Digitalisation: How can we mix the "new oil" and the "old oil? The role of IT Research
1. Digitalization: How can we mix the
”new oil” and the old oil?
The role of IT research
David Cameron
Centre Coordinator, SIRIUS Centre
V November Conference, Rio de Janeiro, 7th November 2017
2. The SIRIUS Centre
Eight years’ financing from RCN
11 Industrial Partners
3 Leading Academic Institutions
Centre for Research-Based Innovation
Funding for 20 Ph.D. students
Innovation through prototypes and pilots
45 affiliated researchers
5. The business context for digitalization
• Digital asset lifecycle
management
• Circular collaborative
ecosystem
• Beyond the barrel
• Energizing new industries
6. Big rewards. Big challenges.
• Digital Asset Management Lifecycle
– Automation: USD 220(230) b benefit, 38000 jobs
“displaced”.
– Analytics: USD 425(525) b benefit
– Connected worker: USD 100 b benefit, 76000 jobs
”displaced”.
• How do we avoid the Integrated Operations
experience?
– OLF 2007 Promised USD 30 billion in 10-year potential
– At oil price of USD 43!
– Coincided with billions in increased cost. Source: WEF Report
OLF presentation on IO, Nov 2007
7. Machines
• Are machine learning and artificial intelligence
oversold?
• Are systems safe? Reliable? Resilient?
• Do we have the data? Is the data good enough?
• How do machines support better work practices?
• How much of our decision-making is decided by
science fiction?
8. Platforms
• Whose platform?
• Whose data?
• Platforms as the new silos?
• Sovereignty, ownership and integrity?
• Security?
• Design of platforms?
• Interoperability?
9. Crowds
• Capturing the contribution of each
engineer, geologist or planner
• Open supply chain to share data
• Work with vendors in co-design
• Expand access to start-ups and SMEs
• Build an agile culture without anarchy
• Self-service access to data and
analytics
11. Rubbish in = Rubbish out
Sensor
Analog
Transmission
A/D
Converter
Data
Processing
Control
System
Historian
Interface
Local
Bus
Historian
True Value
OPC
Data
Processing
Reported
Value
Data
Warehouse
ETL
Sensor
Surroundings
True but unknown.
Barrier between sensor and
quantity measured.
Calibrate and maintain.
Configure properly.
Configure properly.
12. The real world is a hard place
• Dirty, wet, cold, hot
• We haven’t done technology qualification just for
fun.
• Robustness, reliability and resilience is difficult
and expensive
• How many control loops do you have in manual?
13. Data lake or data swamp?
• Integration
• Security
• Integrity
• Maintenance
• Search
• Maintain Meaning and Context
14. Skills and domain knowledge
• Data scientist: a statistician with
attitude? The guru of the new age?
• Ideal is an expert on:
– Statistics
– Computer science
– Domain knowledge (physics,
engineering, geology, business…)
• Might an interdisciplinary team be
more useful?
• Might hybrid modelling be better
than raw empiricism?
15. E&P is an inherently social process
• Interpretation
• Judgment
• Experience
• Dialogue
• Mutual responsibility
• Team work
18. An interdisciplinary computer science approach
Knowledge Representation
Natural Language
Databases
Execution Modelling &
Analysis
Scalable Computing
Work Practices
Data Science
19. With friends with problems … and domains
• Our partners
• Geosciences at the University of Oslo
• Earth Observation in Tromsø
• Automation and computer-aided process engineering
groups
• EU Private-Public Partnerships
– BDVA ⇒ EFFRA, AIOTI
– A.SPIRE
20. From lab bench to products and services
Laboratory Projects
Innovation Projects
Fundamental projects: Oxford, NTNU
and Oslo
Laboratory
Prototyping
Project
Pilot
Project
Product or
Service
21. Optique: Digitalization of Geoscience and NDRs
Student in Petroleum Geoscience:
”I want all Gamma Ray logs from
wells that penetrate Rotliegend
deposits, with porosities larger
than 25% between 3°E-12°E and
50°N-60°N”
DISKOS – CDA – DINO – JUPITER
– German NDR
SEARCH
26. Operations: Planning
Knowledge Representation Semantic representation and manipulation of planning
concepts and data (ILAP).
Natural Language Analysis of free-text fields, manifests and work orders.
Databases
Execution Modelling & Analysis Analysis and optimization of plans and schedules. Simulation
and optimisation. Verification of plans and re-planning.
Scalable Computing HPC and cloud support for planning simulations and
optimisations.
Work Practices Effect of formal methods and simulation-based tools on
planning work processes and practices.
Data Science & Machine Learning Modelling of risk and adaption of plans and models to
observed behaviour.
27. Operations: Digital Twins
Knowledge Representation Semantic support for configuration, interfacing and
integration of digital twins. Ontologies for twins.
Natural Language Conversion from natural language documents to and from
semantic requirements. Interpretation of logs.
Databases Efficient triple-stores for use in digital twin
implementations.
Execution Modelling & Analysis Analysis and optimization of consistency, safety and
correctness of systems and their twins. Handling of events.
Scalable Computing HPC and cloud support for digital twins and supporting tools.
Work Practices Specification and implementation of fit-for-purpose digital
twins.
Data Science & Machine Learning Streaming analytics and events. Reconciliation and
alignment of data and models. Hybrid analytics.
28. Operations: Requirements
Knowledge Representation Semantic representation of requirements for storage,
sharing, transmission and reasoning.
Natural Language Conversion from natural language documents to semantic
requirements
Databases Efficient triple-stores for requirement systems
Execution Modelling & Analysis Analysis of execution of requirement fulfilment and
implementation in design systems
Scalable Computing HPC and cloud support for requirement reasoning systems
Work Practices Collaboration across engineering supply chain
Data Science & Machine Learning Monitoring and continuous improvement of engineering
performance