1 control structure design: what should we measure, control and manipulate? sigurd skogestad...
TRANSCRIPT
1
Control structure design: What should we measure, control and manipulate?
Sigurd Skogestad
Department of Chemical Engineering
NTNU, Trondheim
First African Control Conference, Cape Town, 04 December 2003
2
Outline
• About myself
• Control structure design
• A procedure for control structure design
• Selection of primary controlled variables
• Example stabilizing control: Anti slug control
• Conclusion
3
Sigurd Skogestad
• Born in 1955• 1956-1961: Lived in South Africa (Durban & Johannesburg)• 1978: Siv.ing. Degree (MS) in Chemical Engineering from NTNU (NTH)• 1980-83: Process modeling group at the Norsk Hydro Research Center in
Porsgrunn• 1983-87: Ph.D. student in Chemical Engineering at Caltech, Pasadena, USA.
Thesis on “Robust distillation control”. Supervisor: Manfred Morari• 1987 - : Professor in Chemical Engineering at NTNU• Since 1994: Head of process systems engineering center in Trondheim
(PROST)• Since 1999: Head of Department of Chemical Engineering• 1996: Book “Multivariable feedback control” (Wiley)• 2000,2003: Book “Prosessteknikk” (Tapir)• Group of about 10 Ph.D. students in the process control area
4
Research: Develop simple yet rigorous methods to solve problems of
engineering significance.
• Use of feedback as a tool to1. reduce uncertainty (including robust control),
2. change the system dynamics (including stabilization; anti-slug control),
3. generally make the system more well-behaved (including self-optimizing control).
• limitations on performance in linear systems (“controllability”),
• control structure design and plantwide control,
• interactions between process design and control,
• distillation column design, control and dynamics.
• Natural gas processes
5
Outline
• About myself
• Control structure design
• A procedure for control structure design
• Selection of primary controlled variables
• Example stabilizing control: Anti slug control
• Conclusion
6
Idealized view of control(“Ph.D. control”)
7
Practice: Tennessee Eastman challenge problem (Downs, 1991)
(“PID control”)
8
Control structure design
• Not the tuning and behavior of each control loop,
• But rather the control philosophy of the overall plant with emphasis on the structural decisions:– Selection of controlled variables (“outputs”)
– Selection of manipulated variables (“inputs”)
– Selection of (extra) measurements
– Selection of control configuration (structure of overall controller that interconnects the controlled, manipulated and measured variables)
– Selection of controller type (LQG, H-infinity, PID, decoupler, MPC etc.).
• That is: Control structure design includes all the decisions we need make to get from ``PID control’’ to “Ph.D” control
9
Process control:Control structure design = plantwide control
• Large systems
• Each plant usually different – modeling expensive
• Slow processes – no problem with computation time
• Structural issues important– What to control?
– Extra measurements
– Pairing of loops
10
• Control structure selection issues are identified as important also in other industries.
Professor Gary Balas at ECC’03 about flight control at Boeing:
The most important control issue has always been to select the right controlled variables --- no systematic tools used!
11
Process operation: Hierarchical structure
Need to define objectives and identify main issues for each
layer
PID
RTO
MPC
12
Regulatory control (seconds)
• Purpose: “Stabilize” the plant by controlling selected ‘’secondary’’ variables (y2) such that the plant does not drift too far away from its desired operation
• Use simple single-loop PI(D) controllers
• Status: Many loops poorly tuned– Most common setting: Kc=1, I=1 min (default)
– Even wrong sign of gain Kc ….
13
Regulatory control……...
• Trend: Can do better! Carefully go through plant and retune important loops using standardized tuning procedure
• Exists many tuning rules, including Skogestad (SIMC) rules: – Kc = 0.5/k (1/) I = min (1, 8)
– “Probably the best simple PID tuning rules in the world”
• Outstanding structural issue: What loops to close, that is, which variables (y2) to control?
14
Supervisory control (minutes)
• Purpose: Keep primary controlled variables (y1) at desired values, using as degrees of freedom the setpoints y2s for the regulatory layer.
• Status: Many different “advanced” controllers, including feedforward, decouplers, overrides, cascades, selectors, Smith Predictors, etc.
• Issues: – Which variables to control may change due to change of “active
constraints”
– Interactions and “pairing”
15
Supervisory control…...
• Trend: Model predictive control (MPC) used as unifying tool.
– Linear multivariable models with input constraints
– Tuning (modelling) is time-consuming and expensive
• Issue: When use MPC and when use simpler single-loop decentralized controllers ?– MPC is preferred if active constraints (“bottleneck”) change.
– Avoids logic for reconfiguration of loops
• Outstanding structural issue:– What primary variables y1 to control?
16
Local optimization (hour)
• Purpose: Identify active constraints and possibly recompute optimal setpoints y1s for controlled variables
• Status: Done manually by clever operators and engineers
• Trend: Real-time optimization (RTO) based on detailed nonlinear steady-state model
• Issues: – Optimization not reliable.
– Modelling is time-consuming and expensive
17
Outline
• About myself
• Control structure design
• A procedure for control structure design
• Selection of primary controlled variables
• Example stabilizing control: Anti slug control
• Conclusion
18
Stepwise procedure plantwide control
I. TOP-DOWN
Step 1. DEFN. OF OPERATIONAL OBJECTIVES
Step 2. MANIPULATED VARIABLES and DEGREE OF FREEDOM ANALYSIS
Step 3. WHAT TO CONTROL? (primary outputs, c= y1)
Step 4. PRODUCTION RATE
19
II. BOTTOM-UP (structure control system):
Step 5. REGULATORY CONTROL LAYER
Stabilization and Local disturbance rejection What more to control? (secondary outputs y2)
Step 6. SUPERVISORY CONTROL LAYER
Decentralized control or MPC?
Step 7. OPTIMIZATION LAYER (RTO)
20
Outline
• About myself
• Control structure design
• A procedure for control structure design
• Selection of primary controlled variables
• Example stabilizing control: Anti slug control
• Conclusion
21
What should we control?
y1 = c ? (economics)
y2 = ? (“stabilization”)
22
Optimal operation (economics)
• Define scalar cost function J(u0,d)
– u0: degrees of freedom
– d: disturbances
• Optimal operation for given d:
minu0 J(u0,d)subject to:
f(u0,d) = 0
g(u0,d) < 0
23
Active constraints
• Optimal solution is usually at constraints, that is, most of the degrees of freedom are used to satisfy “active constraints”, g(u0,d) = 0
• Implementation of active constraints is usually simple.
• We here concentrate on the remaining unconstrained degrees of freedom u.
24
Optimal operation
Cost J
Independent variable u(remaining unconstrained)
uuoptopt
JJoptopt
25
Implementation: How do we deal with uncertainty?
• 1. Disturbances d
• 2. Implementation error n
us = uopt(d*) – nominal optimization
nu = us + n
d
Cost J Jopt(d)
26
Problem no. 1: Disturbance d
Cost J
Independent variable uuuoptopt(d(d00))
JJoptopt
d*d*
dd
Loss with constant value for u
27
Problem no. 2: Implementation error n
Cost J
Independent variable u
uuss=u=uoptopt(d(d00))
JJoptopt
dd00 Loss due to implementation error for u
u = us + n
28
Probem: Too complicated
”Obvious” solution: Optimizing control
29
Alternative: Look for another variable c to control
Cost J
Controlled variable cccoptopt
JJoptopt
and keep at constant setpoint cs = copt(d*)
30
Self-optimizing Control
• Self-optimizing Control– Self-optimizing control is when acceptable
loss can be achieved using constant set points (c
s)
for the controlled variables c (without re-
optimizing when disturbances occur).
• Define loss:
31
Controlled variable: Feedback implementation
Issue:What should we control?
32
Constant setpoint policy:Effect of disturbances (“problem 1”)
33
Effect of implementation error (“problem 2”)
BADGoodGood
34
Self-optimizing Control – Illustrating Example
• Optimal operation of Marathon runner, J=T– Any self-optimizing variable c (to control at constant
setpoint)?
35
Self-optimizing Control – Illustrating Example
• Optimal operation of Marathon runner, J=T– Any self-optimizing variable c (to control at constant
setpoint)?• c1 = distance to leader of race
• c2 = speed
• c3 = heart rate
• c4 = level of lactate in muscles
36
Further examples
• Central bank. J = welfare. c=inflation rate
• Cake baking. J = nice taste, c = Temperature
• Biology. J = ?, c = regulatory mechanism
• Business, J = profit. c = KPI = response time to order
37
Good controlled Variables c: Guidelines
• Requirements for good candidate controlled variables (Skogestad & Postlethwaite, 1996):– Its optimal value copt(d) is insensitive to disturbances d (to
avoid problem 1)– It should be easy to measure and control accurately (n small
to avoid problem 1)– The variables c should be sensitive to change in inputs (to
avoid problem 2)– The selected variables c should be independent (to avoid
problem 2)
• Rule: Maximize minimum singular value of scaled G • c = G u
38
Candidate controlled variables
• Selected among available measurements y,
• More generally: Find the optimal linear combination (matrix H):
39
Candidate Controlled Variables: Guidelines
• Recall first requirement:
Its optimal value copt(d) is insensitive to disturbances (to avoid problem 1)
• Can we always find a variable combination c = H y
which satisfies
• YES!! Provided
40
Derivation of optimal combination (Alstad)
• Starting point: Find optimal operation as a function of d:– uopt(d), yopt(d)
• Linearize this relationship: yopt = F d
• Look for a linear combination c = Hy which satisfies: copt = 0
• To achieve
• Always possible if
41
Applications of self-optimizing control
• Distillation
• Tennessee Eastman Challenge problem
• Power plant
• +++
42
Outline
• About myself
• Control structure design
• A procedure for control structure design
• Selection of primary controlled variables
• Example stabilizing control: Anti-slug control
• Conclusion
43
Application: Anti-slug control
Slug (liquid) buildup
Two-phase pipe flow(liquid and vapor)
44
Slug cycle (stable limit cycle) Experiments performed by the Multiphase Laboratory, NTNU
45
Experimental mini-loop
46
p1
p2
z
Experimental mini-loopValve opening (z) = 100%
47
p1
p2
z
Experimental mini-loopValve opening (z) = 25%
48
p1
p2
z
Experimental mini-loopValve opening (z) = 15%
49
p1
p2
z
Experimental mini-loop:Bifurcation diagram
Valve opening z %
No slug
Slugging
50
p1
p2
z
Avoid slugging:1. Close valve (but increases pressure)
Valve opening z %
No slugging when valve is closed
51
Avoid slugging:2. Build large slug-catcher
• Most common strategy in practice
p1
p2
z
52
Avoid slugging:3. Other design changes to avoid slugging
p1
p2
z
53
Avoid slugging: 4. Control?
Valve opening z %
Predicted smooth flow: Desirable but open-loop unstable
Comparison with simple 3-state model:
54
Avoid slugging:4. ”Active” feedback control
PT
PCref
Simple PI-controller
p1
z
55
Anti slug control: Mini-loop experiments
Controller ON Controller OFF
p1 [bar]
z [%]
56
Anti slug control: Full-scale offshore experiments at Hod-Vallhall field (Havre,1999)
57
Poles and zeros
y
zP1 [Bar] DP[Bar] ρT [kg/m3] FQ [m3/s] FW [kg/s]
0.175-0.0034 3.2473
0.0142
-0.0004
0.0048
-4.5722
-0.0032
-0.0004
-7.6315
-0.0004
0
0.25-0.0034 3.4828
0.0131
-0.0004
0.0048
-4.6276
-0.0032
-0.0004
-7.7528
-0.0004
0
Operation points:
Zeros:
zP1 DP Poles
0.175 70.05 1.94-6.11
0.0008±0.0067i
0.25 69 0.96-6.21
0.0027±0.0092i
P1
ρT
DP
FTTopside
Topside measurements: Ooops.... RHP-zeros or zeros close to origin
58
Stabilization with topside measurements:Avoid “RHP-zeros by using 2 measurements
• Model based control (LQG) with 2 top measurements: DP and density ρT
59
Summary anti slug control
• Stabilization of smooth flow regime = $$$$! (or Rand!)
• Stabilization using downhole pressure simple
• Stabilization using topside measurements possible
• Control can make a difference!
Thanks to: Espen Storkaas + Heidi Sivertsen and Ingvald Bårdsen
60
Conclusion
What would I do if I was manager in a large chemical company?
• Define objective of control: Better operation (objective is not just to collect data)
• Define control as a function in my organization
• Define operational objectives for each plant ($ + other..)
• Unified approach (”company policy”):– Stabilizing control. PID rules
– MPC
– Avoid rule-based systems / Artificial intelligens /operator support systems - people are better at this
• Set high standards for acceptable control
My view
More information: Home page of Sigurd Skogestad - http://www.nt.ntnu.no/users/skoge/