Introduction
Researchers agree that one of the most critical technologies that we have today is artificial intelligence (AI) (Brynjolfsson & Mitchell, 2017; Fadziso, 2018). It has entered various industries and created considerable potential for growth and innovation (WIPO-World Intellectual Property Organization, 2019; Bughin et al., 2018). AI has triggered significant changes in healthcare, retail, media, transportation, and finance and has also transformed competition rules in various sectors. According to Iansiti and Lakhani (2020), unlike in the past, when industries relied on traditional business processes centered on human action, they now create value by applying AI solutions. The value creation process is driven by advanced algorithms which rely on enormous datasets; they receive a continuous supply of new data. Whether you consider how Amazon sets the prices for its once-in-a-lifetime offers, Yandex closes its car deals, or Alibaba prioritizes its adverts, you must think AI.
It is not only the digital-oriented companies that profit from AI.
Construction, mining, and oil and gas companies are the latecomers in this digitalization, increasingly depending on AI and machine learning (ML) based solutions frenzy (Kohli & Johnson, 2011; Kane et al., 2015). Scholars including Li et al. (2021), Samoun et al. (2019) and Santamarta et al. (2022) have noted that even though the first AI was applied in the oil and gas sector in the 1970s, is only recently that the industry began to take a bold look into the opportunities that come with integrating AI applications This new impetus in the in sector coincides with AI capabilities’ exponential growth and the sector’s movement towards the 4.0 concept that seeks to attain more excellent value by utilizing advanced technologies (Lu et al., 2019).
Essentially, oil and gas companies have the notoriety of quickly adopting new technologies, though they are slowly experimenting with and changing their business models (Tekic, & Koroteev, 2019). Therefore, AI and other digitalization efforts aim to reduce risks while accelerating processes and improving efficiency (Li et al., 2021, Lu et al., 2019, and Shafiee et al., 2019). This strategic analysis will focus on how AI transforms upstream oil and gas.
Oil and Gas Industry Sectors
The oil and gas sector is not only diverse but also very complex; the discussion will focus on the upstream segment of the industry. The oil and gas industry is divided into upstream, midstream, and downstream. The upstream entails the mining part, which includes exploration, field development, and crude oil and raw natural gas production. The midstream entails transporting the oil and gas, while the downstream entails the refinery processes like producing lubricants, fuels, plastics, and other petroleum products. This section highlights some of the points in the upstream activities where AI solutions have already been applied. Also, it highlights the results and the points where companies can use AI and the expected results.
The upstream consists of the production of crude and natural gas, searching for underwater or underground natural gas reservoirs, how the wells for exploration are drilled, and lastly, how companies drill and operate the wells that are used to bring the raw gas or crude oil to the surface. Essentially, the upstream is the most important of the three sections in the business, entailing the most capital investments, as pointed out by Shafiee et al. (2019). There are many uncertainties and risks that the relevant companies have to mitigate. It is imperative to handle the uncertainties when making decisions involving multi-billion investments focusing on where and how to invest in the next two decades. Traditionally, practitioners have relied on subjective perception and experience in making decisions. Because of the availability of data, it is becoming increasingly popular to use artificial intelligence (Ai) and machine learning (ML) in making decisions in the upstream sections.
Need for AI in the Upstream Sector
The difficulty in recovering oil and gas reserves in the recent past has necessitated the need for novel approaches to operations and business models in exploring and producing hydrocarbons to ensure profitability in the oil and gas industry. This fact applies to the brownfields (well-developed) and green fields (newly discovered) subsurface reservoirs of hydrocarbons. It is important to note that brownfields are generally extensive in size and require ample storage and transport properties. However, the quantity of recoverable oil and gas when the water flooding method is applied is small since most produce more water than oil. The companies are forced to spend vast amounts of money on well treatment, extra drilling, or field-scale enhanced oil recovery procedures to keep the production levels. The situation is no better for green fields; almost all the newly discovered oil and gas reservoirs are complex.
Different factors lead to a problematic green field: (1) location (some are located in harsh environments like the Arctic shelf; (2) others have complex geometry, for example, the rocks saturated with oil made up of thin and winding layers with many cracks; (3) some are found under the salt mineral and seawater whose thick layers hinder operations (like in offshore Brazil); (4) some have poor permeability (the oil and gas are immobile within the reservoir rock ((Lindholt & Glomsrød, 2012)). A company requires costly technologies to develop such a green field, which makes it challenging to make oil production profitable.
Application of AI in the Upstream Sector: Geological Assessment and Drilling
Four significant activities are upstream: geological assessment, drilling, reservoir engineering, and production optimization. For the geological evaluation, the tools applied include extracting geological information from well logs and an automated mapping tool that focuses on reservoir rock properties over a region with oil. It utilizes gradient boosting and non-gradient interpolation and optimization techniques and helps speed up manual mapping to just a few seconds from several weeks. It also defines hydrocarbon targets more accurately by removing the errors caused by wrong mapping (Mao et al., 2019, Cunha et al. 2020). For the drilling section, the tool uses real-time drilling telemetry that detects the drilled rock type and potential failure. Combining machine learning algorithms saves up to 20% of the time and 15% of the money during construction by maximizing the contact between the pay zone and the wellbore. Also, AI enhances the speed, safety, and precision of the drilling process (Hegde & Gray, 2017, Gurina et al, 2020, Klyuchnikov et al., 2019).
Application of AI in the Upstream Sector: Reservoir engineering and Production Optimization
In reservoir engineering, the tool accelerates conventional reservoir simulations using the deep neural networks approach. It accelerates the processes by a factor of between two hundred and two thousand, making it possible for the company to screen through more scenarios of field development, hence selecting the best alternative (Wood, 2020; Meshalkin et al., 2020, Temirchev et al., 2020; Simonov et al., 2018). AI is significant to the up-scaling process in reservoir engineering since there is no scientifically adequate framework for up scaling (Barker & Thibeau, 1997, Farmer, 2002, and Pickup et al., 2005).
The production optimization utilizes a data-driven tool that helps in objectively forecasting the efficiency of well-treatment campaigns. The primary well treatment procedures which are popular are the hydraulic fracturing and chemical treatment processes (Nolen-Hoeksema, 2013; Portier et al., 2007). The tool has an expert-based feature selection and gradient boosting, making it more than a hundred times faster in estimating the well-treatment effect enabling up to twenty percent growth of marginality of the campaign investments. Flovik (2022) noted that by using AI, it is possible to predict the efficiency of hydraulic fracturing jobs. If the algorithms are developed further through programming and optimization math, they can support full-scale recommending systems, which are crucial in assisting decision-makers in selecting specific well-treatment designs for a particular well and planning the appropriate well-treatment initiatives.
Impacts of AI Application in the Upstream Sector
AI application not only impacts cost and de-risking; it also extraordinarily impacts the safety measures of the operations on the oilfields, which involve dangerous maneuvers. Some risks involved in the oilfields’ processes include heavy equipment, high pressure, non-covered rotary equipment, aggressive chemicals, and high-temperature operations. Deep learning-based IT systems can help the safety officers identify a violation of safety protocols. For example, if a worker is not dressed properly for a given task, pattern recognition that entails deep learning and video streams can help raise the alarm. Also, predictive analytics can alert the operators on the status of the health of their equipment, enabling them to take proactive actions, preventing catastrophes that could have devastating consequences on the environment, health, and safety.
More than Just Technology: Need for Strategy
Tekic and Koroteev (2019) advise that for one to succeed in digital competition, one requires more than just technology. AI initiatives cannot fail due to bad algorithms but because the decision-makers lack vision, do not change the company’s business model or modes of operations, and have no data or any form of collaborative initiatives. Strategy, therefore, is a vital ingredient in driving the AI and ML transformation of the oil and gas industry (Kane et al., 2015). The top management must therefore be committed to the change to ensure success in AI and any other effort to transform the sector (Roberts. 1988). Lack of commitment is at the core of the challenges for the oil and gas industry, which is notorious for having a culture of risk-averseness and poor innovation management practices (Hajizadeh, 2019, Najjarpour et al., 2022).
The failure of experiences in the industry, like when oil prices went negative in 2020, should be able to trigger a business model transformation in the industry (Andriole, 2017). Otherwise, the companies risk misunderstanding and mistaking the impressive initial results of using AI as the final goal (Kane et al., 2015). Such a scenario would only lead companies to invest in technology, leading to non-transformative improvements (Tekic, & Koroteev, 2019). Currently, only about 8% of the companies are engaged in the core activities that enhance widespread AI adoption; most of those initiatives are ad hoc and focus on discrete business processes (Fountaine et al., 2019).
Major Assumptions for the Scenarios
Suppose there is no major technological breakthrough in the energy area and no social storm arises to distort the current energy demand trends in the next two decades. In that case, we can perceive three likely scenarios of how artificial intelligence could spread in the upstream sector of the gas and oil industry. We also assume that the Covid-19 pandemic has not irreversibly changed the industry’s main assumptions. We classify the scenarios as positive, realistic, and negative, describing the utilization of potential artificial intelligence developments as shown in the table below.
The three Scenarios
The first positive scenario whose key inputs include proper data platforms and approved data sharing. In this scenario, it is expected that in five years, there will be functional AI testing cutting through different inter-company challenges and people increasing trust the technologies. In 10 years, it is expected that AI tools will support all the decision-making processes which involve cost-intensive decisions, and between 40 and 50% of costs will be saved at E&P. Then, in 20 years, AI tools will support 90% of the operations’ decision-making. The oil century extends because of AI-assisted support of E&P margins.
This positive scenario will depend on the global spread that it is essential to share data across companies and borders. According to Nishant et al. (2020), it requires strong and committed leadership in the companies. If AI is absorbed, as shown in this scenario, it will reduce upstream’s negative footprint.
For the realistic scenario, the key input will be proper platforms for data-sharing platforms will be in set up even though data sharing will be a big challenge. There will also be elevated levels of trust in the technologies in five years. In ten years, artificial intelligence tools shall be fully embraced as the objective expert resulting in ten to fifteen percent savings. In 20 years, the hybrid tools, made up of AI and physics, will take over, and continued strategic investments using AI technologies will continue.
For the negative scenario, the key input is that countries and companies will have no agreements on data sharing. Within five years, AI tools will have poor overall performance since the personnel will lack the relevant training in data handling, which worsen the negative perception they have about AI. Then, in 10 years, artificial intelligence tools will be assisting in some local problems to a small degree but with no significant growth in the margins of the E&P processes. Ultimately, in 20 years, alternative energy sources like wind, solar, and nuclear will dominate.
Conclusion
The lifecycle of an oilfield should change significantly in the AI era compared to the pre-AI era. Two outcomes are expected; improved oil and gas recovery (IOR) and enhanced oil (gas) recovery (EOR) techniques. AI should be able to make exploration and development of active oil fields faster, cheaper, and safer with higher production margins running for long periods.
References
Andriole, S. J. (2017). Five myths about digital transformation. MIT sloan management review, 58(3). Web.
Barker, J. W., & Thibeau, S. (1997). A critical review of the use of pseudo relative permeabilities for upscaling. SPE Reservoir Engineering, 12(02), 138-143. Web.
Brynjolfsson, E., & Mitchell, T. (2017). What can machine learning do? Workforce implications. Science, 358(6370), 1530-1534. Web.
Bughin, J., Seong, J., Manyika, J., Chui, M., & Joshi, R. (2018). Notes from the AI frontier: Modeling the impact of AI on the world economy. McKinsey Global Institute, 4. Web.
Cunha, A., Pochet, A., Lopes, H., & Gattass, M. (2020). Seismic fault detection in real data using transfer learning from a convolutional neural network pre-trained with synthetic seismic data. Computers & Geosciences, 135, 104344. Web.
Fadziso, T. (2018). The Impact of Artificial Intelligence on Innovation. Global Disclosure Of Economics And Business, 7(2), 81-88. Web.
Farmer, C. L. (2002). Upscaling: a review. International journal for numerical methods in fluids, 40(1‐2), 63-78. Web.
Flovik, V. (2022). How do you teach physics to machine learning models?. Medium. Web.
Fountaine, T., McCarthy, B., & Saleh, T. (2019). Building the AI-powered organization. Harvard Business Review, 97(4), 62-73. Web.
Gurina, E., Klyuchnikov, N., Zaytsev, A., Romanenkova, E., Antipova, K., Simon, I.,… & Koroteev, D. (2020). Application of machine learning to accidents detection at directional drilling. Journal of Petroleum Science and Engineering, 184, 106519. Web.
Hajizadeh, Y. (2019). Machine learning in oil and gas; a SWOT analysis approach. Journal of Petroleum Science and Engineering, 176, 661-663. Web.
Hegde, C., & Gray, K. E. (2017). Use of machine learning and data analytics to increase drilling efficiency for nearby wells. Journal of Natural Gas Science and Engineering, 40, 327-335. Web.
Iansiti, M., & Lakhani, K. R. (2020). Competing in the age of AI: strategy and leadership when algorithms and networks run the world. Harvard Business Press. Web.
Kane, G. C., Palmer, D., Phillips, A. N., Kiron, D., & Buckley, N. (2015). Strategy, not technology, drives digital transformation. MIT Sloan Management Review and Deloitte University Press, 14(1-25). Web.
Klyuchnikov, N., Zaytsev, A., Gruzdev, A., Ovchinnikov, G., Antipova, K., Ismailova, L.,… & Koroteev, D. (2019). Data-driven model for the identification of the rock type at a drilling bit. Journal of Petroleum science and Engineering, 178, 506-516.
Kohli, R., & Johnson, S. (2011). Digital Transformation in Latecomer Industries: CIO and CEO Leadership Lessons from Encana Oil & Gas (USA) Inc. MIS Quarterly Executive, 10(4). Web.
Li, H., Yu, H., Cao, N., Tian, H., & Cheng, S. (2021). Applications of artificial intelligence in oil and gas development. Archives of Computational Methods in Engineering, 28(3), 937-949. Web.
Lindholt, L., & Glomsrød, S. (2012). The Arctic: No big bonanza for the global petroleum industry. Energy Economics, 34(5), 1465-1474. Web.