An important source of flow-induced noise comes from the pressure fluctuations beneath a turbulent boundary layer as they interact with a sharp trailing edge. Prediction of this noise using acoustic analogy methods requires knowledge of these pressure fluctuations, which may be obtained using large eddy simulation (LES). However, the computational requirements for wall-resolved LES (WRLES) are prohibitively expensive when applied to boundary layer flows, especially at high Reynolds numbers. Therefore, wall-modelling methods have been investigated to reduce the cost. The first approach (referred to as WMLES) employed a method where the mesh size is reduced using a wall function, and the LES equations are solved throughout the computational domain. The second approach employed the Improved Delayed Detached Eddy Simulation (IDDES) method, where LES is applied only in regions away from walls, while the RANS equations are solved near the walls. The wall function was also introduced for this method. Periodic channel flow was simulated as an initial test case. WRLES, WMLES and IDDES methods were evaluated for predictions of the mean velocity, fluctuating velocity and fluctuating pressure. The power spectral density (PSD) of wall-pressure fluctuations was also calculated. While good accuracy was shown by WRLES, the WMLES method resulted in significant overprediction of the PSD. However, applying the IDDES method, the results were substantially improved, with the final mesh consisting of just 3.9 million cells, compared to 98 million for WRLES. Flow over a NACA0012 aerofoil was also simulated, comparing the WMLES and IDDES methods. The predicted wall-pressure PSD was significantly improved at higher frequencies using IDDES.