Profiling downhole casing integrity using Artificial intelligence

Realizing the growing demand to assess the integrity of downhole casings and proactively predict their failures is of a prime importance. Being able to analyze downhole metal loss data from logged wells is a challenge by itself. Predicting the metal loss profile for non-logged wells is yet more challenging. In huge oil fields, where the number of wells can exceed two thousand, practicality is given a significant weight during the corrosion logging candidate selection process. Downhole corrosion is the main threat to the integrity and upkeep of wellbore completion tubulars. It can be categorized into internal corrosion and external corrosion. Internal corrosion is mostly caused by production/injection fluids. It attacks the inner downhole string namely; tubing or casing for a tubingless completion. External corrosion attacks the outer downhole casings and can cause dramatic loss of assets and production. Water-bearing formations, especially at shallow depths, are the main source of external corrosion. Poor cement bond behind casings promotes and accelerates external corrosion growth. This is mostly observed across loss-circulation zones where cement quality is questionable. Technology development in the field of downhole multiple casings metal loss assessment is still at its infancy. Electromagnetic (EM) tools are on the leading edge in downhole multiple casings inspection yet have their own limitations. An intensive integrity inspection campaign started in Field A using these tools. The objective of this campaign is to identify wells with potential integrity issues and proactively suggest the best course of remedial actions. Artificial intelligence and data-driven models, such as artificial neural networks were employed to predict downhole casing integrity of wells with no corrosion logging data. More than one hundred datasets were collected to build the artificial intelligence model used for this purpose. This paper unlocks a completely new application of artificial intelligence in the field of well integrity surveillance.