2,500-year-old board game inspires AI to tackle engine, data center overheating

A team of scientists led by Associate Professor Jiangtao Cheng at Virginia Tech has developed an AI-driven approach to optimize spray cooling, inspired by the ancient Chinese board game Go and Google’s AlphaGo AI. Recognizing parallels between Go’s interconnected strategic dynamics and the complex parameters of spray cooling systems, the researchers applied machine learning to analyze and predict the most effective cooling strategies. Their goal is to improve thermal management for electrical grids, data centers, engines, computers, and turbines, helping these systems operate efficiently amid rising demand and prevent overheating. The research, published in the journal Artificial Intelligence Review, focuses on the role of water droplets in spray cooling, where rapid evaporation of droplets carries away heat from hot surfaces. The team used AI to analyze data from 25 prior studies, evaluating factors such as optimal droplet size, spray nozzle types, and potential alternatives to water like solvents or engineered mixtures. This machine learning approach allowed them to better understand the thermo-fluid dynamics involved and to propose more effective cooling designs.