The main objective of this thesis is the numerical simulation of hybrid-forming processes in steel production with particular focus on phase transformation. In order to display the specific processes two methods of material modeling, a macroscopic-phenomenological and micromechanical multiscale approach are formulated. The thermodynamically consistent phenomenological multiphase model combines a variety of features such as time- and temperature-dependent phase transformation, austenitisation, transformation plasticity, volume change, temperature- and microstucture-dependent elastoplasticity and viscoplasticity. The FEM simulation of the hybrid-forming process is based on numerical implementation and exhibits good agreement with the structure distribution in the real shaft. Furthermore, it illustrates the possibilities for prediction of the phase distribution by varying the process parameters. A physically motivated, thermodynamic-consistent multiscale model for N-grains and n-bainite variants is developed in the second step, which combines the elasto-viscoplastic behavior with a phase transformation in a polycrystalline structure. This model is capable of capturing both TRIP effects, the contribution due to load-based orientation of bainite-variants (”Magee effect”) and plastic accommodation of the new phase (”Greenwood-Johnson effect”). Finally, these phenomena are evaluated quantitatively for different loads.