New version of LS-DYNA is released for all common platforms.
Release notes for Version R9.2.0.
Herein are summarized most of the new features, enhancements, and significant corrections made since the release of version R9.1.0. These notes are arranged by topic. Understand that in many cases, a particular note may have applicability to multiple topics. Each item is listed only once, under a single category.
- Enhance the robustness of *AIRBAG_INTERACTION to address instability, especially in MPP, when excessive mass exchange is yielded due to numerical noise.
- CPM now supports leakage parameter from *MAT_ADD_AIRBAG_POROSITY_LEAKAGE.
- Change dm_out calculation to better treat trunction error.
- Fix bug affecting *AIRBAG_CPM_INTERACTION if the slave bag has chamber definition. The downstream pressure should use the slave chamber pressure for evaluating the venting from the master bag.
- Fix bug in Structured ALE solver. The bug produced erroneous internal energy increase.
- Fix problem of incorrect motion of rigid part when prescribing multiple translations with *BOUNDARY_PRESCRIBED_MOTION_RIGID while rotational DOF are constrained on Card 2 of *MAT_RIGID.
CESE (Compressible Fluid Solver)
- Corrected an internal problem with BLAST=2 in *LOAD_BLAST_ENHANCED blast waves used as a CESE boundary condition. This was likely an issue only if more than one blast was simultaneously in use.
- Added an error message to clarify that the CESE solver has only BLAST=2 type blasts available as boundary conditions in *LOAD_BLAST_ENHANCED .
- Added real 2D CESE output to d3plot, and this is confirmed to work with LS-PrePost 4.3 and later versions. This also works for d3plot output with the 2D CESE axisymmetric solver.
- For all immersed-boundary CESE solvers, corrected the plotting of the Schlieren number and the chemical species mass fractions.
- Brought in enthalpy-related corrections to the CESE chemistry solvers.
- Enabled output of the timing information for the CESE chemistry solvers.
- Fixed an issue with opening some LSDA-based chemistry results files.
- Fix bugs in reading of *CESE_BOUNDARY_PRESCRIBED_SEGMENT.
- Fix bugs affecting contact in high order shell formulations when:
- using *ELEMENT_SHELL_SHL4_TO_SHL8, or
- using a large part id like 100000001.
- Implement consistency mode for *CONTACT_AUTOMATIC_SURFACE_TO_SURFACE_COMPOSITE when ncpu<0. Applies to SMP only.
- Fix zero contact stiffness if *EOS_TABULATED is using LCC instead of tabulated input cards.
- Allow *CONTACT_TIED_SURFACE_TO_SURFACE to use MAXPAR in Optional Card A It was previously hardcoded to 1.01.
- Fixed SMP segment based contact with CNTCO.ne.0 on *CONTROL_SHELL. Nodal coordinates were being changed. This problem existed in R9.1.0 but not in prior releases.
- Fixed a potential segmentation fault in MPP *CONTACT_2D_AUTOMATIC when a process has no segments.
- Fixed a flaw in the support of SSF on *PART_CONTACT when using segment based (SOFT=2) contact with DEPTH=5. The consequence was to sometimes choose the wrong part when looking up the scale factor so the stiffness was not predictable.
- Fixed MPP’s implementation of the variable SSF in *PART_CONTACT when used with SOFT=2 contact. It was possible that segments passed from one processor to another for contact checking would have a zero stiffness and would therefore penetrate.
- Added a warning message (STR+1392) for those trying to use the ORTHO_FRICTION contact option with SOFT=2 contact, because that option is not available. The contact type is switched to SOFT=1.
- Fixed an error in MPP *CONTACT_2D_AUTOMATIC_SURFACE_TO_SURFACE when used with node sets to define the contact surfaces. The master side was likely to trigger an erroneous error about missing nodes, which terminated the job.
- Add minor change for *USER_INTERFACE_FRICTION: yield stress data of *MAT_TABULATED_JOHNSON_COOK/*MAT_224 provided as input in subroutine usrfrc.
- Fix for *CONTACT_AUTOMATIC_…_TIEBREAK with OPTIONs 10 and 11 (only MPP). Compressive contact forces were erroneously “damaged” as well.
- The MPP version of *CONTACT_AUTOMATIC_BEAMS_TO_SURFACE_ID now supports the contact of the beam nodes from both sides of the contact surfaces.
- Add warning for duplicate *CONSTRAINED_SPR2/*CONSTRAINED_INTERPOLATION_SPOTWELD.
- The SPR models (*CONSTRAINED_SPR2/*CONSTRAINED_INTERPOLATION_SPOTWELD) now support the SPOTDEL option of *CONTROL_CONTACT. That means if shell elements involved in the SPR domain fail, the SPR gets deactivated.
- Make *TERMINATION_DELETED_SOLID work with hex spot weld failures.
- *CONTROL_TIMESTEP: Beam type 9 (spotweld beam) and thick shells are supported for selective mass scaling.
DEM (Discrete Element Method)
- Fix bug affecting *DEFINE_DE_TO_SURFACE_COUPLING capillary force calculation.
- Clarify sometimes incorrect message about spotweld failure.
- Speed up keyword processing for models that use lots of *ELEMENT_SOLID_TET4TOTET10.
- Fix a bug affecting time-dependent slipring friction in 2d seatbelts.
- Fix erroneous 1d seatbelt slipring message that has been there since R8.0.
- Support *element_mass_matrix_node_(set) in MPP.
- Fix memory error when using *NODE_THICKNESS_SET and also fix node set not found error when using *NODE_THICKNESS_SET with *SET_NODE_INTERSECT.
- Fixed the fully integrated membrane element (shell ELFORM=9) when used with NFAIL4=1 on *CONTROL_SHELL and there are triangular elements in the mesh. Triangular elements were being deleted immediately by the distorted element check.
- Modify tolerances on error messages SOL+865 and SOL+866 to prevent unnecessary error terminations. This does not affect results, but allows models to run. SOL+865 is an error that occurs if a discrete element has zero translational mass. SOL+866 is an error message that occurs when a discrete element has zero rotational mass.
- Enable *PART_STACKED_ELEMENTS to be used with adaptivity (ADPOPT).
- Increase accuracy of cohesive elements type 19/21 by using double precision for local coordinates and other internal variables.
- Add alternative mass calculation for critical time step estimate of cohesive elements. This hopefully resolves rarely occuring instability issues. Option ICOH on *CONTROL_SOLID is used for that; see R10 manual for details.
- Fix for solid element form 15 (pentahedron) when used together with *MAT_ADD_EROSION. Inconsistent results were possible since version R9.0.
- Make a few fixes and enhancements for pressure averaged tetrahedron element (*SECTION_SOLID, ELFORM=13), e.g., correctly handle initial volumes from *INITIAL_STRESS_SOLID, use double precision for some internal variables.
- *CONTACT_FORMING_NODES_TO_SURFACE, SOFT=6: Bug fix for duplicate nodes generated for blank edge to gage pin contact.
- Fix in the trimming of sandwich parts – irregular trimmed edge.
- adapt.msh file format fix: Remove an extra blank line output in *ELEMENT_SHELL_THICKNESS.
- Bug fix in *CONTACT_AUTO_MOVE: part did not move as expected.
- *CONTROL_FORMING_OUTPUT_INTFOR: Fix excessive INTFOR file output.
- Frequency domain (BEM acoustics): Fix for running acoustic analysis with multiple b.c. in MPP.
- Frequency domain (SSD fatigue):
- Fixed bug in running SSD fatigue in MPP.
- Fixed a bug in running interpolation on load curve when there is only one frequency in the load curve.
- Fixed a bug in running FREQUENCY_DOMAIN_SSD with hex20.
- Frequency domain (response spectrum):
- Added missing arguments.
- Fixed a bug in extracting modal stress for response spectrum analysis.
- Added von mises stress output for beams.
- Frequency domain (FEM Acoustics)
- Fixed a bug in running FEM acoustics with existing exterior structures.
- Minor fix for writing eigout_ac, with case id.
- Frequency domain (FRF):
- one more fix for reading FRF keyword.
- Minor fix for reading FRF keywords.
- Fix for running frequency domain analysis with higher order elements.
- Frequency domain (FRF, SSD, Random vibration): Fixed a bug in accessing eigenmode databases, when *FREQUENCY_DOMAIN_PATH is defined.
ICFD (Incompressible Fluid Solver)
- Fixed bug that occurred when using birth time for FSI.
- Fixed the stabilization of the thermal problem to avoid some possible MPP issues.
- Skip structural decomposition in MPP if there are no (structural) nodes.
Implicit (Mechanical) Solver
- Fix the marking of the end of the d3mode file when writing stresses. Fix a problem marking the end of the d3eigv file when modal stresses are written.
- Adjust implicit logic for secforc output for arclength method.
- Improve treatment of termination of MPP eigensolver when no eigenmodes are found.
- Implicit was enforcing birth and death times on *BOUNDARY_SPC during dynamic relaxation contrary to the manual. Such times are now ignored with dynamic relaxation.
- Correct output of eigenvalues and frequencies for the unsymmetric eigenvalue problem to match that of rotational dynamics.
- Enhance logic that determines when to write out the last state to d3plot for implicit.
- Fix *PART_MODES in MPP so the part in *PART_MODES is NOT split across processes. A future enhancement is to allowed a part in *PART_MODES to be split across processes.
- When implicit springback followed an explicit transient step, the implicit keywords with the _SPR option were not properly handled. Now such keywords are properly handled.
- Correct initialization of MPP tied contact for implicit mechanics when implicit follows explicit dynamic relaxation.
- Correct linear implicit analysis to allow for inertia relief computations.
- *CONTROL_IMPLICIT_ROTATIONAL_DYNAMICS: Fix a problem when using 4- or 6-noded solids, or 3-noded shells in rotational dynamics analysis.
- Add variable ‘IZSHEAR’ in *INITIAL_STRESS_SECTION to initialize shear stress.
- Fix incorrect initial velocity when ICID.ne.0 in *INITIAL_VELOCITY_GENERATION, and rotational velocity OMEGA is not zero and *PART_INERTIA is also present.
- Fix *INCLUDE_TRANSFORM offset for coordinate systems in *LOAD_NODE_POINT which were using the wrong offset
- Fixed *LOAD_THERMAL_VARIABLE when used with shell element form 2 that has default warping stiffness (BWC=2 on *CONTROL_SHELL), and when OSU=1 on *CONTROL_ACCURACY to activate an objective stress update. Prior to the fix, there was no thermal loading available for this combination of options.
- Box option in *LOAD_SUPERPLASTIC_FORMING now works in MPP.
MAT and EOS
- Fix immediate deletion of elements at time 0 when using *MAT_PIECEWISE_LINEAR_PLASTICITY with FAIL>0.0 and IACC=1 in *CONTROL_ACCURACY.
- Fix spurious element deletion when using table for LCF in *MAT_TABULATED_JOHNSON_COOK/*MAT_224 and *MAT_TABULATED_JOHNSON_COOK_GYS/*MAT_224_GYS.
- Fixed two problems in *MAT_FABRIC forms 0 and 1, both occurring when there are mixed quad and triangular elements in the fabric part:
- when a reference geomeotry is used, a flaw in the strain calculation was leading to possible NaN forces in the elements, and
- when a reference geometry is not used, the forces from triangular elements in mixed element blocks were 2 times too high.
- Correct shear failure behavior in *MAT_LAMINATED_FRACTURE_DAIMLER_CAMANHO/*MAT_262. This will most probably have no effect to any real application, but could be seen in very special 1-element tests.
- Bugfix for shear stiffness behavior in *MAT_058 when using a table definition for GAB and only providing stress-strain-curves for positive shear.
- Bugfix for strain-rate dependent stiffness behavior in *MAT_058 when using a table definition for EA, EB or GAB under compressive loading.
- Add option for *MAT_GLASS/*MAT_280: New parameter FTSCL on Card 2, field 8, is an optional scale factor for the tensile strength: FT_mod = FTSCL * FT. As soon as first crack happens in a part, tensile strength drops to its original value, FT.
- Activate *MAT_GLASS/*MAT_280 for tshell formulations 1 and 2.
- Make AOPT<0 work for *MAT_TRANSVERSELY_ISOTROPIC_CRUSHABLE_FOAM/*MAT_142.
- Bug fix for *MAT_TRANSVERSELY_ISOTROPIC_CRUSHABLE_FOAM/*MAT_142: Tensile behavior under 45 degree to the material direction was incorrect.
- Back port *MAT_TABULATED_JOHNSON_COOK_ORTHO_PLASTICITY/*MAT_264 for use in R9.2.0; see R10 manual for details of this material.
- Add some more history variables for *MAT_TOUGHENED_ADHESIVE_POLYMER/*MAT_252 e.g. current yield stress (5) and triaxiality (7); see R10 manual for details. This affects only the output.
- Fix for history variable organization in *MAT_036 used with shell forms 25-27.
- Fix *MAT_PHS_BMW/*MAT_248 for implicit analysis: convergence speed could have been low since version R9.0.
- Add mid-plane failure option to GISSMO (*MAT_ADD_EROSION): New parameter MIDFAIL=1 (Card 4, field 8) activates a procedure, where critical strain is only checked at the mid-plane integration point(s), i.e. an odd number for NIP should be used. The other integration points compute their damage, but no coupling to the stresses is done. As soon as the mid-plane IP reaches ECRIT/DCRIT, all the other IPs are now also checked. Those of them that are already “above” their critical value immediately start to reduce the stresses. Those who are still “below” critical still do not couple, only if they reach their criterion.
- Fix for rarely occuring divide by zero problem in GISSMO (*MAT_ADD_EROSION). Could only happen sometimes if ECRIT=0 and DCRIT=1.
- Add pre-defined tensors option PDDT and optional damage limitation curve/table LCDLIM to *MAT_ADD_GENERALIZED_DAMAGE; see R10 manual for details.
- Extended option IFLG1 of *MAT_ADD_GENERALIZED_DAMAGE (similar to DMGTYP in GISSMO): IFLG1 = “flag 1” (0 or 1 or 2) + 10 * “history variable slot”.
- Add optional failure criterion FFCAP to *MAT_100_DAMAGE-FAILURE: Failure function limit for OPT = 0 or -1, and DMGOPT = 10, 11, or 12. See R10 manual for details.
- Fix for rarely occuring division by zero (NaN) in *MAT_024 shells (VP=1) with logarithmic strain rate table.
- Add warning message for *MAT_FABRIC with nonzero value for PRAB. Results could have changed between R7 and R8.
- Fix for *MAT_224 plastic behavior with shell elements. Sometimes temperature dependent yield curves (LCKT) were not used correctly.
- Reinstate _STOCHASTIC option for *MAT_123 (shell elements).
- Fix for *MAT_CABLE_DISCRETE_BEAM/*MAT_071 with large values (>2**24) for LCID.
- Fix for implicit analysis with *MAT_058 and shell elements. Convergence speed could have been low in versions R9.0 and R9.1.
- Fix for *MAT_ADD_COHESIVE with INTFAIL > 1. It was possible that elements were not eroded even if enough integration points reached their failure criterion.
- *MAT_036E/*MAT_EXTENDED_3-PARAMETER_BARLAT: The extended Barlat model is implemented, where r00/r45/r90/shear/biaxial hardening and r-values can be specified independently to render more accurate results. Also, hardening flag 7 in the regular Barlat model (*MAT_036/*MAT_3-PARAMETER_BARLAT) is restored to behave as it did in earlier versions.
- Reduce memory footprint (as reported by the OS, not the program) during MPP decomposition.
- Multiple fixes related to memory allocations and array indexing errors.
- Fixed an error that could cause an MPP job to hang in phase 3 decomposition. The error could occur when SOFT=2 contact is used with the periodic intersection check and process 0 does not participate in the contact.
- Improve cost estimation in MPP decomposition for *MAT_030, *MAT_120, *MAT_123, and *MAT_ADD_EROSION with IDAM.ne.0 (GISSMO or DIEM).
- MPP version now reports the eroded hourglass energy in glstat.
- Fix missing plastic strain tensors in d3plot when STRFLG in *DATABASE_EXTENT_BINARY is set and INTSTRN=1 in *INTERFACE_SPRINGBACK.
- Fix for *DATABASE_FATXML to cope with new formats; see R10 manual for details. FATXML is an open, standardized data format based on the Extensible Markup Language (XML) which is developed by the German Research Association of Automotive Technology (Forschungsvereinigung Automobiltechnik – FAT). It is designed for consistent data management in the overall CAE process chain.
- Fix bug in writing *SECTION_*_TITLE to d3prop file. The TITLE in *SECTION_*_TITLE was not written correctly to d3prop.
- Add FILE option to define the interface force file name (*DATABASE_BINARY_INTFOR_FILE).
- Fix corrupt d3plot database triggered by DECOMP=4 in *DATABASE_EXTENT_BINARY.
- Properly start up Explicit LaGrange Multiplier treatment of joints upon restart.
- Fix memory bug during full deck restart when both *INITIAL_VELOCITY_GENERATION and *CHANGE_VELOCITY_GENERATION are in the input file.
- Fix unallocated MPP SPH message buffers during full deck restart.
- Allow change of DT2MS in *CONTROL_TIMESTEP for full deck restart — MPP only.
- Previous value of memory2 is stored in the dump database so that memory2 in a restart will not be set less that stored value, thereby avoiding a possible segmentation fault.
- Fix a bug affecting FTYPE=JOINTSTIF in *SENSOR_DEFINE_FORCE.
SPH (Smooth Particle Hydrodynamics)
- *CONTACT_2D_NODE_TO_SOLID: Corrected SMP contact force output into rcforc ASCII file for the case with multiple definitions of *CONTACT_2D_NODE_TO_SOLID keyword in the input file.
- Fix deadlock in some thermal adaptive problems.
- Fix for input processing error termination due to *SET_NODE_ADD that occurs in input before the sets it references.
- Fix non-effective OPTION DBOX, DVOL, DSOLID, DSHELL, DTSHELL, DSEG for *SET_SEGMENT_GENERAL to delete segments.
- Improve warning/error messages related to duplicate nodes. And make that check equivalent in single and double precision.
- Fix error in reading long format in ALE, *SENSOR, etc.
- *INTERFACE_SSI with blank optional card is now read in correctly. gmbin file is read in correctly in MPP.
- Pertaining to user-defined capabilities via *MODULE:
- A new command line option is added to load one module file without changing the input deck, as “module=filename”. It provides back compatibility to the input without the *MODULE keywords.
- The system paths defined in LD_LIBRARY_PATH are also included for searching module files for those filenames starting with “+”.