Using Artificial Intelligence (AI) to Determine Grain/Particle Size for Completion Selection from Cuttings

Sand control, keeping sand particles out of the wellbore, is one of many factors considered in completion design. Sub-optimal completion design impacts production and rectification is operationally challenging and extremely costly. The predominant technique used for determining the size distribution of sand particles is based analogues or, when present, core and/or sidewall cores. However, the available cuttings data is rarely examined due to contamination of drill cuttings data.

Cuttings are typically collected in every well. Cuttings samples are commonly preserved but contain with surrounding shales or drilling additives alongside fragments of the drilled formation. Whilst it is possible to measure grainsize from these cuttings manually under the microscope, this is process is labour intensive and commonly lacks rigour. A quantitative alternative is laser particle analysis, this is applied where core samples are available but this technique is not easily applied to cuttings samples.

This paper transforms grain/particle analysis for completion selection by demonstrating how images of clean cutting samples examined with an AI can be segmented into thousands of grains. These grains can be filtered removing background grains and contamination leaving a clean statistical grainsize distribution. This method is both rigorous, repeatable and scalable.