CFX Pressure Disibution is Useless

vikramaditya91 · July 14, 2015, 14:55

Hey

I am trying to validate a wedge falling into water using CFX from a certain height according to this publication which seems to be quite standard
A Detailed Assessment of Numerical Flow Analysis (NFA) to Predict the Hydrodynamics of a Deep-V Planing Hull

However I see that the pressure peak as seen in figure 2 of the publication can be obtained only for a particular set of values i.e
1) minimum mesh element size = 0.9 mm
2) time step size = 0.00005 second

It is noted that the publication has used similar values.

The pressure peaks do not match for any other time step and mesh element size (even for a smaller time step or a smaller element size).

It is also noted that the pressure peaks are smaller (less than the experiment) for large elements and larger time steps. And the pressure peaks are higher if the time steps and element sizes used are larger than the values stated above.

Since I know what value I have to obtain, I can set the time step and element size, however if I wanted to carry out the simulation for a different geometry, how would I know what element and mesh size I have to use (to get exact/experimental values)? Is there any other way of doing it?

ghorrocks · July 14, 2015, 19:02

If mesh refinement causes the answer to move away then you should ignore the mesh where you happen to get about the right answer. It was just luck. The whole idea of CFD is that you get a grid-insensitive solution. If you cannot get a grid insensitive solution then you do not have a trust worthy solution.

So if you are not getting the correct answer by mesh refinement you should look at other issues. Have a look at differencing schemes, free surface formulations, time step size, convergence tolerances and everything else.

vikramaditya91 · July 17, 2015, 11:20

Thanks for your reply ghorrocks. But we are using a convergence of 1E-5, surface tension of 0.073 N/m, tried different time deifferencing schemes and we have tried for various turbulence models too.
Are there any other specific aspects that are important to capture pressure peaks?

ghorrocks · July 17, 2015, 18:03

In my experience the free surface model can be optimised to different applications quite a significant amount. So I would list all the settable parameters of the free surface model and see which ones can improve things. Things like free surface smoothing, coupled vs segregated VF equations and those things are what I am talking about.

If you are running surface tension then your simulation will be highly sensitive to mesh quality, and much more sensitive then the default mesh quality check in CFX. What type of mesh have you got? What is it quality?

July 14, 2015, 14:55	CFX Pressure Disibution is Useless	#1
vikramaditya91 New Member eu sou cfd Join Date: Jun 2012 Location: Brazil Posts: 18 Rep Power: 13	Hey I am trying to validate a wedge falling into water using CFX from a certain height according to this publication which seems to be quite standard A Detailed Assessment of Numerical Flow Analysis (NFA) to Predict the Hydrodynamics of a Deep-V Planing Hull However I see that the pressure peak as seen in figure 2 of the publication can be obtained only for a particular set of values i.e 1) minimum mesh element size = 0.9 mm 2) time step size = 0.00005 second It is noted that the publication has used similar values. The pressure peaks do not match for any other time step and mesh element size (even for a smaller time step or a smaller element size). It is also noted that the pressure peaks are smaller (less than the experiment) for large elements and larger time steps. And the pressure peaks are higher if the time steps and element sizes used are larger than the values stated above. Since I know what value I have to obtain, I can set the time step and element size, however if I wanted to carry out the simulation for a different geometry, how would I know what element and mesh size I have to use (to get exact/experimental values)? Is there any other way of doing it?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Pressure Outlet Guage pressure	Mohsin	FLUENT	36	April 29, 2016 17:16
Pulsatile pressure inlet with pressure outlet	a.lynchy	FLUENT	3	March 23, 2012 13:45
reference pressure and relative pressure in CFX and real gas setting	nuimlabib	Main CFD Forum	0	October 12, 2010 20:55
how to define pressure drop with CFX post	alex	CFX	0	September 20, 2007 17:31
negative pressure in cfx	flar.t	CFX	1	December 18, 2006 23:20

July 14, 2015, 19:02		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,703 Rep Power: 143	If mesh refinement causes the answer to move away then you should ignore the mesh where you happen to get about the right answer. It was just luck. The whole idea of CFD is that you get a grid-insensitive solution. If you cannot get a grid insensitive solution then you do not have a trust worthy solution. So if you are not getting the correct answer by mesh refinement you should look at other issues. Have a look at differencing schemes, free surface formulations, time step size, convergence tolerances and everything else.

July 17, 2015, 11:20		#3
vikramaditya91 New Member eu sou cfd Join Date: Jun 2012 Location: Brazil Posts: 18 Rep Power: 13	Thanks for your reply ghorrocks. But we are using a convergence of 1E-5, surface tension of 0.073 N/m, tried different time deifferencing schemes and we have tried for various turbulence models too. Are there any other specific aspects that are important to capture pressure peaks?

July 17, 2015, 18:03		#4
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,703 Rep Power: 143	In my experience the free surface model can be optimised to different applications quite a significant amount. So I would list all the settable parameters of the free surface model and see which ones can improve things. Things like free surface smoothing, coupled vs segregated VF equations and those things are what I am talking about. If you are running surface tension then your simulation will be highly sensitive to mesh quality, and much more sensitive then the default mesh quality check in CFX. What type of mesh have you got? What is it quality?