[Nicles43]

Hi



I am trying to simulate a blunt body hypersonic vehicle



I would like to study only half of it.
The geometry is the one attached.
I divided the conical shape in the mid part. The external boundary is a semi-circonference.



However, when i put in the config file the MARKER_SYM boundary condition, the launched simulation shows to me the following error message:
------------------- Geometry Preprocessing ( Zone 0 ) -------------------
Two dimensional problem.
104284 grid points before partitioning.
206975 volume elements before partitioning.
3 surface markers.
336 boundary elements in index 0 (Marker = FARFIELD).
953 boundary elements in index 1 (Marker = SYMMETRY).
189 boundary elements in index 2 (Marker = CAPSULE).
Executing the partitioning functions.
Building the graph adjacency structure.
Calling ParMETIS... graph partitioning complete (1372 edge cuts).
Distributing ParMETIS coloring.
[nicola-VirtualBox:10959] *** Process received signal ***
[nicola-VirtualBox:10959] Signal: Segmentation fault (11)
[nicola-VirtualBox:10959] Signal code: Address not mapped (1)
[nicola-VirtualBox:10959] Failing at address: 0x91
[nicola-VirtualBox:10959] [ 0] /lib/x86_64-linux-gnu/libc.so.6(+0x43090)[0x7f86f479a090]
[nicola-VirtualBox:10959] [ 1] SU2_CFD(+0x838360)[0x56208804c360]
[nicola-VirtualBox:10959] [ 2] SU2_CFD(+0x839aac)[0x56208804daac]
[nicola-VirtualBox:10959] [ 3] SU2_CFD(+0x184279)[0x562087998279]
[nicola-VirtualBox:10959] [ 4] SU2_CFD(+0x1854ec)[0x5620879994ec]
[nicola-VirtualBox:10959] [ 5] SU2_CFD(+0x185c0c)[0x562087999c0c]
[nicola-VirtualBox:10959] [ 6] SU2_CFD(+0x18b763)[0x56208799f763]
[nicola-VirtualBox:10959] [ 7] SU2_CFD(+0x12ad42)[0x56208793ed42]
[nicola-VirtualBox:10959] [ 8] /lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xf3)[0x7f86f477b083]
[nicola-VirtualBox:10959] [ 9] SU2_CFD(+0x13b96e)[0x56208794f96e]
[nicola-VirtualBox:10959] *** End of error message ***

[Shijian-20]

Hi,

I am working on a similar problem and facing the same error message. I am not sure where the problems really are but this seems to be the problem on volume definition.

[Nicles43]

Originally Posted by Shijian-20

Hi,

I am working on a similar problem and facing the same error message. I am not sure where the problems really are but this seems to be the problem on volume definition.

Hi



I don't know why the MARKER_SYM does not work, yet.

Therefore, i decided to go on with my original complete geoemtry. I managed to reach convergence at first order just reducing the external circular domain from 80 to 50 and refining a little bit more the unstrcutred mesh, near the body.

I'm reaching convergence with RMS = 10^(-7) with a first order scheme (MUSCL = NO).

Now i am trying a second order scheme:
MUSCL = YES
SLOPE LIMITER = VENKATAKRISHNAN_WAN

VENKAT_LIMITER_COEFF= 0.01
ENTROPY_FIX_COEFF= 0.1
CFL_NUMBER= 0.15
CFL_ADAPT= YES
CFL_ADAPT_PARAM= (0.005, 1.005, 0.1, 1)
TIME_DISCRE_FLOW= EULER_EXPLICIT


However, in this case, restarting from the converged 1st order solution, su2 seems to stall at 10^(+2) or even diverges

Let's keep in touch to help each other !

[Shijian-20]

Hi,

How about using a structured mesh? Your case is on 2D domain and employ a block mesh strategy may works. Keep focus on post-cylinder shear layers and boundary layers. Also, allow the mesh on shock to be dense may also help. I am working on Mach 3 case and has not considered NEMO yet, haha.

[Tomjugj]

Originally Posted by Nicles43

Hi



I don't know why the MARKER_SYM does not work, yet.

Therefore, i decided to go on with my original complete geoemtry. I managed to reach convergence at first order just reducing the external circular domain from 80 to 50 and refining a little bit more the unstrcutred mesh, near the body.

I'm reaching convergence with RMS = 10^(-7) with a first order scheme (MUSCL = NO).

Now i am trying a second order scheme:
MUSCL = YES
SLOPE LIMITER = VENKATAKRISHNAN_WAN

VENKAT_LIMITER_COEFF= 0.01
ENTROPY_FIX_COEFF= 0.1
CFL_NUMBER= 0.15
CFL_ADAPT= YES
CFL_ADAPT_PARAM= (0.005, 1.005, 0.1, 1)
TIME_DISCRE_FLOW= EULER_EXPLICIT


However, in this case, restarting from the converged 1st order solution, su2 seems to stall at 10^(+2) or even diverges

Let's keep in touch to help each other !

Hi,

I am also facing similar divergence problem when using SU2-Nemo solver. I don't know whether the problem is the mesh or something else. Did you get convergence solution?