Embedded System for Cavings Detection in Well Drilling Operations Using Computer Vision Techniques to Enhance Real Time Well Stability Analysis

Wellbore instability is a critical challenge in drilling operations, often leading to nonproductive time (NPT), increasing operational costs, and safety risks (Kumar et al., 2012). One of the primary indicators of wellbore instability is the presence of cavings, rock fragments of the formation that detach from the wellbore wall, that were not crushed by the drilling bit, differing in size and shape from traditional drilling cuttings (Nmegbu and Ohazuruike, 2014). The most common causes for the appearance of cavings in shale shakers can be underbalanced drilling, residual stress relief and pre-existing fracture planes in the formations, or even mechanical action related to the drilling process itself, through the action of bottom hole assembly (BHA) for example (Kumar et al., 2012). These cavings can indicate serious operational problems, such as damage to the wellbore wall, and, if not controlled, can cause well collapse and column entrapment, and should therefore be avoided by maintaining a heavy mud with sufficient hydrostatics to maintain the walls of the well and conducting tripping in/out operations with caution. Therefore, the detection and analysis of cavings in well drilling is a very important process, allowing drilling teams to make decisions aimed at ensuring well stability. However, caving detection is usually a non-automated process, which depends on the visual perception of sample catchers and/or geologists, which can make this perception a slow and inaccurate process. In this case, the use of technologies to monitor the occurrence and characteristics of these cavings directly in the shale shaker can provide valuable, real-time insights into geomechanical conditions and alert the drilling team to potential wellbore failures before severe problems develop.