The viscosities of three biodiesel fuels at temperatures up to 300 8C
R.E. Tate, K.C. Watts
* , C.A.W. Allen, K.I. Wilkie Department of Biological Engineering, Dalhousie University, P.O. Box 1000, Halifax, NS, Canada B3J 2X4 Received 12 April 2005; received in revised form 29 September 2005; accepted 25 October 2005 Available online 18 November 2005 Abstract In order to obtain the kinematic viscosities of biodiesel fuels at temperatures up to 300 8C, a modied Saybolt viscometer was designed. The viscometer was used to measure the efux times for 60 ml of methyl esters of canola and soy, and ethyl esters of sh-oil. The Modied Saybolt Viscometer was calibrated using a standard oil and can be used to measure the kinematic viscosity to within 0.056 mm 2 /s with 2% repeatability. Using the measured densities over the same temperature range, the dynamic viscosities were obtained. q 2005 Elsevier Ltd. All rights reserved. Keywords: Biodiesel; Viscosity; Temperature 1. Introduction Viscosity is one of the most important fuel properties. The effects of viscosity can be seen in the quality of atomization and combustion as well as engine wear. The higher viscosity of biodiesel fuel compared to diesel makes it an excellent lubricity additive [1,2]. On the other hand, the high viscosities of biodiesel fuels are reportedly responsible for premature injector fouling [1] leading to poorer atomization. These issues need to be addressed if biodiesel fuel is to become the alternative to petroleum diesel. Biodiesel fuels are the fatty acid methyl or ethyl esters derived from vegetable or animal fats and oils. Fatty acid composition has a signicant effect on the viscosities of fats and oils, and in turn biodiesel fuels. The fatty acid compositions of fats and oils are feedstock dependent and are also affected by factors such as climatic conditions, soil type, plant health, and plant maturity upon harvest [3]. Biodiesel fatty acid compositions and fuel proper- ties can vary signicantly from one supplier to the next, even if it is fromthe same type of plant/animal (i.e. soy bean oil). The quality of fuel atomization is signicantly affected by viscosity. Since viscosity can vary signicantly between different biodiesel fuels, there exists a need to measure the temperature dependant viscosities of biodiesel fuels and verify prediction models used to predict the temperature dependant viscosity based on the fatty acid composition of the fuel. The objective of this paper is to report on the experimentally determined effect of temperature on the viscosity of three commercially available biodiesel fuels. This is part of a larger research project to determine the temperature and pressure dependence of physical properties that affect biodiesel fuel atomization and combustion (viscosity, surface tension, density and droplet size) and verify the prediction models proposed by Allen et al. [4,5] and Yuan et al. [6] over the temperature range of 20 to 300 8C, a temperature just below the boiling point of the lightest oil fraction. 2. Experimental methods The three biodiesel fuels selected were available commer- cially in either their pure or their blended forms. For this work 100% methyl esters of canola (Milligan Bio-Tech, SK) and soy (West Central Soy, IA) and ethyl esters of sh-oil (Wilsons Fuels, NS) were obtained. These are referred to CME, SME and FEE in this paper. The properties of the fuels are given in Table 2 of the companion paper The densities of the three biodiesel fuels at temperatures up to 300C (d.o.i.: 10.1016/ i.fuel.2005.10.024). 2.1. Apparatus Saybolt viscometry was selected because of the adaptability of operation to elevated temperatures. This type of viscometer Fuel 85 (2006) 10101015 www.fuelrst.com 0016-2361/$ - see front matter q 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.fuel.2005.10.015 * Corresponding author. Tel.: C1 902 494 3232; fax: C1 902 423 2423. E-mail address: chris.watts@dal.ca (K.C. Watts). operates under the same principles as other capillary viscometers; the efux time for a known volume of the fuel through a calibrated orice is related to kinematic viscosity. The Saybolt viscometer design [7] was modied to allow for remote operation at temperatures ranging from room temperature to the boiling point of the most volatile oil fraction. An insulated, three part chamber was designed to house the Modied Saybolt system (Fig. 1). In the lower chamber was a receiving ask made of 3/8 in. stainless steel tubing and a 50.8 mm diameter steel sphere. Small holes were drilled at two heights on opposite sides of the tubing (Fig. 1) and were compression sealed with Lexan discs. Two lights, rated at 14-volts, and two photodiodes made up the level detection system. When the liquid level reached the holes the liquid impeded the amount of light transmitted to the diode and a distinct decrease in the signal voltage from the diode was observed. The receiving ask was heated to the test temperature with a 125 W rope heater. Standard Saybolt and Furol tips, 1.765 and 3.150 mm, respectively, were too large in diameter to yield meaningful results at elevated temperatures where viscosities are inherently low. Based on the predicted kinematic viscosity of 0.77 mm 2 /s and the dynamic viscosity of 0.50 mPa s at 300 8C for 100% soybean oil methyl esters by Yuan et al. [6], an orice diameter of 1.016 mm was selected. The efux time for water was measured as 105 Modied Saybolt Seconds (MSS) at 20 8C and 91 MSS at 98 8C. These efux times correspond to kinematic viscosities of 1.004 and 0.295 mm 2 /s and dynamic viscosities of 1.00 and 0.281 mPa s [7]. The selected orice diameter was deemed adequate for the expected measurement range. The oil in the reservoir was heated with two 250 W band heaters controlled by an external onoff temperature controller purchased from Watlow. Three type K thermocouples were used to monitor and control the oil, chamber, and receiving ask temperatures. Type K wire thermocouples, calibrated to 480 8C, were used for the control of the sample temperature and monitoring the receiving ask temperature. A type K thermocouple probe, calibrated to 980 8C, was used for monitoring the sample temperature. The oil was continuously stirred during the heating process to ensure a uniform oil temperature. The nal modication necessary was the remote actuation of the reservoir stop. A solenoid actuated by a solid- state relay was used to control the stop. 2.2. Methodology The experimental procedure was based on ASTM D88 [8], with the exception of the procedures that were automated. Prior to each fuel change, reservoir and receiving ask were cleaned with toluene and dried with acetone. As a precaution, a 100 mm wire mesh was used to lter the uid as it was poured into the reservoir. The apparatus was assembled and a continuously heated ow of nitrogen was started to prevent oxidation of the oil during heating. Once the reservoir was lled and the chamber reassembled, operation of the Modied Saybolt viscometer was controlled via a program developed under LabWindows CVI. A PCI- MIO-16E-4 National Instruments data acquisition board was used for the acquisition of the temperature and light sensor signals. The sample and chamber temperatures were continu- ously monitored. When a stable test temperature was attained an indicator light on the computer monitor was lit. The solid- state relay and solenoid were then manually activated releasing the reservoir plug. The light detecting diode signals were scanned simultaneously and continuously for a predetermined time greater than the estimated efux time. At the completion of this time, the solenoid was deactivated and the reservoir plug closed. Analysis of the collected data within the program required the user to select four points with the cursor beginning with the start time of the low level sensor signal change and ending with the stop time of the high level sensor signal change. Each sensors data array was searched to nd the index points selected. The voltage and time data within these indices were used to determine the maximum slope of each sensor signal and corresponding time at which it occurred. The difference between these two times is equivalent to the efux time of 60 ml of the sample uid in MSS. Each experiment took between 5 and 30 min, depending on the test temperature. Three samples of each fuel were tested at 20 8C intervals between 20 and 300 8C. Fig. 1. Modied saybolt viscometer [10]. R.E. Tate et al. / Fuel 85 (2006) 10101015 1011 2.3. Calibration and repeatability In order to determine the relationship between MSS and kinematic viscosity, calibrations were performed using distilled water, Cannon viscosity oil standard No.1 and each of the biodiesel fuel samples at the specic temperatures for which the kinematic viscosity was known. By using three samples of each of the uids repeatability of the Modied Saybolt viscometer was determined. 2.4. Kinematic viscosity Capillary ow of a Newtonian uid is governed by the Poiseuille equation: dP dl Z 8Qm pr 4 (1) Since the Modied Saybolt Universal viscometer is a modied capillary viscometer, kinematic viscosity is deter- mined by measuring the time for 60 ml of uid to ow through a short capillary tube. Since the ow rate, tube length, tube diameter, and pressure are known, Eq. (1) becomes: m rt Z ghpr 4 8Vl ZC 1 (2) The average head h is considered xed for a specic instrument as the variation is over the same range for all liquids. Flow through the short capillary tube is turbulent and thus the velocity distribution is not established. A correction factor, C 2 /t, is needed. Therefore, the relationship between MSS and kinematic viscosity, including corrections for entrance and exit effects due to the short tube length, is n ZC 1 MSS C C 2 MSS (3) 3. Results and discussion 3.1. Calibration The results of the calibration and accuracy and repeatability analysis are shown in Fig. 2 and Tables 1 and 2. Regression analysis yielded C 1 Z0.028 and C 2 ZK199 with an R 2 of 0.994. The repeatability of the Modied Saybolt viscometer was found to be 2%. The errors associated with the Modied Saybolt Viscometer were less than 0.056 mm 2 /s, on average. 3.2. Kinematic viscosity Figs. 35 show the relationships between the kinematic viscosity and temperature of CME, SME, and FEE, respect- ively. When the ratio of temperature to a liquids critical temperature, T r , is less than 0.7, kinematic viscosity is often represented by the Andrade equation. A modied form of the Andrade equation, as proposed by Tat and Van Gerpen [9], was used lnn ZAC B T C C T 2 (4) where n is the kinematic viscosity, T is the absolute temperature, and A, B, and C are constants specic to the uid. Results of the regressions using Eq. (4) are shown in Table 3. With the lowest R 2 value greater than 0.95, Eq. (4) provides a good t and representation of the data. Fig. 6 compares this experimental data for SME to that of Tat and Van Gerpen [9], who regressed their data to Eq. (4) with an R 2 Z0.999, and the predicted viscosities from Yuan Table 1 Modied saybolt viscometer repeatability results for the oil standard and biodiesel fuel samples Fluid type Tempera- ture (8C) Measured kinematic viscosity (mm 2 /s) Average kinematic viscosity (mm 2 /s) Absolute error (mm 2 /s) Percent error (%) Canon oil Standard No. 1 20 1.298 1.255 0.042 3.368 1.255 0.000 0.007 1.213 0.042 3.375 25 1.335 1.316 0.019 1.467 1.368 0.052 3.971 1.244 0.072 5.438 40 1.017 1.087 0.070 6.399 1.095 0.009 0.792 1.148 0.061 5.607 Fish-oil EE 40 3.999 3.978 0.022 0.543 3.928 0.050 1.257 4.006 0.028 0.714 Soy ME 40 3.989 3.982 0.007 0.171 3.969 0.014 0.342 3.989 0.007 0.171 Canola ME 40 4.608 4.647 0.038 0.827 4.654 0.008 0.166 4.677 0.031 0.661 Average 0.032 1.96 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 80 90 100 110 120 130 140 150 160 170 180 190 200 210 MSS K i n e m a t i c
V i s c o s i t y
( m m 2 / s ) Fig. 2. Modied saybolt viscometer calibration results. $ CME, ,SME, 6 FEE, CCanon oil standard no. 1, Bdistilled water, calibration curve. R.E. Tate et al. / Fuel 85 (2006) 10101015 1012 et al. [6] at low temperature. Tat and Van Gerpen [9] measured the viscosity of soy methyl esters (NOPEC Corporation, FL) over the range of 20 to 100 8C using a Precision Scientic Kinematic Viscosity Bath and following procedures outlined in Table 2 Modied Saybolt viscometer error analysis results for the oil standard and biodiesel fuel samples Fluid type Temperature (8C) Measured kinematic viscosity (mm 2 /s) Standard kinematic viscosity (mm 2 /s) a Absolute error (mm 2 /s) Distilled water 20 0.971 1.004 0.033 25 0.862 0.893 0.030 30 0.745 0.801 0.056 35 0.661 0.723 0.063 40 0.537 0.685 0.147 45 0.564 0.602 0.038 50 0.513 0.553 0.040 55 0.408 0.511 0.103 60 0.491 0.474 0.017 65 0.513 0.433 0.069 70 0.486 0.413 0.073 75 0.409 0.388 0.022 80 0.397 0.365 0.032 85 0.387 0.345 0.043 90 0.393 0.326 0.067 95 0.342 0.309 0.033 Average 0.054 Canon oil Standard No. 1 20 1.087 1.067 0.110 25 1.316 1.269 0.047 40 1.255 0.365 0.049 Average 0.068 Fish-oil EE 40 3.978 3.982 Soy ME 40 3.982 4.200 Canola ME 40 4.647 4.473 Overall average 0.056 a Data were obtained from [7] for distilled water and from the suppliers data sheets for the oil standard and fuel samples. N.B: Due to uncertainties in the manufacturers reported viscosity, the standard error for the fuel samples was not determined. 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 Temperature (C) K i n e m a t i c
V i s c o s i t y
( m m 2 / s ) Fig. 4. Kinematic viscosity of 100% soy methyl ester. 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 Temperature (C) K i n e m a t i c
V i s c o s i t y
( m m 2 / s ) Fig. 5. Kinematic viscosity of 100% sh-oil ethyl ester. 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 Temperature (C) K i n e m a t i c
V i s c o s i t y
( m m 2 / s ) Fig. 3. Kinematic viscosity of 100% canola methyl ester. Table 3 Viscosity correlation constants for three biodiesel fuels over the temperature range of 20300 8C Fuel type A B C R 2 100% Canola methyl ester K0.341 K498 338,301 0.986 100% Soy methyl ester 0.076 K1078 469,741 0.951 100% Fish-oil ethyl ester K0.453 K739 415,238 0.968 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 0 20 40 60 80 100 Temperature (C) K i n e m a t i c
V i s c o s i t y
( m m 2 / s ) Fig. 6. Comparison of experimental and predicted kinematic viscosities of 100% soy methyl ester up to 100 8C. !SME, $Tat and Van Gerpen [8], , Yuan et al. [6]. R.E. Tate et al. / Fuel 85 (2006) 10101015 1013 ASTM D 445 [10]. Yuan et al. [6] predicted the viscosity of soy methyl esters based on chemical composition and structure of the oil up to the critical point of the fuel. The results from this study and the extrapolated results of Tat and Van Gerpen [9] are shown in Fig. 7 and compared to our experimental data. There is very good agreement between the experimental results found in this work and those of Tat and Van Gerpen [9] where the maximum absolute error was 0.61 mm 2 /s at 40 8C. Yuan et al. [6] over estimated the kinematic viscosity by an average of 0.70 mm 2 /s when T!80 8C and under estimated the kinematic viscosity by an average of 0.58 mm 2 /s when TO80 8C. 3.3. Dynamic viscosity Using the measured densities from Tate [11], the dynamic viscosities of each of the biodiesel fuels were calculated (Fig. 8 and Table 4). 4. Conclusions 1 A modied Saybolt Viscometer was shown to give the efux time of 60 ml of biodiesel fuels up to the boiling point of the lightest oil fraction to within 0.056 mm 2 /s and 2% repeatability. 2 The viscosities of three commercial biodiesel fuels, canola methyl esters, soy methyl esters, and sh oil ethyl esters, decreased with temperature, experimentally and as pre- dicted by a modied Andrade equation. 3 The predicted and low temperature experimental values in the literature for kinematic viscosities of soy methyl esters compare very well with the experimental viscosities obtained in this work. Acknowledgements The authors acknowledge the support of the Natural Science and Engineering Research Council (NSERC) by way of an operating grant to K.C. Watts. 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 Temperature (C) K i n e m a t i c
V i s c o s i t y
( m m 2 / s ) Fig. 7. Comparison of experimental and predicted kinematic viscosities of 100% soy methyl ester up to 300 8C. !SME, $Tat and Van Gerpen [8], , Yuan et al. [6]. 0 1 2 3 4 5 6 0 50 100 150 200 250 300 Temperature (C) D y n a m i c
V i s c o s i t y
( m P a - s ) Fig. 8. Dynamic viscosities of three biodiesel fuels from 20 to 300 8C. $CME, ,SME, 6 FEE. Table 4 Kinematic and dynamic viscosities predicted from the correlations of three biodiesel fuels over the temperature range of 20300 8C Temperature (8C) Kinematic viscosity (mm 2 /s) Dynamic viscosity (mPa s) Canola ME Soy ME Fish-oil EE Canola ME Soy ME Fish-oil EE 20 6.665 6.455 6.411 5.751 5.518 5.372 40 4.566 4.153 4.143 3.828 3.447 3.369 60 3.361 2.923 2.915 2.736 2.354 2.298 80 2.615 2.203 2.190 2.065 1.720 1.672 100 2.126 1.752 1.731 1.627 1.324 1.279 120 1.788 1.452 1.424 1.325 1.061 1.017 140 1.546 1.245 1.210 1.108 0.879 0.834 160 1.367 1.095 1.056 0.946 0.746 0.701 180 1.231 0.985 0.940 0.822 0.647 0.601 200 1.125 0.901 0.852 0.724 0.569 0.524 220 1.041 0.837 0.783 0.644 0.508 0.462 240 0.974 0.786 0.729 0.579 0.458 0.412 260 0.919 0.746 0.685 0.524 0.416 0.370 280 0.873 0.713 0.649 0.476 0.380 0.335 300 0.835 0.687 0.620 0.435 0.349 0.304 R.E. Tate et al. / Fuel 85 (2006) 10101015 1014 References [1] Graboski MS, McCormick RL. Prog Energy Combust Sci 1998;24:12564. [2] Schumacher, L.G., VanGerpen, J., Adams, B.T. An ASAE Meeting Presentation Paper 2003; 036036. [3] Swern D. Wiley, New York; 1979. [4] Allen CAW, Watts KC, Ackman RG, Pegg MJ. Fuel 1999;78:131926. [5] Allen CAW, Watts KC, Ackman RG. JAOCS 1999;76(3):31723. [6] Yuan W, Hansen AC, Zhang Q. Trans ASAE 2003;46(6):148793. [7] Weast RC, editor. Boca Raton: CRC Press, 1981. Tate, RE., Watts, K.C., Allen, C.A.W., Wilkie, K.I. Fuel 2006, in press. [8] ASTM D88-94 (99) Philadelphia; 1999. [9] Tat ME, Van Gerpen JH. JAOCS 1999;76(12):15113. [10] ASTM D445-88 Philadelphia; 1986. [11] Tate RE. MASc. Thesis, Dalhousie University, Faculty of Engineering; 2005. R.E. Tate et al. / Fuel 85 (2006) 10101015 1015