Using restart file at higher mesh level

Numerical methods and mathematical models of Elmer
Post Reply
mark smith
Posts: 215
Joined: 26 Aug 2009, 18:20
Location: Peterborough, England

Using restart file at higher mesh level

Post by mark smith »

Hi All,
I have done a thermal+fluid simulation at "Mesh Levels = 1" and wondered if there is any way of interpolating the results from the "Output File = results.dat" as a restart file when I increase the mesh refinement to "Mesh Levels = 2" as trying it in a straight forward way doesn't work ;-(
TIA, Mark
raback
Site Admin
Posts: 4812
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: Using restart file at higher mesh level

Post by raback »

Hi Mark

This would probably require some development. There is a possibility nowadays to use different mesh for startup but still you need to access the two meshes. Probably would not work out of the box as you desire.

Also, consider that time consumption is linear function at best, but probably scales as O(1.5), for example. Then in 3D with increasing mesh levels you get 2^3 bigger problem which uses at least 8 times the time. Hence a good initial start won't help much in time consumption. Well, it might help quite a bit if you have a poorly converging system and can use direct solvers for the lower level, for example. Still, I would focus on finding best possible linear solvers on the fine mesh.

-Peter
mark smith
Posts: 215
Joined: 26 Aug 2009, 18:20
Location: Peterborough, England

Re: Using restart file at higher mesh level

Post by mark smith »

Hi Peter,
Thanks for the reply.
I did a simulation at Mesh levels=2, and saved the results so I could do a restart with different solid material but when I try a restart at Mesh levels=2, the simulation doesn't run see attached log file, the vtu results files from the first run show the results on the refined split mesh I assumed the results file were also on this split mesh and a restart should be possible?
To rerun the case on 8 processors would take 7 hrs where as a restart should be a few 10's of mins so I am keen on doing a restart if at all possible.

N.B. A restart works at mesh level=1 using the results from a mesh level=1 simulation

Best regards
Mark
Attachments
log.txt
log of run console output
(13.22 KiB) Downloaded 224 times
raback
Site Admin
Posts: 4812
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: Using restart file at higher mesh level

Post by raback »

Hi Mark,

Yes, there is a problem. Elmer by default saves the lower mesh levels. They can be used for multigrid levels, or to save results on a coarser mesh etc. But usually not needed. Then set,

Code: Select all

  Mesh Levels = 2
  Mesh Keep = 1
I may tinker the operation...

-Peter
mark smith
Posts: 215
Joined: 26 Aug 2009, 18:20
Location: Peterborough, England

Re: Using restart file at higher mesh level

Post by mark smith »

Hi Peter,
Thanks for the quick reply, i'll try later & report back.
Best regards
Mark
mark smith
Posts: 215
Joined: 26 Aug 2009, 18:20
Location: Peterborough, England

Re: Using restart file at higher mesh level

Post by mark smith »

Hi Peter,
Worked perfectly, thanks ;-)
Regards
Mark
Post Reply