Vibration signal model of an aero-engine rotor-casing system with a transfer path effect and rubbing

Highlights

• Multi-paths modulate vibration transfer signals of aero-engine rotor-casing system.

• Combination of bearing varying compliance frequencies and rotor frequency features.

• Rubbing brings a new vibration transfer path and reduces amplitude of vibrations.

Abstract

This paper addresses modeling of vibration signals produced by rotor-localized excitation and modulated by multi-paths in an aero-engine rotor-casing system. The aim is to provide a better understanding of specific features of measurement point signals at the casing. Multi-paths effects are incorporated, including rubbing and rotor-casing paths through backup and inter-shaft bearings. The vibration waveforms and spectral signatures of signals with and without rubbing are derived and simulated. The most critical features of the resulting vibration signal are a prominent double rotating frequency and smaller overall amplitudes of frequencies due to rubbing, and bearing varying compliance (VC) frequencies due to the rotor-casing paths. A novel pneumatic-driven dual-rotor-casing test bench with a pneumatic rubbing structure is designed to initiate rubbing and acquire vibration signals at the casing. The test results provide sound verification for the proposed signal models. The analysis may prove useful for simulation purposes and rubbing fault diagnosis.

Introduction

The signals collected at casing transducers in the rotor-casing system of aero-engines are complex. The vibration signal produced by rotor-localized excitation can be modulated by multi-paths and its spectral structures can be changed. Analysis of how the vibration signal changes from the rotor to the casing is necessary, especially when faults are present.

Due to demand for improved thrust-to-weight ratio and structural efficiency, aero-engines have frequent faults, including unbalances, rubbing, and bearing faults. Diverse vibration characteristics are used to study the fault mechanisms and diagnose methods of failure. Rotor-stator rubbing usually occurs, leading to waveform cutting and chaotic motion. These rub effects were obtained in theoretical analysis and verified by experiments [1], [2], [3]. For the bearing faults, the vibration waveform is a periodic impulse signal with a bearing passing frequency as a carrier wave [4], [5], [6], [7]. The existing fault diagnosis methods include wavelet analysis and envelope detection [8], ensemble empirical mode decomposition [9], [10], atomic decomposition, and sparse representation [11].

Previous studies on vibration mechanisms and fault diagnosis mainly placed emphasis on the vibration response from the rotor or bearing bracket rather than the casing. However, in most cases, the measurement of vibration signals in aero-engines can only be conducted on the casing. Thus, the spectral characteristics of the casing signals still deserve further investigation. A detailed spectral study of the casing signals is not only helpful for detecting faults but also useful for locating these faults in the aero-engine rotor systems.

The spectra of vibration signals on the casing have a complicated structure. A dual-rotor-bearing-casing system contains a high-pressure rotor, a low-pressure rotor, backup bearings, inter-shaft bearings, a coupling, a cone gear transmission and multiple junctions. Due to these structures, vibration signals from the rotor can experience several paths and multiple modulations before finally reaching the casing. Once faults exist in the rotor or the paths, the vibration signal may be more complicated.

The vibration response on the casing and the relationship between the vibrations of the rotor and the casing both need more study. The spectral structure of the signal captured on the casing has been addressed by existing literature for aero-engines, gears, and reciprocating engines. Chen [12] analyzed casing vibration acceleration signals to extract characteristics of the rubbing faults and found that the impact frequency was the frequency of blades passing the casing. Feng and Zuo [13] built up signal models of gear damage, incorporating amplitude and frequency modulations. Lei et al. [14] and Liu et al. [15] analyzed vibration signals considering the effect of vibration transfer paths in epicyclic gears. Geng et al. [16] developed an analytical vibration model of a non-stationary reciprocating engine taking into account time-varying transfer properties.

In view of the structure of the dual-rotor-casing system in aero-engines, casing vibrations can be much more complicated than the vibrations of gears or reciprocating machinery. However, a search through existing literature implies that few reports present explicit equations to describe frequency components and amplitudes of vibration signals at different casing measurement points of aero-engine dual-rotor-casing systems. The lack of detailed knowledge of the spectral structure is a bottleneck issue that hinders effective fault diagnosis.

To fix this issue, we propose an analysis approach of vibration transfer in the dual-rotor-casing system of aero-engines. Several signal models of the main paths and the resultant signal at the casing are established by analyzing amplitude modulation and frequency modulation (AMFM) caused by the structure paths and the rubbing path. A detailed spectral analysis on the characteristics of a casing signal with rubbing on the rotor is obtained and verified by experimenting on a novel pneumatic-driven dual-rotor-casing test bench. The resulting vibration signal has features of a prominent double rotating frequency, smaller overall amplitudes of frequencies and a bearing varying compliance (VC) frequency. This work addresses the key issue from its root and provides a theoretical guide for rubbing diagnosis of aero-engines via spectral analysis.

This paper is organized as follows: we develop a general vibration signal model and address spectral structures in Section 2. We present main paths of signal transfer from the rotor to the casing. We develop the casing transducer signal models, via primary paths as well as via an instant rubbing path. By applying self-adapting data fusion technology, we acquire the resultant signal at casing transducers. In Section 3, we carry out simulation of the acquired signals to verify the modeling. We analyze both time- and frequency-domain characteristics under unbalance and rubbing and calculate the resultant signal at a middle measurement point. In Section 4, we develop and fabricate a new pneumatic-driven dual-rotor-casing test bench to validate the transducer signal model. We analyze the resultant signal characteristics corresponding to the signal models in 2 Vibration signal model, 3 Analysis of simulated vibration signal models. Conclusions are drawn in Section 5.