This paper presents two methods to estimate the variation coefficient for the accumulated damage, when only one measurement of the load is available. The methods are applied to loads on heavy vehicles travelling on uneven roads. The accuracy and precision of the estimates of the variation coefficient is evaluated using both a standard road profile model (Gaussian with spectrum according to ISO 8608) and a more realistic model that includes irregular sections. Four examples are presented to illustrate the performance of the estimates. Finally, the coefficient of variation is computed as a function of road length, for a real measured gravel road. Such an analysis indicate how long one need to measure, before vehicle damage variability can be neglected. When the model of a random load is fully specified, e.g. if the load is a stationary Gaussian process with a known mean and spectrum, then both expected damage as well as coefficient of variation can be estimated by means of the Monte Carlo method. More precisely, one may simulate a large number of loads, with fixed length, from the model and compute the accumulated damage for each of the loads. Then expected damage and variance can be estimated using standard statistical estimators. (Even the distribution of the damage can be studied in this way.) For very long sequences the variability of the observed damage values is approximately normally distributed and hence mean and coefficient of variation fully describe the variability. However, for high values of the parameter k the convergence to normality can be very slow.
Uncertainty of estimated vehicle damage for random loads
2009
10 Seiten, 7 Bilder, 2 Tabellen, 6 Quellen
Aufsatz (Konferenz)
Englisch
Damage Tolerance and Reliability Assessment under Random Markovian Loads
Tema Archiv | 2013
|Vehicle fatigue damage prediction based on estimated strains and road measurements
BASE
|British Library Conference Proceedings | 2001
|Fatigue damage analysis of the straight switch rail considering complex random loads
Taylor & Francis Verlag | 2023
|