The 3rd Global Business Conferences IIOT & Digital Solutions for Oil & Gas* held in Amsterdam earlier this year heard from a variety of practitioners in the oil and gas digital transformation space. Many current proof-of-concept implementations have struggled to prove their worth. BloombergNEF reported a ‘lack of astonishing results’ to date emanating from the digital transformation movement. For some this is an indication that only a full-blown push for enterprise-wide deployment will lead to transformative success.
McKinsey’s Anosh Thakkar opened the proceedings with a keynote on ‘maneuvering the Industry 4.0 jungle to deliver impact at scale’. McKinsey’s recipe for digital transformation starts with C-Suite/business project ownership (not IT). An approach that Thakkar enigmatically terms a ‘sheltered highlander’. Digital transformation shares a lot with the Lean approach, with ‘Lighthouse’ projects key to demonstrate impacts. There are five principle levers to Industry 4.0: AI-enabled predictive maintenance, production optimization, real time asset performance management, automation of manual processes with robotics and machine vision, end-to-end dynamic optimization. More on McKinsey’s Lighthouses from the 2019 World Economic Forum presentation.
The big challenge however is that only 20% of industry proofs of concept have started to scale, ‘even though their impact was proved’. The technology jungle is another impediment, ‘everyone offers everything’, in a world of 100 interfaces. McKinsey advises drafting a matrix of opportunities, building the business case and following the money. Where many are eager to jump into the technology pool and like to talk technology. Thakkar says, Whoa! Hold off, what is the problem and what is the expected value. Only then will the solution fall into place.
Picking up on the lighthouse theme, Sak Nayagam (BHGE, formerly with Accenture), presented the Pharos digital transformation project, co-created in partnership with BP, Microsoft, Accenture and BHGE. Pharos kicked off some 18 months ago in an ‘ideation’ session that set out to ‘help BP be best in class’. A team was assembled in Houston with instructions to move beyond a ‘land of 1001 pilots’. A few thousand Post-It notes later, the team had worked through predictive analytics for GE’s PowerGen LM2500 gas turbine, a subsea flow assurance use case and more. Business transformation consultants Morphix contributed to the project.
Giacomo Silvestri (ENI) agreed that there is too much focus on the proof of concept. ‘We now understand there is value in the transformation, but this is not a short-term game changer’ ‘I am fed up with the constant PoC approach’. ENI had 20 drone PoCs where two would have done! We need more courage. Investment in these technologies is small beer for the industry. But you need to avoid noise and distraction from the hype. Enter the ENI Digital Agenda, gathering input from tech scouting and open innovation from outside, developing business cases and partnering. Focus now is on safety and asset integrity, enhancing performance, decarbonization and the circular economy. ENI anticipates a billion Euro value from digital over three years and a 7x ROI in ten years. There are currently 165 digital projects underway. These include image recognition in seismic, AI, HMI, IoT, robotics, 3D printing and blockchain. ENI is also working to improve the audit function with help from banks and financial services companies. The scouting activity has scanned some 350 startups and selected 5.
Einar Landre believes that data is to drive Equinor’s next wave of improvements. Equinor’s data roadmap centers on ‘Omnia’, its in-house developed Azure-based subsurface data lake and reservoir engineering platform. Omnia supports operations planning, digital twins, and drilling automation. Landre cited Fraunhofer’ Matthias Naab as stating that ‘since 2010 software can deliver more than expectations’. To date digitization has followed an incremental path. It is now time for new business models and ecosystems and ‘radical new digital services and solutions’. Like others, Equinor is confronted with the innovator’s dilemma. As new tech coming along, when do you jump? This is a tough question. But it may pay to evaluate radical concepts based on their future business value. In which context, Landre presented the Shell-inspired open subsurface data universe OSDU.
Jaco Fok reported on a pilot deployment of a Teradata warehouse at OMV Petrom, the largest energy company in Romania and southeastern Europe. The Teradata integrated, centralized business analytic hub provides a single version of the truth to Petrom’s refiners. The system provides real-time insight into blending performance, throughput and yields and tank farm/terminal operations. The pilot has moved through increasingly promising stages and is now credited with providing a basis for day to day decisions in planning and scheduling. Extended metrics along the hydrocarbon value chain feed into a prediction and recommendation engine. OMV’s Petrobrazi Refinery was home to the Teradata pilot.
Eleonore Lazat from Bloomberg New Energy Finance gave an overview of Bloomberg’s study of emerging technology and new value creation. No surprises in the list of emerging technology, from blockchain to drones and ‘disruption’ from new players and cloud computing. Another study of the impact of digitization in the refinery has found a lack of astonishing results to date in terms of ROI. Perhaps the value of digital is elsewhere, safety?
Noorddin Taj (BP) provided more data on digital underperformance. While eight out of ten companies are on the digital journey, only 14% are able to demonstrate sustainable digital project. ‘They just want to do digital because everyone is doing it!’ So why now it is so important now? Previously companies would just digitize something. Now they are changing their business model to gain improvement. Taj foresees a shift to more ‘agility’ and from ‘asset-intensive’ to ‘idea-intensive’ profitability, citing GE as an ‘asset-intensive failure’*. BP is using more and more APIs, ‘We are on an API journey, all interfaces expose discoverable reusable APIs’. Taj advocates microservices as opposed to a ‘monolithic’ architecture. Also, ‘usability is more important than functionality’. Taj offered one example of a changing business model, petroleum swapping between oil companies. Depending on the distance from the terminal, product can be bought and sold between operators such that BP gas may be delivered to a Shell station. This ‘used to be done in Excel’, now it runs on a blockchain-based exchange. IoT devices manage crypto token exchange and do ML analytics on lifting to predict inventory levels, sales etc. Smart contract and blockchain-based self-service identity allows individuals to check in without registration.
* Our tracking of GE would put it close to the forefront of ‘idea-intensiveness’. GE may not have invented the ‘ideation’ word (its first use was in 1818!) but it certainly popularized its recent use.
Julien Brunel described Linde Engineering as a ‘digital remote plant pioneer’. Linde’s 1,000 plants are run from ten remote operations centers. A ‘Deep Cylinder’ machine vision identifies gas bottle types coming in. The Linde PlantServ portal let customers access information directly and supports a shift away from the old reseller model. Brunel acknowledged that there is still ‘a lot of paper P&IDs, Excel and cut and paste’. Linde is working on a tablet-based P&ID solution.
While others took a somewhat high-level view of digitization, Peter van den Heuvel exposed the nuts and bolts of Shell’s OSIsoft PI center of excellence. A real time architecture leverages C3IoT and PI System across the board. Shell has 15,000 users of PI, 20 years of real time data and 7.5 million connected instruments. C3IoT adds AI and machine learning to the mix, Matlab, R, Python and Seeq also ran although, ‘PI is the standard’. The latest deployment on the 488 m long Prelude FLNG leverages Seeq, C3IoT and PI in the Sky (cloud). The move to the cloud means more flexibility, but also requires getting used to Docker, virtualization, etc.
Vincent Jacquemet (Schneider Electric) has integrated its Aveva and Invensys acquisitions into its EcoStruxure platform. Citing a McKinsey analysis, Jacquement reported that predictive maintenance tops the digital agenda in terms of expected value. Analytics can be delivered in two fashions. Opex-oriented via the cloud or capex-oriented at the edge or embedded in a device. Thermal monitoring of transformers or fault detection in pumps are amenable to the ML in the cloud. Models are trained on ‘normal’ behavior and detect anomalies. Predictive maintenance is a key driver of digital transformation. But it is not exactly new. France’s EDF has been using ML for predictive maintenance for 15 years. Duke Energy has some 14,000 ML models deployed. For successful deployment, data prep is key. Models need to be tested with data playback, to tune thresholds and alarms. Edge computing brings ML to the field and is key to avoiding ‘data decay’? Edge enables the use of high frequency data and just-in-time intervention. Schneider ran a pilot on five Canadian wells installed with Realift controllers and an edge gateway. Schneider’s Vijeo Designer was used to allow human intervention to check analytics and label Dynacard images. The PoC identified three patterns found to be indicative of a faulty load cell, a paraffin issue and rod centering guide problems
John Bell presented Askelos’ finite element analysis (FEA) toolset that performs structural analysis of large offshore platforms ‘1,000 x faster than previously possible’. Askelos came out of MIT and the US Department of Defense with backing from Shell Ventures. Speed-up is achieved by ‘reduced data’ FEA, demonstrated with a ‘full spectral fatigue analysis’ on Shell’s Bonga FPSO. The company is now working with Shell to prove to the regulator that some North Sea assets have much longer life than expected.
Swim.ai presented its ‘distributed data fabric’ that analyzes and learns from time series data, creating a digital twin from the data. Swim.ai replaces ‘gigabytes’ of Hadoop/Spark code with a message broker, a streaming API, dashboard andstorage.
Datumize is ‘unlocking’ industrial data and bridging the IT/OT gap with ‘non-intrusive’ data capture. A data appliance in the DMZ taps into the comms network with a ‘unique network sniffing protocol’ capturing data ‘without being noticed’. The system has been deployed by Cepsa to optimize refinery operations.
Precognize’s SamGuard provides predictive monitoring, spotting failures on catalyzers, heat exchangers and rotating equipment. Unsupervised ML on historical failure data builds a baseline model. This is combined with P&ID structures and plant process behavior, the cognitive bit. SamGuard ‘listens out’ for small changes to show where problems are about to occur. There is ‘no alert fatigue and no need for data scientists’.
Industrial Analytics is a specialist in acoustic/vibration fingerprints of rotating equipment from different suppliers and vintages. Sensors feed into TurboNode, a hybrid physics/ML-based model for processing with work orders delivered to SAP Maintenance Management.
White Space Energy plans to ‘navigate complex decisions’ in oil and gas through game technology-derived AI. Today, there is ‘lots of noise and hype out there’ (in the AI/ML space). White Space is moving slowly to apply ‘superhuman’ gaming to problems such as well trajectory planning, logistics and more.
Atomiton provides edge computing solutions for pipeline monitoring of vandalism/damage and leak prediction (not just detection). Acoustic sensors are processed with ML-derived pattern recognition in a continuous learning process.
* GBC is rebranding the conference which in 2020 will become GO Digital Oil & Gas with two events, one in Abu Dhabi (3-4 March 2020) and the other in Amsterdam (3-4 June 2020).
© Oil IT Journal - all rights reserved.