Dr. Arunkumar Ranganathan | Infosys
Increased use of machine learning technology is driving growth in artificial intelligence (AI) adoption. By 2020, it is estimated the market will be worth over $5 billion. Oil and gas is one of the key industries expected to reap rewards from AI, though adoption is slow compared to other sectors.
The energy exploration sector is acutely aware of the need to identify new assets for exploration while increasing production at existing oil and gas fields. Crucially, companies must ensure the safety of personnel on sites and protect the environment at all times. Increasingly, manual processes are proving unsuitable in the field, but advanced technology solutions such as AI can support growth and eliminate health and safety issues to a large extent.
Oil is found in the minds of men
Risky as it may be, successful exploration requires a combination of visionary theory, technical innovation and commitment – and a bit of luck to be in the right place at the right time.
Geoscientists have the knowledge and experience to locate reserves. As resources become more scarce, AI systems hold the key to pinpointing new drilling sites. Geoscientists retain a wealth of information and asset knowledge but transferring this expertise to the wider organization, and the industry, is ineffective. AI systems play a crucial role in enabling scientists and engineers to remain productive irrespective of their experience.
AI offers a robust data interpretation process that can assist with critical knowledge transfer and decision making. It enables higher productivity and creates opportunities for advancement. While AI has been in use in the upstream lifecycle over the last two decades, there remains a skeptical belief amongst traditionalists that there is no better substitute for the human brain.
Applying AI in the exploration and production (E&P) life cycle
AI system consists of various tools: machine learning, fuzzy logic, artificial neutral networks, and expert systems. These systems transform data into valuable insight that can be applied across various stages of the E&P life cycle, including seismic, geology, drilling, petrophysics, reservoir and production.
Regardless of the business maturity level, AI systems have the ability to automate and optimize data-rich processes. They eliminate duplication of effort and mitigate business risks. As a result, they enhance productivity and minimise the cost of operation. Organizations that have reengineered their strategy and operational models to include AI elements have seen a positive business transform across the enterprise. Use of AI systems will remove redundancy, reducing the cost per barrel.
AI adoption begins with machine learning
Machine learning techniques in oil & gas will require deep insights into the process which it is designed to supports. For example, seismic processing requires the design of various data filtering techniques to enhance signal to noise ratio. These are used in both forward and inverse modeling. By adjusting the parameters within the learning sets and iterative routines, geoscientists can quickly eliminate repeatable procedures and apply this to real-time data for faster feedback.
Establishing processing parameters is an iterative process that improves accuracy. Geological modeling that relies on data alone is meaningless unless the geologist can apply their knowledge to refine the model. For example, the Kriging method is used to interpolate geological models between wells, which helps to refine the process by analyzing additional data input such as seismic lines, well log data, core and cuttings data.
The use of AI techniques helps ensure the process is repeatable in a consistent fashion, as well as enabling the automation of the actual physical task, freeing up staff resource. It also helps preserve the integrity of the analysis even if the geoscientist fails to acquire data in new wells.
Once the foundation is set, fuzzy logic systems can then be applied to support petroleum engineering processes including petrophysics, reservoir characterization, enhanced recovery, infill drilling and well simulation.
Eliminate costly risks in drilling
Drilling is a highly expensive and risky investment. Applying AI in the operational planning and execution stages significantly improves the success rate across the various stages of drilling including: well planning, real-time drilling optimization, estimating fictional drag, and well cleaning prediction. Additionally, geoscientists can better assess variables such as the rate of penetration (ROP) improvement, well integrity, operational troubleshooting, drilling equipment condition recognition, real-time drilling risk recognition, and procedural decision making.
AI techniques can also be applied in other activities such as reservoir characterization, modeling and field surveillance. Fuzzy logic, artificial neural networks and expert systems are used extensively across the industry to accurately characterize reservoirs in order to attain optimum production level.
Today, AI systems form the backbone of digital oil field (DOF) concepts and implementations. However, there is still a lot of scope to develop new techniques to optimize field development and production costs, prolong field life and increase the recovery factor.
AI techniques are most commonly applied in processing and interpreting well log data. In determining the parameters for multi-well processes, facies analysis is performed using quality reservoir data according to the number of wells covering the entire reservoir section. This process involves patter recognition, semi-supervised clustering and grouping.
The data interpretation highlights important geological features such as faults, folds, unconformity and boundaries. This information provided by the AI system is crucial because geoscientists sometimes fail to acquire this critical insight due to poor well conditions and other external factors.
AI techniques enable field experts to automate the system using available data to generate pseudo open whole logs in new wells, either using the Monte Carlo method or case hole logs. Training the system and feeding in well data will enable geoscientists to obtain more accurate log systems.
The use of AI in the oil and gas industry is gaining popularity but overall adoption remains relatively low compared to other sectors. There are plenty of opportunities to develop AI systems to further optimize, automate and improve business and operational efficiencies. Combined with the use of data analytics, there is much value to be gain in the AI market that industry leaders are yet to explore.