[zwu249]

As we all know, CONVERGE has the excellent grid manipulation method. And I know you do have a recommended value to get the grid convergence for LES calculation spray combustion.
Are there some recommendations for the mesh resolution when I want to perform an engine 3D calculation using RANS method? Is there a different value when doing premixed combustion simulation compared to diesel spray combustion simulation? And is it different for SAGE calculation and simplified combustion model?
One more question. When using velocity amr, how can I set an appropriate value for the parameter "amr_vel_sgs_embed" for different engine speed?

[ywang89]

We have several example cases which demonstrate premixed combustion and the user should check these cases periodically. These cases are updated with the latest recommended settings for numerics, spray, combustion and grid.
The grid settings depend on the physics to resolve and the computational cost involved. Having smaller grid sizes are almost always better in CFD, but is limited by the computational cost involved. The following recommendations should only be used as a first approximation and questioning the grid independence of your results is always advisable:

For RANS simulations:
i) base mesh size: 4mm
ii) The combustion chamber should always have cells less than 1mm to resolve turbulence adequately
iii) The velocity gradients can be resolved by using velocity AMR in the intake port and in the cylinder. Use at least 0.5 mm cell sizes
iv) If fuel is injected directly in to the combustion chamber (for example, GDI) or in to the intake port (for example, PFI), use 0.25 mm to 0.5 mm cell sizes to resolve the exit of the injector. The spray velocity gradients due to spray penetration should be resolved via velocity AMR.
v) The grid in the vicinity of the spark plug should be refined sufficiently to resolve turbulence and hence ignition of the mixture. Use 0.125 mm to 0.25 mm cell sizes
vi) Unless the flame speed is specified, the temperature and species gradients should be resolved well for accurate flame propagation speed predictions. Use 0.5 mm cells via temperature AMR for this
vii) A coarser mesh is used typically in the exhaust port, unless any aftertreatment mechanism (ex: urea injection for NOx reduction) is modeled.
NOTE: These recommendations are based on resolving physics accurately. These settings result in ~1-1.5 million cells for an engine whose bore is ~ 10 cm. If the engine size is large and the above settings result in a large cell count which is too expensive to manage, engineering judgement has to be used in coarsening the mesh.

For LES simulations (much more computationally expensive):
i) base mesh size: 0.5 mm
ii) velocity and temperature AMR cell size: 0.0625 mm to 0.125 mm
iii) cell sizes at the injector exit: 0.0625 mm to 0.125 mm
iv) cell sizes in the vicinity of the spark: 0.0625 mm to 0.125 mm

[ywang89]

We have several example cases which demonstrate premixed combustion and the user should check these cases periodically. These cases are updated with the latest recommended settings for numerics, spray, combustion and grid.
The grid settings depend on the physics to resolve and the computational cost involved. Having smaller grid sizes are almost always better in CFD, but is limited by the computational cost involved. The following recommendations should only be used as a first approximation and questioning the grid independence of your results is always advisable:

If the engine and the direct injection process is symmetrical, we recommend running the spray and combustion using only a sector. However, the intake flow should be resolved using the full cylinder geometry with moving valves. The flow field information at intake valve closing can be mapped in to the sector geometry for economical combustion simulations

For RANS simulations:
Intake simulation (full geometry):
i) base mesh size: 4mm
ii) The combustion chamber should always have cells less than 1mm to resolve turbulence adequately
iii) The velocity gradients can be resolved by using velocity AMR in the intake port and in the cylinder. Use at least 0.5 mm cell sizes

Combustion simulation (sector geometry)
i) base mesh size: 1.4mm
ii) The velocity gradients can be resolved by using velocity AMR in the cylinder. Use 0.35 mm cell sizes
iii). Use 0.35 mm cell sizes to resolve the exit of the injector. The spray velocity gradients due to spray penetration should be resolved via velocity AMR.
iv). The temperature and species gradients should be resolved well for accurate flame propagation speed predictions. Use 0.35 mm cells via temperature AMR for this

NOTE: These recommendations are based on resolving physics accurately. These settings result in ~0.5-1.0 million cells for a 60-degree engine sector whose bore is ~ 10 cm. If the engine size is large and the above settings result in a large cell count which is too expensive to manage, engineering judgement has to be used in coarsening the mesh.

For LES simulations (much more computationally expensive):
Intake simulation (full geometry):
i) base mesh size: 0.5 mm
ii) velocity AMR cell size: 0.0625 mm to 0.125 mm

Combustion simulation (sector geometry)
i) base mesh size: 0.5 mm
ii) velocity and temperature AMR cell size: 0.0625 mm
iii) cell sizes at the injector exit: 0.0625 mm

[ywang89]

Originally Posted by zwu249

As we all know, CONVERGE has the excellent grid manipulation method. And I know you do have a recommended value to get the grid convergence for LES calculation spray combustion.
Are there some recommendations for the mesh resolution when I want to perform an engine 3D calculation using RANS method? Is there a different value when doing premixed combustion simulation compared to diesel spray combustion simulation? And is it different for SAGE calculation and simplified combustion model?
One more question. When using velocity amr, how can I set an appropriate value for the parameter "amr_vel_sgs_embed" for different engine speed?

Hi Jack,

Your questions are very typical. Therefore, I posted two answers: one is for SI engines and the other one is for Diesel engines. For different combustion models, usually we use the same grid resolution. The typical value of amr_vel_sgs_embed is 0.5~1.0 for IC engines with RANS.

[zwu249]

I got it. Thanks for your answer and for being so helpful.

[MFGT]

Originally Posted by ywang89

For RANS simulations:
...
ii) The combustion chamber should always have cells less than 1mm to resolve turbulence adequately
...
iv) If fuel is injected directly in to the combustion chamber (for example, GDI) or in to the intake port (for example, PFI), use 0.25 mm to 0.5 mm cell sizes to resolve the exit of the injector. The spray velocity gradients due to spray penetration should be resolved via velocity AMR.
...

regarding ii)
Even if nothing major is happening in the cylinder, should I keep 1mm and not rely on velocity AMR? I had the idea to resolve the area of the valves, valve lift and upper cobustion chamber with 1mm cells or finer, while lower is at 2mm to save runtime when piston is far away from TDC. AMR active with up to 0.5mm cells.

regarding iv)
are there recommendations for the spray embedding cone size in GDI?
I used R1=0.0005, R2=0.0035 and Length=0.02

[ywang89]

Originally Posted by MFGT

regarding ii)
Even if nothing major is happening in the cylinder, should I keep 1mm and not rely on velocity AMR? I had the idea to resolve the area of the valves, valve lift and upper cobustion chamber with 1mm cells or finer, while lower is at 2mm to save runtime when piston is far away from TDC. AMR active with up to 0.5mm cells.

regarding iv)
are there recommendations for the spray embedding cone size in GDI?
I used R1=0.0005, R2=0.0035 and Length=0.02

Hi Tobias,

A 1) The requirement for the 1 mm mesh within the cylinder is to resolve turbulence. You cannot fully rely on velocity AMR to kick in when turbulence is developing.
A 2) This depends on the injection pressure and the cone angle. You need to check the spray plume from Ensight to make sure your embedding cone can cover it.

Thanks,

Yunliang