[Owain_Parry]

Hello,

I'm wandering what the typical configuration people use for gasoline engine simulations with SAGE solver?

1. Does the engine speed directly influence then required time-step during the combustion part of the cycle? If so, what is the typical solver time-step during combustion for a general case at high engine speed?
   
2. Should flame front refinement with Temperature AMR down to 0.25mm be adequate for most gasoline combustion simulations with SAGE?
   
3. My simulation has a lot of fuel left in the intake port during combustion which is causing the solver time-step to change dramatically as it evaporates. Whenever the solver reduces the time-step sharply from 5e-7s to 1e-8s due to 'dt_evap' or 'temperature extrapolation' there is a spike in species mass fractions and chemical heat release rate.
4. Is there a way that the solver recovery step of reducing the time-step could cause a problem between the SAGE and flow solver?

Thank you for your time!

[jlemoine]

Owain Parry,
Thank you for your questions.
1. During the combustion part of the cycle, it is not uncommon for the timestep to be limited by dt_move. dt_move is timestep limiter based on the motion of boundaries,dt_move is then based on the piston speed, if a boundary moves more than 1/3 of the cell dimension CONVERGE limits the timesteps by dt_move. Higher RPM case will have a smaller dt than lower RPM case if the timestep is limited by dt_move.
2. Yes, and we also do recommend to have 1mm refinement in the cylinder to have good turbulence levels in the cylinder. A grid refinement study is encouraged to see if TKE, velocity, tumble etc are converged to engineering expectations. Please refer to our tutorials on our download page for our best practices in terms of grid refinement.
3. This could be issues related to the evaporation model, temperature boundary conditions or even grid refinement. We have a gasoline engine tutorial case, I would recommend you to refer to it. The tutorial may help you to fix this problem. If you still encounter those issues you can email support@convergecfd.com for additional support.
5. No, these are tightly coupled.

[Owain_Parry]

Hello Jerome,

Thank you for your reply.

1) There is no time step being limited by 'dt_move' in the simulation

2) The simulation conforms to all these requirements

3) I followed the gasoline example case in terms of grid set-up, time stepping initially. The predicted cylinder pressure was twice the measured value. My current set-up is very different to the example case and am seeing pressure values in the region of those measured. However my concern is regarding the sensitivity of the simulation to fairly small set-up changes.

4) Do the flow and chemistry solver use the same time-step? If not, how are they coupled?

[jlemoine]

To answer 4), the timestep used for the flow is the same as the chemistry solver. Though, when the combustion is ON there is an additional timestep limiter, dt_chem. CONVERGE will check if the maximum temperature change is 25%(by default) of the cell temperature. If you are interested, there is a more detailed description in CONVERGE manual, chapter 5.6.
For your case, it looks like you have taken the right steps. Please email support@convergecfd.com, and we will help you with your simulation.

[MFGT]

Originally Posted by Owain_Parry

2. Should flame front refinement with Temperature AMR down to 0.25mm be adequate for most gasoline combustion simulations with SAGE?

Hi Jerome,

Regarding AMR, i have always used 0.5mm cells for velocity and Temperature.
However, it might be interessant to use smaller cells.

But if i specify 0.25mm cells, the increase in cells is way too high, increasing simulation run time a lot. Also for Temperature AMR, these small cells might be of interest only during initial stage of combustion. To save CPU time, can i specify a profile to scale AMR back to 0.5mm at lets say 50 CA aTDC?

Or propably it might be better to scale my whole mesh (base mesh 4mm -> 3mm) to increase resolution? This way my AMR would be 0.375mm.

[jlemoine]

Tobias,

CONVERGE doesn't allow a file for the embedded scale for AMR. So if you want to change the overall grid, changing the base grid is an option but this will have a cost in terms of CPU time. You will need to weight between grid resolution and run time. You will have to answer the question : Which one is more important, grid resolution or runtime? I hope this is helpful.

[MFGT]

Originally Posted by jlemoine

You will have to answer the question : Which one is more important, grid resolution or runtime? I hope this is helpful.

Both...
I need more CPU power, i guess.
However, i can chose any base mesh site, unfortunatly 4mm (4->2->1->0.5->0.25->0.125) results in such nice numbers.

I mean, who would start with lets say 3.64mm base mesh, to get AMR of 0.4525mm?

But actually 3.2mm istnt bad either, i will check this