Advanced science.  Applied technology.


NESSUS® Release Notes

NESSUS 9.0 and newer bring many new enhancements. NESSUS 9.8 introduced the NESSUS Response Surface Toolkit. NESSUS 9.7 added a variety of enhancements, including improved support for working with Gaussian Process response surface models, as well as a new dynamically linked user-defined model type. With NESSUS 9.5, the graphical interface has been reworked to greatly simplify the process of setting up and solving a problem.  The following list provides an overview of the highlights in these releases.

New Capabilities and Enhancements

New in Version 9.9

  • Updated format for smx output file. Now uses comma separation, includes variable names, and includes values for all computed variables.
  • Support additional distribution types with correlated variables
  • Updated theory manual for global sensitivity analysis
  • Store GP RSM file when using variance decomposition
  • Bug fix for numerical model input and output file names that include spaces
  • Bug fix related to error checking in maximum number of equations
  • Bug fix for LS-DYNA minimum, absolute minimum, and absolute maximum time selections
  • Abaqus response extraction improvements
  • Fix for delta vector mapping with deterministic variables
  • Switch to the use of 64-bit Java and executable code (user-defined shared libraries should now be compiled in 64-bit mode)

New in Version 9.8

  • Added the NESSUS Response Surface Toolkit, a standalone application for designing, fitting, assessing, and visualizing response surface models
  • Minor improvements to response extraction for the Abaqus interface

New in Version 9.7.4

  • Fix for bug with small LHS sample sizes
  • Fix for locating ext_cmd.cfg on Windows
  • Changes to arguments for running NESSUS from command line

New in Version 9.7

  • Gaussian Process model enhancements:
    • GP model definition saved to file for later re-use without needing to re-fit the model
    • Options for defining and fitting observation noise level
  • New capability to define models using dynamically linked libraries
  • Monte Carlo simulation now supports all distribution types
  • Use of Sobol sequences for global sensitivity analysis to significantly increase efficiency
  • Added p-level and full CDF capabilities for EGRA
  • Updated installation layout for Windows 7 compatibility
  • Fixes for beta distribution
  • Increased limit on number of runs in Deterministic Analysis mode

New in Version 9.6.4

  • Improved error messages for variable mapping with external models
  • Fixes for Matlab interface
  • Added endcycle to WCN function argument list. This variable defines the number of cycles for the endurance limit.

New in Version 9.6.3

  • Improved compatibility with Windows 7 (elevated permissions not required)
  • Minor bug fixes

New in Version 9.6.2

  • Changed GP regression model and RSM_GP method to build GP model with linear trend function instead of constant

New in Version 9.6.1

  • Added polynomial coefficients and goodness of fit statistics to *.out file for regression models
  • Updated Abaqus interface to work with version 6.11
  • Several minor bug fixes for the GUI

New in Version 9.6

  • New capability for global sensitivity analysis using variance decomposition
    • Identify important variables based on their contribution to total variance of the model response
    • Three variance decomposition analysis methods are provided
  • New wizard for pasting and importing data into NESSUS tables
  • Bug fixes:
    • Fixed error when running LHS with triangular distribution
    • Fixed bug when using deterministic analysis with deterministic variables

New in Version 9.5.5

  • CTH interface enhancements and bug fixes:
    • Corrected issue with checking for existing error file for latest compilers
    • Defines last time response value for the restart database when the response is a time history file
  • Enhancements and bug fixes for the WCN model:
    • Added user definition for cycles corresponding to the endurance limit
    • Improved checks for a, c, and K validity limits

New in Version 9.5.4

  • Bug fixes for Abaqus displacement response extraction

New in Version 9.5.3

  • New regression model: Gaussian Process
  • NASTRAN results enhancements:
    • Ability to work with larger models
    • Capabilities for composite materials, including composite failure indices
    • User-defined datablock name
    • Debug option
  • Bug fixes to use correct response level with EGRA and ISMPP

New in Version 9.5

  • Simplified outline layout
    • Problem Definition node begins with problem statement definition instead of system definition (now Fault Tree Definition)
    • Fault Tree, Correlations, and Confidence Bounds are now accessed via popup windows instead of separate outline nodes
    • Consolidated "Deterministic Analysis" and "Parameter Variations Analysis"
    • Consolidated probabilistic analysis type and method configurations
    • Simplified procedure for running analyses and viewing results
  • Improved interfaces for generating finite difference step sizes and deterministic input values (parameter variations)
  • Inverse reliability analysis support for RSMGP
  • Additional advanced options for EGRA
  • Fix for triangular distribution

New in Version 9.1

  • Improved distribution editing interface
    • PDF equations displayed in the GUI
    • Short descriptions of distribution parameters in GUI
  • Support for MacOS X
  • Miscellaneous minor bug fixes

New in Version 9

  • New Reliability Analysis Methods
    • ISMPP: Importance sampling at user-defined MPP
    • EGRA: State-of-the-art surrogate-based method designed for extremely expensive response functions. Unlike MPP-based methods, it is not gradient-based
    • RSMGP: Gaussian process surrogate method
  • LHS enhancements (advanced options)
    • Optimal design based on max-min distance criterion
    • 3 correlation reduction methods for small samples
    • Capability to augment (double) sample size while maintaining Latin Hypercube structure
  • Probability distribution modeling
    • New distributions: beta, exponential GEVmax, GEVmin, gamma, Pareto, triangular, uniform
    • Capability to switch between definition in terms of natural parameters or mean & standard deviation

Operating Systems

  • Windows (XP, Vista, 7)
  • Linux
  • Mac OS X