Submitted to: Journal of Near Infrared Spectroscopy
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: April 4, 2012
Publication Date: May 1, 2012
Citation: Long, D.S., McCallum, J.D., Young, F.L., Lenssen, A. 2012. In-stream measurement of canola (Brassica napus L.) seed oil concentration using in-line near infrared reflectance spectroscopy. Journal of Near Infrared Spectroscopy. 20(3):387-395. Interpretive Summary: The percentage of oil that a crushing plant can recover from an oilseed crop is determined by the oil content of the seed. Because oil recovery varies with seed oil content, it is difficult for plant managers to use one setting universally for all seed and maintain high rates of oil recovery. This study determined how well an optical sensor, having greater shortwave near infrared sensitivity than earlier generation sensors, could detect the seed oil content of canola when its sensor head was placed in direct contact with a moving stream of grain. The sensor was calibrated to grain samples from a diversity of sites in Oregon, Washington, and Montana. It successfully measured the oil content of test samples of the seed with excellent accuracy. The results were sufficiently promising to suggest that such a sensor could be installed in a crushing plant to measure the oil content of the seed before entering the press. This information could then be used to adjust an oilseed press for maximum efficiency in real-time.
Technical Abstract: Natural variation in the seed oil concentration of oilseed crops can impair a crushing plant’s ability to efficiently recover the oil from seed. Consequently, there is interest in using in-line near infrared (NIR) spectroscopy to measure the oil concentration of the seed to be processed and use this information to maximize expeller efficiency. The objective of this study was to determine how well in-line NIR spectroscopy could determine the seed oil concentration of canola (Brassica napus L.) by direct contact of the sensor’s head with the grain stream. High resolution (2 nm) reflectance spectra from 850 nm to 1650 nm were obtained by sliding grain samples of canola directly across the sensor head of the Polytec 1750 NIR reflectance analyzer. Reference analytical results were estimated using the NIR optical spectra as regression estimators. The resulting prediction equation with 14 eigenvectors resulted in a coefficient of multiple determination (R2) of 0.987 and standard error of prediction of 0.4167%. Validation results based on site-year omission confirmed the ability of the instrument to accurately predict seed oil concentration in a grain stream (R2=0.97, SEV=0.601). This creates opportunities for monitoring the oil content of seed entering the expeller and using this information to adjust the expeller for maximum efficiency.