OSIsoft PI System user conference, Berlin

Internet of things? Industrie 4.0? OSIsoft and clients done 'em for decades! Flowserve pump monitored live with AI from SparkCognition. PI marketplace announced. Shell’s PI ’super collectives.’ Mitsui’s PI/SAP cloud. Honeywell on ‘ubiquitous but problematic’ OPC. OSIsoft’s main competition? In-house IT. The PI big data debate. Tullow builds digital oilfield from scratch.

OSIsoft claims to have been doing the industrial internet of things for 35 years already but is quite happy to jump on the current bandwagons of IoT, Industrie 4.0 and ‘digitalization.’ Christoph Papenfuss, welcoming the record 1,200 attendees to the 2016 EU OSIsoft User Conference in Berlin earlier this year, observed that there is a profusion of buzzwords, new technology and companies trying to grab your attention. But for Osisoft’s clients, data underpins all these initiatives. What OSIsoft and its clients have been doing for years is now mainstream. Companies used to give data scientists the wrong tools, spreadsheets. These are ‘like bunny rabbits, lovable but they multiply uncontrollably. Lets get rid of them!’ At the conference, IoT was more than just a paradigm, with an ‘internet of hotel things,’ thermal imaging in conference rooms and RFID chips on the attendees’ name tags, all feeding into a hosted PI Coresight monitor.

Martin Otterson suggested that the PI System means that clients are now tee’d-up for big data analytics. OSIsoft historically has helped break down the automation vendors’ silo walls. But today the major vendors are working towards sensor-to-cloud connectivity and are creating ‘whole new silos.’ Also, data lakes may not provide adequate context to captured data. A better route to the digital transformation is exemplified by OSIsoft’s partnership with Rockwell Automation which has PI technology embedded in its controllers.

A rather compelling use case that the audience appreciated involved a mock-up of a pump from Flowserve which was monitored using technology from OSIsoft, National Instruments and PTC. Data flowed into Labview and CompactRIO with analytics in PTC’s ThingWorks IoT framework. The system demonstrated cavitation issues, impeller and bearing failure, sending out emails when an upset was detected. For more fancy maintenance applications, an augmented reality/CAD overview of pump parts can be used by field engineers (using technology from Vuforia) on smart phones/tablets or even VR headgear. The multi-vendor theme is to be further developed in 2017 with the opening of the OSIsoft Marketplace, a ‘collection of products, applications, and services from over 200 global partners.’ The Marketplace is to launch in 2017 ‘along the lines of the AppStore.’ Of interest to the upstream are the PI Connector for OPC/UA and National Oilwell Varco’s Witsml-to-PI interface.

John de Koning showed how Shell has deployed PI system ‘super collectives’ to funnel real time data from worldwide production sites into its central business analytics function, a.k.a. the ‘Smart solutions platform.’ This leverages OSIsoft tools along with SAP’s Hana enterprise data warehouse. The ‘smarts’ are added with Matlab and R. Shell’s proof of concept runs at its Quest carbon capture and storage facility that runs alongside its Alberta tar sands operations. Data from real time laser strain gauges, acoustics and meteorological feeds stream into Matlab/SSP for model based calculations. Power BI provides users with self-service data discovery and HTML5-based visualization.

Japanese keiretsu Mitsui & Co., which took an equity stake in OSIsoft earlier this year, is now a poster child for ERP-to-operations data integration. Kenji Otake presented the PI/SAP cloud-based integrator proof of concept which is set to become a ‘key component of Mitsui’s digital transformation.’ The IoT integrator for Hana connects the two worlds of equipment and finance and enables collaboration across the supply chain.

Felix Hanish from chemical start-up Covestro’s observed that much of Industrie 4.0 is ‘buzzword bingo’ and someplace ‘between gold rush and panic.’ Even the novel stuff (machine learning à la DeepMind/Go) has been used in the process industries since the 1990s. But the new concepts address real problems in the capital intensive, highly networked chemicals business. Covestro has built its business around the PI System with some 1.6 million PI tags deployed in its StructEese* energy efficiency program. This is lowering CO2 output and energy costs and assures ISO 50001 certification that qualifies Covestro for a ‘big tax rebate.’ Condition monitoring also ran. On the downside, the company struggles with the roll-out of apps across its world-wide operations, spending too much time configuring SAP, document management, SmartPlant P&ID and PI. Another issue stems from different innovation cycle rates. These are slow for the DCS/SIS layer but much faster for management information systems. The lower layer is further challenged with the ‘slow and incomplete integration’ of standards like OPC/UA. There is a need to bypass and tunnel into the DCS to retrofit modern apps and add security.

Honeywell’s Elgonda LaGrange described OPC as the workhorse of plant and process connectivity. It is ‘ubiquitous but problematic.’ OPC/UA improves the situation with multi OS security and is backed by the German IT security organization BSI*. Honeywell’s Matrikon unit provides an OPC/UA software development kit, currently with a plant and process focus but also being extended for an oil and gas major to link offshore data sources to an onshore PI system.

Arie van Boven and Arco Stolk from RWE described the problems of aligning supply and demand in Germany’s diverse energy landscape. An unbalanced electricity grid could disrupt the EU and result in heavy penalties. RWE is reliant on its 24x7 PI infrastructure and is an innovative user of PI Asset Framework, embedded in its in-house developed ‘Dataswitch’ app. The system is now testing on the Amazon cloud.

We sat in on the analyst lunchtime session for an interesting exchange of views between OSIsoft management and the analyst community. PI started out as a tag database but soon matured, adding context to data and becoming ‘a kind of PowerPoint for engineers’ with much more functionality. Its main ‘competition’ has always been the in-house IT department which may be tempted by business warehouse or more recently, by cloud-base repositories from third parties, notably GE. In this complex landscape, PI’s strength is its focus on collecting time series data from a multiplicity of sources. In other words, getting information from the equipment zoo, ‘the hard stuff!’ OSIsoft president Jenny Linton added that the company’s privately held status meant that it could devote more resources to customer support than some, its hot line ‘always gets answered’ and the company can engage on strategic decisions that ‘last for a decade’. But while corporate executives may be au fait with the latest developments in big data and SAP, their own company’s in-house PI system may be under their radar.

We returned to the fray of the mainstream to hear SAP’s Ken Pierce on IT/OT convergence, specifically via the SAP/Hana IoT Integrator. This pipes PI data into the Hana in-memory database. SDKs are available for both Hana and PI. In this presentation at least, SAP is at the focal point of enterprise IT with OSIsoft off to the side. Not a picture that is likely to heighten PI’s status in the eyes of the execs!

Chris Felts’ headed a panel session on innovation in the IoT, with a focus on EU digital manufacturing initiatives under the Industrie 4.0 banner. There are many IoT initiatives around the world with similar objectives, inter alia, to ‘lure industry and academia in.’ But there are differences to the US and EU IoTs. In the US, IoT means data in the cloud combined with advanced analytics. In Industry 4.0, the EU IoT, it is more about interoperability, equipment and standards. For Greg Herr (Flowserve) IoT means doing more than just ‘giving data to the customer.’ Rob Brannan (RtTech) thought IoT was ‘just another elevator pitch!’ NI’s Ian Fountain was more on-message, NI is working with customers such as Flowserve on ‘edge analytics,’ e.g. feature extraction, putting the results into PI.

Some expressed concern that the big data movement might mean that process data may bypass PI and go straight to the cloud and even the operators. Elsewhere, established OSIsoft tools may be eclipsed by tools with a more modern look and feel, ‘but with only 5% of the functionality.’ OSIsoft is addressing this in Coresight 2016 R2 with its ‘high fidelity’ displays, notifications and ad-hoc analytics available from any device. In any event, the PI Integrator for Business Analytics allows PI to cohabit with BI systems, warehouses, data lakes and such.

Stuart Gillen from SparkCognition drilled down further into the Flowserve demonstrator. SparkCognition’s SparkPredict adds deep learning to process data along with natural language processing for human to machine and machine to machine communications. Apparently unconnected data streams are processed to generate testable hypotheses of root causes of upsets. Apache Kafka is used to build real-time streaming data pipelines. Apache Spark also ran, although the company denies a direct link between its name and the open source distributed processing engine. ‘Spark is not a reference to Apache Spark, although we do use it, there is no connection with the name.’

Tullow’s Mark Whitehouse presented on ‘designing and building a digital oilfield solution with no previous experience.’ The target was Tullow’s deep water Jubilee field, located offshore West Africa. Here, data acquisition and analysis is key to proactive field management. Tullow’s digital oilfield solution is branded TAPS, the technology assisted production system. TAPS has the PI System as a central component feeding high frequency raw data from the field to applications that perform integrated asset modelling, hydrocarbon accounting and allocation, production optimization and production reporting. Processed data from these application modules is then fed back into PI for visualization and trending. Whitehouse reports that TAPS has already resolved critical issues, including severe and ‘unprecedented’ productivity impairment in the wells and hydrate blockage in one of the risers. TAPS is now being deployed on Tullows TEN development.

* Structured Efficiency System for Energy.

** The Bundesamt für Sicherheit in der Informationstechnik.

Click here to comment on this article

Click here to view this article in context on a desktop

© Oil IT Journal - all rights reserved.