08 June 2021
Los Angeles, CA


Beyond Limits has harnessed the power of NVIDIA’s platform to solve one of the most substantial problems in the upstream oil and gas industry. Read the full case study HERE.


The Upstream Sector’s Greatest Challenge: Field Planning


Upstream endeavors normally demand an extensive team of geologists, petro-physicists, reservoir engineers, mathematicians and more, working together in order to construct models that take several months, or even years, to complete. It’s a necessity for such models to capture expert input, as well as statical and dynamic data from considerably high-value assets. Historical human knowledge, experience, and expertise are crucial at almost every stage of interpreting historical performance and data.


One of the greatest challenges for the upstream oil and gas sector is the optimization of well placement in oil and gas reservoirs (field planning). As an example, and to give a sense of the scope of the challenge, a game of chess has nearly five million possible moves after the first five. Optimal well placement paints a similar picture. Well placement in a small reservoir can have as many as one hundred trillion possible combinations within only five sequential, non-commutative choices of vertical well-drilling locations. This game of choice is on a level all its own.


Luckily for everyone involved, Beyond Limits is rewriting the rulebook when it comes to solving the well placement optimization problem at scale. By leveraging NVIDIA A100 Tensor Core GPUs, Beyond Limits is developing a deep reinforcement learning (DRL) framework that provides high-performance computing (HPC) infrastructure to support novel AI frameworks making the strategic moves necessary to simplify this intricate field.


Simulated field and well trajectories



Inside Industry Solutions: The NVIDIA x Beyond Limits Approach


Beyond Limits and NVIDIA set out to solve the computational challenge currently facing many companies across the upstream oil and gas sector by developing a new solution that elevates field planning workflows based on a DLR framework with a primary objective to maximize net present value (NPV), i.e., projected dollar value for production capabilities of any given field with a specified number of wells.


Unavoidable additions of well assets inherently result in both higher drilling costs and increased risk for negative environmental impacts. Therefore, simultaneously maximizing returns while minimizing drilling is a capability that’s paramount to the energy sector. Adopting cutting-edge AI solutions is becoming increasingly essential for upstream companies to successfully reduce waste while maximizing productivity and profitability over the projected lifetime of an asset as essential as their most valuable wells.


“Beyond Limits has used the power of the NVIDIA platform to solve one the most substantial problems in the upstream oil and gas industry.” -NVIDIA





Beyond Limits AI Solution for Reservoir Management Accelerated by NVIDIA GPU’s


Beyond Limits AI solution for reservoir management is just one example of an application that supports operators to dramatically accelerate data integration and analysis for pinpointing drilling opportunities, weighing opportunity versus risk, and speeding up time to decisions.


Beyond Limits’ NVIDIA-GPU accelerated reinforcement learning framework enhances the solution with high-performance AI capabilities to optimize well placement decisions by identifying several thousand combinations, reduced from over one hundred trillion possible combinations (mentioned at the outset). Adoption of this advanced solution yields over 250% algorithmic performance improvement in a typical reservoir, leading to an additional 10% improvement in NPV for vertical wells.


Reinforcement learning in real-time


Even the smallest enhancements can make a huge impact. Seemingly minor adjustments help solve challenges at an accelerated speed and increases scale thus reducing well placement (in typical reservoirs) timelines down from months to hours, yielding millions of dollars in additional value and revenue.





Field Planning for Deep Reinforcement Learning: The Net Zero Impact


AI will play a key role in helping the energy industry achieve net zero ambitions by making operations more efficient, increasing productivity, and reducing waste. AI applications such as Beyond Limits’ NVIDIA-GPU reinforcement learning framework can support these initiatives with carbon capture and sequestration, reducing drilling of new wells, and transitioning to low-carbon energy sources. Examples of intentional efforts toward accomplishing those goals include identifying the best CO2 injection candidates across depleted fields, providing optimal planning for existing fields, minimizing drilling length, converting between injection and production weeks, and sequential planning for novel energy use cases broader than field planning.



More Benefits of AI in Upstream: From Deep Reinforcement Learning to Cognitive AI


Beyond Limits advanced AI technology has the ability to help support entire upstream strategies, streamlining processes, and optimizing entire operations. Several benefits for implementing and adopting this technology in the sector include:


+ Capturing Institutional Knowledge & Business Logic


This AI solution bridges the knowledge-loss gap for providers by capturing and encoding veteran operator knowledge then distributing that essential expertise, best practices, and business logic across an entire organization so it’s never lost and easily transferrable to newer, up-and-coming operators within the sector.


+ Reducing Uncertainty when Data is Limited


This technological approach combines expert knowledge with advanced data-driven techniques to overcome limited or incomplete data. By modeling hypothetical paths using advanced conventional machine learning in combination with symbolic methods, Beyond Limits AI is better equipped to more accurately predict scenarios or outcomes and then propose optimal courses of action, yielding more confident decisions even under less-than-ideal data circumstances. 


+ Increasing Prognostic Accuracy


Advanced artificial intelligence’s predictive capabilities balance out the inherent uncertainties associated with operating under inaccurate or incomplete planning models. This solution helps mitigate risks from unforeseen, potentially hazardous, workforce and asset incidents by increasing prognostic accuracy.



Beyond Limits Cognitive AI for Well Health


Beyond Limits Cognitive AI for Well Health is another great example of an artificial intelligence solution making everything easier for the industry. Keeping industrious, valuable wells in optimal health is of the utmost importance. Executing superior diagnosis and prognosis, detecting critical errors, and predicting future problems are vital capabilities for the sector.


More often than not, a narrow field of principal experts is depended on to fill the aforementioned responsibilities, make crucial modifications, and disentangle any and all concerns. However, this is not a scalable answer when each vital well demands day-to-day expert decisions and mediation. Unexpected issues from wear-and-tear or otherwise, that go undiagnosed over a span of time, have huge potential to harm equipment and processes. This innately leads to downtime, drops in production, and surging maintenance costs.


Beyond Limits’ solution is a cognitive decision support system designed to assess data-driven, physics-based material in conjunction with human expertise and best practices. It can translate domain expertise into ideal plans and recommendations to prevent formation damage and ensure well integrity by forecasting and moderating risk while helping stakeholders uphold production objectives. Such capabilities lead to pointedly reduced downtime, optimal well operation, streamlined production planning, and decreased costs.



A Partnership to Keep Your Eye on in this Era of Digital Transformation


High-performance AI fast-tracked by GPUs encourages energy companies to enhance strategies to maximize financial returns, accelerate time to decision, improve decision-maker confidence levels, and optimize well productivity while helping to reduce the environmental impact and hazards caused by drilling.


The adoption and implementation of such AI workflows represent both the vision and the outlook for the upstream oil and gas industry. More importantly, the framework built by Beyond Limits, in partnership with NVIDIA, will continue developing exponentially within this space. With applications that have use cases across the downstream sector, and the energy industry as a whole, alongside power/utilities/grid management, as well as medical/healthcare fields – and a host of others – a partnership as powerful as the combined prowess of these two technological giants has the potential to help almost any enterprise look far beyond its limits.