The Use of Artificial Intelligence (AI) to Give Insight to a Complex Well Tubing Issue to Ensure Suitable Corrosion Mitigation Approaches

Abstract

In one of TotalEnergies North Sea fields, nitrate is continuously injected as the main mitigation of reservoir souring. The nitrate treatment is employed to limit the biogenic generation of H2S, due to reservoir souring. Reservoir souring is caused by the proliferation of SulfateReducing microbes (bacteria and/or archaea). In parallel, severe corrosion was observed in oil producing well tubing conveying fluids from a reservoir subjected to nitrate treatment. It is known that nitrates can catalyze the corrosion of carbon steels through different mechanisms, in particular electrochemical or microbiological. The challenge was therefore to accurately evaluate the impact of nitrate injection, and its presence in the produced fluids, on the corrosion of the producing well tubing. The influence of nitrate treatment towards corrosion of carbon steel well tubing has been singled out using an artificial intelligence algorithm. The algorithm employs a method called "Propensity Score Matching" (PSM). PSM is a statistical method commonly used in the medicine field to assess the efficiency of treatment in biased data. This method has been used to compare two reservoirs, one being subjected to nitrate treatment and another that has never been subjected to nitrate treatment. In this paper a first known use of causal inference in industrial settings is presented. It helped removing biases by comparing only comparable data and to find causal relationship in the data. Outcomes of the multi-faceted study have led to optimized corrosion mitigation approaches and increased well tubing service life.