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 1 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control 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: 1. 2. Gain scheduling Adaptive tuning • Experiments • OLGA Simulations 2 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control 3 Slug cycle (stable limit cycle) Experiments performed by the Multiphase Laboratory, NTNU 3 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control 4 Experimental mini-loop (2003) Ingvald Bårdsen, Espen Storkaas, Heidi Sivertsen 4 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control 5 z p2 Experimental mini-loop Valve opening (z) = 100% p1 SLUGGING 5 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control 6 z p2 Experimental mini-loop Valve opening (z) = 25% p1 SLUGGING 6 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control 7 z p2 Experimental mini-loop Valve opening (z) = 15% p1 NO SLUG 7 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control 8 z Experimental mini-loop: Bifurcation diagram p2 p1 No slug Valve opening z % Slugging 8 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control 9 How to avoid slugging? 9 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control 10 z Design change p2 Avoid slugging: 1. Close valve (but increases pressure) No slugging when valve is closed p1 Valve opening z % 10 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control 11 Design change Avoid slugging: 2. Design change to avoid slugging z p2 p1 11 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control 12 Design change Minimize effect of slugging: 3. Build large slug-catcher z p2 p1 • Most common strategy in practice 12 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control 13 Avoid slugging: 4. ”Active” feedback control PC ref z PT p 1 Simple PI-controller 13 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control 14 Anti slug control: Mini-loop experiments p1 [bar] z [%] Controller ON Controller OFF 14 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control 15 Anti slug control: Full-scale offshore experiments at Hod-Vallhall field (Havre,1999) 15 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control 17 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 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control 18 2009-2013: Esmaeil Jahanshahi, PhD-work supported by Siemens 1st step: New Experimental mini-rig P2 Top-side Air to atm. Valve Seperator 3m Riser P1 safety valve P3 FT water Buffer Tank FT air Mixing Point Pipeline P4 Subsea Valve Water Reservoir Pump Water Recycle 18 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control 19 2nd step: New Simplified 4-state model* Choke valve with opening Z mG1 : mass of gas in the pipeline mL1 : mass of liquid in the pipeline mG 2 : mass of gas in the riser mL 2 : mass of liquid in the riser x3, P2,VG2, ρG2 , HLT P0 wmix,out L3 x1, P1,VG1, ρG1, HL1 x4 wL,in wG,in w L2 h>hc wG,lp=0 wL,lp x2 h L1 θ hc State equations (mass conservations law): *Based on Storkaas model mG1 mL1 mG 2 mL 2 wG ,in wG ,lp wL ,in wL ,lp wG ,lp wG , out wL ,lp wL ,out 19 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control Experiment 20 New 4-state model. Comparison with experiments: Top pressure Subsea pressure Nonlinear: Process gain = slope - approaches zero for large z 20 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control 21 3rd step: Experimental linear model (new approach) Fourth-order mechanistic model: 7 parameters need to be estimated Hankel Singular Values: Model reduction: 4 parameters need to be estimated Stable part has little dynamic effect 21 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control 22 Model Identification: Closed-loop step response using P-controller Experiment 1: Z=20% (valve opening) 22 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control 23 Comparison with mechanistic model. Z=20% 1 10 0 10 Mag [-] Identified model: -1 10 -2 10 Mechanistic model Identified model -3 10 -2 Phase [deg] 10 100 Mechanistic model: -1 0 10 1 10 10 0 -100 Mechanistic model Identified model -200 -2 10 -1 0 10 10 1 10 [Rad/s] 29 Pin [kpa ] 28 Excellent agreement! 27 Experimental data Identified model Mechanistic model 26 25 0 20 40 60 80 100 120 140 t [sec] 23 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control 24 Comparison with mechanistic model. Z=30% 0 10 -1 10 Mag [-] Identified model: -2 10 -3 10 -4 Phase [deg] 10 -2 10 100 Mechanistic model: -1 0 10 1 10 10 0 -100 Mechanistic model Identified model -200 -2 10 -1 0 10 10 1 10 [Rad/s] 27 Pin [kpa ] Mechanistic model Identified model 26 Experimental data Identified model Mechanistic model 25 24 0 20 40 60 80 100 120 140 t [sec] 24 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control 25 IMC Design for Unstable Process Bock diagram for Internal Model Control system IMC for unstable systems: r e + _ C (s) u Plant Model: y IMC controller: 25 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control Experiment 26 IMC controller based on identified model z=30% inlet pressure (controlled variable) 40 open-loop stable Pin [kpa ] 35 30 25 20 15 open-loop unstable 0 2 4 6 8 10 12 14 16 18 20 t [min] actual valve position (manipulated variable) 80 open-loop unstable Zm [%] 60 Controller On Controller Off Controller Off 40 20 0 open-loop stable 0 2 4 6 8 10 12 14 16 18 20 t [min] 26 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control 27 PIDF and PI Tuning based on IMC IMC controller can be implemented as a PIDF controller Bode Diagram 80 ---- IMC/PIDF ---- PI Magnitude (dB) 60 40 20 Phase (deg) 0 225 PI tuning from asymptotes of IMC controller 180 135 90 -4 10 -3 10 -2 10 -1 10 0 10 1 10 Frequency (rad/s) 27 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control 28 PIDF versus PI control. Experiment (z=30%) (IMC) = PIDF controller PI controller inlet pressure (controlled variable) inlet pressure (controlled variable) 40 open-loop stable 30 25 20 15 open-loop unstable 0 2 4 6 8 10 12 14 open-loop stable 35 Pin [kpa ] 35 Pin [kpa ] 40 30 25 20 16 18 15 20 t [min] open-loop unstable 0 2 Controller On Controller Off Controller Off 40 20 0 8 10 12 14 16 18 20 open-loop unstable 60 Zm [%] Zm [%] 80 open-loop unstable 60 6 t [min] actual valve position (manipulated variable) actual valve position (manipulated variable) 80 4 Controller On Controller Off Controller Off 40 20 open-loop stable 0 2 4 6 8 10 t [min] 12 14 16 18 20 0 open-loop stable 0 2 4 6 8 10 12 14 16 18 20 t [min] 28 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control 29 Experiments on medium-scale S-riser Open-loop unstable: IMC controller (PIDF): 29 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control Experiment 30 Experiments on medium-scale S-riser PID-F controller: PI controller: 30 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control 31 Dealing with nonlinearity 1. Gain-scheduling 2. Adaptive controller gain 50 min & max steady-state Pin [kpa ] 40 30 slope = gain 20 10 0 10 20 30 40 50 60 70 80 90 100 Z [%] 1 30 Prt [kpa ] 25 min & max steady-state 20 15 10 5 31 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control 32 Solution 1: Gain-Scheduled PIDF Three identified model from step tests: Z=20%: Z=30%: Z=40%: Three IMC (PIDF) controllers: 32 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control Experiment 33 Gain-Scheduled PIDF Experiment inlet pressure (controlled variable) 40 open-loop stable Pin [kPa] 35 30 25 open-loop unstable 20 15 0 5 10 15 20 t [min] topside pressure open-loop stable Prt [kPa] 15 10 open-loop unstable 5 0 0 5 10 15 20 t [min] actual valve position (manipulated variable) 100 Zm [%] open-loop unstable 50 0 open-loop stable 0 5 10 15 20 t [min] 33 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control 34 Solution 2: Adaptive PI Tuning 50 min & max steady-state Static gain: Pin [kpa ] 40 slope = gain 30 20 10 0 10 20 30 40 50 60 70 80 90 100 Z [%] 1 30 Linear valve: Prt [kpa ] PI Tuning: min & max 25 steady-state 20 15 10 5 0 0 10 20 30 40 50 60 70 80 90 100 Z [%] 1 34 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control Experiment 35 Adaptive PI Tuning. Experiment inlet pressure (controlled variable) 40 open-loop stable 30 25 open-loop unstable 20 15 proportional gain open-loop stable 0 0 5 10 15 20 25 30 t [min] Kc [-] Pin [kpa ] 35 -50 actual valve position (manipulated variable) 80 open-loop unstable -100 0 5 10 20 0 15 20 25 30 20 25 30 t [min] integral time 40 600 open-loop stable 0 5 10 15 t [min] 20 25 30 I [sec] Zm [%] 60 400 200 0 5 10 15 t [min] 35 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control 36 “Direct” nonlinear approaches Solution 3: High-gain observer + state feedback: Did NOT work with bottom pressure (CDC, Dec. 2013) Solution 4: Output linearizing controller + Pcontrol: Worked well, but gain-scheduled IMC more robust with respect to time delay (CDC, Dec. 2013) 36 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control 37 Solution 4: Output-linearizing controller Stabilizing controller for riser subsystem System in normal form: : riser-base pressure : top pressure Linearizing controller: dynamics bounded Control signal to valve: 37 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control Experiment 38 Solution 4: Output-linearizing controller Z=60% riser-base pressure (controlled variable) Prb [kPa] 40 open-loop stable 30 open-loop unstable set-point measurement 20 10 0 5 10 15 20 t [min] Prt [kPa] topside pressure (measurement used by controller) open-loop stable 15 10 open-loop unstable 5 0 0 5 10 15 Gain: 20 t [min] actual valve position (manipulated variable) 50 min & max 100 Controller On Controller Off 50 0 open-loop unstable open-loop stable 0 5 10 15 Pin [kpa ] Zm [%] Controller Off steady-state 40 30 20 20 10 t [min] 0 10 20 30 40 50 60 70 80 90 100 Z [%] 1 30 38 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control min & max steady-state 25 OLGA Simulations 39 Solution 2: Adaptive PI Tuning OLGA Simulations bottom pressure, P 1 68 measurement set-point 1 P [bara] 67.5 67 66.5 0 1 2 3 time [h] 4 5 6 valve opening, Z 0.5 Set-point Z [%] 0.4 0.3 0.2 X: 5.289 Y: 0.2324 0.1 0 0 1 2 3 time [h] 4 5 67.36 67.19 67.07 66.99 66.93 66.88 Valve opening 14 16.1 18.2 20.1 23.24 -- Kc Ti 0.5 0.7 0.94 1.23 1.56 1.93 8400 9600 10800 12000 13200 14400 6 39 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control 40 Conclusions • • • • • A 4-state mechanistic model verified by experiments Identify unstable slug-model from closed-loop step test Good agreement between identified and mechanistic models IMC design works well and gives PIDF controller Nonlinear “fixes” (adaptive gain or gain scheduling) work well Acknowledgement: • SIEMENS: Funding of the project • Master students: Anette Helgesen, Knut Åge Meland, Mats Lieungh, Henrik Hansen, Terese Syre, Mahnaz Esmaeilpour and Anne Sofie Nilsen. 40 Sigurd Skogestad | Closed-loop model identification and PID/PI tuning for robust anti-slug control
© Copyright 2026 Paperzz