Reduction of Casing Failure Risk via SMART Adjustment of Casing Design and/or Downhole Practices

Abstract

Abstract Casing failure incidents have been on the rise since the 70's. An alarming rate of up to 50% of offshore wells in GOM have experienced some degree of casing failure, not to mention, numerous reports of onshore and offshore incidents worldwide. What has been done to mitigate its aftermaths? Most of the attempts to flag or mitigate casing failures were physics-based solutions; analytical, experimental, or numerical, that tackles one or few aspects on a small scale. However, those former studies were found to perform unreliably and did not attain wide-scale execution, mainly, due to the dynamic complexity of the underlying problem. One cannot help but wonder till when the industry is going to turn a blind eye to this global problem. What am I bringing to the table, then? Simply, an automated solution, based on ensemble models focusing on resolving the infamous casing fatigue problem. This avant-garde solution accounts for potential risk factors and their impact on the probability of casing failure, using them to construct the proposed automated casing failure mitigation framework based on a two-step prediction-correction procedure. This research proposes two different schemes or scenarios that can be followed for mitigating casing failure. Namely, adjusting downhole practices (e.g., Doglegs, Frac. Temp., Cement Support, etc.) that impose high risk of casing failure while maintaining the initial casing string design or adjusting casing design (e.g., Size, Length, Grade, etc.) according to encountered downhole conditions while maintaining downhole practices as is. However, the workflow proposed in this paper is based on Scheme I, where downhole practices is adjusted based upon downhole conditions during cyclic operations. As such, the focus of the study is on highlighting potential casing fatigue estimators for different parts of the casing during such operations. The model uses data aggregated from several sources from both failed and non-failed wells and has been tested on the data set managing to reduce the overall outcome occurrence to "Low". It is the feedback step in the adopted "prediction-correction" scheme that gives the edge to the proposed workflow in this study over other data-driven workflows existing in literature, specifically, in the area of tubular and casing failures. Following this two-step procedure, the proposed tool can handle the risk of casing failure proactively rather than reactively. This, in turn, could potentially give the allowance for drillers and drilling engineers to adjust their design specifications in to avoid or mitigate potential casing failure.