A Hybrid Optimization Strategy of Meta-Heuristic and Mathematical Programming Methods for Meshed Power Systems

Grid Planning Benchmark Dataset and Scripts for the Paper "A Hybrid Optimization Strategy Combining Network Expansion Planning and Switching State Optimization"

Instructions

1. Clone the repository
2. Install pandapower and PowerModels.jl (see 'Installation Instructions')
3. Run the optimization, e.g., with python run_greedy.py

Data

This repository contains these subfolders with the following data:

• "power_system_data" containing the power system benchmark cases included additional line and replacement measures
• "scaled_loadcases" contains the data to reproduce the mathematical programming results

Main script to run the hybrid optimization method

The script "run_greedy.py" runs the greedy heuristic combined with the PowerModels.jl optimization framework. An example call starting a combined optimization of switching measures and line measures for the 'brigande' test case is:

python run_greedy.py -grid 'brigande' -kind 'combo' -max_iterations 3 -res_dir './results'

See more options with:

python run_greedy.py -h

If you have trouble using the PowerModels.jl interface, you can just run the script without PowerModels.jl. This gives you a greedy optimization of switching and line measures. Try:

python run_greedy.py -pm ''

Additional script to run the PowerModels.jl Optimization

The repository also contains a script "run_powermodels.py" which runs the PowerModels.jl based results with the pandapower-python to PowerModels.jl-julia interface

The script has several command line options:

• model (str) - The PowerModels.jl power model e.g. "DCPPowerModel"
• solver (str) - The solver to use, e.g. "juniper", "gurobi"
• grid (str) - optional if you only want to calculate one grid, e.g. "brigande"
• kind (str) - the optimizations to run, e.g. "tnep,ots,repl". Can only be a part of these like "tnep,repl"

Example to run the model "DCPPowerModel" with "gurobi" as a solver for the grid "brigande" with the time series loadcases "ts" for the REPl "repl" problem:

python-jl run_powermodels.py --model="DCPPowerModel" --solver="gurobi" --grid="brigande" --kind="repl"

Installation instructions

You need the following software to run the script:

• pandapower: https://github.com/e2nIEE/pandapower
• PowerModels.jl: https://github.com/lanl-ansi/PowerModels.jl/
• tqdm: https://github.com/tqdm/tqdm
• numpy
• pandas

The python-julia interface might require some additional installations. See the following link for details: https://pandapower.readthedocs.io/en/latest/opf/powermodels.html