July 24, 2014, 07:26
|
Systematic FLUENT AMG DIVERGENCE with Delaunay Mesh
|
#1
|
|
New Member
Join Date: Nov 2012
Posts: 6
Rep Power: 12
|
Hi all,
I'm currently have issues with a transient 6DOF simulation using a DELAUNAY type mesh (created by ICEM v13), + remeshing + smoothing with Fluent v13.
In some moment of the simulation, i have an AMG divergence on epsilon whern I run the job on 128 cores (batch, 8 nodes). A high peak on the turbulence residuals appear and it crashes.
Surprisingly, when I run the same job on 16cores (1nodes, graphically, with the Fluent user interface), the peak also appear, but the solver manages tu continue the simulation nad finally reaches good residuals...
Remark : generating an OCTREE type mesh in ICEM, the simulation is completely succesful.
- Do you ever heard that the Fluent remeshing algorithm is inccompatible with DELAUNAY mesher ?
- Have you ever experienced such issues with divergences du to parallelization ?
Thks by advance !
Ben
|
|
|
Similar Threads
Linear Mode
