Non-ideal capabilities for 1-D flame model #1079

gkogekar · 2021-07-29T15:22:43Z

Changes proposed in this pull request

Add non-ideal capabilities to the existing Flame models. Currently, the flame models utilize only ideal gas phase. There has been some discussion in the user group and the feature request (Cantera/enhancements#109) to include 1-D supercritical features in the Flame code. This pull request implements the real-gas flame models (R-K EoS, P-R EoS) along with the modified transport.

Modify StFlow class to include real-gas thermophases (Redlich-Kwong, Peng-Robinson EoS)
Modifications to the Transport / HighP transport models (if required)
Modifications to Python interface
Cantera tests to check real-gas flame capabilities

References
Cantera/enhancements#109

Theory Document
Non_ideal_Flame_equations.pdf

speth · 2022-05-23T22:57:50Z

@gkogekar - If you can, please undo the merge commit and rebase this onto the current main branch instead.

codecov · 2022-06-07T18:16:43Z

Codecov Report

Merging #1079 (a34b39c) into main (e458764) will increase coverage by 0.02%.
The diff coverage is 94.73%.

@@            Coverage Diff             @@
##             main    #1079      +/-   ##
==========================================
+ Coverage   70.44%   70.46%   +0.02%     
==========================================
  Files         375      375              
  Lines       58401    58405       +4     
  Branches    20897    20899       +2     
==========================================
+ Hits        41142    41158      +16     
+ Misses      14238    14227      -11     
+ Partials     3021     3020       -1

Impacted Files	Coverage Δ
src/oneD/StFlow.cpp	`84.16% <94.44%> (+0.23%)`	⬆️
include/cantera/oneD/StFlow.h	`100.00% <100.00%> (ø)`

... and 4 files with indirect coverage changes

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

bryanwweber · 2022-06-21T03:19:23Z

Thanks @gkogekar, I just saw this was ready for review. It looks like there aren't any cases tested where you might expect to see some effect of the non-ideal EOS, can you please add something?

Similarly, the difference with the existing results is somewhat concerning to me. Things are changing in the third and fourth decimal place, (although I don't have a good sense for how much change would be expected). I guess I'm surprised there's so much difference by swapping the enthalpy for the temperature. Can you comment? Can you do a comparison of a simple calculation in the zero-D case to get a sense for how much difference to expect?

gkogekar · 2022-06-21T13:38:02Z

Thanks, @bryanwweber. I had derived the energy equation for non-ideal EoS (the document is attached above). But we decided to move forward with the finite difference approach (I don't exactly remember why. ) In this implementation, the enthalpy flux is calculated based on a first order finite differencing, as opposed to exact calculation of Cp value in case of ideal gas EoS. So the difference in the results is bound to happen depending on the grid size.

I am creating this PR as many people on the user group have asked for this functionality. But I haven't found any particular test case that will show the non-ideal effects. Since we are modeling flames (i.e. high temperatures), the gas-phase is always near ideal condition. Nonetheless, I will create a test case using a non-ideal EoS.

bryanwweber · 2022-06-21T14:29:17Z

Thanks @gkogekar! I think the finite difference makes sense when we don't know anything about the underlying phase, but I wonder if there's room to allow the evaluation to specialize, analogous to Reactor vs. IdealGasReactor? One way I can think of is to add a flag to the constructor like non_ideal_phase (I don't love that name, but I hope it conveys the idea) to allow the user to specify they want to use the finite difference of the enthalpy, which could default to false so that the ideal gas solution is the default. Another way is to auto-detect the phase type; I'm not sure if it's possible to auto-detect the class type since it's a pointer, but that'd be the best case if possible.

As far as examples, I agree that under high-temperature (combustion) conditions, it's really unlikely to see a difference here. That'd be a reasonable test case anyways, if you compare to the ideal gas result. You can also consider something like high-pressure catalysis against a surface, a liquid flow, or something like that.

src/oneD/StFlow.cpp

gkogekar · 2022-06-22T18:15:44Z

@bryanwweber Thanks for your insights. Retrieving the thermo type in StFlow class is possible. So I have just modified the energy equation code, where it will use finite difference only for non-ideal phases. It will use the existing energy residual equation in the case of ideal gas EoS. This is the simplest workaround based on your suggestions. But I can always add an external flag if necessary.

I will add the test for non-ideal EoS soon.

bryanwweber · 2022-06-22T18:22:10Z

But I can always add an external flag if necessary.

Thanks @gkogekar this is a really nice solution, I didn't think it was possible but I'm glad it is! Definitely not something to do here, but I wonder if it makes sense to eliminate some of the class hierarchy in reactors...

speth

Thanks for introducing this PR, @gkogekar. Overall, I think it shows that there isn't all that much required to handle the non-ideal phases in the 1D flame model. Besides the in-line comments below, there are couple of things needed for this PR:

Please eliminate the change to the googletest submodule
I think we need at least a minimal test case for this. I guess that will require adding an input file that has both transport properties and Redlich-Kwong or Peng-Robinson (or critical states) parameters, which I think we currently lack.
A Python example would be nice as well. I don't know what conditions would actually show an effect from the nonideality.
The "science" section of the Cantera website will need to be updated as well. I think for posterity, it would be nice to have a simpler derivation of the equations somewhere. The attached PDF is useful, but it has a lot of digressions and dead ends that aren't really necessary.

src/oneD/StFlow.cpp

include/cantera/oneD/StFlow.h

src/oneD/StFlow.cpp

ischoegl · 2023-03-21T23:17:49Z

Hi @gkogekar ... thanks for updating this PR. Unfortunately, it seems like #1452 introduced a couple of minor merge conflicts since the latest rebase. Apart from this, #1448 introduced a couple of changes that should make the Python side quite a bit easier. Would you mind rebasing so this can be pushed over the finishing line? I'd be happy to assist in case you're facing any hurdles.

gkogekar · 2023-03-22T01:18:15Z

@ischoegl Thanks for letting me know. I will rebase and resolve the merge conflicts.

ischoegl

@gkogekar ... thank you for the rebase, and for taking care of most of the review comments!

Apart from some minor comments for code style guidelines I left below, there currently are some issues with the test suite. As they're consistent across platforms, you can likely reproduce them locally?

*********************************** Testing Summary ************************************

FAILED: One or more tests failed.
Tests passed: 3447
Up-to-date tests skipped: 0
Tests failed: 301
scons: *** [test_results] Explicit exit, status 1

Failed tests:
    - clib.passed ***no results for entire test suite***
    - oneD.passed ***no results for entire test suite***
    - python ***no results for entire test suite***
    - cxx-flamespeed

****************************************************************************************

src/extensions/pythonExtensions.h

include/cantera/oneD/StFlow.h

src/oneD/StFlow.cpp

gkogekar · 2023-03-30T19:54:47Z

Thanks, @ischoegl. I have taken care of most of the failures from the test suite. But one test (Reaction.PythonExtensibleRate) from Kinetics still fails on my Windows machine, and I am not able to figure out the reason for that. But I can see some other tests are now failing on Ubuntu and Mac platforms as well. Is there any way to fix that?

speth

Thanks for updating this, @gkogekar -- I think it's looking pretty close. Please see below for a few suggestions.

data/h2o2.yaml

include/cantera/oneD/StFlow.h

src/oneD/StFlow.cpp

test/python/test_onedim.py

speth · 2023-06-17T16:07:07Z

ext/googletest

This change to the googletest submodule needs to be eliminated from the commit history. If you need help doing that, let me know and I can force-push an update to this branch that takes care of it.

I have no idea how to do that. I tried to search for the related commit, but failed to find it. Can you help me with that?

I just force-pushed the branch with this change here. Please fetch and reset your local branch to match this one (head is commit d655febfe).

@speth Where is this commit d655febfe? The last commit I see is d655feb at https://github.com/gkogekar/cantera/tree/nonIdealFlame

That's the same commit, just truncating the hash at a different point (the full hash is d655febfeebff0c01e6b44effe4b5d64ce4dacba)

Thanks, I just pushed those changes.

speth · 2023-06-17T16:11:23Z

test/python/test_onedim.py

+        gas = ct.Solution("h2o2.yaml", "ohmech-RK")
+        gas.TPX = T_in, 0.05 * ct.one_atm, comp
+        width = 0.2 # m


I think it would be useful to have this regression test at a pressure high enough where the non-ideal model has some effect (if that's even possible for a H2 flame -- I'm not entirely sure).

I have modified the test to run at 10 bar, but truly speaking the temperatures are very high. So there is no difference between the idea gas and real gas models.

You could go farther away from stoichiometry? At the same time, this is h2o2.yaml …

test/python/test_onedim.py

speth · 2023-06-22T16:07:13Z

Please don't introduce the merge commit by using git merge or git pull -- this just brings back in the problematic changes.

With your branch checked out, and with all uncommitted changes stashed or committed (or else they will be lost), you should:

Run git log and note the commit hashes of your commits after d655feb (e.g. cdf9e8c)
Run git reset --hard d655feb
Run git cherry-pick cdf9e8c and repeat for any other commits you want to keep.

The term (cp_k*dT/dt) in the energy equation is replaced by (dh_k/dt).

version

Cantera 3.0 introduces small changes for onedim calculations; as benchmark files were generated with Cantera 2.6, tolerances need to be relaxed.

gkogekar · 2023-06-22T18:17:01Z

Please don't introduce the merge commit by using git merge or git pull -- this just brings back in the problematic changes.

With your branch checked out, and with all uncommitted changes stashed or committed (or else they will be lost), you should:

Run git log and note the commit hashes of your commits after d655feb (e.g. cdf9e8c)

Run git reset --hard d655feb

Run git cherry-pick cdf9e8c and repeat for any other commits you want to keep.

Apologies for that. I am still not getting used to Git advanced commands. Anyway, I have corrected it and also rebased the code with the latest Cantera build.

speth

Thanks, @gkogekar, this looks good to me. With this completed, could you also open a PR for the cantera-website repository to document the updated 1D flame governing equations?

gkogekar · 2023-06-22T21:34:12Z

Thanks, @gkogekar, this looks good to me. With this completed, could you also open a PR for the cantera-website repository to document the updated 1D flame governing equations?

Thanks, @speth
Sure. I will do that.

ischoegl added the 1D Flames label Aug 12, 2021

ischoegl added help wanted and removed help wanted labels Feb 5, 2022

speth mentioned this pull request Mar 15, 2022

Fix Redlich-Kwong partial molar properties #1218

Merged

5 tasks

gkogekar closed this May 23, 2022

gkogekar deleted the nonIdealFlame branch May 23, 2022 22:04

gkogekar restored the nonIdealFlame branch May 23, 2022 22:34

gkogekar reopened this May 23, 2022

gkogekar force-pushed the nonIdealFlame branch from 3336726 to 373da9e Compare June 6, 2022 18:01

gkogekar marked this pull request as ready for review June 15, 2022 17:13

gkogekar force-pushed the nonIdealFlame branch 2 times, most recently from d51cd9c to e0c0768 Compare June 22, 2022 17:42

bryanwweber reviewed Jun 22, 2022

View reviewed changes

src/oneD/StFlow.cpp Outdated Show resolved Hide resolved

speth self-requested a review July 2, 2022 22:16

speth requested changes Jul 29, 2022

View reviewed changes

ischoegl mentioned this pull request Feb 28, 2023

OneD domain factories and cleanup #1445

Merged

5 tasks

gkogekar force-pushed the nonIdealFlame branch from e0c0768 to 6a22daa Compare March 3, 2023 20:46

ischoegl mentioned this pull request Mar 21, 2023

Make 1D flow domains agnostic of thermo/transport model Cantera/enhancements#165

Closed

gkogekar force-pushed the nonIdealFlame branch from 6a22daa to 38b6d3a Compare March 28, 2023 15:19

ischoegl reviewed Mar 29, 2023

View reviewed changes

src/extensions/pythonExtensions.h Outdated Show resolved Hide resolved

include/cantera/oneD/StFlow.h Outdated Show resolved Hide resolved

src/oneD/StFlow.cpp Outdated Show resolved Hide resolved

src/oneD/StFlow.cpp Outdated Show resolved Hide resolved

speth requested changes Jun 17, 2023

View reviewed changes

ischoegl reviewed Jun 20, 2023

View reviewed changes

test/python/test_onedim.py Outdated Show resolved Hide resolved

speth force-pushed the nonIdealFlame branch from 67405fd to d655feb Compare June 20, 2023 21:20

gkogekar force-pushed the nonIdealFlame branch from cdf9e8c to 64b685d Compare June 22, 2023 17:09

gkogekar and others added 18 commits June 22, 2023 13:36

Modify the StFlow class to allow non-ideal thermophase

3990a71

The term (cp_k*dT/dt) in the energy equation is replaced by (dh_k/dt).

Modification to energy equation based on thermo phase type

a75d6d7

Removing some redundant code as recommended in reviewer's comments.

4d5e7f5

Few modifications required after rebasing the code with current Cantera

0fc099b

version

Formatting changes

2e648da

Addition of global arrays m_hk and m_dhk_dz

099e9bd

Few formatting changes

36f3e0b

Removing the ideal gas energy equation code block.

578d870

Modifications to the test outputs

23dd1f9

[unittests] Relax tolerances for legacy benchmarks

14bbd2f

Cantera 3.0 introduces small changes for onedim calculations; as benchmark files were generated with Cantera 2.6, tolerances need to be relaxed.

Formatting changes (Limiting lines to 88 characters)

f5c06fb

Adding a python test for non-ideal flame model

33b161b

Updating the description for R-K parameters

b526c9b

Formatting changes for Doxygen-style comments

6b6f527

Removing redundant variables

0d5e885

Modification to non-ideal flame testcase to run at 10 bar

7617820

Updating 'missing_ok' flag for python>3.8

972fded

Removing the phase check

53c2c8f

gkogekar force-pushed the nonIdealFlame branch from 64b685d to 53c2c8f Compare June 22, 2023 17:39

Updating the failing test

a34b39c

speth approved these changes Jun 22, 2023

View reviewed changes

speth merged commit d717e8b into Cantera:main Jun 22, 2023
40 checks passed

speth mentioned this pull request Jun 22, 2023

Real gas EOS with 1-D flame solutions Cantera/enhancements#109

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Non-ideal capabilities for 1-D flame model #1079

Non-ideal capabilities for 1-D flame model #1079

gkogekar commented Jul 29, 2021 •

edited

speth commented May 23, 2022

codecov bot commented Jun 7, 2022 •

edited

bryanwweber commented Jun 21, 2022

gkogekar commented Jun 21, 2022

bryanwweber commented Jun 21, 2022

gkogekar commented Jun 22, 2022

bryanwweber commented Jun 22, 2022

speth left a comment

ischoegl commented Mar 21, 2023

gkogekar commented Mar 22, 2023

ischoegl left a comment

gkogekar commented Mar 30, 2023

speth left a comment

speth Jun 17, 2023

gkogekar Jun 20, 2023

speth Jun 20, 2023

gkogekar Jun 22, 2023

speth Jun 22, 2023

gkogekar Jun 22, 2023

speth Jun 17, 2023

gkogekar Jun 20, 2023

ischoegl Jun 20, 2023 •

edited

speth commented Jun 22, 2023

gkogekar commented Jun 22, 2023

speth left a comment

gkogekar commented Jun 22, 2023

Non-ideal capabilities for 1-D flame model #1079

Non-ideal capabilities for 1-D flame model #1079

Conversation

gkogekar commented Jul 29, 2021 • edited

speth commented May 23, 2022

codecov bot commented Jun 7, 2022 • edited

Codecov Report

bryanwweber commented Jun 21, 2022

gkogekar commented Jun 21, 2022

bryanwweber commented Jun 21, 2022

gkogekar commented Jun 22, 2022

bryanwweber commented Jun 22, 2022

speth left a comment

Choose a reason for hiding this comment

ischoegl commented Mar 21, 2023

gkogekar commented Mar 22, 2023

ischoegl left a comment

Choose a reason for hiding this comment

gkogekar commented Mar 30, 2023

speth left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ischoegl Jun 20, 2023 • edited

Choose a reason for hiding this comment

speth commented Jun 22, 2023

gkogekar commented Jun 22, 2023

speth left a comment

Choose a reason for hiding this comment

gkogekar commented Jun 22, 2023

gkogekar commented Jul 29, 2021 •

edited

codecov bot commented Jun 7, 2022 •

edited

ischoegl Jun 20, 2023 •

edited