Well logging for oil & gas: machine learning approach

We use deep learning to to detect novelties and anomalies in geochemistry-production data


Steinbeis Consulting AI (STAI) is focused to solve different AI cases. The case study will describe problems & challenges in Oil & Gas which we have faced during a piloting.

Oil & Gas has it's own specific:

Data is highly expensive: in some regions, i.e. in North Seas or Siberia companies spent millions of USD to analyse potential spots.

The main business problem is loss of potential profit, followed by the following four factors:

• Decision making is based on data vendors research

• Interpretations are based on well field operator and external contractors

• Data interpretation is correlated withGIS expertise and the result is not always unambiguous

• Risk of skipping oil & gas saturated intervals

Here is coming a data-driven solution helps you select right data vendor. Machine Learning algorithms can predict accurate data with out spending extra money on research. There are two main things we have made during the pilot:

Find Similar Oil Locations

Data can help client to find similar oil wells and compare then in easy way

Active Learning

The more data we extract, the more piece of information it gives to us. It helps business gain more revenue and save time


I'm as an engineer, is always looking for new Oil locations. These locations has some geochemical properties, GIS data, and meta-information.

Engineers can see oil & gas spots

However, engineers are interested in some carottage, geochemical or any other research information which could be helpful to make a decision about quality and quantity of the given oil location.
Now, engineer can spent less time of making data analysis: AI is ready to help him. Under the interface surface Steinbeis Consulting Center AI has developed a breakthrough algorithm to predict, explore and find similar wells based on well-logging information.

Engineer can run measurements & compare wells

As a part of our data, we use all related oil well data: GIS, in-house measurements and incidents causes, meta-data, engineer's corrections. As a result, we extract all meaningful information from data in an explainable and adaptive algorithm.


• Reduce expenses on data vendors up to 65%

• Increase accuracy (>93%) of prediction having data from sensors

• Expected increase Oil & Gas extraction amount (>8%)

• Optimized II-types errors of finding oil wells

Industry 4.0 is not a novel topic: big players are already transform their industrial processes. Our team has wide expertise in manufacturing, oil&gas and machinery production. Interested in pilot? Contact us

Let us give you a personal tour of the product.

* We never share user details with third parties, period.
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.