Project BALANCE - Data Center Sustainability Optimization

Inspiration

We wanted to use geographical data, with local energy usage, water scarcity, renewable energy availability, and distance to urban centers to generate a sustainability score for siting.

What it does

Generates a sustainability index for predicted sites of data centers, and creates/identifies optimized hotspots

How we built it

Using polygon shapefiles to model the US, and geographical datasets for energy, water, carbon, urban populations.

Challenges we ran into

Issues with noise: no set guidelines, no set zoning laws, very minor public reported databases. Can use extreme to set guideline - zoning laws set minimum at 200ft, in north VA highest load of data centers - sounds heard up to 2-3miles away. Low frequency noise (hum) that propagates further than normal noise from say a stereotypical powerplant or other noise pollution.

Issues with water: really no set reporting, only projections

Issues with %renewable energy: Only way to really get avg peak load is directly from the utility companies, and they won't share without consent from the user themselves. So we have to use the estimated power usage of each data center, which can be either higher or lower, divide by renewable energy data set - less accurate, but will be good enough for prediction

Issues with efficient energy usage - no defined industry standard. Only available data really is MAG7 companies trying to plug their ESG movements, not govt verified, no regulations, not applied to every data center - obviously less prominent ones will have worse efficiency.

Accomplishments that we're proud of

Managed to compile a perfect dataset tor on models on

What we learned

never trust the completeness of a data set

What's next for Data Center Sustainability Optimization

Next steps: as the metrics for individual data centers is not made publicly available, and is not accessible by us, we propose that our sustainability index can be used as a baseline for data center owners - a measure of expected sustainability for a data center, which can then be modified with true values of renewable energy usage, water usage, carbon emissions, and noise pollution.