NASA / CR--2002-211485
Flight Test of Propulsion Monitoring
and Diagnostic System
Steve Gabel and Mike Elgersma
Honeywell Laboratories, Minneapolis, Minnesota
_Jt_,
March 2002
https://ntrs.nasa.gov/search.jsp?R=20020061790 2020-06-24T06:34:13+00:00Z
The NASA STI Program Office... in Profile
Since its founding, NASA has been dedicated to
the advancement of aeronautics and spacescience. The NASA Scientific and Technical
Information (STI) Program Office plays a key part
in helping NASA maintain this important role.
The NASA STI Program Office is operated by
Langley Research Center, the Lead Center forNASA's scientific and technical information. The
NASA STI Program Office provides access to the
NASA STI Database, the largest collection of
aeronautical and space science STI in the world.
The Program Office is also NASA's institutional
mechanism for disseminating the results of its
research and development activities. These results
are published by NASA in the NASA STI Report
Series, which includes the following report types:
TECHNICAL PUBLICATION. Reports of
completed research or a major significant
phase of research that present the results of
NASA programs and include extensive data
or theoretical analysis. Includes compilations
of significant scientific and technical data and
information deemed to be of continuing
reference value. NASA's counterpart of peer-
reviewed formal professional papers but
has less stringent limitations on manuscript
length and extent of graphic presentations.
TECHNICAL MEMORANDUM. Scientific
and technical findings that are preliminary or
of specialized interest, e.g., quick release
reports, working papers, and bibliographiesthat contain minimal annotation. Does not
contain extensive analysis.
CONTRACTOR REPORT. Scientific and
technical findings by NASA-sponsored
contractors and grantees.
CONFERENCE PUBLICATION. Collected
papers from scientific and technical
conferences, symposia, seminars, or other
meetings sponsored or cosponsored byNASA.
SPECIAL PUBLICATION. Scientific,
technical, or historical information from
NASA programs, projects, and missions,
often concerned with subjects having
substantial public interest.
TECHNICAL TRANSLATION. English-
language translations of foreign scientific
and technical material pertinent to NASA'smission.
Specialized services that complement the STI
Program Office's diverse offerings include
creating custom thesauri, building customized
data bases, organizing and publishing research
results.., even providing videos.
For more information about the NASA STI
Program Office, see the following:
• Access the NASA STI Program Home Page
at http://www.sti.nasa.gov
• E-mail your question via the Intemet to
• Fax your question to the NASA Access
Help Desk at 301-621-0134
• Telephone the NASA Access Help Desk at301-621-0390
Write to:
NASA Access Help Desk
NASA Center for AeroSpace Information7121 Standard Drive
Hanover, MD 21076
NASA/CR--2002-211485
Flight Test of Propulsion Monitoring
and Diagnostic System
Steve Gabel and Mike Elgersma
Honeywell Laboratories, Minneapolis, Minnesota
Prepared under Contract NAS3--01105
National Aeronautics and
Space Administration
Glenn Research Center
March 2002
Trade names or manufacturers' names are used in this report foridentification only. This usage does not constitute an official
endorsement, either expressed or implied, by the National
Aeronautics and Space Administration.
NASA Center for Aerospace Information7121 Standard Drive
Hanover, MD 21076
Available from
National Technical Information Service
5285 Port Royal RoadSpringfield, VA 22100
Available electronically at http://gltrs.e-r¢.nasa.gov/GLTRS
Table of Contents
Executive Summary ........................................................................................................................ 1
Section
1.1
1.2
1.3
1.4
1.5
1.6
1. Introduction .................................................................................................................... 2
Background ................................................................................................................... 2
AGATE Program Results ............................................................................................. 2
Program Overview ........................................................................................................ 3PMDS Overview ........................................................................................................... 3
Relevance to Aviation Safety ........................................................................................ 4
PMDS Background Technical Data .............................................................................. 5
Section 2. Flight Testing ................................................................................................................. 6
2.1 Flight Test Aircraft ....................................................................................................... 62.2 PMDS Installation ......................................................................................................... 7
Section 3. Data Analysis ................................................................................................................. 9
3.1 Background ................................................................................................................... 9
3.2 Flight Test Overview .................................................................................................... 93.3 Data Collection ........................................................................................................... 10
3.4 Static Correlation Modeling ........................................................................................ 11
3.5 Linear Dynamic Modeling .......................................................................................... 17
3.6 Nonlinear Dynamic Modeling .................................................................................... 30
3.7 Data Analysis Conclusions ......................................................................................... 40
Section 4. Technology Roadmap .................................................................................................. 42
4.1 Modeling and Fault Diagnosis Algorithms Development .......................................... 42
4.2 Hardware and Software Development ........................................................................ 48
4.3 System Development .................................................................................................. 53
Section 5. New Technology .......................................................................................................... 56
Section 6. Conclusions and Recommendations ............................................................................ 57
Appendix
NASA/CR--2002-211485 iii
List of Figures
Figure 1-1.
Figure 2-1.
Figure 2-2.
Figure 2-3.
Figure 2-4.
Figure 3-1.
Figure 3-2.
Figure 3-3.
Figure 3-4.
Figure 3-5.
Figure 3-6.
Figure 3-7.
Figure 3-8.
Figure 3-9.
Figure 3-10.
Figure 3-11.
Figure 3-12.
Figure 3-13.
Figure 3-14.
Figure 4-1.
Figure 4-2.
Figure 4-3.
Figure 4-4.
Figure 4-5.
Figure 4-6.
Figure 4-7.
Figure 4-8.
PMDS .......................................................................................................................... 4
Flight Test Aircraft ...................................................................................................... 6
PMDS Flight Test Arrangement ................................................................................. 7
PMDS Hardware ......................................................................................................... 7
PMDS Installation ....................................................................................................... 8
PMDS Hardware Arrangement ................................................................................. 10
EGT Fault Model Results, Flight 44 ......................................................................... 13
CHT2 Measured Data and Modeling Data ................................................................ 14
CHT Fault Model Results, Flight 41 ......................................................................... 15
CHT Fault Model Results, Flight 44 ......................................................................... 16
Engine Bay Temperature Sensor Fault Model Results, Flight 44 ............................. 17Linear Dynamic Model Data ..................................................................................... 20
Linear Dynamic Model Results, Flight 41 ................................................................ 22
Linear Dynamic Model Results, Flight 44 ................................................................ 26
Nonlinear Dynamic Model Data ............................................................................. 31
Nonlinear Dynamic Model Results, Flight 41 ........................................................ 33
Nonlinear Dynamic Model Results, Flight 44 ........................................................ 35
Nonlinear vs. Linear Model Results, Flight 41 ....................................................... 37
Nonlinear vs. Linear Model Results, Flight 44 ....................................................... 39
Modeling and Fault Detection ................................................................................... 42
Oil Pressure Failure Detection and Warning Diagnosis ........................................... 44
Modeling and Fault Diagnosis Algorithms Development Roadmap ........................ 47
PMDS Architecture ................................................................................................... 48
PMDS Hardware Architecture .................................................................................. 49
PMDS Hardware ....................................................................................................... 50
Hardware and Software Development Roadmap ...................................................... 52
System Development Roadmap ................................................................................ 55
List of Tables
Table 1-1
Table 3-1.
Table 3-2.
Table 3-3.
Table 3-4•
Table 3-5.
Table 3-6.
Table 4-1.
• PMDS Glossary ............................................................................................................ 5
Flight Tests ................................................................................................................... 9
Linear Model Data ...................................................................................................... 21
Linear Model Plots ..................................................................................................... 21
Nonlinear Model Data ................................................................................................ 32
Nonlinear Model Plots ................................................................................................ 32
Comparison Plots ........................................................................................................ 36
Oil Pressure Fault Modes and Corresponding Detection Tests .................................. 45
NASA/CR--2002-211485 iv
Executive Summary
This report presents the results of the NASA program entitled "Flight Test of Propulsion
Monitoring and Diagnostic System," performed by Honeywell and Aurora Flight Sciences. The
objective of this program was to build on the results of the propulsion monitoring and diagnostic
system (PMDS) technology developed by Honeywell under the NASA Advanced General
Aviation Transport Experiment (AGATE) program and apply them to the broader goals of the
NASA Aviation Safety program. The technical work included two flight tests using PMDS
equipment and analysis of flight test data. The target application for the PMDS is piston-engine-
driven general aviation aircraft.
The PMDS concept is intended to independently monitor the performance of the engine,
providing continuous status to the pilot along with warnings if necessary. Specific sections of
this data would be available to ground maintenance personnel via a special interface. The inputs
to the PMDS include the digital engine controller sensors and other sensors. At its present stage
of development, the PMDS monitors and records engine parameters and stores them into an
engine history database for subsequent processing by off-line diagnostic algorithms. At present,
the system does not compare the parameter values with engine norms to perform on-line
diagnostics and prognostics (this extended functionality would be added in future development).
Technological advances in sensing, processing, and software have resulted in more affordable
and more capable health monitoring technology. The application of health monitoring
technology to aircraft engines has tremendous potential given the complexity, harsh
environmental conditions, and natural degradation that this machinery exhibits. Benefits include
increased safety and reliability and reduced operating costs.
The technical work performed on this program provided the following key results:
• It demonstrated the ability of the PMDS to detect a class of selected sensor hardware
failures.
• It demonstrated the ability of the PMDS hardware to successfully model the engine for the
purpose of engine diagnosis. Not surprisingly, nonlinear dynamic models performed better
than linear dynamic models for the same number of inputs and states.
Future development work for an engine monitoring and diagnostic system should employ the
following elements:
• Engine/aircraft modeling should combine first-principles and empirical approaches.
Empirical methods can be used to calibrate unknown parameter values as needed.
• The monitoring and diagnostic system should employ additional inputs outside the engine,
such as aircraft speed, aileron, elevator, rudder, and flap settings, propeller pitch, etc.
• A prioritized list of engine faults is needed to guide the diagnostic development work.
The monitoring and diagnostic system should be able to gather input data from the full authority
digital engine controller (FADEC) and other systems in the aircraft over a digital avionics bus.
This data sharing capability will enable the use of more sophisticated models and will help tominimize the installed cost of the PMDS.
NASA/CR--2002-211485 1
Section 1. Introduction
This report presents the results of the NASA program entitled "Flight Test of Propulsion
Monitoring and Diagnostic System," performed by Honeywell and Aurora Flight Sciences. The
following sections present the detailed technical results along with a set of conclusions and
recommendations based on experience gained in performing the work.
1.1 Background
A propulsion monitoring and diagnostic system (PMDS) can provide valuable benefits to general
aviation users. The PMDS provides increased confidence in the propulsion system while in flight
for improved safety and provides valuable diagnostic data to ground maintenance technicians for
reduced maintenance costs. The target application for the PMDS is piston-engine-driven general
aviation aircraft. In this program, the PMDS technology developed by Honeywell under the
NASA Advanced General Aviation Transport Experiment (AGATE) program was extended
through additional flight testing and data analysis to demonstrate its current capabilities.
The purpose of the PMDS concept is to provide the pilot with engine health indications and to
inform the pilot when the engine requires preventative maintenance. By providing this
information before in-flight failure of the engine, it greatly enhances flight safety and provides
simplified engine diagnostics for the pilot. The technology developed under AGATE funding is
the core of a future fully functional PMDS. The hardware and software developed under AGATE
funding monitors and records engine parameters and stores them in an engine history database
for subsequent processing by off-line diagnostic algorithms. At present, the unit does not
compare the parameter values with engine norms to perform on-line diagnostics and prognostics
(this extended functionality would be added in future development).
1.2 AGATE Program Results
Honeywell was a participant in the AGATE Propulsion Sensors and Controls Work Package.
During 1999, Honeywell built the PMDS hardware, implemented the firmware that records and
preprocesses the flight data, and with the assistance of Aurora Flight Sciences Corporation,
performed an initial flight test. A simple engine model, developed to postprocess the flight test
data, verified that we could detect a failed (disconnected) exhaust gas temperature (EGT) sensor
signal. A shortfall in funding prevented further test flights under the AGATE program. However,
these results showed the significance of the core PMDS in that it provides both hardware and
software design guidelines for successfully interfacing the PMDS to general aviation aircraft.
Additional test flights and data analysis were performed on this 2001 NASA program to verifythat the design and integration of the core PMDS into the aircraft is a suitable base on which to
build the diagnostic capability.
NASA/CR--2002-211485 2
1.3 Program Overview
The objective of this program was to build on the results of the AGATE work and apply them to
the broader goals of the NASA Aviation Safety program. The technical work included two flight
tests using PMDS equipment and analysis of flight test data.
Our technical approach for the flight testing and data analysis consisted of four steps:
1. Collect data from a set of pertinent engine sensors during a baseline test flight (no
failures).
2. Compute an engine model based on a least-squares fit to flight test data.
3. During the second test flight, introduce sensor faults to test the diagnostic algorithm.
4. Postprocess the flight data to diagnose the faulty sensor/variable.
This program also included the development of a roadmap detailing the recommended next steps
in applying the PMDS technology to general aviation.
1.4 PMDS Overview
The PMDS is a separate subsystem designed to independently monitor the performance of the
engine, providing continuous status to the pilot along with warnings if necessary. Specific
sections of this data are available to ground maintenance personnel via a special interface. The
PMDS also provides a set of data for maintenance event prediction to be used by ground
personnel or for possible impending failure information to be displayed to the pilot. The PMDS
will continuously monitor its own performance to ensure its own integrity and capability.
The PMDS will be able to detect and diagnose the most common engine failures. The set of
failures to be detected will be defined in future development phases. A top level of the system is
shown in Figure 1-1.
The PMDS continuously monitors the performance of the engine in order to detect failures and
predict impending failures (prognosis). Failures and warnings of impending failures are indicated
to the pilot, and the collected engine diagnostic information pertaining to the failures and/or
warnings is also available to a ground maintenance technician. The AGATE PMDS development
effort defined the following performance goals for the system:
• Early detection time: The PMDS is intended to detect and indicate a warning of impending
failure at least 8 flight hours prior to failure.
• High probability of detection and coverage: The PMDS is intended to detect 90% of
impending failures in the engine.
The inputs to the PMDS include the full authority digital engine controller (FADEC) sensors and
other engine sensors. The outputs consist of the pilot warning display and the maintenancedevice interface. The maintenance device interface could be a hand-held interrogation and
display device that would allow maintenance personnel to determine the status of the engine (as
NASA/CR--2002-211485 3
ff
I"/
/
//
/I
I
I
l
tIIl1
PMDS
_\ di splay
\\
,,- ...... ----"FADEC l J'\, ..- --i"/'''"
x ground-based...... _ maintenance data
PMDS i-t ""---... F22S--2
I "_t I
_11 I I II I J
//
engine sensors / \and other inputs //
// maintenanceJ
"-... ./" technician's........ interface
PMDS System
Figure 1-1. PMDS
well as the PMDS itself) and define any necessary maintenance actions. Readout capability is
included in the on-board maintenance panel; however, some installations could forgo this panel,
depending completely on ground-based equipment that would receive data from the interface
port.
Sensor inputs will be received via digital interfaces from the FADEC, single-lever power control
(SLPC), and other digital systems on the aircraft, or via other sensors directly connected to the
PMDS. Examples of these sensor inputs are as follows:
• FADEC: exhaust gas temperature (EGT), cylinder head temperature (CHT), engine speed(RPM), oil pressure, etc.
• SLPC: engine power lever position, throttle command, propeller pitch command, etc.
• Other inputs via an avionics bus: air speed, pilot inputs to control surfaces, etc.
• Other sensors directly connected to the PMDS: vibration sensors, oil particle sensors, etc.
A glossary of terms relating to the PMDS is shown in Table 1-1.
1.5 Relevance to Aviation Safety
Technological advances in sensing, processing, and software have resulted in more affordable
and more capable health monitoring technology. The application of health monitoring
technology to aircraft engines has tremendous potential given the complexity, harsh
NASA/CR--2002-211485
Table 1-1. PMDS Glossary
Early Detection: System predicts that failure will occur in near future, lightswarning light.
Failure Detection: System detects that failure has occurred, lights failure light,(future implementations may take backup action if indicated).
False Alarm System indicates that failure has occurred, although in fact,indicated failure has not occurred.
Impending Failure Engine condition has changed so as to cause a failure in nearfuture.
Diagnostics The process by which a particular fault mode is indicated.
Detection Time The time between the occurrence of either an impending
failure (in the case of early detection) or a failure (in the caseof failure detection) and the corresponding failure or warningindication.
Accuracy RSS (Root Sum Square) value including scale factortolerance, linearity, offset and temperature effects.
Failures A failure is a fault which will require ground maintenanceaction to correct.
FADEC Full Authority Digital Engine Controller
SLPC Single Lever Power Control
EGT Exhaust Gas Temperature
CHT Cylinder Head Temperature
environmental conditions, and natural degradation that this machinery exhibits. Benefits include
increased safety and reliability and reduced operating costs.
The benefits of engine monitoring and diagnostics will be valuable to all segments of general
aviation, from the individual aircraft owner to large fleet operators. All of these potential users
share a common interest in having a capability to increase the availability of their aircraft
engines. At present, we don't know of any commercially available engine monitoring systems for
piston engine aircraft. However, these types of devices are currently offered as optional
equipment in at least one single-engine turboprop aircraft (the Hiatus PC-12). Hiatus offers an
"engine trend monitoring" option with the PC-12 as described at !l_t_t.p://www.pilatus-
aircraft.corn/2 ga commercial/framese! pcl2.htm.
1.6 PMDS Background Technical Data
Under the AGATE program, Honeywell prepared a set of technical documents for the PMDS
concept. These documents define the system requirements, modeling methods, and fault
detection and failure diagnosis methods. This body of information serves as a baseline for future
development of the system.
The PMDS concept is intended to meet standard commercial avionics integration requirements
that apply to the intended application configurations. The PMDS concept is also intended to
conform to the same environmental, electrical, and mechanical standards that apply to
comparable commercial avionics equipment.
NASA/CR--2002-211485 5
Section 2. Flight Testing
The flight testing was performed by Aurora Hight Sciences as described in the following
subsections. Additional information is presented in the Aurora flight test report in the Appendix.
2.1 Flight Test Aircraft
The flight tests were performed using Aurora's twin engine Cessna 0-2 Chiron aircraft, as
shown in Figure 2-1. This aircraft was equipped with a SLPC and FADEC controlling the front
engine (the rear engine was not involved in the flight testing). With the SLPC and FADEC,
electric servo actuators control the throttle and the prop governor; the pilot commands a single
thrust command. Electronic port fuel injection, electronic ignition system with several redundant
feedback loops controls mixture and thermal control (CHT, EGT, exhaust gas oxygen).
Figure 2-1. Flight Test Aircraft
The Chiron front engine (monitored during this test) is a Teledyne IO-360ES. The flight tests
were performed according to flight test plans developed jointly by Honeywell, Aurora, and
NASA. Flight test data was logged on the PMDS. After each flight, the PMDS data was
downloaded via serial interface to a PC and sent to Honeywell via the Internet for data reduction
and analysis.
To test the PMDS diagnostic concept, we simulated a sensor failure by temporarily
disconnecting some noncritical sensors on one of the flights. The flight test approach is shown in
Figure 2-2.
NASA/CR--2002-211485
non-critical sensor
pilot good ______
inputs "_ H failedFADEC Engine
criticalsensors
PMDS
sensordata for
modeling
Figure 2-2. PMDS Flight Test Arrangement
In addition, we noticed some unplanned intermittent sensor faults as described in Section 3 of
this report. This flight testing procedure had no effect on the engine's performance, since the
FADEC recognizes and accounts for faulty EGT signals.
2.2 PMDS Installation
The Honeywell PMDS was mounted in the cabin of the aircraft. Figures 2-3 and 2-4 show the
PMDS hardware and its installation in the aircraft.
Figure 2-3. PMDS Hardware
NASA/CR--2002-211485 7
Figure 2-4. PMDS Installation
NASA/CR--2002-211485
Section 3. Data Analysis
The flight test data files were received from Aurora Flight Sciences and prepared for analysis.
Several technical analyses were performed on the data using Matlab scripts. This technical
analysis work is described in the following subsections.
3.1 Background
The flight testing was performed by Aurora Flight Sciences in a manner similar to that used for
the earlier AGATE work. Aurora's twin engine Cessna 0-2 Chiron aircraft was used for these
tests. The Honeywell PMDS was employed with Aurora's FADEC and electronic SLPC, which
controlled the front engine in the aircraft. The PMDS collected various engine data and other
parameters made available directly from the FADEC via a digital communications link.
3.2 Flight Test Overview
Under this program, two flight tests of the PMDS were conducted. Data was collected at a
variety of altitudes and power settings over the operating envelope of the engine. A top-level
description of the two flight tests is presented in Table 3-1 below.
Table 3-1. Flight Tests
Profile: Flight 41
12ooo F i i / i '_ i i
, , j_ , \ _ ,10000F ..... i- - -
--.--,-.---.-,.,--.--
, ! i _ i i8ooo_.......... '_'+'--_ ' _ '
. , ,- ,i I I _ i
60001- - - - .... : -
4000 I- .... _'_ ....... _ .......... _ _--
2000t- - - -_'-i ............. _--- r-- 5_-.-
/ / ..../ ! I i _ P
0 1000 2000 3000 4000 fi000 6000 7000 8000 9000
lime (see)
Date: April 25, 2001
Setup: Baseline flight
Simulated Faults: none
Intermittent Faults: CHT2 and EGT4
1O0O0
8000
4000
20O0
._..,_ CT "_'_ L
t000 2000
.... / , _
+___________,______ _k.___I _ I I • -I
.... -, - _ ? __.__
.____,_ ,_,_ ..... ____ __
--f.---_ ............... -_q--
J
J
3000 4o00 500o 6000 7000
time (see)
Profile: Flight 44
_2ooo" -- 1---2-- L _ _ L _ _ i _._ _ L _ J _ _
]
i
-4--
i
I
I
J
i
i
Date: April 27, 2001
Setup: Add simulated faults
Simulated Faults:
• Engine bay temperature 2
• Mass airflow sensor
Intermittent Faults: CHT2, EGT1, and EGT4
'f r
8000 90O0
NASA/CR--2002-211485 9
The simulated fault in the mass airflow sensor was not used in the analysis because the sensor
data was not set up for communication to the PMDS. This was an oversight in the preparation of
the test plan, but it did not affect our ability to accomplish the objectives of the flight test and
analysis work. The flight test plans and flight test engineer reports are presented in the Appendix.
3.3 Data Collection
For this project, the PMDS collected engine performance data from the engine FADEC via a
digital communications link. The hardware arrangement is shown in Figure 3-1. (No directly
connected engine sensors were used in this project.)
FADECController
Sensors
SerialInterface
::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
I
LongT_m
Memo_
Central
DiagnosticProcessor
and Memory
Input/OutputController
and
SignalConditiomng
_/////_
y///_
///////_
OutputPort
DiagnosticOutput
Figure 3-1.
PMDS
PMDS Hardware Arrangement
The current PMDS hardware/firmware records the values of 25 variables: time, pressure altitude,
air pressure, fuel pressure, oil pressure, fuel temperature, oil temperature, bay 1 temperature, bay
2 temperature, six EGTs, six CHTs, outside air temperature, manifold air pressure, air charge
temperature, and engine RPM. The 25 variables are measured at 5 Hz. To average out noise and
to fit all the data from many hours of flight into limited memory, the data is reduced in the
following way:
NASA/CR--2002-211485 10
1. Over each 51-second time window, a least-squares-error linear fit is made to each
variable's time history in that window.
2. The slope and intercept of this linear fit is saved, together with the maximum andminimum values of each variable in this window.
A list of the sensor data collected from the FADEC is presented in the Appendix.
3.4 Static Correlation Modeling
A static correlation model can be used to estimate the value of one or more sensors using
measured data from other sensors. While this is not a dynamic model per se, it does employ
measured data across a wide range of operating conditions and is capable of producing a goodestimate of the sensor values of interest. These estimated values can then be compared to
measured sensor data for monitoring and diagnostic purposes. In practice, it would be beneficial
to create several static correlation models using different combinations of inputs in case any one
of the inputs is faulty (thereby causing all of the outputs to be invalid). This approach would
enable the system to check each model to determine which input is faulty.
3.4.1 Static Correlation Modeling Approach
The form of the static correlation model is
y =f(x)
where
y _ R p = output (estimates of desired sensors)
x _ R n = input (measured data for other sensors, including their derivatives)
The model (matrix A) is computed using a least-squares approximation to measured data. For
estimating the five EGT states, matrix A is computed from
AX= Y
where
[ xl0)''''x'(k) R,,×kX= i i
[x,,O).",x,(k)
EGT. 0)"" EaT, (k ) ]
y = : : ] _ R p_s (In this example, p = 5 because one of
G-TpO) Tp--(k)] the EGTs was bad in both test flights.)E ... EG
k = number of samples (period of time) over which the estimate is desired
The correlation matrix A is computed from measured data as follows:
A = yxr(xxr) -l
NASA/CR--2002-211485 11
3.4.2 Static Correlation Modeling of EGT Sensors
A static correlation model for the EGT sensors was computed using the data from Flight 41 as
described above. This model was then applied to measured data from Flight 44 (using sensors
other than the EGTs as input data). The resulting estimated EGT data compared very well against
the actual measured EGT data for Flight 44, as shown in Figure 3-2. As mentioned earlier, EGT1
in Flight 44 showed some intermittent bad data (i.e., an intermittent fault). This bad data is
clearly visible when compared to the estimated values shown in Figure 3-2.
3.4.3 Static Correlation Modeling of CHT Sensors
As mentioned earlier, sensor CHT2 showed some intermittent bad data in both Flight 41 and
Flight 44. The periods of bad data are shown in the top and middle subplots of Figure 3-3. For
modeling purposes, the good data from each flight was combined to create the model. This
model input data is shown in the lower subplot of Figure 3-3. A static correlation model for the
CHT sensors was computed using this data. This model was then applied to measured data from
Flight 41 (using sensors other than the CHTs as input data). The resulting estimated CHT data
compared very well against the actual measured valid CHT data for Flight 41, as shown in
Figure 3-4. The intermittent fault in CHT2 is clearly visible when compared to the estimated
values shown in Figure 3-4. The same analysis was done for Flight 44, which produced similar
results as shown in Figure 3-5.
3.4.4 Static Correlation Modeling of Engine Bay Temperature Sensors
Both engine bay temperature sensors provided valid data in Flight 41. For Flight 44, baytemperature sensor 2 was disconnected to simulate a fault condition. A static correlation model
for the bay temperature sensors was computed in the manner described earlier (using measured
data from Flight 41 ). This model was then applied to measured data from Flight 44 (using
sensors other than bay temperature sensors as input data). The resulting estimated bay
temperature sensor 1 data compared very well against the actual measured data for Flight 44, as
shown in the top subplot of Figure 3-6. The simulated fault condition in bay temperature sensor 2
(consistently low) is clearly visible when compared to the estimated values shown in the lower
subplot of Figure 3-6.
NASA/CR--2002-211485 12
o
G)
#_
EG T I measured (solid), EG T1 estimated (dotted)1200 , i i
) I I ,.|
1000 .... ,J_ -_#_o._ _+E;, - = - =.-'- ...." _i' "'"*. "_,_ -_;._.;_ ' _ x-_ L,",,_: : _" , .__-_.
*ll I i_tj -,_ .... Z _ _J ......800 , ,_,
600 - - i -
i : r L
400 L - i - ,i
2001- - _ ..................b
o[ 'i0 2000 4000 6000 8000
o
EGT2 measured (solid), EGT2 estimated (dotted)
1200 : ] :I ._ I i i st
/ iil , , ,8oo_--- #_! '- ..... " ...... ' .....
] i600_--_ ......... _ ..........
iFI ! _ ...........
400 7 - - _- ,
i I '
0 2_0 40_ 6_0 8_0
ov
EGT3 measured (solid), EGT3 estimated (dotted)1200 i i
i i i
1000 - - _1 ._%.,.*_,-;_- -,_%-._,- - - _;_;_t
800 - - _ ................
600 - - - _ ...................Ii
400 - - _ .................I
I
200 - - ± .....................
00 2000 40_ 6_0 8_0
ov
o
EGT4measured (solid), EGT4estimated(dotted)1200
1000
80O
6OO
4OO
2OO
0
EGT4 is not modeled
due to bad data in both flights
0 2000 4000 6000 8000
G"ov
.o
0 2000 4000 6000 8000
time (sec) PIV[_ Ftight44
ov
o
1000
8OO
6O0
4O0
200
0
EGT6 measured (solid), EGT6 estimated (dotted)1200 , ,
ii
i r
....... _L_ - - - _ ...... ' ..... _"
t_ _ r r
!
--t ......................
1!
_ _ _,." ...................
0 2000 4000 6000 8000
time (sec) F_DS Right44
Figure 3-2. EGT Fault Model Results, Flight 44
NASAJCR--2002-211485 13
700i
i
i
600 .............. '-i
i
o
_ 500
400
CHT2-Flight44 measured (solid)700 i i i
i I i i
i i i i
! ,
600 : ' ' 'i i i
i i i
i
i i
i t i i i, ,_fi ,
i
i .
300 I _ i
0 1000 2000 3000 4000 5000 6000 7000 8000 g000
CHT2 data used for estimation (dotted)700 , r _ ,
600
o t _ i
i i i
500 ....... _ ...... ,...... _....... _ ...... ,....... ,....... _....... ,......
400 _______e ....... , , _ r, ,
i_ i i p i i300 I _ I _0 1000 2000 3000 4000 5000 6000 7000 8000 9000
time (sec)
Figure 3-3. CHT2 Measured Data and Modeling Data
NASA/CR--2002-211485 14
CHT1 measured (solid), CHT1 estimated (dotted)700
600 ........................_-.
o500 ...................
_J _*_ ! k _ .j, \ _ ,_.,_J',
400 [ .... ±v__ _ , .....- ....... i
! it
300 ' '0 2000 4000 6000 8000 10000
CHT3 measured (solid), CHT3 estimated (dotted)700
600G"o
500
4O0
CHT2 measured (solid), CHT2 estimated (dotted)
3O00 2000 4000 6000 8000 10000
CHT4 measured (solid), CHT4 estimated (dotted)700
600 .........................
o5oo ............... i..........
-- _ _-_,"- _ _ L ....400
3000
i
q
I d
2000 4000 6000 8000 10O00
600 ......................
2...o
500 ................. : ....
I_ 400 _- - - _ ............ -_'_± _ _- . .....I
3000
4 I
I
I t
2000 4000 6000 8000 10O00
CHT5 measured (solid), CHT5 estimated (dotted)700
600 .........................
ov
500 .... _ .......... •..........
_" 400 ....................- -_ - _ , _t_
:300 ' ' ' '0 2000 4000 6000 8000 10000
time (sec) PIVlDSRight41
600
o
_ 4oo
CHT6 measured (solid), CHT6 estimated (dotted)700
......................... I
........................ i
L I I I
300 i i 1 1
0 2000 4000 6000 8000 10000
t_n_ (s_) _ Fl_:jht41
Figure 3-4. CHT Fault Model Results, Flight 41
NASA/CR--2002-211485 15
ov
7OO
CHT1 measured (solid), _HT1 estimated (dotted)
600 ........................
500 - - -' ................
400 _
i
300" I i _,0 2000 4000 6000 8000
CHT3 measured (solid), CHT3 estimated (dotted)
700 --
I600,_ ...................
2
500
t-: 400 ---J .................
3000
¢ ,! b
2000 4000 6000 8000
CHT5 measured (solid),700
3000
6O0
o
50o
_- 400
"HT5 estimated (dotted)
2000 4000 6000
time (sec) PMDS Right44
8000
600
o
= 500
I_" 400
CHT2 measured (solid), CHT2 estimated (dotted)700 I
i,i
,iIIi
_i i -?
i ' I
300 _ ' _ -0 2000 4000 6000 8000
CHT4 measured (solid), CHT4 estimated (dotted)
700 _ I
600 .... ; ..... _ ............
, ,
._ 500 ........... " .............
400 - - - _ ........ -' ....... --"_-_'_-- -
/
300 10 2000 4000 6000 8000
6O0
o500
I_" 400
CHT6 measured (solid), CHT6 estimated (dotted)700
i
......................
........................
- - - _ ..... _'_-_-"- -%_.,_x_- -
300 _ _ _0 2000 4000 6000 8O00
time (sec) PIVIDSFlight44
Figure 3-5. CHT Fault Model Results, Flight 44
NASA/CR--2002-211485 16
BayTemp l measured (solid), BayTemp l estimated (dotted)
400 I _,
350 .................................................i i
i i
300 - -i
V..... !...... !.................................._°°- .... i ..... i...................... i................15o_...... ......................................
100 ..... , ...... _
50..... :.................... .... _.................0 1 L
0 1000 2000 3000 4000 5000 6000 7000 8000
time (sec) PIVDS Flight44
BayTemp2 measured (solid), BayTemp2 estimated (dotted)
400 E i t i i
]
350 ...............................................................i E i i i
E p i i
^^^ L_ , , _ , _L ......LSUU - _ - _'_;. ..... _ ....... L....... _ ............. , ---"
250 ............... ,............ _ ................. '- .......o
200 ................................................... ? .......d
_'150
100
50
0
................ r...............
d
.............. 1"- .......
q
i
T ir .......
i iI I
0 1000 2000 3000 4000 5000 6000 7000 8000
time (sec) PIVDS Flight44
Figure 3-6. Engine Bay Temperature Sensor Fault Model Results, Flight 44
3.5 Linear Dynamic Modeling
A linear dynamic model can be used to estimate the value of one or more sensors using a past
history of measured data from other sensors. This type of model is capable of producing a goodestimate of the sensor values of interest. These estimated values can then be compared to
measured sensor data for monitoring and diagnostic purposes.
NASA/CR--2002-211485 17
3.5.1 Linear Dynamic Modeling Approach
The linear dynamic model is constructed in conventional state-space form, with a state equation
and an output equation. The form of the dynamic model is as follows:
_=Ax+Bu
y = Cx + Du
where
x_ R" is the state vector
u _ R'" is the input vector
y _ R _' is the output vector
A _ R "×", B _ R ''_' are the system matrices
Given a set of measured inputs over a specified time period, this model can be used to estimate
the values of all modeled system states and outputs. In our analysis work, we generally did not
compute the C and D matrices. (In a commercialized system, some states could be ignored and
various outputs of interest could be defined and computed for monitoring and diagnostic
purposes.)
Three dynamic models were developed using measured data from Flights 41 and 44. To
compare the models, each used the same set of inputs and state variables. Any sensors that hadsimulated or intermittent faults were not used in the model.
Analysis of the flight test data showed that both test flights exhibited nonlinear characteristics
(i.e., engine RPM was a nonlinear function of SLPC setting). For this reason, the dynamic
models were linearized over the selected operating range. The resulting models were linearized
about a nominal "trim" condition in the following form:
(_: - x_ )= A(x - x _, )+ B(u - u_ )
Since the X_-imterm is assumed to be zero, the above expression can be rewritten as
:_ = Ax + Bu- (Ax_i m + But_)
The last term, (Ax_ + Bu_m _), can be approximated by an additional "bias" input (set to 1.0 in
the u vector), thereby giving the conventional form:
2= Ax + Bu
NASA/CR--2002-211485 18
Themodelwascomputedusinga least-squaresapproximationto measureddata.model is
where
=[A,83*[X;U]
k -- number of samples (period of time) over which the estimate is desired
The form of the
= [k(tj ) ..... k(t_ )] e R ''_k
-x(t,) ..... x(tk) ] R,,,+.,,×k
[X;U]= u(ti) ..... u(tk)J
The system matrices [A, B] were computed as follows:
sc[x;u[[[x;vIx;v l'
3.5.2 Model Data Preparation
The linear dynamic models were developed using the data collected in the two flight tests. The
input data for the models was prepared as follows:
1. Measured data from Flight 41 and Flight 44 were placed into data files; this data was
described in subsection 3.3.
2. Other pertinent data not collected automatically by the PMDS was added to the data files
manually. This manually added data included the SLPC settings and cowl flap data
collected from the flight test engineer's reports.
3.5.3 Linear Dynandc Model Development
Three linear dynamic models were developed using the technical approach described above.
These models are as follows:
• Model 41 created from Flight 41 test data
• Model 44 created from Flight 44 test data
• Model 4144 created from a combination of Flight 41 and Flight 44 test data
These models were developed from the flight test data as described above. The portions of the
flight tests used in developing the models are shown in Figure 3-7.
The flight test data (inputs and states) used in developing the models is shown in Table 3-2.
Altitude slope, or rate of change, was used as a substitute for elevator setting (i.e., to indicate the
load on the engine). This was done because we had no way to collect elevator position or pilot
NASA/CR--2002-211485 19
$v
OJ
<
Flight 41 Altitude
12000 .... r ......... ,....... , ..... / ..... ,-- -, ..........
ortion of flight used/_or dynam+c modeling10000• .... r ...... , : , : _, , - ......
|
i _ ] i :
8000 1 .... , _ ,..........r+.... ...._........++......, I " I I i_
, I ./ I j \
i / I i i ".6000 _ - ...... _- +...... ,+ - - -+..... ,- .... "+,...- ....
, / _ i i : "
+ i - i i ! s
4000 +="=>+-"=+=+-'+-+ - + .... ,.......... + - - _ - +i i i i ! r
i :+ i i i J \_, i
2000 ..... +__ _ _._,L ,_......... ,_ ....... _ ....i / t + i _" ---i +? i , i , ".
i : , i ii J I J I " ....
0 1000 2000 6000 7000 8000 90003000 4000 5000
time (sec) PMDS Right41
Flight 44 Altitude
L ] ...... I [ I +../.. /. [/_ . I [ I12000 r ...... r ,...... ,....... , .... ,- -_'-" -"- ++'__ =_'<- - -, ...... ,......
_ . _ , f , '\ , .
[- : : _, portion of flight used _ dynamlic modeling J : /
'°°°°I, i ......!Ii i / i i _1 i ', i / j j
,i _ / ;_ I, II+-m600or ........... ,-...... + /.... +-...... ,...... +..... ,--+_-4--,_.....
i + , i / + , + , _ I , !
• b i i i i ' i' / , \
< 4000 _-................ +"..++=_...+...z-+c++.z............................. #4- ........ -I
, , , .... ,,I , [2000 ...... '_ .... J_-- - " ..... " ............ " .... "-+-_.--i ...... -_
' ,,/ ..... \, /• i i i i r
0 -------_"-"7---''/ i , i i + '4 i0 1000 2000 3000 4000 5000 6000 7000 8000 9000
time (sec) PMDS Flight44
Figure 3-7. Linear Dynamic Model Data
commands electronically. Similarly, we did not have a means to electronically collect airspeed
data, and that data was not collected manually during the flights. However, some of the other
sensor data effectively serves the same purpose (i.e., a combination of altitude, altitude rate, and
SLPC setting).
Pressure altitude and air charge temperature inputs track other inputs either directly or inversely,
but were used in order to produce a better model.
NASA/CR--2002-211485 2O
3.5.4 Linear Modeling Results
The above linear models were used to simulate a flight using initial conditions taken from either
Flight 41 or Flight 44. Simulated results were compared with actual measured flight test data.
The results are shown in Figures 3-8 and 3-9. The plots in Figures 3-8 and 3-9 were prepared as
described in Table 3-3.
Table 3-2. Linear Model Data
Inputs States Not Used Bad Sensors
Altitude
Altitude Slope
Outside Air Temp
SLPC
Cowl Flaps
Air Charge Temp
Pressure Altitude
(Dummy Input)
Fuel Pressure
Oil Pressure
Fuel Temperature
Oil Temperature
EGT2
EGT3
EGT5
EGT6
CHT1
CHT3
CHT4
CHT5
CHT6
Manifold Air Pressure
Engine RPM
Kollsman
Bay 2 Temperature
Bay 1 Temperature
EGT1
EGT4
CHT2
Figure
3-8
3-9
Table 3-3. Linear Model Plots
Flight
41
44
Models Used
41 and 4144
44 and 4144
Plotted Data
Measured data: heavy line
Linear Model 41 data: dashed
Linear Model 4144 data: solid
Measured data: heavy line
Linear Model 44 data: dashed
Linear Model 4144 data: solid
NASA/CR--2002-211485 21
Flight41 Altitude and SLPC setting
12000[_-....... _,!..............................................." " "J_" " "_ •'" "i i / i ! , i
_ 80o0) ..... :........ ........ I-- - _ ......... _ ..... _.... __\_ .....
<
4000 - - - ....... ;........---...........-r........ _ ....... _ .... r ....... r -'_:- - ._
i i i i i i
0 i 4 i i i , i
0.5_LI I r I I L 0
0 1000 2000 3000 4000 5000 6000 7000 8000
FuelPress
................• il
LI [ I
0 1000 2000 3000 4000 5000 6000 7000 8000
OilPress7O
i/ i i i
_/_" i I J
65 ....... ,........ .... .... - - -v I _ r
r i i i .........._, 6o_-............. ..... ............
r
55 I i j
0 1000 2000 3000 4000 5000 6000 7000 8000
FuelTemp
............... .......!.......!.......!......280 _-..... ,....... "- -
i i I i/ -- measured data ' _'" .... "_'- , _
i_" 275_ ___ Right41model =:..... _ ...... _ ....... , - - " " , - .....
270 I I = _ I _ L0 1000 2000 3000 4000 5000 6000 7000 8000
time (sec) PMDS Right41
Figure 3-8. Linear Dynamic Model Results, Flight 41
NASA/CR--2002-211485 22
360
350.=_=
@
_- 340
f330 L
1060
OilTemp
I I i
r "¢ , _ ,,
i i i i
11300 2000 3000 4000 5000 6000 7000 8000
EGT2
v 1040 ..............¢ 1020 ..........
1000 .......................
980............. ,...........
960 t
1050
1000==
-£- 950
9O0
0 1000 2000
J
' i
i i i
J I I
3000 4000 5000 6000 7000 8000
EGT3
i i i i
i I i I
........ , ............ 7 ........ T ....... T- -
i i Ilk ----
................... r _ i
i i
I I I
1040
_. 1020o
1000
_ 980
_- 96o
940
1000 2000 3000 4000 5000 6000 7000 8000
EG T5
I ......... L ....... L ......
........ i............ -_ ...... _r ...... _ ...... _ -
i J I I /...............1-- measured data t ._ l(; -- _v_ _,'_'_A
, l _,/ I I
Rlght"l model .... t 7 _- _ _ ..... _'" _ -'r- -"- -'--- r ......
Rights41&44 model V ' _ _ i _I I t I I I
1000 2000 3000 4000 5000 6000 7000
time (sec) PMDS Right41
8O0O
Figure 3-8. Linear Dynamic Model Results, Flight 41 (continued)
NASA/CR--2002-211485 23
EG T61020 , , _ , r
! ' ....I I I i I
1000 ...... '- _ ....of_ , ..... , ........... , .... _ ...... T ..... _ .......
980, - ........... - ...... _ ....... _ ..... ,- .... _-......
...... ,.... V .... ........ 7--- .......920' i I J
ov
m'-I
0 1000 2000 3000 4000 5000 6000 7000 8000
CHT1
460 ii [ D# _ I I
400 L i , j _ _
0 1000 2000 3000 4000 5000
CHT3460 I ! I
I I I J "
v_o_(_ 440 .............. It_................. _ !II.... %% _- III -_ rI .... # "
420 , , _c q - -I I ", I I
_X
.e 4uu- ................ _....... -_....... _ .......I-- I I I I
I I I I
I I I
380 I i i I0 1000 2000 3000 4000 5000
6000 7000 8000
6000 7000 8000
CHT4
460[ I I I I ' ' i
44o : - _, -- _....... _......! _ i i
i ,t.
= 420 ........ ,:........ ,"--- -_ _-', , i-'_J-l- - -- -'_= "-'_'_- _- - .....filler' .......... ' ' 1-- measured data / "'" " 'i "_. 400 Right41 model [ .... -_.................
Rights41&44 model ! _ i '".." ',380 I i i l i0 1000 2000 3000 4000 5000 6000 7000 8000
time (s_c) _ Right41
Figure 3-8. Linear Dynamic Model Results, Flight 41 (continued)
NASA/CR--2002-211485 24
CHT5
390r _ r _ _ _
/ ', t I t t_ t t t i
"_ .... 1_ .... L .......
oooV ...............................,,,,..,oo i i A i J i i
," r t t / J t tt _ t t t t
._ 370_-....... . ...... -,_'='_-%-.-L.-_-'_o ---:K4 -'_'_- - .... " l] - - -_
i ' i " i_ i t I i
_. 360 ......................,,, ,,, _,,, _,,, ,_, ,,_ _, --"350 k . R i
0 1000 2000 3000 4000
CHT6430
5000 6000 7000 8000
C" 420
9 410:3
400
E
:3(/)
_r
[ I I I [ l
ti I I _ i
i I I i_ i
...... ,....... _--_ ..... _...... _..... ,.r -_ - - - • ...... T ......i i_ i f I i i i
i /t\, "'- LJv/I \ %.. ,i' \ _, Ii \ _ \ ' I ",
390 ............... ,....... n ....... • .............. r-_ .........
380 _ I I _ I I
0 1000 2000 3000 4000 5000 6000 7000 8000
ManifoldAirPr
100 I i I [ I I It _ t t t t t
I i_ t t t
90 ....... _........ _-""_--_- - _ ....... _ ....... _ ....... _ ....... L .... _ _] t t i t t /
r t _ t i
80 ' _ ' ' _ ' •........ I........ I .......... _ ....... 7" ...... T ....... _-- --
70 , _ .......... _---,,- - '- - .... '-
60 I I _ '0 1000 2000 3000 4000 5000 6000 7000 8000
30OO
" 2800
"o 2600
GO2400
.c_
2200
2000
EngineRPM
........ , ....... _....... _ --..-,_h,._- - -1....... _- r-_ _ _Ir_"__ _,, _ B _ _ _ _
I Rights41&44 model ' : : : :I I I I _ I i
1000 2000 3000 4000 5000 6000 7000 8000
time (sec) RVDS Right41
Figure 3-8. Linear Dynamic Model Results, Fright 41 (concluded)
NASA/CR--2002-211485 25
A
<
Flight44 Altitude and SLPC setting
12000 ........ _........ F....... _ ....... J ....... _ ....... ._._._._:, ......... _ ......
8000 .... ,........ ,........ , ............ __ ...... _'_-"_ ....... , ...... ---.-_- ....
J , , I , I i I '_.
P
0_- ................... i i, ................ i i Ti .......... i '1/L I L J I / 0
0 1000 2000 3000 4000 5000 6000 7000 8000
FuelPress
72 I ] I I _ i
i I I i //
-_'_. ...................... j______/_ ....70 ........... _ .... '
68
3
66
64
62
] i I i
i I I 1
I I I i I I
0 1000 2000 3000 4000 5000 6000 7000 8000
OilPress
75 I I I ', I i /I I _ I I
70 ........ *........ *...... _ ....... "--_ .... -_ ....... _- .... .,_ _ L__ _ _;_ _-_ , . _.-. _ ,, -._1 _ , ;
60 ....... _........ ,-
_ I I I
55 _ _ I _ L _ I I0 1000 2000 3000 4000 5000 6000 7000 8000
Fuel Temp
310
300F ................ _---_ ....... _ ....... _-...... L ......o
290 ...... _........ ,........ r ..... r -
measured data ' _ ' ' "
t_" 280_- ___ Right44model -- --_ ....... _ ....... -_-- . - , .....
l Rights41&44rr_lel ] I : i : :I I I270 " I i I I
0 1000 2000 3000 4000 5000 6000 7000 8000
time (sec) RVlDS Right44
Figure 3-9. Linear Dynamic Model Results, Flight 44
NASA/CR--2002-211485 26
1050
2
_1000
950
I _ I I I
....., .....i-F1_----i.......i.....i....i.....'_......
....-_-_,
_ i I I I0 1000 2000 3000 4000 5000 6000 7000
EGT3
1050
8000
1000,
$_- 950
9O00
1040,
_. 1020 !o.. I¢ 1000!
. 980
96o
9400
; I F [ I
I r F I; _ I J
I \v I
I I
1000 2000 3000 4000 5000 6000 7000 8000
EG T5
i"_ % I ] [ [ I
....... , ...... _-D-_-.C-'_.-j,--_-_ _ ...... _ ..... "-,-.7 .... ,,, ,
....... ,........ ,:_,___- -_,d__-____..__,:,.___:.... __,_:____)_--::\,j -.,-.,.-.__T- -, ........
,_h,,,4mo,_, l_'-'_-- :-"-;---: ...........Rights41&44 model / " " "
I J I I I I
1000 2000 3000 4000 5000 6000 7000 8000
time (sec) PMDS Right44
Figure 3-9. Linear Dynamic Model Results, Flight 44 (continued)
NAS A/CR--2002-211485 27
EGT61020 , f _ _
_-. 1000 - :...... : -_c - -i ............. ', - - T ....... _-.....
980 ....... ........ i ....... _......... - .... _-.... _-I , , _/,,%_v_v/ I ' , , .7 I
, , , , ,96o _............... ..... ....940 ........... :......... -_ ....920 / ; l J I i
ov
D
0 1000 2000 3000 4000 5000 6000 7000 8000
CHT1
480 | F_ !,,6oL....... _........ .--%...... .--.-/--__\I....... _---_-_-_...... _-.......
38o I : , i i i .... i
o
7
J \_/ I
0 1000 2000 3000 4000 5000 6000 7000 8000
CHT3
50o ! _ i ! !I r I I i r
I I _ I I
4501 ....... '........ _---... - -, ........ r ....... r- ......
I i i \ I I I
I \ I I
400 ........ ............... , __ _ =- - - - _ ...... \_ .... r _ - -_- -
350
5O0
I I I I _ /i
I I I I ,
J I I I I
3OOO
450
_)
_- 4o0
0 1000 2000 4000 5000 6000 7000 8000
CHT4
3500
I 1 ; I II J i I , I
I _ I I I I
........ I............ \ -I ........ _ - - T ....... F .......
I \\ I
measured data ] ' " __'- _ _ _ - _"• i I I I % \ I/Right44 model , , , , -.
-- Rights41 &44 model I ' ,' ,' "" - ' ,I l i , I ; I
1000 2000 3000 4000 5000 6000 7000 8000
time (sec) RV[_ Right44
Figure 3-9. Linear Dynamic Model Results, Flight 44 (continued)
NASA/CR--2002-211485 28
400
CHT5
I I I [ Ii I I I i
I I I I
I I I I I
_,} _l_U ...... Pb....... t% _-E-- - _ -i ....... :_'_J_*_ i i J7-- r_ ....... F .....
_4.rl I J I
0 1000 2000 3000 4000 5000 6000 7000 8000
CHT6
4601 I 1 I
I I I
I I I I r
420, ..... ,.... ,- - - :, "_ - _- - ._'_,"- -_, _ - ...... • - -_ - -
__ I I ", 4 / x ) I ]I I _ D_ If_ _i_ / _ _ ___.J ..... C .........400 ............ i.... ,-/ _-" ---'_
I I P _ t X I I /380 .............. ,..... _....... _..... _ .... ----....... ;"......
9,_n I I L I J
0 1000 2000 3000 4000 5000 6000 7000 8000
ManifoldAirPr
100 I l ] l I l lI I I I r I
II I _ P I
-- _ . , _._ _ _ , . ,...... i ...... I_ - - ._I .J_I_ ..... J ....... .L ....... ..- ..... L .... L----
_u ........ _ ........ I........ -2---_Dcr ....... T ...... _ - ---
i i i i _
if)
¢_ 70 ........ ,...... ,....... -_....... _ ....... _-_ -,_'_- - 7. - '- .......
an i i i i
0 1000 2000 3000 4000 5000 6000 7000 8000
EngineRPM
r i ]
...................... _j .... J ........ L ...............
........ II........ I_I .... II.............
measured data t -_ -2 _;; ..... f_ ...... : ..... I' ',. - -Flight44 model _,- [ , ,.
• .to------.. model _ , , , , , .,I i I I L I I
28O0
" 2600
_24O0
o9
•_ 2200
2000'0 1000 2000 3000 4000 5000 6000 7000 8000
time (sec) PIVDS Flight44
Figure 3-9. Linear Dynamic Model Results, Flight 44 (concluded)
NASA/CR--2002-211485 29
3.6 Nonlinear Dynamic Modeling
In reviewing the flight test data, it was observed that there existed a nonlinear relationship
between the engine RPM and the SLPC setting. As the SLPC setting was set to values just below
0.6, there was little change in engine RPM. Presumably in this region, the FADEC is adjusting
the propeller pitch to satisfy the pilot's power level command. We were unable to acquire
technical information about the FADEC control laws to verify this assumption. However, this
measured characteristic of the SLPC and engine RPM response provides an opportunity to create
a piecewise linear model consisting of two operating regions:
• Region 1: SLPC values greater than or equal to 0.6
• Region 2: SLPC values less than 0.6
3.6.1 Nonlinear Dynamic Modeling Approach
The nonlinear dynamic model is constructed in state-space form. The form of the model is asfollows:
k = Ax + Bu + b * min(0, SLPC - 0.6)
y = Cx + Du
where
x = state vector
u = input vector
y = output vector
A,B,C,D = linear system matrices for Region 1 (SLPC > 0.6)
b -- vector of coefficients used to adjust the response for Region 2 (SLPC < 0.6)
The nonlinear model was developed using the following procedure:
1. Using measured data from Region 1, system matrices A, B, C, and D were computed
using the method developed for the earlier linear dynamic models.
2. Using measured data from Region 2 and system matrices A, B, C, and D, the b vector was
computed using a least-squares approximation.
3.6.2 Model Data Preparation
The nonlinear dynamic models were developed using the data collected in the two flight tests.
The flight test data files prepared earlier for the linear modeling were reused for this work.
Region 1 and Region 2 of each flight were determined based on SLPC setting, as shown in
Figure 3-10.
NASA/CR--2002-211485 30
10000
= 5000
<
300O
" 2800
"_ 2600
_ 2400
2200
20000
0.8
.c_
0.6
_ 0.4_J
0.2
Flight 41 Altitude
> _6 ' _ _ \SLF'C O ....... i ........ _ .
rr , rr/ 1, ,rl 1, , ,,-_ ....4" _ _,'_ ,
I: II/ :II i: I I\It
/J', ,, ,',, ,: , , ,................;" I; II :11 1; II I0
0 2000 4000 6000 8000
Flight 41 Engine Speed
I! i [ I I I 11 I I I
, t.} ......... /, I[_,II _! I, II I !:r,t
_._. : _...........I,_Ti ,I! _T-ll "1-j
I: li iii Ii II Ii2000 4000 6000 8000
Flight 41 SLPC Setting
1
_--I_- L $ -_* * _ - L-_ J,--L --4
,_ _ i,l_ tl, |/ /
_ _L__:___a _____a.____a__
/ : tJ ; : l |0
0 2000 4000 6000 8000
time (sec) PIVlDG Right41
<
10000
5OO0
00
3OO0
" 2800
"_ 2600
6o 2400
.__
2200
20O00
0.8
=o0.66O
_ 0.4_J
0.2
Flight 44 Altitude
SLPC > 0_6_o _----_ _--_--_ _',_
_SLF__-._O_ 1-i__, 11_/ .... t-_l-r _
si i ",
_ _ _ ,..... , _ .... :.... _,
:1 !1\!
...................'_:I II : l J ] I t I_2000 4000 6000 I000
Flight 44 Engine Speed
II _ _ _ _ _ i
:t I{ '_,.< : L, Ii ....i. I11
:I 'lI ii -:l.....i iq
;I II ',I III2000 4000 6000 8000
Flight 44 SLPC Setting
1
.... _,__. __ _ L__,_-¢_,
ili _1 _1.... _- ....-4 - - z..., -__ _ _ L-k-
i
0 __i
0 2000 4000 6000 8000
time (sec) PMDS Flight44
i i
IL '
i
...... i ......
i
it
Figure 3-10. Nonlinear Dynamic Model Data
3.6.3 Nonlinear Dynamic Model Development
Three nonlinear dynamic models were developed using the technical approach described above.These models are as follows:
• NL Model 41 created from Flight 41 test data
• NL Model 44 created from Flight 44 test data
• NL Model 4144 created from a combination of Flight 41 and Flight 44 test data
The flight test data used in developing these models is shown in Table 3-4. Since the partitioning
of the flight test data into two regions resulted in fewer data points in each region, we were
forced to reduce the number of inputs and state variables in order to achieve models that were
stable.
NASA/CR--2002-211485 31
Table 3-4. Nonlinear Model Data
Inputs States Not Used Bad Sensors
Altitude
Altitude Slope
Outside Air Temp
SLPC
Cowl Flaps
(Dummy Input)
Oil Pressure
Oil Temperature
EGT3
CHT3
Engine RPM
Kollsman
Bay 2 Temperature
Fuel Pressure
Fuel Temperature
EGT2
EGT5
EGT6
CHT1
CHT4
CHT5
CHT6
Manifold Air Pressure
Air Charge Temp
Pressure Altitude
Bay 1 Temperature
EGT1
EGT4
CHT2
3.6.4 Nonlinear Modeling Results
The above nonlinear models were used to simulate a flight using initial conditions taken from
either Flight 41 or Flight 44. Simulated results were compared with actual measured flight test
data. The results are shown in Figures 3-11 and 3-12. The plots in Figures 3-11 and 3-12 were
prepared as described in Table 3-5.
Table 3-5. Nonlinear Model Plots
Figure Flight Models Used Plotted Data
3-11 41 41 and 4144 Measured data: heavy line
Nonlinear Model 41 data: dashed
Nonlinear Model 4144 data: solid
3-12 44 44 and 4144 Measured data: heavy line
Nonlinear Model 44 data: dashed
Nonlinear Model 4144 data: solid
NASA/CR--2002-211485 32
Flight41 Altitude and SLPC setting
,2ooor....... :.............. !....... ! ..... ......................7_- ---! .......
=" 8°°ol-....... :........ i........ ',--/ ............= .........._--; ..... ,_.... _......._- .....
|= 4ooo,-....... ..................:........ _..... _ .......... ,_--,,=
<
_. 65
(n
6O
i i
0 L .......................................
F ' ' " -- o.s_oL 1 I L 00
70
550
360
350
_- 340
3300
1 O50
°_ 1000
$_- 950
1000 2000 3000 4000 5000 6000 7000 8000
OilPress
[ i [ I F _ rI I I _* I ._ i /
I I _ I ._ i // I I
i i i i " i
i i i i i
i i i i t i
....... i ..... i i i , r _ __
i
1000 2000 3000 4000 5000 6000 7000 8000
OilTemp
I I [
_! j i I
........ J ....... i ....... 7-1- 71 ...... T ....... r .......
I / i Ii i i
..................... i _ --
i i
II I L I
1000 2000 3000 4000 5000 6000 7000 8000
EG T3
900
..... ,___ _ _,_ _,__-_rneasureddata , , ", ] . ._Flight41 model , " _q ,
-- Flights41&44 model ' 'J I I I
0 1000 2000 3000 4000 5000
time (sec) PMDS Right41
1 1
i i
I i
6000 7000 8000
Figure 3-11. Nonlinear Dynamic Model Results, Flight 41
NASA/CR--2002-211485 33
46O
_" 44Oo
.=
._420
_'400#.
3800
3000
" 26O0
"_ 26008.
co 2400
2200
200O0
CHT3
I I r /_. p _ [i i i _ i
J • ._ i i ,
............... , -- -_ ........ 7 ....... 7 .......
, i i i i
I i i i
i , i _ i i i
i i r _ I i \ i i |
i i i r - \ \ I _ _ i I•J I I [ ! I I ' [
1000 2000 3000 4000 5000 6000 7000 8000
EngineRPM
I I ]; I
L I I J L I _ I
1000 2000 3000 4000 5000 6000 7000 8000
time (sec) RvlDS Right41
Figure 3-11. Nonlinear Dynamic Model Results, Flight 41 (concluded)
NASA/CR--2002-211485 34
@
<
Flight44 Altitude and SLPC setting
12000 ................... I ...... _ ....... _ ............... __ ........ _ _ _ _ L ......i i _ i \i i i , 4
]
8000 ..................... - ........................ : ......... '" .......qd / i i] / i i i ",
/ \,I i i4000 ......... *_-_ ........_ ....... _ ..... _ .... r ...... - -_'- - -\
i i i i
i i i ii i
0 ........... - - - - -i ........ i .... i .............. i i _
i i I I t
0 1000 2000 3000 4000 5000 6000 7000
OilPress
75 ', I I ', Ip i _ i i
70 ................... _....... J ....... _ ...... _._ _ _ _.,L......
¢ 65
_- 6o
550
370
_- 360o
350
340I_ 330
32O
/ k
r i _ t _ i I , _ "
I i I I I
1000 2000 3000 4000 5000 6000 7000
OilTemp
I /
t /i
I i
.E
,lh ).0.5_'0
1000
0
1050
1000 2000 3000
!
8000
4000 5000 6000 7000 8000
EGT3
1000
I " Ii i
950 _- _ measured data ..... I ........
- -- Flight44 model _
-- Rights41&44 model _,900 _ i =
0 1000 2000 3000
time (sec)
[ I I
i i i
I I i
• I I........ T ....... Y .......
i i
i
i
I I
4000 5O0O 600O 70O0 8000
_ Right44
Figure 3-12. Nonlinear Dynamic Model Results, Flight 44
NASA/CR--2002-211485 35
CHT3
J
i
i r
...... t- ......i
i
0 1000 2000 3000 4000 5000 6000
EngineRPM28O0 I I [
7000 8000
_" 2600
_[ 2400
•_ 2200
20O00
1 I I [ I I
1000 2000 3000 4000 5000 6000
time (sec) PMDS Right44
7000 8000
Figure 3-12. Nonlinear Dynamic Model Results, Flight 44 (concluded)
3.6.5 Comparison of Nonlinear and Linear Modeling Results
A simulation was performed to compare the quality of the linear and nonlinear modeling
methods. The nonlinear model 4144 compared to a linear model 4144 using the same inputs and
states (as listed in Table 3-4). The results are shown in Figures 3-13 and 3-14. The plots inFigures 3-13 and 3-14 were prepared as described in Table 3-6.
Table 3-6 Comparison Plots
Figure Flight Models Used Plotted Data
3-13 41
3-14 44
Nonlinear model
vs. Linear model
(modeled for
combined Flights41 and 44)
same as above
Measured data: heavy line
Linear Model 4144 data: dashed
Nonlinear Model 41 44 data: solid
Measured data: heavy line
Linear Model 4144 data: dashed
Nonlinear Model 4144 data: solid
NASA/CR--2002-211485 36
"0
Flight41 Altitude and SLPC setting
12000 ...... 1 ............... '....... _ ....... _....... '............_"_ .... ,i i
i i i i _ i \ ii i i i ., r i _
8000 ................ , ..... - ..........._ "=='_"/ - _ ........... _=__- - -i i _ / i i
li 4i b / i i , ,,..../ b i ir ii i t , \
4000 ...... ........ _.................... "_ ...... 7 ...... _ .... _ ..... _ ..... ; - -\i i i i i
i i
0
.E
I ...... i................ i ........ _ .......... i -............ o.5 oL k I I I I 0
0 1000 2000 3000 4000 5000 6000 7000
OilPress70
8000
_. 65
_ 60
55
r i i i r ii _ i i L
i i i i _ r
i p i i i
....... i _ i t _ , i i _ _
i
i
0 1000 2000 3000 4000 5000 6000 7000 8000
OilTemp
360 , __ ,
I i I i i
I I i i350 ....... _........ ,........ _ .... _ ....... _ ......
340 .... ,........... 2"_
#.i I i ii i i i
330 I I _ _0 1000 2000 3000 4000 5000 6000 7000 8000
1050
1000
950
EGT3
9OO0
' 1 I 1 I II I i _ I
I i I II I I i
....... i.......... I- -I ........ T ....... _- ...../ I i i
measured data
- - - Linear Model Rights41&44 I i i " , - iNonlinear Model Rights41 &44 i i ', i
L I i I L I I
1000 2000 3000 4000 5000 6000 7000
time (see) _ Right41
,0130
Figure 3-13. Nonlinear vs. Linear Model Results, Flight 41
NASA/CR--2002-211485 37
CHT3
460_ ] I t _ 4 _ I =.
_ I t i I
:_ t \_ I i i 1
420 ....................... , .......
I I i + \ _,
400 JI
0 1000 2000 3000 4000 5000 6000 7000 8000
300OEngineRPM
2800
v 2600 ........ I........ i-"_ I I I
oo 2400 ,
._ _ measured data __2200 Linear Model Flights41&44 'Nonlinear Model Flights41&44 ..... _ ......
I
2000 J I , J3000 4000 5000
time (sec) PfvDS Flight41
0 1000 2000
Figure 3-13. Nonlinear vs. Linear Model Results, Flight 41 (concluded)
NASA/CR--2002-211485 38
¢)
¢)"0
<
Flight44 Altitude and SLPC setting
12000 .............. J,............ __"-=" '='-_ -"_ -.............. .... f , --_ ...... -_k
8000 .............................. __"'= ..... i'L _ ...... . ...... _.._-.. ...../ p i
\
4000 ................ "_.............. " _ ........ n ...... _ .... r - - - r - - ,.,-
: i i ,
: ' ' ' -jlo ............... __ ..... _ ..... '-.... _o.5,_
l i r 00 1000 2000 3000 4000 5000 6000 7000 8000
OilPress70
"_ 65v
i 60
55
I I I It i ' t _
, _ i i I //
....... i ...... _ ....... I ...... I- - - - ...... T ......
t i / _- r t
...... i ........ r t t _
I I
I
380
_. 370
360
.350
#- 340
330
1000 2000 3000 4000 5000 6000 7000
OilTemp
800O
I I I lb I I !
I i ....... L ...... L ......
t _ t I _ t t
I _ t f f I J
I t x I t I
t r _- t t
I I ...... I .... L ...... L .....
................ )-- I I
p I I
t
t I I t I
I I I I
1050
o 1000
_- 950
9O0
1000 2000 3000 4000 5000 6000 7000
EGT3
8000
I • I
I I
I I
I I
measured data
- - - Linear Model Rights41&44
-- Nonlinear Model Rights41 &44' I I I
I I I
I I I
I , I
........ 1- ....... T ......
i
1000 2O0O 3000 4O00 5000 6000 7000
time (sec) PMDS Flight44
8O00
Figure 3-14. Nonlinear vs. Linear Model Results, Flight 44
NASA/CR--2002-211485 39
CHT3
4801 f I I r I I, __ 460 ....... ,........ ,o "- _--- F .......
440 ...... ,........ ,..... _ - - - _- .......;E
420F....... [..... ; ...................i i i i
400 i i
0 1000 2000 3000 4000 5000 6000 7000 8000
2800EngineRPM
........ ,.... , ......... _- - -_ ....... 7 ...... T ......
........ E........ I - - %.- -_- - T .....I I
measured data , _ \2_" ',k,/
I - - - Linear Model Flights41&44 ...... . - _ - - _ ....... F - - _'- - -
[ Nonlinear Model Rights41 &44 I , I II I I I I I J
_" 2600v
_. 240060
-_ 2200
2O00
0 1000 2000 3000 4000 5000 6000 7000 8000
time (sec) _ Right44
Figure 3-14. Nonlinear vs. Linear Model Results, Flight 44 (concluded)
3.7 Data Analysis Conclusions
Conclusions drawn from the analysis of the flight test data and the resulting static correlation
models and dynamic models are presented in the following subsections.
3.7.1 Static Correlation Modeling Conclusions
Analysis of the static correlation modeling results provided the following observations:
• Disconnected sensors are clearly detectable using a static correlation model that compared
sensor output values with expected values (see Figure 3-6).
• Intermittent sensor faults are also clearly detectable using a static correlation model (see
Figure 3-2).
• Data from multiple flights can be combined to produce an improved static correlation
model (see Figures 3-3, 3-4, and 3-5).
• Static correlation models worked well for detecting sensor faults.
NASA/CR--2002-211485 40
3.7.2 Dynamic Modeling Conclusions
Analysis of the dynamic modeling results provided the following observations:
Combining data from both test flights generally produced better models. This can be seen
for the linear models in the last subplot in Figure 3-9 (comparing model 4144 and model
44).
• Nonlinear models were better than linear models for the same number of inputs and states.
This can be seen by comparing the nonlinear and linear model results in the last subplot in
Figure 3-13. The nonlinear models do, however, require more empirical data to generate.
• As expected, more data gives a better model. This can be seen by comparing the linear
model in the last subplot of Figure 3-8 with that in the last subplot of Figure 3-13.
• Extending the dynamic model results from these two flight tests to future development
(having many more flight tests) will produce much improved dynamic models. Combining
the empirical modeling approach with a physics-based modeling approach also has the
potential for improved accuracy.
• The two flight tests performed on this program demonstrate that dynamic models of the
engine/aircraft can be produced using relatively simple and inexpensive instrumentation
such as would be found in a commercializable on-board engine diagnostic system.
• Dynamic modeling has the potential to detect mechanical faults internal to the engine aswell as sensor faults.
NASA/CR--2002-211485 41
Section 4. Technology Roadmap
The PMDS concept shows promise as a means to improve the pilot's awareness of the condition
of the engine. This technology will require more development before it can be commercialized
and broadly applied in the aviation marketplace. Honeywell has prepared an overview of the key
technology areas that require further development. This information is presented for the purpose
of guiding the direction of future development. The key technology development areas are
• Modeling and fault diagnosis algorithms development
• Hardware and software development
• System development
The following subsections discuss these three key areas and provide a roadmap for future
development of each area.
4.1 Modeling and Fault Diagnosis Algorithms Development
The PMDS concept employs model-based diagnostic technology. Mathematical models for the
engine, sensors, and related equipment are created from first-principles analysis and from
empirical data collected from flight and ground-based testing. These models are used to detect
faults in the engine, sensors, and related equipment. Fault information is used to make failure
diagnoses. This process (for empirical models) is shown in Figure 4-1.
Make Empirical / Physics-basedPlant Model
ModelIdentification
Plant ModelSimulationModel
U_ Use Plant Model for FaultDetection and Isolation
input plant Modelvariables _JSimulation
I _ IModel
Data variables
at
prediction_ residual
IFault Estimation
-_]Algorithms
fault estimates
Figure 4-1. Modeling and Fault Detection
NASA/CR--2002-211485 42
4.1.1 Background
Modeling-System modeling for engines can be done in various ways. Empirical modeling uses
input/output data to derive a model of the engine. First-principles modeling uses physics and
thermodynamics, etc. to derive a model of the engine. The pros and cons of each approach, as
well as that of a combined approach, are discussed below.
The advantage of empirical models is that they can be very simple and general enough to apply
to a variety of systems with little change from system to system. The disadvantage of empirical
models is that they typically require many unknown coefficients that need to be identified using
extensive empirical data. A general rule, based on the Cramer-Rao bounds, is that the number of
required sample points is proportional to the square of the number of unknown coefficients.
Given enough sensors and sample points, it is possible to evaluate all the coefficients, but if any
system dynamics change, it may be necessary to start over. For example, given a system with
x _ R", numerically differentiated state derivatives _ R" and u _ R m , coefficient matrices
A _ R ''×'', B _ R ''_' , and a linear empirical model of the form .4"= [A, B]*[x; u], the unknown
coefficient matrices [A,B] can be evaluated using k samples of each of the n+m measurements of
x and u. Let
F = [_(tl ) ..... )?(tk )] _ R ''×k
x(tl) ..... x(t k)] R _'+''_×k= E
G [u(t l), ,u(t_)
Then F = [A,B]*G and [A,B] = F*G'*inv(G*G').
Since the n+m coefficients in each row of [A,B] only affect the derivative of the corresponding
state, those n+m parameters need to have k > (n+m)^2 samples of that state derivative in order to
satisfy the quadratic Cramer-Rao bounds. With n = 5 and m = 5, that requires k > (5+5)^2 = 100.
With n = l0 and m = 5, we would need k > (10+5)^2 = 225 samples. With the test flights used in
this study, we had 110 good test points in each of the two flights, for a total of k -- 220. This
means that a single flight, with 1 l0 samples, is enough to identify a five-state model, while a ten-
state model would require the data from both flights together.
The advantages of a first-principles model include the ability to do "what if" experiments with
the model and the ability to adapt the model to new untested situations. Another advantage of
first-principles models is that they typically have fewer parameters than empirical models. It is
possible to take advantage of the structure imposed by multiple time scale data and the natural
separation of the system dynamics to create multiple subsystem models.
The parameters in a first-principles model are closely related to things that can be measured in
isolation from the rest of the system. This makes it possible to combine the general structure of
the first-principles model with empirical data to calibrate a few unknown parameter values. The
disadvantage of first-principles models is that they take considerable time to derive and must be
tailored to each specific application.
In our application, the only nonempirical information that we used was the nature of the
nonlinearity in the SLPC. The SLPC input controls a combination of RPM and propeller pitch.
The system behaves essentially like one linear system above SLPC = 0.6, and like another linear
NASA/CR--2002-211485 43
system when SLPC is at lower levels. Rather than doubling the number of parameters, we kept
the same A matrix and simply added one column to the B matrix, which multiplied the added
input: rain(0, SLPC - 0.6).
Failure Detection and Fault Diagnosis-During the AGATE program, a set of failure detection
and fault diagnosis algorithms was developed. A set of engine models is used in computing an
estimate of each of the various engine performance parameters needed for fault diagnosis. An
example failure detection algorithm for oil pressure is shown in Figure 4-2.
Oil L__ Filtering, A/D _P°f _0_[Pressure [ /] and unit cony.
[Sens°rI I ,Olo;+tOo lO]II i-ll Tol&anceI
RPM _ I------ _ Detection
Oil Pressure IOil Temp"----_ Estimate
Oil Quant-----_
APop
Periodic sampling
at a period of tm
Average
Average
Store Instantaneous
) out-of-tolerance
value
>
J Store APopave
Issue
Maint.
-x tol Warning
Early Wamin I
Tolerance &
Detection
>
Store Popave
Figure 4-2. Oil Pressure Failure Detection and Warning Diagnosis
The anticipated failures and the associated detection for each of the engine subsystems are
arranged in a matrix of probable faults with the associated detection methods. An example fault
diagnosis matrix related to oil pressure is shown in Table 4-1.
NASA/CR--2002-211485 44
Table 4-1. Oil Pressure Fault Modes and Corresponding Detection Tests
_$
O
O
o
O
O
"O
¢D
, i
m
' E :i • t
it
: cn
. = : .=, Q. t_
_ , ¢-
: __ , =>J :3 "_
ta
: : "8
i• i•'o :, i
, ,,,
i. !•,
,
Detection Test Methodology
Difference between oil pressureestimate and oil pressure abovedetection threshold
Difference between oil delta Pestimate and oil delta P above
detection threshold
LL
.r"
0 o')
1:* 0, 0..t
J
UJ O O}
o _1 E< oEL
_9
"5 ._- o_
._ .-n ,, N
(Other tests...) : : :
(Other tests...) : : ,
• ',•
e- ,,1=
The current fault detection amounts to noting when a measured output deviates (by more than
some threshold amount, for several samples) from the value of the model output, when the model
is being fed the same inputs as the actual system. The current fault diagnosis assumes that the
fault lies with the sensor whose value is deviating from the model prediction. A more detailed
fault diagnosis would have to include engine faults as well as sensor faults.
An empirical way to calibrate an engine fault model would be to record sensed variables with
and without each expected fault. Another option would be to derive a first-principles model of
how each fault affects each sensor output. It would likely be necessary to combine the methods,
using a first-principles model and using empirical data to calibrate the remaining unknown
coefficients associated with the faults.
4.1.2 Future Development
Professor Giorgio Rizzoni and Gary L. Parker at Ohio State University have developed a first-
principles individual-cylinder model of an internal-combustion engine under the AGATE
program. This model was implemented in Simulink. In future applications of our PMDS
technology, we would like to use our existing flight data to evaluate some of the coefficients in
such a first-principles model. For example, an empirical modification is often needed for the
first-principles model of the relationship between throttle setting and manifold pressure. We
NASA/CR--2002-211485 45
would also like to get empirical data for the most common types of faults experienced by general
aviation engines.
A PMDS also would benefit greatly from additional inputs from outside the engine, such as
aircraft speed, aileron, elevator, rudder, and flap settings, propeller pitch, etc. With a
combination of a first-principles model and the availability of sufficient measurements of engine
and aircraft variables, it will be much easier to achieve a good calibration of the modelcoefficients.
Future development of an engine monitoring and diagnostic system will require a prioritized list
of engine faults to guide the diagnostic development work. Ideally, this technical information
should come from an engine manufacturer. Data from more than one engine manufacturer would
be even more helpful in advancing the technology.
While at present most general aviation operators are using gasoline-powered piston engines,
current development programs in alternative-fuel engines, such as diesel engines, can offer a
means to collect the engine fault information needed for the eventual development of an engine
monitoring and diagnostic capability for these new engine technologies.
The application of vibration monitoring is an obvious area of interest in the subject of engine
monitoring and diagnostics. Certain engine conditions such as bearing wear are best detected
through vibration monitoring. The AGATE PMDS hardware was designed to optionally take
information from a vibration sensor mounted directly on the engine, process the vibration data,
and determine prognostics from that data. A university-led study performed under the Honeywell
AGATE program has shown that, for piston engines, vibration monitoring can be used to detect
engine conditions such as bearing wear. However, the team discovered that it was not sufficient
to analyze the frequency spectrum of the vibration data as in traditional vibration monitoring
methods, but rather to use a direct sample of the vibration time signal. This preliminary study
used 5000 samples/second for one second of engine operation in order to detect engine
conditions sufficient for prognostic prediction. Further research and development will be
necessary before such optional vibration data can be used in a reliable fashion for engine
prognostics.
The future development steps described above are shown in Figure 4-3. At this time, the scope of
the work, the timing, the source of development funding, and the makeup of the development
team are undefined. These planning issues will be addressed as the general aviation community
continues to dialog about engine diagnostics technology and market needs.
NASA/CR---2002-211485 46
@
r-II<II
//
//
//
//
I_
.L-
_,._
ILPI
I..I
i"L
IIII
I\
\III
¢-
II
•
¢-
.......I
I
e-,
¢-
I
II
I:=
f,.t.,U..,
E
r,,
<<z_,t-,l
1.1..1
<
@@"i:@@l:iIlli:Ii
't,,i,II
NA
SAJC
R--2002-211485
47
4.2 Hardware and Software Development
During the AGATE program, Honeywell developed a prototype design for the PMDS. This hardware
and software design will require additional development to bring it to a commercializable form. The
following subsections describe the key hardware and software development areas ahead.
4.2.1 Background
The AGATE prototype system architecture concept is shown in Figure 4-4.
Bus Interface
Airframe
Displays andWarnings
Power
Supplies
Avionics Bus
Bus Interface ThrottleController
PropellerController
FADEC
IgnitionN,knifle
Engine
Single LeverPower Control
Propell_Govemor
Pilot
Figure 4-4. PMDS Architecture
PMDS Hardware Design-The PMDS hardware architecture is shown in Figure 4-5. (The hardware
arrangement used in our flight testing under this program and in the earlier AGATE flight testing did not
use individual engine sensors, but relied on a serial communication interface to the FADEC. The test
setup is described in Section 3 of this report.) The present PMDS hardware is a prototype design
implemented under the AGATE program. This PMDS hardware is shown in Figure 4-6.
NASA/CR--2002-211485 48
During the design of the PMDS hardware in the AGATE program, Honeywell solicited guidance from
the AGATE community about electronic design standards. At the time, the AGATE electronic design
standards were still evolving, so the Honeywell design team opted to design the prototype PMDS
hardware using then-available best practices for guidance (i.e., for lightning, EMI, thermal, shock, and
other design criteria). The resulting PMDS hardware has performed flawlessly in all flight testing to
date.
Engine
Sensors
FADECController
Sensors
I ll
I t, I
III
Serial Ilnt_face I
II I
tI t, I
IIIIII
LongTerm
Memory
Central
DiagnosticProcessor
and Memory
Input/OutputController
andYagr_
Conditioning
'//'///_
"/////_
'/.////,_
MaintenanceOutput
Port
DiagnosticOutput
_:i_:_ DisplaySerialInterface
PMDS
Figure 4-5. PMDS Hardware Architecture
PMDS Software Design-The present PMDS software was implemented under the AGATE program and
consists of the following key elements:
• Executive program with supporting modules
• Communications to a PC for data transfer
• Communications to the FADEC for sensor data collection
• Data conversion and storage
• Built-in-test and power-up sequencing
NASA/CR--2002-211485 49
The present PMDS software resides in two places: the run-time portion that runs on the PMDS
hardware, and a retrieval portion that runs on a PC and receives saved data from the PMDS at PMDS
startup.
The PMDS software runs on "bare metal," i.e., without an operating system, as a single-thread
application. At power-on, it initializes the hardware, conducts startup tests, and then sends any
previously stored data out the maintenance port. A PC connected to the maintenance port and ready to
receive the data will have the data sent over in a "raw" form, but be able to convert this data to a moreusable form.
Figure 4-6. PMDS Hardware
After startup, the universal asynchronous receiver/transmitter (UART) for the port connected to the
FADEC is assigned an interrupt handler and enabled, and the main thread goes into an endless loop.
FADEC data is sent in records. As each byte of FADEC data arrives over the port, the interrupt handler
builds the record that is being sent. When a record is complete, the main loop is given access to the
record. Statistical information is computed for each field in the record and saved for later computations.
With each record, a counter is incremented; when the counter reaches a threshold, the statistical
information is written to the EEPROM for long-term storage.
The stored data is sent over the maintenance port whenever the PMDS powers up. To download the
data, a PC must be attached via serial cable to the PMDS and the PC-side software must be running,
waiting for the data to be transmitted. The data is sent in "raw" format; the PC software converts it to ahuman-readable format.
NASA/CR--2002-211485 50
4.2.2 Future Development
The PMDS hardware was originally designed several years ago under conditions that are markedly
different than today. At that time, the hardware price goal drove designers to go with a dual-processor
design using relatively low-power processors and forego expensive dual-port RAM for communication
between the processors in favor of a more complicated approach with cheaper memory. These
approaches and the lack of commercial alternatives led to a platform that would be difficult to upgrade
in the future as processor capabilities and memory capacity increase while prices decrease.
In the time since those design decisions were made, the state of the art for small, embedded processors
and cards has changed substantially. In addition to the costs of processors and memory going down and
the power and capacity of processors and memory going up (Moore's Law), there are many more off-
the-shelf components available. Processors substantially more powerful than the present PMDS's
Motorola ColdFire 5206 are available, such as the Intel Pentium (II/III/IV) series, the PowerPC series,
and MIPS and ARM. Many different single-board computers are commercially available in the PC-104
format (the same card format used in the PMDS). Also interesting is the increasing number of system-
on-a-chip systems, which integrate a higher degree of functionality on a single chip, replacing functions
that would otherwise be found on the system board. All of these developments can allow a more
capable system for less investment. More important, they can provide the same processing capability in
one processor as two ColdFire processors. Not only is this a simpler hardware design, but it also
simplifies the software. One factor, however, must be kept in mind: candidate processors and boards
must be chosen to meet applicable environmental requirements for temperature, humidity, and vibration.
Faster, more common processors with more memory will allow more flexibility with regard to software,
both in the size of the applications and in their sophistication. The PMDS used a single-threaded
application with interrupt service routines that receive data from the FADEC. This approach was
dictated by the limited memory on the PMDS and the desire to keep things simple. Adding more
complex algorithms to the PMDS may require a more capable infrastructure, such as an operating
system (OS) might provide. Options range from open-source OSs such as eCOS
(http://sources.redhat.com/ecos/) and Linux to commercial OSs such as Windows CE, VxWorks, and
LynxOS. By using these OSs, standardized programming interfaces and off-the-shelf tools can be used.
Advancements in digital avionics will also benefit the PMDS concept. It was noted in Section 4.1 that
the PMDS would benefit greatly from additional inputs outside the engine, such as aircraft speed,
aileron, elevator, rudder, and flap settings, propeller pitch, etc. These other sources of data (outside the
FADEC) could provide information to the PMDS over a digital avionics bus in the aircraft. This data-
sharing capability will enable the use of more sophisticated models (through more sensory information)
and will help to minimize the installed cost of the PMDS. The ability for different systems on board the
aircraft to share digital data is a key enabler for the development and commercialization of the PMDS
concept.
The future development steps described above are shown in Figure 4-7. At this time, the scope of the
work, the timing, the source of development funding, and the makeup of the development team are
undefined. These planning issues will be addressed as the general aviation community continues to
dialog about engine diagnostics technology and market needs.
NASA/CR----2002-211485 51
....-_E
EJ7
e_
..__
_
k_
-iIII_l_
II_.--
_--_
ii_
mm
l_Im_
o+
-
<u_-._
e-_.=
¢-,IT">"_
i
T<<zl1-<
Ii
ie#]
Ii
eI
NA
SA
/CR
--2002-21148552
4.3 System Development
During the AGATE program, Honeywell developed a prototype system design for the PMDS concept.
This system design will require additional development to bring it to a commercializable form. The
following subsections describe the key system development areas ahead.
4.3.1 Background
The current PMDS technology embodies the early set of system requirements defined during the
AGATE program. We expect these system requirements to evolve as the development continues. Some
of the key system requirements that will guide future PMDS development are
• Applications
• Certification issues
• Pilot interface
• Ground technician interface
These topics are briefly described below.
Applications-The PMDS flight testing performed to date has been done using the Aurora Flight
Sciences Chiron aircraft with the Teledyne IO-360ES engine. Thus, all of our results are specific to that
engine. Future development must, of course, address other engine models and other engine
manufacturers. As market studies are made, a target market of key engine applications will be identified.
These applications will be the focus of the next steps in the system development process
Certif'wation Issues-The PMDS is intended strictly as a monitoring system and diagnostic-aiding
system. The PMDS shall not provide test inputs to the propulsion system in order to perform. Rather, it
will only monitor the propulsion system. Its outputs in flight will be limited to engine failure and
warning indications. All decisions as to required pilot actions are strictly the purview of the pilot. All
maintenance decisions are strictly the purview of the ground maintenance technician. The PMDS is not
intended to replace these judgments. In view of these considerations, certification of the PMDS need
only be to nonessential levels. The communication interface with the FADEC (and any other avionic
equipment for collecting input data) will be one-way only. In this sense, the PMDS is isolated and
cannot interfere with the operation of other critical systems on the aircraft. More definition of system
requirements with respect to certification will be made as the development continues.
Pilot Interface-The PMDS shall provide two elements to the pilot interface: an engine failure indication
and an engine warning indication. An engine failure indication means that the PMDS has detected an
engine failure that may immediately affect the engine's power output or cause immediate harm to the
engine. An engine warning indication means that the PMDS has detected an impending failure and the
pilot should initiate a maintenance action before the next flight. More definition of pilot interface
requirements will be made as the development continues.
Ground Technician Interface-During ground maintenance, the PMDS shall have a data interface for
ground maintenance technicians, allowing them to interrogate the PMDS and determine fault and
impending fault indications. This capability will enable technicians to download the diagnostic
information that was used by the PMDS to determine the fault or warning indication. This shall include
a fault history and information on the signals that caused the indications to be made. This information is
NASA/CR--2002-211485 53
limited to fault andearlydetectioninformationandis not intendedto performasaflight datarecorder.Theinterfaceshallpresenttimesof failureandearlydetections,valuesof therelevantsignals,detailsofwhichtestscausedtheindications,andanindicationof thepossiblefault modesthatled to theindication,all in aform thatgroundmaintenancepersonnelcanreadilyuse.
4.3.2 Future Development
The above system requirements will be incorporated into future development work. Honeywell
envisions that the next stage of development could be accomplished using two approaches: testing with
a pending failure and fleet testing.
Testing with a Pending Failure-This type of testing would be accomplished as either:
Limited flight tests, for failures that are not safety risks, such as failures of noncritical sensors or
equipment, and/or
Ground tests, for failures that pose a safety risk. This would also provide a means to achieve
greater breadth and depth of testing.
Fleet Testing-This type of testing would consist of long-term flight testing with a fleet of aircraft to
diagnose a specific class of failures (i.e., those that can be detected and diagnosed using data from
EGTs, CHTs, and/or other recorded PMDS parameters). This type of testing will enable the
development team to examine a wide range and long duration of actual operating conditions.
These two approaches could be used independently or in parallel, depending on the makeup of the
development team. Future development steps using these approaches are shown in Figure 4-8. At this
time, the scope of the work, the timing, the source of development funding, and the makeup of the
development team are undefined. These planning issues will be addressed as the general aviation
community continues to dialog about engine diagnostics technology and market needs.
NASA/CR--2002-211485 54
E
II
e-
1
•=_
_
i_
.,..._,"
,'_'°'!_
Ii_-,o
_-_
I'_,_
_'"
Ii_
Im
••
I
Eo_EIIII
rt
0
ff
Jf
ff
J
"_:..,,,____
_
I
$
E
fml_
....
Iu_o
n,_I_
I
i_I_I_I_
__
"=o
!g_......._.-
<<z_
<
Eo_
NA
SA/C
R--2002-211485
55
Section 5. New Technology
Per the requirements of the NASA contract, this section is intended to "identify all nonpatentable
discoveries such as improvements, innovations, and computer codes; and all patentable
inventions, whether developed or discovered during performance of this contract. Possible
secondary applications of reported new technology should also be included in this section."
In performing the technical work on this program, the Honeywell team did not create any new
technology. As described throughout this report, our work on this program was primarily an
extension of work that was begun on the AGATE program. This program's primary results are
the completion of AGATE flight testing and the demonstration of the viability of PMDS
technology. As such, there were no discoveries that fit any of the descriptions in the paragraph
above. (During the AGATE program, Honeywell did prepare an invention disclosure covering
various facets of the PMDS concept. Honeywell is currently pursuing patent protection for that
intellectual property.)
NASA/CR--2002-211485 56
Section 6. Conclusions and Recommendations
This NASA program has taken the results of Honeywell's AGATE work and completed the
initial evaluation of the capability of the prototype PMDS hardware. This work provided the
following key results:
It demonstrated the ability of the PMDS to detect a class of selected sensor hardware
failures. Disconnected sensors were clearly detectable using a static correlation model that
compared sensor output values with expected values. Intermittent sensor faults were also
clearly detectable using a static correlation model.
It demonstrated the ability of the PMDS hardware to successfully model the engine for the
purpose of engine diagnosis. The two flight tests performed on this program demonstrated
that dynamic models of the engine/aircraft can be produced using relatively simple,
inexpensive instrumentation such as would be found in a commercializable on-board
engine diagnostic system. Not surprisingly, nonlinear dynamic models performed better
than linear dynamic models for the same number of inputs and states. Also as expected,
the greater the number of test data points, the better the quality of the resulting model. (A
full-scale development project would involve many sets of flight test data, thereby
resulting in improved dynamic models.) Dynamic models could be used to detect faults
internal to the engine as well as sensor faults.
Future development work for an engine monitoring and diagnostic system should employ the
following elements:
• Engine/aircraft modeling should combine first-principles and empirical approaches. This
strategy offers the advantage of using fewer parameters as required by first-principles
models, while using empirical methods to calibrate unknown parameter values as needed.
• The monitoring and diagnostic system should employ additional inputs outside the engine,
such as aircraft speed, aileron, elevator, rudder, and flap settings, propeller pitch, etc. This
strategy will result in an improved dynamic model to be used for fault detection.
• A prioritized list of engine faults is needed to guide the diagnostic development work.
Ideally, this technical information should come from an engine manufacturer. Data from
more than one engine manufacturer would be even more helpful in advancing the
technology.
• The monitoring and diagnostic system should be able to gather input data from the
FADEC and other systems in the aircraft over a digital avionics bus. This data-sharing
capability will enable the use of more sophisticated models (through more sensory
information) and will help to minimize the installed cost of the PMDS.
NASA/CR--2002-211485 57
Appendix
The Appendix contains the following items:
• List of PMDS sensors
• Flight 41 documentation (consisting of Flight Test Plan 8F34, and Flight Test Engineer's
Report 8F35)
• Flight 44 documentation (consisting of Flight Test Plan 8F35, and Flight Test Engineer's
Report 8F36)
NASA/CR--2002-211485 59
PMDS Sensor Inputs
Ch. Data Ch.
1 labels
2 Altitude intercept 38
3 Altitude slope 39
4 Altitude min 40
5 Altitude max 41
6 Kollsman intercept 42
7 Kollsman slope 43
,8 Kollsman min 44
9 Kollsman max 45
10 FuelPress intercept 46
11 FuelPress slope 47
12 FuelPress min 48
13 FuelPress max 49
14 OilPress intercept 50
15 OilPress slope 5116 OilPress min 52
17 OilPress max 53
18 FuelTemp intercept 54
19 FuelTemp slope 55
20 FuelTemp min 56
21 FuelTemp max 57
22 OilTemp intercept 58
23 OilTemp slope 59
24 OilTemp min 60
25 OilTemp max 61
26 BaylTemp intercept 62
27 Bay1Temp slope 63
28 Bay1Temp min 64
29 BaylTemp max 65
30 Bay2Temp intercept 66
31 Bay2Temp slope 67
32 Bay2Temp min 68
33 Bay2Temp max 69
34 EGT1 intercept 70
35 EGT1 slope 7136 EGT1 min 72
37 EGT1 max 73
Data
EGT2 intercept
EGT2 slope
EGT2 min
EGT2 max
EGT3 intercept
EGT3 slopeEGT3 min
EGT3 max
EGT4 intercept
EGT4 slopeEGT4 min
EGT4 max
EGT5 intercept
EGT5 slope
EGT5 min
EGT5 max
EGT6 intercept
EGT6 slopeEGT6 min
EGT6 max
CHT1 intercept
CHT1 slopeCHT1 rain
CHT1 max
CHT2 intercept
CHT2 slopeCHT2 min
CHT2 max
CHT3 intercept
CHT3 slopeCHT3 min
CHT3 max
CHT4 intercept
CHT4 slope
CHT4 min
CHT4 max
Ch. Data
74 CHT5 intercept
75 CHT5 slope
76 CHT5 min
77 CHT5 max
78 CHT6 intercept
79 CHT6 slope80 CHT6 min
81 CHT6 max
82 OutsideAirTemp intercept
83 OutsideAirTemp slope
84 OutsideAirTemp min
85 OutsideAirTemp max
86 ManifoldAirPr intercept
87 ManifoldAirPr slope88 ManifoldAirPr min
89 ManifoldAirPr max
90 AirChargeT intercept
91 AirChargeT slope
92 AirChargeT min
93 AirChargeT max
94 EngineRPM intercept
95 EngineRPM slope
!96 EngineRPM min
97 EngineRPM max
98 PressureAItitude* intercept
99 PressureAItitude* slope100 PressureAItitude* min
101 PressureAItitude* max
* PressureAItitude = mbar/10
Flight Test Data Collected Manually
Ch. Data
102 SLPC
103 Cowl flaps
NASA/CR--2002-211485 60
SLPC CHIRON AU-008
[AGATE Integrated Flight Test Plan AGATE 8F34
N427AU
Operations Number: 8F34Date: 04/02/01
Estimated Flight Time: 2.5 Hours
Range Time: 5:00 to 7:00Proposed Engine Start: 4:45
Flight Crew: Pilot in Command:
Flight Test Engineer:
Bill Weber
Ken Zugel
Test Objectives: flight with PMDS, SLPC-FADEC to 4,000 ft; 8,000 fl and 12,000 ft. Several single-lever powersettings. Sensors all nominal.
Test Event Summary:Taxi.
Takeoff in SLPC mode.Set Power = 100%.
Climb to 4,000 feetSet Power = 40% cruise.
Set Power = 60% cruise.Set Power = 80% cruise.
Flight - see Test Card 8F33
Reach steady state.
Reach steady state.Reach steady state.
Set Power = 100% climb to 8,000 ft.Set Power = 40% cruise. Reach steady state.
Set Power = 60% cruise. Reach steady state.Set Power = 80% cruise. Reach steady state.
Set Power = 100% climb to 12,000 ft.
Set Power = 40% cruise. Reach steady state.
Set Power = 60% cruise. Reach steady state.Set Power = 80% cruise. Reach steady state.
Pull power back to 40%, descend to 4,000 ft.Set Power = 75% for enroute climb to 8,000 ft.
Level off at 8,000 fl at Power = 75%.Continue enroute climb to 12,000 ft.
Descend to airport.
Conduct approach.Landing.Debrief.
Call Signs: Aircraft Call Sign:
Frequencies: Manassas TowerAdditional Information:WX:
Support Equipment: N/A
Aircraft Configuration:
N427AU
SLPC - FADEC active, FTC inactive.
NASA/CR--2002-211485 61
FCU Software:
ECU Software:Fuel:
Operating Limits:
Special Precautions:
MAF sensor OPERATIVE, Tbay sensor OPERATIVE
(SLPC system). Honeywell PMDS active.
CH8.10
Full Main Tanks
Conditions: wet/dry, daylight, VFRMax. altitude: 12,500 feet
Secure Ramp Area at Aurora Hangar.
r.. ]
SLPC CHIRON AU-008 IN427AUlIAGATE Integrated Flight Test Plan AGATE 8F34 I
SIGNATURES FOR FLIGHT APPROVAL:
Quality Assurance:
Director of Engineering:
Aurora FRR Board Chairman:
Director of Flight Ops:
Project:
Date:
Date:
Date:
Date:
Date:
NASA/CR--2002-211485 62
IPMDS Flight Test 8F34 TEST CARD SLPC CHIRON AU-008 I
IN427AUI FLIGHT DATE:
Event: Test Description:
1. CHECKLIST PROCEDURE AND START FRONT ENGINE IN SLPC MODE.
2. START AND CHECK SLPC DATA LOG AS PER PROCEDURE.
3. START PMDS DATA LOG AS PER PROCEDURE.
4. WARMUP.
5. TAKEOFF. POWER = 100%.
6. CLIMB TO 4,000 FEET.
7. ADJUST COWL FLAPS AS NECESSARY, MAKE NOTES.
8. SET 40% POWER. REACH STEADY STATE.
9. SET 60% POWER. STEADY STATE.
10. SET 80% POWER. STEADY STATE.
11. SET 100% POWER, CLIMB TO 8,000 ft.
04/12/01
12. SET 40% POWER. REACH STEADY STATE.
13. SET 60% POWER. STEADY STATE.
14. SET 80% POWER. STEADY STATE.
15. SET 100% POWER, CLIMB TO 12,000 ft.
16. SET 40% POWER. REACH STEADY STATE.
17. SET 60% POWER. STEADY STATE.
18. SET 80% POWER. STEADY STATE.
19. SET 40% POWER OR MIN., DESCEND TO 4,000 ft.
20. SET 75% POWER, LEVEL FLIGHT.
21. INITIATE ENROUTE CLIMB TO 8,000 ft.
22. LEVEL OFF, CONSTANT 75% POWER.
23. INITIATE ENROUTE CLIMB TO 12,000 ft.
24. DESCEND TO AIRPORT.
25. APPROACH.
26. INITIATE LANDING PATTERN, LAND.
27. SWITCH OFF SLPC DATA LOG PRIOR TO FADEC SHUTDOWN.
28. SHUTDOWN.
NASA/CR--2002-211485 63
Flight Test Engineer Report
SLPC/PMDS Flight Test 8F35 (ref: Flight Test Plan "AGATE 8F34" 04/02/01)25 April 2001Bill Weber- Pilot
Ken Zugel - Flight Test Engineer
Narrative:
This was a repeat of the previous flight after replacing the propeller governor.
Bill and I manned the aircraft at 1215 and started the engines at 1230. We completed the checklist and poweredup SLPC/PMDS system. The Manassas weather was clear, 10 miles visibility, winds were from 360 at 10 knots,and the altimeter setting was 30.28"hg. Following the runup and system checks, we took off at 1245 in SLPC
mode and departed toward the southwest. We experienced problems with the landing gear system during the
climbout. The gear doors did not close after the gear retracted. The gear warning horn activated during the first
gear up cycle and the gear warning circuit breaker popped. Bill recycled the gear and it retracted normally andthen he reset the breaker. Once the gear problems were resolved Bill continued the departure and configured the
aircraft for cruise climb between 20 and 65% power. He made slow climb to 4000 ft due to airspace and air trafficcontrol limitations. The test area was roughly between Culpeper, Charlottesville, and New Market due to airtraffic concerns.
Upon reaching 4000 ft we proceeded with the cruise test points. We reversed the order of the test points from the
previous flight. The altitude varied +/- 100 ft. We experienced light to moderate turbulence. The 80% power testpoint began at 1300 and was nominal. The cowl flaps were closed once the engine temperatures stabilized. The
60% test point began at 1306 and was nominal as well. The 40% power setting began at 1313 and was nominal aswell.
The climb to 8000 ft was performed at 90% power vice the 95% on the test card due to engine/propeller
limitations. The cowl flaps were opened for the climb. We leveled off at 8000 ft and started the 80% power pointat 1325. The cowl flaps were closed after the temperatures stabilized. The 60% power test point was started at
1331 and was nominal. The 40% power setting was started at 1338. We did not experience any of the propellerRPM oscillations that we did on the previous flight.
The climb to 12000 ft was performed at 100% power and the climb rate varied between 300 and 500 feet perminute. Once we reached altitude and the engine temperature stabilized, the cowl flaps were closed and we started
the 80% power test point at 1353. The 60% power test point was started at 1359 and was nominal. The 40%power test point was started at 1405 and was also nominal.
From the last test point Bill initiated the descent to 8000 ft at 40% power. We leveled off at 8000 ft for the 75%
power test point and it was successfully completed at 1424. We continued the descent to 4000 ft at 40% power.Bill leveled off at 6000 ft to clear terrain and then continued the descent to 4000 ft once we were east of the
mountains. He continued the descent at 40% power until we entered the pattern. We returned to Manassas andconfigured for landing. Bill made a normal landing at 1451. We taxied in and shutdown the aircraft at 1457.
Conclusions:
The SLPC system worked better than the last flight with the new propeller governor.
Recommendations:
Proceed with the next SLPC/PMDS test flight Friday at 1200.
NASA/CR--2002-211485 64
SLPC CHIRON AU-008 N427AU[ AGATE Integrated Flight Test Plan AGATE 8F35
Proposed Engine Start: 1200Date: 04/25/01
Range Time: 1230 to 1430Estimated Flight Time: 2.5 Hours
Flight Crew: Pilot in Command:
Flight Test Engineer:
Bill Weber
Ken Zugel
Test Objectives: PMDS, SLPC-FADEC to 4,000 fl: 8,000 fl and 12,000 ft. Several single-lever power settings.
Test Card:
Power up and start data logFADEC Start and Taxi.
Runup and test SLPCTakeoff in SLPC mode.Set Power = 100%
Reduce to 95% leaving pattern
Climb to 4,000 feet
Set Power = 80% cruise. Reach steady state. Time for 5 minutesSet Power = 60% cruise. Reach steady state. ""
Set Power = 40% cruise. Reach steady state. ""
Set PowerSet Power
Set PowerSet Power
= 95% climb to 8,000 ft.
= 80% cruise. Reach steady state. "
= 60% cruise. Reach steady state. '"'= 40% cruise. Reach steady state. ""
Set PowerSet Power
Set PowerSet Power
= 100% climb to 12,000 ft.
= 80% cruise. Reach steady state. '"'
= 60% cruise. Reach steady state. '"'= 40% cruise. Reach steady state. '"'
Descend to 8,000 ft. at 40%,
Level off at 8,000 ft at 75%. Reach steady state. Time for 5 minutes.Reduce power to 40%, continue descent to pattern altitudeDescend to airportConduct approach and Go-Around (if needed) in SLPC mode
LandingStop data log and shutdownDebrief.
Call Signs: Aircraft Call Sign: N427AUFrequencies: Manassas Tower: 133.1
Mission at 1400: 123.45
Aircraft Configuration: SLPC - FADEC active, FTC inactive.MAF sensor INOP, Tbay sensor INOP.
(SLPC system). Honeywell PMDS active.FCU Software:ECU Software: CH8.10
Fuel: Full Main Tanks, Full Aux. Tanks
Operating Limits:
Special Precautions:
Conditions: wet/dry, daylight, VFRMax. altitude: 12,500 feet
Secure Ramp Area at Aurora Hangar.
NASA/CR--2002-211485 65
SIGNATURES FOR FLIGHT APPROVAL:
Quality Assurance:
Director of Engineering:
Aurora FRR Board Chairman:
Director of Flight Ops:
Project:
Date:
Date:
Date:
Date:
Date:
NASA/CR--2002-211485 66
IPMDS Flight Test 8F35 TEST CARD SLPC CHIRON AU-008 J
lN427AU FLIGHT DATE: 04/25/01
Event: Test Description:
29. CHECKLIST PROCEDURE AND START FRONT ENGINE IN SLPC MODE.
30. START AND CHECK SLPC DATA LOG AS PER PROCEDURE,
31. START PMDS DATA LOG AS PER PROCEDURE.
32. RUNUP/SWITCH TO FADEC MODE
33. TAKEOFF in SLPC POWER = 100%.
34. CLIMB TO 4,000 FEET.
35, ADJUST COWL FLAPS AS NECESSARY, MAKE NOTES.
36. SET 80% POWER, REACH STEADY STATE.
37. SET 60% POWER. STEADY STATE.
38. SET 40% POWER. STEADY STATE.
39. SET 95% POWER, CLIMB TO 8,000 ft.
40. SET 80% POWER. STEADY STATE.
41. SET 60% POWER. STEADY STATE.
42. SET 40% POWER. STEADY STATE.
43, SET 100% POWER, CLIMB TO 12,000 ft.
44. SET 80% POWER. STEADY STATE.
45. SET 60% POWER. STEADY STATE.
46. SET 40% POWER. STEADY STATE.
47. SET 40% POWER, DESCEND TO 8,000 ft,
48. SET 75% POWER, LEVEL FLIGHT.
49. LEVEL OFF, CONSTANT 75% POWER.
50. SET 40% POWER, DESCEND TO 4000 ft.
51. APPROACH AND GO-AROUND IN SLPC MODE,
52. INITIATE LANDING PATIERN, LAND,
53. SWITCH OFF SLPC DATA LOG PRIOR TO SHUTDOWN.
54.SHUTDOWN.
NASA/CR--2002-211485 67
Flight Test Engineer Report
SLPC/PMDS Flight Test 8F36 (ref: Flight Test Plan "AGATE 8F35" 04/25/01)27 April 2001Bill Weber- Pilot
Ken Zugel - Flight Test Engineer
Narrative:
This flight was a repeat of the previous flight with the T-bay temperature and Mass Airflow sensors disconnected.
The Manassas weather was 9000ft overcast, 10 miles visibility, winds were from 300 at 7 knots, the temperaturewas 21 degrees Celsius, and the altimeter setting was 30.08"Hg.
Bill and I manned the aircraft at 1345 and started the engines at 1401. We completed the checklist and poweredup SLPC/PMDS system. Following the runup and system checks, we took off at 1416 in SLPC mode and
departed toward the southwest. We didn't have any problems with the landing gear system during the climbout.Bill pulled the gear warning circuit breaker to prevent a false alarm and keep the gear down bulb from burning out
as had occurred on the previous flight. Bill continued the departure and configured the aircraft for cruise climbbetween 75% power. He made a slow climb to 4000 ft due to airspace and air traffic control limitations. The testarea was roughly between Culpeper, Charlottesville, and Fredericksburg due to ceilings.
Upon reaching 4000 fl we proceeded with the cruise test points. The altitude varied +/- 200 ft. We experienced
continuous light to moderate turbulence. The cowl flaps were kept open due to the higher outside air
temperatures. The 80% power test point began at 1426 and was nominal. The 60% test point began at 1432 andwas nominal. The 40% power setting began at 1438 and was nominal as well.
The climb to 8000 fl was performed at 90% power due to engine/propeller limitations and resulted in a 500 fpmrate of climb. The cowl flaps were open for the climb and the entire 8000 ft test block due to the higher OAT. We
leveled off at 8000 ft and started the 80% power point at 1451. The manifold pressure (MAP) stabilized at 22" Hgand the RPM was at 2450. The 60% power test point was started at 1457 and was nominal. The MAP was 21" andthe RPM was 2350. The 40% power setting was started at 1503 and was nominal as well. The MAP was 20.75"and the RPM stabilized at 2325.
The climb to 12000 fl was performed at 100% power and the climb rate varied between 300 and 500 feet perminute. The indicated power command was only 96%, but appears to have been caused by a change in the Single
Power Lever position sensor. We encountered wake turbulence from a B-727 descending through our altitude,which triggered several warnings and caused two momentary upsets. Once we reached altitude and the engine
temperature stabilized, the cowl flaps were closed and we started the 80% power test point at 1520. The MAP was19" and the RPM stabilized at 2525. The 60% power test point was started at 1526 and was nominal. The 40%
power test point was started at 1533 and was also nominal.
From the last test point Bill initiated the descent to 8000 ft at 40% power. We leveled off at 8000 fl for the 75%
power test point at 1545. Once it was completed we continued the descent at 40% power until we entered thepattern. The rate of descent was between 500 and 1000 fpm due to airspace and traffic limitations. The weather at
Manassas had changed: the altimeter setting was 30.00"Hg, the skies were clear, and the temperature was 25
degrees Celsius. We returned to Manassas and made a normal landing at 1601. We taxied in and shutdown theaircraft at 1607.
Conclusions:
The SLPC system functioned nominally.
The Single Power Lever position sensor should be recalibrated if additional SLPC flights are planned.
Recommendations:
The erroneous gear warnings and malfunctions seem to be switch related and may need to be fixed prior to the
next flight test.Proceed with the next SLPC/PMDS test flight if requested and the schedule will allow it.
NASA/CR--2002-211485 68
REPORT DOCUMENTATION PAGE FormApprovedOMB No. 0704-0188
Public reportingburdenfor this collectionof information is estimated to average 1 hour per response, includingthe time for reviewinginstructions,searchingexistingdata sources,gathering and maintainingthe data needed, and completingand reviewingthe collectionof information. Send comments regardingthis burden estimate or any other aspect of thiscollection of information,includingsuggestionsfor reducingthis burden, to WashingtonHeadquarters Services,Directoratefor InformationOperations and Reports, 1215 Jefferson
! Davis Highway, Suite 1204, Arlington,VA 22202-4302, and to the Office of Management and Budget,Papenvork ReductionProject(0704-0188), Washington, DC 20503.I
I 1 AGENCY USE ONLY (Leave b/ank) 2. REPORT DATE 3. REPORT TYPE AND DATES COVERED
March 2002 Final Contractor Report5. FUNDING NUMBERS4. TITLE AND SUBTITLE
Flight Test of Propulsion Monitoring and Diagnostic System
6. AUTHOR(S)
Steve Gabel and Mike Elgersma
7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES)
Honeywell Laboratories
3660 Technology Drive
Minneapolis, Minnesota 55418
9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES)
National Aeronautics and Space Administration
Washington, DC 20546-0001
WU-728-30-20-00
NAS3-01105
8. PERFORMING ORGANIZATIONREPORT NUMBER
E-13253
10. SPONSORING/MONITORINGAGENCY REPORT NUMBER
NASA CR--2002-211485
11. SUPPLEMENTARY NOTES
Pr_ectManager, Donald L. Simon, VehicleTechnology Directorate, NASA Glenn Research Center, organization
code 0300,216---433-3740.
12a. DISTRIBUTION/AVAILABILITY STATEMENT
Unclassified - Unlimited
Subject Category: 07 Distribution: Nonstandard
Available electronically at httt)://eltrs._rc.nz_sa._ov/GLTRS
This publication is available from the NASA Center for AeroSpace Information, 301--621-0390.
12b. DISTRIBUTION CODE
13. ABSTRACT (Maximum 200 words)
The objective of this program was to perform flight tests of the propulsion monitoring and diagnostic system (PMDS)
technology concept developed by Honeywell under the NASA Advanced General Aviation Transport Experiment
(AGATE) program. The PMDS concept is intended to independently monitor the performance of the engine, providing
continuous status to the pilot along with warnings if necessary as well as making the data available to ground maintenance
personnel via a special interface. These flight tests were intended to demonstrate the ability of the PMDS concept to detect
a class of selected sensor hardware failures, and the ability to successfully model the engine for the purpose of engine
diagnosis.
14. SUBJECT TERMS
Aircraft engines; General aviation aircraft; Systems health monitoring; Flight safety;
Modeling; Diagnostics
17. SECURITY CLASSIFICATION 18. SECURITY CLASSIFICATION 19. SECURITY CLASSIFICATION
OF REPORT OF THIS PAGE OF ABSTRACT
Unclassified Unclassified Unclassified
15. NUMBER OF PAGES
7516. PRICE CODE
20. LIMITATION OF ABSTRACT
NSN 7540-01-280-5500 Standard Form 298 (Rev. 2-89)
Prescribed by ANSi Std. Z39-1B298-102