In the current gas turbine market, the traditional design-test-redesign loop is not a viable solution to deploy new products within short timeframes. Hence, to keep the amount of testing to an absolute minimum, theoretical simulation tools like Finite Element Modelling (FEM) have become a driving force in the design of blades to predict the dynamic behaviour of compressor and turbine assemblies in high-speed and unsteady flows. The predictions from these simulation tools need to be supported and validated by measurements. For the past five years, Rolls-Royce Blade Tip Timing (BTT) technology has been replacing rotating Strain Gauge systems to measure the vibration of compressor blades, reducing development times and costs of new aero engine programmes. The overall aim of the present thesis is to progress the BTT technology to be applied to aero engine turbine modules. To this end, the two main objectives of this project are: i. To improve the current validated Rolls-Royce BTT extraction techniques, through the development of novel algorithms for single/multiple asynchronous and responses. ii. To validate the improved extraction using simulated and real engine test data in order to bring the Turbine BTT technology to a Rolls-Royce Technology Readiness Level (TRL) of 4 (i.e. component and/or partial system validation in laboratory environment). The methodology adopted for the development of the novel algorithms is entirely based on matrix algebra and makes extensive use of singular value decomposition as a means for assessing the degree optimisation achieved through various novel manipulations of the input (probe) raw data. The principle contributions of this thesis are threefold: i. The development of new BTT matrix-based models for single/multiple non-integral and integral engine order responses that removed certain pre-processing assumptions required by the current method. ii. The development of BTT technology to operate under the constraint of having equally spaced probes, which is unavoidable in turbines and renders current BTT methods unusable for turbine applications. iii. The development of methods for extracting measurement uncertainty and signal to noise ratios that are based solely on the raw data, without reliance on simulated reference data. Following the verification and validation of the new processing algorithms against simulated data and against validated software with numerous examples of actual engine test data, a Rolls-Royce's Research & Technology (R&T) Critical Capability Acquisition and Capability Readiness (CCAR) review has accredited the novel techniques with a TRL of 4.

No / This paper presents and compares the most important and often used methods to measure turbine blade vibrations: use of strain gauges and telemetry system which is an intrusive method or, on the other site. The Blade Tip Timing (BTT) method known as Non-Intrusive Stress Measurement (System) NSMS. Both methods have advantages and disadvantages which are described below. This paper focused on synchronous vibrations, which are more important in terms of turbine blades fatigue prediction and design optimization.

Dans le cadre de la certification de ses moteurs d'hélicoptères, Safran Helicopter Engines réalise des essais en fonctionnement lors desquels les réponses vibratoires de turbomachines (compresseurs et turbines) sont mesurées. Les réponses vibratoires contiennent des modes (ou raies spectrales) dont les fréquences et amplitudes doivent être caractérisées. Les mesures sont réalisées par la technologie tip-timing qui permet d'observer les vibrations de toutes les pales d'un aubage en rotation.Cependant, la technologie tip-timing présente deux spécificités importantes. Premièrement, l'échantillonnage des signaux de vibrations est irrégulier quasi-périodique. Deuxièmement, l'ordre de grandeur des fréquences de vibration est généralement supérieur à la fréquence d'échantillonnage équivalente. Ces deux caractéristiques donnent lieu à des artefacts des composantes fréquentielles sur les spectres des signaux de vibrations. Ceux-ci gênent alors fortement l'identification du contenu spectral et perturbent donc l'interprétation du comportement vibratoire des pales.La nouvelle méthode d'analyse spectrale proposée s'appuie sur une modélisation parcimonieuse des signaux tip-timing et prend en compte les variations de la fréquence de rotation. L'analyse spectrale des signaux est alors réalisée par la minimisation d'un critère des moindres carrés linéaires régularisé par une pénalisation de "norme-l0" par l'algorithme Block-OMP.À l'aide de résultats numériques sur signaux synthétiques, il est démontré que cette méthode fournit de bonnes performances d'estimations des composantes spectrales et réalise une réduction importante de leurs artefacts. La prise en compte des variations de la fréquence de rotation permet en effet de tirer profit de l'utilisation de longues durées d'observation afin de réduire significativement les artefacts des composantes fréquentielles contenus dans les spectres. Par ailleurs, avec des performances légèrement meilleures à celles de l'ESMV (méthode reconnue pour l'analyse spectrale des signaux tip-timing), la méthode proposée est environ cent fois plus rapide.Deux cas de données réelles sont étudiés. À travers une détection de crique de pale, le premier cas d'étude montre que la méthode proposée est pertinente et réalise des estimations comparables aux méthodes industrielles. Le second cas d'étude présente plusieurs vibrations synchrones et asynchrones simultanées. Cela met en avant la capacité de réduction des artefacts des composantes fréquentielles de la méthode développée afin de faciliter l'interprétation du contenu vibratoire complexe de ce signal.L'optimisation du placement des sondes tip-timing est également étudiée pour faciliter l'identification des composantes synchrones. À partir de résultats numériques, il est démontré qu'éloigner les capteurs améliore l'estimation des amplitudes ce type de composantes. / As part of the certification of its helicopter engines, Safran Helicopter Engines performs operational tests in which the vibrations responses of turbomachines (compressors and turbines) are measured. The vibratory responses contain modes (or spectral lines) whose frequencies and amplitudes must be characterized. The measurements are provided by the tip-timing technology which can observe the vibrations of all the blades while rotating.However, tip-timing technology has two important features. Firstly, the sampling of the vibrating signals is irregular quasi-periodic. Secondly, the vibrating frequencies are generally higher than the equivalent sampling frequency. These two characteristics generate frequency components artefacts onto the vibrating signals spectrum. As a consequence, they strongly hinder the identification of the spectral content and thus disturb the interpretation of the blades vibratory behaviour.The proposed new spectral analysis method relies on sparse modelling of the tip-timing signals and considers the variations of the rotational frequency. The spectral analysis of the signals is then performed by the minimization of a linear least squares criterion regularized by a penalty of "norm-l0" by the Block-OMP algorithm.Using numerical results from synthetic signals, it is shown that this method provides good spectral component estimation performances and achieves a significant reduction of their artefacts. Considering the variations of the rotational frequency allows to take advantage of the use of long observation periods in order to significantly reduce the frequency components artefacts contained in the spectrum. In addition, with slightly better performances than the ESMV (acknowledged method for the tip-timing signals spectral analysis), the proposed method is about a hundred times faster.Two cases of real data are studied. Through a detection of a blade crack, the first studied case shows that the proposed method is relevant and makes equivalent estimates with respect to industrial methods. The second studied case presents several simultaneous synchronous and asynchronous vibrations. That highlights the ability to reduce the frequency components artefacts of the developed method in order to simplify the interpretation of the complex vibratory content of this signal.The optimization of the positioning of the tip-timing probes is also studied in order to simplify the identification of synchronous components. From numerical results, it is demonstrated that moving away the probes improves the amplitudes estimation of this type of components.

This dissertation proposes hybrid models for (i) diagnosis and (ii) remaining useful life estimation of a single fatigue crack in a low-pressure turbine blade. The proposed hybrid methods consist of physics-based methods and data-driven methods. In this dissertation, blade tip timing is used to measure the relative tip displacement of a rotor blade. The natural frequency of the blade is determined by detecting the critical speeds of the blade using a newly derived least squares spectral analysis method. The method shares its origin from the Lomb-Scargle periodogram and can detect resonance frequencies in the blade’s displacement while the rotor is in operation. A Campbell diagram is then used to convert the critical speed into a natural frequency. Two kinds of shaft transients are considered, a run-up run-down crossing the same critical speed, is used to test the new method. This dissertation shows that the relative displacement of the blade tip is comparable to those simulated from an analytical single degree of freedom model. It is also shown that the newly proposed resonance detection method estimates the natural frequency of the blade to a high degree of accuracy when compared to the measurements from a modal impact hammer test. The natural frequency obtained from the real time measurement is then used in a pre-constructed hybrid diagnostics model. The diagnostics model provides a probability density function estimation of the surface crack length given the measured natural frequency. A Gaussian Process Regression model is trained on data collected during experiments and finite element simulations of a fatigue crack in the blade. The final part of this dissertation is a sequential inference model for improving the estimation of the crack length and the prediction of the crack growth. The suggested model uses an unscented Kalman filter that improves estimations of the crack length and the rate of crack growth from Paris’ Law coefficients. The model is updated each time a diagnosis is performed on the blade. The RUL of the blade is then determined from an integration of Paris’s Law given the uncertainty estimates of the current damage in the blade. The result of the algorithm is an estimation of the remaining number of cycles to failure. The algorithm is shown to improve the overall estimation of the RUL; however, it is suggested that future work looks at the convergence rate of the method. / Dissertation (MEng)--University of Pretoria, 2019. / Eskom Power Plant Engineering Institute (EPPEI) / Mechanical and Aeronautical Engineering / MEng / Unrestricted

This dissertation presents an approach to use blade tip timing measurements with finite element analysis to predict the fatigue life of a low pressure steam turbine last stage blade under high backpressure and low flow conditions. Material fatigue properties were determined through the extended universal material law for FV566 material, along with different temper scenarios. A finite element model of a blade with damping pins was developed, using the principle of cyclic symmetry for a perfectly tuned model. Pre-stress modal analysis was conducted, incorporating damping via friction and plasticity for initial 20% overspeed test. The finite element model was verified by two experimental tests: the first being a blade impact test and the second a telemetry strain gauge test in a balance pit. Fatigue life analysis was conducted under the assumption that non-synchronous vibration is experienced by the blade and that only one mode is dominant in the vibration. The results from the fatigue analysis corresponded to the location of the cracks experienced on the blades. The results show twelve orders of magnitude lower life at low load, high backpressure conditions, compared to high load high pressure conditions. The research was further extended to check the same vibratory response on the first three modes, up to their tenth nodal diameters. This was done to analyse fatigue life in a case that a different mode was excited. / Dissertation (MEng)--University of Pretoria, 2020. / Eskom Rotek Industries / Mechanical and Aeronautical Engineering / MEng / Unrestricted

This research developed two innovative test methods that were used to experimentally evaluate the performance of a novel blade tip timing (BTT) system from Prime Photonics, LC. The research focused on creating known blade tip offsets and tip vibrations so that the results from a BTT system can be validated. The topic of validation is important to the BTT field as the results between many commercial systems still are not consistent. While the system that was tested is still in development and final validation is not complete, the blade tip offset and vibration frequency validation results show that this BTT system will be a valuable addition to turbomachinery research and development programs once completed. For the first test method custom rotors were created with specified blade tip offsets. For the blade tip offset alternate measurement, the rotors were optically scanned and analyzed in CAD software with a tip location uncertainty of 0.1 mm. The BTT system agreed with the scanned results to within 0.13 mm. Tests were also conducted to ensure that the BTT system identified and indexed the blades properly. The second developed test method used an instrumented piezoelectric blade to create known dynamic deflections. The active vibration rotor was able to create measureable deflection over a range of frequencies centered on the first bending mode of the blade. The results for the 110 Hz, 150 Hz, 180 Hz first bending resonance, 200 Hz, and 1036 Hz second bending resonance cases are presented. A strain gage and piezoelectric sensor were attached to the active blade during the dynamic deflection tests to provide an alternate method for determining blade vibration frequency. The BTT system correctly identified the active blade excitation frequencies as well as a 120 Hz frequency from the drive motor. This thesis also explored applying BTT methods and testing to more realistic blade geometry and vibration. Blade vibrations are usually classified by their frequency relative to the rotation speed. Synchronous vibrations are integer multiples of the rotational speed and are often excited by struts or vanes fixed to the engine case. For this reason, special probe placement algorithms were explored that use sine curve fitting to optimize the probe placement. Knowing how the blade will vibrate at operation before testing is critical as well. In preparation for future research, ANSYS Mechanical was used to predict the first three modes of a PT6A-28 first stage rotor blade at 1,966, 5,539, and 7,144 Hz. These frequencies were validated to within 4% using scanning laser vibrometry. The simulation was repeated at speed to produce a Campbell Diagram to highlight synchronous excitation crossings. / Master of Science

Methods of determining natural frequencies of the D76D88, B76D88, A86E93, C86G90, C86L90 and C125L89 turbine wheel designs for various environmental conditions were investigated by application of Finite Element Analysis and beam theory. Modelling and simulation methods were developed ; the first method composed of 15 finite element simulations ; the second composed of 15 finite element simulations and a set of experimental frequency survey results; the third composed of 5 simulations , an incorporated mathematical model and a set of experimental frequency survey results. Each of these methods was designed to allow prediction of resonant frequency changes across a range of exhaust gas temperature and shaft rotational speed. For the new modelling and simulation methods, an analysis template and a plotting tool were developed using Microsoft Excel and MATLAB software. A graph showing a frequency-temperature-speed variations and a Campbell Diagram that incorporates material stiffening and softening effects across a range of rotational speeds was designed, and applied to the D76D88, B76D88, A86E93, C86G90, C86L90 and C125L89 turbine wheel designs. New design methodologies for turbine wheels were formulated and validated, showing a good agreement with a range of data points from frequency survey, strain-gauge telemetry and laser tip-timing test results. The results from the new design method were compared with existing single compensation factor methodology, and showed a great improvement in accuracy of prediction of modal vibration. A new nomenclature for the mode shapes of a turbocharger’s blade was proposed, designed and demonstrated to allow direct identification of associated mode shape. It is concluded that Finite Element Analysis combined with the frequency survey is capable of predicting changes in turbine natural frequencies and, when incorporated into the existing turbine design methodology, resulted in a major improvement in the accuracy of the predictions of vibration frequency.

Methods of determining natural frequencies of the D76D88, B76D88, A86E93, C86G90, C86L90 and C125L89 turbine wheel designs for various environmental conditions were investigated by application of Finite Element Analysis and beam theory. Modelling and simulation methods were developed ; the first method composed of 15 finite element simulations ; the second composed of 15 finite element simulations and a set of experimental frequency survey results; the third composed of 5 simulations , an incorporated mathematical model and a set of experimental frequency survey results. Each of these methods was designed to allow prediction of resonant frequency changes across a range of exhaust gas temperature and shaft rotational speed. For the new modelling and simulation methods, an analysis template and a plotting tool were developed using Microsoft Excel and MATLAB software. A graph showing a frequency-temperature-speed variations and a Campbell Diagram that incorporates material stiffening and softening effects across a range of rotational speeds was designed, and applied to the D76D88, B76D88, A86E93, C86G90, C86L90 and C125L89 turbine wheel designs. New design methodologies for turbine wheels were formulated and validated, showing a good agreement with a range of data points from frequency survey, strain-gauge telemetry and laser tip-timing test results. The results from the new design method were compared with existing single compensation factor methodology, and showed a great improvement in accuracy of prediction of modal vibration. A new nomenclature for the mode shapes of a turbocharger’s blade was proposed, designed and demonstrated to allow direct identification of associated mode shape. It is concluded that Finite Element Analysis combined with the frequency survey is capable of predicting changes in turbine natural frequencies and, when incorporated into the existing turbine design methodology, resulted in a major improvement in the accuracy of the predictions of vibration frequency. / Additional data files have been restricted by request.

No description available.

This thesis has been a collaboration between the Royal Institute of Technology (KTH) and Siemens Energy which invest in the research facility at KTH. The objective was to investigate the use of eddy current sensors in real-time monitoring of turbine blades. The main focus has been on finding a use for blade tip clearance and a correlation for the insufficient sampling that eddy current sensors suffer from. At the same time, it was desirable to also investigate the use of the same sampled data for blade tip vibration. The research on eddy current sensors is important for their relative low price compared to other instruments and how resistance it is to contamination found in turbines, enabling real-time monitoring.  The testing has been conducted at the Energy Technology department which utilizes a scaled version of a full-sized turbine to investigate performance measurements. It is scaled to have the same stage loading for both blisks investigated. Two different blisks have been used for this project, one with thicher but fewer blades and one with thinner but more blades. On each blisk different types of sampling have been done in order to capture suitable data for both tip timing and tip vibration. This resulted in sampling with static RPM and sweeps over certain regions as well as full sweeps from design RPM to standstill.  A computer model was developed to evaluate the sampled data. In the model, the sample points were interpolated to compensate for the insufficient sampling, enabling tip gap measurements. Measurements and calibration were done on the blisks for the possibility of using a compensation curve in order to be able to compensate for the signal error. The results show that eddy current sensors and setup used here have a good capability of capturing the tip clearance with precision in the range of hundreds of millimeter on the blisk with thicker blades and up to a certain rotational speed on the blisk with thinner blades. In regards to the tip vibration, eddy current system had problems capturing the time of arrival with sufficient precision correctly. / Den här uppsatsen har varit ett sammarbete mellan Kunglig Tekniska Högskolan (KTH) och Simens Energy vilka investerar i forskningen som bedrivs på KTH. Målet var att undersöka användningen av eddy current sensors för övervakning av turbinbliskar. Huvudfokus har varit att hitta användning av sensorerna för topspelsmätningar och ta fram en korregering av den otillräckliga insamlingen av data som eddy current sensorer lider av. Samtidigt var det önskvärt att samtidigt undersöka samma insamlade data för att utvärdera bladvibrationer. Forskning på eddy current sensorer är viktig för dess relativt låga pris jämfört med andra alternativ samt att de dess höga motståndskraft mot smuts som ofta finns i miljöer där turbiner används. Testerna har gjorts på instutitionen för Energiteknik vilka använder en nedskalad versioner av den verkliga storleken på turbinen för att utföra mätningar på. Två olika bliskar har använts för detta projekt, en med grövre men färre blad samt en med tunnare och fler blad. Stegbelastningen är dock samma för båda. På båda bladen har olika typer mätningar gjorts för att kunna fånga passade data för båda topspelsmätningar och bladvibrationer. Detta gav data med statiskt varvtal, långsamma svepningar över specialla regioner och svepningar över från designvarvtal ner till stillastående. En datormodell har utvecklats för utvärdera insamlade data. I modellen sker en interpolering som kompenserar för de låga antalet samplade punkter på bladet. Mätnigar och kalibreringar är gjorda på bliskarna för att skapa en kompenseringskurva åt signalfel i utdatan. Resultatet visar att eddy current systemet har goda möjligheter att visa rätt toppspel med god precition för blisken med tjocka blad och upp till ett visst varvtal på den med tunnare blad. När det kommer till bladvibrationer hade sensorerna och datormodellen svårt att fånga rätt ankomst tid för bladet med tillräckligt hög precision.