Problem description

Simulations involving IonFlow have some remaining convergence issues: e.g. the example ion_burner_flow.py packaged with 2.4 shows poor convergence (with a potential band-aid fix in #912). Additional issues surfaced while reviewing the code. Overall, the following should be addressed:

• Convergence of simulations appears to depend on domain size, which is not intuitive
• The Python attribute IonFlameBase.electric_field_enabled (as well as the C++ counterparts) has no effect on the actual solution - the implementation is incomplete as the C++ vector m_do_electric_field is set but never used.
• It is unclear why the Poisson equation shouldn't be solved at individual grid points as apparently intended by the boolean vector IonFlow::m_do_electric_field (is this physical?)
• It is not possible to impose an external electric field (the upstream residual enforces zero electric field at the inlet)
• The solution stage approach is not consistent with simulations of other effects (e.g. soret is enabled by a boolean flag): now that there are no longer three solution stages (as in early IonFlow versions), simply toggling the electric_field_enabled switch should suffice.

System information

• Cantera version: 2.5.0a4
• OS: any
• Python/MATLAB/other software versions: any

Attachments

Issues: #909, #911 (superseded by this issue), #912 (most relevant changes were transferred to #905)