[Octagon]

Tried to search the forum for a similiar problem but couldn't find anything.

I try to model the first stage of a compressor turbine. The problem itself is quite simple, steady state, no cooling, constant total pressure/temperature profile at the inlet and constant averaged static pressure at the outlet. I do, however, need to resolve the boundary layers and to fulfill 0.01<y+<8. The mesh is structured and the tip clearance is modeled which I believe is where the problem occurs.

The problem is that the residuals converges ok to about constant values for ~ 500 iterations then suddenly crashes, see output from the last two iterations below. I monitor the Mach number which stabilzes with the residuals until the last iterations where the Mach number increases from 1.3 to 1e12. I know that the flow is subsonic in the whole domain except at the tip clearance. I have tried both to coarsen and to refine the mesh locally at the tip but without any success.

The solution is initially quite sensitive and is run with a very small timescale at the begining (0.01) but successively ramped up to 10 later on. Niether the residuals or the Mach number seems to change with change in timescale which led me to believe that the solution was converged before the crash.

Does anyone have any idea how to prevent this, or if the pre-crash result could be used?

OUTER LOOP ITERATION = 747 CPU SECONDS = 6.050E+05
----------------------------------------------------------------------
| Equation | Rate | RMS Res | Max Res | Linear Solution |
+----------------------+------+---------+---------+------------------+
| U-Mom | 0.97 | 2.3E-04 | 2.8E-02 | 2.0E-02 OK|
| V-Mom | 1.01 | 1.9E-04 | 2.3E-02 | 5.9E-02 OK|
| W-Mom | 1.03 | 9.9E-05 | 2.6E-02 | 2.9E-02 OK|
| P-Mass | 0.99 | 1.6E-05 | 4.0E-03 | 8.0 5.6E-02 OK|
+----------------------+------+---------+---------+------------------+
| H-Energy | 1.03 | 2.6E-04 | 3.9E-02 | 8.0 6.0E-02 OK|
+----------------------+------+---------+---------+------------------+
| K-TurbKE | 1.00 | 8.1E-05 | 1.9E-02 | 8.0 2.2E-02 OK|
| O-TurbFreq | 1.01 | 1.9E-05 | 3.9E-03 | 7.9 4.2E-03 OK|
+----------------------+------+---------+---------+------------------+
================================================== ====================
OUTER LOOP ITERATION = 748 CPU SECONDS = 6.057E+05
----------------------------------------------------------------------
| Equation | Rate | RMS Res | Max Res | Linear Solution |
+----------------------+------+---------+---------+------------------+
| U-Mom | 0.03 | 6.1E-06 | 4.2E-03 | 9.3E+10 * |
| V-Mom | 0.02 | 4.1E-06 | 3.2E-03 | 1.6E+11 * |
| W-Mom | 0.06 | 6.2E-06 | 3.5E-03 | 7.4E+09 F |
| P-Mass | 0.00 | 1.9E-08 | 1.1E-05 | 15.0 1.2E+10 * |
+----------------------+------+---------+---------+------------------+
+--------------------------------------------------------------------+
| ****** Notice ****** |
| A wall has been placed at portion(s) of an OUTLET |
| boundary condition (at 55.1% of the faces, 44.1% of the area) |
| to prevent fluid from flowing into the domain. |
| The boundary condition name is: Outlet Outlet. |
| The fluid name is: Air Ideal Gas. |
| If this situation persists, consider switching |
| to an Opening type boundary condition instead. |
+--------------------------------------------------------------------+
| H-Energy |23.54 | 6.0E-03 | 4.4E-01 | 8.0 5.5E-04 OK|
+----------------------+------+---------+---------+------------------+
| K-TurbKE |38.03 | 3.1E-03 | 2.0E-01 | 8.0 6.5E-04 OK|
| O-TurbFreq |82.87 | 1.5E-03 | 1.0E+00 | 77.5 5.9E-13 OK|
+----------------------+------+---------+---------+------------------+

... and so on ---> Floating point exception: Overflow

[ghorrocks]

FAQ: http://www.cfd-online.com/Wiki/Ansys...do_about_it.3F

In this case as you have had a run which has gone for quite some time then suddenly crashed. This is usually because something has been slowly developing as the simulation progresses, and has now reached a critical point of the simulation which cannot handle it. Examples could be a heat plume reaching the exit, a shock wave reaching the exit or reflecting off a critical component. In this case I suspect a shock wave - so I would have a close look at a result saved just before the crash and see if you can see a shock wave about to hit something.

[Cosme]

Hi Glenn

I have the same situation in the domain that I am modeling.

In my case, I model the inside of a boiler, and my residual begin to oscillate slowly. It will be because I gave an initial solution that is not fully converged (The User Points from the previous simulation are stable).

Greetings.

[ghorrocks]

This is a FAQ as well: http://www.cfd-online.com/Wiki/Ansys...gence_criteria