Emerson Exchange 365

Tuning a Flow Loop with Unstable PV

Hello, 

Today I attempt to tune a flow loop that is normally set in AUTO. 

When I place the loop in MAN to stabilize the flow before step change, I notice the PV (actually flow) was still oscillating and not stable. Normally when I place a loop in MAN, the valve is constant output and PV stays stable. 

I believe, I need to check the flow meter for issues and make sure the valve is not moving.

Has this happen to you before?

Thanks,

DCS Newbie

• 1 Jun 2016 3:29 AM
• Cancel

7 Replies

• leirbag

  

  • 1 Jun 2016 9:58 AM

  1. Check first if your flow is single in the line (there is no other flow from the same pipe).
  2. Check if there is no other control element (automatic or manual) on the line.
  3. Check the flowmeter and increase the damping time (more than 2-3 sec) if necessary.
  4. Check the control valve if oscillating due to mechanical problems, positioner malfunction problems or booster relay gain problems (if booster relay exist).

  • Cancel
• DCS Newbie

  

  • 1 Jun 2016 12:28 PM

  In reply to leirbag:

  leirbag said:

  1. Check first if your flow is single in the line (there is no other flow from the same pipe).
  2. Check if there is no other control element (automatic or manual) on the line.
  3. Check the flowmeter and increase the damping time (more than 2-3 sec) if necessary.
  4. Check the control valve if oscillating due to mechanical problems, positioner malfunction problems or booster relay gain problems (if booster relay exist).

  Thank you leirbag for response. I will definitely check the steps above before tuning.

  How does damping time effect the flow measurement?

  Thanks,

  • Cancel
• leirbag

  

  • 1 Jun 2016 5:47 PM

  In reply to DCS Newbie:

  The traditional signal filtering method for getting a stable output in noisy applications is to add damping time. Emerson magmeter transmitter default damping time is 2 sec. While this makes the output look stable, the flowmeter’s response time to actual process changes is compromised, if this constant is very high. I sometimes use such constant up to 10 sec. This is a decision to be taken only when there is a disturbing factor in the process that produces those sudden swings in flow.

  • Cancel
• DCS Newbie

  

  • 2 Jun 2016 12:54 PM

  In reply to leirbag:

  leirbag said:

  The traditional signal filtering method for getting a stable output in noisy applications is to add damping time. Emerson magmeter transmitter default damping time is 2 sec. While this makes the output look stable, the flowmeter’s response time to actual process changes is compromised, if this constant is very high. I sometimes use such constant up to 10 sec. This is a decision to be taken only when there is a disturbing factor in the process that produces those sudden swings in flow.

  Thank you Ieirbag,

  How does damping time in the transmitter relates to the PV Filter? Does this mean the actual flow measurement can be damp on the DCS side as well?

  

  Thank you for help. 

  • Cancel
• Andre Dicaire

  

  • 2 Jun 2016 5:30 PM

  In reply to DCS Newbie:

  As Leirbag explained, adding damping to suppress noise adds to the overall lag in the process response as seen by the PID algorithm. The process may appear stable, when it is not. Addressing the noise directly and closest to the source is best, but often not feasible. Damping in the device is the next best place, but this affects the process response and the loop may need to be retuned when filter is changed. Using the PV Filter is an option, but you should first make sure your channel filter is set to mitigate signal aliasing at the module.
  
  Each AI channel in DeltaV has a hardware anti-aliasing filter designed to work with the IO channel scan rate, which is hard coded I DeltaV. There is a channel filter that users should set based on the scan rate of the control module to mitigate aliasing at the module. Set this to 1.35 Sec for a 1 second module. The filter has fixed choices based on module scan rate.
  
  If additional filtering is desired at the module, the AI block has a filter, as does the PID, and one can add a filter block as needed. The PV Filter TC parameter in a PID Detail will typically point to the PID/PV_FTIME parameter. Any of these filters result in additional lag in the process as seen by the PID, and too much filtering will degrade the performance of the loop. A smaller filter time allows higher frequencies to pass.
  
  Anyway, start with transmitter damping and channel anti aliasing filter. Add module filtering only if needed.
  
  For process oscillation, look to address this at the source. Otherwise, your loop will be constantly moving, which wastes energy. Oscillations may be caused by upstream integrating process like level and pressure where those control loops are tuned with too much integral action. Adding a filter on this loop to mask oscillation will result in a perceived stable loop, but the oscillation will pass through to the downstream process.

  Andre Dicaire

  • Cancel
• Gareld Butler

  

  • 3 Jun 2016 2:35 AM

  A couple more things to look for: Is the final control element a VFD? In the past I have found a poorly tuned VFD (or positioner) that was introducing its own disturbance, even though the output of the PID loop was steady. Another possibility is if there is a positive displacement pump upstream of the flow. If you have this, and the disturbance is a consistent pulse, that may be your problem. It is very difficult to tune a loop (and impossible to tune using most auto-tuning software) if the flow is not stable when the loop output is constant. You need to find what is causing the disturbance and find some way to make this steady while you are tuning this loop.

  Gareld Butler

  • Cancel
• neilrbrown

  

  • 25 Jun 2016 5:13 AM

  You are right to check the flow meter (‘garbage in, garbage out’).

   

  Other possibilities you should check:

   

  - oscillation is caused by the process (or another controller) upstream or downstream of the control valve

  - control valve has stiction and/or hysteresis, and valve positioner has integral control action (this will cause the valve to move even when the loop is in MAN)

  - valve positioner is simply faulty

  - what you see as oscillation is actually ‘noise’ (process and/or measurement) – try applying a mild PV filter (1 or 2 seconds; flow loops are usually quite fast so you can’t filter them too much)

   

  Hope that helps,

   

  Neil Brown

   

   

  • Cancel