[aja1345]

Hi,

Is it possible to have different numerical results with different timescales?

In fact, if I change timescale, Do results change?

if I change timescale and results change, what is problem?

Can you explain about timescale briefly?

Thanks.

[aja1345]

Should I use different timescales before and after stall in turbine simulation or not?

[Opaque]

For steady state calculations, a converged solution should be unique for any value of the timescale. Different values of timescale produce the following results sorted as:

- Diverging results, you must reduce your timescale
- Converged results with a lot of struggle, if converged well, your results are just fine, but you can try reducing your timescale a bit for the next simulations.
- Converged results with an optimal/ideal convergence, do nothing
- Converged results after many many iterations, if converged well, your results are just fine, but you can try increasing the timescale a bit for the next simulations.
- Residuals seems stuck, and they seem to not progress at all, try increasing your timescale and see if it falls in one of the cases above.

The above applies to the Physical Timescale, Auto Timescale options only.

Hope the above helps,

[aja1345]

Originally Posted by Opaque

For steady state calculations, a converged solution should be unique for any value of the timescale. Different values of timescale produce the following results sorted as:

- Diverging results, you must reduce your timescale
- Converged results with a lot of struggle, if converged well, your results are just fine, but you can try reducing your timescale a bit for the next simulations.
- Converged results with an optimal/ideal convergence, do nothing
- Converged results after many many iterations, if converged well, your results are just fine, but you can try increasing the timescale a bit for the next simulations.
- Residuals seems stuck, and they seem to not progress at all, try increasing your timescale and see if it falls in one of the cases above.

The above applies to the Physical Timescale, Auto Timescale options only.

Hope the above helps,

Thanks.

I almost understand your statement. Unfortunately My English language is not very good.

What is your exact meaning of "struggle" and "stuck"?

For steady state calculations, a converged solution should be unique for any value of the timescale.

if I change timescale and results change, What is the reason?

[Opaque]

Originally Posted by aja1345

Thanks.

I almost understand your statement. Unfortunately My English language is not very good.

What is your exact meaning of "struggle" and "stuck"?

Solution does not converge monotonically

if I change timescale and results change, What is the reason?

In most cases the solution is unique (definitely for linear systems) if the solution is converged, i.e. residual are small enough that the solution remains unchanged even if the residual are reduced further.

However, you must keep in mind that because the flow equations are non-linear, there is the possibility of multiple solutions and the software may have converged towards one of them. In such situations, you may have to dig deeper to understand if the solution makes sense, is it stable ? is it the most probable solution ?

Hope the above helps,