Doubly-Fed Induction Generators (DFIG) are a key component of grid-connected wind energy conversion systems. Therefore, co-simulation studies of the generator and its connected power system are of great importance for reliable, economical and stable operation. However, co-simulation studies have been always a challenge. In order to reduce the simulation time to a reasonable value, lumped-parameter electric ma- chines models are commonly used in electric power system modeling software packages to avoid the heavy computational burden of more accurate modeling methods, at the expense of less accuracy. The Finite Element Method (FEM) is well-known for high accuracy in modeling the electromagnetic behavior of electric generators including topology complexities and performance nonlinearities. However, this comes with the cost of long simulation times and heavy computational requirements which restrict the utilization of FEM to the design and offline performance assessment of electric machines. In this thesis, a new method is proposed to reduce the simulation time required for co-simulating the DFIG using FEM with an external power system. The method merges the Dynamic Phasor Modeling technique with FEM by representing the state variables of the electromagnetic machine’s model and the power system’s electric voltage equations by time-varying dynamic phasors. The proposed method solves the machine and system equations simultaneously in a time-frequency environment capable of representing the main frequency components of the system along with the time-variation of the system inputs. The mathematical formulation of the proposed method is presented in detail along with reviewing the traditional time-domain FEM equations to be used as a basis for testing the performance of the new method. A custom-written C++ code has been developed to simulate a grid-connected DFIG using the proposed method. For fair comparison, the traditional time-domain FEM has also been adopted in a custom-written code whose validity is proven by comparing its results to an open-source FEM software package. In order to verify the accuracy and the computational performance of the proposed method as a tool for co-simulation studies of DFIG and the grid, three simulation case-studies are presented. Firstly, the proposed Dynamic Phasor FEM method is used to model the start-up performance of grid connected DFIG which is usually a complicated task due to the severe starting transients. The philosophy of dynamic phasor modeling implies solving the FEM equations in frequency-domain at each time-window. This enables the utilization of both the virtual blocked rotor method and the traditional physical mesh movement for modeling rotor speed. A proposed technique to merge the two methods is used to model the start-up performance of the grid connected DFIG. Because the proposed method models the machine in the time-frequency domain, the internal harmonics caused by the winding topology are not reflected in the line current making it less effective for power quality assessment studies. Therefore, in the second simulation study, a proposed procedure is presented to mitigate this problem by calculating the winding space harmonics within the proposed Dynamic Phasor FEM method. Thirdly, the convenience of the proposed method is further investigated by using it to model the machine performance under system disturbances (sudden voltage dip and grid-voltage harmonic pollution). For each of the intended simulation studies, the DFIG performance is modeled using both the proposed Dynamic Phasor FEM method and the traditional time-domain FEM method as a comparison base. The proposed method succeeded in producing comparable results to the traditional method with significantly reduced simulation times. This superior computational performance of the proposed Dynamic Phasor FEM method makes it a good candidate for the electric machine grid-connection co-simulation studies.