In recent years, learning factories have proven to be an effective instrument for developing competencies, especially in lean production and digitization. The concept of learning factories has been enriched in the recent past by elements and training units in virtual reality (VR). This enrichment allows an expansion of the mapping abilities of different training environments and value streams in the context of lean education. Nevertheless, learning factory developers are faced with the challenge of selecting suitable scenarios in terms of content and scope. An approach for the competency-oriented and structured design of such scenarios will be presented in this publication and illustrated by means of an application example of the research project PortaL (Virtual action tasks for personalized, adaptive learning).