OREDA - Offshore Reliability Data Handbook 2002- 4th Edition.iso
Click Here >> https://geags.com/2t9Pmt
Statistical and numerical uncertainty analysis is performed in the optimization procedure as a part of the reliability data assessment. Constraints defined in the optimization are used to mitigate variability in the results (Molshakov & Sofronov 2015; Slawinski 2011). Therefore, it is crucial to evaluate the uncertainty that exists in the reliability data and to estimate the contribution of each reliability data source to the overall uncertainty of the reliability model. The main sources of uncertainty are the reliability data, which are represented by the chosen failure rates (i.e., the frequency per year), the reliability and mean time to repair (MTTR) data, and the statistical confidence interval (C.I.) for the reliability data (Ambhl et al. 2015). The first source of uncertainty has been estimated using the Monte Carlo probabilistic method with a large number (up to 10,000) of samples. The second source of uncertainty has been estimated using the Monte Carlo simulations in which the reliability data are distributed around the best-fit model by a C.I. for the reliability data. In this paper, the proposed method for the C.I. estimation is the use of the standard error (SE) and coefficient of variation (CV) to calculate the C.I. for the reliability data. The validity of the method has been verified by examining the probability density function (PDF) of reliability data sources, particularly for the devices having a small number of data samples (Ambhl et al. 2015; Mantzaris et al. 2007; Slawinski et al. 2004). Finally, the third source of uncertainty is the deviation in the mean time to repair (MTTR) data. Two approaches to estimate the deviation in the MTTR data have been used: (i) the bootstrapping method has been used to estimate the variance in the MTTR data for the devices having a small number of data samples, and (ii) the interval estimation of the MTTR data has been estimated from the historical data for the devices having a large number of data samples. In this paper, the approach (i) has been used.
where t is the device or plant lifetime, n (n>0) is the number of failure cycles, N is the estimated number of customers, and Z is the failure rate of the device. The reliability of a machine is then estimated using the cumulative distribution function (CDF) of the probability mass function, defined as 827ec27edc