IGTI/ASME Turbo Expo, Munich, Germany, May 2000

Advanced Diagnostics and Prognostics for

Gas Turbine Engine Risk Assessment
Michael J. Roemer and Gregory J. Kacprzynski
Impact Technologies, LLC
125 Tech Park Drive
Rochester, NY 14623
716-424-1990
[email protected]

Abstract- Real-time, integrated health monitoring of gas of enhanced diagnostic and prognostic algorithms that can
turbine engines that can detect, classify, and predict predict, within a specified confidence bound, time-to-failure
developing engine faults is critical to reducing operating and of critical engine components can provide many benefits
maintenance costs while optimizing the life of critical including:
engine components. Statistical-based anomaly detection
algorithms, fault pattern recognition techniques and • Improved safety associated with operating and
advanced probabilistic models for diagnosing structural, maintaining gas turbine engines.
performance and vibration related faults and degradation • Reduced overall life cycle costs of engines from
can now be developed for real-time monitoring installation to retirement.
environments. Integration and implementation of these • Ability to optimize maintenance intervals for specific
advanced technologies presents a great opportunity to engines or fleets of engines and prioritization of tasks to
significantly enhance current engine health monitoring be performed during the planned maintenance events.
capabilities and risk management practices. • Increased up-time/availability of all engines within a
fleet.
This paper describes some novel diagnostic and prognostic • Provides engineering justification for scheduling
technologies for dedicated, real-time sensor analysis, maintenance actions with corresponding economic
performance anomaly detection and diagnosis, vibration benefits clearly identifiable.
fault detection, and component prognostics. The
technologies have been developed for gas turbine engine The development of enhanced diagnostic and prognostic
health monitoring and prediction applications which strategies built upon existing engine condition monitoring
includes an array of intelligent algorithms for assessing the platforms can allow for continuous monitoring and prediction
total ‘health’ of an engine, both mechanically and of component failure rates and degraded engine performance.
thermodynamically. A block diagram illustrating how prognostic technologies can
be integrated within existing diagnostic system architectures
TABLE OF CONTENTS is shown in Figure 1.

1. INTRODUCTION
2. TECHNICAL APPROACH
3. ENHANCEMENTS TO ENGINE HEALTH
MONITORING AND DIAGNOSTIC S YSTEMS
4. COMPONENT PROGNOSTIC MODELING
5. CONCLUSIONS
6. REFERENCES

1. INTRODUCTION

There is a great opportunity for military jet engines to

become more reliable, operationally available and
economically maintained through the use of enhanced
diagnostic and prognostic strategies such as those Figure 1 Integration of Prognostics within Diagnostic
presented in this paper. The development and integration Framework
 IGTI/ASME Turbo Expo, Munich, Germany, May 2000

From Figure 1, the integration of prognostic technologies consequential and fixed costs can then be used to optimize
within existing diagnostic systems begins with validated the maintenance intervals or alter operational plans.
sensor information currently measured on the engine being
feed directly into the diagnostic algorithms for fault 2. TECHNICAL APPROACH
detection/isolation and classification. The ability of an
enhanced diagnostic system to fuse information from
multiple diagnostic sources together to provide a more In order to evolve purely diagnostic health monitoring
confident diagnosis is emphasized along with a system’s systems to those that are capable of more robust diagnostics
ability to estimate confidence and severity levels and failure prognostics, a probabilistic framework is
associated with a particular diagnosis. In a parallel mode, advantageous. Certainly, a prognostic system output that
the validated sensor data and real-time current/past only reports a specific time-to-failure without having any
diagnostic information is utilized by the prognostic confidence (uncertainty) bound associated with the prediction
modules to predict future time-to-failure, failure rates would be unwise. This is true for simple prognostic
and/or degraded engine condition (i.e. vibration alarm approaches that only utilize historical reliability data (such as
limits, performance margins, etc.). The prognostic Weibull distributions) to the more advanced prognostic
modules will utilize physics-based, stochastic models modeling approaches that take design parameter and
taking into account randomness in operation profiles, operating condition uncertainties into account.
extreme operating events and component forcing. In The prognostic modeling approach implemented in this paper
addition, the diagnostic results will be combined with past takes advantage of the directly sensed parameters, fused and
history information to train real-time algorithms (such as a diagnostic EHM system results, as well as inspection and
neural networks or real-time probabilistic models) to historical reliability data to provide critical inputs for
continuously update the projections on remaining life. The producing accurate failure predictions. As shown in Figure
specific approaches and algorithms for determining these 2, this process begins with a comprehensive evaluation of the
component prognostic results are described in this paper. sensor data including multiple techniques that are fused
together for identifying incipient sensor failure modes. Once
Once predictions of time-to-failure or degraded condition the possibility of a sensor malfunction has been minimized,
are determined with associated confidence bounds, the multiple diagnostic algorithm outputs are combined with
prognostic failure distribution projections can be used in a diagnostic fusion techniques to enhance the fault
risk-based analysis to optimize the time for performing identification capabilities of the engine health monitoring
specific maintenance tasks. A process which examines the system. An example of diagnostic fusion might be
expected value between performing maintenance on an intelligently processing information from vibration data,
engine or component at the next opportunity (therefore performance parameters, and oil related measurements to
reducing risk but at a cost of doing the maintenance) versus detect bearing faults with higher certainty. Next, in order to
delaying maintenance action (potential continued increased obtain the prognostic projections on component failures or
risk but delaying maintenance cost) can be used for this unsatisfactory engine operation, different levels of prognostic
purpose. The difference in risk between the two strategies can be utilized. Approaches to be considered in
maintenance or operating scenarios and associated this paper include probabilistic, physics-based models that
take design parameter and operating condition uncertainties
into account.

Figure 2 Prognostic Enhancements to Diagnostic Systems

2
 IGTI/ASME Turbo Expo, Munich, Germany, May 2000

3. ENHANCEMENTS TO ENGINE HEALTH Fused Sensor Validation and Diagnostics

MONITORING AND DIAGNOSTIC S YSTEMS
A necessary front-end to all engine health monitoring
systems must insure the integrity of the measured
A few technologies for improving the robustness of existing parameters. Sensor problems such as ground loop faults,
engine health monitoring (EHM) and diagnostic systems sensor drift or electrical noise can often appear as the onset
will be discussed next which include; fused sensor
of a performance or vibration fault and must be isolated and
validation and recovery, probabilistic anomaly detection and detected properly. The sensor analysis enhancements
diagnosis and stochastic vibration fault classification. These described herein must validate the integrity of sensor signals
real-time diagnostic enhancements are capable of
with multiple techniques that isolate particular failure
accounting for uncertainties from engine transient modes, therefore also providing a level of fault diagnosis for
conditions, random measurement fluctuations and modeling the sensor system itself.
errors associated with model-based diagnostic procedures.
Besides providing more robust diagnostic results, the The sensor validation and diagnostic process is performed
techniques yield direct confidence and severity levels using multiple and collaborative techniques that offer
associated with particular diagnoses that can be directly
advantages for isolating and detecting specific sensor failure
utilized by prognostic algorithms which are also modes (Figure 3). Some available techniques that have
probabilistic in nature. been implemented with success include; trained neural
networks, fuzzy logic analysis, signal auto and cross
correlation, and high-pass filtering.

Figure 3 Sensor Validation and Fault Diagnostics

The sensor validation and diagnostic techniques fall in two because system response characteristics may be used to aid
categories; signal processing-based and physics-based. The in fault identification in highly dynamic situations.
signal processing techniques accomplish their tasks
(detecting and diagnosing sensor anomalies) independent of The more advanced detector schemes are rooted in the
a monitored system’s characteristics. Conversely, the engine’s physical characteristics. The fuzzy logic based
physics-based techniques accomplish more advanced tasks sensor analysis continuously assesses the “normal” bands
(sensor recovery and virtually sensing parameters) because associated with each sensor signal at the current operating
they are developed from a-priori knowledge of the system condition. When a signal goes outside these bands, while
characteristics. Signal auto-correlation and high-pass others remain within, an anomaly is detected associated with
filtering are two signal processing techniques used for those specific sensors. The neural network operates by
identifying grounding and intermittent spikes at high levels comparing the physical relationships between signals as
of sensitivity. The correlation matrix tracks the degree of determined from a gas path model of the engine’s
co-linearity between signals in real-time and can detect performance parameters. The neural network has the ability
clipping, noise and multiplexer cross-talk. This technique to recovery sensors that have failed or “virtually” sense
bridges signal processing and physics-based approaches critical diagnostic parameters that is not monitored. All of
these parallel algorithms may be combined in a data fusion
3
 IGTI/ASME Turbo Expo, Munich, Germany, May 2000

process that determines the final confidence levels that a

particular sensor has either failed or has suspect operation.
[3].

Figure 4 shows an example of some engine test cell results

where the fuel flow sensor began “spiking” at full throttle.
Both the high pass filter and autocorrelation technique were
able to autonomously detecting this sensor fault.

Figure 5 Engine Signature Model and Statistical Anomaly

Detection

Engine performance anomaly detection algorithms are

designed to statistically detect the manner in which
performance parameters are shifting over time and to
analyze the confidence intervals associated with engine
Figure 4 Detection of Spikes in Fuel Flow Sensor
performance degradation issues.

The algorithms statistically trend and analyze how the

Statistical Engine Baseline Models for Robust Fault
measured parameters compare against the engine baseline
Detection
“signature” model. A sample parametric plot is shown in
the bottom left of Figure 5. The bottom right of Figure 5
A critical aspect of most model-based diagnostic or
illustrates the normal deviation from a baseline engine
prognostic processes is the physics-based engineering
signature model as the PDF or Probability Density Function
analysis and/or computer simulation of the engine and
identified with “baseline mean” under a specific operating
associated baseline data that is used to develop the “baseline
condition. When a sufficient amount of new data at this
signature” model. The analysis model will be used to model
condition has been collected, its PDF (identified by “new
the engine’s “normal” behavior, and any statistical deviation
mean”) may be statistically compared with the baseline PDF
from that model will used to detect engine faults. Baseline
to determine the confidence that the mean value has shifted.
data is typically analyzed for each engine over the entire
As an example, we may be 75% confident that the mean has
envelope of operating conditions. The corrected parametric
shifted by more than 3%. Continuously tracking significant
curves are then developed as the “engine signature” models.
parameters such as flows, vibration fault frequency
Subsequent engine performance assessments depends on
amplitudes, etc. in this statistical manner can yield robust
evaluating the distribution shifts associated with the normal
and pertinent diagnostic information regarding degraded
deviations from these baseline parametric models. This
engine operation.
concept is illustrated at the top of Figure 5.
Fused Fault Pattern Recognition and Probabilistic Engine
Diagnostics

Utilizing collaborative probabilistic [2] and pattern

recognition techniques [3] to diagnose particular fault error
patterns with associated confidence and severity levels can
significantly enhance current engine diagnostics. The
probabilistic fault identification process utilizes known
information on how measured parameters degrade over time
and compares them with the current parameter distribution
shifts (calculating the degree of overlap between the known
fault joint probability density function and currently
measured joint PDF) to identify potential fault scenarios.

4
 IGTI/ASME Turbo Expo, Munich, Germany, May 2000

Pattern recognition of a fault pattern has been successfully The function Ö( ) is the standard normal cumulative
performed using an unsupervised neural network (Kohonen distribution. The notation â is the reliability index. If F and
Map) for clustering fault types in series with a standard back C are not normal or lognormal variables then the probability
propagation neural network to classify specific fault patterns of failure can be determined only by using a computer
in 2-D space [4]. The diagnostic predictions made by these algorithm as described in [4].
two techniques are then fused together using a Bayesian
inference algorithm that either increases or decreases the Two different fault measures have been developed from this
confidence of a particular diagnosis based on the outputs stochastic technique [1]. First, a cumulative sensitivity
agreement or disagreements. A discussion and example of index defined by the global non-dimensional variation from
the probabilistic fault diagnostic method is provided in this the initial state, at time 0, to the final state, at time t (over
paper. the time interval [0,t]). Next, an evolutionary sensitivity
index defined by the local non-dimensional variation from
As mentioned, the probabilistic fault identification process an intermediary state, at time ti, to another intermediary
utilizes known information on how measured parameters state, at time ti+1, over the time interval [ti,ti+1]. Both of
degrade over time and compares them with the current these indices are given below.
parameter distribution to track and identify degraded fault
conditions. The amount of each parameter’s shift from the Cumulative Sensitivity Index:
expected baseline operation at a particular operating
condition is continuously analyzed along with the associated βt − β0
statistical confidence level. The shifted distributions are C0 ,t = − (2)
β0
compared against the baseline parameter’s probability
distribution for all known faults. Comparing the measured
distributions against all the known fault distributions yields Evolutionary Sensitivity Index:
the confidence of a known performance degradation
problem. β ti+1 − β ti
Eti ,ti +1 = − (3)
β ti
A stochastic interpretation of the measured and known
parameter distributions can provide a powerful means of where β is the reliability index related to the Euclidean
calculating multiple fault contributions to a current engine distance between the current conditional distribution and a
operating condition and envisioning the most likely given fault distribution.
evolution of the fault. A manipulation of the First Order
Reliability Method (FORM) can be used to gage a fault’s Figure 6 illustrates the fault detection and classification
evolution in state vector space where the faults and the process in a two-dimensional parameter space. Starting at
current condition are not known with complete certainty. the origin, (representing initial normal engine operation) the
The conceptual framework for structural reliability and measured parameter distributions begin to shift as some type
more specifically the FORM algorithm is provided by of performance degradation begins to occur. After several
classical reliability theory [5]. missions when the anomaly level is reached an anomaly
detection warning is issued. As shown in Figure 6, the
In essence, the probabilistic fault diagnostic process current measured PDF moves to point T1 then to T2 as time
involves assigned PDF’s to performance error patterns progresses, ultimately approaching the 4% fault. The
associated to known faults in N-dimensional space. described situation in Figure 6 can be handled with
Similarly, the current error exists as a PDF in the space as mathematical accuracy by using the proposed probabilistic
well. The probability that the current condition (C), may be fault diagnostic calculations previously described.
attributed to a given fault (F) is determined by their joint
probability density function. If C and F can be assumed to
be normally distributed, the probability of association (Pa)
can be found using:

F −C (1)
p a = 2 Φ( − ) = 2Φ ( − β)
σ +σ
2 2
f c

where:

F, C = the mean of the distributions F and C respectively

σ f ,σ c = the standard deviation of the F and C distributions

5
 IGTI/ASME Turbo Expo, Munich, Germany, May 2000

Figure 6 Probabilistic Fault Diagnosis Process collected from accelerometers at specified locations on the
Figures 7 and 8 provide a further example of the evolution machine can be enhanced using neural-fuzzy diagnostic
of a performance error pattern. Three different performance techniques. Domain knowledge associated with particular
faults at a 2% level were identified using a gas path model vibration fault frequencies, fixed frequency ranges, per-rev
of an engine. In this example, and shown in Figure 7, the excitations, structural resonance’s, etc. are extracted from
PDF of the current error pattern initially evolves towards a the vibration spectrums and used to develop a knowledge
2% HP Compressor efficiency fault. This is also shown in base from which the fuzzy logic membership functions and
Figure 8 from the fact that from T=0 to T=3 the Euclidean rulebases are developed. This feature extraction process is
distance between the current PDF and the HPC fault gets illustrated in Figure 9. Non-vibration related data such as
smaller. However, as time goes by, the current condition the performance parameters, oil analysis data, etc. can also
evolves towards, and eventually past, the HPC fault. From be integrated (knowledge fusion) into the fuzzy expert
Figure 8, at T=5 the current PDF is closest to the 2% HP system to collaborate on a particular diagnosis.
Turbine Efficiency fault. Figure 7 illustrates that the
engine’s degradation has indeed evolved beyond association
with the HP Compressor fault to high association with the
HP Turbine fault. In this example, the final position rests at
9.98% association with the HPC fault and 22.8% association
with the HPT fault.

Figure 9 Feature Extraction from Vibration Spectrums

Most current vibration anomaly detection schemes rely

solely on amplitude levels. However, the vibrational
analysis scheme presented here investigates amplitude
levels, critical spectral features and utilizes a shape-based
statistical analysis of the tracked order coupled with a
Figure 7 Degree of Fault Association intelligent rulebase to detect and diagnose mechanical
faults. The combination of these 3 techniques allows for
robust and more sensitive diagnostic capability. Figure 10
shows the average shape of a HP shaft tracked order and +/-
2 standard deviations determined from testing of multiple
engines. The bold line in Figure 10 shows a simulated
tracked order of an engine with a different structural
resonance. In this example, a simple amplitude level band
on the racked order would not detect a problem, however, a
statistical analysis of the shape of the tracked order is able to
detect a fault.

Figure 8 Fault Evolution

Enhanced Vibration Fault Detection and Diagnostics

Real-time assessment of mechanical faults (bearing,

rotordynamic, structural, etc.) utilizing vibration signatures

6
 IGTI/ASME Turbo Expo, Munich, Germany, May 2000

useful component life as a function of uncertainties in

component strength/stress or condition for a particular fault.
The results from such a model can then be used to create a
neural network or probabilistic-based autonomous system
for real-time failure prognostic predictions. Other
information used as input to the prognostic model includes
diagnostic results, current condition assessment data and
operational profile predictions. This knowledge-rich
diagnostic information is generated from multi-sensory data
fusion combined with in-field experience and maintenance
information obtained from data mining processes.

A prognostic model must have ability to predict or forecast

the future condition of a component and/or system of
components given the past and current information. The
Figure 10 – Shape-based Vibration Diagnostics realm of prognostics is sometimes divided into failure and
condition prognostics. Failure prognostics often refers to
the continuous accumulation of damage and/or life on
Database Analysis components or systems of components, with or without the
presence of any identified faults. Components governed by
Utilizing probabilistic and artificial intelligence methods to mechanical wear and failure often fit into this category (i.e.
record and trend critical component life usage, prediction of crack initiation without the presence of a fault
instrumentation problems, as well as vibration and detected). In contrast, condition prognostics is most often
performance faults over the life of the machine is an associated with a fault being diagnosed prior to a vibration
important feature of an advanced diagnostic and prognostic or performance related limit being exceeded. A detected
system. A database would at a minimum consist of error fault must be isolated and assessed for severity so that the
pattern trend charts, life accumulation charts, as well as remaining useful life can be determined. This useful life is
selected data and anomoly detection logs. The error pattern defined by the operating time between detection and an
trend chart shown in Figure 11 is a “to-date” snap-shot of unacceptable level of degradation.
how performace or vibration faults have manifested
themselves over the life of the machine. Similarly, the life Failure Prognostics using Physics-Based Models
accumulation chart would show how the remaining life for a
given critical compent has reduced over time. A physics-based stochastic model for failure prognosis
typically incorporates mechanical (finite element) or
thermodynamic (through-flow model) deterministic models
as their basis. The probabilistic procedure for addressing
inherent modeling uncertainties must be built into these
models using statistical distributions of the parameters that
most directly effect the component life limiting factors.
Some of these factors include the material properties,
dynamic forcing, and process variability. The distribution
on the current remaining life in a component life prediction
may be determined by calculating all possible combinations
of these life-limiting factors in a stochastic process given
past maintenance and operating conditions. Operating hours
can be statistically analyzed, trended and projected into the
future to provide the prognosis of remaining life. More
advanced stochastic models that represent failure mode
Figure 11 Error Pattern Trending uncertainties, projected operational parameters, and
rare/random events can be used to help predict failure mode
propagation.
4. COMPONENT PROGNOSTIC MODELING This analytical model should be calibrated using in-service
data to clearly reflect the root cause of the in-service failure
A physics-based stochastic model is a technically mode experiences. In the case where a finite element model
comprehensive modeling approach that has been can be used (gearing, blading, impellers, or rotors for
traditionally used for component failure mode prognostics. example) crack initiation regions should agree with any in-
It can be used to evaluate the distribution of remaining field experience and inspection data. More empirically

7
 IGTI/ASME Turbo Expo, Munich, Germany, May 2000

analyzed components such as bearings should have clearly associated with this steady stress due to variations in the
identified relationships between diagnosed fault severity and operating environment, temperatures, manufacturing
life consumption. tolerances, etc must be accounted for with a statistical
distribution computed by testing and field service data.
Example of a Stochastic Physics-Based Model Another factor critical in predicting turbine blade life is the
dynamic stress as a function of the uncertainty in harmonic
Sophisticated fracture mechanics and damage accumulation excitation and response characteristics. The strength or
analysis have shown that accelerated crack nucleation and resistance capability of the material must also be considered
micro-crack formation in components can occur due to start- as a function of the uncertainty in material properties.
ups and shutdowns, transient load swings, higher than
expected intermittent loads, or defective component
materials. More commonly, normal wear causes
configuration changes (loose fit of assembled parts, work
hardened surfaces, and reduced structural section areas) that
contribute to increased or unexpected dynamic loading
conditions. High cycle dynamic and transmission loads
cause micro-crack incubation and formation at material
grain boundaries in stress concentrated regions (especially
between hardened surfaces and softer subsurface material
interfaces, and at acute changes in component material
geometry). The majority of crack growth evolves in a sub-
critical propagation process of crack tip blunting, unstable
crack formation, and crack elongation. As super-critical
loading in the cracked material region is approached, growth
accelerates resulting in material dislocation and detachment.
Sub-critical crack evolution is highly dependent on a
component’s material, geometry, loading conditions, and the
particulars of the unique component crack growth cycle.
This kind of failure-mode knowledge is often times
overlooked in determining the potential usefulness of a
particular prognostic or diagnostic algorithm.

The available time to take corrective or compensatory

actions during specific periods of micro-crack incubation,
formation, and sub-critical propagation in the material of a
faulted component must be considered. Based on this
understanding, either of two beneficial actions could be Figure 12 Physics-Based Stochastic Model
taken: a corrective one to perform maintenance to repair or
replace the part, or a compensatory one to reduce system In an effort to describe the process of prognostic modeling
operational loads to extend the life of the faulted part. The for a specific component failure mode (only one aspect of
informed decision exists only if the diagnostic/prognostic this turbine blade model) the LCF fatigue life for the root
system has the ability to detect that the fault exists, isolate it location that experiences stress cycling in excess of the
to the specific component, and assess its severity in a timely material’s yield strength will be described in the equation
manner. (4) below [6].

A stochastic physics-based model of a turbine blade will be  1   1   −1 

1      
used to describe the modeling approach described above and Nf 1L = ⋅σL (true )  ( n − c )  ⋅ K  ( n −c )  ⋅ Ef  c  (4)
2
is shown below in Figure 12. Although each component
prognostic modeling procedure is different based on the
All of the parameters involved in calculating LCF life have
failure modes being predicted, a process that utilizes the
levels of uncertainty associated with them and are therefore
raw, database and processed diagnostic data through a
physical-based model is still applicable. given as probability distributions that may or may not be
Gaussian. The distributions are combined using a Monte-
Carlo simulation. The Monte-Carlo simulation is an
The factors and associated level of uncertainty that most
directly effect the remaining useful life on a component automatic process that randomly selects thousands of
different values from each of the life-limiting factor
must be identified in this physical model. One of these
distributions. Over the entire simulation, the randomly
factors specific to a turbine blade is the steady stress at the
chosen values are combined to generate the distribution of a
critical locations in the root region. The uncertainty
parameter that may have been very difficult or impossible to

8
 IGTI/ASME Turbo Expo, Munich, Germany, May 2000

calculate in a strict analytical sense. The result of the remaining life and predicting future usage associated with
simulation is also shown in Figure 12. critical components.

The damage due the low cycle fatigue may be given by a 6. REFERENCES
non-linear damage accumulation rule proposed by Gary
Halford at NASA Langley [7]:
[1] Bowerman and O’Connel, Forecasting and Time Series,
r1 Duxbury Press, 1993.
 n 
Damage =  1  (5)
 Nf 1L  [2] M.J. Roemer, and B. Atkinson, “Real-Time Engine
Health Monitoring and Diagnostics for Gas Turbine
The complete turbine blade prognostic model must further Engines,” Proceedings of the International Gas Turbine &
account for the other failure modes at other critical locations Aeroengine Congress and Exhibition, Orlando, 1997
on the blade, however that is outside the illustrative scope of
this paper. The net result, however, is the path that’s been [3] M.J. Roemer, and D.M. Ghiocel, ”A Probabilistic
taken to determine the current component life consumption Approach to the Diagnosis of Gas Turbine Engine Faults”
as shown in Figure 13. 53rd Machinery Prevention Technologies (MFPT)
Conference, Virginia Beach, VA, April 1999

[4] T. Kohonen, Self Organizing and Associative Memory,

New York, Springer-Verlag, 1987

[5] P. Bjerager, “On Computational Methods for Structural

Reliability Analysis”, Proceedings of the International
Workshop on Structural System Reliability, Boulder, CO,
1988

[6] G.J. Kacprzynski, and M.J. Roemer, “Outage Interval

Optimization of a Large Fossil Unit Using Stochastic Lifing
Techniques for Blade Failure Risk Evaluation”, Proceedings
of the IJPGC, Baltimore, MD, 1998

[7] G. Halford, “Cumulative Fatigue Damage Modeling –

Crack Nucleation and Early Growth” First International
Conference on Fatigue Damage, Sept, 1996
Figure 13 Damage Accumulation and Projected
Dr. Michael J. Roemer is the Director of Engineering at
Remaining Useful Life
Impact Technologies in Rochester, NY and Adjunct
Professor of Mechanical Engineering at the Rochester
When statistics on the past operating profile of the machine
Institute of Technology. He has over 12 years experience in
are tracked, projected future operating conditions and
the development of real-time, state-of-the-art, automatic
maintenance actions can be estimated and utilized by the
monitoring and diagnostic systems for large turbomachinery
prognostic model in order to forecast the remaining life in
including large steam turbines and gas turbine engines. He
the blade.
has developed several diagnostic and prognostic capabilities
for turbomachines utilizing probabilistic methods that are
5. CONCLUSIONS directly linked to maintenance planning and operation of
turbomachines. He is also currently the Chairman of the
An integrated set of turbomachinery health monitoring, Machinery Failure Prevention Technology (MFPT) Society
diagnostic and prognostic technologies have been presented, a Division of the Vibration Institute.
that when implemented will offer significant potential for
reducing current turbomachinery Life Cycle Costs (LCC). Mr. Gregory J. Kacprzynski is a Project Manager at
These technologies can be implemented across the entire Impact Technologies with 5-yrs. experience in the
spectrum of turbomachines from mid-sized pumps to land- development and testing of diagnostic/prognostic R&D
based gas and steam turbines as well as aircraft engines. systems for compressors, pumps, transmissions, gas and
Implementation of these technologies is advantageous in steam turbines. He has been involved in developing real-
nearly eliminating sensor problems, improving maintenance time, intelligent health monitoring systems for gas turbine
decision effectiveness by providing early warning of engines for on-wing and test cell applications. Greg has
incipient performance and vibration faults and gauging published papers and developed technologies in the areas of

9
 IGTI/ASME Turbo Expo, Munich, Germany, May 2000

turbomachinery health monitoring, prognostics, and

maintenance optimization.

10