[FLASH-USERS] problems with refinement in FLASH3.1
Tomasz Plewa
tplewa at fsu.edu
Mon May 11 16:16:50 EDT 2009
Seyit and All -
Data at refinement level boundaries, as mentioned by Latif, does
necessarily come from interpolation. If variation in the solution is
substantial, interpolation may produce unphysical state.
FLASH2.5 used different interpolation module than FLASH3. I have
originally implemented monotonic interpolation for the AMRA code; the
interpolation driver has later been ported to FLASH2 and used only for
in-house applications. It seems it has been made (with some
modifications in the driver part) a default/only interpolation method
for FLASH3.
The bottom line is monotonic interpolation differs from default FLASH2.
Tomek
--
Seyit Hocuk wrote:
> Tomek, it was not like this in Flash2.5. You can even see this if you
> use different refinement levels with the Jeans setup making
> perturbations slightly stronger. It is easily reproducible. I think
> guard cells get their information from parents and we know that they
> are not evolved in Flash.
>
>
> Tomasz Plewa wrote:
>> Seyit, Latif, and All -
>>
>> It is not advisable to place f-c (refinement) boundaries in regions
>> of strong gradients. If we do, things will very likely go badly.
>>
>> As a rule of thumb, we would like to have a dozen of zones per scale
>> height implying perhaps no more than 10% jumps in critical variables
>> (density, pressure, velocity).
>>
>> Tomek
>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: tplewa.vcf
Type: text/x-vcard
Size: 296 bytes
Desc: not available
URL: <http://flash.rochester.edu/pipermail/flash-users/attachments/20090511/d4679532/attachment-0001.vcf>
More information about the flash-users
mailing list