METHOD AND SYSTEM FOR IDENTIFYING GAS TURBINE ENGINE FAULTS
BACKGROUND OF THE INVENTION
[0001] The invention relates generally to the field of gas
turbine engine modeling. More specifically, the invention
relates to methods and systems that distinguish between turbine
case cooling (TCC) and high pressure turbine (HPT) performance
faults.
[0002] Gas turbine performance diagnostics concerns itself with
tracking changes in engine module performance measures
(typically efficiency and flow parameters) as the engine
deteriorates over time. The primary sources of information
driving this methodology are measurements taking along the
engine's gas path, such as temperatures, pressures, speeds, etc.
These measurements are typically monitored during stable cruise
conditions and stored for the purpose of performing a Module
Performance Analysis (MPA). Because of the inherent limitation
of available measurements in commercial and military
aero-engines, there is a difficulty in differentiating between
various faults in the turbine section of the engine. Two typical
faults that fall in this category are turbine case cooling (TCC)
and high pressure turbine (HPT) performance faults.
[0003] Current MPA methods use steady state cruise data to
perform fault isolation. The parameters that are monitored are
rotational speeds, temperatures, and pressures taken at various
stages along an engine's gas path. When a shift in these
measured quantities is detected, a percent A is computed for
each gas path parameter, capturing the level and direction of
1A

the shift. The resulting vector of measurement parameter As is
used to compute the MPA.
[0004] The calculation is effectively a pattern matching
methodology, wherein the analysis compares the computed percent
A vector to a series of other vectors representing known faults,
and the best match is selected. This type of analysis has many
methodologies and variants known in the art. Unfortunately, the
signatures of HPT and TCC performance faults are nearly
identical in terms of the commonly measured gas path parameters
and are indistinguisnable within the confines of this analysis,
no matter what particular methodology is employed.
[0005] Current methods in performance tracking cannot
differentiate between HPT performance faults and TCC faults.
Although TCC faults are more likely to occur than HPT faults,
the benign nature of a TCC fault (increased fuel consumption
penalty) if improperly diagnosed may result in the engine
remaining on wing with a potential for an in-flight shutdown
(IFSD) or catastrophic event if the underlying cause had been an
HPT problem instead. What is needed is a method and system that
mitigates the risk of a TCC/HPT misdiagnosis.
SUMMARY OF THE INVENTION
[0006] Although there are various methods and systems employing
performance tracking to differentiate between HPT and TCC
performance faults, such methods and systems are not completely
satisfactory. The inventor has discovered that it would be
desirable to have methods and systems that distinguish between
2

turbine case cooling (TCC) and high pressure turbine (HPT)
performance faults.
[0007] One aspect of the invention provides a method for
distinguishing between gas turbine engine case cooling (TCC) and
high pressure turbine (HPT) performance faults. Methods
according to this aspect of the invention preferably start with
acquiring a predetermined number of in-flight gas path data
samples corresponding to a predetermined number of engine
parameters, determining if a percent A shift has occurred in a
gas path engine parameter, if a percent A signature shift has
occurred, determining whether the shift is from an HPT or TCC
performance fault comprising, extracting exhaust gas temperature
shift magnitudes from the predetermined number of engine
parameters, calculating an exhaust gas temperature differential
between in-flight and take-off exhaust gas temperature margin
calculations, wherein the exhaust gas temperature differential
is the difference between the change in exhaust gas temperature
margin and the exhaust gas temperature shift magnitudes,
calculating a TCC event likelihood and a non-TCC event
likelihood, wherein if the TCC event is greater than or equal to
the non-TCC event likelihood, declaring a TCC fault.
[0008] Another aspect of the method includes determining whether
the shift is from an HPT or TCC performance fault. This aspect
further comprises obtaining take-off exhaust gas temperature
margin calculations for the present flight and from a previous
flight, and calculating a change in exhaust gas temperature
margin calculations between the present flight and a previous
flight.
3

[0009] Yet another aspect of the method includes calculating the
likelihood of a TCC or non-TCC event. This aspect further
comprises obtaining a mean and standard deviation for gas
turbine engines not experiencing TCC faults, obtaining a mean
and standard deviation for gas turbine engines experiencing TCC
faults, and calculating an exhaust gas temperature difference
between cruise and cake-off conditions.
[0010] The details of one or more embodiments of the invention
are set forth in the accompanying drawings and the description
below. Other features, objects, and advantages of the invention
will be apparent from the description and drawings, and from the
claims.
BRIEF DESCRIPTION OF THE DRAWINGS
[0011] FIG. 1A is an exemplary plot showing a gas path
measurement parameter percent A trend.
[0012] FIG. 1B is an exemplary plot showing a gas path
measurement parameter percent A trend that experienced a fault
event (AA) .
[0013] FIG. 2 is a block diagram of an exemplary method that
distinguishes between turbine case cooling (TCC) and high
pressure turbine (HPT) performance faults.
[0014] FIG. 3A is an exemplary plot showing an exhaust gas
temperature Gaussian distribution for a non-TCC event.
4

[0015] FIG. 3B is an exemplary plot showing an exhaust gas
temperature Gaussian distribution for a TCC event.
[0016] FIG. 3C is an exemplary plot showing an exhaust gas
temperature differential Gaussian distribution for a TCC event
and a non-TCC event.
[0017] FIG. 3D is an exemplary plot showing an exhaust gas
temperature differential Gaussian distribution region where TCC
and non-TCC event separation is not observable.
5

DETAILED DESCRIPTION
[0018] Embodiments of the invention will be described with
reference to the accompanying drawing figures wherein like
numbers represent like elements throughout. Further, it is to be
understood that the phraseology and terminology used herein is
for the purpose of description and should not be regarded as
limiting. The use of "including," "comprising," or "having" and
variations thereof herein is meant to encompass the items listed
thereafter and equivalents thereof as well as additional items.
The terms "mounted," "connected," and "coupled" are used broadly
and encompass both direct and indirect mounting, connecting, and
coupling. Further, "connected" and "coupled" are not restricted
to physical or mechanical connections or couplings.
[0019] The invention is not limited to any particular software
language described or implied in the figures. A variety of
alternative software languages may be used for implementation of
the invention. Some components and items are illustrated and
described as if they were hardware elements, as is common
practice within the art. However, various components in the
method and system may be implemented in software or hardware.
[002 0] The invention is a modular framework and may be deployed
as software as an application program tangibly embodied on a
program storage device. The application code for execution can
reside on a plurality of different types of computer readable
media known to those skilled in the art.
[0021] Current: MPA methods monitor rotational speeds,
temperatures, and pressures taken at various stages along an
engine's gas path to conduct performance estimation tracking and
6

fault isolation. The data is time-averaged at a stable cruise
flight condition, normalized to standard reference conditions
and compared to a reference baseline model to produce a vector
comprised of percent As. The vector represents a time sequence
of gas path performance history and is generally trended to aid
in detecting health deviations and fault events.
[0022] FIG. 1A. shows a plot of a typical trend for an arbitrary
measurement A parameter such as fuel flow, exhaust gas
temperature, or other engine gas path parameter. The gradual
upward trend may be indicative of engine degradation. Individual
parameter measurement A samples 101 and a superimposed average
trend line 102 are shown. This is in contrast with FIG. 1B that
shows a sudden shift 103 in the parameter measurement Δs that
may indicate some type of temporal fault event. The method of
the invention analyzes these types of perturbations.
[0023] When a sudden shift 103 in one or more measured gas path
parameter As is observed, it usually indicates an underlying
fault. A single fault assumption is a common hypothesis since it
would be improbable that several engine components, or a
multiple component failure, would occur. The underlying fault
may be one of many possible faults, such as a component
performance fault (rapid change in a major component efficiency
or flow due to foreign or domestic object damage), an engine
system fault or failure such as an actuator failure controlling
variable geometry guide vanes within the engine, or a leak, or
failure of one of many engine bleeds used for stability and
off-board service such as aircraft air conditioning, etc.
7

[0024] The invention addresses TCC faults. The TCC system
controls air flow from a compressor stage bleed. The air is
circulated around the outside case of the high pressure turbine
(HPT) in an attempt to cool it. Since the air that is bled from
the compressor is relatively cool with respect to the HPT case,
the bleed air cools the HPT case and allows it to contract. This
reduces the HPT turbine blade tip clearances within the case and
reduces parasitic bypass losses thereby increasing efficiency,
and reducing fuel burn.
[0025] Thus, it is not surprising that the signature of measured
parameter As associated with a TCC failure is like that which
would be observed from other faults in the turbine such as blade
erosion, blade: damage, etc., that also affects HPT efficiency. A
TCC fault is a more benign fault than a damaged HPT in the sense
that it may only affect fuel efficiency and shorten the life
expectancy of the turbine in the long term. HPT damage may lead
to an in-flight shutdown of the engine and potential
catastrophic engine failure if left undiagnosed and unattended.
It is critical to be able to distinguish between these two types
of faults.
[0026] Since "he signatures of the measured parameter As are
very similar, it is usually not possible to differentiate
between them using current gas path analysis methods. The
invention uses independent information typically available in
aero-engine monitoring systems to effect diagnosis.
[0027] Shown in FIG. 2 is one embodiment of the invention. A
measurement percent A parameter vector at discrete time k is
8

monitored (step 202). The vector consists of percent As derived
using a nominal baseline reference, in typically monitored gas
path parameters such as rotational speeds, temperatures, and
pressures, etc. Comparing this data sample to previously
monitored data samples at times k -1, k - 2, k -3,..., k - n , a
determination is made whether or not a sudden shift has occurred
in any one of the parameters (step 2 03) as in FIG. IB. Any
reliable methodology may be employed to detect sudden, temporal,
shifts in the data. If no sudden shift is detected, the method
returns to normal gas path analysis calculations for trending
the performance health of the engine.
[0028] If a shift is detected (step 203), then the magnitude of
the shift is determined (step 2 04). For example, referring to
FIG. IB, this would be the AA 104 between the present data point
at time k and the previous average level. A AA is calculated for
every measurement parameter in a measurement A vector. If no
shifts occur, that parameter would be zero. This produces a
vector of measurement AAs at time k , denoted as ΔΔk . There are
numerous isolation methods known that may be employed to
determine the single fault yielding the closest match to the
observed AA shift ΔΔk . Isolation methods using Kalman filters,
weighted least squares, probabilistic neural networks, trained
feed-forward artificial neural networks, and others may be used
to provide isolation. This yields the fault type most likely
responsible for the shift (step 204).
[0029] As described above, if the actual underlying fault
responsible for the shift is either a TCC or an HPT performance
fault, there will be ambiguity in the isolation determination
9

(step 2 04). The method determines whether the engine has
experienced an HPT fault, a TCC fault, or that it may have been
either fault type (seep 2 05; . If that is the case, the EGTΔΔ
component from the ΔΔk. vector (step 206) is examined. Since EGT
is a flight critical parameter, its monitored value is always
available. The EGTAA component is denoted by EGTΔΔk . Since the
measurement parameter A vectors are in percent, the computed AA
vector will be in percent and hence EGTΔΔk will be in percent.
This value may be converted co degrees, AA °R or AA °K, (step
2 0 7) by
[0030]
[0031] where EGTΔΔkdeg is the EGTAA in degrees and EGTbasek is the
nominal baseline reference level for EGT in absolute degrees (°R
or °K) from which the kth measurement A for EGT was calculated.
[0032] Additional, independent information may be used to
differentiate between a TCC and HPT fault. This information may
be in the form of an EGT margin calculation that may be
performed during aircraft take-off, either on-board or
off-board, and is a standard procedure for commercial and
military applications.
[0033] EGT margin is the delta between the observed (takeoff)
EGT and a pre-defined threshold typically specified by the
engine manufacturer. It. is calculated by subtracting observed
values from threshold values during take-off. A positive margin
10

210). The calculation may require several calculations based on
empirically derived statistical observations of how EGT and EGT
margin distributions react to a TCC event and a non-TCC event.
[0040] A statistical sample of engines that have not experienced
a TCC failure may be analyzed, and the mean and standard
deviation for "heir measurement EGTAA taken at cruise may be
calculated. The EGTAA is from a vector where one of the
parameters experienced a shift. The mean and standard deviation
for the change in EGT margin, i.e., ΔEGTMargin , may be calculated
from the take-off data for this same sample of engines.
[0041] Experience has indicated that the distributions for the
cruise and take-off data for non-TCC events are very similar.
This is shown in FIG. 3A where the distributions for EGTAA 301
and ΔEGTMargin 3 02 assume a normally distributed population. In
contrast, FIG. 3B shows similar statistics from engine samples
having experienced TCC events. In this case, the change in EGT
margin, ΔEGTMargin, and the cruise EGTΔΔ shifts have significantly
different mean values. The invention exploits these properties.
[0042] Since EGTDiff is the sum of EGTAA and ΔEGTMargin , its
distribution is centered at zero degrees for a non-TCC event 303
and offset at a positive mean for a TCC event 3 04. These are
shown in FIG. 3B and the distributions are a direct consequence
of the distributions obtained empirically for EGTΔΔ 304 and
ΔEGTMargin 303.
12

[0043] From empirical observation, the mean and standard
deviation for EGTDiff , for a non-TCC event, is denoted by αNon_TCC
and βNon-TCC respectively, and the mean and standard deviation for
EGTDiff , for a TCC event, denoted by αTCC and βTCC , respectively.
[0044] The likelihood that the observed shift is due to a TCC
event is denoted by EventTCC , and is calculated (step 210) by
[0045]
[0046] where e is 2.713.
[0047] The likelihood that the observed shift is due to a
non-TCC event is denoted by EventNon-TCC , and is calculated (step
210) by
[0048]
[0049] It is possible that these two magnitudes may be close in
value, i.e. that their absolute difference could be less than
some predefined Threshold ,
[0050]
13

[0051] This occurs when the calculated EGT differential,
EGTDiff , is in a region overlapping both tails of the TCC vs.
non-TCC distribution. This is shown in FIG. 3D 307. In this
instance, the TCC and non-TCC event hypothesis is not observable
(step 211) and an alert; is made to a user that the ambiguity
could not be resolved (step 212).
[0052] If the absolute difference is greater than the predefined
threshold, a check is performed to see if the likelihood of the
TCC event hypothesis. EventTCC is greater than the likelihood of
a non-TCC event hypothesis, EventNon-TCC , (step 213) . If it is, the
event is a TCC fault (step 214), otherwise it is an HPT
performance fault (step 215).
[0053] One or more embodiments of the present invention have
been described. Nevertheless, it will be understood that various
modifications may be made without departing from the spirit and
scope of the invention. Accordingly, other embodiments are
within the scope of the following claims.
14

CLAIMS
What is claimed is:
1. A method for distinguishing between gas turbine engine case
cooling (TCC) and high pressure turbine (HPT) performance faults
comprising:
acquiring in-flight gas path data samples corresponding to
a predetermined number or engine parameter percent Δs from
nominal;
determining if a percent Δ shift has occurred in a gas path
engine parameter percent A ;
if a percent Δ signature shift has occurred, determining
whether the shift ΔΔ is from an HPT or TCC performance fault
comprising:
extracting exhaust gas temperature shift magnitudes
from the predetermined number of engine parameters;
calculating an exhaust gas temperature differential
between in-flight and take-off exhaust gas temperature margin
calculations, wherein the exhaust gas temperature differential
is the difference between the change in exhaust gas temperature
margin and the exhaust gas temperature shift magnitudes; and
calculating a TCC event likelihood and a non-TCC event
likelihood, wherein if the TCC event is greater than or equal to
the non-TCC event likelihood, declaring a TCC fault.
2. The method according to claim 1 wherein determining whether
the shift is from an HPT or TCC performance fault further
comprises:
obtaining take-off exhaust gas temperature margin
calculations for the present flight and from a previous flight;
and
calculating a change in exhaust gas temperature margin
calculations between the present flight and a previous flight.
15

3. The method according to claim 2 wherein calculating the
likelihood of a TCC or non-TCC event further comprises:
obtaining a mean and standard deviation for gas turbine
engines not experiencing TCC faults;
obtaining a mean and standard deviation for gas turbine
engines experiencing TCC faults; and
calculating an exhaust gas temperature difference between
cruise and take-off conditions.
4 . The method according to claim 3 further comprising trending
the predetermined number of gas path engine parameters while
in-flight.
5. The method according to claim 4 further comprising
converting the exhaust gas temperature shift magnitudes from a
percent measurement to degrees.
6. The method according to claim 5 wherein EGT margin is a
delta between the observed (takeoff) EGT and a pre-defined
threshold.
7. The method according to claim 6 wherein a positive margin
indicates that the engine is within the pre-defined threshold
value.
8. The method according to claim 7 wherein a zero or negative
margin indicates that the engine requires maintenance.
9. The method according to claim 8 wherein TCC is not active
during take-off.
16

17
10. The method according to claim 9 wherein ΔEGT margin is
approximately the same, but having opposite polarity, as the
ΔEGT shift when TCC is not an underlying problem.
11. The method according to claim 10 further comprising
calculating confidence metrics for TCC event likelihoods and
non-TCC event likelihoods to obviate the effect of measurement
non-repeatability.
12. The method according to claim 11 wherein the confidence
metrics may be a fuzzy confidence measure.

An isolation method and system is described for distinguishing
between turbine case cooling (TCC) and high pressure turbine
(HPT) performance faults. A trend is observed in gas path
parameter data during cruise and a resulting percent Δ signature
across the shift in the gas path parameters is assignable to
either an HPT or TCC performance fault. During either fault,
exhaust gas temperature (EGT) will shift upward. Since take-off
EGT margin is calculated from take-off data, the shift, or lack
of shift in EGT margin may be used to differentiate between TCC
and HPT faults.