Rolling bearings are a leading cause of equipment breakdowns in electrical machines, underscoring the significance of predictive maintenance strategies. However, the given methods require high-quality big data, which is challenging to acquire, especially for faulty cases. Simulation models offer an alternative by generating large data sets to complement experimental data. However, bearings involve complex contact-related phenomena, such as slipping and clearance. Therefore, generating realistic data comparable to the real-world necessitates accuracy. Our study presents a multibody simulation system of a motor bearing, incorporating a geometry-based polygonal contact method (PCM), which accurately captures nonlinear bearing dynamics and allows for the simulation of various contact geometries. We introduce a systematic approach to adjust the PCM contact properties for rolling bearings, referencing the well-established Hertzian theory. Both healthy and faulty bearings with a local outer ring fault were simulated. The simulated output was a relative shaft displacement, experimentally validated using a capacitive sensor. Our model successfully demonstrates the potential to employ geometry-based contacts for generating realistic data on faulty bearings with the aim of predictive maintenance.