1. Stabilizing control and controllability:
Control solutions to avoid slug flow in pipeline-riser systems
Espen Storkaas
Trondheim

2. Thesis summary Introduction
Controllability analysis of a two-phase pipeline-riser systems at riser slugging conditions
A low-dimensional dynamic model of severe slugging for control design and analysis
Implications of input rate limitations on controllability and controller design
Stabilization of multiphase flow in pipelines with single-loop and cascade controllers
Model-based anti-slug controllers
Extended slug control – An industrial application
Conclusions and further work

3. Outline Introduction Slug flow in pipeline-riser systems
Modelling of pipeline-riser systems for control applications
Controllability analysis
Effect of input rate limitations
Controller design
Extended slug control
Conclusions

4. Introduction Oil producing wells also produce gas and water
Longer multiphase tie-in lines in offshore oil production from increases flow-related challenges
Flow assurance technology plays an increasingly important role
Hydrates
Wax
Corrosion
Flow regimes

5. Slug flow in pipeline-riser systems
Riser slugging
Hydrodynamic slugging
Terrain slugging
Transient slugging
*Pictures from SINTEF Multiphase flow laboratory

6. Riser slugging and control - History
From design challenge to control objective
First relevant publication : Schmidt et al (1979)
Experimental work by Hedne & Linga (1990)
Simulations studies and experimental work from several sources (Total, Shell, ABB, Statoil)
First industrial application: Hod-Valhall (Havre et al 2000), more has followed in later years
Included in design of new projects (riser slugging potensial at design conditions)

7. Outline Modelling of pipeline-riser systems for control applications
Introduction
Slug flow in pipeline-riser systems
Modelling of pipeline-riser systems for control applications
Controllability analysis
Effect of input rate limitations
Controller design
Extended slug control
Conclusions

8. Main case study Test case for riser slugging in OLGA
Simplified geometry
Two-phase flow
Constant feed
Constant pressure behind choke

9. Bifuracation diagram for riser slugging

10. Modelling (1) – Lessons learned from two-fluid model
Two-fluid model used to investigate system caracteristics
Transition to instability through Hopf bifurbation
Complex unstable poles
Controllability analysis gives information about measurement selection
Simpler model should be used

11. Modelling (2)- Design specs for simplified model
The model must:
Describe the dominant dynamic behavior of the system for the time scales for which control is to be effective
The model should :
be continuous
be simple (low state dimension)
contain few empirical coefficients

12. Modelling (3) – Simplified 3-state model
Three dynamical states
From entrainment model
Given by valve equation:

13. Modelling (4) – Entrainment model

14. Modelling (5) –Properties of 3-state model
Phenomenological model with 3 dynamical states
Based on bulk properties
Describes both riser slugging and unstable stationary operating points
4 empirical parameters – easy to tune
Hopf bifurcation, complex unstable poles
Very useful for controllability analysis and controller design

15. Modelling (6) – Model comparison

16. Outline Controllability analysis Introduction
Slug flow in pipeline-riser systems
Modelling of pipeline-riser systems for control applications
Controllability analysis
Effect of input rate limitations
Controller design
Extended slug control
Conclusions

17. Controllability analysis
Investigation into a plants achievable control performance
Independent of controller
Inverse response
Step in valve opening:

18. Measurement evaluation
Achievable performace can be represented by lower bounds on closed-loop transfer functions such as
Sensitivity function S
Complementary sensitivity function T
Input usage KS
Bound computed from 3-state model
Small numerical value for lower bounds on closed loop transfer functions indicate a good measurement candidate

19. Measurement evaluation
Achievable performace can be represented by lower bounds on closed-loop transfer function
For example:
Small numerical value for lower bounds on closed loop transfer functions indicate a good measurement candidate
Chen (2000)
Glower (1986)

20. Measurement evaluation
Unstable system at 30% valve opening pi=0.0007±0.0073 y
RHPZ
MS=MT PI - 1
0.11 DP
0.016
1.9
0.25 Q
0.09
Low steady-state gain
Similar results from two-fluid model

21. Conclusions from controllability analysis
Inlet or riserbase pressure well suited for stabilizing control
Time delay may prevent the use of inlet pressure for long pipelines
Pressure at top of riser not suitable for stabilizing control due to unstable zero dynamics
Flow measurement at riser outlet can be used for stabilization but has lacking low-frequency gain
Best used as a secondary measurement in a cascade or in combination with another measurement

22. Outline Effect of input rate limitations Controller design
Introduction
Slug flow in pipeline-riser systems
Modelling of pipeline-riser systems for control applications
Controllability analysis
Effect of input rate limitations
Controller design
Extended slug control
Conclusions

23. Effect of input rate limitations
Limitation on input rate can limit performance for control systems
Explicit lower bounds on required input rate derived
Stabilization
Disturbance rejection
Controller design with limited input rates

24. Controller design Controllers design based on simplified 3-state model
Stabilizes both two-fluid model and OLGA model
Measurement selection from controllability analysis confirmed
Controllers based on upstream pressure measurement robust and effective
Controllers based on only a flow measurement tends to drift off
A flow measurement combined with another measurement can be used for stabilizing control

25. Single-loop controllers
PID controller with measured outlet flow
PID controller with measured riser base pressure
H∞controller with measured inlet pressure
PID controller with measured inlet pressure

26. Cascade and MISO controllers
Cascade controller, y1=DP, y2=Q
H∞-controller, y=[DP Q]
Cascade controller, y1=PI, y2=Q

27. Extended slug control
Anti-slug control combined with functionality to mitigate surge waves and startup slugs

28. Summary
Simplified model of pipeline-riser systems at riser slugging condisons for controllability analysis and controller design
Controllability analysis gives clear recommendations for measurement selection for stabilizing control
Input rate limitations may be important
Controller design
Extended control application
Further work
Effect of water, different geometries
New measurements