In this work, the predictive capabilities of Flash Thermography, a non-destructive testing method, in detecting the presence of defects in metallic parts are investigated using the Finite Element Method (FEM). Finite-element simulations of flash thermography on an artifact under inspection are carried out to understand the effect that various thermal and process parameters have on the temperature distribution in the artifact. The 1D analytical models for the finite and semi-infinite plate are studied to get an insight into the heat conduction process. Numerical models for finite width and semi-infinite plate are developed in ABAQUS Explicit, and the results arecompared with the analytical model for validation. The thermal signal processing techniques, "Thermal Signal Reconstruction" (TSR) and "Coefficient Maps" (CM) are reviewed in this work to critique their extent of applicability for different process parameters.TSR’s dependability on the selection of the higher order polynomial and requirement of well-defined flash duration makes it unsuitable for in-situ measurement in an additive manufactured part. CM algorithm is tested on the numerical models having a flash duration longer than the diffusion time to signify the presence of the defect. Several numerical models are developed with different parameters for this study. Lastly, a qualitative comparison highlighting the unique features and the limitations of both the algorithm is presented.