Training a supercomputer
In collaboration with IBM, at the Repsol Technology Lab, we are developing novel ways to apply the latest in cognitive technology to the oil industry on a worldwide scale. We are bringing together human intelligence and artificial intelligence to make the most accurate decisions.
With the Pegasus Project, we want to create a much more effective interaction between humans and computers.This tool is capable of analyzing hundreds of thousands of documents, reports, and data, and looking for trends and connections between them in order to propose the most appropriate hypotheses and solutions at any given time.
The world’s first cognitive laboratory
The multiple technological projects that we carry out at the Repsol Technology Lab to create oil models, like Excalibur, have improved the safety, efficiency, and profitability of Upstream operations. And now with the Pegasus Project, it’s possible to construct more fluid conceptual and geological models, and to explore hypothetical scenarios to guarantee the best decisions in the search for hydrocarbons . A new world of possibilities is therefore opened.
We are developing this technology jointly with a mixed team from Repsol and IBM. This team works in the first ever cognitive lab in the world owned by IBM in New York, and in the Repsol Technology Lab in Móstoles, located outside of Madrid.
Accessing new reserves is an increasingly difficult task. The subsoil is a great unknown. Drilling and making large financial investments are risky, difficult decisions. To minimize error and make the correct decisions at Repsol, we have decided to let technology help us to make those decisions. And the Excalibur Project is born as a result.
Excalibur uses mathematical methods to explore reservoirs in-depth prior to operating. We simulate them with 3D technology, making it easier to decide whether to drill or not, and if so, how to go about it.
We can perform a comprehensive evaluation of reservoirs in order to characterize their uncertainty and optimize development, minimizing the inherent risk. The tool also can be used to organize hierarchically a portfolio of reservoirs quickly and accurately and to identify new investment opportunities which are difficult to detect with traditional techniques.
When reservoirs have limited information and present a higher level of uncertainty, reservoir evaluation is more difficult and traditional techniques are less useful.
The Excalibur Project has been an outstanding source of innovation and intellectual property. In the testing conducted in the North Sea reservoir, Excalibur obtained an improvement of 9% over the best solution published to date by other companies and reputable research institutions.
With a small amount of data, the analogous reservoir model enables unknown properties of the field in question to be ascertained by comparing it with other known fields. Thus, it is possible to attain a probabilistic distribution of all the properties defining the geological model, which are otherwise difficult to access, consequently, increasing the rate of success.
The analogous reservoir model has been tested successfully at Repsol with several cases. The hierarchical organization attained showed a high degree of similarity with the analogous reservoirs previously provided by experts.
For a second module, new simulation techniques were developed to generate a group of 3D geological models. This technology includes the information obtained with the analogous reservoir tool and the information provided by the subsoil studies. With this information, it is possible to characterize the structural and petrophysical uncertainty of the reservoir, quantifying reserves more accurately than can be done with current methodologies.
Finally, based on the group of geological tests performed as well as the latest risk evaluation techniques, Excalibur can create a plan for optimal development of the field using new optimization algorithms created exclusively at Repsol. This provides a clear advantage over competitors as it can identify new opportunities for investment.