How to Load the Coretransp CPO

Where to Put the Ds and Vs additional to the Fluxes

The CPO form for each transport channel contains diff_eff, vconv_eff, and flux as the three pieces of relevance to us.

As the simplest example we take the electron temperature channel (conductive heat flux). The entire section is coretransp%te_transp which is of type transcoefel. The contents are then given by the relevant section of euitm_schemas which is

type type_transcoefel
  real(DP),pointer  :: diff_eff(:) => null()
  real(DP),pointer  :: vconv_eff(:) => null()
  real(DP),pointer  :: flux(:) => null()
  type (type_offdiagel) :: off_diagonal
  integer  :: flag=-999999999
endtype

So for these the first three elements are all dimensioned to nrho which is the number of radial points on the turbulence code's grid. It has been decided (Cyprus 2011 Code Camp) that the IMP4 codes should not try to interpolate onto another CPO grid, because the transport modelling should be left with the flexibility (ie, take the core portion from a global code and the edge layer from an edge code which might have different physics). The off-diagonal elements are left blank by turbulence codes.

The next is the electron density channel (main particle flux). The entire section is coretransp%ne_transp which is of type ne_transp. The contents are then given by the relevant section of euitm_schemas which is

type type_ne_transp
  real(DP),pointer  :: diff_eff(:,:) => null()
  real(DP),pointer  :: vconv_eff(:,:) => null()
  real(DP),pointer  :: flux(:) => null()
  type (type_offdiagel) :: off_diagonal
  integer  :: flag=-999999999
endtype

which is just like the transcoefel except the first two elements have an extra index (the last one). This is dimensioned 1:3 and tells the transport solver which coefficient appears with convective heat fluxes: the multiplier is 0, 3/2, or 5/2, and the relevant D and V go into position 1, 2, or 3, respectively. It is possible to have two channels with different multipliers. The ExB transport has multiplier 3/2 due to the Poynting cancellation, but the parallel velocity component of magnetic flutter has multiplier 5/2. So the conductive D and V coefficients are added together into the te_transp channel, and the particle D and V coefficients from the ExB and flutter pieces are put into positions 2 and 3, respectively. Note that although the part of magnetic flutter transport arising from parallel current and radial magnetic fluctuations vanishes in the average, the ion parallel velocity is left over and can contribute in unusual situations.

The ion species are done just like the electrons, but the types are transcoefion and ni_transp with an extra second index in all variables, for the ion species index. For a pure plasma (e and i only), the dimension of the second index is nion = 1 and the density D and V now have a third index which is the one for the 0, 3/2, or 5/2 multipliers. All other considerations are as for the electrons.

Examples for how these should be loaded appear in ~bscott/public/HDF5_files/global/ and ~bscott/public/HDF5_files/fluxtube/ for global and fluxtube codes, respectively, with the filename being itmfluxes.f90