Tokamaks loose heat faster than our best gyrokinetic models suggest they should. It’s a mystery. In the last post I talked about the difference between MHD and gyrokinetics. Here I’m going to say a little about some recent developments.

A few weeks ago two papers by the same group of authors popped up.

  1. Multi-scale gyrokinetic simulation of tokamak plasmas: enhanced heat loss due to cross-scale coupling of plasma turbulence. link
  2. Synergistic cross-scale coupling of turbulence in a tokamak plasma. link

The big news here is that the researchers have tweaked an existing piece of gyrokinetic software called GYRO to model the turbulence of both ions (Which is typically what people study) and also electrons. They found that adding electrons into the mix drastically increases heat flow through the plasma. Apparently the electrons form structures called streamers which radiate out and join the electron turbulence to the ion turbulence.

You might be wondering why, if plasma is a mix of ions and electrons, had people not previously tried simulating electron turbulence alongside ion turbulence. Well they had.

But in an effort to simplify the computations they set the masses of the ions and electrons closer than they actually are. In atomic mass units the electron has a mass of


so for a deuterium plasma the ion/electron mass ratio should be



Earlier work used values anywhere form 400 to 1800.

Adding in a realistic mass ratio came at a cost. These simulations took 37 days each! And that’s running on a 17,000 processor supercomputer. The NERSC Edison. For a grand total of 100 million CPU hours.

Previously there had been doubt that electron turbulence could influence ion movement in a big way because the gyro-radius for electrons is so much smaller due to their lower mass. See the previous post for the details.

Why this matters

This work is important. We have known that heat moves through plasmas faster than predicted for years. These papers seem to have explained why.

The transport of heat through a plasma decisively influences one of the most important parameters in fusion physics; the energy confinement time usually denoted by the Greek letter tau. The energy confinement time measures how long heat remains in the plasma before conducting through its surface into the walls of the reactor. If the confinement time is too short the plasma will cool down and fusion reactions will stop. If the confinement time is too long… No one has ever had that problem.

So if we want to correctly model the heat flow in ITER or the Wendelstein x7 these results might come in handy. It’s always nice to know what’s going to happen when you switch one of these things on.


[Image courtesy of JPL A rare type of atmospheric wave on Jupiter:  NASA/ESA/GSFC/UCBerkeley/JPL-Caltech/STScI]