The oil industry is known to overestimate the volume in prospects and understate the probability of success. This has been confirmed by the NPD’s analyses.
The sector has long been working to achieve more accurate estimates. Miscalculations can undermine exploration decisions, and thereby reduce value creation for society.
Resource Exploration Report 2018, NPD
Automizing formation evaluation, by integrating domain expertise and advanced data analytics. Think of the resulting algorithms as smart data robots …
… producing innovative products that will help subsurface teams doing their job faster, more cost-efficiently and with higher precision.
Cost efficiency and time savings
A petro physicist will evaluate one well manually in days. A robot will perform the same task in seconds.
Since robots are unbiased and also capable of estimating confidence levels of their predictions
Robots can identify complex relationships in data which can be exploited further and innovate work processes. Identification of source rocks for finding new plays is an example.
Robots are well suited for making databases containing key subsurface properties. Having access to such databases, better decisions can be taken earlier in the exploration process e.g. by ranking prospects based on reservoir parameters.
Domain expertise is crucial
Domain expertize can fill information gaps when limited data is available.
Domain expertise is important for understanding models and gaining reliability in their predictions.
Well data contains systematic errors that may arise from physical limitations, operational constraints, or geological reasons. Domain expertise is critical in identifying and dealing with such errors.
Domain expertise helps properly preparing the input data (feature engineering) to get most out of the available information.
I hold a PhD in physics, and has participated in succesful commercialisation of several products.
I started working for the E&P industry in 2004 when I joined Rocksource as a senior geophysicist doing exploration with CSEM. The last 10 years I have spent working in Atlantic Petroleum/MVestEnergy as senior petrophysicist.
I have published a number of articles and conference papers, including one about machine learning to be published at EAGE London 2019, and hold two patents.