a scalable semidefinite relaxation approach to grid scheduling · • semidefinite programming...
TRANSCRIPT
![Page 1: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/1.jpg)
Ramtin Madani, UT Arlington
A Scalable Semidefinite Relaxation Approach to Grid Scheduling
Ramtin Madani
Department of Electrical EngineeringThe University of Texas at Arlington
Ali Davoudi
Department of Electrical EngineeringThe University of Texas at Arlington
Alper Atamtürk
Industrial Engineering and Operations ResearchUniversity of California, Berkeley
![Page 2: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/2.jpg)
Ramtin Madani, UT Arlington 2
Generators Loads
PowerSystem
Determine:
1. On/off status of generators,2. Power injections,3. Voltages.
Day-ahead Scheduling
![Page 3: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/3.jpg)
Ramtin Madani, UT Arlington 2
Generators Loads
PowerSystem
Determine:
1. On/off status of generators,2. Power injections,3. Voltages.
Subject to:
1. Unit constraints,2. Network constraints.
Day-ahead Scheduling
![Page 4: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/4.jpg)
Ramtin Madani, UT Arlington 2
Generators Loads
PowerSystem
Determine:
1. On/off status of generators,2. Power injections,3. Voltages.
Subject to:
1. Unit constraints,2. Network constraints.
Proposed Approach: Third-order Semidefinite Programming (TSDP).• In-between the SDP and SOCP relaxations.
Day-ahead Scheduling
![Page 5: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/5.jpg)
Ramtin Madani, UT Arlington 2
Generators Loads
PowerSystem
Determine:
1. On/off status of generators,2. Power injections,3. Voltages.
Subject to:
1. Unit constraints,2. Network constraints.
Proposed Approach: Third-order Semidefinite Programming (TSDP).• In-between the SDP and SOCP relaxations.
Distinctive Features:• Compatible with Nonlinearity: Joint Unit Commitment AC Optimal Power Flow,
Day-ahead Scheduling
![Page 6: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/6.jpg)
Ramtin Madani, UT Arlington 2
Generators Loads
PowerSystem
Determine:
1. On/off status of generators,2. Power injections,3. Voltages.
Subject to:
1. Unit constraints,2. Network constraints.
Proposed Approach: Third-order Semidefinite Programming (TSDP).• In-between the SDP and SOCP relaxations.
Distinctive Features:• Compatible with Nonlinearity: Joint Unit Commitment AC Optimal Power Flow,
• Massively Scalable: 24-hour problem with 4000 units in a 13000-bus network,
Day-ahead Scheduling
![Page 7: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/7.jpg)
Ramtin Madani, UT Arlington 2
Generators Loads
PowerSystem
Determine:
1. On/off status of generators,2. Power injections,3. Voltages.
Subject to:
1. Unit constraints,2. Network constraints.
Proposed Approach: Third-order Semidefinite Programming (TSDP).• In-between the SDP and SOCP relaxations.
Distinctive Features:• Compatible with Nonlinearity: Joint Unit Commitment AC Optimal Power Flow,
• Massively Scalable: 24-hour problem with 4000 units in a 13000-bus network,
• Small Gap: Less than 2% away from global optimality.
Day-ahead Scheduling
![Page 8: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/8.jpg)
Ramtin Madani, UT Arlington 3
Experimental Results24-hour problem based on IEEE and European grid data:
![Page 9: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/9.jpg)
Ramtin Madani, UT Arlington 3
Experimental Results
Feasible solutions with less than 2% distance from the globally optimality
24-hour problem based on IEEE and European grid data:
![Page 10: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/10.jpg)
Ramtin Madani, UT Arlington 3
Experimental Results
Feasible solutions with less than 2% distance from the globally optimality
24-hour problem based on IEEE and European grid data:
• Several relaxation schemes have been proposed for UC and OPF:
SDPRelaxation
SOCPRelaxation
![Page 11: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/11.jpg)
Ramtin Madani, UT Arlington 3
Experimental Results
Feasible solutions with less than 2% distance from the globally optimality
24-hour problem based on IEEE and European grid data:
• Several relaxation schemes have been proposed for UC and OPF:
SDPRelaxation
SOCPRelaxation
TSDP Relaxation
An ideal balance between strength and complexity.
![Page 12: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/12.jpg)
Ramtin Madani, UT Arlington
Problem Formulation
4
• Consider a problem with time periods and units:
![Page 13: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/13.jpg)
Ramtin Madani, UT Arlington
Problem Formulation
4
• Consider a problem with time periods and units:
• The on/off status of unit at time :
![Page 14: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/14.jpg)
Ramtin Madani, UT Arlington
Problem Formulation
4
• Consider a problem with time periods and units:
• The on/off status of unit at time :
• Active power injection of unit at time :
![Page 15: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/15.jpg)
Ramtin Madani, UT Arlington
Problem Formulation
4
• Consider a problem with time periods and units:
• The on/off status of unit at time :
• Active power injection of unit at time :
• Reactive power injection of unit at time :
![Page 16: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/16.jpg)
Ramtin Madani, UT Arlington
Problem Formulation
4
• Consider a problem with time periods and units:
• The on/off status of unit at time :
• Active power injection of unit at time :
• Reactive power injection of unit at time :
• Operating cost of unit at time :
![Page 17: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/17.jpg)
Ramtin Madani, UT Arlington
Problem Formulation
4
OverallCost
• Consider a problem with time periods and units:
• The on/off status of unit at time :
• Active power injection of unit at time :
• Reactive power injection of unit at time :
• Operating cost of unit at time :
![Page 18: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/18.jpg)
Ramtin Madani, UT Arlington
Problem Formulation
4
UnitConstraints
• Consider a problem with time periods and units:
• The on/off status of unit at time :
• Active power injection of unit at time :
• Reactive power injection of unit at time :
• Operating cost of unit at time :
![Page 19: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/19.jpg)
Ramtin Madani, UT Arlington
Problem Formulation
4
• Consider a problem with time periods and units:
• The on/off status of unit at time :
• Active power injection of unit at time :
• Reactive power injection of unit at time :
• Operating cost of unit at time :
• Unit feasible sets:
![Page 20: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/20.jpg)
Ramtin Madani, UT Arlington
Problem Formulation
4
NetworkConstraints
• Consider a problem with time periods and units:
• The on/off status of unit at time :
• Active power injection of unit at time :
• Reactive power injection of unit at time :
• Operating cost of unit at time :
• Unit feasible sets:
![Page 21: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/21.jpg)
Ramtin Madani, UT Arlington
Problem Formulation
4
• Consider a problem with time periods and units:
• The on/off status of unit at time :
• Active power injection of unit at time :
• Reactive power injection of unit at time :
• Operating cost of unit at time :
• Unit feasible sets:
• Network feasible sets:
![Page 22: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/22.jpg)
Ramtin Madani, UT Arlington
Problem Formulation
4
Discrete Parameters
• Consider a problem with time periods and units:
• The on/off status of unit at time :
• Active power injection of unit at time :
• Reactive power injection of unit at time :
• Operating cost of unit at time :
• Unit feasible sets:
• Network feasible sets:
![Page 23: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/23.jpg)
Ramtin Madani, UT Arlington
Problem Formulation
4
Nonlinear Equations
Discrete Parameters
• Consider a problem with time periods and units:
• The on/off status of unit at time :
• Active power injection of unit at time :
• Reactive power injection of unit at time :
• Operating cost of unit at time :
• Unit feasible sets:
• Network feasible sets:
![Page 24: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/24.jpg)
Ramtin Madani, UT Arlington
Unit Constraints
5
![Page 25: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/25.jpg)
Ramtin Madani, UT Arlington
Unit Constraints
5
![Page 26: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/26.jpg)
Ramtin Madani, UT Arlington
Unit Constraints
5
![Page 27: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/27.jpg)
Ramtin Madani, UT Arlington
Unit Constraints
5
![Page 28: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/28.jpg)
Ramtin Madani, UT Arlington
Unit Constraints
5
![Page 29: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/29.jpg)
Ramtin Madani, UT Arlington
Unit Constraints
5
![Page 30: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/30.jpg)
Ramtin Madani, UT Arlington
Unit Constraints
5
![Page 31: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/31.jpg)
Ramtin Madani, UT Arlington
Unit Constraints
5
![Page 32: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/32.jpg)
Ramtin Madani, UT Arlington
Unit Constraints
5
![Page 33: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/33.jpg)
Ramtin Madani, UT Arlington
Unit Constraints
5
![Page 34: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/34.jpg)
Ramtin Madani, UT Arlington
Unit Constraints
5
![Page 35: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/35.jpg)
Ramtin Madani, UT Arlington
Unit Constraints
5
![Page 36: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/36.jpg)
Ramtin Madani, UT Arlington
Unit Constraints
5
![Page 37: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/37.jpg)
Ramtin Madani, UT Arlington
Unit Constraints
5
![Page 38: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/38.jpg)
Ramtin Madani, UT Arlington
Network Constraints
6
![Page 39: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/39.jpg)
Ramtin Madani, UT Arlington
Network Constraints
6
![Page 40: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/40.jpg)
Ramtin Madani, UT Arlington
Network Constraints
6
![Page 41: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/41.jpg)
Ramtin Madani, UT Arlington
Network Constraints
6
![Page 42: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/42.jpg)
Ramtin Madani, UT Arlington
Network Constraints
6
![Page 43: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/43.jpg)
Ramtin Madani, UT Arlington
Network Constraints
6
![Page 44: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/44.jpg)
Ramtin Madani, UT Arlington
Network Constraints
6
![Page 45: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/45.jpg)
Ramtin Madani, UT Arlington
Network Constraints
6
![Page 46: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/46.jpg)
Ramtin Madani, UT Arlington
Network Constraints
6
![Page 47: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/47.jpg)
Ramtin Madani, UT Arlington
Network Constraints
6
![Page 48: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/48.jpg)
Ramtin Madani, UT Arlington
Network Constraints
6
![Page 49: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/49.jpg)
Ramtin Madani, UT Arlington
Network Constraints
6
![Page 50: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/50.jpg)
Ramtin Madani, UT Arlington
Network Constraints
6
![Page 51: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/51.jpg)
Ramtin Madani, UT Arlington
Network Constraints
6
![Page 52: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/52.jpg)
Ramtin Madani, UT Arlington
Convex Relaxation
7
• Consider a problem with time periods and units:
![Page 53: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/53.jpg)
Ramtin Madani, UT Arlington
Convex Relaxation
7
• Consider a problem with time periods and units:
![Page 54: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/54.jpg)
Ramtin Madani, UT Arlington
Convex Relaxation
7
• Consider a problem with time periods and units:
![Page 55: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/55.jpg)
Ramtin Madani, UT Arlington
Convex Relaxation
7
• Consider a problem with time periods and units:
• A convex relaxation can be created by replacing and with theirconvex surrogates.
![Page 56: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/56.jpg)
Ramtin Madani, UT Arlington
Convex Relaxation
7
• Consider a problem with time periods and units:
• A convex relaxation can be created by replacing and with theirconvex surrogates.
• Unit feasible sets:
![Page 57: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/57.jpg)
Ramtin Madani, UT Arlington
Convex Relaxation
7
• Consider a problem with time periods and units:
• A convex relaxation can be created by replacing and with theirconvex surrogates.
• Unit feasible sets:
• Network feasible sets:
![Page 58: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/58.jpg)
Ramtin Madani, UT Arlington
Convex Relaxation
7
The proposed relaxation
• Consider a problem with time periods and units:
• A convex relaxation can be created by replacing and with theirconvex surrogates.
• Unit feasible sets:
• Network feasible sets:
![Page 59: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/59.jpg)
Ramtin Madani, UT Arlington
Relaxation of Unit Constraints
8
• Non-convex constraints:
![Page 60: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/60.jpg)
Ramtin Madani, UT Arlington
Relaxation of Unit Constraints
8
• Non-convex constraints:
![Page 61: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/61.jpg)
Ramtin Madani, UT Arlington
Relaxation of Unit Constraints
8
• Non-convex constraints:
• Lifting: Define a number of auxiliary variables:
![Page 62: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/62.jpg)
Ramtin Madani, UT Arlington
Relaxation of Unit Constraints
8
• Non-convex constraints:
• Lifting: Define a number of auxiliary variables:
• The new variables and are defined to linearize the non-convex constraint.
![Page 63: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/63.jpg)
Ramtin Madani, UT Arlington
Relaxation of Unit Constraints
8
• Non-convex constraints:
• Lifting: Define a number of auxiliary variables:
• The new variables and are defined to linearize the non-convex constraint.
![Page 64: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/64.jpg)
Ramtin Madani, UT Arlington
Relaxation of Unit Constraints
8
• Non-convex constraints:
• Lifting: Define a number of auxiliary variables:
• The new variables and are defined to linearize the non-convex constraint.
• Whereas and are used to create a stronger relaxation.
![Page 65: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/65.jpg)
Ramtin Madani, UT Arlington
Relaxation of Unit Constraints
8
• Non-convex constraints:
• Lifting: Define a number of auxiliary variables:
• The new variables and are defined to linearize the non-convex constraint.
• Whereas and are used to create a stronger relaxation.
• The goal is to design a set of linear and conic inequalities that partially describe theconvex hull of all feasible variables:
![Page 66: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/66.jpg)
Ramtin Madani, UT Arlington
Relaxation of Unit Constraints
9
• Linearized version of the cost equations:
![Page 67: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/67.jpg)
Ramtin Madani, UT Arlington
Relaxation of Unit Constraints
9
• Linearized version of the cost equations:
• Binary requirement can be captured through the “McCormick constraints”:
![Page 68: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/68.jpg)
Ramtin Madani, UT Arlington
Relaxation of Unit Constraints
9
• Linearized version of the cost equations:
• Binary requirement can be captured through the “McCormick constraints”:
• 2 third-order Semidefinite (TSDP) inequalities:
![Page 69: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/69.jpg)
Ramtin Madani, UT Arlington
Relaxation of Unit Constraints
9
• Linearized version of the cost equations:
• Binary requirement can be captured through the “McCormick constraints”:
• 2 third-order Semidefinite (TSDP) inequalities:
• 24 linear inequalities:
![Page 70: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/70.jpg)
Ramtin Madani, UT Arlington
Relaxation of Unit Constraints
9
• Linearized version of the cost equations:
• Binary requirement can be captured through the “McCormick constraints”:
• 2 third-order Semidefinite (TSDP) inequalities:
• 24 linear inequalities:
![Page 71: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/71.jpg)
Ramtin Madani, UT Arlington
Relaxation of Unit Constraints
9
• Linearized version of the cost equations:
• Binary requirement can be captured through the “McCormick constraints”:
• 2 third-order Semidefinite (TSDP) inequalities:
• 24 linear inequalities:
![Page 72: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/72.jpg)
Ramtin Madani, UT Arlington
Relaxation of Unit Constraints
10
• 4 additional inequalities can be designed in terms of the minimum down time constraints:
![Page 73: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/73.jpg)
Ramtin Madani, UT Arlington
Relaxation of Unit Constraints
10
• 4 additional inequalities can be designed in terms of the minimum down time constraints:
• 4 additional inequalities can be designed in terms of the minimum up time constraints:
![Page 74: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/74.jpg)
Ramtin Madani, UT Arlington
Relaxation of Network Constraints
11
![Page 75: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/75.jpg)
Ramtin Madani, UT Arlington
Relaxation of Network Constraints
11
![Page 76: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/76.jpg)
Ramtin Madani, UT Arlington
Relaxation of Network Constraints
11
![Page 77: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/77.jpg)
Ramtin Madani, UT Arlington
Relaxation of Network Constraints
11
![Page 78: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/78.jpg)
Ramtin Madani, UT Arlington
Relaxation of Network Constraints
11
• The network constraints can be cast linearly w.r.t. .
![Page 79: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/79.jpg)
Ramtin Madani, UT Arlington
Relaxation of Network Constraints
11
• The network constraints can be cast linearly w.r.t. .
• Semidefinite programming (SDP) relaxation:
![Page 80: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/80.jpg)
Ramtin Madani, UT Arlington
Relaxation of Network Constraints
11
• The network constraints can be cast linearly w.r.t. .
• Semidefinite programming (SDP) relaxation:
• Decomposed SDP: A collection of overlapping subsets are obtainedfrom a graph-theoretic analysis of the network:
![Page 81: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/81.jpg)
Ramtin Madani, UT Arlington
Relaxation of Network Constraints
11
• The network constraints can be cast linearly w.r.t. .
• Semidefinite programming (SDP) relaxation:
• Decomposed SDP: A collection of overlapping subsets are obtainedfrom a graph-theoretic analysis of the network:
• Second-order cone programming (SOCP):
![Page 82: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/82.jpg)
Ramtin Madani, UT Arlington
Relaxation of Network Constraints
12
SDPRelaxation
SOCPRelaxation
• SDP relaxation is computationally expensive while SOCP has poor performance:
![Page 83: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/83.jpg)
Ramtin Madani, UT Arlington
Relaxation of Network Constraints
12
SDPRelaxation
SOCPRelaxation
TSDP Relaxation
• SDP relaxation is computationally expensive while SOCP has poor performance:
• The proposed TSDP relaxation:
![Page 84: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/84.jpg)
Ramtin Madani, UT Arlington
Relaxation of Network Constraints
12
SDPRelaxation
SOCPRelaxation
TSDP Relaxation
• SDP relaxation is computationally expensive while SOCP has poor performance:
• The proposed TSDP relaxation:
• More scalable than SDP and more likely to result in a feasible solution compared to SOCP.
![Page 85: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/85.jpg)
Ramtin Madani, UT Arlington
Convex Relaxation
13
• Consider a problem with time periods and units:
![Page 86: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/86.jpg)
Ramtin Madani, UT Arlington
Convex Relaxation
13
• Consider a problem with time periods and units:
• Third-order semidefinite relaxation:
![Page 87: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/87.jpg)
Ramtin Madani, UT Arlington
Convex Relaxation
13
• Consider a problem with time periods and units:
• Third-order semidefinite relaxation:
![Page 88: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/88.jpg)
Ramtin Madani, UT Arlington
Conclusions• Third-order semidefinite relaxation:
• Joint UC-AC-OPF.
14
4-bus power network:
Feasible region of voltage angles:
![Page 89: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/89.jpg)
Ramtin Madani, UT Arlington
Conclusions• Third-order semidefinite relaxation:
• Joint UC-AC-OPF.
• Advantages:• Compatible with the AC model of networks,• Massively scalable,• Near-globally optimal feasible points.
14
4-bus power network:
Feasible region of voltage angles:
![Page 90: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/90.jpg)
Ramtin Madani, UT Arlington
Conclusions• Third-order semidefinite relaxation:
• Joint UC-AC-OPF.
• Advantages:• Compatible with the AC model of networks,• Massively scalable,• Near-globally optimal feasible points.
• Future directions:• Incorporation of uncertainties and security
consideration,• Implementation on GPU: Linear algebra on
3 x 3 matrices has closed-form solutions.
14
4-bus power network:
Feasible region of voltage angles:
![Page 91: A Scalable Semidefinite Relaxation Approach to Grid Scheduling · • Semidefinite programming (SDP) relaxation: • Decomposed SDP: A collection of overlapping subsets are obtained](https://reader035.vdocuments.site/reader035/viewer/2022071423/611d8875b93a11306349a2c3/html5/thumbnails/91.jpg)
Ramtin Madani, UT Arlington
Thank you