[en] Existing standard fault tree assessments of systems do not include an assessment of the effects of dependencies in an integrated fashion, but simply add on a ''common cause cut-off''. To support the values used for this cut-off, cut-sets involving certain groups of components susceptible to dependent failure can be assessed using the Distributed Failure Probability method. These rank one contributions do not cover all the possibilities, however, so there is an outstanding need for an integrated procedure for dependent failure assessment of systems which allows for all ranks of cut-set. The purpose of this note is to provide such a procedure which builds upon the standard approach to fault tree analysis. In this standard approach, only a limited number of cut-sets is found, and they are evaluated assuming independence of their components. So, some cut-sets are neglected which could be important contributors to the system failure probability if their components are not independent of each other. The procedure developed therefore deals with this truncation problem and with dependency together. The result is a practical and efficient method for bounding system failure probabilities. The method is a progressive one, whereby this bound is reduced as necessary by a more refined analysis. A simple example is used to illustrate the procedure. (author)

[en] The application field of the fault-tree techniques has been explored in order to assess whether the AND-OR structures covered all possible actual binary systems. This resulted in the identification of various situations requiring the complete AND-OR-NOT structures for their analysis. We do not use the term non-coherent for such cases, since the monotonicity or not of a structure function is not a characteristic of a system, but of the particular top event being examined. The report presents different examples of complete fault-trees, which can be examined according to different degrees of approximation. In fact, the exact analysis for the determination of the smallest irredundant bases is very time consuming and actually necessary only in some particular cases (multi-state systems, incidental situations). Therefore, together with the exact procedure, the report shows two different methods of logical analysis that permit the reduction of complete fault-trees to AND-OR structures. Moreover, it discusses the problems concerning the evaluation of the probability distribution of the time to first top event occurrence, once the hypothesis of structure function monotonicity is removed

No abstract available

[en] The essential part of Monte Carlo calculation on neutron radiation effect is nuclear cross section data. However, due to large amount of these data, it becomes a major obstacle to program development on microcomputer which has only limited memory size. One solution is to keep the majority of data in external memory, while the core memory will store only the most frequently needed data. As far as the speed of data access to the external memory is concerned, it can be significantly improved by employing unformatted binary data type together with assembly code for data access. As a result from some test runs of the program, which is written in Fortran language, on 386SX computer with 33 MHz clock employing Ram Drive as the external memory, it is found that the overall speed is quite satisfactory. Specifically, it takes approximately 1.5 hours to simulate 100,000 primary neutrons at 14 MeV employing 40x40 cm and 10 cm thick shielding which is a mixture of lead and polyethylene

No abstract available

[en] Fault trees are a method of deductive analysis and a means of graphic representation of the reliability and security of systems. The principles of the method are set out and the main points illustrated by many examples of electrical systems, fluids, and mechanical systems as well as everyday occurrences. In addition, some advice is given on the use of the method

No abstract available

[en] Fault tree analysis is being widely used for reliability and safety analysis of systems encountered in the nuclear industry and elsewhere. A disadvantage of the fault tree method is the voluminous fault tree graphical representation that conventionally results from analysis of a complex system. Previous methods for shortening the fault tree graphical representation include (1) transfers within the fault tree, and (2) the use of the SAMPLE (K out of N logic) gate, the MATRIX gate, and the SUMMATION gate. The purpose of this presentation is to introduce TABULATION gates as a method to abbreviate the fault tree graphical representation. These new gates reduce the cost of analysis and generally increase the system behavior visibility that is inherent in the fault tree technique

[en] This paper proposes an analysis of a hybrid Bayesian-Importance model for system designers to improve the quality of services related to Active Assisted Living Systems. The proposed model is based on two factors: failure probability measure of different service components and, an expert defined degree of importance that each component holds for the success of the corresponding service. The proposed approach advocates the integration of expert-defined importance factors to enrich the Bayesian Fault Tree Analysis (FTA) approach. The evaluation of the proposed approach is conducted using the Fault Tree Analysis formalism where the undesired state of a system is analyzed using Boolean logic mechanisms to combine a series of lower-level events.

No abstract available