Personal tools
 

LS-DYNA R8.1.0 (R8.105896) released

Filed under:
New version of LS-DYNA is released for all common platforms.

Release notes for Version R8.1.0.

Herein are summarized enhancements and bug fixes made since the release of version R8.0.0. The items are arranged by category. Understand that in many cases, a particular item may have applicability to multiple categories, but in the interest of brevity, each item is listed only once, under a single category.

.

Airbag
  • Calculate CPM airbag volume based on a reference node from the bag instead of using the global origin. As a result, the bag volume will be independent of coordinate transformation.
  • Add IAIR=4 option to *AIRBAG_PARTICLE. This otion enables a special logic to treat the region mixed with air and inflator particles. The collision number of air and inflator particles for each segment is stored. Air pressure based on this ratio is applied on the segment against ambient air pressure. An optional time TSTOP may be specified to switch IAIR=4 to IAIR=2.
  • Bugfixes for *AIRBAG_PARTICLE:
    • IBLOCK > 1 options
    • Adjust the leakage pressure for porous leakage of *MAT_034 for internal airbag part. The FAC curve is based on the absolute pressure, the internal part needs to add a reference pressure (PATM) 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, the job terminates with an error message.

.

ALE
  • Fix a bug in ALE dbfsi force reporting introduced in r66482. This caused triangular segment forces to not be reported in dbfsi.

.

Boundary
  • Fix bug whereby MPP was dropping SPCs during adaptivity in some cases.
  • Fix erroneous results if SET_BOX option is used for *BOUNDARY_PRESCRIBED_MOTION.
  • Fix *BOUNDARY_PRESCRIBED_ACCELEROMETER_RIGID for MPP. It could error terminate or give wrong results if more than one of this keyword were used.
  • Fix segmentation fault when using *BOUNDARY_PRESCRIBED_ORIENTATION with vad=2, i.e. cubic spline interpolation.
  • Fix incorrect behavior if multiple *BOUNDARY_SPC_SYMMETRY_PLANE, i.e. >1, are used.
  • Fix for problem of nodal SPCs possibly getting lost during adaptivity.
  • Fix incorrect motion if *BOUNDARY_PRESCRIBED_MOTION_RIGID_LOCAL is on a rigid part which is merged with a deformable part that has been switched to rigid using *DEFORMABLE_TO_RIGID.
  • Fix memory allocation for the velocity matrix of *BOUNDARY_PRESCRIBED_ACCELEROMETER.

.

Blast
  • Bug fix for *LOAD_BLAST_ENHANCED in MPP: Put ALE ambient element type 5 and its adjacent element used in *LOAD_BLAST_ENHANCED option together into same processor to treat this option correctly and remove unnecessary communication between MPP processors. Also speedup initialization phase for this type of simulation.

.

Contact
  • MPP contact: Increase the # of buckets in the initial processing of tied contacts, from 4096 to 1% of numnp. This SIGNIFICANTLY speeds up the initialization of large tied contacts in big models. Also, increase the old global MPP # of buckets from 4096 to 10000.
  • Skip constraint based interfaces when computing the stable contact time step size.
  • Fix for MPP *CONTACT_FORCE_TRANSDUCER when used together with groupable contact.
  • Correct internal memory overwrite for explicit using tiebreak contact in MPP.
  • Speed up SOFT=2/DEPTH=45 contact by about 15%.
  • Fix potential problem in *CONTACT_AUTOMATIC_GENERAL whereby the small scale of coordinate values could result in penetration for beam to beam contact.
  • Fix spurious repositioning of nodes when using *CONTACT_SURFACE_TO_SURFACE for SMP. Broken in r91056.
  • Fix input error if *CONTACT_AUTOMATIC_ONE_WAY_SURFACE_TO_SURFACE_TIEBREAK_THERMAL is used with CT2CN>0.0.
  • Delete hex spotwelds if SPOTDEL=1 in *CONTROL_CONTACT for *CONTACT_TIED_SHELL_EDGE_TO_SURFACE_ID_CONSTRAINED_OFFSET.
  • Fix thermal MPP segment-based contact. The message passing of thermal energy due to friction was being skipped unless peak force data was written to the intfor database.
  • Fix contact for fabric material (*MAT_034) when neither EB nor EC are input as zero.
  • Fix segment-based contact in a case where all deformable parts are switched to rigid at the start of the calculation. The time step grows enormous and the contact stiffness is decreased to compensate for the large time step. Later, the switched parts were switched back to deformable and the time step plummets. Unfortunately, the contact is not designed to stiffen as the time step drops as this could lead to a bad energy balance. Consequently, the contact is not nearly stiff enough to prevent penetration. The setting of the contact time step has been postponed until contact starts, at which time the time step is small again so the contact is stiff enough.  To allow back compatibility in most cases, this is only done if the DTINIT>0 on *CONTROL_TIMESTEP.
  • Fix MPP *CONTACT_2D_FORCE_TRANSDUCER when both master and slave surfaces are defined. The message passing was not done so forces were wrong. Also, found a blunder in the MPP bucket sort for *CONTACT_2D_AUTOMATIC_SINGLE_SURFACE and SURFACE_TO_SURFACE that could cause unpredictable behavior such as a force spike or penetration.
  • Fix 2-surface force transducers for 2D automatic contact when the smp parallel consistency option is active. Also, corrected the mass data in the rcforc file.
  • Fix a flaw in segment-based contact with DEPTH=25 that could allow penetration to occur. Put protection against the same error in DEPTH=5, but it's likely this second change will have no effect.
  • The d3plot output for rigid surface contact was incorrect for MPP.
  • Fix bug in *CONTACT_RIGID_NODE_SURFACE broken in r86847. The bug was in reading *NODE_RIGID_SURFACE.
  • Fix memory clobber in MPP's *CONTACT_AUTOMATIC_BEAMS_TO_SURFACE.
  • MPP fixes for constrained tied contact when used with adaptivity. The behavior of the slave nodes in adaptive constraints was not correct if they were also master nodes of a tied interface. This has been fixed, and support for the rotations required for *CONTACT_SPOTWELD have also been added.
  • The optional extra cards in *CONTACT_MPP can now be indicated with "OPTCARD" in the first field, in addition to the traditional "&" in column 1
  • Update MPP handling of AUTOMATIC_TIEBREAK option 5 to release the slave nodes (and report them as having failed) when the damage curve reaches 0.
  • Fix untied contacts when using *CONTACT_TIED... with *MAT_ANISOTROPIC_ELASTIC_PLASTIC/*MAT_157.
  • Fixed two unitialized variables in initialization code for 2-surface force 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.
  • Fixed MPP ncforc output when used with eroding contact. When the number of nodes dropped to zero in a partition, we were outputting force data for nonexistant nodes.
  • Fixed glstat's sliding interface energy in the MPP version to retain energy even after the interface death time has been reached.
  • Improved segment based contact with DEPTH=5. There was a potential problem with edge forces being nonzero at the start of the calculation which could lead to energy growth. This change should have negligable effect on most problems.
  • Fixed a serious MPP error in the sliding option of *CONTACT_2D_AUTOMATIC.
  • Fixed a problem in the contact energy calcualtion that could lead to abnormal terminations. Also made the process of searching for nodes to tie more robust as some problem was found with nodes being missed.
  • Added master side output in the MPP version for 2-surface force transducers when used with segment based (SOFT=2) contact.
  • Improved segment based contact when SHAREC=1 to run faster when there are rigid bodies in the contact interface.
  • Fixed *CONTACT_2D_AUTOMATIC_TIED_ONE_WAY which had a possible memory error during initialization.
  • Fixed implicit tied contact by *CONTACT_2D_TIED_CONTACT which has been broken since r83535 when fixing force transducers. The penalty constraint had a NaN stiffness so only jobs that could run with all kinematic constraints were working.
  • Fixed a bug in MPP segment based *CONTACT_ERODING_SURFACE_TO_SURFACE that has existed since r72456. After decompostion, solid parts may have surfaces that appear to be exterior on the decomposition boundary, but are really interior if adjacent partitions are considered. These segments are removed as they can cause sliding parts to catch and not slide well. At r72456, a change was made that caused segments on the matster surfaace to not be removed if that partion did not also have slave segments.  This is now fixed.
  • Fixed a possible problem durning initialization of segment based contact. Options that use neighbor segment data such as the sliding option and edge-to-edge checking could access bad data if the same nodes were part of both the slave and master surfaces. This would not be a normal occurence, but could happen.
  • Fixed glstat output to omit 2D force transducer energy from the global sliding interface energy.  Also, corrected rcforc output for 2D_AUTOMATIC_TIED contact.
  • Fix for output of tiebreak damage as "contact gap" in intfor database. In the SMP version, this value could have been wrong if other contacts without friction (FS=FD=0) were defined.

.

Constrained
  • Fix bug in *DAMPING_RELATIVE. If the rigid part PIDRB is the slave part in *CONSTRAINED_RIGID_BODIES, the damping card did not work correctly. The work-around was to set PIDRB to the master part in *CONSTRAINED_RIGID_BODIES, not the slave part.
  • 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 incorrect results when using *CONSTRAINED_SHELL_TO_SOLID with MPP.
  • Improve determination of interpolation functions for *CONSTRAINED_SPR2/*CONSTRAINED_INTERPOLATION_SPOTWELD with INTP=0.
  • Add some results from *CONSTRAINED_SPR2/*CONSTRAINED_INTERPOLATION_SPOTWELD to d3plot and d3thdt files in the places where post-processors expect to find the forces and moments for resultant beams.
  • Reduce initialization time for *CONSTRAINED_TIED_NODES_FAILURE in case there are many of them.
  • With *CONSTRAINED_JOINTS, if radius of part A for a gear joint is not specified then compute it. If ratio of radii not specified compute radius of part B and then compute the ratio.
  • Error terminate with message, SOL+700, if CIDA and CIDB is not defined for *CONSTRAINED_JOINT_STIFFNESS_GENERALIZED.
  • Fix incorrect constraints on rotary DOF for adaptivity.
  • Fix incorrect motion if NRBF=2 in *DEFORMABLE_TO_RIGID_AUTOMATIC and if any of the *CONSTRAINED_NODAL_RIGID_BODY nodes belong to a solid element.
  • Fixed an problem with *CONSTRAINED_BEAM_IN_SOLID when used with segment based (SOFT=2) eroding contact in the MPP version. Some constraints were failing due to contact changing the adjacent element array. This combination now works.

.

Control
  • Fixed bug in *CONTROL_CHECK_SHELL if PSID.lt.0 (part set ID) is used.
  • Fix spurious deletion of elements when using TSMIN.ne.0.0 in *CONTROL_TERMINATION, ERODE=1 in *CONTROL_TIMESTEP, and initialized implicitly in dynamic relaxation.
  • Fix ineffective FRFREQ in *CONTROL_OUTPUT for MPP.
  • Use correct element ID for output of failed solid elements when *MAT_ADD_EROSION (GISSMO) is used with *CONTROL_DEBUG.
  • *CONTROL_RIGID: Avoid singularity for 2- and 3-noded nodal rigid bodies when using RBSMS=1, consistent rigid body mass scaling.
  • SMP now writes d3msg file correctly when MSGFLG=1 in *CONTROL_OUTPUT.
  • Fix bug affecting *CONTROL_STAGED_CONSTRUCTION with beam elements of ELFORM=2: when these beams are dormant, they could still control the timestep. This is now fixed.
  • Add new keyword *CONTROL_PREVENT_INCLUSION which causes an error abort if it occurs inside an include file.
  • Fix a bug for pore air flow analysis (*CONTROL_PORE_AIR) that, because of numerical round off, can cause unexpected air flow even when the structure undergoes no deformation.
  • Fix spurious error, STR+755, if using *DAMPING_FREQUENCY_RANGE with *CONTROL_ADAPTIVE.
  • Fix ineffective NLQ parameter in *CONTROL_SOLUTION.
  • Add new feature to *CONTROL_SOLUTION, LCACC, to truncate load curve to 6 significant figures for single precision & 13 significant figures for double precision. The truncation is done after applying the offset and scale factors.
  • Fixed the W-MODE option in *CONTROL_SHELL for deletion of shell elements. The criterion was inadvertantly scaled by 16.

.

Database
  • Fix for output of bndout data for joints in MPP, which was writing out incorrect data in some cases.
  • Minor change to how pressure is computed for triangles in the INTFOR database for contact.
  • MPP fix for INTFOR contact database output, which had garbage in some of the element deletion flags, resulting in segments missing from the output.
  • Fix disbout output for discrete beams in a restart.
  • *DATABASE_CURVOUT: Fix zero strain values output to curvout for *DEFINE_CURVE_FUNCTION using function, ELHIST, for solid elements.
  • Fix incorrect damping energy computation for glstat.
  • Fix NaN output to elout_det and spurious element deletion if NODOUT is set to STRAIN, STRAIN_GL, ALL, or ALL_GL.
  • Fix centroid locations written to the secforc data when used with Belytschko-Schwer beams. They were broken at rev 93744.
  • Fix stress output to the d3hsp file for triangular shell elements that are swithed to C0 triangle elements (ESORT=1) on *CONTROL_SHELL. The stress output was much too small.
  • Increase the length of header to 80 for the following files in binout for matsum, nodout, spcforc, ncforc.
  • Fix d3plot output for DCOMP=5 in *DATABASE_EXTENT_BINARY.
  • Fix incorrect curvout values when using BEAM(id,jflag,comp,rm) for *DEFINE_CURVE_FUNCTION and if the beam formulation is type 3, i.e. truss.
  • Fix incorrect output to curvout file if using ELHIST in *DEFINE_CURVE_FUNCTION for shells.
  • Output stresses for all 4 intg points to eloutdet for cohesive element type 19 & 20.
  • Fix incorrect rotational displacment to nodout when REF=2 in *DATABASE_HISTORY_NODE_LOCAL. Affects MPP only.
  • Fix incorrect strains output to elout for shell type 5 and when nip>1.
  • Fix incorrect acceleration output to nodout file when IACCOP=1 in *CONTROL_OUTPUT and IGRAV=1 in *ELEMENT_SEATBELT_ACCELEROMETER.
  • Fix corrupted D3PLOT when RESPLT=1 in *DATABASE_EXTENT_BINARY and idrflg.ge.5 in *CONTROL_DYNAMIC_RELAXATION.
  • Fix missing element connectivities in nastin file when using *INTERFACE_SPRINGBACK_NASTRAN_NOTHICKNESS.
  • Fix seg fault when using *DATABASE_BINARY_D3PART with *CONTACT_TIED_SHELL_EDGE_TO_SURFACE. This affects SMP only.
  • Fixed an ELOUT error of outputting the wrong number of integration points for some element types when OPTION2 > 0 on *DATABASE_ELOUT (history variable output) and ESORT > 0 on *CONTROL_SHELL. The problem existed for element types that are not sorted, such as triangle elements, membrane elements, and 2D sections.
  • Enabled 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.

.

Discrete Element Method
  • Fix time step calculation bug for *DEFINE_DE_INJECTION. Bug was introduced in r84485.
  • Output ASCII format for demrcf if BINARY.eq.3.

.

Element
  • Fix for seatbelt skew angle calculation.
  • Fix an MPP bug for affecting type 9 seatbelt pretensioner.
  • Prevent inactive shell elements (from *ELEMENT_SHELL_SOURCE_SINK) from controlling the solution time step.
  • Omit null shell elements from warning about badly shaped elements (STR+1129).
  • Fix all thick shells to work with anisotropic thermal strains which can be defined by *MAT_ADD_THERMAL. Also, this now works by layer for layered composites.
  • Fixed solid element formulation 4 so that rigid body translation will not cause strain and stress due to round-off error.
  • Fix torsion in linear beam formulation 13. A failure to add the torsional moment at node 2 caused an inability to reach equilibrium in the torsional mode.
  • Fix thick shells with TSHEAR=1 behavior for user defined materials and isotropic materials.
  • Avoid division by zero when *ELEMENT_BEAM_SOURCE is used. Also, modify original length of first beam element at *ELEMENT_BEAM_SOURCE just to ensure correct beam prism representation in LS-Prepost.
  • Fix for MPP not handling element deletion properly in some cases at decomposition boundaries.
  • Fix minor loss of precision for *ELEMENT_SEATBELT_RETRACTOR anchor node displacement calculation.
  • *ELEMENT_DIRECT_MATRIX_INPUT: Enhance reading of nastran dmig files to allow for LS-DYNA type comment lines starting with '$'.
  • Added an error termination (STR+1339) if someone uses shell form 18 in an explicit analysis.
  • Changed *ELEMENT_TSHELL so that both the COMPOSITE and BETA options can be read at the same time. Prior to the fix, only the first one would be read.
  • Added an error termination (SOL+1307) if an invalid integration option is used with linear shell form 18. Only Gauss integration with 10 or fewer points is valid. In the the next major release (R(), all of the regualar shell options will be available.
  • Fixed SMP consistency for  solid form 20.
  • Changed 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.
  • Fixed the s-axis and t-axis orientation of beam spotwelds in the MPP version when those beam weld elements are defined with a 3rd node. The 3rd node was being discarded prior to initializing the beam orientation so the s and t-axes were being randomly assigned as if the 3rd node had not been assigned.
  • Added new SCOOR options for discrete beam form 6 (*SECTION_BEAM). A flaw was found in how the discrete beam accounts for rigid body rotation when SCOOR=-3, -2, +2, and +3. A correction for this is made and introduced as new options, SCOOR=-13, -12, +12, and +13. A decision was made to leave the existing options SCOOR=-2, +2, -3 and +3 unchanged so that legacy data could run without changes. The flaw in SCOOR=-3,-2,+2, and +3 is in the strain modification for rigid body rotation of the element.  The average of nodal rotations is used as an estimate of element rotation, but his may be a poor assumption since the beam rotation may be based on the rotation of node 1 (SCOOR=-3), node 2 (SCOOR=3), or the r-axis may depend on the nodes (SCOOR=-2 or +2) with the s-axis and t-axis depending on the nodal rotations. The corrected forms, SCOOR=-13, -12, +12, and +13 directly use the element orientation tensor in the current and previous configuration to calculate element rotation. As a result, when SCOOR=-12 or +12, the transverse strains in the s-direction or t-directions are zero.
  • Fix the change of connectivity in case of cohesive pentahedron solid elements.
  • Fix product moments of inertia (IST) for *SECTION_BEAM section types SECTION_03 (L-shape) and SECTION_19 (Box shape).
  • Add missing cross section forces (secforc) for linear shell form 18.
  • Bug fix for tetrahedron type 13 in implicit analyses.
  • *SECTION_SOLID: Cohesive elements 19-22 supported for SMP consistency, ncpu<0.
  • Fix parsing error in *SECTION_BEAM_AISC

.

Forming
  • Some improvements to blanksize development.

.

Frequency Domain
  • *FREQUENCY_DOMAIN_RANDOM_VIBRATION_FATIGUE: Fix a bug in getting SN fatigue curve type when multiple SN curves are present in random vibration fatigue analysis.
  • *FREQUENCY_DOMAIN_ACOUSTIC_BEM: Fix a bug in running Burton-Miller dual BEM, for irregular frequency problem for exterior acoustics.
  • *FREQUENCY_DOMAIN_SSD_FATIGUE: Update fatigue analysis to avoid damage ratio computation with tiny stresses which may cause numerical error.
  • Frequency domain (BEM acoustics):
    • Fix a bug in running BEM acoustics when multiple boundary condition (ssd + impedance).
    • Revert the normal directions of elements for exterior problems so that the formulation for interior problems can be used. Only applicable to collocation BEM.
  • Frequency domain (FEM acoustics): Fix a bug in FEM Acoustics with linear tet elements.
  • Frequency domain (SSD): Fix a bug in running ssd with non-classical damping or damping matrix.

.

Implicit
  • Improve Implicit's treatment of constrained joints to account for rounding errors.
  • Correct Implicit's treatment of prescribed motion constraints defined by a box.
  • Correct interaction of beam damping for explicit with intermittent eigenvalue analysis.
  • Reinstate the option of MAXREF < 0 on *CONTROL_IMPLICIT_SOLUTION.
  • Fix Implicit's collecting of damping terms for beams that have certain types of reference nodes.
  • Add support for *CONSTRAINED_LINEAR for 2D implicit problems. It was already supported for standard 3D problems.
  • Fix generation of part id for displaying superelements in LS-PrePost.
  • Correct loading of SPR constraints for implicit.
  • Apply the enhancements for processing *CONSTRAINED_INTERPOLATION constraints in implicit with large numbers of independent nodes.
  • Adjust Implicit's handling of sw1. and sw3. sense switches to properly handle the dumping. If sw1. sense switch and not at equilibriuma, then reset time and geometry to that at the end of last implicit time step. If sw3. sense switch waits until equilibrium is reached before dumping and continuing.
  • Fix *PART_MODES to correctly handle constraint modes by eliminating removed rigid body modes and providing the correct construction of reduced stiffness matrix.
  • Add the caseid to the implicit scratch files.
  • Fix *CONTROL_IMPLICIT_STATIC_CONDENSATION and *CONTROL_IMPLICIT_MODES to properly work in the environment created by *CASE.
  • Allow for unsorted node/dof sets for *CONTROL_IMPLICIT_STATIC_CONDENSATION and *CONTROL_IMPLICIT_MODES. As well as properly handle the suppression of rotational dofs in the node/dof sets for the solid element only case.
  • Skip the explicit computation that overwrites the accelerations for slave nodes on rigid bodies in an implicit analysis as implicit already does that correctly.
  • Apply a correction about overlapping use of an internal scratch common block while loading damping matrix terms for global mass damping in implicit.
  • Enhance the error handling for input for *PART_MODES. Fix implicit solutions with thick shells with *MAT_057 when there are also solid elements in the model that use *MAT_057. Thick shells support only the incremental update of the F tensor but a flag was set incorrectly in the material model.
  • *CONTROL_IMPLICIT_GENERAL, *CONTROL_ACCURACY: Make sure accuracy flags INN and OSU always apply in an implicit simulation.
  • Fix bug associated with redundant constraint rows in MPP (implicit).
  • For implicit springback, zero out the forces being reported to RCFORC for those contact interfaces disabled at the time of springback.
  • *CONTROL_IMPLICIT_EIGENVALUE: Explicit with intermittent eigenvalue analysis was getting incorrect results after the eigenvalue analysis because an incorrect time step was used for the implicit computations. This, apparently, disturbed the material data. So for this scenario implicit now uses the explicit time step.
  • Fix collecting of resultant forces for SPC constraints (implicit).
  • Fix collecting of any nodal forces for things like nodal force group with more that 1024 nodes (implicit).
  • Fix collecting of stiffness damping terms for implicit.
  • Fix storage allocation for MPP eigenvector post processing for small problems.
  • Set accelerations to zero for output phase during implicit statics.
  • Properly turn off contact with death time for implicit springback.
  • Fix out-of-range array access for implicit mass damping on rigid bodies.
  • Fix a problem with implicit initialization in MPP with 2 or more superelements.
  • Added initialization of all dynamic memory allocated by Metis (used in Implicit Mechanics, Thermal, and wherever sparse direct solvers are used in LS-DYNA).
  • Correct the output of resultant forces for Implicit Linear analysis.
  • Adjust implicit mechanical time step for the case of switching from explicit to implicit so as not to go past the end time.
  • Add warning for implicit mechanics when the product of ILIMIT and MAXREF (two parameters on *CONTROL_IMPLICIT_SOLUTION) is too small. And for the special case when the user changes the default of ILIMIT to 1 to choose Full Newton and does not change MAXREF I reset MAXREF to 165 and issued a warning to the user.
  • Fix Implicit handling of *CONSTRAINED_TIE_BREAK constraints in MPP.
  • Correct the loading of mass damping terms when collecting damping terms for postprocessing.
  • Correct the loading of consistent mass matrix terms for implicit eigenvalue problems and matrix dumping.
  • *CONTROL_IMPLICIT_ROTATIONAL_DYNAMICS:
    • Update the choice of reference frame. When IREF=0 and 2, it is in rotating reference frame (the rotor seems to be stationary in this frame), but the rotor will be manually rotated for the purpose of visualization when IREF=2. When IREF=1, it is in fixed rotating reference frame (it is global coordinate system and the rotor rotates in this frame). For symmetric rotor and support, both coordinate systems work; for symmetric rotor but un-symmetric support, only fixed coordinate system can be used; for un-symmetric rotor but symmetric support, rotating coordinate system should be chosen.
    • The rotating speed can be defined both by a positive value or a negative integer. If it is a negative integer, then the integer is the curve ID to define rotating speed.
    • For solid element and tshell elements, analysis in a rotating reference frame is required for gyroscopic effects to be considered.  In a fixed reference frame, gyroscopic effects are considered only on nodes with rotational DOF, e.g., in shells.
    • Update the number of eigenvalues in eigout and eigenmodes in d3eigv, so that it is the same as defined in keyword (NEIG).
    • Update the mode tracking algorithm to track different modes when studying the rotational speed effects on eigenvalues, for example, plotting a Campbell Diagram.
  • Fixed implicit solutions with solid elements of *MAT_221 which were broken since r94169.

.

Initial
  • Fix several instances of overwriting the initial velocities of any interface nodes read in from a linking file (SMP only).
  • The initial strain and energy is now calculated for *INITIAL_FOAM_REFERENCE_GEOMETRY.
  • *INCLUDE_TRANFORMATION: Fix incorrect transformation of *DEFINE_BOX which resulted in incorrect initial velocities if the box is used in *INITIAL_VELOCITY.
  • Fix incorrect *INITIAL_STRAIN_SHELL output to dynain if INTSTRN=1 in *INTERFACE_SPRINGBACK.
  • Fix stress initialization (*INITIAL_STRESS_SECTION) for type 13 tet elements. The pressure smoothing was causing incorrect pressure values in the elements adjacent to the prescribed elements.
  • Fix for *INITIAL_STRESS_BEAM when used with spotweld beam type 9. It was possible that error/warning message INI+140 popped up even if number of integration points matched exactly.
  • Fix incorrect initial velocity when using *INITIAL_VELOCITY with NX=-999.
  • Do not transform the translational velocities in *INITIAL_VELOCITY if the local coordinate system, icid, is defined.
  • Fix seg fault when using *INITIAL_INTERNAL_DOF_SOLID_TYPE4 in dynain file.
  • Fix uninitialized velocities when using *INITIAL_VELOCITY_GENERATION with STYP=2, i.e. part id, for *ELEMENT_SHELL_COMPOSITE/*ELEMENT_TSHELL_COMPOSITE.
  • Fix for *INITIAL_STRESS_SOLID. When ELFORM=4 on *SECTION_SOLID and ESORT=1 on *CONTROL_SOLID was used, this element was unfortunately sorted to type 10 for reading and writing (dynain) of initial stresses.

.

Isogeometric Elements
  • Fix bug in added mass report for *ELEMENT_SHELL_NURBS_PATCH in MPP.
  • Fix bug for *ELEMENT_SHELL_NURBS_PATCH when MAT_RIGID is used.

.

Load
  • Fix incorrect pressure applied if the directional cosines, v1/v2/v3, for *LOAD_SEGMENT_SET_NONUNIFORM is not a unit vector.
  • Fix incorrect behavior when using arrival time, AT, or box, BOXID, in *LOAD_ERODING_PART_SET.

.

Material
  • Fix incorrect behavior for *MAT_LINEAR_ELASTIC_DISCRETE_BEAM/*MAT_066 when using damping with implicit (static) to explicit switching.
  • Fix bug when using table for EA & EB for in *MAT_058.
  • Fix incorrect forces/moments when preloads are used for *MAT_067/*MAT_NONLINEAR_ELASTIC_DISCRETE_BEAM and the strains change sign.
  • Fix a bug in *MAT_034 in which extrapolation of either the load or unload curve was done incorrectly in order to find an intersection point, resulting in unpredictable behavior.
  • Fix spot weld smoothing (NF>1) on *MAT_SPOTWELD when SPOTSTP>0 on *CONTROL_CONTACT. This was broken at rev. 92366.
  • Fix a *MAT_SPOTWELD error introduced at rev. 92116. This fix changes the failure function calculation for spot weld assemblies and affects swforc output when damage is initiated by the failure function (OPT=10,11,12 on *MAT_SPOTWELD). It also could delay the time of failure when damage is initiated by plastic strain.
  • Fix shell elements with *MAT_USER_DEFINED with IORTHO=1. The material direction had been wrong since rev. 66723.
  • Fix implicit solutions with shell elements that use *MAT_040 and laminated shell theory.
  • Fix *MAT_219 when used with thick shell types 3, 5, and 7. A failure to initialize terms for the time step could cause an incorrect time step.
  • Store tangential and normal separation (delta_II & delta_I) as history variables 1 & 2 for *MAT_COHESIVE_MIXED_MODE (*MAT_138).
  • Fix bug in *MAT_225 (*MAT_VISCOPLASTIC_MIXED_HARDENING) when using table definition together with kinematic hardening.
  • Fix bug in using *MAT_157 with IHIS.gt.0 for shells. Thickness strain update was not correct and plasticity algorithm failed.
  • Fix bug in *MAT_058 (*MAT_LAMINATED_COMPOSITE_FABRIC) when used with strain-rate dependent tables for stiffnesses EA, EB and GAB and LAMSHT=3.
  • Fix bug in *MAT_157 with IHIS.gt.0 (thin shells only).
  • *MAT_CWM (*MAT_270): Add support for thermal thick shell formulation.
  • *MAT_244:
    • Fix energy computation when CRSH=1 that resulted in a drop of the energy ratio to 0.5.
    • Fix phase change computations.
  • *MAT_249:
    • Fix input in long format.
    • Enable elastic fiber shearing using locking angle definition.
    • Fiber stiffness and shear response of the fibers now accept input of tables to define a temperature dependent behavior.
  • Put *MAT_100DA's "failure function" value to history variable 18.
  • Add optional in-plane failure strain to *MAT_169 (*MAT_ARUP_ADHESIVE). New input parameter FSIP on Card 6, Column 3.
  • Fix for contact stiffness of *MAT_157.
  • Improve error message for unsuccessful *MAT_103 curve fitting.
  • Add exit option for plasticity algorithm of *MAT_168 in case of non-convergence to avoid infinite loop.
  • Fix for *MAT_ADD_COHESIVE: INTFAIL=0 now really hinders element deletion.
  • Now *MAT_169 with functions should also work for solids with nip>1.
  • Some parameters of *MAT_169 can be defined via *DEFINE_FUNCTIONs.
  • Add new history variable "averaged Lode parameter" to GISSMO.
  • Fix for combined usage of SIGVM<0 and MXEPS<0 in *MAT_ADD_EROSION.
  • Reset first history variable of some forming materials (*MAT_036, *MAT_122, *MAT_190, *MAT_233, *MAT_243, *MAT_260) to the default value if *INITIAL_STRESS does not come up with a reasonable value.
  • Fix for stress oscillations in *MAT_089.
  • *MAT_026: Fix bug in implicit tangent matrix for implicit convergence improvement.
  • *MAT_153: Enhance the robustness and accuracy for shell material 153, in particular for implicit.
  • Fixed bug in *MAT_172, which occurred when ELFORM=1 (Hughes-Liu shell formulation) was combined with Invariant Numbering (INN>0 on *CONTROL_ACCURACY). In this case, the strain-softening in tension did not work: after cracking, the tensile strength remained constant.
  • Fix bug in *MAT_ARUP_ADHESIVE (*MAT_169). The displacement to failure in tension was not as implied by the inputs TENMAX and GCTEN. For typical structural adhesives with elastic stiffness of the order of 1000-10000 MPa the error was very small. The error became large for lower stiffness materials (i.e. when the elastic displacement to yield was of the same order as the element thickness).
  • Fix bug in *MAT_211 (*MAT_SPR_JLR) that caused spurious results or unexpected element deletion if TELAS=1.
  • Revised output variables for *MAT_211 (*MAT_SPR_JLR). The definition of Displacement Ratio has changed. This affects Extra History Variables 3 through 8 for solid elements, and "plastic strain" for solid elements (which is the maximum out of Extra HVs 3 through 8), and the failure parameter written to swforc file (which is the same as the "plastic strain"). These output variables now vary from 0 to 1 before peak force is reached, then from 1 to 2 until the onset of damage, then from 2 to 3 when failure occurs.
  • Fix ineffective *MAT_ADD_EROSION if the MID is defined using a alphanumeric label.
  • Fix seg fault when using *MAT_PIECEWISE_LINEAR_PLASTIC_THERMAL/*MAT_255 for solids.
  • Zero the pressure for *MAT_JOHNSON_HOLMQUIST_JH1/*MAT_241 after it completely fractures, i.e. D>=1.0, under tensile load.
  • Fix incorrect element failure when using EPSTHIN and VP=0 for *MAT_123/*MAT_MODIFIED_PIECEWISE_LINEAR_PLASTICITY.
  • Fix error termination when using adaptive remeshing for 2D analysis with *MAT_015/*MAT_JOHNSON_COOK and NIP=4 in *SECTION_SHELL and elform=15.
  • Fixed orthotropic user-defined materials when used with thick shell elements.
  • Enabled thick shell forms 3, 5, and 7 to work with *MAT_103, *MAT_104, and *MAT_108.
  • Fixed an error in reading the AOPT parameter for *MAT_CODAM2 (*MAT_219). This error has existed since r89750.
  • Fixed the initialization of *MAT_CODAM2 (*MAT_219) when used with thick shell forms 3, 5, or 7. The 3D thick shell routine uses only 2 terms for the tranformation and therefore needs unique initialization of the transformation data.
  • Enable *MAT_021 to work with thick shell types 3 and 5.
  • Rearranged history data for the solid element *MAT_PLASTICITY_WITH_DAMAGE_ORTHO_RCDC so that damage values would be history values 23 and 24 as they were before the rev 94804 update. This should not change results, only output.
  • Fixed a mistake in r92116 that caused DMGOPT=1,2,11, and 12 of *MAT_FABRIC to not work correctly.
  • The AOPT=0 option of ortho/anisotropic materials when use with solid elements has be corrected when elements are skewed. Previously, the material direction was initialized to be equivalent to the local coordinate system direction. This is not consistent with the manual for skewed elements which states that the material a-axis is in the 1-2 directions for AOPT=0.  This is now fixed and the manual is correct. Also, tetrahedral element forms 10, 13, and 44 had a completely wrong material direction for AOPT=0.
  • Enabled materials that use equation of states to run with thick shell forms 3 and 5.
  • Made change so that NF on *MAT_SPOTWELD will work for solids and solid assemblies without the other options being needed.
  • Bugfix for *MAT_244 in bainite reaction.
  • Fixed a serious bug in automatic switching between cooling and heating for HEAT.eq.2 in *MAT_244. It might have spoiled the results of the phase change computation completely at least for certain domains of the part.
  • Fixed an input bug for *MAT_244: In some cases two instances of the material were merged even though the material input was not completely the same.
  • *MAT_249:
    • bugfix for input of anisotropic transverse shear stiffness
    • allow for more iterations of plane stress iteration
  • Fixed a bug for defining different orientation angles through the thickness of TSHELL elements (forms 2 and 3) and *MAT_021.
  • Fix missing offset when using *DEFINE_TRANSFORMATION with load curve LCID-P in *MAT_SAMP-1 (*MAT_187).
  • Materials *MAT_224 and *MAT_225 now also working in implicit with 2D solids (*SECTION_SHELL, ELFORM = 13, 14, 15).
  • Add tangent stiffness to *MAT_124 for implicit analyses with shell elements.
  • Fix element deletion issue with *MAT_187 that could lead to too many failed elements in rare cases of bad input data.
  • Minor changes for *MAT_ADD_EROSION damage model GISSMO:
    • Allow for DMGEXP values smaller than 1
    • ECRIT<0 can now also refer to a table
  • Bug fix for *MAT_ADD_EROSION damage model GISSMO related to regularization reduction factors SHRF and BIAXF.
  • Fix for failure behavior of *MAT_107 (*MAT_MODIFIED_JOHNSON_COOK), it was erratic before, e.g. depending on element nodal numbering. Element is now deleted when all integration points failed.
  • Correct failure behavior in *MAT_184. Integration points were still able to resurrect even they already reached FN/FT_FAIL.
  • Allow initial temperatures for *MAT_224 to be set via *INITIAL_STRESS_...
  • Correct the behavior of *MAT_ADD_EROSION's parameter NUMFIP>0 for GISSMO. The number of integration points to fail an element was limited by the maximum number of layers, but it should be limited by the total maximum number of integration points.
  • Default to robust Ridder's method in *MAT_224 if BETA<0 is used.
  • Fix option BETA<0 of *MAT_224/*MAT_TABULATED_JOHNSON_COOK. It was not really working correctly before.
  • *MAT_184/*MAT_COHESIVE_ELASTIC: set cohesive forces to zero in case of failure.
  • Fix for *MAT_188/*MAT_THERMO_ELASTO_VISCOPLASTIC_CREEP with CRPM<0 (and curve i with negative ordinate values) and CRPLAW=1. Out-of-range forces were likely to occur.
  • *MAT_181: AVGOPT<0 means that |AVGOPT| is a time interval over which strain rates are averaged.
  • *MAT_106: Plasticity algorithm is made more robust.
  • *MAT_034: Fix paradoxical behavior in transverse shear for bending option (ECOAT,SCOAT,TCOAT).
  • *MAT_ADD_EROSION: The DIEM damage model, IDAM<0, recoded for efficiency. Furthermore, a damage initiation option DITYP=4 added based on the principal stress. The NUMFIP and NCS parameters can be used to control the way erosion and damage incrementation is performed.
  • Automatically add damping to fluid *MAT_PML for earthquake simulations.
  • Fix initialization issues so that *MAT_PML models can be run with CASE commands. From dev 97124.

.

MPP
  • Revert r86884 at customer request, which was: 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 now the transformations will NOT apply to these keywords
  • Modify behavior of *CONTROL_MPP_DECOMPOSITION_AUTOMATIC so that if the initial velocity used is subject to *INCLUDE_TRANSFORM, the transformed velocities are used.
  • MPP fix to honor decomp { automatic } in the pfile.
  • Fix MPP issue with initialization when command line arguments are > 256 characters long.
  • Two csv files will be generated after each MPP execution:
  • load_profile.csv
    cont_profile.csv
    • These will give general information about load balancing and cost profile of the job. User can use this info to adjust the input to get better execution performance.
  • *CONTROL_MPP_DECOMPOSITION, *SECTION_SOLID The CPU cost of solid elements -1 and -2 is properly accounted for when doing the domain decomposition.
  • Allow memory2 option in *KEYWORD, for example,
  • *KEYWORD memory=1G memory2=200m
    • This option will be ignored while running SMP mode.
  • Fix MPP hang up when using *CONTACT_ENTITY.

.

Restarts
  • Fix for MPP restart of LSDA based interface linking file.
  • Fix for MPP full deck restart in case there are non-matching contacts in the two runs, and MPI error could have happened.
  • Fix incorrect behavior of *CONTACT_ENTITY in full deck restart.
  • Correct bug for restart job that had any change in node/element. This bug existed since R6 version.
  • Fix incorrect full deck restart analysis if initial run was implicit and the full deck restart run is explicit.

.

Sensor
  • *SENSOR_CONTROL can now control parts with pore air properties.
  • Fix a bug affecting JOINTSTIF of *SENSOR_CONTROL

.

SPH
  • SPH memory handlers for MPP fulldeck restart now work properly.

.

Thermal
  • Thermal Solver:
    • Fix handling of start time defined with *CONTROL_START for thermal solver.
    • Fix memory location of thermal history variables defined in the *INITIAL_STRESS keywords.
    • Make sure AOPT value is correctly written into structured input for all thermal user materials.
    • Mark material as anisotropic in case the IORTHO variable is not 0.
  • Fix factor storage estimation for thermal fluid coupling. Thermal bug fix when writing dump files (d3dump) at specific time intervals and using a thermal time step defined by a curve LCTS>0 on *CONTROL_THERMAL_TIMESTEP.
  • Thermal bug fix for a full deck restart. The nodal temperature was initialized to 0 instead of temperature in dump file.

.

Miscellaneous
  • Fix bug affecting *DEFINE_STAGED_CONSTRUCTION_PART when used with *PART_COMPOSITE. The bug occurred when different material types were used for different layers within the same part, and that part becomes active during the analysis. The symptom of the bug was that stresses and/or history variables were not always zero when the part became active.
  • Fix for *CASE handling, if there are more than one extra command line arguments used.
  • Allocate keyword memory in larger chunks to speed up input processor of *SET_SEGMENT
  • Fix bugs affecting *INCLUDE_TRANSFORM:
    • when adaptive elements are defined in a file included by *INCLUDE_TRANSFORM, or
    • when *DEFINE_BOX is included by *INCLUDE_TRANSFORM, or
    • when *BOUNDARY_SPC_SYMMETRY_PLANE is included by *INCLUDE_TRANSFORM.
  • Error terminate with message, KEY+1083, if *DAMPING_GLOBAL and *DAMPING_PART_MASS are used together.
  • Fix seg fault when using *DEFINE_HAZ_TAILOR_WELDED_BLANK with *DEFINE_HAZ_PROPERTIES.
  • Fix bugs affecting *INCLUDE and *INCLUDE_TRANSFORM with adaptivity.
  • Fix bug in *PARAMETER definition in long format.
  • Fix bug in which *DAMPING_FREQUENCY_RANGE_DEFORM gave incorrect results when large rigid body rotations occurred.
  • Fix bug affecting damping energy output from *DAMPING_FREQUENCY_RANGE_DEFORM. If RYLEN on *CONTROL_ENERGY = 2, the energy associated with this damping should be included in the Internal Energy for the relevant part(s). This energy was being calculated only if there was also *DAMPING_PART_STIFFNESS in the model but this is now fixed.
  • Allow jobid > 63 characters.
  • Fix writing and reading of DYNAIN data in LSDA format, particularly *INITIAL_STRESS_SHELL data.
  • When writing the LSDA format for the DYNAIN file, if temperatures are present, write those temperatures into the LSDA file instead of an external ascii file.
  • Fix for input processing hang that could occur for some inputs with encryption.
  • Fix issue related to encrypted inputs: encryption was not being properly shut down on detecting an *END statement, resulting in possible problems if there was any data in the encrypted file past the *END statement.
  • The shells used for visualisation of *RIGIDWALL_PLANAR_MOVING_DISPLAY and *RIGIDWALL_PLANAR_MOVING_DISPLAY in d3plot were not moving with the rigidwall. This is now fixed.

.

Filed under: