Emerson Exchange 365

• Not Answered

Split Range Controller Tuning

We are using a single PID block connected to a splitter which outputs to 2 flow control valves (large & small).  Is there documentation available to help configure the PID tuning parameters and splitter block?

Jane Kiekhaefer

• 16 May 2018 1:30 PM
• Cancel

6 Replies

• James Beall

  

  • 16 May 2018 2:48 PM

  There are a couple of different reasons to use spilt range valves. One is to simply provide a wider range of the manipulated variable. In this case, the two valves will have different process gains. This gain difference is also impacted by the setting of the split range (0-50%, 50-100% vs. 0-30%, 30-100%). The "gain scheduling" feature in the library can help deal with this by providing up to 3 different tuning sets depending on a state variable, the PID/OUT in this case. Another case is for a cooling valve and a heating valve. A similar application is pressure control with a valve that increases pressure and one that decreases pressure. Another case is using the little valve for fine control and the big valve to keep the small valve in an operating range (e.g. 20-70%). Lot's to talk about on this topic! There is some information on this topic in Books Online. Let me know if you have more specific questions.

  • Cancel
• James Beall

  

  • 17 May 2018 9:29 PM

  In reply to James Beall:

  You can search for "spilt range" in DeltaV Books On-Line and you will see several useful references to using the Splitter block. You can search for "PID_GAINSCHED" to find out how to use the Gain Scheduling module (or copy functionality). Note there is a special faceplate and detail display for the Gain Scheduling module. The faceplate and detail contained in Emerson's PCSD configuration standard library will automatically pickup and display the parameters for gain scheduling.

  • Cancel
• John Rezabek

  

  • 18 May 2018 12:56 PM

  In reply to James Beall:

  In lieu of gain scheduling, why not put the "splitter" (in the form of a CTSEL (sp?) block upstream of two PID's, one for the small flow and one for the large flow. Then each can be tuned for the loop's specific dynamics and gain. Give the "big" flow loop a SP of zero until the little one is at the limits of its useful capacity . . .
  With DeltaV, PID is Free. Up to a point (controller capacity) anyhow . . .
  Advice given w/o warranty or the endorsement of my employer - PHA and testing is recommended.

  • Cancel
• James Beall

  

  • 18 May 2018 1:13 PM

  In reply to John Rezabek:

  Good idea if you have the two flows. Often there is no flow or one total flow in which case you canmoy implement the two individual flow loops. If you had the two flows, and implemented two slave flow loops, all you need is the SPLITTER block (no need for the Control Selector) unless I am not thinking of the situation you have in mind. Even then you might need to adjust the tuning on the Master loop if the Primary PV response is significantly different for the two "linear" slave flow loops. FYI, the Gain Scheduler is FREE also! It's in the Library of composite modules (Module Templates/Analog Control). If you have InSight, you could use Adaptive Control (license per loop) to provide up to 5 regions of tuning. As you can tell, this is a broad topic because there are many different cases of using "two valves"!

  • Cancel
• Lou Heavner

  

  • 18 May 2018 1:15 PM

  At the risk of taking this in an undesired direction or sounding like a one trick pony, I would suggest that MPC be considered in lieu of a splitter in many, if not most applications. Instead of making a hand-off from one vale or another, it can effectively coordinate and optimize the control. For the rangeability problem, each flow (or control valve where flowmeter does not exist) is an MV and the valve position of the fine control valve (or perhaps a high or low select of all of the valve positions) becomes a process output along with the control objective (total flow, temperature, composition, etc). In the case where you are using utility streams (heating and cooling) you could optimize the utility use by minimizing the cost of the utility usage, particularly if one utility is more expensive than the other. Splitters are a useful and effective tool for advanced regulatory control, and they require no additional licensing. However, they do require some skill and time to deploy properly, so they are not free. Typical applications would have relatively fast dynamics, and could be executed at a high frequency with model predictive control. Similarly, identifying the required models and commissioning would not take long, either. In the case where dynamic response is quite different (such as with an extruder temperature controller), it can be very beneficial to consider MPC.

  • Cancel
• James Beall

  

  • 18 May 2018 1:45 PM

  In reply to Lou Heavner:

  There is a very nice Advanced Application note on using MPC for multiple valves. Search for "DeltaV Advanced-Application Note" and look for DeltaV Advanced Application Note 2, "MPC Implementation Methods for the Optimization of the Response of Control Valves to Reduce Variability". As Lou points out, the MPC will have an individual model (gain and time constants) for each valve!

  • Cancel