At-line and inline prediction of droplet size in mayonnaise with near-infrared spectroscopy

Abstract

A novel use of near-infrared (NIR) spectroscopy and chemometrics is presented for non-invasive prediction of droplet-size of mayonnaise. Both at-line and inline monitoring capabilities were explored. At first during the offline experimentation, two different batches of mayonnaise were prepared under pilot plant manufacturing conditions. The mayonnaise was manufactured with different fat content levels and different milling speeds to induce differences in the droplet size during the process. The reference droplet-sizes were measured using pulsed-field gradient nuclear magnetic resonance (pfg-NMR). NIR data were calibrated with reference droplet-size measurements by partial least-square (PLS) regression. The results of at-line analysis showed that the NIR models reached high performance to predict droplet-size in mayonnaise. Furthermore, separate, and global models were used for droplet size prediction for different fat content mayonnaise. For the second part of study, a diode-array NIR spectrometer was directly integrated in the process line for mayonnaise manufacturing and real-time NIR measurements were performed. The process was modulated with different milling speeds and the effect on NIR measurements was explored. The chemometric analysis performed on the NIR process data allowed following the droplet size changes due to changing milling speed. Overall, NIR provided a good correlation with droplet-size of mayonnaise and can support applications such as real-time monitoring of droplet-size during mayonnaise manufacturing process to optimize process and product properties.