Download presentation
Presentation is loading. Please wait.
Published byCecily Morris Modified over 5 years ago
1
Combination of Measurements as Controlled Variables for Self-optimizing Control
Vidar Alstad† and Sigurd Skogestad Department of Chemical Engineering, Norwegian University of Science and Technology, Trondheim, Norway Presented at ESCAPE 13, Lappeenranta, Finland, June †
2
Outline Introduction optimal operation
Implementation of optimal operation Strategies Self-optimizing control Introduction Illustrating example Selection of controlled variables Optimal linear combination of measurements Examples Toy example Gas allocation in oil production Escape 13 - Lapperanta - June
3
Introduction and Motivation
Optimal operation for a given disturbance d Generally two classes of problems Constrained: All DOF (u’s) optimally constrained → Implementation easy by active constraint control Unconstrained: Some DOF (u’s) unconstrained (Focus here) Escape 13 - Lapperanta - June
4
Implementation Real-time optimization
Requires detailed on-line model Self-optimizing control (feedback control) easy implementation Escape 13 - Lapperanta - June
5
Self-optimizing Control
Define loss: Self-optimizing Control Self-optimizing control is when acceptable loss can be achieved using constant set points (cs) for the controlled variables c (without re-optimizing when disturbances occur). Escape 13 - Lapperanta - June
6
Self-optimizing Control – Illustrating Example
Optimal operation of Marathon runner, J=T Any self-optimizing variable c (to control at constant setpoint)? Escape 13 - Lapperanta - June
7
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 Escape 13 - Lapperanta - June
8
Controlled variables Controlled variables c to be selected among all available measurements y, Goal: Find the optimal linear combination (matrix H): Escape 13 - Lapperanta - June
9
Candidate Controlled Variables: Guidelines
Requirements for good candidate controlled variables (Skogestad & Postlethwaite, 1996) Its optimal value copt(d) is insensitive to disturbances It should be easy to measure and control accurately The variables c should be sensitive to change in inputs The selected variables should be independent Escape 13 - Lapperanta - June
10
Optimal Linear Combination -
Linearized where “sensitivity” Want optimal value of c insensitive to disturbances: To achieve Always possible if: Escape 13 - Lapperanta - June
11
Example – Toy example Consider the scalar unconstrained problem
The following measurements are available Controlling y1 gives perfect self-optimizing control. Is there a combination of y2 and y3 with the same properties? (Yes, should be because we have Escape 13 - Lapperanta - June
12
Example – Toy example (cont.)
Select y2 and y3: Gives the optimal controlled variable: Loss Escape 13 - Lapperanta - June
13
Example – Gas Lift Allocation - Introduction
Wells produce gas and oil from sub-sea reservoirs Gas injection: used to increase production Additional cost of compressing gas Limited gas processing capacity top-side Limits the rate of gas from the reservoirs and injection Case studied 2 production wells Gas injection into each well 1 transportation line Escape 13 - Lapperanta - June
14
Example – Gas Lift Allocation (cont.)
Escape 13 - Lapperanta - June
15
Example – Gas Lift Allocation (cont.)
Objective Maximize profit Constraints Maximum gas processing capacity Valve opening Escape 13 - Lapperanta - June
16
Example – Gas Lift Allocation (cont.)
Escape 13 - Lapperanta - June
17
Example – Gas Lift Allocation (cont.)
Evaluation of loss for different control structures The loss for cLC with the combined uncertainty is due to non-linearities Escape 13 - Lapperanta - June
18
Choice of measurements y
Escape 13 - Lapperanta - June
19
Conlusion Controlled variables: Derived simple method for optimal measurement combination Find sensitivity of optimal value of measurements to disturbances Select the controlled variables as: Illustrated on two examples Toy example Gas injection in oil production Escape 13 - Lapperanta - June
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.