In this application, a highly scalable and user-friendly simulation framework for phase-field models is to be realised in materials sciences, which enables an extremely fine spatial and temporal resolution, and thus the mapping of all relevant physical effects. This leads to extremely high requirements for the computing power. In oder to achieve a high energy efficiency, future development of simulation software must focus on absolute performance, and thus on the lowest possible consumption of resources at a specified simulation accuracy. In addition to excellent scaling, an increasing part of the computing power must be performed by finely granulated concurrency within the computing nodes. For this purpose, the framework waLBerla is developed at the Chair for System Simulation, Erlangen, which already could demonstrate an excellent performance for an initial prototypical phase-field simulation. There, the recipe for success lies in the specialisation of both the used models and the numerical algorithms, as well as the data structures, and in the systematic optimisation for the respective platform, typically a manycore architecture. Thereby, the project also focuses on portability, flexibility and productivity, by using a structured and sustainable development process.