In my previous article “Reducing Process Variation Pays Off”, I discussed the importance of variation and how you can increase production and profit by reducing variation in your Bayer plant. In this article, I will expand on how to implement feedback alumina to caustic ratio control (A/C ratio control).
What is A/C ratio?
The A/C ratio is one of the most important process parameters in an alumina refinery. It measures how much alumina can be dissolved into the spent liquor. The alumina is then extracted in the precipitation section of the plant. The more alumina that can be dissolved into the spent liquor the better, but there is a limit. If too much alumina is digested initially as the pregnant liquor cools down, the alumina will precipitate out in the wrong section of the plant (ie not in the precipitation section) wasting a lot of money and causing other process issues.
A/C Ratio Feedback Control
In the past, A/C ratio control has been done through taking a sample every few hours, analysing it and then changing the ratio of bauxite slurry (which contains the new alumina) to spent liquor to compensate. As you can imagine this manual control led to large variations in the actual A/C achieved. To ensure that the A/C didn’t reach a level where precipitation in the settler was an issue, the target A/C was reduced to allow for variation thus reducing potential production.
PLA has developed instrumentation (AL-DCIK A/C Conductivity Sensor) that, when utilised with specially designed correlations can provide a continuous measurement of the A/C ratio in the stream which leaves digestion. This allows for continuous adjustment of the ratio of bauxite slurry to spent liquor which alters the A/C measurement. Because this is continuously occurring, variation is reduced. This is an example of feedback ratio control. The control is considered feedback because it relies on A/C measurement to change before the ratio is changed. An example of A/C ratio control is shown in Figure 1.
Figure 1. Typical A/C Control set up
How dead time affects A/C ratio feedback control
Feedback A/C ratio control is a common strategy in industry but there is a complication with this application, dead time. Dead time is the difference in time between when a change in made to the controlled variable (the ratio between the slurry and spent liquor flow in this application) and when the effect of this change is observed by the A/C instrument. In process terms, this is the time it takes for the slurry stream to be digested.
Dead time in a digestion unit can be calculated by :
1. Measuring the total flow rate (slurry + spent liquor) through the digestion unit (in m3/hr)
2. Measuring the diameter (in m) of the pipework and calculating the area (in m2)
3. Measuring the total pipe length in the digestion unit (m)
4. Dividing the flow rate (1) and cross sectional of the pipe (2) which will provide the velocity of the flow (in m/s).
5. Dividing the velocity of the flow (in m/s) by the length of the pipe (m) giving a time in seconds.
6. Any residence time in any vessels should also be added. This can be calculated by dividing the active volume of the vessel (in m3) by the total flow rate (in m3 / hr).
In a typical high pressure digestion circuit this may in the order of 20 minutes.
The dead time makes feedback control difficult as changes are occurring based on the way the system was, not necessarily as it currently is. To solve this issue the two most commonly used methods are:
1. A Smith Predictor (as shown in Figure 2). For more information on Smith Predictors a good resource is ControlEng.com.
2. Neural Network. These tend to be more complicated but can deliver better results than a Smith predictor.
Without dead time compensation A/C ratio control becomes ineffective and can increase the variation in the A/C. With dead time compensation a substantial reduction in variation is possible and has been achieved at many alumina refineries across the world.
Figure 2. Diagram of generic Smith Predictor (Source : https://www.controleng.com/single-article/overcoming-process-deadtime-with-a-smith-predictor/8918a2bc933af5cb40967d5427c0afd7.html)
Once you achieve successful implementation of feedback ratio control with dead time you will see a substantial reduction in your A/C variation which allows A/C set point increases and more production. Once this system is stabilised and working properly, the next step is to implement a feedforward component for your A/C ratio control. This will be described in the next article.
For more information contact our technical team at firstname.lastname@example.org or call us on +61 3 9786 1711
Category: Process Control