Document
www.nasa.gov Donald L. Simon Cleveland, OH 44135 21000 Brookpark Road NASA Glenn Research Center Cleveland, OH July 28-30, 2014
Engine Test Results
AIAA Joint Propulsion Conference 2014 Aidan W. Rinehart Vantage Partners LLC Brook Park, OH 44142 3000 Aerospace Parkway
Integrated Architecture for Aircraft Engine
Performance Monitoring and Fault Diagnostics: www.nasa.gov
Overview
Background Architecture Application Results Conclusion
• • • • •
www.nasa.gov Data Transfer Denotes notional “snapshot” measurement point Example Aircraft Engine Flight Data Altitude Fan Speed Temperature Fuel Flow Ground Station Exhaust Gas
aircraft engine performance trend
–
monitoring and gas path fault diagnostics
Background
Ground-based Processing of “snapshot” measurements post-flight Enables estimation and trending of engine performance and gas path fault diagnostics Early diagnosis of incipient fault conditions with minimal latency can be challenging Advances in on-board processing and flight data recording capabilities are enabling new diagnostic approaches Acquisition of full-flight streaming/continuous measurement data now possible Requires new approaches to analyze expanded quantity and format of data – – – – – – – Conventional Approach: Emerging Approach: • • www.nasa.gov
Monitoring and Fault Diagnostics
Architecture for Engine Performance
Estimation and trending of deterioration-induced engine performance changes Detection and isolation of gas path system faults
– –
Designed for processing real-time continuous (streaming) engine measurement data to provide:
•
www.nasa.gov
Real-Time Self Tuning Model
Application for underdetermined estimation problems Minimizes mean squared estimation error in parameters of interest – – Self-tuning piecewise linear Kalman filter design Applies NASA-developed optimal tuner selection Provides real-time estimates of unmeasured engine performance parameters
• • •
www.nasa.gov which is used to improve model-to-engine tracking , r y u.
Performance Baseline Model
Actuator commands, Power reference parameter, capability. Periodic model tuning parameter updates from RTSTM to account for gradual degradation effects.
– – – Piecewise linear state space model design, open-loop with inputs: PBM provides a baseline of recent engine performance
• •
www.nasa.gov ~ y R T
~ y j ˆ ~ y ~ y
R WSSR T
j ˆ ~ y ~ y
signal.
Fault Diagnostics
j (WSSR) WSSEE Monitors residuals between sensed engine outputs and PBM estimated outputs Fault detection is performed by calculating and monitoring a weighted sum of squared residuals Upon fault detection, fault classification is performed by identifying the candidate fault signature that most closely matched the observed residual in a weighted least squares sense.
• • • www.nasa.gov Sensed Measurement Estimate
are calculated
WSSR Anomonly Detection Threshold
WSSEE
and
Time (sec)
WSSR
Fault Diagnostics
Gas Path Parameter WSSR 5: LPT 6: B25 2: LPC 4: HPT 1: FAN 3: HPC 0: No Fault
At each time sample a new
www.nasa.gov Fault Identification value is classified as the fault type.
Fault Diagnostics
WSSEE
0.1 0.2 0.3 0.05 0.15 0.25 0.35 WSSEE
The smallest
www.nasa.gov Boeing C-17 Globemaster III Pratt & Whitney F117 Turbofan Engine th Faults include station 2.5 bleed valve and 14 stage bleed valve faults
Propulsion Research (VIPR) Engine Test Data
o Ongoing at NASA Armstrong / Edwards Air Force Base Partners include NASA, US Air Force, Pratt & Whitney, and others Boeing C-17 Globemaster III Equipped with Pratt & Whitney F117 high- bypass turbofan engines A series of nominal and seeded faulted engine test cases Data collected over a range of engine power settings including steady-state and transient operating conditions
Application Example: Analysis of Vehicle Integrated
– – – – – – VIPR is a series of ground-based, on-wing engine tests to mature engine health management sensors and algorithms Test vehicle: VIPR ground tests include:
• • •
www.nasa.gov Description fuel flow variable stator vanes station 2.5 bleed valve Wf VSV BLD25 Actuator Commands Symbol
Architecture
Commercial Modular Aero-Propulsion System Simulation 40k (C-MAPSS40k)
Model-Based Gas Path Diagnostic
Description fan speed core speed low pressure compressor exit total pressure low pressure compressor exit total temperature high pressure compressor exit static pressure high pressure compressor exit total temperature low pressure turbine exit total pressure low pressure turbine exit total temperature
Architecture Designed Based on NASA C-MAPSS40k Engine Model
P5 T5 N1 N2 P25 T25 Ps3 T35 Symbol
Gas Path Sensor Measurements •
www.nasa.gov
Architecture
Commercial Modular Aero-Propulsion System Simulation 40k (C-MAPSS40k)
Model-Based Gas Path Diagnostic
Architecture Designed Based on NASA C-MAPSS40k Engine Model
• 6 state variables (1 rotor speeds, 5 metal temperatures) 8 engine sensors (2 rotor speeds, 3 pressure and 3 temperature) 6 engine performance deterioration tuning parameters 7 state variables (2 rotor speeds, 5 metal temperatures) 8 engine sensors (2 rotor speeds, 3 pressure and 3 temperature) 6 engine performance deterioration tuning parameters Description Description RTSTM Kalman Filter Estimated Parameters PBM Estimated Parameters www.nasa.gov
Wf
T5 P5
Stg Bleed
th
T35 Ps3 14
VSV
T25* P25*
N1 N2
Considered in this Study
Engine Sensors and Commands
Station 2.5 Bleed
*Sensors unique to VIPR II tests www.nasa.gov
Fan
Fault Types
Stage Bleed Valve
th
Variable Stator Vane
Low Pressure Turbine
High Pressure Turbine
Station 2.5 Bleed Valve
Low Pressure Compressor
High Pressure Compressor
Fault Types
1 2 3 4 5 6 7 8
Fault Index
www.nasa.gov WSSR Anomaly Detection Threshold Sensed Measurement PBM Predicted Measurement RTSTM Predicted Measurement Time (sec)
VIPR I Baseline Results
Gas Path Parameter WSSR 5: LPT 6: B25 8: B14 2: LPC 4: HPT 7: VSV 1: FAN 3: HPC 0: No Fault Diagnosed Fault ID www.nasa.gov WSSR Anomaly Detection Threshold Sensed Measurement PBM Predicted Measurement RTSTM Predicted Measurement Time (sec)
VIPR I Station 2.5 Bleed Valve Fault Results Gas Path Parameter WSSR
5: LPT 6: B25 8: B14 2: LPC 4: HPT 7: VSV 1: FAN 3: HPC 0: No Fault Diagnosed Fault ID www.nasa.gov WSSR Anomaly Detection Threshold Sensed Measurement PBM Predicted Measurement RTSTM Predicted Measurement Time (sec) Gas Path Parameter WSSR
VIPR II Station 2.5 Bleed Valve Fault Results
5: LPT 6: B25 8: B14 1: FAN 2: LPC 4: HPT 7: VSV 3: HPC 0: No Fault Diagnosed Fault ID www.nasa.gov WSSR Anomaly Detection Threshold Sensed Measurement PBM Predicted Measurement RTSTM Predicted Measurement Time (sec)
Stage Bleed Valve Fault Results
th
VIPR I 14
Gas Path Parameter WSSR 5: LPT 6: B25 8: B14 2: LPC 4: HPT 7: VSV 1: FAN 3: HPC 0: No Fault Diagnosed Fault ID www.nasa.gov WSSR Anomaly Detection Threshold Sensed Measurement PBM Predicted Measurement RTSTM Predicted Measurement Time (sec)
Stage Bleed Valve Fault Results
th
Gas Path Parameter WSSR
VIPR II 14
5: LPT 6: B25 8: B14 2: LPC 4: HPT 7: VSV 1: FAN 3: HPC 0: No Fault Diagnosed Fault ID www.nasa.gov
Conclusion
Architecture was found to provide reliable steady- state fault detection and isolation Addition of station 2.5 sensor provided fault detection at lower power settings Future work will include improved matching of model to engine dynamics The architecture’s ability to estimate deteriorated engine performance will be evaluated during the follow on VIPR III test
• • • •
www.nasa.gov
Acknowledgments
Research conducted under the Vehicle Systems Safety Technologies Project of NASA’s Aviation Safety Program www.nasa.gov
Backup Slides
) a y www.nasa.gov N1corrected Original model Re-trimmed model a y Original and re-trimmed PWLM (parameter ) a y
Enhancements
N1corrected Re-trimmed piecewise linear model to match F117 engine performance Updated model thermocouple dynamics Original model Engine steady-state data point Steady-state data polynomial curve fit
Model-Based Gas Path Diagnostic Architecture
o o
Model-based gas path diagnostic architecture designed based on NASA C-MAPSS40k model. Model updates were necessary due to notable mismatch between F117 engine and C-MAPSS40k model: a y acquired steady-state data (parameter
• •
Original model and polynomial curve fit through
j
ˆ
y ~
y
~
www.nasa.gov
y
~
R j T
R
H T
y
~
j
1
ˆ
y
~
j
R
H T
y
~
y
~
j R
T j
ˆ j
y ˆ
i m
j
y j ~ m H
Hm H
y
j
j , i j ˆ
y
~
ˆ
y m y
~ ~
H
WSSR
WSSEE
Equations
Measurement residuals: Weighted sum of squared residuals: Theoretical sensor residual: Fault influence matrix: Estimated fault magnitude: Estimated sensor residual: Weighted sum of squared estimated error: www.nasa.gov WSSR Anomaly Detection Threshold Sensed Measurement Performance Baseline Model Predicted Measurement Real Time Self Tunning Model Predicted Measurement Time (sec)
VIPR II Baseline Results
-3 x 10 Gas Path Parameter 0 1 2 0.5 1.5 WSSR 5: LPT 6: B25 8: B14 2: LPC 4: HPT 7: VSV 1: FAN 3: HPC 0: No Fault Diagnosed Fault ID