CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   Adequate timestep selection for multidomain problem (

gerardosrez November 27, 2010 15:37

Adequate timestep selection for multidomain problem
I'm setting up a multidomain (solid and fluid) transient simulation of a furnace (involving combustion) where I wanna know how much time is needed to get certain temperature in part of a solid domain (according to the experimental data is 4 hours).

I selected a timestep of the order of 1e-04 as the initial timestep and a factor of 10 to increase it, but after severeal timesteps calculations this timestep hasn't been increased yet (after 30 timesteps aprox.).

So my question is, how can i achieve the result i desire using a higher timesteps with a good phisical accuracy, because if the initial timestep doesn't increase i will never get the Total time of the simulation (aprox. 4 hours).

Thanks in advance.

ghorrocks November 27, 2010 20:58

What are you using as your adaptive time step critereon?

gerardosrez November 27, 2010 21:20

Adequate timestep selection for multidomain problem
I'm using the Number of Coefficient Loops method....

I tried switching to a timestep list with the number and the value of timesteps over my whole simulation and combustion never started...I'm using EDM and I don't know why combustion doesn't starts, as a solution I included a source point with the double of the activation energy as the source value with any satisfactory result...

Any advice on my problem will be very welcome...thanks in advance ghorrocks...

joey2007 November 28, 2010 15:05

always the same misunderstandings :mad:. First I would use "pure" EDM. Check the model equations in order to understand why pure EDM for premixed flames cannot be ignited by temperature or enthalpy.

gerardosrez November 28, 2010 15:24

Adequate timestep selection for multidomain problem
Yes....well i have already solved that isssue turning -1. the B coefficient (producto limiter), i had a value of 0.5 to control combustion

ghorrocks November 28, 2010 17:50

For single phase simulations you usually aim to get 2-5 or 3-5 coeff loops per iteration. For multiphase and combustion you will probably have to increase it, maybe 5-10 loops per timestep. But you might want to have some fine and constant time steps when you start the combustion.

gerardosrez November 28, 2010 18:50

Adequate timestep selection for multidomain problem
Thanks for the help i'll struggle to get this done efficiently :)

All times are GMT -4. The time now is 16:34.