Prediction of external corrosion for steel cylinders at the Paducah Gaseous Diffusion Plant: Application of an empirical method

Abstract

Comparison of mean and standard deviation of age-averaged corrosion rates (mils/yr) for C-745-F and C-745-G yards calculated in Lyon (1995)

Prediction of external corrosion for steel storage cylinders

“…Some activities planned may substantially reduce the rate of corrosion, in which case the results presented here are conservative. The results presented here are intended to supercede those presented previously (Lyon 1995(Lyon ,1996, as quality of several of the datasets has improved. System Requirement 1 .…”

“…Using the F-test (see, e.g., Snedecor and Cochran 1978, pp.116-117), one can accept that the variances of the logarithms of the pit depths for the top and bottom row populations are equal with 5% significance, whether or not the breaches are included. Similarly, using the same test as described in Lyon (1996), namely the t-test with unequal variances (Casella and Berger 1990), one can conclude that the medians of the distributions are the same; i.e., there is not a statistically significant difference between the medians of the distributions for the top and bottom row cylinders. Both of these results support treating the top and bottom rows as a single population, whether or not the breaches are included.…”

“…In Lyon (1995 and1996) this was dealt with in a simple fashion. In this report, the initial thickness is modeled using a distribution that accounts for the variability in the initial thickness.…”

“…This method can be rather sensitive to deviations from the "leveling ofY pattern usually expected for the penetration depth, and so must be used with caution. Indeed, in cases where no "leveling off" is observed (due, for example, to a narrow age range for cylinders), the simpler method used in Lyon (1995Lyon ( ,1996 Given that the data consist of what itre considered to be maximum pit depths, it is natural to apply extreme-value statistics to this problem. Indeed, application of this method (without confidence limits) is discussed in several papers in the literature and has also been suggested for use within this program in Rosen and Glaser (1996).…”

Ultrasonic thickness sampling plan for the depleted uranium hexafluoride program

“…For example, by using upper 95% confidence limits for each term, one cannot in general conclude with 95% confidence that the resulting s u m is an upper 95% confidence limit. Thus, the confidence limits reported in Lyon (1995Lyon ( ,1996 as upper 95% confidence limits are more accurately described as simply upper bounds using upper 95% confidence limits for each term.…”