Improving real-time measurement of phosphorus using spectroscopy and non-linear mathematics

Abstract: The quantification of phosphorus in fats and oils processing, whether for edible oils or as a feedstock for renewable diesels and sustainable aviation fuel, is an area of significant interest. Traditional methods use atomic or x-ray spectroscopy and measure the P atom itself. These require laboratory instruments and off-line sampling, and are slow and expensive approaches to take. Here we present the successful use of static-optics FTIR spectroscopy to measure P on-line, and in real time – regardless of oil type or phospholipid type.

We will outline the spectral features of phospholipids, how these change between different phospholipid end groups, and how the chelation of Mg2+ or Ca2+ ions changes them. We will share the calibration approaches taken for a range of different oil types and approaches – ranging from lab samples through to industrial installations in edible oil refineries.

Lastly, we will show the differences between linear and non-linear calibration, as well as machine learning based approaches. The different types of calibration will be compared, and the benefits and drawbacks of different model types explained, as well as how this performed in real-world applications and installations.