How Synthetic Data Can Power the Future of EV Charging
More people than ever are switching to electric vehicles (EVs) as they become more affordable and accessible.
Planning where to build EV charging stations and how to distribute energy efficiently remains a major challenge, largely due to the limited availability of detailed, real-world charging demand data.
That’s the problem the ORSOL-Synthetic Data team from Purdue University tackled in the Curiosity Cup 2025.
They came up with a brilliant solution: when real data wasn’t available, they generated realistic synthetic data instead.
Using SAS® Viya for Learners, the team tested different approaches to simulate EV traffic patterns.
They compared three methods of synthetic data generation, each with its own strength:
- Capturing peak traffic flows
- Preserving seasonal trends
- Reproducing daily usage patterns
TimeGAN, a generative adversarial network designed for time-series data, outperformed other models by producing the most realistic simulations of California traffic patterns, an especially relevant benchmark given California's leadership in U.S. electric vehicle adoption.
With this enriched dataset, the team was able to consider additional factors such as energy prices and weather, creating a far more complete picture of EV charging needs.
Their study confirmed that synthetic data isn’t just a workaround but a powerful tool for enabling smarter decisions on charging station placement, energy distribution, and dynamic pricing.
Team ORSOL-Synthetic Data was formed by Jaewon Cho, Alexa Noto, Ankitha Mallekav, and Akanksh Rao, with faculty advisor Sivaranjani Seetharaman. Congratulations!
Their full paper in PDF is available under the main headline of this article.
