How AI Impacts Oil & Gas Organizations
The impact of Artificial Intelligence on an organization is an important topic when considering machine learning. In a previous blog entitled, What Artificial Intelligence in Oil and Gas Brings to Production, we discussed a summary of key solutions that AI can target for problems facing most oil and gas organizations utilizing Artificial Lift. As we mentioned, these solutions will not be implemented overnight, but the effect on production and operations in an organization will be substantial and worth the effort.
When considering that improving data quality is a marathon, and not a sprint, we should take the first step and consider how an organization might be impacted by an artificial intelligence journey and how that interweaves with data quality. The following are three considerations of the impact of AI on an organization.
First, the initial consideration is an operational change or shift. The impact of this shift is based on where the organization is currently in the digital oil field transformation. If the organization has yet to make strides toward a digital oil field, this might be the first time these impacts have been considered.
Collection and storage of sensor streams or well designs is well understood in organizations today. Ensuring that they are accurately updated may not have historically been considered primary activity. Having a current and accurate view of the surface and subsurface equipment for each well is of paramount importance. Many modeling techniques require knowledge of well equipment and replacement schedules.
Detailed maintenance and failure logs may not have been critical in the past, but artificial intelligence initiatives demand varied high-quality data. Curating this data requires coordination and cooperation across many departments with a significant shift in the importance of its maintenance.
The second consideration of the impact of artificial intelligence on an organization is that the introduction of new techniques in Artificial Lift monitoring, surveillance or optimization can be viewed skeptically by some operational teams. It is vital to reinforce that artificial intelligence solutions are enabling tools to aid in focusing scarce operational or production teams’ to wells that are most critically impacting production. Predictive diagnostics provide substantially earlier indication to failures than manual reviews or waiting until production is impacted. Preventative maintenance analytics can show where wells are not running optimally and what mitigations can be considered to improve production.
If an organization does not have a strong basis for operating from data, then it is possible that their team will question the validity of the source data. It can be difficult for some teams to trust the diagnostics or recommendations provided by machine learning solutions, if they don’t first trust the source data on which it is based.
The third consideration is that organizations must honestly assess where improvement in collection and curation of source data is needed. We will observe possible dimensions to review and evaluate the quality of the source data. In many cases, cataloging the current sources of data available can be the initial step. There may even be quick wins by tackling items like consolidating the naming conventions used for wells or locations across all of the systems. It is reasonable to question the source data being reviewed as it is easy to overlook a serious issue in source data that is collected. Anomalies in signal might be just an anomaly or they might be an indication of some other issue.
To consider the impacts of these source data issues, we suggest that you request our whitepaper, entitled “Data Quality Fuels the AI Race.” Feel free to comment or ask questions. We would love to hear your thoughts and begin a conversation.
