Analytics in Schlumberger

Mohit Mittal
20th Nov 2020
Our History – More than 90 years of Growth

History of Innovation

History of Expansion

History of Acquisition

1930 1950 1970 1990 2010

 Oil & Gas Sectors

3
 Data Driven Equipment Monitoring – The Big Picture

VISION
Data driven decision making with zero
unplanned down time

Management Failure Prediction

MISSION

❖ AVOID or MITIGATE Non-Productive Time

❖ REDUCE Total Cost of Ownership
❖ IMPROVE Efficiency

Components/Issues and Trends

The Analytics Team

Global footprint Multidisciplinary

❖ Agile, Highly Competent Data

Comp Elect
Science Eng. Science & Digital Product
Clamart, France Development Team
Houston, US Stats
Mech
❖ Top schools
Pune, India Eng.

❖ Global Tech Centers

Control
Systems
Optimi-
zation
❖ Product Deployment Across
Segments

Working on high-impact industrial problems

Across Schlumberger Product Lines
 Application 1: Predictive Maintenance for Logging-While-
Drilling (LWD) tool
Problem description & relevance
• Multifunction Logging-While-Drilling (LWD) tool –
EcoScope
• Focus subsystem: Pulsed Neutron Generator
PNG - self-contained particle accelerator
• Typical cycle: 100 hours at temp 150C, pressures • Acquire 1,000 tool channels during runs
of 20,000psi, significant shock & vibration
• After each job, technicians in Maintenance Base
use tool data to decide on maintenance needed
before tool is sent for next job
• PNG functionality is critical to core tool
measurements.
• Any failure can have catastrophic consequences

Ref: Data-drive Fault Detection for Neutron Generator Subsystem in Multifunction Logging-While-Drilling Service – by Mosallam, Laval, Youssef, Fulton, Viassolo.
2018 PHM European Conference
 LWD tool maintenance: Old Vs New
Previous solution – manual data analysis New solution approach – advanced analytics
by experts • Construct Health Indicator (HI) from PNG sensor data
• Manual Data analysis - very time consuming, prone • Stat summary of selected channels
to error
• First principal component (PC1) from features
• Large number of potential failure modes, many
intermittent or only evident under the stress of down- • Empirical Mode Decomposition: from PC1 to intrinsic
hole conditions mode functions + residual signal

• Personal Biases reducing objective approach • Residual signal shows rate of change in variance with
time - developing PNG degradation
• Decision-tree, trained on HIs of different runs labeled as
healthy or faulty by expert

Channel Health Indicator

Preprocessing Modeling
Selection Construction

Ref: Data-drive Fault Detection for Neutron Generator Subsystem in Multifunction Logging-While-Drilling Service – by Mosallam, Laval, Youssef, Fulton, Viassolo.
2018 PHM European Conference
 LWD tool: Health Index
New solution approach – Advanced Analytics

Ref: Data-drive Fault Detection for Neutron Generator Subsystem in Multifunction Logging-While-Drilling Service – by Mosallam, Laval, Youssef, Fulton, Viassolo.
2018 PHM European Conference
 Application 2: Automated condition monitoring for BOPs
BOP is safety-critical drilling rig component; final
layer of protection from a hazardous blowout

Annular BoP Ram BoP

Ref: Towards Automated Condition Monitoring of BOP Wellbore Packers – by Park, Baker, Kumar Bade, Viassolo.
2018 PHM Conference
 Automated condition monitoring for BOPs: Problem description

• 45 out of 59 rigs operating in Gulf of

Mexico reported component failures of
well control related equipment in 2017
(BSEE)
• The integrity of packers (leaks due to
aging) must be assessed regularly by
simulating pressures on them
Wellbore Packers isolate pressure from well in
event of a ‘kick’. Packers during operation
(above) and during test or ‘kick’ (below) • Needs:
1) Test planning – to ensure time efficient
testing of all critical components
2) Once test data available, generate
accurate Health Index for each
component
BOP Stack with 19x components: 1x Annular, 2x
Shear Ram, 3x Pipe Ram, 11x Gate valves and 2x
Line isolation valves
Ref: Towards Automated Condition Monitoring of BOP Wellbore Packers – by Park, Baker, Kumar Bade, Viassolo.
2018 PHM Conference
 Automated condition monitoring for BOPs: Health Index
Health index evaluation Configurations

Automated Pressure Point

identification from test data

Health indices definition

HI(s) = max(Ts - s,0) / |Ts|
HI(d) = max(Td - d, 0) / d
HI = HI(s) + HI(d)

11
Ref: Towards Automated Condition Monitoring of BOP Wellbore Packers – by Park, Baker, Kumar
Bade, Viassolo. 2018 PHM Conference
 Automated condition monitoring for BOPs: Implementation

Condition Monitoring Process Flow Health indices for Middle Pipe Ram
for BOP Wellbore Packer over Mach 2017 – Feb 2018

Data Source Selection & Data Processing

Identification of test events

Feature Extraction

Health Index Evaluation

Classification of Test Configuration

Benchmarking and Health Decision

Ref: Towards Automated Condition Monitoring of BOP Wellbore Packers – by Park, Baker, Kumar Bade, Viassolo.
2018 PHM Conference
Application 3: Frac Pumps
Application 4: Reuse and Repair of Drill Bits
Drilling a Well requires:
• to overcome the resistance of the rock, crushing it into small
particles
• to remove the rock particles, while still acting on fresh material
• to maintain the stability of the walls of the hole
• to prevent the fluids contained in the drilled formation from
entering the well

Drill Bit

Rig Key component in BHA, as it can

enhance Drilling Economics

Bottom Hole Assembly (BHA)

• Lowest part of the drill string, extending from drill bit to surface
• BHA defines drilling objective of well, from formation evaluation to
maneuvering the direction of well to untap O&G reserves
Drill Bit Industry Challenges

• Manual repair instructions

• Hard copy repair packet
• No possible data analytics

Manual Inspection

• Manual cutter placement

• 7 out of 12,000 cutter
combinations
• Minimal TCO optimization
Bit Design
Digital Analytics on Drill Bits
BT

Light precise adaptive CNN Model

Custom build cutter tracker
State of the art accuracy for Mobile

2/8

High precision pixel level analysis

New segmentation algorithm

SL NO: AAXC231J

COSD

Type

Lifecycle
Custom powerful conv filters feature extraction
Using robotic arm for cutter analytics

US20190279356A1
System for Characterizing Oilfield Tools
Digital Journey for Drill Bits
Digital Enablers Implementation

• AI enabled objective and HD inspection

Digital Repair
• Digital and customized repair packet
Recommendations
• Full Digital Twin

Hi Def Data for • Damage data linked to client/geology

Design • Cutter placement optimized for:
• Performance
• Reliability
• Total Cost of Ownership
• SLB Cutter Manufacturing
Cutter
• Link damage to: geology, cutter types
Analytics
• Benchmark cutter performance
 Portfolio Snapshot

Bit Cutter Analytics HR – Retention Advisor

Downhole Tools PHM M&S Forecasting Engine

Supply chain & Business Optimization

Prognostics & Health Management
Vision Analytics

Text Processing & NLP

Rig Intelligence Platform
Contract Analysis Tool
Pump PHM Smart Field Material Transfer

Parts Inspection
HR Chatbot

ROTF Asset Analytics Lead Time Analysis Engine

Digital Factory
Performance Advisor
Closing Remarks
❖ Asset Health Management is a key enabler for:
➢ Lowering Total Cost of Ownership for O&G Tools and Equipment
➢ Improve Service Quality
➢ Reduce NPT
➢ Enhanced Availability & Productivity
➢ Situational Awareness and Planning

❖ New Business Models can bring new revenue streams for OEMs via Service
Agreements – pay by performance, pay by uptime, lease business models, etc.

❖ Though there are bottlenecks, recent advances in IoT, Cloud and Machine Learning
have made it easier to acquire, transmit, analyze and forecast equipment conditions,
enhancing predictive planning & resource management abilities
 Thank
You

Please post your queries in the QnA Chat Box