In the EU post the 2014 downturn, industry health has been ‘mixed’ and dependent on roles and geographies. There is action in the fields of artificial intelligence, machine learning and around plug and abandoning and non-invasive inspection. All fields where GIS is, or could be a part of the solution. Esri’s Danny Spillman observed that the advent of the digital transformation, data and analytics means doing things differently. A Dresner Advisory Services survey found that two thirds of respondents believe location intelligence to be critical for their business. While GIS may be losing its central role to data/analytics, ‘it is still part of everything you want to do in a digital transformation’.
Shell’s Rob Dunfrey spoke to the ‘geographic advantage’ in Shell’s digital transformation, which he dates back to 2013. Today, this is ‘gathering pace’ with the creation of a digital center of excellence and a company-wide digital strategy. Dunfrey agreed that geospatial has a significant role in digital transformation, the internet of things (IoT) and in artificial intelligence (AI) applications.
Spillman reported on other companies’ usage and the interplay between geography and digital transformation. ConocoPhillips ‘uses Esri to make maps’, but is also having conversations around Hadoop, R and Python, and has convinced its transformation folks that there is more to Esri than maps. GIS may not be the main player here, but it clearly has a role as a component of a system of reference. Additionally, as ArcGIS capabilities are recognized, the question becomes ‘how to use spatial analytics to drive insights’. In which context, surveillance imagery from drones needs a GIS/data strategy, as does geographically disperses IoT/sensor data. And finally mobile, where ESRI has no less than three strategies/offerings. Another new direction is the marriage of GIS and building information management (BIM) and visualization of the real world in 3D. You can now work with BIM files in GIS, adding context as ‘BIM and GIS coalesce in capital projects’. Esri is partnering with Autodesk to, for instance, design well pads with Revit. Format conversion is facilitated with the OGS’ I3S, indexed 3D sharing format (more from Esri and on Github) that retains Revit attribute information.
Spillman wound up commenting that digital big data is ‘overhyped and misunderstood’. In reality, it comprises business intelligence, analytics, machine learning and artificial intelligence. All of which may be owned by different groups in an organization. Increasingly, Esri’s offering addresses real time situational awareness, for instance in oil and gas, for competitor analysis. Elsewhere, Esri’s Python API is used for real time activity and object detection. ArcGIS for the IoT has been used to map power outages, track assets or the weather and monitor production. Location intelligence is playing a key role in the digital transformation.
ArcGIS Enterprise Sites (AES) lets you build custom websites with your own content. Enterprise Sites create a tailored web page from corporate GIS data that is accessible to non-GIS users.
AG Analytics for IoT will be out real soon now. Analytics for IoT collates data from millions of sensors into a constellation of big data microservices and containers (scala, kafka, spark play for map services) and spatio-temporal storage (under evaluation). The promise is for real time/big data GIS ‘as a service’. A ‘battery-efficient’ location tracking service leverages a NoSQL data store. A demo showed an inept golfer shanking left and right around the course, being tracked with ‘a Strava-like functionality’.
Another NoSQL data store powers high performance big data visualization with ‘quantization’ that adapts display sampling on zoom. The promise is for streaming performance of large data sets such as IHS Markit’s 4.7 million wells. The environment is a toolkit for ‘really big’ data that requires spatial awareness sans data movement. A ‘program-to-the-data’ philosophy leverages a Hive data lake with WebHDFS; U-SQL, Spark and HD Insights, all viewed in ArcGIS naturellement.
A GeoAI virtual machine (developed with Microsoft) combines AI with Esri geospatial adding AI/ML to ArcGIS. A demo use case involved a communications exercise with residents living near a pipeline. A buffer area is created with standard geoprocessing. Inside the buffer zone, aerial imagery is used to create training samples tagged with ‘house’, ‘hospital’ and so on. The data set is exported for deep learning and analysis by a data scientist. The resulting ML-derived model is imported back into ArcGIS which identifies similar structures in imagery. One GeoAI test identified 6,000 well pads from Sentinel II satellite imagery.
Rob Dunfey described digitalization as being ‘in Shell’s DNA’. Shell has a digital ‘center of excellence’, and a coherent approach that regards data as an asset. The business owns digital, and operates an in-house digital capability where the customer/user is central. Shell takes inspiration from new players in energy, Tesla and even Apple (selling surplus energy from its data centers). Disruption is both a threat and an opportunity. Current focus is on dig tech that is ‘reaching an inflection point’, such as the IoT. Here, Shell is working with Esri on a modern geospatial IoT platform embedding Geoevents into Shell’s MyMaps along with Scala; mongodb, kafka elastic, python and spark. Other aspects of the geo-digital transformation include robotics and drones for inspection and surveillance, mobile apps to locate pipelines and wearables with 360° photogrammetry for mobile Hazid*. 3D and BIM has been used at Shell’s Deer Park Refinery with a model hosted and accessed via MyMaps. Finance and competitive intelligence, aka ‘Facit’, leverages ML, machine vision and earth observation to map and monitor hot spots (as hedge funds are said to use in monitoring unconventional activity). In conclusion, geospatial is key to the successful deployment of several of these technologies. In the Q&A Dunfey was asked what new skills are required of the geospatial community. He suggested TensorFlow. Another questioner observed that many business apps have inbuilt spatial functionality, how does one avoid duplication of effort? This is a question Shell asks itself. ‘Does it matter what the endpoint is as long as underlying data is accessible?’
* Hazard identification studies.
Dominic Bull provided an update BP’s OneMap. BP’s global GIS community of practice is working to sustain and grow the platform. There are huge opportunities for geospatial across the modern oil company. BP uses Esri technology to devolve some aspects of GIS innovation and development to ‘citizen development’, although ‘we don’t want people to run wild’. Not everybody is an expert in all aspects of every platform. BP now has standard roles for GIS professionals and analysts and has ‘moved away’ from its earlier geographical organization. BP has built a geographic information science and technology body of knowledge, a university consortium for GIS leveraging Yammer and SharePoint.
BP has established IT&S standards for GIS data with quality controlled coordinates and data products in a ‘tips and tricks’ area. Github is used ‘quite extensively’ as are shared FME workspaces. Advanced analytics involves more interaction between GIS, geoscience and data science. Recent developments include a vehicle tracking service for Khazzan and ‘simops’ (simultaneous drilling and production operations) for Indonesia.
2018 saw the rollout of OneMap 4.0 with analytics, the cloud, big data and GeoIoT. In the Q&A Bull stated that Voyager Search was used to scan geo data for compliance. Citizen app development has proved very successful, ‘the results better than our own, it’s scary!’. Asked about the challenges of opening-up GIS systems Bull acknowledged that people need to be aware of what they are doing, ‘most work atop of our managed data.’
Tom Royston (now with DGTop) used to work for Addax Petroleum. Addax was acquired by China’s Sinopec Group, which eventually moved the whole company to Beijing. Addax’ data was housed in a range of applications (EnergySys, IDS, NeuraDB’s well header database, EDB, PSApphire, SeisQuest), most all visible from an Esri front end. To move Addax’ data, Royston used a combination of the Azure cloud, ArcGIS and a PPDM databasde. An initial proof of concept involved a move to the cloud for the EnergySys and IDS data. PingOne identity management was found to be a key enabler for working in the cloud. Following the 2017 POC on Azure, things accelerated will the early closure of the Geneva HQ.
Royston did some work on cloud comparison but in the event, Office365 proved an unavoidable front door-cum-trojan horse into the Azure cloud. This required a major network upgrade with dual connections especially for Addax’s African units. Local internet breakouts allowed access points to be securely distributed, with Azure used as a ‘virtual’ data center. The migration involved a choice between ‘lift and shift’ and a re-platform. Addax went for the latter with a move from Oracle on-site to SQL Server in the cloud. ArcGIS was re-installed with ESRI’s Cloud Builder and the ArcGIS Desktops moved to Azure virtual machines. Now all is up and running on Azure in Hong Kong, accessed by the Africa and China teams. The original desktop app running in cloud can be used across CNPC/Petrochina (not just Addax). Geodatabases are now in Azure so there is 'no more need to back up’. Lessons learned included the fact that you ‘can’t just jump in’. You need to consider exactly what the objectives are. Regarding the business case, the cost of the cloud may hold some surprises re cloud providers and price plans! If you can afford it, re-platform. In the present case, the editor of the accounting systems ‘would not play ball’. The cloud is not ‘out of the box’.
Cécile Noverraz works with the team building the controversial Nord Stream 2 gas pipeline from Russia to the EU. An onsite-deployed Esri ArcGIS/Portal master database is at the heart of the project with a loosely-coupled document control system from Easy. The comprehensive development tracks a variety of cultural information on cable crossings, UXO, berms and more. A Geoevent server integrates AIS ship tracking data from ExactEarth, pulling data every 5 minutes. Hundreds of boats are involved (30 in a single dredging operation) on the project so ‘proximity incident reporting’ is key. The project is creating a high-quality data set that can be handed over to inspection and maintenance on project completion. Wish Software’s VisualGIS server (an ArcGIS extension) integrates ROV inspection videos with the mapping system. Nord Stream 2 has some 100TB of video imagery to manage and share. Noverraz is also planning to manage pipeline geometry in ArcGIS Pipeline Referencing. The Esri Portal and GIS technologies have been ‘deliverables’ from the start of the project and are now ensuring long-term data integrity.
Jeff Allen (Esri) presented on the latest technology in pipeline data management. ArcGIS pipeline referencing and web services make it easier for partners and users to address particular issues. An initial focus is asset management (GIS, MAOP, TVC…) but these are also used by other groups (integrity, operations, HSE, business). APR, ArcGIS pipeline referencing, manages point and linear data as a dense dataset atop of the pipeline rout. Core tables from APR can be embedded in PODS Lite and/or Esri’s utility and pipeline data model (UPDM). A web app allows for event editing, query and QC with rule-based protection of the underlying database. Different apps for different parts of the organization takes the pressure off GIS editors. A use case involved an inline inspection pig survey, switching between schematics and imagery with above ground markers. Schematics can be captured the utility network model. This, along with rule-based asset packages, can be used for oil, gas, liquid, electricity and more. The UPDM 2018 file geodatabase can be downloaded along with an oil and gas profile/package.
Paul Cleverley’s research has led him to beware of cognitive
bias, i.e. ‘group think’ that favors consensus over critical thinking.
He has observed geoscientists talking about their work and
observing what papers (in OnePetro etc) they click on. He is less than
enthusiastic about current text-search technology and advocates rule
based, supervised and unsupervised ML for text mining. 100 years of
Geoscience World has been analyzed and spatialized from the text. This
includes extracting numerical information, like ‘0.4ppm’ or 500mybp
from ‘dark’ data i.e. numerical data hiding in text. Word2Vec and GeoDeepDive
from U Wisconsin also ran. The approach lets you ‘see big patterns and
develop new theories’. Cleverley acknowledged that there is a lot of
technology ‘propaganda’ about. It is unlikely that machines can read as
well as humans, but they can read a lot more. In narrowly-defined
tasks, this can be very useful. In the Q&A, Cleverley observed that
‘training data dominates the equation’ and that PDF is a ‘pretty awful’
source for text. It is easier with Word but better still, extract
the text itself, ‘curation and normalization are key’.
Keith Winning presented a paper at the 2017 EPUG on data modeling of a major gas pipeline. Since Pipeline Data Solutions' early adoption of the data modeling process the company has built 11 data models for some 500km of pipeline and 1.5 million records. Historically, pipeline modeling for operations and integrity management was performed post construction. PDS proposes a ‘digital twin’ approach, building the model concurrently. This allows the model to be used in construction/project management and to assure data quality post-handover. But there are challenges, both internal (schema changes, how complete is the model), and external issues, of data quality, out-of-sequence data arrival and managing change at the attribute level. To be sure, ‘you need to check and check’ and, if you can, incentive contractors to provide data. Excel data, written weld numbers scribbled onto a pipe in the field and many other issues make the process very error prone. Data acceptance and review is key although, ‘you can’t ask a contractor to focus on trivia’, a pragmatic approach is needed for secondary data. PDS uses a PODS schema with an extra cross reference table to map different contractors’ tags. ESRI Survey 123 and Collector also ran. But really, ‘the construction industry has to change and incentivize good data delivery’. Preferably delivered prior to releasing a section for hydrotesting.
Jostein Bjerkan described AkerBP’s MapInsight GIS portal. MapInsight supports geology, seabed and operations, interfacing with Office365 and SharePoint. AkerBP’s spatial services team has automated many data migration and ingestion processes. The move from ED50 to WGS84 brought a 3x speedup in mapping performance. AkerBP now creates its master data in GIS. Data can be consumed via a standards-based open source API for analytics. All is rolled-up in the Cognite data platform and a ‘massive data lake’. AkerBP likes to play with its data and has ‘thrown everything’ into Unity 3D, a gaming/rendering technology that connects GIS with Petrel Studio and enables 3D visualization of reservoir models, interpretation results, seabed data, pipelines and more. The video was underwhelming, ‘of course this is not implemented as a fully working tool yet…’
Mark Jones reported on Shell’s trials with Insights for ArcGIS, Esri’s ‘web-based data analytics for advanced location intelligence’. Jones sees synergies and commonality of purpose between GIS and AI. Insights for ArcGIS allows drag combination of spatial data into an analytics environment. Shell has used the tool to mash-up its own proprietary data with data from third parties including Westwood Global Energy Analytics and Woodmac PetroView. This allows for spatial queries on licenses by round, aggregate reserves by operator, resource type and so on. (ie all Oracle-style queries, apparently sans AI). It turns out that authoring dashboards is easy and Insights offers good enterprise IT integration. But is it really deployment-ready? For Jones the answer is … yes and no. It currently lacks reporting functionality and there are labelling issues as the solution moves away from full-blown GIS functionality. The close collaboration with advanced analytics can evolve though. Geospatial analysts are also data analysts. Insights can be a way of democratizing geospatial. Let the business ask questions we don’t know. Geospatial is a key lever in analytics and is ‘perfectly positioned’ to participate in the revolution. In the Q&A, Jones was asked how Insights compared with LogiInfo or Spotfire. These tools offer similar functionality but with less spatial functionality.
Rui Menezes provided an update on iGas Energy’s use of M-Files, comprising a ‘GIS and intelligent IM’ solution. A joint venture between Esri Finland and M-Files (also Finnish) has developed and embedded M-Files with ArcGIS Pro. Conversely, ArcGIS can now be launched from inside M-Files.
Advances in AI make searching historical data easier and more performant. Voyager Search helps connects the portals/dots, spanning SharePoint, ArcGIS Online, files and stuff in the cloud. This is achieved without move or copying data, just by index building. Voyager adds ‘where’ to conventional ‘who what when’ search. This is achieved with a built-in machine learning-based enrichment platform. Natural language processing also ran. Voyager has, inter alia, indexed Reuters’ news articles and extracted location information from place names. More from Voyager.
Geocap’s Seismic Portal
offers ‘Petrel to Esri’ integration, bringing seismic and well data from
Petrel Studio to ArcGIS Pro. A subsurface 3D viewer pops up and
displays geoscience data along with rig/platform and seabed kit.
Exprodat is helping clients migrate to Pro and showed-off its skills with a drone video embedded in Pro. Pro was released in 2014 to replace ArcMap (which will run to 2023). It is ‘time to think about getting into Pro’. Exprodat offers migration services and customization with Python and SDJ code to replace Arc Objects. More from Exprodat.
Conterra provides a security manager for ArcGIS that allows fine tuning of usage such that ‘everyone in the company uses what they need to do their job’. It can be hard to map all services to all users. ArcGIS Pro Server security provides service-level control. Conterra’s security manager extends the security concept with object/layer/spatial access control and intersecting spatial filters as required. Conterra also works on apps. The Conterra admin interface is integrated with the ArcGIS server manager. More from Conterra.
Woodmac PetroView for ArcGIS Pro provides intuitive data exploration along with business intelligence à la Spotfire. PetroView installs in parallel with Desktop and runs off the same data. Alternatively, the tool can run from inside Pro as a tab/add-in. PetroView performs impressive semi-intelligent queries across a big North Sea dataset. This seemed to show that North Sea production would be finished by 2027.
Thierry Gregorius (Getech) compered a Menti.com quiz by way of a closing ceremony. Menti allows the audience to ask free-form questions and vote in their answers. It was established that 24% were PUG first times, 11% had attended one event before and 66% serial ‘many’ EPUG attendees. ESRI were surprised to learn that in the parallel sessions there was a 10 to 1 preference for case studies over ‘technology’. Asked ‘Who is a data scientist’, almost all answered ‘Something I’m aware of but can’t do myself’ none admitted to being a ‘true’ data scientist. Around half were part of a ‘digital transformation’ the remainder split equally between not being part of such a team and not even having such a team. Over half were moving to the cloud out of choice, the rest were either not moving or moving because they had to. There was a consensus that the key skills to learn today are python and machine learning. Finally, a distinct preference was expressed for Paris as next year’s EU PUG location.
The presentations from the 2018 EU PUG are available here.
© Oil IT Journal - all rights reserved.