Initial CR model coupling #5
Draft
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…scharge code. The CR model is a high-fidelity plasma kinetics model that allows you to choose the number of excited levels to be included in the model. For now, we assume a Maxwellian EEDF. The jaccobian is evaluated numerically. The problem has been vectrorized to increase performance. The scheme is fully implicit which may allow for larger timesteps.
NB: Due to a bug fix in the elastic collision term, the 6 species reference solution was updated.
Modify path settings in `chebSolver.py` so that we are able to run from arbitrary place but still set python search paths appropriately. Is this the best way? I don't know.
…acobian only once at each time step. This lead to a performance speedup of 2.5. The jacobian doesn't change much so there was no need to evaluate it at each iteration in the Newton-Raphson loop. NB: This feature is optional (see the input flag `freeze_jacobian` in the `step` method of the `timeDomainCollocation` class).
…is the glowDischarge model coupled with the CR model. Now, the model can run in both CPUs and GPUs.
….dot. This lead to significant speed up.
…s and the rhs in the CR model.
…ved small values and the exp(ion/Te) gave overflow.
…vity at Te <2 ev. That caused many problems at the sheath. The code used to crash at the edges. Now is more stable.
…ation for the EEDFs. I tested the periodic solver and it works fine. I needed to remove some statements like U2[U2<0] = 0 because they were causing the periodic solve to fail. The time marching case worked ok.
…5 degrees instead of 10. I have made also some changes in the spatial_jacobian where I added the if staements with the charge[i] !=0.0. I have added some terms SEC_U that we forgot to add although their contribution is negligible.
…rgy transport coefficient are read functions of Te read in the case set-uo file. It failed to reach a converged solution so I need to come back to that. There might be a bug.
…fficients. I aslo added the derivative of diffusion coefficient wrt temperature when Einstein relation is employed with variable mobility. This term was missing and was causing convergence issues.
…cients and get smoother gradients. Stiff gredients were causing convergence issues. Now, I managed to get a converged solution. I added variable Ion Molbility (function of Te from Raja's paper).
…an read crs from Biagi along with BSR. The names of the species still need to follow Racah notation. I set the code up to run cases with 30 excited states.
…s in the CRModel. I added diffusion losses in the radial direction.
…e changes in the mobility_U and diffusivity_U functions. I added more files for the EEDFs. I added files for cross-sections from different databases.
…all transitions that I have information for the oscillation strength. Ar(i) + Ar(1) <-> Ar(j) + Ar(1).
… Unfortunatly, this causes convergence issues with the periodic solver. This commit changes the default ion flux BC implementation from weak (where the ion boundary flux is modified by the BC but the PDE is retained at the boundary point) to strong (where the PDE is discarded at the boundary point and replaced by the BC equation). To allow the original behavior to be recovered, the behavior of the `--weakbc` flag has been modified. Previously the `--weakbc` flag changed the electron flux BC implementation from strong to weak. However, in practice, this option was never used. Thus, it has been changed here so that `--weakbc` now modifies the ion flux BC. This enables the original behavior to be reproduced even though the default has changed.
…eat 3-body recombination cross sections in the CR model. I chnaged the default rtol and atol of the timePeriodicSolver. I added some comments for the BC in chebSolver.
…. This is the model by Gogolides 1992. You can activate the model with the IonEffEField flag. We aslo now treat ion flux BC strongly.
…) instead of Te. This is becasue EN recieves the highest values at the electrodes and mu_e is min at max EN. However, due to the electron energy BC, Te is min at the electrodes and max at the midle. This is causing some stability issues. From Bolsig we expect that Te increases monotonically with EN. This 1-1 mapping is not representative of the discharge along the interelectrode direction.
… is expressed as a function of Te. The BC of electron energy are changes and now we prescibe a flux which is more consistent with the BC for electron number density. The ion BC is treated strongly. We can now handle MOLECULES. You need to specify their Cp in the input properties file. I have added reactions for molecular argon ion Ar2+ and molecule excimers Ar2*.
…l background electron source term to prevent electron density from becoming too small or negative at the sheaths.
…dding an jac frequency var in an attempt to make the solver more robust. NB: jac_frequency defaults to 1 but can be reset via the command line as desired.
…he linear iterations. It's slightly faster.
…e last part where we solve one period we call a function called solve_one_period and uses fixed timesteps. I also added a flug for adaptive timesteping but it is not finihed yet.
…e transient part, not the shooting solver. I needed to clip the Te to prevent it from becaome zero or negative. The solver is more robust now but not sure of this Te clip can cause issues to the outer newton solver
…r so that it can find a good starting point that reduces the residuals.
… whether to use time adaptivity or fixed timesteps for the shooting method. Fixed timesteps is more stable. I haven't managed to get a solution with time adaptivity when using the shooting method.
…ion if the values drop below these floors. I added a PID controller for growing or shrinking the timestep. I hope this will make the solver more robust. Note that the clipping can be turned off using the `clip_state` flag in the `modelClosures` class.
…tables with transport properties of electrons and ions from the LXCat database.
1) Add 6 species and CR tests to single run script 2) Clean up output slightly
trevilo
force-pushed
the
cr-model-coupling
branch
from
75cfc1e to
96efb85
Compare
trevilo
force-pushed
the
cr-model-coupling
branch
from
ceabc75 to
9911e2d
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
The PR pulls in all the CR model developments.