New version of LS-DYNA is released for all common platforms.
Release notes for Version R7.1.3.
Herein are summarized most of the new features, enhancements, and significant corrections made since the release of version R7.1.2. These notes are arranged by topic. Understand that in many cases, a particular note may have applicability to multiple topics.
Airbag
Fix restart bug for particle airbags.
Backport UP (Uniform Pressure) airbag chamber interaction.
CPM (Particle) airbag block=1 option was accidentally disabled.
Output both thermal pressure and particle pressure to CPM database.
Update chamber pressure and temperature by the total enthopy and iterate through each gas component to get better agreement with bag state data.
To keep consistent airbag volume calculation independent of coordinate transformation, the vent hole center has been shifted. This will make enhanced venting work correctly.
Adjust porous leakage for internal fabric parts using CPM.
Bug fix for reading negative number for gas property ID.
Fix bug in *AIRBAG_PARTICLE when the following options ID is longer than 6 digits:
*DEFINE_CPM_VENT
*DEFINE_CPM_GAS_PROPERTIES
Correct damping energy calculation for CPM airbag internal part.
Bug fix for CPM’s IBLOCK > 1 options.
Add option of death time for CPM gas front tracking algorithm.
Adjust the leakage pressure for porous leakage of *MAT_34 for internal airbag part. The FAC curve is based on the absolute pressure, the internal part needs to add a reference pressure (p_atm) to its pressure difference to get correct leakage velocity.
Evaluate CPM bag volume based on related coordinates to avoid truncation error after transformation.
Check airbag part list in *AIRBAG_PARTICLE. If duplicated parts are found, it will error terminate the job.
Add iair=4 in *AIRBAG_PARTICLE for window airbag. For iair=4, store collision history for every airbag segment to get proper air ratio and apply force to avoid bag being collapsed by ambient pressure.
Enable hybrid for CPM p2p contact — MPP only.
Apply BAGREF only if _DECOMPOSITION option is used in CPM .
Increase weight of initial air particle while using IAIR=4. This will keep about the same pressure for curtain bak of front and back fabric.
Contact
Fix bug affecting multiple contact force transducers that share segments.
MPP contact:
Increase the # of buckets in the initial processing of tied contacts, from 4096 to 1% of num np. This SIGNIFICANTLY speeds up the initialization of large tied contacts in big models.
Fix for MPP problem with *CONTACT_FORCE_TRANSDUCER when combined with groupable contact — in some cases this could have resulted in segfault.
Fix two uninitialized variables for 2-surface forces transducers when used with segment based (SOFT=2) contact. This apparently went unnoticed since it would ususally behave fine if the values were zero. Howevever, the variables affected memory pointers so the consequences could be segmentation faults.
Bug fix for NCFORC output with eroding contact.
Fix uninitialized variables for mortar contact. — MPP only
Filter duplicate slave nodes for penalty/constraint contacts using IPBACK in SMP.
SOFT=2 eroding contact fix.
MPP groupable contact: fix copy error that may result in bad behavior, but in practice was very unlikely to give incorrect results if the problem actually ran to normal termination
Enable hybrid for CPM p2p contact — MPP only.
Improve MPP handling of groupable AUTOMATIC_*_TO_SURFACE contact.
MPP fix for possible deadlock in groupable contact.
MPP groupable contact: fix memory clobber when a master node is in a 6 DOF constrained interface and ALSO in a 3 DOF constrained interface which appears later in the input.
Fix for MPP IPBACK initialization issue. (IPBACK invokes the option to automatically create a backup penalty-based contact for a tied constraint-based contact. In some situations, a segmentation fault would occur for this option.)
Fix force transducers with MPP segment based contact when segments are involved with multiple 2 surface force transducers. The symptom was that some forces were missed for contact between segments on different partitions.
Constrained
Modify warnings related to *CONSTRAINED_INTERPOLATION.
Fix for rebar constraint coupling with only a few solid elements.
Make initialization improvements for *CONSTRAINED_TIED_NODES_FAILURE.
Control
Straighten out some silist/sidist issues in MPP decomp: silist and sidist outside of a “region” in the pfile are now deprecated. They have been undocumented for several years (since “region” was introduced). Also, fix the keywords *CONTROL_MPP_DECOMPOSITION_CONTACT_DISTRIBUTE and *CONTROL_MPP_DECOMPOSITION_CONTACT_ISOLATE. Which were not treating each contact interface individually (as the manual states), but collectively.
General DEFLR flag in *CONTROL_SHELL.
MPP: change to the decomposition behavior of *CONTROL_MPP_DECOMPOSITION_PARTS_DISTRIBUTE, *CONTROL_MPP_DECOMPOSITION_PARTSET_DISTRIBUTE, *CONTROL_MPP_DECOMPOSITION_ARRANGE_PARTS in the case where a decomposition transformation is also used. Previously, any such regions were distributed without the transformation being applied. This has been fixed so that any given transformation applies to these regions also. So the transformations will NOT apply to these regions
Attempt to correct/clarify/extend the use of DT2MSLC on *CONTROL_TIMESTEP, issue warnings for various combinations of DT2MSLC and DT2MS as appropriate. Manual modified as well, and the goal with this is to make it clear and still maintain back compatibility.
Apply BAGREF only if _DECOMPOSITION option is used in CPM .
Change method of computing volume of solid element types 10, 13, 44 when PSFAIL.ne.0 in *CONTROL_SOLID or ERODE=1 in *CONTROL_TIMESTEP.
Fix bug on shell element deletion based on W-MODE in *CONTROL_SHELL. Update to W-MODE failure to restrict to only parts in PSNFAIL, if PSNFAIL is set.
Fix for the negative volume check.
Database
Fix for output of large D3PLOT states in single precision.
Fix for output of BNDOUT data for joints in MPP, which was writing out incorrect data in some cases.
Modify NCFORC_FILTER.
Fix of a 2D belt force issue.
Correct problem with single precision output for NCFORC filtering.
Various changes to *DATABASE_NCFORC_FILTER.
Various improvements to output timing profile — MPP only.
Enable solid elements with user-defined orthotropic materials to work with the INTOUT and NODOUT options on *DATABASE_EXTENT_BINARY. The transformation matrix was stored in the wrong place causing strain and stress transformations to fail.
Bug fix for NCFORC output with eroding contact.
The section force for seatbelts is recoded to provide more robust and accurate results.
Fixed bug for running with dyna.str and load curve is used to control D3PLOT output. Added load curve id to LSDA database.
Define
Add option of POS6N for *DEFINE_TRANSFORM to define transformation with 3 reference nodes and 3 target nodes. Issue error message and terminate the simulation when illegal ACTION is used for *DEFINE_TRANSFORM
Fix bug in *AIRBAG_PARTICLE when the following options ID is longer than 6 digits: *DEFINE_CPM_VENT, *DEFINE_CPM_GAS_PROPERTIES
Allow more than 5 digits of user load curve ID.
Do not transform the translational velocities in *INITIAL_VELOCITY_GENERATION if the local coordinate system, icid, is defined.
Dynamic relaxation
Restructure and simplify the readin routine of nodal displacement data during DR — MPP only.
Fix a bug when EFG shells are used in seatbelt model.
Element
Merge in seatbelt related fixes:
Fix MPP seatbelt bug that can happen when buckle pretensioner is modeled as type-9 pretensioner.
Fix solid element 4 so that rigid body translation will not cause strain and stress due to round-off error.
Fix bug for 2D belt that does not consider 2D seatbelt element IDs when creating null shell for *RIGIDWALL_PLANAR and hence results in a conflict.
Fix thick shell type 2 with mass scaling.
Add SMP consistency for cohesive element.
Change the MPP behavior of discrete beams (ELFORM=6) when attached to elements that fail. They were behaving like null beams, in the sense that it was possible for beam nodes to become dead due to attached elements failing, and discrete beams would be no longer visualized even if the beams themselves had not failed. With this change, the MPP discrete beams now behave like other beams in that the beams have to fail before they are removed. MPP and SMP behavior is now consistent.
Backport of cohesive elements 19-22.
Change method of computing volume of solid element types 10, 13, 44 when PSFAIL.ne.0 in *CONTROL_SOLID or ERODE=1 in *CONTROL_TIMESTEP.
Fix a bug when EFG shells are used in seatbelt model.
IGRAV of *ELEMENT_SEATBELT_ACCELEROMETER can be a curve defining gravitation flag as a function of time.
Include
decomp { automatic } can be used in the pfile and will take into account the effect of *INCLUDE_TRANSFORM.
Initial
Fix for *INITIAL_STRESS_BEAM.
Do not transform the translational velocities in *INITIAL_VELOCITY if the local coordinate system, icid, is defined.
Do not transform the translational velocities in *INITIAL_VELOCITY_GENERATION if the local coordinate system, icid, is defined.
Interface
Fix for problem with *INTERFACE_LINKING which was completely dropping the linked interface if there were some nodes not tied due to tolerance AND there were some nodes not tied due to conflicting constraints (they were master nodes of a tied interface).
Fix for MPP restart and output of LSDA interface linking file.
Material
Save some memory in hex weld generation in keyword section to allow using single precision in MPP.
Committed fixes for *MAT_34 bending treatment, rbsms fixes and *MAT_100 fixes.
Fix bug for *MAT_242.
Fix for binout output of swforc file which can get the data vs. ids out of sync when some solid spotwelds fail.
Fix shell elements with *MAT_USER_DEFINED with IORTHO=1.
Fix bug where by when searching for the intersection point of the load and unload curves (used by *MAT_34 with negative unload curve option), and extrapolation of one of the curves was needed to find the intersection point, the extrapolated stress was calculated incorrectly causing unpredictable behavior.
Add possibility to use yield curve or table in *MAT_100 (*MAT_SPOTWELD) for solid elements if SIGY.lt.0 is used.
Add a solid spot weld material option to treat the stress state as uniaxial. This option is available for solid assemblies also. The uniaxial option is used by setting the elastic modulus to a negative number on *MAT_SPOTWELD where the absolute value of E is the elastic modulus.
Material changes for:
*MAT_169 with *DEFINE_FUNCTION capabilities.
*MAT_169 with in-plane failure strain criterion.
*MAT_100_DA with output of “failure function” on hisvar 18.
*MAT_ADD_EROSION, GISSMO output of Lode values.
Minor fix for warning message INI+140.
Adjust porous leakage for internal fabric parts using CPM.
Include all improvements to *MAT_181.
Enable solid elements with user-defined orthotropic materials to work with the INTOUT and NODOUT options on *DATABASE_EXTENT_BINARY. The transformation matrix was stored in the wrong place causing strain and stress transformations to fail.
Adjust the leakage pressure for porous leakage of *MAT_34 for internal airbag part. The FAC curve is based on the absolute pressure, the internal part needs to add a reference pressure (p_atm) to its pressure difference to get correct leakage velocity.
*MAT_SPR_JLR changes to output variable.
Fix for energy calculation in GISSMO material model.
Fix incorrect reading of *MAT_DAMAGE_2/*MAT_105 parameters EPSD, S, DC.
Save some memory in hex weld generation in keyword section to allow using single precision in MPP.
Fix problem of MPP pre-decomposition that can occur if the local directory specified in the pfile has very different lengths in the initial run vs theactual run. The difference resulted in a line count difference in the size of the structured files created, throwing off the reading of theb file in the acutal run.
Straighten out some silist/sidist issues in MPP decomp: silist and sidist outside of a “region” in the pfile are now deprecated. They have been undocumented for several years (since “region” was introduced). Also, fix the keywords *CONTROL_MPP_DECOMPOSITION_CONTACT_DISTRIBUTE and *CONTROL_MPP_DECOMPOSITION_CONTACT_ISOLATE. Which were not treating each contact interface individually (as the manual states), but collectively.
MPP contact: Increase the # of buckets in the initial processing of tied contacts, from 4096 to 1% of num np. This SIGNIFICANTLY speeds up the initialization of large tied contacts in big models.
Various improvements to output timing profile — MPP only.
MPP: change to the decomposition behavior of *CONTROL_MPP_DECOMPOSITION_PARTS_DISTRIBUTE, *CONTROL_MPP_DECOMPOSITION_PARTSET_DISTRIBUTE, *CONTROL_MPP_DECOMPOSITION_ARRANGE_PARTS in the case where a decomposition transformation is also used. Previously, any such regions were distributed without the transformation being applied. This has been fixed so that any given transformation applies to these regions also. So the transformations will NOT apply to these regions
Fix MPP issue with initialization when command line arguments are > 256 characters long.
Fix for MPP not handling element deletion properly in some cases at decomposition boundaries.
Clean up temporary LSDA files used in MPP decomposition, which should fix an issue with very large long running adaptive problems.
Fix possible MPP pre-decomposition memory overwrite
Fix issue that could cause unpredicatable behavior during MPP decomposition if the input uses *NODE_MERGE.
Save hex weld creation orientation to the pre-decomposition file so that the subsequent run generates the welds in the same way.
Restart
Fix restart bug for particle airbags.
Fix for MPP restart and output of LSDA interface linking file.
Rigid
Fix bug for 2D belt that does not consider 2D seatbelt element IDs when creating null shell for *RIGIDWALL_PLANAR and hence results in a conflict.
Set Options
Fixes for DPART in *SET_SEGMENT_GENERAL.
Speed up keyword processing of *SET_SEGMENT.
Miscellaneous
Update shell and solid cost parameters for “newcost” option.
Various improvements to output timing profile — MPP only.
decomp { automatic } can be used in the pfile and will take into account the effect of *INCLUDE_TRANSFORM.
Output general and contact cost profile to csv format for load balance graph — MPP only.
Allow memory2 option in *KEYWORD line of input.
Fix for input processing hang in some inputs with encryption.
Avoid hybrid consistency problem.
Clean up temporary LSDA files used in MPP decomposition, which should fix an issue with very large long running adaptive problems.
Update the internal CASE driver to better handle JOBID strings.
Fix issue related to encrypted inputs: encryption was not being properly shut down on detecting an statement, resulting in possible problems if there was any data in the encrypted file past the statement.
Add MPP message truncation debugging with debug=3 command line flag.
