The Undigitized Frontier: Automating Shale Shaker Analysis with Computer Vision

The abstract introduces “camera-as-a-sensor technology” to enhance real-time situational awareness during drilling. With the push towards automated drilling, the shale shakers—the point of separation between drilled rock and mud—remain “conspicuously undigitized,” relying on intermittent, analog processes like the “bucket test”.

The approach positions a camera at the exit of each shale shaker to image solids, using embedded edge GPUs for pre-processing. Computer vision then applies a five-step process to detect, characterize, and classify objects as cuttings, cavings, or unidentified falling objects (UFOs). This data enables crucial calculations and workflows, such as:

1. Calculating volumetric cuttings return rate and comparing it to the theoretical rock volume drilled.
2. Generating alerts for cavings or UFOs, including a 5-second video clip for the wellsite geologist (WSG).

The deployment of this technology has validated its capability to quantify cuttings return rate and detect cavings earlier than conventional methods. Pathways to value creation include:

1. Optimizing drilling time by adjusting parameters based on observed hole cleaning effectiveness.
2. Saving minutes on each cleanup cycle by pinpointing when shakers are clean.
3. Detecting the onset of cavings to mitigate incipient borehole instability.
4. Providing critical real-time cutting size to automated drilling algorithms.

In conclusion, digital shale shaker surveillance shows “considerable promise” for reducing hidden non-productive time and decreasing the frequency of “costly stuck-pipe incidents” by improving holecleaning and borehole-stability management.