CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   CFX (https://www.cfd-online.com/Forums/cfx/)
-   -   false time step (https://www.cfd-online.com/Forums/cfx/18544-false-time-step.html)

icmyn June 3, 2001 22:47

false time step
 
Dear ALL:

Who knows the meaning of 'false time step'?

Thanks in advance.

Li Xinfeng June 4, 2001 03:28

Re: false time step
 
As an alternative to setting under-relaxation factors, you may requestfalse time-stepping using keywords of the subcommand >>FALSETIMESTEPS. In this case, the equations are modified in the same wayas if each outer iteration is a time step, with two differences: the ¡®old¡¯time values used are the most recently calculated values of each of the variables; and the time increment may be chosen arbitrarily to achieve the best convergence rate. This procedure may also be applied at each (true) time step of a transient calculation, in which case the equations are modified twice. Different false time steps can be used to accelerate each equation independently. These time steps should reflect the time scale over which each variable changes. Then every variable develops at its own optimum rate. The main drawback of false time stepping is that a knowledge of time scales is required, whereas under-relaxation factors may take default values for a wide range of problems. However, the more physical character of false time stepping may prove beneficial, particularly when several separate influences are driving the flow, e.g. a forced flow with buoyancy.

FROM¡¡CFX4.3 MANUAL

Mukhopadhyay June 5, 2001 02:58

Re: false time step
 
When you have an elliptic equation to be actually solved, you may formulate the problem as a (time) parabolic one and then solve it for t=infinity. While time marching, the time steps involved are 'false' because you are not looking at the temporal solution - you are looking forward to the solution of the elliptic problem. Hope I have been able to clarify.


All times are GMT -4. The time now is 12:59.