Improvements to Shape Control Using Real-Time Optimization of Feedforward Coil Current Trajectories

POSTER

Abstract

Shape control in many tokamaks relies on offline estimates of the poloidal field coil currents needed to produce the desired plasma shape, combined with a set of PID controllers to correct the shape in real time. Due to deficiencies in the predictive capabilities for factors such as plasma resistivity, these offline estimates often require significant feedback adjustments requiring high integral gains, leading to potential windup issues; this is even more pronounced in long pulse tokamaks such as KSTAR and EAST. To lessen this burden, the Real-Time Feedforward algorithm was developed to periodically recalculate the optimal feedforward coil currents during the shot based on updated plasma resistivity measurements. This algorithm was previously tested on KSTAR and was shown to significantly decrease the required feedback corrections. In this work, the Real-Time Feedforward algorithm has been improved to take into account changes in the plasma current distribution throughout the shot, as well as to adapt to changes in the shape target in real time. This updated version has been shown to be effective in simulation for both KSTAR and EAST. Work is currently ongoing to implement this change in the PCS for both KSTAR and EAST, and experiments to test the updated algorithm are planned for later this year.

*Supported by the US DOE under award numbers DE-SC0023399 and DE-SC0010685.

Presenters

  • Shira Morosohk

    Oak Ridge Associated Universities

Authors

  • Shira Morosohk

    Oak Ridge Associated Universities

  • Jayson L Barr

    General Atomics - San Diego

  • Michael L Walker

    General Atomics

  • Nicholas W Eidietis

    General Atomics

  • Andrea M. Garofalo

    General Atomics

  • Sang-hee Hahn

    Natl Fusion Res Inst

  • Bingjia Xiao

    Hefei Institutes of Physical Science

  • Zhengping Luo

    ASIPP

  • Qiping Yuan

    ASIPP