Matteo Boscato and Giuseppina Tomei presented on the optimization of the water treatment system on ENI’s ‘complex’ Armada Olombendo FPSO offshore Angola. The solution leverages machine learning on top of PI System data to provide predictions and anomaly detection at an ‘Integrated Operations Center’. The ‘digital plant’ displays information on onboard processes including four AI/ML algorithms for water treatment. EDOF (Eni’s Digital Oilfield, a PI System development) acts as a single source of real time data. The ML toolset includes components from Deepmind/Impala and Apache Hive. Data display uses Appian BPM linked to the PI System.
Virginie Segard and Gaël Cottet presented TotalEnergies’ approach to PI System data management. TE has been using PI for over 20 years and has developed many local use cases and governance best practices. The aim now is to scale these up, with strong data management and central governance, to create a ‘robust data contextualization platform’ supporting PI AF/Vision for data sharing, deployment and maintenance. TE is developing digital twins of its plants from generic site templates, localized with site-specificities such as UOM, language and historic tag naming conventions. Scalability is achieved with a distributed AF architecture running on central and local servers. On the niceties of data management vs. data governance, the authors pointed to a Tableau reference. Templates and naming conventions have been mapped and standardized and now, all generic use cases are developed by a team of AF developers and process engineers working on an Azure DevOps platform. To date some 400 AF master templates cover various activities. These have been instantiated to local sites for processes such as boilers, furnaces and reformers all with PI Vision-based displays. An AF to AF Manager is used to replicate between sites, with particular attention to units of measure,… ‘we guarantee that everyone works with the same UoM conversions’. Other custom tools have been developed for AF data validation and to detect differences between AF source and reference data. Work is carried-out in a new, consolidated ‘OneTech’ organization with some 3,400 engineers, researchers, technicians and support teams.
Azim Ezani bin Muzamli explained how ExxonMobil is moving from decentralized implementations of PI Process Book to a consolidated deployment on PI Vision. The 2021 release of PI Vision is an operational visualization tool that integrates with PI Asset Framework. ExxonMobil is leveraging PI Vision to manage load balance, failovers and offer a centralized visualization platform.
Amir Sadeghi and Yassine Messaoud presented OMV’s
‘Helius’ data ecosystem, underpinned by a PI System-based ‘single
source of truth’ for operations. Helius is the central point of access
for all E&P data for application developers, citizen developers and
end users. Other Helius components include OSDU, Cognite Data Fusion
and Esri ArcGIS. The system provides interfaces to other applications
from Seeq (analytics), Halliburton (Decision Space), Siemens and Black
& Veatch. One use case involved compressor membrane failure
identification using Seeq to spot anomalous pressure and instability
before failure happens. The deployment is a component of OMV’s DigitUP digital journey. More from OMV and Oil IT Journal.
Vishal Mahna and KK Chong presented Schlumberger’s
enterprise visualization platform for asset performance management. The
Schlumberger EVP corrals distributed data to provide contextual
insights and optimal decisions. Reading between the lines of the
presentation, it would appear that Schlumberger here is playing the
role of a systems integrator, building its EVP on top of multiple Aveva
tools running under the control of the Aveva Unified Operations Center.
Inpex Corp.’s first use of the PI System was as an enterprise historian. Since then, usage has evolved, as Naoto Yamabe and Kohei Kawamura described. PI played an essential role during the commissioning of the Ichthys LNG asset. Inpex has migrated its Japanese PI System infrastructure to the Microsoft Azure cloud, streamlining maintenance workloads and optimizing its corporate IT infrastructure. With help from systems integrator MKI Japan the whole on-site PI infrastructure was migrated to Azure in about a year, including an upgrade from Process Book to PI Vision. Various options for data migration were investigated. Negotiations with stakeholders were necessary to accommodate the eight hours of downtime during the switchover. A detailed cost breakdown showed that over a five year period, total cost of ownership is about the same for on-prem as in the cloud. Interestingly, the hardware cost for on prem is much less than the cost of the Azure virtual machines. But there are advantages in that on-prem implies extra costs in server setup and IT labor. IT can now ‘focus on supporting process engineers working on advanced analysis’. Use cases include analytics/ML with Databricks and visualization with PowerBI. Note that this migration was limited to Inpex’ Japanese data. Migrating the Australian Ichthys LNG systems is a more complex task that is still under study. The plan is for a scaled-up ‘hot’ cutover for Ichthys that adheres to the joint venture’s ‘zero data loss’ philosophy.
Abdulaziz Alzahrany and Reham Faqehi presented on the role that the PI System plays in Saudi Aramco’s Forth Industrial Revolution Center (4IRC), first unveiled in 2019. PI System is deployed to automate rotating equipment switchover. Tools used include PI System, Aveva InTouch and Unified Operations Center. The Aramco 4IRC presentation included a glimpse of a spectacular Aveva Video Wall project, predictive analytics using Aveva Aventis Prism also ran.
Tom Jacobs Stephan Leufke from Evonik (a German headquartered specialty chemicals company) presented the company’s OneCAE project to transform its existing plants with a ‘harmonized’ CAE* landscape. Evonik currently has a heterogeneous and complex system landscape with over 164 engineering IT systems, some paper based work processes and poor data governance. The OneCAE project sets out to fix this with a digital twin (DT), new data-driven processes, data quality assurance and an alignment between the physical plant and associated digital information. Evonik’s earlier digitization work has leveraged standards including PlantXML, DEXPI and most recently, the Asset Lifecycle Data Model from the German Namur standards body. The latter is the basis (inspiration?) of OneCAE whose target landscape is an assembly of software tools from Aveva, SAP and others. The user interface is to be Aveva Engage. Currently the project is at the MVP* development stage. A substantial migration and digitization of existing plant documents and data including laser scans is to create a virtual as-build representation of the plant. The target DT platform includes multiple Aveva tools along with SAP.
* Computer-aided engineering.
** Minimal viable product.
Finally we have to mention Gianmarco Rossi and Luca Cadei’s presentation on the use of ‘data democracy’ as an enabler for ENI’s upstream digital transformation. The paper reprises Eni’s work on the e-DOF (above) to show how this is being extended with data science and PI Vision graphics, accessed from an Integrated Operations Center (IOC). But what most caught our fancy was the literary endnote to the presentation, a snippet from poet William Blake’s ‘Marriage of Heaven and Hell’ viz …
Man has no Body
distinct from his Soul for that call’d Body is a portion of Soul
discern’d by the five Senses, the chief inlets of Soul in this age.
Energy is the only life and is from the Body and Reason is the bound or outward circumference of Energy.
Energy is Eternal Delight.
All this may or may not be
relevant to the e-DOF, but the Blake snippet is actually from the
chapter titled "The Devil’s Voice", and the three lines are a rebuttal
to the Devil’s arguments.
More from the PI World home page.
© Oil IT Journal - all rights reserved.