Intelligent Application of Computer Vision and Data Analytics to Optimize the Separators Cleaning for Unconventional Reservoirs

Abstract

The reliability of the production operations depends not only on the well performance but also on the effectiveness of the surface facilities to transport and separate the produced fluids. In the case of the unconventional reservoirs, the completion treatments placed to stimulate the long horizontal wells require large volumes of proppant and water. During flowback and even later in the life of the well, fractions of the proppant makes its way to the surface and into the separators. Large accumulations of sand reduce the ability of the separators to perform as designed, impacting production, and requiring complete shut-down for cleaning to restore their original capability. The work introduces an intelligent end-to-end workflow integrating computer vision and data analytics to automatically interpret thermographic images, identifying when a production separator needs condition-based maintenance. The new approach leverages infrared thermography pictures taken from hundreds of separators in an unconventional asset and automates a labor-intensive process to make objective maintenance decisions. Contrasted to the manual method, where vessels were taken offline, visually inspected, and cleaned out on time-based maintenance schedules, this work provides an accurate visualization of the sand level using computer vision. The study demonstrates who how integration of digital technologies such as computer vision and data analytics enable optimization of maintenance work. The application showcases the business impact not only through cycle time reduction and effort, by also enables better decision making and optimization of resources.