Requirements

Python 3.7 or 3.8

Numpy, Pandas, Sklearn, Argparse, Statsmodels, CVXPY, Mosek, Matplotlib, Dask, Fastparquet, S3fs

Note Mosek optimization solver requires a license to run: https://www.mosek.com/products/academic-licenses/

Installation

installation via pip

Download OpenDSS network files from:
For San Francisco area networks: https://egriddata.org/dataset/bay-area-synthetic-network
For Iowa network: https://wzy.ece.iastate.edu/Testsystem.html
For Vermont network: https://sourceforge.net/p/electricdss/code/HEAD/tree/trunk/Distrib/EPRITestCircuits/epri_dpv/J1/
For Sacramento network: https://sourceforge.net/p/electricdss/code/HEAD/tree/trunk/Distrib/IEEETestCases/123Bus/
For Arizona network: https://sourceforge.net/p/electricdss/code/HEAD/tree/trunk/Distrib/IEEETestCases/34Bus/
The various network files can be reformatted to be made compatible with the rest of the code using the methods in standardize_network_files.py
Download solar data profiles from:
https://www.nrel.gov/grid/solar-power-data.html
Download NREL projections from:
https://data.nrel.gov/submissions/126 (We use high electrification with moderate technology growth in our study)
For flexible load: https://data.nrel.gov/submissions/127 (We use high electrification in our study)
Projections can be extracted from downloaded files using read_energy_projections.py
Download NREL load profiles using dl_nrel_raw_data.py
Note that the raw data can be >10GB per network

Each python file contains code related to a different aspect of the simulation. A short description of each is as follows.

DD_models.py: Training the data driven linear power flow models
dl_nrel_raw_data.py: Extracting and processing the original load profile datasets
read_energy_projections.py: Extracting the NREL projections from their datafile
opendss_interface.py: Interfacing the data with the OpenDSS simulator
EV_Profiles.py: Processing the data for the EV charging windows
global_optimization.py: Code for running the centralized controller optimization
local_controller.py: Code for running the local controller
process_data.py: Code for generating the DER scenarios including PV, EV, and storage placement
read_results.py: Code for reading the results of the power flow simulation
standardize_network_files.py: Code for interpreting between different OpenDSS data types

Download datafiles as shown in Datafiles section
run process_data.py to generate the scenario including electrified loads, PV, EV, and storage placement
run local_controller.py to simulate the local controller on the scenario
run read_results.py to evaluate the results of the power flow simulation under local control
run DD_models.py to train the models used by the centralized controller
run global_optimization.py to simulate the centralized controller on the scenario
run read_results.py to evaluate the results of the power flow simulation under centralized control

Note that running local_controller.py, read_results.py, and global_optimization.py for the full year horizon has a very long runtime