Neo4j in Oil & Gas: Industry Use Cases and Impac

Neo4j in Oil & Gas: Industry Use Cases and Impac

1. Focus Use Cases: Oil
& Gas
Ezhil Vendhan
Principal Partner Architect APAC

2. HAS_LINKED_IN
About Me
2

3. You are no Stranger to Graphs!
3

4. And all these…
The Internet The Metro Molecules
Financial Transaction Defence Ops Plan
4

5. Employees
Suppliers
Product
Customer
Transactions
Process
Model your data
like your business
Network & Security

6. Neo4j Graph Data Platform
DATA SOURCES USE CASESINGEST
Structured
Unstructured
DATA ANALYTICS
DATA MANAGEMENT
Journey Analytics
Risk Analytics
Churn Analysis
What-if Analysis
Feature Engineering
& ML
Fraud
Recommendations
Data Fabric
Data Compliance
Data Governance
Data Provenance
Data Lineage
Next Best Case
Ontologies
Neo4j
Bloom
Neo4j
GDS Library
PRODUCT COMPONENTS
APOC
VISUALIZE
DRIVERS & APIs

7. Neo4j Graph Data Science: 70+ Algorithms
Pathfinding &
Search

•A* Shortest Path
•All Pairs Shortest Path
•Breadth & Depth First Search
•Delta-Stepping Single-Source
•Dijkstra's Single-Source
•Dijkstra Source-Target
•K-Spanning Tree (MST)
•Minimum Weight Spanning Tree
•Random Walk
•Yen’s K Shortest Path
Centrality &
Importance

•ArticleRank
•Betweenness Centrality & Approx.
•Closeness Centrality
•Degree Centrality
•Eigenvector Centrality
•Harmonic Centrality
•Hyperlink Induced Topic Search (HITS)
•Influence Maximization (Greedy, CELF)
•PageRank
•Personalized PageRank

Community
Detection

•Conductance Metric
•K-1 Coloring
•K-Means Clustering
•Label Propagation
•Leiden Algorithm
•Local Clustering Coefficient
•Louvain Algorithm
•Max K-Cut
•Modularity Optimization
•Speaker Listener Label Propagation
•Strongly Connected Components
•Triangle Count
•Weakly Connected Components

Supervised
Machine Learning

•Link Prediction Pipelines
•Node Classification Pipelines
•Node Regression Pipelines






… and more!
Heuristic Link
Prediction

•Adamic Adar
•Common Neighbors
•Preferential Attachment
•Resource Allocations
•Same Community
•Total Neighbors
Similarity

•K-Nearest Neighbors (KNN)
•Node Similarity
•Filtered KNN & Node Similarity
•Cosine & Pearson Similarity Functions
•Euclidean Distance Similarity Function
•Euclidean Similarity Function
•Jaccard & Overlap Similarity Functions
Graph
Embeddings

•Fast Random Projection (FastRP)
•FastRP with Property Weights
•GraphSAGE
•Node2Vec
•Collapse Paths
•One Hot Encoding
•Pregel API (write your own algos)
•Property Scaling
•Split Relationships
•Synthetic Graph Generation

8. Graph is enterprise-grade
Up
to
Vertical
scaling
Unified DB for
analytical
& operational
workloads
Guarante
ed
Uptime
SLA
TB Graphs
with
Sharding
Integrations
with
Any cloud.
Any
workload
Encrypted
Data
SOC2
Type 2
HIPAA
Compliant
Graph
Algorithms

9. Neo4j on Azure
Connector for Apache Kafka
Connector for Apache Spark
Connector for BI
Graph Data Science Graph Database
Bloom
Data Warehouse
Connector
Azure ML
Azure Data Factory
Azure Databricks
Azure Synapse Analytics

10. Graphs in Oil & Gas
X360°
●Asset 360°
●Capital Projects Total Cost Visibility
●Digital Twin / IoT / Predictive Maintenance

EDF (Enterprise Data Fabric)
●Data Discovery
●Entity Resolution/Identity Management
●IOT Smart Grid

Process
●Energy Trading
●Opportunity development
●Worker Safety

Risk
●Regulatory Compliance / Internal Audit
●Supply Risk / Dependency / Distribution
●Cybersecurity

11. Asset Integrity and Predictive
Maintenance with Operational
Digital Model
11

12. “Imagine if every piece of equipment in our operations
could communicate its status in real time, alert us before
failures occur, and allow us to test changes before
implementing them. This isn't the future—it's happening
now with Operational Digital Model and Neo4j”

13. Refinery Reliability &
Predictive Maintenance

● 92% of maintenance-related shutdowns
were unplanned, primarily due to issues like
leaks in piping and other critical
components.

● Unplanned Outages reduce capacity,
increasing margins by between $6 and $12
per barrel of crude oil

● Reliability-related lost profit opportunities
can range from $20 million to $50 million
per year for mid-size refineries



McKinsey & Company
13

14. Existing Systems & The Why
●Static Models & Simulations
●Partial real-time integration
●Isolated systems
Operational Digital Model
Predictive Maintenance
Asset Integrity Management
Data Silos & Integration
complexity
●Condition-based monitoring
●Schedule predictive interventions
●Standalone Predictive tools
●Periodic Inspections & Compliance Checks
●Risk based approaches
●Use of Asset Management Systems
Inadequate relationship
modeling
High costs &
technical expertise
Limited real-time
visibility
Manual Processes &
Human error
False Positives
Reactive approaches

15. EnXChange
●Up to 30% reduction in
energy consumption

●10-15% to as low as 2%
predicted reduction in
utility line loss

●Reduced peak demand
costs

“You cannot create transformational change without transformational technology. That’s what
graph represents in this industry.”

– David Swank, CEO, enxchange

Operational Digital Model & Predictive Maintenance for Energy Grids
https://neo4j.com/case-studies/enxchange/
https://www.youtube.com/watch?v=Xo_2xUijD4Y

22. Supply Chain & Logistics
Optimization
22

23. ●Static Models & Simulations
●Partial real-time integration
●Isolated systems
Supply Chain
Data Silos & Integration
complexity
Inadequate relationship
modeling
Inefficient Routing
Lack of real-time
visibility
Risk of Disruptions
Existing Systems & The Why
Complex Decision Making

24. Supplier
Inventory
Product
Uncover patterns in your

Suppliers
Demand Shaping
● Customer Segmentation
● Price & Promotion Planning
● Demand Forecasting

Inventory Positioning
● Material Flow Plannings
● Procurement Doc Analysis
● Inventory Planning/ Optimization
● Risk Management
Perfect Fulfilment
● Distributed Order Planning
● Assisted Order Picking
● Packing Visual Inspection
● Fleet Routing Optimization
● Multi Resource Ops Planning
● Contact Center AI

Sustainability
● Sustainability risk forecasting
● Greenhouse gas emission monitoring
and forecasting

25. Supply Chain as a Graph
Knowledge
Graph
Data
Silos
IoT
Sensor Data
Use
Cases
Demand Shaping
Inventory
Positioning
Fulfilment
Demand SupplyManufacturingWarehousing Order
Fulfillment
Transportation 3
rd
Party
Data

26. Supply Chain/ Logistics: How We Help
Customer Pain Points
Unable to complete whole
network impact analysis

Slow Root Cause Analysis

Costly and Slow BOM
Management
Impact
Unable to plan for scenarios such as
geopolitical events, extreme
weather, supplier risk, etc





Increases the impact to customers
of an issue in the network





Inefficient ordering and shipping of
components






Slow Tracking and Routing
Business unable to scale operations
based on legacy system






Customer Results with Neo4j
Companies can simulate 1000’s of scenarios and
the impact on their whole network in real time.





Complex analysis is 7.5x faster resulting in
quicker, cheaper and more accurate ordering of
components





Knowledge Graph of logistics network allowed for
real-time routing with a single source of truth
resulting in reduce costs and increased network
capacity





Identify the root cause of an issue not just where
the symptom occurs in real time

27. Fraud Detection & Operational
Technology (OT) Network
Protection
27

28. Relational Database Systems Fall Short
Inability to
Uncover
Complex Fraud
Patterns
Difficult to
Adapt to
Evolving
Threats
Can’t Scale to
Deliver Rapid
Analysis
Analyzing data without
relationships doesn’t work

Pathfinding and entity
resolution need
specific algorithms

Identifying complex,
cross-entity fraud becomes
difficult or impossible

Finding recursive fraud patterns
doesn’t scale

Fraud evades detection because
queries take too long

Hard to connect disparate data
sets without refactoring the
whole database

Coding fraud patterns in SQL is
time-consuming and results in
brittle queries

29. Expose Intermediaries and Find Fraud
Resolve entities using cypher queries and
graph algorithms like Node Similarity and
Weakly Connected Components
Uncover connections between fraudulent
actors and intermediaries using pathfinding
algorithms like Yen’s, Delta-Stepping
Single-Source,
and Dijkstra Source-Target
Make linkage predictions between
identities in the graph, create unified
views of individual identities, and group
and classify communities of nodes using
node similarity and community detection
algorithms
Known
Fraudster 1
Registered
Address
Registered
Phone Number
Registered
Business
Known
Fraudster 2
Name: Johan Nordberg
Age: 32
Occupation: Pilot
Employer: SAS
Address 1: Arlanda, Sweden
Address 2: Uppsala, Sweden
Bank Acct: XXXXX934
Name: Johan Nordberg
Age: 32
Occupation: Pilot
Employer: SAS
Address: Uppsala, Sweden
Bank Acct: XXXXX934
Name: Johan Nordberg
Age: 32
Occupation: Pilot
Employer: SAS
Address: Arlanda, Sweden
Bank Acct: XXXXX934

30. Why Cypher?
Find larger fraud patterns without changing
queries using Cypher’s Quantified Path
Patterns and recursive pattern matching
Incorporate new fraud patterns into the
code quickly using Cypher queries that
are shorter than the equivalent SQL
queries or programmatic code
--Find Fraud Ring 3-10 txns

MATCH ring=(a:Account)
(()-[:PERFORMS]->()-
[:BENEFITS_TO]-()){3,10}(a)
RETURN ring;

31. …or Spend Time Writing SQL
--Find Fraud Ring 3-10 txns

select a.account_id, a.first_name, a.last_name,
a.city, a.state, a.zip,
t1.*,
a2.account_id, a2.first_name, a2.last_name,

32. …or Spend Time Writing SQL

33. Fraud Detection: How We Help
Customer Pain Points
High false-positive & False
negative rate

Slow processes to detect
fraud

Lack of fast, explainable
results
Impact
Legitimate transactions being
declined, loss in revenue &
customer satisfaction





Risk severe financial losses and
reputational damage





Higher time to resolution resulting
in customer frustration






Slow application response
Higher database operational cost;
high compute and storage cost for
application






Customer Results with Neo4j
25% reduction of false positives





Increased Fraud Detection by 300%
Provides fraud indications in under 200ms


Identified €1M of new, real fraud during the
pre-sales POC

34. Uncover patterns in your

Assets
Location
Access
Patterns
Network & Security
Account/Identity Control
Adaptive & Intelligent
Reputation Scoring
Threat Detection
Access Control
Zero Trust

35. OT Network Security with Neo4j
With Neo4j Cybersecurity will have:

●Comprehensive Network Modeling
○ Model the entire OT network as a graph, capturing all
drilling, refining, and pipeline management.

●Advanced Threat Detection
○ Apply graph algorithms to identify unusual patterns and
anomalies indicative of cyber threats.

●Real-Time Monitoring and Alerts
○ Set up alerts for specific patterns that match known
attack signatures

●Compliance and Reporting
○ Provide audit trails by tracking changes and incidents
over time

●Risk Analysis and Mitigation
○ Identify critical nodes and vulnerabilities

“Defenders think in lists. Attackers think in
graphs. As long as this is true, attackers win.”

John Lambert,
General Manager,
Microsoft Threat Intelligence Center

36. Uncover patterns in your

Assets
Location
Access
Patterns
4 min
to determine
blast radius
Used to take hours
or even days
100M
Customers
Protected
Network & Security
Account/Identity Control
Adaptive & Intelligent
Reputation Scoring
Threat Detection
Access Control
Zero Trust

37. Regulatory Compliance & Risk
Management
37

38. ●Manual Processes & Spreadsheets
●Disparate Systems
●Periodic Inspections & audits
Regulatory Compliance
Complexity of regulations
Inadequate relationship
modeling
Regulatory Changes
& Updates
Siloed systems
Reactive Risk
Management
Existing Systems & The Why
Data silos & fragmentation

39. Carbon Management Compliance Needs Neo4j
Features

●Integrated Emission Data Management

●Real-Time Impact Analysis

●Automated Compliance Checks

●Proactive Regulatory Compliance
Management

●Risk Relationship Analysis

●Streamlined Reporting and Auditing
Carbon Emissions
Natural Sources
Human Sources
Ocean / Atmosphere
Exchange
Volcanism
Fossil Fuels
Industrial Processes
Respiration &
Decomposition
Land Use
Facilities
Equipment
Sensors
Data
Now What?
Infrastructure

40. Reference Architecture
Azure IoT
Hub
Raw
Stream
Sensor Alerts & Sampled
Stream
Neo4j
ODBC
Connector
ETL
Pipeline OLTP
Azure TS
Insights
Azure Blob
Store
PowerBI
Reports
Azure
Cosmos DB
Azure Data
Warehouse
Azure SQL
Azure
Stream
Analytics
Notification
Services
Event
Sources
Azure Device
Provisioning
Neo4j Digital
Twin Graph
Neo4j
Bloom
Visualizat
ion
Unstructured Data, JSON Documents, Structured Data
Raw
Stream
Hot Path
Warm Path
Web Apps /
GraphQL API
Neo4j Secure
BOLT Driver
Power BI Server
Enriching
data
sources
Neo4j
Graph Data Science
Neo4j Operational Digital Model integrates a
wide variety of data sources (beyond BOM +
Sensor Data) to add additional analytical
context to the graph.
●Vendors
●Costs
●Compliance
●Schematics
●Service Records

41. Key Takeaways
Neo4j based solution

•Real-Time Insights and Analytics

•Advanced Predictive Maintenance

•Simulation and Scenario Analysis

•Integrated Digital Twin Modeling

•Regulatory Compliance and Safety
Amplified Benefits to Petronas

•Operational Excellence

•Cost Savings

•Enhanced Safety and Compliance

•Innovation and Competitive Advantage

•Supply Chain Optimization

42. Demo

45. Thank You