CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CFX (http://www.cfd-online.com/Forums/cfx/)
-   -   Transient analysis time step (http://www.cfd-online.com/Forums/cfx/113677-transient-analysis-time-step.html)

kmgraju February 25, 2013 04:12

Transient analysis time step
 
Dear all
I would like to clarify the time steps.
I have a sinusoidal function and the frequency is 1 Hz. I did the transient simulation for one cycle
one of my friends told me and insist me to do the multiple cycle simulations.Is it necessary to do the simulations for multiple cycles?. If I do the multiple cycles the solution time is too large. Is there any way to short cut.

Thank you

Regards

ghorrocks February 25, 2013 05:15

If you want the startup transient then you do not need to do multiple cycles. If you want the repeating cycle it depends how long it settles to a repeating pattern. It could be 1 cycle, it could be hundreds - it depends on how stable the system is and how long it takes to settle.

But you will need to run it long enough to show that you have run long enough - so the change in the last few cycles is small enough that you can call it converged.

Lance February 25, 2013 06:55

Glenn is (as always) correct, you will need to perform a sensitivity analysis on the number of cycles in order to avoid initial effects - otherwise the initial conditions will affect your solution.

kmgraju March 2, 2013 21:01

Quote:

Originally Posted by Lance (Post 409906)
Glenn is (as always) correct, you will need to perform a sensitivity analysis on the number of cycles in order to avoid initial effects - otherwise the initial conditions will affect your solution.

Dear all

I have a set of data from 0 s to 2 s. I used one dimensional interpolation and activated extended minimum and maximum.

My question is

How to run this simulation for repeated cycles.( at least for 4 cycles) ?

Thank you

Regards

Govind

ghorrocks March 3, 2013 06:19

The simple nasty approach is just to repeat the 1D function 4 times.

The better approach is to use the mod function, so the time input to the interpolation is t-mod(t,2) which should return a function which increase to 2, then goes to zero and increases to 2 again when t is 4, and repeats that forever.

kmgraju March 3, 2013 14:14

Quote:

Originally Posted by ghorrocks (Post 411145)
The simple nasty approach is just to repeat the 1D function 4 times.

The better approach is to use the mod function, so the time input to the interpolation is t-mod(t,2) which should return a function which increase to 2, then goes to zero and increases to 2 again when t is 4, and repeats that forever.

Dear Mr Glenn

Thank you . It works

Govind


All times are GMT -4. The time now is 13:38.