Process Derivative Action in an AC Drive


Thread Starter

Vineet Parikh

We plan to impliment AC Drive for control of Fluid Pump to control Pressure.

Allen Bradley AC Drive provides only PI Control action while many other manufactures provide PID action. We already have a few AB drives in the plant and would prefer buying AB.

Can anybody please guide me if Derivative action is really required? AB feels we do not need it in most application. Im afraid that I might buy AB and it won't work on our application.

Please Help!?!
In general, you need a derivative gain term in your control loop if the derivative of the control quantity "wants" to maintain its value, AND there is no inner loop explicitly controlling that quantity.

In a position loop, the derivative quantity of velocity wants to maintain its value due to momentum, so you either need derivative position-loop gain, or an inner velocity loop.

In a velocity loop, the derivative quantity of acceleration does not want to maintain its value, so you usually don't see derivative gain used in a velocity loop. Note that the motor current level, which is pretty much proportional to acceleration, does want to maintain its value due to inductance, but there is always a current loop in any serious drive. Besides, in an AC motor, you have AC currents for DC torque.

I am presuming that you are talking about a PI velocity loop. Unless there is a process effect that introduces a need for derivative gain, I think the PI will be sufficient.

Curt Wilson
Delta Tau Data Systems


Dear Vineet:

For your application, if you must have a very fast response then you must use PID controller, because the derivative term speeds up the
controller responce . In this case you can order the ABB ACS600 AC drive and if the fast response is not needed then you can use the PI controller only and it will be sufficient.

Best regards,

[email protected]
When you use a variable speed pump to control the pressure in your liquid distribution system, you definitely DO NOT WANT derivative in your control
I have done dozens of such systems (at IBM and other places) and liquids being incompressible, the pressure measurement is too fast and found without exception, that this application is too noisy for using derivative.
In the past, all the successful variable speed pump pressure control algorithms were the PI type with tuning ranges of: 0.2 to 1.0 gain and 0.5 to
2 secons/repeat integral.

Béla LiptákInc.

Rajesh Mehta

Dear Vineet,

If we go back to the control system fundamentas, you need the derivative action to inculcate predictive nature in the loop.

In real life terms what it means is that a small deviation in the measured variable will result in a large change the output. The derivative action comes into picture when you have changes occuring very fast.

So, you tend to keep value of D very low so that any small noise/ deviations do not cause unnecessary destabilization of process.

All said and done, it is a nice feature to have but seldom used in practice until you have no noise and want to have rock steady output.
Derivative action is basically how fast the deviation from PV to setpoint is changing. With typical centrifugal or similar loads only PI control is generally necessary. Derivative action is helpful in certain slow reacting systems such as thermal reactions (furnaces, etc.). Unless you have or will have some very unique application requirements, PI generally will do a good job.

John Catch
[email protected]