PowerPoint Presentation

[Audio] Hello guys! I'm going to present a second set of scenarios that we've put together for your DataHub product demo. We have in Petrobras a set of internal applications for Topside Integrity Management and we have to deal with a lot of data management challenges. We´ve brought some of them here in the form of use cases and we would like to see how your DataHub product features could contribute. We hope this material gives you the necessary information for implementing the demos. We are looking forward to seeing your product working on the challenges. Have a nice job!.

[Audio] Here we present the four demo use cases we'd like to see, all linked to topside integrity management. They are named as scenarios, starting from 5, as there were four previous cases already sent to you with other challenges, right? In scenario 5 we present a use case for monitoring Topside equipments. Here we would like to know better the DataHub features that make these kind of applications easier to develop, such as dashboards that monitor the health of plant elements with operational data coming from OSI PI and Engineering databases. In scenario 6 we present a maintenance services planning use case. We would like to understand DataHub capabilities that better address maintenance data from a variety of data sources, especially SAP-PM. This case address the formation of a knowledge base that facilitate queries through a related view of the concepts involved. Scenario 7 brings an offshore gallery idea. We would like to understand how DataHub would manage unstructured data such as 360 images and videos that are obtained by image campaigns on the oil & gas platforms. Last but not least, we would like to understand the features and strategies of the DataHub working with industry standard models such as OSDU and CIFHOS, including their interoperability with internal models at Petrobras. This case is represented as scenario 8..

Disclaimers. 1. The scenarios in this material bring application use cases that could take advantage of a DataHub product in order to solve their data requirements. They are simplified when compared to our practical needs but work here to give some taste of real challenges we have to face in the development of Topside Integrity Management applications. 2. Some implementations would be needed inside the applications so that the end-users would have their problem solved, but here we are more interested in the DataHub features demonstration. However, we need to understand the boundary. We would like to get a clear view of what are the capabilities of the DataHub product that could contribute to the solution to the end-user and what will would still have to be implemented inside the applications. This decision depends on the product features and also on data strategy rules. You can make comments on this in your presentation, mentioning pros and cons if there are. 3. We need that your implementations follow the Demo Plan row by row, as strictly as possible, and that you solve all your doubts before the presentation. But be free to make comments on other alternatives of solution you envision. 4. Some items in the Demo Plan we tagged as “extra challenges”, meaning that we do not expect they to be core capabilities of a DataHub, but we would like to see at what extent does the product could contribute to their solution..

Scenario 5 Monitoring Topside Equipments. Ver a imagem de origem.

Scenario 5 Data comes from different sources. Data source Description Data of interest Technology Architecture PI PI System measurement values x measurement units (addressed by TAGs) OSISoft PI Distributed PI servers (per operational site) SPPID Engineering Schematics Database instruments x pipelines; instruments x equipments Oracle RDBMS Central database SPI Instrumentation Database instruments x pipelines; instruments x equipments; o perational limits for alarms Oracle RDBMS Central database S3D 3D Models Database Platforms x modules x location TAGs Oracle RDBMS Central database Equipments control systems Equipments control systems events associated with physical equipments Variate Distributed at automation network.

Scenario 5 Other information. For historical reasons, measurement TAGs do not have the same name pattern among PI Servers and have to be mapped to ANSI/ISA-5.1. For example, the same type of instrument has TAG P75_AST-A116001 at P-75 and 36030G_100_AST_001 at P-58. The number of position of modules varies among platforms, depending on how they are designed. Location TAGs are standardized and unique between all platforms. Measurements of interest to the monitoring dashboard are given by the instruments installed near the equipment – SPPID or SPI databases can provide this information..

Scenario 5 Challenges for developing the dashboard.

Scenario 5 Demo Plan. Demonstration How? Describe in PowerPoint the dashboard solution usingresourcesfrom the DataHub. Describe the proposed architecture of com onents in the DataHub. Commenton the assum Dons made. Simul ate the datasources from scenario 5 and how the sol uti on could gather and join data from them in order to serve data to moni tor all the turbomachines from some specific pl ataform, consideringsome minimal setof interesting measurements asin the exam lein the first slide . 1m lementan information model thatcontextualizes the data and abstracts the com lexi of havin multi le datasources. mpl ement the logi cs/mappings/rules described in the example to transform data and show how the outputs could be consume d b the dashboard develo er. Demonstrate how to stream the temporal series curves from Pl to Power gI It also needs to have flexibiliti in order to choose if needs raw or summarized data in different time frequency (1 second, Imi mute, 1 hour...). Disouss pros and cos of different strate i es for servin this information to thea licahon. Demonstrate how thea lication could et asttime series re ardin some time interval. Demonstrate how to deal properly with Engineeringlnformation whi ch rarely changes, but that could trigger updates in the I nformation model consumedby the application. Develop and show a simple dashboardin Power Bl to exercise frie idea of monitoring turbomachines of any oil & gas platform, considering the chall enges described and using the DataHub to se rve data. Discuss on whatneeds to be developed at Power 81 Comment on avail able optonsfor defining the equipment condi tion he alth (green, yellow, red) as a KPI. Implement a simple rul to serve this KPI. Di scuss on whatneeds to be developed at Power gI side * DataHub side. C)emonstrate how the DataHub could help in managing alarms (se tti mg parame ters, sending events to the application, storing event history, pri oritizing, Commenton available options for defi ng the automated rules based on SPI operational limits to tri er the alarms. Discuss on whatneedsto be develo edat Power side * DataHub side. Demonstrate how the DataHub could help in sendingalarm eve nts based on smartmodelsfor anomaly detecbon, predicbon (A1 or oåner es These models could be acessedb APIs lu ed in the DataHub or not bein served outside . Demonstrate how to implement automated rules which define commands to send to the equipments through their control systems after each alarm event Example: according to operational conditi ons you may have to starta sendby machine after one has failed: "if tem erature of Ourbomachine 1 is above f', sto turbomachine 1 and start turbomachine 5". Demonstrate how the DataHub could hel in servin smartmodels that im lementaction recommendations to the user as in the Demonstrate how to communicate with control systems using OPC-DA, OPC-LIA and/or MQTT protocols to gather informati on directl from the turbomachines. Demonstrate how to indude an attribute to the information model that ivesa rade to the data uali of each monitored Demonstrate how to serve a smart model that ives the above rade based on other monitorin variables from Pl real time Demonstrate the available mechanisms for control the access of data, expl ai ni ngusers and roles. Comment on how to preserve descentrali zed authentcation/authorization rules, ei ther you have appl i cati on acounts or end-user accounts..

Scenario 6 Planning Maintenance Services. A user wants to know all the maintenance demands that are in his backlog concerning some search/filter criteria. He wants to gather the details of each demand and to make overall calculations, for example of mean resolution times. With this information he will be able to compose a prioritized service portfolio. Petrobras wants to develop an application that democratizes the access of maintenance information from various data sources. This application builds a knowledge base defining logical concepts and their relationships. It also make enrichments by implementing some business rules..

Demand. Maintenance request. RTI. Maintenance order.

Scenario 6 Creating the Maintenance Knowledge Base.

Scenario 6 Data comes from different sources. Data source Description Data of interest Technology Architecture SAP PM SAP Plant Maintenance Maintenance orders, maintenance requests, RTI reports, equipment and location information (technical objects), RTI attachments SAP One server per development function and one server for production PB1 Proprietary Base 1 Operational activities connected to orders and job permissions Oracle RDBMS Central database PB2 Proprietary Base 2 Thickness measurements – for integrity tracking Oracle RDBMS Central database.

Concept Data source Rule Technical object SAP PM Equipment + functional location Maintenance request SAP PM Maintenance request RTI SAP PM RTI x Maintenance request items x Object description x RTI status Maintenance order SAP PM Maintenance order Thickness measurement PB2 Thickness measurement x functional location Operational activity PB1 Operational activity Job permission PB1 Job permission TAG N/A RTI text attribute 1 Demand N/A Maitenance request + maintenance order.

Scenario 6 Enrichments. Concept Enrichment Rule Maintenance request Add “situation” attribute Process values from attribute “STATUS” and map to a text using a table FROM-TO as reference RTI Add “elapsed time” attribute If RTI is yet open, process the difference between now and the opening time. If RTI is already closed, process the difference between the closing time and the opening time..

Scenario 6 Querying and searching in the Knowledge Base.

Scenario 6 Expanding the Knowledge Base. Now we would like to expand the Knowledge Base including information from equipment materials. These information are in S3D (the same datasource in Scenario 5).

Scenario 6 Expanding the Knowledge Base. There is a challenge in order to join S3D information with SAP-PM. S3D identifies the equipments by their location TAG and is unaware of SAP functional location. In RTIs some Petrobras sites have oriented the users to mention the location TAG inside RTI attribute 1, but the information is not always filled. Sometimes users fill the same information in field 2. Sometimes they use a general description textbox..

Scenario 6 Demo Plan. Item Demonstration How? Describe in PowerPoint how to crede the m ai ntenance knowl edge graph for queries/search es usi ng resou rces of the DataHub. Describe the pro posed End-to-end overview Of the solution achitecture of com onentsinthe DataHuh. Comment on the assum tions made. Dem onstrate the development of am ainte nance centrd repository with structured data i ngest ed from SAP-PM and the other RDMS sources Consider the specific Information gven in the slides(orders, maintenance requests, RTI reports equipment data operational activities, job pe missions, measurement records) and Sim ulate this inform ationcomingfrom datatables of the datasources. Here we wo uld like to see a raw layer popul ated with the original inform ation from the datasources, but if you already have a ption layer from SAP- PM, you can this as part Of the first layer. In this case show how this I er is built and ke u •to •date ust do the extra in stion from the other ROMS nurces. Demonstrate how to create data relationshi sin the Datal-i_lb. Connect the datain Sted Demonstrate how to do transformations on the s litin mer n cleanin Develo some to illustrate. Dem onstrate how to define and m æ-'age schemas in the central repository. Propose and imp lem ent a schema organizati on to the m ai ntenance m Odel in the DataHub. Dem onstrate how to histo and version data treate a maintenance know led* base Query the knowl base Extractingi nform on from unstructure d daa Create the mantenance knowledge base with the concepts shown in Scenario 6, Demonstrate how to create the logcal concepts. Show ho w to link them Sh ow alternatives of logical and physical represe ntation (I.e. knowl edge graph). Dem o nstrate how to im pl em ent the transform ation rul es descri bed in the slidesin order to conceive the conte tS. Dern onstrate how to enrich the repository by includingnew logical concepts or new dtributes based on mappingrules. use the slides as reference for enrichm ent exam les Demonstrate how to kee there ositor u -to-d*e concernin chan sin the data sources and comment on the strate es. Demonstrate how to add unstructured documentstothe maintenance central re osito readin filesattached in RTI re ortsat_ SAP PM Demonstrate how to link the RTlre ortfilesvvith the structured m aintenance data Sh ow if there IS any form of autom cally I lnk Inglogc concepts based on keys or attribut es that addresses the same concept in different sources. F or le technical 0b ectsthat .e referencedinthe measurements database xtechnlcal ob•ectsof maintenance re uests. Dem onstrate how to send uerl es to the m antenance knowled database to recel ve direct answers, Dern onstrate queries on the know base that could be interestingforthe maintenMce to do. He could for example do queries in order to check the maintenance backlo for aset Of e ui ments. Demonstrate how user ueri es to the knowled e base .etrd"slatedb the DataHub. Create som e fictitious docum ents (PDFs) of an RTI report with plant im ages, descriptive texts m entio ning equipments and TAGS, pl atform narne a.d tabl es with measurements. Store them in the mai ntenance central re osito asthe were in stedfrom SAP Dem onstrate an extraction pi peline that docum ents (batch), apply rules/m Odel Sfor recogn iti on of entiti es/attributes and enrich the knowledge base with this information in a structured manner Show how to add external models in order to do recognitions mentioned above. Show how to deploy th ese modelsinside the DataHub. Show how to connect to external models de 10 ed in Petrobras on- rem Ise environment inside an extraction i ellne. Show if ou have a catalo e Of existent modelsthat extract information from PDFS includin NLP techni u es. Pick one to demonstrate the i eline usin it. Dem onstrate exam oles of auervin ethe knowledee base in order to gather the inform ation e ktræ:ted in PDF ioinewith other inform ation from the mo del.

Scenario 6 Demo Plan. Item 5 6 7 How? Demonstrate how toex and the knowled e base with e ui ment materials comin from S3D datasource. Demonstrate how the LhtaHubcould solve the •oinin challen e mentioned in the scenario. Comment on how to build and store a Ina in table. Demonstrate a ue to atherall the endin maintenance re uests forcarEX)n steel i ui ments oft i e" with material 'API 5LGrB'• E_xpanding the knowledge base Create a virtual maintenance Demonstrate how the DataHubcould hel in fillin the ma in table with automatic extraction axle from RTI attributes. (extra challe nge) Denu)nstrate how to deal if the situation where the refere nce to the TAG is is nowhere except inside the RTI dcxuments. Could use this information to use asthe ma in k ? Howdo ou deal if the situation that uhave man TAGS mentioned in theRTlre Its. How to find the "tar et" TAG? Demonstrate how to create a maintenance virtual knowledge base with the same data (orders, mai ntenance requests, RTI reports, equipment data, o erational activities •ob rmissions measurement records without havin to co data if su (ted the (Xluct. Comment on the benefits, limitations and drawbacks of this alternative. knowledge base (extra challenge) Demonstrate how to ue data from differentori inal without havin to ex lici define the data source and how to •oin data. mmonstrate how to ue data from differentori inal but ex lici definin the data source and how to •oin data. Demonstrate if the DataHubcan help on indexing the maintenance data to allow queryi ng on text and documents by keywords, giving answers by relevance regarding the pieces of information found. The search needs to consider text inside structured attributes and also text inside the unstmctu red docume nts (RTI attachments in the exam le . on the knowledge base (extra Show some exam les on doin text searches in the knowled base su 'ted the indexin challenge) search features like score fuzz re ex hi hli ht. Demonstrate examples of search ing the knowledge base in orde r to gather the information extracted in PDFtogether with other information (only necessary if the search feature was develo ed . how to reco nize tammar structure in the texts art of s ech Demonstrate how to incorrx)rate ontology Is to the DataHub in orderto enrich the userqueries. mmonstrate on how to do disambiguations like U BRA mantics (extra cha Ilenge for a field or LIBRA for a curren Show other semantics features of the DataHub that could help in solving queries/searches in the maintenance knowledge base (if existent)..

Scenario 7 Offshore Reality Capture Media Gallery.

Scenario 7 Data. Reality capture media may include 360 photos, videos and point cloud data. Media is associated to context (which platform, module, floor, administration site, logical area…) Metadata may include campaigns info (name, data), coordinates, calibration parameters of the capturing device etc. One campaign can result in 70GB of photos (10.000), 15GB of videos and 60GB of point clouds (150). For all the campaigns in a year, the estimation is 11TB of photos, 23TB of videos and 9T of cloud points. Because of this large amount of data, transfer and storing optimization strategies are mandatory. After each campaign all media is saved to folders in a Unix File System, organized by platform and campaign date. Access to media need to be controlled by implementing authentication and authorization strategies..

Scenario 7 Enrichments. Some applications will also need to enrich the metadata, adjust the images or add user annotations to the repository. They can access data, process them and store back or execute some conversion or intelligent functions than can be plugged in the DataHub to serve more applications. The DataHub could also provide some built-in functions based on AI models and/or logical rules. Examples of functions: Recognize the technical object identifications (TAGs) from the images, given that they are appear on the scene (either painted or printed on plates) Link the TAGs extracted from the images with the equipment identifications in the Engineering Databases, dealing with inconsistency issues Apply AI algorithms for semantic segmentation – classifying turbomachines, sensors, instruments, pipes, people… Apply AI algorithms for corrosion detection Convert image coordinates to 3D space of the oil & gas platform Associate context information to media Include or update metadata information of media.

Scenario 7 Access to media in the gallery. Users may want to query by filters or search media by text concerning: Context Metadata Spatial envelope Applications would need APIs to access the media in order to manipulate them if needed, and also to get the associated metadata. Users could want to play or download the videos. The user is given a link to the video. Streaming support is needed in order to play and control the videos online from outside applications through APIs. As there are external users that not have access to Petrobras on-premises network, the gallery need to be served in the cloud. Media from the last campaign is more accessed than media from previous ones..

Scenario 7 Security. The captured images are sensible and the access to them need to be controlled by rules implemented in the DataHub. The roles need to consider: Company Administration site Oil & gas platform The list of platforms, companies, administration sites and relationships are registered in Petrobras databases and the access validation need to be automatized reading these bases online or consuming and API that receives the user and the platform and returns yes/no as the access right. One administration site has many platforms. A user is associated to some company. If the user company is Petrobras, it is associated to one or more administration sites and can see the images of all platforms from these administration sites. If not, it is associated to one or more platforms directly and has more controlled access..

Scenario 7 Demo Plan. rtpm Demonstrati.ml Describe in PowerPoint how to create the offsh ore reality capture media gallery using resou rces of the DataHu b. Describeth e proposed architecture of components in the DataHub. Ccnment on the assum ions made. Demonstrate th e ingestion of the offshore reality capture m eda from asi mulated campaign archiveto th e DataHub. Consider that the mediafil eswere origindly sued to folders in a Unix Filesystem, organized by liatform an d cam pägn date, so this I sthe datasouræ. The origind m etadata need to be preserved and accessible for queries. ton sider the volumet neratedb each carn ai n in order to definethe arch itecture. Demonstrate ho.v to sto version and m an e thet esof media mentioned inScenari07 30 hotos videos and ointclouddata Demonstrate hov.'toman ethe metadata assodatedtc each media In order to u date or creae extra attributes. Demonstrate h o." to associate context datato media (platform, m odule, floor, adrninl stration site, logical area) Te context needsto be managed in the OataHub as a relational model in estedfrom En ineerin databases, It needs to ass throu h automatic data u alit rocessi based on rules. Simulate this eline. th e DataHub an dis alreac$ availabl e/impl emented. Consider the enrichment examples described in the slides forScenario 7, mention th e ones you have or that you could Demonstrate ho.v to use of these built-in intelli ntfunctions. TJernonstrate hot" to addPetr0brasAl modes as functions in the OataHub. Demonstrate how tou date them if th are retrained. Demonstrate how to li nkTAGs extracted from mages with th e equi pment iden tification TAGS in th e Engi neeri ng Databases, deali ngwith incon siRen issues. Sim ulate a fu nction/integration in the DataHub that receives an and returns a list of TAGS. Demonstrate how to use this featureto create a link between the images an dthe technical o -ectsfrom theEn 'neerin Databases. Show the alternatives forim lementin thelinks. Demonstrate th e existent alternatives for an outside application to make aqustments to th e reality captu regal I ery images, processing th em an d storing back. Demonstrate how toinclude andu date metadata from inside usin the D2aH ub fu nction alibes Demonstrate th e existent alternatives for an outside application to include and update metadaa. Demonstrate th e existent alternatives for an outside a licatim to chan e media context. Demonstrate h ow to store pixel labels or ot*ect bounding boxes generated by A1 models or users in the repository (semanti c sem entati on). Demon strate h owto link them to the im ages. Demonstrate an exam leaf ue in media context. Demonstrate an exam leof ue media metadata attributes. Demonstrate an exam leaf searchin media b text in themetadata or context, Demonstrate tofindall I m es s atial envelo coordinates are In the metadata . For uam le mnsiderfindin all im es inside a ficticious cube volume. Demonstrate how tofind all images of tu rbomachines from a specific module. The list of modules and technical objects and their classification asturbornach ines are in the En ineerin Database. The im eswere reviousl linkedto the technical o •ects item Describe how to rovide APIsin the DataHubto man e and ue search the alle usi DataHubresources and host them tobeavailableto outside a lic2ions. Demonstrate ho.vtothe DataHub could offer video streamin su ottto I orcontrol videos. Describe available services andAPlsthat control video streami Comment on on• emises Strat iesfor hostin and servin the all to users. Media from the ast campaign is m ore accessed than media frorn previous ones. Demonstrate inssible dl anges in the architecture to take this into consideration and be more 0 imited. Corn ment a bait aforrnance and cost. Demonstrate how to cre2e the offshore redity capture gall stohng information in a media repository outside the DataHub, but acessi ble as a virtual repository from the DataHub. All th e queri es/searches to media needto be answered by the DataH ub but reading the repository as a datasource. Simu late thi s datasource and demon strate exam esof uenes search esto the DataHub in order to thermedia. mplementthe mles in the DataHub for control ling the access to the media in th egallery con cerni ngthe requirements described in the sl idesforScenario 7. The roles n eed to comply: company, administration site and platform. Consider both altem atives (readine a rel ation al Petrobras datasource or implementi ne an API - in side or outsi de).

Scenario 8 Industrial Standard Hub. Suppose Petrobras is willing to look at industry standards like OSDU, PPDM, RESQML, CFIHOS etc. in other to exchange digital data with other Oil & Gas companies. It also wants to use third-party applications that import/export data in these formats. For that, it has to translate its data to the industrial standard models (both directions), which requires some effort from Petrobras applications. The DataHub could help making this conversions, serving the industrial standard data models integrated with the internal models. It can also be the client for integration to remote repositories like OSDU..

[Audio] We are very excited to see your demo.. Scenario 8 Demo Plan.

[Audio] Hello guys! I'm going to present a second set of scenarios that we've put together for your DataHub product demo. We have in Petrobras a set of internal applications for Topside Integrity Management and we have to deal with a lot of data management challenges. We´ve brought some of them here in the form of use cases and we would like to see how your DataHub product features could contribute. We hope this material gives you the necessary information for implementing the demos. We are looking forward to seeing your product working on the challenges. Have a nice job!.