Last Updated: March 22, 2019 (see Change Log for release notes)
Initial Release: October 30, 2018
We recommend running check_data.py (see Evaluation Source Code) on these datasets to identify problems.
The Original Dataset 1 consists of two parts:
- Original_Dataset_Offline_Edition_1 with Network Models
- Original_Dataset_Real-Time_Edition_1 with Network Models
These two parts are the same except for the base case operating point. This point is represented by 5 variables (bus voltage magnitude in pu,VM; bus voltage angle in degrees, VA; generator real power output, PG; generator reactive power output, QG; and initial switched shunt susceptance, BINT) described in Appendix D.4 of the Formulation. These variables are mapped to the raw file in Appendices C.3 (Bus Data; A2.2), C.6 (Generator Data; A2.5), and C.9 (Switched Shunt Data, A2.8) This information can be used as a starting point for various algorithms. In general we make no assertion of feasibility or optimality for this starting point. We make no assertion of feasibility for the Offline version but the Real-Time version, which is derived from an economical unconstrained solution, should be feasible but not optimal.
The Real-Time version has a runtime limit of 10 minutes and is used for Divisions 1 and 3 scoring.
The Offline version has a runtime limit of 45 minutes and is used for Divisions 2 and 4 scoring.
Each of the two parts consists of 5 separate network models with 10 scenario instances each; Network_01*-10, Network_03*-10, Network-05*-10, Network_07*-10, and Network_10*-10. The last two digits indicate the number of scenario instances associated with each network (ten). The * can be either R for Real-Time or O for Offline. The files in both parts are the same except for the raw files, which contain the starting information variables VM, VA, PG, QG, and BINT. The first scenario from each network model is available as a Real-Time dataset in the Sandbox, along with additional configuration information. Please use the Sandbox version for algorithm development before running all 100 scenarios in this dataset.
Associated with each scenario instance are four case files (case.con, case.inl, case.raw, case.rop). Two of the case files change with each scenario instance (case.con and case.raw; also case.inl for Network_10*-10). These are kept in the scenario_n folders (n is 1 to 10). Case files that do not change (case.inl, except for Network_10-10, and case.rop) are kept in the network folder. The file inputfiles.ini in the network folder summarizes where each type of case file is to be found. Within each network folder is a readme.txt file with network configuration information (the number of buses, loads, generators, transformers, contingencies, etc.) and an explanation of inputfiles.ini.
These network models are the same as the ones with the same name in the Original Dataset 2, but with different scenario instances. The same will be true of future datasets such as Trial 1, Trial 2, and Final.
The use of the term scenario is not to be interpreted as involving time coupling or stochastic variation. These scenarios are separate instances and not time coupled—they are completely independent and need to be solved individually. There is only one case.raw for each scenario instance--no separate base case.