Thunderhead Engineering Forum

Please login or register.

Login with username, password and session length
Advanced search  

News:

Forum moved to https://forum.thunderheadeng.com

Author Topic: Mesh Sensitivity Testing  (Read 7556 times)

Kseow

  • New Member
  • *
  • Posts: 5
    • View Profile
Mesh Sensitivity Testing
« on: July 14, 2010, 04:40:00 am »

Dear All,

I have a question regarding mesh sensitivity for my FDS model. I am currently running several tests with different mesh size and number of cells. The intention to refine the mesh until there is no appreciable differences in the results. Can somebody tell me what is the appropriate simulation time to be used when conducting a mesh sensitivity study? The objective of my FDS modelling exercise to observe the results of a fire scenario at 30 minutes after a fire start. Does it mean I have to run 30 minutes for each mesh sensitivity test also? How can it affect the result of a 30minutes model if I only run 15 minutes for the purposes of mesh sensivity study? I would be grateful if somebody can response to my question.

Thank you.

Regards,
Ken
Logged

Bryan Klein

  • Thunderhead
  • *****
  • Posts: 103
    • View Profile
    • Thunderhead Eng. Site
Re: Mesh Sensitivity Testing
« Reply #1 on: July 15, 2010, 10:58:39 am »

This is a very important topic to consider when running models like FDS.  The answer is not simple, but I will do my best to point you in the right direction and hopefully give you some information and tools to determine for yourself if your FDS solutions are adequate for your needs. Note: This information may require an updated version of FDS and some will not be in the User Guide until FDS_6 is released.  But, I think it is important to answer your questions.

There was an extensive discussion on the FDS Discussion Group on a very similar topic that will be informative to review.
http://groups.google.com/group/fds-smv/browse_thread/thread/bcc91ccdd404022e

Since then, Randy McDermott at NIST has worked to develop quality assessment tools for LES calculations, specifically for FDS.  You can read about this in more detail in a paper he wrote entitled "TOWARD FDS 6: COMPLEX GEOMETRY, EMBEDDED MESHES, AND QUALITY ASSESSMENT" for presentation at V European Conference on Computational Fluid Dynamics (ECCOMAS CFD 2010) in June of 2010.
http://www2.bfrl.nist.gov/userpages/wmell/PUBLIC/TALKS_PAPERS/mcdermott_etal_FDS6_CFD2010.pdf

Pay particular attention to section 5 (page 10), where Randy covers metrics in that can be used in FDS to determine quality of the FDS mesh resolution.

To use MTR you must set CHECK_KINETIC_ENERGY -- (LOGICAL) to .TRUE. on the MISC line, which will estimate the subgrid kinetic energy after the corrector step and calculate MTR(x,t) -- the "measure of turbulence resolution".  This quantity may be output to a Slice through the area of interest by using QUANTITY=''TURBULENCE RESOLUTION'' on a SLCF line.

To do quick comparisons of SLCF output, Glenn Forney has developed a tool called SmokeDiff.  You can read more about how to use it in the Smokeview User Guide, Chapter 18. http://fds-smv.googlecode.com/svn/trunk/FDS/trunk/Manuals/All_PDF_Files/SMV_5_User_Guide.pdf
SmokeDiff is also mentioned in Randy''s paper linked above.

In the upcoming version of the FDS User Guide, there will be new information about how to use these metrics.  This version of the FDS User Guide is available in the FDS-SMV Subversion repository as a .tex file.  I have attached it to this message.  Please, keep in mind that it is not an official release of the manual, and should be taken as it is in draft form.

I hope this helps,
-Bryan Klein
Old Guy from NIST/New Guy at Thunderhead Eng.
Logged