Complexity, safety relevance and long life time make nuclear power plants heavily dependent on information and skilled people. The knowledge accumulated during design, construction and operation is a critical asset that must be preserved for continuous improvement and future modifications. This paper proposes an approach that supports the safety and reliability of nuclear power plants based on cross-disciplinary methods and tools to create, maintain and exploit a structured knowledge basis for long-term operation, maintenance and up-grade. It proposes advanced modelling and justification methods to capture design information that currently often remains incompletely documented. In particular, it aims at making safety properties, dependencies and reasons behind design solutions understandable to all stakeholders. As a step towards its vision of model-based knowledge repositories, it extends traditional system descriptions with assumptions, operational constraints, design rationales and safety justifications. The knowledge base also links design solutions to requirements in regulations and standards. In addition, advanced plant models enable use of computer tools e.g. for automated dependency analysis and simulation. Focusing on design models for plant operation, the approach works in a multidisciplinary fashion and looks at reactor islands as socio-technical systems. Within the context of knowledge management and organizational learning, it integrates process and plant engineering, instrumentation and control system design and human factors engineering. On the basis of the current state-of-the-art and industrial needs, a common framework is defined. The general principles are refined towards more formal modelling languages and software tools. Finally, the framework and its implementations are tested in industrial case studies, and the experiences are used to refine the outputs of the project.