Artificial Intelligence in a Carbon Light Future

Authors: Gareth Taylor and Barry Zhang

Our environmental footprint and consequences of acting without data

Consumers and financial investors are increasingly calling on the oil and gas industry to reduce its environmental footprint. In the Upstream sector, oil companies have traditionally focused on health and safety but the environmental aspect in HSE never received the same priority. Additionally, it has been challenging for companies first, to quantify the CO2 impact of their field operations and second, to find viable alternatives. While billions of dollars in value have been created by new technologies, the fundamental process of drilling a well and lifting hydrocarbons has presented daunting challenges to those aiming to reduce carbon footprint.

If we examine the Deepwater Horizon blowout, an event with catastrophic environmental consequences and a deeply troubling and longstanding impact on consumer perceptions of the Upstream industry, we can identify clear examples of how today’s artificial intelligence (AI) technology could have perhaps mitigated both the environmental consequences and saved lives.

Fig1. NOAA image of Deepwater Horizon Disaster and Contamination

While most have a perception of the blowout being due to a poor cement job, the post-mortem analysis sheds light on the root cause. The cement job failure was actually the result of drilling from highly over-pressured intervals into a less pressured interval which led to loss of hydrostatic pressure immediately following shut down of the cement job and before the wellhead seals were set and locked. Furthermore, according to the technical postmortem report of the Deepwater Horizon blowout, a sharp decrease in the pore pressure at the M56 reservoir sand (Figure 2) was accompanied by an increase in the average mudstone velocity, resistivity and density reflecting increased compaction due to increased effective stress. This, in combination with the extreme pore pressures within overlying strata, drastically narrowed the range of safe operational borehole pressures. These geologic phenomena produced challenging conditions for drilling, prevented successful temporary abandonment of the well, and contributed to the well’s failure.

Fig.1. Pressure vs Depth through the M56 reservoir in the Macondo well. Note the modeled mudstone pressure, ums, (blue line) falls abruptly at Macondo (adapted from Pinkston & Fleming,20191)
Fig2. Macondo Pressure

AI technology is now available to provide real-time, ahead-of-the-bit awareness of the reservoir property changes that led to the explosion and caused the tragic deaths of eleven people and, in the three months after the blowout, an estimated 4 million barrels of oil leaked into the Gulf of Mexico. AI is also able to extract rock property information out of the vast amounts of drilling data being collected on the rig. These AI-based well logs provide real-time sonic properties over 100ft ahead of the LWD sonic tools, which translates into at least an hour of additional lead time that could have helped the decision-makers identify unanticipated wellbore stability issues. These examples illustrate how AI could have perhaps provided the oil and gas professionals at BP and its service companies with the right information earlier.

However, CO2 reduction initiatives and improved safety are multiple times more impactful if methods to deliver the right information earlier are adopted routinely across all the global field operations of oil companies, rather than either not at all, or only in a few, high spec drilling operations such as those in deep and ultra-deep water.

From a historical perspective numerous attempts have been made to assist drilling engineers with “seeing-ahead-of-the-bit” capabilities while drilling. These attempts range from calculating resistivity several meters ahead of the drill bit (Constable, et al., 2016) to “ahead-of-the-bit” pore pressure prediction using VSP while drilling (Xi et al., 2010), to the deployment of “seismic-while-drilling” technology for salt flank and fault plane imaging (Dethlof and Petersen, 2007). These solutions have been either too complex, too costly, too slow, too inaccurate or too poor resolution.

Quantico Energy Solutions provides AI technology that offers real-time formation evaluation logs (QLog) and earth model updates (QEarth). The latter provides real time updates of geological structure and the following properties – in high resolution – utilizing the latest data collected from LWD tools or QLog.

It is the Class 3 and 4 properties that are most impactful. Previously unanticipated aberrations in these properties could lead to significant drilling risks, including shallow water flow, shallow gas, anomalous high-pressured pockets, salt/sediment proximity, heterogeneity within salt (sand lenses) and sub-salt pore pressure, etc. Traditionally, drilling engineers have focused on the safety consequences surrounding these issues as well as the financial impact caused by non-productive time, stuck pipe, lost circulation, hole deviation, bit wear, etc.

Fig3. QEarth permian earth model, cross-section of results

The potential gains from carbon light field operations

Consider a future for Upstream comprised of carbon light field operations. A drilling engineer watches the passage of the drill bit through a 3D earth model that shows a map of unconfined compressive stress (UCS) thousands of feet ahead of where the bit is currently located. Continuous feeds of gamma ray, resistivity and drilling dynamics measurements are being streamed to a central monitoring site, where AI software autonomously processes the data and predicts compressional sonic, shear sonic and density, which in turn are used to calculate UCS in the wellbore. These AI logs subsequently are ingested into additional AI software which autonomously updates a high resolution earth model hundreds of feet ahead of and around the drill bit.

QEarth and QLog are software solutions that don’t require the presence of two to ten personnel on site – translating into fewer hours of safety exposure as well as less fuel and logistics relating to lodging and transportation to and from the rig. Additionally,

nuclear sources can be eliminated, which means that if down-hole equipment becomes stuck in the well, there are no undue environmental consequences from leaving radioactive material in the well or increasing the CO2 footprint further as part of additional cementing operations to isolate the radioactive material from the surrounding environment.

With recent commitments from major oil companies such as BP, Shell and Total regarding their plans to achieve net-zero greenhouse gas emissions by 2050, the deployment of artificial intelligence has clear implications for the Upstream industry. Environmental sustainability considerations include not only the avoidance of unsafe drilling and oil spills, but also carbon light drilling and logging operations that when adopted as routine practice, add up to substantial reduction in CO2 and perhaps equally importantly, demonstrate to global consumers and investors that the industry is rising to the call to action.


1. Pinkston and Fleming, 2019. Overpressure at the Macondo Well and its impact on the Deepwater Horizon blowout. Scientific Reports, Nature.

2. Taylor, et. al., 2020. Seeing-ahead-of-the-bit: A game changer enabled by Machine Learning: ARMA, American Rock Mechanics Association.

132 views0 comments

Recent Posts

See All