Artificial Intelligence Improves Business Outcomes for Offshore E&P
In offshore exploration and production (E&P) operations, data remain largely an untapped asset. Platform data are particularly far from being ready for any artificial-intelligence (AI) applications. However, the data landscape aboard an offshore platform provides a good starting point for learning how these applications potentially can transform the offshore E&P enterprise, whether for a single platform or an entire fleet.
Pumping Data to Nowhere
Despite platforms having streams, if not floods, of data about the condition of topside equipment as well as drilling gear, electrical submersible pumps, and other wellhead mechanisms, most experts agree that only a small fraction of such data is used. Compounding the problem is the fact that most data currently are siloed; their sources are sensors on the various components of platform equipment and tools of their suppliers. Data are thus further fragmented and isolated as they find their way into spreadsheets on individual personal computers and historian servers that often become final data repositories. And, if data are needed for decision support, the chores associated with collection, cleaning, and normalization can be enormously time-consuming, whether those doing the work are on a platform—where labor, living, and transportation costs are highest—or in onshore control rooms or offices.
Evolving Technology Solutions
Among the technologies offering the most promise in addressing these issues are advanced data analytics for so-called big data; accessible, scalable cloud platforms; and AI that drives machine learning (ML). Complementing these at the field level are increasingly “smart,” self-calibrating sensors that provide offshore equipment and components with the ability to communicate status to higher-level systems. In addition, well-established global standards are facilitating machine-to-machine communication. Finally, cybersecurity standards and layered, defense-in-depth models have grown in response to the increasing frequency and sophistication of cyber threats that continue to endanger critical infrastructure, especially energy.
AI and ML: A Closer Look
Data fuel most AI applications, with ML among them. The latter is an application that runs data through various statistical models to find patterns and, in effect, learn from the data and adapt its functions without specific programming. The more data processed, the smarter the program or machine becomes. With ML, users can find out what they do not know and refine what they think they know. When these applications operate, they strive constantly to quantify patterns that humans cannot detect in streams of a billion data points every second.
Benefits in the North Sea
How can these technologies benefit offshore E&P operators? One North Sea platform more than 100 miles off the Norwegian coast and just 450 miles south of the Arctic Circle is securely transmitting real-time equipment data more than 600 miles through an undersea cable network to an onshore control room. The data allow the company to monitor equipment conditions as if staff were aboard the platform. Instead of scheduled maintenance, however, remote condition-based maintenance (CBM), a more-precise, safer, and lower-cost approach, is conducted.
Maintenance can be conducted as needed, not as scheduled. The former approach reduces costly disruptions by extending maintenance cycles. As data accumulate, AI can be employed to look for anomalous machine behavior to conduct predictive maintenance. The number of platform maintenance staff personnel is cut in half, reducing the number of individuals exposed to platform dangers both aboard and en route. Costs are also lower because spare parts are reduced, as well as staff living and transportation expenses.
After its first full year of production in 2017, CBM contributed to a 70% reduction of personnel on the platform and a 30% operational cost savings.
Enabling New Business Models
Remote CBM illustrates just one facet of digitalization that AI can enhance, allowing offshore E&P operators or rig owners to optimize equipment performance while lowering costs. As stakeholders in the offshore E&P industry, origjnal equipment manufacturers (OEMs) also can extend the concept to deploy new business models that potentially can continue to lower costs in the face of lower-for-longer market-pricing conditions. Large compressors aboard platforms are a good example of this principle. They can be extremely expensive and, like most rotating equipment, complex to maintain. By applying AI across a vast data set drawn from hundreds or thousands of units in their installed base, an OEM can assess failure risk to its models accurately and assign a value-based price for buying compression but not compressors.
Two Strategic Considerations
Organizational Adaptability. Is the E&P organization set up to use AI and ML, especially with regard to the transformations that both can drive, such as eliminating functional silos, reducing latencies, and creating new responsibilities and accountabilities? With much of their baby-boomer workforce retiring, offshore E&P operators are losing valuable experience-based knowledge and learned intuition about platform equipment and circumstantial situations requiring decisions. This timing coincides with the rise of AI and ML applications, which can simplify the acquisition of needed ML proficiency and associated organizational changes.
The specific changes required of an organization depend on a company’s circumstances and current capacity to adopt new ways of working. The referenced North Sea operator has a publicly stated goal of lowering production costs to $7/bbl while pursuing a strategic policy of low-manned platforms to achieve that goal. Thus, the company’s organization is more likely to embrace change than those without such a clearly stated goal and strategy.
Organizational Discipline. Does the E&P organization have the strategic and tactical discipline needed to identify and address specific cases that can liberate business value, then pragmatically expand their application to other areas of operation? Best-practice deployments of AI and ML will seek rapid success through use cases that begin with relatively simple, achievable goals that can show immediate value and return on investment. A sustained focus is needed at both management and operational levels to drive these use cases to fruition and then identify new opportunities elsewhere in the offshore E&P enterprise. For all use cases, these questions must be answered:
What does this project aim to achieve? What does the organization aim to learn?
What are the project’s success criteria?
What organizational function(s) will employ the use-case outcome?
Do those function(s) know how to use this outcome?
In starting their journey, offshore operators can lay the foundation for AI and ML by starting with relatively easy use cases. These cases could involve monitoring energy consumption, motor temperatures, or motor vibration. The point is to become comfortable with channeling specific platform data into an analytic model for validation while building historical data sets that AI and ML can use.
This first step also helps in determining bandwidth requirements. For example, sending raw data to the cloud through satellite connections from the Gulf of Mexico, North Sea, or Persian Gulf can be expensive relative to onshore connectivity; in these cases, preprocessing the data using edge analytics can be more cost-effective. Edge analytics can also be required for extremely low-latency applications; data transit times to onshore processing facilities take too long.
It would then be desirable to identify use cases that deliver quick wins by providing greater operational transparency. Examples of these could be on/off visibility to such platform equipment as motors, pumps, and valves. Following these use cases are ones that, on the basis of growing data sets, engage basic analytics to identify parameter trending, threshold detection, basic anomaly detection, and performance prediction using simple inputs. If a pump motor’s current threshold is 30 amps and it operates at 15–20 amps, then platform operators can decide between the benefits and drawbacks of running the motor harder.
Finally, once large-enough data sets have accumulated, offshore operators can much more fully reap the rewards of AI and ML by applying advanced analytics methods. Use cases can include predictive maintenance, anomaly detection, pattern recognition, and root-cause analyses. These, in turn, can help to build an institutional knowledge base about operating infrastructure that can span a fleet of platforms worldwide, leveraging cloud technology for worldwide coverage to minimize latencies.
With this knowledge base, conditions can be monitored against common operating baselines, as well as ones adjusted for geographic and environmental conditions. An anomaly found on one platform can serve as an alert for other platforms across an operator’s fleet.
Compressor-Specific, Remote Diagnostic Capabilities
Four examples of compressor-specific, remote diagnostic capabilities are compressor-performance monitoring, seal-gas monitoring, gas-composition monitoring, and bearing monitoring. These data can fuel AI and ML models to improve decision support and business outcomes for offshore operators. When measurements of various parameters exceed their limits or show anomalous variances from expectations, the CBM-based system can alert human operators so they can examine the data and make decisions about root causes and appropriate responses. Also, when these parameters trend against as-designed or as-tested performance signatures of the equipment, the system can flag these for investigation.
Decommissioning Work To See Increased Spending; North Sea Region To Lead the Way
Even before the global pandemic impacted markets, decommissioning work in the North Sea region was expected to increase. One research group estimates global decommissioning projects could reach $24 million by 2024.
Mammoet Awarded Unloading, Transport, Installation Contract for Arctic LNG 2
The contract is Mammoet’s biggest in terms of aggregate tonnage. The Arctic LNG 2 project reached FID in September 2019.
Prudent Ways to Restart Big Projects in Post-Pandemic World
Delays and disruptions have stymied project plans in the past few months. How will companies prevent current stresses from morphing into catastrophic issues?
Don't miss out on the latest technology delivered to your email every two weeks. Sign up for the OGF newsletter. If you are not logged in, you will receive a confirmation email that you will need to click on to confirm you want to receive the newsletter.
20 May 2020
20 May 2020