Of all the changes coming in the oil and gas industry, artificial intelligence is likely to be one of the most significant.
“I think we are living in a truly unprecedented time that AI technologies have the potential to transform the exploration and production processes,” said Ulisses T. Mello, director of IBM Research – Brazil (BRL).
Mello is an IBM distinguished engineer and led the Smarter Natural Resources and Discovery area at BRL until 2012. He also is an IBM Global Research executive for the chemicals and petroleum industry sector.
IBM has developed an AI-based adviser to aid and enhance seismic interpretation in partnership with Galp, a Portuguese energy group with a global footprint.
The tool can facilitate creation of enhanced geological models, risk assessment of new prospects and optimization of the placement of new oil wells, Mellow said.
Petrogal Brasil and IBM Research-Brazil undertook the three-year research project under the Brazilian National Petroleum Agency R&D incentive regulatory framework.
Exploration “is very knowledge intensive and the way every expert does his or her work varies significantly, depending on the expert´s previous experience and his or her tacit knowledge,” said Mello.
“This project is not about pure application of AI technologies (like machine learning and deep learning). It has a strong component of Knowledge Representation and Reasoning, as well as the use of continuous learning from the professionals,” he added.
“I would say that capturing and normalizing the level of knowledge involved in seismic interpretation was a big challenge,” Mello said.
“Relatively reduced data volume was also an issue,” he added.
While the oil and gas industry as a whole has massive amounts of data, the amount of information available on a particular asset can be limited in most companies. To get good results, this data must be augmented with other forms of knowledge, Mello explained.
Easing the Workload
The new tool uses AI and other state-of-the-art technologies to interact with scientists.
The scientists can use the tool to create and enhance geological models, devise faster and more efficient risk assessment and optimize placement of new wells, he said.
The tool uses knowledge gleaned from previous interpretations and experiences that are captured on a practice-driven knowledge representation system. It uses new data and interaction with users to learn and continuously improve its capabilities.
“The prototype integrates relevant information from multiple sources, including seismic images, academic papers, notes, and reports. Using AI techniques, it presents suggestions to geoscientists with supporting evidence,” Mello said.
He said the tool has been well accepted by geoscientists working on the project.
“Most scientists tend to have a workload that involves both mechanical and intellectual work. Most would prefer to work on the intellectually challenging work. The adviser helps in both ways by accelerating the completion of the mechanical work and pointing out information and knowledge that could be helpful for a more comprehensive, efficient and rewarding work. With that understanding, most of the users that we have worked with us in this project have not only accepted but embraced the technology. This technology is coming to help and improve productivity and effectiveness, and most professionals really value that,” Mello said.
He said the tool also can help young professionals develop their expertise more quickly and can help avoid individual bias in interpretation. It also helps users compare information from different projects, different geographies and different decades.
Evolving the Technology
Another goal is to make the tool easily accessible.
“The technology is supposed to be a sidekick that can be embodied using some rich forms of interfaces, ranging from mobile apps to chatbots. At this point we have a traditional web-based prototype and we are doing digital ethnography studies to define what is the best way to have the tool being consumed by scientists,” Mello said.
Mellos aid the adviser has been applied to several real cases in Brazil and worldwide.
“We have early indication of very significant gains. Now we are formalizing a process for assessment and benchmarking to have the value quantified with rigor using well defined business and technical key performance indicators,” he said.
Mello said the researchers hope to add many more capabilities to the adviser.
“The exploration process has phases in which very limited data is available and as the processes advance, more data and knowledge are acquired. The value of information varies along the evolution of the asset, so there is a lot of room to use AI techniques such as Transfer Learning from mature analog assets to less mature ones. This a direction that we would like to grow,” he said.
He said the specific project with Galp had about 10 people assigned to it but suggested that figure could be misleading because the project leveraged the support of technologies developed by IBM research and Watson. IBM research employes about 3,000 scientists, with about 100 people in natural resources, including oil and gas, Mello said.
Mello said AI is poised the transform the oil and gas industry.
“We are in the infancy of this transformation. There are substantial challenges ahead of us but some of the benefits of narrow AI are ready now. Some may see this as hype, but this is coming for real,” he said.