1 / 39

Closed-loop model identification and PID/PI tuning for robust anti-slug control

Closed-loop model identification and PID/PI tuning for robust anti-slug control Esmaeil Jahanshahi Sigurd Skogestad Department of Chemical Engineering, NTNU, Trondheim, Norway. Outline. Introduction New 4-state nonlinear model New procedure to identify linear unstable slug-model

crolando
Download Presentation

Closed-loop model identification and PID/PI tuning for robust anti-slug control

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Closed-loop model identification and PID/PI tuning for robust anti-slug control Esmaeil Jahanshahi Sigurd Skogestad Department of Chemical Engineering, NTNU, Trondheim, Norway

  2. Outline • Introduction • New 4-state nonlinear model • New procedure to identify linear unstable slug-model • IMC Design for unstable slug process • PID(F) and PI tuning • Dealing with nonlinearity: • Gain scheduling • Adaptive tuning • Experiments • OLGA Simulations

  3. Slug cycle (stable limit cycle) Experiments performed by the Multiphase Laboratory, NTNU

  4. Experimental mini-loop (2003) Ingvald Bårdsen, Espen Storkaas, Heidi Sivertsen

  5. z p2 p1 Experimental mini-loopValve opening (z) = 100% SLUGGING

  6. z p2 p1 Experimental mini-loopValve opening (z) = 25% SLUGGING

  7. z p2 p1 Experimental mini-loopValve opening (z) = 15% NO SLUG

  8. z p2 p1 Experimental mini-loop:Bifurcation diagram No slug Valve opening z % Slugging

  9. How to avoid slugging?

  10. z p2 p1 Design change Avoid slugging:1. Close valve (but increases pressure) No slugging when valve is closed Valve opening z %

  11. z p2 p1 Design change Avoid slugging:2. Design change to avoid slugging

  12. z p2 p1 Design change Minimize effect of slugging:3. Build large slug-catcher • Most common strategy in practice

  13. ref PC PT Avoid slugging:4. ”Active” feedback control z p1 Simple PI-controller

  14. Anti slug control: Mini-loop experiments p1 [bar] z [%] Controller ON Controller OFF

  15. Anti slug control: Full-scale offshore experiments at Hod-Vallhall field (Havre,1999)

  16. Summary anti slug control (2008)* • Stabilization of desired non-slug flow regime = $$$$! • Stabilization using downhole pressure simple • Stabilization using topside measurements difficult • Control can make a difference! • “Only” problem: Not sufficiently robust *Thanks to: Espen Storkaas + Heidi Sivertsen + Håkon Dahl-Olsen + Ingvald Bårdsen

  17. 2009-2013: Esmaeil Jahanshahi, PhD-work supported by Siemens 1st step: New Experimental mini-rig 3m

  18. 2nd step: New Simplified 4-state model* Choke valve with opening Z x3, P2,VG2, ρG2 , HLT P0 wmix,out L3 x1, P1,VG1, ρG1, HL1 x4 L2 h>hc wG,lp=0 wL,in wG,in w x2 wL,lp h L1 hc θ State equations (mass conservations law): *Based on Storkaas model

  19. New 4-state model. Comparison with experiments: Experiment Top pressure Subsea pressure Nonlinear: Process gain = slope - approaches zero for large z

  20. 3rd step: Experimental linear model (new approach) Fourth-order mechanistic model: Model reduction: 7 parameters need to be estimated Hankel Singular Values: 4 parameters need to be estimated Stable part has little dynamic effect

  21. Model Identification: Closed-loop step response using P-controller Experiment 1: Z=20% (valve opening)

  22. Comparison with mechanistic model. Z=20% Identified model: Mechanistic model: Excellent agreement!

  23. Comparison with mechanistic model. Z=30% Identified model: Mechanistic model:

  24. y r u e Plant + _ IMC Design for UnstableProcess Bock diagram for Internal Model Control system IMC for unstable systems: Model: IMC controller:

  25. Experiment IMC controller based on identifiedmodel z=30%

  26. PIDF and PI Tuning basedon IMC IMC controller can be implemented as a PIDF controller ---- IMC/PIDF ---- PI PI tuning from asymptotes of IMC controller

  27. PIDF versus PI control. Experiment (z=30%) (IMC) = PIDF controller PI controller

  28. Experimentsonmedium-scale S-riser Open-loop unstable: IMC controller (PIDF):

  29. Experiment Experimentsonmedium-scale S-riser PID-F controller: PI controller:

  30. Dealing with nonlinearity • Gain-scheduling • Adaptive controller gain slope = gain

  31. Solution 1: Gain-Scheduled PIDF Three identified model from step tests: Z=20%: Z=30%: Z=40%: Three IMC (PIDF) controllers:

  32. Experiment Gain-Scheduled PIDFExperiment

  33. Solution 2: Adaptive PI Tuning Static gain: slope = gain Linear valve: PI Tuning:

  34. Experiment Adaptive PI Tuning. Experiment

  35. “Direct” nonlinear approaches Solution 3: High-gain observer + state feedback: Did NOT work with bottom pressure (CDC, Dec. 2013) Solution 4: Output linearizing controller + P-control: Worked well, but gain-scheduled IMC more robust with respect to time delay (CDC, Dec. 2013)

  36. Solution 4: Output-linearizing controllerStabilizing controller for riser subsystem System in normal form: : top pressure : riser-base pressure Linearizingcontroller: dynamics bounded Control signal to valve:

  37. Experiment Solution 4: Output-linearizing controllerZ=60% Gain:

  38. OLGA Simulations Solution 2: Adaptive PI TuningOLGA Simulations

  39. Conclusions • A 4-state mechanisticmodelverified by experiments • Identify unstable slug-model from closed-loop step test • Good agreement betweenidentified and mechanisticmodels • IMC design works well and gives PIDF controller • Nonlinear “fixes” (adaptive gain or gain scheduling) work well Acknowledgement: • SIEMENS: Fundingoftheproject • Master students: Anette Helgesen, Knut Åge Meland, Mats Lieungh, Henrik Hansen, Terese Syre, MahnazEsmaeilpour and Anne Sofie Nilsen.

More Related