Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
© 2000-2016 New York Independent System Operator, Inc. All Rights Reserved. © 2000-2016 New York Independent System Operator, Inc. All Rights Reserved.
Small Resource Integration Challenges for Large Scale Day-Ahead Security
Constrained Unit Commitment Cuong Nguyen
Senior Market Technologies Research Engineer New York Independent System Operator
Control at Large Scales: Energy Markets and Responsive Grids University of Minnesota -- Institute for Mathematics and its Applications
Minneapolis, MN May 11, 2016
© 2000-2016 New York Independent System Operator, Inc. All Rights Reserved. 2
NYISO Facts & Figures New York State population 19 million
2015 Peak Load 31,138 MW
2015 Required Installed Capacity 39,273 MW
Record peak 33,956 MW (July 19, 2013)
Power Generation 700+ units
High-Voltage Transmission 11,000+ circuit-miles
Average Annual Market Transactions $7.5 Billion
Market Participants 400+
© 2000-2016 New York Independent System Operator, Inc. All Rights Reserved. 3
Unit Commitment is at the Heart of NYISO Market Systems
Input • Bids/Offers • Forecasts
Output • Least cost solution • Meet reliability needs
SCUC
Bids & Offers • GEN Supply • LSE • Virtual • Regulation • Reserve • Bilaterals • Import/Exports • Demand Response
Post Results • Clearing Prices
• LBMP (GEN & Zonal) • Reserve • Regulation
• GEN Schedule • Transaction Schedule • Bilateral Transactions • Reserve Commitment • Regulation Commitment
Security Constrained Economic
Commitment and Dispatch
© 2000-2016 New York Independent System Operator, Inc. All Rights Reserved. 4
Core Innovation & R&D Distribution System Platform Providers New York State Community PartnershipREV Campus Challenge
New York Green Bank: $1 Billion
Energy Storage R&D/Commercialization:NY-BEST & Brookhaven National Lab
Five Cities Energy Plans: 20% Reduction in Municipal Energy Consumption by
2020
Other Renewable Initiatives:K-Solar
Shared Renew ablesOffshore Wind Initiative
Community Choice AggregationAGILe:
$35 Million Lab - Smart Grid Technologies for Eff iciency, Reliability, Resilience
New York Sun Initiative:10-Year, $1 Billion
3,000 MW
Long Island "Utility 2.0" Plan:Smart Solar, Vehicle to Grid, Eff iciency
Smart Generation & Transmission:440MW Increased Pow er Flow on Existing Lines
Large-Scale Renewables:10-Year, $1.5 Billion
Bundled PPAs?
REV Business Model Demonstrations BuildSmart New York:20% Energy Use Reduction in State-Ow ned Buildings
Energy Efficiency Programs:NYSERDA, Utility-Based
New York Prize Community Microgrids:$40 Million
Energy Highway
Reforming the Energy Vision: Foundational Element of State Energy PlanGuiding Principles: Market Transformation; Community Engagement; Efficiency; Private Sector Investment; Innovation & Technology; Customer Value & Choice
2015 New York State Energy PlanComprehensive Roadmap to Build a Clean, Resilient, Affordable Energy System for All New Yorkers
Clean Energy Fund: $5 Billion 2016-2026Attract Private Capital
Greater Deployment/Maturity of Clean Energy TechnologySignif icant Greenhouse Gas Reductions
REV Proceeding:Facilitate Expansion of Distributed Energy Resource (DER) and Align
Utility BusinessModels to Support Clean Energy
NYPA Leadership:Inform pow er supply, and demand-side programs
Addressing Key Challenges for Goal Attainment:Clean, Resilient, Affordable, Regulatory Reform, Environmental Justice, Clean & Reliable Transportation
2030 Greenhouse Gas Goal:40% Emissions Reduction Relative to 1990 Levels
2030 Renewable Energy Goal:50% of Electricity Generation from Renew able Energy Resources
2030 Energy Efficiency Goal:23% Reduction in Building Energy Consumption
Majority of state funding to directly influence markets
http://energyplan.ny.gov/Plans/2015
© 2000-2016 New York Independent System Operator, Inc. All Rights Reserved. 6
Small Resources • Transmission Operator Distribution Operator
Contemplates many, smaller transactions Currently 1 MW minimum for NYISO
• SCUC optimization branch and bound technique leaves “MIP gap” Initial unit commitment may not be physically feasible, and SCUC must iterate to
achieve a least-cost unit commitment while respecting all system constraints Problem is bounded at a production cost limit The impact of small resources on the solution may be less than the production
cost limit. In this circumstance, a branch-and-bound Mixed Integer Programming (MIP) solution does not determine when such resources’ commitment will enhance efficiency
• Commitment of small resource will require: more processing power new optimization techniques, and/or an active role for distribution operators
© 2000-2016 New York Independent System Operator, Inc. All Rights Reserved. 7
Small Resource Integration
Notes: • Line impedances are equal • Units offer two equal MW segments between [Pmin, Pmax]
G1 500 1270 150 300 10 12G2 700 1660 205 305 10 11G3 200 310 105 135 12 21G4 200 316 105 135 12 19G5 300 430 55 95 13 23G6 3 8 1 3 10 11G7 2 3 1 2 12 20G8 6 8 1 2 13 20
Unit Startup Cost ($)
Mingen Cost ($/h)
Pmin (MW)
Pmax (MW)
Incremental Cost ($/MWh)
© 2000-2016 New York Independent System Operator, Inc. All Rights Reserved. 8
Small Resource Integration
© 2000-2016 New York Independent System Operator, Inc. All Rights Reserved. 9
Classic Branch-and-Bound MIP Solution Procedure
• Solution technique Branch-and-bound method is used to illustrate the solution procedure Depth-first search strategy is applied for the node selection operation Least fraction strategy is adopted to choose binary variables for the branching operation
© 2000-2016 New York Independent System Operator, Inc. All Rights Reserved. 11
Decomposition Approach for Solving
Small Resources
© 2000-2016 New York Independent System Operator, Inc. All Rights Reserved. 12
Tighter Reformulation Approach for Solving Small Resources*
Convex hull of an MIP problem is the smallest polyhedron that contains all integer feasible solutions. Convex hull formulation provides a tighter lower bound, and in turn could reduce the search effort of the branch-and-bound algorithm
Ideal: An MIP formulation is called ideal if vertices of the corresponding LP relaxation satisfy integrality requirements. That is the optimal solution to the original MIP problem can be directly obtained by solving the LP relaxation problem. However it’s hard to find the explicit convex hull formulation of the entire MIP problem
Local: convex hull formulation is sought for a specifically selected portion, which could help tighten the lower bound and in turn reduce the search effort of the BAB algorithm, especially when such a subset includes key binary decision variables of the SCUC problem.
* Lei Wu, “Accelerating NCUC Via Binary Variable-Based Locally Ideal Formulation and Dynamic Global Cuts,” IEEE Transactions on Power Systems, December 2015
© 2000-2016 New York Independent System Operator, Inc. All Rights Reserved. 13
Tighter Reformulation Approach for Solving Small Resources*
• General purpose branch-and-bound solvers like CPLEX, GUROBI may not be aware of the specific SCUC problem structure Unnecessarily branch on one of the many variables Resort to heuristic rounding approaches for obtaining integer feasible solutions
• Locally ideal reformulation Commitment variable Iit is closely related to dispatchable variable Pit Iit is largely dependent on the operation cost Cit Tighter reformulation can be achieved to dramatically reduce computational burden
* Lei Wu, “Accelerating NCUC Via Binary Variable-Based Locally Ideal Formulation and Dynamic Global Cuts,” IEEE Transactions on Power Systems, December 2015
© 2000-2016 New York Independent System Operator, Inc. All Rights Reserved. 14
Locally Ideal Reformulation for Branch-and-Bound MIP Solution Procedure
• Solution technique Branch-and-bound method is used to illustrate the solution procedure Depth-first search strategy is applied for the node selection operation Least fraction strategy is adopted to choose binary variables for the branching operation
© 2000-2016 New York Independent System Operator, Inc. All Rights Reserved. 15
Technical Challenges • System boundaries
How far we would like to push the boundary of the large scale day-ahead security constrained unit commitment?
• Solution techniques Decomposition approach? Tighter MIP reformulation to exploit special characteristic of energy system? Smarter cuts?
• Computing resources Hi-Performance Computing Cloud computing
© 2000-2016 New York Independent System Operator, Inc. All Rights Reserved. 16
The Mission of the New York Independent System Operator, in collaboration with its stakeholders, is to serve the public interest and provide benefit to consumers by:
• Maintaining and enhancing regional reliability
• Operating open, fair and competitive wholesale electricity markets
• Planning the power system for the future
• Providing factual information to policy makers, stakeholders and investors in the power system
www.nyiso.com