Scientific validation of a new system reliability analysis code by means of an existing software

In the frame of the BRITE-EURAM project BRPR-CT96-0211, financed by the European Commission, a new system reliability analysis code, based on Monte-carlo techniques, was created by Siveco, a French software company partner in the project, specialised in the development of CMMS (Computer Maintenance Management Systems). This code, called RELSYAN, is part of a more complex software tool and it allows the direct simulation of the effects of different maintenance policies on existing engineering systems, taking into consideration their availability and costs. The RELSYAN module was tested by the software producer using elementary analytical tests, but, in consideration of its importance in the economy of the main software tool, a more convincing and complete validation of the code (also considering complex cases, such as failure rates varying in the time, pre-failure conditions, protections, stand-by components and induced failures etc.), resulted necessary. To this aim, another independent Montecarlo software, SPAR (Malchi Science), tested by a long previous experience, was used. In order to allow the execution of such tests, modifications into the user interface and into the logic of SPAR had to be introduced so that a maintenance simulation in a way as closer as possible to RELSYAN was performed by it.The comparisons were carried out oil different simple cases, obtaining similar results. It means that, even if the different approach, the new RELSYAN code may be profitably applied for the maintenance simulation and improvement. Obviously the obtainment of different results from the two codes would have meant that software or modelling problems existed in at least one of them.The paper includes some details on the algorithms used in the two codes for the simulation of maintenance and it is concluded by the results obtained in one of the comparisons carried-out.