How to define the time step size using CFL number

Mudassir khan

Dear Experts 

Greetings 

I am working on the transient dam break multiphase flow. For faster and accurate simulation I need to set the optimum mesh size and time step. how to decide the time step using CFL number. Suppose I have to simulate the transient flow of with maximum front velocity 2m/s. I could not understand what should be the value of maximum velocity fector and mesh scale. Further it is transient turbulent flow whose CFL number must be less than 1. Plz guide to choose correct time step in my case.

 

Thanks in Anticiaption

Payal Pawar_20288

Hi

Please let me know what Altair product you are using so that I can direct your question to the proper forum.

Mudassir khan

I am using HWCFD

Jagan Adithya Elango

Step Size = (Smallest Mesh Length*CFL number)/Velocity

if you using Hyperworks CFD you could use the inbuilt calculator present in the Physics Tab.

Good Luck.

acupro

As a general rule, you won't need to limit this to CFL=1.  The solution will remain stable and accurate with a much higher CFL number - so long as you have the mesh to resolve flow gradients, and time step to resolve the changes in time.

Ingeniorator

acupro_21778 said:

As a general rule, you won't need to limit this to CFL=1.  The solution will remain stable and accurate with a much higher CFL number - so long as you have the mesh to resolve flow gradients, and time step to resolve the changes in time.

I have always wondered which CFL number is acceptable for Acusolve. Since it apparently uses rather stable (implicit?) schemes that allow for CFL numbers much larger than 1, what's the limit or at least what's a good rule of thumb?

acupro

Ingeniorator said:

I have always wondered which CFL number is acceptable for Acusolve. Since it apparently uses rather stable (implicit?) schemes that allow for CFL numbers much larger than 1, what's the limit or at least what's a good rule of thumb?

Correct - AcuSolve uses an implicit solver scheme.  Typically we would go through a sensitivity on time increment and see how the results of interest change.  I've not really used a CFL-based approach to determine the time step.  If you have an idea of a frequency in the flow - maybe start with 30 time steps per 'cycle' then go from there.

Ingeniorator

acupro_21778 said:

Correct - AcuSolve uses an implicit solver scheme.  Typically we would go through a sensitivity on time increment and see how the results of interest change.  I've not really used a CFL-based approach to determine the time step.  If you have an idea of a frequency in the flow - maybe start with 30 time steps per 'cycle' then go from there.

Okay, that makes sense. Thanks for clearing it up.