Prediction of External Corrosion for Steel Cylinders 2003 Report

Depleted uranium hexafluoride (UF{sub 6}) is stored in over 60,000 steel cylinders at the East Tennessee Technology Park (ETTP) in Oak Ridge, Tennessee, at the Paducah Gaseous Diffusion Plant (PGDP) in Paducah, Kentucky, and at the Portsmouth Gaseous Diffusion Plant (PORTS) in Portsmouth, Ohio. The cylinders range in age from six to 52 years. Although when new the cylinders had wall thicknesses specified to within manufacturing tolerances, over the years corrosion has reduced their actual wall thicknesses. The UF{sub 6} Cylinder Project is managed by the United States Department of Energy (DOE) to safely maintain the UF{sub 6} and the cylinders containing it. The requirements of the Project are delineated in the System Requirements Document (LMES 1997a), and the actions needed to fulfill those requirements are specified in the System Engineering Management Plan (LMES 1997b). This report documents activities that address requirements and actions involving forecasting cylinder wall thicknesses. Wall thickness forecasts are based on models fit to ultrasonic thickness (UT) measurement data. First, UT data collected during FY02 is combined with UT data collected in earlier years (FY92-FY01), and all of the data is inventoried chronologically and by various subpopulations. Next, the data is used to model either maximum pit depth or minimum thickness as a function of cylinder age, subpopulation (e.g., PGDP G-yard, bottom-row cylinders), and initial thickness estimates. The fitted models are then used to extrapolate minimum thickness estimates into the future and to compute estimates of numbers of cylinders expected to fail various thickness criteria. A model evaluation is performed comparing UT measurements made in FY02 with model-fitted projections based only on data collected before FY02. The FY02 UT data, entered into the corrosion model database and not available for the previous edition of this report (Schmoyer and Lyon 2002), consists of thickness measurements of 48'' thin-wall cylinders: 102 cylinders at Paducah, 104 at ETTP, and 117 at Portsmouth; and 72 thick-wall cylinders at Portsmouth. However, because of missing values, repeated measures on the same cylinders, outliers, and other data problems, not all of these measurements are necessarily used in the corrosion analysis, and some previous measurements may simply be replaced with the new ones. In this edition of the report, cylinder subpopulation definitions and counts are updated using the latest (as of June 2003) information from the Cylinder Inventory Database (CID). Cylinders identified in the CID as painted during the last ten years are excluded from subpopulations considered at-risk of failing minimum thickness criteria, because it is assumed that painting fully arrests corrosion for ten years. As in the previous edition of the report, two different approaches to corrosion modeling are pursued: (1) a direct approach in which minimum thickness is modeled directly as a function of age, subpopulation, and initial thickness estimates; and (2) an indirect approach, in which maximum pit depth is modeled, and the pit-depth model is then combined with a model of initial thickness to compute estimates of minimum thickness