Synthetic Power Grid Models: What are They, How Theyre Made, and - PowerPoint PPT Presentation

Synthetic Power Grid Models: What are They, How They’re Made, and Why They Matter Tom Overbye University of Illinois at Urbana-Champaign (overbye@illinois.edu) PSERC Webinar March 15, 2016

Acknowledgments and Thanks • Work presented in these slides is based on the results of several projects including • PSERC S-62G (Seamless Bulk Electric Grid Management with EPRI) • PSERC T-57 (High Impact) • BPA project TIP 353 (Improving Operator Situation Awareness by PMU Data Visualization • ARPA-E Grid Data Synthetic Data for Power Grid R&D • Support is gratefully acknowledged! • Thanks also to Adam Birchfield, Kathleen Gegner, Ti Xu, Komal Shetye, Richard Macwan, Profs Bob Thomas, Anna Scaglione, Zhifang Wang and Ray Zimmerman 2

Presentation Overview • Access to data about the actual power grid is often restricted because of requirements for data confidentiality (e.g., critical energy infrastructure) • Focus here is on high voltage power flow, optimal power flow, transient stability models, SCADA, PMUs • Some data is public, some is available by NDAs, and some is essentially unavailable to those outside of power system control centers • Focus of talk is on the creation of synthetic (fictional) models that mimic the complexity of the actual grid cases but will contain no confidential data and can be publicly available 3

A Few Initial Thoughts • The reason why this matters is to help spur innovation in the electric grid software • Algorithms tested on synthetic models applied to actual • In 2000 the NAE named Electrification (the vast networks of electricity that power the developed world) as the top engineering technology of the 20th century • automobiles (2), airplanes (3), water (4), electronics (5) • Our challenge in this century is to develop a sustainable and resilient electric infrastructure for the entire world 4

A Few Initial Thoughts • "All models are wrong but some are useful,“ George Box, Empirical Model-Building and Response Surfaces , (1987, p. 424) • “The use of nondisclosure agreements or NDA’s to obtain data, while useful in many instances, is not useful if the world community is to engage in research that adheres to the scientific principle of reproducibility of results by other qualified researchers and to use important findings to advance their own work“ PSERC Founding Director Bob Thomas, 2015 5

Overall Goals • The development of entirely synthetic transmission system models and scenarios that match the complexity and variety of the actual grid • Models that incorporate both the average characteristics and outlier characteristics of the actual grid • Models and scenarios suitable for security constrained optimal power flow (SCOPF) studies; they will also be set for use in transient stability and geomagnetic disturbance analysis • All models will have embedded geographic coordinates • Scenarios will be SCOPF validated • We want to partner with industry! 6

The Need • Few, if any, of the existing public models (such as the IEEE 300 bus) match the complexity of the models used for actual large-scale grids • Issues include size, with the 3 176 MW 105 Mvar 3 3 100 MW 96 MW 75 Mvar 47 MW 7 Mvar 3 8 MW 26 Mvar 131 MW 3 Mvar 3 96 Mvar 3 61 MW 3 30 Mvar 3 42 MW 14 Mvar 3 3 3 3 3 -114 MW 33 Mvar 77 MW 22 MW 16 Mvar 38 MW 77 Mvar 3 3 13 Mvar 3 3 3 3 171 MW 61 MW 70 Mvar 30 Mvar Eastern Interconnect models 3 3 328 MW 3 188 Mvar 29 MW 43 MW 14 Mvar 14 Mvar 3 3 3 3 255 MW 149 Mvar 3 3 428 MW 29 MW 3 232 Mvar 27 MW 14 Mvar 173 MW 3 12 Mvar 99 Mvar 3 3 3 269 MW 35 MW -23 MW 410 MW 12 Mvar -17 Mvar 40 Mvar 157 Mvar 3 3 3 3 3 572 MW 489 MW 800 MW 244 Mvar 53 Mvar 3 72 Mvar 64 MW 404 MW 21 Mvar 212 Mvar 3 3 7 MW 3 223 MW 2 Mvar 148 Mvar 3 3 3 3 3 3 159 MW 10 MW 96 MW 3 Mvar 3 38 MW 595 MW now more than 70,000 buses, 107 Mvar 46 Mvar 13 Mvar 448 MW 3 120 Mvar 143 Mvar 3 3 7 MW 58 MW 1 3 0 MW 2 Mvar 10 Mvar -5 Mvar 3 3 148 MW 3 33 Mvar 1 1 12 MW 77 MW 2 Mvar 1 1 Mvar 538 MW 3 3 3 83 MW 369 Mvar 3 1 1 3 21 Mvar 72 MW 24 Mvar 3 1 1 353 MW 130 Mvar 1 120 MW 1 41 Mvar 3 2 58 MW -33 MW 14 Mvar 1 1 81 MW 2 96 MW 1 2 -29 Mvar 561 MW 220 Mvar 160 MW 56 MW 43 Mvar 23 Mvar 60 Mvar -21 MW 1 15 Mvar -14 Mvar 69 MW 13 Mvar 1 41 MW 1 1 1 2 2 14 Mvar 18 MW 28 MW 1 and also model complexity 1 1 1 205 Mvar 0 Mvar 0 Mvar 9 Mvar 1 1 482 MW 127 MW 23 Mvar 1 20 MW 2 MW 78 MW 2 1 90 MW -13 Mvar 1 Mvar 2 1 31 MW 1 1 49 Mvar 0 Mvar 1 285 MW 1 45 MW 2 2 100 Mvar 1 1 1 12 Mvar 44 MW 276 MW 300 MW 0 Mvar 1 26 MW 28 MW 59 Mvar 535 MW 96 Mvar 0 Mvar 1 1 7 Mvar 2 55 Mvar 1 2 1 1 100 M W 2 2 29 M var 2 764 MW 70 MW 21 MW 1 1 -100 M W 291 Mvar 183 MW 17 MW 58 MW 7 Mvar 116 MW 44 Mvar 30 Mvar 0 Mvar 12 Mvar 2 34 M v ar 2 1 1 -24 Mvar 2 2 2 1 195 MW 1 515 MW 381 MW 2 1 50 MW 37 Mvar 827 MW 60 MW 2 98 MW 1 29 Mvar 1 17 Mvar 1 2 83 Mvar 1 20 Mvar 2 135 Mvar 24 Mvar 1 2 229 MW 2 1 224 MW 2 58 MW 2 1 1 71 Mvar 12 Mvar 5 Mvar 60 MW 145 MW 1 1190 14 MW 24 Mvar 50 Mvar -35 Mvar 2 2 1 1 56 MW 28 MW 1200 2 2 2 20 Mvar -20 Mvar 1 650 Mvar 124 MW 2 116 MW 1 20 MW -24 Mvar 73 MW 1 38 Mvar 1 2 2 0 Mvar 1 14 MW 0 Mvar 2 2 200 MW 1 33 MW 2 137 MW 50 Mvar 1 22 MW 1 48 MW 1 1 3 Mvar • Public models also lack extra 17 Mvar 0 MW 10 Mvar 1 60 MW 40 MW 14 Mvar 1 17 Mvar 0 Mvar 2 2 0 Mvar 0 Mvar 1 169 MW 1 42 Mvar 2 75 MW 1 1 1 -15 MW 57 MW 2 70 MW 274 MW 2 26 Mvar 5 Mvar 2 55 MW 1 1 19 Mvar 100 Mvar 61 MW 78 MW 1 18 Mvar 2 28 Mvar 0 Mvar 1 2 2 40 MW 1 1 17 MW 4 Mvar 10 MW 9 Mvar 2 14 MW 1 1 1 86 MW 1 Mvar 85 MW 0 Mvar 1 2 2 1 Mvar 1 74 MW 24 Mvar 28 Mvar 1 -5 MW 155 MW 67 MW 1 1 83 MW 1 30 MW 388 MW 2 1 18 Mvar 0 Mvar 37 MW 35 MW 218 MW 5 Mvar 0 Mvar 1 1 Mvar 15 Mvar 115 Mvar 227 MW 106 Mvar 1 13 Mvar 2 2 69 MW 89 MW 1 110 Mvar 1 1 1 55 MW 3 Mvar 1 -11 MW 1 2 35 Mvar -1 Mvar 2 92 MW 1 1 63 MW 1 26 Mvar 6 Mvar 1 1 2 2 145 MW 25 Mvar 2 208 MW 58 Mvar 2 1 1 32 MW 107 Mvar data like transient stability 0 Mvar 2 2 2 2 1 1 241 MW 39 MW 5 MW 89 Mvar 427 MW 0 Mvar 16 MW 595 MW 174 Mvar 1 9 Mvar 0 Mvar 112 MW 0 Mvar 83 Mvar 28 MW 1 12 Mvar 69 MW 2 1 -1 Mvar 25 MW 63 MW 0 Mvar 2 1 2 49 Mvar 2 50 MW 1 26 MW 2 5 MW 0 Mvar 1 1 1 MW 0 Mvar 4 Mvar 24 MW 2 14 Mvar 9 MW 0 Mvar 0 Mvar 1 1 2 2 41 MW 2 19 Mvar 1 163 MW 43 Mvar 2 2 74 MW 2 MW 1 1 1 1 29 Mvar 1 46 MW -4 Mvar -21 Mvar 66 MW 1 2 2 0 Mvar 1 1 1 176 MW 1 30 MW 83 Mvar 85 MW 23 Mvar 32 Mvar 4 MW 2 MW 2 1 Mvar 1 Mvar 1 Mvar 56 MW 3 MW 25 Mvar 1 1 1 1 2 3 MW 1 1 Mvar • Innovation is hindered by not 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 MW 0 Mvar 1 MW 1 MW 3 MW 1 MW 2 MW 2 MW 2 MW 2 MW 1 0 Mvar 0 Mvar 1 Mvar 0 Mvar 1 Mvar 1 Mvar 1 Mvar 1 Mvar 1 MW 5 MW 0 Mvar 2 Mvar 1 1 1 1 1 1 2 MW 1 MW 1 MW 1 2 MW 1 Mvar 0 Mvar 0 Mvar 1 Mvar 0 MW 0 Mvar 0 Mvar 0 MW being able to compare results for complex models Image: IEEE 300 Bus case downloaded from http://icseg.iti.illinois.edu/ieee-300-bus-system/ 7

What Makes a Model Real? • The challenge is to capture the essence of what makes actual grid models different • Actual grid models are quite diverse • Statistics can be used to quantify some of the characteristics • topology, parameters for buses, generators, loads, transmission lines, transformers, switched shunts, transient stability and GMD parameters • System-wide metrics are also needed 8

Complexity Examples • A recent 76,000 bus Eastern Interconnect (EI) power flow model has 27,622 transformers including 98 phase shifters • Impedance correction tables are used for 351, including about 2/3 of the phase shifters; tables can change the impedance by more than two times over the tap range • The voltage magnitude is controlled at about 19,000 buses (by Gens, LTCs, switched shunts) • 94% regulate their own terminals with about 1100 doing remote regulation. Of this group 572 are regulated by two or more devices, 277 by three or more, twelve by eight or more, and three by twelve devices! 9

How to Make Realistic, Geographically- Based, Synthetic Models • Our approach is to make models that look real and familiar by siting these synthetic models in North America, and serving a population density the mimics that of North America • The transmission grid is, however, totally fictitious • Goal is to leverage widely available public data: • Geography • Population density (easily available by post office) • Load by utility (FERC 714) and state-wide averages • Existing and planned generation: Form EIA-860 contains information about generators 1 MW and larger; data includes location, capacity and fuel type 10

Example: 2100 Bus Texas Case Frequency Response 11

EIA-860 Generator Data • Online at www.eia.gov/electricity/data/eia860/ Since our goal is to make entirely synthetic models, no existing company names will be used. We may be changing the actual generator capacity values as well. 12