Hello everyone: During the unsteady computation, can the increased time step size affect the results and postprocessing? If I set the time step size 1e-05 initially and 1e-04 finally , are the results reasonable?

What is your numerical Scheme ? If the scheme is implicit the difference should not be so much . In an explicit scheme the difference can be considerable even for simple problems.

Right. Also consider the physical time scale of the system you're modeling. You should use a time step smaller than it.

Best regards, ap

Hi mini;

The time step - according to my knowledge - is dependent on: 1- the cell length. 2- the {|u+a|} term which express the maximum speed with which the information can travel. 3-and the CFL. If the cell has del(x) != del(y), you can take the smaller to compute the time step, although there are some other tricks that could be used.

Having this said, the time step should be calculated and not given. Of course if you give a little bit smaller time step than that computed, your solver will work. this is equvalent to decreasing the CFL, but you will increase; meanninglessly, your executaion time. The same arguement apply -to some extent- for a little bit increase.

Hoping it will help. cheers ;-)

This is true if you use an explicit method, while if you use a fully implicit one, the method is stable for each value of the time step.

However, this doesn't mean you can use a time step as big as you like. Obviously you still have the physical constraints on your dt to obtain an accurate solution.

ap

implicit scheme, thanks.

Yes. Thanks.