Aker BP Hackathon 2023: A celebration of data science!

Last week, Aker BP organized a Hackathon* in our offices in Fornebu. Six teams worked for 36 hours solving six challenges using coding and data facilitation: AkerAgent, FlashLiberation, LibraryChat, MaxOil, PDFinders and Wellbert. A panel of judges named three winners: Best project, Best community contribution, and People’s award.

Winner of Best Project, AkerAgent, was a chatbot with access to a number of Aker BP’s underground databases and APIs. With the same design layout as ChatGPT or Bing Chat, you could ask questions that were directly linked to Aker BP’s well data and internally developed machine learning models, and get quantitative answers about things like reservoir quality, stratigraphy, and even more involved topics like whether there were signs of hydrocarbons the area.

The winner of the People’s Prize was PDFinders. The team made better use of the invoice information we receive in PDF format from subcontractors. Today we get pages of details on invoices that are never used. Using python and several different libraries, they extracted tables from invoices in PDF format, and used this data to analyze new ways of categorizing cost.

Winner of Best community contribution was LibraryChat. The solution will make it possible to retrieve data from all Aker BP’s internal Sharepoint pages, make it searchable, and feed this information into ChatGPT, which will be able to provide detailed answers about what is published on Sharepoint.

FlashLiberation addressed phase separation into liquids and gas, which often occurs when hydrocarbons are produced and transported through pipelines. Calculations can be made for how much of the liquid will evaporate during such transportation – today this is usually carried out by third parties. The FlashLiberation project allows anyone in AkerBP to do these calculations and build new applications leveraging these calculations without the help of third parties.

MaxOil looked at how to maximize oil production by optimizing well controls using explainable and causal AI.

The Wellbert team investigated how recent advances in language models could be applied for analysing measurements taken in boreholes when exploring and producing hydrocarbons.
This paves the way for potentially leveraging the vast data we have on the NCS and building better predictive models.

– Events like this are important to use the knowledge and creativity of skilled employees to solve difficult and valuable challenges for Aker BP. We have a very strong environment for analytics and data science in Aker BP. In a Hackathon, the analysts build networks, and manage to contribute to developing more efficient ways of working in Aker BP, says Hege Fjell Urdahl, VP Digital Insight. She was also on the panel of judges during this year’s Hackathon, and is a sponsor of the Data Science environment in Aker BP.

* A Hackathon can be described as a social coding event, where computer programmers and other interested parties gather to either improve already existing software or build new software.