What's New

View new features for OptiStruct 2024.

Altair OptiStruct 2024 Release Notes

Highlights

  • Multiple Results Output for the same Output request
  • Temperature-dependent Cohesive Material
  • Johnson-Cook Failure Criterion for Explicit Analysis
  • Hyperelastic materials for shells in Explicit analysis
  • Global-Local Analysis support for Direct and Modal Transient Analysis
  • Fully Coupled Electro-Thermal-Mechanical Analysis
  • ESL Optimization for Transient Analysis
  • SLIPRING joint for Implicit Nonlinear Analysis

New Features

Stiffness, Strength and Stability
Skip non-FREEZE contact in Linear Analysis
Contact interfaces which are non-FREEZE/non-TIE can now be deactivated in Linear Analysis using CONTPRM,DEACTLIN,NOTFRZ. This applies to any linear analysis subcase which does not contain NLSTAT preloading. The default (CONTPRM,DEACTLIN,NONE) continues to retain all CONTACT interfaces in linear analysis.
Linear Elastic property from Nonlinear Explicit material in Implicit Analysis
Certain Nonlinear materials such as Johnson-Cook, Crushable Foam, Cowper-Symonds, and Johnson-Holmquist materials are currently only supported for Explicit analysis. Now for these materials, if present in Implicit analyses, then only the corresponding linear elastic part of these materials are used for these non-Explicit analysis types. This also allows the use of these material models in models which contain both Implicit and Explicit analysis subcases.
Temperature-dependent Cohesive Material
Temperature-dependent Cohesive materials are now supported. The following properties can now be temperature-dependent:
  • The elasticity moduli for Mode I, Mode II, and Mode III deformation modes can now be temperature-dependent and are specified using the KI_i, KII_i, KIII_i fields for each corresponding temperature value of X_i on the MCOHED Bulk Data.
  • The maximum values of strain or traction depending on the cohesive damage initiation type can now be temperature-dependent and can be specified via the V1_i, V2_i, and V3_i fields for each corresponding temperature value of X_i on the DMGINI Bulk Data.
  • The corresponding Damage evolution curves, defined via ALPHA_i and W1_i, W2_i, W3_i parameters can now be temperature-dependent for each corresponding temperature value of X_i on the DMGEVO Bulk Data.
SLIPRING for JOINTG for Implicit Nonlinear Analysis
SLIPRING has already been available in the previous release for Explicit analysis. This is now available also for Implicit analysis. SLIPRING is available on the JOINTG Bulk Data Entry to model belt and pulley joints. Linear and Nonlinear Elasticity, Friction and Friction Angle, Mass and Damping are considered properties for SLIPRING joints. It is now supported for large displacement implicit nonlinear static/transient analysis and Explicit Dynamic analysis. JTYPE on JOINTG entry should be set to SLIPRING and various properties can be defined for DOF 1 for the slipring joint using the ELAS, NELA, FRIC, MASS, and DAMP property types on the PJOINTG entry. SLIPRING has an additional degree of freedom, flow, which is unique for this joint type.
Kinematic Hardening and Mixed Hardening for Plane Stress Elements
Kinematic Hardening (HR=2 on MATS1 Bulk Data) and Mixed Hardening (HR=3) are now supported for Plane Stress Elements (CTPSTS and CQPSTS) for both Small and Large displacement nonlinear analysis.
Explicit Dynamic Analysis
TEMP(INIT) support for Explicit Analysis
Initial temperature support is now supported for Explicit Analysis. TEMP(INIT) can point to a TEMP/TEMPD Bulk Entry which identifies the initial temperature field. This temperature field is used to look up the temperature-dependent material data on the corresponding TABLEMD entry referenced on the MATS1 Bulk Data.
MPC support for Explicit DDM
MPCs were already supported for SMP jobs for Explicit Dynamic Analysis in previous versions of OptiStruct. MPCs are now additionally supported for Domain Decomposition method (DDM) for Explicit Analysis.
Johnson-Cook Failure Criterion for Explicit Analysis
Johnson-Cook Failure Criterion for elasto-plastic material failure for stress triaxiality effect is now supported for Explicit Analysis. It considers an exponential decrease of the material ductility with increasing stress triaxiality. It can be activated by setting CRITERIA field to JOHNSON on the MATF Bulk Data Entry. The Johnson-Cook parameters D1, D2, and D3 can be defined via the V1, V2, and V3 fields on the MATF Bulk Data Entry.
Enhancement to PLAS Failure Criterion for Explicit Analysis
The maximum plastic strain (PLAS) failure criterion for Explicit analysis has now been enhanced:
  • Previously, PLAS criterion was only based on a constant maximum equivalent plastic strain value at failure (defined via V1 field).
  • This failure criterion has now been enhanced by adding:
    • Failure based on thinning strain (defined via V2 field) and its interpretation depends of the sign of the input value. If V2 > 0.0, then it corresponds to εzz total strain (ZZ component of the total strain tensor), or if V2 < 0.0, then it corresponds to εzz plastic strain (ZZ component of the plastic strain tensor).
    • Failure based on major strain (defined via V3 field). It is the failure computed based on the maximum positive value of the principal total strains. If multiple options are specified together for PLAS option, then the maximum damage is retained for the output
Hyper-elastic Materials for shell elements in Explicit Analysis
Hyper-elastic materials are now supported for shell elements in Explicit analysis. Hyper-elastic materials were already previously supported for shell elements in Implicit analysis.
Edge-to-Edge contact turned off by default for shells in Auto-Contact for Explicit analysis
Edge Criteria default for shell elements in auto-contact has now been changed from 45.0 degrees to 179.0 degrees. This basically turns off edge-to-edge contact by default, which can help improve performance as the edge is not considered. Note that the shell element boundary edges are still considered. The default remains 45.0 degrees for solid elements.
Switch the default element formulation for first order TETRA elements in Explicit Analysis
The default element formulation for first order CTETRA elements has now been switched from nodal pressure averaged formulation to 1 point formulation in Explicit Analysis.
Penalty-based RBE3 supported
Penalty-based RBE3 is now supported for Explicit analysis. The traditional kinematic RBE3 formulation is automatically switched to the Penalty-based RBE3 formulation in the following cases:
  • Incompatibility (Over-constraint): When Reference grid of RBE3 is also:
    • Dependent grid of an RBE2
    • Secondary side grid of TIE
    • Part of SPC
  • Hierarchical RBE3s exist:
    • Wherein the Reference grid of an RBE3 is also the independent grid of another RBE3.
  • Hierarchical RBE3 and RBE2 exist:
    • Wherein the Reference grid of an RBE3 is also the independent grid of another RBE2.
Noise and Vibration
Significant reduction in disk space usage for Superelement generation runs using AMSES
The disk space usage of Superelement generation runs using AMSES has been significantly improved. This will allow for less resource usage and disk I/O requirements. Note that PARAM,AMSE4CMS,YES should be active (which is active by default).
GPU Support for Direct Frequency Response Analysis
GPU Support is now available for Direct Frequency Response Analysis.
Reflective Surface outside External Acoustic Cavity for APML External Acoustics
Reflective surfaces can now be defined outside the External Acoustic cavity for APML. The reflective planes are defined perpendicular to the location on the corresponding axis defined by the XRFL, YRFL, and ZRFL fields on the PACPML Bulk Data. There can only be 1 reflection plane perpendicular to each axis. The corresponding reflection factor for each of the reflection planes can be defined by the XFAC, YFAC, and ZFAC fields of the PACPML Bulk Data. The contribution of the reflection planes to far field pressure can be calculated based on the mirror rule incorporating the corresponding reflection factor.
Preloading support for Response Spectrum Analysis
Preloading is now supported for Response Spectrum Analysis. The STATSUB(PRELOAD) entry can be defined in the Response Spectrum subcase and it can point to the corresponding preloading subcase.
Absolute Maximum Principal Strain output for Random Response Analysis
Absolute Maximum Principal Stress output has already been available in previous versions when PARAM,PSDPRINC,YES in H3D format. Similarly, PARAM,PSDPRINC,YES now also activates output for Absolute Maximum Principal Strain output in H3D format for random response analysis.
Enhanced AMSES performance for MPI runs
AMSES performance has now been enhanced for both single-node and multi-node MPI runs. The improvement in performance shows good scaling when compared to the previous versions of OptiStruct for AMSES when run in DDM mode. The following explains in more detail about the feature by providing a simple example:
  • For single-node runs, where a model which uses AMSES for eigen-extraction is run in DDM mode, assuming the run configuration is 1x8x4, where the model runs with 8 MPI processes and 4 SMP threads on a single node, then when AMSES solver starts, the main MPI process will suspend the remaining 7 MPI processes and deploy AMSES to run on all physical cores on the node which were available for the job. In this case, 8x4=32 cores are now available for the AMSES run and AMSES will run on 32 cores SMP until the eigen-extraction is done. Then the main MPI process will revive the other 7 MPI processes and the DDM job continues normally from this point.
  • For multi-node runs, where a model which uses AMSES for eigen-extraction is run in DDM mode, assuming the run configuration is 2x8x4, where the model runs with 8 MPI processes and 4 SMP threads on each node of a 2-node cluster, then when AMSES solver starts, the main MPI process will suspend the remaining 15 MPI processes across both nodes and deploy AMSES to run on the node where the main MPI process runs and will use all physical cores on this node which were available for the job. In this case, on the node where the main MPI process runs, 8x4=32 cores are now available for the AMSES run and AMSES will run on 32 cores SMP until the eigen-extraction is done. Then the main MPI process will revive the other 15 MPI processes which were spread across the 2 nodes and the DDM job continues normally from this point.
Note that running AMSES under DDM mode can lead to some overhead due to communication between the secondary MPI processes and the main MPI process to retrieve the local matrices for AMSES to be run on the main MPI process. Therefore, AMSES may still be run optimally on pure SMP mode, but for cases where models using AMSES need to be run in DDM mode (like when there are NLSTAT preloading subcases), then it can help improve AMSES performance by allowing it to use all available physical cores on the node containing the main MPI process.
Fluid Grid loading now supported in OLOAD output
Fluid Grid loading via SLOAD, ACSRCE load is now output when OLOAD I/O Entry is specified.
Fatigue
Fatigue Analysis based on Steady-State subcase
Fatigue Analysis can now be performed based on a Steady-State Subcase. The LCID field on the FATLOAD entry should point to a Steady-State subcase to activate Fatigue with steady-state analysis. The TID field on the FATLOAD entry should be left blank.
MultiPhysics
Joule Output based on property, component, and SET
Joule loss density output has already been available via the HEAT I/O Entry for electrical analysis. Now the property/component/SET-based outputs are now included with the corresponding grid-based output via the PROP, COMP, and SET group options in the HEAT I/O Entry. Similarly, OPROP, OCOMP, and OSET options outputs only property, component, and SET-based output, respectively. This is currently only supported for H3D format.
Electrical Contact is now defined using Conductance per unit Area
Previously, the electrical contact definition via PCONTEC and PGAPEC were specified using resistance per unit area. This has now been switched to Conductance per unit area.
Fully Coupled Electro Thermal Mechanical Analysis
Electrical, Thermal, and Mechanical subcases can now be fully coupled. For this purpose, connectivity is updated in the electrical and thermal domains, based on the results of the mechanical domain. After a subsequent electrical analysis, Joule heating, as well as mechanical induced heating (for example, inelastic strain), is considered during a thermal analysis. The resulting temperature field is in turn used to update the temperature-dependent material properties and to compute thermal expansion in the mechanical analysis. The subcases which can be coupled are:
  • Nonlinear Implicit (Static or Transient) structural subcase
  • Nonlinear Transient Heat Transfer subcase
  • Multi-Steady Electrical Analysis subcase
The COUPLE(HEAT)=<ID> entry should be specified within the Nonlinear Direct Transient subcase where the ID points to the subcase ID of the Nonlinear Transient Heat Transfer subcase to be coupled. Similarly, the COUPLE(ELEC)=<ID> entry should also be specified in the Nonlinear Direct Transient subcase where the ID points to the subcase ID of the Multi-Steady Electrical Analysis. Note that COUPLE(ELEC) and COUPLE(HEAT) can also be used separately for corresponding coupling.
For electrical subcase in coupling, DDM is currently not supported. For heat-transfer subcase, cavity radiation is not currently supported. For structural subcase, NLCTRL is mandatory for this coupling (NLPARM is not supported).
When coupled, all results, including results for heat transfer and electrical subcase are printed to the _impl.h3d file which is generated from the Nonlinear Direct Transient subcase. PARAM,IMPLOUT,YES should be specified.
Optimization
Sensitivity output for Free-Shape Optimization
Sensitivity output is now supported for Free-Shape optimization. It can be activated using OUTPUT,H3DSHAPE. The ALL, STRESS, and NOSTRESS options are supported.
Enhancements for Draw Direction with thickness gradient Constraint in Free-Size optimization
Draw Direction with thickness gradient constraint in free-size optimization has now been made more robust and additionally enhanced with more options. The Draw direction along which the thickness gradient is to be applied should typically be orthogonal to the rib shell element normals, and OptiStruct, by default, checks the draw direction by comparing it to the element normals (within a specific tolerance). Based on this check, OptiStruct will issue an ERROR if the user-defined draw direction is beyond the specified tolerance.
Figure 1.


The COPT field on DSIZE entry can be set to:
  • CHECK: (Default) Checks the draw direction
  • ADJUST: Checks and adjusts the draw direction
  • SKIP: Does not check the draw direction
If COPT is set to CHECK, OptiStruct will check the draw direction, and:
  • Issue an ERROR if it is beyond the specified tolerance.
If COPT is set to ADJUST, OptiStruct will check the draw direction, and:
  • Adjust the draw direction, as long as it is within the specified tolerance.
  • Adjust the draw direction, and then issue an ERROR if the adjustment value was more than the specified tolerance
The TOLDIR field is the tolerance (in degrees, default=5.0) for the user-defined draw direction.
The TOLBOT field is the tolerance (in degrees, default=10.0) to determine if a side edge should be treated as a bottom edge.
The TOLTOP field is the tolerance (in degrees, default=10.0) to determine if a side edge should be treated as a top edge. If the angle between an edge and the draw direction is lower than TOLBOT/TOLTOP, then it is considered as a side edge.
ESL optimization for Transient Analysis
The Equivalent Static Loads (ESL) method can now be employed within OptiStruct to solve linear transient optimization problems. The ESL method creates a number of equivalent linear static auxiliary sub-problems based on the solution of the corresponding Linear Transient analysis. Those auxiliary problems are solved in a nested loop. The analysis and the nested loop are repeated in an outer loop until convergence is achieved.
Supported analysis types are:
  • Linear Direct Transient Response (DTRAN)
  • Linear Modal Transient Response (MTRAN)
The method is activated using DOPTPRM,NESLOPT,#ET, where #ET specifies the number of ESL auxiliary load-cases created. The time represented by each auxiliary load case can be specified using the ESLTIME card, for example, such that the extremes of a specified response are captured. Moreover, the parameters DOPTPRM,DESMAX and DOPTPRM,ESLMAX control the maximum number of outer loops. Currently, Stresses, Displacement and Compliance are supported as responses.
Note: The number of ESL-times should be kept low to enhance runtime performance.
General
Multiple Results Output for the same Output request
Multiple outputs of the same output type can now be requested for single or multiple subcases. The following options are now supported.
  • For the same output, multiple requests can now be specified, with different options, such as SET IDs, result types, such as VON, DIRECT and so on.
  • For each such output request, the FILE=filename option should be specified to identify the separate filename to which this result is to be output. Output with different options can be printed to different files or to the same file, based on the specified filename in the FILE=filename option. The FILE option is not currently supported for OP2 format.
  • The SYSSETTING(MULTIPLEOUTPUT=YES) option should be specified to activate this feature.
  • It is currently supported for:
    • Linear static, normal modes, direct and modal frequency response, direct and modal transient response analyses types.
    • Displacement, Stress, Strain, and Force results.
    • The H3D, OP2, PUNCH, and OPTI formats.
SDISP, SVELO, and SACCE output for HDF5 format
SDISP, SVELO, and SACCE outputs are now supported in HDF5 format for Modal Frequency Response and Modal Transient Response analyses. These were already supported in H3D, OP2, and PUNCH formats in previous releases.
Global-Local Analysis is now supported for Direct and Modal Transient Analyses
Both combinations are supported, but the majority of use-cases would likely be Modal Transient in the global model and Direct Transient in the local model. Currently only the two-step modelling approach is supported. The results from the global model are imported from the H3D file using ASSIGN,H3DRES. The specific subcase in the global model from which the results are to be imported is specified using IMPORT,SUB. A grid SET is created in the local model to identify the interface/transfer zone and SPCD is defined on these grids with field D set to M to indicate that they are used for mapping. Since this is transient analysis, the SPCD should be referenced by TLOAD1 entries.
Threshold Support for STRAIN output
Threshold options RTHRESH, THRESH, RTOP, and TOP are now supported for STRAIN output. This is supported for both Total and Neuber strain. It is currently supported for H3D format only. It is supported for Linear Static, Nonlinear Static, Normal modes, Frequency response, and Transient analyses.
Threshold Support for Neuber STRESS output
Threshold options RTHRESH, THRESH, RTOP, and TOP are now supported for Neuber STRESS output. Note that threshold options were already supported for stress output without Neuber in previous releases. It is currently supported for H3D format only. It is supported for Linear Static, Nonlinear Static, Normal modes, Frequency response, and Transient analyses.
Statistics output for STRAIN output in Linear Transient Analysis
Statistics output for STRAIN output is now supported for both direct and modal Linear Transient Analysis. This is supported for both total strain and Neuber strain. They are activated using the STATIS or OSTATIS options. STATIS output regular strain along with statistics over time, while OSTATIS outputs only statistics over time.
Statistics output for ERP output in Linear Transient Analysis and Steady-State Analysis
Statistics output for ERP output is now supported for both direct and modal Linear Transient Analysis and Steady-State analysis. They are activated using the STATIS or OSTATIS options. STATIS output regular ERP along with statistics over time, while OSTATIS outputs only statistics over time. This is currently only supported for H3D format.
Applied Temperature output via OLOAD for HDF5 format
Applied temperature load output is now supported for HDF5 format when OLOAD is defined in Linear Static analysis. If additional structural loads exist along with temperature load, then they are also output along with the applied temperature load output.
MAT8 Output in HDF5 format
When any HDF5 specific output request is defined in the model, for example, DISP(HDF5)=ALL, or if OUTPUT,HDF5,YES is defined, in a model which contains MAT8 material, then this MAT8 material is output to the HDF5 file.
Energy output for Linear Static analysis with Neuber correction
Energy output is now available for Linear Static analysis with Neuber correction. The NLENRG Bulk/Subcase pair can be used in Linear Static analysis where Neuber correction is used. Note that the NLENRG Subcase Entry should have an additional NLENRG(NEUBER)=ID option defined for this output. The internal energy and plastic dissipation energy values for the full model and/or the specified element SET in the filename_e.out file.

Resolved Issues

  • For Explicit Auto-Contact, the specified PCONT with specific ID will be used now for the particular contact interface, even if PCONT with option ALL is defined.
  • Only Large Displacement nonlinear analysis is supported for OptiStruct-AcuSolve FSI coupling. Previously, using small displacement nonlinear analysis would lead to incorrect results. This has now been fixed by printing a clear error requesting the user to switch to large displacement nonlinear analysis in OptiStruct.
  • The following issues were resolved for Draw Direction in free-size optimization:
    • For ribs which contain overhangs, there are multiple bottom and/or top edges along the draw direction. This is now properly handled and the correct base is now identified for such ribs.
    • Unpredictable results could be previously observed when DSIZE with Draw-direction was combined with another DSIZE without any manufacturing constraints. This has now been fixed.
    • A crash or unpredictable results were previously observed when DSIZE with regular draw-direction and DSIZE with draw-direction with thickness gradient optimization were combined in the same model. This has now been fixed.
  • An explicit analysis model with TIE contact previously crashed with data manager check. This has now been fixed.
  • The FSTHICK file for free-size optimization contained incorrect nodal thicknesses. This has now been fixed.
  • When one of the parts connected by FREEZE, CONTACT is removed using MODCHG (with PARAM,MCHGRMV,1) but the associated contact definition is not removed, the run previously failed with ERROR 4772. This has now been fixed.
  • Incorrect Displacement and Stress results were generated for elements which are close to rigid elements for SMDISP Inertia Relief models. This has now been fixed.
  • OptiStruct-Flux Optimization with both EM and Mechanical responses previously required a large amount of memory. This has now been fixed.
  • Parameter Optimization with load design variable and SPCFORCE moment response previously did not show a progression in the optimization, and soft converged in a single iteration. This has now been fixed.
  • Coincident RBE2s led to double-dependency error even if the second RBE2 was excluded using MODCHG. This has now been fixed.
  • When multiple MATMDS with different number of constituent materials (for example, are associated with a Laminate), unpredictable results were observed. This has now been fixed.
  • Incorrect results were generated when TLOAD2 entries were referenced by DLOAD in an Implicit Nonlinear analysis. This has now been fixed.
  • A model with nonlinear heat transfer analysis with higher number of SMP threads previously led to a memory insufficient error. This has now been fixed.
  • For OSTTS optimization, when the initial temperature for the transient heat transfer subcase is zero, previously the run failed with an error that all sensitivities for the objective are zero. This has now been fixed.
  • An incorrect ERROR was generated when TEMP(LOAD) in a Nonlinear Transient subcase with CNTNLSUB was pointing to a different temperature load set from the TEMP(INITIAL). This has now been fixed.
  • For electrical optimization with non-electrical responses (such as volume, mass, and so on) without any structural material present, previously the model failed with a programming error. This has now been fixed and a clear error is now issued which mentions that a valid structural material is required for structural responses.