Published: 2015-Sep-11 Author: Sebastian Gault
Artificial intelligence is shaping the future of energy, but don t forget the risks
Data is the new oil. One often hears that saying these days, but what does it really mean? For some people, oil has long been the ultimate symbol of wealth and power, but some like to suggest its role is being replaced by information.
Economic clout is shifting from petro-capitals like Houston and Riyadh to tech centres like Silicon Valley and Shenzhen. On June 30, Apple s market capitalization exceeded $724 billion, effectively double that of the world s second-largest company, ExxonMobil. Google, Microsoft and IBM look down on energy giants like PetroChina, Chevron and Royal Dutch Shell. Others interpret the expression more concretely. They point out that to optimize any upstream, midstream or downstream operation, you must get better at collecting data and extracting value from it. The reason? Better use of data means more oil at a lower price. More and more, oil and gas companies are morphing into IT companies. With producers desperate to cut costs as commodity prices languish, tools from the realm of big data, artificial intelligence (AI) and robotics present exciting opportunities. So hold onto your hats this isn t your father s oilfield anymore.
automation[1] to autonomy
Machines are getting faster and smarter. Oxford scholars Carl Frey and Michael Osborne famously estimate that by 2035, 47 per cent of all American jobs could be automated.
The oil and gas industry has mixed feelings about this prediction. For instance, Mark Reese, president of rig solutions at National Oilwell Varco, sums up the industry s old view on just about any industry problem: We need more and more people and experience, and that s the only way to accomplish this task. The cover story of the August issue of The Atlantic warns of a world without work and shows where automation[2] is taking us. It s not all good. Until recently, the problem in the energy sector was that there simply weren t enough skilled people around high labour costs soured the economics of many a project. But things have changed. Economic malaise and technological automation[3] might put oil and gas employment in double jeopardy. In this new era, industry professionals could end up going from asking whether their jobs are safe to whether their jobs will even exist in a few years.
We re thankful when smart technologies help us make better decisions. But what if these technologies become so smart they start replacing humans en masse? There are two main views about the future of AI-driven automation[4]. One view sees technology such as data mining, advanced analytics and visualization tools augmenting the work of professionals be they petroleum engineers, geophysicists or financial modellers. The second view is gloomier; it sees automation[5] technologies developing along an unstoppable pathway that leads to the replacement of human labour by self-learning machines. Take a hauling vehicle for example. If you outfit it with anti-locking brakes, you ll improve the driver s performance. But if you buy a self-driving truck, you can do without him.
Data-driven world
To explain our brave new world of data, Microsoft guru Jim Gray talks about the evolution of scientific discovery in terms of four paradigms. Science was initially about building theories, and then later the experimental method became the focus. With the advent of computers, the next paradigm involved scientists using computational models to advance knowledge. And now, during the last decade, we ve entered the fourth paradigm of data-intensive scientific discovery.
The data used to serve us; today we serve the data, says Shahab Mohaghegh, professor of petroleum and natural gas engineering at West Virginia University. His company, Intelligent Solutions, applies cutting-edge AI systems and predictive analytics to the exploration and production industry. The approach behind his firm s reservoir-modelling software reflects the data-centric approach of fourth-paradigm thinking: We let the wells and the reservoir speak for themselves and impose their will on the model, instead of imposing our current understanding of the geology and physics on the model, he says. The model is then validated by testing it with blind data during post-modelling analysis. Before the fourth paradigm era, data was sparser. You worked with statistical sampling, stochastic modelling and probability theories if you wanted to predict the impact of your decision on the future.
Now in the so-called big data world, information is proliferating at a geometric rate. There are more types of data at greater volumes, and it s coming at us faster than ever. Programmers have met the data challenge by turning to human cognition for inspiration.
The mind-inspired fields of artificial neural networks, fuzzy logic and genetic algorithms are better suited to cope with what Mohaghegh calls nature s complexity and non-linearity. Tolerant of imprecision and uncertainty, soft computing methods as opposed to uncertainty-intolerant hard computing are crunching the numbers from surface and sub-surface data streams, shortcutting resource-intensive analytical processes, revealing insights in a fraction of the time. Describing his work as science and IT meet operations, Dariusz Piotrowski is a partner in strategy and analytics at IBM Calgary, providing big data solutions for the energy sector.
We are finding that our oil and gas clients operate in areas where the underlying variables for the models are in a state of constant flux, he says. So the models have to evolve in time, and that s why we design them to be machine-learning models.
Over time, as new and improved data arrives, the hypotheses can be [automatically] verified with the actual results, and the models keep improving.
The leading producers are enthusiastically instrumenting their oil and gas fields with all manner of sensors. As Scott Fawcett, a global account executive at GE Heavy Oil Solutions in Calgary, points out, sensors are now everywhere. Just downhole there are flowmeters and pressure, temperature, vibrations gauges, as well as acoustic and electromagnetic sensors. Employing predictive analytic software and self-learning models, operations personnel can monitor the data feeds, helping them forecast pending breakdowns whether drilling[6] equipment, compressor[7]s or valves and schedule preventative maintenance. Turning to reservoir management tools as examples of intelligent systems, Mohaghegh explains their pattern recognition power to predict hydraulic fracture details in shale. For these tools, advances in sensors and other devices enable the production of more types and greater volumes of hard data proppant and fluid type, amount, injection and breakdown pressure, and so on so that the more subjective soft data conductivity, simulated reservoir volume, height and width plays a lesser role. The future is all about smarter and smarter systems that use more and more accurate measurements.
Smart fields, smart organizations
Just because an operation is automated does not mean that it is smart, says Mohaghegh, warning that automated fields may have lots of sensors and other jazzy hardware, but AI is what makes this stuff intelligent. To fully realize this, one must subscribe to a complete paradigm shift.
Remote SCADA and video surveillance are just small parts of creating smart oilfields.To shift with the times, oil and gas companies need to re-think their organizational structures. Traditionally, the different links in the value chain of an oil and gas company land, regulatory, pad construction, drilling[8], completions, operations have worked more or less independently. The trend is now toward more integrated operations and a breakdown of these boundaries.
Across the industry, you can see experimentation in new structures, processes and decision rights, according to a recent report, Big Data Analytics in Oil and Gas, by Bain & Company. Executives are trying to optimize the organization model to encourage timely, cross-functional collaboration and put the right data in the hands of decision makers [at the right time]. If the segments have different IT systems and incompatible data models, it becomes very difficult to have a clear integrated view of what is happening in the field.
As a result, some companies are moving away from the functional organizational model toward an asset-based one where all field functions report into one geographically-based organizational structure.
We are at the dawn of a new generation of business systems, says Kadri Umay, director of industry technology strategy at Microsoft, in an email. With the advent of unlimited computing capacity in the cloud as well as new rich data platforms that have the ability in real-time to reason over data, we now can build systems of operational intelligence. Asked about the trend toward new organizational models, Umay explains his company s Azure platform. The vision behind this platform is about building a fully connected upstream value chain that provides a feedback loop with exploration, appraisal, development, drilling[9] and production.
This feedback loop helps us take all of the digital information we have and make it much more real-time in terms of how we can drive both performance and efficiency, he says. As an example, he describes one of Microsoft s connected rig solutions, where we are ingesting real-time rig data and analyzing it utilizing stream analytics and machine learning models to provide real-time alerts for stick and slip detection while maximizing the rate of penetration.
I, Robot
Several energy companies are doing research and development on technology that aims to remove people from the most repetitive, dangerous and time-intensive parts of oilfield work. Robotic drilling[10] Systems in Norway, for instance, has collaborated with NASA on the design of an autonomous robotic drill floor. Roald Valen, a control system manager at the company, sees the integration of autonomous abilities into machines as reducing expensive human error.
Robotic drilling[11] Systems of Norway has developed an autonomous robotic drill floor. The first reason to have the autonomous function is to prevent the big stops, the critical stops, he says.
New technology is making drill bits more intelligent and able to respond instantly to conditions they encounter, such as extreme temperatures or high pressures. Apache aims to produce a drill bit that has the ability to make its own decisions and communicates with surface equipment that controls drilling[12] speed and direction. Eric van Oort, a professor of petroleum engineering doing research in the area of drilling[13] automation[14] at the University of Texas, real-time operations monitoring and remote command and control, sees a day when automated rigs roll onto a job site based on satellite coordinates, implant 14-story-tall steel reinforcements, drill a well, then pack up and move to the next site. A few years ago, few people thought the idea of self-driving vehicles zooming around our city streets was nothing more than science fiction. Today, most people see it as inevitable. It s bound to become popular first in places like San Francisco, but the idea is taking root up in Fort McMurray, where oilsands companies are, by necessity, very open to exploring futuristic ideas if they can trim operating costs.
For some time, rumours have circulated that Suncor prepared to follow the example of global mining firms like Rio Tinto and BHP Billiton, which are implementing autonomous haulage systems. Sneh Seetal, Suncor s media relations manager, is a little cagey about answering questions related to this sensitive subject. She emphasizes that, yes, the technology is proven, but it has worked in other parts of the world where there are hard rock mining conditions, such as Chile and Australia.
Our oilsands conditions are soft rock mining conditions, she says. A commercial and sustainability evaluation is ongoing at Suncor. Testing is expected to begin in the fourth quarter of 2015. Bringing the future to an oilsands mine, soft rock and all, is clearly not a plug-and-play affair, she says. If we proceed with the technology, she cautions, implementation would take place over a number of years.
The big crew change
Robots and other autonomous systems are on the up and up, and the rise of AI is coming at a time when a whole generation of oil and gas employees is getting ready to collect their gold watches. When the big crew change takes place, what happens to all that valuable grey matter? Piotrowski believes IBM s cognitive systems offer the means to capture within the corporation the expertise and accumulated know-how of these outgoing professionals. The end product creates a kind of at-the-ready armoury of information tools that allows every engineer to capture the [legacy] knowledge and be the best engineer in the field.
Umay likewise sees the urgency for the oil and gas industry to respond to the big crew change. He feels the only way to do so, without introducing operational inefficiencies and significant risks, is to introduce expert systems and automation[15].
It s no longer sufficient to provide red and green lights, he says. We also need to provide actionable insights on what the problem is and what are the best actions to take based on experience. The capturing and refining of knowledge and making it accessible becomes very important. Intelligent data-driven drilling[16] systems are becoming more and more important. Microsoft is now collaborating on the development of smart drilling[17] systems, which provide real-time guidance based on analytical models. Still, the analytic talent required to develop, operate and maintain such intelligent systems is a scarce resource at the best of times. Typically, geologists and engi neers don t have expertise in the latest analytic tools and the learning curve is steep. What s more, firms can t turn to the IT technicians responsible for the firm s architecture. As cloud and open-source architectures become increasingly popular, even they are often not adequately familiar with the latest technologies. And yet, faced with these challenges, the oil and gas industry is certainly not gripped with a sense of despair. Quite the contrary.
Oil and gas has long been seen as the quintessential big data industry. In fact, it has been at the forefront of applying computer technology, going back to the 1970s. It can be expected that as AI technologies continue to reshape the economic landscape, hydrocarbons and information will keep mixing and the boundaries between oil and data will blur.
References
^ Find more Automation services on COSSD! (www.cossd.com)
^ Find more Automation services on COSSD! (www.cossd.com)
^ Find more Automation services on COSSD! (www.cossd.com)
^ Find more Automation services on COSSD! (www.cossd.com)
^ Find more Automation services on COSSD! (www.cossd.com)
^ Find more Drilling Products & Services on COSSD (www.cossd.com)
^ Find more Compressor services on COSSD (www.cossd.com)
^ Find more Drilling Products & Services on COSSD (www.cossd.com)
^ Find more Drilling Products & Services on COSSD (www.cossd.com)
^ Find more Drilling Products & Services on COSSD (www.cossd.com)
^ Find more Drilling Products & Services on COSSD (www.cossd.com)
^ Find more Drilling Products & Services on COSSD (www.cossd.com)
^ Find more Drilling Products & Services on COSSD (www.cossd.com)
^ Find more Automation services on COSSD! (www.cossd.com)
^ Find more Automation services on COSSD! (www.cossd.com)
^ Find more Drilling Products & Services on COSSD (www.cossd.com)
^ Find more Drilling Products & Services on COSSD (www.cossd.com)
The post Artificial intelligence is shaping the future of energy, but don’t forget the … appeared first on Trucker World.