Upload
raymundo-simcox
View
216
Download
1
Tags:
Embed Size (px)
Citation preview
Controls and Dynamics Branch at Lewis FieldGlenn Research Center
Engine Performance Deterioration Mitigation Control- A retrofit approach
Dr. Sanjay GargBranch Chief
Ph: (216) 433-2685FAX: (216) 433-8990
email: [email protected]://www.lerc.nasa.gov/WWW/cdtb
Presented at: Aerospace Guidance and Control System Committee MeetingBoulder, CO, March 1, 2007
Research Performed by: Jonathan Litt – Army Research LabShane Sowers – Analex
Controls and Dynamics Branch at Lewis FieldGlenn Research Center
OverviewOverview
• Motivation
• Architecture Description
• Steady State Evaluation
• Transient Evaluation
• Piloted Simulation
• Conclusions
Controls and Dynamics Branch at Lewis FieldGlenn Research Center
Source: AIA PC 342 Committee on Continued Airworthiness Assessment Methodology Initial Report on Propulsion System and APU Related Aircraft Safety Hazards 1982 Through 1991
Propulsion Related Accidents & Incidents 1982 - 1991
Includes all Part 25 Category Transports Aircraft Data - Turboprop, Low Bypass, High Bypass Turbofans. (Does not include data from former Soviet Union and satellite countries’ products.)
0
5
10
15
20
25
30
35
40
Nu
mb
er
of
Even
ts
Level 4 - Severe Consequenses
Level 3 - Serious Consequenses
Uncontained
Propulsion System Malfunction + Inappropriate Crew Response
(PSM+ICR)
Controls and Dynamics Branch at Lewis FieldGlenn Research Center
Example PSM+ICR Turbofan AccidentsExample PSM+ICR Turbofan Accidents
Rejected Takeoff Events at or above V1 (30 Turbofan Events, 5 Hull Losses, 1 Fatal)
• 13 June 1996; Garuda Indonesian Airways DC10-30; Fukuoka, Japan (Contributing event: fracture of a HPT stage 1 blade)
• 19 October 1995; Canadian Airlines DC10-30ER; Vancouver, Canada (Contributing event: progressive HPC blade failures)
Shutdown / Throttle Wrong Engine (27 Turbofan Events, 2 Hull Losses, 1 Fatal)
• 8 January 1989; British Midland Airways 737-400; near East Midlands Airport, UK (Contributing event: fan blade failure)
Loss of Control (14 Turbofan Events, 11 Hull Losses, 7 Fatal)
• 24 November 1992; China Southern Airlines 737-300; Guangzhou, China (asymmetric thrust - stuck throttle)
• 31 March 1995; Tarom Romanian Airlines A310; near Balotesti, Romania (asymmetric thrust - stuck throttle)
Autonomous Propulsion System TechnologyAutonomous Propulsion System Technology- Reduce PSM+ICR incidents
Reduce/Eliminate human dependency in the control and operation of the propulsion system
Diagnostics/PrognosticsAlgorithms Are Being Developed
Demonstrate Technology in a relevant environment
Vehicle Management System
Self-Diagnostic Adaptive Engine Control System
• Performs autonomous propulsion system monitoring, diagnosing, and adapting functions
• Combines information from multiple disparate sources using state-of-the-art data fusion technology
• Communicates with vehicle management system and flight control to optimize overall system performance
Engine Condition/Capability
Performance Requirement
Model-Based Fault Detection
Fuzzy Belief
Network
Data Fusion
p
FADEC
u
controlsignals
Sensoroutput
PLA
Real EngineClosed Loop Control
NMPCmin J
Optimizer
State/Parameter Est. +
-
xP ˆ,
y
y
vw
Eng. Model
Eng. Model
REF
p
FADEC
u
controlsignals
Sensoroutput
PLA
Real EngineClosed Loop Control
NMPCmin J
Optimizer
State/Parameter Est. +
-
xP ˆ,
y
y
vw
Eng. Model
Eng. Model
REF
Controls and Dynamics Branch at Lewis FieldGlenn Research Center
PILOT WORKSHOP at GRC - 2002PILOT WORKSHOP at GRC - 2002
OBJECTIVE: Get direct input from pilots that will be used to help define the APST project plan
GOALS:• Under all flight regimes, identify what processes or procedures associated with propulsion system management could be candidates for autonomous operation• Identify what propulsion system information or control features will be helpful in managing the integration of propulsion with flight control for normal and abnormal operations• Identify what “sensory” information, other than the engine instruments, is used by the pilots in operation and control of the propulsion system for all flight regimes
Controls and Dynamics Branch at Lewis FieldGlenn Research Center
• The conclusions of 2002 NASA Glenn Pilot Workshop fell into three main categories– Control
• Thrust asymmetry control• Thrust response rate variation between engines• Propulsion Controlled Aircraft• Operating envelope expansion for emergency operation
– Diagnostics• Fault detection and isolation for vibration and potential
engine shutdowns• Health and usage monitoring
– Indications to pilots• Fault signals• Vehicle status under autopilot, especially concerning
throttle movement and split throttles
Results from PILOT WORKSHOPResults from PILOT WORKSHOP
• Engine Control Logic Is Developed Using A “Nominal” Engine Model…But “Nominal” Engine Does Not Exist
TimePLA
Thr
ust
Nominal Engine withFixed Control Normal
Variation
NormalVariation
Degraded Engine withFixed Control
Mea
sure
of
Per
form
ance
Typical Current Engine ControlTypical Current Engine Control
Control Logic
Limit Logic
EngineFan Speed Schedule
PLA N2c
N2
eN2 WFc WF y
FADEC – Full Authority Digital Engine Control
+
-
• Since Thrust cannot be measured, another parameter such as Fan Speed (N2), which correlates to Thrust, is regulated
Asymmetric Thrust Accident InformationAsymmetric Thrust Accident Information• Aircraft asymmetric thrust accidents have been identified as a concern in the
AIA/AECMA study on PSM+ICR [1]: “A further area of concern was power asymmetry resulting from a slow power loss, stuck
throttle, or no response to throttle coupled with automatic controls. Flying aids, such as the auto-pilot and auto-throttle, can mask significant power asymmetry until a control limit is reached. At this point, the flight crew has to intervene, understand the malfunction, and assume control of an airplane which may be in an upset condition. Better indications and/or annunciations of power asymmetry could warn crews in advance and allow them time to identify the problem and apply the appropriate procedures.”
• The following description of past asymmetric thrust accident is taken from an FAA Policy Statement on aircraft thrust management systems (TMS) [2]:
1. Sallee, G.P., and Gibbons, D.M., “AIA/AECMA Project Report on Propulsion System Malfunction Plus Inappropriate Crew Response (PSM+ICR), Volume I,” (Aerospace Industries Association and The European Association of Aerospace Industries, November 1, 1998).
2. FAA Policy Statement, “FAA Policy on Type Certification Assessment of Thrust Management Systems,” FAA Policy Statement Number ANM-01-02, March 2002. http://www.airweb.faa.gov/Regulatory_and_Guidance_Library/rgPolicy.nsf/0/0f670523ec44af9f86256ce9004c4539
March 31, 1995, Tarom Airbus Model A310-300, Bucharest, Hungary: The airplane crashed shortly after takeoff. The Romanian investigating team indicated that the probable cause of the accident was the combination of an autothrottle failure that generated asymmetric thrust and the pilot's apparent failure to react quickly enough to the developing emergency.
Report Conclusion: Data from these accident investigations have provided evidence that it is incorrect to assume that the flightcrew will always detect and address potentially adverse TMS effects strictly from inherent operational cues.
Controls and Dynamics Branch at Lewis FieldGlenn Research Center
Model-Based Controls and DiagnosticsModel-Based Controls and Diagnostics
GroundLevel
Engine Instrumentation• Pressures• Fuel flow• Temperatures• Rotor Speeds
Actuator Commands• Fuel Flow• Variable Geometry• Bleeds
Ground-Based Diagnostics• Fault Codes• Maintenance/Inspection
Advisories
On-Board Model & Tracking Filter
• Efficiencies • Flow capacities• Stability margin• Thrust
Selected Sensors
On Board
SensorValidation &
Fault Detection
Component Performance
Estimates
Sensor Estimates
Sensor Measurements
Actuator Positions
Adaptive Engine Control
• Applicable only to future systems• Still in research mode with many technical changes to overcome
Controls and Dynamics Branch at Lewis FieldGlenn Research Center
THE NEEDTHE NEED
• There is a need to develop a “simplified” approach to maintaining throttle to thrust relationship in the presence of engine degradation, and detecting thrust asymmetry situations. The approach “shall”:
• Be retrofitable to existing FADEC systems
• Leverage the extensive investment in existing FADEC control logic – specially in terms of limits imposed for operational life and safety
• Be mostly software/logic additions – not require any new sensors or actuation hardware
• Have “reasonable” development, verification and implementation costs
Control Logic
Limit Logic
EngineFan Speed Schedule
PLA
T_des
N2
eN2 WFc WF y
FADEC – Retrofit
Thrust Model
Thrust Estimator
+-
N2c Modifier
delN2c
N2cmod +
+
-
T_est
Addition to Existing FADEC Logic
Engine Performance DeteriorationEngine Performance Deterioration Mitigation Control (EPDMC)Mitigation Control (EPDMC)
• The proposed retrofit architecture:
• Adds the following “logic” elements to existing FADEC:• A model of the nominal throttle to desired thrust (T_des) response• An estimator for engine thrust (T_est) based on available measurements• A modifier to the Fan Speed Command (delN2c) based on the error between desired and estimated thrust
• Since the modifier appears prior to the limit logic, the operational safety and life remains unchanged
EPDMC Testbed ArchitectureEPDMC Testbed Architecture
• Engine– Full envelope, nonlinear
Component Level Model– Represents a large
commercial turbofan engine
Parts of EPDMC Testbed ArchitectureParts of EPDMC Testbed Architecture
• Engine Control– Typical Full Authority
Digital Engine Control (FADEC) type controller
– PLA in, fuel flow out
– Fan speed is controlled
Parts of EPDMC Testbed ArchitectureParts of EPDMC Testbed Architecture
• Nominal Engine Model– Piecewise linear model– Scheduled on percent
corrected fan speed
Parts of EPDMC Testbed ArchitectureParts of EPDMC Testbed Architecture
• Thrust Estimator– Piecewise linear Kalman filter– Based on Nominal Engine Model– Provides optimal estimation of variables in a least
squares sense subject to sensors selected
Parts of EPDMC Testbed ArchitectureParts of EPDMC Testbed Architecture
• PI Control with Integrator Windup Protection– Performs outer loop PLA
adjustment– Stops integrating error when
PLA limit is reached
Controls and Dynamics Branch at Lewis FieldGlenn Research Center
EPDMC EvaluationEPDMC Evaluation
• The purpose of the evaluation is to determine– The steady state accuracy of the thrust estimator at
many operating points and degradation levels with various types of uncertainty (model mismatch, nonlinearities, noise)
– How well the outer loop control is able bring the thrust back to the nominal level in steady state
– How well the outer loop control is able to maintain a nominal thrust response over a typical flight trajectory with a deteriorated engine
Controls and Dynamics Branch at Lewis FieldGlenn Research Center
• Evaluation was performed in two phases– Steady State– Transient
• Assumptions– 10 health parameters, two each (efficiency and flow
capacity) for each of the five major components– Worst case degradation 5% in each health parameter– Health parameters degrade at their own pace, pretty
much independent of each other no restrictions placed on simulated deterioration except upper limit of 5%
EPDMC EvaluationEPDMC Evaluation
Outer Loop Control off
Steady State EvaluationSteady State Evaluation• Thrust performance deterioration with engine degradation
• Thrust estimation error is << Thrust deterioration => Thrust estimate can be used effectively for performance recovery
Outer Loop Control on
Steady State EvaluationSteady State Evaluation
Outer Loop Control off
• EPDMC maintains “close” to nominal thrust performance- even with high levels of engine degradation
Controls and Dynamics Branch at Lewis FieldGlenn Research Center
Transient EvaluationTransient Evaluation
• Trajectory is takeoff/climb/cruise
• It passes through or near the linearization points
• No airframe is included, the engine is operating as if it were in a wind tunnel
Controls and Dynamics Branch at Lewis FieldGlenn Research Center
Transient EvaluationTransient Evaluation
• Nominal Engine with and without Outer Loop Control
Controls and Dynamics Branch at Lewis FieldGlenn Research Center
Transient EvaluationTransient Evaluation• Degraded Engine with and without Outer Loop Control
Controls and Dynamics Branch at Lewis FieldGlenn Research Center
Flight SimulatorFlight Simulator
THROTTLESTICK
PEDALS
INSTRUMENTATIONDISPLAY
HEADS UPDISPLAY
SCREEN
““Piloted” Evaluation of ArchitecturePiloted” Evaluation of Architecture
Segment 1 2 3 4 5
Fan Speed 86% 90% 88% 82% 86%
Indicated Airspeed
290 knots 290 knots 290 knots 290 knots 290 knots
Heading 270º 270º 270º 270º 270º
Altitude 32,000 feet
Climb 33,000 feet
descend 32,000 feet
Duration 3 minutes - 3 minutes - 3 minutes
• Pilot-in-the-loop in a fixed-base simulator• Maintain airspeed and heading while following profile - Three cases: Nominal, 1 engine degraded – OLC Off/On
Controls and Dynamics Branch at Lewis FieldGlenn Research Center
Pilot Workload During Transient FlightPilot Workload During Transient Flight
Very ClearIncrease inWorkload WithOuter LoopControl Off
Controls and Dynamics Branch at Lewis FieldGlenn Research Center
ConclusionsConclusions• Developed a controls architecture that would maintain throttle
to thrust relationship as the engine degrades– Addresses one of the major issues of propulsion related workload
identified during a pilot workshop– Requires “minor” additions to existing FADEC logic– Preliminary simplified simulation results encouraging
• Current research focusing on implementing the architecture on the fan speed correction over the whole engine operating envelope and performing more detailed evaluations
• Need to address some of the potential challenges for implementation:– Pilots are used to relating throttle setting to fan speed– Acoustics issues related to two engines running at different but very
close fan speeds (Beat frequency)