Re: Tuning integrating CVs in DMC+

[ Follow Ups ] [ Post Followup ] [ Control Consulting ] [ FAQ ]

Posted by Dan O'Connor on May 01, 2001 at 12:01:21:

In Reply to: Tuning integrating CVs in DMC+ posted by Keith Landells on May 01, 2001 at 07:35:41:

: Has anyone developed any ways to tune integrators with large capacity in DMC so that the buffer volume is fully utilised ? We have plants where MVs/CVs span across buffer tanks that hold many hours of production. We want to be able to fully use the tanks so that we can tolerate rises in level over 8hrs, say, in order to deal with downstream CV disturbances without significantly changing the upstream MVs (reactors). The only way we can approach the response we want using Simulate is to use the programmed imbalance facility - but this has no way to drive the level back to our ideal comfort value and just ramps to max or min.

: DMC+ just loves to balance ramps quickly, but we have a Tss of 180mins and a tank that can run imbalanced for many shifts without causing operational problems.

: Should we place so much emphasis on Simulate results or should we put the controller on the plant and see what happens ?!

Keith,
Two points first: 1. You are absolutely right to be using a programmed imbalance ramp. A traditional ramp must balance at steady state. For large vessels, this means that
the controller isn't able to run the MV's with enough material imbalance to get to the target level. The programmed imbalance runs with a material imbalance based on the level pv,
the upper and lower limits, and the ramp horizon.
2. I would put the emphasis on Simulate - Although you may find that during the day, the controller must cut back, and that at night, it may increase rates again, an thus the level will
drift rather than hit a high or low limit. Whether it drifts, or whether it runs at the level limits probably depends on how much upstream feed is in the plant. If you are truly maximizing rate,
you hit different limits. Again, maybe during the day, the downstream part limits, and at night, the upstream. For this case, the level will fill during the day, and fall at night. This may be a
very desirable (and profitable) behavior.

If you still believe there are times that you would like to drive the ramp back towards a level setpoint which is in between the CV limits of the ramp, I may have a solution for this as well. The idea is to put the ramp in the controller again as a
pseudo-ramp, rank 1000 CV. In this way, the controller can give up on the less important ramp for economics. If you have some questions on this, feel free to give me a call. Make sure your rotation factors
are the same (and very small) for the two ramps. Make sure you only bring the PV into the controller interface once, and pass the value over to the other level CV, so the predictions will be identical.

Hope this helps.
Dan

Follow Ups:

Post a Followup

Name    :

E-Mail  :

Subject :

Comments:

: : Has anyone developed any ways to tune integrators with large capacity in DMC so that the buffer volume is fully utilised ? We have plants where MVs/CVs span across buffer tanks that hold many hours of production. We want to be able to fully use the tanks so that we can tolerate rises in level over 8hrs, say, in order to deal with downstream CV disturbances without significantly changing the upstream MVs (reactors). The only way we can approach the response we want using Simulate is to use the programmed imbalance facility - but this has no way to drive the level back to our ideal comfort value and just ramps to max or min.
: : DMC+ just loves to balance ramps quickly, but we have a Tss of 180mins and a tank that can run imbalanced for many shifts without causing operational problems.
: : Should we place so much emphasis on Simulate results or should we put the controller on the plant and see what happens ?!
: Keith,
: Two points first: 1. You are absolutely right to be using a programmed imbalance ramp. A traditional ramp must balance at steady state. For large vessels, this means that
: the controller isn't able to run the MV's with enough material imbalance to get to the target level. The programmed imbalance runs with a material imbalance based on the level pv,
: the upper and lower limits, and the ramp horizon.
: 2. I would put the emphasis on Simulate - Although you may find that during the day, the controller must cut back, and that at night, it may increase rates again, an thus the level will
: drift rather than hit a high or low limit. Whether it drifts, or whether it runs at the level limits probably depends on how much upstream feed is in the plant. If you are truly maximizing rate,
: you hit different limits. Again, maybe during the day, the downstream part limits, and at night, the upstream. For this case, the level will fill during the day, and fall at night. This may be a
: very desirable (and profitable) behavior.
: If you still believe there are times that you would like to drive the ramp back towards a level setpoint which is in between the CV limits of the ramp, I may have a solution for this as well. The idea is to put the ramp in the controller again as a
: pseudo-ramp, rank 1000 CV. In this way, the controller can give up on the less important ramp for economics. If you have some questions on this, feel free to give me a call. Make sure your rotation factors
: are the same (and very small) for the two ramps. Make sure you only bring the PV into the controller interface once, and pass the value over to the other level CV, so the predictions will be identical.
: Hope this helps.
: Dan
:

Optional Link URL:

Link Title:

Optional Image URL:

[ Follow Ups ] [ Post Followup ] [ Control Consulting ] [ FAQ ]