| ||Format||Pages||Price|| |
|PDF Version||6||$25||  ADD TO CART|
Variability in experimental data on fatigue crack growth kinetics reflects contributions not only from material property variations and environmental and other uncontrolled variables but also from measurement precision used in determining the primary data (that is, crack length versus elapsed cycles) and the subsequent data processing procedures in determining rates from the primary data. To assess the contributions of measurement precision and data processing procedures to variability, computer simulation of primary fatigue crack growth data was made for a number of measurement intervals and precisions. These simulated data were analyzed by several data processing procedures.
Variability in the derived growth rate data was found to depend strongly on the magnitude of the measurement interval relative to the measurement precision. This variability was reduced by data processing procedures, such as incremental polynominal methods, that fitted a smooth curve through portions of the primary data. Such procedures, however, can introduce statistically significant bias into the derived data on crack growth kinetics; this bias may or may not cause large errors in predicted fatigue life. Because variability associated with crack length measurement interval and precision and with data processing procedures can be quite large and is incorporated in an unknown way into much of the published data on fatigue crack growth rate, caution should be exercised in drawing statistical inferences regarding material differences from such data. Suitable procedures that would allow assessment of material differences from experimental crack growth data need to be developed.
Professor, Lehigh University, Bethlehem, Pa.
research assistant, University of Illinois, Urbana-Champaign, Ill.
Engineer, Homer Research Laboratories, Bethlehem Steel Corp., Bethlehem, Pa.
Stock #: JTE11207J