Public Data and Code

I have a number of projects for which we are able to offer open-source code and/or data to other researchers. Please find the details below.

Reinforcement Learning for Battery Trading

Primary developer: Caleb Ju

This package defines a reinforcement learning environment to train agents to perform energy arbitrage based on historic price signals.

You can find the code here.

Global Demand Data from 2019

Primary developer: Constance Crozier

This is a dataset of the hourly electricity demand in 2019 for 155 transmission systems, spanning five continents. Data is provided in UTC.

You can download the data here.

We request that you cite this paper in any resulting works. [Bibtex entry]

GridLearn: Grid-aware Multi-Agent Reinforcement Learning

Primary developer: Aisling Pigott

This package adds functionality to the CityLearn package, which simulates a multi-agent reinforcement learning environment for demand response. Added features include: coupled pandapower grid environment, sub-hourly intervals, phase shift enabled smart inverters, synchronous action selection.

You can find the code here.

We request that you cite this paper in any in any resulting works. [Bibtex entry]

Storage Cost and Optimization of Renewable Electricity Systems (SCORES)

Original developer: Constance Crozier

More recent contributions: Cormac O’Malley, Chris Quarton

This is a modeling framework for simulation and sizing of a renewables based electricity system. The model takes hourly data for weather and demand and can simulate the reliability of a system with given solar, wind, hydro, battery, thermal storage, and hydrogen assets.

You can find the code here.

A user guide is provided here.

We request that you cite this paper in any resulting works. [Bibtex entry]