Arno van den Haak presented Amazon Web Services’ cloud-enabled technologies as a ‘paradigm shift’ for oil and gas that promises ‘50% reduction in IT cost, new solutions and different ways of working’. AWS poster child is Shell’s co-developed data platform for upstream innovation which has apparently solved the age-old problem of geoscientists spending ‘up to 70% of their time finding and collating data’. Shell has migrated its subsurface deep learning research program to the platform and ‘achieved a 3X performance improvement over internal benchmarks’. More from Amazon.
Very large data sets can be shipped on disk via Amazon’s Snowball service for delivery to a regional AWS facility and uploaded to client’s cloud portal. A more esoteric data collection option is the Insitu ScanEagle drone which can stay aloft over for 24 hours carrying video cameras, large format still imagers, lidar and other signal detectors. Live information is streamed to the enterprise operation center for real-time monitoring and information sharing with other operational functions.
Another AWS client is BP downstream which has ported its Schneider Electric Spiral Suite linear programming software to the cloud. For some reason, BP, ‘couldn’t take full advantage of the software’s potential power while running in its on-premises data centers’. The move to the cloud means that Spiral Suite analytics are now executed ‘in minutes, not hours’ and now provide a single source of truth for worldwide company data. More from Amazon. van den Haak concluded saying ‘We are at the dawn of a new era which requires a new way of working. It will be digital. The cloud is the new normal.’
Comment: While this may seem rather obvious, ‘digital’ is certainly not ‘new’. Back in the 1970s, seismics was digital, well logging was digital and digital GIS was starting too with Intergraph’s dual screen monsters. Widespread use of ‘digital’ in the office environment began back in the early 1980s with the arrival of the IBM PC. Even the cloud is not really new. Our first encounter with a cloud-based solution for oil and gas was in 1999 with our visit to the GeoQuest PowerHouse.
Bernt Tysseland presented Equinor’s new Integrated Operation Centre. The IOC vision is to safely produce to the limits, to prevent breakdowns before they happen and enable remote support from experts in production optimization and predictive maintenance. The IOC acts as a data hub for time series data coming in from offshore platforms and routed to Equinor’s cloud-based data platform. A monitoring and support dashboard offers machine learning smarts and ‘clear, common targets’ for a multi-disciplinary workforce. The overarching philosophy for the IOC mode of operations derives from the Gemba Walk lean management approach. Equinor’s Decision support tools include one of the widest screen Excel displays we’ve ever seen, showing Petex’ OpenServer well surveillance solution in action. Those interested in Equinor’s data acquisition should checkout Alfonse Reynes’ exploration of the now public domain Equinor Volve data set.
Chika Uduma outlined how Blue Gentoo combines physics-based modeling along with artificial intelligence in its HydraSens gas hydrate early warning system. Despite best efforts at improving software for simulating fluid conditions in pipeline systems, critical data on inhibitor concentration, water salinity and hydrate formation are often lacking. As industry moves into deep and ultradeep waters, ‘every offshore gas well has potential hydrate blockage problems’. Blue Gentoo advocates considering production optimization and hydrate management early in the planning processes. Blue Gentoo’s HydraSens ‘Smoke Alarm’ system for hydrates formation system has been used on Total’s Northern Underwater Gas Gathering, Export, and Treatment System (Nuggets) project as reported in SPE 166596.
Chris Lenzsch presented on ‘connected digital operations’ a Dell EMC/Arundo/Wipro joint venture that sets out to deploy artificial intelligence ‘at the edge’, more specifically with a Dell Edge OPC client running an Arundo edge agent. One use case is the PixelVelocity ‘Event Velocity’ monitoring technology that ensures that flare stacks are burning, beam pumps pumping or, using thermographics, can monitor tank levels from outside. Arundo Enterprise analytics also ran in an ‘augmented PLC/SCADA’ context for pump condition and performance monitoring.
Chris Tolleson (Wipro) reported on the ‘Intelligent Drilling Advisor’ which leverages a DARPA project on ‘explainable artificial intelligence’ (XAI). XAI aims to create a suite of machine learning techniques that produce more explainable models, while maintaining a high level of learning performance. These will help users understand and trust new AI solutions. Tolleson considers the goal of a completely automated drilling system as ‘very important’. Currently, the IDA is a BP-owned patent that exposes AI-derived results to scrutiny. The system distinguishes between measured and calculated values and provides reliability estimates for various drilling parameters.
Dell EMC’s David Holmes opined that petro-technical computing will evolve at a faster pace over the next five years than the previous thirty, that expectations of our present and future workforce are changing dramatically, and that technologists ‘deserve and require’ a seat at the big table. The digital transformation and connected operations will see collaborations between digital natives and citizen data scientists. The balance between vendor-provided and in-house developed solutions, and the relative importance of proprietary and open-source software is a ‘conundrum’. Holmes presented an IT landscape with Dell Edge devices feeding a Dell/Boomi Atom hub atop of a Dell EMC upstream data lake.
Simon Copping and James Innes presented Wood Group’s Proevx, an AI-inspired extension to Wood’s iWit well integrity data management solution. iWit was first developed in 2006 and is used today by 20 operators across some 14,000 wells. The acquired data set provides a basis for analytics and machine learning opportunities such as actuator integrity management with valve signature machine learning. Here, monitoring of supply pressure, function pressure and stroked accumulator volume can provide early warning or failing actuators due to degrading internal components. Proevx was also used to perform regression modelling of Reid vapor pressure at a gas processing plant, derived from 5 years of historical data and some 240 input parameters. The resulting ML-derived model is deployed via Proevx.
Rotimi Alabi presented RAB Microfluids’ ‘point-of need’ analysis of engine oils. RAB’s technology is a hardware and software bundle that combines physical analysis of engine oils with analytics software. More from RAB Microfluids.
Finally, Graham Gaston presented Sensalytx’ data analytics and visualization toolset for distributed temperature survey data. Sensalytx’ Q-DOS AI-based software for automated event recognition provides a neat color-coded quasi-3D view of DTS downhole data. Q-DOS offers added ‘AI in the cloud’ and data visualization technology from Fraps.
© Oil IT Journal - all rights reserved.