Emerson Exchange 365

I hate to see knowledge exchange left in the obscurity of an email inbox. Here's a recent exchange I wanted to share in case it helps someone with similar questions.

Question: 

I am having some trouble with the Cascaded Master controller tuning in the DeltaV. The masters are usually looking at level or pressure and cascaded to flows. I usually use low to moderate Proportional (2 to 12) , very little if any Integral( >400) and then Derivative anywhere between 5 and 50.

I notice that that even though the error is small in the master controller, often the output will change will change rapidly up or down. I can sometimes see in my historian that there may have been a small noise or bump in the master Controller process variable.

I am wondering if the standard control algorithm where P& I work on the error and Derivative looks at the PV may be causing this? I am going to change equations in the Master to the P,I & D work on the error collectively, to see if I get a more buffed response from the algorithm.

From what I have described, do you have any thoughts?

I shared this question with Emerson's James Beall, whom I've featured many times on the Emerson Process Experts blog. Here's a link to one of his Emerson Exchange presentations, Interesting and Useful Features of the DeltaV PID Controller.

James responded:

I’ll be glad to visit with you regarding your questions on Cascade control in DeltaV. Here are some comments on the topic in preparation for a discussion if needed.

• Changing the Derivative to work on Error instead of PV will not make any difference in the response when the Setpoint is NOT being changed. However, Derivative on Error will cause an output spike up/down (or down/up) when the Setpoint is changed. I generally do not apply Derivative to the error, especially on Master loops.
   
• Noise on the PV will cause the output to jump due to both the Gain and Derivative action. Note that the Derivative is applied after the gain is applied so a large gain and a large derivative (Rate) setting will increase the output movement.
   
• A separate, first order filter is applied to the signal to which the derivative action is applied. The time constant of this filter is equal to the PID Parameter “Alpha” times the Rate setting. Alpha has a default value of 0.125.
   
• If there is noise on the PV, a shorter (faster) the module execution time can “amplify” the derivative action because the rate of change is calculated over the scan time. Often, level and slow pressure loops can be executed at a 5 or 10 second rate with no loss of control performance.
   
• A PV filter can be applied if there is excessive noise on the PV but there are guidelines about this that we can discuss.
   
• Most level loops and some pressure loops have an “dead time plus integrating” (vs. a self-regulating) response. Derivative action is only helpful if there is a significant “lag” in addition to the dead time. Perhaps Derivative is not needed.
  
  
• Make sure the ARW low and high limits on the PID block are slightly OUTSIDE the out scale values.  The default for the ARW limits is 0-100  but if the output scale range is changed to something outside of 0-100, like 0-5000 lb/hr, the ARW limits are not automatically changed.