Altair OptiStruct 2019 Release Notes

Highlights

  • Nonlinear Axisymmetric Analysis (Elasto-plasticity and large displacement nonlinearity)
  • Combination Joints via JOINTG
  • Static Analysis for Rotor Dynamics
  • Mode Tracking and Rotor Energy output from Rotor Dynamics with Complex Eigenvalue Analysis
  • Symbolic Substitution
  • New Free-shape optimization(GRID-based) – Beta version
    • Allow the nodes to move normal to shell plane
    • More flexible shape changes than the classical free-shape method
    • Support for both shells and solids
  • Multi-Model Optimization (MMO) enhancement:
    • DGLOBAL with MMO

New Features

Stiffness, Strength and Stability
Axisymmetric analysis
Support for Elasto-Plasticity and Large Displacement nonlinearity
Support for QUAD element type (4 and 8 node CQAX1 Bulk Entry)
New combination JOINTG types
JTYPE = AXIAORIE (Axial + Orient)
JTYPE = INLICARD (Inline + Cardan)
JTYPE = RLINORIE (Rigid Link + Orient)
CGAP(G)
Can be included in large displacement nonlinear analysis
Damping stabilization can be activated via CNTSTB Bulk and Subcase entries
Incremental output for CSTRESS, CSTRAIN and CFAILURE (NLOUT)
Incremental output support via NLOUT Bulk and Subcase entries.
Search distance output in .out file
When no contact is generated due to the inappropriate search distance, warning #4735 is printed with estimated search distance (SRCHDIS) which is the largest distance between a pair of grids on master/slave.
CONTF I/O Entry
Contact results output for Linear Analysis
CBEAM/CBAR
Pin Flags elements are supported for Large Displacement Nonlinear Analysis.
Currently, only 4, 5, 6, 56, and 456 degrees of freedom are available for LGDISP.
Thickness change output
Due to large displacement nonlinear analysis is output in .h3d file.
MONITOR feature for Nonlinear Analysis
A quick convergence summary is output by default in the .monitor file for nonlinear analysis. This file contains a summary of number of cutbacks in current step, the number of contact iterations and so on. The displacement value on particular degrees of freedom can also be monitored and output in the .monitor file. MONITOR Bulk and Subcase entries can be used to specify the degree of freedom of interest for displacement monitoring during nonlinear runs.
Rotor Dynamics
Static Analysis
Rotor effect (RGYRO) can be included in static analysis with centrifugal force
Mode Tracking
Now available for rotor dynamics with complex eigenvalue analysis.
Rotor Energy
From the complex eigenvalue analysis is available in the .rengy file via the RENERGY output request.
Noise and Vibration
Modal results output from PFMODE
Displacement and Strain/Kinetic Energy results on eigenvectors with large participation identified by PFMODE are directly available in Modal Frequency response.
SPCD
Now supported on fluid grids for Modal Frequency Response Analysis.
FASTFR
Multiple modal spaces are now supported with FASTFR.
PBUSHT
Frequency dependent bushing is now available with PARAM,FASTFR,YES.
Inclusion of missing mass effect for Response Spectrum Analysis
Inclusion of missing mass effect will be achieved by adding the spectrum acceleration (at zero period), on the RSPEC entry via the ZPA fields. This inclusion enhances the accuracy of acceleration output.
Inclusion of MFLUID mass to mass summary in the Grid Point Weight Generator output
The mass of MFLUID is included in the Grid Point Weight Generator output in the .out file if both PARAM,VMOPT,1 and PARAM,VMMASS,YES are specified. The MFLUID mass is not included by default (PARAM,VMMASS,NO is the default) when PARAM,VMOPT,1 is present.
Fatigue Analysis
Default Structural SN curve
Seam Weld: If the SN curve is not defined by the user, a default Structural SN curve for Seam Weld is used based on a Stress Ratio of R=-1.0. The Stress Ratio “R” cannot be modified for Seam Weld Fatigue.
Spot Weld: If the SN curve is not defined by the user, a default Structural SN curve for Spot Weld is used based on a Stress Ratio of R=0.0. The Stress Ratio “R” can be modified on the SPWLD continuation line of the MATFAT Bulk Data Entry. Note that the Stress Ratio “R” can only be modified for Spot Weld Fatigue and it can only be set to 0.0 or -1.0 (default SN curve is always input at R=0.0 for Spot Weld Fatigue).
Stress Ratio
SN curve for Spot Weld can be defined based on Stress ratio R=0.0. Stress ratio (R=-1.0 or 0.0) can be defined on the SPWLD continuation line of the MATFAT Bulk Data Entry. Default is R=-1.0 if the user inputs SN curve and the Default is R=0.0 if the SN curve is not input by user and OptiStruct uses the default curve for Spot Weld Fatigue.
Default value of Tref in PFATSMW and PFATSPW
Changed to a mm-based value and set to 25 units.
Optimization
New GRID-based Free-shape optimization with more flexible shape changes for shells and solids
A new option, GRID, is available on the 3rd field of DSHAPE Bulk Entry that allows more flexible shape changes (including normal to the shell surface) as each DSHAPE grid has its own design variable.
The old method (now called as “CLASSIC“) is still the default.
Powerflow optimization response
Available for Topology, Sizing, and Shape optimization
Principal Stress/Strain optimization responses
Available for Frequency Response Analysis.
DGLOBAL
Supported with Multi-Model Optimization (MMO)
Optimization response (DRESP1)
A SET of elements can be used to define the optimization response.
Maximum member size (MAXDIM)
Available with OVERHANG constraints.
Lattice sizing cleaning
Some penalty is imposed during the optimization phase to minimize such discrepancy between the optimization and re-analysis.
Large Shape Change
Supported with Domain Decomposition Method (DDM).
Auto-normalization of Weighted Compliance Response
The Weighted Compliance Response can now be auto-normalized based on the user-defined subcase-dependent weights using DOPTPRM,AUTOWGHT,YES. This allows the user-defined weights to influence the optimization, as intended, regardless of the value of the compliance at Iteration 0. When DOPTPRM,AUTOWGHT,YES is specified, the user-defined weights are normalized with the corresponding compliance values at Iteration 0 for each subcase and subsequent new weights are determined for each subcase. These new weights are then utilized for the entire optimization without further changes. DOPTPRM,AUTOWGHT,NO is the default.
Output
Normal, Shear, and von Mises Stress output
For CBEAM and CBAR in Frequency Response Analysis is supported
Sound Pressure Level (SPL) output
In punch format with RADSND analysis is supported
PSDM output request
For STRESS can be used to request the output in Random Response Analysis. Axial stress for CBEAM and CBAR elements are also output.
Group Energy output
By SETs of elements is supported via the boolean operator, OR.
SET (each SET + elemental energy) and OSET (only each SET) options are introduced for ESE, EKE and EDE output requests for energy output by SET.
Stress and Force output
Available for Fourier-based Transient Analysis
ESE/EKE
Available for Complex Eigenvalue Analysis
SPCF/MPCF/GPCFORCE output
Supported for Linear and Nonlinear Transient Analysis
MPCFORCE output
Supported for Modal Transient Analysis with SDAMP
Mass output with Virtual Fluid Mass
PARAM,VMOPT,1 is available in the .out file with PARAM,VMMASS,YES.
SECTION output
Available for analysis only (for solids).
SECTION results
Automatically available (for solid pretension bolts) in the .out file, as well as separate .secres file.
Acceleration output
Available from Response Spectrum Analysis
ERP output
Available from Complex Eigenvalue Analysis
MPCF output
Available from CMSMETH in .op2 file
Solvers and Performance
Parallelization (SMP)
Kinetic Energy calculation is now SMP-enabled to improve the speed with multiple processors (-nt).
DDM support for GPFORCE calculation in Transient Analysis
For GPFORCE calculations in Transient Analysis
DDM support for Preloaded Complex Eigenvalue Analysis
Now supported for Preloaded Complex Eigenvalue Analysis with preload coming from Nonlinear Static Analysis subcase. The preloading Nonlinear subcase is now DDM-enabled; however, the subsequent preloaded Complex Eigenvalue Analysis will only run on one MPI process. This provides DDM support for Brake Squeal Analysis, as well.
SCOTCH and PTSCOTCH ordering methods
Supported for 64-bit OptiStruct runs (-i64)
64-bit OptiStruct runs (-i64)
Supported with Platform MPI
GPU support for PCG solver
Now supported for GPU runs. It is recommended for typical solutions where PCG solver is recommended, for instance, linear static analysis with solid models and compliance-based optimization. Good speedup is observed for PCG models with GPU, for example, a speedup of 2.6 times was achieved for a model run on 16 CPU cores + 1 GV100 GPU when compared to just a 16 CPU cores run.
Multiple GPU Support for PCG solver
Multiple GPU cards are supported for the PCG solver via the -ngpu # run option, where # specifies the number of GPU cards to be used. This is helpful for large models which otherwise cannot be solved by a single GPU card. OptiStruct will terminate with an error if a particular model is too large to be solved by a single GPU card. In such cases, multiple GPU cards can be used.

Enhancements

PELAST
Allows negative table input
RBE2GS
Defines an RBE2 connecting the two closest grids to a grid point, GS.
TIC
Defining Initial conditions is now supported for Modal Tranisent Analysis
PARAM,AMSE4CMS,YES
Support for GM and Guyan Method. The condensation will be accelerated by AMSES when this parameter is used. Significant speedup is possible, especially for a model with a large number of interface degrees of freedom. Note that when AMSES is selected in the CMSMETH Bulk Data, this parameter is automatically turned on.
SET for GRIDs
Can be specified based on the specific properties or materials.
Automatic EIGRL-based Inertia Relief
Can be activated for all static subcases with SPC’s in the model.
Skip the FRF Solver calculation
Random Response analysis is post-procesing of FRF. FRF subcase’s solver calculation can be skipped by importing FRF displacement from the previous FRF or Random Response Analysis. In order to import FRF displacement, a new ASSIGN option, H3DRES, is introduced.
Symbolic substitution
Provides flexibility to modify the input file to use parameterized input to define various data fields across the model. Currently, only real-valued data fields of entries in the Bulk Data section are supported for parameterization using symbolic substitution. Data fields in the Bulk Data section can be controlled using such parameterized input variables, which allows for rapidly manipulating certain key model attributes to conduct studies on how the response of the model varies based on these attributes.
Symbolic substitution can be defined using the following keywords:
%setrepsym <variable_name> = <variable_value> 
%defrepsym <variable_name> = <variable_value> 
%setrepsym creates a variable for symbolic substitution and %defrepsym creates a default variable for symbolic substitution. The symbolic variables can be unset using the following keywords, respectively.
%unsetrepsym <variable> 
%undefrepsym <variable> 
The syntax for using the symbolic variables in the fields of a Bulk Data Entry is:
%<variable>% 
Multiple variables are allowed, and the same variable can also be used for multiple parameterizations on multiple Bulk Data Entries. Additionally, multiple parameterizations on the same Bulk Data Entry is also supported.
Acceleration loading (ACCEL, ACCEL1, and ACCEL2) from reduced DMIG mass matrices
Generation of acceleration loading (ACCEL, ACCEL1, and ACCEL2) from reduced DMIG mass matrices is available in the residual run. This can be controlled via PARAM,CMSALOAD,YES/NO.

Resolved Issues

  • ELFORCE results were incorrect for nonlinear springs with PELAST.
  • Improved pre-processing time on ACCEL1.
  • Non-Smooth bore distortion results could occur earlier when there is even a small offset in the Z-coordinate of grids in a layer of bore cylinder. More accurate identification of grids of each layer of bore cylinder is implemented to produce the smooth results.
  • CFAST force results could be incorrect, if GS option is used for modeling.
  • Inaccurate SPCF attached to MPC in Large Displacement Nonlinear Analysis.
  • Programming errors for datablock “ielmlstsol” could occur for shape optimization with Contact or CGAP(G).
  • Inaccurate results occurred for Direct Frequency Response, if SPCD was applied on fluids grids with low fluid density specified.
  • Singularity error 3400 could occur when both MATS1 and composite failure criteria are specified.
  • Frequency response results with MFLUID could be incorrect with DDM mode.
  • CSTRESS sensitivities for free-size optimization could be incorrect if the local system is assigned on grids.
  • Internal programming error in “gapss_stabnl.F”.
  • OptiStruct no longer errors out when the unloading curve of gasket is incomplete, that is, the starting point of unloading curve is below yield point of loading curve.
  • Thermal loading for CBEAM/CBAR in large displacement nonlinear analysis is correctly supported.
  • Topology optimization combined with discrete sizing variables could cause a crash in approximation module.
  • Internal programming Error #252 could occur for optimization runs when GPSTRESS is requested for non-static subcase.
  • SMCORE option on ply-based modeling (STACK and PCOMPP) was handled incorrectly.
  • Inaccurate sensitivity for the composite stress response with free-size. This issue in old version only occurs when GRID entry has output coordinate system defined (CD field).