Speaker
Francesca Poli
Description
See the full Abstract at http://ocs.ciemat.es/EPS2018ABS/pdf/P4.1078.pdf
How predict-first will change our approach to experimental planning
F. Poli1, B. Grierson1, M. Podestà1, Z. Wang1, J. Ferron2, C. Holcomb3, B. Victor3, K. Thome4
1
Princeton Plasma Physics Laboratory, NJ 08543, USA
2
General Atomics, San Diego, CA 92121, USA
3
Lawrence Livermore National Laboratory, Livermore, CA 94551, USA
4
Oak Ridge Associated Universities, Oak Ridge, TN, 37831, USA
Time-dependent, transport simulations of experiments ahead of time can improve the
efficiency of our experimental studies and might become a game changer. High-fidelity,
validated models are critical for the success of the predict-first approach, which relies entirely
on the fidelity of the models used to evolve transport and magnetic equilibrium.
We are going to show an example of successful application of the predict-first approach on
DIII-D, for the optimization of access to steady-state operation with sustained high qmin at
mid-radius. A feed-forward scheme has been proposed using free-boundary time-dependent
simulations with TRANSP, combining EC and NBI injection in the ramp-up to delay the
relaxation of the safety factor profile. Simulations indicate that a combination of Electron
Cyclotron heating and current drive for pre-heating in L-mode and Neutral Beam injection
sustain a broad and flat safety factor profile in the flattop phase, which has been a posteriori
verified in the experiment, a successful demonstration of the predict-first approach towards
experimental planning and optimization of runtime resources.
The limits of predictive modeling will be discussed and examples will be provided to show
what assumptions are critical for the success of the predict-first approach. In particular, self-
consistent particle transport, with realistic feedback control on the line-averaged density, like
it is done in experiments, is critical for experimental projections.
These examples provide directions for improvement of modeling capabilities in time-
dependent solvers.
This work is supported by the U.S. Department of Energy, Office of Science, Office of Fusion Energy
Sciences under contract numbers DE-AC02-09CH11466 and DE-FC02-04ER54698.