DIAGNOSTICS, PROGNOSTICS AND HEALTH MONITORING FOR ... · monitoring for operational gas turbine...
Transcript of DIAGNOSTICS, PROGNOSTICS AND HEALTH MONITORING FOR ... · monitoring for operational gas turbine...
DIAGNOSTICS, PROGNOSTICS AND HEALTH
MONITORING FOR OPERATIONAL GAS TURBINE
ENGINES
Konstantino Kouris
November 3, 2005
RESTRICTIONPAGES OF THIS DOCUMENT MARKED WITH THE NOTICES SET FORTH BELOW CONTAIN PRATT & WHITNEY PROPRIETARY DATA WHICH IS SUBMITTED IN CONFIDENCE. NO DISCLOSURE OR USE OF THIS INFORMATION IS PERMISSIBLE WITHOUT PRIOR WRITTEN CONSENT OF UNITED TECHNOLOGIES CORPORATION EXCEPT FOR OFFICIAL PURPOSES WITHIN THE U.S. GOVERNMENT.
NOTICESPRATT & WHITNEY PROPRIETARY – MIGHT CONTAIN LOCKHEED MARTIN / NORTHROP GRUMMAN / BOEING PROPRIETARY INFORMATION
October 2005UTC Proprietary 2
Diagnostics, Prognostics, and Health
Management
Diagnostics:
Estimate failure state at
time now
Health Management:
Take action based on
diagnostics/prognostics
Health Management: The capability to make
intelligent, informed, & appropriate decisions
about maintenance and logistics actions based
on diagnostics/prognostics information,
available resources and operational demand.
Prognostics:
Predict degraded state at
time > now
Prognostics: Material condition assessment which
includes predicting and determining the useful life and
performance life remaining of components
Diagnostics: The process of determining
the state of a system to perform its
functions.
With definitions borrowed from a Lockheed Martin presentation
October 2005UTC Proprietary 3
Optimizing Removals And Workscope
Early
Late
Time on wing (EFH)
MA
INT
EN
AN
CE
CO
ST
S
($ /
EF
H)
Monitor Health and Safety -- Prevent Unscheduled Engine Removals (UERs), Optimize performance
Proactively Manage Configuration – Extend Average Time On Wing (ATOW)
Plan Removals -- Minimize Disruptions
Optimize Workscopes – Reduce Shop Visit Cost (SVC) and Minimize Turn Around Time (TAT)
BUSINESS PROPOSITION:
LIFE CYCLE COST MANAGEMENT
REMOVALTODAY
IMP. PRODUCTCAPABILITY
EXTENDED TIME ON WING
RE
DU
CE
D M
RO
CO
ST
OPTIMIZEDREMOVAL
October 2005UTC Proprietary 4
Engine Health Monitoring Roadmap
2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2014
Sys
tem
Ca
pa
bil
ity
Transitioning Capability Programs
On-board Anomaly
DetectionLife
modelsLife
models
Oil Debris
Vibrations
Vibrations
Vibrations
Oil Debris
Oil Condition
Oil Condition
Oil Debris
APHM
Board
Filter
HealthFuel/Oil Filter
Health
Life
models
Off-board
Diagnostics
Mature Set of
Diagnostics Solutions
Off-Board
Gather engine data
reduce data
variation
Validate model
accuracy and false
detect rate
Execute
October 2005UTC Proprietary 5
Technology Transition Process
Measure of Technical Readiness
Technology
Demonstration
Fielded
Systems
Technology
Maturation
& TransitionNew
F100-PW-229
F119
Technology
Readiness
Level
9
8
7
6
5
4
3
2
1
TRL
Levels
In Service
Qualification/
Certification
Flight Test
System-level
Validation
Rig/Core
(Expanded
Design Space)
Rig Test (Minimal
Design Space)
Proof-of-Concept
Technology
Application
Basic Principles
October 2005UTC Proprietary 6
Off Board Diagnostics Facilitates On-
Board PHM Systems
A Mature Set of Diagnostics Solutions
Data collected few
times per flight
Filter, remove
outliers, validate data
Model parameters
Kalman Filter
based isolation
Off-board diagnostics
algorithms hosted within
secure database
Infrastructure facilitates
engine/fleet management
solutions as well
Detect trends
KF estimation of
deterioration
Report on the
web
October 2005UTC Proprietary 7
P&W Tools to Support DPHM
On Ground
Diagnostics and
Prognostics
Isolation
Trending
Early Warning
Detection
On-Condition
Management
Configuration and
Utilization Tracking
Modification
Standardization
Fleet Alert Summary
and Watchlists
Removal and Shop
Visit Planning
Shop Maintenance
Instructions
PWxxxx Operator: xxx
PW4074 Shipped Configuration: Bare
8848:52 Engine Time Since Last Shop Visit (TSLV): 2048:03
7451 Engine Cycles Since Last Shop Visit (CSLV): 1734
Document Number:
Revision Date:
Oil Type: Mobil Jet Oil 254
Test As Received: No
Preservation Method: Method II
Contract TAT Contract term 110
days from receipt.
Next Shop Visit SMI-
Planned O/H Location
Cheshire Other
Requirements:
Photos Required. Fit Requirements. Oil Sample
Analysis. Cleaning Requirement.
Engine Time Since New (TSN):
PW4000 112-Inch Fan Engine Maintenance Planning Guide
05-Oct-01
4. Life Limited Parts Summary (Optional - Provided Electronic
Attachment of Data is accomplished)
9. Last Shop Visit Information - Engine Test Results / MAP Results (Optional)
Test 9, Test as PW4077 and ship as PW4074. Post test MAP analysis required.
Engine removed for high time to perform Heavy Maintenance.
Expectations are to Heavy Maintenance the major modules so the engine can be returned to service for a full run interval of 7500 cycles or greater.
Additionally reliability and durability mods are called out for incorporation per Worksheet "11. SB Requir
Hold material for review with FMP Representative
5. Engine and Module Time and Cycle Information (Optional -
Provided Electronic Attachment of Data is accomplished)
3. Known Shortages at Time of Shipment (Optional)
1. Summary and General Information
Shop Maintenance Instruction - P777xxx
Records Included
Scrap Material
Requirements:
Engine Cycles Since New (CSN):
Summary and General Information
Shop Visit Expectations
All work performed in accordance with the following worksope specification
Engine Type:
Engine Serial Number:
Special Items
13. Service Bulletin Incorporation Status
12. Industry Item List from Production New (Optional)
11. Service Bulletin Incorporation Requirements
10. Engine and Module Basic Shop Requirements
Reason for Removal
2. Document Revision
Test
Requirements:
16. Additional Requirements (Optional)
7. Airworthiness Directive / Alert SB Compliance Status
6. Engine Condition Monitoring Trends (Optional) 14. ETOPS Requirements (Optional)
15. Records Requirements
8. Last Shop Visit Information - Engine and Module Work
(Optional)
Removal Planning /
Event Tracking
Modification Incorporation
On-board
Anomaly Detection
Models
Sensors
Algorithms
On Board
October 2005UTC Proprietary 8
ADEM Portal for Trending and Alerting
Engine Data
AD
EM
: A
dva
nce
d D
iagn
ostics a
nd
En
gin
e M
an
age
ment
October 2005UTC Proprietary 9
Engine Life Algorithms
Crack growth models
Thermal/stress analysis
Gas pathparameters
Missionanalysis
Advanced sensor data if any
Engine control parameters
Measure factors affecting actual life remaining
Engine used differently than worst-case
assumption
Aircraft flown differently
than expected
Engine-to-engine
variations
Finite element models
Thermal / stress algorithms
Reduction
Methods
On-board host (DEEC)
On-Board Life Management Algorithms
Flight Data
Life models
(probabilistic)
October 2005UTC Proprietary 10
On-Board Anomaly Detection
DEEC (with PHM board) has
new STORM
Normal continuous data & physics
used to build real-time engine models
Transient data and
real-time models used
to detect anomalies
Anomaly Detection:
• Provides trigger to capture real-time data for subsequent off-board analysis
• Additional data to support fault isolation
• Can facilitate long-term performance deterioration tracking
Causes for anomalies
• Deterioration
• Physical faults
• Operational abnormalities
• Flight domain excursions
Anomalies: Deviations from “Normal” Operation
October 2005UTC Proprietary 11
Advanced PHM Sensors
Early Detection of Various Problems
• Oil condition monitor
• Oil debris sensor
• Oil and fuel filter health
• Inductive / microwave probes
• Vibration pickups
October 2005UTC Proprietary 12
Oil Health Monitoring
• Sensors installed on production military
engines
• Condition monitoring added to OLS
• Tubes form capacitors for level and
conductivity
• A single probe provides lube oil capacity,
condition, and calibration signal to the box
• ODM located on supply side of main oil
pump
Technology Matured
Oil Level Sensor
Oil Debris Monitor
Transition process
• Analyze & develop algorithms
• Increase inputs in APHM board
• Prototype, test, and qualify APHM board
• Develop software
• Test and qualify system
October 2005UTC Proprietary 13
Data Normalization Leads to Enhanced
Fault Diagnostics
Normalized
inputs
Empirical prediction
models
Predicted
Deterioration
+ -
X (predicted)
Flight data is
gathered from
engine
Xnom
Ground Station
Monitored
Parameters
October 2005UTC Proprietary 14
Examples Of Data Normalization
Empirical models allowed reduction in variance by as much as
17 times while maintaining fault sensitivity
Reduction in variance for DWF = 4.7
DT6 = 17.5
Reductions for other variables: DT41 = 6.7
DN2 = 15.5
October 2005UTC Proprietary 15
Hybrid Modeling: eSTORM
eSTORM Implementation Phase
SVM
EngineANN
KF
+-Inputs ( u )
Residuals ( r )
Estimated residualsr̂
Tuners )ˆ( x
+
+
Adjusted EstimatedParameters
}*ˆ{ P
MeasuredParameters
}{ P
EstimatedParameters
}ˆ{ P
Standard STORM configuration
with feedback
Trained ANN will estimate Model/Engine
difference and adjust the estimated
parameters accordingly
Residuals will now track
performance changes
Hybrid Model
October 2005UTC Proprietary 16
Diagnostics Using Hybrid Models:
eSTORMSTORM vs eSTORM : Performance Fault Visibility
eSTORM disabled
eSTORM enabled
Performance
deviation
becomes visible
Module Performance
Deltas (Tuners) absorb
mismatch
October 2005UTC Proprietary 17
Diagnostics / PHM Needs
Diagnostics & Prognostics Analysis Tools• Signal processing (primarily vibrations)
• Empirical modeling techniques (Neural Networks, Support Vector
Machines, Partial Least Square, Nonlinear regression, etc.)
• Anomaly detection and fault isolation
• Directed troubleshooting
• Data mining
Diagnostics Sensors and Computing Hardware• Oil condition, debris, level – preferably low cost
• Robust clearance monitoring
• Acoustic sensors
• Aerospace qualified wireless and self powered sensors
• Distributed computing for health management and controls
• Data compression, storage, and secure transmission
Military engine Applications
Solutions Partially
availableSolutions available
Solutions Needed