Methodology for automated Petri Net model generation to support Reliability Modelling

As the complexity of engineering systems and processes increases, determining their optimal performance also becomes increasingly complex. There are various reliability methods available to model performance but generating the models can become a significant task that is cumbersome, error-prone and tedious. Hence, over the years, work has been undertaken into automatically generating reliability models in order to detect the most critical components and design errors at an early stage, supporting alternative designs. Earlier work lacks full automation resulting in semi-automated methods since they require user intervention to import system information to the algorithm, focus on specific domains and cannot accurately model systems or processes with control loops and dynamic features. This thesis develops a novel method that can generate reliability models for complex systems and processes, based on Petri Net models. The process has been fully automated with software developed that extracts the information required for the model from a topology diagram that describes the system or process considered and generates the corresponding mathematical and graphical representations of the Petri Net model. Such topology diagrams are used in industrial sectors, ranging from aerospace and automotive engineering to finance, defence, government, entertainment and telecommunications. Complex real-life scenarios are studied to demonstrate the application of the proposed method, followed by the verification, validation and simulation of the developed Petri Net models. Thus, the proposed method is seen to be a powerful tool to automatically obtain the PN modelling formalism from a topology diagram, commonly used in industry, by: - Handling and efficiently modelling systems and processes with a large number of components and activities respectively, dependent events and control loops. - Providing generic domain applicability. - Providing software independence by generating models readily understandable by the user without requiring further manipulation by any industrial software. Finally, the method documented in this thesis enables engineers to conduct reliability and performance analysis in a timely manner that ensures the results feed into the design process.