Modelling and Analysis of Reliability and Availability by stochastic modelling techniques considering continuously the statistical quality of the initial data

• Applicant: Professor Dr.-Ing. Bernd Bertsche
• Subject Area: Engineering Design, Machine Elements, Product Development
• Term: from 2018 to 2022
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 405325157

Today, modeling and simulation of reliability and availability of technical systems are carried out with efficient methods. The methods are not restricted to specific systems and enable the modelling of different aspects. Nevertheless, the calculation of reliability and availability is based on mean value distributions. The statistical quality of the data basis of the calculations, i.e. the confidence levels of the distribution functions can neither be integrated into the input data of the models nor be considered within the analysis yet. Thus, the results earned by the analysis are mean values, too.Usually, limited information to describe the systems’ behaviour are given, e.g. limited sample sizes of failure times of the system. Hence, the continuously consideration of the statistical quality of the input data is necessary, i.e. the confidence level of the input data need to be considered within the analysis of reliability and availability.Therefore, objective of the research project is a continuously integration of the statistical quality within the holistic methodology, i.e. the continuously consideration within the modelling as well as within the analysis of technical systems. The development of such a universal calculation and simulation method represents a practically relevant research goal and has a serious scientific aim. The method combines the performance of the modelling methods with the expressiveness of confidence levels. It has a high potential to be used in varied applications, e.g. the planning and allocation of reliability, the test planning and the demonstration of general reliability and availability values as well as the area of modelling-based prognostics as given by the Prognostic Health Management (PHM). The uncertainty of actual reliability and availability prediction and demonstration procedures can be reduced significantly. A new quality of reliability and availability analysis can be achieved.At the end of the research project, a new methodology is given as a new Petri net class. Prototypical software engineering works with MATLAB enable a first application of the new method. Furthermore, they are the basis of new research. The added value of the new methodology for a system manufacturer are clarified by a few applications.

• DFG Programme: Research Grants
• Co-Investigator: Professor Dr.-Ing. Peter Zeiler