FrontISTR v4.2c Release Notes¶

August 2, 2013

Analysis and Response to Bug Reports Reflected in FrontISTR Ver.4.2c (130928)

Suenori / Okuda

Document last updated: 131203

(1) From Mr. O of an information technology company

About FrontISTR defects.pdf (About defects in FrontISTR_V42b(130628))

1. Spelling of messages (Ver.3.4)

2. Error at the array boundary check (Ver.3.4)

3. Thermal stress analysis (Ver.3.4) examples exF F300.cnt terminates with an error message.

4. About UTABLE (Ver.3.4)

5. Termination by MPI_Abort (Ver.3.4) examples/dynamic/exW

6. Element group warning message (Ver.3.4) eigenvalue analysis of tutorial/17_freq_beam

7. Frequency response analysis (Ver.3.4) frequency response analysis of tutorial/17_freq_beam

Terminated with a segmentation fault.

8. Parallel contact analysis using MUMPS (Ver.3.4) tutorial/11_contact_2beam contact analysis with MUMPS terminated with a segmentation fault.

9. Femap neutral file I/O (Ver.3.4)

9.1. Femap neutral format output of the analysis results for 2D elements

9.2. Femap neutral format output of shell element stresses

9.3. Errors in neu2fstr

10. When FrontISTR 4.2 is compiled using OpenMPI (Ver.4.2) Running fistr2 causes a segmentation fault.

11. Refinement in pressure load analysis (Ver.3.4) When using an element group and load type P1, and further performing refinement one level, the displacement becomes 2x that of when refinement was not performed.

12. Visualization data output in parallel analysis on multiple nodes (Ver.3.4) When node-specific disks are created in the working directory, is visualization data output in the AVS format etc. possible?

1 and 2: These are bugs. Resolved.

3: This is a bug. Resolved.

4: The UTABLE function is not public. Also, it is out of scope for output cleanup, and no final test has been performed.

Corrected the cnt files in examples

5: This is a bug. Resolved.

The reason hecmw_abort was called is

Displacement increment too large, please adjust your step size!

Kept only the data in examples/dynamic that does not produce errors, and deleted the rest

6: Data correction for tutorial

That the message display differs from the cause of the error is a bug. Resolved.

7 and 8: These are bugs. Resolved.

9.1: We have decided not to perform FEMAP output maintenance.

However, the implementation is such that if SECOPT in the mesh file's !SECTION is set correctly, the corresponding element types will be output.

9.2: We have decided not to perform FEMAP output maintenance.

However, separately from this point, the node output order of the 232 and 732 shell elements that was pointed out has been corrected.

9.3: We have decided not to perform neu2fstr maintenance.

This point alone cannot be resolved, and another error occurs.

However, since it runs to completion on WINDOWS, it was bundled with the WINDOWS version

10: hecmw2+fistr2 works only with

gnu + mpich2 (on cygwin)

intel + intelMPI

and not with anything else. Problems remain in hecmw2.

11: Since elements are duplicated top and bottom, the number of faces doubles. A response is difficult.

12: This can be addressed by placing hecmw.ini in advance.

(2) From Mr. A of Manufacturer I: FrontISTR_Thermal stress calculation with changed calculation conditions.pdf Defects in visualization when using the refiner

When using the refiner, the only file that can be used as-is for post-processing is COMPLETE_SEPARATE_AVS (UCD output per domain).

Please carry out the method explained in "Serial and Parallel Analysis" of the 1st FrontISTR Study Group.

(3) From Mr. U of S Construction

(3-1) First email of 130722.docx

(1) Points noticed when compiling FrontISTR_V42a.tar.gz.

We are aware of the warnings when compiling hecmw2, but the current situation is that there are a large number of them and they have not been addressed. The partitioner is incomplete.

(2) Ran example as a serial job. Suspected bugs.

(2)-1 Output of the FEMAP Neutral file: the order of the nodes on the edges of quadratic triangular elements (ietyp=232, 731) is different.

This is a bug that has been around for a long time. Fixed.

(2)-2 Out-of-range reference

(The one you pointed out) will be resolved by the countermeasure for the first 2.1.

(3) Looking at the results of examples/static/exA/A231 and A232, out-of-plane deformation is occurring.

This is a bug in FEMAP output. Not addressed.

(3-2) Bug report from the 130731 email

1: In F231, F232, F241, and F242 of example/static/exF, the input temperature data does not seem to be reflected in the program.

This is a bug. Resolved.

2: In F341, F342, F351, F352, F361, and F362 of example/static/exF, it terminates after outputting the error "Fails in fetching expansion coefficient!".

This is a bug. Resolved.

3: A variable named " M_EXAPNSION" is used — is this a spelling mistake?

The spelling is indeed wrong, but since it is a program, it will not be corrected.

4: The results of a model using first-order prism elements differ by 1% from the values in the manual.

It was fixed four years ago, and the computation results themselves are the same as Abaqus.

5: The number of arguments is different

It is added automatically in FORTRAN calls, so there is no problem.

5 to 7: Handling of the " declared implicitly" warnings (errors with the -Werror option)

Fixed.

8: I tried fixing hecmw_fstr_output_femap.c.

Thank you. We will use it as a reference. It has not been reflected in the public version.

End