Gas Turbine Sensor Faults

POLITECNICO DI TORINO
Master’s Degree in Aerospace Engineering
Master’s Degree Thesis
INVESTIGATION OF SENSOR FAULTS

IN A GAS TURBINE
Supervisors Candidate
Prof. Andrea FERRERO
Giuseppe SOLARO
Prof. Christian KOCH
Ottobre 2022
Summary
When testing gas turbine engines, a variety of flow parameters must be monitored.
The precise measurement of total pressure and total temperature values is critical
for evaluating the performance of the entire engine as well as individual engine
components. Moreover, these parameters are used to define the control law and
the engine health monitoring in an installed engine in operation conditions. The
actual sensor design is often challenging due to the high temperatures and flow
velocities within a gas turbine. In this study, a detailed literature review is reported,
in order to describe the inaccuracy of the two systems, the temperature and the
pressure measurement system, given by the contributions of each component in
the chain. This review allowed a deeper understanding of the different parameters
involved in defining the system’s inaccuracy and, moreover, the formulation of
inaccuracy forecast highlighted in a modern engine. Regarding the temperature
measurement system inaccuracy, a thermocouple in MIMS configuration with an
exposed junction was considered. The indicated temperature is primarily influenced
by systematic errors caused by the design of the sensor probe, whereas thermocouple
drift has a lower impact on the deviation from true temperature: the errors in the
individual components of the chain are in the same order as the inaccuracy of the
type K thermocouple, therefore, they cannot be neglected. Regarding the pressure
measurement system inaccuracy, the probe’s design is critical, instead the errors
from other components of the pressure chain have little impact, which could be
omitted due to the high pressure inside the engine.
ii
Acknowledgements
Desidero ringraziare il Dott.Ing. Christian Koch, mio tutor e guida presso l’ILA
(Institut für Luftfahrtantriebe), che con grande disponibilità e competenza ha
supportato il mio lavoro grazie ai suoi suggerimenti e le sue preziose indicazioni.
Un sentito ringraziamento al Prof. Andrea Ferrero per la fiducia e il sostegno che
mi ha dimostrato in questi mesi.
A mio padre,
per essere stato senza saperlo la mia costante forza di volontà.
A mia madre, esempio di forza e caparbietà,
per essere sempre la casa in cui posso tornare e sentirmi al sicuro.
Questo grande traguardo è anche Vostro.
iii
Table of Contents
List of Tables vii
List of Figures viii
Acronyms xi
Introduction 1
1 V2500 and CFM56 Gas Turbine Engine 4

1.1 V2500 and CFM56 Propulsion Unit . . . . . . . . . . . . . . . . . 4
1.1.1 Thermodynamic Stations . . . . . . . . . . . . . . . . . . . . 6
1.1.2 Sensed signals . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.1.3 Details . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Wiring harness design . . . . . . . . . . . . . . . . . . . . . 9
Pressure transducer . . . . . . . . . . . . . . . . . . . . . . . 11
2 Signal processing for CFM56 engine 13

2.1 Full Authority Digital Engine Control - FADEC . . . . . . . . . . . 13
2.1.1 Signal processing and fault strategy for CFM56 . . . . . . . 14
Data validity and range check . . . . . . . . . . . . . . . . . 15
Cross- channel validation . . . . . . . . . . . . . . . . . . . . 17
2.1.2 Range check and Cross - channel validation for EGT inputs 18
3 Temperature and pressure measurement systems 20

3.1 Sensor chain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.2 Temperature Measurement System . . . . . . . . . . . . . . . . . . 23
3.2.1 Mineral Insulated Metal Sheathed - MIMS . . . . . . . . . . 27
3.2.2 Error source: BIAS thermocouples . . . . . . . . . . . . . . 30
Velocity error . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Conduction error . . . . . . . . . . . . . . . . . . . . . . . . 34
Radiation error . . . . . . . . . . . . . . . . . . . . . . . . . 36
v
3.2.3 Error source: Aging/Drift and Inhomogeneities of the ther-
mocouple . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.2.4 Other error source . . . . . . . . . . . . . . . . . . . . . . . 41
3.2.5 Maintenance . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.3 Pressure Measurements System . . . . . . . . . . . . . . . . . . . . 44
3.3.1 Error source: flow characteristics . . . . . . . . . . . . . . . 48
4 Test case 52
4.1 Test case 1 - Velocity, conduction and radiation error . . . . . . . . 53
4.1.1 On-design . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
Velocity error . . . . . . . . . . . . . . . . . . . . . . . . . . 55
Conduction error . . . . . . . . . . . . . . . . . . . . . . . . 55
Radiation error . . . . . . . . . . . . . . . . . . . . . . . . . 56
Considerations . . . . . . . . . . . . . . . . . . . . . . . . . 56
4.1.2 Off-design . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
4.2 Test case 2 - Aging/Drift and Inhomogeneities of the thermocouple 59
4.3 Test case 3 - Lost of one thermocouple . . . . . . . . . . . . . . . . 61
Conclusions 62
A MATLAB script 64
A.1 Dipendence of the M aJ to AE/AB ratio . . . . . . . . . . . . . . . 64
A.2 Effect of the L/D ratio on conduction error . . . . . . . . . . . . . . 65
A.3 Velocity, conduction and radiation error on-design . . . . . . . . . . 66
A.4 Effect of AE/AB ratio on velocity, conduction and radiation error . 68
A.5 emf - Temperature relationship for type K thermocouple . . . . . . 70
Bibliography 71
vi
List of Tables
1.1 Propulsion Unit - Data V2500-A1 Engine, [2] [3] . . . . . . . . . . . 5

1.2 Propulsion Unit - Data CFM56-5A3 Engine, [4] [5] . . . . . . . . . 6
1.3 Thermodynamic stations . . . . . . . . . . . . . . . . . . . . . . . . 7
1.4 V2500-A1 - sensed signals . . . . . . . . . . . . . . . . . . . . . . . 8
1.5 CFM56-5A3 - sensed signals . . . . . . . . . . . . . . . . . . . . . . 8
3.1 Typical thermocouple response time . . . . . . . . . . . . . . . . . . 29
4.1 Thermocouple chain Inaccuracy . . . . . . . . . . . . . . . . . . . . 57

4.2 Type K Thermocouples Coefficients . . . . . . . . . . . . . . . . . . 59
vii
List of Figures
1 Gas turbine trend, [1] . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1 IAE V2500 Engine - the responsibility, [2] . . . . . . . . . . . . . . 5

1.2 V2500-A1 - Thermodynamic stations, [2] . . . . . . . . . . . . . . . 7
1.3 CFM56-5A3 - Thermodynamic stations, [4] . . . . . . . . . . . . . . 7
1.4 CFM56 - EGT Thermocouples Assembly, [4] . . . . . . . . . . . . . 10
1.5 V2500 - EGT Thermocouples Assembly, [3] . . . . . . . . . . . . . . 11
1.6 CFM56 - Pressure pick-ups, [5] . . . . . . . . . . . . . . . . . . . . 12
2.1 CFM56 - control system components, [6] . . . . . . . . . . . . . . . 14

2.2 EEC Design, [6] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.3 Signal processing,[6] . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.4 Data validity and range check, [6] . . . . . . . . . . . . . . . . . . . 17
2.5 Data selection and fault setting, [6] . . . . . . . . . . . . . . . . . . 18
2.6 EGT signal processing, [6] . . . . . . . . . . . . . . . . . . . . . . . 19
3.1 Temperature measurement system chain . . . . . . . . . . . . . . . 20

3.2 Pressure measurement system chain . . . . . . . . . . . . . . . . . . 21
3.3 Schematic drawing of a thermocouple, [8] . . . . . . . . . . . . . . . 23
3.4 Thermal emf of thermoelements, [7] . . . . . . . . . . . . . . . . . . 26
3.5 Mineral Insulated Metal Sheathed- MIMS, [8] . . . . . . . . . . . . 27
3.6 MIMS thermocouple with exposed Bare Wire Junction,[7] . . . . . . 28
3.7 Transient response of a thermocouple, [7] . . . . . . . . . . . . . . . 30
3.8 Schematic cross-section of total temperature sensor,[14] . . . . . . . 31
3.9 Dependence of the M aJ to AE/AB ratio . . . . . . . . . . . . . . . 33
3.10 Effect of the L/D ratio on conduction error . . . . . . . . . . . . . . 35
3.11 Error due to radiation heat transfer, [14] . . . . . . . . . . . . . . . 36
3.12 Sections of the thermocouple affected by short range order and
oxidation, [10] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.13 Effect of Cr on the Seebeck coefficient of Ni, [8] . . . . . . . . . . . 39
viii
3.14 Drift for bare wire KP and KN thermoelements (3.2 mm wire diam-
eter) in low oxygen content atmosphere - short-range order effect,
[10] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.15 Missing probe tips, [16] . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.16 A comparison among different probes’ insensitivity ranges, [19] . . . 44
3.17 Kiel head yaw and pitch angle, [19] . . . . . . . . . . . . . . . . . . 45
3.18 Pressure Transducer Thermal Zero and Sensitivity Shift, [22] . . . . 47
3.19 Semi-infinte tube, [23] . . . . . . . . . . . . . . . . . . . . . . . . . 48
3.20 Correlation between critical angle and local Mach number for Kiel-
head probes, [19] . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.21 A general speed vector composition useful to analytically threat
bi-dimensional turbulence . . . . . . . . . . . . . . . . . . . . . . . 49
3.22 Correlation between total pressure read by Kiel-head probe and the
turbulence intensity with a CFD evaluation, [24] . . . . . . . . . . . 50
4.1 Best M aJ to reduce the velocity, conduction and radiation error . . 53

4.2 Schematic of the EGT sensor - CFM56, [4] . . . . . . . . . . . . . . 54
4.3 Deviation of a typical total temperature measurement and influence
of separate effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.4 Effect of AE/AB ratio on velocity, conduction and radiation error
to 1188 K . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
4.5 emf - Temperature relationship for type K thermocouple . . . . . . 60
4.6 Lost of one thermocouple . . . . . . . . . . . . . . . . . . . . . . . . 61
ix
Acronyms
EGT
Exhaust Gas Temperature
EEC
Electronic Engine Control
FADEC
Full Authority Digital Engine Control
HPC
High Pressure Compressor
HPT
High Pressure Turbine
LPT
Low Pressure Turbine
OGV
Outlet Guide Vane
CCDL
Cross-Channel Data Link
RTDs
Resistance Temperature Devices
TSR
Temperature Sensitive Resistor
xi
emf
electro-motive force
MIMS
Mineral Insulated Metal Sheathed
TJ
total Temperature of the thermocouple Junction
TT
Free-stream Total Temperature
TS
Free-stream Static Temperature
R
Recovery factor
M aF
Free-stream Mach number
M aJ
Junction Mach number
Yv
Velocity error
AE
Area Kiel entry
AB
Area bleed hole
r
overall recovery factor
R̄
corrected recovery factor
xii
Yc
Conduction error
L
Length of thermocouple exposed
D
wire Diameter
Yr
Radiation error
FS
Full Scale
SIT
Semi-Infinte Tube
DAQ
Data Acquisition system
xiii
Introduction
Modern turbine engine control and health management depend on sensing a wide
range of quantities throughout the engine, including temperatures, pressures, and
vibration, all with varying redundancy, reliability, and accuracy requirements.
These values can be used as inputs to control laws as well as to estimate the
performance and health of various engine components. The sensors must meet
different requirements depending on their function: control sensors require proven
extended reliability, accuracy, and speed of response, which makes them extremely
expensive for turbine engine applications, while engine health management sensors,
on the other hand, do not have the same high-reliability requirements and are
frequently not redundant.
This work focuses on technology for temperature and pressure measurements
inside the section of the engine; the intent is to provide an overview of the difficulties
and issues in obtaining accurate measurements.
The Exhaust Gas Temperature EGT is one of the most important parameters
of the turbine section. The severe circumstances in which the turbine runs are
most of the difficulties encountered in taking measurements inside the turbine area.
This is not surprising given that the turbine is exposed to gases directly from the
combustor, which is near the thermodynamic peak temperature and pressure.
Gas turbine trend, which including higher cycle temperatures and pressures, is
shown in figure 1.
In addition to the fact that turbine conditions are becoming harsher, there is
also a push to reduce uncertainty because uncertainty and transients force the need
for margins in the design of turbine engines. Reducing uncertainty is one tactic
to increase the level of integration that is practical within a complex propulsion
system, as each component or subsystem must be able to function in the worst-case
scenario with other components.
The main reasons to monitor continuously the gas turbine signals, pressure and
temperature, are:
• cycle performance optimization and power control engine through all the
different operational conditions
1
Introduction
Figure 1: Gas turbine trend, [1]
• health monitoring of high-temperature components limiting over-temperature

for increased hot section life. Due to the harsh operating conditions of the
turbine, monitoring temperature readings helps to guarantee that material
limits aren’t exceeded and that the wear can be calculated
• performance evaluation because the knowledge of inlet and exit signals allows
to calculate the efficiency of compressors and turbines
• improved engine transient response within thermal cycling limits
Chapter I provides a brief description of the CFM56 and V2500 gas turbine
engines, which are chosen like reference engines, along with basic performance data
and the engine’s leading particulars about sensor instrumentation.
Chapter II presents for the CFM56 engine how the sensor signal from the
Electronic Engine Control EEC is handled in terms of utility and fault detection
strategy.
Chapter III presents the temperature measurements system and the pressure
measurements system with the related reliability, accuracy, and related problems
attached.
Chapter IV presents the test case developed with an evaluation of the forecast
results, according to the problems highlighted in the temperature measurements
system and pressure measurements system chapter.
Chapter V summarizes and provides conclusions about the work that has been
done. A comparison between values predicted in the test case model developed
2
Introduction
and data provided by the literature is presented, identifying existing problems, and
proposing further work to settle unsolved discrepancies.
Motivation
The Institute of Aircraft Propulsion Systems operates a remote-controlled model
helicopter powered by a gas turbine. In teaching, it is utilized as a test object for
engine performance model development based on flight test data. It also serves as
a test vehicle for the research of transient condition monitoring. The gas turbine
used is a commercial product called PHT-3. The PHT-3 engine on the ground test
bed is used as a real-world data generator for the validation of fault and damage
detection algorithms. There is currently no logic in place to detect sensor faults or
measurement system degradation.
The unknown behavior of malfunctioning sensors and the precise fault variation
picked from a list of probable problems are the starting points for this problem.
Objectives
The aim of this thesis is to find out possible sensor faults, how they will affect the
sensor signal and how such faults could be simulated on the test bed.
3
Chapter 1
V2500 and CFM56

Gas Turbine Engine
1.1 V2500 and CFM56 Propulsion Unit
The V2500 and CFM56 are twin-spool, axial flow, high by-pass ratio turbofans series
designed primarily for 150 seats, short to medium range aircraft. Both engines
incorporate several advanced technology features which include Full Authority
Digital Electronic Control - FADEC, with two channels.
The first one was developed in 1983 by five of the world’s leading aerospace
manufacturers who signed a 30-year collaboration agreement to produce an engine
for the single isle aircraft market with the best proven technology that each could
provide and it was certified in 1988. Each of the shareholder companies is given
the responsibility for developing and delivering one of the five engine modules, as
shown in figure 1.1. However, the engines are assembled by the major partners,
Rolls Royce and Pratt and Whitney.
4
V2500 and CFM56 Gas Turbine Engine
Figure 1.1: IAE V2500 Engine - the responsibility, [2]
The second one is a Franco-American family of high-bypass turbofan aircraft

engines manufactured by CFM International. CFMI is a 50-50 joint venture
between France’s Safran Aircraft Engines (formerly known as Snecma) and the
United States’ GE Aviation.
Both are in charge of component production, and each has its own final assembly
line: GE manufactures the High-Pressure Compressor HPC, combustor, and High-
Pressure Turbine HPT, while Safran manufactures the Fan, gearbox, exhaust, and
Low-Pressure Turbine LPT. Avio of Italy and Honeywell of the United States
manufacture some components.
The propulsion unit - Data is shown below, respectively for the two families of
engines: for both engine families, the engine installed on the A320 is taken as a
reference.
T.O Thrust 111.205 KN

Total Airflow 355 Kgs/second
By-pass ratio 5.42:1
Overall Pressure ratio 29.4:1
Fan Diameter 160 cm
Engine overall length 320 cm
Engine Weight 2242 Kgs
Table 1.1: Propulsion Unit - Data V2500-A1 Engine, [2] [3]
5
T.O Thrust 117.877 KN

Total Airflow 397 Kgs/second
By-pass ratio 6:1
Overall Pressure ratio 31.3:1
Fan Diameter 173 cm
Engine overall length 242.31 cm
Engine Weight 2266 Kgs
Table 1.2: Propulsion Unit - Data CFM56-5A3 Engine, [4] [5]
Both engines, used as a comparison to comprehend which are the instrumented

stations and the primary technologies implemented in terms of sensors in today’s
engine, share a number of similar characteristics.
1.1.1 Thermodynamic Stations

The flow in a turbofan engine follows two distinct and independent paths: primary
and secondary flows. The primary flow, also known as core flow, enters the engine
at the inlet and passes through the Fan near its root, generating a little increase in
pressure. The primary air then enters the core via the Booster and HPC, where
its pressure is greatly increased (approximately 30 times ambient pressure for
both engines), after which it is a mixture with fuel and burned in the combustion
chamber, passes through the HPT and LPT, where it expands, and is discharged
via the core nozzle.
The secondary flow, also known as by-pass flow, enters the engine at the intake,
passes through the Fan on the periphery of its blades, causing a pressure increase,
and is then steered by the Outlet Guide Vanes OGVs before being discharged. This
flow accounts for roughly 80 % of the total thrust generated in the engine.
The V2500 is a Mixed Exhaust Turbofan while the CFM56 is a Separate Flow
Turbofan, which makes a significant difference between the two engines.
• V2500: the core engine ’hot’ exhaust and the ’cool’ by-pass flow are ’mixed’
in the Common Nozzle Assembly before passing through the single propelling
nozzle into the atmosphere.
• CFM56: the core engine ’hot’ exhaust and the ’cool’ by-pass flow pass through
a different propelling nozzle into the atmosphere.
An engine map for each one is presented, showing the location of the stations
respectively in figure 1.2 for V2500 and in figure 1.3 for CFM56. In table 1.3 are
reported for each station the corresponding parts of the engine.
6
Figure 1.2: V2500-A1 - Thermodynamic stations, [2]
Figure 1.3: CFM56-5A3 - Thermodynamic stations, [4]
- V2500 CFM56
Intake/Engine inlet interface 1 1
Fan inlet 2 2
Fan exit 12.5 13
LPC OGV exit 2.5 2.5
HP Compressor exit 3 3
Combustion section exit 4 4
HP Turbine exit 4.5 4.2
LP Turbine exit 4.9 5
Table 1.3: Thermodynamic stations
When the engine’s performance are evaluated, measurements are made at certain
locations inside the engine. These locations coincide very often with thermodynamic
stations, and the measurements are typically temperatures and pressures, static or
total.
7
1.1.2 Sensed signals

For scheduling power setting systems and trend monitoring, the Electronic Engine
Control EEC senses the following pressure and temperature signals. [2][3][4][5]
Fan inlet P2 T2
LPC OGV exit - T2.5
HP Compressor exit P3 T3
LP Turbine exit P4.9 T4.9
Fan Exit P12.5 -
Table 1.4: V2500-A1 - sensed signals
Fan inlet P12 T12

LPC OGV exit P2.5 T2.5
HP Compressor exit PS3 T3
LP Turbine stage 2 inlet - T4.95
LP Turbine exit - T5
Fan Exit PS13 -
Table 1.5: CFM56-5A3 - sensed signals
EGT is one of the most essential factors in engine performance studies since it
is the most important temperature in the engine’s hot sector.
The hot region of the engine has three main temperatures:
• burner exit temperature T4
• HP Turbine exit temperature, identified in T4.5 for V2500 and T4.2 for CFM56
• EGT, identified in in T4.9 for V2500 and T4.95 for CFM56
The burner exit temperature’s value of T4 is impacted by the amount of cooling
air used by the HPC. Furthermore, it is difficult to install a sensor due to the
extreme conditions encountered (high temperature), as these conditions affect the
sensor output signal generating sensor biases (systematic errors) which will be
discussed in the following chapter.
For the same reasons stated above, the HPT exit temperature is not sensed.
The EGT for the engine V2500 is detected in the LPT exit (station 4.9), but
the EGT for the CFM56 is identified in the LP Turbine inter-stage, specifically at
the stage 2 inlet.
To do a better analysis of engine condition, some more parameters are potentially
accessible for the CFM56 engine in comparison to the V2500 engine: sensor signals
8
PS13, P2.5, and T5 are used for monitoring the trend of the CFM56 engine in
addition to the signals of sensors P12.5, T2.5 and T3 also used for monitoring the
trend of the V2500.
Engine condition monitoring is possible by the ability of the FADEC to broadcast
the engine parameters through the ARINC 429 bus output.
1.1.3 Details
It is possible to note that both engines have a very identical array of sensors, except
for the optional sensors described above for the CFM56.
Both engines are equipped with a resistance probe at the LP compressor exit,
going into more detail it is a platinum resistance probe: the operating principle of
the sensor is based on the properties inherent to metals (in this case platinum),
being that their resistance varies in relation to temperature. Therefore, current
supplies to the probe resistor and its signals are modified by the temperature
surrounding the probe.
In the other stations, the temperature signals are perceived through thermocou-
ples type K (Chrome (+) - Alumel (-)): the operating principle of the sensor is
based on the generation of an electromotive force between two dissimilar conductors,
submitted to a temperature gradient.
The thermocouples will be discussed in more detail in the next chapter.
Thermocouples are the most popular sensors used for gas temperature measure-
ments nowadays, owing to their robustness, ease of manufacture, tolerated reaction
time, and low cost of use.
Resistance Temperature Devices RTDs are increasingly being used, although
each has limitations in size, ruggedness, signal processing, or cost when compared
to the more popular thermocouples. Thermistors are not commonly utilized due
to their low-temperature range, which is typically less than 200°C, indeed in our
references engines are only used in LPC exit.
The greatest difference between the two reference engine was observed in the
choice of thermocouple wiring harness design in the station identified as EGT and
the transducer type used to read the pressure signal.
Wiring harness design

For the CFM56, nine thermocouple probes are positioned in the T4.95 plane of the
LPT stage 2 inlet, which sense and average the EGT.
The T4.95 thermocouple wiring harness, as shown in figure 1.4, has the following
elements:
• three identical and interchangeable thermocouple lead assemblies with two

probes
9
• one thermocouple lead assembly with three probes

• one upper extension lead
• one lower extension lead
• one main junction box assembly
Figure 1.4: CFM56 - EGT Thermocouples Assembly, [4]
The electromotive force is averaged in each individual lead assembly (2 probes

and 3 probes), therefore at the level in the main junction box assembly relating
to the interface connector, and in a second time, for each individual assembly the
averaged electromotive force is sent to the EEC via harness, but the right-side
sector and left-side sector lead assembly groups are sent in two different channels.
As regards V2500, four thermocouple probes, in double redundancy, are posi-
tioned in the T4.9 plane of the LPT exit, which sense and average the EGT.
There is one main junction box assembly where all the thermocouple lead
assemblies are connected and the electromotive force is averaged. Through one
connector the average electromotive force is sent to the EEC, but the redundant
signal, with respect to the main one, is managed by a different channel, as shown
in figure 1.5
10
Figure 1.5: V2500 - EGT Thermocouples Assembly, [3]
In way of comparison to the CFM56 engine, the two channels manage different
signals but at the same points of the section, while in the CFM56 the two channels
manage signals from different points of the section.
Pressure transducer
In the V2500 are installed piezoresistive transducers, silicon capacitive type, while
in the CFM56 are installed piezoelectric transducers, quartz capacitive type.
Piezoresistive based transducers rely on the piezoresistive effect which occurs
when the electrical resistance of a material change in response to applied mechanical
strain. The piezoresistive effect in silicon is due primarily to changes at the atomic
level and is approximately two orders of magnitude larger than in metals. When
piezoresistors are placed in a Wheatstone bridge configuration and attached to
a pressure-sensitive diaphragm, a change in resistance is converted to a voltage
output which is proportional to the applied pressure.
Piezoelectric based transducers rely on the piezoelectric effect, which occurs when
a crystal reorients under stress forming an internal polarization. This polarization
results in the generation of charge on the crystal face that is proportional to the
applied stress. Therefore, an electric charge proportional to the applied force is
generated when a piezoelectric material is stressed.
In both engines, in each channel is installed a single pressure transducer for each
pneumatical line. Consequentially, in the station where there are different rakes
mounted in a different positions, the pressure is collected and averaged through a
pneumatical line that runs around the section, as shown in figure 1.6.
11
Figure 1.6: CFM56 - Pressure pick-ups, [5]
The pneumatical line, coming from the engine section within the EEC, through
a shear plate is routed to the channel A and B transducers which compute the
actual pressures.
Only for the optional monitoring pressures, for the CFM56 engine, the pneu-
matical lines are supplied by a single channel.
12
Chapter 2
Signal processing for CFM56

engine
2.1 Full Authority Digital Engine Control -

FADEC
Each engine features a FADEC system. The FADEC functions as data multiplexer
for the propulsion system, making engine data available for condition monitoring:
it provides the engine system regulation and scheduling to control the thrust and
optimize the engine operation controlling the fuel flow, the compressor airflow, and
turbine clearance.
Furthermore, the FADEC provides over-speed protection for N1 and N2, in
order to prevent the engine from exceeding certified limits, and also monitors the
EGT.
The FADEC is made up of:
• the EEC containing two identical computers, designated channel A and channel
B that cross-talk
• the Hydro-Mechanical Unit HMU which converts electrical signals from the
EEC into hydraulic pressures to drive the engine’s valves and actuators
• the peripheral components and sensors that are used for control and monitoring
The EEC is the first component of the engine control system and each channel
can control different engine system components. Channels A and B are usually
always operational but they operate independently from each other: if one fails,
the system automatically switches to the other one. The EEC governs the engine
in response to thrust command inputs from the airplane and provides information
to the airplane for the flight compartment indication, maintenance reporting, and
engine condition monitoring.
13
Signal processing for CFM56 engine
Figure 2.1: CFM56 - control system components, [6]
The EEC receives engine inlet condition data from the Air Data Computers
ADCs and operational commands from the Engine Interface Unit EIU.
As shown in figure 2.1, operating condition data from dedicated engine sensors
such as T12, PS12, P0, N1, N2, PS3, T25, T3, and temperature case are sent to
EEC: through these parameters, it computes the required fuel flow, Variable Stator
Vane VSV, Variable Bleed Valve VBV, HPT clearance control, LPT clearance
control, and rotor active clearance control valve positions.
The FADEC is categorized as “on condition” equipment which means they are
removed only when a failure is indicated.
2.1.1 Signal processing and fault strategy for CFM56

Each channel has all the necessary sensors, engine and aircraft interfaces, a central
CPU, and output drivers: the single channel continuously examines and processes
its own inputs, and to make sure there are no anomalies, it compares its data with
those of the other channels via a cross-channel data link.
The active/stand-by principle is used by the two EEC channels to control their
output drivers. This means that while both channels constantly receive and process
inputs, only the active channel, also known as ’the channel in control’, delivers
control outputs. Active and Stand-by selection is based upon the health of the
channels (no fault identified): each channel determines its own health state, and
when both channels have an equal health status, the Active/Stand-by channel
14
selection rotates as each engine starts.

The dual redundant architecture’s goal is to lessen the impact of errors in the
control system on engine performance.
Figure 2.2: EEC Design, [6]
The different inputs from sensors, switches, and aircraft are checked at several
stages within the EEC before the values received are finally picked to be used in
the control law computations.
After being converted to digital format, the values must go through signal and
range check logic to then be compared across the Cross-Channel Data Link CCDL
prior to being chosen for the control law calculations.
After the values have been calculated and processed in the control law logic, they
pass through the output stage for transmission to the engine, or aircraft systems.
As shown in 2.3, both EEC channels validate their inputs, process the data, and
check their outputs identically.
Data validity and range check

When EEC converts parametric analog inputs into a digital representation and
subsequently verifies whether the conversion is successful, an internal EEC fault
could happen: the signal moves on to the next stage for data validity and range
check if it passes the conversion test, therefore if it is considered valid.
With parametric values are meant values that are not fixed but can change
within a certain range, as opposed to discrete values, which can only have one of
two values, such as on/off or open/closed. The data validity and range test logic
receive discrete inputs directly because most of them are treated as digital signals.
15
Figure 2.3: Signal processing,[6]
Checking the maximum and minimum bounds against predetermined values is

the simplest range test for a parameter and it is used for sensed signals T12, T25
and T5. The EEC then compares the sensor output signal to preset criteria and
creates a sensor failure message for the defective channel if:
• the sensed temperature/pressure is not within the maximum and minimum

ranges
• the fault persists for more than 4.8 seconds
Although the signal’s fault isolation would seem so simple, some inputs take
into account other parameters in the calculation phase, making fault isolation more
difficult like sensed signal P25, PS13, PS3, and T3. By way of showing, the sensor
faults logic for the T3 sensor signal is described: T3 sensor signal is a control input
used to calculate the demand on the HPT Active Clearance Control HPTACC
valve, in this case, the fault strategy used to isolate the sensor fault is much more
stringent because when the engine is running, the EEC checks only the minimum
limit, more precisely -60 °C for more than 4.8 seconds, while if the engine is running,
the EEC compares if T3 value is less or equal to the selected value of T25, or T3
is sensed as being above 725°C. In both case the error must be present for 4.8
seconds. It is evident that the sensor fault strategy used depends on the operating
conditions of the engine.
If the parameter fails the range tests, the EEC sets a flag like ’invalid’ and holds
the parameter at its last valid value, and the EEC generates an error designed as:
’Signal out of the range’.
16
Figure 2.4: Data validity and range check, [6]
Due to all the potential engine failures, isolating sensor faults is not always easy.
The logic examples described above, while not giving specific instructions on how
to do so, do help understand how challenging it is to isolate sensor errors and how
heavily it depends on the engine’s control and monitoring logic.
Cross- channel validation
After passing through the signal and range check logic, these values are compared
throughout the CCDL: validated input signals and validation status data are sent
from one EEC channel to the next via a set of digital discrete control/status signals.
This cross-channel data can then be used in the data selection process.
The opposite channel’s data is used by the local channel through CCDL if the
local channel’s digital data is invalid or fails a maximum/minimum range check,
provided that the data has passed validity and range tests. If it fails the validation
and maximum/minimum range checks also in the stand-by channel, the EEC then
chooses either a fail-safe value or, for certain parameters, a model value generated
mathematically by the EEC from other parameter values: sensor inputs like PS3,
T25, and T3 have an EEC calculated models.
When both channels are operational and cross-channel data is available, the
validity of channel A and channel B dual sensor measured values is given by the
fact that EEC checks the absolute difference between the two inputs within a
predetermined range: if the difference between the inputs is less than the specified
tolerance, the average of the values is selected.
The value that is closest to the input model is chosen if the delta between the
two inputs is beyond the set range. In the absence of a model for the sensor, a fail
safe value is chosen but in either situation, the EEC will produce the following
error message: ’Signal disagree’.
The figure 2.5 schematically summarizes what has just explained above.
17
Figure 2.5: Data selection and fault setting, [6]
2.1.2 Range check and Cross - channel validation for EGT

inputs
The range check and cross-channel validation for EGT inputs are described as a
clarifying illustration.
In the previous chapter, it is highlighted in detail the design of the nine thermo-
couples that are supplied in the CFM56 engine for the EGT, located in the T4.95
plane of the LPT stage 2 inlet.
The nine probes are grouped to make up four sectors: the right sectors are
connected to channel A while the left sectors are connected to channel B.
As long as all sectors are valid, the average value is selected but in the event
that one of the sectors is invalid, the selected value is calculated from the weighted
average of the valid sectors.
To validate/invalidate the various sectors is implemented an internal logic:
• if the engine is not running, the EEC checks to see if the temperature felt is
less than -60°. If a sector’s temperature falls below this level, the EEC sets a
fault flag and generates an error message’Signal out of the range’.
• if the engine is running, the EEC conducts additional examinations on the
sectors. If one of the sectors detects a temperature below the specified value
of T25 or above about 1300°C, the EEC sets a fault flag and generates an
error message. ’Signal out of the range’.
18
An error message ’Signal disagree’ is generated by the EEC if one sector has 200°C
discrepancy compared to the average of the other sectors.
Figure 2.6: EGT signal processing, [6]
There is no true cross-channel validation in the scenario under consideration,

since the two channels use separate signals. However, the CCDL performs a
discrepancy check in relation to the other sectors.
19
Chapter 3
Temperature and pressure

measurement systems
3.1 Sensor chain

The measurement system should have the capability to accommodate a variety of
inputs including both low-level and high-level signals. Typical low-level signals
include thermocouples, transducers, and RTDs, while devices such as Linear
Variable Displacement Transformers LVDTs produce high-level signals.
In this chapter, it will be discussed only and exclusively low-level signals ther-
mocouples and transducers.
The main elements of a measurement system are the sensor, thermocouples/-
transducer, signal transmission/distribution network, signal conditioning, and data
acquisition and processing systems.
A schematic representation of the chain for the temperature and pressure
measurement systems is provided in figure 3.1 and 3.2.
Figure 3.1: Temperature measurement system chain
20
Temperature and pressure measurement systems
Figure 3.2: Pressure measurement system chain
Data collection and processing systems offer the link between the instrumentation
systems and their user’s data. These components obtain the outputs from the
signal conditioning units and process these electrical signals to generate output
data in the specified digital and/or analog formats. This element is identified as
EEC. The data acquisition and processing systems have already been thoroughly
addressed in the previous chapter.
The intrinsic portion of the instrumentation system for temperature and pressure
measuring systems are discussed in details below.
• The sensor is the fundamental element in the hierarchy of a measuring sys-

tem that takes energy from a physical phenomena and sends it for either a
mechanical or electrical transformation to convert the energy into a usable
form.
• The thermocouple/transducer is the fundamental link in the measurement

system chain and converts the energy of the measure received from the sensor
into an electrical quantity that can be amplified and conditioned. These
devices are categorized as either active (commonly called "self-generating")
because due to a physical input they generate an output without the need of
external electrical excitation.
• Any measurement system must have a signal transmission. Simply said,

transmission means that measurement can be taken at one physical point,
and the signal can then be sent to another location for extra processing, such
as signal conditioning or data gathering and processing. The main goal is to
convey the necessary information accurately without allowing outside factors
to reduce the accuracy of the learned information. In a typical real-time
system, signal distribution to the system’s various components is accomplished
using cables.
21
• To transform a thermocouple or transducer output into a signal suitable for

an input to a data acquisition system, signal conditioning is necessary. The
signal conditioning is primarily composed of a filter and an amplifier; the
former is used to remove unwanted frequency components or to enhance other
frequency components of the signal, the latter provides voltage amplification
to normalized inputs to a multiplex system. Since the output is in the millivolt
range, amplification of some kind is necessary to increase the amplitude to a
level appropriate for further processing.
The connected issues for the two systems, temperature and pressure measurement
systems, that are associated with each of the parts mentioned above will then
be looked into and highlighted. On the other hand, simply an indication of the
related error will be provided for the signal conditioning, the data acquisition and
processing systems, which can be neglected.
22
3.2 Temperature Measurement System

The thermocouple sensor is the initial component in the chain of the temperature
measurement system, as shown in figure 3.1.
Usually, the thermocouple sensor consists of the sensing element, the thermo-
couple itself, surrounded by a stagnation tube, the Kiel head, where the flow
velocity is reduced to a level where no further compressibility effects occur and the
temperature under total conditions can be measured.
A thermocouple is a device made by two different wires joined at one end, called
junction end or measuring end. The two wires are called thermoelements or legs
of the thermocouple: the two thermoelements are distinguished as positive and
negative ones.
Being a self-generating device, a thermocouple directly produces a voltage
that can be used as a measure of temperature. That terminal voltage used in
thermometry results only from the Seebeck effect which is due to the association
of two physical phenomena: the Peltier effect, or generation of an electromotive
force due to contact between two dissimilar conductors, and the Thomson effect,
or generation of an electromotive force in a metal bar submitted to a temperature
gradient.[7]
The internal electrical potential difference or electromotive force, which is
outwardly perceived as a voltage between the terminals of a thermocouple, is
known as the Seebeck electromotive force emf. Even if it is not coupled to a circuit,
this Seebeck source emf can be found in any electrically conducting substance that
is not at a constant temperature. However, for thermocouple thermometry, the net
voltage between paired dissimilar materials is always employed due to practical
considerations. The Seebeck emf doesn’t happen at the thermocouple connections
but within thermocouple legs: the emf is generated in the thermocouple wires only
in the temperature gradient existing between the “hot” and “cold” junctions and
not in the junctions themselves.
The temperature vs voltage relationship is given by:
Ú T2 Ú T2
emf = S12 dT = (S1 − S2 ) dT (3.1)
T1 T1
Figure 3.3: Schematic drawing of a thermocouple, [8]
23
The thermocouple temperature measurement is also a differential measurement

because T1 and T2 are being measured at two different thermocouple’s stations
with T2>T1: a monotonic emf versus junction end temperature T2 relationship,
which is the one that is desired, is required to have a usable thermocouple for
measurement, ensuring that a distinct voltage is generated at the tail end for each
temperature at the junction end. However, from the integral in Equation 3.1, it
can be understood that the emf depends on both T1 and T2: a monotonic emf vs
T2 relationship cannot be defined if the tail end temperature is not constant since
T1 and T2 can change independently.
Only one of them may be unknown. Therefore, the temperature of the measuring
junction must be measured according to the known actual temperature of one
reference junction. Due to this, the tail end is kept in an ice bath that is created
by mixing water and crushed ice reference temperature of 0°C: this reference is
used in the rig testing engine.[7]
In the installed engine, the ice bath is typically replaced by an integrated circuit
called a cold junction compensator, where the tail end is at a known temperature
and temperature fluctuations are tolerated. The voltage of the thermocouple at the
tail end can be added to the voltage produced by the cold junction compensator,
which is actually equal to the voltage of the thermocouple between 0°C and
ambient temperature, to recreate the voltage versus temperature relationship of
the thermocouple. A compensating circuit contains a combination of fixed resistors,
and a Temperature Sensitive Resistor TSR, which have a similar variation of emf as
the temperature of the TSR is varied. If the reference junction of the thermocouple
is coupled thermally to the TSR and the compensating circuit is connected in
series with the thermocouple, so that its change in temperature/emf opposes to
that caused by the reference junction, the thermocouple behaves as if the reference
junction temperature were held constant.
In conclusion, with a cold junction compensator, the temperature of the reference
junction is allowed to vary, and a compensating emf is introduced into the circuit
or accounted for by calculation: the Seebeck coefficient is sufficiently matched to
allow the whole instrument’s accuracy to be typically 0.25% of full scale throughout
a tolerable range of temperatures within EEC.[7]
As shown in the previous chapter, in both reference engines are installed K type
thermocouples, which have a range operation until about 1260°C with 3.25 mm
wire size like upper temperature limit, and until 760°C with 0.25 mm wire size
like upper temperature limit: this range to be considered in a conventional design
closed-end protecting tube.
The approximate compositions for positive and negative thermoelements of type
K thermocouple are Nickel - 10% chromium (+) versus nickel - 2% aluminum, 2%
manganese and 1% silicon (—): approximate compositions because most modern
manufacturers now adopt compositions with alterations, which are usually made
24
to improve the stability and/or manufacturing processing of their wire, that can
depart significantly from those of the original formulation.[9]
Large calibration shifts will occur if exposed to a marginally oxidizing atmosphere
at a temperature above 800°C due to Green-rot corrosion, such corrosion results
from preferential oxidation of chromium when the oxygen content of the atmosphere
surrounding the thermocouple is low.[10] This aspect will be explored later.
This thermocouple type presents an accuracy of ± 2.2°C or 0.75% for category
2, while accuracy of ± 1°C or 0.4% for category 1.
Figure 3.4 shows the voltage against temperature relationship for some of the
most common thermocouples (letter designated thermocouples) distinguishing
positive (P) and negative thermoelements(N): from the graph, it’s obvious that for
the temperatures inside the reference engines, other thermocouple types could also
be installed.
Consequently, it will be clarified why it was chosen type K thermocouple in the
reference engines, in accordance with [7]: basically, thermocouples can be divided
into two groups: thermocouples with Ni-base thermoelements (type E, J, K, and N)
called base-metals thermocouples and thermocouples with Pt-base thermoelements
(type R, S and B) called nobles-metals thermocouples.
Type R, S, and B thermocouples employ Pt-base thermoelements and can
withstand temperatures of up to 1700°C; however, they are more expensive and
have a lower voltage output than type K. Ni-based thermocouples, on the other
hand, can operate at lower temperatures than Pt-based thermocouples, but for
the engine taken as reference, the temperature range, at which the two types of
thermocouples can operate, does not affect the choice. The main difference that
influenced the choice is that the continued use of Types R and S thermocouples at
high temperatures results in excessive grain growth, which can lead to mechanical
breakdown of the platinum element, resulting in less durability than type K
thermocouples.
As regards thermocouples with Ni-base thermoelements, the only two thermo-
couples with Ni-base thermoelements that can be ruled out are type J and type
E, the former because the rate of oxidation of the iron thermoelement (positive
thermoelement) is rapid above 540°C, and the use of heavy-gauge wires is recom-
mended when long life is required at higher temperatures, the latter because type
K oxidation resistance characteristics are better than type E.
Regarding type N thermocouple, it works well with reference engines because
it varies from type K because it has silicon added to both the NP and NN wires
as well as more chromium in the NP wire. The NN wire also contains roughly
0.15% magnesium. This thermocouple offers lower drift rates around 1000°C, does
not exhibit the short-range ordering of type K materials and is less vulnerable to
preferential oxidation effects compared to type K thermocouples.
Between the measuring junction and the reference junction extension wires are
25
Figure 3.4: Thermal emf of thermoelements, [7]
inserted, which have nearly identical thermoelectric properties as the thermocouple

wires they are used with. Some advantages of using extension wires are:
• improvement in mechanical or physical properties of the thermoelectric circuit
• thermoelectric circuitry cost reduction. For instance, in cases when the

reference junction is located far from a noble metal, some base metal extension
wires may be used in place of noble metal wires thermocouple
• interchangeability of the thermocouple without having to replace the entire

thermocouple chain
26
Assuming that an extension wire category 1 was employed in the thermocouple

design stage, which means alloys identical to those used in thermocouples, there
are various potential sources of temperature measurement inaccuracy in the use of
extension wires in thermocouple circuits.
The main source of mistake can occur if there is a temperature difference between
the two thermoelement-extension wire junctions: an undesired emf will occur across
the two junctions.[7] [11]
This ∆emf will be perceived as an error in the thermocouple output that is
being measured. For every degree of ∆T between the thermocouple-extension wire
junctions, such errors do not exceed approximately one degree at the measurement
junction. Equilibrating the two junctions’ temperatures, this means isolating the
thermoelement-extension wire junctions part carefully, the inaccuracies will be
eliminated.
3.2.1 Mineral Insulated Metal Sheathed - MIMS

The simplest thermocouple is constructed from two thermoelements linked at one
end and left directly exposed to the operational environment’s atmosphere. The
working environment may be an oxidizing, reducing, or inert atmosphere, and
using thermoelements in certain of these environments may be restricted due to
incompatibility between the materials used to make the thermoelements and the
gases, which may cause rapid deterioration. For instance, green-rot corrosion in
type K thermocouples due to oxidizing atmospheres in temperatures above 800°C.
The Mineral Insulated Metal Sheathed MIMS configuration can solve the issues
caused by the interaction of thermoelements with the atmosphere: the thermoele-
ments are shielded from the environment by a metallic sheath, that provides
mechanical protection, and a ceramic material provides electrical insulation be-
tween the thermoelements and the sheath.
Figure 3.5: Mineral Insulated Metal Sheathed- MIMS, [8]
27
Although in a MIMS thermocouple the thermoelements are working in the

environment made by the insulator and the sheath, the interaction with the
insulator and the sheath may affect the stability of the signal and cause drift.[12]
The major functions of the sheath material are to keep the ceramic compact,
shield the thermoelements and ceramic from the environment, and offer the assembly
mechanical strength.
Sheathed thermocouples often use magnesium oxide as the insulation material.
Magnesium oxide is a widely used thermocouple insulator due to its general
compatibility with common thermoelements and sheathing materials, as well as its
accessibility, reliability, and cheap cost.
The thermoelement’s properties are identical to those of other thermocouples
described above but, since they are contained in a protective sheath, therefore,
small diameter wires can be subjected to high temperatures for longer than the
bare wire of the same size with less damage, provided they are constructed and
sealed appropriately.[10] [12]
Although MIMS configuration can provide wire protection, the time response of
MIMS thermocouples to changes in the operating environment’s temperature is
slower than traditional bare wire configurations. The MIMS with an exposed Bare
Wire Junction configuration can solve the issues: in this type of junction, the sheath
and insulating material are removed to expose the thermocouple wires exposed
for a specific length, as shown in figure 3.6.This cause the wires are subject to
mechanical damage because are exposed to the environment while the thermocouple
will have a fast response.
Figure 3.6: MIMS thermocouple with exposed Bare Wire Junction,[7]
A thermocouple’s response time is typically defined as the time taken for

the thermal voltage (output) to reach 63.2% of maximum for the step change
temperature in question, and this corresponds to the time when the temperature
difference between the thermocouple and the environment has been reduced by
e−1 (36.8%) of the initial difference.
The response time is affected by a number of factors, including thermocouple
dimension, construction, tip configuration and the nature of the medium in which
the sensor is placed.
To provide an order of magnitude, typical thermocouple response times are
shown in table 3.1.
28
Sheat outside diameter Junction type response time [s]

250 [°C] 430 [°C] 850 [°C]
insulated 4 5 16
6.0 mm
exposed 0.4 0.5 0.16
insulated 1.1 1.4 4.5
3.0 mm
exposed 0.11 0.14 -
insulated 0.18 0.21 0.73
1.0 mm
exposed - - -
Table 3.1: Typical thermocouple response time
For exposed measuring junctions, the thermocouple response time is about 10

time less. With conductor diameter in the range 0.025mm to 0.81mm, response
times in the region of 0.05 to 0.40 seconds can be realised.[13]
The transient response time of a thermocouple can be defined from a first
order, linear differential equation that results from the heat balance between the
thermocouple and the surrounding fluid.[7]
The thermocouple energy balance, neglecting thermal radiation transport, bal-
ances the thermocouple convective loss with the time rate of change in the thermo-
couple stored energy.
dT
hA (Te − T ) = mc (3.2)
dt
Where:
• m = thermocouple mass
• c = specific heat capacity of the thermocouple material
• h = heat transfer coefficient
• A = thermocouple heat transfer surface area
Eq. 3.2 is rewritten in the form of a first-order ordinary differential equation.
dT
τ + T = Te (3.3)
dt
Where the time constant, τ , for thermocouple is defined as:
mc
τ= (3.4)
hA
The thermocouple temperature can be solved as a function of time for the
immediate change in fluid temperature at time zero:
29
t 1Út −t
T = Ce τ + Te e τ dT (3.5)
τ 0
Where:
• T = thermocouple temperature
• C = integration costant
• t = time
• Te = environment temperature
Figure 3.7 shows the transient response of a thermocouple. While 63.2% of the
final temperature is reached after one time constant, it takes 5τ for the value to
reach 99.3% of the new temperature value.
Figure 3.7: Transient response of a thermocouple, [7]
3.2.2 Error source: BIAS thermocouples

Sensing elements have limitations that must be understood in order to ensure
measurement accuracy. This accuracy will be determined not only by the sensing
element itself, but also by the design of the probe into which it will be inserted. As a
result, the probe design will always be a compromise between the probe dimensions
limited by available space, strength criteria, and the best possible aerodynamic
probe characteristics to meet the required measurement accuracy.
The aim of the thermocouple probe design is to produce an environment that
allows the thermocouple to measure gas temperature with the required accuracy.
Temperature readings from thermocouples junction, TJ , submerged in flowing
gas via the walls of a gas duct, differs from the total temperature of the gas, TT ,
which is the thermodynamic parameter relevant to the engine. TJ is the result of
thermal equilibrium between the thermocouple and the surroundings environment.
30
According with [11] and [14], the most important thermodynamic factors affecting
thermocouple probe design are:
• conduction along the wires and the sheath of the thermocouple

• radiation to/from the walls and the blades/vanes surfaces
• convection at the boundary layer around the thermocouple
Figure 3.8 shows the schematic cross-section of such a sensor design with the
basic heat transfer phenomena indicated above.
Figure 3.8: Schematic cross-section of total temperature sensor,[14]
Conduction and radiation cause two types of measurement errors: conduction

error and radiation error.
Furthermore, getting the total temperature of a flowing gas is required in order
to transform the kinetic energy of the flowing gas into a temperature increase,
which means to stop the gas adiabatically, In practice, this is only an ideal process:
as a result, the measured total temperature at the thermocouple junction TJ shows
to be lower than the actual total temperature TT of the gas. A measurement error
arises, called velocity error.
Summarizing, aerodynamic and thermodynamic factors generate three types of
systematic errors in thermocouples:
• velocity error
• conduction error
• radiation error
The errors mentioned above will be discussed detail, in according with [11], [14]
and [15].
31
Velocity error
The difference between the total temperature of the thermocouple junction TJ , and
the total temperature of the gas TT , is described with the so called recovery factor
R and is defined as:
Tj − TS
R= (3.6)
TT − TS
This is the ratio of the actually measured to the total thermal energy available
from the flow’s adiabatic deceleration.
A good thermocouple design is achieved if the recovery factor is very close to 1.
The recovery factor R of a bare wire thermocouple is a function of the thermocouple
geometry and can be varied only between narrow limits. For bare thermocouple
wires recommended values of R for probe designs are:
R = 0.68 ± 0.07 wires normal to f low (3.7)
R = 0.86 ± 0.09 wires parallel to f low (3.8)

Due to the incomplete deceleration, a velocity error remains that can give as:
γ−1
M a2J
Yv = TT − TJ = (1 − R) 2
TT (3.9)
1 + γ−1
2
M a2J
The Mach number inside the Kiel at the thermocouple junction M aJ is extremely
important in velocity inaccuracy: this Mach number strongly depends on the free-
stream Mach number M aF .
Being the free-stream Mach number M aF an engine parameter, the only way to
alter the thermocouple junction Mach number M aJ is to change the area ratio of
the Kiel head intake AE and exit bleed holes AB.
1
γ−1
4−
M aF
3
γ−1
M aJ = 1+ M a2F (3.10)
AE
AB
1
In conclusion, the velocity error can be increased, or reduced, by changing the
internal flow Mach number M aJ , regulating the inlet-to-exit area ratio: a higher
inlet-to-exit ratio minimizes velocity error.
The variation of thermocouple junction Mach number M aJ concerning Kiel
entry AE, to bleed hole AB, area ratio, for different free-stream Mach number M aF
is shown in figure 3.9
On design, in practical cases, is prudent to keep M aJ between 0.05 and 0.15 to
minimize the velocity recovery error.
32
Figure 3.9: Dependence of the M aJ to AE/AB ratio
The aerodynamic recovery factor R can be used in conjunction with the overall
recovery factor, which serves as a measure of how well the total temperature probe
under test recovers the total thermal energy of the flow, to perform a correction
factor ∆.
This correction factor ∆ is defined in equation 3.11 and is related to the
aerodynamic recovery factor and ratio of static to total temperature. This correction
factor is subsequently added to the overall recovery factor defined in equation 3.12
to give the flow total temperature through the corrected recovery factor R̄ in
equation 3.13.
TS
3 4
∆ = (1 − R) 1− (3.11)
TT
TJ
r= (3.12)
TT
γ−1
M a2F
R̄ = r + ∆ = r + (1 − R) 2
(3.13)
1 + γ−1
2
M a2F
This corrected recovery factor R̄ is used to correct errors due to Mach number
effects.
33
Conduction error
It is assumed that the heat energy transferred from the hot gas to the thermocouple
hot junction through forced convection is equal to the heat energy transferred
through conduction along the thermocouple wire. The thermal energy loss due to
conduction is given by the following equation:
TT − TM
Yc = TT − TJ = 5 1 21 6 (3.14)
cosh L 4 hc
D kS
2
Where:
• TM = mount temperature
• L = length of thermocouple exposed
• D = wire diameter
• hC = convective heat transfer coefficient
• kS = thermal conductivity of thermocouple material
The conduction error Yc is determined by the temperature differential between

the gas and the probe wall, the L/D ratio of the thermocouple, the convective heat
transfer coefficient hC , and the thermal conductivity of the thermocouple material
kS .
It is reasonable to consider as fixed the mount temperature TM and the thermal
conductivity of thermocouple kS , because the first one is determined by the action
of the external environment on the support and stagnation tube structure, and
the designer can control the effects of the duct wall temperature by insulation
between probe and duct wall, while the second one depends from the material of
the thermocouple.
What has been said leads us to conclude that the only way to increase, or reduce,
conduction error is a variation of the convective heat transfer coefficient hC or L/D
ratio.
The convective heat transfer coefficient hC is calculated from the Nusselt number
N u which in turn is calculated from the Reynolds number Re.
N u kg
hC = (3.15)
D
The Nusselt number can be evaluated through empirical equation:
N u = (0.44 ± 0.06) Re0.5 wires normal to f low (3.16)
N u = (0.085 ± 0.009) Re0.674 wires parallel to f low (3.17)

34
Where
ρV D
Re = (3.18)
µ
It is possible to notice that a possible strengthening of the convective heat
transfer coefficient is directly coupled with a higher flow velocity inside the sensor,
which returns as a counterpart increases in the velocity error.
The figure 3.10 illustrates the effect that the ratio L/D can have on conduction
error for a thermocouple normal to flow.
Figure 3.10: Effect of the L/D ratio on conduction error
In conclusion, the conduction error can be increased, or reduced, by changing

the heat transfer coefficient hC or L/D ratio: a higher hC reduces the conduction
error at the expense of the velocity error because a higher hC corresponds to an
increased thermocouple junction Mach number M aJ , in the worst case, while a
higher L/D ratio minimizes the conduction error. Another way to have higher hC
to reduce the conduction error is a smaller wire diameter.
35
Radiation error
Due to the flow slowdown towards the total condition, the thermocouple is at a
greater temperature than the adjacent components: the stagnation tube has a
temperature near the static temperature due to a large heat transfer to the free
stream. Furthermore, due to flame radiation or excessive air cooling, the engine
parts in front of the sensor may have a greater or lower temperature.
This results in a significant radiation heat transfer that can be expressed by:
1 σϵ 1 2
Yr = TT − TJ = T 4
− T 4
(3.19)
hc 1 + AT C
AW
(1 − ϵ) T W
Where:
• σ = Boltzmann constant = 5.67 10−8 W/(m4 K 4 )

• ϵ = emissivity of thermocouple = 0.2
• TW = wall temperature
• AW = wall surface
Other than hC , all other parameters are fixed. The only way to increase, or
reduce, the radiation error is by changing hC : a higher hC reduces the conduction
error but corresponds to an increased thermocouple junction Mach number M aJ ,
which, as seen for the conduction error, implies an increase of velocity error.
The emissivity ϵ is basically fixed by the thermocouple material, but it changes
during engine running in an unpredictable way due to oxidation or deposition.
Figure 3.11: Error due to radiation heat transfer, [14]
36
Figure 3.11 illustrates the influence of the radiation heat transfer. It is clear
that radiation has an impact on temperature starting at about 500K. This effect
becomes noticeable around 1000 K, even when compared to other errors like the
thermocouple error. The assumption that the maximum temperature of all engine
components is constrained, results in a significant change of the gradient around
1300 K. Above this point, the surfaces immediately around the sensor will be cooled
and the temperature difference will increase. As a result, the radiation heat transfer
has risen sharply.
Another type of radiation is radiation error induced by gas and flame radiation,
but is generally considered to be negligible at atmospheric pressure and temperatures
experienced in the gas turbine.
37
3.2.3 Error source: Aging/Drift and Inhomogeneities of

the thermocouple
When a thermocouple is used to measure an environment’s temperature, it is
expected that the obtained voltage will remain constant if the environment’s
temperature is constant. Even though the environment’s temperature remains
constant, the voltage might actually fluctuate over time; this phenomenon, known
as drift, is a source of mistakes in thermocouple measurement and generates a
decalibration of the thermocouple.
The decalibration can be caused by changes in the metallurgical state of the
thermocouple materials or in the composition of the thermoelements:[7] [10] [12]
• metallurgical changes can be caused by mechanical deformations of the ther-

moelements, the annealing procedure, or solid-state phase deformation
• the composition changes may be caused by reactions between the thermocouple
wires and the insulation, by impurities in the gases surrounding the wires or
impurities from the sheath
According to [10] [12], the main phenomena responsible for type K thermocouple
drift are oxidation and short-range order transformations.
According to [7], short-range ordering contributes to drift in the temperature
range of 200-600°C. On the contrary, oxidation becomes considerably above 800°C.
As a result, it can be assumed that sections of the thermocouple below 600°C are
primarily affected by short-range order and sections above 600°C are primarily
affected by oxidation: this approximation allows to decouple the contribution to
drift from oxidation and short-range order transformation, as shown in figure 3.12.
These approach is confirmed in [10]
Figure 3.12: Sections of the thermocouple affected by short range order and
oxidation, [10]
38
• Oxidation: when type K bare wires thermoelements are exposed to air, they
oxidize. Above about 800°C the effect of oxidation on Ni alloys becomes
important. The oxidation produces an extensive change in composition as a
result of the formation of oxide scales on the surface of the thermoelements.
By way of showing, it is considered the effect of oxidation on the positive
thermoelement KP (Ni-Cr alloy). In the positive thermoelements KP, Cr is
oxidized preferentially forming Cr-rich oxides. The Cr content of the alloy is
not enough to form a continuous external chromium oxide, so an external Ni
oxide is formed with an underlying Cr rich oxide layer.
Because Cr is depleted from the base metal in order to generate Cr rich oxides,
the Cr oxide has a higher Cr concentration than the alloy. As a result, the
Seebeck coefficient is reduced, causing drift.
Figure 3.13: Effect of Cr on the Seebeck coefficient of Ni, [8]
• Short-range order effect: in general, the phenomenon arises because, in the

range of temperatures between 200°C and 600°C, the elements of the KP
thermoelement, locally developed ordered atomic arrangements. The exposed
part becomes heterogeneous causing an increase in the Seebeck coefficient.
The Seebeck coefficient change is time-dependent, and at each temperature,
it increases until a maximum value, which is identified at about 400°C. For
higher temperatures, it decreases, up to regenerating the disordered structure
of the KP thermoelement for temperatures above 600 °C.
39
As a result, it may be stated that the Seebeck coefficient changes reversibly as

a result of the crystal structure’s short-range ordering.
The rate of these changes is affected by a variety of factors, including the
temperature, the composition of the thermoelements, the composition of the
surrounding elements, and the size of the thermocouple wires.
Both, low-temperature and high-temperature, phenomena impacting the Seebeck
coefficient need to be considered to account for drift:
Ú
dT
emf = E0 + ∆emfshort−range + ∆emfoxidation = S dx (3.20)
dx
A change in the Seebeck coefficient is a necessary condition to have drift, but
it is not a sufficient condition for it because the change in the Seebeck coefficient
needs to occur in a region of temperature gradient according to the equation 3.20.
In the MIMS with exposed bare wire junctions configuration, the only ther-
mocouple part subjected to oxidation is the junction, being the only part of the
thermocouple exposed to the surrounding environment.
But changes at the junction of the thermocouple usually do not play any role in
drift because it can be assumed that the junction is at a constant temperature.[8]
For this reason, only the change caused by the short-range order effect contributes
to drift in the MIMS with exposed bare wire junctions configuration.
According to [12], the behavior of the MIMS thermocouple is very similar to a
type K bare wire thermocouple tested in a low oxygen content atmosphere, which
is affected only by short-range ordering, once the effect of oxidation is removed.
This confirms that at 1000°C for MIMS thermocouples the short-range order is
the main drift contribution, and the drift curves, in figure 3.14, can be used to
estimate drift.
The only remaining potential cause of drift is the interaction with the sheath,
but for the same reasons that the contribution of oxidation to drift is neglected, it
will not be taken into consideration.
Drift curves for positive KP thermoelement and negative KN thermoelement at
600°C, 800°C, and 1000°C are shown in Figure 3.14: these drift curves are obtained
in a low-oxygen content atmosphere to simulate only the short-range order effect.
In figure 3.14, short-range order effects are evident even for temperatures above
600°C: the exposition of the junction of a type K thermocouple at a temperature
above 600°C brings a portion of the positive thermoelement to undergo the short-
range order transformation and drift will take place.
The contribution to drift from the KP thermoelement is predominant with
ranges between +100 and +120 µV after 400 h exposure. As temperature increases,
a decrease in KP drift occurs. The contribution from KN thermoelement is
between +5 and +15 µV but no temperature dependence can be detected for KN
thermoelement.
40
Figure 3.14: Drift for bare wire KP and KN thermoelements (3.2 mm wire
diameter) in low oxygen content atmosphere - short-range order effect, [10]
As a result, it may be confirmed that the KP thermoelement is the primary

cause of drift between 0 and 600°C, based on evidence of a short-range order
transformation in the KP thermoelement, well documented in the literature.
3.2.4 Other error source

The most common types of thermocouple damage that set the stage for a sensor
fault are:[16]
• missing probe tips, which is potentially the most serious condition of all from
the standpoint of the possible effects on the control system. In this case, the
probe tip is completely missing from the affected thermocouple.
Figure 3.15: Missing probe tips, [16]
If the thermocouple is inserted directly in the flow, the thermocouple is

41
subjected to foreign object damage, whatever the damage that separated

probe tip may inflict on another part of the engine on its way through the
turbine. The presence in the system of a thermocouple without a probe tip
has a marked effect on the average temperature signal, such that the complete
loss of the thermocouple can be considered.
• open sensor circuits, which occur when a thermocouple develops an open circuit,
which generally occurs adjacent to the weld point of the two thermoelements
in the sensing element.
Such damage can occur if, for example, hot spots develop in the turbine inlet
because of improperly functioning fuel nozzles. Hot spots in this location can cause
a dramatic increase in erosion and sulfidation of metal surfaces, which greatly
reduces the service life of thermocouples.[17]
Also that even under the best of conditions, the effects of heat, erosion, and
sulfidation will eventually exact their toll on all thermocouples. This is why
thermocouples should be inspected at regular intervals.
When a thermocouple develops an open circuit, it generally occurs adjacent to
the weld point of the two thermoelements in the sensing element.
3.2.5 Maintenance
Due to the fact that the manuals for the two reference engines are not available,
the following chapter summarizes the information provided by the Operation and
Maintenance Manual for Allison 250-C18.[18]
Temperature measurement system maintenance consists in inspection and re-
placement of the thermocouple assembly.
The inspection consist in two phases:
• visual check
• electrical check
The visual check is executed through a strong light and 10X magnification to
identify wear, broken leads or other damage. The inspection of the thermocouple
tip is vital to identify cracks, erosion, and scale.
Thermocouple assembly will be replaced for any of the following reasons:
• excessive wear or damage to the leads

• probes or tips that indicate melting or show other evidence of over-temperature
• tips which evidence erosion to the point that either wire has been reduced in
area an estimated 25% or more
42
• tips having scale on either wire affecting 25% of the cross-sectional area. If
the thermocouple is otherwise satisfactory, it is possible to remove the scale
by soft grit blast.
The electrical check of the thermocouple assembly consists in a continuity check

and an insulation check.
The continuity check has to be carried out in order to determine the resistance of
the thermocouple. An open circuit in the harness assembly will cause this resistance
to be high, while a short circuit will cause it to be low.
The insulation check has to be carried out acceptable resistance to ground. In
presence of carbon buildup on the probe tip, it causes a short to ground, therefore
resistance to ground is less than the minimum acceptable. In this case, it is
necessary to clean carbon deposits from the probe using a clean soft cloth or by
salt/grit blast.
Low resistance can be caused by carbon formation on the probe tip, but also by
oxide insulation shorting the probe to ground.
Furthermore, special inspections are scheduled in case of over-temperature
operation or hot start encountered.
Because thermocouple types, temperature ranges, construction, application,
and intensity of use are so numerous and varied, there are no formally specified
frequencies for recalibration. To meet its requirements and experience, an in-house
quality management scheme should evolve a checking and recalibration program.
Where thermocouples are installed for an extended period of time, calibration checks
are best performed in-situ by allowing for the insertion of a standard alongside
the working thermocouple, in alternative a thermocouple can be substituted for a
standard thermocouple and their emf compared.
A change in the emf and calibration of a thermocouple as the result of use,
or even as the immediate result of calibration, can be quantified by immersing
the thermocouple in a thermally stabilized bath or furnace held at an appropriate
temperature and measuring the output at a series of immersion depths spanning
the normal working depth. If the thermocouple is substantially over-immersed,
i.e. beyond any previous working depth, the measured emf should closely approx-
imate the value shown on the (first) calibration certificate at the corresponding
temperature, therefore the calibration system is validated.
For base-metal thermocouples, a replacement with a calibrated thermocouple
rather than a recalibration is often the best solution. In practice, a periodic
replacement program may be preferable.
43
3.3 Pressure Measurements System

The probe is the initial component in the chain of the pressure measurement system,
as shown in the figure 3.2.
There are many probe types available in the market but only a limited number
of them can be used in turbomachinery due to the presence of various constraints
that make this field very demanding in terms of robustness and frequency response.
Especially in fully unsteady environments, fluctuations and turbulence amplitude,
frequency, and vortex scale are the most crucial factors to consider in order to
safeguard a steady-calibrated probe submerged in a fully unsteady environment
from reading an untrue pressure.
The most popular probe is the Pitot device, through which the dynamic pressure,
which is the difference between stagnation pressure, recorded through the probe
intake, and the static pressure, captured by lateral holes not exposed to the main
flow stream, may be measured. Although due to the limited ducts’ section in which
they have to be installed, there is the need to miniaturize the probe, then separate
total and static tubes have to be harnessed.
A flow misalignment, even a slight deviation, can produce a significant mistake
in total pressure reading in this type of probe, therefore is used to get a flow field
measurement almost parallel to flow direction, the direction along which the probe
error is null.
Figure 3.16: A comparison among different probes’ insensitivity ranges, [19]
44
The study carried out by Chue [19] provides an important conclusion that
supports the claim that even though there are various Pitot probe designs, each
with a distinct yaw angle insensitivity range, they all have considerably lower
critical insensitive angles than Kiel-head probes, making Pitot probes unsuitable
for highly unstable environments.
The critical angle is the incidence, referred to either yaw or pitch, at which the
stagnation pressure error overtakes the amount of 1% of the pressure measured at
a null angle of attack.
The Kiel-head probe is nothing more than a Pitot probe whose intake is protected
by a shroud, and it is helpful to obtain stagnation pressure because of the extremity
probe that derives from the Pitot probe.
It can be used if the probe axis alignment and the flow direction have a relative
imprecision or incidence of almost ± 35° ÷ 60°, in function of the probe geometry,
resulting in a higher yaw/pitch applicable range than Pitot probes, as shown in
figure 3.16.
Figure 3.17: Kiel head yaw and pitch angle, [19]
Considering keeping the same probe, results that the probe insensitivity range
rises by increasing the external diameter, therefore the inner geometry does not
affect the probe performance in terms of yaw and pitch angle insensitivity. [20]
Therefore, a degradation of the external surface could give an error reading
pressure from the true pressure.
To obtain a pressure measurement at a specified point in a gas stream, therefore,
the pressure is usually sensed by a probe, which is usually a Kiel-head probe due
to its great range of insensitivity, and then conveyed to a transducer.
As seen previously, the two reference engines have different transducers types:
in the V2500 are installed piezoresistive transducers, silicon capacitive type, while
in the CFM56 are installed piezoelectric transducers, quartz capacitive type.
Production aerospace transducers generally report a total error band. Total
error band specifications are meant to convey the installed performance when any
combination of realistic environmental factors may be present.
Refer to [21], below are shown the characteristics between the piezoresistive
45
transducer and piezoelectric transducer.

The piezoresistive transducer presents a total error band ± ≤ 0.1%F S (Full
Scale) while the piezoelectric transducer presents a total error band ± ≤ 1%F S.
The main contributing factors to total band error are described below., according
with [22]
Total error band is inclusive of the static error band:
ñ
static error band = nonlinearity 2 + hysteresis2 + repeatibility 2 (3.21)
• Non-linearity is defined as the maximum deviation of any calibration point on

a specified straight line, during any one calibration cycle
• Hysteresis is defined as the maximum difference between output readings, at

any pressure value within the specified range, when the value is approached
first with increasing and then with decreasing pressure
• Repeatability is defined as the ability of a transducer to reproduce output

readings when the same pressure value is applied to it consecutively, under
the same conditions, and in the same direction
The overall contribution of hysteresis and repeatability to the static error band is
typically less than 10% in silicon capacitive type, due to the nearly perfectly elastic
nature of the mono-crystalline silicon pressure sensor combined with the low level
of stress applied at full scale. Therefore, Non-linearity is the largest contributing
component of the static error band. The static error band of a pressure transducer
is typically ±0.1% FS.
A great contribution to the total error band is given from thermal zero shift,
which is normally the single largest error source in general purpose pressure
transducers together with thermal sensitivity shift.
Thermal sensitivity shift and thermal zero shift define the effects on sensitivity
and zero measurement output of operation at operating temperatures other than a
normal ambient temperature of 24°C.
The thermal zero drift, which is measured as µV /°C, causes a constant error that
exists over the full range of measurement of the instrument and it can be treated
as a bias and periodically tared. Therefore, the thermal zero drift is normally
removable by calibration.
Figure 3.18 shows the effect on the output of the thermal zero shift and the
thermal sensitivity shift.
46
Figure 3.18: Pressure Transducer Thermal Zero and Sensitivity Shift, [22]
In conclusion, the temperature effect on a pressure sensor will be negligible when

used at the same laboratory temperature at which it is calibrated. This, however,
is not practical for engine measurements.
The pressure transducers can be optimized for different temperature conditions
through temperature compensation active or passive.
Regarding the piezoresistive transducer, in passive compensation the span
compensation is accomplished by using a series resistor with constant voltage
excitation to compensate for the loss of strain sensitivity of silicon piezoresistors
with increasing temperature, reaching a deviation of ±0.01% FS/°C, while in the
active compensation where the piezoresistive pressure sensors are conditioned by
the use of embedded digitally-programmed electronics, temperature sensitivity may
be controlled to within ±0.001% FS/°C.
Regarding the piezoelectric transducer is possible with the addition capacitor to
give a temperature sensitivity in the range ±0.03 ÷ 0.2% FS/°C, which is slightly
higher than the piezoresistive transducer.
Regarding the drift, both types of transducers show small value: ≤ 0.1%FS/year
for the piezoresistive transducer, while ≤ 0.5%FS/year for the piezoelectric trans-
ducer.
The operating temperature capability for both types of transducers is about
540°C, but, as shown in the reference engine, there are regions of a gas turbine,
which are hotter, where dynamic gas path pressures are required to be measured.
In these high-temperature environments, the preferred way of measuring the
pressure is use of a non-resonant semi-infinite tube SIT system which removes the
pressure transducer from the very hot environment by a distance of up to 1 m. [23]
47
In figure 3.19 a schematic representation of an SIT system is given.
Figure 3.19: Semi-infinte tube, [23]
Where:
• LM is about 90 cm maximum
• LM is about 548 cm minimum
The theory behind the SIT design is that for a sufficiently long tube (semi-
infinite), pressure fluctuations at the measuring station will have attenuated to
small enough values at the far end that their reflection back to the measuring
station will be very small, giving negligible measurement errors.
Despite this, in the engines taken as a reference, this technology was not found.
A possible explanation for this choice by the manufacturer could be that the length
of the pressure line is not consistent, therefore the error given from not using an
SIT system is negligible.
3.3.1 Error source: flow characteristics
Regarding the influence of flow characteristics on probe reading the only two
effects that can have a significant influence in this instance are compressibility and
turbulence in the flow field.[24] [19]
Regarding all Kiel-head probes there is a local speed range in which the critical
angle remains constant, this means that the probe in this range is not sensitive to
compressibility, as shown in figure 3.20.
48
Figure 3.20: Correlation between critical angle and local Mach number for Kiel-
head probes, [19]
A steady-calibrated probe in an unsteady and turbulent environment may result

in a reading as erratic as lower is the pitch and yaw insensitivity range of the
probe itself. This is due to the primary effect of turbulence, which features a
time distribution of additional velocity components generated by bi-dimensional
vortexes.
As Chue [19] reported, "For the usual turbulence intensity encountered in pipe
and wind-tunnel flow, the turbulence error is small, e.g. for a turbulence intensity
of 20%, which is too high for practical situations, the impact pressure reads high by
only 2% if the static pressure is assumed to be correctly measured".
According with [20], the bi-dimensional turbulence intensity level of the flow,
which layout is reported in figure 3.21, can be written as:
v
i[%] = (3.22)
U
Figure 3.21: A general speed vector composition useful to analytically threat

bi-dimensional turbulence
Where:
• U = free-stream flow speed

49
• v or w = are the turbulent vortexes induced speed components on the plane

orthogonal to the probe longitudinal axis
• α0 = actual angle of attack
Therefore, considering the limit case of α0 equal to the maximum positive or

negative critical angle, it is possible to define the turbulence intensity above which
the probe’s accuracy decreases:
ilim = tan αcr (3.23)
This leads to the conclusion that until the flow turbulence intensity remains
under 70%, which is the limit turbulence intensity as calculated through equation
3.23 considering a medium Kiel-head critical angle of 35°, the Kiel-head probe is
not sensitive to turbulence induced errors.
A correlation between total pressure read by Kiel-head probe and the turbulence
intensity with a CFD evaluation is reported in figure 3.22
Figure 3.22: Correlation between total pressure read by Kiel-head probe and the
turbulence intensity with a CFD evaluation, [24]
The stagnation pressure in figure 3.22 features a peak at 17 % of intensity

followed by a steep decrease that is due to a possible separation.
Nevertheless, not considering this peak due to separation,the total pressure felt
differs from the free-stream one at almost 10% of turbulence intensity.
This result does not agree with the previously specified limit intensity but, in
first time, very different turbulence models have been used because the figure
50
features a three-dimensional model, and in second time, in the previous study a lot
of real aspects, as well as phenomena, have been neglected.
It is critical to understand the relationship between total pressure read by the
probe and turbulence because, for example, downstream-the-rotor mounted probes
are subjected to important pressure fluctuations owing to wake vortexes generated
by the relative motion of rotor blades ahead the probe axis.
In general, fluctuations in turbo-machines can be subdivided into three categories.
• Periodic, which in general, are produced by the rotation of the blades. Other
factors that may cause this kind of fluctuations may be the presence of shock
waves and the interaction among secondary flow rates, which may include
bleed and bypass ones, and the core primary flow
• Nearly periodic, which are caused by large domain unsteady phenomena, such
as rotating stall as well as compressor’s surge
• Stochastic, which are caused by phenomena such as intermittent blade flutter,
boundary layer separation, unsteady transition as well as turbulence
51
Chapter 4
Test case
In this chapter, for each test case presented, the parameters used to define the
error for each type of problem will be briefly summarized before moving on to the
construction of the test case in question. A comparison with the literature will also
be presented, if available.
Another main objective of this is to choose in which area of the engine to apply
the test case, taking into account in which station the error could occur or in which
station the error occurs with greater magnitude.
For all test cases presented the station engine identified is the EGT.
• Test case 1 - The temperature and the internal flow Mach number M aJ have
the greatest influence on the error highlighted in this study. Because the mach
junction number is independent of the section, the only parameter left to
define the section where the test case should be applied is the temperature.
The EGT section shows a higher temperature between the sections sensed.
• Test case 2 - Temperature and exposure hours have the greatest influence on
the error highlighted in this study. With only three temperature drift curves,
it was decided to apply the test case in the region where the temperature was
closest to the drift curves, the EGT section is identified like it.
• Test case 3 - Because hot spots can dramatically increase the erosion and
sulfidation of metal surfaces, the temperature is the parameter that influenced
the error highlighted in this study. and as identified previously, the EGT
section has a higher temperature than the other sections.
52
Test case
4.1 Test case 1 - Velocity, conduction and radia-

tion error
The total error in temperature measurement is calculated from three individual
errors, namely velocity error, conduction error, and radiation error. As all these
errors reduce the measured gas temperature, uncertainty analysis leads to the
direct addition of all these errors to the measured value TJ to get the actual gas
temperature TT .
The velocity error can be increased, or reduced, by changing the internal flow
Mach number M aJ , regulating the inlet-to-exit area ratio of the probe.
The conduction error can be increased, or reduced, by a variation of the convec-
tive heat transfer coefficient hC , which is directly coupled to flow velocity inside
the sensor.
The radiation error can be increased, or reduced, by a variation of the convective
heat transfer coefficient hC , like the conduction error.
As a result, the main factor is the Mach number of the flow within the sensor
tube that depends on the flow area ratio of the tube inlet AE and the bleed holes
AB (see figure 3.9).
Figure 4.1: Best M aJ to reduce the velocity, conduction and radiation error
As shown in figure 4.1, Kiel’s internal Mach number M aJ is a trade-off between

velocity error on one side and conduction and radiation error on the other side: to
get an optimum result the internal Mach number should be small enough to keep
the velocity error small, but high enough to avoid large conduction and radiation
errors.
During the probe design phase, a minimized recovery error in realistic cases can
be realized with internal Mach numbers between 0.05 and 0.15 but in applications
with very high radiation and conduction errors (high temperatures) the internal
Mach number, which is the most important factor should be chosen in a range of
around 0.15.
Another parameter to be considered in the design is the L/D ratio that influences
the conduction error: in practical probe design, L/D = 10÷15 seems to be achievable
and sufficient regarding the mechanical stability of the junction for wires normal
to flow.
53
Test case
4.1.1 On-design
Before proceeding with the development of the test case, it is necessary to define
the installation of the thermocouple in the reference engine CFM56, which is chosen
between the two reference engine to simulate the test case.
The EGT, which presents a value equal to 1188 K, is not measured in the main
stream of the LPT air, it is measured by a thermocouple located in the interior of
the vane as shown in figure 4.2.
Figure 4.2: Schematic of the EGT sensor - CFM56, [4]
Considering that the real local Mach number in the application is not exactly
known and that for the reference engine CFM56 the EGT is in the 2nd stage of LP
turbine, therefore, a free-stream Mach number is assumed to be 0.8 like a possible
turbine’s Mach design.
To obtain an accurate measurement of the conduction and radiation errors,
a heat transfer analysis of the probe would be required to obtain the mount
temperature TM and the wall temperature TW .
Following it will present a method to obtain valid values in first approximation,
without to recur necessarily to heat transfer analysis.
The following analyses will be conducted using 3.2 mm wire diameter thermo-
couple, being the same diameter used to characterize the drift for bare wire KP
and KN thermoelements in low-oxygen content atmosphere, shown in figure 3.14
Moreover, taking into account that the EGT work under the effect of a gas
′
combustion, specific heat gas combustion cp is taken equal to 1184 KgJ K , the
′
R*(constant gas/molar mass) is taken equal to 293.77 KgJ K , and the γ is taken
equal to 1.33.
54
Test case
Velocity error
Taking into account the considerations made previously, to obtain a Mach number
junction M aJ , in accordance with the range that has been highlighted previously,
a ratio AE/AB equal to 4 is defined.
This ratio, referring to equation 3.10, return a Mach number junction M aJ
equal to 0.137.
As shown in 4.2, the flow it will be considered parallel to the wire. Under
these conditions the recovery factor recommended for bare thermocouple should
be in the range R = 0.86 ± 0.09, but not having characterized the probe, in first
approximation, is assumed the rake shaft like a cylindrical body under cross-flow
conditions, which the recovery factor value is given to 0.68. [14]
Referring to the equation 3.9, a velocity error results equal to 1.18 K.
Conduction error
Taking into account the considerations made previously, the ratio L/D equal to 10
is defined.
Considering a wire diameter equal to 3.2 mm, is obtained a length of thermo-
couple exposed equal to 320 mm.
From Sutherland’s formulas 4.1, a dynamic viscosity equal to 4.60 10−5 mKgs is
obtained for the temperature of 1188 K.
T 3/2
µ=S (4.1)
χ+T
Where:
• S = 1.4610−6 Kg
m s K 1/2
• χ = 110K
An approximation of the velocity inside the probe is given from equation 4.2
and 4.3: considering the Mach number junction M aJ , which is defined as equal to
0.137, and the total temperature, which is defined as equal to 1188 K, a velocity
inside the probe is equal to 93 m/s.
T0
T = γ ′ −1
(4.2)
1+ 2
M a2J
ñ
V = M aJ γ ′ R∗ T (4.3)
Having defined the velocity inside the probe and the dynamic viscosity, it is
possible to define Reynolds number, and consequently, the Nusselt number and the
convective heat transfer coefficient hC , which results equal to 411 mW2 K .
55
Test case
Regarding the mounting temperature, it is assumed that with a bleed from the
first stages of the HPC, which have the same pressure to be able to thus the cooling,
it is possible to maintain the mounting temperature at 150 k less than the total
temperature inside the probe.
The thermal conductivity of thermocouple material kS is 100 mWK while for the
gas, considering the pressure and the temperature in operation condition, a value
of the thermal conductivity is given equal to 0.09 mWK .
Referring to the equation 3.14 is obtained a conduction error equal to 2.09 K.
Radiation error
Referring to the equation 4.5, it is necessary to estimate the surface wall and the
surface thermocouple exposed: the surface wall is estimated 1700 mm2 considering
an internal diameter equal to about 5.2 times the diameter of wires for MIMS
configuration [7], while the surface thermocouple exposed is estimated 643 mm2 .
The rake shaft temperature rises from a low value, at the cooled engine casing,
to a stable value between the static and total temperatures, which is determined
by the shaft’s recovery factor and the radiation heat exchange with surrounding
surfaces, which will be neglected in first approximation.
This analysis allows to arrive at an approximation of the rake’s temperature
equal to 1151 K, following the equation 4.4, where free-stream static temperature
TS is defined through the isentropic equation and equal to 1074.5 K.
γ′ − 1
A B
TW = TS 1+R M a2F (4.4)
2
Referring to the equation 4.5, a radiation error equal to 2.94°C is obtained.
Considerations
Figure 4.3 shows the influence of the different effects on the measurement of total
temperature. A deviation from the total temperature of 0.53% is detected.
The deviation is mainly caused by the radiation which results 0.25% in according
with the figure 3.11.
A correction with the calibrated recovery factor only eliminates the velocity and
the conduction error. Therefore an error of 0.25% will remain after the correction.
This is in the same order as the thermocouple error and must not be neglected.
The comparison between the results of this investigation and data from the
literature shows an overall deviation of this work in good approximation valid. The
overall deviation from the total temperature in [14] is equal to 0.9%. The low
difference can be related to the different temperatures and to the different free-
stream Mach numbers, on which the influence of separate effects is studied. Strongly
56
Test case
deviation is detected for the radiation error being this work evaluated about 200
K less, while the velocity and the conduction error give the same percentage of
deviation.
Figure 4.3: Deviation of a typical total temperature measurement and influence

of separate effects
The other errors like the thermocouple error still have to be added. The table
4.1 shows the single parts identified in the chain of the thermocouple with the
relative inaccuracy. The ∆T obtained in the analysis is related to the temperature
of the EGT in °C.
Kiel’s probe 6.21 °C 0.68%

Type K thermocouple 3.66°C 0.4%
TC extension wire 0.91°C 0.1%
Reference junction 2.28°C 0.25%
DAQ System - «1 - neglected
tot. 13.06°C 1.43%
Table 4.1: Thermocouple chain Inaccuracy
Overall, is detected an inaccuracy from the real total temperature equal to 13.06
°C, which corresponding in percentage to 1.42% respect 915°C EGT temperature.
57
Test case
4.1.2 Off-design
Considering a geometry change of the Kiel’s probe inlet AE and the bleed holes
AB, due to erosion or thermal dilatation, in figure 4.4 is presented the percentage
variation of the velocity, conduction, and radiation error.
Figure 4.4: Effect of AE/AB ratio on velocity, conduction and radiation error to
1188 K
The change of AE/AB ratio is followed by a change of the Mach number junction
M aJ , that as shown in the figure 4.1, is followed by a redefinition of inaccuracy
percentage due to the velocity, conduction, and radiation error.
As shown in figure 4.4, considering the absurd situation in which AE /AB changes
its value to +50 % from the design point, an overall ∆%overall equal to 0.04%
occurs regarding the design configuration, returning a Kiel’s probe inaccuracy equal
to 6.77 K, while for -50 % from the design condition, an overall ∆%overall equal
to 0.15% regarding the design configuration, returning a Kiel’s probe inaccuracy
equal to 8.07 K.
This low variation from design conditions is justified because the most predom-
inant error in high-temperature measurement is radiation error, which increases
exponentially above 1300 K, while the velocity and the conduction error are strongly
influenced by the Mach number junction M aJ , which have no significant values,
going only outside the ideal range and, at the same time, outside the range in
which the speed error increases consistently.
58
Test case
4.2 Test case 2 - Aging/Drift and Inhomogeneities

of the thermocouple
The only thermocouple part subjected to oxidation in the MIMS with exposed
bare wire junctions configuration is the junction, which is the only part of the
thermocouple exposed to the surrounding environment. But nevertheless, changes
at the thermocouple junction usually have no effect on drift because the junction
is assumed to be at constant temperature. As a result, only the change caused by
the short-range order effect contributes to drift in the MIMS configuration with
exposed bare wire junctions. For the same reason, the interaction with the sheath
does not contribute to thermocouple drift.
Referring at the figure 3.14, after 400h for exposition at 1000°C, results a drift
equal to +100 µV for the KP thermoelement, while a drift equal to +15 µV for
the KN thermoelement.
Overall the short-range order effect produces a positive contribution to drift of
115 µV after 400 h.
From the NIST online database [25] has been obtained the type K Thermocouples
Coefficients, shown in table 4.2.
c0 -0.17600413686 10−1 c7 -0.320207200030 10−18

c1 0.389212049750 10−1 c8 0.971511471520 10−22
c2 0.185587700320 10−4 c9 -0.121047212750 10−25
c3 -0.994575928740 10−7 a0 0.118597600000 100
c4 0.318409457190 10−9 a1 -0.118343200000 10−3
c5 -0.560728448890 10−12 a2 0.126968600000 103
c6 0.560750590590 10−15 - -
Table 4.2: Type K Thermocouples Coefficients
The emf-Temperature relationship is obtained with the following equation and

shown in the figure 4.5:
n
2
1 2
emf = ti90 + a0 ea1 (t90 −a2 ) i = 0, ..., n (4.5)
Ø
ci
i=0
59
Test case
Figure 4.5: emf - Temperature relationship for type K thermocouple
At the temperature of 915°C results a thermocouple output equal to 37.9249

µV : considering that the same type k thermocouple has a contribution to drift of
+115 µV after 400 h, the overall output result to be equal to 38.0399 µV , which
corresponds to ∆T equal to 2.5°C.
The contribution of the drift for a thermocouple in configuration MIMS with
exposed junction, with wire diameter 3.20 mm after 400 h to 1000°C results to be
0.27%.
Adding these results to the overall error by velocity, conduction, and radiation
error, a value prediction equal to 15.56°C is obtained, which corresponds to 1.70%
with respect to 915°C EGT temperature.
60
Test case
4.3 Test case 3 - Lost of one thermocouple

Assuming the loss of a thermocouple due to the missing probe tip or open sensor
circuits, as discussed previously, the average value will be calculated from the
weighted average of the valid sectors because an error message Signal disagree’ is
generated by the EEC due the sector that has 200°C discrepancies compared to
the average of the other sectors.
Figure 4.6: Lost of one thermocouple
In the case in which the sector is not excluded, for example, because the logic
has not been implemented, is obtained an untrue temperature read compared to
the real total temperature equal to 1037 K, according to the the following equation:
N qn
Ti
EGTread = i=1
i = 1, ..., n j = 1, ..., N (4.6)
Ø
j=1 nj
Where:
• n = number of thermocouples in each sector

• N = number of sector
61
Conclusions
This chapter represents the end of the study carried out, by the undersigned, at
the ILA (Institut für Luftfahrtantriebe) sited in the University of Stuttgart.
Within the proposed work, the author tried to analyze the contributions of each
component in the chain of temperature and pressure measurement systems, to the
inaccuracy of the two systems.
In addition, it has been conducted a forecast campaign about three test cases
developed, according to the errors highlighted in the two chain systems.
In other words, the present study dealt with the comprehension of how much
and in which way the instrumentation, in terms of temperature and pressure
measurements, used in today’s engine can affect the precision of the real temperature
and pressure inside the gas turbine engine
In this way, the aim of this study consisted in performing, firstly, a deeper
understanding of the contributions of different parameters involved in defining the
system’s inaccuracy and, secondly, the formulation of inaccuracy forecast for the
errors highlighted during this work.
Another feature on which the attention has been focused was been to get an
idea of how modern engine identifies and insulates sensor faults. This analysis led
to the conclusion that due to all the potential engine failures, isolating sensor faults
is not always easy, because some inputs take into account other parameters in the
calculation phase. Therefore, the manufacturers don’t give specific instructions on
the fault strategy. The only conclusion that can be highlighted is how the fault
strategy heavily depends on the engine’s control and monitoring logic.
During the first part of these six months, detailed bibliography research has
been conducted in order to gather sufficient data to focus the study but, owing
to the very broad and variegated field such as this one debated, the link at single
details for each component of the chain it turned out to be necessary.
For the temperature measurement system, despite the results are not highly
representative of the real physics of the studied phenomena as some strong approx-
imations, the same results constitute a valid tool to state that:
• at temperature close to 1000°C the radiation error is the main cause of the
temperature deviation, together with thermocouple accuracy and the error
62
Conclusions
identified for the reference junction. Considering the absurd situation of a

change of ±50% from the design point AE/AB ratio, the plus deviation respect
to the intrinsic one is irrelevant.
• in MIMS configuration with exposed junction, the short-range order is the
main drift contribution, because the junction part exposed to flow can be
assumed that is at a constant temperature, so it does not play any role in
drift. With a diameter of 3.2 mm, the deviation is forecast to be equal to
0.27%, which being in the same order as the thermocouple type K error must
not be neglected.
• the lost of one thermocouple due to missing probe tip or open sensor circuits
leads the internal logic to exclude the sector because the discrepancy compared
to the average of the other sectors is above the limit.
For the pressure measurement system, the largest source of the error is given
from the probe, identified in Kiel’s probe design as it has large ranges of yaw
and pitch insensitivity. The influence of flow characteristics on probe reading
could lead to deviations from the real pressure. Regarding the sensitivity to
compressibility, for Mach number upper 0.5 a critical angle decreases while for
the effect of the turbulence, keeping in mind that in turbo-machinery turbulence
intensities won’t overwhelm a 10% value, the errors will be negligible every time.
Being that the Kiel’s head yaw and pitch insensitivity depend on the external
diameter, degradation due to erosion of the external surface of the shroud could be
the cause of a deviation from real pressure.
63
Appendix A
MATLAB script
A.1 Dipendence of the M aJ to AE/AB ratio
1 gamma=1.409
2 Ma_F= 0 : 0 . 1 : 1 ;
3 r a t i o=l i n s p a c e ( 1 , 1 0 , 2 0 0 )
4 f o r i =1: l e n g t h (Ma_F)
5
6 %Ma_j t o AE/AB r a t i o
7 Maj=Ma_F( i ) . / r a t i o . ∗ ( 1 + (gamma−1) /2∗Ma_F( i ) ) .^( −1/(gamma−1) ) ;
8 end
9
10 %r e c o v e r y f a c t o r : reccomends f o r w i r e t h e r m o c o u p l e normal t o f l o w
11 R=0.9
12 Tt=l i n s p a c e ( 2 8 8 , 1 4 9 0 , 1 0 ) ;
13 f o r i =1: l e n g t h ( Tt )
14
15 % V e l o i t y e r r o r − [K]
16 Ve=(1−R) ∗ ( ( ( gamma−1) /2∗Maj . ^ 2 ) . / ( 1 + (gamma−1) /2∗Maj . ^ 2 ) ) . ∗ Tt ( i )
64
MATLAB script
A.2 Effect of the L/D ratio on conduction error
1 kg = 0 . 1 2 ; %[W/ (m∗K) ]
2 r a t i o =[ 5 , 1 0 , 1 5 , 2 0 ]
3 D=0.005 %[m]
4 L=D∗ r a t i o %[m]
5
6 %Reynold number −−> Re
7 Re = 2 0 0 : 1 0 0 : 6 0 0 0
8
9 %Valutiamo i l numeor d i N u s s e l t −−> Nu
10 Nu= ( 0 . 4 4 + 0 . 0 6 ) ∗Re . ^ ( 1 / 2 ) ;
11
12 %Valutiamo h e a t t r a n s f e r c o e f f i c i e n t −−> h_c
13 hc=Nu . ∗ kg . /D;
14 ks =100 ; %[W/ (m∗K) ]
15 f o r i =1: l e n g t h ( r a t i o )
16 f o r j =1: l e n g t h ( Re )
17
18 %Conduction e r r o r n o r m a l i z z a t o
19 Ce_norm ( i , j ) =1/( c o s h (L( i ) ∗ ( 4 ∗ hc ( j ) / (D∗ ks ) ) ^ ( 1 / 2 ) ) )
20 end
21 end
65
MATLAB script
A.3 Velocity, conduction and radiation error on-

design
1 r a t i o _ 1 =4; %DESIGN
2
3 %Valutiamo g a s combusti
4 gamma= 1 . 4 0 9 ;
5 cp_primo =1184; %[ J / ( kg K) ]
6 R_primo = 2 9 3 . 7 7 ; %[ J / ( kg K) ]
7 gamma_primo=(cp_primo /R_primo ) / ( cp_primo /R_primo−1) ;
8
9 %r e c o v e r y f a c t o r : reccomends f o r w i r e t h e r m o c o u p l e p a r a l l e l t o f l o w
10 R=0.68 ;
11 Ma_F=0.8 ; %Supposed l i k e Ma_D LPT
12 Tt=915+273; %[K]
13 Maj=Ma_F/ r a t i o _ 1 . ∗ ( 1 + ( gamma_primo−1) /2∗Ma_F) ^( −1/(gamma_primo−1) ) ;
14
15 %V e l o c i t y e r r o r
16 Ve=(1−R) ∗ ( ( ( gamma_primo−1) /2∗Maj ^2) . / ( 1 + ( gamma_primo−1) /2∗Maj . ^ 2 ) ) ∗Tt
17
18 kg = 0 . 0 9 ; %[W/ (m∗K) ]
19 r a t i o =10
20 D=0.0032 %[m]
21 L=D∗ r a t i o %[m]
22 c h i =110; %[K]
23
24 S=1.46∗10^ −6; %[ kg / (m∗ s ∗K^1/2) ]
25
26 %S u t h e r l a n d law −−>dinamic v i s c o s i t y
27 mu=S∗Tt ^ ( 3 / 2 ) / ( c h i+Tt ) ; %[ kg / (m∗ s ) ]
28
29 Ts=Tt /(1+(gamma_primo−1) /2∗Maj ^2)

30 c o s t =292.43
31
32 %V e l o c i t y i n s i d e t h e probe
33 V=Maj ∗ ( gamma_primo∗ c o s t ∗Ts ) ^ 0 . 5 %[m/ s ]
34
35 rho = 0 . 3 2 ; %[ Kg/m^ 3 ] −−>rho=P/ (Ra∗T)
36
37 %Reynold numeber −−> Re

38 Re=rho ∗V∗D/mu
39
40 %N u s s e l t number −−> Nu
41 %Nu= ( 0 . 4 4 ) ∗Re ^ ( 1 / 2 ) %w i r e s normal t o f l o w
42 Nu=0.085∗Re ^ ( 0 . 6 7 4 ) %w i r e s p a r a l l e l t o f l o w
43
66
MATLAB script
44 %Heat t r a n s f e r c o e f f i c i e n t
45 hc=Nu∗ kg /D;
46 ks =100 ; %[W/ (m∗K) ]
47
48 %Conduction e r r o r
49 Ce_norm=1/( c o s h (L) ∗ ( 4 ∗ hc / (D∗ ks ) ) ^ ( 1 / 2 ) )
50 d i f f T =150 %[K]
51 Ce=d i f f T ∗Ce_norm %[K]
52
53 eps =0.2;
54 sigma =5.67∗10^ −8;
55 Ts_w=Tt /(1+(gamma_primo−1) /2∗Ma_F^2)
56 Tw=Ts_w∗(1+ R∗ ( gamma_primo−1) /2∗Ma_F^2)
57
58 %S u r f a c e t h e r m o c o u p l e
59 ATC=2∗ p i ∗D/2∗L∗2 %[m^ 2 ]
60
61 %S u r f a c e w a l l
62 AW=2∗ p i ∗ (D/ 0 . 1 9 ) / 2 ∗ (L) %[m^ 2 ]
63
64 %R a d i a t i o n e r r o r
65 Rer=sigma ∗ e p s / hc ∗ ( Tt^4−Tw^4) ∗1/(1+ATC/AW+(1−e p s ) )
66
67 %P e r c e n t a g e e r r o r
68 err_per_V=Ve/Tt ∗100
69 err_per_Ce=Ce/Tt ∗100
70 err_per_Rer=Rer /Tt ∗100
71 t o t=Ve+Ce+Rer
67
MATLAB script
A.4 Effect of AE/AB ratio on velocity, conduc-

tion and radiation error
1 ratio_1 = 1 : 0 . 1 : 9 ;
2
3 %Valutiamo g a s combusti
4 gamma= 1 . 4 0 9 ;
5 cp_primo =1184; %[ J / ( kgK ) ]
6 R_primo = 2 9 3 . 7 7 ; %[ J / ( kgK ) ]
7 gamma_primo=(cp_primo /R_primo ) / ( cp_primo /R_primo−1) ;
8
9 %Recovery f a c t o r : recommends f o r w i r e t h e r m o c o u p l e p a r a l l e l t o f l o w
10 R=0.68 ;
11 Ma_F=0.8 ; %Supposed l i k e Ma_D LPT
12 Tt=915+273; %[K]
13 Maj=Ma_F. / r a t i o _ 1 . ∗ ( 1 + ( gamma_primo−1) /2∗Ma_F) ^( −1/(gamma_primo−1) ) ;
14
15 %V e l o c i t y e r r o r
16 Ve=(1−R) . ∗ ( ( ( gamma_primo−1) . / 2 ∗ Maj . ^ 2 ) . / ( 1 + ( gamma_primo−1) /2∗Maj . ^ 2 ) )
∗Tt
17
18 kg = 0 . 0 9 ; %[W/ (m∗K) ]
19 r a t i o =10
20 D=0.0032 %[m]
21 L=D∗ r a t i o %[m]
22 c h i =110; %[K]
23 S=1.46∗10^ −6; %k [ g / (m∗ s ∗K^1/2) ]
24
25 %>S u t h e r l a n d law −−> dinamic v i s c o s i t y
26 mu=S∗Tt ^ ( 3 / 2 ) / ( c h i+Tt ) ; %[ kg / (m∗ s ) ]
27
28 Ts=Tt . / ( 1 + ( gamma_primo−1) . / 2 ∗ Maj . ^ 2 )

29 c o s t =292.43
30
31 %V e l o c i t y i n s i d e t h e probe
32 V=Maj . ∗ ( gamma_primo∗ c o s t ∗Ts ) . ^ 0 . 5 %[m/ s ]
33
34 rho = 0 . 3 2 ; %K[ g /m^ 3 ] −−>rho=P/ (Ra∗T)
35
36 %Reynolds number −−> Re

37 Re=rho ∗V. ∗D. /mu
38
39 %N u s s e l t number −−> Nu
40 %Nu= ( 0 . 4 4 ) ∗Re . ^ ( 1 / 2 ) ; %Wires normal t o f l o w
41 Nu=0.085∗Re . ^ ( 0 . 6 7 4 ) %Wires p a r a l l e l t o f l o w
42
68
MATLAB script
43 %Heat t r a n s f r c o e f f i c i e n t
44 hc=Nu . ∗ kg /D;
45 ks =100 ; %[W/ (m∗K) ]
46
47 %Conduction e r r o r
48 Ce_norm=1./( c o s h (L) . ∗ ( 4 ∗ hc . / (D∗ ks ) ) . ^ ( 1 / 2 ) )
49 d i f f T =150 %[K]
50 Ce=d i f f T ∗Ce_norm
51
52
53 eps =0.2; sigma =5.67∗10^ −8;
54 Ts_w=Tt /(1+(gamma_primo−1) /2∗Ma_F^2)
55 Tw=Ts_w∗(1+ R∗ ( gamma_primo−1) /2∗Ma_F^2)
56
57 %S u r f a c e t h e r m o c o u p l e
58 ATC=2∗ p i ∗D/2∗L∗2 %[m^ 2 ]
59
60 %S u r f a c e w a l l
61 AW=2∗ p i ∗ (D/ 0 . 1 9 ) / 2 ∗ (L) %[m^ 2 ]
62
63 %r a d i a t i o n e r r o r
64 Rer=sigma ∗ e p s . / hc . ∗ ( Tt^4−Tw^4) ∗1/(1+ATC/AW+(1−e p s ) )
65
66 %P e r c e n t a g e e r r o r
67 err_per_V=Ve . / Tt ∗100
68 err_per_Ce=Ce . / Tt ∗100
69 err_per_Rer=Rer . / Tt ∗100
70 t o t=Ve+Ce+Rer
71 t o t _ p e r c=t o t /Tt ∗100
69
MATLAB script
A.5 emf - Temperature relationship for type K

thermocouple
1 %Type K c o e f f i c i e n t s
2 c0 = −0.176004136860∗10^( −1)
3 c1 =0.389212049750∗10^( −1)
4 c2 =0.185587700320∗10^( −4)
5 c3 = −0.994575928740∗10^( −7)
6 c4 =0.318409457190∗10^( −9)
7 c5 = −0.560728448890∗10^( −12)
8 c6 =0.560750590590∗10^( −15)
9 c7 = −0.320207200030∗10^( −18)
10 c8 =0.971511471520∗10^( −22)
11 c9 = −0.121047212750∗10^( −25)
12 a0 =0.118597600000∗10^0
13 a1 = −0.118343200000∗10^( −3)
14 a2 = 0 . 1 2 6 9 6 8 6 0 0 0 0 0 ∗ 1 0 ^ ( 3 )
15 x=0:0.5:1372
16 f o r i =1: l e n g t h ( x )
17 y ( i )=c0+c1 ∗x ( i )+c2 ∗x ( i )^2+c3 ∗x ( i )^3+c4 ∗x ( i )^4+c5 ∗x ( i )^5+c6 ∗x ( i )^6+c7 ∗
x ( i )^7+c8 ∗x ( i )^8+c9 ∗x ( i )^9+a0 ∗ exp ( a1 ∗ ( x ( i )−a2 ) ^2)
18 end
19
20 %emf t o 1188 K
21 t 1 =915
22 y0=c0+c1 ∗ t 1+c2 ∗ t 1^2+c3 ∗ t 1^3+c4 ∗ t 1^4+c5 ∗ t 1^5+c6 ∗ t 1^6+c7 ∗ t 1^7+c8 ∗ t 1^8+
c9 ∗ t 1^9+a0 ∗ exp ( a1 ∗ ( t1−a2 ) ^2)
23 y1 =37.9249+115∗10^( −3)
70
Bibliography
[1] Alexander Von Moll, Dr. Alireza R. Behbahani, Gustave C. Fralick, John D.
Wrbanek, and Dr. Gary W. Hunter. «A Review of Exhaust Gas Temperature
Sensing Techniques for Modern Turbine Engine Controls». In: Integrated
Distributed Engine Controls. Cleveland, Ohio, July 2014 (cit. on p. 2).
[2] V2500 A1 A5/D5 Familiarization. MTU Maintenance (cit. on pp. 5, 7, 8).
[3] I.A.E V2500 course notes (cit. on pp. 5, 8, 11).
[4] ATA 71-80 engine CFM56-5A. Lufthansa technical training. 1999 (cit. on
pp. 6–8, 10, 54).
[5] Training manual - Engine systems, CFM56-5A. CFMI. 2000 (cit. on pp. 6, 8,
12).
[6] Training manual - Fault detection and annunciation, CFM56-5A. CFMI. 2003
(cit. on pp. 14–19).
[7] Helen M. Hoersch, ed. Manual of the use of the thermocouples in temperature
measurements. ASTM editor, 1993 (cit. on pp. 23–30, 38, 56).
[8] M. Scervini. Thermoelectric materials for thermocouples. url: https://www.
msm.cam.ac.uk/utc/thermocouple/pages/MicheleScerviniMainPage.
html (cit. on pp. 23, 27, 39, 40).
[9] E. S. Webster. «Drift in Type K Bare-Wire Thermocouples from Different
Manufacturers». In: TEMPMEKO. 2016, p. 70 (cit. on p. 25).
[10] M. Scervini and C. Rae. «The Contributions to Drift of Positive and Negative
Thermoelements in Type K Bare-Wire Thermocouples». In: Temperature: Its
Measurements and Control in Science and Industry. Vol. 8. 2013, pp. 564–569
(cit. on pp. 25, 28, 38, 41).
[11] Recommended Practices for Measurement of Gas Path Pressures and Tem-
peratures for Performance Assessment of Aircraft Turbine Engines and Com-
ponents. Advisory Report 245. Advisory Group for Aerospace Research and
Development - AGARD, 1990 (cit. on pp. 27, 31).
71
BIBLIOGRAPHY
[12] M. Scervini and C. Rae. «The Contributions to Drift of Positive and Negative
Thermoelements in Type K MIMS Thermocouples». In: Temperature: Its
Measurements and Control in Science and Industry. Vol. 8. 2013, pp. 570–575
(cit. on pp. 28, 38, 40).
[13] Thermocouple theory and practice. Tech. rep. labfacility - Temperature process
technology (cit. on p. 29).
[14] A. Zeisberger. Total temperature probes for turbine and combustor applications.
American Institute of Aeronautics and Astronautics. 2007 (cit. on pp. 31, 36,
55, 56).
[15] Steven Tyler Englerth. «An Experimental Conduction Error Calibration
Procedure for Cooled Total Temperature Probes». MA thesis. Blacksburg,
Virginia: Virginia Polytechnic Institute, 2015 (cit. on p. 31).
[16] D. L. Fish. «Thermocouples and TIT». In: Service news 9 (1982), pp. 147–223
(cit. on p. 41).
[17] I.N.Niranjan Kumar V.Naga Bhushana Rao and*, K.Bala Prasad, N.Madhulata,
and Naresh Gurajarapu. «Failure mechanisms in turbine blades of a gas tur-
bine Engine –an overview». In: International Journal of Engineering Research
and Development 10 (2014), pp. 48–57 (cit. on p. 42).
[18] Operation and Maintenance Manual - revision 7. Rolls–Royce Corporation.
2003 (cit. on p. 42).
[19] S. H. Chue. «Pressure probes for fluid measurement». In: Progress in Aerospace
Sciences 16 (1975), pp. 147–223 (cit. on pp. 44, 45, 48, 49).
[20] Gerardo Russo. «Experimental analysis of pneumatic probes in an unsteady
environment». MA thesis. Turin: Politecnico di Torino, 2018 (cit. on pp. 45,
49).
[21] Steve Carter, Alex Ned, John Chivers, and Andy Bemis. Selecting Piezore-
sistive vs. Piezoelectric Pressure Transducers. Tech. rep. AN-102. NJ: Kulite
Semiconductor Products, Inc, 2018 (cit. on p. 45).
[22] Steve Carter. Understanding the Pressure Transducer Datasheet: Total Error
Band vs. Accuracy. Tech. rep. TD 1012. NJ: Kulite Semiconductor Products,
Inc, 2020 (cit. on pp. 46, 47).
[23] Application Information. Tech. rep. Kulite Semiconductor Products, Inc, 2018
(cit. on pp. 47, 48).
[24] S. Bauinger, A. Marn, E. Göttlich, and F. Heitmeir. «Influence of Pressure
Fluctuations on the Mean Value of Different Pneumatic Probes». In: Inter-
national journal of Turbomachinery Propulsion and Power (2017) (cit. on
pp. 48, 50).
72
BIBLIOGRAPHY
[25] NIST. Tables of Thermoelectric Voltages and Coefficients - ITS90. url:

https://srdata.nist.gov/its90/type_k/kcoefficients.html (cit. on
p. 59).
73

Gas Turbine Sensor Faults

Uploaded by

Copyright:

Available Formats

Gas Turbine Sensor Faults

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Gas Turbine Sensor Faults

Uploaded by

Copyright:

Available Formats

POLITECNICO DI TORINO

Master’s Degree in Aerospace Engineering

Master’s Degree Thesis

INVESTIGATION OF SENSOR FAULTS

List of Tables vii

List of Figures viii

1 V2500 and CFM56 Gas Turbine Engine 4

2 Signal processing for CFM56 engine 13

3 Temperature and pressure measurement systems 20

1.1 Propulsion Unit - Data V2500-A1 Engine, [2] [3] . . . . . . . . . . . 5

3.1 Typical thermocouple response time . . . . . . . . . . . . . . . . . . 29

4.1 Thermocouple chain Inaccuracy . . . . . . . . . . . . . . . . . . . . 57

1.1 IAE V2500 Engine - the responsibility, [2] . . . . . . . . . . . . . . 5

2.1 CFM56 - control system components, [6] . . . . . . . . . . . . . . . 14

3.1 Temperature measurement system chain . . . . . . . . . . . . . . . 20

4.1 Best M aJ to reduce the velocity, conduction and radiation error . . 53

Figure 1: Gas turbine trend, [1]

• health monitoring of high-temperature components limiting over-temperature

• improved engine transient response within thermal cycling limits

V2500 and CFM56

1.1 V2500 and CFM56 Propulsion Unit

Figure 1.1: IAE V2500 Engine - the responsibility, [2]

The second one is a Franco-American family of high-bypass turbofan aircraft

T.O Thrust 111.205 KN

Table 1.1: Propulsion Unit - Data V2500-A1 Engine, [2] [3]

T.O Thrust 117.877 KN

Table 1.2: Propulsion Unit - Data CFM56-5A3 Engine, [4] [5]

Both engines, used as a comparison to comprehend which are the instrumented

1.1.1 Thermodynamic Stations

Figure 1.2: V2500-A1 - Thermodynamic stations, [2]

Figure 1.3: CFM56-5A3 - Thermodynamic stations, [4]

Table 1.3: Thermodynamic stations

1.1.2 Sensed signals

Table 1.4: V2500-A1 - sensed signals

Fan inlet P12 T12

Table 1.5: CFM56-5A3 - sensed signals

Wiring harness design

• three identical and interchangeable thermocouple lead assemblies with two

• one thermocouple lead assembly with three probes

Figure 1.4: CFM56 - EGT Thermocouples Assembly, [4]

The electromotive force is averaged in each individual lead assembly (2 probes

Figure 1.5: V2500 - EGT Thermocouples Assembly, [3]

Figure 1.6: CFM56 - Pressure pick-ups, [5]

Signal processing for CFM56

2.1 Full Authority Digital Engine Control -

Figure 2.1: CFM56 - control system components, [6]

2.1.1 Signal processing and fault strategy for CFM56

selection rotates as each engine starts.

Figure 2.2: EEC Design, [6]

Data validity and range check

Figure 2.3: Signal processing,[6]

Checking the maximum and minimum bounds against predetermined values is

• the sensed temperature/pressure is not within the maximum and minimum

Figure 2.4: Data validity and range check, [6]

Cross- channel validation

Figure 2.5: Data selection and fault setting, [6]

2.1.2 Range check and Cross - channel validation for EGT

Figure 2.6: EGT signal processing, [6]

There is no true cross-channel validation in the scenario under consideration,

Temperature and pressure