MIT researchers have unveiled a prediction method that blends a physics-based plasma model with machine learning to manage tokamak rampdowns more safely and reliably. The approach targets disruption avoidance when plasma current is reduced, a critical step for future grid-scale fusion plants.
Tested on Switzerland's TCV device using several hundred plasma pulses, the hybrid model accuratel by Jennifer Chu, MIT News
Cambridge MA (SPX) Oct 09, 2025
MIT researchers have unveiled a prediction method that blends a physics-based plasma model with machine learning to manage tokamak rampdowns more safely and reliably. The approach targets disruption avoidance when plasma current is reduced, a critical step for future grid-scale fusion plants.
Tested on Switzerland's TCV device using several hundred plasma pulses, the hybrid model accurately forecast plasma evolution during rampdown with relatively little data. Only a couple hundred low-performance pulses and a small handful of high-performance pulses were sufficient for training and validation.
An accompanying control algorithm translates predictions into actionable trajectories for magnets and heating, enabling automated, stable shutdowns. On TCV, the team demonstrated rampdowns that were faster and free of damaging instabilities compared with standard procedures.
"For fusion to be a useful energy source it's going to have to be reliable," says lead author Allen Wang of MIT's Plasma Science and Fusion Center Disruption Group. "To be reliable, we need to get good at managing our plasmas."
"At some point the plasma will always go away, but we call it a disruption when the plasma goes away at high energy. Here, we ramped the energy down to nothing," Wang notes. "We did it a number of times. And we did things much better across the board. So, we had statistical confidence that we made things better."
Co-authors include Cristina Rea (PSFC), Oswin So, Charles Dawson, and Chuchu Fan (LIDS), Mark (Dan) Boyer (Commonwealth Fusion Systems), and collaborators from the Swiss Plasma Center. Support came from CFS, EUROfusion via the Euratom Research and Training Program, and the Swiss State Secretariat for Education, Research, and Innovation. The work aligns with CFS's SPARC pathway to compact, net-energy fusion.
Research Report:"Learning Plasma Dynamics and Robust Rampdown Trajectories with Predict-First Experiments at TCV"
Related Links
Massachusetts Institute of Technology
Powering The World in the 21st Century at Energy-Daily.com