A study of tip-timing measurement techniques for the determination of bladed-disk vibration characteristics
All turbomachinery blades experience vibration under operating conditions. Given that theoretical tools are often limited in producing reliable quantitative predictions, most engine development certification programs still rely on contacting vibration measurements. The main objective of this research work is the development of robust and validated real-time data analysis procedures for vibration signals captured on rotating assemblies using non-contacting optical probes. This process is referred to as blade tip-timing and was broken down into two distinct steps, first, vibration parameter derivation from blade passing times at the optical probes and second, processing of this parameter to characterise the blade vibration, the second step forming the main part of the work reported. It was decided to focus on synchronous assembly response, though a chapter has been dedicated to asynchronous response. There are established procedures for the characterisation of the latter whilst the accurate identification of the former is a long-standing problem. A multi-degree-of-freedom numerical simulator, which includes the bladed-disk assembly properties, the external forcing terms and the characteristics of the optical probe, was developed to assess the reliability of the various data processing techniques to identify the vibration characteristics of bladed-disk assemblies. The inherent limitations of the existing data analysis techniques were discussed and new approaches were suggested. A method for identifying the resonance frequency using two probes was introduced. The proposed method was shown to be easy to use and its robustness was discussed for a range of cases, including relatively high resonance amplitudes, the presence of close modes and blade mistuning. Further direct and indirect analysis methods were developed by considering up to four probes. The real-time implementation of the four-probe algorithm was performed using a transputer network and it was concluded that dual processor TRAMs are required for the benefits of parallel processing architectures and fast floating point calculations to be achieved. The work reported in this thesis has led to development of three new robust synchronous response analysis techniques and it has been shown that blade vibration characteristics can be satisfactorily identified using a minimum of four measurements at the same axial position.