LS-OPT 3.3

The following new features are available in Version 3.3:

• Radial Basis Function networks are available to create surrogate models. The building of these metamodels is significantly faster than Neural networks.

• Multi-Objective optimization (MOO) for converging to the Pareto optimal front is supported as both a direct and metamodel-based solver.

• Robust parameter (Taguchi) design is supported. The variation of a response can be used as an objective or a constraint in the optimization process.

• Mapping of results to the FE mesh of the base design the results are considered at fixed coordinates. These capabilities allow the viewing of metal-forming robustness measures in LS-PREPOST.

• The ANSA morpher is supported as a preprocessor.

• The truncated normal distribution is supported.

• Extra input files can be provided for variable parsing.

• A library-based user-defined metamodel is supported.

• Interfacing to a Windows Compute Cluster Server (CCS) is supported.

• Statistics can be calculated for checkpoints using an existing metamodel.

• The 3-D Metamodel plotting including point plotting has been significantly enhanced.

• Constraint plotting on 3-D metamodels is available with color-coded feasible/infeasible regions.

• Regression statistics for metamodels (Mean, PRESS, R-squared) integrated with metamodel "Accuracy" plots.

• The file viewing option (menu-bar) has been significantly enhanced. "lsopt_report" summary file refined and expanded to more types of analyses.