The safety assessment of Safety Critical Systems (SCSs) is a challenging task since it involves different actors and a combination of several knowledge domains. This increases the complexity of the integration of safety requirements into the design model. Consequently, there is a need for a shared model with an unambiguous terminology aiming to avoid misunderstandings between both safety and design teams. In this paper, we propose a model-based system engineering approach in order to support the goal-oriented safety reasoning and to provide a common model between both safety and requirement engineering driven by goals. Furthermore, the present study considers the safety rules development process based on the Organization-based Access Control (Or-BAC) model, which is normally used to improve the security of the information systems. Then, the common vocabulary proposed for the interpretation of the considered notions of domains is defined. Moreover, safety requirements are expressed with a high level of abstraction according to the required railway knowledge and the requirement traceability process is considered through an up-bottom reasoning using the Unified Modeling Language (UML) diagrams. The proposed approach aims to provide a methodology able to identify safety conditions in order to anticipate risks and to make better safety-related decisions. Finally, the proposed methodology is evaluated through a real accident scenario analysis in order to validate its adaptability to represent real critical situations.