Modular plants are an essential step towards agile production via smart manufacturing in the process industry. An important challenge, however, is that they represent an ecosystem with distributed knowledge between process equipment assembly (PEA) manufacturers and owner/operators (O/Os). This complicates the application of familiar simulation methods. To address this issue, this paper reviews typical simulation tasks in the plant life cycle and investigates the distribution of knowledge in modular plants. While the O/O typically has extensive knowledge about the physical property system of a process, a PEA manufacturer can be expected to develop high equipment related expertise. It is suggested to use a digital twin (DT) to integrate these knowledge parts considering intellectual property concerns. The DT will be provided by the PEA- manufacturer, must be customizable through standardized interfaces, and should be applicable to different simulation tasks in the plant life cycle. To realize this concept, quality requirements specifications for DTs related to these simulation tasks must be developed. In addition, standardized interface specifications for the integration of physical property packages, the communication between different DTs and the simulation of DTs must be matured. Furthermore, hierarchical and horizontal DT structures are required to foster re-use of sub-models for different simulation tasks.