Research by Peter Kaznacheev
The Next Frontier of Innovation After Shales. Application of Artificial Neural Networks in the Oil and Gas Industry. Peter Kaznacheev (EUCERS Newsletter No.66, page 6)

Скриншот 2017-07-31 14.26.41

“The Next Frontier of Innovation After Shales. Application of Artificial Neural Networks in the Oil and Gas Industry”

By Peter Kaznacheev

Peter Kaznacheev is a UK-based non-resident Senior Research Fellow at EUCERS and Director of the Centre for Resource Economics at the Russian Academy for National Economy and Public Administration (RANEPA). He is also Lead Economist at Khaznah Strategies Ltd., a consulting firm that specialises in energy market research. Previously, he worked as a Business Development Advisor at BP, Assistant Representative at the Group of Eight (G8), and consultant with the World Bank. He received a Master’s degree in international economics from the Johns Hopkins School of Advanced International Studies (SAIS) and a BA and a PhD in political science and philosophy from Moscow State University.

The energy industry is in need of new technological solutions that would allow it to adapt to lower crude prices, increase efficiency and maintain operational safety and environmental security. One of the areas of intense innovation is the use of artificial neural networks. This article1 provides a brief overview of the three main areas where artificial neural networks are applied in the oil and gas sector, namely: interpretation of geological data, automation in field management, and market research.

A New Technological Frontier

The Japanese celebrity scientist Michio Kaku at a global energy conference in Abu Dhabi drew a possible picture of the future: “We are witnessing a new generation of oilfields. If something breaks down in the oilfield, your contact lenses will identify it and order the new part by simply blinking. Very soon we will have intelligent wallpaper which will be inscribed with artificial intelligence…”.2

That may sound more like science fiction but, in fact, artificial intelligence is already changing the shape of the industry. For many oil and gas producers, innovation is not a trendy buzzword anymore but a matter of survival. Today, when the oil price is less than half what it was in 2014, there is very strong demand for technologies that would allow companies to cut costs and dramatically increase efficiency.

Over the last three years, United States’ shale producers have taken the lead in transformation. They demonstrated impressive adaptability – most observers expected a much stronger decline in shale production. The unexpected resilience of U.S. oil output was the result of two main factors: intense cost cutting and improvements in efficiency. In 2015 – just in three years – upstream costs were brought 25%-30% below their 2012 levels.3 Efficiency improvements included, among others, shorter drilling and completion times, and increased well productivity.

At the moment, the global oil industry beyond U.S. shales is in desperate search of new technological solutions that can boost efficiency. Hence, the time for advancing artificial intelligence is arguably more favourable right now than ever before. Artificial intelligence is used in various segments of the energy industry value chain. For the purposes of this overview we will focus on three areas where, in our opinion, artificial intelligence is adding the most significant value. These three areas are: interpretation, automation and research.

Interpretation: Geological Data Analysis

Artificial intelligence comes in different varieties – Support Vector Machines, genetic algorithms, Bayesian networks, fuzzy logic etc. – but the most popular among them are artificial neural networks (ANNs). At their core, artificial neural networks are a mechanism of information processing – a computer system with many connected and interacting processors (artificial neurons) which is capable of learning. The advantage of ANNs is that they can handle large volumes of complex multi-format data (including missing and erroneous observations), work with nonlinear relationships, analyse non-stationary and volatile data series, adapt to changing conditions, generalise and learn on past data.

A schematic depiction of an artificial neural network

The area where ANNs were first applied in the energy industry almost 30 years ago was interpreting data acquired through geological exploration. It includes both imagery from seismic studies [Kuroda et al., 2012] and data from well logs. Neural networks are used to analyse various geophysical parameters and reservoir properties, such as porosity and permeability [Huang, 1996], water saturation [Nakutnyy, 2008], and fluid contents [Akinyokun et al., 2009]. Such data evaluated with the help of ANNs is then used for digital reservoir modelling.

Some studies demonstrate that digital modelling of soil layers using artificial neural networks can lead to very high levels of prediction accuracy about hydrocarbons in geological formations – as high as 90% compared with data from test wells [Choobbasti et al., 2015]. Efficient data interpretation saves costs due to fewer wells that need to be drilled and less time spent on data interpretation by geoscientists. The use of neural networks reduces the cost of research, accelerates geological evaluation and improves the accuracy of forecasts.

Automation: Digital Fields and Monitoring

At the next stage of technological innovation, the application of ANNs went beyond geological data interpretation and into field operations. Automation in field management allows to both reduce costs of production and to monitor field development in order to maintain safety and environmental standards. Some of the areas where ANNs help to reduce human involvement in the process include: monitoring of the drilling process, wells construction and well testing, analysis of hydraulic fracturing, optimising gas lift, monitoring oil production [Bello et al., 2015].

Several major oil companies have introduced holistic solutions to automation and developed what is often referred to as “digital fields”. International oil companies have invested significant resources in such technologies and even given them brand names: for example, Shell develops “Smart Fields”, and BP – “Fields of the Future”. Some national oil companies have followed suit and advanced digital field solutions – for instance, Saudi Aramco and Petrobras. In Russia, a joint venture between Shell and Gazprom Neft operating the Salym group of fields in West Siberia applied ANN-based automation and digital control of wells and reservoirs. As a result, operating costs were reduced, downtime was lowered, and production increased at a rate 2-2.5% a year.4

Automation is also used beyond field management – for instance, in oil and gas transportation ANNs are applied to corrosion monitoring, leakage detection and pipeline diagnostics. All of this contributes not only to gains in efficiency but also to accident prevention and increased operational safety.

Research: Planning and Strategy

The uncertainty which followed the oil market shift of 2014 created demand for deeper structural analysis of energy markets and new approaches to forecasting. So far, artificial intelligence methods in market research have been primarily the domain of traders – mostly in price forecasting of highly volatile products (stocks, commodities, derivatives etc.). More recently, ANNs are starting to make their way into the realm of oil and gas companies.

One of the main advantages of analytical methods based on artificial intelligence is that they are better capturing the non-linear nature of market behaviour than traditional statistical models. ANNs are also more adaptable to changing market trends; they are more tolerant to errors and incomplete data sets; and they have the ability to learn based on new available data [Sehgal et al., 2015]. In addition, machine learning algorithms can be applied to big data, for instance, as semantic analysis of news reports or so called “refined text mining” in order to determine changes in market sentiment and how such mood swing among investors can affect the price trajectory of curtain commodities [Yu et al., 2005].

The application of neural networks to market research in oil and gas companies has so far been less visible than in data interpretation and automation (as described above). But it carries a lot of potential. It elevates artificial intelligence to a new level: from an operational instrument in exploration and field management to an analytical tool for corporate strategy and planning. More accurate forecasts of supply and demand of crude oil, oil products and natural gas can have a direct positive effect on oil companies’ performance and strategic development. In addition, in the public sector the use of ANNs could allow to better estimate the future energy balance and make forecasts that strengthen supply security in the long run.


As there are fewer easily reachable oil deposits left worldwide, oil companies have to move into areas where hydrocarbons are contained in complex geological formations several kilometres underground or deep beneath the ocean floor. This is becoming ever more challenging at a time of lower oil prices. The use of artificial neural networks allows companies to increase efficiency, cut costs, and strengthen operational and environmental safety. The time is right to advance artificial intelligence as the energy industry is in search of innovative technological solutions that would help adapt to the new reality of cheaper oil.


Akinyokun O. C., Enikanselu P. A., Adeyemo A. B., Adesida A. Well log interpretation model for the

determination of lithology and fluid contents. The Pacific Journal of Science and Technology, 2009, vol.

10, no. 1, pp. 507-517.

Bello O., Holzmann J., Yaqoob T., Teodoriu C. Application of artificial intelligence methods in drilling

system design and operations: A review of the state of the art. JAISCR, 2015, vol. 5, no. 2, pp. 121-139.

Choobbasti A. J., Farrokhzad F., Mashaie E. R., Azar P. H. Mapping of soil layers using artificial neural network (case study of Babol, northern Iran). Journal of the South African Institution of Civil Engineering, March 2015, vol. 57, no. 1, pp. 59-66.

Kuroda M. C., Vidal A. C., Almeida de Carvalho A. M. Interpretation of seismic multiattributes using a neural network. Journal of Applied Geophysics, 2012, vol. 85, pp. 15-24.

Nakutnyy P., Asghari K., Torn A. Analysis of waterflooding through application of neural networks. Conference Paper, Canadian International Petroleum Conference, Calgary, Alberta, 17-19 June, 2008.

Sehgal, N., Pandey, K.K. Artificial Intelligence Methods for Oil Price Forecasting: A Review and Evaluation. Energy Systems, 2015, Vol. 6, No. 4, pp. 479-506.

Yu, L., Wang, S., Lai, K.K. A Rough-Set-Refined Text Mining Approach for Crude Oil Market Tendency Forecasting. International Journal of Knowledge and Systems Sciences, 2015, Vol. 2, No. 1, pp. 33-46.


The views expressed in this Newsletter are strictly those of the authors and do not necessarily reflect those of the European Centre for Energy and Resource Security (EUCERS), its affiliates or King’s College London.


1 This article is an updated and shortened version of a study published in October 2016 in Russian: Kaznacheev, P., Samoilova, R., Kjurchiski, N., Improving Efficiency of the Oil and Gas Sector and Other Extractive Industries by Applying Methods of Artificial Intelligence. Economic Policy Magazine, 2016, No 5, pp. 188-197. (In Russian: Primenenie metodov iskusstvennogo intellekta dlya povysheniya ehffektivnosti v neftegazovoj i drugih syr’evyh otraslyah // Ekonomicheskaya Politika). Online:

2 Graves, L., Physicist sees augmented reality playing key role in hydrocarbon industry’s future. The National. 2014. Online:

3 A joint study by the U.S. Energy Information Administration and IHS

4 Smart Fields of Salym // Russian Oil and Gas Technologies Magazine, 2014.