varying timestep size

[newuser]

I wonder if varying timestep size can be set for a simulation?

for example- If the total simulation time is 11 sec.

 timestep size of 0.01 sec for first 100 timesteps and then 0.1 sec for next 100 timesteps.

In other way timestep size of 0.01 for 0.01 x 100 = 1 sec and then 0.1 for next 0.1 x 100 = 10 sec.

Thanks

[Kevin]

You can either fix the time step to be a constant, which is not recommended for most apps, or let the time step be adjusted automatically according to the CFL criteria.

[newuser]

Hello Kevin,

istn't there any way to do, so that the computation time can be reduced. 

[Kevin]

 If we could just choose a larger time step to make our calculations run faster, why wouldn't we all just do this? Please take some time to learn about CFD, in particular LES, before using FDS.

[newuser]

Hi,

in Fluent solver it is possible to set varying timestep addition to no. of iterations per timestep and my question was in FDS is it possible or not? If concerned with timestep, i amn't going to set timestep of  1 or 5 sec for simulation. It could be 0.01 or less than it depends on the CFL no. I carried a simulation in FDS and time difference between two timesteps i observed was approx. 0.01. 

[Kevin]

Given that you have chosen the username "newuser", I have to assume you do not know much about FDS. If that is the case, the last thing I want you to be doing is mucking around with the time step. FLUENT is a RANS model and FDS is LES. Two very different forms of CFD.

[newuser]

Hi,

yes I am new to FDS but not to CFD. I am totally agreed  with what you said that in LES timesteps should be smaller to maintain stability so leave it on FDS. 

As you said it isn't recommendable but a fixed timestep can be set then why two different timesteps one after the other cannot be set. If there is no possibility of doing this in FDS then no question from my side. 

If others can also answer my query then that would be appreciable.

[dr_jfloyd]

The primary reason we have the fixed time step option is for verification and testing purposes. For any real calculation there is little to no value in specifying a fixed time step. We don't have any inputs to allow you to vary the fixed time step.

[newuser]

 It is clear to me now. Thank you so much.

[shostikk]

However, there are some special applications where it would be useful to change the maximum time step over the course of the simulation. I do not recommend this for new FDS users, but there are. For instance, I some cases where I have an extremely dynamic event in the beginning of the calculation that requires forcing a small time step to maintain the calculation stable. Later in the simulation, a longer time step would suffice.

There are of course other possibilities to tackle this kind of stability problem, and thus I have never started implementing a DT_RAMP.

[newuser]

Hello,

so that's the twist. Thank you.