time step
 

when i run it with time step 0.05s it stops saying ' Fatal Overflow '. then i tried with time step 0.005 and it is working and seems creeping from one iteration to the next very very slowly. Is it because the smalles mesh scale (0.005m)is very very small ???

Re: time step
 

Change the scheme!

Re: time step
 

Also depends on what flow physics you're trying to model. What kind of Courant number are you getting?

Re: time step
 

Hi,

Generally this error is caused by a overflow error in the linear solver. The fix is to use a smaller timestep, at least when you are starting up.

Glenn Horrocks

Re: time step
 

A good way to estimate timestep is is by using the equation dt = 2U/L, where U is the estimated fluid velocity and L is the length of your domain. This is a good estimation of your advection time. I will usually use this method to estimate an initial timestep for a simulation and after it runs will check the *.out file for the computed advection time. I typically like to use a timestep on the order of 1/3 of that value (small timestep).

Re: time step
 

Try using adaptive time step.

Re: time step
 

"Try using adaptive time step."

Cool!

Re: time step
 

If it is a transient simulation, be careful with the adaptive time stepping scheme - especially if the time step undergoes large changes

Re: time step
 

Courant number is changing in each timesept...where should i check for a Courant number...???

Re: time step
 

The Courant Number (Co in CFX, often Cn) is in the .out file of your run, both RMS Co and Max Co are shown.

Re: time step
 

You can reduce the factor by which the time step will change.

Re: time step
 

i still have that problem... the courant number i am getting is Name : Default Domain Modified Water.RMS Courant Number = 5.5075E-01,

Water.Maximum Courant Number = 6.3199E+00

Domain Name : Domain 1 Water.RMS Courant Number = 1.0387E-01,

Water.Maximum Courant Number = 8.8954E-01