| Popis: |
The work reported in this thesis was carried out in the School of Engineering Science, University of Warwick, between October 1967 and April 1971. It establishes that by using high sensitivity eddy current crack detection, fatigue damage accumulation for both random and constant amplitude loading can be represented as occurring in two stages:- Stage A:- Microcrack Initiation and Propagation. Stage B:- Macrocrack Propagation. The results of specimen sectioning and optical microscope work suggest that there is some correlation between the observed stage behaviour and that proposed by Forsyth (17). An empirical relationship is restated for constant amplitude loading whereby the proportion of life spent in either stage can be estimated from the elastic stress concentration factor. This relationship is based on the behaviour observed from a wide range of specimen configurations. It is established under random loading that for a given specimen configuration the proportion of time spent in Stage A behaviour for a given fatigue life remains unaltered for changes in waveform irregularity factor. This statement applies to two series of tests for loading waveforms of unaltered fundamental p.s.d. shapes and Gaussian amplitude probability density distributions. It is suggested that the signal maximum peak/rms. ratio is a significant factor in determining the proportion of life spent in crack initiation. On-line computer techniques are described which enable data acquisition for the accurate definition of the statistical properties of random loading waveforms. The computer programs which perform off-line analysis of data and provide power spectral density, peak probability density and amplitude probability density descriptions of random signal properties are described. Cumulative damage predictions are made using Miner's Hypothesis on a basis of positive peak stresses to failure for Stage A and Stage B lives, and overall fatigue life. The results confirm that Miner seriously underestimates the damage contribution of low stresses on fatigue life as a whole, and they also show that the damage contribution of low stresses is underestimated for the crack initiation phase of life. It is suggested that the inadequacies in Miner's Hypothesis cannot always be accounted for by adopting the "hypothetical S/N curve" concept. An investigation into size effect behaviour under random and constant amplitude loading is presented, and indicates a striking similarity in Stage A/Total life behaviour for both large and small geometrically similar specimens. The fundamentals involved in the design of electro-hydraulic servo controlled fatigue test rigs are outlined and applied to the three servohydraulic rigs which were built in order to carry out the described experimental program. |
