AI slashes the time needed to design better heat-harvesting devices
Researchers at the National Institute for Materials Science in Japan have developed an AI tool called TEGNet that predicts the performance of thermoelectric generators with over 99% accuracy while reducing computational design time by roughly 10,000 times. By using machine learning to emulate complex physics, the system enables rapid testing of thousands of device configurations in minutes instead of months. TEGNet's modular design allows it to simulate advanced architectures by combining models of individual materials. The approach led to the creation of two highly efficient prototypes with energy conversion efficiencies of 9.3% and 8.7%.
Opening excerpt (first ~120 words) tap to expand
April 28, 2026 report AI slashes the time needed to design better heat-harvesting devices by Sam Jarman, Phys.org Sam Jarman contributing writer Meet our staff & contributors Learn about our editorial standards edited by Sadie Harley, reviewed by Robert Egan Sadie Harley scientific editor Meet our editorial team Behind our editorial process Robert Egan associate editor Meet our editorial team Behind our editorial process Editors' notes This article has been reviewed according to Science X's editorial process and policies. Editors have highlighted the following attributes while ensuring the content's credibility: fact-checked peer-reviewed publication trusted source proofread The GIST Add as preferred source Testing thermoelectric materials for optimal performance. Credit: Airan Li et al.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at Phys.org.