Theory Weekly Highlights for December 2006

December 15, 2006

The newly developed theory of transient flow in a jet tube was applied, along with the 2-D CFD code “FLUENT”, to ITER-like massive gas injection (MGI) systems. Knowledge of gas delivery rates has enabled us to more accurately infer the particle mixing efficiency Ymix in DIII-D. With our new average-Z model, we find Ymix ~ 3% for argon. Mixing occurs predominately on the Thermal Quench (TQ) time scale, which may not be enough time to deliver “massive” quantities of electrons to suppress runaways in ITER. A general theme emerges on the efficacy of “standard” MGI. Soon after the gas valve opens, the plasma seems to collapse by the very first tenuous gas particles at the leading edge of the gas front that impinge on the plasma well before “full flow” in the jet tube is established. These “early arrivals” are responsible for triggering the thermal collapse. The collapse outruns the more sluggish jet and consequently limits further density increase. Assuming that the TQ time in ITER is ~10 ms, a pipe length of 4 m and valve diameter 20 mm, the number of electrons delivered to the plasma is a small fraction (~ 5%) of the critical number needed for runaway suppression. with either argon or hydrogen gas). New schemes with faster particle delivery rates are being considered for experimentation on DIII-D.

December 08, 2006

The Monte-Carlo Hamiltonian guiding center drift code ORBIT-RF has been upgraded by partially re-writing the Message Passing Interface (MPI) program, and has been successfully tested using up to 1 million test particles on the Linux cluster DROP at GA. The ORBIT-RF is one of the options for interfacing in the Simulation of Wave Interaction with MHD (SWIM) project. In order to provide a more accurate particle distribution function, simulations with multi million particles are required and the recent simulations are a first step toward that goal. The simulation for 1 million particles took about 1 week using 40 nodes. It shows improved statistics, such as better accuracy and a much smoother particle distribution function. For more efficiency in computation time, ORBIT-RF is also ported to the supercomputer Cray XT-3, Jaguar, at ORNL. This will allow massive computations within a reasonable wallclock time limit. Testing on Jaguar is underway.

Dr. Valerie Izzo, a research scientist from UCSD, is our newest on-site long-term collaborator and will be working on nonlinear extended MHD in close cooperation with DIII-D experimentalists.

December 01, 2006

Modifications were completed to the GATO ideal MHD stability code to eliminate the stable Alfven continuum by replacing the full kinetic energy norm by a norm that includes only the normal component of the displacement. Previously, the continuum, which is numerically destabilized by the Finite Hybrid Element (FHE) Method, was restabilized by addition of a numerical correction term (see Highlight from April 07 2006). This was a prerequisite to modifying the kinetic energy norm to remove the continuum modes completely; otherwise the use of the FHE method results in unacceptable spurious unstable modes appear in the spectrum for all mesh sizes (i.e. spectral pollution). With the new energy norm and the numerical correction term, initial tests indicate that the restabilized continuum is now fully eliminated with no sign of spectral pollution.

Dr. S.C. Guo from Consortzio RFX, Italy, is currently visiting GA for a month to work with M. S. Chu on feedback stabilization of MHD modes in RFP.

Dr. M.S. Chance from PPPL is currently visiting GA for two weeks to work with M. S. Chu on feedback stabilization of resistive wall mode utilizing the normal mode approach.

These highlights are reports of research work in progress and are accordingly subject to change or modification