CFD Online Logo CFD Online URL
www.cfd-online.com
[Sponsors]
Home > Forums > Main CFD Forum

Time-marching scheme diverges

Register Blogs Members List Search Today's Posts Mark Forums Read

Reply
 
LinkBack Thread Tools Display Modes
Old   October 12, 2013, 16:42
Default
  #21
Member
 
SergeAS's Avatar
 
Serge A. Suchkov
Join Date: Oct 2011
Location: Moscow, Russia
Posts: 73
Blog Entries: 5
Rep Power: 5
SergeAS is on a distinguished road
Send a message via Skype™ to SergeAS
1) In your case outlet BC should be d[rho,U,V,h]/dx=0
2) Negative values ​​of pressure, temperature and density suggest that your solution is unstable, In this case for chosen numerical method I would suggest to reduce the time step or CFL
...or use more recent numerical methods.

PS: extrapolation is bad way, because in this case you are replacing the physical laws of conservation of computational tricks
SergeAS is offline   Reply With Quote

Old   October 12, 2013, 17:02
Default
  #22
Member
 
Obad
Join Date: Sep 2013
Posts: 31
Rep Power: 3
Obad is on a distinguished road
Hy,

alright, then I will try it with zero gradient instead of extrapolation.

Does the reduction of the CFL really make the code much more stable. I understand the reason for CFL and of course it should not be to big and in my case I see that I definitely shouldn't choose Courant numbers above 0.1. But I mean, is there a certain point where the code becomes stable?

Another question is, how does the spacing of the grid influence the stability?
My grid has a uniform spacing in x and y direction of about 5mm. Is that small enough? Or does the overall size of the domain influence the stability.

As I already stated, I use the Maccormack technique because it is pretty simple. In the 1970s they calculated stuff like the space shuttle with that, so why shouldn't it work in 2013. But maybe you could tell me some other easy to program more modern techniques. Maybe I'll give it a try

I really appreciate your help SergeAS!
Obad is offline   Reply With Quote

Old   October 12, 2013, 17:41
Default
  #23
Member
 
SergeAS's Avatar
 
Serge A. Suchkov
Join Date: Oct 2011
Location: Moscow, Russia
Posts: 73
Blog Entries: 5
Rep Power: 5
SergeAS is on a distinguished road
Send a message via Skype™ to SergeAS
The grid spacing will influence the stability in terms of how much will interact different BC. For example in the problems of supersonic aerodynamics bow shock wave can interact with the inlet boundary, moreover mesh size affects the resolution accuracy shocks. As to the Courant number should be noted that it is calculated on the basis of the conditions of linear stability, while the problem with the shock waves are very nonlinear. If we talk about the McCormack method, then it has both pluses and minuses. The most significant downside I just would like to mention a strong dispersion error in the high-gradient area and hence stability problems in these areas.
Personally, I do not use the McCormack method for about 15 years.
SergeAS is offline   Reply With Quote

Old   October 20, 2013, 18:14
Default
  #24
Member
 
Obad
Join Date: Sep 2013
Posts: 31
Rep Power: 3
Obad is on a distinguished road
Ok SergeAS you convinced me.
I will try it with a different technique. Although my current code might work properly, I can't really check it because the calculation would take days...

Something is confusing me, I read a little bit about the modern shock capturing methods and they are always applied to equations that are discretized with the finite volume technique.
Is it necessary in my case to use the finite volume discretization? My geometry is simple and I of course use a structured grid for it.
Is there also a modern shock capturing technique that can be used with a finite difference discretization?

Could someone suggest me a good technique for unsteady hyperbolic PDEs?

Cheers!
Obad is offline   Reply With Quote

Old   October 21, 2013, 03:43
Default
  #25
Senior Member
 
Filippo Maria Denaro
Join Date: Jul 2010
Posts: 1,642
Rep Power: 23
FMDenaro will become famous soon enough
Quote:
Originally Posted by Obad View Post
Ok SergeAS you convinced me.
I will try it with a different technique. Although my current code might work properly, I can't really check it because the calculation would take days...

Something is confusing me, I read a little bit about the modern shock capturing methods and they are always applied to equations that are discretized with the finite volume technique.
Is it necessary in my case to use the finite volume discretization? My geometry is simple and I of course use a structured grid for it.
Is there also a modern shock capturing technique that can be used with a finite difference discretization?

Could someone suggest me a good technique for unsteady hyperbolic PDEs?

Cheers!
FV discretizations starts from the weak form of the hyperbolic equations, for this reason they are used in presence of non-regular solutions (shock).
Try a reading to the book of LeVeque
FMDenaro is offline   Reply With Quote

Old   October 21, 2013, 17:06
Default
  #26
Member
 
Obad
Join Date: Sep 2013
Posts: 31
Rep Power: 3
Obad is on a distinguished road
Which book do you mean the book Numerical Methods for Conservation Laws?

What does weak form of the equations mean?

For the FDM I used the Euler equations in strong conservation form (this is how Anderson calls them).
Can't I just discretize this equation with the FVM for example by using a second order upwind scheme?
I enclosed how I would discretize the Euler equation in strong conservation form. Would this discretization be correct and could I solve it simply explicitly for every node?
Attached Images
File Type: jpg finite volume 1.jpg (87.2 KB, 5 views)
File Type: jpg finite volume 2.jpg (91.8 KB, 3 views)
File Type: jpg finite volume3.jpg (91.7 KB, 3 views)
Obad is offline   Reply With Quote

Old   October 21, 2013, 18:27
Default
  #27
Senior Member
 
Filippo Maria Denaro
Join Date: Jul 2010
Posts: 1,642
Rep Power: 23
FMDenaro will become famous soon enough
have a reading of http://books.google.it/books/about/F...UC&redir_esc=y
FMDenaro is offline   Reply With Quote

Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are On


Similar Threads
Thread Thread Starter Forum Replies Last Post
same geometry,structured and unstructured mesh,different behaviour. sharonyue OpenFOAM Running, Solving & CFD 13 January 2, 2013 23:40
High Courant Number @ icoFoam Artex85 OpenFOAM Running, Solving & CFD 9 January 3, 2012 09:06
Orifice Plate with a fully developed flow - Problems with convergence jonmec OpenFOAM Running, Solving & CFD 3 July 28, 2011 05:24
Convergence moving mesh lr103476 OpenFOAM Running, Solving & CFD 30 November 19, 2007 15:09
Modeling in micron scale using icoFoam m9819348 OpenFOAM Running, Solving & CFD 7 October 27, 2007 00:36


All times are GMT -4. The time now is 19:45.