Advancing Fusion Theory
PPPL members of the H-mode team: Physicists Seung-Hoe Ku, Robert Hager, Choong-Seock Chang, and Randy Michael Churchill
Unlocking the mystery of high confinement that sustains fusion reactions
PPPL scientists have unlocked the mysterious process that bottles up plasma in doughnut-shaped tokamaks that confine the plasma in powerful magnetic fields. The new understanding models the sudden transition of plasma from low to high confinement, marking a critical step toward creating “a star in a jar” by capturing fusion energy.
Physicists have long sought to explain the workings of this transition to create it on demand in present and future tokamaks. Researchers at PPPL have now produced the first complete model of the principles behind the process, which begins with the swirling of plasma at the edge of the fusion facility. When heated to high enough temperature, the turbulent edge plasma spontaneously enters a high-confinement (H-mode) phase that sustains fusion reactions.
The simulation took three days and full capacity of Titan, the fastest U.S. supercomputer, which can perform 27 million billion (10^15 ) calculations per second. “After 35 years, the fundamental physics has been simulated, thanks to the rapid development of the computational hardware and software,” said PPPL physicist C.S. Chang, leader of the nationwide team that developed the extreme-scale code and produced the model.
Full understanding of the spontaneous transition will be essential for ITER, the international tokamak under construction in France to demonstrate the feasibility of fusion power. Operators of the massive seven-story, 23,000-ton machine must achieve H-mode to reach the goal of producing 10 times more energy than it will take to heat the plasma. Moving to H-mode will keep the turbulent edge from cooling the center of the plasma and slowing or halting fusion reactions.
Coming enhancements of the code will become part of the Exascale Computing Project, a nationwide program to develop computers that will run up to 100 times faster than today’s fastest machines, improving U.S. security, economic competitiveness and scientific capability. PPPL leads an initiative that will develop the first complete model of an entire fusion plasma, deepening understanding of the magnetically confined matter that could fuel a promising new era of energy production.
The blob that ate the tokamak: how bubbles drain heat and reduce fusion efficiency
Like rapidly boiling water, superhot plasma produces numerous bubbles, or blobs. These blobs arise at the edge of the plasma and carry off heat that sustains fusion reactions.
A team based at PPPL has now produced simulations that provide new insight into how the troublesome blobs behave. These simulations, using a PPPL-developed code called“XGC1,”depict two different regions of the plasma edge simultaneously, creating a fuller picture of how heat escapes from the plasma and can damage tokamak walls. Such a picture could lead to improved control of the problematic blobs,
“In simulations, we often separate two areas at the plasma edge known as the pedestal and the scrape-off layer and focus on one or the other,” said PPPL physicist Michael Churchill, lead author of a paper describing the findings in the journal Plasma Physics and Controlled Fusion. “XGC1 is unique because it is able to simulate both regions simultaneously, showing how they affect each other.”
Researchers used the code to simulate plasma in high-confinement mode, or H-mode, a state that helps plasma retain its heat. Results showed how a large number of blobs form in H-mode between the pedestal and the scrape-off layer and move towards the outer edge and the wall, carrying vital heat with them. “The big picture is that blobs can pull energy and particles out of the plasma, and you don’t want that,” Churchill said. “These simulations can help us understand the process and potentially how to avoid it.”
Future research will focus on the formation of the blogs and how the design of a tokamak can affect their behavior. Also to be determined is the impact on blobs of the density and temperature of the plasma and the electromagnetic forces inside the tokamak.
Physicist Michael Churchill
Simulation of plasma turbulence generating positive (red) and negative (blue) stress that drives sheared rotation Inset: comparison between measured and simulated rotation profile
Physicist Brian Grierson at DIII-D
Discovering a surprising self-organized flow of fusion plasmas
Simulations of experiments on the DIII-D National Fusion Facility that General Atomics operates for the U.S. Department of Energy in San Diego could open the door to improved control of fusion reactions in ITER, the international experiment under construction in France. The simulations, led by PPPL physicists Brian Grierson and Weixing Wang, revealed a surprising self-organized flow of the superhot plasma that fuels fusion reactions — a flow that could improve the stability and performance of fusion devices.
The new research found that sufficient heating of the core of the plasma generates a special type of turbulence that produces a twisting force that causes regions of the plasma to rotate in stability-enhancing sheared, or opposite, directions. Researchers have traditionally produced such shearing by injecting high-energy beams of neutral atoms into the plasma. The new findings are directly relevant to future reactors, since neutral beam injection will create only limited rotation in the huge plasmas inside such facilities.
A key next challenge will be to extrapolate the heating process for ITER. Such modeling will require massive simulations that will push the limits of the high-performance supercomputers currently available. “With careful experiments and detailed simulations of fundamental physics, we are beginning to understand how the plasma creates its own sheared rotation,” said Grierson. “This is a key step along the road to optimizing the plasma flow to make fusion plasmas more stable, and operate with high efficiency.”
A plasma instability that can make itself vanish
A type of instability called edge-localized modes (ELMs) occurs in short bursts in plasma and can damage the inside of tokamaks that house fusion reactions. Now physicist Fatima Ebrahimi has used advanced models to simulate the cyclical behavior of ELMs, a development that could lead to improved understanding of the behavior of plasma and help scientists more reliably produce plasmas for fusion reactions. The new findings could also provide insight into solar flares, the eruptions of enormous masses of plasma from the surface of the sun into space.
ELMs occur around the outer edge of high-confinement, or H-mode, plasmas due to strong currents that develop at the edge. Ebrahimi showed how ELMs go through a repeated cycle in which they form, develop, and vanish. “This research both reproduces and explains the burst-like, or quasi-periodic, behavior of ELMS,” she said. “If ELMs occur in large tokamaks in the future, the bursts could damage plasma-facing components, and understanding them could help prevent that damage.”
The model demonstrates that ELMs can form when a steep gradient of current exists at the plasma edge. The instability then forms a filament, or current-carrying thread, that moves around the tokamak producing electrical fields that interfere with the currents that caused the ELMs to form. With the original currents disrupted, the ELM dies. “In a way,” Ebrahimi said, “an ELM extinguishes itself by its own motion.”
Ebrahimi’s findings are consistent with observations of the cyclic behavior of ELMs in tokamaks around the world. Her next step will involve investigating the impact of differences in plasma pressure on the cyclic behavior of ELMs.
Fatima Ebrahimi
Deep learning developers. Standing: Willliam Tang, left, and Eliot Feibush, with Alexey Svyatkovskiy
Artificial intelligence helps predict disruptions that can halt fusion reactions
Before scientists can effectively capture and deploy fusion energy, they must learn to predict major disruptions that can halt fusion reactions and damage the walls of tokamaks. Timely prediction of disruptions, the sudden loss of control of the hot, charged plasma that fuels the reactions, will be vital to triggering steps to avoid or mitigate such large-scale events.
Researchers at PPPL and Princeton University are employing artificial intelligence to improve predictive capability. Led by William Tang, a PPPL physicist and lecturer with the rank of professor at Princeton University, the scientists are developing a computer code for predictions for ITER, the international experiment under construction in France.
The new software, called the Fusion Recurrent Neural Network (FRNN) code, is a form of “deep learning” — a newer and more powerful version of modern machine- learning software, an application of artificial intelligence. “Deep learning represents an exciting new avenue toward the prediction of disruptions,” Tang said. “This capability can now handle multi-dimensional data.”
Members of the PPPL and Princeton team are the first to systematically apply a deep learning approach to the problem of disruption forecasting in tokamak fusion plasmas. Using this approach, the team has demonstrated the ability to predict disruptive events more accurately than previous methods have done.
The team now aims to reach the challenging goals that ITER will require. These include producing 95 percent correct predictions when disruptions occur, while providing fewer than 3 percent false alarms when there are no disruptions. Members of the team include Eliot Feibush, a computational scientist at PPPL, Alexey Svyatkovskiy, a Princeton big data researcher, and Julian Kates-Harbeck, a graduate student at Harvard University and holder of a Computational Science Graduate Fellowship sponsored by the U.S. Department of Energy, who was chief architect of the FRNN code.