Measuring Water, Methanol and Total Glycerin in B100 Samples
May 25, 2007
BY Mike Beauchaine
Biodiesel is produced from vegetable oils and animal fats through a process called transesterification. During this process, the vegetable oil or animal fats react with methanol in the presence of a catalyst to produce a mixture of free fatty acid methyl esters (FAME) and glycerin. In order for biodiesel to be sold as a fuel or blending stock, it must meet a set of requirements defined in the American Society for Testing Materials (ASTM) D 6751 standard or the European Committee for Standardization (CEN) EN 14214 standard. These standards specify the maximum allowable concentrations of contaminants in pure biodiesel (B100), including water, methanol and total glycerin.
Fuel meeting these specifications should be clear in appearance and free of water. The presence of water generally indicates poor fuel handling practices. Water can have a great impact on diesel engines, including the corrosion of steel fuel system components. In addition, water can cause improper combustion and microbial growth. The level of water specified in the regulations is within the solubility level of water in fuel and, as such, does not represent free water. High methanol content in B100 can also lead to engine problems, including long-term corrosion and adverse effects on injector performance since it increases its volatility. Total glycerin content is a byproduct of biodiesel production. High levels of glycerin in B100 or biodiesel fuel blend can result in fuel separation, material incompatibility, injector carbon buildup, storage difficulties and transportation problems.
The ASTM and CEN standards have established limits to allow measured results of water, methanol and total glycerin to be compared with a maximum level acceptable for proper engine operation. More specifically, the maximum allowable concentration for water is 0.05 percent by volume, 0.2 by volume for methanol and 0.24 by volume for total glycerin. Any B100 batch not meeting these requirements can contaminate the entire supply chain with catastrophic economic consequences. Therefore, biodiesel producers must test their finished product to ensure compliance. However, the cost of third-party laboratory testing can significantly add to the cost of producing a single batch. In fact, the cost of sending noncompliant samples for third-party laboratory testing can be as high as $1,200 per sample.
When an acceptable test method for water, methanol and total glycerin in a fuel blend is used, a limit value can be established. ASTM and CEN regulations both specify, a centrifuge method, flash point tester and a high temperature gas chromatographer for this analysis. However, spectroscopic techniques, such as near infrared spectroscopy, are being increasingly utilized for quality control purposes to measure multiple components with one simple test method.
Testing with Spectroscopy
Encoded photometric near-infrared (EP-NIR) spectroscopy facilitates a simple, yet process-rugged, photometric design whereby the incoming near-infrared beam from a sample is imaged onto a diffraction-grating-based spectrograph. The dispersed radiation from the grating is then imaged across an aperture onto the surface of an encoder disk, which is spinning at 6,000 revolutions per minute (rpm) at 100 hertz (Hz), providing ultra-fast, real-time detection. The encoder disk has a series of reflective tracks spatially located within the dispersed grating image to correspond to the wavelengths and wavelength regions used for the analysis. Each track has a pattern that produces a reflected beam with a unique sinusoidal modulation for each individual wavelength. The reflected beams are brought to an image on a single detector, which generates a signal similar to a discrete interferogram. The intensity contribution for each wavelength component is obtained by applying a Fourier transform to the interferogram.
EP-NIR spectroscopy features an ultra-fast scanning capability of 100 scans per second, enabling simultaneous measurement of multiple components in low parts per million (ppm) ranges. The end result is high sample throughput, real-time quality-control monitoring and extreme sensitivity through spectral averaging. In addition to the ultra-fast sample processing capability, EP-NIR spectroscopy is also capable of covering a spectral range of 1,375 to 2,750 newton meters (nm) compared with conventional NIR systems that normally stop at 2,100 nm. As a consequence, EP-NIR 1,375 to 2,750 can meet the analytical requirements of a wide range of applications, such as protein purification and separation, protein and fat analysis, petrochemicals, and alcohols or starch.
Although EP-NIR spectrometers have been designed to deliver full spectrum information, they are free from environmentally sensitive components. The units don't utilize any hygroscopic optical components or internal lasers. In addition, EP-NIR technology has met military certification standards for vibration resistance. A high-frequency vibration resistance test was performed to determine the effect of vibration on component parts of the EP-NIR analyzers in frequency ranges of 0.5 to 30 Hz. The units demonstrated no degradation in electrical or photometric performance during or after the test. The photometric performance was further tested by collecting spectra as vibrations were applied to the analyzer. Even under the stress of such vibrations, EP-NIR analyzers retained photometric performance and met root mean square (RMS) signal-to-noise specifications greater than 50,000 to 1.
Return on Investment for B100 producers
EP-NIR spectroscopy offers B100 producers many significant benefits, including speed of analysis, non-destruction of samples, ease of use and low cost of ownership since the use and subsequent disposal of chemicals is virtually eliminated. EP-NIR spectroscopy, as with other secondary analytical methods, wasn't designed to replace the official analysis method. Nonetheless, as a screening method, it can deliver immediate results in a cost-effective manner to exonerate as many batches of B100 as possible from the requirement of costly and lengthy reference method analyses by third-party laboratories.
A study was designed to test the accuracy of EP-NIR spectrometry for the simultaneous measurement of water, methanol and total glycerin in B100 samples. A multi-component 2750 EP-NIR analyzer was used in conjunction with an external halogen NIR source and an extended range two millimeter path length process transmission multimode fiber probe. Synthetic biodieselsamples were prepared using an orthogonal experimental design in order to achieve maximum variability of B100 composition with a minimum number of calibration standards, while avoiding colinearities between components. Samples were prepared from four fractions, namely B100 (0.19 percent total glycerin, 0 percent methanol and 0 percent water), methanol (collected at the B100 production site), total glycerin (collected at the B100 production site) and distilled water.
Concentration ranges covered were 0 percent to 0.1 percent by volume for water, 0.1 percent to 0.3 percent by volume for methanol, and 0.19 percent to 0.3 percent by weight for total glycerin. All samples were analyzed in four replicates. Calibration models using principle least square (PLS) and principle components regression (PCR) were developed using trademarked software suite GRAMS AI 8.0. In all cases, spectra pretreatment consisted only of a mean centering to enhance small spectral variations and an automated two-point baseline correction of all the absorbance spectra to improve stability of the signal. F-test statistical analysis was performed to identify outliers. The graphing of the standard error of cross validation as a function of the number of factors in the method and a prediction residual error sum of equals were used as criteria for the determination of the number of factors to include in each calibration model.
The Pearson's Product Moment Correlation (R2, indicating the quality of the fit of the calibration model) and the root mean of the standard error of calibration (RMSEC) were the statistical outputs used to quality the accuracy and precision of the methods.. An accuracy of 95 percent confidence was defined as plus or minus two multiplied by RMSEC, and precision was defined as the average of standard deviation of the residual calculated concentrations for each set of four replicate results.
Results and Discussion
Figure 1 reports the accuracy and precision levels of the EP-NIR method using three different algorithmic approaches. Water contamination was measured at 0.038 percent by volume or less. Methanol contamination was estimated at 0.184 percent by volume or less, and total glycerin contamination was measured 0.221 percent by volume or less.
Figure 1
Figure 2
Figure 3
Figure 4
With regards to the measurement of water in B100, best analytical results were obtained using the peak cell rate algorithmic approach (Figure 2). R2 was 97.86 percent, RMSEC was 58 ppm (resulting in a 95 percent confidence accuracy measurement of plus or minus 117 ppm), and precision was 58 ppm. The PLS1 algorithmic approach was proven best for measuring methanol and glycerin (Figures 3 and 4). In the case of methanol, R2 was 98.87 percent, RMSEC was 95 ppm (resulting in a 95 percent confidence accuracy in measurement of plus or minus 189 ppm), and precision was 91 ppm. For glycerin, R2 was 91.28 percent, RMSEC was 145 ppm (resulting in a 95 percent confidence accuracy in measurement of plus or minus 290 ppm), and precision was 144 ppm. Overall, glycerin was the most difficult compound to model. The difficulty of the measurement can be better understood when looking at the chemical nature of the compound in relationship to the other molecules present in the matrix.
From a chemometrics viewpoint, it appeared that the preprocessing of the absorbance spectra (mean centering and two-point automated baseline correction) had much more impact on improving the accuracy and precision of the measurement than switching from one multivariate algorithmic approach to the other (PLS1, PLS2 and PCR).
In conclusion, EP-NIR spectroscopy achieved a level of accuracy of 95 percent confidence for all measurements. As a result, EP-NIR spectroscopy can amply meet the ASTM-defined analytical requirements for the measurement of water, methanol and total glycerins in B100. Assuming that manufacturing of the synthetic samples was constant, it was observed that the least interfered chemical (water) produced the most accurate results, whereas the most interfered chemical (glycerin) produced the least accurate results. Furthermore, from a chemometrics viewpoint, it appeared that the preprocessing of the absorbance spectra had much more of an impact on improving the accuracy of measurement than switching from one multivariate algorithmic approach to the other.
Experimental results prove that EP-NIR spectroscopy is capable of producing accurate results for the simultaneous measurement of water, methanol and total glycerin in B100 samples. Prescreening the samples using EP-NIR spectroscopy eliminates the costly number of noncompliant samples sent for third-party analysis, thus achieving significant cost savings for biodiesel producers.
Mike Beauchaine is the international sales engineer for Aspectrics Inc., a Pleasanton, Calif.-based company providing original equipment manufacturers with infrared spectroscopy technology. For more information, please contact Celine Callender at (+44 1 606 837 787) or pr@scottmail.co.uk.
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).
Fuel meeting these specifications should be clear in appearance and free of water. The presence of water generally indicates poor fuel handling practices. Water can have a great impact on diesel engines, including the corrosion of steel fuel system components. In addition, water can cause improper combustion and microbial growth. The level of water specified in the regulations is within the solubility level of water in fuel and, as such, does not represent free water. High methanol content in B100 can also lead to engine problems, including long-term corrosion and adverse effects on injector performance since it increases its volatility. Total glycerin content is a byproduct of biodiesel production. High levels of glycerin in B100 or biodiesel fuel blend can result in fuel separation, material incompatibility, injector carbon buildup, storage difficulties and transportation problems.
The ASTM and CEN standards have established limits to allow measured results of water, methanol and total glycerin to be compared with a maximum level acceptable for proper engine operation. More specifically, the maximum allowable concentration for water is 0.05 percent by volume, 0.2 by volume for methanol and 0.24 by volume for total glycerin. Any B100 batch not meeting these requirements can contaminate the entire supply chain with catastrophic economic consequences. Therefore, biodiesel producers must test their finished product to ensure compliance. However, the cost of third-party laboratory testing can significantly add to the cost of producing a single batch. In fact, the cost of sending noncompliant samples for third-party laboratory testing can be as high as $1,200 per sample.
When an acceptable test method for water, methanol and total glycerin in a fuel blend is used, a limit value can be established. ASTM and CEN regulations both specify, a centrifuge method, flash point tester and a high temperature gas chromatographer for this analysis. However, spectroscopic techniques, such as near infrared spectroscopy, are being increasingly utilized for quality control purposes to measure multiple components with one simple test method.
Testing with Spectroscopy
Encoded photometric near-infrared (EP-NIR) spectroscopy facilitates a simple, yet process-rugged, photometric design whereby the incoming near-infrared beam from a sample is imaged onto a diffraction-grating-based spectrograph. The dispersed radiation from the grating is then imaged across an aperture onto the surface of an encoder disk, which is spinning at 6,000 revolutions per minute (rpm) at 100 hertz (Hz), providing ultra-fast, real-time detection. The encoder disk has a series of reflective tracks spatially located within the dispersed grating image to correspond to the wavelengths and wavelength regions used for the analysis. Each track has a pattern that produces a reflected beam with a unique sinusoidal modulation for each individual wavelength. The reflected beams are brought to an image on a single detector, which generates a signal similar to a discrete interferogram. The intensity contribution for each wavelength component is obtained by applying a Fourier transform to the interferogram.
EP-NIR spectroscopy features an ultra-fast scanning capability of 100 scans per second, enabling simultaneous measurement of multiple components in low parts per million (ppm) ranges. The end result is high sample throughput, real-time quality-control monitoring and extreme sensitivity through spectral averaging. In addition to the ultra-fast sample processing capability, EP-NIR spectroscopy is also capable of covering a spectral range of 1,375 to 2,750 newton meters (nm) compared with conventional NIR systems that normally stop at 2,100 nm. As a consequence, EP-NIR 1,375 to 2,750 can meet the analytical requirements of a wide range of applications, such as protein purification and separation, protein and fat analysis, petrochemicals, and alcohols or starch.
Although EP-NIR spectrometers have been designed to deliver full spectrum information, they are free from environmentally sensitive components. The units don't utilize any hygroscopic optical components or internal lasers. In addition, EP-NIR technology has met military certification standards for vibration resistance. A high-frequency vibration resistance test was performed to determine the effect of vibration on component parts of the EP-NIR analyzers in frequency ranges of 0.5 to 30 Hz. The units demonstrated no degradation in electrical or photometric performance during or after the test. The photometric performance was further tested by collecting spectra as vibrations were applied to the analyzer. Even under the stress of such vibrations, EP-NIR analyzers retained photometric performance and met root mean square (RMS) signal-to-noise specifications greater than 50,000 to 1.
Return on Investment for B100 producers
EP-NIR spectroscopy offers B100 producers many significant benefits, including speed of analysis, non-destruction of samples, ease of use and low cost of ownership since the use and subsequent disposal of chemicals is virtually eliminated. EP-NIR spectroscopy, as with other secondary analytical methods, wasn't designed to replace the official analysis method. Nonetheless, as a screening method, it can deliver immediate results in a cost-effective manner to exonerate as many batches of B100 as possible from the requirement of costly and lengthy reference method analyses by third-party laboratories.
A study was designed to test the accuracy of EP-NIR spectrometry for the simultaneous measurement of water, methanol and total glycerin in B100 samples. A multi-component 2750 EP-NIR analyzer was used in conjunction with an external halogen NIR source and an extended range two millimeter path length process transmission multimode fiber probe. Synthetic biodieselsamples were prepared using an orthogonal experimental design in order to achieve maximum variability of B100 composition with a minimum number of calibration standards, while avoiding colinearities between components. Samples were prepared from four fractions, namely B100 (0.19 percent total glycerin, 0 percent methanol and 0 percent water), methanol (collected at the B100 production site), total glycerin (collected at the B100 production site) and distilled water.
Concentration ranges covered were 0 percent to 0.1 percent by volume for water, 0.1 percent to 0.3 percent by volume for methanol, and 0.19 percent to 0.3 percent by weight for total glycerin. All samples were analyzed in four replicates. Calibration models using principle least square (PLS) and principle components regression (PCR) were developed using trademarked software suite GRAMS AI 8.0. In all cases, spectra pretreatment consisted only of a mean centering to enhance small spectral variations and an automated two-point baseline correction of all the absorbance spectra to improve stability of the signal. F-test statistical analysis was performed to identify outliers. The graphing of the standard error of cross validation as a function of the number of factors in the method and a prediction residual error sum of equals were used as criteria for the determination of the number of factors to include in each calibration model.
The Pearson's Product Moment Correlation (R2, indicating the quality of the fit of the calibration model) and the root mean of the standard error of calibration (RMSEC) were the statistical outputs used to quality the accuracy and precision of the methods.. An accuracy of 95 percent confidence was defined as plus or minus two multiplied by RMSEC, and precision was defined as the average of standard deviation of the residual calculated concentrations for each set of four replicate results.
Results and Discussion
Figure 1 reports the accuracy and precision levels of the EP-NIR method using three different algorithmic approaches. Water contamination was measured at 0.038 percent by volume or less. Methanol contamination was estimated at 0.184 percent by volume or less, and total glycerin contamination was measured 0.221 percent by volume or less.
Figure 1
Figure 2
Figure 3
Figure 4
With regards to the measurement of water in B100, best analytical results were obtained using the peak cell rate algorithmic approach (Figure 2). R2 was 97.86 percent, RMSEC was 58 ppm (resulting in a 95 percent confidence accuracy measurement of plus or minus 117 ppm), and precision was 58 ppm. The PLS1 algorithmic approach was proven best for measuring methanol and glycerin (Figures 3 and 4). In the case of methanol, R2 was 98.87 percent, RMSEC was 95 ppm (resulting in a 95 percent confidence accuracy in measurement of plus or minus 189 ppm), and precision was 91 ppm. For glycerin, R2 was 91.28 percent, RMSEC was 145 ppm (resulting in a 95 percent confidence accuracy in measurement of plus or minus 290 ppm), and precision was 144 ppm. Overall, glycerin was the most difficult compound to model. The difficulty of the measurement can be better understood when looking at the chemical nature of the compound in relationship to the other molecules present in the matrix.
From a chemometrics viewpoint, it appeared that the preprocessing of the absorbance spectra (mean centering and two-point automated baseline correction) had much more impact on improving the accuracy and precision of the measurement than switching from one multivariate algorithmic approach to the other (PLS1, PLS2 and PCR).
In conclusion, EP-NIR spectroscopy achieved a level of accuracy of 95 percent confidence for all measurements. As a result, EP-NIR spectroscopy can amply meet the ASTM-defined analytical requirements for the measurement of water, methanol and total glycerins in B100. Assuming that manufacturing of the synthetic samples was constant, it was observed that the least interfered chemical (water) produced the most accurate results, whereas the most interfered chemical (glycerin) produced the least accurate results. Furthermore, from a chemometrics viewpoint, it appeared that the preprocessing of the absorbance spectra had much more of an impact on improving the accuracy of measurement than switching from one multivariate algorithmic approach to the other.
Experimental results prove that EP-NIR spectroscopy is capable of producing accurate results for the simultaneous measurement of water, methanol and total glycerin in B100 samples. Prescreening the samples using EP-NIR spectroscopy eliminates the costly number of noncompliant samples sent for third-party analysis, thus achieving significant cost savings for biodiesel producers.
Mike Beauchaine is the international sales engineer for Aspectrics Inc., a Pleasanton, Calif.-based company providing original equipment manufacturers with infrared spectroscopy technology. For more information, please contact Celine Callender at (+44 1 606 837 787) or pr@scottmail.co.uk.
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).
Advertisement
Advertisement
Upcoming Events
