Comparative Analysis of CFD and PIV Base Buffeting Data Considering Decomposition Techniques and Filtering Effects in the Measurement Chain

Abstract

Significant dynamical loads on the nozzle structure of launch vehicles during ascent can cause deformations and eventually complete failure of rocket engines. This effect, known as base buffeting, has been observed for the Ariane V vehicle and is caused by a strong flow seperation at the base of the main body and subsequent reattachement on the nozzle. In order to investigate this phenomenon and to assess mitigation strategies, wind tunnel experiments and numerical simulations of a subscale wind tunnel model at transonic flight conditions have been conducted. This work compares an experimental and numerical dataset of a launch vehicle tail flow field in terms of Proper Orthogonal Decomposition (POD) and Dynamic Mode Decomposition (DMD). Both methods allow to isolate dominant flow features and, in case of DMD, to determine an oscillation frequency associated with the mode. In order to account for the different spatial and temporal resolutions of the datasets, the numerical simulation data is subsampled and spatially filtered to match the experimental Particle Image Velocimetry (PIV) dataset as closely as possible. This approach also allows to assess the effect of spatio-temporal filtering on the CFD mode structure. Additionally, POD and DMD modes of the numerical surface pressure distribution are compared to high-resolution experimental surface pressure measurements.