[kevinthomas93]

Hello Guys!

Im encountering CVODE Error while experimenting SAGE combustion in Converge CFD.

[CVODES ERROR] CVode
At t = 7.65329e-07, mxstep steps taken before reaching tout.

(Rank 1) WARNING: ncyc 3397, CVode error -1, CV_TOO_MUCH_WORK, The solver took mxstep internal steps but could not reach tout


The practice setup I am working on is the following:

-> O2 /CH4 Combustion SAGE model.

-> Mech.dat 10 species 6 reaction ( used ChatGPT to generate the mechanism
Since I could not find any simplified file elsewhere)

-> KH-RT model for parcel injection.

-> Simplified thin Cross section of a rocket engine combustion chamber (chamber +CD nozzle).

-> Grid size X= 0.001m , Y=0.001m , Z= 0.001m (z direction thickness of the CAD model is 0.001m )
Around 400K cells with AMR

-> Direction of Flow is in x axis.

-> Transient simulation , RANS k epsilon

-> Ignition Source : tried various options available in converge.
Source: Energy per unit volume time.

***********
Results:

Case 1:
Source value: 100,000,000 j/m3s
No error and No CO2 generation
Which I believe combustion is not starting - I’m monitoring temperature and species mass out.


Case 2:
Source value: 550,000,000 j/m3s
CVODE error occurs and struggles and CO2 generation is noted.
Which I believe combustion is starting.

I tried playing with the numbers and source values it’s either CVODE error with CO2 generation or everything goes smooth but No combustion(CO2 or H2O not generated ).

I’ve spent several weeks on it and couldn’t figure it out.
I would really appreciate if anyone could provide your expertise.

Thank you
Kevin

[drowinsk]

Hello Kevin, thanks a lot for your question. Regarding this issue it sounds like the mechanism is the cause of the problem. I have seen some of the automatically generated mechanisms and they can be questionable at best. For a scenario like this, I would recommend either the USC 12 species O2-CH4 mechanism or the RAMEC 17 species O2-CH4 mechanism. Additionally, Converge's C3 mech with C-H-O extracted will be a bit larger (50 species) but a reliable alternative, as well as the classic GRI 3.0 which is included in Converge in some of the Methane example cases (you can remove the N species to reduce the size to about 30 species). Feel free to let me know directly if you have any further questions.

Best regards,
David

[kevinthomas93]

Hello David!

Thank you for the response. I tried using the mech.dat shown below, is the USC 12 species O2-CH4 mechanism simpler than the one below ? Can I use that in SAGE model ?


By using the below mechanism on thin slice engine geometry , I noticed the CVODE error has disappeared. So now i modeled a 3D cross sectional 1/4th geometry of liquid rocket engine. with two sides symmetry

The current problem that I face with the following mechanism and running the simulation is , I see combustion happening when i use very low mass flow of fuel and oxidizer, as soon as i increase the mass flow to realistic values (design fuel oxidizer), the simulation crashes with fatal error, or is stuck. I tried cases by starting with low mass flow and gradually increasing mass flow - how ever the moment the actual mass flow comes in (1/4 of total mass flow) the error appears :

FATAL ERROR Something is wrong in function get_temp_from_table_masssfrac: upper_energy_value_must be greater than lower energy value! Upper energy_value is -nan(ind) and lower energy value is -nan(ind)

********mech.dat*********************

! Trimmed LOX-CH4 Reaction Mechanism for Rocket Engine Simulation ADDED N2
! Based on GRI-Mech 3.0
! Reduced for numerical stability in CONVERGE CFD (CVODE-friendly)

ELEMENTS
C H O N
END

SPECIES
CH4 O2 CO2 H2O CO H2 OH O H CH3 HO2 CH2O H2O2 N2

END

REACTIONS
! Key CH4 combustion pathways

CH4 + O2 => CO + H2 + H2O 5.0E13 0.0 30000
CH4 + OH => CH3 + H2O 1.0E08 1.6 3120
CH3 + O2 => CH2O + OH 2.2E13 0.0 15000
CH2O + OH => CO + H2O 1.0E13 0.0 3000
CO + OH => CO2 + H 2.5E13 0.0 5000


! Hydrogen oxidation paths

H2 + O2 => OH + OH 1.0E14 0.0 30000
H2 + OH => H2O + H 2.1E08 1.5 3430
H + O2 => OH + O 2.0E14 0.0 15000

! Termination steps

O + H2 => OH + H 5.1E04 2.67 6290
H + OH + M => H2O + M 2.2E22 -2.0 0
H + O2 + M => HO2 + M 2.0E18 -0.8 0
HO2 + H => OH + OH 1.5E14 0.0 1500
HO2 + OH => H2O + O2 2.5E13 0.0 2000
HO2 + HO2 => H2O2 + O2 1.3E11 0.0 1000
H2O2 + H => H2O + OH 1.0E13 0.0 2000

END
STOP





Looking forward! Thank you
Kevin

[drowinsk]

Hello Kevin, thanks for your follow-up. There might be several different reasons for the behavior you observed, either related to the flow, the chemistry, or a combination. Getting a good idea of the case's behavior leading up to the crash point might help better understand what is going on. It could be that the higher flow rates lead to greater pressure or temperature variability and that causes problems with the reduced mechanism or the flow itself. I would recommend to take a look at some output quantities like the min/max pressure and temperature, the heat release rate, and the flame development in space to see if that can shed any light on the nature of the problems. Feel free to email us at support@convergecfd.com if a more intensive look would help.

Best regards,
David

[kevinthomas93]

Hi David,

Thank you for the response. As you advised, I compared two cases—Case E and Case E1
(see attached figures below this thread). Both cases are identical except for the ramping rate of fuel and oxidizer injection.

Upon comparison of the heat release rate (HRR), temperature, and pressure fields:

- The value ranges of HRR and temperature are nearly identical.

- However, the maximum pressure shows a noticeable overshoot in Case E compared to Case E1, which appears to be due to the higher mass flow rate.

I'm wondering if there might be something incorrect or unstable in the KH-RT parcel injection behavior. Could the issue be related to evaporation dynamics? Specifically, as injection mass flow increases, the sudden introduction of cold liquid propellants may lead to delayed vaporization and accumulation of vapor in the chamber. This could potentially result in a rapid, bulk combustion event—leading to the observed pressure spike.

Does this seem plausible to you? I’d appreciate your insights on what might be causing this behavior and how it could be mitigated or stabilized.

Looking forward to your feedback. Thanks
Kevin

[kevinthomas93]

Attachments