AI technologies like machine learning and large language models hold promise for complex and consequential upstream workflows, like seismic interpretation—if you can figure out how best to apply them. Here’s how ThinkOnward is approaching it.

In our recent blog post we identified crucial success factors for the open talent model (OTM) to work best in upstream oil and gas, and deliver the innovation needed to meet modern, global energy challenges. This post takes a closer look at one of those success factors: the technology environment, which, for optimum results, should be provided and maintained by the OTM provider.

The platform for performance and innovation

The ThinkOnward cloud-based technology workspace has these main purposes: 

• It provides a secure, reliable, state-of-the-art environment in which on-demand geoscience professionals can perform project work for clients, quickly and securely. The platform is equipped with leading oil and gas applications such as Petrel™, MOVE, PaleoScan™, and others.
• It is a catalyst for collaboration. Cloud-based workspaces enable real-time collaboration and seamless integration with other productivity tools, enhancing overall workflow and efficiency.
• It is a springboard for innovation. Artificial intelligence (AI) technologies such as machine learning (ML) and large language models (LLM) are implemented, tested, and experimented with to determine how to best apply them to upstream workflows.

The platform is developed, managed, supported, and enhanced by a diverse team of technology professionals. This team—a global brain trust of IT, AI, and oil and gas domain professionals—also provides training and support to the on-demand geoscience professionals who do project work on the cloud-based workspace.

The workspace provides opportunities where the team can experiment with new technologies, in various combinations, and collaborate with domain professionals to determine how to best apply them to real-world workflows and problems. Additionally, ThinkOnward collaborates with startup companies, making the platform a community and clearinghouse for new products and ideas, and a proving ground to experiment with these innovations for measurable advancements in oil and gas workflows.

When sufficiently mature, these technologies can be used in project work by the on-demand geoscience professionals.

Breakthroughs for seismic interpretation workflows

A crucial technique used in the oil and gas industry is seismic interpretation. The sheer volume and complexity of seismic data makes this process an excellent candidate for transformation by applying AI techniques such as ML and LLM.

Traditionally, interpreting seismic data has been a very human-centric process. In the early years it involved using printed maps, cross sections, and colored pencils to analyze huge volumes of processed data aiming to separate the signal from the noise to determine significant geological features critical to finding economically viable hydrocarbon deposits. This manual approach took 4 to 6 months or more to develop a first practical interpretation. This timeframe was, of course, dependent on the complexity of the geology, data quality, and scope.

In the early 2000s, digital tools provided some advancements, which significantly reduced time. However, the process was still very human-centric, and an interpreter was required to guide the interpretation process. Also, a lot of time was needed for data preparation, for example, to digitize the data so it could be loaded into the interpretation software. Another limitation was the lack of compute power for performing algorithms that could process such huge and complex graphical and spatial data sets.

The advent of cloud computing paved the way to use modern AI technologies to create advanced seismic data processing models. Machine learning uses huge volumes of seismic data to “train” models to learn what to look for, to analyze the data, find the signal in the noise, identify those faults and horizons, and then digitize them for analysis—tasks that computers can do far easier and faster than humans. Large language models specialize in processing human language, which allows professionals to talk about and to the model, to question the model, so that the LLM can learn more about the data and capture in those natural language dialogs the knowledge of the professionals asking the questions.

Then geoscientists can pump huge volumes of data through these new more sophisticated algorithms to produce interpretations faster. How much faster? Average times are about 2 to 3 weeks to produce a base interpretation. But speed isn’t the only benefit. 

“Whoever sees the most rocks wins”

This axiom of geology suggests that having a wide range of experiences can help to unlock understanding to solve problems faster. This is also the promise of AI: to perform tasks that computers can do faster and more easily than humans, to accumulate knowledge, reduce cycle times, and not only generate models faster, but produce greater quantities, varieties (such as models from different products or approaches), and reliability. This approach can make it possible for geoscientists to see more rocks and more models, which can ultimately lead to better and quicker decision-making.

Additional resources

• Identifying missed pay zones in an area of declining production.  For a mature area that had been in production for more than three decades, this 26-week, part-time project generated specific technical insights that were previously unconsidered by the client. Download the case study here.
• Unlocking exploration potential in an underexplored active margin. In this 10-week project, the insights generated by the team made it possible for the client to make informed decisions in less time, reducing decision time in the exploration screening process. Download the case study here.

Ready to unlock the future of seismic interpretation?

With so many other challenges to address, we're here to help. Our cloud-based technology workspace is designed to empower our on-demand geoscience professionals with state-of-the-art tools and AI technologies. Reach out to us today and let’s explore how we can collaborate to help you meet your subsurface interpretation challenges.