The idea of using hydrogen as a medium for storing excess renewable energy is gathering momentum but to deliver, several technological advances are required. One of the main challenges is boundary layer flame flashback (BLF) as a key operability issue for low NO_x premixed combustion of hydrogen-enriched fuels; thus design of tools to predict flashback propensity is of interest. Extensive research has already been done, and the result is three main BLF models, promise to predict the initiation of boundary layer flame flashback accurately. The first model was developed in the Paul Scherrer Institute (PSI) and was applied to an axial-dump burner operating at elevated pressure and temperature conditions with fuel mixtures of H₂-N₂ and H₂. The second model was developed at the University of California, Irvine (UCI) and was based on the flashback experimental data of 100% H₂ fuel (similar geometry with UCI) at elevated pressure and temperature conditions. Finally, the most recent BLF model was developed at TU Delft based upon the 'generalized' Stratford's turbulent boundary layer separation criterion, which is a further development of the model initially developed in TU Munich (TUM) for atmospheric confined H₂ flames in a horizontal channel burner. In this thesis, the performance of the state-of-the-art BLF models is evaluated in a number of academic burners operating with H₂-rich and pure H₂ fuels at atmospheric and elevated operating conditions. While the TU Delft model showed good agreement with 100% H₂ flashback experiments under atmospheric conditions and simple geometries, further modifications are proposed to perform under conditions relevant for gas turbines (increased pressure and temperature). A new turbulent flame speed correlation to capture the effect of pressure on flashback propensity is applied to the model and validated in high-pressure academic burners. Furthermore, the TU Delft model is validated for a lab-scale size of a gas turbine burner geometry, operating at atmospheric conditions and tested in the TU Delft laboratory. This validation shows that it is essential to review the outcome of the boundary layer flashback model at different locations in the boundary layer, and with minor modifications, the model can capture the test results adequately. The sensitivity of the TU Delft model on different fuel compositions is also investigated. Proper selection of the turbulent flame speed correlation is found to improve the model's performance for fuel mixtures less rich in H₂.