Data Analysis and reduction in Hanford's corrosion monitoring systems

Abstract

A project to improve the Hanford Site's corrosion monitoring strategy was started in 1995. The project is designed to integrate EN-based corrosion monitoring into the site's corrosion monitoring strategy. In order to monitor multiple tanks, a major focus of this project has been to automate the data collection and analysis process. Data collection and analysis from the early EN corrosion monitoring equipment (241-AZ-101 and 241-AN-107) was primarily performed manually by a trained operator skilled in the analysis of EN data. Thousands of raw data files were collected, manually sorted and stored. Further statistical analysis of these files was performed by manually stripping out data from thousands of raw data files and calculating statistics in a spreadsheet format. Plotting and other graphical display analyses were performed by manually exporting data from the data files or spreadsheet into another plotting or presentation software package. In 1999, an Amulet/PRP system was procured and employed on the 241-AN-102 corrosion monitoring system. A duplicate system was purchased for use on the upcoming 241-AN-105 system. A third system has been procured and will eventually be used to upgrade the 241-AN-107 system. The Amulet software has greatly improved the automation of waste tank EN data analysis. In contrast with previous systems, the Amulet operator no longer has to manually collect, sort, store, and analyze thousands of raw EN data files. Amulet writes all data to a single database. Statistical analysis, uniform corrosion rate, and other derived parameters are automatically calculated in Amulet from the raw data while the raw data are being collected. Other improvements in plotting and presentation make inspection of the data a much quicker and relatively easy task. These and other improvements have greatly improved the speed at which EN data can be analyzed in addition to improving the quality of the final interpretation. The increase in data automation offered by the Amulet software is necessary if multiple tanks are to instrumented and analyzed at the Hanford Site. Although advances in the automation of data analysis have been great, Hanford EN data analysis still demands a highly trained corrosion expert. Neural networks could de-skill the post-data collection analysis procedure and broaden the range of users able to understand and interpret corrosion data. Ultimately, the ability to de-skill data the data analysis process will make or break the use of EN as a plant monitoring tool on a wide scale.