The sensitivity of hybrid RANS-LES methods like Improved Delayed Detached Eddy Simulation (IDDES) to numerical model parameter variations related to generic space launch vehicle aft-body flows is investigated. In particular, the changes resulting from the choice of the time-step size, the turbulence model, the fluid modelling, the circumferential grid resolution, the filter length definition, and the data collection period is considered. The results are also compared to experimental and numerical data taken from the available literature. The sensitivity to the time-step size and the turbulence model is minuscule with respect to the obtained mean flow field, wall pressure distributions, azimuthal modes, and wall pressure frequency spectra. However, circumferential resolution, fluid model, and filter length definition affect the solution to a higher extent. Buffeting spectra are very sensitive to the data collection period.