Applying Lux's AI Framework for Successful Outcomes
Back in May we shared our outcome-focused framework that can increase a company's overall chances of success for artificial intelligence projects. This framework, explained in the report, "Artificial Intelligence: A Framework to Identify Challenges and Guide Successful Outcomes," recommends that companies begin the AI thought process with three factors in mind: the end goal or outcome of the project, capabilities of today's AI tools, and solutions and challenges associated with AI implementation.
At Lux we are applying this framework to several different industrial settings to help uncover novel insights and guidelines on successful deployment of AI. In this blog, we focus on the upstream oil and gas industry. The following are the use cases that we will analyze using our AI framework.
- Seismic data interpretation: The process of understanding and mapping subsurface geology using processed seismic data
- Well planning for optimized economics: Optimized well plans and completion designs that offer higher production returns
- Formation testing fluid sampling: Utilizing historic and incumbent data to predict contamination levels of fluid samples during formation testing
- Frack fluid and proppant injection design: Leverage visual analytics and machine learning on real-time well data for optimizing injection of proppants and frack fluid
- Crude logistics and trading optimization: Optimize the value of a barrel of crude by combining oil quality data with market data, commodity prices, etc.
- Synthetic well logs: Produce premeditated well logs using dynamic drilling data
- Interpret distributed acoustic and temperature measurements: Process distributed fiber data using machine learning to understand downhole fluid properties and movement
- Reservoir management: Reducing risk in the drilling of new production or development wells
- Rod pump optimization: Predicting and reducing failures in artificial lift equipment like rod pumps for higher well production
These applications and use cases are a mix of potential ideas and already-proven solutions. The amount of reasoning and environmental complexity each of these use cases require in the upstream oil and gas industry is quite varied. As a result, we encourage companies to segment the applications on a granular level when conducting similar exercises internally. With that said, we used the qualifying criteria below to define the two axes for the upstream sector along with industry-agnostic metrics:
Below are the results from the AI framework for the prior mentioned use cases based on our methodology for the two axes in upstream:
Summarizing the information from our analysis above, the following three takeaways are evident:
- Use cases in exploration remain long-term: Use cases in exploration: reservoir management, and seismic data interpretation end up in quadrants 3 and 4, which currently do not have mature AI tools that will help in implementation. Despite this limitation, startups like Earth Science Analytics continue to raise investments from the likes of Saudi Aramco.
- Solutions that are complex to deploy are in late stages of development: Synthetic well logs, a use case currently in early stages of large-scale commercialization by Quantico Energy Solutions, require a high amount of reasoning in a complex drilling environment. Nonetheless, the company has had trials with Shell, with current plans to deploy the solution in EMEA, potentially with new operators.
- Low-hanging fruits have yet to be resolved: Outside applications in artificial lift, use cases that are lower on both reasoning and complexity are still in early stages of development by startups like Validere's crude trading solution and Sensalytx's distributed sensing platform. In other words, it appears the upstream sector has overlooked low-hanging fruits in its pursuit of complex applications.
Alongside the above takeaways, it can be concluded that the upstream industry has pursued mostly big wins. Oil and gas has been finding million-dollar opportunities for AI to resolve, compared to applications that actually produce tangible benefits. This is particularly evident for applications in exploration, where reducing the risk of dry wells can save E&P companies in the hundreds of millions of dollars. However, given that most of these applications sit in quadrants 3 and 4 of our framework, they are long-shots and consequently might not produce tangible ROIs, at least in the near future.
Another case in point can be made when we compare the two startups Ambyint and Quantico Energy. Ambyint is a Canadian startup that incorporates deep learning algorithms to optimize the performance of artificial lift equipment like crank rods, ESPs, and stock tanks. Quantico, on the other hand, produces premediated well logs via machine learning and reduces the substantial costs associated with downhole logging tools. In its decade of existence, Quantico has generated revenue in the single-digit millions with no disclosed deployment of its solution outside a pilot with Shell. On the other hand, Ambyint has surpassed $10 million in revenue and has a long list of customers, including Equinor, BP, Husky Energy, and Canadian Natural Resources. As evident from our analysis, Ambyint's solution is catered to the current level of AI tools available in the industry compared to Quantico and has subsequently experienced higher growth and success.
In conclusion, we would recommend oil companies to start small and go for quick wins by utilizing the framework above and targeting applications in quadrants 1 and 2. Not only will this approach produce ROIs in favorable timelines, it will facilitate an easier cultural transformation of the organization by highlighting the financial gains AI can bring to the company. The latter issue in particular is highly pertinent in the upstream oil and gas industry today, where change management is considered the biggest hurdle in digital transformation.
This does not mean that applications that make it to quadrants 3 and 4 should not be given any impetus at all; instead, clients should be aware that such use cases will either fail or only bring tangible value in a much longer time frame. However, given the phrase "cost is king" is heavily relevant to the upstream industry; investing in such long-shots for AI development would be a luxury only the bigger players like the oil majors can afford.
In the coming months, Lux plans on publishing similar analysis of several use cases across multiple industry sectors. Please do not hesitate to reach out to us to share your feedback or discuss how other use cases and applications might fit in this framework.
- Insight: Applying Lux's AI framework for succesful outcomes: Uncovering insights in the upstream oil and gas industry (Members Only)
- Report: Artificial Intelligence: A Framework to Identify Challenges and Guide Successful Outcomes (Members Only)
- Blog: Navigating the AI Hype: An Outcome-Focused Framework
- Blog: #LuxTake: Emerging Challenges & Solutions in AI