Serigne Gaye Senior Industry Consultant TERADATA
Transcript of Serigne Gaye Senior Industry Consultant TERADATA
Lessons learned in Heavy Assets Predictive Maintenance
Serigne Gaye
Senior Industry Consultant
TERADATA
5 Octobre 2017
– Demand is increasing while budgets are shrinking
– Infrastructure aand Rollong Stock are under stress
– Regulatory bodies are more demanding on Quality
of Service
– Trains have to be on-time and mission ready
– MRO costs need to be drastically reduced
– Sacrifying Customer Service for Cost reduction is not an option
Challenges
Doing more with Less
Safety Availability Costs Maturity
Air Forces X X X
Airlines X X X
Aircraft Manufacturer X X X
Aircraft Engines Manufacturer X X X
Railway Company X X X
Rolling Stock Manufacturer X
Turbine Manufacturer X X X
Truck / Tractor manufacturer X
Manufacturing Process X X
Electrical Grid X
Telco X
Mining Industry X X X
Facility Management
Ol & Gas
Car manufacturer
Maturity driven by safety, availability and TCO
Even though Railway
pays lot of attention
to availability and
TCO, it’s still lagging
behind on predictive
maintenance
Hard constraint
Average constraint
Soft Constraint
Use Cases
• Predictive Maintenance program has produced for a fleet of 120 helicopters: – 11 % increase in fleet availability – 41 000 Maintenance Man Hours saved
in a year
US Army
Caterpillar
• 3350 trains/day on a 32,000 miles network
• Infra red sensors installed every 20 miles – 20 millions readings / day – Monitioring of mechanical wear,
overheating and derailments – 1500 problems qualified every day
• Reduction of damages, delays and customers claims
• Monitoring and analysis of Truck engines sensor data has led to 100 M$ savings in executing Performance Based Contracts
Navistar
• Telematics data is enabling – Failures predicton – Fleet operator notification – Proactive routing od spare parts
• Logistic response enhanced from 7 to 2 days
More availability and 5-10 % reduction in life cycle costs
Union Pacific
Siemens Vision 2020
http://www.teradata.com/Resources/Videos/Siemens-Capitalizing-on-
Digitization,-the-In
Teradata wins a Stevie® Award Teradata was named a Bronze Stevie Winner for the video "Siemens: Speeding Down the Path of a Successful Future with Big Data and Analytics" in the “sales & technology video” category.
Identifying risks and anticipating breakdowns
Sensors continuously gather a breadth of data
Most of the data is analysed on the fly
Drafts are identified
Alerts are triggered on abnormal situations
Predictive Maintenance
• «Performance-based-contracting
• One delay out of 2300 trips
• Valero E est the reliable RENFE High Speed Train
• Big market share on Madrid – Barcelone leg
Identifying risky situations
Market share Railway vs Airlines
Benefits
Diagnostics Health & Performance Awareness
Working OK Failure occurs Collateral damage Precursor Faults Drift starts Fa
ilure
cyc
le
An
aly
tic
al
fu
nc
tio
na
lity
Monitor and trend fault progression
Predict Continuously Remaining Life
Assess health
Prognostics & Health Management
Incipient Fault propagation
Detect and isolate precursor faults Identify drifts
Explore events prior to failure
Unveil failure signature
Impact of Predictive Maintenance
Before After
Preventive Corrective Preventive
Predictive
Corrective
The amount and mix of Maintenance workload will change
− Develop a deeper knowledge on
equipment health and aging effects
− Improve Equipmement reliability and
maintainability
− Optimise Maintenance Planning and
Execution
– Make better trade off between
Repair vs. Leave in field with Partial
Mission readinesss
– Increase the sharing ot equipmenent
life cycle data
– Process Reengineering
Focus on Predictive maintenance
Health and performance Surveillance Scrutinise asset operations & pathologies
Material Management Predict spare parts demand and align supply & inventories
Maintenance & Repair Efficiency
Accelerate MRO tasks completion
Intelligent Asset Tracking Analyse asset demography
LifeCycle Costing Planning and controlling MRO costs over the asset life
Maintenance Planning Fine tune the maintenance ailor maintenance pro
Asset Improvement & retrofit Upgrading and extending asset capabilities
Tra
ck
Pla
n
Execu
te
Value first ! Post Maintenance
Test Hours
Gains
Operations interruptions
Over Maintenance
Maintenance Scheduling
False alarms & No Faults
Found
OPEX
CAPEX
Spares
Maintenance Man Hours
Energy Consumption
Inspection Hours
Maintenance Induced Damage
Collateral Damages
Event Driven
Maintenance
Supply Chain
Pipe Line
Inventory Sizing &
Mix
Repair vs.
Replace
Obsolescence
Asset Life
Fleet size
Logistic footprint
“Waiting for Parts”
Turn Around Times
Focus first on organs that actually drive failures
Shafts
Gears
Bearings
compressors, turbines, pumps, motors, etc
Leverage any kind of data
BEARING FAILURE
ODM, EBM, VIBES, RLI, SWAN
COMBUSTOR/HPT EROSION
EDMS
BEARING FAILURE
ODM, EODM, EBM, VIBES, RLI,
SWAN
FAN 1 ROTOR TIP/ CRACKS
(ECS, BVM, AFD, IDMS,
EDMS, VIBES, RLI, SWAN
BULK OIL
DEGRADATION
OCM, ODM, EBM
VAPOR PHASE COKING
OCM, ODM, EODM, EBM
BULK OIL CONTAMINATION
OCM, EBM, ODM, EODM
OVERFLOWING / BLOCKED
FUEL NOZZLE
EDMS, VIBES, RLI, SWAN)
GEARBOX FAILURE
VIBES, RLI, SWAN
LPT BLADE RUB
EDMS,VIBES, RLI,SWAN
RLI Robust LASER Inferometer
SWAN Stress Wave Analysis
PZT Piezoceramic Patch Crack Detection
AFD Acoustic FOD Detector
IDMS Ingested Debris Monitoring
BVMX Blade Vibration Meter
ODM Oil Debris Monitor
EBM Electrostatic Bearing Monitor
EDMS Engine Distress Monitoring
ECS Eddy Current Blade Sensor
VIBES Vibrations sensor
EODM Eletrosatatic Oil Debris Monitor
Failure modes
Understand the dynamic of the machinery
Sensor Data Data characteristic
Shafts
Gears
Bearings
• Failure of outer/inner races, rollers or cage
• Tooth pitting
• Gear crack
• Wear
• Misalignment
• Unbalance
• Bend
• Crack
• Rubbing
• Vibration
• oil debris
• acoustic
• emission
• Vibrations
• high noise in raw data
• High noise
• Signal modulatIion (coupling with bearing, shaft transmission path)
• Vibration signal is relatively clean and harmonic
• frequency components of rotating speed can indicate the defects
• Vibration
• oil debris
• Acoustic emission
Pick and train relevant analytical technique
Fourier Transform Analysis
STFT Analysis
Wavelet Packet Decomposition
Autoregressive Modeling
Feature extraction
Pattern Recognition Forecasting (Remaining Life)
Diagnosis
Feature Map
Hidden markov Model
Bayesian Network
Neural network
Logistic Regression
Feature Map
Statistical Pattern Recognition
Hidden Markov Model
Autoregressive Moving Average
Match Matrix Prediction
Fuzzy Logic Prediction
Neural Network Predicton
Visualise and share
Flux
Hiérarchie
Accord
Affinité
Enlarge progressively your data scope
Operational data Engineering Data Repair Shop DAta Supply Chain Data
PDM data CRM Data Key Functional Scope
Track Schedule Disptach
Equipment Environment
Client Operations
Subsystem
Technical event
Maintenance Task
Maintenance Report
Maintenance Work Order
Maintenance Plan
Configuration Envelope
Engineering Change
Maintenance Policy
Maintenance Instructions
Supplier
Organ Manufacturer Delivery
Transport Localisation
Logistic Provider
Purchase Order
Inventory
Warranty
Claim
17
Adopt a step by step approach
Assess Plan Prioritize
Data Readiness Identify data
requirements and their
availability, quality and
timeliness
Prioritisation
Assess BIO value
versus deployability
9
3
5
7 10
Deplo
yability
Business impact
+
- -
6
11
2 8
4
12
13
Low
High
Ea
sy
Ha
rd
Low
High
High Value
Easy to deploy
Data ready
Data Readiness
Process Change
Deployment Roadmap
Reliability
Availability
Inventory Sizing
Turn Around Times
Service Chain Visibility
Obsolescence
Health
Maintenance schedulIng
Q1 Q2 Q8
Time
. . .
1
+
Develop/mature BIO
content
Cash
flo
w
Cost
reduct
ion
Cost
avoid
ance
Revenue
Business Impact Identify & valuate
Business improvement
Opportunities (BIOs)
Bu
sin
ess
va
lue
In-service Performance
Detectability
Impacts of Mods
Maintainability
MRO policy
Q5
Prioritisation
Contract Fulfilment
Asset contribution
Segmentation
Costs drivers
Profitability
Lifecycle costs
Inventory Sizing
Costs drivers
Segmentation
Q & A
Thank you !