Tank Level - Integrating or Self Regulating


Thread Starter


Hi all,

I have a level controlled surge tank, that discharges via a centrifugal pump with VSD.

Most literature I have read describes pumped tank level control as an integrating or non self regulating process. But for a fixed control output (pump speed), the discharge flow rate is affected by the tank level (increasing level = less system head = increased flow per pump curve). This adds a degree of self-regulation to the system, correct? (Unlike say a PD pump, where discharge flow rate is tied only to the control output).

How does this then affect tuning of a PID controller?
If I read your description correctly, the level in your surge tank is controlled by varying the speed of the VSD on the output pump. You do not have a fixed control output because the output pump changes speed. You still have an integrating process.

If the feed to the surge tank changes (load change) then the level in the tank will rise or fall until detected by the level transmitter.

This is a classic level control problem and calls for the PID controller to be set with very little integral action, no derivative action, and a proportional control doing most of the work. Do not expect the level to be controllable close to the setpoint - it will drift. As long as the level stays within the upper and lower control limits, it is properly tuned.
Thanks for your reply. Yes, we are controlling level by varying the speed of the VSD.

But if we have an upset condition, and the tank level increases, then the pump flowrate also naturally increases, before we even send any signal to the VSD. Say we send no signal to the VSD, the level will continue to increase but then should eventually settle at some new equilibrium where the pump has moved right along its curve until its flowrate matches the new incoming flow (assuming there was no upper constraint on the tank level of course). Is this still an integrating process?

In the article:
http://www.chemicalprocessing.com/articles/2011/neglect-level-control-at-your-peril/?start=1 it mentions for an integrating process "Changes in level (and, thus, head) don't affect the discharge flow", but with a centrifugal pump this is not the case.

You are correct - as the surge tank level increases, the pump head increases and therefore the flow through the pump will increase. However, I think you will find that the rate of flow increase due to the change in pump head will not be sufficient to achieve a self-regulating process. It's a matter of how much. To test it, all you need to do is put the level controller in MANUAL and see where the level settles out; as if the process operator will let you try this.

Most level tanks work this way. We always speak of level control as being an integrating process and theoretically no RESET is required, but in fact a little bit of RESET is usually necessary to keep the settling point of the level within the bounds of the tank. The error most beginner control engineers make is to try to make the level control to always balance out exactly at the SETPOINT - but it will not. To try to force the level to be at the SETPOINT is asking too much of the PID controller.