CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   CFX (https://www.cfd-online.com/Forums/cfx/)
-   -   Transient simulation with high courant number possible? (https://www.cfd-online.com/Forums/cfx/160043-transient-simulation-high-courant-number-possible.html)

euschelino12345 September 30, 2015 09:59

Transient simulation with high courant number possible?
 
Hi everybody,

im working on my bachelor thesis in cfx about a thermal storage system which is defined as a porous medium.
Now im trying to run a transient simulation of the heating stadium. For testing purposes
i have set the courant Number (for a simulation time of 0.5 seconds) to 100, 500, 1000 and
10000 with similar results and a very less computing time at a courant Number of 10000
(which is way to high). Is it now possible to use the courant Number of 10000 at a
Simulation time of 600 seconds without making a big mistake? Do you have any other ideas
to accelerate the Simulation without making big mistakes?


I apologize for my bad english! :D

ghorrocks September 30, 2015 17:53

CFX is an implicit solver so Courant number is not as critical as it is for explicit solvers. So for some simulations you can run large Courant numbers and have no ill effects.

So do not dismiss a simulation based on the Courant number alone. You should do a time step sensitivity check, and determine time step size from that.

euschelino12345 October 1, 2015 07:48

thank you for your quick response.

Sorry but im a beginner in the simulation with cfd can you give me a quick explanation about your answer.
Do you mean i should set the value of the initial Timestep a bit higher and run some testing simulations?

my settings in the heading Analysis Type are:

- Time Duration: [Option: Total Time; Total Time: 600s;]

- Time Steps: [Option: Adaptive; First Update Time: 0s; Timestep Update Freq.: 1; initial Timestep: 0.001s;]

- Timestep Adaption:[ Option: RMS Courant Number; Maximum Timestep: 0.1s; Minimum Timestep: 0.0000001s; Courant Number: 10000]

- Initial Time: [Option: Automatic;]

ghorrocks October 1, 2015 17:05

Adaptive time steps are a good approach. But I would not recommend adapting based on Courant number, but rather on coefficient loops. Adapt to 3-5 coeff loops per iteration. This ensures the solver runs efficiently with smallish time steps it can solve quickly, and also ensures a time step size independent solution in my experience.


All times are GMT -4. The time now is 23:03.