It is vital, when optimising a control loop, to establish at the outset if the process is self-regulating or integrating, as not only do these two process types behave completely differently ,but also they are tuned very differently. A typical self-regulating process is a flow where if you make a step-change to the input to the process, the output moves to a new value and then remains constant at that value. An integrating process on the other hand is one where if the input to, and the output from, the process are different, then the output is always either ramping up or down. It is only constant when the output is equal to the input. It is known as a balancing process. A very good example of an integrating process is the control of a level in a vessel.
Unfortunately, it is sometimes very difficult to decide if a process is actually self-regulating or integrating. For example, certain processes, like certain temperature and pressure control of compressible fluids like gases and steam, sometimes seem to exhibit both characteristics. A good example of this is level control of a liquid in a tank that has not got a pump on the output of the tank. We call this a gravity feed tank. The output flow is dependent on the force of gravity acting on the height of the liquid in the tank. Now, if at a certain point in time the input and output flows are equal, the level remains constant. If one now makes a small step increase in the output flow, the level immediately starts ramping down, but as the level decreases, the head of liquid in the tank is getting smaller and hence the output flow starts decreasing as well. If you have not made too big a step, the level will actually eventually balance itself out. The resultant total curve recorded in the level now looks exactly like a self-regulating process. However, level is definitely integrating. So how do we define it?
Basically what has effectively occurred is that the process started reacting as an integrator and then changed into a self-regulator. There are quite a few cases where this happens, and there are also some odd cases where a process starts as self-regulating and turns into an integrator. The important rule for tuning is that in these cases you always tune on the initial reaction.
It is sometimes quite difficult to decide if a process is better tuned as an integrator or as a self-regulator, and it sometimes takes quite a long time, especially on very slow processes like some temperature control processes. We teach how best to deal with these cases in our courses.
Generally, level is always integrating, but there are occasional cases where the tests make it look self-regulating. An example of this is the gravity feed tank level described above. However, I recently came across an unusual case of level control where the test initially appeared to be giving a truly self-regulating response. This was in the control of levels of cooling towers on a mineral processing plant.
Figure 1 is a recording of an As Found Closed Loop test on one of the towers, where a setpoint (SP) step change is made on the controller, which is in automatic, with the original tuning parameters set in it. The level was in a small continuous slow cycle. It can be seen that the response after the SP was stepped was also rather slow and very cyclic. This type of action is very typical of cycling on integrating processes and is usually caused by either bad tuning (commonly as the result of too fast an Integral (I) setting, or by hysteresis in the valve, and with P+I action in the controller.
Figure 2 shows the Open Loop test we performed. We first got the process into balance with the PD (controller output) at a value where the level PV remained constant. With the controller in manual, a step change was made on the PD. It can be seen that the level started ramping down, but after a while the ramp rate started decreasing and eventually the PV became constant. This would certainly appear to be a self-regulating process, however level is nearly always definitely integrating.
So why did this become self-regulating? We could not initially see any reason for this behaviour, as the level was not gravity fed. However, the process was configured in rather an unusual manner. The liquid feeding the cooling tower was pumped into the tower via the control valve. Then there was also a pump on the output of the vessel which one would think would keep the flow out of the tank pretty constant. This, on the face of it, definitely appears to be a true integrating process.
The only explanation for the behaviour of the actual response that we could think of would be that the output pump’s characteristics were affected by the head of liquid in the tower. Unfortunately in the time available it was not possible to get details of the pump’s characteristics, but it seems to be the only logical explanation.
In any event we did try tuning it both as a self-regulator and then as an integrator. The tuning as an integrator worked much better. Figure 3 shows the window we used for the integrator tuning.
Figure 4 shows a section of the final Closed Loop test with the new tuning. The response was still a little cyclic but the PV settled out after one cycle and stuck to SP at normal conditions without cycling. It was working much better.
The new tuning had much the same P gain as the old, but the integral time was double.
This was an interesting example of the difficulties one can encounter when trying to identify if a process is integrating or self-regulating.
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 now presents courses and performs optimisation over the internet. His work has taken him to plants all over South Africa, and also to other countries.
About Michael Brown
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 now presents courses and performs optimisation over the internet.
His work has taken him to plants all over South Africa and also to other countries. He can be contacted at: Michael Brown Control Engineering CC,
| Email: | michael.brown@mweb.co.za |
| www: | www.controlloop.co.za |
| Articles: | More information and articles about Michael Brown Control Engineering |
© Technews Publishing (Pty) Ltd | All Rights Reserved
printer friendly version
