2021 IEEE-NASPI Oscillation Source Location Contest

This webpage hosts data and models to be used for the 2021 IEEE-NASPI Oscillation Source Location (OSL) Contest. For complete contest info and registration, please visit: https://www.naspi.org/node/890


  • 08/06/2021: Released full simulation models and data of all 13 cases used by this Contest (Go to download page), and the solution key.
  • 07/21/2021: Award Announcement - The contest committee is happy to announce the top three finishers of the contest, all of whom had very impressive scores:
  • 05/10/2021: Added three additional cases.
  • 04/19/2021: Released the final contest data set and solution template, updated the one-line diagram, provided explanation on how VA is calulated in TSAT.
  • 03/26/2021: Added the small-signal analysis result of the test system.
  • 03/19/2021: Added a one-line diagram of the test system and a list of monitored buses and branches.
  • Contest Objective:

    Oscillations are a significant concern for reliable power system operation. This contest was organized to advance the electric power industry's ability to respond appropriately when oscillations occur. Contestants will be presented with realistic and challenging cases to help them evaluate and refine their methods. The contest will highlight robust methods that can be adopted by system operators and reliability coordinators to improve system reliability.

    With these objectives in mind, solutions should focus on providing actionable information for mitigating oscillations in the online environment. Based on available PMU measurements, successful methods will correctly characterize oscillations and identify their sources (if applicable) in a way that is practical for real-world deployment. This requires that the source be identified as accurately as possible given the available data, whether that's to a specific generator, load, power plant, substation, or area. Successful methods will handle real-world constraints such as limited system observability, limited topology information, imperfect measurements, and the simultaneous occurrence of system disturbances. The test cases have been designed to reflect these real-world challenges. Reliable methods that are robust to these conditions will provide a significant benefit to the reliability of power systems.

    About the Test System:

    All test cases used for this contest were generated by simulating a WECC 240-bus test system [1] developed by NREL based on reference [2]. The system has 243 buses, 146 generating units at 56 power plants (including 109 synchronous machines and 37 renewable generators), 329 transmission lines, 122 transformers, 7 switched shunts and 139 loads. A base case of the test system can be accessed from NREL model release webpage.

    Note that the contest committee has made the following changes to the original base case:

    Included in the NREL model webpage download:

  • PSS models were added to generate test cases suitable for this contest.

  • Not Included in the NREL model webpage download:

  • The contest cases used a different power flow dispatch and topology;
  • HVDC dynamic model;
  • Some control settings may be modified to produce the desired oscillatory phenomena (see details in the case description).

  • The following provides a one-line diagram on the entire system topology, except for the CA area whose detailed topology will be posted here in a separate figure before April 19, 2021.


    About Test Cases:

    There will be a dozen test cases for the contest. For each case:

  • Synthetic PMU measurements of bus voltage and branch current phasors from multiple locations of the test system is provided;
  • There is a 30 seconds leading window before the event and 60 seconds time window after that, total 90 seconds of data;
  • White noise is added to the load during simulation to mimic random load fluctuations;
  • Powertech Lab's TSAT software was used for the time-domain simulation;
  • EPRI's PMU Emulator is used to process the simulation results to mimic PMU device performance, a mix of P Class (2-cycle window) and M Class (6-cycle window) PMUs were used;
  • EPRI's Synchrophasor Data Conditioning Tool is used to process the synthetic PMU data to introduce data quality problems;
  • Data will be in the IEEE OSL TF Test Case Library Format. Each data set consists of four text files for Bus voltage magnitudes, Bus voltage angles, Branch current magnitudes, and Branch current angles. For your convenience, a Matlab program to read data into Matlab work space is also provided;
  • Voltage angle measurements are referenced to the swing bus ("3933 | TESLA 20.0") angle at t=0 and evolve according to the base frequency of 60 Hz.

  • Observability:

    The system consists of four areas: NORTH, SOUTH, CALIFORNIA, and MEXICO. Synthetic PMU measurements provide partial observability to the system in all four areas including the following. Monitored buses and branches are also summarized in an EXCEL file.

  • All 23 tie-lines between all four areas. For convenience of sign notations, all tie-lines to CALIFORNIA are monitored from CALIFORNIA side, while tie-lines between SOUTH and NORTH are monitored from the SOUTH side;
  • AC lines connecting HVDC terminals at CELILO and SYLMARLA;
  • 35 transmission lines (mainly 500 KV) within NORTH, SOUTH and CALIFORNIA areas;
  • Step-up transformers connecting 23 power plants to the network (measured at the low-voltage side) in all four areas. Thus, only 23 of 56 power plants are monitored by PMUs and no individual generator within a power plant is monitored by a PMU.
  • Pro tips:

  • Sustained oscillations may be forced or due to a poorly damped natural mode;
  • Forced oscillations are generated by adding a periodic signal into a load, governor or exciter of a synchronous machine, control of HVDC, or any combination of the above;
  • No forced oscillations originate from renewable generators;
  • Frequency and amplitude of a forced oscillation may be time-varying;
  • A forced oscillation may resonate with a natural mode;
  • There may be more than one oscillation sources;
  • Source(s) of an oscillation, if exist, may not be monitored by or close to a PMU;
  • A short-circuit fault and/or a line tripping event may happen as an oscillation initiating event or as a coincidental event to a forced oscillation.
  • Sample Data Set:

    Sample data set is for you to get familiar with the data format and the system. The final contest data set only contains synthetic PMU data.

  • Click here to download the sample data set.

  • System Model is based on NREL 240-bus WECC reduced model.

  • Contest Data Set and Solution Submission Instructions

  • Click here to download the contest data set.

  • Click here to download three additional cases.

  • Click here to download the contest solution template.

  • Q&A

    Q1: Can the small-signal analysis (SSA) result for the test system be released?

    A1: Yes. We have prepared the SSA result for the test system using SSAT by Powertech Labs. Note that frequency range (0.01Hz, 10Hz) and damping ratio range (0%, 20%) were used as filters, and the resulting 97 modes were included in the first sheet. Mode shapes, participation factors and left eigenvectors of 30 modes whose frequencies<1.5Hz, among these 97 modes, were respectively included in the second and following sheets.

    Q2: What do the machine ID acronyms mean?

    A2: If there is only one letter, this letter represents its fuel type. If there are two letters, except for SC and DP, the first letter represents the owner info (not relevant to this contest), the second letter represents the fuel type, e.g.: B - Biomass; G - Gas; H - Hydro; S - PV; W - Wind; C - Coal; E - Geothermal; N - Nuclear; SC - Synchronous Condenser; R - Renewable (not differentiate wind or PV); P - Pump Storage; D - Demand Response; DP - Distributed PV.

    Responses to some other questions:

  • In case of conflict, use the data in the model file, not the one-line.
  • Line flow from bus# 4009 to bus #4104 was inevitably duplicated in the PMU data, they are the same.
  • In filling out the solution sheet, any additional nuisance or wrong answers may lower your scores.

  • References

    [1] J. E. Price and J. Goodin, "Reduced network modeling of WECC as a market design prototype," IEEE Power Energy Soc. Gen. Meet., 2011.

    [2] H. Yuan, R. Sen biswas, J. Tan, Y. Zhang, "Developing a Reduced 240-Bus WECC Dynamic Model for Frequency Response Study of High Renewable Integration," 2020 T&D