion on the soap molecule. After the exchange takes place, the free fatty acid portion of the soap molecule is allowed to pass through the resin bed. 4. Glycerin/Soap Interaction: Soap has a stronger affinity for the glycerin portion of the biodiesel, therefore glycerin can aid in the removal of soap. As glycerin becomes adsorbed on the surface of the purification media, soap is entrapped in the glycerin layer and removed from the biodiesel stream. These four modes of biodiesel purification encompass the mechanisms and interactions that allow impurities to be removed from biodiesel. In the technical literature there is no consensus about the mechanisms that are active in biodiesel purification. This theory clarifies the way impurities are removed from biodiesel. 3.2 Purification Processes Purification of biodiesel involves a series of separation processes that may occur in combination. These separation processes include filtration, physical adsorption, and ion exchange adsorption. Filtration is a mechanical process where insoluble particles are removed from a fluid-particle suspension. The filtration media is placed in a column and the fluid- particle suspension is passed through the column and the particles become attached to the external surface of the of the filtration media, removing them from 16 the fluid. In the crude biodiesel solution, insoluble soap and glycerin particles can be filtered. There are two main types of filtration, surface filtration and depth filtration. Surface filtration occurs on the surface of the filtration media layer. Depth filtration occurs when multiple layers of filtration media are available and provides a greater surface area for particle removal. Physical adsorption is a process where molecules in the fluid bond onto the internal surface of the adsorbent media and are removed. Adsorption differs from absorption in that adsorption is the accumulation of solute molecules on the surface of the adsorbent media. Absorption is the accumulation of solute molecules in the bulk volume of the absorbent media. Adsorption uses interactions similar to van der Waal bonds, to attract and bind solute molecules to the surface of the adsorbent. Van der Waal bonds are formed by a series of repulsive and attractive forces that reach a point of equilibrium when adsorption occurs. Adsorption bonding occurs when there is a concentration gradient between the fluid and the adsorbent. This bonding is weak and can be reversed by washing the adsorbent with a clean solvent (Cooney, 1998). Figure 3.1 depicts the mechanics of physical adsorption. In this figure it can be seen that an adsorbent particle is surrounded by a fluid film layer. The solute molecules must first diffuse into the fluid film layer and then transfer onto the adsorbent particle surface. The diffusion from the fluid is driven by a difference in concentration. If the fluid has a relatively high concentration of solute, diffusion will occur toward the particle and adsorption will occur. If the concentration of solute is 17 high on the particle, then desorption will occur. Desorption is the process used to regenerate adsorbent materials. Figure 3.1 Physical adsorption mechanics (Cooney, 1998). Ion exchanging is a special type of separation process where ions are adsorbed from the fluid surrounding the ion exchange material. This adsorption releases ions of the same charge from the ion exchange material into the fluid. Figure 3.2 depicts the mechanics of an ion exchange resin. In figure 3.2 it can be seen that the basic adsorption mechanics for ion exchange are similar to the physical adsorption depicted in figure 3.1. A diffusion film layer surrounds the ion exchange particle which is suspended in a homogeneous fluid. In the ion exchange adsorption, sodium ions diffuse from the fluid flow into the film layer and then onto the ion exchange particle. The sodium 18 ions are exchanged with the hydrogen ions from the ion exchange particle. The hydrogen ion diffuses into the film layer and then into the fluid flow. (Arden 1968) Figure 3.2 Ion exchange mechanics (Arden, 1968). There are four categories of ion exchange resins: Strong and weak acid cation exchange resins and strong and weak base anion exchange resins. Strong acid cation exchange resins are the type used to purify biodiesel. Strong acid cation exchange resins are manufactured by the sulphonation of styrene-divinyl benzene copolymers. The process is performed by agitating a suspension of styrene, divinyl benzene and a stabilizer such as polyvinyl alcohol. The rate of agitation determines the size of the resin beads that polymerize. The resin beads are then sulphonated using sulfuric acid. Figure 3.3 shows the process of strong acid cation exchange resin manufacture (Osborn, 1956). 19 Figure 3.3 Manufacture of strong acid cation exchange resins (Osborn, 1956). Ion exchange resins have several important properties that affect their performance. Ion exchange resins swell or shrink with adsorption or desorption of solvents. The internal cross-linking of the resin’s structure controls the extent of swelling. Cross-linking improves the resin’s mechanical properties but reduces its ion exchanging rate and capacity (Inczedy, 1966). Figure 3.4 shows the structure of a strong acid cation exchange resin. Manufacturers primarily make two types of ion exchange resins for purification of biodiesel, “gel” and macroporous resins. Gel resins are the most common and are made by Purolite, Dow, Thermax, and Rohm and Haas. Gel resins have low cross-linking and have a translucent appearance. Macroporous resins have higher cross-linking and are opaque in appearance. Thermax currently is marketing a macroporous resin specifically for purification of biodiesel. 20 Figure 3.4 Structure of a strong acid cation exchange resin. -----, Polystyrene chain; - ----, divinyl-benzene cross-link. (Albright and Yarnell, 1987). 21 Chapter 4. Experimental Methods and Equipment Two main approaches were used to compare media for the purification of biodiesel. The first involves passing raw biodiesel through a packed bed of purification media. The second involves mixing a purification media with the biodiesel for a period of time and then pumping it through a filter to separate the material from the biodiesel. Ion exchange resins were tested using the packed bed method. Adsorbents were tested by the mix and filter method. 4.1 Packed Bed Purification The equipment used to evaluate the packed bed method consisted of glass columns ranging from 1 to 3 inches in diameter. In the bottom of the column was an 80 mesh screen that contained the purification media but allowed biodiesel to pass through. At the top of the column was an inlet and overflow outlet. The fuel level was maintained by a float switch controller and a peristaltic pump to provide a constant head over the bed. Flow rate was controlled by a needle valve and measured by a rotameter. The cleaned biodiesel was then collected in a separation funnel. A Sigma 900 Max automatic water sampler was used to pump the cleaned biodiesel from the separation funnel into sample bottles at a set time interval. Figure 4.1 shows the equipment used in the packed bed purification experimentation. 22 Figure 4.1 Ion exchange test setup. 4.2 Adsorbent Purification Adsorbents, such as magnesium silicate, were tested by first removing the methanol from the crude biodiesel. Two percent by weight of adsorbent was mixed into the crude biodiesel. The biodiesel adsorbent mixture was then agitated for 10 min at a temperature of 60°C and then poured into a Büchner funnel containing a #4 Whatman filter. Vacuum was applied to suction the biodiesel through the filter. The equipment used for this process is depicted in figure 4.2. 4.3 Feedstock Preparation The biodiesel used in this experiment was prepared from crude mustard seed oil and canola oil crushed at the University of Idaho pilot plant. Biodiesel was prepared in a large batch using sodium hydroxide and methanol. The biodiesel was then water washed to remove impurities. This was used as the test fuel. 23 For each test, the clean biodiesel was treated with glycerin, sodium methoxide and methanol to produce the desired level of soap and glycerin. The test fuel was prepared in 5 and 10 gallon batches and was typically used within 1 to 2 days. Appendix B shows the soap and glycerin addition calculation. Figure 4.2 Adsorbent purification equipment. 4.4 Purification Media The adsorbent media used in this study was Magnesol from the Dallas Group. The ion exchange resins used in this study were T45BD and T45BDMP from the Thermax Company and BD10Dry from Rohm and Haas. Table 4.1 shows the product data from several ion exchange resin manufacturers. 24 Table 4.1 Comparison of ion exchange resin product data. 4.5 Measurement Techniques In order to determine the effectiveness of purification, measurements were performed on the soap level, free fatty acid value, and free glycerin percentage before and after the processing. Soap level was determined using AOCS method Cc 17-79. One hundred ml of acetone containing 2% deionised water was used as a solvent for each sample. To provide a blank, each batch of acetone solvent and bromophenol blue indicator was titrated with NaOH or HCl to the point of a slight color change to faint yellow prior to adding the biodiesel. Then, after adding a measured quantity of fuel, the mixture was titrated with 0.001 N HCl to the blue-to- yellow endpoint of the bromophenol blue. Soap level was calculated using the following equation: Soop (ppm) os soJium olcotc = ECI (ml) × Normolity (ECI) × Su4,4uu Somplc Hoss (g) Later in the test program, an automated method was developed to determine the soap level using a Metrohm 848 Titrino Plus Autotitrator (Herisau, Switzerland). The autotitrator uses an electrode to measure the pH of the solution, and then it adds small increments of 0.01 or 0.001 normality HCl. As the solution Ion Exchange Resin Product Data T45BD T45BDMP PD206 DR-G8 BD10 Manufacturer Thermax Thermax Purolite Dowex Rohm & Haas Type Gel Macroporous Gel Gel Gel Matrix Structure polystyrene polystyrene polystyrene polystyrene polystyrene Functional Groups Sulfonic acid Sulfonic acid Sulfonic acid Sulfonic acid Sulfonic acid Ionic Form Hydrogen Hydrogen Hydrogen Hydrogen Hydrogen Particle size 0.4-1.2 mm 0.4-1.2 mm Not available 0.3-1.2 mm Not available Density (a) 80 g/ml 40 g/ml 80 g/ml 80 g/ml 80 g/ml 25 becomes neutralized, a titration curve is generated. The titrator then analyzes the second derivative of the titration curve to determine the equivalence point, which is the point where all the soap is neutralized. The point on the titration curve that has the largest second derivative is considered to be the equivalence point. Acid value was measured using the ASTM D974 method. A mixture of 50% toluene, 0.5% water and 49.5% isopropyl alcohol was used as a solvent for each sample. Phenolphthalein was added to a sample of solvent and biodiesel and then titration was performed with 0.1 KOH in iso-propanol to the clear-pink endpoint of the phenolphthalein. An automated method was developed to determine the acid value using the Metrohm autotitrator. Acid value was calculated using the following formula: Iotol AciJ Numbcr ( mg K0E g ) = K0E (ml) × Normolity (K0E) × S6.1 Somplc Hoss (g) Free glycerin was measured using gas chromatography with an Agilent 6890N gas chromatograph, according to the ASTM D6584 GC method. 0.1 gram of biodiesel was mixed with 100 µL each of Butanetriol, Glyceryl-tridecanoate, and N- Methyl-N-trifluoroacetamide (MSTFA) and the solution was allowed to set for 20 min. 8 mL of heptane was then added as a solvent to the solution. 1 µL of this solution was then injected to perform the GC analysis. 26 Figure 4.4 Gas chromatograph 4.6 Experimental Procedures In order to evaluate the effectiveness of packed bed media such as ion exchange resins, several tests were developed. Resin life tests were developed to determine how the resin performs as it becomes exhausted. Dry resin was weighed and poured into the column. Clean biodiesel or methanol was pumped into the outlet of the column to classify the resin bed. Classification of the resin bed caused the resin to be distributed with larger resin beads on the bottom and smaller beads on the top. Air pockets are also removed by classification. After classification was completed, the biodiesel was drained until it was level with the top of the resin bed. Raw biodiesel was then pumped into the column. 1000 ml samples were taken at 1 to 2.5 hour intervals for flow rates of 2 to 4 bed Figure 4.3 Autotitrator 27 volumes per hour. Samples were analyzed to determine the effectiveness of purification at intervals of 5 to 10 liters. To determine the effectiveness of purification, the soap level, acid value and free glycerin level were measured before and after passing through the bed. When free glycerin levels leaving the resin bed exceeded the ASTM specification level of 0.02%, the resin bed was washed with clean methanol to regenerate the resin bed. Methanol washing involved passing 5 to 10 bed volumes of methanol through the resin bed to remove the glycerin bonded to the resin. This washing was performed at flow rates ranging from 2 to 4 bed volumes per hour. Soap levels were monitored to determine when 50 ppm soap in the product was reached. Soap levels above 50 ppm cause the biodiesel ASTM specification for sulfated ash (0.02% weight) to be exceeded. The column was run until soap levels were well above the specification levels. Acid values were also measured to determine the amount of soap removal due to ion exchange in the resin. The effect of flow rate and bed height was measured by varying the bed height from 1 to 3 inches in several columns. For each column, the flow rate was varied from 1 to 7.5 bed volumes per hour. Samples were taken at each flow rate and soap levels were determined. The effect of processing biodiesel with ion exchange resin on the acid value was determined by passing biodiesel through the resin at several soap levels. For each soap level, the acid value of the outgoing fuel was measured. To determine the effectiveness of purification of biodiesel with adsorbents such as magnesium silicate or diatomaceous earth, the mass of the adsorbent was 28 varied, and for each treatment level, the soap and glycerin levels were measured. The experimental setup was similar to that used in ion exchange resin experiments. 29 CHAPTER 5. Results and Discussions This chapter discusses the experimental results of tests conducted to validate the four mode theory of soap and glycerin removal. It also describes the operating parameters needed to achieve optimal performance from biodiesel purification products. The performance of adsorbents, filtration media, and ion exchange resins for their ability to purify biodiesel are compared. 5.1 Four Mode Theory Validation Experiments were conducted to validate the four mode theory of soap and glycerin removal from biodiesel which was discussed in chapter 3. The four mode theory states that soap and glycerin can be removed by a combination of filtration, physical adsorption, ion exchange and purification by the interaction between soap and glycerin. 5.1.1 Filtration Mode The first mode that was tested was soap removal by filtration. Figure 5.1 shows the effect of filtration on soap removal. In this experiment, the particle retention size of the filters was varied from 2 to 40 µm and the soap levels after filtration were found to vary from 600 to 1200 ppm. In this initial test, the biodiesel feedstock contained no methanol or glycerin and was at room temperature. The sodium-catalyzed biodiesel feedstock contained 1500 ppm of soap and the potassium-catalyzed biodiesel feedstock contained 1300 ppm. incre the f can Figu part This cont keep prec The resu eases, the filterability o be filtered ure 5.1 Effec A visual iculate in b s experimen taining sod p soap in so cipitation. F ults of this e amount of of sodium a more effec ct of filter p analysis w iodiesel sa nt involved ium and po olution with Figure 5.2 s experiment soap in the and potassi ctively than porosity on s was also per mples whic the vaporiz otassium so h the biodie shows the e Coffee Whatman #4 Whatman #1 Whatman #2 Whatman #42 t show that e effluent als ium soaps. potassium soap remov rformed to ch containe zation of me oaps. Since sel, methan experimenta Filter Proper Particle Rete 40 22 11 8 2 2.5 as the filte so increase It can be s soap. val. compare th d sodium a ethanol from e methanol nol remova al setup for rties ention (μm) Typ 0 2 Qu 1 Qu 8 Qu 5 As r particle re es. This fig seen that s he amount o and potassiu m two biodi acts as a c al should ca r this analys pe ualitative ualitative ualitative shless etention size gure compa odium soap of soap um soaps. iesel sampl cosolvent to ause soap sis. 30 e ares p les o 31 The visual analysis and filtration experiments show that there are distinct differences in the physical characteristics of sodium and potassium soaps. When methanol was removed it was found that sodium soaps formed a “gel-like” particulate layer in the biodiesel. Potassium soaps did not appear to form this particulate layer. It was found that during methanol removal, the biodiesel samples with sodium soaps generated foaming, whereas, the potassium soaps did not. This foam is likely formed by methanol vapors bubbling through the sodium soap particulate layer. From the visual observations of the differences between sodium and potassium soaps in biodiesel and the soap filtration data, in figure 5.1, it appears that sodium soap is less soluble in biodiesel than potassium soap. Figure 5.2 Methanol removal biodiesel containing sodium and potassium soap (Left: Potassium, Right: Sodium). 32 5.1.2 Ion Exchange Mode Experiments were conducted to determine the quantity of soap that is removed to due to the ion exchange mode. An increase in free fatty acid (FFA) levels when soap is removed indicates that the sodium portion of the soap molecule is being held by the resin, releasing the FFA into the effluent biodiesel. Figure 5.3 shows the effect of soap removal, by ion exchange, on the effluent biodiesel free fatty acid levels. The incoming soap level was varied from roughly 500 ppm to 3500 ppm. Free fatty acid measurements were taken on the effluent biodiesel at each soap level. From this figure it can be seen that as the crude biodiesel soap level is increased, the effluent biodiesel free fatty acid value also increases. Three ion exchange resins, Amberlyte BD10 by Rohm & Haas and T45BD and T45BDMP by Thermax, were tested. In this experiment, the experimental data were compared to the theoretical 100% ion exchange curve. The 100% ion exchange curve is the case where all of the soap entering the column is converted to FFA. For each ion exchange resin, complete soap removal was achieved but 100% conversion from soap to free fatty acid was not achieved at soap levels greater than 1000 ppm. At soap levels below 1000 ppm the experimental data shows close to 100% ion exchange. At soap levels from 1500 to 3500 ppm, the experimental curve and the 100% ion exchange curve diverge. These data show that at lower soap levels, soap removal is performed primarily by ion exchange. At higher soap levels, soap removal is performed by a combination of ion exchange, filtration and adsorption. Figu exch calc conf ana belo ure 5.3 Effec To ensu hange curv culated. Fig fidence inte lysis, it can ow the 100% ct of biodies ure that the e was statis gure 5.4 sho ervals fall b be assume % ion excha sel soap lev difference stically sign ows that th elow the 10 ed, with 90 ange curve vel on acid between th nificant, a 9 e lines corr 00% ion exc % confiden e. value. he experime 90% confide responding change cur nce that the ental and 1 ence interva to the uppe rve. From t e experimen 00% ion al was er and lowe this statistic ntal data fal 33 er cal lls Figu 5.2 optim Res only tests 5.2. resin was ure 5.4: Stat Ion Exc Tests we mal operati sin exhausti y, glycerin o s were com 1 Soap O Figure 5 n, with no g s used to pu tistical anal change Res ere conduc ng parame on tests we only and a c mpared to th nly 5.5 shows th glycerin pre urify biodies ysis of exp sin Exhaus cted to dete eters and th ere conduc combination he performa he resin so esent. In th sel containi erimental d stion ermine the i e effect of t ted for biod n of soap a ance of othe ap exhaust is test, a co ng 2000 pp data. on exchang the four mo diesel feeds nd glycerin er competit tion curve f olumn with pm of soap, ge resin’s e odes over th stock conta n. The resu tive product for Thermax 200 ml of s , 5% metha effective life he resin life ining soap ults from the ts. x’s T45BD swelled resi anol and no 34 e, the e. ese gel in o 35 glycerin. This test was conducted using a flow rate of 4 bed volumes per hour. Biodiesel was passed through the column until about 500 bed volumes (BV) were processed. The effluent soap level, after 500 BVs, was about 900 ppm. The effluent biodiesel soap level exceeded 50 ppm of soap after about 250 BVs. The small decrease in soap level at 100 BVs and 350 BVs and the acid value at 100 BVs was due to a fluctuation in flow rate and does not affect the overall results of this experiment. In figure 5.5, the effluent free fatty acid was also measured for samples up to about 500 BV. At 50 BVs, the FFA was measured to be about 2100 ppm and at 500 BVs the FFA measured 600 ppm. The soap and acid curves in figure 5.5 were compared to determine the effect of the four mode theory over the life of the resin. There were two distinct regions where the mode of soap removal changed. These regions were from 0 to 225 BVs and 225 to 500 BVs. It can be seen that the soap level is low from 0 to 225 BVs but the FFA values decrease, significantly, over this range of bed volumes, this trend shows that as the resin bed ages ion exchange gradually decreases and adsorption contributes more to soap removal. This shift in purification mode is likely due to the ion exchange sites being covered with soap particulate. From 225 to 500 bed volumes the soap level begins to rise in the effluent biodiesel and the FFA continues to drop, this indicates that ion exchange, adsorption and filtration capacity are being exhausted, allowing soap to pass with the effluent biodiesel. Figu of th exch sites filtra adso lead 5.2.2 glyc ure 5.5 Effec To summ he resin bed hange to ad s being cov ation and ad orptive pore ding to com 2 Glycerin Figure 5 cerin remov ct of resin a marize figu d. As the re dsorption. T vered with s dsorption g es and bed plete bed e n Only 5.6 shows th al with no s age on soap re 5.5, the esin bed ag The ion exc soap particu radually be filtration ar exhaustion. he resin ex soap presen p removal (R soap remov ges, soap re change gra ulate. In th ecome exha rea are fille xhaustion cu nt. In this t Resin: T45B val occurs emoval mo adually decr e second p austed. As ed, more so urve of the test, biodies BD). in two phas ode shifts fro reases due period, ion e the ion exc ap is allowe T45BD gel sel containi ses over the om ion to the activ exchange, change site ed to pass, resin for ing 0.08% f 36 e life ve es, free glyc flow bed rege the Figu grad conv rem mea cerin, 4% m w rate for thi volumes a eneration ca life of the re ure 5.6 Glyc From fig dually, at a versations w oved by ad asurements ethanol and s test was nd then the apacity. Th esin. cerin remov gure 5.6, it c relatively c with the Th dsorption in s that did no d no soap w about 3.5 B e resin bed he effluent al with no s can be see constant rat hermax Com the ion exc ot increase. was purified BV/hour. B was washe biodiesel fr soap presen n that the e e, as the re mpany, it is change res . d with 200 Biodiesel wa ed with met ree glycerin nt (Resin: T effluent glyc esin bed ag generally u in; this is su ml of swelle as purified f thanol to te n level was T45BD). cerin level i ges. Accord understood upported by ed resin. T for about 31 st its measured o ncreases ding to that glycer y free fatty 37 The 16 over rin is acid 38 During this experiment the ability to regenerate the ion exchange resin for glycerin removal was tested. Figure 5.6 shows that washing the ion exchange resin with methanol at 316 bed volumes, after glycerin exhaustion, can reduce effluent glycerin levels. In this figure, glycerin levels were reduced from 0.08% to 0.02% by washing with methanol. 5.2.3 Soap and Glycerin Combined In normal production, crude biodiesel contains both soap and glycerin. Tests were conducted to determine the resin life and active modes for biodiesel containing soap and glycerin. These tests compared gel and macroporous resins, from different manufacturers, to determine if resins from different manufacturers perform differently. 5.2.4 Macroporous Resin Test Figure 5.7 shows the resin exhaustion curve for soap removal and the corresponding free fatty acid value, for Thermax’s T45BDMP macroporous ion exchange resin. For this test, crude biodiesel containing 1200 ppm of soap, 0.08% glycerin and 4% methanol was purified by 180 ml of swelled resin, at a flow rate of 2.5 BV/hour. This test was conducted until about 1000 bed volumes had passed through the resin at which time the incoming and outgoing soap levels were approximately equal. At 50 BV, the soap and FFA values were 27 ppm and 1800 ppm, respectively. At 1000 BV, the soap and FFA values were 1070 ppm and 1260 ppm, respectively. A soap level of 50 ppm was exceeded at about 550 BV. Figu in fig To c The 5.8, 50 p the volu purif ure 5.7: Soa Compar gure 5.5, it compare the crude biod a soap lev ppm soap le 1200 ppm c umes neede The rela fication of c ap removal f ring the soa can be see ese curves diesel in figu el of 1200 p evel is reac case. Basi ed to achiev ationship be crude biodie from biodie ap exhausti en that simi the initial s ure 5.5 con ppm was u ched at 250 cally, if the ve a given s etween the esel contain esel contain on curve in lar soap re soap level m ntained 200 sed. From BV for the soap level soap level a soap remo ning soap a ning soap an n figure 5.7 moval char must be tak 0 ppm of so these figur 2000 ppm is doubled appears to oval and free and glycerin nd glycerin to the 0% g racteristics ken into con oap, where res, it can b case and a the numbe be halved. e fatty acid n follows ro (T45BDMP glycerin cur were achie nsideration. eas, in figure be seen tha at 550 BV f er of bed values for oughly the s 39 P). rve eved. . e at the for same 40 pattern as the 0% glycerin case. From figure 5.7, it can be seen that most of the soap is removed but the acid value gradually decreases as the influent increases from 0 to about 500 BV. This region corresponds to the shift from ion exchange to adsorption that occurs as the resin ages. From 550 to 1000 BV, the FFA value continues to drop and the soap level climbs rapidly. This region is the phase where ion exchange, adsorption and filtration are all becoming exhausted and soap is allowed to pass into the effluent. The fluctuations in FFA value occurring at 50, 580 and 650 BV volumes are due to a decrease in flow rate leading to more efficient ion exchange causing an increase in FFA. An important difference between the ion exchange resin exhaustion curves depicted in figures 5.5 (soap only) and 5.7 (soap and glycerin) is the slope and final value of the free fatty acid curve. In figure 5.5, the FFA curve has a much steeper slope and the final FFA value is around 500 ppm. In figure 5.7, the FFA curve has a smaller slope and the final FFA value is around 1300 ppm. These differences indicate that when glycerin is present, the ion exchange sites may become blocked off during the purification process reducing the ion exchange action. The glycerin removal tests were also performed with the macroporous resin. Glycerin removal was not found to be effective compared to the gel resin tests. Macroporous resins are used in industry for biodiesel purification and are claimed to be effective. Additional experiments should be conducted to whether verify glycerin removal occurs with macroporous resins. 5.2.5 BD1 cont flow resin glyc Figu to a 50 p regio 5 Parallel Figure 5 10 gels resi taining 100 w rate of 2 B n bed was w cerin remov ure 5.8 Com In figure bout 700 B ppm. The r on of the cu Gel Resin 5.8 shows th ns, for com 0 ppm of so BV/hour, by washed wit al. mparison of e 5.8, the ef Vs. From 0 resins of the urve. At 40 n Test he soap ex mbined soap oap, 4% me 200 ml of s th methano gel resins f ffluent biodi 0 to 400 BV e two manu 00 BVs, a m xhaustion cu p and glyce ethanol and swelled res ol to test the for soap rem iesel soap Vs the soap ufacturers p methanol wa urve of the erin remova d 0.08% gly sin. At abou e resin’s ab moval (BD1 level was m p level grad performed in ash was co T45BD and al. In this te ycerin was ut 400 bed ility to be re 0, T45BD). measured fo ually increa n a similar onducted. T d Amberlyte est, biodiese purified, at volumes, e egenerated or samples ased to abo manner for The methan 41 e el a each d for up out r this nol 42 wash was performed at too fast of a flow rate on the Amberlyte BD10 column causing the column to partially fluidize and become unstable. Soap particulate was dislodged and leached into the effluent biodiesel raising the soap level during the 400 to 600 BV region of the Amberlyte BD10 curve. After 600 BV, the column stabilized and the performance of the two resin columns became similar again. The resin soap removal exhaustion curve from crude biodiesel, containing both soap and glycerin, is similar to the soap exhaustion curve from biodiesel containing only soap. The crude biodiesel used in the soap-only test, depicted in figure 5.5, contained soap levels of about 2000 ppm. The effluent biodiesel soap level exceeded 50 ppm at about 225 BV in this test. The crude biodiesel used in the soap and glycerin combination test, depicted in figure 5.8, contained soap levels of about 1000 ppm. The effluent biodiesel soap level exceeded 50 ppm at about 400 BV in this test. From these figures, it can be seen that doubling the soap level of the crude biodiesel entering the column will reduce the number of bed volumes that can be treated, for a given effluent soap level, by approximately half. Comparing figures 5.5, 5.7 and 5.8 it can be seen that the resin exhaustion curves are of roughly the same shape. In figures 5.7 and 5.8, the crude biodiesel contained about half the soap level as was used in figure 5.5, therefore, about double the bed volumes are needed to achieve the same effluent soap level. From this comparison, it appears that the presence of glycerin in the crude biodiesel does not affect the resin’s soap removal life. Figure 5.9 shows the glycerin exhaustion curve from the same test as figure 5.8. In this test, biodiesel containing 1000 ppm of soap, 4% methanol and 0.08% glyc abo resin glyc proc abo BV. effec met of 2 Figu cerin was pu ut 400 bed n’s ability to In figure cerin values cedures. E ut 150 BV. It was fou ct on the ef hanol wash to 5 bed vo ure 5.9: Com urified, at a volumes, e o be regene e 5.9, the gl s after the m ffluent glyc The ASTM nd that the ffectiveness h flow rate s olumes per mparison of flow rate o each resin b erated for g ycerin leve methanol wa cerin values M specificat flow rate o s of washin should be r r hour. f gel resins of 2 BV/hou bed was wa glycerin rem l was meas ash are not s varied from tion level of of methanol ng. From fig roughly the for glycerin r, by 200 m ashed with moval. sured from t typical due m about 0% f 0.02% wa during rege gure 5.6, it same as th n removal (B ml of swelled methanol t 17 to 700 B e to incorre % at 75 BV t s exceeded eneration h can be see he biodiese BD10, T45B d resin. At o test the BV. The ect wash to 0.01% a d at about 2 has a signif en that the l run flow ra BD). 43 t 200 icant ate 44 The glycerin exhaustion curves of the two gel resins were of roughly the same shape and their performance can be considered comparable. The slight variability between the two curves is likely due to differences in the flow rate and flow dynamics through the resin bed. These parameters were different due to variability in the flow control valves and differences in methanol evaporation between the two columns. The manufacturer’s estimate of the life of the ion exchange resin, at an initial soap level of 1000 ppm, is 1 kg of resin is needed to purify 1600 kg of biodiesel. In this study, it was found that, at an initial soap level of 1000 ppm, 1 kg of resin is needed to purify 500 kg of biodiesel. This discrepancy might be due to differences in the expectations for quality of the purified biodiesel since these are not stated by the manufacturers. 5.3 Ion Exchange Resin Operating Parameters Tests were conducted to determine the proper flow rate and bed aspect ratio parameters to achieve optimal performance of the ion exchange resins. Figure 5.10 shows the effect of flow rate and bed aspect ratio on soap removal. For this test, the flow rate was varied from 0 to 250 unswelled bed volumes per hour, for unswelled bed heights of 1 to 3 inches in 0.5 inch increments. The crude biodiesel that was used for this test contained 1000 ppm of soap, 5% methanol and glycerin. The exact amount of glycerin was not measured. This data shows that as flow rate increased, soap removal decreased and as the bed height increased, the soap removal increased. It can be seen that for this range of flow rates and 2-3 inches bed heights, the soap removal is roughly the 45 same. Breakthrough of soap, as indicated by elevated outgoing soap levels, occurs almost immediately with the 1 inches bed height and at about 75 BV/hour for the 1.5 inches height. It can be assumed that if additional tests were conducted with greater flow rates, breakthrough would occur in the larger bed heights in the same way it occurs in the 1 and 1.5 in bed heights. It was not possible to run these tests with the current equipment because it would have required pressurizing the bed and this was not possible. From this figure, it can be seen that for a dry resin bed aspect ratio of 2 to 1 or greater, the flow rate should be kept below 100 bed volumes per hour to obtain outgoing soap levels of below 50 ppm. The Thermax company recommends that a flow rate of about 2-5 swelled or 4 to 10 unswelled BV/hour be used. This data shows that the manufacturer’s flow rate recommendations are conservative. Figure 5.11 shows the effect of flow rate on glycerin removal. In this test, crude biodiesel containing 0.09% glycerin, 4% methanol and no soap was passed through 140 ml of swelled resin. The flow rate was varied from about 0 to 30 BV/hour. It can be seen that as the flow rate increased, the glycerin breakthrough increased. These data show that the ASTM specification for glycerin was not reached for the maximum flow rate tested, which was significantly greater than the manufacturer’s recommended flow rate of 2 to 5 swelled BV/hour. Figu Figu ure 5.10 Effe ure 5.11 Effe ect of flow r ect of flow r rate and asp rate on glyc pect ratio o cerin remov on soap rem val (Resin: T moval. T45BD). 46 47 5.4 Adsorbent Media Effectiveness Tests were conducted to determine the effectiveness of adsorbents, such as Magnesol in removal of soap and glycerin from biodiesel. Crude biodiesel, at 60°C, containing 1200 ppm of soap, 0.08% free glycerin and no methanol, was treated with Magnesol and diatomaceous earth. Magnesol is specially treated by the manufacturer to have the ability to adsorb polar compounds. Diatomaceous earth is not supposed to have the ability to attract or adsorb compounds but merely acts as a depth filter as it accumulates on the surface of the media. Before testing the adsorbents, the crude biodiesel was passed through a paper filter to determine the filter’s ability to removal soap and glycerin. The filtered biodiesel contained soap and glycerin levels of 700 ppm and 0.06%, respectively. For this test, the weight percentage of purification media was varied from 0.1 to 5% and the soap and glycerin values were measured. The details of this experimental setup were discussed in more detail in chapter 4. In figure 5.12, it can be seen that Magnesol, the commercial magnesium silicate adsorbent media, removed soap to a lower concentration than diatomaceous earth. Magnesol reduced the effluent soap levels to 100 ppm at 0.1% weight. From 1 to 5% weight, the soap level was consistently about 30 ppm. Diatomaceous earth reduced the effluent soap levels to about 375 ppm at 0.1%. The effluent soap level gradually decreased as the weight percentage was increased from 0.1% to 2%. From 2 to 5% weight, the soap level was consistently about 70 ppm. Figu glyc 0.04 as th glyc appa and weig 0.05 5% ure 5.12 Soa In figure cerin than d 45% at 0.01 he Magnes cerin level w arent increa should not Diatoma ght. From 0 5% to 0.03% did not red ap removal e 5.13, it ca iatomaceou 1%. The gly sol percenta was below t ase in glyce t be conside aceous eart 0.1 to 0.5% %. Increasi uce glyceri by adsorpt n be seen t us earth. M ycerin leve age was inc the detectio erin level fr ered signific th reduced % of diatoma ing the perc n levels be tion and filtr that Magne Magnesol re l decreased creased fro on threshold rom 0.25 to cant. effluent gly aceous eart centage of low 0.03%. ration. esol is more educed the d from 0.04 m 0.1 to 2% d and consi 0.5% was ycerin levels th the glyce diatomaceo . The incre e effective a effluent gly 45% to unde %. From 2 idered to be due to exp s to 0.05% erin levels r ous earth fr eases in gly at removing ycerin level etectable le to 5%, the e 0.005%. erimental e at 0.1% reduced fro rom 0.5% to ycerin levels 48 g s to evels The error om o s at 1 an sign corr perc abo effec Figu that biod sign nd 2% were nificant. Figure 5 responds to centage is v ut 800 to 30 ct in improv ure 5.13 Gly The com Magnesol diesel than d nificantly red e proably du 5.14 shows o the same varied from 00 ppm. Th ving the qua ycerin remo mparison an is significan diatomaceo duce biodie ue to exper the effect o test as figu 0.5% to 5% his figure s ality of biod val by adso nd evaluatio ntly more e ous earth. esel free fat rimental err of Magneso ures 5.12 an % and the F hows that M diesel, by re orption and on of Magn effective at r It was foun tty acid valu ror and sho ol on the ac nd 5.13. Fo FFA over th Magnesol c educing its filtration. esol and di removing s nd that Mag ues. All of t uld not be c cid value. T or this test, his range de can have a free fatty a iatomaceou oap and gly gnesol can a he testing i considered This figure the Magne ecreased fr significant cid value. us earth fou ycerin from also n this 49 esol rom und m expe likel biod Figu 5.5 Mag sign incre incre for t mak eriment wa y produce b diesel, thus ure 5.14 Red Sulfur Table 5. gnesol on th nificant effec ease in sulf eased sulfu he macropo ke a signific s conducte better soap making the duction in a .1 shows th he biodiese ct on sulfur fur level in t ur levels fro orous resin cant change ed using bio p removal d em easier to acid value b he effect of el sulfur leve r levels. Ion the initial fu om 12.5 ppm n after 5 BV e in sulfur le odiesel at 60 ue to the so o filter. by processin purification els. Magne n exchange uel process m to 22 ppm V. After 24 evel, sulfur 0°C. Lowe oaps being ng with Mag n by ion exc esol purifica e resin prod sed. The io m for the ge BV, the ion levels retu er temperatu g less solub gnesol. change resi ation did no duced a sign n exchange el resin and n exchange rned to abo ures would le in the ns and ot have a nificant e resin d up to 35 p e resin did n out 12.5 pp 50 ppm not m. 51 In order to determine if the differences were significant between each purification method and the crude biodiesel, statistical analysis using the SAS software package was performed. The procedure used for this analysis was the t test (LSD) Least Significant Difference. Using a 95% confidence level it was found that each sample had a significant difference in sulfur level as compared to the crude biodiesel except for the Macroporous resin after 24 BV, therefore, it can be assumed that ion exchange resins do not provide a significant change in sulfur levels after 24 BV. Biodiesel sulfur levels are high in the low BV period due to sulfur residue leaching from the resin. Sulfur levels reduce after all the sulfur residue is leached from the resin. This sulfur residue is a by-product from the sulphonation process used in the manufacturing of the ion exchange resins. Table 5.1 Effect of purification media on biodiesel sulfur level. Sample Sulfur (ppm) Crude Biodiesel 13.11 Gel Resin (5 BV) 22.05 Macroporous Resin (5 BV) 35.26 Macroporous Resin (24 BV) 12.29 Magnesol 10.02 52 CHAPTER 6. CONCLUSIONS 6.1 Conclusions This chapter discussed the purification of biodiesel with ion exchange resins and adsorbents. Results supporting the four mode theory of soap and glycerin removal were presented and discussed. Tests were conducted to determine the operating parameters such as the bed aspect ratio and flow rate required to produce optimal performance. Ion exchange resin exhaustion curves were generated for the gel and macroporous resins. The performance of adsorbents were evaluated and compared. Two tests were performed to validate the four mode theory of soap and glycerin removal. It was found that soap can be removed by a paper filter and that increasing the filter’s fineness produces more soap removal. Sodium soaps were found to be more filterable than potassium soaps. Ion exchange resins were evaluated to determine the modes that are active in their operation. It was found that at low crude biodiesel soap levels ion exchange resins remove soap predominately by the ion exchange mode. At higher soap levels, soap is removed by a combination of ion exchange, adsorption and filtration. This test also showed that as the crude biodiesel soap level is increased the effluent free fatty acid value increases. Ion exchange resin exhaustion curves were generated for gel and macroporous resins from two manufacturers. These curves were generated for crude biodiesel containing only soap, only glycerin and a combination of soap and 53 glycerin. It was found that, as the ion exchange resin ages, the effluent free fatty acid value decreases and the effluent soap level increases. It was observed that there are two phases of soap removal that occur sequentially during the exhaustion of a resin. First, ion exchange gradually decreases and combines with adsorption. Second, ion exchange, adsorption and filtration all gradually become exhausted leading to more soap passing into the effluent biodiesel. Glycerin was found to be removed by a combination of adsorption and filtration. Glycerin levels increase at a relatively constant rate over a glycerin exhaustion curve; upon reaching the incoming level, the effluent glycerin content levels off. Glycerin removal capability can be regenerated by washing the ion exchange resin with methanol. This process was found to be sensitive to high methanol flow rate. High methanol flow rate does not allow time for the glycerin to be desorbed into the methanol. The recommended methanol flow rate for glycerin regeneration is 2-5 BV/hour. The shape of the soap exhaustion curve is roughly the same, regardless of the crude biodiesel glycerin levels. It was found that there is a linear relationship between the crude biodiesel soap level and the number of bed volumes that can be processed to reach exhaustion. If the crude biodiesel soap level is doubled, the exhaustion bed volume is halved. Tests were conducted to compare ion exchange resins of different types and different manufacturers. Gel and macroporous resins were found to provide similar performance for soap removal. Gel resins from Rohm and Haas and Thermax were found to have similar performance for soap and glycerin removal. 54 Tests were conducted to determine the bed aspect ratio and flow rate to obtain optimal performance. Bed aspect ratio has a significant effect on the optimal flow rate through the ion exchange resin bed. It was found that as the flow rate increased, the effluent soap and glycerin levels increased. As the bed aspect ratio increased, the effluent soap levels decreased. For the range of flow rates tested it was found that aspect ratios from 2 to 1 through 3 to 1 performed similarly. Tests were conducted to evaluate and compare the effectiveness of adsorbents and filtration media, such as diatomaceous earth, Magnesol and saw dust, in removing soap and glycerin. It was found that paper filters alone, can reduce soap levels significantly. Magnesol is more effective in removing soap and glycerin than diatomaceous earth. Magnesol can reduce the free fatty acid value of biodiesel. Saw dust was also found to be effective in removing soap and glycerin from crude biodiesel. Tests were conducted to determine the effect of biodiesel purification products on biodiesel sulfur levels. When ion exchange resins are first used the initial biodiesel contained increased sulfur levels. After about 20 bed volumes, this value decreases to the crude biodiesel level. Magnesol was found to have a small effect in reducing crude biodiesel sulfur levels. 6.2 Recommendations for Future Work Future studies on this topic should develop methods to obtain better control of process variables. Biodiesel methanol concentration was found to be a difficult variable to control due to the volatility of methanol. The columns and biodiesel reservoir should be designed to minimize methanol leakage. A new column design 55 should be used that incorporates flow rate control using a feedback loop and temperature control. Soap and glycerin measurements should be taken as soon as possible after processing with the purification media. Samples were found to change over time when stored. Soap and glycerin measurements should be performed at least three times and averaged to obtain sound values. Additional tests should be conducted to compare purification performance using potassium and sodium catalyst. Tests should also be conducted comparing several different initial soap, glycerin and methanol levels. The effects of temperature should also be tested on resin performance and life. 56 REFERENCES Albright, R.L. and Yarnell, P.A. 1987. Ion Exchange Polymers: Encyclopedia of Polymer Science and Engineering. New York: Wiley. Arden, T.V. 1968. Water Purification by Ion Exchange. New York: Plenum Press. Berrios, M. and Skelton, R.L. 2008. Comparison of Purification Methods for Biodiesel. Chemical Engineering Journal. 144: 459-465. Bruice, Paula. 2004. Organic Chemistry. Upper Saddle River: Prentice Hall. Bryan, Tom. 2008. Adsorbing it All. Biodiesel Magazine March. Cooney, David. 1998. Adsorption Design for Wastewater Treatment. Boca Raton: Lewis Publishers. He, H.Y. et al. 2006. Comparison of Membrane Extraction with Traditional Extraction Methods for Biodiesel Production. J. Am. Oil Chem. Soc. 83(5): 457-460. Inczedy, J. 1966. Analytical Applications of Ion Exchangers. Oxford: Pergamon Press. Kucek, K.T. 2007. Ethanolysis of Refined Soybean Oil Assisted by Sodium and Potassium Hydroxides. J. Am. Oil Chem. Soc. 84: 385-392. Ma, F., and Hanna M.A. 1999. Biodiesel Production: A Review. Bioresource Technology. 70: 1-15 McCormick, R.L. and Graboski, M.S. 1998. Emission Data on 2-stroke and 4 stroke Engines-Combustion of Fat and Vegetable Oil Derived Fuels in Diesel Engines. Prog. Energy Combust. Sci. 24: 125-164. Osborn, G.H. 1956. Synthetic Ion-Exchangers. New York: The Macmillan Company. Predojevic, Z.J. 2008. The Production of Biodiesel from Waste Frying Oils: A Comparison of Different Purification Steps. Fuel. 87: 3522-3528. Tien, Chi. 1994. Adsorption Calculations and Modeling. Boston: Butterworth- Heineman Series in Chemical Engineering. Van Gerpen, J and Canakci, M. 2003. A Pilot Plant to Produce Biodiesel from High Free Fatty Acid Feedstocks. Transactions of the ASABE 46(4): 945-954. 57 Van Gerpen, J. 2005. Biodiesel Processing and Production. Fuel Processing Technology 86(10): 1097-1107. Wang, Y. et al. 2009. Refining of Biodiesel by Ceramic Membrane Separation. Fuel Processing Technology. 90: 422-427. Yoris, J.C. 2007. Deglycerolization of Biodiesel Streams by Adsorption Over Silica Beds. Energy & Fuels. 21: 347-353. Yoris, J.C. 2008. Adsorptive Properties of Silica Gel for Biodiesel Refining. Energy & Fuels. 22: 4281-4284. 58 APPENDICES A. Saw dust tests Figure A.1 shows the effectiveness of saw dust in removing soap from crude biodiesel. Crude biodiesel containing 2000 ppm of soap, 4% methanol and no glycerin was passed through the saw dust and the outgoing soap level was measured. The saw dust used in this experiment was fine saw dust obtained from a cabinet shop. This experiment found that soap was reduced to levels below the detection limit of about 10 ppm, up to a biodiesel to saw dust ratio of about 45 gram biodiesel to 1 g saw dust. An additional large scale test was conducted and it was found that sufficient soap and glycerin removal was achieved for about 14.6 g biodiesel per g saw dust. Saw dust is able to remove soap and glycerin from crude biodiesel due primarily to filtration. When packed in a column, fine saw dust acts as an excellent surface filter and depth filter. Surface filtration is the particulate deposition on the top surface of a filter. Depth filtration is the particulate deposition throughout the volume of multiple layers of filtration media. Figu ure A.1 Soap p removal wwith saw du ust 59 60 B. Soap and Glycerin Addition Calculations Problem Statement: How much sodium methoxide, methanol and glycerin should be added to obtain 500 ppm of soap and 0.05% glycerin? Assumptions: Initial soap and glycerin levels are 0. MW of sodium soap = 304.4, use 25% sodium methoxide (NaOCH 3 ). Soap: Suu (ppm) = Suu g soop 1u 6 g bioJicscl × 1 molc soop Su4.4 g soop = 1.64S molc soop 1u 6 g bioJicscl u.2S g Na0CB 3 g cotolyst mix × 1 molc Na0CB 3 S4 g Na0CB 3 × 1 molc soop 1 molc Na0CB 3 = u.uu46S molc soop g cotolyst mix 1.64S molc soop 1u 6 g bioJicscl u.uu46S molc soop g cotolyst mix = SS4.9 g cotolyst mix 1u 6 g bioJicscl = u.SSS g cotolyst mix kg bioJicscl Glycerin: Hoss 0lyccrin = % 0lyccrin 1uu × Hoss BioJicscl 61 C. Autotitrator Procedure 1. Prepare samples a. Label i. Weight ii. Sample number iii. Test (ie. T45BD) b. Weigh fuel into cup c. Add acetone 2. Sampler a. Put cups into sampler 3. Titrator a. Menu/enter/sample table b. Sample table i. Method-Soap or Acid ii. ID1 – Sample Number iii. ID2- Test iv. Weight c. Esc to main screen 4. Starting titration a. Press start on sampler i. Number of samples ii. Starting sample position Run calibration daily 1. Tirator a. Sample table i. 1 entry, method “calibration” 2. Sampler a. Start b. 3 samples Purge buret-daily or when changing titrant (HCL or KOH) 1. Put sampler on empty cup 2. Titrator a. Menu/manual control/dosing 62 D. GC Procedure Step 1: Measure 0.1 g of biodiesel into sample bottle. Step 2: Add 100 μL of S1, S2, MSTFA solutions then wait 20 minutes. Step 3: Add 8 ml of heptane to sample bottle. Step 4: Pipet 1000 μL of prepared sample into GC sample bottle. Step 5: In GC program, go to Sequence/Sequence parameters/directory and select appropriate directory. Step 6: Name vials in desired sequence in GC program. 63 E. Sample Titrator Calibration Calibration was performed each day the titrator was used. Three buffers were used with pH values of 4, 7 and 10. The titrator was considered to be calibrated when the Slope was between 96.0% and 100%. Figure E.1 Sample titrator calibration 64 F. Sample GC Calibration GC calibration was performed each time the GC column was replaced. 65 G. Ion Exchange Resin Bed Start-Up Procedure 1. Weigh resin into column. 2. Pump methanol or clean biodiesel into the outlet of the column. Resin should become fluidized. 3. Allow resin to settle and drain the methanol or clean biodiesel until it is just covering the top of the resin bed. 4. Fill column with crude biodiesel. Run several bed volumes through column to remove methanol before taking measurements. 66 H. Methanol Wash Procedure 1. Drain crude biodiesel until it is just covering the resin bed. 2. Fill column with methanol. Pass about 5 BVs of methanol through the column at a flow rate of 2 to 5 BVs per hour. 3. Drain methanol unit it is just covering resin bed. Fill column with crude biodiesel and continue processing. Allow several bed volumes to pass before taking measurements. ii AUTHORIZATION TO SUBMIT THESIS This thesis of Jacob Wall submitted for the degree of Master of Science with a major in Biological and Agricultural Engineering and titled “COMPARISON OF METHODS FOR THE PURIFICATION OF BIODIESEL” has been reviewed in final form. Permission, as indicated by the signatures and dates given below, is now granted to submit final copies to the College of Graduate Studies for approval. Major Professor ______________________ Dr. Jon Van Gerpen ______________________ Dr. B. Brian He ______________________ Dr. W. Daniel Edwards _________________ Date _________________ Date _________________ Date Committee Members Department Administrator ______________________ Dr. Jon Van Gerpen Discipline’s College Dean ______________________ Dr. Don Blackketter _________________ Date _________________ Date Final approval and acceptance by College of Graduate Studies _______________________ Dr. Margrit von Braun _________________ Date iii ABSTRACT Biodiesel is a proven alternative to petroleum diesel fuel. Once produced, biodiesel contains several impurities such as soap and glycerin. The free fatty acids in the oil react with the sodium or potassium catalyst to form soaps. After the biodiesel and by product glycerin are separated, trace amounts of glycerin remain in the biodiesel. Soap and glycerin impurities in the biodiesel can lead to engine wear and fuel storage problems. Traditionally, soap and glycerin are removed from the biodiesel by water washing. Water washing has several disadvantages such as producing large amounts of waste water that requires treatment and causing plant operational problems such as emulsion. Recently, several alternative “waterless” purification procedures have been used, such as ion exchange resins and Magnesol, a brandnamed adsorbent. The objective of this study was to compare these two “waterless” purification methods by their mode of operation, effectiveness, life, and its ease of use. Experiments were developed in order to compare each purification method. Ion exchange resins were tested in packed bed columns and Magnesol was tested by mixing and filtering. During this research it was found that soap and glycerin can be removed by a combination of four modes. These modes are filtration, physical adsorption, ion exchange and soap removal by glycerin affinity. It was found that the four mode theory was supported by the experimental data. Ion exchange resin resins can reduce glycerin levels from 0.iv It was also found that ion exchange resins can reduce soap levels from 1200 ppm to below 50 ppm for about 550 BV of processed biodiesel.02% for about 200 BV of processed biodiesel.08% to below 0.08% to below 0.02% at a Magnesol concentration of 1%. . Magnesol can reduce soap levels from 1000 ppm to below 50 ppm and glycerin levels from 0. I would like to thank my committee members Dr.v ACKNOWLEDGEMENTS I thank my major professor. W. want to thank my wife Christina Wall for her patience and support. Thompson for his ideas and support. especially. Daniel Edwards for their guidance and support. I. Joseph C. Dr. . I would like to thank my lab assistants Sweta Khanal and Sohana Khanal for their help preparing for and conducting the experiments. I would like to thank Dr. I also want to mention his help with the development of equipment and procedure to make this project possible. Jon Van Gerpen for the opportunity to investigate a practical problem in the biodiesel production process. I would like to thank Jim Sabzali and Vivek Naik from the Thermax Company for their visits and technical and financial support. Brian He and Dr. iii Acknowledgements…………………………………………….xi Chapter 1.3 4...26 .14 Purification Processes……………………………………………………15 Chapter 4.4 4.2 4.21 Adsorbent Purification…………………………………………………….………….. Chapter 2.5 Filtration……………………………………………………………………...1 4.1 3... 2.21 Packed Bed Purification…………………………………………………... 3.5 4.v List of figures…………………………………………………………….7 Adsorbents………………………………………………………………….2 Theory.22 Feedstock Preparation……………………………………………………22 Purification Media…………………………………………………………23 Measurement Techniques……………………………………………….…………………………………4 Water wash…………………………………………………………………..2 2.6 Experimental methods and equipment………...4 Introduction….8 Ion Exchange…………………………………………………………….12 Chapter 3.….……………………………………….....14 Four Mode Theory………………………………………………………..……………………………………………….ii Abstract….x Abbreviations………………………………………………………………. 4....vi Table of Contents Authorization to submit thesis…………………………………………..24 Experimental Procedures……………………………………………….3 2..1 Literature review……..1 2.………………………………………………………………….viii List of tables………………………………………………………………. …………………………………………………………….2.34 5.3 5.5 Parallel Gel Resin Test………………………………………….5 Chapter 6..vii Chapter 5.41 Ion Exchange Resin Operating Parameters……………………………44 Adsorbent Media Effectiveness………………………………………….……………………..65 Methanol Wash Procedure………………………………………………66 .4 Macroporous Resin Test…………………………………………38 5..3 Soap and Glycerin Combined……………………………………38 5.29 5.2 Conclusions…. Saw Dust Tests……………………………………………………………58 Soap and Glycerin Addition Calculations………………………………60 Autotitrator Procedure……………………………………………………..2.. D.34 5.1 6.1 Filtration Mode…………………………………………………….. F.1.47 Sulfur……………………………………………………………………….26 Four Mode Theory Validation……………………………………………29 5..1 Results and discussions…….. H....1.2.4 5.2 Ion Exchange Mode………………………………………………32 Ion Exchange Resin Exhaustion………………………………………. 5.62 Sample Titrator Calibration………………………………………………63 Sample GC Calibration………………………………………………….54 References…...………………………………………………………………58 A..52 Recommendations for Future Work………………………………..52 Conclusions……………………………………………………………….64 Ion Exchange Resin Bed Start-up Procedure………………………….2 Glycerin Only………………………………………………………36 5.. E.2.56 Appendices.61 GC Procedure…………………………………………………………….50 5. 6.….2.2 5. G. B.………………………………………..1 Soap Only…………………………………………………………. C. ..8 Figure 3.9 Comparison of gel resins for glycerin removal……………………………43 Figure 5.11 Effect of flow rate on glycerin removal……………………………………46 .31 Figure 5..26 Figure 4.34 Figure 5.36 Figure 5...33 Figure 5.3 Autotitrator…………………………………………………………………….2 Adsorbent purification equipment………………………………………….39 Figure 5.viii LIST OF FIGURES Figure 1.3 Manufacture of a strong acid cation exchange resin…………………….22 Figure 4.46 Figure 5.41 Figure 5.17 Figure 3.6 Glycerin removal with no soap present…………………………………….5 Effect of resin age on soap removal……………………………………….2 Ion exchange mechanics……………………………………………………18 Figure 3.2 Methanol removal from biodiesel containing sodium and potassium……………………………………………………………….1 Diagram of ceramic membrane filtration for purification of biodiesel…….10 Effect of flow rate and aspect ratio on soap removal………………….7 Soap removal from biodiesel containing soap and glycerin…………….1 Ion exchange test setup…………………………………………………….19 Figure 3.1 Physical adsorption mechanics…………………………………………….1 Figure 2...4 Gas Chromagraph……………………………………………………………26 Figure 5..8 Comparison of gel resins for soap removal……………………………….3 Effect of biodiesel soap level on acid value……………………………….1 Saponification reaction……………………………………………………….....23 Figure 4.4 Structure of a strong acid cation exchange resin…………………………20 Figure 4.37 Figure 5.1 Effect of filter porosity on soap removal……………………………………30 Figure 5.4 Statistical analysis of experimental data…………………………………. 13 Glycerin removal by adsorption and filtration……………………………49 Figure 5.48 Figure 5.ix Figure 5.50 .14 Reduction in acid value by processing with Magnesol………………….12 Soap removal by adsorption and filtration……………………………….. 1 Effect of purification media on biodiesel sulfur level………………………51 .....4 Table 2.8 Table 2.9 Table 2.x LIST OF TABLES Table 2.4 Effect of Magnesol purification on biodiesel properties………………….10 Table 4.1 Comparison of ion exchange resins product data…………………………24 Table 5...3 Monoglyceride.2 Results of purification with a ceramic membrane…………………………. bound glycerin content from two step reaction………….1 Effect of impurities on biodiesel and engines……………………………….. xi ABBERVIATIONS The following abbreviations were used throughout this thesis: FFA: Free fatty acid GC: Gas Chromatography PPM: Parts per million BV: Bed volume . by phase separation. Biodiesel has been shown to reduce engine exhaust emissions leading to less pollution of the environment. Figure 1. Soap is the product of saponification reactions that occur in parallel to tranesterification. If the oil contains free fatty acids. (McCormick et al. INTRODUCTION Biodiesel is a well known and proven alternative to petroleum diesel fuel. Figure 1. After the transesterfication reaction has been completed and the glycerin has been separated and removed.. and heated until a separation occurs between the biodiesel and the glycerin. the raw biodiesel still contains a small percentage of free glycerin and soap.1 CHAPTER 1. 1998) Biodiesel is typically produced using a transesterification reaction. these will be converted to soap by the . Plant oils or animal fats are mixed with alcohol and a base catalyst. such as sodium hydroxide. It can be used in most diesel engines with few modifications.1 depicts the saponification reaction. Soap is generally formed because of free fatty acids reacting with the catalyst.1 Saponification reaction (Bruice. The majority of the glycerin is then removed from the solution. 2004). The water contained in the reactants of transesterification causes an increase in free fatty acids by hydrolysis of the oil and therefore more soap will be created. several methods are available to remove soap and glycerin from biodiesel. such as calcium and magnesium that could be transferred to the fuel causing the fuel to be out of specification for those compounds. Other methods for removing impurities . the fuel must be dried. wash water must be deionized to remove metal ions. The soap and glycerin are extracted into the water phase and removed when the water is separated from the solution. After water washing. Prior to use. • Water washing with a high soap level biodiesel can lead to emulsions that can cause significant yield loss and other plant operational problems. (Ma et al.. Water washing introduces water to the fuel that can cause it to degrade. multiple washes and water-biodiesel separation. the soap and trace amounts of glycerin are removed by washing with water. These impurities can be removed by passing the fuel through a bed of ion exchange resin. This causes an increased energy cost. 1999) Traditionally. (Bryan.2 saponification reaction. This involves mixing water with the biodiesel. • Water washing can require a great deal of time for drying. • • • Water washing produces a large amount of waste water that must be treated. 2008) As an alternative to water washing. and then allowing them to separate. The use of water to wash the biodiesel feedstock causes many problems. agitating them gently. . effectiveness. life. Each method will be compared by its mode of operation. 2007).. and ease of use. .3 involve the use of adsorbent compounds such as magnesium silicate (Berrios et al. Kucek et al. The operating parameters and recommendations for use will also be presented. The object of this research is to compare ion exchange resins to adsorbents for the purification of biodiesel. 2008. high sulfated ash (filter blockage). Table 2. 2008. methanol.. Predojevic et al. The impurities contained in crude biodiesel include free fatty acids. and metal compounds such as soap and catalyst. Yoris et al. water. crystallization Injector deposits.. Impurity Free Fatty Acid (FFA) Water Methanol Glycerides Metals(soap.. 2008). Increases in aldehydes and acrolein exhaust emissions . Table 2. abrasive engine deposits Settling problems. 2009. injector deposits. 2008. corrosion High viscosity.. glycerides. magnesium silicate and ion exchange resin are the common materials used in these studies. filtration with a membrane. bacterial growth (filter blockage) Low density and viscosity. catalyst) Free Glycerin Effect Corrosion. Crude biodiesel contains impurities that can cause problems in its use and storage.1 shows how the impurities affect the biodiesel and engines. Silica gel.. Researchers have used water washing. adsorbents and ion exchange resins to remove impurities such as soap and glycerin (Berrios et al. 2006. 2008). Kucek et al. low flash point.. corrosion. low oxidative stability Formation of FFA.4 CHAPTER 2.1 Effect of impurities on biodiesel and engines (Berrios et al. 2007. free glycerin. This chapter discusses the results of these studies in purifying biodiesel. He et al.. Wang et al. Literature Review Various purification methods have been proposed in the literature to purify biodiesel. water content. cetane index and yield loss were evaluated. (2003. Studies by Van Gerpen et al. Also. . 2003).1 Water Wash Washing the crude biodiesel with water is the traditional way to remove impurities. acid value. water content. kinematic viscosity. glycerin removal required longer wash times with acidified water compared to deionised water. Their results showed that increased wash water temperatures improved soap removal but glycerin removal was not significantly improved by temperatures elevated above ambient. Some cases required as many as six individual washes (Van Gerpen et al. Berrios et al. the effect of wash method on the biodiesel density. There are several variations to water washing. saponification number.5 2. including washing with deionised water. iodine number. and using a membrane to prevent emulsions with water washing. saponification number. In this study. 2005) have shown that adding acid to the wash water aids in removing soaps. cetane index between a distilled water wash and a phosphoric acid wash. However. Predojevic et al. (2008) compared water washing with hot distilled water to a 5% phosphoric acid solution. It was found that the phosphoric acid wash decreased the acid value to a greater degree. multiple washes are required to achieve satisfactory soap removal. washing with acidified water. iodine number. kinematic viscosity. (2008) tested water washing with deionised water and acidified water and found that acidified water performed better in removing soaps.. and the phosphoric acid wash produced higher yields than the distilled water wash. It was found that there was no significant change in density. Then. Two membranes were tested. (2006) conducted experiments using several methods of water washing. and then the water was removed. The distilled water wash involved adding a 1:1 volume ratio of water to biodiesel. These wash methods included distilled water. the solvent was removed and the solution was dried over sodium sulfate. The HCl wash was performed by adding a 1:1 volume ratio of HCl (pH =1) to biodiesel and agitated and separated similar to the distilled water wash. This study compared yield loss for each method but did not give information about impurity removal. allowed to phase separate. This process was repeated three times. HCl acid wash. The hollow fiber membrane extraction involved using a 1 m long by 1 mm diameter hollow fiber membrane. It was found that the highest biodiesel yield was achieved at a water temperature of 50°C and an agitation rate of 125 rpm. The solution was then dried over sodium sulfate. a hydrophilic . solvent extraction. The biodiesel was then washed with distilled water two additional times and then the solution was dried over sodium sulfate. Evaporation of the solvent also lead to yield loss. The solvent extraction was performed by adding petroleum ether to biodiesel at a 1:1 volume ratio. and hollow fiber membrane extraction.6 He et al. The solution was then water washed three times. The solution was agitated in a water bath oscillator for 20 min. It was found that the addition of acid led to reduced biodiesel yield loss compared to a distilled water wash at the same temperature. Addition of solvent led to a yield loss due to emulsion formation at the biodiesel/water interface. The acid value increased due to the addition of acid. whereas washing with 20°C distilled water provided the highest yield loss of 15. 0. magnesium.2 Filtration Wang et al. The inside of the tube contained 19 three mm channels giving a total filtration area of 0. These results were compared for each membrane pore size used. The purified biodiesel was then dried over sodium sulfate. To begin the test.1% weight. This study involved passing raw biodiesel through a ceramic membrane after methanol removal. a sample was collected and analyzed for potassium.. Biodiesel was then passed through the membrane at 60 °C. The effects of temperature and transmembrane pressure were also tested at the 0.2 µm. . 8. (2009) found that soaps and glycerin could be removed by filtration using a membrane.1 shows the setup for filtration using the ceramic membrane tubes.045 m2.1 µm pore size.1 µm. The hydrophilic polysulfone membrane provided the lowest yield lost of the study. Each membrane was placed in a beaker filled with distilled water and biodiesel was pumped through it. methanol at 30 °C was passed through the membrane to provide initial cleaning. filtration pore sizes of 0. To test the effectiveness of the ceramic membrane. After 3 minutes. The membrane was configured in a tube shape with an outside diameter of 26 mm and a length of 250 mm.2% weight. calcium. (He et al. the mass transfer occurred without phase mixing with each other. and 0. 2006) 2. It was found that when using membrane separation. Figure 2. sodium.6 µm were evaluated. and free glycerin.7 polysulfone membrane and a hydrophobic polyacrylonitrile membrane. 45 0..0179 2. 2009).95 0. .0257 Pore: 0.261 Pore: 0.18 0.8 Figure 2.26 0.0276 Pore: 0.2 Results of purification with a ceramic membrane (Wang et al. Table 2.25 0.2 shows the results comparing the three pore sizes and water washing for the crude biodiesel.15 0. Magnesium silicate and silica gel were the materials used.2 0.25 0. Metals Content (mg/kg) Potassium Sodium Calcium Magnesium Free Glycerin (%wt) Crude Biodiesel 160 8.. The potassium was in the form of soap and catalyst.55 0.2 µm 2. From this table it can be seen that as the pore size decreases the effectiveness of removal of potassium and free glycerin were enhanced significantly.88 0. Table 2.36 0. Kucek et al.98 1.68 0. 2009).7 1.64 0.6 µm 4.7 0.0152 Water Wash 2.33 0.1 Diagram of ceramic membrane filtration for purification of biodiesel (Wang et al.1 µm 1.46 1.3 Adsorbents Solid adsorbents have shown success in the removal of impurities from biodiesel.41 0. 9 (2007) conducted a study on analyzing the use of the commercial magnesium silicate product. The biodiesel fuels were sampled after 10.. MAG (wt%) Bonded Glycerin (wt%) NaOH 1. The first stage involved producing biodiesel with ethanol and KOH or NaOH.97 0. Use of Magnesol provided no significant yield loss. Biodiesel fuels from two different sources were compared. 2007).3 Monoglyceride (MAG). a second reaction was performed and the biodiesel was water-washed. In this study biodiesel was produced in a two stage reaction. Table 2.3 Berrios et al. All other biodiesel impurities were below the Brazilian biodiesel standard. After the first stage. the solution was allowed to separate and the glycerin was drained off and the ethanol was removed. bonded glycerin content from a two step reaction (Kucek et al. The Magnesol was mixed with the biodiesel fuels for 30 min and removed by filtration.25 KOH-Magnesol 1. Magnesol. 2% by weight of Magnesol was added to the biodiesel and stirred continuously for 20 min at 65°C.65 0.89 0. Next. (2008) evaluated the use of Magnesol for the purification of biodiesel.48 NaOH-Magneso 0. Table 2.42 KOH 1. Using this procedure it was found that the use of Magnesol provided a significant reduction in monoglycerides and bound glycerin compared to a two stage reaction with a water wash between the stages.3 shows the reduction in monoglyceride and bond glycerin obtained from this study. 20 and 30 min of agitation to determine .19 0. After the Magnesol was filtered from the biodiesel. To dry the biodiesel.02 Monoglyceride ((%(m/m)) 0. Table 2.43 5. water content. Raw Fuel Magnesol Treated Biodiesel 1 Biodiesel 2 Biodiesel 1 Biodiesel 2 EN 14214 Maximum Methyl Ester Content (%(m/m)) 95. 0.8 0.3 6 Acid Value (mg KOH/g sample) 0. soap.8 Diglyceride ((%(m/m)) 0.10 the effect of residence time on purification effectiveness.11 0.05 0.2 0.35 5. 2008).24 0.1 Free Glycerol (%(m/m)) 0.2 0.21 0.4 Effect of Magnesol purification on biodiesel properties (Berrios et al.3 90. water content. It can be seen that the free glycerin. and cetane index and yield loss. It was found that silica gel produced yields of about 94% compared to 90% from water washing with .25.19 0.71 0.69 0. kinematic viscosity.46 0.21 1. 0. a 1 cm layer of sodium sulfate was placed above the silica gel.11 0. iodine number.21 1.68 0. acid value and oxidative stability. Each of the samples was evaluated for methanol content. iodine number.5. soap.4 shows the averaged performance of Magnesol purification on the properties of raw biodiesel.51 0.2 Soap (ppm) 909 1240 250 340 Water (mg/kg) 1050 1050 1050 1050 500 Oxidative Stabilty (h) 1.2 Triglyceride ((%(m/m)) Methanol ((%(m/m)) 1. saponification number or cetane index due to purification by silica gel.4 1. It was found that there was no significant change in density.73 0.7 3. acid value.3 1. and methanol content are reduced significantly.18 0. This study compared the results from purification with silica gel to that of water washing. kinematic viscosity. Four experiments were conducted using 0.1 0. This study analyzed the biodiesel density.5 Predojevic et al. saponification number. It was found that soap and glycerin were removed satisfactorily by the 10 min case.8 3.41 0. free glycerol.03 0.75 and 1 % by weight of Magnesol in the biodiesel. (2008) evaluated the use of silica gel in a column with bed dimensions of 2 cm by 2 cm.. Table 2.32 0. Stirring was performed at 25°C for 2 hours. test solutions were created by adding glycerin to clean biodiesel and also solutions prepared containing water. Silica gel reduced the acid value by greater than 90% compared to a 75% reduction by distilled water washing. . was found to exceed 1 hour at a space velocity of 14 cm3/min.1 . (2007.2% was passed through the bed at a space velocity of 3-11 cm3/min. Sodium stearate and Glyceryl monosterate were used to simulate soap and monoglycerides.139 grams glycerin per grams silica gel. For this test. The breakthrough time. a test solution was prepared by adding glycerin to clean biodiesel with no methanol. 2008) to evaluate the effectiveness of silica gel for the removal of glycerin from biodiesel.25 inch inside diameter and 30 cm in length was loaded with silica gel. In the second study. Two studies were conducted by Yoris et al. which is the time needed to reach the ASTM standard for glycerin. 3 grams of silica gel were added to the solution. After the test solution was prepared. It was also found that the glycerin adsorption capacity could be restored by flushing the bed with 4 bed volumes of methanol. From this curve it was found that the maximum adsorption capacity was 0. A column of 0. methanol.11 distilled water. The silica gel was then allowed to settle and the liquid phase was sampled for analysis. monoglycerides and soaps in addition to glycerin.0. Tests were conducted to regenerate the bed by washing with methanol. a breakthrough curve was generated. Biodiesel containing glycerin levels of 0. In the first study. Soap and glycerin were reduced significantly. The ion exchange resins were tested their ability effect methanol. Two ion exchange columns were tested in parallel.5 bed volumes per hour were used.4.12 The samples were analyzed to determine the effect of soaps. it can be seen that biodiesel fuels #1 and # 2 had about the same glycerin content but biodiesel fuel #2 . water. Berrios et al. From this test it was found that ion exchange resin has little effect on methanol concentration after the bed has reached equilibrium. 2. soap. The two resins were found to perform similarly and the purification capacity was found to be 500 L biodiesel /kg resin for the biodiesel #1 and 720 L biodiesel/kg resin for the biodiesel #2. methanol and monoglycerides on the adsorption of glycerin.25 L/h or 2. Methanol and monoglycerides were shown to reduce the adsorption capacity of the silica gel for glycerin. Oxidative stability did not seem to be affected. Purification with the ion exchange resin caused the acid value of the purified biodiesel to increase slightly. acid value and oxidative stability parameters of the biodiesel. (2008) evaluated Ion exchange resins from Purolite and Rohm and Haas to determine their effectiveness. It was found that soaps and water did not affect the adsorption of glycerin. Two raw biodiesel fuels were used for testing. glycerin.4. as was shown in Table 2. In table 2. The 3 cm diameter resin columns were loaded with 80 grams of dry resin and flow rates of 0.4 Ion Exchange Berrios et al. (2008) tested the use of ion exchange resins on the purification of biodiesel. 13 had about 300 ppm more soap. This difference in initial soap level is the reason why the resin had longer life processing biodiesel fuel #1. . Adsorption: Adsorption is the removal of soluble impurities by a chemical action. Filtration: Filtration is the removal of impurities that are insoluble in biodiesel by a mechanical action. Filtration can be used to remove both insoluble soap and glycerin. Filtration is most effective when biodiesel methanol levels are low. The soap and glycerin adsorption capacity of ion exchange resins can be regenerated using a methanol wash.1 Four Mode Theory In the course of the research conducted for this project. these byproducts are mainly soap and free glycerin. the authors proposed that the removal of trace amounts of soap and free glycerin from biodiesel involves four different operating modes. 2. The various media and methods of purification take advantage of at least one of these modes. Ion Exchange: Ion exchange is the removal of impurities by exchanging an ion from the ion exchange material. The pores on the adsorbent are charged and attract the impurities. This chapter discusses theory behind the removal of these impurities. for the metal portion of the impurity . 3. The four modes of purification are: 1. Adsorption can be used to remove both soluble soap and glycerin. Theory The production of biodiesel produces byproducts that can have a negative impact on the biodiesel’s end use. 3. The impurities become chemically bonded to the charged surface of the adsorbent. In most cases. This mechanical action can take the form of surface filtration or depth filtration.14 CHAPTER 3. both types of filtration are involved. soap is entrapped in the glycerin layer and removed from the biodiesel stream. After the exchange takes place. Filtration is a mechanical process where insoluble particles are removed from a fluid-particle suspension. These separation processes include filtration. Ion exchange is the chemical breakdown of the impurity. 4. therefore glycerin can aid in the removal of soap. This theory clarifies the way impurities are removed from biodiesel. These four modes of biodiesel purification encompass the mechanisms and interactions that allow impurities to be removed from biodiesel. In the technical literature there is no consensus about the mechanisms that are active in biodiesel purification. The filtration media is placed in a column and the fluidparticle suspension is passed through the column and the particles become attached to the external surface of the of the filtration media. and ion exchange adsorption.15 contained in the raw solution. 3. Glycerin/Soap Interaction: Soap has a stronger affinity for the glycerin portion of the biodiesel. As glycerin becomes adsorbed on the surface of the purification media. removing them from . physical adsorption. Ion exchange can be used to remove soap from biodiesel by exchanging the hydrogen ions from the ion exchange material for the sodium or potassium ion on the soap molecule. the free fatty acid portion of the soap molecule is allowed to pass through the resin bed.2 Purification Processes Purification of biodiesel involves a series of separation processes that may occur in combination. The solute molecules must first diffuse into the fluid film layer and then transfer onto the adsorbent particle surface. Depth filtration occurs when multiple layers of filtration media are available and provides a greater surface area for particle removal. surface filtration and depth filtration. There are two main types of filtration. In the crude biodiesel solution. The diffusion from the fluid is driven by a difference in concentration. Adsorption bonding occurs when there is a concentration gradient between the fluid and the adsorbent. Absorption is the accumulation of solute molecules in the bulk volume of the absorbent media. Adsorption differs from absorption in that adsorption is the accumulation of solute molecules on the surface of the adsorbent media. Figure 3. If the fluid has a relatively high concentration of solute. Surface filtration occurs on the surface of the filtration media layer. 1998).1 depicts the mechanics of physical adsorption. In this figure it can be seen that an adsorbent particle is surrounded by a fluid film layer. Adsorption uses interactions similar to van der Waal bonds. diffusion will occur toward the particle and adsorption will occur. Physical adsorption is a process where molecules in the fluid bond onto the internal surface of the adsorbent media and are removed. Van der Waal bonds are formed by a series of repulsive and attractive forces that reach a point of equilibrium when adsorption occurs. to attract and bind solute molecules to the surface of the adsorbent. insoluble soap and glycerin particles can be filtered.16 the fluid. If the concentration of solute is . This bonding is weak and can be reversed by washing the adsorbent with a clean solvent (Cooney. In the ion exchange adsorption. Figure 3. Desorption is the process used to regenerate adsorbent materials.1 Physical adsorption mechanics (Cooney.2 it can be seen that the basic adsorption mechanics for ion exchange are similar to the physical adsorption depicted in figure 3. The sodium .2 depicts the mechanics of an ion exchange resin. Figure 3.17 high on the particle. A diffusion film layer surrounds the ion exchange particle which is suspended in a homogeneous fluid. 1998). This adsorption releases ions of the same charge from the ion exchange material into the fluid. sodium ions diffuse from the fluid flow into the film layer and then onto the ion exchange particle. In figure 3. then desorption will occur. Ion exchanging is a special type of separation process where ions are adsorbed from the fluid surrounding the ion exchange material.1. Strong acid cation exchange resins are manufactured by the sulphonation of styrene-divinyl benzene copolymers.18 ions are exchanged with the hydrogen ions from the ion exchange particle. divinyl benzene and a stabilizer such as polyvinyl alcohol. The hydrogen ion diffuses into the film layer and then into the fluid flow.3 shows the process of strong acid cation exchange resin manufacture (Osborn. Figure 3.2 Ion exchange mechanics (Arden. . 1968). There are four categories of ion exchange resins: Strong and weak acid cation exchange resins and strong and weak base anion exchange resins. 1956). (Arden 1968) Figure 3. The process is performed by agitating a suspension of styrene. Strong acid cation exchange resins are the type used to purify biodiesel. The resin beads are then sulphonated using sulfuric acid. The rate of agitation determines the size of the resin beads that polymerize. Dow. Ion exchange resins swell or shrink with adsorption or desorption of solvents. Gel resins have low cross-linking and have a translucent appearance. Ion exchange resins have several important properties that affect their performance. “gel” and macroporous resins. Cross-linking improves the resin’s mechanical properties but reduces its ion exchanging rate and capacity (Inczedy.3 Manufacture of strong acid cation exchange resins (Osborn. 1956).4 shows the structure of a strong acid cation exchange resin. . Thermax currently is marketing a macroporous resin specifically for purification of biodiesel. 1966). Figure 3.19 Figure 3. Manufacturers primarily make two types of ion exchange resins for purification of biodiesel. and Rohm and Haas. Thermax. Macroporous resins have higher cross-linking and are opaque in appearance. The internal cross-linking of the resin’s structure controls the extent of swelling. Gel resins are the most common and are made by Purolite. 20 Figure 3. -----. . divinyl-benzene cross-link. Polystyrene chain.4 Structure of a strong acid cation exchange resin. 1987). ----. (Albright and Yarnell. A Sigma 900 Max automatic water sampler was used to pump the cleaned biodiesel from the separation funnel into sample bottles at a set time interval. 4. The second involves mixing a purification media with the biodiesel for a period of time and then pumping it through a filter to separate the material from the biodiesel. The cleaned biodiesel was then collected in a separation funnel. .21 Chapter 4.1 shows the equipment used in the packed bed purification experimentation. In the bottom of the column was an 80 mesh screen that contained the purification media but allowed biodiesel to pass through. Figure 4. The fuel level was maintained by a float switch controller and a peristaltic pump to provide a constant head over the bed.1 Packed Bed Purification The equipment used to evaluate the packed bed method consisted of glass columns ranging from 1 to 3 inches in diameter. Ion exchange resins were tested using the packed bed method. Flow rate was controlled by a needle valve and measured by a rotameter. At the top of the column was an inlet and overflow outlet. The first involves passing raw biodiesel through a packed bed of purification media. Adsorbents were tested by the mix and filter method. Experimental Methods and Equipment Two main approaches were used to compare media for the purification of biodiesel. The equipment used for this process is depicted in figure 4. Vacuum was applied to suction the biodiesel through the filter.2 Adsorbent Purification Adsorbents.1 Ion exchange test setup. were tested by first removing the methanol from the crude biodiesel. such as magnesium silicate.3 Feedstock Preparation The biodiesel used in this experiment was prepared from crude mustard seed oil and canola oil crushed at the University of Idaho pilot plant. The biodiesel adsorbent mixture was then agitated for 10 min at a temperature of 60°C and then poured into a Büchner funnel containing a #4 Whatman filter. 4.22 Figure 4. The biodiesel was then water washed to remove impurities. 4. Biodiesel was prepared in a large batch using sodium hydroxide and methanol.2. Two percent by weight of adsorbent was mixed into the crude biodiesel. This was used as the test fuel. . . Figure 4. the clean biodiesel was treated with glycerin. sodium methoxide and methanol to produce the desired level of soap and glycerin.2 Adsorbent purification equipment. The ion exchange resins used in this study were T45BD and T45BDMP from the Thermax Company and BD10Dry from Rohm and Haas.23 For each test.4 Purification Media The adsorbent media used in this study was Magnesol from the Dallas Group. Table 4. 4.1 shows the product data from several ion exchange resin manufacturers. Appendix B shows the soap and glycerin addition calculation. The test fuel was prepared in 5 and 10 gallon batches and was typically used within 1 to 2 days. 2 mm Density (a) 80 g/ml T45BDMP Thermax Macroporous polystyrene Sulfonic acid Hydrogen 0.24 Ion Exchange Resin Product Data T45BD Manufacturer Thermax Type Gel Matrix Structure polystyrene Functional Groups Sulfonic acid Ionic Form Hydrogen Particle size 0. each batch of acetone solvent and bromophenol blue indicator was titrated with NaOH or HCl to the point of a slight color change to faint yellow prior to adding the biodiesel. after adding a measured quantity of fuel.400 Later in the test program. 4. and then it adds small increments of 0. As the solution . The autotitrator uses an electrode to measure the pH of the solution.4-1. free fatty acid value.3-1. One hundred ml of acetone containing 2% deionised water was used as a solvent for each sample. Soap level was determined using AOCS method Cc 17-79.001 normality HCl.5 Measurement Techniques In order to determine the effectiveness of purification.2 mm 80 g/ml BD10 Rohm & Haas Gel polystyrene Sulfonic acid Hydrogen Not available 80 g/ml Table 4.01 or 0. and free glycerin percentage before and after the processing. To provide a blank.4-1.2 mm 40 g/ml PD206 Purolite Gel polystyrene Sulfonic acid Hydrogen Not available 80 g/ml DR-G8 Dowex Gel polystyrene Sulfonic acid Hydrogen 0.001 N HCl to the blue-toyellow endpoint of the bromophenol blue. Switzerland). an automated method was developed to determine the soap level using a Metrohm 848 Titrino Plus Autotitrator (Herisau.1 Comparison of ion exchange resin product data. Then. the mixture was titrated with 0. Soap level was calculated using the following equation: 304. measurements were performed on the soap level. a titration curve is generated. Glyceryl-tridecanoate. Acid value was measured using the ASTM D974 method. 0. 0. Phenolphthalein was added to a sample of solvent and biodiesel and then titration was performed with 0.25 becomes neutralized.1 gram of biodiesel was mixed with 100 µL each of Butanetriol.5% water and 49.5% isopropyl alcohol was used as a solvent for each sample. The titrator then analyzes the second derivative of the titration curve to determine the equivalence point. 1 µL of this solution was then injected to perform the GC analysis.1 KOH in iso-propanol to the clear-pink endpoint of the phenolphthalein. The point on the titration curve that has the largest second derivative is considered to be the equivalence point. 8 mL of heptane was then added as a solvent to the solution. . Acid value was calculated using the following formula: 56. according to the ASTM D6584 GC method. which is the point where all the soap is neutralized. and NMethyl-N-trifluoroacetamide (MSTFA) and the solution was allowed to set for 20 min. A mixture of 50% toluene.1 Free glycerin was measured using gas chromatography with an Agilent 6890N gas chromatograph. An automated method was developed to determine the acid value using the Metrohm autotitrator. After classification was completed. Air pockets are also removed by classification. Classification of the resin bed caused the resin to be distributed with larger resin beads on the bottom and smaller beads on the top. the biodiesel was drained until it was level with the top of the resin bed. Raw biodiesel was then pumped into the column.3 Autotitrator Figure 4. Dry resin was weighed and poured into the column.5 hour intervals for flow rates of 2 to 4 bed . Clean biodiesel or methanol was pumped into the outlet of the column to classify the resin bed. several tests were developed.4 Gas chromatograph 4.26 Figure 4. Resin life tests were developed to determine how the resin performs as it becomes exhausted.6 Experimental Procedures In order to evaluate the effectiveness of packed bed media such as ion exchange resins. 1000 ml samples were taken at 1 to 2. To determine the effectiveness of purification of biodiesel with adsorbents such as magnesium silicate or diatomaceous earth.02% weight) to be exceeded. Methanol washing involved passing 5 to 10 bed volumes of methanol through the resin bed to remove the glycerin bonded to the resin.02%. the resin bed was washed with clean methanol to regenerate the resin bed. Samples were analyzed to determine the effectiveness of purification at intervals of 5 to 10 liters. acid value and free glycerin level were measured before and after passing through the bed. Soap levels were monitored to determine when 50 ppm soap in the product was reached. Soap levels above 50 ppm cause the biodiesel ASTM specification for sulfated ash (0. the soap level. When free glycerin levels leaving the resin bed exceeded the ASTM specification level of 0. the acid value of the outgoing fuel was measured. The effect of flow rate and bed height was measured by varying the bed height from 1 to 3 inches in several columns. For each column. The effect of processing biodiesel with ion exchange resin on the acid value was determined by passing biodiesel through the resin at several soap levels. The column was run until soap levels were well above the specification levels. Samples were taken at each flow rate and soap levels were determined. To determine the effectiveness of purification. For each soap level. the mass of the adsorbent was .27 volumes per hour. the flow rate was varied from 1 to 7. Acid values were also measured to determine the amount of soap removal due to ion exchange in the resin. This washing was performed at flow rates ranging from 2 to 4 bed volumes per hour.5 bed volumes per hour. the soap and glycerin levels were measured. and for each treatment level. The experimental setup was similar to that used in ion exchange resin experiments. .28 varied. 5. ion exchange and purification by the interaction between soap and glycerin. In this experiment.1.1 shows the effect of filtration on soap removal.1 Filtration Mode The first mode that was tested was soap removal by filtration.1 Four Mode Theory Validation Experiments were conducted to validate the four mode theory of soap and glycerin removal from biodiesel which was discussed in chapter 3. Results and Discussions This chapter discusses the experimental results of tests conducted to validate the four mode theory of soap and glycerin removal. the biodiesel feedstock contained no methanol or glycerin and was at room temperature. Figure 5. . The four mode theory states that soap and glycerin can be removed by a combination of filtration. The performance of adsorbents. the particle retention size of the filters was varied from 2 to 40 µm and the soap levels after filtration were found to vary from 600 to 1200 ppm. The sodium-catalyzed biodiesel feedstock contained 1500 ppm of soap and the potassium-catalyzed biodiesel feedstock contained 1300 ppm.29 CHAPTER 5. In this initial test. and ion exchange resins for their ability to purify biodiesel are compared. 5. filtration media. It also describes the operating parameters needed to achieve optimal performance from biodiesel purification products. physical adsorption. 30 The resu of this e ults experiment show that as the filter particle re t etention size e incre eases, the amount of soap in the effluent als increase This fig e so es. gure compa ares the f filterability of sodium a potassi o and ium soaps. It can be s seen that sodium soap p can be filtered more effec ctively than potassium soap. Filter Proper rties Particle Rete ention (μm) 40 0 22 2 11 1 8 2.5 5 Typ pe Qu ualitative Qu ualitative Qu ualitative As shless Coffee Whatman #4 Whatman #1 Whatman #2 2 Whatman #42 Figu 5.1 Effec of filter p ure ct porosity on s soap remov val. A visual analysis w also per was rformed to compare th amount o soap he of particulate in biodiesel samples whic contained sodium a potassiu soaps. ch and um This experimen involved the vaporiz s nt zation of me ethanol from two biodi m iesel sampl les cont taining sodium and po otassium so oaps. Since methanol acts as a c e cosolvent to o keep soap in so p olution with the biodiesel, methan remova should ca h nol al ause soap prec cipitation. Figure 5.2 s F shows the e experimenta setup for this analys al r sis. 31 The visual analysis and filtration experiments show that there are distinct differences in the physical characteristics of sodium and potassium soaps. When methanol was removed it was found that sodium soaps formed a “gel-like” particulate layer in the biodiesel. Potassium soaps did not appear to form this particulate layer. It was found that during methanol removal, the biodiesel samples with sodium soaps generated foaming, whereas, the potassium soaps did not. This foam is likely formed by methanol vapors bubbling through the sodium soap particulate layer. From the visual observations of the differences between sodium and potassium soaps in biodiesel and the soap filtration data, in figure 5.1, it appears that sodium soap is less soluble in biodiesel than potassium soap. Figure 5.2 Methanol removal biodiesel containing sodium and potassium soap (Left: Potassium, Right: Sodium). 32 5.1.2 Ion Exchange Mode Experiments were conducted to determine the quantity of soap that is removed to due to the ion exchange mode. An increase in free fatty acid (FFA) levels when soap is removed indicates that the sodium portion of the soap molecule is being held by the resin, releasing the FFA into the effluent biodiesel. Figure 5.3 shows the effect of soap removal, by ion exchange, on the effluent biodiesel free fatty acid levels. The incoming soap level was varied from roughly 500 ppm to 3500 ppm. Free fatty acid measurements were taken on the effluent biodiesel at each soap level. From this figure it can be seen that as the crude biodiesel soap level is increased, the effluent biodiesel free fatty acid value also increases. Three ion exchange resins, Amberlyte BD10 by Rohm & Haas and T45BD and T45BDMP by Thermax, were tested. In this experiment, the experimental data were compared to the theoretical 100% ion exchange curve. The 100% ion exchange curve is the case where all of the soap entering the column is converted to FFA. For each ion exchange resin, complete soap removal was achieved but 100% conversion from soap to free fatty acid was not achieved at soap levels greater than 1000 ppm. At soap levels below 1000 ppm the experimental data shows close to 100% ion exchange. At soap levels from 1500 to 3500 ppm, the experimental curve and the 100% ion exchange curve diverge. These data show that at lower soap levels, soap removal is performed primarily by ion exchange. At higher soap levels, soap removal is performed by a combination of ion exchange, filtration and adsorption. 33 ure ct sel vel Figu 5. a 9 90% confide ence interva was al calc culated. To ensu that the difference between th experime ure he ental and 100% ion exch hange curve was statis stically sign nificant. . nce e ntal lls belo the 100% ion excha ow % ange curve e. Fig gure 5.3 Effec of biodies soap lev on acid value. From t this statistic cal analysis. it can be assume with 90% confiden that the experimen data fal ed.4 sho ows that the lines corr responding to the uppe and lowe er er conf fidence inte ervals fall below the 10 00% ion exc change cur rve. 1 Soap Only Figure 5. anol and no o .2 Ion Exc change Res Exhaus sin stion Tests we conduc ere cted to dete ermine the ion exchang resin’s e ge effective life the e. optim operating parame mal eters and the effect of t four mo the odes over th resin life he e. Res exhaustion tests we conducted for biod sin ere diesel feeds stock containing soap only glycerin only and a c y. g esent. In this test. 5.4: Stat 5.5 shows th resin soap exhaust 5 he tion curve f Thermax T45BD gel for x’s resin with no glycerin pre n. ults ese tests were com s mpared to th performa he ance of othe competit er tive product ts.2. 5% metha sel pm . o combination of soap and glycerin The resu from the n n.34 ure tistical analysis of experimental d data. a co olumn with 200 ml of s swelled resi in was used to pu s urify biodies containing 2000 pp of soap. Figu 5. the effluent free fatty acid was also measured for samples up to about 500 BV. was about 900 ppm. the FFA was measured to be about 2100 ppm and at 500 BVs the FFA measured 600 ppm. These regions were from 0 to 225 BVs and 225 to 500 BVs. It can be seen that the soap level is low from 0 to 225 BVs but the FFA values decrease. The effluent soap level. this trend shows that as the resin bed ages ion exchange gradually decreases and adsorption contributes more to soap removal. after 500 BVs. In figure 5. From 225 to 500 bed volumes the soap level begins to rise in the effluent biodiesel and the FFA continues to drop. Biodiesel was passed through the column until about 500 bed volumes (BV) were processed. The effluent biodiesel soap level exceeded 50 ppm of soap after about 250 BVs.35 glycerin. over this range of bed volumes. The soap and acid curves in figure 5. This shift in purification mode is likely due to the ion exchange sites being covered with soap particulate. adsorption and filtration capacity are being exhausted. allowing soap to pass with the effluent biodiesel. .5 were compared to determine the effect of the four mode theory over the life of the resin. this indicates that ion exchange. This test was conducted using a flow rate of 4 bed volumes per hour. The small decrease in soap level at 100 BVs and 350 BVs and the acid value at 100 BVs was due to a fluctuation in flow rate and does not affect the overall results of this experiment. There were two distinct regions where the mode of soap removal changed. significantly. At 50 BVs.5. the soap remov occurs in two phas over the life val ses e of th resin bed As the re he d.2 Glycerin Only 2 n Figure 5. ed lead ding to complete bed e exhaustion. As the ion exc change site es. T ion exc The change gra adually decr reases due to the activ ve sites being cov s vered with s soap particu ulate. soap re emoval mo shifts fro ion ode om exch hange to ad dsorption.2. filtra ation and ad dsorption gradually be ecome exha austed. biodies containi 0. adso orptive pore and bed filtration ar are fille more soap is allowe to pass. esin bed ag ges.5 Effec of resin a on soap removal (R ure ct age p Resin: T45B BD). es rea ed. To summ marize figure 5.5. 5. In the second p period.08% f sel ing free . ion e exchange. test.6 shows th resin ex 5 he xhaustion cu urve of the T45BD gel resin for glyc cerin removal with no s soap presen In this t nt.36 Figu 5. as the re esin bed ag ges. Th effluent biodiesel fr glycerin level was measured o he ree n over the life of the re esin. Figu 5. s ot . Accord ding to conv versations with the Th w hermax Com mpany. 4% methanol and no soap w purified with 200 ml of swelle resin. .37 glyc cerin. at a relatively c constant rate. it c be seen that the e can effluent glyc cerin level increases grad dually. it is generally u understood that glycer is rin removed by ad dsorption in the ion exc change resin.6 Glyc ure cerin removal with no s soap presen (Resin: T nt T45BD).6. From fig gure 5. B Biodiesel wa purified f about 31 as for 16 bed volumes and then the resin bed was washe with met e ed thanol to test its rege eneration ca apacity.5 B w BV/hour. this is su upported by free fatty acid y mea asurements that did no increase. T d was d ed The flow rate for this test was about 3. 2. from different manufacturers.7 shows the resin exhaustion curve for soap removal and the corresponding free fatty acid value.2. 5.3 Soap and Glycerin Combined In normal production. A soap level of 50 ppm was exceeded at about 550 BV. the soap and FFA values were 27 ppm and 1800 ppm.4 Macroporous Resin Test Figure 5.08% to 0.02% by washing with methanol. Tests were conducted to determine the resin life and active modes for biodiesel containing soap and glycerin.6 shows that washing the ion exchange resin with methanol at 316 bed volumes. the soap and FFA values were 1070 ppm and 1260 ppm. at a flow rate of 2. glycerin levels were reduced from 0.5 BV/hour. respectively. Figure 5. These tests compared gel and macroporous resins. 0. At 50 BV. for Thermax’s T45BDMP macroporous ion exchange resin. after glycerin exhaustion. .08% glycerin and 4% methanol was purified by 180 ml of swelled resin. crude biodiesel containing 1200 ppm of soap. At 1000 BV. to determine if resins from different manufacturers perform differently. can reduce effluent glycerin levels. In this figure. For this test. respectively. crude biodiesel contains both soap and glycerin.38 During this experiment the ability to regenerate the ion exchange resin for glycerin removal was tested. This test was conducted until about 1000 bed volumes had passed through the resin at which time the incoming and outgoing soap levels were approximately equal. 5. The crude biod diesel in figu 5. . Basically. Compar ring the soa exhaustion curve in figure 5. it can be see that similar soap removal char en racteristics were achie eved. To c compare the curves the initial s ese soap level m must be tak into con ken nsideration.7 to the 0% g ap n glycerin cur rve in fig gure 5. From these figur res. if the soap level is doubled the numbe of bed c er umes neede to achiev a given s ed ve soap level a appears to be halved.8. volu The rela ationship be etween the soap remo oval and free fatty acid values for e purif fication of crude biodie c esel contain ning soap a glycerin follows ro and n oughly the s same . where eas.5. it can b seen tha the be at 5. in figure e ppm was used. a soap level of 1200 p 50 p ppm soap le evel is reac ched at 250 BV for the 2000 ppm case and a 550 BV f at for the 1200 ppm case.7: Soa removal f ure ap from biodie esel contain ning soap an glycerin (T45BDMP nd P).5 con ure ntained 2000 ppm of so oap.39 Figu 5. This region corresponds to the shift from ion exchange to adsorption that occurs as the resin ages. the FFA curve has a smaller slope and the final FFA value is around 1300 ppm.40 pattern as the 0% glycerin case. .7. adsorption and filtration are all becoming exhausted and soap is allowed to pass into the effluent. In figure 5. The glycerin removal tests were also performed with the macroporous resin. These differences indicate that when glycerin is present. This region is the phase where ion exchange. the ion exchange sites may become blocked off during the purification process reducing the ion exchange action. Glycerin removal was not found to be effective compared to the gel resin tests. From figure 5.7 (soap and glycerin) is the slope and final value of the free fatty acid curve.7. Macroporous resins are used in industry for biodiesel purification and are claimed to be effective. From 550 to 1000 BV. In figure 5. it can be seen that most of the soap is removed but the acid value gradually decreases as the influent increases from 0 to about 500 BV. An important difference between the ion exchange resin exhaustion curves depicted in figures 5. the FFA curve has a much steeper slope and the final FFA value is around 500 ppm.5 (soap only) and 5. the FFA value continues to drop and the soap level climbs rapidly. Additional experiments should be conducted to whether verify glycerin removal occurs with macroporous resins.5. The fluctuations in FFA value occurring at 50. 580 and 650 BV volumes are due to a decrease in flow rate leading to more efficient ion exchange causing an increase in FFA. The resins of the two manu r e ufacturers p performed in a similar manner for this n r regio of the cu on urve. the ef e ffluent biodi iesel soap level was m measured fo samples up or to about 700 BVs.41 5. for com 10 mbined soap and glyce remova In this te biodiese p erin al. T45BD). T methan The nol . Figu 5.8. At abou 400 bed volumes.2. e ut each resin bed was washed wit methano to test the resin’s ability to be re n w th ol e egenerated for d glyc cerin removal. a m 00 methanol wa was co ash onducted. From 0 to 400 BV the soap level gradually increa Vs p ased to abo out 50 p ppm.5 Parallel Gel Resin Test 5 n Figure 5.8 shows th soap ex 5 he xhaustion cu urve of the T45BD and Amberlyte d e BD1 gels resins. by 200 ml of s w B swelled res sin. at a flow rate of 2 BV/hour.08% gly d ycerin was purified. At 40 BVs. In figure 5.8 Com ure mparison of gel resins f soap rem for moval (BD10. 4% me ethanol and 0. el cont taining 1000 ppm of so oap. est. therefore. by approximately half. From these figures. Figure 5. contained soap levels of about 2000 ppm. Soap particulate was dislodged and leached into the effluent biodiesel raising the soap level during the 400 to 600 BV region of the Amberlyte BD10 curve. it appears that the presence of glycerin in the crude biodiesel does not affect the resin’s soap removal life. for a given effluent soap level. 4% methanol and 0.8. it can be seen that doubling the soap level of the crude biodiesel entering the column will reduce the number of bed volumes that can be treated. depicted in figure 5.8 it can be seen that the resin exhaustion curves are of roughly the same shape. The effluent biodiesel soap level exceeded 50 ppm at about 400 BV in this test. biodiesel containing 1000 ppm of soap.5. the crude biodiesel contained about half the soap level as was used in figure 5.8.42 wash was performed at too fast of a flow rate on the Amberlyte BD10 column causing the column to partially fluidize and become unstable. The crude biodiesel used in the soap-only test. The crude biodiesel used in the soap and glycerin combination test. From this comparison. is similar to the soap exhaustion curve from biodiesel containing only soap.7 and 5. about double the bed volumes are needed to achieve the same effluent soap level. In figures 5. the column stabilized and the performance of the two resin columns became similar again.7 and 5.08% . After 600 BV. 5. The effluent biodiesel soap level exceeded 50 ppm at about 225 BV in this test. The resin soap removal exhaustion curve from crude biodiesel. In this test.5. Comparing figures 5.8. depicted in figure 5. contained soap levels of about 1000 ppm.9 shows the glycerin exhaustion curve from the same test as figure 5.5. containing both soap and glycerin. . e each resin b was wa bed ashed with methanol to test the resin ability to be regene n’s o erated for g glycerin rem moval. In figure 5. The glyc cerin values after the m s methanol wa are not typical due to incorre wash ash t e ect proc cedures.6. r of 2 to 5 bed vo Figu 5. it can be see that the en methanol wash flow rate s h should be r roughly the same as th biodiesel run flow ra he ate olumes per hour. the glycerin level was meas e sured from 17 to 700 B BV. It was found that the flow rate o methanol during rege of eneration h a significant has effec on the ef ct ffectiveness of washin From fig s ng.9.02% was exceeded at about 2 f d 200 BV.9: Com ure mparison of gel resins for glycerin removal (B f n BD10. at a flow rate o 2 BV/hour.43 glyc cerin was pu urified. The ASTM specificat M tion level of 0.01% at s m % to about 150 BV. At of ml d about 400 bed volumes. gure 5. T45B BD). Effluent glyc cerin values varied from about 0% at 75 BV t 0. by 200 m of swelled resin. Figure 5. is 1 kg of resin is needed to purify 1600 kg of biodiesel.3 Ion Exchange Resin Operating Parameters Tests were conducted to determine the proper flow rate and bed aspect ratio parameters to achieve optimal performance of the ion exchange resins. The crude biodiesel that was used for this test contained 1000 ppm of soap. soap removal decreased and as the bed height increased.44 The glycerin exhaustion curves of the two gel resins were of roughly the same shape and their performance can be considered comparable. the flow rate was varied from 0 to 250 unswelled bed volumes per hour. for unswelled bed heights of 1 to 3 inches in 0. 5. This discrepancy might be due to differences in the expectations for quality of the purified biodiesel since these are not stated by the manufacturers. it was found that. The exact amount of glycerin was not measured.5 inch increments.10 shows the effect of flow rate and bed aspect ratio on soap removal. 5% methanol and glycerin. the soap removal is roughly the . For this test. It can be seen that for this range of flow rates and 2-3 inches bed heights. In this study. at an initial soap level of 1000 ppm. the soap removal increased. These parameters were different due to variability in the flow control valves and differences in methanol evaporation between the two columns. 1 kg of resin is needed to purify 500 kg of biodiesel. at an initial soap level of 1000 ppm. The manufacturer’s estimate of the life of the ion exchange resin. This data shows that as flow rate increased. The slight variability between the two curves is likely due to differences in the flow rate and flow dynamics through the resin bed. These data show that the ASTM specification for glycerin was not reached for the maximum flow rate tested. the glycerin breakthrough increased. Breakthrough of soap. The flow rate was varied from about 0 to 30 BV/hour. which was significantly greater than the manufacturer’s recommended flow rate of 2 to 5 swelled BV/hour.5 inches height.5 in bed heights. occurs almost immediately with the 1 inches bed height and at about 75 BV/hour for the 1. as indicated by elevated outgoing soap levels. It can be assumed that if additional tests were conducted with greater flow rates. 4% methanol and no soap was passed through 140 ml of swelled resin. it can be seen that for a dry resin bed aspect ratio of 2 to 1 or greater.11 shows the effect of flow rate on glycerin removal. From this figure.09% glycerin. It can be seen that as the flow rate increased. Figure 5. In this test. The Thermax company recommends that a flow rate of about 2-5 swelled or 4 to 10 unswelled BV/hour be used. the flow rate should be kept below 100 bed volumes per hour to obtain outgoing soap levels of below 50 ppm. This data shows that the manufacturer’s flow rate recommendations are conservative.45 same. crude biodiesel containing 0. breakthrough would occur in the larger bed heights in the same way it occurs in the 1 and 1. It was not possible to run these tests with the current equipment because it would have required pressurizing the bed and this was not possible. . 11 Effe of flow r ure ect rate on glyc cerin remov (Resin: T val T45BD). .46 Figu 5.10 Effe of flow r ure ect rate and asp pect ratio o soap rem on moval. Figu 5. 47 5. the weight percentage of purification media was varied from 0. such as Magnesol in removal of soap and glycerin from biodiesel. .1% weight. 0.1%.08% free glycerin and no methanol. the soap level was consistently about 30 ppm. The filtered biodiesel contained soap and glycerin levels of 700 ppm and 0. The details of this experimental setup were discussed in more detail in chapter 4. Crude biodiesel. The effluent soap level gradually decreased as the weight percentage was increased from 0. the crude biodiesel was passed through a paper filter to determine the filter’s ability to removal soap and glycerin. at 60°C. the soap level was consistently about 70 ppm.1 to 5% and the soap and glycerin values were measured. From 1 to 5% weight. Before testing the adsorbents. was treated with Magnesol and diatomaceous earth. respectively.06%. Diatomaceous earth is not supposed to have the ability to attract or adsorb compounds but merely acts as a depth filter as it accumulates on the surface of the media. the commercial magnesium silicate adsorbent media. it can be seen that Magnesol.4 Adsorbent Media Effectiveness Tests were conducted to determine the effectiveness of adsorbents. From 2 to 5% weight. For this test. Magnesol reduced the effluent soap levels to 100 ppm at 0.12. Magnesol is specially treated by the manufacturer to have the ability to adsorb polar compounds. containing 1200 ppm of soap. removed soap to a lower concentration than diatomaceous earth. Diatomaceous earth reduced the effluent soap levels to about 375 ppm at 0.1% to 2%. In figure 5. 04 d 45% to unde etectable le evels as th Magnes percenta was inc he sol age creased from 0.1 to 0.1 to 2% From 2 to 5%.48 Figu 5. The e appa arent increa in glyce level fr ase erin rom 0.12 Soa removal by adsorpt ure ap tion and filtr ration.05% at 0. In figure 5.05 to 0. the %.03% Increasi the perc 5% %. Diatoma aceous eart reduced effluent gly th ycerin levels to 0.5% was due to experimental e error and should not be conside t ered signific cant. From 0.25 to 0. glyc cerin level was below t detectio threshold and consi w the on d idered to be 0. The gly ycerin level decreased from 0.03%. ing centage of diatomaceo earth fr ous rom 0. M us Magnesol re educed the effluent gly ycerin levels to 0. eases in gly ycerin levels at s .01 1%. The incre . it can be seen t e that Magne esol is more effective a removing e at g glyc cerin than diatomaceou earth.13.04 45% at 0.1% s ght.005%.5% of diatoma 0 % aceous eart the glyce levels r th erin reduced fro om weig 0.5% to o 5% did not reduce glycerin levels below 0. Figure 5. by re educing its free fatty acid value. T 5 of ol cid This figure corr responds to the same test as figu o ures 5. the Magne nd or esol perc centage is varied from 0.49 1 an 2% were proably du to exper nd e ue rimental err and should not be c ror considered sign nificant. Th figure shows that M 00 his Magnesol c have a significant can ct ving the qua ality of biod diesel.13. All of the testing in this . Fo this test.12 an 5. effec in improv Figu 5.5% to 5% and the F v % FFA over th range de his ecreased fr rom about 800 to 30 ppm.13 Gly ure ycerin removal by adso orption and filtration. It was foun that Mag d ous nd gnesol can a also biod sign nificantly red duce biodie esel free fat acid valu tty ues. The com mparison an evaluatio of Magnesol and di nd on iatomaceou earth fou us und that Magnesol is significan more e ntly effective at r removing soap and gly ycerin from m diesel than diatomaceo earth.14 shows the effect o Magneso on the ac value. 5 ppm to 22 ppm for the ge resin and up to 35 p ur om m m el d ppm for the macropo orous resin after 5 BV After 24 BV. out .5 ppm.5 Sulfur Table 5. Lowe temperatu er ures would likely produce better soap removal due to the so b p oaps being less soluble in the g biod diesel.1 he n change resins and Mag gnesol on th biodiese sulfur leve he el els.50 expe eriment was conducte using bio ed odiesel at 60 0°C. shows th effect of purification by ion exc . 5. em o Figu 5. n e not mak a signific ke cant change in sulfur le e evel.14 Red ure duction in a acid value b processin with Mag by ng gnesol. thus making the easier to filter. Ion exchange resin prod ct r n e duced a sign nificant incre ease in sulf level in t initial fu process fur the uel sed. the ion exchange resin did n n V. Magne esol purifica ation did no have a ot sign nificant effec on sulfur levels. sulfur levels returned to abo 12. The ion exchange resin e incre eased sulfu levels fro 12. 26 Macroporous Resin (24 BV) 12. Sample Sulfur (ppm) Crude Biodiesel 13.05 Macroporous Resin (5 BV) 35. This sulfur residue is a by-product from the sulphonation process used in the manufacturing of the ion exchange resins.29 Magnesol 10. therefore. Biodiesel sulfur levels are high in the low BV period due to sulfur residue leaching from the resin.02 . Sulfur levels reduce after all the sulfur residue is leached from the resin.11 Gel Resin (5 BV) 22. statistical analysis using the SAS software package was performed. Using a 95% confidence level it was found that each sample had a significant difference in sulfur level as compared to the crude biodiesel except for the Macroporous resin after 24 BV. it can be assumed that ion exchange resins do not provide a significant change in sulfur levels after 24 BV.51 In order to determine if the differences were significant between each purification method and the crude biodiesel. Table 5.1 Effect of purification media on biodiesel sulfur level. The procedure used for this analysis was the t test (LSD) Least Significant Difference. Two tests were performed to validate the four mode theory of soap and glycerin removal. Ion exchange resins were evaluated to determine the modes that are active in their operation. Ion exchange resin exhaustion curves were generated for gel and macroporous resins from two manufacturers. only glycerin and a combination of soap and . Results supporting the four mode theory of soap and glycerin removal were presented and discussed. It was found that soap can be removed by a paper filter and that increasing the filter’s fineness produces more soap removal. This test also showed that as the crude biodiesel soap level is increased the effluent free fatty acid value increases. soap is removed by a combination of ion exchange.1 Conclusions This chapter discussed the purification of biodiesel with ion exchange resins and adsorbents. These curves were generated for crude biodiesel containing only soap. CONCLUSIONS 6. adsorption and filtration. It was found that at low crude biodiesel soap levels ion exchange resins remove soap predominately by the ion exchange mode. Ion exchange resin exhaustion curves were generated for the gel and macroporous resins. At higher soap levels. Tests were conducted to determine the operating parameters such as the bed aspect ratio and flow rate required to produce optimal performance.52 CHAPTER 6. The performance of adsorbents were evaluated and compared. Sodium soaps were found to be more filterable than potassium soaps. First. the effluent free fatty acid value decreases and the effluent soap level increases.53 glycerin. . adsorption and filtration all gradually become exhausted leading to more soap passing into the effluent biodiesel. High methanol flow rate does not allow time for the glycerin to be desorbed into the methanol. Second. upon reaching the incoming level. ion exchange gradually decreases and combines with adsorption. If the crude biodiesel soap level is doubled. Gel resins from Rohm and Haas and Thermax were found to have similar performance for soap and glycerin removal. Glycerin was found to be removed by a combination of adsorption and filtration. The shape of the soap exhaustion curve is roughly the same. as the ion exchange resin ages. Glycerin removal capability can be regenerated by washing the ion exchange resin with methanol. the effluent glycerin content levels off. Glycerin levels increase at a relatively constant rate over a glycerin exhaustion curve. It was found that. Gel and macroporous resins were found to provide similar performance for soap removal. It was found that there is a linear relationship between the crude biodiesel soap level and the number of bed volumes that can be processed to reach exhaustion. ion exchange. The recommended methanol flow rate for glycerin regeneration is 2-5 BV/hour. This process was found to be sensitive to high methanol flow rate. regardless of the crude biodiesel glycerin levels. the exhaustion bed volume is halved. It was observed that there are two phases of soap removal that occur sequentially during the exhaustion of a resin. Tests were conducted to compare ion exchange resins of different types and different manufacturers. As the bed aspect ratio increased. It was found that paper filters alone. Magnesol was found to have a small effect in reducing crude biodiesel sulfur levels. It was found that as the flow rate increased. Biodiesel methanol concentration was found to be a difficult variable to control due to the volatility of methanol. After about 20 bed volumes. can reduce soap levels significantly. such as diatomaceous earth. For the range of flow rates tested it was found that aspect ratios from 2 to 1 through 3 to 1 performed similarly. 6. Tests were conducted to determine the effect of biodiesel purification products on biodiesel sulfur levels. Magnesol and saw dust.2 Recommendations for Future Work Future studies on this topic should develop methods to obtain better control of process variables. A new column design . Magnesol can reduce the free fatty acid value of biodiesel.54 Tests were conducted to determine the bed aspect ratio and flow rate to obtain optimal performance. The columns and biodiesel reservoir should be designed to minimize methanol leakage. the effluent soap and glycerin levels increased. this value decreases to the crude biodiesel level. in removing soap and glycerin. the effluent soap levels decreased. Tests were conducted to evaluate and compare the effectiveness of adsorbents and filtration media. Saw dust was also found to be effective in removing soap and glycerin from crude biodiesel. Magnesol is more effective in removing soap and glycerin than diatomaceous earth. Bed aspect ratio has a significant effect on the optimal flow rate through the ion exchange resin bed. When ion exchange resins are first used the initial biodiesel contained increased sulfur levels. glycerin and methanol levels. Soap and glycerin measurements should be performed at least three times and averaged to obtain sound values. . Samples were found to change over time when stored. Additional tests should be conducted to compare purification performance using potassium and sodium catalyst. Soap and glycerin measurements should be taken as soon as possible after processing with the purification media. The effects of temperature should also be tested on resin performance and life. Tests should also be conducted comparing several different initial soap.55 should be used that incorporates flow rate control using a feedback loop and temperature control. 2003. 83(5): 457-460. Prog. Cooney. Soc. H. 84: 385-392. Am. Upper Saddle River: Prentice Hall. 1966. and Skelton. and Hanna M.T. 1998. Adsorption Calculations and Modeling. T. Arden. R. J. Transactions of the ASABE 46(4): 945-954. . M. Sci. Osborn. 1994. 1956. Adsorbing it All. Ma. He. K. 2004.56 REFERENCES Albright. Comparison of Membrane Extraction with Traditional Extraction Methods for Biodiesel Production. Tom. 87: 3522-3528. and Yarnell. Synthetic Ion-Exchangers. Z. Inczedy. Bruice. M. Emission Data on 2-stroke and 4 stroke Engines-Combustion of Fat and Vegetable Oil Derived Fuels in Diesel Engines. 1968. Chemical Engineering Journal.. Chi. P.H.L. Boca Raton: Lewis Publishers.V. et al. 2008. Am. New York: The Macmillan Company. 144: 459-465. Oil Chem. G. Water Purification by Ion Exchange. Boston: ButterworthHeineman Series in Chemical Engineering. Bryan. Biodiesel Magazine March. Analytical Applications of Ion Exchangers. David. J and Canakci.S. 2008. Berrios. 70: 1-15 McCormick. Paula. Ethanolysis of Refined Soybean Oil Assisted by Sodium and Potassium Hydroxides. Oxford: Pergamon Press. 2006. Bioresource Technology.Y. Organic Chemistry. A Pilot Plant to Produce Biodiesel from High Free Fatty Acid Feedstocks. Comparison of Purification Methods for Biodiesel.L. Soc. Biodiesel Production: A Review. Ion Exchange Polymers: Encyclopedia of Polymer Science and Engineering.L.A. Kucek. and Graboski. 1999. The Production of Biodiesel from Waste Frying Oils: A Comparison of Different Purification Steps. Adsorption Design for Wastewater Treatment. Energy Combust. J. 24: 125-164. Tien. Van Gerpen. M. Oil Chem. 2007. New York: Wiley. 1987. F. 2008.J. R. Fuel. New York: Plenum Press. R.A. Predojevic. 1998. J. Fuel Processing Technology. 2007. Biodiesel Processing and Production.57 Van Gerpen. Deglycerolization of Biodiesel Streams by Adsorption Over Silica Beds. Energy & Fuels. Yoris. Fuel Processing Technology 86(10): 1097-1107. Wang.C. 21: 347-353. Adsorptive Properties of Silica Gel for Biodiesel Refining. Refining of Biodiesel by Ceramic Membrane Separation. 2009. 22: 4281-4284. J. J. 90: 422-427. 2008. 2005. . Yoris.C. Y. J. Energy & Fuels. et al. The saw dust used in this experiment was fine saw dust obtained from a cabinet shop. .58 APPENDICES A. Saw dust is able to remove soap and glycerin from crude biodiesel due primarily to filtration. up to a biodiesel to saw dust ratio of about 45 gram biodiesel to 1 g saw dust. Depth filtration is the particulate deposition throughout the volume of multiple layers of filtration media.1 shows the effectiveness of saw dust in removing soap from crude biodiesel. fine saw dust acts as an excellent surface filter and depth filter. Crude biodiesel containing 2000 ppm of soap. Saw dust tests Figure A.6 g biodiesel per g saw dust. An additional large scale test was conducted and it was found that sufficient soap and glycerin removal was achieved for about 14. Surface filtration is the particulate deposition on the top surface of a filter. When packed in a column. 4% methanol and no glycerin was passed through the saw dust and the outgoing soap level was measured. This experiment found that soap was reduced to levels below the detection limit of about 10 ppm. 1 Soap removal with saw du ure p ust .59 Figu A. Soap: 500 500 10 1 304.643 10 0.00463 1. methanol and glycerin should be added to obtain 500 ppm of soap and 0. Soap and Glycerin Addition Calculations Problem Statement: How much sodium methoxide. use 25% sodium methoxide (NaOCH3).9 10 0.05% glycerin? Assumptions: Initial soap and glycerin levels are 0.4.60 B.25 NaOCH 1 54 NaOCH NaOCH 1 1 NaOCH 0.4 1.00463 354. MW of sodium soap = 304.355 Glycerin: % 100 .643 10 0. 1 entry. Autotitrator Procedure 1. Starting sample position Run calibration daily 1. method “calibration” 2. Label i. Sample table i. ID2. Menu/enter/sample table b. Prepare samples a.Test iv. Test (ie. Sampler a. Add acetone 2. Menu/manual control/dosing . Weigh fuel into cup c. Titrator a. 3 samples Purge buret-daily or when changing titrant (HCL or KOH) 1. Sample number iii. Start b. Method-Soap or Acid ii. Weight c. Tirator a. Number of samples ii. Sample table i. Weight ii. ID1 – Sample Number iii. Put sampler on empty cup 2. Esc to main screen 4. Titrator a. Starting titration a. Sampler a. Press start on sampler i. T45BD) b.61 C. Put cups into sampler 3. Step 2: Add 100 μL of S1. Step 6: Name vials in desired sequence in GC program. S2. Step 5: In GC program. GC Procedure Step 1: Measure 0. . MSTFA solutions then wait 20 minutes. Step 3: Add 8 ml of heptane to sample bottle.1 g of biodiesel into sample bottle. go to Sequence/Sequence parameters/directory and select appropriate directory. Step 4: Pipet 1000 μL of prepared sample into GC sample bottle.62 D. 7 and 10.0% and 100%. Sample Titrator Calibration Calibration was performed each day the titrator was used. Figure E.63 E. The titrator was considered to be calibrated when the Slope was between 96. Three buffers were used with pH values of 4.1 Sample titrator calibration . Sample GC Calibration GC calibration was performed each time the GC column was replaced.64 F. . 2. . Resin should become fluidized. Fill column with crude biodiesel. Pump methanol or clean biodiesel into the outlet of the column.65 G. 4. 3. 1. Allow resin to settle and drain the methanol or clean biodiesel until it is just covering the top of the resin bed. Ion Exchange Resin Bed Start-Up Procedure Weigh resin into column. Run several bed volumes through column to remove methanol before taking measurements. Drain methanol unit it is just covering resin bed. 1. 3. Fill column with methanol. Fill column with crude biodiesel and continue processing.66 H. Allow several bed volumes to pass before taking measurements. Pass about 5 BVs of methanol through the column at a flow rate of 2 to 5 BVs per hour. Methanol Wash Procedure Drain crude biodiesel until it is just covering the resin bed. 2. .