Initial low Mach reacting flow capabilities #281

* rough skeleton of new features added but nothign hooked up * added mixture, transport, and chemistry instantiation in reactingFlow constructor * added rx to assert check:

…OT TESTED

* runs but temp does not converge in first step * valgrind errors galore

…es, and Rgas init error

…lem with production

The primitive state was being filled with mass fractions for the species entries and then passed to GetConservativesFromPrimitives, but this function expects that the species entries of the primitive state contain molar densities. Because of this, the returned conserved state was incorrect and the resulting molar densities returned by computeNumberDensities were also incorrect, which then made the species production terms incorrect.

Three bug fixes here: 1) The units of the formation energy from the input file is J / mol, so we need to divide by the species weight (kg / mol) here to get the formation energy per unit mass. 2) The vector hw_ was sizes to have entries for all species and the contributions were not summed. We need hw_ to be a scalar field with the contributions from all species summed. 3) Sign error Note that this commit does not add the sensible enthalpy contribution (see Poinsot and Veynante, Theoretical and Numerical Combustion, 2005, eq. 1.62).

Species history update was out of order.

to energy eqn rhs in ReactingFlow::heatOfFormation. Not general b/c Cv is hardcoded. Just for investigation. Don't merge this.

…t vs full dof bugs

Three bugs were found and fixed while working on the low Mach analog to argon_minimal.binary.test. The fixes are 1) In ReactingFlow::speciesStep, set the initial guess for the iterative linear solve to the mass fraction for the current species from the previous time step. Because of how Yn_next_gf_ is managed, that was not the case and sometimes the electron species solve in particular struggled because of it. 2) Multiply the diffusivities returned by computeMixtureAverageDiffusivity by the mixture density (i.e., so that the diffision term in the species equation is div( \rho D \grad Y) ). 3) Populate the species entries of the primitive state in ReactingFlow::updateDiffusivity with the mole densities, rather than the mass fraction.

Two mods: 1) Set velocity to 0. Because of a bug, the test was generated with 0 velocity rather than what the input file said. 2) Tighten linear solver tolerance.

Was always too tight. Bug fix in LQ_form_ DiffusionIntegrator coefficient in CaloricallyPerfectThermoChem was enough to push it past tolerance. Given linear solver tolerances, updated values seem reasonable.

Need to check with shaering about this.

It appears that the computeMixtureAverageDiffusivity functions have tripped up the cuda support for virtual functions, as we have encountered prieviously (e.g., #203). For the moment, I hack around the problem by introducing an unused argument in computeMixtureAverageDiffusivity, which makes its signature sufficiently unique that the problem doesn't occur.

To more closely match the original compressible path test.

To better match original compressible path test

We don't (yet) support ambipolar (quasi-neutral) or two-temperature on the low Mach reacting path. Die if detected.

…ion term now consistent with Poinsot book eq 1.67.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Initial low Mach reacting flow capabilities #281

Initial low Mach reacting flow capabilities #281

Commits on Jun 4, 2024