Many different technical areas are involved in the design process of large gearless drives for the mining industry, aiming at correctly describing the electrical-mechanical-thermal behavior of the drive. So far, these various technical areas are being treated more or less separately, and no descriptions or references are found concerning the modeling of these large drives using coupled multi-physics models, which allow an overall optimization of this kind of machinery. In this framework, the goal of this thesis is to create reliable and coherent interdisciplinary mathematical models based on a multi-physics approach. The originality of the thesis is to be found in the full interlinked model of a gearless drive. The use of ”Evolution Strategies” in the optimization is also an original contribution to the field of gearless drive design. The proposed mathematical multi-physics model incorporates the key technical areas of a gearless drive. These technical areas are electromagnetism, structural dynamics, heat and mass transfer. Such technical areas are closely linked to each other, with results from one area affecting the others and vice-versa. The multi-physics model of the drive is connected to a meta-heuristic optimization procedure based on ”Evolution Strategies”. The electromagnetic, thermal and structural behaviors have been modeled using the Finite Element Method in 2D and 3D. The mass transport has been described by means of a discrete model and solved using the Newton-Raphson method. One of the major challenges has been to simplify the different sub-models to minimize calculation time without losing accuracy of the final results. This has allowed the models to be utilized in an iterative optimization process. It is shown that the proposed multi-physics model leads to different results than the decoupled models previously used, as the decoupled models use constant values from the several technical areas even though these values are interdependent. The multi-physics model therefore leads to a more precise determination of the design parameters. The thesis gives a clear overview of the necessity of the coupled models and highlights the vital design parameters. As already mentioned, the main contribution to the modeling of gearless mill drives is to be found in the full integration of the various technical areas, enabling a more accurate determination of the different physical parameters that characterize mill drive behavior. This enables an overall optimization that would not have been possible by other means. In this work, the optimization is based on the minimization of the mass of the drive components and losses in the drive, leading to a minimization of purchase and operating costs. The optimization resulted in a mass reduction of 4.0% and a decrease of losses of 9.9% compared to the original drive design. The thesis also opens new research fronts and highlights three new necessary research aspects for further development of the design processes of large gearless drives based on a multi-physics approach: a) experimental tests on the physical mill drives for verification and adjustment of the presented models are crucial, since the multi-physics models have only been compared with other mathematical models; b) 3D simulation of the thermal part in order to investigate the effect of the axial heat flux; and finally c) investigation of the effect of the end-windings and the cooling-air channels in a frame with a detailed 3D CFD model.