A digital twin for alloy 718 aircraft parts with numerical estimation capabilities in terms of integrated computational materials engineering (ICME) of local material properties is an essential tool for lightweight design, geometry optimization and for a significant reduction of development and experimental characterization costs. This leads to a demand for numerical models to capture initial microstructural inhomogeneities, describe the thermo-mechanical processing and reflect the local microstructure and mechanical properties of the final product. Therefore, a digital twin for multi-step forging process was developed in order to reproduce and evaluate the resulting local microstructure through the whole processing chain. Since the microstructure determines the mechanical properties like yield stress and fracture toughness, a multi-class grain size model was implemented to describe the local evolution of the relevant microstructural constituents like grain size distribution, volume fraction of delta phase (�) and strengthening phases like gamma prime (��) and gamma double prime (���). A fundamental basis for modeling microstructure and calculation of mechanical properties comprises the knowledge of the initial microstructural state of the pre-material, the complete thermo-mechanical history of the component and the final heat treatment. The initial microstructure varies locally throughout the volume of the billet and has a major influence on the final structure. Therefore, pre-material from different manufacturers and production routes (re-melting types) as well as different billet sizes was analyzed and digitized in this work. The characterization of several feed stock samples showed a pronounced variation of the microstructure due to different production routes involving processes like vacuum arc re-melting, homogenization, upsetting and cogging. The digital part of the twin is a finite element model of the forging process, which takes the inhomogeneities of the billet material into account and provides quantities like local temperature, strain and strain rate. Based on the initial microstructural data and the thermomechanical history, a newly developed multi-class grain size model, parametrized by a series of specially designed experiments, reconstructs the complete microstructure evolution during the different forming processes. In this simulation recrystallization, grain growth and precipitation processes are taken into account. A special focus is set on the recrystallization behavior, which can be subdivided into dynamic, meta-dynamic and static recrystallization. The locally different occurrence and interaction of the previously mentioned phenomena significantly impacts the resulting microstructure in terms of the grain size distribution. As the fracture toughness of alloy 718 results from the specific combination of microstructural constituents such as the grain size distribution, size and amount of precipitations and �-phase fraction, the local fracture toughness can then be estimated in dependence of load direction and assumed crack orientation. For validation of the digital twin and the fracture toughness model, a series of tensile and fracture tests was performed. These tests were carried out not only on the final part, but also on the billet material in order to determine the initial property values and inhomogeneities at different positions. The investigated positions are then tracked through the finite element (FE) simulation and local thermomechanical data are used as input for the multi-class grain size model, which finally yields the resulting microstructure at the tracked material points. This modelling approach establishes the link between the initial microstructure of the billet, the forging process and the final distribution of the fracture toughness and mechanical properties in the component.