-
Notifications
You must be signed in to change notification settings - Fork 13
Home
The project OneSeismic is part of the umbrella project NGRMDF (Next Generation Reservoir Monitoring Data Flow. The project aims to create a scalable solution for storing and accessing seismic data in the cloud. NGRMDF is a project led by Johan Sverdrup Petec, owned by the Digital Centre of Excellence (DCoE) and financed by a consortium of assets, namely Snorre, Johan Sverdrup, Grane and Johan Castberg. Their main objective is to have access to more efficient ways for analysing 4D seismic data.
For any building workflows around seismic data in a cloud setting, an efficient way to store and access these volumes needs to be in-place. This includes an ingest pipeline, a storage format, “nearly” random access mechanisms, as well as a consistent and scalable information model. Lastly, a more abstract layer of APIs should serve the needs of any application interfacing with seismic data (e.g. for visualization purposes). Thus, building a visualization component utilizing this infrastructure is equally part of the scope. The project could be described as laying the fundament for seismic data in the cloud.
The scope of the project is centred around the core functionality – being able to programmatically extract values from a regular cube along an arbitrary surface quickly and in a cloud setting. This functionality should serve two purposes: 1. Enable entirely new workflows with seismic data from a data science perspective and 2. Facilitate the development of any application using seismic data for e.g. (co-) visualization, reporting, etc.
In order to serve as an enterprise solution, it needs to be scalable in terms of storage, processing and the data model. Initially, the scope should focus on 3D datasets, but in order to not accumulate too much technical depth, generalize enough to extend the number of dimensions at a later stage.
- An ingest pipeline for seismic data (segy)
- Interface for loading and attaching metadata to seismic cube
- Loading and conversion into storage format (“subcubes”)
- Lower level APIs for slicing the seismic cubes along axes (in-, crossline, time/depth), along a wellsection, along an arbitrary horizon
- Higher level APIs for resampling and visualization purposes
- Visualization component that can be integrated into other applications in x.y and z,y (not 3D)
- Alignment with OSDU initiative (evaluation of OVDS format)
- Software Innovation Bergen – Team lead Jean-Paul Balabanian
- Petec Oseberg - Matteo Ravasi
- Enterprise Data Management (EDM) –Ivar Leander Johannessen Sørheim
The project should deliver a prototype including ingest and lower level APIs for testing in Q4 2019.
By Q2 2020 the PRM assets should be able to upload their seismic volumes and build simple applications using the lower-level APIs. The number of cubes included in the solution will then be around 20-100.
Q3 2020 should see the usage of the visualization component in another web-based application utilizing seismic data (e.g. REP, etc.).
Throughout 2020, the architecture should be iterated on and optimised while scaling it to more assets and exploration.
The project is constrained by a relatively diverse stakeholder picture, which currently only exists of PRM assets, but will quickly expand once an MVP is released. This will then include EXP next to all Petec assets (essentially all projects working with seismic data). The solution aims to be an enterprise wide solution for using seismic data (in a cloud setting)
The main challenge for this project is to strike the balance between fast data access mechanisms and cost-efficiency. As the data volumes in question are high (several PB at full scale), optimizing the architecture can have a large cost-impact.
Equally, a large number of datasets require an information model that is consistent and flexible to accommodate all different types of datasets. This needs to be developed by EDM and is crucial for scaling the project.
Maximilian Georg Schuberth (Project Lead)
PNF JS PH2