anisotropic mesh adaptation
([2]) which construct completely new mesh in the sense
that the interpolation error is uniformly distributed over the whole
computational domain
Both of these method are applied for the flow through GAMM channel and
they are compared with the results obtained without local mesh adaptation.
There exist a shock wave for inlet Mach number M_{int} = 0.67.
The shock wave is a discontinuity but in numerical realisation it is only
a jump of quantity with a very high gradient.
Our aim was to obtain for all three method results with high quality,
which is meassured as the magnitute of gradient of Mach number on shock
wave.
Computational results
Method |
number of elements |
CPU times |
gradient of shock waves |
Links to results |
global refinement |
108 240 |
21 600 s |
69.5 |
GLOBAL |
multlevel mesh refinement |
2 813 |
240 s |
535.5 |
MMR |
anisotropic mesh adptation |
2 470 |
780 s |
298.1 |
AMA |
This table shows that apllying adaptive strategies (AMA or MMR) we
have results of higher quality (higher gradient) than for global
refinement. Also the CPU-time and number of elements was multiple smaller.
All three strategies captured the Zierepp singularity (local minimum behind
the shock wave).
Global mesh refinement
The triangulation and isolines of Mach number.
Some details of results are available here.
Multilevel mesh refinement
The triangulation and isolines of Mach number.
Some details of results are available here.
Anisotropic mesh adaptation
The triangulation and isolines of Mach number.
Some details of results are available here.
The summary results are available here.
Bibliography