Biodiesel fuels have been in commercial use in many countries for up to 20 years. Production and utilization is growing in almost every country. Biodiesel use has been rapidly growing in the United States due to high crude oil prices and the desire to reduce dependence on foreign oil while improving the environment and providing opportunities for American agricultural products.
Biodiesel is a domestic, renewable fuel used in diesel engines and derived from natural fats and oils such as soybean oil. It must meet the specifications of ASTM D 6751, "Biodiesel Fuel Blend Stock (B100) for Middle Distillate Fuels
1." According to ASTM D 6751, biodiesel is defined as "a fuel comprised of mono-alkyl esters of long chain fatty acids derived from vegetable oils or animal fats." Under D 6751, biodiesel, or B100, which meets this specification, may be used in blends with petroleum diesel at up to 20 percent (B20) by volume.
The use of B20 and lower blends made with D 6751-grade B100 has been largely trouble-free since the onset of the biodiesel industry in the United States more than 10 years ago. However, unexpected filter plugging took place in Minnesota during the winter of 2005-2006 in vehicles using B20 or lower biodiesel blends, even though the cold flow properties (cloud point and cold filter plugging point) of the biodiesel did not indicate a filter plugging potential, according to the "Remember December" article in the October 2006
Biodiesel Magazine.
Our initial testing of material recovered from plugged filters identified a preponderance of saturated monoglycerides in the organic component. In response to the filter problems encountered in Minnesota, a cold incubation ("cold soak") filter test was instituted for B100 used in making biodiesel blends in the state. The cold soak filter test consists of chilling the sample for a specified time and then allowing it to warm to room temperature prior to filtering in a modified ASTM D 6217 "Particulate Contamination in Middle Distillate Fuels by Laboratory Filtration" procedure
2. Although not part of the official ASTM D 6751 requirements for B100, the cold soak filter test has become commonly used in the industry as a measure of filterability. In addition, some companies have also implemented other filter tests, such as ASTM D 2068 "Filter Plugging Tendency of Distillate Fuel Oils
3."
Statistically designed experiments are powerful tools for the characterization of complex systems. Through proper application of established statistical design methods, they allow the experimenter to obtain the most information from the smallest possible number of experimental runs. The goal of a statistical design is to develop a polynomial model that adequately describes the effect of the variables on a result of interest. The model can be used to provide an estimation of the effects of components within the range tested without having to test every possible combination of the components. In addition, the statistical model can help explain the effects of interactions which are not easily accessible by direct observation only, leading to a deeper understanding of the fundamental mechanisms of the observed effects and thus better process control and consistent product quality. Modern, commercially available statistical design packages are robust and provide guidance in proper experimental design and interpretation of results.
The effects of various components and processing residuals on the properties of finished soy biodiesel are largely unknown. To better understand the effects, a statistically designed experiment was developed and conducted to determine the influence of monoglycerides, sterol glucosides, soaps and water on the cloud point and filterability after a cold soak treatment. The range of concentration for each component was chosen to reflect possible expected values found in commercial soy biodiesel meeting D 6751 specifications. The influence of these components reported for soy biodiesel may not describe their influence in biodiesel made from other feedstocks due to differences in composition.
Table 1. Combinations of components added to distilled soy biodiesel versus cloud points and filter times obtained. Mono=Monoglycerides; SG=Sterol glucosides
Materials and Methods
Distilled soy methyl ester biodiesel (Steposol SB-D, provided by Stepan, Northfield, Ill.) was obtained and analyzed to confirm the absence of sterol glucosides, sodium/soap, diglycerides, triglycerides and free glycerol. Only a trace (0.02 percent or 200 parts per million (ppm)) of monoglycerides was found. The distilled soy methyl ester was dried under vacuum to remove any traces of water, and spiked with measured quantities of monoglycerides, sterol glucosides, soaps and water (Table 1).
Monoglycerides were prepared from soy oil. Refined, bleached and deodorized soybean oil (provided by Archer Daniels Midland Co., Decatur, Ill.) was subjected to glycerolysis with sodium methoxide, followed by wiped-film distillation to remove methyl esters, free fatty acids, methanol and water. The purified product contained 95 percent by weight monoglycerides and a small amount of diglycerides, with a typical soy fatty acid profile.
Sterol glucosides were obtained from a filter used in the commercial production of soy biodiesel. Residue from the filter contained 21.5 percent sterol glucosides, with the remainder consisting of methyl esters (77.2 percent by weight), mono- and diglycerides (1.0 percent by weight), and water (0.3 percent by weight). The filter residue was solvent washed, filtered and dried overnight in a 70 degrees Celsius (158 degrees Fahrenheit) oven to obtain 99.8 percent by weight pure sterol glucosides. The soy sterol glucosides were dissolved in pyridine (provided by Sigma, St. Louis, Mo.) as a 1 percent weight by volume solution.
Soaps were mixed from commercially obtained materials (Nu-Chek Prep, Elysian, Minn). Sodium palmitate, sodium stearate, sodium oleate and sodium linoleate were combined in the ratio of 0.105/0.041/0.232/0.621, respectively, to simulate the fatty acid profile of soy oil. Sodium linoleate was used to represent all the polyunsaturated fatty acid soaps due to the oxidative instability of sodium linolenate. The soap mixture was dissolved in dry methanol (less than 50 ppm moisture, provided by Acros, Morris Plains, N.J.,) as a 0.5 percent weight by volume solution.
Neutral pH water obtained from the facility's deionization system.
Experimental Procedures
Four common components of soy biodiesel were chosen for this study. The components were added to distilled soy biodiesel at levels relevant to commercial soy biodiesel: 0-10,000 ppm (0 percent to 1 percent) monoglycerides, 0-40 ppm sterol glucosides, 0-40 ppm soaps and 0-500 ppm water. The levels were chosen based on previous filter test experience and the expectation that filter test responses would be meaningful compared to a time limit of 900 seconds. A constrained mixture statistical experimental design consisting of a set of 30 experiments was constructed and analyzed using commercial software (provided by Stat-Ease Design Expert, Minneapolis). The system was modeled as a quadratic equation with first order interactions. The trials were run in randomized order.
Trial batches were prepared by adding sterol glucoside and soap solutions as required to the weighed distilled soy methyl esters. The solvents from the sterol glucoside and soap solutions as well as any moisture were removed by heating the mixture under vacuum to 90 degrees Celsius (194 degrees Fahrenheit) and stirring for 20 minutes. The mixture was cooled to 70 degrees Celsius, vacuum was broken, monoglycerides were added as required and stirred for five minutes. Finally, water was added as required by means of a volumetric pipette and stirred for five minutes. After stirring, each trial batch was visually inspected to verify that the batch was clear and all components were completely dissolved at 70 degrees Celsius. Each trial batch was allowed to cool to room temperature and stored under ambient conditions prior to testing.
The cloud point of each trial batch was measured per ASTM D 5773 "Cloud Point of Petroleum Products (Constant Cooling Rate Method)
4" using a PCA-70X Automated Pour and Cloud Point Analyzer (provided by Phase Technology, Richmond, British Columbia). Samples of 0.15 milliliters (ml) at room temperature were cooled at 1.5 degrees Celsius (2.7 degrees Fahrenheit) per minute to optically determine the cloud point as indicated by the scattering of light caused by the formation of crystals.
The cold soak treatment and filtration were carried out on 300 ml samples. After mixing each trial batch, a 300 ml sample was transferred to a 16-ounce glass jar with a Teflon-lined lid and allowed to cool to room temperature. The sample was then placed in a 4.4 degrees Celsius (40 degrees Fahrenheit) bath for a 16-hour (plus or minus 10 minutes) "cold soak."
Figure 1. Modified ASTM D 6217 filtration method
Cold soak filter times were measured using a modified ASTM D 6217 filtration procedure. After the 16-hour cold soak, the sample was removed from the bath and allowed to warm to 20-22 degrees Celsius (68-72 degrees Fahrenheit) for 2.5 hours (plus or minus 15 minutes). The sample was briefly swirled to disperse any precipitates which had settled to the bottom of the jar and then was immediately filtered using a modified ASTM D 6217 filtration procedure. The filtration procedure consisted of pouring the entire 300 ml sample into a 650 ml stainless steel funnel with a glass fiber filter (Whatman GF/F, 47-millimeter diameter, 0.7-micron pore size) supported by a stainless steel filter support (Figure 1). A vacuum of 23 inches of mercury (585 torr) was applied, and the time required for the entire 300 ml sample to pass through the filter was measured.
Results and Discussion
The combinations of monoglycerides, sterol glucosides, soap and water added to distilled soy biodiesel and the cloud points and filter times obtained are shown in Table 1. Each response was the result of a single measurement. The distilled soy biodiesel with no added components had a cloud point of 0.3 degrees Celsius (33 degrees Fahrenheit) and a cold soak filter time of 88 seconds.
Table 2. Predicted cloud points and filter times at set levels of monoglycerides and variable levels of sterol glucosides, soap and water.
Statistical analysis (Analysis of Variance, ANOVA) was performed using commercial software (Stat-Ease Design Expert). Terms that were not statistically significant were eliminated from the models and selected point predictions of the cloud point and filter time at various combinations of components were made (Table 2). As typical soy biodiesel contains 6,000 ppm monoglycerides, predictions were calculated at this level and at two bracketing levels (3,000 ppm and 9,000 ppm). It is important to note that predictions cannot be made for levels of components which are outside the ranges studied.
The cloud point model developed in the study shows that addition of monoglycerides alone, especially at 6,000 ppm (0.6 percent) or higher, is predicted to produce the largest increase in cloud point (Table 2). In addition, soaps or water in combination with monoglycerides at 4,000-5,000 ppm and higher would also increase the cloud point. Sterol glucosides at the levels tested would not affect the cloud point.
The filter test model indicates that the filter test response was particularly sensitive to the presence of sterol glucosides and soaps. Approximately 10 ppm sterol glucosides or soaps would cause a similar undesirable increase in filter time as 3,000-4,000 ppm monoglycerides using the modified ASTM D 6217 filtration method.
Per the model, the presence of monoglycerides or sterol glucosides caused the filter time to increase as the amount of added water increased, but the presence of water decreased the filter time when soap was present at levels of 20 ppm or more. The interaction of water with other components bears further examination. While the presence of monoglycerides, sterol glucosides and soaps can be controlled at the production stage, the moisture content in biodiesel may increase during transportation, handling and storage. These data emphasize the importance of careful biodiesel handling to prevent water contamination, even after biodiesel leaves the manufacturing facility.
Monoglycerides have been generally viewed as the single most harmful component affecting filter plugging. Although we observed that monoglycerides negatively affected filter times under the conditions of the cold soak filter test, filter test results were also strongly affected by the presence of other components and by interactions between the components. It should be noted that the negative effect of sterol glucosides and soaps would not be limited to cold conditions, but can cause filter plugging problems even in warm weather because these components can be present as solid particles at warm temperatures after prolonged storage. The cold soak is utilized to accelerate the precipitation of soaps and sterol glucosides.
It must be emphasized that the monoglyceride effects in this study reflect the monoglycerides from soy oil feedstock. Soy oil has about 15 percent saturated fatty acids
5 and therefore soy biodiesel and the monoglycerides present in soy biodiesel also have about 15 percent saturates. Soy biodiesel with 6,000 ppm (0.6 percent) monoglycerides contains 900 ppm (0.09 percent) saturated monoglycerides. The cloud point of soy biodiesel increases significantly as the content of saturated monoglycerides increases
6. Exposure of soy biodiesel to cold temperatures during storage is known to cause crystallization of saturated monoglycerides, which will concentrate at the bottom of storage vessels. Biodiesel drawn from the bottom of a storage tank may contain much higher levels of saturated monoglycerides than those typically found in soy biodiesel. Some fats or oils, such as tallow or palm, have saturated fatty acid contents in excess of 40 percent5 and the impact of this level of saturated fatty acids on cloud point and filterability of biodiesel made from these feedstocks or soy biodiesel containing atypical levels of saturated fatty acids cannot be predicted from this study.
For cloud point analysis, the model had an adjusted R-squared value of 0.92 and had a precision, or signal-to-noise ratio, with a relatively high value of 20. A good model was developed for cloud point, and, overall, the cloud point model described the experimental results and can be used to make reasonable predictions within the experimental constraints. Cloud point temperatures ranged from 0.3 to 3.9 degrees Celsius (33 to 39 degrees Fahrenheit), while repeatability of the test instrument was 1.3 degrees Celsius (2.3 degrees Fahrenheit).
To facilitate analysis of the filter time response data, it was transformed to log
10(filter time). The analysis of the data yielded a model with a high adjusted R-squared value of 0.86 and a precision of 17. However, the model did not pass the "lack of fit" test because the quadratic model used in modeling the system is inadequate to completely explain the variation in the experimental design space. Modeling with a cubic equation would address this issue but would require significantly more experimental runs. However, the quadratic model does an adequate job of explaining the relationships between the components in the center of the design space and only develops problems when multiple components are present at their maximum or minimum values. Therefore, the filter time model was considered suitable for predictions and trending the effects of the components studied, provided we are aware of the increased possible error at high and low levels of more than one component.
Table 3. Repeatability of test results and comparison of predictions with test results
Finally, several trials were repeated to compare with the predicted results and initial test results. Good repeatability was obtained (Table 3). The predicted values matched the trends seen in the actual results and were also in reasonable agreement with the actual values (within the 95 percent confidence interval).
Conclusions
Overall, monoglycerides, especially at 6,000 ppm (0.6 percent) or higher, had the most significant negative impact on cloud point. In addition, soaps or water in combination with monoglycerides at 4,000-5,000 ppm and higher had a negative impact on cloud point. Sterol glucosides did not have a negative effect on cloud point. A good statistical model was developed for cloud point.
The filter test response was particularly sensitive to sterol glucosides and soaps. Filter times also increased with monoglyceride concentration, but the effect was much less dramatic. The filter test response was overall negatively affected by water, but interesting interactions between water and other components were observed. The negative effect of water emphasizes the need for proper handling of biodiesel after the production stage. The model for filter time after cold soak had some limitations, but was found to be useful in trending the effect of the components studied. While this study clearly showed the interaction between the four components studied, further close examination of the interactions between the components is necessary to wholly understand the effects of these components on biodiesel filterability. Proposed future work includes further examination of component interaction using an experimental design for cubic equation modeling, and testing of actual engine performance using biodiesel containing the components examined in this study.
This research originally appeared in the September 2007 supplement to Inform
magazine and is reprinted with permission of its publisher, the American Oil Chemists' Society.
1. Annual Book of ASTM Standards, Vol. 05.04, American Society for Testing and Materials, West Conshohocken, Pa. D 6751-07a Standard Specification for Biodiesel Fuel Blend Stock (B100) for Middle Distillate Fuels.
2 Annual Book of ASTM Standards, Vol. 05.03, American Society for Testing and Materials, West Conshohocken, Pa. D 6217-98 Standard Test Method for Particulate Contamination of Middle Distillate Fuels by Laboratory Filtration.
3. Annual Book of ASTM Standards, Vol. 05.01, American Society for Testing and Materials, West Conshohocken, Pa. D 2068-04 Standard Test Method for Filter Plugging Tendency of Distillate Fuel Oils.
4. Annual Book of ASTM Standards, Vol. 05.03, American Society for Testing and Materials, West Conshohocken, Pa. D 5773-05 Standard Test Method of Cloud Point of Petroleum Products (Constant Cooling Rate Method).
5. Hammond, E. G., Introduction to Fats and Oils (O'Brien, R. D., W. E. Farr, P. J. Wan, eds.), AOCS Press, pp. 50-51 (2000).
6. Yu, L., I. Lee, E. Hammond, L. Johnson, and J. Van Gerpen, The Influence of Trace Components on the Melting Point of Methyl Soyate, J. Am. Oil Chem. Soc. 75:1821-1824 (1998).
The claims and statements made in this article belong exclusively to the author(s) and do not necessarily reflect the views of Biodiesel Magazine
or its advertisers. All questions pertaining to this article should be directed to the author(s).