Mistuned Forced Response Of A Bladed Drum

Abstract

In the last decades, blade mistuning in engine structures has become a problematic of importance in industries. Blade mistuning, defined as the blade-to-blade variations of mechanical properties, is all the more dangerous for the engine as the coupling between the blades and the drum and between the stages of the drum increase, which is the case in Bladed druMs (BluM). The need to determine the impact of mistuning on the structure behaviour led to the development of a software, AEROMECA, able to perform mono-stage analysis of engine rotoric and statoric parts. In this context, the work performed for this Master's Thesis consists in validating the software for the very first time against real experimental measurements of a bladed drum rotor stage. With this aim in view, ping-test experiments of the rotor stage obtained for the first bending mode were post-processed using three different methods - peak-picking, Inverse Component Mode Mistuning (ICMM) in SAMCEF and ICMM in ANSYS - to obtain three sets of frequencies. The finite element model of the BluM stage was set up in WORKBENCH. The centrifugal stiffening was computed for the rotation speed of interest in SAMCEF. The modal basis was constructed in SAMCEF. Simulations in AEROMECA were then launched with the three sets of ping-test frequencies and the results were compared to the tip-timing measurements performed on the BluM in rotation. Influence studies were performed in order to determine the entry data to which AEROMECA is sensible. The drum damping as well as the blades individual values of the centrifugal stiffening have no impact on the results while the values of the frequencies and the blades damping have a high influence on the AEROMECA predictions. The results of this study show that to this date, the best correlation achievable is of 54.07\% and is obtained for ping-test frequencies determined with the inverse component mode mistuning method implemented in SAMCEF when the individual blades damping are considered. The AEROMECA predictions overestimate the relative maximum displacement of the critical blade by 20.53\% compared to the tip-timing measurements. Eventually, a random mistuning simulation was performed using the Weibull distribution in order to determine the sensibility of the maximum amplification factor to the error on the ping-test frequencies. This study shows that the maximum amplification factor obtained for an imposed mistuning has to be multiplied by 1.15 in order to certify with 99.9\% confidence that all the blades will have an amplification factor smaller than this value.