Application note: Static-optics FTIR for in-process monitoring of free fatty acid (FFA) neutralisation process
The majority of edible oil refining methodologies require the neutralisation of free fatty acids (FFAs) using an alkali such as sodium hydroxide (“caustic”, or NaOH). This turns the FFAs into soaps which are readily removed with water washing. However, just the right amount of NaOH is required: add too little and there will be residual FFA present, and add too much and the oil product itself (triglycerides) will be hydrolysed into soaps too – thereby decreasing overall yield.
Because FFA concentration can vary dramatically from feedstock to feedstock, or even day to day due to environmental and storage differences, it is important to continually monitor it. Traditional approaches for measuring FFA involve taking a sample to the laboratory for titration testing. This is a laborious process, and only informs the operators what the FFA content was at the time when the sample was taken in the process. Since there is a time delay to receive results from off-line titration, the results may not be truly representative of the current state of the process which may have changed. Furthermore, titrations are an expensive use of resources, as they are slow and difficult to perform.
Various approaches have been taken over the years to try and move FFA testing on-line, using near infrared spectroscopy (NIR) with minimal success. This is because NIR instruments offer limited information because of the fundamental physical chemical processes happening in the NIR wavelength range. Mid-infrared (FTIR) spectroscopy is significantly more informative but traditional FTIR instruments are extremely fragile and not suitable for on-line analysis.
Here we present the use of a static-optics FTIR instrument for FFA measurement to enable real-time dosing of NaOH. We also show the instrument’s suitability for measuring other chemicals of interest, namely copper (Cu), iron (Fe) and water.
This work was performed using an on-line installation of the IRmadillo spectrometer, installed directly into the feed line of the refinery. Spectra were acquired with a 120 s acquisition time. Each reference point was calibrated with triplicate spectra to reduce impacts of noise (assuming the rate of change of the process is < 360 s).
Off-line reference data were provided by the customer using their standard process: titration for FFA and an ICP-OES instrument for copper and iron measurements. Chemometric calibrations were built using Eigenvector Research Inc’s Solo software, using a range of partial least squares (PLS) and locally weighted regression (LWR) models for quantitative analysis and principal component analysis (PCA), partial least squares discriminant analysis (PLS-DA) and support vector classification (SV-C) models for qualitative analysis.
Results & Discussion
The collated results for 4 weeks of data are shown in figure 1 – with measurements of FFA, copper, iron and moisture displayed on the same graph. (FTIR instrumentation can measure an almost unlimited number of chemicals at the same time – so multiple outputs can be sent to the DCS or control room simultaneously).
Typical measurement errors for these chemicals are:
- FFA: ± 0.1 %
- Moisture content: ± 0.013 %
- Copper: ± 0.017 ppm
- Iron: ± 0.34 ppm
The results show very good agreement between the reference data and the measurements, and highlight how chemical components that may be considered fairly stable throughout a batch process can change quite dramatically.
It is also possible to see at least one instance where sample handling issues or human error has likely led to a mistake in the reference data as seen, for example, on the 15th November. The laboratory reference for this day does not agree with any of the measurements – suggesting either a sample contamination issue or some other error associated with laboratory sampling. But the results reported are within expected values, so it would have been unlikely to have been identified and understood using an infrequent laboratory sampling methodology. This shows the power of continuous and on-line measurement, regardless of the technology used to achieve this.
The use of on-line measurement in this installation allows the operators to control the addition of NaOH downstream from the instrument – a “feed forward” installation. This is a very effective way to use on-line analytics. It does assume a consistent efficiency in the subsequent chemical reaction, which may not be the case. It is also possible to measure FFA further downstream and then “feed back” the result to fine control the addition rate or residence time in a reactor (a so-called “feed back trim”).
An example of this type of measurement is shown in figure 2 – where the installation of the IRmadillo is downstream of most of the processing steps and is measuring residual FFA. Here the average error was ± 0.005 % (50 ppm). These results show that a consistently low level < 0.1 % was measured until a step change carries this up to 0.16 %. This change is not noticed immediately for several hours until a routine off-line test is performed, the process is almost immediately adjusted to compensate, and an over-compensation is made. A second off-line sample is taken, and the system then allowed to return to ~ 0.1 %.
These results show another benefit of on-line and real-time measurement: identifying quickly when processes move out of specification and taking action early enough to prevent issues.
Figure 1: Measurements over time for moisture, iron, copper and FFA in an oil feed line for a period of ~ 4 weeks with laboratory sampling shown for reference. NB: The refinery operates on batches and the “spikes” shown occur during a batch changeover and are not measurement artefacts.
Figure 2: FFA measurements from on-line installation of refined oil over 3 weeks
This work has shown a single installation in an oil inlet measuring FFA, copper, iron and moisture simultaneously over a 4-week period with the same calibration and background scan. The static-optics FTIR instrument – the IRmadillo – has given the performance of a laboratory grade mid-infrared instrument but with the robustness and resilience one would expect from a process ready NIR instrument.
The same instrument has been used to measure residual FFA in refined oils with a completely different concentration range over a similar time period. This work shows that spectroscopic measurements of edible oils can be used for closed-loop control of chemical dosing – opening the way for automated NaOH addition and optimisation of FFA neutralisation.