Probabilistic Corrosion Assessment for Natural Gas Storage Wells

Abstract

Regulations for gas storage wells require that operators perform initial and subsequent mechanical integrity evaluations as determined using risk assessment [1], incorporated by reference in 49 CFR 192.12 for U.S. operators [2]. As a well ages, metal loss on the casing can grow, increasing the probability of a failure from corrosion. Inspection and repair programs manage this probability by reducing uncertainty in the casing condition and repairing significant metal loss anomalies. However, performing a casing inspection involves a considerable amount of risk, which can vary depending on the well configuration [3]. The benefit of inspection and repair needs to be balanced with the inspection risk to determine the interval that minimizes the overall risk. This paper demonstrates that if detailed information about a well’s configuration, loading, and existing corrosion population is considered, a probabilistic corrosion analysis can be completed to determine a reinspection date that minimizes the overall risk of a release. A probabilistic implementation of the Level II analysis found in API 579 Fitness-For-Service is described and recommended for these assessments [4]. The deterministic version of this model is the most accurate for predicting burst pressures of casings with metal loss under a wide range of loading conditions [5]. The measurement error of inspection tools and their reporting thresholds relative to typical corrosion rates presents many challenges in calculating corrosion rates deterministically. Calculating unrealistically high growth rates and apparent negative growth rates using inspection data is common. Using the tool measurement error and the distribution of calculated growth rates across several wells, a Bayesian updating approach is described, grouping anomalies in similar environments to develop credible growth rate distributions specific to each joint on a well. This paper provides several assessments of realistic storage well configurations and corrosion populations to demonstrate how the probabilistic corrosion assessment can determine an inspection interval that minimizes the overall risk and ultimately inform integrity maintenance plans on a well-by-well basis. The examples span a wide range of well conditions to illustrate that the optimal inspection and repair program depends on each well’s configuration, loading, and existing corrosion population. The effect of a corrosion anomaly’s depth in the well and the cement quality on the expected release rate and the resulting risk is also examined.