[ghorrocks]

Does having very fine mesh near wall boundaries affect the time step selection by adaptive time stepping?

Yes. Finer mesh results in finer time steps. So as you refine the mesh the run times increase very quickly.

the mass flow should be 2.49 kg/s and the difference between them is minimal, do you think I should run it longer?

That is up to you. If you are happy with that accuracy then proceed. If not then you need to do more validation and verification work to improve accuracy.

Isn’t adaptive time step algorithm supposed to adjust the time step during a run?

Yes, that is exactly what it does.

I am not sure why CFX uses very fine time steps such as 4e-7 since the start.

It has decreased since the initial value but it has not increased at all.

It is saying you need very fine time steps to resolve this flow. Alternately, it could be saying that your simulation is numerically unstable and it needs very fine time steps to converge. Either way it probably needs it.

So the problem is not your time step settings, but the rest of your simulation. the first thing to do is to improve mesh quality, that always helps. You can also do a time step and convergence criteria sensitivity study to see if you can loosen then at all.

But overall, this is how CFD works. When you try to do accurate simulations the simulation get very big with long run times. That is why a good deal of the world's supercomputers are built to do CFD simulations. CFD was never meant for cheap desktop computers.

[Julian121]

Originally Posted by ghorrocks

Yes. Finer mesh results in finer time steps. So as you refine the mesh the run times increase very quickly.

What else affects the selection of time step by adaptive time stepping?

I am using SST turbulence model and I have kept y+ under 1 which has made the mesh very fine.

If I use k-e model which needs a y+ more than 11 and does not need a fine mesh, will the solution be inaccurate in terms of mass flow/pressure ratio prediction?

Right now, the adaptive time step is using 2.6e-6. When it uses such a fine time step, will anything coarser than that certainly yield inaccurate result?

[ghorrocks]

Anyway which affects convergence will affect the time step determined by adaptive time stepping. This is just about everything

Will a k-e simulation with y+>11 be inaccurate? That depends on your situation. k-e is a very good model in many cases, that is why it is the default turbulence model in many other CFD codes.

Also, do not assume that you must have y+<1 when you use the SST model. This is a common furphy. SST can handle coarser meshes than that in many cases. No matter what turbulence model you use, you should determine the wall resolution by sensitivity analysis.

If you use a coarser time step then you won't be converging every time step. If you don't converge the time steps then your time history will be wrong.

[Julian121]

Originally Posted by ghorrocks

Also, do not assume that you must have y+<1 when you use the SST model. This is a common furphy. SST can handle coarser meshes than that in many cases.

How big can I try the y+?

Do other transformation methods such as TT or FT have better convergence rate compared to the PT?

[ghorrocks]

There is no general upper limit to y+. If your flow is not much affected by the boundary layer you can make it what ever you like and it will be fine. So just do a sensitivity study and see how sensitive you are to it.

I do not understand your second question. Please do not use acronyms, especially ones which are not in common usage. I do not know what your acronyms mean.

[Julian121]

Originally Posted by ghorrocks

There is no general upper limit to y+. If your flow is not much affected by the boundary layer you can make it what ever you like and it will be fine. So just do a sensitivity study and see how sensitive you are to it.

How should I quantify the effect of boundary layer?

Does it have significant effect on global parametrs such as mass flow or stage pressure ratio?

Originally Posted by ghorrocks

I do not understand your second question. Please do not use acronyms, especially ones which are not in common usage. I do not know what your acronyms mean.

Does time transofrmation reach convergence sooner than the other methods?

PT: Profile Transformation

TT: Time Transformation

FT: Fourier Transformation

[ghorrocks]

Does it have significant effect on global parametrs such as mass flow or stage pressure ratio?

If the boundary layer is important then yes. If the boundary layer is not important then no. You appear to be modelling a blade cascade and in those cases the boundary layer is important most of the time, so you probably need to be careful about the boundary layer.