Case History 152: Control problems with airflows in flotation banks
January 2017, System Integration & Control Systems Design
I have written quite a few articles specifically dealing with control problems in mineral processing plants that are largely due to poor understanding, and apparently complete lack of concern of control practice by the plant designers. As a rule, the control systems in these types of plants perform very badly. Apart from the lack of knowledge, the control equipment used in these plants is often purchased with price as the chief criterion, and little thought is given to the long-term effects of poor control on the profitability of the plant.
The problem of airflow control in flotation banks
To offer a simplified description of one problematic process: flotation banks are used to separate the desired minerals from other constituents in a slurry. To achieve this, the slurry is treated with additives and passed through a series of tanks, where air is bubbled through from the bottom forming a ‘bubbly’ froth on top of the slurry. The desired minerals adhere to the bubbles and the ‘rich’ froth overflows the top of the tank. The remaining liquid below the froth gravitates to the next tank where the process is repeated. The three main controls in each tank are pH, liquid level and airflow. The last two are particularly critical for good recovery.
In earlier days, level was regarded as the most important parameter. However, in recent years the metallurgists have realised that the airflow is also very important, and it is becoming common for online froth analysers to be used to control the airflow.
In the course of my career, I have performed optimisation in many of these plants, and of all the controls I have ever come across, the airflow control is the one least likely to work. At first sight this seems strange, as probably flow is the easiest of all processes to control. Even without a scientific tuning package, flow is relatively easy to tune as it has fast response, and one generally finds most flow control loops in plants are working in automatic, even though they are generally not working efficiently in most plants. (The reason for this is, of course, that very few people know how to analyse loop problems and tune the loops properly.)
Why then do the airflow loops on flotation plants seldom operate correctly? The two main reasons are poor measuring transmitters and control valves that are very often installed in these plants. Firstly, plant suppliers often go for a cheap make of flowmeter, which suffers badly from drift and is not at all accurate. They also often seem to stop working properly at high ambient temperatures and give false readings. Secondly, the suppliers install low cost and really bad valves with positioners and actuators that are not good for modulating control systems.
What is quite interesting is that the majority of flotation plant personnel, including the metallurgists, nearly always tell me that their flow controls are working perfectly, when in fact they are really not working at all. Why is that? The reason is that not many plants are yet using the airflow controls cascaded down from froth bubble analysis. The operators put the controllers into manual and get the flow to where they want, and then put the controllers back into automatic, and the flow stays where it is as nothing changes. This makes great trends. The flow on setpoint! Great control! We call these trends ‘tramlines’.
Two practical examples
I was recently working on flotation bank control at a mine in Portugal, where again they were convinced that the airflow controls were working splendidly, but they will be using higher level control soon and wanted to check how good the controls really are. Figure 1 is the recording taken from an open loop test on one of the airflow control loops, and shows just how terrible the valve was. It can be seen that it is not doing what the controller is asking. The controller was in manual and equal steps are being made on the way up, although larger steps (equal in size) are being done on the way down. The valve sticks at times. It sometimes overshoots and cycles. On one occasion, it even moves in the wrong direction. On other occasions, the valve responds with differing size steps for the same size steps on the controller output. All of these things mean that the valve is non-repeatable and does not do what the controller is asking. There is no way that any reasonable automatic control could be achieved with a valve like this.
To compound the problem, every single valve that we tested on the banks was as bad, if not worse, than this. This type of thing is really common on many of the float banks I have worked on. It makes one wonder why so much money is wasted on control, and even more so, as it shows up how plant personnel have so little idea of what is going on with their controls.
To illustrate again how people don’t seem to be aware of what is going on with their controls, the second example is of airflow on a float cell, where everyone was quite happy that the control was fine. Figure 2 shows an open loop flow test. The operators just set the setpoint at a particular value, and left it. Nobody could have checked anything for as can be seen in the test the valve was jammed solidly in one position and although the controller’s output was stroked from zero to 100%, the flow never moved, and was stuck at 4%. These things are hard to believe. Surely, the cell’s performance must have been bad, but no one was worried about it.
I have often said that feedback control is a grey area, of which few people have any real practical knowledge, and these examples illustrate what I mean.
Michael Brown received some feedback from Alastair Nuttall, a control engineer in the Western Cape, on the Case History 150 article which appeared in the September issue. We have decided to publish it along with Michael’s reply, as there may be other readers out there who will benefit.
“Good Day Michael,
First of all I’d like to commend you on your articles in the SAI&C magazine, I don’t always get the chance to read every issue, but when I do get to one, your article is always my first port of call. I find them very well written and so easy to follow.
In my career I haven’t had much (if any) exposure to the control of distillation columns, but I often read about them and was taught about them at tech. I spent my first few ‘appy’ years working as an instrument tech for Illovo Sugar, but have subsequently found my calling in Building Management Systems (BMS).
In the BMS field, we encounter quite a few heat exchangers in most of our projects, but nothing on the scale of what you’d be writing about. Nevertheless the principle is still the same.
You’ve probably investigated these things but here are some of the questions that I found myself asking when reading your article:
1. What flowmeter are you using on the MV steam side? I’ve had instances where a dp flowmeter has had some integral settings incorrectly left in the meter itself, which gave me incorrect readings.
2. What is the valve position during your sudden rise of temperature? Is the valve operating at its lower end of stroke indicating an oversized valve?
3. This next question is hard for me to ask without losing any integrity I may have gained up to this point, but someone has to ask the stupid question at some stage: In the construction of your heat exchanger, is there any mechanical way that the primary side could be leaking/passing into the secondary side and affecting your readings? As silly as it may seem, this is the only way I can account for both the temperature and the pressure increasing at the moment a setpoint change occurs. It may constantly be passing ever so slightly into the secondary side that under normal operation your pressure controller is always compensating for it. And because your flowmeter is installed downstream of the HX this slight flow isn’t being picked up. A differential temperature measurement across the HX primary side could help indicate there is some flow with the valve closed.
I really have so many questions for you at this stage but I’ll leave it at that for now. I look forward to hearing back from you as to what the problem turned out to be.
Sorry to have taken so long to get back to you, but I have been away overseas, and very busy on my return.
I really welcome your comments, and appreciate the time and thoughts. In answer to your points which are interesting, and are about things that people often don’t think about:
1. The flowmeter is indeed a dp cell and orifice. I don’t think there were any problems with the measurement; however, I was also worried that the measurement was on the downstream side of the exchanger. As far as I can remember all other systems I have worked on have had the flowmeter installed on the upstream steam flow. It may well be worthwhile changing the measurement point. However, practically it is unlikely the plant will do this as there are so many other priorities, and we did get the control working reasonably well.
2. The valve was definitely not oversized and was working between 50% and 70% of opening. Normal flow tests on the valve showed it was working well with no noticeable problems. As shown in Figure 1 in the article it did have a cascade flow secondary to the temperature, and the cascade flow secondary will normally take care of all the valve problems.
3. I have also queried the possibility of leakage from the one to the other side of heat exchangers in the past, and this would obviously have an immense influence on the dynamics, but in this particular case it would be unlikely to have caused the effect we encountered. A small leakage would be very hard to detect on a working exchanger, and I would guess could only be detected with tests when the plant was offline. Naturally they will never take a plant offline if they can still make product. As mentioned above, the process people don’t worry too much as long as you can get them reasonable control.
You mention that it would be interesting to get feedback once it is determined as to what caused the problem. Unfortunately I seldom ever get feedback myself from plants. The plant people live in a high pressure environment and spend much of their time ‘fighting fires’. My job is to uncover problems in their control systems for them, but it is extremely seldom that they ever come back and tell me if they have fixed the problems. They are just too busy keeping the plants running.
It was really great getting your contribution Alastair. I really welcome comments from readers. The points you have raised could also possibly help other people in the field with problems of their own. As I said earlier, problems caused by things like incorrect measurement, oversized or incorrectly characterised valves, are often completely overlooked. The mantra in many people’s minds is to try and solve everything by tuning.
Michael Brown is a specialist in control loop optimisation with many years of experience in process control instrumentation. His main activities are consulting, and teaching practical control loop analysis and optimisation. He gives training courses which can be held in clients’ plants, where students can have the added benefit of practising on live loops. His work takes him to plants all over South Africa and also to other countries. He can be contacted at Michael Brown Control Engineering cc, +27 (0)82 440 7790, firstname.lastname@example.org, www.controlloop.co.za