The design of high-performance electrical machines requires a tightly coupled consideration of electromagnetic behavior, thermal management, and mechanical integrity. In this work, an integrated simulation framework is employed to investigate the interdependencies between e-magnetic flux distribution, cooling channel layout, and the mechanical response of the rotor and stator assembly. Choosing the optimal cooling concept for traction e-motors requires a holistic view of electromagnetic losses, thermal pathways, design space constraints, and vehicle-level duty cycles. This work presents a rigorous workflow in GT-SUITE to compare candidate cooling concepts—water jacket, stator slot oil, rotor oil-spray/jet, and shaft cooling—using integrated electromagnetic loss prediction and coupled flow-thermal simulation. Speed- and load-dependent copper, iron, and magnet losses are generated from a 2D FE electromagnetic model and mapped into a quasi 3D-equivalent thermal network of the windings, stator yoke/teeth, rotor, bearings, housing, and end-winding regions. Hydraulic models of the coolant circuits (pump, water jackets, oil galleries, jets, oil cooler) are included to capture parasitic pumping power and pressure/flow distribution. Resulting thermal load distributions, together with centrifugal forces and interference fits, are subsequently used in a mechanical stress analysis to assess von Mises stress levels, press-fit behavior, and burst speed margins. The combined approach reveals critical design trade-offs—such as the influence of cooling geometry on magnetic performance and the impact of interference stresses on allowable operating speeds—and supports the optimization of reliable, compact, and efficient machine designs. Key performance indicators include peak winding temperature, ΔT across end-windings, bearing and magnet temperatures, pumping power, mass/packaging, and demagnetization. A Design of Experiment (DoE) is conducted, sweeping across geometrical and electromagnetic parameters associated with the electric motor to understand the impact of the design variables on the performance and efficiency of the electric machine. Additionally, GT-SUITE`s in-built Machine Learning Assistant (MLA) is leveraged to create a metamodel based on the DoE investigation. The metamodel is utilized for easy trade-off investigations, and to improve the e-motor efficiency and performance. The study delivers a repeatable methodology to select and justify the cooling concept early, reducing prototypes and accelerating e-drive development. The approach extends naturally to inverter/gearbox integration and to digital-twin deployment for on-road monitoring.