Skip to content

Port compositional freeflow model and tests to new staggered

Ned Coltman requested to merge feature/compositional_freeflow into master

passing tests:

  • maxwellstefan (changed base epsilon)
  • Add back the maxstefan gnuplot output
  • channelnc mole diffusion
  • channelnc mass diffusion
  • channelnc advection
  • channelnc advection nocaching
  • channelncni advection
  • channelncni diffusion (replaced reference solution)
  • density driven flow

The ported channelncni-diffusion test doesn't pass with the old reference solution (using fieldcompare package), with errors in X^Air_liq, x^Air_liq, deltaP, velocity_liq (m/s). As fieldcompare seems to output only the error of the first violation of thresholds, the following error magnitudes are computed using the legacy comparison backend (note that with the legacy backend also the advective tests fail with errors in velocity_liq (m/s), see comment).

field X^Air_liq x^Air_liq deltaP velocity_liq (m/s)_0 velocity_liq (m/s)_1
relative 4.41% 4.41% 100% 189.16% 189.16%
reference 5.73852e-05 3.56985e-05 0.0 -2.05625e-13 7.43444e-15
test 5.48522e-05 3.41227e-05 1.45519e-11 1.83336e-13 -6.67723e-15
Edited by Mathis Kelm

Merge request reports