Simulation Softwares

The KB3 workbench that EDF has been developing since the 90s allows automation of dependability studies by capitalizing on the knowledge acquired on systems through knowledge bases written in an object-oriented modelling language called FIGARO.

From the graphic input of a model, made possible by the definition of a knowledge base dedicated to the problem to be treated, the tool KB3 automatically builds reliability models. These models can be static, such as fault-trees, or dynamic, exploitable by means of markov methods or by Monte-Carlo simulation.

The KB3 workbench also allows the reliability analysts to build graphically all their usual models:

• fault-trees
• reliability block diagrams
• markov graphs
• petri nets
• reliability networks
• many extensions of these standard models. In particular the KB3 workbench fully supports the formalism BDMP (Boolean logic Driven Markov Process)^®, that is extremely powerful and general

Models built with KB3 can be processed in various ways with several calculation tools, developed either by EDF (FIGSEQ, YAMS, FIGMAT-SF for dynamic models), or by other companies (RISKSPECTRUM, GRIF-Tree for the fault-trees generated by KB3). These tools allow making calculations of reliability, availability, productivity, costs... They also give precious qualitative indications to improve the performances of the studied systems: minimal cut sets or dominating sequences, importance factors...

KB3 offers a very user-friendly graphic interface, with generic functions that are useful in all applications: levels of detail, display modes in colour, interactive simulation, optimized management of variants of models, copy and paste of graphs and data towards Microsoft Office tools.

• time savings in the implementation of studies
• accessibility of the tool to non-experts in reliability, thanks to knowledge bases dedicated to applications. A knowledge base (in French) contains the generic description of the classes of components which one can meet in the systems of a given category. For example, in a knowledge base dedicated to the study of thermohydraulic systems, one will find the generic description of components of type pump, valve, heat-exchanger, etc.
• quality improvement of studies obtained by a guarantee of consistency between the various models built with the same knowledge base and the traceability of the data
• the possibility of launching several processings from a unique model

Among the possible applications of KB3 we can quote :

• availability studies performed to optimise the design or the exploitation of industrial systems : production installations, public networks (electricity, transport, communication)
• reliability assessment of high technology systems (on-board systems, medical devices...)
• the control of the safety of risky industrial processes, e.g. in nuclear, chemical, petrochemical industries
• security evaluation of "cyber-physical" systems. It is possible to generate the scenarios of accidental failure or attack from a single model

The KB3 installer containing the demonstration version and several examples of the knowledge bases and the system models can be downloaded here.

The only limitation of the KB3 demonstration version is that the size of studies that can be created, loaded etc. is limited to 80 objects.

The database "Pedagogical v2-x-y kit" contains the following knowledge bases (in French and English), with several examples of systems models associated to each knowledge base:

• RBD: dependability diagrams with parts in standby redundancy, shared repair teams, gates k/n
• PETRI: stochastic Petri nets, extended by the ability to condition certain transitions by Boolean expressions defined graphically with the help of logic gates
• BDMP (Boolean logic Driven Markov Processes)®: this knowledge base includes the possibility of the 2 previous one, and more. See the BDMP inset for more information
• Telecom_basic: communication network made up of data sources, transmission nodes, mono-directional and bi-directional edges
• Miniplant: production facility with variable capacity depending on component capacity and whether or not they are out of order. This knowledge base contains a FIFO stack model to manage component repairs
• Skelbo: knowledge base dedicated to thermohydraulic systems modelling

Also find databases containing varied examples of BDMP for safety and security on the site.

Visual Figaro has been developed as a plugin for the Jedit text editor (in Java, Open Source). This environment allows you to easily modify the example knowledge bases, even develop your own.

Visual Figaro itself, as well as the examples of instructional knowledge bases, can be downloaded on Sourceforge.net These instructional bases are those found already present in the "Pedagogical v2-x-y kit". They allow:

• reliability block diagrams (DIAGFIA)
• modelling systems that function with variable production levels.

MINIPLANT 1, 2 and 3, in growing complexity, are tutorials that lead to a knowledge base that allows the resolution of the 4 variants of the MINIPLANT test case: computation of reliability, availability, and system equivalent availability.

Read the article describing the MINIPLANT test case with a detailed explication of how the Miniplant3 knowledge base functions.(in French)

• Petri nets (PETRI)
• simple electric systems, described in the following article (in French)
• communication networks (Telecom_basic)
• thermohydraulic systems (Skelbo)

These knowledge bases are a starting point to be enriched by Visual Figaro users, why not you?

The dependability study of complex systems with reconfiguration capacities, standby redundancy (normal and backup functioning), or other dependency types between components require the use of behavioural (or dynamic) models, in which we explicitly model the random process that makes the system evolve from state to state, until it reaches an undesirable state.

BDMP ("Boolean logic Driven Markov Processes"), created in 2002, are a new graphic formalism that allows these models to be easily defined. They also have mathematical properties that facilitate probabilistic calculations.
The graphic representation of BDMP is very close to that used in fault trees, which makes them easy to use. The following document contains various examples that demonstrate BDMP modelling power and ease of use.

Note :

A training session is available for the use of the KB3, BDMP, FIGSEQ and YAMS tools.

Check the ITech training catalogue (in French)