Synchronous vibrations, which are caused by periodic excitations, can have a severe impact on the service life of impellers. Blade Tip Timing (BTT) is a promising technique for monitoring synchronous vibrations due to its non-intrusive nature and ability to monitor all blades at once. BTT generally employs a Once-per-Revolution (OPR) sensor that is mounted on the shaft for blade identification and deflection calculation. Nevertheless, OPR sensors can be unreliable, as they may be affected by shaft vibrations, and their implementation can be restricted by space constraints. Moreover, the low number of BTT sensors typically leads to under-sampled deflection signals, which consequently hinders the estimation of the vibration frequencies due to aliasing problems. For this reason, BTT is commonly accompanied by strain gauge (SG) measurements on some blades. In this paper, improved BTT techniques are presented, which enable the determination of vibration properties of synchronous vibrations without the need for an OPR sensor and ensure a reliable frequency assessment. Specifically, the blades are identified by unique characteristics resulting from manufacturing tolerances, while the blade deflections are calculated through a novel method, which relies on the impeller’s circumferential position. The proposed method enables accurate OPR-free calculation of blade deflections, by accounting for speed variations within a revolution and considering the actual blade positions on the impeller. By completely eliminating the need for an OPR sensor, the accuracy of BTT is enhanced, as the blade deflections are no longer affected by shaft vibrations, while speed variations within a revolution can be accounted for. Moreover, the implementation possibilities of BTT are improved, allowing its application in systems, where an OPR sensor cannot be instrumented due to space constraints. Subsequently, the vibration frequencies are accurately estimated, by employing an improved Multi-Sampling method based on Non-Uniform Fast Fourier Transform. This approach enables the blind analysis of BTT measurements and can identify multiple vibration frequencies. The proposed method expands the capabilities of BTT through a reliable assessment of vibration frequencies from under-sampled BTT signals. Therefore, it is no longer necessary to accompany BTT measurements with SG measurements for frequency identification. Finally, the vibration properties are determined using regression models. The proposed BTT techniques are validated through comparison with SG measurements as well as a commercial BTT system, using experimental data from a test bench of a turbocharger used for marine applications. The vibrations were recorded under real operating conditions, thus demonstrating the industrial applicability of the proposed BTT evaluation procedure.
Read full abstract