Heyns, P.S. (Philippus Stephanus)2015-01-192015-01-192014/12/122014Booysen, C 2014, Fatigue life prediction of steam turbine blades during start - up operation using probabilistic concepts, MEng Dissertation, University of Pretoria, Pretoria, viewed yymmdd <http://hdl.handle.net/2263/43250>M14/9/415http://hdl.handle.net/2263/43250Dissertation (MEng)--University of Pretoria, 2014.Fatigue in low pressure (LP) turbine blades has been recognised to be one of the primary causes of steam turbine blade failures worldwide. As a result various methods for predicting the fatigue life of the blades have been proposed. Application of these methods has traditionally been performed using deterministic models which often require overly conservative assumptions. However, given the range of uncertainty in key variables such as material properties, loading and damping; the question is then raised about the subjectivity in the selection of these parameters. An alternative approach is to incorporate probabilistic modelling which can eliminate overly conservative assumptions and allow for uncertainty in key variables to be accounted for. This necessitates the need for development of a probabilistic model which can be used in fatigue life calculations. This dissertation presents a sequential approach used in fatigue life prediction of a LP steam turbine blade during resonance conditions encountered during a turbine start-up by incorporating probabilistic principles. The approach makes use of a probabilistic fatigue life model to account for uncertainty in material fatigue strength parameters and stress cycle history arising from transient stress operation due to variability in blade damping. The fatigue life consumption during a turbine start-up is performed using two well-known theories applied in discrete life and probabilistic models. Mechanical and material fatigue properties required for finite element and life prediction models are determined through experimental testing of scrapped LP blade material X22CrMoV12-1. Load controlled uniaxial fatigue testing data required statistical modelling using regression analysis to interpret the - diagram from which the fatigue properties could be determined. A high coefficient of determination indicated that the regression line fitted the data very well. A finite element model of a free-standing LP blade was developed using the principle of sub-structuring which enabled the vibration characteristics and stress response of the blade to be determined. Steady-state and dynamic transient simulations were required for this. Calculated blade natural frequencies compared favourably with measured data from which a Campbell diagram could be developed for resonance identification. A pressure excitation is formulated through approximations from the static steam forces and used in obtaining the transient stress response during the passing of a critical speed of assumed resonance. The stress response is significantly influenced by variations in damping which are also shown to limit the peak dynamic stress. Fatigue damage calculations necessitated the need to perform rainflow cycle counting on the transient stress histories to obtain a set of simple stress reversals for ease of use. Random curve fitting routines performed on the data ensured the selection of the random variables used in fatigue life calculations is stochastic in nature. The random vectors are selected from a multivariate normal distribution. It was shown that the use of confidence intervals in the probabilistic fatigue life model worked effectively in being able to account for uncertainty in the material fatigue strength parameters and varying stress in the blade root. The probabilistic model was successful in being able to predict the fatigue life of the blade as results are shown to reinforce those of the discrete life model. This research proves that probabilistic fatigue life modelling can be successfully implemented to determine the life of a steam turbine blade during start-up conditions.en© 2014 University of Pretoria. All rights reserved. The copyright in this work vests in the University of Pretoria. No part of this work may be reproduced or transmitted in any form or by any means, without the prior written permission of the University of Pretoria.Multivariate normal distributionFree-standing bladeSteam turbineHigh cycle fatigueFinite element analysisUCTDDissertation12378012