Welcome to the official repository of AlphaDataCenterCooling, a framework designed for enhancing and optimizing data center cooling systems. This virtual testbed is split into two integral components:

1. AlphaDataCenterCooling Docker Service: Encapsulates the data center cooling system's simulation model. Operates within a Docker container for consistent, reproducible simulations.
2. AlphaDataCenterCooling Gym Environment: Located within the AlphaDataCenterCooling_Gym directory. It nterfaces with the Docker service, offering a standardized environment for applying and testing various control algorithms, specifically designed for optimizing data center cooling strategies.

AlphaDataCenterCooling is an open-source virtual testbed specifically designed to facilitate the development and evaluation of control strategies for data center cooling systems. This virtual testbed is built using Python and Modelica, and it is wrapped using the standardized interface of Farama-Foundation Gymnasium (previously known as OpenAI Gym). The Modelica model of the cooling system is constructed based on the architecture and the annual historical operational data of a real data center's cooling system.

• /AlphaDataCenterCooling_Gym: This folder contains the gym environment for the AlphaDataCenterCooling.
• /Resources: This folder contains models, boundary condition data, and other data required to load the model. This is crucial for the initialization and normal operation of the environment.
  • AlphaDataCenterCooling_FMU.fmu: This is a Functional Mock-up Unit (FMU) representing the cooling system model.
  • Disturbance.csv: Boundary condition data.
  • Initialization_actions.csv and Initialization_observation0.csv: Data used for initializing the environment.
  • mlp.pth: This is a Multilayer Perceptron (MLP) neural network model used for calculating the required head (pressure)
  • version.txt: Indicates the current version of the AlphaDataCenterCooling environment.
• /docs: This directory contains documentation and figures related to the AlphaDataCenterCooling environment.
• /testing: This directory includes notebooks for REST API interaction.
  • test_REST_API.ipynb: Demonstrates how to interact with the AlphaDataCenterCooling environment using the REST API.
  • validate_Pumps_Power.ipynb: Validates whether the results obtained using Matlab Simulink match those simulated using the PyFMI library.

1. Download and navigate to the repository:

git clone https://github.com/wfzheng/AlphaDataCenterCooling.git
cd AlphaDataCenterCooling

2. Install Docker and Docker Compose.
3. To construct and initiate the AlphaDataCenter Service, execute the subsequent command within the /AlphaDataCenterCooling root directory:

docker-compose up

4. Validate the functionality of the AlphaDataCenter docker environment using the Jupyter notebook located at test/test_REST_API.ipynb

Feel free to send any questions/feedback to: Zhe Wang

If you use our code, please cite us as follows: