Chalco evaluated various options to get better transparency on their bauxite composition which they receive from a Guinea based supplier.
What is the benefit of real time information?
In the Bayer Process, one of the key process parameters is the alumina to caustic ratio (A/C ratio) at the back end of digestion. This parameter is controlled by varying the ratio of spent liquor and bauxite slurry entering the digestion train. If the A/C ratio is too high precipitation of alumina occurs in the settler which causes process problems. To compensate for the variation in the process the set point is reduced so that the A/C ratio doesn’t reach a point where precipitation occurs in the settler. By doing this, there is always a safety margin in place to avoid precipitation in the settler. The cost of the safety margin, however, is that some of the alumina will not be recovered.
If the set point can be increased without precipitation occurring in the settler (due to the reduced variation) then there will be an increase in production. The easiest way to do this is to control the A/C ratio continuously in a preventive way: by knowing the available alumina (AA) and reactive silica (RSi) content going into the process and hence, being able to compensate for the variation in the material as it moves through the process.
With the information on AA and RSi available in real time to the DCS, the feed of caustic soda can be optimized in real time.
Figure 1. The Bayer Process (curtesy of PLA)
Additional recovery of Al2O3 from the bauxite ore is not the only benefit. Quantification of the reactive silica will also help find the necessary volume and concentration of caustic soda to be added, to ensure a sufficient digestion process.
Having covered the “WHY”, we can move on to the “HOW”. Chalco looked at different options, from sampling to different online analytical methods.
The first criteria for Chalco was the possibility to analyze the AA and RSi. The use of a radioactive (PGNAA or PFTNA / CNA) or laser based (LIBS) analysis technologies were consequently ruled out. Both technologies are online elemental analyzers: their sources (radioactive material or laser) emit a strong energy which interacts with the elements of the material passing by on the belt. These technologies are only able to provide inaccurate qualitative information on AA or RSi by way of assumed correlations, which become more unreliable with any variation in the mineralogy of the bauxite.
The remaining options were an auto sampling system or real-time near infra-red (NIR) online analysis, which is sensitive to mineral phases.
After careful consideration, the auto sampling was ruled out with the main reasons being:
There was no real price advantage to an online system, especially considering the increased lab work (or automation level) over time;
Representative sampling of coarse/lumpy bulk material on fast moving conveyor belts is difficult and requires labour-intensive maintenance on the equipment;
There is still a considerable time delay in the analysis and dependency on the availability of the lab equipment and staff; and finally
The necessitated dual conveyor belt setup (found in many plants) to ensure redundancy and allow for maintenance, and consequent cost, labour and time.
The final option, SpectraFlow’s NIR real-time analysis technology fulfilled all the requirements:
Reduced sampling requirement
The technology is sensitive to mineralogy, organics and moisture – all in one detector
Single analyzer installation can be made over a dual conveyor belt setup (pictured below)
Figure 2. Chalco SFA Crossbelt Analyzer installation view.
One important follow on question is on how NIR technology can measure the mineral phases, and hence AA and RSi in real time, compared to other measurement technologies?
SpectraFlow uses halogen light bulbs as sources. These bulbs emit light. Part of that light energy is in the near infra-red wave length range (wave length 0.5 to 2.5 µm). Light is a completely safe enegery without no major hazards associated. The light energy is thus of too low intensity to interact with the individual elements of the material mix on the belt. Instead, the energy is absorbed in the bonding structures of the minerals and organics of the material mix. As the bonding structures of Gibbsite, Boehmite, Kaolinite, etc. differ from each other, they each have a different absorbance of energy in the NIR wavelength range. This can best be demonstrated by spectra of pure Gibbsite, Diaspore, Kaolinite, etc on the webpage of the US Geological Survey (https://crustal.usgs.gov/speclab/QueryAll07a.php) where one can find their comprehensive and up to date spectral NIR library. By leveraging NIR technology and deep learning models, SpectraFlow is able to measure the differences in the behavior of the various aluminium minerals (Gibbsite, Diaspore and Kaolinite).
Figure 3. Gibbsite, Kaolonite and Diaspore spectra, from left to right respectively.
SpectraFlow combines 150 Spectra collected over a minute to output a combined “fingerprint” (spectra) for the absorbance of the varying minerals and organic molecules during that time period. This fingerprint is then interpreted by the calibration model to output calibrated concentration predictions of total alumina (in oxide form, as is commonly used in the lab: Al2O3), available alumina, SiO2, reactive silica, Fe2O3, TiO2, K2O, Na2O (or even sodium oxelate – Na2C2O4), total Carbon, CaO, Mgo, etc. These results are sent on a minute-by-minute basis to the client's DCS. With this knowledge in the control room, Chalco can now start adjusting the caustic soda / spent liquor and dosing adjustments, first manually and then, once confidence is formed, in a closed or automated loop control.
In this way, Chalco benefits from an optimized aluminium recovery leading to increased production, decreased costs and reduced waste.