Advanced materials have been identified as a one of the crucial enabling technologies. The vision in material science consist in the integration of materials modelling methodologies and tools with materials informatics, and advanced optimization approaches and tools to explore for new materials and structures. A digital twin technology is rapidly growing field addressing the needs providing a virtual model of a physical object that spans its life-cycle, is updated from real-time data, and uses simulation and machine learning to design optimization and decision support. The behavior of complex materials requires to take into account their heterogeneous, multi-scale nature and strong dependence on processing conditions. The modelling of such complex problems is itself a very challenging task, requiring to combine knowledge from different fields. On top of that, the challenge is the integration of existing models into complex simulation workflows to enable their documented, automated, and repeatable execution. In this contribution, we will present the original design of MuPIF simulation platform, focusing on its design and key features. The focus will be on its Data Management System, which is capable of representing a user-defined, semantically defined digital counterpart of the problem (digital twin model). The integration of digital twin model with simulation platform offers many advantages, including capability to run simulation workflows with inputs and outputs directly linked to entities and their attributes in digital twin model, full traceability of simulations or integration with high performance computing resources. The use will be illustrated on selected applications from organic electronics and composite material domains.