Author(s): Suwan Park; Sang Kil Park

Linked Author(s):

Keywords: ROCOF; Log-linear ROCOF; Power law process; Water main failure data; Threshold break rate; Optimal replacement time

Abstract: This paper presents applications of the log-linear ROCOF and the power law process to model failure rate and to estimate optimal replacement time of individual pipes in a water distribution system in the U. S. The performance of the two ROCOFs in the assessment of the failure rate is examined by using the maximized log-likelihoods for different modeling approaches in relation to varying data types. The optimal replacement time equations for the two ROCOFs are developed by applying the methodology of Loganathan et al. (2000) for the case in which rescaled failure times are used. It was found that the ‘failure-time-based’ method is superior compared to the ‘failure-number-based’ method for the water mains under study, implying that recording each failure time is more helpful in better modeling of ROCOF than observing failure numbers in some time intervals. Furthermore, the log-linear ROCOF showed better performance than the power law process in representing the empirical failure rates of the pipes for the ‘failure-time-based’ method.

DOI:

Year: 2005