|
[Sponsors] |
August 5, 2014, 13:45 |
Under-relaxation factors
|
#1 |
New Member
Josef Camilleri
Join Date: Mar 2014
Posts: 27
Rep Power: 12 |
Hi all,
I am modelling a wedge free falling into an initially calm water surface and measuring the pressures acting at different positions along the bottom surface (surface average report of pressure on a specified area) and the vertical force acting on the bottom surface. As far as I know, my model is correct and I am not getting divergence or anything but I thought of playing around with the pressure, velocity and VOF under-relaxation factors to see what effect they have on my simulation. I did a number of systematic studies where I kept two of the three URFs constant and varied the other. The thing is that the peak pressure values and the time history of the vertical force change when changing the URFs. The difference is not that big but still. My time-step is relatively fine for this particular grid (CFL around 1) and I am using 20 inner iterations. Changing the URFs should not affect the solution but only the way convergence is achieved right? Is this something I should expect for transient simulations? Also, is there a way how I can make sure that the solution is converged within each time step? Thank you. Regards, Josef |
|
August 8, 2014, 03:09 |
|
#2 |
Senior Member
Gajendra Gulgulia
Join Date: Apr 2013
Location: Munich
Posts: 144
Rep Power: 13 |
Hello Jcamillari
In transient cases, the CFL has to be greater than 5 and URF is not a necessary criteria for convergence. URF's will increase the robustness of your solution but then you need to increase the iterations/timesteps for acheiving the desired convergence. If you are not getting divergence in your original setup, then I suggest you leve the solver settings to it's default values. |
|
August 11, 2014, 04:38 |
|
#3 |
Member
David
Join Date: Nov 2011
Posts: 39
Rep Power: 14 |
are you sure that for transient cases, the CFL has to be greater than 5?
|
|
August 11, 2014, 05:20 |
|
#4 |
New Member
Josef Camilleri
Join Date: Mar 2014
Posts: 27
Rep Power: 12 |
Hi ggulgulia, dai 549592484,
That's what I was thinking. Yes implicit schemes allow larger time step however for my case since everything is happening so fast and in a very short time period (around 0.5s) I need a very fine time step to make sure that all the physical phenomena of interest are resolved. Any ideas? |
|
August 11, 2014, 05:55 |
|
#5 |
Member
David
Join Date: Nov 2011
Posts: 39
Rep Power: 14 |
Hi jcamilleri
Have you done grid independent test? Are you using first order time discritization method or second order? If second order is used you need to make sure your maximum CFL is not larger than 0.5. To check whether your solution is converged or not, you can increase your maximum inner iteration to a large value, say 100, and monitor your pressure, plot your pressure against number of iteration. You can use the number of iteration where the pressure does not change anymore. |
|
August 11, 2014, 06:22 |
|
#6 | ||
New Member
Josef Camilleri
Join Date: Mar 2014
Posts: 27
Rep Power: 12 |
Hi dai549592484,
I am at this part of the problem at the moment where I am dividing both the base size and the time step by the same factor (2 in this case) and looking at the pressure and vertical force. Again, with refinement I am getting different peak pressures and vertical force plots. I might be that the theoretical pressure values are much higher Quote:
Quote:
|
|||
August 11, 2014, 07:46 |
|
#7 |
Member
David
Join Date: Nov 2011
Posts: 39
Rep Power: 14 |
I mean you should plot the pressure against number of iteration. and check the pressure variation within 1 time step. Ideally, when the impact happened. How did you monitor the pressure?
|
|
August 11, 2014, 07:56 |
|
#8 |
New Member
Josef Camilleri
Join Date: Mar 2014
Posts: 27
Rep Power: 12 |
Yes that is what I did - I changed the trigger (in Monitors) and x-axis
Monitor (in plots) from time step to iteration. I am attaching a pressure plot. P1 - P6 refer to six different positions along the bottom surface of my wedge where I am measuring the pressure - P1 being located most closely to the initial point of impact. The number of inner iterations in this case is 20. When the wedge is initially in air (up to approx 60000 iteration) I can see that the pressure plots reach an asymptotic limit however as the wedge impacts the water I am not able to see that. |
|
August 11, 2014, 08:29 |
|
#9 |
Member
David
Join Date: Nov 2011
Posts: 39
Rep Power: 14 |
do you use skype?
You should first make sure your interation is finished in 1 time step, that means time is not marching so that you can judge whether solution is converged in one time step. |
|
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
Relaxation Factors for Transient solvers | philippose | OpenFOAM Running, Solving & CFD | 19 | March 20, 2014 04:39 |
Purpose of relaxation factors | Mohsin | FLUENT | 5 | April 30, 2010 11:57 |
relaxation factors and time accuracy | Mike | Main CFD Forum | 7 | May 21, 2005 12:41 |
Relaxation Factors | Tim | Phoenics | 3 | June 30, 2004 02:03 |
relaxation factors adjust | zhujianguo | Phoenics | 1 | July 15, 2003 11:11 |