Dear Support and Users,
I’m trying to make calculations of flow with combustion in a simple 2D geometry. The “cold” flow calculation runs fine but when I set PC combustion (using GUI) I get the following error file:
cs_base.c:809: Fatal error.
Sub-routine calling rasize: MEMTRI
Memory needed in ra (number of reals): 1311229
available: 693840
----> Define rasize to a value at least equal to 1311229 reals).
Call stack:
1: 0x7fed0d99f055 <rasize_+0xc5> (libsaturne.so.0)
2: 0x7fed0dae57a7 <memtri_+0x5bf> (libsaturne.so.0)
3: 0x7fed0d979807 <caltri_+0x7b7> (libsaturne.so.0)
4: 0x7fed0d9594db <cs_run+0x83b> (libsaturne.so.0)
5: 0x7fed0d9597c5 <main+0x1f5> (libsaturne.so.0)
6: 0x7fed09ef8abd <__libc_start_main+0xfd> (libc.so.6)
7: 0x4007a9 ? (?)
End of stack
Could anyone give some idea what can be the source of it?
Hi Norbert,
This is a matter of memory management. In the Fortran part, we allocate two big arrays (one for integers, or for reals) and we try to guess their size at the start of the calculation. However, for some calculations like combustion ones (where the number of equations increases), the user has to adapt the size of the arrays. This can be done in the graphical interface or at the bottom of the usini1 subroutine (assuming you are using 2.0 version).
The error message you get can give you a clue on how much memory you need for each array.
David
Hi David,
Thanks for the reply. I have another question regarding coal combustion. My solution seems to converge but there are some strange effects concerning the temperature contours (take a look on the attachment). Another thing is that in some cells the total enthalpy falls below zero. The problem appears both with and without radiation model. Do you have any idea what can be the reason? Is it possible that solution domain beeing 2D (cell dimension = 1 in z direction, symmetry BC in X||Y planes) not 3D has influence on this error?
Hi Norbert,
Concerning the enthalpy going below zero, this is (generally) due to the non-respect of the min/max principle at the discrete level. The 2nd-order centered scheme is precise but does not respect this principle, contrary to the 1st-order upwind scheme. So the solution is either to switch to the upwind scheme (for the enthalpy variable) or to clip the enthalpy to zero by minimum values (this might be done by default, I don’t remember).
As regards the temperature issue, this is somewhat weird… I need to talk with the combustion guys 
David
Hi David,
I have used the upwind scheme for the enthalpy but it still falls below zero. In which subroutine can I clip the enthalpy to zero? Can’t find it.
Norbert
Previously Norbert Modlinski wrote:
Hi David,
Thanks for the reply. I have another question regarding coal combustion. My solution seems to converge but there are some strange effects concerning the temperature contours (take a look on the attachment). Another thing is that in some cells the total enthalpy falls below zero. The problem appears both with and without radiation model. Do you have any idea what can be the reason? Is it possible that solution domain beeing 2D (cell dimension = 1 in z direction, symmetry BC in X||Y planes) not 3D has influence on this error?
Hi David.
The enthalpy of a hydrocarbon fuel generally has a negative value. So having negative value of the enthalpy in the field is nothing unexpected.
Johannes
Thank you for the reply.
I have one more issue concerning coal combustion. When running the simulation for one size class of coal the calculation runs fine but when I set 2 or more size classes i receive the following error: SIGSEGV signal (forbidden memory area access) intercepted! I set boundary conditions in GUI. Please take a look on listing file attached. I shall be very glad for any suggestions and comments.
Best regards
listing-07250952.txt (3.76 KB)
dp_fcp-07250952.txt (3.59 KB)
Hi
I’m trying to modelling coal combustion and I have the same problem. If i choose two or more coal classes calculations stops with an error. Any ideas how to solve this issue ? Have a look into listing file.
Pawel
listing-08031835.txt (3.67 KB)
Hello,
Look like you have the same issue. Could one of you send me your xml and dp_* file ? (I won’t need the mesh at this stage, as the crash seems to occur before the mesh is read). The crash is probably due to a NULL string (in other words, an inconsistency beween what the GUI produces and what the code expects to read, or a bug in the XML reader), and should be easy enough to find (the earlier and more systematic the crash, the easier to locate).
Best regards,
Yvan
Hi Yvan,
So far I have found out that the crash occurs in “cs_gui_model_property_post” referenced by CS_PROCF (csenso, CSENSO) (line 4048 of cs_gui,c). Bellow you can find the attached files.
Best regards
Norbert
dp_fcp.txt (3.59 KB)
case.txt (48.9 KB)
Hello,
I can trace the crash being related to the fact that label Ga_HET_O202 is not defined in the XML file (Ga_HET_O201 is defined, but no Ga_HET_02 is found for x > 1.
In the GUI, if I do not start with the dp_FCP file, only one class of coal appears, but if I add another, the label is not added. In addition. If I rename your dp_fcp to dp_HCP, the GUI is consistent with your file (1 coal, 8 classes), but at least one label for class2 (Ga_HET_O202) does not appear.
This seems like a GUI bug leading to a reader crash, but I don’t know much about the combustion modeling version, so I’ll try to check with Alexandre tomorrow.
Best regards,
Yvan
Hi all,
If Alexandre is not around these days, could you open a new ticket in the bug-tracker so that we don’t forget this issue?
Thanks in advance anyway.
David
