Process-induced fiber orientation has a significant influence on the mechanical performance of injection-molded short fiber-reinforced thermoplastic parts. Simulative estimation of fiber orientation is then a key factor for a successful virtual design of injection-molded parts. However, this numerical task is still challenging because the available models have strong limitations in the concentrated regime, which is the one of industrial interest. In his daily work, a simulation engineer faces several difficult situations as the choice of a fiber orientation model and the determination of the respective model parameters. We propose a multi-scale simulation approach that circumvents the mentioned issues, and at the same time guarantees highly accurate fiber orientation estimations. The method consists of an offline micro-scale simulation (virtual shear test using a discrete element method) and an online Autodesk® Moldflow simulation of the injection molding process, which solves the fiber orientation problem with a single fiber orientation model implemented via Solver API. The goal of the offline step is to identify the material-dependent parameters of the macroscopic fiber orientation model for the injection molding simulation. We discuss the proposed workflow in front of simulations of fiber orientation for several industrial parts and with different materials.