COMREL performs time-invariant reliability analysis of individual failure modes based on advanced FORM/SORM methodology. Several algorithms to find the most likely failure point (ß-point) are implemented including a gradient free algorithm for non-differentiable failure criteria (state functions). Complementary or alternative computational options are Mean Value First Order (MVFO), Monte Carlo simulation, Adaptive Sampling, Spherical Sampling, several Importance Sampling schemes and Subset Simulation.

FORM/SORM techniques allow to compute a rich set of sensitivity measures showing the impact on reliability of individual basic random variables, of distribution parameters and of other constant parameters. Provided characteristic values are specified partial safety factors for all basic variables are another straightforward result.

COMREL can deal with arbitrary dependence structures in the stochastic model (Rosenblatt, Hermite and Nataf-models). The full set of stochastic models offered by STATREL is supported (44 models at present) and can be input either in parameter form or in terms of the first two moments and additional parameters if necessary. The models can be truncated and new user defined models can be added. You can make distribution parameters dependent on other variables, parameters and even functions. Dependencies can also be described in terms of correlations when this is theoretically admissible. The increased versatility in stochastic modeling certainly is one of the strength of COMREL/SYSREL.

In COMREL several failure criteria can be defined in one job. State functions can be either easily implemented in the Graphical User Interface or called from external programs. State functions can be specified in normal mathematical notation. Names for variables and parameters can be chosen freely and are automatically transferred into the stochastic model and vice versa. Important constants are predefined. Built-in functions include all elementary, trigonometric, hyperbolic, logarithmic and some special functions like the Gaussian distribution function and its inverse, Bessel and Gamma functions. Several alternatives for numerical integration, differentiation and root finding are available as well as comparative operators and test functions. Auxiliary user defined functions and reference functions can be defined.

Time-variant reliability is computed by the outcrossing approach also based on FORM/SORM methodology for stationary or non-stationary cases. Available random process models are regular or intermittent rectangular wave processes and differentiable Gaussian and non-Gaussian translation processes (Hermite or Nataf processes). Both models can be scalar processes and vector processes. All random process types can be combined with each other including the possibility to combine intermittent and non-intermittent processes. Sensitivity measures and partial safety factors are provided similar to time-invariant analysis. In addition, various other exceedance measures like excursion time, hazard rate and point-in-time non-availability are evaluated.

SYSREL covers system reliability evaluation including event updating. The graphical user interface offers easy and efficient way to define the model. The definition of the stochastic model (basic random variables) is the same as in COMREL with all features to model dependencies. The logical model in SYSREL is connected with the failure criteria and the stochastic model (basic random variables) for a fully interactive control. System modeling includes not only the representation by a (minimal) set of parallel systems in series but also the important case of conditional events (observations, event updating).

When working with FORM/SORM in SYSREL, you have access to several efficient and reliable algorithms searching for the ß-point (most likely failure point) with special solution strategies. An alternative reliability method available in SYSREL is Monte Carlo simulation.

COSTREL covers both minimization of a user definable objective function (cost function) under a reliability constraint (max. allowed failure probability) as well as minimization of the failure probability under a cost constraint and other constraints. The reliability computation is based on the FORM/SORM methodology. The optimization task is performed by two alternative algorithms (NLPQL & Joint5) using the so called "one-level" approach (optimization to find the beta-point and optimization of the objective are done in one level). Besides the solution for the cost variables together with elasticities w.r.t. total cost also sensitivity measures and partial safety factors for all random variables are provided as in COMREL and SYSREL.

STATREL is a program with many special features for statistical reliability-oriented data analysis but also covers standard statistical analyses. It provides a useful set of tools to perform basic descriptive statistics with many graphical facilities. For all models included in other STRUREL modules STATREL performs parameter estimation by different methods, confidence interval and quantile estimation as well as hypothesis testing including tests for sample validity, distribution functions and parameters. Simple analysis of variance and regression is also included. Several Bayesian methods are implemented.

Results are made visible in terms of numerous graphical representations such as histograms, cumulative frequencies, bivariate plots, bar charts, probability paper plots. Import of data from spreadsheet programs like Excel and export of stochastic models to COMREL and SYSREL is possible. Schemes for simulation of random numbers, random vectors and random time series (stationary, non-stationary, Gaussian, Hermite)allow numerical experiments. Different generation techniques such as ARMA or by simple and fast Fourier transforms can be used. Many predefined spectra can be selected and various forms of non-stationarities can be defined.

The implemented features for time series analysis provides rich tools for setting up and testing models. The features include numerous graphical representations of the time series such as histograms, probability paper plots, scatter diagrams, moving averages, Husid functions, mean value crossings and periodograms. Many transformation and filter techniques including trend removals, smoothing, differentiation and integration, amplitude and frequency modulation are provided. Non-parametric spectral estimation and ARMA modeling as well as amplitude and frequency trend estimation are available.

STATREL v.3.14 is freeware which can be downloaded from the Demo part of this website. It is a 32-bit application.