Using Modeling and Data Analytics To Signal Change: Can We Avert a "Midlife Crisis"?

Abstract

Abstract More often than not we have more data than we know what to do with. The availability of large amounts of data should enhance decision making but often has the opposite effect. The sheer amount of data, combined with the pressure to make smarter and faster decisions often overwhelms decision-makers. This holds true in the world of corrosion as much as any other field. The cost of acquiring data and the risks of misinterpretation are significant. Any manipulation of data which can reveal aspects/patterns that are linked to corrosion will help to minimize risks and assessment of the condition of an operating system. Increased capability to manage risk has the potential to add significant value through optimization of effectiveness for monitoring, inspection and chemical deployment as well as timely identification and selection of equipment for repair/replacement. Given the critical importance that estimation of corrosion rate has in determining risk of future failure and in the estimation of the time to repair/replacement, this paper outlines the concepts of alternative and complementary means of estimating this rate/condition using data analytical methods. In addition to consideration of local estimates of corrosion rate (i.e. spot ultrasonic data or in-line inspection data) the activity has involved processing a large set of rate estimates to search for predictive patterns based on attributes of the lines, process conditions, coupon/probe data, similarity of service conditions and use of inhibition chemicals.