TRANSP NBCD bug Clive Challis With thanks to

TRANSP NBCD bug Clive Challis With thanks to: Doug Mc. Cune and Yuriy Baranov 1 Clive Challis - TRANSP User Group Meeting - 21 May 2004

Fast Ion Current • Beam current driven by tangentially injected fast ions • Fast ions that can circulate toroidally provide a current which depends on: – – Beam energy Number of fast ions Toroidal velocity component Fast ion slowing-down time beam neutrals ionisation plasma • Fast ions can be trapped in B field gradient if v||/v too small – Trapped if |v||/v |<(2 )0. 5 (approx – Wesson, =r/R) – Trapped ions do not drive current 2 Clive Challis - TRANSP User Group Meeting - 21 May 2004 fast ions

Back Electron or Shielding Current • Thermal electrons dragged by collisions with fast ions • Bulk electron rotation given by momentum balance with fast and thermal ions • For: – Zbeam=Zplasma; Mbeam=Mplasma – ve>>vi – R>>r (no particle trapping) thermal ions (nion) thermal electrons (ne, <ve>) • Momentum balance gives: – nfast(v||fast-<ve>)-nion<ve>=0 – Jnet=nfastev||fast-nee<ve> – ne=nion+nfast Jnet=0 3 Clive Challis - TRANSP User Group Meeting - 21 May 2004 fast ions (nfast, v||fast)

Net Current • Thermal electrons can be trapped in B field gradient if v||/v too small – Trapped electrons do not carry current – Trapped electrons drag on rotating thermal electron – Shielding current reduced by factor 1. 0 -1. 46 0. 5 (approx – Wesson) • electron-ion coulomb cross-section scales like Z 2 ion, so higher Z ions preferentially drag on electrons to prevent complete cancellation of fast ion current • Resulting net current scales roughly as: – Jnet~Jfast{1. 0 -(Zfast/Zeff)(1. 0 -1. 46 0. 5)} • For Zbeam=1, high Zeff and/or large advantageous 4 Clive Challis - TRANSP User Group Meeting - 21 May 2004

TRANSP NBCD Bug • TRANSP Zfast used to calculate the shielding current was accidentally reset by mixed H & He beams, or non-NBI fast ion calculations (e. g. fusion products, including He) • When alpha-simulations were switched on (typical for JET, but not for MAST) the beam driven current was miscalculated using Zfast=2 rather than 1 to give: – Jnet~Jfast{1. 0 -(2/Zeff)(1. 0 -1. 46 0. 5)} • This reduces the current significantly in the core of low Zeff plasmas, but the effect diminishes at large • On-axis current drive vanishes for Zeff 2, and the current profile becomes hollow (normally peaked for JET) 5 Clive Challis - TRANSP User Group Meeting - 21 May 2004

Typical Case • In this typical JET ITB plasma the TRANSP neutral beam driven current profiles are compared with and without the bug at different times • Better agreement with PENCIL is achieved without the bug • The corrected TRANSP calculation is better than PENCIL due to superior physics model • In this case the bug resulted in a 40% reduction in total beam driven current 6 Clive Challis - TRANSP User Group Meeting - 21 May 2004

History and Checks • Doug Mc. Cune, who promptly located the bug from my symptoms, has determined the history of the bug: – Introduced 18 March 2002 approx (30 May 2002 at JET - Jim) – Removed 19 April 2004 approx • For 2004 runs the absence of the bug can be confirmed by the presence of the following signals in the multiplot JBFACS: – JBFAC, JBFACZ 1, JBFACZ 2 (as opposed to JBFAC, JBFAC 0, JBFACNC with the bug) • Alternatively, the shielding current calculation can be checked directly: – JBFAC( =0)=1. 0 -Zbeam/Zeff (Zbeam=2 for H, D or T beams being the fault condition) 7 Clive Challis - TRANSP User Group Meeting - 21 May 2004