decreasing computation time

80 Views
Last Post 13 January 2019
Topic Is Solved

prashant posted this 01 January 2019

I am using VOF(3 phases liquid nitrogen, water, air) along with solidification and melting model, keeping k-epsilon for turbulence, but since the velocity at inlet is 70 m/s ,so I have to choose very low time step(0.0000001) in order to satisfy the courant number. Refining mesh increase the computation time. Selecting implicit scheme (in VOF) is not giving the proper result.

SUGGEST ANY REMEDY TO DECREASE THE COMPUTATION TIME.

WHY CAN'T WE USE IMPLICIT WITH VOF?

#Permalink
0
0
0

Attached Files

4 Comments

Order By: Standard | Newest | Votes

prashant posted this 13 January 2019

Thanks Raef and Roowlhow for your response,I have simplified the mesh and now time step has reduced.

Raef I have learned alot from your videos on youtube, I am a beginner in ansys your videos are helpful to me, so thanks again for posting such vedios.

#Permalink
0
0
0

Moderator

Raef.Kobeissi posted this 10 January 2019

Maybe you can try and reduce the mesh resolution without jeopardizing the integrity of the simulation result. you will have to do some trials until you get the best combination(mesh/accuracy).

Cheers

Raef Kobeissi

#Permalink
0
0
0

prashant posted this 05 January 2019

I HAVE TRIED WITH IMPLICIT SCHEME WITH VOF BUT SOLUTION IS GETTING DIVERGED AND LITERATURE ALSO SUGGEST TO USE EXPLICIT SCHEME WITH VOF WHEN DEALING SUCH PROBLEM.

ACTUALLY LIQUID NITROGEN IS GETTING EVAPORATED AS IT COMES IN CONTACT WITH AIR DUE TO ITS PROPERTY.

ANY OTHER APPROACH TO MODEL THIS PROBLEM.

#Permalink
0
0
0

ANSYS

rwoolhou posted this 03 January 2019

We can use Implicit VOF, but there are other features that need to be turned on: read up on the numerics to see what you need.

Why are you using three phases, or so you expect the liquid nitrogen to not mix with the water? Please post an inline image as we can see what you're trying to model.

Computational time can be reduced by reducing cell count, simplifying the model or using more cpu (parallel processing - there are limits to this). Otherwise it's a case of waiting: CFD isn't an instant process and even the simple tutorials take time to run: more complex models can take days or weeks to converge.