Mistuning identification in a mono-stage bladed disk using an Inverse CMM method

Abstract

Bladed disks (or blisks) used in jet engines' axial compressors are supposed to be cyclically symmetric. But some unavoidable deviations from blade to blade (called mistuning) due to manufacturing tolerances, the presence of defects or operational wear break the cyclic symmetry of blisks. It results in considerable changes in the dynamic behavior of the structures such as frequency splitting, dynamic strain energy localization and amplification of the forced response.

Companies developing cyclically symmetric structures, especially in the aero-space industry, are more and more interested in including mistuning early in development stages, in order to optimize the mass of parts by avoiding oversizing of the structures for design purposes.

This thesis aims to bring the knowledge, tools and procedures to Safran Aero Boosters in the identification of the mistuning in mono-stage blisks using an Inverse Component Mode Mistuning (Inverse CMM or ICMM) previously implemented using the \emph{Python} programming language. Some influence studies are also performed to assess the robustness on an industrial case.

First, the identification of mistuning in this thesis is carried out on noiseless, simulated FRFs of a lowly damped, randomly mistuned numerical model of an industrial blisk made up of 36 sectors. The result of this identification, which is based only on the first blade modes family, is as correlation of 99.95\% between the mistuning pattern identified with the ICMM method and the known mistuning pattern introduced in the numerical model.

Then, influence studies on the amount of measured modes, the amount of tuned-system modes retained for the generation of the Inverse Reduced Order Model (IROM) used for the identification of mistuning, the introduction of simulated noise in the FRFs to get as close as possible to experimental measurements, and the choice of another blade-modes family, prove that the method is really robust when enough modes can be extracted from the modal analysis.