[Federal Register Volume 88, Number 118 (Wednesday, June 21, 2023)]
[Notices]
[Pages 40234-40253]
From the Federal Register Online via the Government Publishing Office [www.gpo.gov]
[FR Doc No: 2023-13168]


-----------------------------------------------------------------------

DEPARTMENT OF ENERGY

Federal Energy Regulatory Commission

[Docket No. AD10-12-014]


Increasing Market and Planning Efficiency through Improved 
Software; Second Supplemental Notice of Technical Conference on 
Increasing Real-Time and Day-Ahead Market and Planning Efficiency 
Through Improved Software

    As first announced in the Notice of Technical Conference issued in 
this proceeding on February 7, 2023, Commission staff will convene a 
technical conference on June 27, 28, and 29, 2023 to discuss 
opportunities for increasing real-time and day-ahead market and 
planning efficiency of the bulk power system through improved software. 
Attached to this Second Supplemental Notice is the agenda for the 
technical conference and speakers' summaries of their presentations.
    While the intent of the technical conference is not to focus on any 
specific matters before the Commission, some conference discussions 
might include topics at issue in proceedings that are currently pending 
before the Commission, including topics related to capacity valuation 
methodologies for renewable, hybrid, or storage resources. These 
proceedings include, but are not limited to:


[[Page 40235]]


PJM Interconnection, L.L.C., Docket No. EL21-83-000
California Independent System Operator Corp., Docket No. ER21-2455-004
New York Independent System Operator, Inc., Docket No. ER21-2460-003
ISO New England, Inc., Docket No. ER22-983-002
PJM Interconnection, L.L.C., Docket No. ER22-962-003
Southwest Power Pool, Inc., Docket No. ER22-1697-001
Midcontinent Independent System Operator, Inc., Docket No. ER22-1640-
000
ISO New England, Inc., Docket No. EL22-42-000
Southwest Power Pool, Inc., Docket No. ER22-379-000
PJM Interconnection, L.L.C., Docket No. ER22-1200-000
California Independent System Operator Corp., Docket No. ER23-1485-000
California Independent System Operator Corp., Docket No. ER23-1533-000
California Independent System Operator Corp., Docket No. ER23-1534-000
Midcontinent Independent System Operator, Inc., Docket No. EL23-28
Midcontinent Independent System Operator, Inc., Docket No. ER23-1195
Midcontinent Independent System Operator, Inc., Docket No. EL23-46

    The conference will take place in a hybrid format, with presenters 
and attendees allowed to participate either in-person or virtually. 
Further details on both in-person and virtual participation will be 
available on the conference web page.\1\ Foreign nationals attending 
in-person must register through the Commission's website on or before 
June 2, 2023. We also encourage all other in-person attendees to also 
register through the Commission's website on or before June 2, 2023, to 
help ensure Commission staff can provide sufficient physical and 
virtual facilities and to communicate with attendees in the case of 
unanticipated emergencies or other changes to the conference schedule 
or location. Access to the conference (virtual or in-person) may not be 
available to those who do not register.
---------------------------------------------------------------------------

    \1\ https://www.ferc.gov/news-events/events/increasing-real-time-and-day-ahead-market-and-planning-efficiency-through.
---------------------------------------------------------------------------

    The Commission will accept comments following the conference, with 
a deadline of July 28, 2023.
    There is an ``eSubscription'' link on the Commission's website that 
enables subscribers to receive email notification when a document is 
added to a subscribed docket(s). For assistance with any FERC Online 
service, please email [email protected], or call (866) 208-
3676 (toll free). For TTY, call (202) 502-8659.
    FERC conferences are accessible under section 508 of the 
Rehabilitation Act of 1973. For accessibility accommodations please 
send an email to [email protected] or call toll free (866) 208-
3372 (voice) or (202) 502-8659 (TTY), or send a fax to (202) 208-2106 
with the required accommodations.
    For further information about these conferences, please contact:

Sarah McKinley (Logistical Information), Office of External Affairs, 
(202) 502-8004, [email protected]
Alexander Smith (Technical Information), Office of Energy Policy and 
Innovation, (202) 502-6601, [email protected]

    Dated: June 14, 2023.
Debbie-Anne A. Reese,
Deputy Secretary.
[GRAPHIC] [TIFF OMITTED] TN21JN23.069

Technical Conference: Increasing Real-Time and Day-Ahead Market 
Efficiency Through Improved Software

Agenda

AD10-12-014

June 27-29, 2023

Tuesday, June 27, 2023

9:15 a.m. Introduction
Elizabeth Topping, Federal Energy Regulatory Commission (Washington, 
DC)
9:30 a.m. Session T1 (Commission Meeting Room)
Probabilistic Energy Adequacy Assessment under Extreme Weather Events
    Jinye Zhao, ISO New England (Holyoke, MA)
    Stephen George, ISO New England (Holyoke, MA)
    Ke Ma, ISO New England (Holyoke, MA)
    Steven Judd, ISO New England (Holyoke, MA)
    Eamonn Lannoye, EPRI (Dublin, Ireland)
    Juan Carlos Martin, EPRI (Madrid, Spain)
Transmission Outage Probability Estimation Based on Real-Time Weather 
Forecast
    Mingguo Hong, ISO New England (Holyoke, MA)
    Xiaochuan Luo, ISO New England (Holyoke, MA)
    Slava Maslennikov, ISO New England (Holyoke, MA)
    Tongxin Zheng, ISO New England (Holyoke, MA)
Overview of MISO and PJM Hybrid Multiple Configuration Resource Model 
Implementation Within PROBE Software
    Qun Gu, PowerGEM (Clifton Park, NY)
    Boris Gisin, PowerGEM (Clifton Park, NY)
    Anthony Giacomoni, PJM Interconnection (Audubon, PA)
    Chuck Hansen, Midcontinent ISO (Carmel, IN)
Optimizing Combined Cycle Units in PJM's Wholesale Energy Markets using 
a Hybrid Multiple Configuration Resource Model
    Anthony Giacomoni, PJM Interconnection (Audubon, PA)
    Danial Nazemi, PJM Interconnection (Audubon, PA)
    Qun Gu, PowerGEM (Clifton Park, NY)
    Boris Gisin, PowerGEM (Clifton Park, NY)
11:30 a.m. Lunch
12:30 p.m. Session T2 (Commission Meeting Room)
Enhancements to Ramp Rate Dependent Spinning Reserve Modeling
    Shubo Zhang, New York ISO (Rensselaer, NY)
    John L. Meyer, New York ISO (Rensselaer, NY)
    Iiro Harjunkoski, Hitachi Energy (Mannheim, Germany)
Determining Dynamic Operating Reserve Requirements for Reliability and 
Efficient Market Outcomes: Tradeoffs and Price Formation Challenges
    Matthew Musto, New York ISO (Rensselaer, NY)
    Kanchan Upadhyay, New York ISO (Rensselaer, NY)
    Edward O Lo, Hitachi Energy (San Jose, CA)
Operational Experience with Nodal Procurement of Flexible Ramping 
Product
    Guillermo Bautista-Alderete, California ISO (Folsom, CA)
    George Angelidis, California ISO (Folsom, CA)
    Yu Wan, California ISO (Folsom, CA)

[[Page 40236]]

    Kun Zhao, California ISO (Folsom, CA)
Impact of DERs on Load Distribution Factors in Forecasting
    Khaled Abdul-Rahman, California ISO (Folsom, CA)
    Hani Alarian, California ISO (Folsom, CA)
    Trevor Ludlow, California ISO (Folsom, CA)
    Chiranjeevi Madvesh, California ISO (Folsom, CA)
Increased Congestion in SPP and Optimization in the Day Ahead Market 
with Gurobi
    Seth Mayfield, Southwest Power Pool (Little Rock, AR)
    Yasser Bahbaz, Southwest Power Pool (Little Rock, AR)
3:00 p.m. Break
3:30 p.m. Session T3 (Commission Meeting Room)
MISO Operations Risk Assessment and Uncertainty Management
    Congcong Wang, Midcontinent ISO (Carmel, IN)
    Long Zhao, Midcontinent ISO (Carmel, IN)
    Jason Howard, Midcontinent ISO (Carmel, IN)
Market Simulation Tools and Uncertainty Quantification Methods to 
Support Operational Uncertainty Management
    Nazif Faqiry, Midcontinent ISO (Carmel, IN)
    Arezou Ghesmati, Midcontinent ISO (Carmel, IN)
    Bing Huang, Midcontinent ISO (Carmel, IN)
    Yonghong Chen, Midcontinent ISO (Carmel, IN)
    Bernard Knueven, National Renewable Energy Laboratory (Golden, CO)
Pumped Storage Optimization in Real-time Markets under Uncertainty
    Bing Huang, Midcontinent ISO (Carmel, IN)
    Arezou Ghesmati, Midcontinent ISO (Carmel, IN)
    Yonghong Chen, Midcontinent ISO (Carmel, IN)
    Ross Baldick, University of Texas at Austin (Austin, TX)
Forecasting Aggregate Electricity Demand on a 5-minute Basis using 
Machine Learning
    Yinghua Wu, PJM Interconnection (Audubon, PA)
    Laura Walter, PJM Interconnection (Audubon, PA)
    Anthony Giacomoni, PJM Interconnection (Audubon, PA)
Long-Term Outlook for the ERCOT Grid
    Pengwei Du, Electric Reliability Corporation of Texas (Austin, TX)
6:00 p.m. Adjourn

Wednesday, June 28, 2023

9:00 a.m. Session W-A1 (Commission Meeting Room)
Uncertainty-Informed Renewable Energy Scheduling: A Scalable Bilevel 
Framework
    Dongwei Zhao, Massachusetts Institute of Technology (Cambridge, MA)
    Vladimir Dvorkin, Massachusetts Institute of Technology (Cambridge, 
MA)
    Stefanos Delikaraoglou, Axpo Solutions AG (Zurich, Switzerland)
    Alberto J. Lamadrid L., Lehigh University (Bethlehem, PA)
    Audun Botterud, Massachusetts Institute of Technology (Cambridge, 
MA)
Enhancing Power System Resilience and Efficiency through Proactive 
Security Assessments and the Use of powerSAS.m: A Robust, Efficient, 
and Scalable Security Analysis Tool for Large-Scale Systems
    Yang Liu, Argonne National Laboratory (Lemont, IL)
    Feng Qiu, Argonne National Laboratory (Lemont, IL)
    Jianzhe Liu, Argonne National Laboratory (Lemont, IL)
Stochastic Unit Commitment and Market Clearing in Julia with 
UnitCommitment.jl
    Alinson Santos Xavier, Argonne National Laboratory (Lemont, IL)
    Og[uuml]n Yurdakul, Technische Universit[auml]t Berlin (Berlin, 
Germany)
    Aleksandr M. Kazachkov, University of Florida (Gainesville, FL)
    Jun He, Purdue University (West Lafayette, IN)
    Feng Qiu, Argonne National Laboratory (Lemont, IL)
Reduced-order Decomposition and Coordination Approach for Markov-based 
Stochastic UC with High Penetration Level of Wind and BESS
    Niranjan Raghunathan, University of Connecticut (Storrs, CT)
    Peter B. Luh, University of Connecticut and National Taiwan 
University (Alexandria, VA)
    Zongjie Wang, University of Connecticut (Storrs, CT)
    Mikhail A. Bragin, University of California, Riverside (Riverside, 
CA)
    Bing Yan, Rochester Institute of Technology (Rochester, NY)
    Meng Yue, Brookhaven National Laboratories (Upton, NY)
    Tianqiao Zhao, Brookhaven National Laboratories, (Upton, NY)
Learn to Branch and Dive for Large-scale Unit Commitment Problem
    Jingtao Qin, University of California, Riverside (Riverside, CA)
    Nanpeng Yu, University of California, Riverside (Riverside, CA)
    Mikhail Bragin, University of Connecticut (Storrs, CT)
9:00 a.m. Session W-B1 (Hearing Room One)
Stochastic Nodal Adequacy Pricing Platform (SNAP)
    Richard D. Tabors, Tabors Caramanis Rudkevich (Newton, MA)
    Aleksandr Rudkevich, Newton Energy Group (Newton, MA)
    Russel Philbrick, Polaris Systems Optimization (Seattle, WA)
    Selin Yanikara, Newton Energy Group (Newton, MA)
Assessing Nodal Adequacy of Large Power Systems
    F. Selin Yanikara, Newton Energy Group (Newton, MA)
    Russ Philbrick, Polaris Systems Optimization (Seattle, WA)
    Aleksandr M. Rudkevich, Newton Energy Group (Newton, MA)
    Sophie Edelman, The Brattle Group (New York, NY)
Comparison of Flexibility Reserve and ORDC for Increasing System 
Flexibility
    Phillip de Mello, Electric Power Research Institute (Niskayuna, NY)
    Erik Ela, Electric Power Research Institute (Boulder, CO)
    Nikita Singhal, Electric Power Research Institute (Palo Alto, CA)
    Alexandre Moreira da Silva, Lawrence Berkeley National Laboratory 
(Berkeley, CA)
    Miguel Heleno, Lawrence Berkeley National Laboratory (Berkeley, CA)
ABSCORES, A Novel Application of Banking Scoring and Rating for 
Electricity Systems
    Alberto J. Lamadrid L., Lehigh University (Bethlehem, PA)
    Audun Botterud, Massachusetts Institute of Technology (Cambridge, 
MA)
    Jhi-Young Joo, Lawrence Livermore National Laboratory (Livermore, 
CA)
    Shijia Zhao, Argonne National Laboratory (Lemont, IL)
Recent Developments in the Day-ahead and Real-time Electricity Market 
Design and Software Caused by the Higher Energy Costs and Emerging 
Technologies--European Experience
    Petr Svoboda, Unicorn Systems A.S. (Prague, Czech Republic)

[[Page 40237]]

11:30 a.m. Lunch
12:30 p.m. Session W-A2 (Commission Meeting Room)
System Resilience through Electricity System Restoration and Related 
Services
    Douglas Wilson, General Electric (Edinburgh, United Kingdom)
    James Yu, ScottishPower Energy Networks (Glasgow, United Kingdom)
    Ian Macpherson, ScottishPower Energy Networks (Glasgow, United 
Kingdom)
    Marta Laterza, General Electric (Glasgow, United Kingdom)
    Marcos Santos, General Electric (Glasgow, United Kingdom)
    Richard Davey, General Electric (Glasgow, United Kingdom)
Coordinated Cross-Border Capacity Calculation Through The FARAO Open-
Source Toolbox
    Violette Berge, Artelys Canada (Montr[eacute]al, Canada)
    Nicolas Omont, Artelys (Paris, France)
Advanced Scenario Selection Methods for Probabilistic Transmission 
Planning Assessments
    Eknath Vittal, Electric Power Research Institute (Palo Alto, CA)
    Anish Gaikwad, Electric Power Research Institute (Palo Alto, CA)
    Parag Mitra, Electric Power Research Institute (Palo Alto, CA)
Incorporating Climate Projections into Grid Models: Bridging the Data 
Gap to Capture Weather Dependent Representative and Extreme Events and 
Corresponding Uncertainties
    Zhi Zhou, Argonne National Laboratory (Lemont, IL)
    Neal Mann, Argonne National Laboratory (Lemont, IL)
    Yanwen Xu, University of Illinois at Chicago, Urbana-Champaign 
(Champaign, IL)
    Zuguang Gao, University of Chicago (Chicago, IL)
    Akintomide Akinsanola, University of Illinois at Chicago (Chicago, 
IL)
    Todd Levin, Argonne National Laboratory (Lemont, IL)
    Jonghwan Kwon, Argonne National Laboratory (Lemont, IL)
    Audun Botterud, Senior Energy Systems Engineer, Argonne National 
Laboratory (Lemont, IL)
12:30 p.m. Session W-B2 (Hearing Room One)
Enhancing Decision Support for Electricity Markets with Machine 
Learning
    Yury Dvorkin, Johns Hopkins University (Baltimore, MD)
    Robert Ferrando, University of Arizona (Tucson, AZ)
    Laurent Pagnier, University of Arizona (Tucson, AZ)
    Zhirui Liang, Johns Hopkins University (Baltimore, MD)
    Daniel Bienstock, Columbia University (New York, NY)
    Michael Chertkov, University of Arizona (Tucson, AZ)
Boosting Power System Operation Economics via Closed-loop Predict-and-
Optimize
    Lei Wu, Stevens Institute of Technology (Hoboken, NJ)
    Xianbang Chen, Stevens Institute of Technology (Hoboken, NJ)
Synergistic Integration of Machine Learning and Mathematical 
Optimization for Sub-hourly Unit Commitment
    Jianghua Wu, University of Connecticut (Storrs, CT)
    Zongjie Wang, University of Connecticut (Storrs, CT)
    Yonghong Chen, MIDCONTINENT ISO (Carmel, IN)
    Bing Yan, Rochester Institute of Technology (Rochester, NY)
    Mikhail Bragin, University of California, Riverside (Riverside, CA)
Privacy-Preserving Synthetic Dataset Generation for Power Systems 
Research
    Vladimir Dvorkin, Massachusetts Institute of Technology (Cambridge, 
MA)
    Audun Botterud, Massachusetts Institute of Technology (Cambridge, 
MA)
2:30 p.m. Break
3:00 p.m. Session W-A3 (Commission Meeting Room)
Parallel Interior-Point Solver for Security Constrained ACOPF problems 
on SIMD/GPU Architectures
    Mihai Anitescu, Argonne National Laboratory (Lemont, IL)
    Fran[ccedil]ois Pacaud, Ecole des Mines (Paris, France)
    Michel Schanen, Argonne National Laboratory (Lemont, IL)
    Sungho Shin, Argonne National Laboratory (Lemont, IL)
    Daniel Adrian Maldonado, Argonne National Laboratory (Lemont, IL)
The Need for More Rigorous Calculation of Shadow Prices and LMPs
    Xiaoming Feng, Hitachi Energy (Raleigh, NC)
Real-Time Market Enhancements for Reliability and Efficiency
    Mort Webster, Pennsylvania State University (University Park, PA)
    Anthony Giacomoni, PJM Interconnection (Audubon, PA)
    Aravind Retna Kumar, Pennsylvania State University (University 
Park, PA)
    Sushant Varghese, Pennsylvania State University (University Park, 
PA)
    Shailesh Wasti, Pennsylvania State University (University Park, PA)
Economics of Grid-Supported Electric Power Markets: A Fundamental 
Reconsideration
    Leigh Tesfatsion, Iowa State University (Ames, IA)
3:00 p.m. Session W-B3 (Hearing Room One)
Simulation of Wholesale Electricity Markets with Capacity Expansion and 
Production Cost Models to Understand Feedback between Short-Term Market 
Procedures and Long-Term Investment Incentives
    Jesse Holzer, Pacific Northwest National Laboratory (Richland, WA)
    Abhishek Somani, Pacific Northwest National Laboratory (Richland, 
WA)
    Brent Eldridge, Pacific Northwest National Laboratory (Bel Air, MD)
    Diane Baldwin, Pacific Northwest National Laboratory (Richland, WA)
Making the Right Resource Choice Requires Making the Right Model Choice
    Rodney Kizito, Ascend Analytics (Wheaton, MD)
    Gary W. Dorris, Ascend Analytics, CEO (Boulder, CO)
    David Millar, Ascend Analytics (Boulder, CO)
Transmission Shortage Pricing By MW-Mile Based Demand Curve
    Sina Gharebaghi, Pennsylvania State University (University Park, 
PA)
    Xiaoming Feng, Hitachi Energy (Raleigh, NC)
Grid OS--A Modern Software Portfolio for Grid Orchestration
    Renan Giovanini, General Electric (Edinburgh, UK)
    Joseph Franz, General Electric (Melbourne, FL)
5:00 p.m. Adjourn

Thursday, June 29, 2023

9:30 a.m. Session H1 (Commission Meeting Room)
Integration of DER Aggregations in ISO-Scale SCUC Models
    Brent Eldridge, Pacific Northwest National Laboratory (Bel Air, MD)
    Jesse Holzer, Pacific Northwest National Laboratory (Richland, WA)
    Abhishek Somani, Pacific Northwest National Laboratory (Richland, 
WA)
    Eran Schweitzer, Pacific Northwest National Laboratory (Richland, 
WA)
    Rabayet Sadnan, Pacific Northwest National Laboratory (Richland, 
WA)
    Nawaf Nazir, Pacific Northwest National Laboratory (Richland, WA)

[[Page 40238]]

    Soumya Kundu, Pacific Northwest National Laboratory (Richland, WA)
Current-Voltage AC Optimal Power Flow for Unbalanced Distribution 
Network
    Mojdeh Khorsand Hedman, Arizona State University (Tempe, AZ)
    Zahra Soltani, Arizona State University (Tempe, AZ)
    Shanshan Ma, Arizona State University (Las Vegas, NV)
Empowering Electricity Markets through Distributed Energy Resources and 
Smart Building Setpoint Optimization: A Graph Neural Network-Based Deep 
Reinforcement Learning Approach
    You Lin, Massachusetts Institute of Technology (Cambridge, MA)
    Audun Botterud, Massachusetts Institute of Technology (Cambridge, 
MA)
    Daisy Green, Massachusetts Institute of Technology (Cambridge, MA)
    Leslie Norford, Massachusetts Institute of Technology (Cambridge, 
MA)
    Jeremy Gregory, Massachusetts Institute of Technology (Cambridge, 
MA)
Multi-timescale Operations of Nuclear-Renewable Hybrid Energy Systems 
for Reserve and Thermal Products Provision
    Jie Zhang, University of Texas at Dallas (Richardson, TX)
    Jubeyer Rahman, University of Texas at Dallas (Richardson, TX)
11:30 a.m. Lunch
12:30 p.m. Session H2 (Commission Meeting Room)
Optimizing Stand-Alone Battery Storage Operations Scheduling Under 
Uncertainties in German Residential Electricity Market Using Stochastic 
Dual Dynamic Programming
    Pattanun Chanpiwat, University of Maryland & Aalto University 
(College Park, MD; Espoo, Finland)
    Fabricio Oliveira, Aalto University (Espoo, Finland)
    Steven A. Gabriel, University of Maryland (College Park, MD)
Integration of Hybrid Storage Resources into Wholesale Electricity 
Markets
    Nikita Singhal, Electric Power Research Institute (Palo Alto, CA)
    Rajni Kant Bansal, Johns Hopkins University (Baltimore, MD)
    Erik Ela, Electric Power Research Institute (Palo Alto, CA)
    Julie Mulvaney Kemp, Lawrence Berkeley National Laboratory 
(Berkeley, CA)
    Miguel Heleno, Lawrence Berkeley National Laboratory (Berkeley, CA)
Predicting Strategic Energy Storage Behaviors
    Yuexin Bian, University of California (San Diego, CA)
    Ningkun Zheng, Columbia University (New York City, NY)
    Yang Zheng, University of California--San Diego (San Diego, CA)
    Bolun Xu, Columbia University (New York, NY)
    Yuanyuan Shi, University of California--San Diego (San Diego, CA)
Energy Storage Participation Algorithm Competition (ESPA-Comp)
    Brent Eldridge, Pacific Northwest National Laboratory (Bel Air, MD)
    Jesse Holzer, Pacific Northwest National Laboratory (r)
    Abhishek Somani, Pacific Northwest National Laboratory (Richland, 
WA)
    Kostas Oikonomou, Pacific Northwest National Laboratory (Richland, 
WA)
    Brittany Tarufelli, Pacific Northwest National Laboratory (Laramie, 
WY)
    Li He, Pacific Northwest National Laboratory (Richland, WA)
2:30 p.m. Break
3:00 p.m. Session H3 (Commission Meeting Room)
Congestion Mitigation with Transmission Reconfigurations in the Evergy 
Footprint
    Pablo A. Ruiz, NewGrid (Somerville, MA)
    Derek Brown, Evergy (Topeka, KS)
    Jeremy Harris, Evergy (Topeka, KS)
    German Lorenzon, NewGrid (Somerville, MA)
    Grant Wilkerson, Evergy (Kansas City, MO)
Optimal Transmission Expansion Planning with Grid Enhancing 
Technologies
    Swaroop Srinivasrao Guggilam, Electric Power Research Institute 
(Knoxville, TN)
    Alberto Del Rosso, Electric Power Research Institute (Knoxville, 
TN)
The Key Role of Extended ACOPF-based Decision Making for Supporting 
Clean, Cost-Effective and Reliable/Resilient Electricity Services
    Maria Ilic, Carnegie Mellon University (Pittsburgh, PA)
    Rupamathi Jaddivada, SmartGridz (Boston, MA)
    Jeffrey Lang, Massachusetts Institute of Technology (Cambridge, MA)
    Eric Allen, SmartGridz (Boston, MA)
Data & API Standards for Clean Energy Solutions and Digital Innovation
    Priya Barua, Clean Energy Buyers Institute (Washington, DC)
    Ben Gerber, M-RETS (Minneapolis, MN)
Mine Production Scheduling under Time-of-Use Power Rates with Renewable 
Energy Sources
    Daniel Bienstock, Columbia University (New York, NY)
    Amy Mcbrayer, South Dakota School of Mines (Rapid City, SD)
    Andrea Brickey, South Dakota School of Mines (Rapid City, SD)
    Alexandra Newman, Colorado School of Mines (Golden, CO)
5:30 p.m. Adjourn

Conference Abstracts

Day 1--Tuesday, June 27

Session T1 (Tuesday, June 27, 9:30 a.m.) Commission Meeting Room

Probabilistic Energy Adequacy Assessment Under Extreme Weather Events
Dr. Jinye Zhao, Technical Manager, ISO New England (Holyoke, MA)
Stephen George, Director, ISO New England (Holyoke, MA)
Dr. Ke Ma, Senior Analyst, ISO New England (Holyoke, MA)
Steven Judd, Manager, ISO New England (Holyoke, MA)
Dr. Eamonn Lannoye, Program Manager, Electric Power Research Institute 
(Dublin, Ireland)
Juan Carlos Martin, Senior Engineer, Electric Power Research Institute 
(Madrid, Spain)

    As intermittent and limited energy resources become a larger 
portion of the region's generation resource mix, and as the region's 
demand becomes increasingly electrified, it has become increasingly 
important to understand the operational risks associated with future 
weather extremes. To better inform the region's understanding of these 
risks, ISO New England in collaboration with EPRI, has developed a 
probabilistic energy adequacy assessment framework. This approach of 
stress testing the system's energy adequacy focuses on generating 
comprehensive extreme weather scenarios for the New England region and 
performing risk analyses across these scenarios. The framework offers a 
tailored approach to identify unique energy adequacy risks faced by the 
New England power system and enables us to analyze related stressors 
under extreme events.
Transmission Outage Probability Estimation Based on Real-Time Weather 
Forecast
Dr. Mingguo Hong, Principal Analyst, ISO New England (Holyoke, MA)
Dr. Xiaochuan Luo, Manager, ISO New England (Holyoke, MA)

[[Page 40239]]

Dr. Slava Maslennikov, Technical Manager, ISO New England (Holyoke, MA)
Dr. Tongxin Zheng, Director, ISO New England (Holyoke, MA)

    Extreme weather patterns including both winter and summer storms 
have been posing increasing threats to power transmission security in 
the New England area. Being able to accurately predict their impacts 
will benefit both power system operation and planning. In recent years, 
the ISO New England has been developing machine-learning algorithms for 
estimating the probability of transmission line outage in real-time, 
given weather forecast variables such as wind, temperature, snow, and 
rain precipitation, etc. This presentation will share our study 
findings and on-going software implementation experience.
Overview of MISO and PJM Hybrid Multiple Configuration Resource Model 
Implementation Within PROBE Software
Dr. Anthony Giacomoni, Manager, Advanced Analytics, PJM Interconnection 
(Audubon, PA)
Dr. Danial Nazemi, Operations Research Engineer II, PJM Interconnection 
(Audubon, PA)
Dr. Qun Gu, Principal Consultant, PowerGEM (Clifton Park, NY)
Dr. Boris Gisin, President, PowerGEM (Clifton Park, NY)

    For the past three years, PJM, MISO and PowerGEM have been working 
jointly on developing an advanced SCUC algorithm to prepare for the 
full-scale implementation of a Multiple Configuration Resource (MCR) 
model in their energy markets. PJM currently uses aggregate models for 
MCRs that do not accurately capture their true operating 
characteristics. Often MCRs may need to overestimate costs to ensure 
cost recovery, underestimate costs to ensure selection or offer reduced 
operating ranges to be able to accurately reflect their operating 
capabilities. This presentation will focus on the impacts to PJM's 
energy markets from optimizing the multiple configurations and 
components of their combined cycle units. The optimization of multiple 
configurations and components is very challenging due to the additional 
integer variables and constraints that impact the solution time and may 
lead to performance challenges. A prototype full-scale MCR model has 
been implemented in the PROBE Day-Ahead software, which is currently a 
critical component of PJM's Day-Ahead Market (DAM) clearing process. 
The prototype MCR model has the ability to perform energy and ancillary 
service co-optimization for combined cycle units with multiple 
configurations and components. The developed model has no practical 
limits on the number of configurations that each unit can have and the 
model allows for simultaneously enforcing configuration and component 
level constraints. Benefits of the new model include enhanced modeling 
flexibility and accuracy, which allows combined cycle participants to 
submit bids that align with their units' physical operating 
constraints, better alignment with the real-time model and market 
outcomes with increased social benefits. To quantify the impacts of the 
MCR model on PJM's energy markets, PJM gathered configuration and 
component data from a large number of combined cycle units in its 
footprint. Simulations using one year of historical DAM data were then 
performed to measure the impacts of the MCR model on the clearing 
engine's computational performance and market outcomes. Results clearly 
demonstrate significant potential bid production cost savings of over 
$100 million per year with a very modest increase in solution time. The 
MCR model is currently being implemented in PJM's DAM for the 
optimization of synchronous condensers. It is planned that after 
successful implementation of the MCR model for synchronous condensers 
the same model will be implemented for combined cycle units and 
possibly for hybrid resources as well.

Session T2 (Tuesday, June 27, 12:30 p.m.) (Commission Meeting Room)

Enhancements to Ramp Rate Dependent Spinning Reserve Modeling
Dr. Shubo Zhang, Energy Market Engineer, New York ISO (Rensselaer, NY)
John L. Meyer, Senior Energy Market Engineer, New York ISO (Rensselaer, 
NY)
Iiro Harjunkoski, Researcher, Hitachi Energy (Mannheim, Germany)

    In a joint effort between the NYISO and Hitachi Energy, a Ramp Rate 
Dependent (RRD) formulation of spinning reserve scheduling that 
utilizes Multiple Response Rates (MRR) across a Combined Cycle Gas 
Turbine (CCGT) generator or other dispatchable resource's range of 
output has been developed. To provide more flexibility to Market 
Participants, a ``Limited Participation'' conceptual strategy is also 
included that would allow a CCGT or other dispatchable resource to 
selectively provide spinning reserves or regulation for a certain range 
of output. This presentation will discuss the market basis and design 
of Limited Participation in spinning reserves and regulation, in the 
context of Ramp Rate Dependent Spinning Reserve Modeling.
Determining Dynamic Operating Reserve Requirements for Reliability and 
Efficient Market Outcomes: Tradeoffs and Price Formation Challenges
Matthew Musto, Technical Specialist--Market Solutions Engineering, 
NYISO (Rensselaer, NY)
Kanchan Upadhyay, Senior Energy Market Engineer--Market Solutions 
Engineering, NYISO (Rensselaer, NY)
Edward O Lo, Consultant, Hitachi Energy (San Jose, CA)

    With increasing intermittent resources in the generation mix, the 
need for more economic responsiveness and operational flexibility while 
maintaining system reliability is growing. The NYISO and Hitachi Energy 
have been working on advanced design and techniques for calculating 
operating reserve requirements dynamically for each reserve region 
while simultaneously optimizing the dispatch solution in the market 
clearing engine. A key benefit of the dynamic reserves formulation is 
the functionality to determine the least-cost generation and reserve 
mix to meet load. This dynamic determination of reserve requirements in 
New York Control Area (NYCA) and all reserve regions within the NYCA 
creates new tradeoffs between energy schedules and reserve 
requirements. This presentation will discuss these tradeoffs and 
highlight the associated price formation challenge.
Operational Experience with Nodal Procurement of Flexible Ramping 
Product
Dr. Guillermo Bautista-Alderete, Director, Market Analysis & 
Forecasting, California ISO (Folsom, CA)
George Angelidis, Executive Principal--Power Systems and Market 
Technology, California ISO (Folsom, CA)
Yu Wan, Power Systems Engineer, California ISO (Folsom, CA)
Kun Zhao, Market Engineering Specialist Lead, California ISO (Folsom, 
CA)

    The CAISO's market procures flexible ramping capacity to manage 
weather-based uncertainty realized in real time. The CAISO introduced 
this product in 2016 using a procurement requirement at the system 
level. Using a system-level procurement requirement, the market 
frequently procured flexible ramping capacity from locations impacted 
by

[[Page 40240]]

congestion, thereby stranding the flexible ramping capacity. The CAISO 
has enhanced the design of the flexible ramping product using a 
formulation that observes transmission constraints. This approach 
considers congestion management as part of the procurement of flexible 
ramping capacity helping to ensure the CAISO can deploy this capacity 
when uncertainty arises. This new design poses additional complexity 
because the market clearing process now considers transmission 
constraints for energy and for flexible ramping capacity. The CAISO 
will provide an update on the performance of its flexible ramping 
product under this new design.
Impact of DERs on Load Distribution Factors in Forecasting
Dr. Khaled Abdul-Rahman, Vice President, Power System and Market 
Technology, California ISO (Folsom, CA)
Hani Alarian, Executive Director of Power Systems Technology 
Operations, California ISO (Folsom, CA)
Trevor Ludlow, Specialist Lead of Power Systems Technology Operations, 
California ISO (Folsom, CA)
Chiranjeevi Madvesh, Lead Engineer of Power Systems Technology 
Operations, California ISO (Folsom, CA)

    The calculation of load distributing factors (LDFs) is 
traditionally performed based on a collection of historical state 
estimator calculated values and stored in libraries for use when 
simulating power system operations in look-ahead market and reliability 
applications. The inherit assumption is that bus loads are accurately 
estimated from the aggregate system load forecast using LDFs, and 
generation quantities are deterministically known. Accordingly, it is 
assumed that there is a strong correlation between the system load and 
individual bus loads. However, the proliferation of behind-the-meter 
distributed energy resources, solar rooftops, batteries, hybrid 
resources, as well as the use of behind the-meter demand response 
utility programs, and electric vehicles introduces a non-conforming 
load component at locations that were previously conforming loads.
    This issue requires a more accurate forecast of non-conforming 
loads by taking into consideration the probabilistic nature of bus 
loads and variable/intermittent generation. The CAISO's enhanced LDF 
forecast algorithm takes into account not just the average hour of the 
day and the day of the week but includes machine learning ability to 
distinguish between flows that scales up with load in both a non-linear 
and linear fashion. It also includes a new fusion-forecasting model 
that improves forecasting accuracy. Additionally, the CAISO's algorithm 
uses data engineering and preprocessing options to increase the 
accuracy of the proposed model. The CAISO analyzes load data to verify 
that the proposed methodology provides higher forecasting accuracy with 
lower error indices.
Increased Congestion in SPP and Optimization in the Day Ahead Market 
With Gurobi
Seth Mayfield, Manager of Market Support & Analysis, Southwest Power 
Pool (Little Rock, AR)
Yasser Bahbaz, Director of Markets Development, Southwest Power Pool 
(Little Rock, AR)

    SPP has seen substantial increased congestion in recent years. 
These trends have numerous reliability and economic impact. In the Day-
Ahead Market, SPP has noticed high transmission activation leading to 
longer optimization runtimes. High activations results in large 
increases in the mathematical growth, which then results in slower 
Mixed Integer Program (MIP) runtimes. Other factors include increasing 
market rules complexity (such as uncertainty product) and additional 
market resource registrations. SPP performed a study where we evaluate 
swapping our existing optimization engine (IBM's CPlex) with Gurobi's 
optimization engine. The study reran every approved DAMKT SCUC 
operating day for 2021 (365 cases). Gurobi solved the cases 41% faster 
than CPlex using Gurobi without tuning. A very light discussion with 
Gurobi resulted in a few tuning suggestions which pushed the runtime 
reduction to 43%. SPP is in the process of acquiring Gurobi licenses 
and will work with our software vendor to incorporate the engine into 
our market. Phase 1 will include simultaneously running both CPlex and 
Gurobi as we believe this will give us the best/fastest results for 
each day. It is expected that there will be a transition to using more 
Gurobi instances than CPlex as time goes on.

Session T3 (Tuesday, June 27, 3:30 p.m.) (Commission Meeting Room)

MISO Operations Risk Assessment and Uncertainty Management
Dr. Congcong Wang, Lead, Operations Risk Assessment, Midcontinent ISO 
(Carmel, IN)
Dr Long Zhao, Senior Advisor of Operations Risk Assessment, 
Midcontinent ISO (Carmel, IN)
Jason Howard, Director of Operations Risk Management, Midcontinent ISO 
(Carmel, IN)

    Fleet transition is driving a new risk profile at MISO. Uncertainty 
and Variability are increasing in their intensity, diversity, and 
volatility. While probabilistic forecasting has made progress for wind 
and solar, its integration into operations and markets is uneven. 
Furthermore, uncertainty comes in more sources than just renewable 
energy such as generation and transmission outages, fuel scarcity 
especially during extreme weather events, resulting in challenges for 
the RTO to manage the aggregated or net uncertainty. This presentation 
will outline MISO's operations risk assessment and uncertainty 
management initiatives including: (1) Characterize Risks--transform 
traditional deterministic renewable, load and ``net'' load forecasts to 
probabilistic forecasts in production systems; and assess generation 
and fuel risks to better capture the unknowns; (2) Integrate risks into 
Operations Situational Awareness and Operations Planning--provide 
control room a dynamic and geographically granular visualization of 
operating reserve margin; and visibility of weather driven operations 
risks; (3) Automate risk management through market products with 
dynamic reserve requirements--assess net uncertainty across different 
timeframes; and predict risks to establish a daily target for procuring 
market-based reserves using analytical and meteorological techniques. 
This work is done in collaboration with R&D through the joint 
Uncertainty Roadmap.
Market Simulation Tools and Uncertainty Quantification Methods To 
Support Operational Uncertainty Management
Dr. Nazif Faqiry, R&D Engineer, Midcontinent ISO (Carmel, IN)
Dr. Arezou Ghesmati, R&D Engineer, Midcontinent ISO (Carmel, IN)
Dr. Bing Huang, R&D Engineer, Midcontinent ISO (Carmel, IN)
Dr. Yonghong Chen, Consulting Advisor, Midcontinent ISO (Carmel, IN)
Dr. Bernard Knueven, Research Scientist, National Renewable Energy 
Laboratory (Golden, CO)

    Portfolio evolution and more frequent extreme weather events are 
introducing more challenges to MISO Market Operations with new risk 
profiles. To improve market efficiency and generate efficient price 
signals for operational and investment decisions, it is increasingly 
important to align market

[[Page 40241]]

design with reliability and risk management needs. This work presents 
the Electrical Grid Research & Engineering (EGRET) market simulation 
tool adapted and enhanced at MISO to evaluate existing and future 
system, and a novel netload ramp uncertainty prediction and scenario 
generation method to support stochastic simulation and reserve 
requirement settings. First, it presents a multi-periods market 
simulation tool and its capabilities, including rolling real-time unit 
commitment and economic dispatch (UCED), followed by the results of 8 
GW solar penetration study. Then, it presents a novel method that is 
developed to predict and generate scenarios for uncertainties across 
different lead times. The scenarios can be used as inputs to the market 
simulation tool for stochastic simulation. The two parts together may 
lead to multi-scenario stochastic unit commitment in the future. In the 
near term, the stochastic market simulation can help to validate market 
design and operational procedures. The uncertainty predication and 
scenario generation may help operational situational awareness and 
better define reserve requirements and operational margins.
Pumped Storage Optimization in Real-Time Markets Under Uncertainty
Bing Huang, Research Engineer, Midcontinent ISO (Carmel, IN)
Arezou Ghesmati, R&D Scientist, Midcontinent ISO (Carmel, IN)
Yonghong Chen, Consulting Advisor, Midcontinent ISO (Carmel, IN)
Ross Baldick, Emeritus Professor, University of Texas at Austin 
(Austin, TX)

    Pumped storage hydro units (PSHU) can provide flexibility to power 
systems and may especially be valuable with increasing shares of 
intermittent renewable resources. However, the scheduling of PSHUs, 
particularly in the real-time market, has not been thoroughly studied. 
To enhance the use of PSH resources and leverage their flexibility, it 
is important to incorporate the uncertainties to properly address the 
risks in the real-time market operation. In this work, first a 
deterministic PSHU model that incorporates the state of charge in the 
Day-ahead market optimization is introduced. Second, two pumped storage 
hydro (PSH) models that use probabilistic price forecasts are proposed 
for Look-ahead commitment (LAC) in the real-time market operation. A 
risk neutral stochastic PSH model and a risk averse robust optimization 
PSH model are developed using the probabilistic price forecasts to 
capture the real-time market uncertainties. Numerical studies in Mid-
continent Independent System Operator (MISO) system demonstrate that 
the proposed models improve market efficiency and reduce PSH real time 
risk compared to the current approach. Probabilistic forecast for Real 
Time Locational Marginal Price (RT-LMP) is created and embedded into 
the proposed stochastic and robust optimization models, a statistically 
robust approach is used to generate scenarios for reflecting the 
temporal inter-dependence of the LMP forecast uncertainties.
Forecasting Aggregate Electricity Demand on a 5-Minute Basis Using 
Machine Learning
Dr. Yinghua Wu, Senior Lead Data Scientist, PJM Interconnection 
(Audubon, PA)
Laura Walter, Senior Lead Data Scientist, PJM Interconnection (Audubon, 
PA)
Dr. Anthony Giacomoni, Manager--Advanced Analytics, PJM Interconnection 
(Audubon, PA)

    PJM currently has two load forecasts used in dispatch and real-time 
operations. These forecasts are comprised of the short-term forecast, 
which is the forecasted hourly average load for the next seven days, 
and the very short-term load forecast, which is the forecasted 5-minute 
load averages for the next six hours. The very short-term load forecast 
is constantly fed into the real-time dispatch software for optimal 
power flow calculations and real-time market pricing. It is of crucial 
importance that these forecasts closely match the actual load in the 
near future to maintain system frequency and voltage. If not, 
dispatchers must take action to quickly intervene and adjust the load 
up or down. The load profiles generally follow temporal patterns, but 
are also driven by weather and other usage patterns. Given the recent 
rapid growth of machine learning technologies, this presentation will 
survey a collection of some of the most representative and innovative 
methods that are suitable to time series predictions such as load 
forecasting, e.g., gradient boosting, recurrent neural network, causal 
convolution, etc. We will also revisit some traditional methods such as 
generalized linear models and automatic regressive moving average 
(ARMA) methods to explore whether they can capture the load shape in 
short horizons. We will survey and analyze these new technologies for 
their power of prediction to see if these methods provide the potential 
to improve on current forecasting practices.
Long-Term Outlook for the ERCOT Grid
Pengwei Du, Supervisor--Economic Analysis & Long Term Planning Studies, 
The Electric Reliability Council of Texas (Austin, Texas)

    The bulk transmission network within ERCOT consists of the 60-
kilovolt (kV) and higher transmission lines and associated equipment. 
ERCOT conducts a forward-looking study to understand long-term 
reliability and economics need to ensure continued system reliability 
and efficiency. This talk will present the key challenges and findings 
from the most recent long-term system assessment planning study, which 
accounts for the inherent uncertainty of planning the system in the 10- 
to 15-year planning horizon.

Day 2--Wednesday, June 28

Session W-A1 (Wednesday, June 28, 9:00 a.m.) (Commission Meeting Room)

Uncertainty-Informed Renewable Energy Scheduling: A Scalable Bilevel 
Framework
Dr. Dongwei Zhao, Postdoctoral Associate, Massachusetts Institute of 
Technology (Cambridge, MA)
Dr. Vladimir Dvorkin, Postdoctoral Fellow, Massachusetts Institute of 
Technology (Cambridge, MA)
Dr. Stefanos Delikaraoglou, Data Scientist, Axpo Solutions AG (Zurich, 
Switzerland)
Dr. Alberto J. Lamadrid L., Associate Professor, Lehigh University 
(Bethlehem, PA)
Dr. Audun Botterud, Principal Research Scientist, Massachusetts 
Institute of Technology (Cambridge, MA)

    The fast-growing variable renewable energy sources (VRES) in 
electricity markets are creating challenges to uncertainty management. 
This work addresses these challenges by adopting an uncertainty-
informed adjustment toward VRES bidding quantities in the day-ahead 
market and minimizing expected system costs under the sequential 
market-clearing structure. However, implementing this mechanism 
requires solving a bilevel optimization problem, which is 
computationally difficult for practical large-scale systems. To 
overcome this challenge, we propose a novel technique based on strong 
duality and McCormick envelopes. This approach relaxes the original 
problem to a linear program, enabling efficient computation for large-
scale systems. We conduct case studies on the 1576-bus NYISO systems 
and compare our bilevel VRES-adjustment model with the myopic strategy 
where VRES producers bid the forecast value in the day-ahead market. 
The results

[[Page 40242]]

demonstrate that under a future high VRES penetration level (e.g., 
40%), our bilevel framework can significantly reduce the expected 
system cost and the volatility of the market prices, participants' 
revenues, and real-time re-dispatch adjustments, by efficiently 
optimizing VRES quantities in the day-ahead market. Furthermore, we 
found that increasing transmission ability may incur a much higher 
system cost under the myopic strategy while a lower cost under the 
bilevel model) because of the lack of flexible generators or reserves 
in real time to deal with uncertainty.
Enhancing Power System Resilience and Efficiency Through Proactive 
Security Assessments and the Use of powerSAS.m: A Robust, Efficient, 
and Scalable Security Analysis Tool for Large-Scale Systems
Dr. Yang Liu, Postdoctoral Appointee, Argonne National Laboratory 
(Lemont, IL)
Dr. Feng Qiu, Principal Computational Scientist, Argonne National 
Laboratory (Lemont, IL)
Dr. Jianzhe Liu, Energy Systems Scientist, Argonne National Laboratory 
(Lemont, IL)

    Power system security assessment is directly related to increasing 
real-time and day-ahead market and planning efficiency because it helps 
ensure the reliable and secure operation of the power system, which is 
essential for efficient market and planning activities. Without proper 
security assessments, the power system is vulnerable to a variety of 
threats, including cyber attacks, natural disasters, and equipment 
failures, which can disrupt the operation of the system and lead to 
market inefficiencies and planning uncertainties. By performing 
security assessments and identifying potential vulnerabilities, system 
operators can take proactive measures to mitigate risks and improve the 
reliability and efficiency of the power system, which, in turn, 
supports the goals of real-time and day-ahead market and planning 
efficiency. Additionally, advanced software tools and models can be 
used to support security assessments, enabling operators to better 
anticipate and respond to potential security threats and further 
improve the efficiency and reliability of the power system. Existing 
tools (commercial or open-source) work fine for routine security 
analysis under normal operating conditions. However, in resilience 
analysis, which studies the system security and reliability under 
stressed scenarios, existing tools often experience various numerical 
issues, significantly impacting operators' assessment of system 
resilience. A recent example is the non-convergence issues with PSS/E, 
one of the best commercial power system analysis tools used in the DOE 
Puerto Rico resilience project led by Argonne. The numerical issues 
forced the team to give up more advanced analysis. A robust and 
efficient security analysis tool is imperative for resilience study in 
large-scale systems. In this talk, we will introduce a recently 
released open-source power system security analysis tool called 
powerSAS.m. The powerSAS.m is a robust, efficient, and scalable power 
grid analysis framework based on semi-analytical solutions (SAS) 
technology. The talk will cover the following two critical aspects and 
discuss how they are directly related to increasing real-time and day-
ahead market and planning efficiency. First, we will introduce the 
fundamentals of the SAS technology and the major functionalities of the 
powerSAS.m, including (1) Steady-state analysis, including power flow, 
continuation power flow, and contingency analysis. (2) Dynamic security 
analysis, including voltage stability analysis, transient stability 
analysis, and flexible user-defined simulation. (3) Hybrid extended-
term simulation provides adaptive quasi-steady-state-dynamic hybrid 
simulation in extended term with high accuracy and efficiency. We will 
also introduce some ongoing functionalities, including the SAS-based 
electromagnetic transient (EMT) simulation and multi-scale simulations. 
Second, we will present some use cases to demonstrate the key features 
and performance of the SAS technology and powerSAS.m tool, including: 
(1) High numerical robustness. Backed by the SAS approach, the PowerSAS 
tool provides much better convergence than the tools using traditional 
Newton-type algebraic equation solvers when solving algebraic 
equations/ordinary differential equations/differential-algebraic 
equations. (2) Enhanced computational efficiency and scalability. Due 
to the analytical nature, PowerSAS provides model-adaptive high-
accuracy approximation, which brings significantly extended effective 
range and much larger steps for steady-state/dynamic analysis. PowerSAS 
has been used to solve large-scale system cases with 200,000+ buses.
Stochastic Unit Commitment and Market Clearing in Julia With 
UnitCommitment.jl
Dr. Alinson Santos Xavier, Computational Scientist, Argonne National 
Laboratory (Lemont, IL)
Og[uuml]n Yurdakul, Ph.D. Candidate, Technische Universit[auml]t Berlin 
(Berlin, Germany)
Dr. Aleksandr M. Kazachkov, Assistant Professor, University of Florida 
(Gainesville, FL)
Jun He, Professor, Purdue University (West Lafayette, IN)
Dr. Feng Qiu, Principal Computational Scientist, Argonne National 
Laboratory (Lemont, IL)

    UnitCommitment.jl (UC.jl) is a comprehensive open-source 
optimization package for the Security-Constrained Unit Commitment 
Problem (SCUC), providing an extensible and fully-documented data 
format for the problem, Julia/JuMP implementations of state-of-the-art 
mathematical formulations and solution methods, as well as a diverse 
collection of realistic and large-scale benchmark instances. This talk 
focuses on two major features recently introduced to the package. 
Firstly, the package now supports modeling and optimizing two-stage 
stochastic versions of the problem, in addition to the deterministic 
SCUC. Compared to existing implementations, UC.jl allows a broader set 
of network parameters to be treated as uncertain, including not only 
demands and generation limits, but also production costs, network 
topology, transmission limits, among others. Benchmark scripts are 
provided to accurately evaluate the performance of different stochastic 
solution methods. Secondly, the package now includes various 
functionalities for market clearing, such as the computation of 
generator payments and locational marginal prices (LMPs) using 
different methods proposed in the literature. In this talk, we will 
discuss the usage of these new features, technical challenges 
associated with them, and the potential simulations or studies that 
they enable.
Reduced-Order Decomposition and Coordination Approach for Markov-Based 
Stochastic UC With High Penetration Level of Wind and BESS
Niranjan Raghunathan, Ph.D. Student, University of Connecticut (Storrs, 
CT)
Dr. Peter B. Luh, Professor, University of Connecticut and National 
Taiwan University (Alexandria, VA)
Dr. Zongjie Wang, Professor, University of Connecticut (Storrs, CT)
Dr. Mikhail A. Bragin, Professor, University of California, Riverside 
(Riverside, CA)
Dr. Bing Yan, Professor, Rochester Institute of Technology (Rochester, 
NY)
Dr. Meng Yue, Research Staff Electrical Engineer, Brookhaven National 
Laboratories (Upton, NY)

[[Page 40243]]

Dr. Tianqiao Zhao, Renewable Energy Group, Brookhaven National 
Laboratories (Upton, NY)

    With the growing need to achieve carbon neutrality, integrating 
renewable energy (e.g., wind and solar) and battery energy storage 
systems (BESSs) into the grid is an urgent and challenging enterprise. 
At the day-ahead stage, unit commitment (UC) decisions need to account 
for uncertainties of geographically distributed renewable generation. 
BESS integration can help mitigate intermittence and reduce curtailment 
by storing energy during high renewable generation periods and 
releasing energy when needed, thus improving the cost efficiency of 
grid operation. Therefore, ensuring economic and reliable grid 
operations with the significant rise in renewable energy penetration 
necessitates the consideration of spatially distributed uncertainties 
and BESS in UC. To achieve this, a risk-neutral approach (i.e., 
scenario-based stochastic UC and Markov-based stochastic UC) is 
preferred over risk-averse approaches (e.g., robust optimization and 
interval optimization), as the latter yields overly conservative 
solutions. Between the risk-neutral approaches, Markov-based approaches 
have two advantages over scenario-based approaches: (1) Due to the 
Markov property, where stochastic information at the next time step 
depends only on the information at the current time step, the 
uncertainty can be compactly modeled by wind generation states at each 
time step and state transitions between subsequent time steps. 
Consequently, the overall number of possible states and transitions in 
the Markov model increases linearly with the number of intervals in the 
optimization horizon, whereas the number of possible scenarios 
increases exponentially. (2) Reduced Markov models preserve the 
volatility of wind generation, the underlying spatio-temporal 
correlation structure, and low-probability, high-impact events more 
effectively in uncertainty sets compared to scenarios. Therefore, the 
problem is formulated as Markov-based stochastic UC. With distributed 
wind, however, the number of possible wind states grows exponentially 
with the number of wind farms in different locations considered, posing 
major computational difficulties. To reduce complexity, an innovative 
decomposition and coordination framework is developed, where 
approximate area subproblems are formulated by utilizing area-
perspective, reduced-order Markov models. In these models, the 
variability of local (in-area) windfarms is emphasized while that of 
nonlocal (out-of-area) windfarms is approximated by using Principal 
Component Analysis (PCA) to reduce dimensionality while preserving the 
maximum amount of variation. This is a reasonable approximation because 
variations at the local level have more impact on the behavior of local 
units and power flow through local transmission lines compared to 
variations at distant locations. The objective of an approximate area 
subproblem is to optimize in-area resources based on its area-
perspective Markov model. The approximate area subproblems are solved 
iteratively while their solutions are coordinated using Surrogate 
Absolute-Value Lagrangian Relaxation (SAVLR), a state-of-the-art dual 
method with faster convergence than traditional Lagrangian Relaxation 
(LR)-based methods. To improve performance, an online filtering method 
for removing redundant transmission capacity constraints at each 
iteration is implemented in parallel by utilizing multiple cores. The 
solutions are validated using Monte Carlo simulations. Testing results 
based on the 118-bus system with 5 distributed wind farms show the 
effectiveness of the method in finding low-cost and robust UC solutions 
in a timely manner for multiple cases with different volatilities of 
wind generation and simulated extreme weather events. Analysis of the 
operation of BESSs shows that they absorb excess energy during high 
wind periods and release the energy during low wind periods, thus 
reducing wind curtailment and overall costs.
Learn To Branch and Dive for Large-Scale Unit Commitment Problem
Jingtao Qin, Research Assistant, University of California, Riverside 
(Riverside, CA)
Nanpeng Yu, Associate Professor, University of California, Riverside 
(Riverside, CA)
Mikhail Bragin, Assistant Research Professor, University of Connecticut 
(Storrs, CT)

    Unit commitment (UC) problems are typically formulated as mixed-
integer program (MIP) and solved by the branch-and-bound (B\&B) 
paradigm. The recent advances in graph neural network (GNN) motivate 
the application of GNN in learning to dive and branch for B\&B 
algorithm in modern MIP solvers. Existing GNN models are mostly 
constructed from B\&B trees, which are computationally expensive when 
dealing with large-scale UC problems. In this paper, we propose a 
physical network information-based hierarchical graph convolution model 
for neural diving that leverages the underlying features of various 
components of power systems to find high-quality variable assignments. 
Furthermore, we adopt the B\&B tree-based graph convolution model for 
neural branching to select the optimal variables for branching at each 
node of the B\&B tree. Finally, we integrate neural diving and neural 
branching into a modern MIP solver to establish a novel neural MIP 
solver that is specially designed for large-scale UC problems. Numeral 
studies show that our proposed model has better performance and 
scalability than the baseline B\&B tree-based model on neural diving. 
Moreover, the neural MIP solver yields the lowest MIPGap for all 
testing days after combining it with our proposed neural diving model 
and baseline neural branching model.

Session W-B1 (Wednesday, June 28, 9:00 a.m.) (Hearing Room One)

Stochastic Nodal Adequacy Pricing Platform (SNAP)
Dr Richard D. Tabors, Partner and President, Tabors Caramanis Rudkevich 
(Newton, MA)
Dr. Aleksandr Rudkevich, President, Newton Energy Group (Newton, MA)
Russel Philbrick, President, Polaris Systems Optimization (Seattle, WA)
Dr. Selin Yanikara, Analyst, Newton Energy Group (Newton, MA)

    The Stochastic Nodal Adequacy Pricing Platform (SNAP) software 
system provides an implemented methodology to calculate the probability 
and value of RESOURCE INADEQUACY of electricity supply on an hourly 
basis for a period of one to five days ahead of real time. The 
stochasticity of SNAP is driven by the stochastic weather forecasts 
available and provided by IBM The Weather Company on a i5 day forward 
basis for a 4x4km grid worldwide (SNAP uses at most 5). Forecasts are 
developed from 76 different numerical weather prediction models (and 
their ensemble members) as inputs to their forecast system. Bayesian 
model averaging is used to correct for systematic errors (bias). 
Results are rearranged to create 100 synthetic weather system scenarios 
through the use of Ensemble Copula Quantile-Coupling technique. The 
result is a probabilistic forecast within which each of the scenarios 
is equally likely. As the electric supply system moves toward greater 
dependence on renewable sources both in front of and behind the meter 
and as weather conditions are evolving, the stochastics of weather have 
become a, if not the

[[Page 40244]]

driving force in forecasting power system adequacy. SNAP is developed 
as an information/assist tool for operational planning at the utility 
system level. SNAP has been developed with funding from the Department 
of Energy's ARPA-E PERFORM program. SNAP uses the individual components 
of the weather forecast scenarios to create 100 probabilistic scenarios 
of the output of individual wind and solar locations as well 
forecasting of demand incorporating behind the meter generation. Based 
on the probability of renewable supply, demand, and the probability of 
outage of traditional supply sources and transmission, SNAP runs 
100,000 Monte Carlo SCED/SCUC runs of the commercially available cloud-
based ENELYTIX software system to identify the existence of resource 
inadequacy, the nodal location of that inadequacy, its cause and 
potential solutions. The objective is to present the structure of the 
computational and analytic processes that allow for running and 
evaluation of 100,000 scenarios for each individual forecast hour. The 
presentation will discuss the cloud-based structure the allows the 
analysis to be completed in under 50 minutes using 500 virtual machines 
at a costs of $120 at spot rates.
Assessing Nodal Adequacy of Large Power Systems
Dr. F. Selin Yanikara, Energy Analyst, Newton Energy Group (Newton, MA)
Russ Philbrick, President, Polaris Systems Optimization (Seattle, WA)
Aleksandr M. Rudkevich, President, Newton Energy Group (Newton, MA)
Sophie Edelman, Electricity Research Analyst, The Brattle Group (New 
York, New York)

    Extreme weather events, increasing electrification, and integration 
of wind and solar power pose significant challenges for reliable 
operation of the power grid. Quantitative evaluation of these impacts 
is critical for making efficient policy and investment decisions and in 
designing markets and engineering controls. This presentation will 
summarize the theoretical foundation for nodal probabilistic assessment 
of resource adequacy and its applications to modern electrical systems 
with a significant penetration of weather dependent variable energy 
resources and storage technologies. In addition, this presentation will 
address the need for, and will present, new adequacy metrics that 
reflect an economically justified contribution of each system asset--
generation, transmission, or demand resource to system adequacy. The 
analysis relies on the Monte Carlo based methodology using new 
computationally efficient and statistically accurate methods. We 
illustrate the numerical results and computational performance of our 
approach using the ENELYTIX[supreg] powered by PSO SaaS and our 
standard dataset for the ERCOT market.
Comparison of Flexibility Reserve and ORDC for Increasing System 
Flexibility
Phillip de Mello, Senior Technical Leader, Electric Power Research 
Institute (Palo Alto, CA)
Erik Ela, Program Manager, Electric Power Research Institute (Boulder, 
CO)
Nikita Singhal, Technical Leader, Electric Power Research Institute 
(Palo Alto, CA)
Alexandre Moreira da Silva, Research Scientist, Lawrence Berkeley 
National Laboratory (Berkeley, CA)
Miguel Heleno, Research Scientist/Engineer, Lawrence Berkeley National 
Laboratory (Berkeley, CA)

    Power system composition changes are making flexible resources more 
important to balance the increasing variability and uncertainty. System 
operators often look to increase the amount of flexibility available to 
give real time operations greater control. Two common methods for 
increasing flexibility are to create new reserve products that are 
targeted towards flexibility and ramping capability or using an 
extended operating reserve demand curve (ORDC) to procure more of an 
existing reserve when the additional value exceeds costs. Detailed 
operation simulations to mimic day ahead and real time markets were 
conducted to compare flexibility reserves and ORDCs. Benefits to 
reliability were measured by a reduction in shortages of reserves and 
energy experienced across the system. The extra reserves generally 
increased the costs of running the system, but it was lower than the 
penalty prices of the shortages relieved. Some periods showed a 
reduction of system costs with added reserves, suggesting that more 
efficient designs of reserves could not only increase system 
reliability but also reduce costs. Both methods increase the 
flexibility on the system, but function differently in typical 
deployments in current ISO/RTO practice. The different parameters 
defining each technique was explored to understand how their 
differences manifest in improving reliability. Most differences reflect 
the tradeoff between flexibility in designing a new product versus ease 
of implementation of procuring more of an existing product. The key 
difference of the techniques results due to the sharing of generator 
ramp rates between different reserve products. Most existing 
implementations require dedicated capacity for each reserve product but 
often do not require dedicated ramp capability. Using a new flexibility 
reserve that can share ramp rates will typically be able to schedule 
more reserve for a certain available generator capacity than applying 
an ORDC to an existing product. This impacts the cost and effectiveness 
of those reserves particularly in periods of system stress. Toggling 
the ramp sharing constraint can be used to make either implementation 
perform similarly as the other.
ABSCORES, A Novel Application of Banking Scoring and Rating for 
Electricity Systems
Alberto J. Lamadrid L., Associate Professor, Lehigh University 
(Bethlehem, PA)
Audun Botterud, Principal Research Scientist, Massachusetts Institute 
of Technology (Cambridge, MA)
Jhi-Young Joo, Research Scientist, Lawrence Livermore National 
Laboratory (Livermore, CA)
Shijia Zhao, Energy Systems Scientist, Argonne National Laboratory 
(Lemont, IL)

    This presentation discusses the basis for the establishment of an 
Electric Assets Risk Bureau. We are developing different scores 
customized according to the application required. We study the use of 
financial models to determine the risk associated to individual assets 
in the system. We present a model focused on managing operational risk, 
and outline the methodology for risk metrics applied to high impact, 
low probability (HILP) events. We distinguish between, first, public 
risk, related to the physical provision of supporting services required 
for the stability of the electricity system (i.e., ancillary services); 
and second, financial risk, derived from positions taken by 
participants with pecuniary repercussions. A key paradigm of our 
framework is a focus on implementability of the approach (under 
existing regulatory structures) and a method for dispute resolution 
given potential decisions taken with the metrics proposed.
Recent Developments in the Day-Ahead and Real-Time Electricity Market 
Design and Software Caused by the Higher Energy Costs and Emerging 
Technologies--European Experience
Petr Svoboda, Engineer, Unicorn Systems a.s. (Prague, Czech Republic)

    Europe has been dealing with the imbalance of production and

[[Page 40245]]

consumption for years. This has led to the development of the single 
de-regulated electricity market to solve the barriers between the 
individual states and provide the most cost-effective way to ensure 
secure, sustainable, and affordable energy supply to the customers. 
Recent changes in the market caused by the increase of the energy costs 
and emergence of the new technologies have caused the fundamental 
shifts in the market design and software enabling its operations. In 
our presentation we would like to discuss the latest developments in 
the areas of: 1. New algorithms of transmission capacity calculation 
that have proven to increase the efficiency of capacity usage and 
relevant economic welfare. 2. Development of the HVDC interconnectors 
and their impact on the market efficiency and transmission costs. 3. 
15-minute day-ahead markets. 4. Emergence of the integrated real-time 
markets, new reserve products and multi-interval market clearing. 5. 
Introduction of the flexibility instruments to the energy markets. 6. 
Successful implementation of the hourly renewable certificates as the 
next step towards clean energy transition.

Session W-A2 (Wednesday, June 28, 12:30 p.m.) (Commission Meeting Room)

System Resilience Through Electricity System Restoration and Related 
Services
Douglas Wilson, Principal Analytics Engineer, GE (Edinburgh, United 
Kingdom)
James Yu, Head of Future Networks, ScottishPower Energy Networks 
(Glasgow, United Kingdom)
Ian Macpherson, Senior Innovation Manager, ScottishPower Energy 
Networks (Glasgow, United Kingdom)
Marta Laterza, Power Systems Engineer, General Electric (Glasgow, 
United Kingdom)
Marcos Santos, Senior Power Systems Engineer, General Electric 
(Glasgow, United Kingdom)
Richard Davey, Senior Project Manager, General Electric (Glasgow, 
United Kingdom)

    Electricity system restoration following a partial or system-wide 
outage is an essential service in the power system. There is a need to 
apply new resources based on renewable resources to replace the 
services that up to now have depended on fossil fuel generation. This 
presentation describes a project led by SP Energy Networks in 
collaboration with GE to demonstrate a co-ordinated restoration 
approach in the distribution grid using a novel control approach 
applied to a controlled zone with multiple resources. Live trials of 
the approach in the SP Energy Networks power system are presented, as 
well as results of testing the approach extensively in a hardware-in-
the-loop environment. The emerging weaknesses of the traditional 
methodology were recognised in UK electricity regulation, which was 
recently changed to include a requirement for 60% of customer load to 
be restored within 24 hours on a regional basis, with all supplies 
restored within 5 days (Electricity System Restoration Standard, 2021). 
Previous restoration requirements were less onerous on the timeframes 
and did not define geographic requirements. Since some regions now lack 
large transmission-connected blackstart-capable plant for the 
traditional top-down restoration approach, there is a need to harness 
the capabilities that renewable and distributed generation and storage 
can offer to address the deficit of system restoration capability. The 
new service being developed and trialled involves starting distributed 
generation and growing an island with customer load within the 
distribution network. This island can be sustained by automated control 
through managed load pickup as well unplanned disturbances with 
existing distributed energy resources, battery storage and demand 
response providing the control capability to keep the island in 
balance. The blackstart zone may then be reconnected to the 
transmission network if this is energised and can then contribute to 
managing the power balance as the restoration of the wider system 
continues. If appropriate, neighbouring islands can be connected 
together, and the resulting larger island is capable of greater block 
load pickup of active and reactive loads. One of the distinctive 
benefits of the approach taken is that it uses diverse resources of 
existing generation, storage and demand response capability that is 
present and operational in the network for other day-to-day purposes. 
These resources can be harnessed to provide the new electricity system 
restoration services with few additional power assets. Inherently, some 
devices can provide faster response than others, and large 
instantaneous power, and some may be able to sustain an energy supply 
while others have limited energy resource. Voltage support and short 
circuit current are also considerations. A diversity of renewable 
resources is useful to mitigate against individual resources being 
unavailable e.g. low wind or low solar conditions. A key requirement 
for the co-ordination of an electricity system restoration zone is a 
wide area monitoring and control system that manages the power 
balancing and switching of the network to automate the process of 
growing and sustaining the power island. The approach being trialled 
includes a SCADA/distribution management system with the topology 
information for network switching, together with a synchrophasor based 
wide area control system that manages the balancing, frequency control 
and resynchronization alignment of the network. Since the island is 
small in comparison to the normal interconnection, a rapid response to 
disturbances is required to maintain a stable frequency. Once a 
distribution zone is instrumented with the measurement, communication 
and control equipment to deliver the service, it is possible to use the 
same infrastructure to offer further services to manage grid stability 
in the more common circumstance of disturbances during grid-connected 
operation.

Coordinated Cross-Border Capacity Calculation Through The FARAO Open-
Source Toolbox

Violette Berge, Vice President, Artelys Canada (Montr[eacute]al, 
Canada)
Dr. Nicolas Omont, Vice President of Operations, Artelys (Paris, 
France)

    Cross-borders exchanges have taken a major role in European 
strategy to achieve climate goals. The European Commission set a target 
of 15% interconnections in 2030, meaning that each country should have 
the physical capability to export at least 15% of their production. 
Increasing exchanges makes short term planning more complex. In this 
context, the French TSO (RTE) released an open-source toolbox FARAO to 
perform Coordinated Capacity Calculation (CCC) and ensure the security 
of supply. Artelys is a consultancy expert in power systems 
optimization and carries out various projects around TSO operational 
coordination in Europe. FARAO performs the optimization of both 
preventive and curative remedial actions, including HVDC lines, phase-
shifter transformers and counter-trading but also topological actions. 
It is operationally used for the exchanges between Italy and its 
northern neighbors as well as between France, Spain and Portugal. 
Artelys will present the algorithms of the FARAO toolbox and how they 
are actually used to enable greater operational coordination amongst 
the countries.

[[Page 40246]]

Advanced Scenario Selection Methods for Probabilistic Transmission 
Planning Assessments
Dr. Eknath Vittal, Principal Technical Leader, EPRI (Palo Alto, CA)
Anish Gaikwad, Senior Program Manager, Electric Power Research 
Institute (Palo Alto, CA)
Parag Mitra, Senior Technical Leader, Electric Power Research Institute 
(Palo Alto, CA)

    Given the temporal and spatial characteristics of extreme weather 
events, developing transmission planning scenarios, i.e., snapshots of 
instantaneous operational conditions, is a challenging problem. It 
requires a multi-model assessment that links long-term planning models 
that capture the operational performance of the system (resource 
adequacy and production cost modeling) to the future meteorological 
projections that will inform the impacts of weather and extreme events. 
Scenario generation and analysis is computationally and labor 
intensive. Identifying snapshot conditions for future system states can 
be challenge. This presentation will highlight and detail an EPRI 
application that helps transmission planners identify critical power 
flow conditions from operational simulations such as production cost 
simulations. The EPRI High-Level Screening (HiLS) for Data Analytics 
tool allows planners to apply statistical analysis to large dataset 
that capture the operational performance of the system. The tool allows 
for the data to be organized into clusters of similar operating 
conditions reducing the dimensionality of the state space. As an 
example, an operational simulation of 8760 hours can be reduced to 10 
operating hours that capture 95% of the variability seen over the 
course of the year. As uncertainty and variability increase on both the 
generation and load, developing methods and processes to understand the 
conditions that present the most challenging reliability and stability 
conditions will be critical. The HiLS tools, provides transmission 
planners a platform that can help them organize and visualize data 
representing future operational conditions of the system that considers 
both load variability and generator availability.
Incorporating Climate Projections Into Grid Models: Bridging the Data 
Gap To Capture Weather Dependent Representative and Extreme Events and 
Corresponding Uncertainties
Dr. Zhi Zhou, Principal Computational Scientist, Argonne National 
Laboratory (Lemont, IL)
Dr. Neal Mann, Energy Systems Engineer, Argonne National Laboratory 
(Lemont, IL)
Yanwen Xu, Graduate Student, University of Illinois at Chicago, Urbana-
Champaign (Champaign, IL)
Zuguang Gao, Graduate Student, University of Chicago (Chicago, IL)
Dr. Akintomide Akinsanola, Assistant Professor, University of Illinois 
at Chicago (Chicago, IL)
Dr. Todd Levin, Team Lead, Argonne National Laboratory (Lemont, IL)
Dr. Jonghwan Kwon, Energy Systems Engineer, Argonne National Laboratory 
(Lemont, IL)
Dr. Audun Botterud, Senior Energy Systems Engineer, Argonne National 
Laboratory (Lemont, IL)

    It is crucial to consider high-fidelity weather data and climate 
projections in grid models in order to capture future weather trends, 
extremes, and uncertainties. However, traditional power system studies 
often overlook many of these considerations and rely solely on 
historical weather data. To address this challenge, we develop a 
computationally manageable framework to process high-quality 
representations of climate data for use with power system models. The 
framework includes a three-stage architecture to select representative 
regions and periods, and also identify periods of extreme weather 
conditions after translating climate data (temperature, wind-speed, 
etc.) into grid inputs (load, power generation profiles and outage 
probabilities). The framework also models and represents uncertainty of 
future weather events based on ensembles of climate model simulations. 
The outcome of the framework is a set of processed grid inputs in time 
series format that capture the impact of climate features on the 
system. This includes grid inputs directly converted from weather 
variables at the cell level, as well as those from representative 
regions and time periods, those representing the impact from extreme 
weather events, and their associated uncertainties. We apply this 
computational framework to translate downscaled climate projections 
generated by three different global climate models, encompassing over 
60 different weather variables at 12-km geographic and 3-hour temporal 
resolution for all North America. We then demonstrate how consideration 
of high-quality climate-driven grid inputs in electricity system models 
impacts optimal long-term planning decisions. Capturing future weather 
conditions and associated uncertainties is becoming important as power 
systems, and their associated markets, are being impacted by both 
efforts to decarbonize the effects of a changing climate. These are 
also important considerations when updating market designs to maintain 
reliability and economic efficiency as the underlying power system 
evolves. In addition, capturing weather uncertainty is critical for 
risk-aware decision making. Therefore, this work provides a valuable 
resource for power system modelers by bridging the gap between climate 
models and grid models to help ensure that long-term system planning 
decisions are informed by the impacts of future climate conditions.

Session W-B2 (Wednesday, June 28, 12:30 p.m.) (Hearing Room One)

Enhancing Decision Support for Electricity Markets With Machine 
Learning
Yury Dvorkin, Faculty, Johns Hopkins University (Baltimore, MD)
Robert Ferrando, Graduate Assistant, University of Arizona (Tucson, AZ)
Laurent Pagnier, Assistant Professor, University of Arizona (Tucson, 
AZ)
Zhirui Liang, Ph.D. Student, Johns Hopkins University (Baltimore, MD)
Daniel Bienstock, Professor, Columbia University (New York, NY)
Michael Chertkov, Professor, University of Arizona (Tucson, AZ)

    This presentation describes how machine learning can be leveraged 
to enhance computational speed of day-ahead and real-time unit 
commitment and optimal power flow routines, which are at the core of 
market-clearing procedures in US ISOs. Our machine learning 
architecture embeds both power flow physics and market design 
properties (e.g., cost recovery and revenue adequacy) into the training 
stage, which increases accuracy of computations and preserves a 
relationship between primal (dispatch) and dual (prices) variables. The 
accuracy and scalability of the proposed method is tested on a 
realistic 1814-bus NYISO system with current and future renewable 
energy penetration levels. We also demonstrate ~100x gain in 
computations relative to traditional optimization approaches.
Synergistic Integration of Machine Learning and Mathematical 
Optimization for Sub-Hourly Unit Commitment
Jianghua Wu, Ph.D. Candidate, University of Connecticut (Vernon, CT)
Dr. Zongjie Wang, Assistant Professor, University of Connecticut 
(Storrs, CT)
Dr. Yonghong Chen, Consulting Advisor, Midcontinent ISO (Carmel, IN)

[[Page 40247]]

Dr. Bing Yan, Assistant Professor, Rochester Institute of Technology 
(Rochester, NY)
Dr. Mikhail Bragin, Assistant Project Scientist, University of 
California, Riverside (Riverside, CA)

    The integration of intermittent renewables into power systems 
presents significant challenges for operators due to increased 
uncertainties and greater intra-hour net load variability. Sub-hourly 
Unit Commitment (UC) has been suggested as a solution to quickly 
respond to changes in electricity supply and demand, which is more 
complicated than hourly UC because of a higher number of time periods, 
and higher dependencies among coupled periods. Traditional optimization 
methods could be time-consuming while machine learning (ML) may have 
additional feasibility concerns. To address these challenges, a hybrid 
approach based on synergistic integration of ML and optimization is 
developed. This novel approach adopts our recent decomposition and 
coordination Surrogate Absolute-Value Lagrangian Relaxation (SAVLR) 
method with efficient coordination and accelerated convergence. ML is 
then used to quickly predict SAVLR subproblem solutions. Compared to 
those of the original overall problem, subproblem solutions are much 
easier to learn. Nevertheless, predicting ``good'' subproblem solutions 
is still challenging because of the ``jumps'' of binary decisions and 
many types of unit-level constraints. To overcome these issues, a 
generic ML model, embedding recurrent neural networks (RNNs) and the 
Attention mechanism in the encoder-and-decoder structure, is developed. 
Because of the features of RNNs and Attention, this generic model can 
learn different subproblem solutions to reduce the training effort, and 
can provide time-based predictions to capture dependencies. In 
addition, to resolve the limitation of ML in handling constraints, a 
rule-based feasibility layer is incorporated in the predicting process, 
ensuring feasibility with respect to unit-level constraints. Testing on 
the IEEE 118-bus system demonstrates the effectiveness of our approach, 
providing feasible and accurate subproblem solutions quickly, and 
obtaining near-optimal overall solutions efficiently.
Boosting Power System Operation Economics Via Closed-Loop Predict-and-
Optimize
Dr. Lei Wu, Anson Wood Burchard Chair Professor, Stevens Institute of 
Technology (Hoboken, NJ)
Xianbang Chen, Ph.D. Candidate, Stevens Institute of Technology 
(Hoboken, NJ)

    By and large, power system operations are implemented by 
Independent System Operators (ISO) in an open-loop predict-then-
optimize (O-PO) process. First, the uncertainty realizations (e.g., 
renewable energy availability) are predicted as accurately as possible. 
Taking the predictions as inputs, day-ahead unit commitment and real-
time economic dispatch problems are then optimally resolved for 
determining the operation plan (i.e., optimization). The operation goal 
is to achieve the minimum system operation cost, i.e., the optimal 
operation economics. However, the operation economics could suffer from 
the open-loop process because its predictions may be myopic to the 
optimizations, i.e., the predictions seek to improve the immediate 
statistical prediction errors (i.e., accuracy-oriented) instead of the 
ultimate operation economics. To this end, we propose to improve 
operation economics by closing the open loop between the prediction and 
the optimization, i.e., a closed-loop predict-and-optimize (C-PO) idea. 
Specifically, two C-PO frameworks are designed, including a feature-
driven C-PO framework and a bilevel mixed-integer program (MIP) C-PO 
framework. Their core is to feed the induced operation cost back for 
training the predictor and measuring the prediction quality with the 
operation cost (i.e., cost-oriented). As a result, the prediction and 
the optimization can be implemented jointly in a single framework. 
Based on real-world data, the feature-driven C-PO is compared to the 
traditional O-PO, showing noticeable improvement in operation 
economics, although with slightly compromised prediction accuracy for 
certain cases. The experiments on a large-size system show that the C-
PO has potential in a real-world application. The bilevel MIP C-PO is 
more versatile than the feature-driven C-PO. Based on an IEEE 118-bus 
system, the bilevel MIP C-PO is compared to the state-of-the-art 
methods of handling uncertainties, i.e., stochastic programming and 
robust optimization. The case studies show that the bilevel MIP C-PO is 
economically competitive with the state-of-the-art methods but is more 
compatible with the current operational practice.
Privacy-Preserving Synthetic Dataset Generation for Power Systems 
Research
Dr. Vladimir Dvorkin, Postdoctoral Fellow, Massachusetts Institute of 
Technology (Cambridge, MA)
Dr. Audun Botterud, Principal Research Scientist, Massachusetts 
Institute of Technology (Cambridge, MA)

    Power systems research heavily relies on the availability of real-
world power system datasets (network parameters, time series, etc.). 
However, data owners, such as system operators, are often hesitant to 
share their data due to valid security and privacy concerns. To 
overcome these challenges, we have developed state-of-the-art 
algorithms that enable the synthetic generation of optimization and 
machine learning datasets for the power systems industry. Our 
algorithms take real-world datasets as input and output their 
synthetic, perturbed versions that maintain the accuracy of the 
original data on specific problem classes, such as power system 
dispatch and wind power forecasting. Importantly, the original data 
remains undisclosed, effectively controlling the privacy risk in data 
releases. To ensure privacy preservation, we employ rigorous 
perturbation techniques of differential privacy that strictly control 
the amount of privacy loss. Furthermore, we preserve the accuracy of 
original data through post-processing convex optimization. Our 
algorithms have many applications, including synthetic generation of 
transmission parameters and renewable generation records. We have open-
sourced our algorithms to encourage their use by interested parties. 
For more information, please visit our GitHub repository at https://github.com/wdvorkin/SyntheticData.

Session W-A3 (Wednesday, June 28, 3:30 p.m.) (Commission Meeting Room)

Parallel Interior-Point Solver for Security Constrained ACOPF Problems 
on SIMD/GPU Architectures
Dr. Mihai Anitescu, Senior Mathematician, Argonne National Laboratory 
(Lemont, IL)
Fran[ccedil]ois Pacaud, Assistant Professor, Ecole des Mines (Paris, 
France)
Michel Schanen, Computer Scientist, Argonne National Laboratory 
(Lemont, IL)
Sungho Shin, Postdoctoral Scientist, Argonne National Laboratory 
(Lemont, IL)
Daniel Adrian Maldonado, Assistant Energy Systems Scientist, Argonne 
National Laboratory (Lemont, IL)

    We investigate how to port the standard interior-point method for 
security constrained ACOPF problems, which are block-structured 
nonlinear programs with state equations, on SIMD/GPU architectures. 
Computationally, we decompose the interior-point algorithm into two

[[Page 40248]]

successive operations: the evaluation of the derivatives and the 
solution of the associated Karush-Kuhn-Tucker (KKT) linear system. Our 
method accelerates both operations using two levels of parallelism. 
First, we distribute the computations on multiple processes using 
coarse parallelism. Second, each process uses a SIMD/GPU accelerator 
locally to accelerate the operations using fine-grained parallelism. 
The KKT system is reduced by eliminating the inequalities and the state 
variables from the corresponding equations, to a dense matrix encoding 
the sensitivities of the problem's degrees of freedom, drastically 
minimizing the memory exchange. Our experiments on SIMD/GPU with 
security-constrained AC optimal power flow problem show that the method 
can achieve a 50x speed-up compared to the state-of-the-art method.
The Need for More Rigorous Calculation of Shadow Prices and LMPs
Dr. Xiaoming Feng, Research Fellow, Hitachi Energy (Raleigh, NC)

    LMPs (locational Marginal Prices) are used in nodal electricity 
markets to determine payments or charges to market participants. Due to 
the great monetary impact, it is imperative LMP is defined rigorously 
and calculated consistently. It has been observed the current method of 
shadow price and LMP calculation could produce values that are non-
unique under certain conditions, which might signal non-economic 
incentives to the market. We start with formal definitions for shadow 
price and LMP and present the properties of the perturbation functions 
and their computational consequences. We use simple examples to 
illustrate the discrepancy between theoretical shadow price and the 
shadow price calculated by state-of-the-art optimization solvers. From 
the discussion, we make the case for more rigorous calculation of both 
shadow prices and LMPs.
Real-Time Market Enhancements for Reliability and Efficiency
Dr. Mort Webster, Professor of Energy Engineering, Pennsylvania State 
University (University Park, PA)
Dr. Anthony Giacomoni, Manager, Advanced Analytics, PJM Interconnection 
(Audubon, PA)
Aravind Retna Kumar, Ph.D. Candidate, Pennsylvania State University 
(University Park, PA)
Sushant Varghese, Ph.D. Candidate, Pennsylvania State University 
(University Park, PA)
Shailesh Wasti, Ph.D. Candidate, Pennsylvania State University 
(University Park, PA)

    The projected trends in the U.S. power system, increasing wind and 
solar generation and retiring fossil fuel generation, will increase the 
net load variability and forecast uncertainty over the next several 
decades. There has been considerable research focusing on how to 
provide more flexibility to the power system. Within this line of 
research, numerous market design proposals have been explored: multi-
interval dispatch, ramp products, stochastic market clearing, an 
increase in flexible resources (virtual power plants (VPP), energy 
storage). Although flexibility is often cited as an objective the 
outcomes of concern are reliability (unserved demand and reserve 
shortages), efficiency (reducing bid production cost and uplift 
payments), curtailment of renewable generation, and incentives for 
future flexible resources (i.e., price formation). In the U.S., 
Independent System Operator (ISO) and Regional Transmission 
Organization (RTO) real-time market clearing and operations have the 
following properties: they operate on a rolling horizon basis 
throughout the operating day, face changing forecasts throughout the 
day with forecast errors, and frequently solve a real-time unit 
commitment (RUC), which is separate from the real-time dispatch. In 
contrast, most of the analysis and academic literature on market design 
enhancements neglect one or more of these characteristics in their 
analysis framework. The separation of commitment from dispatch raises 
the question: which market enhancement in which clearing engine? In 
this work, we present a simulation framework for the PJM wholesale 
energy markets with a rolling horizon and forecast errors. 
Specifically, we simulate the solution of the day-ahead market, 
followed by PJM's Intermediate-Term Security Constrained Economic 
Dispatch (IT-SCED) (real-time commitment process) every 15 minutes and 
PJM's Real-Time Security Constrained Economic Dispatch (RT-SCED) (real-
time dispatch) every 5 minutes throughout the operating day. Net load 
forecasts change every 5 minutes. We use this framework to simulate 
several of the commonly discussed market enhancements applied to either 
IT-SCED, RT-SCED, or both. We consider multi-interval dispatch, ramp 
products, and stochastic market clearing. Our results demonstrate that 
market design changes are most successful if they addresses both 
commitment (bringing enough capacity and operating range online) and 
dispatch (using the online operating range effectively).
Economics of Grid-Supported Electric Power Markets: A Fundamental 
Reconsideration
Dr. Leigh Tesfatsion, Research Professor of Economics, Courtesy 
Research Professor of Electrical & Computer Engineering, Iowa State 
University (Ames, IA)

    U.S. RTO/ISO-managed wholesale power markets operating over high-
voltage AC transmission grids are transitioning from heavy reliance on 
fossil-fuel based power to greater reliance on renewable power. This 
presentation highlights four conceptually-problematic economic 
presumptions reflected in the legacy core design of these markets that 
are hindering this transition. The key problematic presumption is the 
static conceptualization of the basic transacted product as grid-
delivered energy (MWh) competitively priced at designated grid delivery 
locations during successive operating periods, supported by ancillary 
services. The presentation then discusses an alternative conceptually-
consistent ``Linked Swing-Contract Market Design'' that appears well-
suited for the scalable support of increasingly decarbonized grid 
operations with more active participation by demand-side resources. 
This alternative design entails a fundamental switch to a dynamic 
insurance focus on advance reserve procurement permitting continual 
balancing of real-time net load. Reserve consists of the guaranteed 
availability of diverse power-path production capabilities for possible 
RTO/ISO dispatch during future operating periods, as protection against 
volumetric grid risk. Each reserve offer submitted by a dispatchable 
power resource m to a forward reserve market M(T) for a future 
operating period T is a two-part pricing swing-contract in firm or 
option form that permits m to ensure its revenue sufficiency.

Session W-B3 (Wednesday, June 28, 3:30 p.m.) (Hearing Room One)

Simulation of Wholesale Electricity Markets With Capacity Expansion and 
Production Cost Models To Understand Feedback Between Short Term Market 
Procedures and Long Term Investment Incentives
Dr. Jesse Holzer, Mathematician, Pacific Northwest National Laboratory 
(Richland, WA)
Dr. Abhishek Somani, Electrical Engineer, Pacific Northwest National 
Laboratory (Richland, WA)
Dr. Brent Eldridge, Electrical Engineer, Pacific Northwest National 
Laboratory (Bel Air, MD)

[[Page 40249]]

Diane Baldwin, Project Manager, Pacific Northwest National Laboratory 
(Richland, WA)

    Wholesale electricity markets are undergoing rapid changes, 
including variability and uncertainty and low prices from wind and 
solar, load flexibility and price responsiveness, distributed energy 
resources, energy storage, and revenue adequacy concerns. In response 
to these changes, enhancements to electricity market procedures have 
been proposed, including new reserve product, sloped reserve demand 
curves, multi-settlement forward markets, and stochastic modeling in 
market clearing optimization engines. These enhancements have the 
potential to improve operational outcomes in the short term time scale 
of hours to days by enabling better market responses to the changing 
market conditions. But they also affect the long run incentives for 
investment in grid equipment that ultimately result in the mix and 
capacity of various grid technologies. This mix in turn influences 
short term market conditions. We use linked models of capacity 
expansion and production cost to explore this feedback between short 
term and long term market conditions and to shed light on how this 
feedback affects the assessment of market enhancements to address 
changing market conditions.
Making the Right Resource Choice Requires Making the Right Model Choice
Dr. Rodney Kizito, Senior Manager, Ascend Analytics (Boulder, CO)
Gary W. Dorris, Ph.D., CEO, Ascend Analytics (Boulder, CO)
David Millar, Director of Consulting Services, Ascend Analytics 
(Boulder, CO)

    Production cost modeling simulates the operation of electric 
systems. It provides a lens into a highly uncertain future, allowing 
utilities to craft strategy and make critical decisions for their 
customers, shareholders, and stakeholders. The power and acuity of this 
lens will determine what resources will be deemed the most economic to 
provide a reliable, lower-carbon supply portfolio. Resource planning 
using production cost models that simulate the operation of power 
systems, once a straightforward exercise of deciding how many new power 
plants would be needed to meet future load growth, has become a much 
more complicated and challenging enterprise. The dramatic decline in 
the cost of renewables and storage technologies and the societal push 
for decarbonization means planners must model more complex and 
uncertain portfolio options. Renewables and their meteorologically 
determined fuel supply are creating new dynamics that highlight the 
need for more powerful modeling tools to capture the increasing 
variability in the power supply and the ensuing effect on market price 
volatility. This presentation highlights the benefits of using a new 
class of resource planning models to plan for a decarbonized future. 
Utilities, regulators, independent system operators, and other industry 
stakeholders rely heavily on modeling to support decision making for 
the allocation of scarce capital resources, as well as to ensure that 
the right resources are available to maintain a high level of 
reliability and resilience. This presentation argues that the older 
generations of models that remain widely in use today fail to capture 
the emerging dynamics of a power grid supplied primarily by renewable 
energy. For this reason, industry decision makers are unknowingly 
burdened by ``model-limited choice,'' which can lead to imprudent 
investments in assets liable to become functionally useless and 
ultimately disallowed. This presentation also provides a new 
terminology to classify a model's ability to capture the new market 
dynamics, high-definition production cost models (HD PCMs) versus 
traditional production cost models (PCMs). HD PCMs use simulation to 
capture the stochastic nature of load and electricity production 
generated by renew able energy sources, as well as to drill down to a 
5-minute level of temporal and spatial (i.e., nodal) granularity to 
capture the flexibility requirements of renewable integration. Further, 
HD PCMs mimic real-world uncertainty by simulating imperfect foresight 
of future system conditions between the day-ahead forecast and the 
real-time dispatch. Traditional PCMs are highly simplified because they 
were developed when computing power was a significant limitation. 
Today, resource planners can take advantage of the rapid increase in 
computing power provided by distributed computing to upgrade their 
analytical platforms to enable HD PCMs that provide more robust 
analysis.
Transmission Shortage Pricing By MW-Mile Based Demand Curve
Sina Gharebaghi, Graduate Research Assistant, Pennsylvania State 
University, Hitachi Energy (University Park, PA)
Dr. Xiaoming Feng, Research Fellow, Hitachi Energy (Raleigh, NC)

    ISOs use transmission demand curves (TDC) in security constrained 
unit commitment (SCUC) to relax transmission constraints when no 
feasible solution exists with hard transmission constraints. TDC is a 
penalty curve administratively specified as a function of the amount of 
MW violation of the transmission line's limits. Use of TDCs to ensure 
non-empty feasible solution space can result in excessively high LMP 
when multiple TDCs are active. Researchers have studied a transmission 
constraint screening approach to remove `redundant constraints' of 
serially connected transmission lines before the pricing run to avoid 
the accumulation of high shadow prices over multiple redundant 
constraints for LMP calculation. The screening approach alleviates to a 
large degree the occurrence of excessive LMP but has subtle and 
significant unintended consequences with respect to SCUC solution 
stability. We propose an alternative approach using MW-Mile based TDC 
to solve the transmission constraint violation problem and eliminate 
the root cause of excessive LMP without the need to remove redundant 
constraints. We discuss the economic justification of the MW-Mile based 
TDC approach and its advantage of solution stability with illustrative 
examples.
Grid OS--A Modern Software Portfolio for Grid Orchestration
Renan Giovanini, Ph.D., MBA, Transmission Product Marketing Director, 
General Electric (Edinburgh, United Kingdom)
Joseph Franz, Senior Marketing Manager, General Electric (Melbourne, 
FL)

    The 21st century has brought new challenges for Transmission and 
Distribution Operators that were hardly perceived in the turn of the 
century. There have been fast increases in bulk and micro renewable 
resources in conjunction with international agreements on 
CO2 emission targets. Severe droughts, and more frequent 
floods happening in the same country are driving needs also. An 
increasing number of changing weather patterns creating disruptions at 
several levels. Data tsunami has been created due to increasing types 
and number of sensors installed in the field. The grid itself was 
initially designed in the early 1900s based on a uni-directional flow 
requirement now is called to become bi-directional. Previous electric 
software solutions were created very organically since late 1970s/early 
1980s addressing

[[Page 40250]]

use-cases from that era. New tools were created over time, but always 
bolted-on to existent solutions. Energy Management Systems became more 
and more complex and started to present challenges in terms of 
scalability and maintainability leading to increasing staff and costs. 
Previous well defined siloes between Generation, Transmission and 
Distribution are becoming more blurred. In order to address all of 
these challenges, utilities and software companies started a journey to 
re-invent itself. Based on the most recent digital technologies, these 
companies created new modular and composable solution prepared for 
ultra-scaling and immense amounts of data ready to leverage the most 
modern mathematical algorithms and artificial intelligence methods 
available to date for assisted and automated control. The need for 
project executions in months as opposed to years has been taken 
carefully in consideration, creating a software solution ready for 
faster time-to-value. These solutions are already in production at a 
few customers and a number of new use-cases are currently under proof-
of-concept, development or available for productization. The 
presentation will cover some of these latest software developments and 
highlight regulatory challenges to slowing the adoption of these 
technologies by utilities: 1. A new market system prepared to validate 
& clear more frequent and increasing number of bids with smaller 
amounts of power; 2. Digital twin technologies such as digital dynamic 
line ratings ready to integrate electrical and weather data to provide 
real-time and forecast ampacity for transmission lines integrated to 
real-time and look-ahead security assessment systems; 3. Advanced 
forecasting solutions based on AI for (1) renewable power production at 
T&D levels and (2) outage predictions for improved crew allocation and 
faster restoration times; 4. Optimal system restoration management in 
real-time in assisted and automated modes; 5. Exploration of 
Distributed Energy Resource to supply grid services at transmission 
level such as grid stabilization and blackstart restoration.

Day 3--Thursday, June 29

Session H1 (Thursday, June 29, 9:30 a.m.) (Commission Meeting Room)

Integration of DER Aggregations in ISO-Scale SCUC Models
Dr. Brent Eldridge, Electrical Engineer, Pacific Northwest National 
Laboratory (Bel Air, MD)
Jesse Holzer, Mathematician, Pacific Northwest National Laboratory 
(Richland, WA)
Abhishek Somani, Economist, Pacific Northwest National Laboratory 
(Richland, WA)
Eran Schweitzer, Electrical Engineer, Pacific Northwest National 
Laboratory (Richland, WA)
Rabayet Sadnan, Electrical Engineer, Pacific Northwest National 
Laboratory (Richland, WA)
Nawaf Nazir, Electrical Engineer, Pacific Northwest National Laboratory 
(Richland, WA)
Soumya Kundu, Electrical Engineer, Pacific Northwest National 
Laboratory (Richland, WA)

    FERC issued Order 2222 in September 2020, which will require all 
ISOs in the U.S. to implement participation models for DER aggregators. 
Among other requirements, this rule required ISOs to lower the 
participation threshold for wholesale market participation to 0.1 MW. 
Wider participation of these resources can bring significant benefits 
to the grid, such as by locating energy supply closer to demand, 
opening up more participation from the demand side, and providing an 
additional flexibility source to balance intermittent renewables. 
However, DER aggregations will have unique characteristics that may 
pose challenges to the large-scale security-constrained unit commitment 
(SCUC) software used by ISOs. This presentation will focus on the 
formulation of a new mathematical model to represent the internal 
constraints of a DER aggregator and the study design that is intended 
to better understand the challenges associated with DER integration.
Current-Voltage AC Optimal Power Flow for Unbalanced Distribution 
Network
Dr. Mojdeh Khorsand Hedman, Assistant Professor, Arizona State 
University (Tempe, AZ)
Zahra Soltani, Ph.D. Candidate, Arizona State University (Tempe, AZ)
Dr. Shanshan Ma, Postdoctoral Research Scholar, Arizona State 
University (Las Vegas, NV)

    With proliferation of distributed energy resources (DERs), 
distribution management systems (DMSs) need to be advanced in order to 
enhance the reliability and efficiency of modern distribution systems. 
This work proposes novel nonlinear and convex AC optimal power flow 
(ACOPF) models based on current-voltage (IVACOPF) formulation for an 
unbalanced distribution system with DERs. In the proposed formulation, 
untransposed distribution lines, shunt elements of distribution lines, 
and detailed representation of distribution transformers and DERs are 
modeled. The proposed nonlinear IVACOPF model is linearized and 
convexified using the Taylor series. The performance of the proposed 
nonlinear and convex IVACOPF approaches is compared with OpenDSS and 
the widely used LinDistFlow method for modeling unbalanced distribution 
systems. The proposed accurate convex IVACOPF model has multiple 
applications for distribution system management, planning, and 
operation. Applications of the proposed model on two key parts of 
advanced DMS, (i) DERs scheduling and (ii) simultaneous topology 
processor and state estimation, will be presented. Two models are 
developed including Quadratic Programming (QP) and linear programming 
(LP) for performing the distribution state estimation. The performance 
of the methods is compared. The proposed models are tested using 
distribution feeder of an electric utility in Arizona.
Empowering Electricity Markets Through Distributed Energy Resources and 
Smart Building Setpoint Optimization: A Graph Neural Network-Based Deep 
Reinforcement Learning Approach
Dr. You Lin, Postdoctoral Associate, Massachusetts Institute of 
Technology (Cambridge, MA)
Dr. Audun Botterud, Principal Research Scientist, Massachusetts 
Institute of Technology (Cambridge, MA)
Dr. Daisy Green, Postdoctoral Associate, Massachusetts Institute of 
Technology (Cambridge, MA)
Dr. Leslie Norford, Professor, Massachusetts Institute of Technology 
(Cambridge, MA)
Dr. Jeremy Gregory, Executive Director of Climate and Sustainability 
Consortium, Massachusetts Institute of Technology (Cambridge, MA)

    Smart buildings play a pivotal role in the electricity market by 
boosting energy efficiency and demand flexibility by implementing 
advanced control strategies. In this study, a setpoint optimization 
model is proposed using a graph neural network-based deep reinforcement 
learning (DRL) algorithm that considers thermal exchanges among various 
zones within buildings. By intelligently scheduling the day-ahead 
temperature setpoints and adjusting the real-time setpoints in response 
to dynamic conditions and price signals, DRL-based controllers can 
optimize energy consumption while reducing overall costs. This 
strategic energy

[[Page 40251]]

management not only benefits building occupants but also bolsters the 
electricity grid through load balancing and the provision of essential 
grid services. Through the testbed of MIT campus buildings, it is 
demonstrated that smart buildings employing DRL for setpoint 
optimization contribute to a more efficient, reliable, and sustainable 
electricity market.
Multi-Timescale Operations of Nuclear-Renewable Hybrid Energy Systems 
for Reserve and Thermal Products Provision
Jie Zhang, Associate Professor, University of Texas at Dallas 
(Richardson, TX)
Jubeyer Rahman, Ph.D. Student, University of Texas at Dallas 
(Richardson, TX)

    This talk will present an optimal operation strategy of a nuclear-
renewable hybrid energy system (N-R HES), in conjunction with a 
district heating network, which is developed within a comprehensive 
multi-timescale electricity market framework. The grid-connected N-R 
HES is simulated to explore the capabilities and benefits of N-R HES of 
providing energy products, different reserve products, and thermal 
products. An N-R HES optimization and control strategy is formulated to 
exploit the benefits from the hybrid energy system in terms of both 
energy and ancillary services. A case study is performed on the 
customized NREL-118 bus test system with high renewable penetrations, 
based on a multitimescale (i.e., three-cycle) production cost model. 
Both day-ahead and real-time market clearing prices are determined from 
the market model simulation. The results show that the N-R HES can 
contribute to the reserve requirements and also meet the thermal load, 
thereby increasing the economic efficiency of N-R HES (with increased 
revenue ranging from 1.55% to 35.25% at certain cases) compared to the 
baseline case where reserve and thermal power exports are not 
optimized.

Session H2 (Thursday, June 29, 12:30 p.m.) (Commission Meeting Room)

Optimizing Stand-Alone Battery Storage Operations Scheduling Under 
Uncertainties in German Residential Electricity Market Using Stochastic 
Dual Dynamic Programming
Pattanun Chanpiwat, Doctoral Candidate, University of Maryland (College 
Park, MD) & Aalto University (Espoo, Finland) (Silver Spring, MD)
Fabricio Oliveira, Ph.D., Associate Professor, Aalto University (Espoo, 
Finland)
Steven A. Gabriel, Ph.D., Full Professor, University of Maryland 
(College Park, MD)

    We present a new variation of the stochastic dual dynamic 
programming (SDDP) algorithm for solving multistage, convex stochastic 
programming problems considering uncertainties such as electricity 
prices, variable renewable energy generation, and residential demand in 
the electricity market. We approximate the convex expected-cost-to-go 
functions via a linear policy graph, to obtain optimal operational 
strategies for the battery storage usage of residential households. We 
develop a heuristic algorithm (i.e., executable on edge-computing 
devices located at the households) of a residential electricity network 
with a flexible structure that allows residents to efficiently hedge 
their electricity consumption via community-shared battery storage 
while accounting for uncertainties and limitations of the energy 
system. We provide an economic assessment and insights into battery 
storage scheduling strategies and the model capabilities through case 
studies on a test network model of Southern German residential 
households. The results are compared with other mathematical models 
including a multistage stochastic convex optimization model with the 
assumptions of a perfect information case and/or a business-as-usual 
case.
Integration of Hybrid Storage Resources Into Wholesale Electricity 
Markets
Dr. Nikita Singhal, Technical Leader, Electric Power Research Institute 
(Palo Alto, CA)
Rajni Kant Bansal, Ph.D. Candidate, Johns Hopkins University 
(Baltimore, MD)
Dr. Erik Ela, Program Manager, Electric Power Research Institute (Palo 
Alto, CA)
Dr. Julie Mulvaney Kemp, Research Scientist, Lawrence Berkeley National 
Laboratory (Berkeley, CA)
Dr. Miguel Heleno, Research Scientist, Lawrence Berkeley National 
Laboratory (Berkeley, CA)

    Electric storage resources and other technologies that are co-
located and share a common point of interconnection are presently being 
incorporated into bulk power systems in increasing numbers, with more 
hybrid storage resources planned and under study within interconnection 
queues. Such hybrid storage resources are predominantly seen being 
combined with variable energy resources and are either being operated 
as two separate resources or as a single integrated resource. Market 
designers and system operators are presently researching ways to 
effectively integrate hybrid storage resources into their existing 
system operations and scheduling processes given the ambiguity around 
their impacts, particularly when high levels of hybrid resources are 
present. This research explores advanced market participation modeling 
options for integrating utility-scale hybrid storage resources into 
market clearing software in addition to discussing the economic and 
reliability implications of the different modeling options. This 
includes the consecutive impact of the participation models on the 
market clearing software solution and the dispatch and revenue of 
hybrid battery projects. The alternate participation models evaluated 
in this research include two separate resources ISO-managed co-located 
participation model, single integrated resource self-managed hybrid 
participation model and two separate resources ISO-managed linked co-
located participation model.
Predicting Strategic Energy Storage Behaviors
Yuexin Bian, Ph.D. Student, University of California, San Diego (San 
Diego, CA)
Ningkun Zheng, Ph.D. Student, Columbia University (New York City, NY)
Yang Zheng, Assistant Professor, University of California, San Diego 
(San Diego, CA)
Bolun Xu, Assistant Professor, Columbia University (New York, NY)
Yuanyuan Shi, Assistant Professor, University of California, San Diego 
(San Diego, CA)

    Energy storage are strategic participants in electricity markets to 
arbitrage price differences. Future power system operators must 
understand and predict strategic storage arbitrage behaviors for market 
power monitoring and capacity adequacy planning. This paper proposes a 
novel data-driven approach that incorporates prior model knowledge for 
predicting the behaviors of strategic storage participants. We propose 
a gradient-descent method to find the storage model parameters given 
the historical price signals and observations. We prove that the 
identified model parameters will converge to the true user parameters 
under a class of quadratic objective and linear equality-constrained 
storage models. We demonstrate the effectiveness of our approach 
through numerical experiments with synthetic and real-world storage 
behavior data. The proposed approach significantly

[[Page 40252]]

improves the accuracy of storage model identification and behavior 
forecasting compared to previous blackbox data-driven approaches.
Energy Storage Participation Algorithm Competition (ESPA-Comp)
Dr. Brent Eldridge, Electrical Engineer, Pacific Northwest National 
Laboratory (Bel Air, MD)
Jesse Holzer, Mathematician, Pacific Northwest National Laboratory 
(Richland, WA)
Abhishek Somani, Economist, Pacific Northwest National Laboratory 
(Richland, WA)
Kostas Oikonomou, Electrical Engineer, Pacific Northwest National 
Laboratory (Richland, WA)
Brittany Tarufelli, Economist, Pacific Northwest National Laboratory 
(Laramie, WY)
Li He, Electrical Engineer, Pacific Northwest National Laboratory 
(Richland, WA)

    Energy Storage Participation Algorithm Competition (ESPA-Comp) is 
an upcoming pilot competition that will challenge participants to 
develop innovative algorithms for energy storage participation in 
wholesale electricity markets. Energy storage technologies will play a 
critical role in making sure we have access to reliable and low-cost 
electricity. However, optimizing energy storage systems in wholesale 
electricity markets is a complex task that requires sophisticated 
algorithms to accurately predict electricity prices and account for the 
physical constraints of energy storage technologies. ESPA-Comp aims to 
bring together researchers, engineers, and students with expertise in 
AI/ML, optimization, and economics to develop algorithms that can 
effectively address these challenges. In this competition, participants 
will ``operate'' an energy storage resource in a simulated wholesale 
electricity market and will be awarded based on the profits they earn. 
Participants will need to submit algorithms that generate strategic 
offer curves, taking into account factors like weather, market 
competition, and network congestion. Competition results will help us 
to understand how different market designs can affect storage 
incentives and support the efficient use of storage resources.

Session H3 (Thursday, June 29, 3:00 p.m.) (Commission Meeting Room)

Congestion Mitigation With Transmission Reconfigurations in the Evergy 
Footprint
Dr. Pablo A. Ruiz, CEO and CTO, NewGrid, Inc. (Somerville, MA)
Derek Brown, Regulatory Affairs Manager, Evergy (Topeka, KS)
Jeremy Harris, Transmission Operations Planning Manager, Evergy 
(Topeka, KS)
German Lorenzon, Senior Engineer, NewGrid (Somerville, MA)
Grant Wilkerson, Director of Business Development, Evergy (Kansas City, 
MO)

    Transmission needs are becoming more variable and are rising 
rapidly, as shown by significant increases in congestion management 
costs and in the frequency of transmission overloads. Further, 
transmission capability has been critical during recent extreme events, 
to support power transfers from less affected areas to the more 
affected ones. Topology optimization software is a grid-enhancing 
technology that identifies reconfiguration options to re-route power 
flow around transmission bottlenecks employing less utilized facilities 
and satisfying reliability criteria. These reconfigurations provide 
cost savings to power customers and increase the transmission network 
performance from both reliability and market efficiency perspectives. 
At the same time, the use of reconfigurations remains limited. For 
example, the usual practice in the Southwest Power Pool is to employ 
known reconfigurations as a last resort, after resource redispatch is 
exhausted and constraints are breached. This presentation will discuss 
the reliability and cost saving impacts of reconfigurations implemented 
in the Evergy footprint to mitigate congestion under the current SPP 
practice, as well as illustrate additional benefits that could be 
obtained if topology optimization opportunities were used proactively 
to address congestion.
Optimal Transmission Expansion Planning With Grid Enhancing 
Technologies
Swaroop Srinivasrao Guggilam, Senior Engineer, Electric Power Research 
Institute (Knoxville, TN)
Alberto Del Rosso, Program Manager, Electric Power Research Institute 
(Knoxville, TN)

    The power system is evolving with a rapid increase in demand. It 
provokes rethinking ways to increase generation and expand the system's 
capacity to support it. This combination of fast-paced demand growth 
and supply has made the planning and expansion of the transmission 
system challenging in recent years. The futuristic hyperactive power 
system grid needs to be versatile. The grid should be able to host a 
variety of renewable energy resources, adapt to various system 
conditions, be highly secured under extreme events, and be dynamically 
responsive to make the power system reliable. All this is to be 
achieved at minimal cost to the customers and efficiently. The 
traditional transmission solutions will continue to be the backbone of 
the power system transmission grid, but upcoming state-of-the-art grid-
enhancing technologies can significantly aid in supporting these ever-
changing power system grid requirements with optimal cost and improved 
efficiency. Various grid-enhancing technologies include power flow 
control devices such as SmartValve devices and phase shift 
transformers, dynamic and adaptive transmission line ratings, and 
optimal topology control. The increasing penetration of distributed 
energy resources such as batteries also activates a different avenue to 
pursue being able to support transmission expansion planning needs. The 
term around the battery as a viable alternative is coined as a non-wire 
alternative solution. In many utilities, it's necessary to assess the 
non-wire alternative solutions such as batteries to meet FERC 
requirements. Developing and analyzing these various modern 
transmission solutions that work in tandem is challenging. One needs 
proper technical characterization of these technologies and assess the 
technology readiness. One also needs to evaluate its performance under 
normal and extreme conditions, the flexibility to deploy and install 
these technologies, calculate capital and operational costs, understand 
different available control options for these devices, and analyze 
potential limitations. Suitable analytical methods and high-performing 
software tools are needed to run the optimization simulations to enable 
integration and efficient use of these grid-enhancing technologies. 
EPRI has developed a software tool called CPLANET (Controlled PLANning 
Expansion Tool) that helps identify effective and low-cost solutions 
for mitigating thermal overloads in a power system over various 
operating scenarios. An optimal solution is determined from a given set 
of candidate projects, including various grid-enhancing technologies 
and traditional transmission expansion projects such as installing new 
transmission lines or upgrading existing substations. The software uses 
a mixed-integer linear programming formulation in the optimization 
engine to identify the least-cost solution for the grid's various 
physical and operating needs. The scope and goal of this presentation 
are to discuss the ongoing efforts at EPRI's forefront around grid-
enhancing technologies. Showcase the current capabilities of the 
CPLANET tool and

[[Page 40253]]

discuss case studies and share existing challenges and future goals.
The Key Role of Extended ACOPF-Based Decision Making for Supporting 
Clean, Cost-Effective and Reliable/Resilient Electricity Services
Maria Iilic, Professor Emerita, Carnegie Mellon University (Pittsburgh, 
PA)
Rupamathi Jaddivada, Director of Innovation, SmartGridz (Boston, MA)
Jeffrey Lang, Vitesse Professor, Massachusetts Institute of Technology 
(Cambridge, MA)
Eric Allen, Director of Engineering, SmartGridz (Boston, MA)

    Societal objectives are rapidly moving towards decarbonized, 
affordable, and reliable/resilient electricity services. In this talk 
we first revisit these objectives by identifying basic changes and the 
related challenges taking place. In particular, decarbonization 
requires planning and operations of the changing electric energy 
systems so that seamless integration of clean resources, ranging across 
wind, solar, nuclear, geothermal, and hydro, is enabled. Notably, this 
must be done with an eye on generation adequacy. Also, these new 
resources present locational issues (NIMBY) in operating the existing 
power grid. Finally, the end users still must be served without 
interruptions and without being exposed to wide-spread blackouts. 
Similar challenges are related to ensuring cost-effective and reliable/
resilient services. Second, we show how an extended (robust, adaptive, 
multi-temporal) ACOPF is essential for meeting these societal 
challenges. Pretty much any of the new software needed (for wind 
integration, resilient service, and preventing blackouts) requires 
effective optimization tools for identifying the main bottlenecks/
obstacles to physical implementation and for advising operators and 
planners regarding the most effective remedial actions (new investments 
and/or flexible utilization). We illustrate potential benefits from 
utilizing ACOPF as a basic means of supporting software tools needed 
for meeting the societal challenges. We offer a taxonomy of such badly 
needed tools and illustrate the role of extended ACOPF estimated 
benefits on several real-world systems based on our work to-date.
Data & API Standards for Clean Energy Solutions and Digital Innovation
Priya Barua, Director of Market Policy and Innovation, Clean Energy 
Buyers Institute (Washington, DC)
Ben Gerber, President & CEO, M-RETS (Minneapolis, MN)

    There is an opportunity for energy attribute certificate (EAC) 
issuing bodies in the U.S. and abroad to enable next generation carbon-
free electricity (CFE) procurement solutions that accelerate grid 
decarbonization investments by capturing more attributes and better 
serving as a digital ``platform of platforms''. Energy customers who 
buy clean energy rely on EACs to assert ownership claims over each 
megawatt-hour of CFE they procure for auditing, reporting, and 
marketing purposes. EAC issuing bodies promote CFE procurement 
integrity and validation by issuing, tracking, and canceling EACs, 
which each represent a unique standardized tradable instrument 
representing one megawatt-hour of verified CFE generation. By adopting 
open data and automated programming interface (API) standards, EAC 
issuing bodies can improve data access and solutions for customers. 
This session will explore opportunities for EAC issuing bodies to 
establish consistent, modern automated programming interfaces (APIs), 
template legal agreements, and other tools that will make it easier for 
data providers to deliver data and for users to update the status of 
EACs through connected digital trading platforms-- enabling innovation 
for CFE procurement solutions.
Mine Production Scheduling Under Time-of-Use Power Rates With Renewable 
Energy Sources
Dr. Daniel Bienstock, Professor, Columbia University (New York, NY)
Amy Mcbrayer, Ph.D. Candidate, South Dakota School of Mines (Rapid 
City, SD)
Andrea Brickey, Professor, South Dakota School of Mines (Rapid City, 
SD)
Alexandra Newman, Professor, Colorado School of Mines (Golden, CO)

    Renewable energy use on active and reclaimed mine lands has 
increased dramatically in recent years. With mining companies focused 
on increasing efficiencies, reducing carbon intensity, and developing 
sustainable mining practices, opportunity exists to integrate data on 
electricity usage and demand into mine production schedules to 
capitalize on alternative energy sources and to take advantage of 
favorable pricing strategies. Utilizing real data from an active coal 
mine that has already integrated electric equipment into their loading 
fleet, we show the impacts of (i) seasonal power price fluctuations on 
a medium-term production schedule; and, (ii) hourly power price 
fluctuations on a short-term extraction schedule. Results reveal the 
economic potential both for: (i) the integration of renewable energy 
sources on reclaimed and active mine lands; and (ii), the corresponding 
synchronization of a production schedule with time-of-use energy 
pricing contracts.

[FR Doc. 2023-13168 Filed 6-20-23; 8:45 am]
BILLING CODE 6717-01-P