Sign In
2021 Results:

Crude oil
Million Tons


Gas exploited
Billion M3


Billion KWhs


Thousand Tons


Million Tons

VPI applied machine learning model and artificial intelligence to predict fractures in the basement rock

The Vietnam Petroleum Institute (VPI) said that it has applied artificial intelligence (AI) and machine learning (ML) algorithms to readily identify the presence of fractures in basement rock with an accuracy of more than 80%, saving drilling time and cost.

The discovery and successful putting into production of oil reservoirs in fractured granitoid basement of Bach Ho and other fields in Vietnam have changed the conventional view, forming a new perspective in oil and gas prospecting and exploration in the region and the world as well. 

To identify fractured basement rock by conventional methods requires specialized tools, which may have to stop drilling wells temporarily, prolong the rental period of the rig. Identifying the fracture zone right during the drilling process will help the operator to recognize drilling risks and take appropriate treatments timely when drilling through the fractured and faulting zones, helping the process be safe and efficient, shortening the rig rental period, and thus saving drilling costs (the current rental is around 65 – 68,000 USD/day).

To optimize this process, based on drilling parameters and using machine learning models and artificial intelligence, VPI has studied and developed a solution to predict fractures in basement rock, the VPI Application of Fracture Prediction, which helps to pinpoint the presence of fracture systems based on real-time data.

Drilling parameters (such as torque, weight of bit, flow rate, standpipe pressure, rotor rotational speed, etc.) are used as input for the supervised machine learning algorithms; then the models will be screened, ranked, and evaluated to determine the optimal one for fracture prediction.

This prediction model was tested with drilling data of 12 wells from several fields that have analogous geological structures, resulting in a fracture prediction accuracy of over 80% for those wells. The output is displayed on the MLOps platform, which helps to improve drilling efficiency, particularly to determine accurately the depth ranges where fractures appear, supporting operators to make decisions promptly, saving drilling time and cost. The cost savings will depend on the specific drilling plan of each operator, treatment options during the drilling, if any, but according to primary estimations it could be up to hundreds of thousands of USD at prices of equipment and specialized manpower quoted in 2022.

Le Ngoc Anh – VPI Data Director said, the fractured basement rock prediction model exploits the availability of relevant data of the Vietnam oil and gas industry to develop optimal algorithms, creating novel values, ensuring the required confidentiality and safety.

Implementing the strategy of the Vietnam Oil and Gas Group on digital transition “to support and accelerate the business model transformation, optimize operation methods and improve operational management capacity", VPI has recently built and announced the Oilgas AI Ecosystem to integrate, represent and analyze in-depth oil and gas data for products such as crude oil, gasoline, LPG and natural gas.

The Oilgas AI Ecosystem creates a competitive advantage through optimal analysis, safe exploitation and usage of data, application of the models and algorithms developed by VPI scientists in collaboration with other leading experts in the oil and gas industry.

The energy market products and services available on the Oilgas AI Ecosystem will provide solutions to help companies make faster and more efficient decisions in day-to-day operation and business activities, as well as in the development of action plans and long-term strategies.


Results of fracture prediction by the VPI machine learning model and artificial intelligence