[overture] Re: False Shown Physical Boundaries
- From: matthew fuller <matt.l.fuller@xxxxxxxxx>
- To: overture@xxxxxxxxxxxxx
- Date: Mon, 13 Jul 2009 10:36:42 -0400
Bill,
Thanks for the prompt reply! I think i can understand where you are coming
from on the representation of convex corners. There really is no way to
fully represent these areas with mappings.
I had another question about modifying the grid.C file. I'm in unknown
waters here with remaking Overture. I attempted to enter make in the
overture directory and I get a "no targets specified no makefile found". I
do notice a Makefile.in and a Makefile.save. Should i run one of those
instead? Within them they state that environmental variables should be
defined prior to running though.
Thanks for your help.
Matt
On Sun, Jul 12, 2009 at 11:21 PM, Bill Henshaw <henshaw@xxxxxxxx> wrote:
> Dear Matt,
>
> At some point the plotting of mixed boundaries was broken. Here is the
> fix:
>
> Change Overture/ogshow/grid.C around line 463 from
>
> --- old:
>
> if( parameters.gridOptions(grid) &
> GraphicsParameters::plotInteriorBoundary )
> ISinteriorBoundaryPoint=0;
> else
> ISinteriorBoundaryPoint=MappedGrid::ISinteriorBoundaryPoint;
>
> ---- new:
>
> if( parameters.gridOptions(grid) &
> GraphicsParameters::plotInteriorBoundary )
> ISinteriorBoundaryPoint=MappedGrid::ISinteriorBoundaryPoint;
> else
> ISinteriorBoundaryPoint=0;
> -------
>
> After making the change just type 'make' from the main Overture directory.
>
> As a note, to send *.cmd files just rename them to *_cmd for example.
>
>
> Convex corners are a contentious issue in my mind. I am
> going to rant a bit here but don't take it personally :)
> The bottom line is that you are fooling yourself if you
> think you are getting an accurate answer near the sharp convex
> corner.
>
> I always recommend that the corners be rounded off slightly.
> As the mesh is refined the corner can be made sharper if
> desired. In this case more and more grid points are added
> to resolve the corner where the solution changes rapidly.
> You thus will get an accurate answer to a slightly wrong geometry
> rather than the wrong answer to the right geometry.
>
>
> Regards,
> Bill
>
> matthew fuller wrote:
>
>>
>> Hi,
>> I have a few questions about some false physical boundaries which
>> are plotted when using cgins.
>> To begin, i am mapping a flatback airfoil. Essentially it is an
>> airfoil with the trailing edge chopped off. This in itself proved
>> difficult to grid given my limited knowledge of ogen mappings and
>> ogens seemingly limited ability to map around sharp corners. None
>> the less i was able to get a very sharp flat trailing edge using
>> shared sides, mixed boundaries, and no hole cutting. Ogen indicates
>> my mappings overlap properly. But, when i read in this grid file to
>> cgins the mixed boundary that i specified is still plotted as a
>> physical boundary. To be clear i can tell by the flow field that
>> these "false" physical boundaries are not actually recognized by
>> cgins. They do however look strange and raise questions to their
>> origin and significance. I have found that by disabling boundary
>> plotting completely whilst in cgins gets rid them but it also gets
>> rid of the airfoil boundary. This make its harder to see where the
>> airfoil stops and the flow field begins.
>>
>> My question is, is there a way to change the representation of these
>> mixed boundary boundaries in ogen to prevent them from showing up
>> when running a simulation? I have attached a screen shot to show
>> what i mean. Unfortunately
>> the email client i am using (Gmail) will not let me send .cmd files
>> (for fear of malware i suppose) so i can not send my ogen and cgins
>> command files.
>>
>> I would appreciate any insight you can provide on this issue.
>>
>> Thanks!
>> Matt
>>
>>
>>
>> ------------------------------------------------------------------------
>>
>>
>
Other related posts:
