
[Sponsors] 
November 28, 2003, 06:45 
VOFConvergence

#1 
Guest
Posts: n/a

Hi,
I have a convergence problem with VOF, when simulating 1mm water drop free falling in air. In order to ensure convergence I need to decrease the time step down to 10^(9) sec. throughout the calculations. It seems that I am not likely to reach steady state in a reasonable time. My mesh is about 10 micron to resolve well the inner and outer fields. I define velocity inlet BC with zero velocity at the bottom and pressure outlet BC at the top of the drop. Does anyone have an idea what's wrong with that? 

November 28, 2003, 07:48 
Re: VOFConvergence

#2 
Guest
Posts: n/a

Hi, Could explain a bit more your BC conditions. I do not understand what you mean by 'bottom and top of the drop'. Your drop is falling down, therefore the oulet will be at the bottom. So why do you have pressure inlet there ?
Then you say you are not reaching a steady state in a reasonable time. Do you mean ? computer time ? or Physical time ?. If you are simulating a drop starting form a zero velocity, Did you check the relaxation time of your drop ? Thomas 

November 28, 2003, 09:18 
Re: VOFConvergence

#3 
Guest
Posts: n/a

Hi,
Actually I patch my drop in the middle of a rectangular domain, where the bottom edge is defined as velocity in BC with v=o m/s, the top of the domain defined as pressure out BC, and the to sides are defined as symmetry BC. What I mean by time to reach stst is the computation time for the computer to arrive at t=t(stst) since the time steps are extremely small. Could you please explain a bit more what you mean by relaxation time of the drop 

November 28, 2003, 15:45 
Re: VOFConvergence

#4 
Guest
Posts: n/a

Hi,
Relaxation time is the time your drop need to achieve a steady state, means constant velocity. If you start with a drop from v=0, your drop will need a certain time to reach this stst. This time is what I call a physical time. Depending on your time step and computer, you will need a certain computation time to achieve the relaxation time(= a certain number of time step. If you start from a drop with v=0, check if you calculated an enought number of time step to achieve this relaxation time. If note it means you are not in a steady state yet. I hope it is clear enought Just 2 questions: What is the maximum speed of you outter and inner fluid? How do you define your time step ? thomas 

November 29, 2003, 07:14 
Re: VOFConvergence

#5 
Guest
Posts: n/a

Hi, I just want to bring a little correction on what I said about relaxation time. When you write your force balance on your particule and you integrate it to find the velocity evolution of your particule you will find an exponential expression with a time constant in it. It is this time constant that we call 'relaxation time of your particule'. This relaxation time represents around 67% of the time that the particule needs to achieve its terminal velocity (means steady state). So the relaxation time of your particule is not exactly the time you need to achieve the stst but just a sort of approximation. Also it allows to calculate the terminal velocity of your particule.
thomas. 

November 29, 2003, 21:51 
Re: VOFConvergence

#6 
Guest
Posts: n/a

Are you using georeconstruct as the algorithm for the interface ? This is scheme is always seems to cause me grief as well, typically requiring very small timesteps to prevent divergence. I would recommend using some of other schemes first, such as the explicit or implicit euler. These schemes allow you to use a larger timestep, albeit at the expense of losing accuracy along the interface.


November 30, 2003, 08:18 
Re: VOFConvergence

#7 
Guest
Posts: n/a

Hi,
Actually, the problem itself is the long computational time needed for the computer to reach the stst time of drop. If we assume that terminal v= 3 m/s and g=10 m/s^2, then neglecting drag force, the time to reach stst for drop is 0.3 m/s. If a need to go with time steps of 10^(9)sec., where each time steps needs about 20 itterations each taking about 1 second, then I think I will be in trouble!! About the inner & outer boundaries: I have very small velocities here since only source for the flow in the domain is the movement of my drop. Actually I am not sure how to define the inlet & outlet boundaries for my system where a drop is falling in an initially stagnant evironment. 

November 30, 2003, 08:27 
Re: VOFConvergence

#8 
Guest
Posts: n/a

Hi,
Yes, I am using GeoReconst. scheme. Thanks for the advice, I will try the other algorithms. 

December 2, 2003, 06:11 
Re: VOFConvergence

#9 
Guest
Posts: n/a

do you have body force formulation on in your VOF settings and what schemes are you using for pressure and momentum? I have generally found that presto and second order respectively are best in VOF


December 3, 2003, 11:10 
Re: VOFConvergence

#10 
Guest
Posts: n/a

hi,
well I follow the Fluent's recommendations, and use body force weighted for pressure, and QUICK for momentum. I also have body force formulation. But, any way i never be able to converge with time steps greater than 10^(5) seconds. Another problem is that, my drop is insistingly deforming, under phsical conditions which it should stay spherical (I include surface tension in the calculations). 

Thread Tools  
Display Modes  


Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
Convergence  Centurion2011  FLUENT  27  February 7, 2016 12:58 
Force can not converge  colopolo  CFX  13  October 4, 2011 22:03 
Convergence of CFX field in FSI analysis  nasdak  CFX  2  June 29, 2009 01:17 
VOF model convergence  lana  FLUENT  11  October 1, 2005 05:23 
VOF convergence  CFD Newbie  FLUENT  3  November 24, 2002 02:14 