Condition monitoring case study

Engine Condition Monitoring as a Route to Savings
PGA – Portugália Airlines as a case study
Gonçalo Matos dos Santos Marques
A thesis in fulfilment of the requirements for the degree of
Master of Aerospace Engineering
Jury
President: Prof. Fernando José Parracho Lau
Advisor: Prof. Pedro da Graça Tavares Álvares Serrão
External Examiner: Prof. António José Nobre Martins Aguiar
November 2010

1
Acknowledgements/Agradecimentos
The language that will be used here in the acknowledgements will be Portuguese, for obvious reasons.
Não menosprezando qualquer uma das pessoas que me apoiaram ao longo dos anos, gostaria de dedicar
a realização deste trabalho, e portanto o concluir de uma fase importante da minha vida, à minha mãe Maria da
Piedade Mendes, à minha avó Maria Angélica Pato e ao meu avô José Luís Pato, pela sua importância na minha
construção como pessoa, no apoio e no carinho incondicionais. Agradeço também ao meu pai, Acílio Mendes
pelo apoio, pelos conselhos e ensinamentos, que fizeram de mim uma melhor pessoa e um melhor cidadão do
mundo. Obrigado a todos por possibilitarem a minha educação, tanto formal como pessoal. Agradeço também à
restante família e especialmente aos meus irmãos João e Diana, pelas brincadeiras e sorrisos, que me alegraram
nos momentos mais sombrios.
Uma palavra especial de agradecimento para Susana Serra, pelo seu papel fundamental ao longo da
realização deste trabalho e dos últimos anos do curso. As palavra amigas e de encorajamento que me deste para
enfrentar os momentos difíceis, são apenas pequenos vislumbres da enorme pessoa que és, da pessoa que
respeito e amo.
Um agradecimento sentido também para muitos colegas e amigos, sem os quais a conclusão deste
mestrado teria sido com certeza impossível, nomeadamente para os meus companheiros de estágio e amigos de
aventuras João Ribeiro, Tânia Trindade e Pedro Martins, e também para os não menos importantes, Cátia
Palmeiro, Pedro Pereira, João Lisboa, Henrique Escórcio, Rita Teixeira e Júlio Luta.
Finalmente gostaria de agradecer ao professor Pedro Álvares Serrão, pelo apoio e conselho na
realização do presente trabalho, e por me dar a possibilidade de realizar um estágio numa empresa de
reconhecido valor no mundo da aviação. Uma palavra de agradecimento também para todos na PGA – Portugália
Airlines, especialmente para o meu co-orientador Pedro Figueira, pelas ajudas e esclarecimentos, e Marta
Boavida pela simpatia e companheirismo sempre disponíveis.

2
Abstract
The successive economical and financial world crises in the last decade are taking a toll in the sensitive
aviation sector. The increased competition from new low-cost companies and the rising awareness towards the
impact of the aviation industry in the environment, which will in a near future materialize in the EU Emissions
Trading Scheme, are forcing established airline companies to compete by reducing costs through the
optimization of their operations.
The main objective of this master thesis is to evaluate the importance of well-fitted trend monitoring
tools, particularly regarding engine condition and fuel consumption, to the optimization processes companies
want to enforce. PGA – Portugália Airlines, a Portuguese Regional Airline, has provided the means to develop
important tools and to conduct the present analysis.
One objective is reviewing the flight profile for the Fokker 100 fleet, regarding engine life-limited
mandatory parts. While not being able to improve the flight profile, the analysis provides good indicators to
future improvements.
A study is conducted to confirm the value of Engine Condition Monitoring as a useful optimization
tool. Through an ECM tool, COMPASS, the idea is to study the different analyses that can be conducted in terms
of ECM, and see what kind of conclusions could be withdrawn. Several real examples are given and the value of
ECM is confirmed, due to its versatility.
An additional tool was developed to actively monitor the fuel consumption across both PGA fleets,
proving to be invaluable, linking maintenance to flight operations, and thus achieving optimization in both
departments.
Keywords: Flight Profile; Regional Airline; ECM – Engine Condition Monitoring; Turbofan; Fuel
Consumption; Operational Optimization.

3
Resumo
As sucessivas crises económicas e financeiras da última década estão a ter um profundo impacto no
frágil sector aeronáutico. A crescente competição de novas companhias low-cost e a crescente sensibilização
para o impacto ambiental da indústria aeronáutica, estão a forçar empresas de aviação já estabelecidas a
competir, reduzindo custos através da optimização das suas operações.
O principal objectivo desta tese de mestrado é avaliar a importância da utilização de ferramentas de
monitorização bem adaptadas para auxiliar os processos de optimização que as empresas desejam iniciar,
sobretudo em termos da condição de motores e do consumo de combustível. A PGA – Portugália Airlines,
providenciou os meios para desenvolver ferramentas importantes e para realizar o presente estudo.
Um objectivo deste trabalho prende-se com a revisão dos perfis de voo da frota Fokker 100,
relativamente a peças de motor de vida limitada. Embora não tenha melhorado o perfil de voo actual, a análise
resulta em bons indicadores para futuras melhorias.
Um estudo é realizado para confirmar o valor da Monitorização da Condição de Motores como uma útil
ferramenta de optimização. Através da ferramenta computacional COMPASS, pretende-se mostrar as diferentes
análises e conclusões que podem ser retiradas através da monitorização de motores. Vários exemplos reais de
análises são apresentados e o valor da monitorização é confirmada, dada a sua versatilidade.
Uma outra ferramenta computacional foi desenvolvida para monitorar activamente o consumo de
combustível em ambas as frotas, provando ser valiosa, por estabelecer uma ligação entre os departamentos de
manutenção e de operações de voo, optimizando a operação de ambos no processo.
Palavras-chave: Perfis de Voo; Companhia Aérea Regional; Monitorização Condição Motor; Turborreactor
Duplo Fluxo; Consumo de Combustível; Optimização Operacional.

4
Contents
Acknowledgements/Agradecimentos.......................................................................................................................1
Abstract..................................................................................................................................................................................2
Resumo...................................................................................................................................................................................3
Contents.................................................................................................................................................................................4
Index of Figures..................................................................................................................................................................5
Index of Tables....................................................................................................................................................................8
Acronyms Abbreviations and Terms......................................................................................................................10
1 – Objectives - Optimization as a Philosophy....................................................................................................12
2 – Portugália Airlines as a Case-Study..................................................................................................................14
3 – Flight Profile ..............................................................................................................................................................17
3.1 – Importance in Guaranteeing Airworthiness............................................................................17
3.2 – Determining Method..........................................................................................................................19
3.3 – Fleet Statistical Analysis...................................................................................................................21
3.4 – Results......................................................................................................................................................24
3.5 – Change in Thrust Mode.....................................................................................................................30
3.5.1 – Results..................................................................................................................................................30
3.5.2 – Consequences – Cost: Time and Fuel vs. Maintenance ....................................................32
4 – Engine Condition Monitoring..............................................................................................................................35
4.1 - Monitoring as a Route to Safety.....................................................................................................35
4.2 –Engine Monitoring Method and COMPASS................................................................................36
4.3 – Data Input...............................................................................................................................................41
4.4 – Results......................................................................................................................................................50
4.4.1 – Fleet Comparison with ECM........................................................................................................50
4.4.2 – ECM as a Problem Identification Tool – Drop in ITT Margin.........................................54
4.4.3 – ECM as a Pre-emptive Tool - Vibrations High Pressure Shaft.......................................57
4.4.4 – Overhaul or Midlife Influence in Engine Performance.....................................................60
5 – Fuel Monitoring........................................................................................................................................................64
5.1 - Monitoring as a Route to Savings..................................................................................................65
5.2 – A Fuel Monitoring Tool.....................................................................................................................67
5.3 – Acquired Results..................................................................................................................................69
5.3.1 – Comparison between fleets/routes .........................................................................................69
5.3.2 – Engine Offline Washing.................................................................................................................73
6 - Conclusions/Future Work ....................................................................................................................................82

5
References..........................................................................................................................................................................84
Bibliography......................................................................................................................................................................84
Appendix I – ECM Trend Guideline Chart.............................................................................................................86
Appendix II – Embraer ERJ145 and AE3007 Specifications..........................................................................87
Appendix III – Fokker 100 and TAY650-15 Specifications............................................................................88
Appendix IV – IATA and ICAO Codes of Relevant Airports............................................................................89

6
Index of Figures
Figure 2.1 – Fokker 100 (CS-TPD) from PGA Portugália Airlines landing at Schiphol Airport ......................... 14
Figure 2.2 – Embraer 145 (CS-TPI) from PGA Portugália Airlines landing at Brussels Airport ......................... 15
Figure 2.3 – Rolls-Royce TAY 650-15 engine ..................................................................................................... 15
Figure 2.4 – Rolls-Royce Allison AE3007 engine................................................................................................ 16
Figure 3.1 – Plan B datum flight profile set, for low-pressure engine speed (N1) and high-pressure engine speed
(N2), as defined in the RR TLM, Chapter 05-10-01............................................................................................. 18
Figure 3.2 – Flight profile assessment process...................................................................................................... 21
Figure 3.3 – Distribution of flights for each aircraft of the Fokker 100 fleet, for the year 2009........................... 22
Figure 3.4 – Distribution of flights in some routes throughout the Fokker 100 fleet, for the year 2009............... 23
Figure 3.5 a) – Datum flight profiles from Plan A to Plan D and analysis results for N2..................................... 24
Figure 3.5 b) – Datum flight profiles from Plan A to Plan D and analysis results for N2 (detail)........................ 24
Figure 3.6 b) – Datum flight profiles from Plan A to Plan D and analysis results for N2 .................................... 25
Figure 3.6 b) – Datum flight profiles from Plan A to Plan D and analysis results for N2 (detail)........................ 25
Figure 3.7 a) – Evolution of flight profile defining point N1 for Take-off, from 2005 to 2009............................ 27
Figure 3.7 b) – Evolution of flight profile defining point N2 for Take-off, from 2005 to 2009 ........................... 27
Figure 3.8 a) – Evolution of flight profile defining point N1 for Climb, from 2005 to 2009................................ 27
Figure 3.8 b) – Evolution of flight profile defining point N2 for Climb, from 2005 to 2009 ............................... 27
Figure 3.9 a) – Evolution of TGT corresponding to defining point N1 for Take-off, from 2005 to 2009 ............ 28
Figure 3.9 b) – Evolution of TGT corresponding to defining point N1 for Climb, from 2005 to 2009................ 28
Figure 3.10 a) – Evolution of TGT corresponding to defining point N1 for Take-off on TAY650-15 SN17392
engine, from 2005 to 2009 .................................................................................................................................... 29
Figure 3.10 b) – Evolution of TGT corresponding to defining point N1 for Climb on TAY650-15 SN17392
engine, from 2005 to 2009 .................................................................................................................................... 29
Figure 3.11 – Example of a Thrust Mode Select Panel (TMSP), not necessarily the one equipped on the Fokker
100 fleet ................................................................................................................................................................ 30
Figure 3.12 a) –Thrust mode impact in one specific flight, on defining point N1, according to FL..................... 31
Figure 3.12 b) –Thrust mode impact in one specific flight, on defining point N2, according to FL..................... 31
Figure 4.1 – ECM process for both Embraer 145 and Fokker 100 fleets.............................................................. 36
Figure 4.2 – COMPASS ECM software for the TAY650-15 and AE3007 engines, by Rolls-Royce................... 39
Figure 4.3 – Equipment Directory COMPASS tool.............................................................................................. 40
Figure 4.4 – Values for smoothed DFF parameter, from January to February 2010, in Embraer CS-TPI............ 44
Figure 4.5 – Values for smoothed DFF parameter, from January to February 2010, in Fokker CS-TPA............. 44
Figure 4.6 – Values for smoothed DTGT parameter, from January to February 2010, in Fokker CS-TPA ......... 45
Figure 4.7 – FF variation of values for a flight in CS-TPA, on the 26th
February 2010 ....................................... 46
Figure 4.8 – FF variation of values for a flight in CS-TPA, on the 26th
February 2010 (detail)........................... 47
Figure 4.9 – FF results with and without filtering, for CS-TPA flights, in January/February 2010...................... 49

7
Figure 4.10 – FF approximate linear trends the Fokker 100 fleet, from January to August 2010......................... 50
Figure 4.11 – FF approximate linear trends the Embraer 145 fleet, from January to August 2010 ...................... 51
Figure 4.12 – Comparison of the scatter of DN2, DTGT/DITT and DFF parameters between Fokker 100 and
Embraer 145, from January 2010 to July 2010 ..................................................................................................... 51
Figure 4.13 – DFF values and approximate linear trend, for both engines of the CS-TPJ Embraer aircraft and
comparison with fleet average, from January to July 2010................................................................................... 52
Figure 4.14 – DN2 values and approximate 6th
degree polynomial trend, for both engines of the CS-TPJ Embraer
aircraft and comparison with fleet average, from January to July 2010................................................................ 53
Figure 4.15 – DITT values and approximate 6th
degree polynomial trend, for both engines of the CS-TPJ
Embraer aircraft and comparison with fleet average, from January to July 2010................................................. 53
Figure 4.16 – Margin parameters; interpretation guide to the engine limitation depending on SLOATL ............ 55
Figure 4.17 – Variation of smoothed SL N2 margin parameter, for the Embraer CS-TPJ, in August 2010......... 55
Figure 4.18 – Variation of smoothed SL ITT margin parameter, for the Embraer CS-TPJ, in August 2010........ 55
Figure 4.19 – Boroscope inspection to high pressure turbine blades in the SN311088 AE3007 engine .............. 56
Figure 4.20 – Boroscope inspection to high pressure turbine blades in the SN311088 AE3007 engine (detail).. 57
Figure 4.21 – HP shaft vibration values and approximate linear trend, during Takeoff in CS-TPM.................... 58
Figure 4.22 – HP shaft vibration values and approximate linear trend, during Cruise in CS-TPM...................... 58
Figure 4.23 – DFF variation with engine change in CS-TPE Fokker 100 aircraft, in 29/04/2010........................ 60
Figure 4.24 – DN2 variation with engine change in CS-TPE Fokker 100 aircraft, in 29/04/2010 ....................... 60
Figure 4.25 – DTGT variation with engine change in CS-TPE Fokker 100 aircraft, in 29/04/2010 .................... 61
Figure 4.26 – DN1 values before and after overhaul of engine SN17317; approximate logarithmic trend for post-
overhaul engine operation, from February 2006 to August 2007.......................................................................... 63
Figure 4.27 – TGT values for critical N1 and approximate logarithmic trend, from 2006 to 2009 ...................... 63
Figure 5.1 – Evolution of kerosene prices, in US cents per US gallon, from June 1986 to June 2008 (analysis in
August 2007) [7]................................................................................................................................................... 65
Figure 5.2 a) – Form1, form presented to users of the fuel monitoring tool, to introduce analysis criteria.......... 67
Figure 5.2 b) – Form1, example of correctly introduced analysis criteria ............................................................ 67
Figure 5.3 – Output of the fuel monitoring tool, corresponding to the criteria of Figure 5.2................................ 68
Figure 5.4 – Normal distributions of the DFC metric, in the LIS-OPO route, for the Fokker 100 fleet, from
January to July 2010 ............................................................................................................................................. 69
Figure 5.5 – Normal distributions of the DFC metric, in the LIS-OPO route, for the Embraer 145 fleet, from
January to July 2010 ............................................................................................................................................. 69
Figure 5.6 – Normal distributions of the DFC metric, in the OPO-BRU route, for the Embraer 145 fleet, from
January to July 2010 ............................................................................................................................................. 71
Figure 5.7 – Passenger efficiency, i.e. fuel spent per passenger, for both fleets and several routes, with a 15-day
period per average point, from January to July 2010............................................................................................ 72
Figure 5.8 – Time-based fuel efficiency for several routes for the Fokker 100 aircraft........................................ 73
Figure 5.9 – Example of an action of offline engine washing, not necessarily the exact equipment/method used in
PGA’s washes ....................................................................................................................................................... 74
Figure 5.10 – Dust removal from pre-wash (on the left) to post-wash (on the right) from turbine blades ........... 75

8
Figure 5.11 – Effect of engine wash on DTGT, for the TAY650-15 SN17277 and SN17276 engines ................ 77
Figure 5.12 – Effect of engine wash on smoothed DFF, for the TAY650-15 SN17277, SN17276 and SN17318
engines, in comparison with the non-washed SN17317 engine............................................................................ 78
Figure 5.13 – Variation of TFC with engine wash, for CS-TPA and CS-TPB Fokker 100 aircraft, in the LIS-OPO
route, from January to July 2010........................................................................................................................... 78
Figure 5.14 – Variation of DFC with engine wash, for CS-TPA and CS-TPB Fokker 100 aircraft, in the ...... LIS-
OPO route, from January to July 2010.................................................................................................................. 78

9
Index of Tables
Table 3.1 – Lives of some engine life-limited parts, depending on the current Flight Profile.............................. 18
Table 3.2 – Yearly results of N1 and N2 values, for each flight phase, from 2005 to 2009; flight profile limits
and global results .................................................................................................................................................. 26
Table 3.3 – TGT values corresponding to the maximum values of N1 and N2 for Climb and Takeoff, from 2005
to 2009 .................................................................................................................................................................. 28
Table 3.4 – Obtained results and improvement in Flight Profile after change in Thrust Mode............................ 32
Table 4.1 – Entry and exit conditions used to define the representative take-off points (RTOP)......................... 37
Table 4.2 – Entry and exit conditions used to define the representative cruise points (RCP)............................... 38
Table 4.3 – Part of the trend guideline chart presented on APPENDIX I............................................................. 43
Table 4.4 – Compilation of FF results with and without filtering and respective improvement........................... 49
Table 5.1 – Typical operational profile and PGA example for the Embraer 145 aircraft; savings analysis.......... 66
Table 5.2 – Compilation of Normal Distributions results for the DFF, DN2 and DTGT parameters, before and
after wash, and the correspond absolute and relative improvement...................................................................... 76

10
Acronyms Abbreviations and Terms
APU – Auxiliary Power Unit
AI – Absolute Improvement
BH – Block Hours
CDU – Control-Display Unit
CG – Centre of Gravity
CMC – Central Maintenance Computer
Critical Points – Moments when the N1 or N2 values are maximized for the take-off and climb flight phase,
and are used for the determination of a flight profile.
DFC – Distance-based Fuel Consumption
DME – Direcção de Manutenção e Engenharia – Maintenance and Engineering Department
ECM – Engine Condition Monitoring
E&M – Engineering and Maintenance
EASA – European Aviation Safety Agency
ETA – Estimated Time on Arrival
ETS – Emissions Trading System
EU – European Union
FAA – Federal Aviation Administration
FC – Fuel Consumption
FDR – Flight Data Recorder
FH – Flight Hours
FL – Flight Level
FMS – Flight Management System
FP – Flight Profile
Gal – Gallon
GPU – Ground Power Unit
H/h – Hour
HPT – High Pressure Turbine
IATA – International Air Transport Association
ICAO – International Civil Aviation Organization
IPCC – International Panel for Climate Change
ips – Inches per second
ISA – International Standard Atmosphere
ITT – Interstage Turbine Temperature
kg – Kilogramme – Base unit of mass in SI
l – Litre
lb – Pound
M – Mach Number
MSL – Mean Sea Level
MAXTOW – Maximum Take-Off Weight

11
N1 – Ratio between actual and maximum rotational speed of the low-pressure shaft
N2 – Ratio between actual and maximum rotational speed of the high-pressure shaft
OAT – Outside Air Temperature
OEW – Operational Empty Weight
Pax – The same as passengers
PGA – Portugália Airlines
RCP – Representative Cruise Point
R&D – Research and Development
RI – Relative Improvement
RJ – Regional Jet, which for the purpose of this work, is assumed to be an aircraft capable
of flying up to medium-haul routes, carrying no more than 100 passengers.
RP – Representative Point
RR – Rolls Royce
RTOP – Representative Take-off Point
s – Second, the base unit of time in SI
SI – International System of Units – “Système International d’Unités”
SL – Sea Level
TAP Portugal – Major Portuguese Airline
TFC – Time-based Fuel Consumption
TGT – Turbine Gas Temperature, a temperature measured at the first stage of the low
pressure turbine nozzle guide vanes
TMSP – Thrust Mode Switch Panel
TOC – Top of Climb
TOD – Top of Descent
TOW – Take-Off Weight
TPA TPF – References to the tail numbers of the Fokker F28 Mk 100 aircraft that constitute one of
PGA’s fleets, namely CS-TPA, CS-TPB, CS-TPC, CS-TPD, CS-TPE, CS-TPF
TPG TPN – References to the tail numbers of the Embraer ERJ145 aircraft that constitute one of PGA’s
fleets, namely CS-TPG, CS-TPH, CS-TPI, CS-TPJ, CS-TPK, CS-TPL, CS-TPM, CS-TPN
TSFC – Thrust Specific Fuel Consumption
TP – Turbo Props
US – United States

12
1 – Objectives - Optimization as a Philosophy
It is a global world. In great part thanks to the aviation industry. And now, that global world commercial
aviation has so proudly created has become an impossible dichotomy: it is both its livelihood and its worst
adversary. The recent global crisis, the effort to decelerate global warming to avoid a global catastrophe, and
rampage fierce global competition between airlines while the passenger increases only feebly are leaving their
mark in the aviation industry, with bankruptcies being almost commonplace. All these factors, associated with
the introduction in the last few years of a number of new low-cost, low-service airline companies, have forced
established airlines to adapt and find new ways to compete, by focusing or widening their operation, through
undesirable personnel downsizes, with mergers between some of the largest companies and company
acquisitions, and naturally by trying to reduce costs in all departments and aspects of operation.
The aeronautical industry is probably the one industry that has been suffering the most with the constant
increase of international oil prices in the last decade. Fuel has always represented a significant part of the airline
companies’ expenses, but now, more than ever, operators have started to take action and make investments in
order to implement preventive and corrective measures, with the objective of reducing unnecessary fuel
consumption.
The environmental impact associated with the burn of fossil fuels is also becoming an increasing factor
of concern to the aviation industry. According to the IPCC, in the last ten years, the number of passengers in
commercial scheduled flights has increased about sixty percent, and even with oscillations due to economical
crisis, the prediction is that it will continue to grow at an average rate of five percent per year for the next ten to
fifteen years. Although aviation only contributes to the production of green house gases by two or three percent,
there is a will to maintain these contributions to the minimum possible, and some airline companies are already
taking some measures in that sense. Presently, the possibility of savings through reduction in fuel consumption is
only focused on the cost of the fuel itself. However, in a near future operators will also have to support,
proportional to the fuel consumption, the additional cost associated with the mandatory compensation for
resulting carbon dioxide emissions, as a consequence of the implementation of the ETS in the EU. This
perspective is giving companies one more reason to optimize their engines’ and operations’ efficiency, which
should result in less fuel consumption and less pollutant emissions.
Besides the investments in equipment and procedures which have the objective of reducing fuel
consumption, there is an increasing bet in systems to monitor the aircraft’s fuel consumption. These systems
allow airline companies to follow the evolution of fuel consumption and quantify the effects of the measures
undertaken within projects of fuel consumption reduction. Additionally, in order to achieve significant
improvement regarding fuel-costs, it is important to maintain a close monitoring on the degradation of each fleet,
so that the consumption of each single flight can be optimized.

13
The monitoring of aircraft and engines consists in a continuous compilation process of flight data,
which is then analysed in different ways to assess the level of degradation and performance at that moment.
Particularly for engines, the continuous following of an engine’s health can allow significant savings in several
ways. Engine Condition Monitoring (ECM) is becoming used on a daily basis in many airline companies, for
they have understood the potential savings that can be achieved, in fuel, parts and maintenance actions. ECM
continuously assesses the health of an engine, and therefore the performance it will present in terms of fuel
consumption. If this information could be complemented with a Fuel Monitoring tool that would compile and
analyse all data regarding routes, flight time, aircraft weight, number of passengers, take-off weight (TOW), etc,
the operator could establish a direct correlation between every engine maintenance action and fuel consumption,
assessing their cost-effectiveness and allowing to define long term strategies to reduce costs.
PGA – Portugália Airlines, a Portuguese Regional Airline is one of those companies that have grasped
the importance of ECM and fuel monitoring as tools to optimize its operation and reduce costs all around. The
present work is the result of one of the first steps towards ECM full implementation process.
The initial objective was to review the flight profile for the Fokker 100 fleet, regarding engine life-
limited mandatory parts, in order to assess if there could be an improvement in the fleet’s flight profile. Such an
improvement would translate into increased lives for critical engine parts, which would reflect in significant
changes for the company. That analysis has been performed and achieved good results, with the process and
results being thoroughly presented in Chapter 3. Also in this analysis there was an assessment of engine
condition, but in this case specifically concerning life-limited parts.
It was concluded that a broader study should be conducted to confirm the value of ECM as a versatile
optimization tool. An engine trend monitoring tool, COMPASS, which was supplied by the engines’
manufacturer Rolls-Royce, should be implemented at 100% although it was already used by PGA’s DEM
(M&E). Furthermore, a pc-tool was to be developed to facilitate and automatize the process of introduction of
adequate flight data into COMPASS. After uploading the necessary information, through COMPASS the
objective was to study the possibilities of conducting different analysis in terms of ECM, and see what kind of
conclusions could be withdrawn from such analysis. This process is described in Chapter 4 of this work, where
ECM and COMPASS software are properly presented, several real examples of analysis where ECM can be
invaluable are given and conclusions are withdrawn regarding the potential and capabilities of ECM
implementation into PGA’s operations.
Chapter 5 of this thesis will be about the introduction to a simple fuel monitoring tool created by the
author, whose objective is to create a bridge between flight operations and maintenance practices, and
furthermore, a direct link between maintenance practices and fuel consumption savings. The tool will be briefly
explained and its capabilities presented, in order to show the different conclusions that can be withdrawn from
the statistical analysis on fuel consumption, from different viewpoints and considering different variables. This
tool will be used together with COMPASS to assess whether offline engine washing is cost-efficient for the
PGA’s particular case.

14
2 – Portugália Airlines as a Case-Study
PGA Portugália Airlines is a Portuguese Regional Airline company based at Lisbon International
Airport operating scheduled international and domestic routes from Lisbon and Oporto. While officially
established on the 25th July 1988, it only began its operation two years later because of a delay on the
liberalization of the Portuguese commercial aviation scenario. The first flight was from Lisbon to Oporto,
followed by the immediate integration of other domestic destinations. In June 1992 PGA flew for the first time
an international route, starting to operate European routes to Strasbourg, Cologne and Turin. As the destinations
were becoming more varied and the demand was rising, PGA continuously increased its Fokker 100 fleet, which
was completed in 1993, acquiring then from 1997 to 2000 eight Embraer ERJ-145 to satisfy the demand imposed
by several new destinations in Spain.
On June 2007 PGA was acquired by the Portuguese national carrier TAP Portugal, marking the
beginning of a new era in the company. Though there is a strict business relationship between the two
companies, PGA kept its own flight crews, maintenance personnel, human resources and engineering
department, but in this new phase the company provides a service to TAP by covering TAP’s necessities in the
short-haul domestic, Iberian and European scenarios
PGA has imposed itself as a reference in the European market, marked by its innovative spirit and
excellence in customer service. An effective management of its operations, an enthusiastic team and a
commitment towards quality and safety have granted PGA international acknowledgement through numerous
awards, namely “Best European Regional Airliner” for six years in a row.
Figure 2.1 - Fokker 100 (CS-TPD) (source http://www.flickr.com)
PGA currently possesses two fleets, the Fokker 100 (Figure 2.1) and the Embraer ERJ-145 (Figure 2.2),
with cabin layouts of 97 and 49 passengers respectively. Although both aircraft models fly to most destinations
according to demand, the Fokker 100 is more suited for the medium-haul routes to destinations in France, Italy,
Switzerland, Holland and Belgium, while the Embraer 145 is more efficient covering Iberian destinations. Both

15
models have been continuously updated in terms of navigation and safety systems, allowing both PGA fleets to
keep up with the increasing demands from national and international regulators.
Figure 2.2 - Embraer 145 (CS-TPI) (source http://www.flickr.com)
In a new phase of its operation, and in a particularly difficult international economical scenario,
Portugália Airlines is looking for ways of optimizing its operations, in order to reduce costs and maintain its
valid contribution as an asset within the TAP group. Because maintenance costs and fuel costs constitute a very
large part of the yearly total expenses, the validation of practices to reduce costs with engine parts, with
maintenance inefficiency and with fuel consumption, was important for PGA and a challenge presented to the
author, who accepted it through practical development of optimization tools and then wrote a compilation of the
obtained results in the form of a master thesis. All the data analyzed in this thesis is collected from PGA’s FDR
and CMC records, and so the realization of this thesis would be impossible without PGA’s support.
Figure 2.3 - Rolls-Royce TAY 650-15 engine (source http://commons.wikimedia.org)
Because the engines present in PGA’s fleets will be the focus of this work, they will be now briefly
presented. The Tay 650-15 engine that powers the Fokker 100 fleet is an axial flow, by-pass engine with two
compressor spools. The Rolls-Royce Tay is the combination of two highly successful engines. The high pressure
system comes from the RB183-555 engine of the Fokker F28 and the low pressure system comes from the
RB211-53E4 engine. The combination of reliability, low fuel burn and low noise makes the Tay engine family
very popular in RJs. The biggest advantage of the 650 relatively to the other TAY models was the better
performance at higher altitudes and a better climb rate. The Tay 650 on the Fokker 100 aircraft provides

16
increased maximum thrust for take-off, climb and cruise, plus efficiency improvements through small increases
in fan diameter and an advanced high-pressure turbine.
Figure 2.4 - Rolls-Royce Allison AE3007 engine (source http://www.fly-corporate.com)
The Allison AE3007 is the powerplant of choice for the Embraer 145 and has been a fuel efficiency
leader since its introduction in 1995, making it the greenest and one of the quietest engines in its class. The
engine produces 7.200 pounds of thrust at sea level and has been certified at an altitude of 51.000 feet. The
AE3007 turbofan core is derived from the AE1107 turboprop engine and it was developed to provide a turbofan
member of the AE common core family for the growing regional jet and medium/large business jet markets.
Designed with excellent reliability, maintainability and performance in mind, the capability and versatility of the
AE 3007 turbofan is demonstrated by its use in regional, corporate and military applications. Over the first seven
years of operation on Embraer’s 37 to 50 seat Regional Jets the AE 3007A achieved over 10 million hours
service experience, powering over 750 Embraer deliveries.

17
3 – Flight Profile
3.1 – Importance in Guaranteeing Airworthiness
The type of record and the level of control applied over its life vary according to the part, depending on
its importance to the engine’s performance, condition and ultimately to the aircraft’s airworthiness. For this
purpose, parts are divided into Groups A – Critical Parts, B – Sensitive Parts and C – Unclassified Parts.
Furthermore, Group B parts are divided into mandatory parts and non-mandatory parts. The definition of the
engine’s flight profile is associated with Group A and mandatory Group B parts, therefore only these will be
considered henceforth.
The failure of a critical part in an engine can have serious consequences. These parts are together
termed Group A parts: a mandatory (not to be exceeded) operating life is declared for each of them. Group A
parts comprise those major rotating parts of an engine the failure of which could, if uncontained, affect the
airworthiness of the aircraft. They are the only parts of the engine for which an Airworthiness Authority certified
life is required and notification of the lives must be made to each aircraft operator. These parts are thus subjected
to strict life limitations which must not be exceeded in service. These limitations are made as a result of
experience collected by the manufacturer through cyclic ring testing, development engine testing, metallurgical
investigation and other techniques, including sampling of service run parts. The service life achieved on a part
may prove to be less than that permitted by the manufacturer for airworthiness consideration. For example,
corrosion or wear may result in rejection of the part at an earlier life than the quoted limitation.
Mandatory Group B parts are the ones which test and service experience have shown require special
attention in service, in order to avoid failures which:
May be serious in terms of engine disruption but are classed as considerably less significant than
the possible consequences of a Group A part failure;
Though not hazardous in their own right, may threaten the integrity of a Group A part.
The lives of Group A and mandatory group B parts are expressed in terms of Flight Cycles. This is the
most direct and accurate way of controlling the service life or parts. The life is calculated from the rate of fatigue
life usage per datum flight. A flight cycle can be defined as a normal take-off to landing cycle or a “touch and
go” situation. In Table 3.1 some examples of engine life-limited parts can be found, as well as each part life
limitation cycle-wise, depending on which flight profile the fleet has. For these particular parts, an improvement
from Plan B to Plan A would result in a life increase of 10%, therefore about 10% of cost reduction with limited-
life parts. Taking into account the large number of parts (especially blades) and the elevated cost of each of those
parts, it’s easy to understand the operator’s will to upgrade its flight profile.

18
Part Description Group Plan A Plan B Plan C Plan D
LP compressor rotor discs assembly A 23 000 21 000 20 000 20 000
HP compressor rotor disc stage 2 A 22 000 20 000 20 000 20 000
HP turbine shaft and seal assembly A 23 000 21 000 21 000 20 000
HP turbine rotor disc stage 1 A 23 000 21 000 18 000 14 250
LP compressor blade B 22 000 20 000 20 000 20 000
T able 3.1 - Lives of some engine life-limited parts, depending on the current Flight Profile
A Flight Profile is a graphical representation of the actual service operations over an entire fleet and
route structure and the operator is responsible for its determination. Flight profiles are classed from A to D, A
being the best flight profile, associated with the largest life-limits and D the worst a fleet can present in terms of
life-limits. At regular intervals, as agreed between the operator and his local airworthiness authority, usually
annually, the operator must review the fleet representative flight profile data and send it to the manufacturer for
analysis. If the operator thinks that his operation is no longer represented by the reference datum flight profile,
the changed fleet representative flight profile data must be supplied to the manufacturer. This was the case of
Portugália Airlines.
Figure 3.1 - Plan B datum flight profile set for N1 and N2
In the last couple of years, PGA has increased its effort to stride forward when optimization is
concerned, always trying to find new ways of reducing costs without compromising flight safety or passenger
comfort. Maintenance is responsible for a very significant part of the company’s global annual expenses, in part
because of the inevitability of replacing very expensive life-limited critical parts. The possibility of extending the
use of those parts is therefore very appealing and worth looking into. In the last years measures have been taken
to reduce engine wear, both in the Flight Operations department, where pilots have been given strict instructions
to reduce power when possible, namely in take-off and climb and also in the Maintenance department, where
trend-monitoring and frequent inspections and overhauls keep the engines in the best condition possible. The
enforcing of these measures gave PGA the confidence to assess if they could improve its long-lasting Flight
Profile “B” through a more detailed analysis from FDR data.

19
The most common methods used to collect data on flight profiles are manual cockpit flight data
recording or automatic cockpit flight data recording (using an FDR or DFDR), from which the latter is preferred
in this particular case, because it’s more practical, more accurate and reduces the pilot’s workload. The operator
must make sure that the recorded operational flight data contains all areas in which the datum flight profile could
be more than the limit. An example of a flight profile reference datum is displayed in Figure 3.1 and although the
whole flight is represented, in most cases the most extreme points are sufficient to attribute a fleet’s flight
profile, through comparison with the reference data flight profiles. The minimum required parameters are:
Maximum speed of each rotor (N1, N2) during the take-off phase
Maximum speed of each rotor (N1, N2) during the climb phase.
To find out if it agrees with a datum flight profile, the recorded peak rotor speeds must be averaged over
a twelve month period. The average of the recorded peak rotor speed must be less than or equal to the
corresponding value in the reference flight profile, as given in the RR Time Limits Manual, Chapter 05-00-02.
In association with the maximum rotor speeds, other information should be sent for further analysis
such as TGT, FF, flight conditions at the moment like Mach, TAT and altitude, not to mention the hour and date
when the maximum values appeared. In terms of data representativeness, some cautions must be taken when
choosing the data that will be processed to result in a flight profile classification, namely for the case study:
For a fleet size of 6 aircraft, all aircraft should be sampled, as evenly as possible and the minimum
total number of flights to be recorded per twelve month period is 60 flights.
The data collected must be representative of the total operation. Samples are to be taken throughout
the year, which contain the complete route structure and must include any extremes of the
operation, which may affect engine speed.
3.2 – Determining Method
In the previous sub-chapter, the definition of a flight profile was given, conditions were imposed and
the objective was set. In this section it will be described how the analysis was exactly conducted, to determine
the fleet’s current flight profile, while making an effort to provide the most accurate and representative results
possible in relatively short time.
The Flight Data Recorder, as previously stated, was the selected source of information from which the
flight profile would be assessed. FDR data is easy and practical to obtain, making it possible to have records
from every aircraft, throughout the year and in any conditions. Also, apart from any FDR or sensor malfunctions,
FDR data is easy to work with and usually accurate and reliable. PGA has a relatively large FDR data archive,
which is the result of continuous data downloads and archiving throughout the last years. To provide with an

20
accurate and complete flight profile assessment, it was determined that analysing flight profile status and FDR
records from 2005 onwards would suffice.
Firstly, because the flight profile analysis is made in an annual basis, the recorded files were processed
and analysed by years, namely 2005, 2006, 2007, 2008 and 2009. The year of 2010 wasn’t considered because
this analysis was concluded in May 2010 and the data acquired until that point wasn’t representative enough of
the year 2010, to include in an annual-based analysis. According to the requirements presented before, a 6-
aircraft fleet should present at least 60 FC to correctly determine the fleet’s flight profile, however to achieve the
most accurate result possible, more flights were analysed, as calculated below (approximate values):
Total ൬
FC
year
൰ ൌ 25 ൬
FC
aircraft ൈ quarter
൰ ൈ 4ሺquarterሻ ൈ 6ሺaircraftሻ ൌ 600 ൬
FC
year
൰
Equation 3.1
About 25 flights were considered per aircraft per quarter, which results in about 600 flights per year and
about 3000 FC for total analysed flights. The records were divided in quarters to assure that all flight and
operational conditions were covered, but after that, for a given quarter, the choosing process of the FCs was
completely random and without any criteria, with the objective of not adulterating the result.
The determination of the flight profile is represented in Figure 3.2 and commences with the FDR pre-
selected analysis files, which have to be introduced into a software program that changes the .DLU files
downloaded from the FDR into ASCII (.txt for example) files. This tool also selects which parameters are
important for the upcoming analysis. In this case the following parameters are: Aircraft, Date/Time, Flight
Phase, Pressure Altitude, Airspeed, Groundspeed, TAT and for each engine TGT, EPR, N1 and N2.
The process continues by processing the .txt files with a program developed in Visual Basic for
Applications language under Microsoft Access, which detects all the files to be processed in a user-selected
folder. Only files of the same year should be processed together. Afterwards, some filtering has to be done in
order to eliminate sensor reading errors and other data spikes and discrepancies. The maximum high-pressure
and low-pressure rotor speeds (N1 and N2 respectively) are obtained, and the points where these maximum
values take place become Critical Points, for they correspond to the moments when engine operation is the most
extreme. All information associated with these Critical Points is recorded into a Microsoft Excel sheet for further
processing.

Figure
The result of the previous steps for any given year is two Excel Workbooks, one for the Take
and another for the Climb phase, each of them with two worksheets, one with the Critical Points information
when N1 was maximum and the other with the Critical Points data when N2 was maximum. The last step
consists on making a simple average of the low
each worksheet, in other words, an average per speed, per flight phase, per year. If more tha
analysed, like in the present case, a final simple or weighed average
phase to achieve one final flight profile.
so they can analyse these results and attribute a flight profile classification from A to D.
description of this final process will be presented in
3.3 – Fleet Statistical Analysis
The concept of one flight profile being able to describe an entire fleet can be puzzling.
a couple of numbers be sufficient to represent the condition of a fle
in different moments in terms of their maintenance program, whose engines have just been overhauled or haven’t
been refurbished for a while? In the flight operations department, with the multitude of destination
is the flight profile a trustworthy representation of all the different routes? Is this over
In this section, these questions will be answered,
Figure 3.2 - Flight profile assessment process
The result of the previous steps for any given year is two Excel Workbooks, one for the Take
b phase, each of them with two worksheets, one with the Critical Points information
consists on making a simple average of the low-pressure rotor speed N1 and high-pressure rotor speed N2
each worksheet, in other words, an average per speed, per flight phase, per year. If more tha
, a final simple or weighed average should be done for each parameter and fli
phase to achieve one final flight profile. Finally the results and data should be sent to the engine’s manufacturer,
so they can analyse these results and attribute a flight profile classification from A to D.
process will be presented in section 3.4.
Fleet Statistical Analysis
The concept of one flight profile being able to describe an entire fleet can be puzzling.
a couple of numbers be sufficient to represent the condition of a fleet, whose planes have different ages,
In the flight operations department, with the multitude of destination
is the flight profile a trustworthy representation of all the different routes? Is this over-simplifying the problem?
will be answered, by pointing out what requirements the fleet must fill,
21
The result of the previous steps for any given year is two Excel Workbooks, one for the Take-off phase
b phase, each of them with two worksheets, one with the Critical Points information
pressure rotor speed N2 in
each worksheet, in other words, an average per speed, per flight phase, per year. If more than one year is
should be done for each parameter and flight
Finally the results and data should be sent to the engine’s manufacturer,
so they can analyse these results and attribute a flight profile classification from A to D. A more detailed
The concept of one flight profile being able to describe an entire fleet can be puzzling. How can literally
et, whose planes have different ages, can be
In the flight operations department, with the multitude of destinations PGA flies to,
simplifying the problem?
what requirements the fleet must fill, by

explaining the concept from a statistical
PGA’s Fokker 100 fleet.
There are various explanations for the valid questions
related with the aircraft’s age. PGA’s Fokker 100 fleet has several characteristics that help validating the use of a
single flight profile per fleet:
The aircraft all have roughly the same age, the maximum discrepancy being
which in theory would result in similar
when the maintenance program is concerned
There is an effort to equally utilize all aircraft, as can be seen in Figure 3.3, so they also have
similar behaviour in terms of flight
If scheduled inspections are strictly carried out and engine trend
persistently enforced, the engines shoul
immediately detected, investiga
Some fleets from other airline companies won’t be able to present t
probably have more difficulty in justifying the single flight profile for that fleet.
Figure 3.3 - Distribution throughout the Fokker 100
In terms of flight operations,
flight profile. Different destinations result in different climates, runway lengths and altitudes,
altitudes and noise-related directives. Also longer routes mean more fuel and thus more weight, which forces the
engine rotor speeds to increase, degrading the associated flight profile. The same weight increase happens in
busier routes. Flight profile assessment
these variations, while unavoidable, statistically speaking will be averaged and in consequence down
although this aspect must be accounted for in the final stages of the pro
this particular case, it’s easy to see that PGA’s fleet won’t be too affected in terms of flight
of its regional operation status. PGA’s longest route is only about 800 NM, most destinations ar
16,9%
17,1%
16,4%
Flight Distribution in Fokker 100 Fleet
e concept from a statistical perspective to ultimately justify the use of a single flight profile for
There are various explanations for the valid questions raised above and one of the most importa
PGA’s Fokker 100 fleet has several characteristics that help validating the use of a
The aircraft all have roughly the same age, the maximum discrepancy being
in theory would result in similar behaviour and even a relative closeness between aircraft
aintenance program is concerned
in terms of flight-cycle maintenance
f scheduled inspections are strictly carried out and engine trend-monitoring is actively and
persistently enforced, the engines should all be in very good condition, with small deviations being
immediately detected, investigated and when possible corrected
Some fleets from other airline companies won’t be able to present these argument
probably have more difficulty in justifying the single flight profile for that fleet.
Distribution throughout the Fokker 100 fleet for the year 2009
tions, different routes imply numerous differences that will affect the calculated
flight profile. Different destinations result in different climates, runway lengths and altitudes,
related directives. Also longer routes mean more fuel and thus more weight, which forces the
Flight profile assessment must obviously consider all routes, despite their length or occupancy and
these variations, while unavoidable, statistically speaking will be averaged and in consequence down
although this aspect must be accounted for in the final stages of the process. Although these differences exist, in
this particular case, it’s easy to see that PGA’s fleet won’t be too affected in terms of flight
PGA’s longest route is only about 800 NM, most destinations ar
16,5%
16,2%
16,8%
16,4%
Flight Distribution in Fokker 100 Fleet
TPA
TPB
TPC
TPD
TPE
TPF
22
stify the use of a single flight profile for
above and one of the most important is
PGA’s Fokker 100 fleet has several characteristics that help validating the use of a
The aircraft all have roughly the same age, the maximum discrepancy being around two years,
behaviour and even a relative closeness between aircraft
monitoring is actively and
, with small deviations being
arguments and therefore, will
the year 2009
imply numerous differences that will affect the calculated
flight profile. Different destinations result in different climates, runway lengths and altitudes, obstacle clearing
related directives. Also longer routes mean more fuel and thus more weight, which forces the
must obviously consider all routes, despite their length or occupancy and
these variations, while unavoidable, statistically speaking will be averaged and in consequence down-rated,
cess. Although these differences exist, in
this particular case, it’s easy to see that PGA’s fleet won’t be too affected in terms of flight-operations, because
PGA’s longest route is only about 800 NM, most destinations are in mild

Western Europe and most destination runways are almost at sea
differences. Furthermore, long runways allow PGA’s relatively small aircraft to take
engine power, which improves operational flight profile. Also in this area the
concept makes sense and can be applied as long as the existing variations in flight operations are not forgotten
and are included in the final safety margin
After understanding that statistically
which present relatively mild climates, we mustn’t forget that with statistical averages comes an error or
deviation, and that the worst-case deviation is the
one specific aircraft does almost exclusively
all year, especially in the summer, from Lisbon to these tropical locations, Madeira and Casablanca
with high humidity, very high temperatures
aircraft engines, resulting in worse flight profile results than the rest of the fleet. In this situation, the lif
critical parts of these engines, while part of the fleet, aren’t expected to last as long as the rest, so the fleet’s
flight profile doesn’t represent these engines
profile is to expose all engines to all operating conditions.
routes (thus not exposing the engine to differing route operating conditions), an individual operational flight
profile monitoring must be made for the group of engines operat
fleet. PGA has taken this into account and, also for maintenance reasons, tries to assign each plane to cover all
routes in order to average engine and aircraft behaviour. This effort can be visualised
the most frequent flown routes are displayed because of their importance from a statistical perspective, but the
rest of the routes also follow the same distribution model
Figure 3.4 - Distribution of flights in some routes through t
0,0
5,0
10,0
15,0
20,0
25,0
LIS-OPO LIS-MAD
Flights(%)
Flight Distribution per Route
Western Europe and most destination runways are almost at sea-level, which reduces some of the referred
differences. Furthermore, long runways allow PGA’s relatively small aircraft to take-off with
perational flight profile. Also in this area the single
margin.
that statistically almost all of PGA flights are made for European destinations,
case deviation is the one we should worry about. As an example, let us imagine that
one specific aircraft does almost exclusively four routes: LIS-FNC, LIS-PXO, LIS-CMN
summer, from Lisbon to these tropical locations, Madeira and Casablanca
very high temperatures and over large stretches of ocean, puts additional stress on the
aircraft engines, resulting in worse flight profile results than the rest of the fleet. In this situation, the lif
, while part of the fleet, aren’t expected to last as long as the rest, so the fleet’s
ese engines properly. This is why one of the requirements for a
all operating conditions. If the operator uses dedicated aircraft on individual
profile monitoring must be made for the group of engines operating a common route as though it were a single
PGA has taken this into account and, also for maintenance reasons, tries to assign each plane to cover all
routes in order to average engine and aircraft behaviour. This effort can be visualised in Figur
rest of the routes also follow the same distribution model.
Distribution of flights in some routes through the Fokker 100 fleet for
MAD LIS-BCN OPO-LGW LIS-LYS OPO-LUX OPO-AMS OPO-FCO
Routes
Flight Distribution per Route
23
level, which reduces some of the referred
off without requiring full
single flight profile per fleet
are made for European destinations,
As an example, let us imagine that
CMN and LIS-CAS. Flying
summer, from Lisbon to these tropical locations, Madeira and Casablanca respectively,
puts additional stress on the
aircraft engines, resulting in worse flight profile results than the rest of the fleet. In this situation, the life-limited
, while part of the fleet, aren’t expected to last as long as the rest, so the fleet’s
properly. This is why one of the requirements for a single flight
If the operator uses dedicated aircraft on individual
ing a common route as though it were a single
PGA has taken this into account and, also for maintenance reasons, tries to assign each plane to cover all
in Figure 3.4, where only
fleet for the year 2009
FCO
TPA
TPB
TPC
TPD
TPE
TPF

24
In conclusion, the fleet’s age, the up-close and consistent monitoring of engine behaviour, the
compliance of mandatory inspections and other maintenance tasks, associated with pilot standardized
procedures, flight operations optimization and even the airline’s own regional status, justify the use of only one
Flight Profile to accurately describe PGA’s Fokker 100 fleet, when critical engine parts are concerned.
3.4 – Results
The process itself to obtain the fleet’s flight profile, although simple, was a bit time-consuming, which
was expected given the number of flights involved, spread throughout five years. However the normal annual
flight profile analysis should be relatively simple and quick to complete, depending on the used sample.
As described before and illustrated in Figure 3.1, flight profile reference data is given for all the
duration of an example flight, however in a first analysis only the points where the rotor speeds reach maximum
values are of interest to compare with these reference profiles and attribute a flight profile to the fleet, because
these are the operational Critical Points. Only if these Critical Points prove to be insufficient to provide a clear
flight profile result, should other points from other moments of the flight be considered.
Figure 3.5 a) and b) - Datum flight profiles from Plan A to Plan D and analysis results for N1
20
30
40
50
60
70
80
90
100
0 10 20 30 40 50 60
N1(%)
Minutes
N1 Flight Profile Data
A
B
C
D
AVG
75
77
79
81
83
85
87
89
91
93
0 5 10 15 20 25 30
N1(%)
Minutes
N1 Flight Profile Data Detail
A
B
C
D
AVG

25
As explained in section 3.1, flight profile A is the most demanding, in the sense that the maximum rotor
speeds should be the lowest, which means that engine operation must be very smooth. To easily grasp the
differences between profiles at every flight stage, in Figure 3.5 and 3.6 the reference flight profiles are
graphically displayed. PGA’s obtained Critical Points for the conducted analysis are also represented in the
Graphs, to give an idea of the fleet’s situation and to make predictions about the manufacturer’s future flight
profile attribution.
Analysing these graphical representations, particularly in Figure 3.5 b), where the differences between
reference profiles at high rotation speeds can be easily seen, it becomes clear the growing level of demand in
terms of engine health and operational profile from D all the way to A, represented by the increasing limitation
in the maximum speeds of the low-pressure rotor. It is also noticeable how the C and D reference profiles are
almost identical except during take-off, which is the most demanding and critical phase of the flight engine-wise
and where it was therefore expectable that a difference between the reference profiles would exist.
`
The yellow dots in Figures 3.5 and 3.6 represent the N1 and N2 respective Critical Points which
characterize PGA’s Fokker 100 fleet for the considered period of time, in terms of flight profile. The obtained
50
60
70
80
90
100
110
0 10 20 30 40 50 60 70
N2(%)
Minutes
N2 Flight Profile Data
A
B
C
D
AVG
90
92
94
96
98
100
102
0 5 10 15 20 25 30
N2(%)
Minutes
N2 Flight Profile Data Detail
A
B
C
D
AVG
AVG
Figure 3.6 a) and b) - Datum flight profiles from Plan A to Plan D and analysis results for N2

26
results are systematized in Table 3.2, where the 5-year analysis is divided by years, as it was already referred in
section 3.2., and then by parameters and flight phase.
Flight Phase Parameter Limits 2005 2006 2007 2008 2009 Average
Take-off
N1 A < 84,5
83,99 83,91 84,00 83,45 82,96 83,66
A A A A A A
N2 94,9 < B < 96,4
95,96 95,85 96,07 95,47 95,52 95,77
B B B B B B
Climb
N1 A < 89,0
87,97 88,81 88,85 88,13 88,42 88,44
A A A A A A
N2 94,6 < C < 96,6
95,36 95,45 95,49 95,18 95,37 95,37
C C C C C C
Table 3.2 - Yearly results of N1 and N2 values; flight profile limits and global results
A simple average was used in this case, although a weight average was also considered, but because the
difference between the results obtained by each approach was very small, the weight average was discarded.
With further observation of Figure 3.6 b), there’s a strange situation that draws the observer’s attention.
Around the 25 minute mark, when the aircraft’s engines reach the maximum values of high-pressure rotor speed,
it comes to attention that the blue line, which represents reference flight profile B, is lower than the red line,
which represents reference flight profile A, which means that in that important instant, reference flight profile B
is actually more demanding (requires lower maximum values of rotor speed to be applied) than reference flight
profile A. When calculating average values from the 5-year data of the N2 parameter for the climb flight phase
(see Table 3.2), the result is 95.37%, which is 0.77% over the reference profile A and 1.23% below reference
profile C. The fact that the profile B isn’t half-way between profiles A and B as usual, forces PGA’s Critical
Point for N2 climb phase to an apparently unfair flight profile C.
Although understanding that this kind of analysis isn’t scientifically correct, but just to give an idea of
the damage the inexistence of a reference profile B in the referred point, considering that a numerical scale
would exist from 1 to 4, which would correspond to the existing classification, from D to A (in other words,
if D ൌ 1, C ൌ 2, B ൌ 3 and A ൌ 4), the “average” final result with and without reference profile B would be:
FP ൌ
୅ା୆ା୅ା୆
ସ
ൌ
ସାଷାସାଷ
ସ
ൌ
ଵସ
ସ
ൌ 3,5 ൌ A and FP ൌ
୅ା୆ା୅ାେ
ସ
ൌ
ସାଷାସାଶ
ସ
ൌ
ଵଷ
ସ
ൌ 3,25 ൌ B.
Equation 3.2
Despite the lack of scientific value of this analysis, and more in a qualitative basis, assuming that 3.5
would round to 4 then PGA’s flight profile would have an A classification, which would represent huge savings
for the company. Moreover, because the take-off phase is more demanding for the high-pressure rotor system
than the climb phase, an increased importance should be put in the results of the most demanding phase, and the
good behaviour demonstrated at take-off should be taken into account. Further discussion about this situation

27
will be presented at the end of this section, when the final flight profile attribution by the manufacturer will be
analysed.
Turning the focus to the historical evolution of the studied parameters from 2005 to 2009, in order to
better visualize the results presented in Table 3.2, graphical representations of the results were obtained and are
displayed in Figures 3.7 a) through d), where the evolutions of flight profile defining points N1 and N2 for Take-
off and N1 and N2 for Climb throughout the last five years can be more easily analysed.
Figure 3.7 a) and b) - Evolution of flight profile defining point N1 and N2 for take-off
Figure 3.8 a) and b) - Evolution of flight profile defining point N1 and N2 for climb
Observing the graphical representations displayed in Figure 3.7 a) and b) and Figure 3.8 a) and b), it is
clear that the maximum high and low pressure rotor speeds didn’t suffer great variations in the last five years,
varying in about 1% at the most. Furthermore, the registered variations were never sufficient to make any of the
determined Critical Points change their position relatively to the reference flight profiles, a trend that can be
easily confirmed by checking Table 3.2. Despite the relatively small variations and the immutability of the flight
profile classification, two aspects should be pointed out.
Firstly, if we take into consideration for instance that in the take-off phase between reference profiles A
and B, both for N1 and N2, there are differences of less than 2%, it becomes evident that a variation of even 1%
or 0.5% can result in a big difference. An improvement of about 2.1% can represent upgrading a C profile to an
A profile and in consequence saving 15% or even 28% in the example parts shown in table 3.1. This possibility
of great savings with apparent little improvement is the main reason airline companies take measures to reduce
82,50
83,00
83,50
84,00
84,50
2005 2006 2007 2008 2009
N1(%)
Year
Take-off N1
94,80
95,30
95,80
96,30
96,80
2005 2006 2007 2008 2009
N2(%)
Year
Take-off N2
87,50
88,00
88,50
89,00
89,50
2005 2006 2007 2008 2009
N1(%)
Year
Climb N1
94,20
94,70
95,20
95,70
96,20
2005 2006 2007 2008 2009
N2(%)
Year
Climb N2

28
engine wear and request the update of their flight profile status frequently. Because a small difference can make
a huge difference, this means that the representativeness of the considered data and the accuracy of the
conducted analysis are essential to provide correct results.
Secondly, although the variations in rotor speed are relatively small, observing graphical representations
of Figure 3.7 and Figure 3.8, it is clearly visible that every parameter shows a trend to decrease the rotation
speed, thus showing a slight improvement in engine condition, or at least it is approximately stabilized – Climb
N2 is an example. It should be noted that the natural evolution of the rotor’s speed is to increase, corresponding
to the natural degradation of the engine throughout its parts’ lives. The fact that the graphical representations
show the opposite trend is explained by the good practices enforced by the company, such as directives for the
pilots to use low power whenever possible, the effort of the maintenance team to maintain engine health by
regular inspections and immediate part substitution when any faulted part is lowering the engine’s performance,
etc. Many of these directives have been more fiercely enforced since late 2007 and the results can be seen
graphically by the drop registered in both rotor speeds for both flight phases from 2007 to 2008. In 2009 rotor
speeds stabilized or increased slightly, while still maintaining the same policies, which is somewhat puzzling.
Naturally, the more the engine improves, the less margin for improvement exists, and so it’s comprehensible that
the rate of improvement would diminish, and on the other hand, knowing that engine speed is highly dependent
on the weather and particularly on the TAT, the fact that 2009 was one of the warmest years in history helps
explaining this behaviour.
TGT 2005 2006 2007 2008 2009 Average
Take-off
N1 701,0 694,4 696,3 690,6 691,7 694,8
N2 700,6 693,5 694,9 689,2 690,8 693,8
Climb
N1 702,0 701,7 701,1 699,9 702,0 701,3
N2 704,3 704,6 704,3 702,3 706,1 704,3
Table 3.3 - TGT values corresponding to the maximum values of N1 and N2
Figure 3.9 a) and b) - Evolution of TGT corresponding to defining point N1 for Take-off and Climb
Shaft rotation speed is one of the most important parameters to take into account when studying an
engine’s condition, which is why flight profile is classified according to its values. Other important type of data
is the engine’s TGT, a parameter very sensitive to variations in the engine’s operation, as a raise in TGT is
686,0
691,0
696,0
701,0
706,0
2005 2006 2007 2008 2009
TGT(ºC)
Year
TGT Take-off N1
695,0
697,0
699,0
701,0
703,0
705,0
2005 2006 2007 2008 2009
TGT(ºC)
Year
TGT Climb N1

29
usually the first sign that something isn’t working as it should inside the engine or that its performance is for
some reason degrading. It makes then perfect sense to study these two parameters together and analyse their
trends as one. Just like with rotor speed evolution, similar tendencies of improvement until 2008 and stagnation
or increase in 2009, generally in the engine wear rate, can be observed in Figure 3.9 a) and b) where TGT for
both flight phases is graphically represented. Also noticeable is the almost 10 degree global decrease in TGT at
take-off, which is a good symptom of engine health and a sign that the measures being taken to improve engine
condition are being successful. Table 3.3 compiles all the evolutions of the TGT parameter for the 2005-2009
period.
The engine’s sensitivity to variations of the outside temperature, in terms of fuel flow, TGT, shaft
speeds and performance in general, makes the analysis at hand more difficult to conduct if a statistical method
wasn’t used.
Throughout the year, many climate characteristics such as humidity and winds at high altitudes
continuously change, sometimes in unpredictable ways, directly affecting flight conditions and therefore engine
performance. The effect of humidity on engine performance is a much discussed subject, because the amount of
water in the intake air can affect the air’s temperature and density. Although some investigators believe it does
affect engine performance [3][4], it is commonly agreed that the impact will be reduced when comparing to
other variables.
One of the characteristics that notoriously affect the process of flight profile determination is the
temperature, in a more predictable way: when temperature increases, so does TGT, rotor speeds and fuel flow.
An example for a specific TAY650-15 engine is displayed in Figure 3.10 a) and b), where graphical
representations can be found, which show the TGT variation depending on the time of the year.
Figure 3.10 a) and b) - Evolution of TGT corresponding to defining point N1 for Take-off on a specific
TAY650-15 engine
Inspecting the graphics in Figure 3.10 a) and b), in the horizontal axis the letter corresponds to the
aircraft where the engine was installed in that moment (from CS-TPA to CS-TPF), which is followed by the date
when the FDR data was downloaded, in the DD/MM/YY format, each download corresponding to about twenty
650,0
660,0
670,0
680,0
690,0
700,0
710,0
720,0
730,0
740,0
B310105
B260205
B040805
D160606
D170906
D131206
D010307
D140507
D240707
D111007
D140208
D200608
D130808
D141008
D080309
D240409
D110809
D211009
TGT(ºC)
Aircraft + Date
TGT N1 Take-off
670,0
680,0
690,0
700,0
710,0
720,0
730,0
740,0
B310105
B260205
B040805
D160606
D170906
D131206
D010307
D140507
D240707
D111007
D140208
D200608
D130808
D141008
D080309
D240409
D110809
D211009
TGT(ºC)
Aicraft+Date
TGT N1 Climb

30
consecutive flights. The main reason for the “saw-shaped” display is the temperature and eventually humidity
variation as the year progresses, with low values of TGT during the cold and dry winter months and higher
values during hotter late-spring or summer-months. From 2007 onwards each stage of the oscillations is clear,
with medium points corresponding to the spring or autumn and in 2005 and 2006, due to lack of stored data, only
three points are presented, but the tendency is naturally the same. Through observation of the linear trend lines
representing the global TGT evolution, it is also possible to see that overall the TGT increases in the five years
of operation, which is a consequence of the natural degradation of the engine’s condition, with average
variations of almost 20ºC, which can lead to diminished performance and even structural problems, once again
proving the importance of up-close monitoring of engine health.
3.5 – Change in Thrust Mode
3.5.1 – Results
The importance of an accurate determination of a fleet’s flight profile, with all the associated difficulties
already described, justifies the analysis of a large sample of data, hence the usual utilization of FDR data, which
can be easily collected from each single flight. However, flight profile assessment can also be conducted based
on the cockpit manual flight data recording with an added workload to the pilot.
The power output of an engine is influenced by many factors such as outside conditions, the engine’s
bleed status which is controlled by the ECS, the anti-ice status and the Thrust Mode. The engine’s thrust is
controlled by the Thrust Management Computer for reference EPR computation. The Thrust Mode Select Panel
(TMSP), an example of which is represented in Figure 3.11, allows for selecting the Reference Thrust:
TO/GA – Selects TO (takeoff) Mode on the ground or GA (go-around) Mode in flight;
CLB – Selects CLB (Climb) Mode;
CRZ – Selects (cruise) Mode;
CON – Selects CON (max continuous/economical) Mode.
Figure 3.11 - Example of a Thrust Mode Select Panel (TMSP)
After takeoff, Thrust Mode will naturally change from TO to CLB, and stay that way until TOC is
reached, where it will change to CRZ, which isn’t as demanding to the engine as the Climb Thrust Mode. A
question was then raised: could the CRZ (cruise) Thrust Mode be applied while in climb flight phase? That
would theoretical decrease engine wear and consumption but would it be possible operation-wise? It would be

31
very interesting to have an idea about the savings that could be accomplished both in fuel and in maintenance
while making a balance between savings and eventual added costs or problems and also limitations to its
implementation.
To measure the impact of changing the Thrust Mode, a directive would have to be approved, namely
safety-wise, then transmitted to all the flight crews and finally systematically collected and then processed. A
measure of this type, implemented in a whole fleet for a period of time, only to assess the possibility of an
eventual gain, is too expensive and generates many changes to the flight crews’ routine, which could even
compromise flight safety. A crew was then asked to manually record the engines’ rotor speeds in two flights in
the same routes and similar conditions (consecutive days, with the same departure hour, same aircraft and
roughly the same weight), one with the usual CLB Thrust Mode and other with the CRZ Thrust Mode. In Figure
3.12 a) and b), one can find the difference in N1 and N2 respectively, between the two Thrust Modes as the
aircraft climbs, for both engines.
Figure 3.12 a) and b) - Thrust mode impact in one specific flight, on defining point N1 and N2
Observing the graphical representations of Figure 3.12 a) and b), it becomes clear that the change of
Thrust Mode results in a very significant difference in terms of both high and low pressure rotor speeds.
80
82
84
86
88
90
92
100 120 140 160 180 200 220 240 260 280 300 320 340
N1(%)
Flight Level
Thrust Mode Influence on N1
Eng1CLB
Eng1CRZ
Eng2CLB
Eng2CRZ
92
93
94
95
96
97
98
99
100
100 120 140 160 180 200 220 240 260 280 300 320 340
N2(%)
Flight Level
Thrust Mode Influence on N2
Eng1CLB
Eng1CRZ
Eng2CLB
Eng2CRZ

32
Although the trends for the two thrust modes are similar for each motor, there is a clear offset between CLB and
CRZ, representing a well defined reduction for the complete climb phase. The improvement is always larger than
1% and for some flight moments or flight levels the reduction sizes to about 4%. However, as already pointed
out in section 3.2, for the calculation of a flight profile only the maximum speed values of the climb phase are
required and so the maximum values of the line charts were collected and systematized in Table 3.4. In a
separate note, there is a significant difference between engines which should be further investigated.
Flight
Phase
Rotor
Speed
Limits Average
Current
Profile
Engine CLB CRZ
Improv.
CLB/CRZ
Equiv.
Improv.
New
Profile
Climb
N1 ‫ܣ‬ ൏ 89 88,44 A
1 91,3 88,8 2,74 86,02
A
2 91,6 88,7 3,17 85,64
N2 94,6 ൏ ‫ܥ‬ ൏ 96,6 95,37 C
1 97,7 95,6 2,15 93,32
A
2 95,9 93,9 2,09 93,38
Table 3.4 – Obtained results and improvement in Flight Profile after change in Thrust Mode
In the previous table, columns CLB and CRZ have the maximum speed values for each engine/rotor
combination with the “Climb” and “Cruise” Thrust Modes respectively. The following column displays the
speed reduction when changing Thrust Mode, in percentile. If we would assume that the whole fleet would have
the behaviour of this engine, and thus have the improvement values displayed in this column, then the equivalent
improvement of the fleet would be the same. Applying that improvement to the average of the fleet presented in
Sub-section 3.4.1, which is displayed in the “Average” column in Table 3.4, would result in the values shown in
the second to last column, “Equiv. Improv.”, which corresponds to Equivalent Average Improvement. When
comparing these new results with the limits which are the criteria to attribute flight profiles, in column “Limits”,
it is easy to realize that for the high-pressure shaft speed, the current flight profile of C would transition to an A,
with huge consequences. Also in terms of the low-pressure rotor speed, the distance for the A reference profile
would increase substantially, giving more confidence in the current A profile.
Before taking conclusions about how beneficial such a change could be in terms of life-limited engine
part costs, all other aspects of the airline’s operations which would be involved with such a change have to be
taken into account and consequences have to be predicted. This will be the discussion of the next Sub-section.
3.5.2 – Consequences – Cost: Time and Fuel vs. Maintenance
The objective of this Sub-section is to point out some of the potential problems that could arise if the
Thrust Mode in the climb phase was changed from CLB to CRZ, which would assumedly result in an upgrade of
the fleet’s flight profile. Firstly, the consequences of the flight profile upgrade will be addressed, followed by a
discussion over the change in the Thrust Mode.
As stated before in a case like this, when an operator believes that his operation is no longer represented
by the reference datum flight profile, the new changed fleet representative flight profile data must be supplied to

33
the manufacturer. Any operator changing from one Life Profile Operation to another must notify the
manufacturer of this change and the residual lives of all the Group A and Group B mandatory parts must be
calculated again, and entered in the engine maintenance records. The formula is as follows:
RL ൌ FPLL ൈ ൬1 െ
CSN
IPLL
൰
Equation 3.3
With RL = Residual Life; FPLL = Final Plan Life Limit; IPPL = Initial Plan Life Limit; CSN = Cycles
Since New. When the flight profile is improved, in this case from “B” to “A”, the residual life of the critical parts
will increase, providing a better flight cycle per part cost ratio, which was the intended effect. This easy
calculation and the consequent information update of the airline’s databases is the only real task that would have
to be performed to have the new flight profile, obviously besides maintaining all the practices and policies that
granted the airline this flight profile in the first place. As the flight profile improves, so should the level of
maintenance, with frequent inspections and trend-monitoring to continuously assess engine condition, processes
that also come with its costs.
However, because this improvement was achieved mainly based on low-power and reduced climb rate
policies, this means that the flights will be a bit longer, which can affect passenger satisfaction, mean more fuel
consumption and an increase in time-related costs.
The climb phase has a huge impact on fuel consumption when considering short and medium range
flights since it represents from 20% to 40% of the trip time, registering fuel flows 40% greater than on cruise
phase. Climbing with reduced thrust will increase fuel consumption because it would extend time spent at lower
altitudes where the fuel flow is higher. Therefore, reducing thrust during climb will not save fuel. On the other
hand, using thrust settings higher than CLB to make the climb faster would as already seen severely penalize
engine life. So in terms of overall fuel consumption, changing the Thrust Mode wouldn’t be beneficial.
In terms of flight safety there are also some limitations and concerns that should be taken into account.
In order to ensure that air traffic controllers can accurately predict flight profiles to maintain standard vertical
separation between aircraft, pilots of aircraft commencing a climb or descent in accordance with an ATC
Clearance should inform the air traffic controller if they anticipate that their rate of climb or descent during the
level change will be less than 500 ft per minute, or if at any time during such a climb or descent their vertical
speed is, in fact, less than 500 ft per minute, as referred in point 2.4.1 of UK’s AIP General Rules and
Procedures [2], a limit internationally accepted. This means that the reduced thrust could be used, and if the
aircraft’s ROC would be less than 500ft/min, the ATC had to be informed and ATC indications strictly followed.
Also from an operational point of view, about 30% of PGA’s flights cover distances under 200 NM, in
other words a significant part of the flights take 30-45 minutes in total, with sometimes only about 15-20
minutes in cruise at the selected flight level. This means that changing the Thrust Mode, thus flattening the

34
climbing profile, prolongs the climb flight phase, reaching TOC later and therefore further shortening the cruise
phase, which can be impossible or impractical in short-haul flights like LIS-OPO.
Passenger satisfaction is always a top principle to any airline company, especially to PGA which has
always made passenger satisfaction and comfort top priorities, having earned numerous awards and international
recognition for it. Therefore any change in the company’s operational profile will have an effect in the
costumers’ opinion that must be taken into consideration. An eventual change in Thrust Mode would influence
their satisfaction in two ways. The lower rate of climb implies more time climbing and a longer flight in general,
so this increase in flight duration would have to be measured to have an idea if it would make passengers less
satisfied. An increase of just a couple of minutes shouldn’t damage PGA’s reputation in terms of passenger
satisfaction, because it’s not very significant an increase from 47 min to 49 min for example on a LIS-OPO
flight. On the other hand, lower thrust and lower rate of climb result in a more silent and smoother flight, with
slower pressure variations, which together produce a more enjoyable and comfortable flight. Balancing both
effects, it is predicted that a change in Thrust Mode would have an impact by slightly increasing passenger
comfort and satisfaction, depending on the increase of flight time relatively to its original duration.
A change in the Thrust Mode during climb and the consequent flight profile improvement is being
considered because of the savings in flight-cycle dependent maintenance costs. However there are other costs
that can be influenced by the increase in the climb flight time, one of them is fuel cost which would increase like
it was already explained before in this Sub-section, and also time-related costs would also be affected and
therefore it’s important to assess if the change in Thrust Mode would pay-off after the overall costs balancing
was completed. Maintenance time-related costs should be the first to be analysed, although this would probably
prove to be a difficult exercise: rust and wear inspections would be slightly affected, the 5000-hour, 10000-hour,
etc checks would in theory be more frequent, approximately in the same amount as the average increase of flight
time. Regarding the flight/cabin-crew costs, they would remain unaffected because PGA’s flight and cabin crew
personnel are paid not by the hour but on a per-flight-basis or have a monthly fixed salary, thus unrelated with
the flight’s duration. Finally there are some costs that depend on the aircraft’s flight hours, such as insurance,
aircraft rental, interest and other company related costs. These would also need to be accounted for to assess the
penalty associated with this change in the operational profile.
The results presented in Table 3.5, although without statistical value, cannot be overlooked given their
definite behaviour and the gain margins involved. The potential savings in maintenance costs justifies in a first
stage the undertaking of a serious and thorough study about an eventual earlier change in Thrust Mode during
climb, estimating gains and losses in all the departments affected by this change and assessing overall benefits
from this measure. If it was predicted that such a measure would be advantageous, in a second stage it should be
enforced in an experimental period, applying this change to several aircraft and different routes for a sufficient
period of time while continuously collecting the FDR data as usual, comparing the results with the ones obtained
in the same period in the previous year and assessing if an improvement was achieved. If so in a third stage this
change would then be applied to all aircraft and all routes, or at least the routes in which it had proved to be
beneficial.

35
4 – Engine Condition Monitoring
4.1 - Monitoring as a Route to Safety
Everyone agrees that there is no price for a human life, nor does it exist for a company’s reputation, and
no eventual savings can justify the destruction of either one. For these reasons, ever since the beginning of
commercial aviation, passenger and crew safety has always been the number one priority. Because an engine
failure during a commercial flight is very likely to have a devastating effect, failures just can’t be tolerated,
which is why most of the maintenance practices are conducted with the objective of guaranteeing safety. In the
aeronautical industry a specific maintenance method is usually enforced, which consists of changing an engine
or major modules in need of inspection and repair with a new or refurbished engine or modules. This method is
commonplace in aviation for it allows the aircraft to remain in service as much as possible, without
compromising flight safety.
In the primordial years of aviation history, manufacturers and operators began to operate engines to
failure, which means that the engine was left on wing until something failed, usually with catastrophic results.
Since then, companies operating gas-turbine engines have tried to minimize its high maintenance costs by
avoiding potential engine failure through preventative maintenance actions at fixed intervals. This practice
ensured safety for the most part, but still wasn’t completely safe and wasn’t economically efficient, sometimes
with little actual gain in engine health or performance.
In recent years however, an Engine Condition Monitoring (ECM) approach has been adopted by many
engine manufacturers and operators, in which intelligent real-time data analysis systems are employed to assess
the condition of engine components. The objective is then to make maintenance needs be determined according
to the engine’s operating condition, rather than maintenance being performed at fixed periods of time. ECM
involves both “manual” practices such as MCD, oil consumption and vibration monitoring, and computer
methods based on performance analysis and mechanical parameter monitoring using an adequate software tool.
These monitoring systems for engine health typically process data from engine-mounted sensors, whose recent
evolution in robustness and versatility have made the implementation of ECM possible.
ECM may be used for three different approaches, one pre-emptive, one reactive and one more analytical
and educational. Through the condition monitoring, early warning of potentially hazardous engine conditions
may result in the identification of the precursors to component failure in advance of the actual failure. This is a
prognostic approach to condition monitoring, and is useful for types of faults that may be prevented if identified
soon enough. An example of this type of approach will be presented in Sub-Section 4.4.3. Faults for which there
are no such precursors like a “bird strike" require a diagnostic approach. Those monitoring systems
automatically identify engine faults that have occurred, and may recommend restorative maintenance actions
appropriate to the type of fault. The analysis to the monitoring of an unplanned event resulting in some
mechanical damage will be discussed in Sub-Section 4.4.2. Finally, ECM can be used as a post-event tool, not to
repair eventual damage, but to understand the nature of the event or why it happened, and what was the cost or

benefit associated with it. This type of approach should also allow operators to construct better and more
accurate models and processes for their ECM, to make
4.2 –Engine Monitoring Method
In a global, fast-paced and continuously changing world, the aviation and aerospace industries are
definitely among the fastest evolving fields in human
technological characteristics and also because of today’s constant battle for optimization and cost reduction. In
the aviation world, much can change in ten years, in terms of components, sensors, materials,
equipment and even philosophies. This idea of constant improvement and evolution can be seen while
comparing the two PGA aircraft models. The Embraer ERJ
Fokker 100 and that gap of time, associ
with different philosophies coming from different continents, makes the Embraer a different and in general more
modern aircraft than the Fokker 100. Many of these differences are
instruments each aircraft has, despite most of those equipments being similar in use or in function for both
models. In terms of engine monitoring, the process to evaluate the Embraer
and quicker than the evaluation of the Fokker 100 engines, as it is shown in Figure 4.1.
Figure 4.1- ECM process for Embraer 145 and Fokker 100 fleets
The diagram above clarifies the difference between the two PGA fleets, regarding the process
trend monitoring. From the Embraer 145 flight data, the CMC automatically calculates the Takeoff and Cruise
representative points of each flight and immediately builds the .dig files that will be introduced into the trend
accurate models and processes for their ECM, to make it even more reliable and cost-efficient in the future.
Method and COMPASS
paced and continuously changing world, the aviation and aerospace industries are
definitely among the fastest evolving fields in human activity, due to its demanding, cutting edge and
the aviation world, much can change in ten years, in terms of components, sensors, materials,
comparing the two PGA aircraft models. The Embraer ERJ-145 is approximately ten years more recent than the
Fokker 100 and that gap of time, associated with different requirements due to their different sizes and ultimately
modern aircraft than the Fokker 100. Many of these differences are related to the avionic equipment and flight
models. In terms of engine monitoring, the process to evaluate the Embraer engines’ condition is much
than the evaluation of the Fokker 100 engines, as it is shown in Figure 4.1.
ECM process for Embraer 145 and Fokker 100 fleets
The diagram above clarifies the difference between the two PGA fleets, regarding the process
36
efficient in the future.
paced and continuously changing world, the aviation and aerospace industries are
activity, due to its demanding, cutting edge and
the aviation world, much can change in ten years, in terms of components, sensors, materials, avionics,
145 is approximately ten years more recent than the
ated with different requirements due to their different sizes and ultimately
related to the avionic equipment and flight
’ condition is much simpler
The diagram above clarifies the difference between the two PGA fleets, regarding the process of engine

37
monitoring tool COMPASS. In opposition, for the Fokker 100 the process is somewhat more complex and time-
consuming, because the Fokker doesn’t have a CMC or capability to produce the .dig files which are the input of
COMPASS.
The representation of Figure 4.1 clearly shows the steps that have to be taken to proceed with the
Fokker 100 engines’ trend monitoring, starting with the download of the FDR data, which will be later processed
by ADRAS software to obtain an ASCII type file, both steps already done for flight profile determination
purposes (see Section 3.1), although some of the processed parameters may not be the same. In the next step
those ASCII files, which have all the data of all the recorded flights, will be processed with a tool enclosed in the
TrendMonitoring.mdb database file. The objective of this tool is, for each flight, find one point that correctly
represents the Takeoff flight phase, and another point which correctly represents the Cruise flight phase – the
Takeoff and Cruise Representative Points respectively. This is done to assess the average engine condition and
performance in each flight, during two crucial flight phases that have a huge impact, both in engine deterioration
and in engine performance and consumption.
Ideally all the moments of each flight should be analysed, but that is unfeasible and furthermore the
objective of the engine trend monitoring is not to study the engine’s condition evolution throughout one
particular flight, or in one particular moment, but to detect trends in the period of a month or a year and then
make comparisons between engines to measure deterioration and abnormal behaviour. Therefore only one point
per flight phase per flight will be considered, with the best accuracy and representativeness possible and do
determine these representative points, a moment had to be defined and in consequence some conditions had to be
imposed. As already referred, the Fokker 100 doesn’t have the inherent ability to produce the representative
points, reason why the conditions imposed by the Embraer 145’s CMC served as a good baseline to define the
conditions to be applied to Fokker flights. The conditions used to define the desired points for each flight phase,
for both fleets, are displayed in Table 4.1 and 4.2.
Takeoff Trend – Entry Conditions
Embraer 145 Fokker 100
Parameter Condition Parameter Condition
Thrust Rating Mode TAKEOFF
Active Thrust Mode TAKEOFF
WOW GROUND
TLA ൒ 72°
Air/Ground Switch GND
Mach 0.17 േ 0.01
Takeoff Trend – Exit Conditions
WOW AIR Air/Ground Switch FLT
Table 4.1 - Entry and exit conditions used to define the representative take-off points (RTOP)

38
Cruise Trend – Entry/Exit Conditions
Embraer 145 Fokker 100
Parameter Condition Parameter Condition
Pressure Altitude (ft) 15000 േ 100 Pressure Altitude (ft) 15000 േ 100
Mach ൐ 0.4 േ 0.015 Mach ൐ 0.4 േ 0.015
N2 (%) ൐ 80 േ 0.3 N2 (%) ൐ 80 േ 0.3
Thrust Rating Mode CRUISE Active Thrust Mode CRUISE/BLANK
Anti-ice OFF All Anti-ice OFF
Stabilized conditions for 5 minutes Stabilized conditions for 5 minutes
Table 4.2 - Entry and exit conditions used to define the representative cruise points (RCP)
On a first note, it should be noticed that it isn’t very important exactly which criteria are defined to
determine the representative points for the Fokker fleet, as long as those criteria remain unchanged for every
flight throughout the analysis, so that they can be correctly compared. Nevertheless there are some minimum
pre-requisites in the conditions imposed, so that the Trend Monitoring Tool actually presents reasonable and
meaningful results.
Observing Table 4.1 where the Takeoff entry and exit conditions are displayed and comparing the
criteria for both fleets it is clear that some changes or simplifications have been done from the Embraer starting
point to the Fokker case. The representative point for Takeoff in the Fokker fleet was determined in a simplistic
way, as the moment the aircraft lifts-off, which is one of the most critical moments in any given flight, and that
moment was detected by the change of the parameter Air/Ground Switch from “Ground” to “Flight”, while
obviously in “TAKEOFF” Active Thrust Mode.
For the Cruise flight phase, the representative points for the Fokker fleet were defined with conditions
similar to those imposed to the Embraer flights, which are enumerated in Table 4.2. The representative point is
determined when the Active Thrust Mode reads “CRUISE” or “BLANK” (the latter we know from experience to
be the same as CRUISE), when the Mach and N2 are bigger than 0.4 and 80% respectively, when the pressure
altitude is larger than 15000 ft and all the anti-ice systems are OFF. After these conditions are verified and
remain stable and true for five minutes, the three hundredth second (300th
) will be recorded as a Cruise
representative point.
When a representative point is determined, all the flight and engine data associated with that moment
are recorded and written in a specific format required by the manufacturer, a format which COMPASS will be
able to read to extract the information it needs. The TrendMonitoring database tool is also responsible for this
data formatting and later assembly in files with a .dig extension, which can be uploaded into COMPASS by one
of its data uploading tools.

Figure
In order to better comprehend the uploading process into COMPASS, how the data is processed and the
type of analysis that this software tool conducts, the COMPASS Engine Trend Monitoring program will be
briefly presented. COndition Monitoring
mainframe in 1985 and is in service since 1988, with multiple updates being released throughout the years.
COMPASS Navigator is a set of software ‘tools’
uploading and management and results visualization through graphical representation:
Manual Data Entry
Automatic Data Entry
run as a ‘hands off’ system.
System Properties
Data Management
databases.
Equipment Directory
COMPASS.
Alert Status
Graphical Plotting
one or more aircraft simultaneously or even the whol
Semi-Graphical Plotting
Figure 4.2 - COMPASS ECM software, by Rolls-Royce
onitoring Performance Analysis Software System was first developed for
COMPASS Navigator is a set of software ‘tools’ - each of which performs a specific function
uploading and management and results visualization through graphical representation:
Manual Data Entry - configurable, aircraft oriented data entry screens.
Automatic Data Entry - automatically processes data and all
run as a ‘hands off’ system.
System Properties - provides an easy means of configuring COMPASS
Data Management - allows the user to manage the data stored on the Navigator
Equipment Directory - ‘Explorer’ type view of all the
Alert Status - allows users to view alerts generated for engines
raphical Plotting - user can plot graphical trend plots and define new plots, analyse
one or more aircraft simultaneously or even the whole fleet.
Graphical Plotting - plots semi-graphical trend plots.
39
ystem was first developed for
s a specific function, such as data
data entry screens.
and allows COMPASS to be
COMPASS.
allows the user to manage the data stored on the Navigator
equipment known to
for engines within the fleet.
plots and define new plots, analyse

Some of COMPASS Navigator analytical features are:
Trending of cruise
Trending of engine vibration
Trending of take-off
Trending of take-off TGT and shaft speed margins
Maintenance action data handling
Alert generation
Secondary Data Analysis. (Maiden, Compressed, Fleet Averaging)
Sophisticated smoothing algorithms
After setting up the interface and databases, the first ste
aircraft to each fleet and assigning the respective engines to each aircraft, with the assistance of the Equipment
Directory COMPASS tool, represented in Figure 4.3
recorded, so that for each moment, a determined engine is placed in the right aircraft.
Figure
The .dig files containing the representative points were uploaded into COMPASS using the Automatic
Data Entry feature, because it processes large quantities of data in a short amount of time and because it’s
automatic, which means other tasks can be done while the data is
facilitate this operation. It can even be said that
would be very slow, which would probably result in most cases in a loss of representativeness or even imply not
using the program and thus not proceeding with the engine trend monitoring. T
chronologically from the beginning of 2010 until the end of August, month for which all data was available at
the end of the analysis.
The most utilized feature of the COMPASS Trend Monitoring program was the Graphical Plotti
for obvious reasons, since it was responsible for all the graphical representations and ultimately all the results
acquired for this analysis and presented in section 4.4. The Graphical Plotting tool can represent every chosen
Some of COMPASS Navigator analytical features are:
Trending of cruise TGT, fuel flow and shaft speed
Trending of engine vibration
off TGT and shaft speed SLOATL's
off TGT and shaft speed margins
aintenance action data handling
Secondary Data Analysis. (Maiden, Compressed, Fleet Averaging)
histicated smoothing algorithms
After setting up the interface and databases, the first step was to construct both fleets, inserting the
, represented in Figure 4.3. Engine placement history was also taken into
recorded, so that for each moment, a determined engine is placed in the right aircraft.
Figure 4.3 - Equipment Directory COMPASS tool
Entry feature, because it processes large quantities of data in a short amount of time and because it’s
automatic, which means other tasks can be done while the data is being uploaded, two aspects that greatly
facilitate this operation. It can even be said that without such a tool, data uploading and therefore data processing
using the program and thus not proceeding with the engine trend monitoring. The data had to be introduced
The most utilized feature of the COMPASS Trend Monitoring program was the Graphical Plotti
presented in section 4.4. The Graphical Plotting tool can represent every chosen
40
p was to construct both fleets, inserting the
. Engine placement history was also taken into account and
Entry feature, because it processes large quantities of data in a short amount of time and because it’s
uploaded, two aspects that greatly
without such a tool, data uploading and therefore data processing
he data had to be introduced
The most utilized feature of the COMPASS Trend Monitoring program was the Graphical Plotting tool,
presented in section 4.4. The Graphical Plotting tool can represent every chosen

41
parameter or combination of parameters that the user requires, during any period of time and for one or more
aircraft, as long as that data has been previously uploaded into the program. This feature then generates graphical
representations in Excel sheets, whose data can then be altered at will. The observation of the graphical
representations and the analysis of the trends that these representations may reveal constitute the last step in the
trend monitoring process, which can be obviously followed by any maintenance or inspection actions if the
acquired results suggest that a problem could exist in one of the analysed engines. After the maintenance action,
a new trend analysis should be conducted to see if the problem remains or if the studied parameters show an
improvement in the engines condition.
The Data Management and System Properties tools were used only rarely when needed to erase
incorrect data which was uploaded or to customise the other more utilized tools. The Semi-Graphical Plotting
tool wasn’t utilized at all, given its limitations compared to the Graphical Plotting tool and because there was no
necessity of getting additional representations of the results.
4.3 – Data Input
Determining the degradation of a gas turbine in operation requires a certain level of instrumentation and
a methodology to compare data taken under various ambient and operating conditions. The most accurate way of
correcting the operation of gas turbines to datum conditions is by using manufacturer software that models the
gas turbine or manufacturer-supplied performance curves. There are some good practices in terms of which data
should be utilized to effectively monitor an engine’s condition, one of them is to exclude data taken under
transient conditions, which won’t truthfully represent the engine’s real status in terms of performance. Also the
operation profile and the flight conditions have a great impact in the engine’s power output and response. Gas
turbine performance is very sensitive to ambient conditions, such as pressure and temperature, power turbine
speed and additionally the heat rate of the gas turbine is sensitive to part load. The results can be quite different
for different loads, particularly for gas turbines that bleed air at part-load operations such as for emissions
control. Therefore, the information about the power produced is useless for condition monitoring if it is not
corrected for the prevailing operating conditions.
Valuable information can be gathered by looking at engine trend plots. Trend plots can indicate if
engine performance is deteriorating as expected or if there has been a major change in performance because of
hardware failure. The Engine Condition Monitoring can be based in the analysis of the behaviour of multiple
parameters, such as engine temperatures, engine vibration, high and low-pressure rotor speeds, engine pressure
ratio (EPR), etc. Naturally, the more parameters are evaluated, the more correct and representative results are
acquired, which will lead to a more global perception of the engine’s condition. However, some parameters have
less predictable behaviours than others, often resulting in variations that aren’t really significant, that couldn’t be
foreseen, which can’t be avoided or sometimes not even explained. Furthermore, proceeding with a detailed
analysis for many parameters simultaneously is very time-consuming, so a better approach would be to identify a

42
few Key Parameters that are more sensitive regarding the engine’s condition, which will therefore present
significant, visible changes in their value when an event occurs inside or around the engine, affecting its
operation. These Key Parameters had already been identified, namely by the manufacturer, as the engine’s TGT,
the high-pressure shaft rotation speed N2 and the fuel flow FF into the combustion chamber. Parameter
characteristics can be summarized as follows:
Turbine Gas Temperature (TGT) is sensitive to engine deterioration
Fuel flow (FF) is comparable to TGT but less sensitive
HP shaft speed (N2) will increase or decrease depending on the LP or HP system
In fact, these are the characteristics of the engine’s operation that will be consistently considered and
analysed throughout the study developed in this chapter. When the evolutions of these Key Parameters are
analysed and a possible event is detected, then a more complete and thorough study, with all the available
parameters, can be conducted to collect the most information possible about that precise event and hopefully
identify possible causes. Therefore, other parameters, such as alarms, vibrations, bearing temperatures, oil
temperatures and pressures, etc, need to be recorded as well, as they can give valuable correlations in the search
for causes of performance degradation, in particular if the degradation is unexplained and rapid.
By observing the variations the analyst can interpret which component may be deteriorated. In addition,
the trends of all three parameters (TGT, FF, and N2) can determine if there is an aircraft measurement or an
engine measurement problem. Using the fault diagnosis diagrams in the Trend Guideline Chart presented in
Figure 4.4 as a reference, it becomes possible to associate a certain variation or combination of variations with a
possible cause, thus identifying engine problems. However, when changes occur in more than one component it
may be difficult to diagnose the precise problem. This is often the case when engines deteriorate over a period of
time and a mixture of dirt, erosion, tip clearances, worn bearings, ... etc. combine to produce trends that reflect
overall deterioration rather than a specific component problem. When a significant engine problem occurs, such
as hardware failure, it will primarily affect a single component in a short period of time. In these circumstances,
analysis of the trend data may show the cause of the problem.
In the represented Trend Guideline Chart, Inter Turbine Temperatures (ITT) are used as one of the Key
Parameters, however ITT and TGT temperatures are measured in close proximity (see the Glossary of Terms)
and have similar values and behaviours, and the use of one in detriment of another usually depends on the engine
model or manufacturer. Because of its great similarities the example diagnosis diagrams displayed in Table 4.3
and the complete Guideline Chart in Appendix I, can be used in our case without any problem.

43
Table 4.3 - Part of the trend guideline chart presented in APPENDIX I
COMPASS Navigator compares engine performance data recorded in flight against model predicted
performance levels and calculates the differences as a 'Parameter Delta'. This is a very interesting feature,
because not only do these delta parameters enable the analyst to work with relative values in percentage, which
are easier and more practical to compare and quantify, as these parameters also measure the deviation between
the in-flight acquired results and the theoretical-model expected values. For these reasons our analysis will be
done based not on the exact parameters previously referred, but rather on their variations to the expected model
values: DeltaTGT, DeltaN2 and DeltaFF – Delta Key Parameters. Naturally all the trend behaviours presented by
these delta parameters will be identical to those displayed by the original Key Parameters, therefore the use of
these delta parameters only add practicability and the possibility to assess theoretical/empirical relation without
affecting the sensitiveness of the parameters to engine condition changes.
Other parameters that will be found in this study are the “Smoothed parameters”, usually the Smoothed
Delta Key Parameters. The COMPASS Navigator program has sophisticated smoothing algorithms which:
Reduce the effect of noise and data scatter
Routines employ optimal estimation techniques
Produce smoothed data, which is used to detect longer term trends in the data
This smoothing and estimation capability is one of the most important and valuable features that make
the COMPASS software a true trend monitoring tool. The Smoothed Delta Key Parameters were used
continuously during the current study and will be represented in this chapter concurringly, sometimes
accompanied by the true Delta Key Parameters or by themselves.
Now that the Key Parameters that will be used in the following analysis are presented, let us see how
the results are presented for a random aircraft by COMPASS’s Graphical Plotting Tool, while presenting and
discussing a potential problem that was detected during the course of the analysis. In Figure 4.4, the evolution of

44
the parameter Delta Fuel Flow (DFF) throughout January and February 2010 is represented for both engines of
an Embraer 145 aircraft, along with the respective smoothed parameters, DFF__L.
Figure 4.4 - Values for smoothed DFF parameter in Embraer CS-TPI
Observing the graph above for the Embraer aircraft, it is clear that the FF for the installed engines is
slightly above the theoretically predicted values, which is to be expected, but in overall terms there is a clearly
stabilized profile, with smoothed variations (variations of the smoothed parameter) of about 0.5%-1%. In terms
of the DFF, there are no great spikes or steps, the greatest variation from one flight to the next being of about
2%. These results were acquired for the CS-TPI aircraft, for the months of January and February, but in general
all Embraer aircraft displayed stable, low-noise and clean results throughout the year, as far as the fuel flow is
concerned. Bearing these results in mind, let us now analyse the behaviour of the same parameters DFF and
DFF__L, for the same months, in a Fokker 100 aircraft, which are displayed if Figure 4.5.
Figure 4.5 - Values for smoothed DFF parameter in Fokker CS-TPA
Analysing the graphical representation displayed in Figure 4.5 for the Fokker 100 aircraft, and
comparing it with the one on Figure 4.4, containing results for an Embraer 145, it is clear that the two plots and
-20-1001020
01-01-2010
03-01-2010
06-01-2010
09-01-2010
11-01-2010
15-01-2010
18-01-2010
22-01-2010
25-01-2010
28-01-2010
30-01-2010
03-02-2010
05-02-2010
08-02-2010
23-02-2010
26-02-2010
DeltaFuelFlow
Date
Smoothed Delta Fuel Flow
310098: DFF (I) 310071: DFF (I) 310098: DFF___L (I) 310071: DFF___L (I)
-20-1001020
01-01-2010
02-01-2010
04-01-2010
08-01-2010
09-01-2010
12-01-2010
17-01-2010
20-01-2010
22-01-2010
28-01-2010
31-01-2010
02-02-2010
05-02-2010
07-02-2010
09-02-2010
20-02-2010
22-02-2010
27-02-2010
DeltaFuelFlow(%)
Date
Smoothed Delta Fuel Flow
17277: DFF (I) 17276: DFF (I) 17277: DFF___L (I) 17276: DFF___L (I)

45
their trends are quite different. In both Fokker engines we can see constant, high-frequency, large amplitude
variations for the exact parameters which reflect themselves in the smoothed parameters. The smooth and exact
fuel flow parameters present variations up to 4% and 10% respectively, from one flight to the next. In the
particular case of engine 17276, the DFF__L varies in a range of 8.5% and the DFF in the range of a whopping
20.1%, against the ranges of 2% or 3% verified for both parameters within the Embraer results. If it is taken into
account than each of the spikes represented in Figure 4.5 are supposed to represent a stabilized flight in terms of
fuel flow, the implications of the results above could be huge. Saying that an engine is consuming 10% more
than it should in theory, for the whole cruise flight phase, is alarming at the least, because fuel costs sky-rocket
and also because this may be a symptom of a serious problem in that specific engine. Such erratic behaviour
should then raise a red flag regarding this engine and the engine should probably be inspected.
Figure 4.6 - Values for smoothed DTGT parameter in Fokker CS-TPA
The other Delta Key Parameters were analysed for this engine trying to pinpoint the exact cause of such
variations. The DTGT and DTGT__L were represented in Figure 4.6 and the results were surprising. TGT and
the FF usually present the similar behaviours, but the TGT parameter is known to be more sensitive to changes
in the engine’s operation than FF, so how is it possible that such large variations in FF aren’t reflected by the
TGT’s graphical representation? The DTGT relatively stabilized profile makes a hardware failure unlikely so the
problem likely has to do with the FF sensors or FF data recording process for example. After analysing all the
Fokker 100 aircraft throughout the last year, it became evident that every engine of the Fokker fleet presented
similar quick, high-amplitude variations, although the case of engine 17276 is probably the most extreme within
the fleet. These results confirm that the most probable cause may be the way the flight data is filtered in the
Fokker fleet, although a further study should be conducted to explain these variations.
One of the possible causes for this baffling behaviour could be related with the definition of the
Representative Point for the cruise flight phase, namely the representative points might have been recorded
before the Top of Climb (TOC) or before the cruise was completely stabilized, but after investigating a great
number of flights flown with all aircraft, it became evident that wasn’t the case. The representative points were
well inside the “cruise plateau”, like in the example-flight showed in Figure 4.7, a random flight which took
-20-1001020
01-01-2010
02-01-2010
04-01-2010
08-01-2010
09-01-2010
12-01-2010
17-01-2010
20-01-2010
22-01-2010
28-01-2010
31-01-2010
02-02-2010
05-02-2010
07-02-2010
09-02-2010
20-02-2010
22-02-2010
27-02-2010
DeltaTGT(%)
Date
Smoothed Delta TGT
17277: DTGT (I) 17276: DTGT (I) 17277: DTGT__L (I) 17276: DTGT__L (I)

46
place on the 26th
February 2010. It is easy to see that the cruise flight phase is comprised in the period roughly
between 16h00 and 18h00 and that the representative points for cruise for both engines, represented by the
yellow and purple dots, are inside that period. However, after closer inspection at the “stable cruise plateau” it
becomes clear that the points which belong to that plateau range between the 700 kg/h to 1050 kg/h, which
represents a maximum variation of about 30% for an average value of about 900 kg/h. Well this level of
variation makes it impossible to consider this cruise stabilized in terms of fuel flow.
Figure 4.7 - FF values variation for a flight in CS-TPA, on the 26th February 2010
The evolution of the FF parameter during cruise flight phase, according to the representation above, is
not stable but continuous high-frequency variations that will result in points with huge deviations from a mean
value, up to 20%. One can then conclude that the representative cruise points that are being recorded aren’t
actually representing the flights they belong to, unless they are close to the mean value of the cruise plateau data.
This is a probable explanation for the odd behaviour registered in Figure 4.5, where variations within each flight
result in variations in the yearly evolution of the fuel flow parameter. Having identified a possible cause to the
parameter’s behaviour throughout the year, it was then necessary to characterize the variations during the cruise
phase and find a possible source. In Figure 4.8, a close-up view from a small part of the cruise belonging to the
flight of Figure 4.7 makes it possible to better visualize the nature of the detected variations.
0
500
1000
1500
2000
2500
15:43:12
15:50:24
15:57:36
16:04:48
16:12:00
16:19:12
16:26:24
16:33:36
16:40:48
16:48:00
16:55:12
17:02:24
17:09:36
17:16:48
17:24:00
17:31:12
17:38:24
17:45:36
17:52:48
18:00:00
18:07:12
18:14:24
FuelFlowkg/h
Time
Fuel Flow CS-TPA afternoon flight
17277
17276
17277 RP
17276 RP

47
Figure 4.8 - FF values variation for a flight in CS-TPA, on the 26th February 2010 (detail)
Observing Figure 4.8, one can realize that the nature of the in-cruise variations were now clearly visible
and the analyst can visualize that the behaviour of the fuel flow parameter is almost sinusoidal, with a virtually
constant amplitude of about 120kg/h and a well defined period of about 30s. With this information, it would be
probably easier to identify the source of such variations and make efforts to reduce them. On the other hand this
representation is also valuable to easily visualize how a change of a few seconds in the requirements imposed to
identify a flight’s representative cruise point can change the final results and the final FF plot dramatically. The
utilized criteria for cruise are displayed in Table 4.2, and one of them is that the aircraft has to be in stabilized
flight in the cruise phase for 5 minutes or 300 seconds. However these were the conditions for the Embraer 145,
so this stabilizing period could be of 295, 305 or 400 seconds for example, which would produce a completely
different representative point to describe the same flight. Assuming that the 5-minute period is kept, the
objective was to find a cause and possible solution for this behaviour, while trying to reduce the incurred error
on the definition of the representative flight point.
Since the erratic behaviour described above only affects the Fokker 100 fleet, the responsible for such
variations must be something that differs between the two fleets. Because the two models are very different in
several aspects, as was explained before, the culprit may be the engines themselves, the controlling avionics and
the associated sensors. Studying the differences between all these components would be important to completely
explain these results, but one big difference between both models’ engine-related control systems springs to
mind.
The Autothrottle system provides thrust control from takeoff through landing. Autothrottle mode and
speed selection is controlled from the MCP and the TMSP and utilizes the Thrust Management Computer
(TMC). When in VNAV, the FMC selects Autothrottle modes and targets thrust values. The Autothrottle can be
operated without using the flight director or the autopilot and can be manually overridden or disconnected by
using either Autothrottle disconnect switch. The Autothrottle system moves both thrust levers together,
maintaining their relative positions, to control speed or thrust, depending on the engaged mode. Thrust levers can
be manually positioned without disconnecting the Autothrottle. After manual positioning the Autothrottle system
0
200
400
600
800
1000
1200
1400
16:04:48
16:05:14
16:05:40
16:06:06
16:06:32
16:06:58
16:07:24
16:07:49
16:08:15
16:08:41
16:09:07
16:09:33
16:09:59
16:10:25
16:10:51
16:11:17
16:11:43
FuelFlowkg/h
Time
Fuel Flow CS-TPA detail afternoon flight
17277
17276
17277 RP
17276 RP

48
repositions the thrust levers to comply with the computed thrust requirements of the engaged mode. Therefore
when the Autothrottle is engaged it constantly corrects the thrust to maintain the selected thrust or speed, which
is important to maintain a smooth flight for the passengers and to provide an easier and more relaxed flight to the
pilot, while assuring that the flight will have the duration that was expected. However, this constant and quick-
response correction by the Autothrottle originates also quick and systematic variations of the fuel flow into the
engines, in order to respond to the always-changing thrust necessity. This is the probable explanation to the
phenomenon discussed in this Section and represented in the Figure 4.5. Of course, the lack of an Autothrottle
feature in the Embraer 145 explains the smooth FF results presented for this aircraft.
This visualization problem can be a real issue because the fast fuel flow variation can affect both flight
efficiency and engine monitoring capability. In terms of air resistance and inertia, constant accelerations and
decelerations will definitely penalize the flight’s efficiency and therefore increase fuel consumption, possibly in
a significant way. Regarding the engine’s condition monitoring, the added “noise” provoked by the Autothrottle
can make much more difficult to interpret the fuel flow graphical representations, in search of spikes, steps or
other variations. Fuel flow being an important parameter to analyse engine’s health and efficiency, it seems that
the capability of effectively predict engine behaviour and detect potential serious problems will be severely
impaired.
To solve this problem, two solutions can be at first glance appointed. After the undertaking of a focused
and systematic study to really assess whether more fuel is burnt than necessary due to the Autothrottle operation,
a course of action should be chosen depending on the result of this study. If Autothrottle is not damaging the
flight in terms of fuel-burn efficiency and the problem becomes reduced to the poor graphical visualization of the
FF parameter in COMPASS, then a solution would be changing the reports uploaded into the trend monitoring
tool. After the referred cruise conditions were met, and after the five minute period given to ensure that cruise
was stable, instead of taking the data of just one second, an average could be made from a given period, in order
to reach a mean value, thus eliminating the undesired “noise”. The exact time period had to be carefully
evaluated, because a very short period might still not eliminate the quick variations and a long period would
eliminate not only the variations due to the Autothrottle but also other real variations like steps, which can be a
symptom of the existence of a problem, and therefore eliminating much of the parameter’s sensitivity and in
consequence diminishing once again the value of the engine condition monitoring. If after analysis the results
show that the Autothrottle operation is in fact contributing negatively to the fuel consumption in a significant
way, an enquire should be made to the manufacturer if there would be a possibility to either tone down the
Autothrottle sensitivity to changes in flight speed or thrust, or reduce its response to such variations, in order to
achieve more efficient flights.
The described method to reduce the “noise” associated with FF representation was applied with an
example period of 50 seconds, period chosen with the requirement of being larger than the period of the small
oscillations of Figure 4.8. The results are presented below in Figure 4.9 and systematized in Table 4.4.

49
Figure 4.9 - FF results with and without filtering, for CS-TPA flights, in January/February 2010
Parameter_Period Average (kg/hr) Deviation (kg/hr) Noise Reduction (%)
FF1_0 966,37 138,6523
18,19477
FF1_50 976,33 113,4248
FF2_0 943,51 144,0379
21,0825
FF2_50 950,43 113,6711
Table 4.4 - Compilation of FF results with and without filtering
It should be noted that there is always a certain amount of inherent scatter in any recorded flight data.
Deviations due to normal scatter should not be regarded as 'real changes' and a certain amount of judgement
must be exercised. However, this method to reduce the noise influence in the expected results had a significant
impact. The calculation of a simple average during a 50 second period, in cruise phases that can be of 7200s (two
hours) or more, which corresponds to an averaging period of 0.69% of cruise time, results in a significant
improvement of about 20% in terms of data scatter. Therefore this method proves to be effective in improving
the quality of the results and should be applied to facilitate graphical visualization and trend identification in the
future.
The remaining deviations that can be visualized in Figure 4.9 will probably have other causes, the most
important being that when calculating the representative points of each flight and comparing them all together
throughout the year, the weight of the aircraft at takeoff isn’t taken into account, an aspect that varies much more
in the Fokker fleet than in the Embraer fleet, due to its size, and affects the thrust and therefore fuel flow demand
dramatically. When in longer, busier flights, in which the aircraft’s weight is near MTOW, the required thrust
and consequently the expected fuel flow will be much larger than in shorter and lighter flights, which does not
mean that the engine is in poor condition. This situation shows that to be fully effective engine trend monitoring
has to take into account all flight aspects, especially important ones as the weight and the selected route. The
departure and arrival airports also have a small impact in the representative points generated, especially for the
takeoff RP, for which aspects like runway length and weather may significantly change the required point.
Another very important aspect that must be taken into account when determining a flight’s RP for cruise is the
0
200
400
600
800
1000
1200
1400
1600
0 20 40 60 80 100 120 140 160
FuelFlow(kg/hr)
Nr Flight
Fuel Flow Noise Reduction
FF2_0
FF2_50

50
flight level. The performance of the engine varies greatly depending on the altitude of the aircraft, for the levels
of oxygen present are crucial in determining the strength of the combustion. On the other hand lower FLs imply
more air density and thus more aerodynamic resistance. The required thrust to overcome the aerodynamic
resistance has to increase, thus increasing the fuel flow in the engines. Given the great influence of altitude in
engine performance and power demand, this aspect should also be taken into account whenever possible.
4.4 – Results
4.4.1 – Fleet Comparison with ECM
In this Section 4.4, the potential of ECM as an optimization tool will be demonstrated through some
practical examples, which represent the different ways airline companies can use this potential to their
advantage. All the results presented in this Section were produced by the COMPASS Graphical Plotting Tool,
which processed data acquired from FDR and CMC for the Fokker 100 and Embraer fleets respectively.
In the current Sub-Section, the objective will be to give an overview of both PGA fleets regarding
engine condition, namely to investigate the approximate evolutions of the previously identified Key Parameters
and compare them between the two fleets. This general analysis will show the importance of using ECM as a
means to rapidly assess the current status of a fleet in terms of engine condition and to predict its future
evolution, thus providing valuable information for long-term operational decisions.
In Figure 4.10 and Figure 4.11, the FF evolution of the Fokker 100 and the Embraer ERJ145 fleets are
respectively represented, to give the reader an idea of the current condition of the engines from both fleets. It is
however difficult to make assumptions about each engine’s fuel consumption, due to the influence of numerous
parameters, which make most engines seem to be improving throughout the year, against what was expected.
Figure 4.10 - FF approximate linear trends the Fokker 100 fleet
-3-2-1012345
01-01-2010
18-01-2010
06-02-2010
01-03-2010
24-03-2010
12-04-2010
05-05-2010
31-05-2010
23-07-2010
08-08-2010
27-08-2010
DeltaFuelFlow(%)
Date
DFF Fokker 100 Fleet
Linear (17216)
Linear (17217)
Linear (17218)
Linear (17228)
Linear (17276)
Linear (17277)
Linear (17283)
Linear (17317)
Linear (17318)
Linear (17353)
Linear (17392)
Linear (17420)
Linear (17205)

Figure 4.11 -
By comparing the two fleets directly through the FF parameter, it is possible to see that in this aspect
the Embraer fleet will be globally more efficient, for it presents smaller FF deviations between the registered
values and the nominal ones, so the added FF consumption due to engine wear will be higher on the Fokker fleet.
The AE3007 engines also present much more stable profi
corresponding to engine degradation, which is also a result of the rising temperatures as the year progresses. The
AE3007 engines are also comprised in the 0
deviation from the theoretical FF results.
In order to further compare the condition of both fleets’ engines, as well as the engine’s reliability to
deliver consistent results in terms of fuel flow, turbine temperature and high pressur
values for these parameters have been put together in the plot displayed in Figure 4.12.
Figure 4.12 - Comparison of the scatter of DN2, DTGT/DITT and DFF parameters between fleets
The results discussed above regarding
scatter corresponding to the TAY engines is much larger than the one presented by the AE3007 engines. This
elevated variation for the TAY engines happens for all parameters, which can be explain
0
1
2
3
4
5
6
7
01-01-2010
21-01-2010
10-02-2010
DeltaFuleFlow(%)
-7
-5
-3
-1
1
3
5
23-12-2009
12-01-2010
01-02-2010
21-02-2010
DeltaParameters(%)
- FF approximate linear trends the Embraer 145 fleet
be globally more efficient, for it presents smaller FF deviations between the registered
also present much more stable profiles in terms of FF, with a small increase in Delta values
AE3007 engines are also comprised in the 0-3% interval, with several engines presen
esults.
deliver consistent results in terms of fuel flow, turbine temperature and high pressure rotor speed, the daily mean
values for these parameters have been put together in the plot displayed in Figure 4.12.
Comparison of the scatter of DN2, DTGT/DITT and DFF parameters between fleets
The results discussed above regarding the DFF can be confirmed by Figure 4.12, where especially the
elevated variation for the TAY engines happens for all parameters, which can be explain
10-02-2010
02-03-2010
22-03-2010
11-04-2010
01-05-2010
21-05-2010
10-06-2010
30-06-2010
20-07-2010
Date
Linear (310037)
Linear (311079)
Linear (311079)
Linear (311089)
Linear (310085)
Linear (311077)
Linear (310097)
Linear (310070)
Linear (310086)
Linear (310086)
Linear (310098)
Linear (310071)
Linear (310057)
DFF Embraer 145 Fleet
21-02-2010
13-03-2010
02-04-2010
22-04-2010
12-05-2010
01-06-2010
21-06-2010
11-07-2010
31-07-2010
Date
Fleet Scatter Comparison
DFF AE3007
DN2 AE3007
DITT AE3007
DFF TAY650
DN2 TAY650
DTGT TAY650
Linear (DFF TAY650)
Linear (DN2 TAY650)
Linear (DTGT TAY650)
51
FF approximate linear trends the Embraer 145 fleet
be globally more efficient, for it presents smaller FF deviations between the registered
les in terms of FF, with a small increase in Delta values
3% interval, with several engines presenting less than 1% of
e rotor speed, the daily mean
Comparison of the scatter of DN2, DTGT/DITT and DFF parameters between fleets
gure 4.12, where especially the
elevated variation for the TAY engines happens for all parameters, which can be explained by the previously
Linear (310037)
Linear (311079)
Linear (311079)
Linear (311089)
Linear (310085)
Linear (311077)
Linear (310097)
Linear (310070)
Linear (310086)
Linear (310086)
Linear (310098)
Linear (310071)
Linear (310057)
DFF AE3007
DN2 AE3007
DITT AE3007
DFF TAY650
DN2 TAY650
DTGT TAY650
Linear (DFF TAY650)
Linear (DN2 TAY650)
Linear (DTGT TAY650)

referred limitation of the Fokker fleet regarding data collecting and filtering
problem affects mainly the FF parameter
degree the DTGT and DN2 parameters are also affected. The mean values concerning the DN2 parameter are
stable and low for both fleets, where the TAY engines present the best results.
present relatively stable behaviours, with frequent var
conclusion that the AE3007 engines are in general more reliable, more stable and present smaller deviations to
the nominal values than the TAY engines.
Figure 4.13 - DFF values and approximate linear
Observing once again Figure 4.11, one can see that there is only one engine that stands out within the
Embraer fleet, engine 310085, which presents clearly more elevated DFF values than the rest of the fleet, along
with a constant tendency to increase these values. This evolution represents a possible high rate of degradation of
the engine’s condition and therefore the engine should be closely monitored. In Figures 4.13, 4.14 and 4.15 the
behaviour of this engine will be further analysed and compared to the other on
Embraer 145 aircraft and the fleet’s average evolution.
In Figure 4.13, the FF parameter is only represented for
the CS-TPJ aircraft, namely engine 310085 which presented an odd behaviour in Figure 4.11, and an additional
engine with more usual behaviour, placed in position #2. The fleet average results and the ones
311077 work as a baseline for “normal” operation, presentin
in DFF over a seven months period, which corresponds to expected engine degradation. However, the engine
310085 plot shows an increase in DFF of about 2%, which has to be taken into consideration. Even more
because this tendency seemed to be installed for some time because at the end of July the engine had a DFF
value of about 6.5%, which is significant when comparing to the rest of the fleet.
-1
0
1
2
3
4
5
6
7
8
23-12-2009
12-01-2010
01-02-2010
DeltaFuelFlow(%)
referred limitation of the Fokker fleet regarding data collecting and filtering – see Section 4.3.
mainly the FF parameter, because it stacks up with the Autothrottle action,
nd DN2 parameters are also affected. The mean values concerning the DN2 parameter are
stable and low for both fleets, where the TAY engines present the best results. The DTGT and DITT parameters
present relatively stable behaviours, with frequent variations of small deviation. This analysis allows the
the nominal values than the TAY engines.
DFF values and approximate linear trend, for both engines of CS
h a constant tendency to increase these values. This evolution represents a possible high rate of degradation of
e further analysed and compared to the other on-wing engine in the CS
Embraer 145 aircraft and the fleet’s average evolution.
, the FF parameter is only represented for the fleet average and for
t, namely engine 310085 which presented an odd behaviour in Figure 4.11, and an additional
engine with more usual behaviour, placed in position #2. The fleet average results and the ones
311077 work as a baseline for “normal” operation, presenting slight increases of approximately 0.3% and
period, which corresponds to expected engine degradation. However, the engine
value of about 6.5%, which is significant when comparing to the rest of the fleet.
01-02-2010
21-02-2010
13-03-2010
02-04-2010
22-04-2010
12-05-2010
01-06-2010
21-06-2010
11-07-2010
31-07-2010
Date
Fleet Average
310085
311077
Linear (Fleet
Average)
Linear (310085)
CS-TPJ Delta Fuel Flow
52
see Section 4.3. This input
, because it stacks up with the Autothrottle action, but also in a smaller
nd DN2 parameters are also affected. The mean values concerning the DN2 parameter are
The DTGT and DITT parameters
This analysis allows the
trend, for both engines of CS-TPJ
h a constant tendency to increase these values. This evolution represents a possible high rate of degradation of
wing engine in the CS-TPJ
the fleet average and for the engines placed in
t, namely engine 310085 which presented an odd behaviour in Figure 4.11, and an additional
engine with more usual behaviour, placed in position #2. The fleet average results and the ones for engine
g slight increases of approximately 0.3% and 0.8%
period, which corresponds to expected engine degradation. However, the engine
Fleet Average
Linear (Fleet
Linear (310085)

Figure 4.14 - DN2 values and approximate 6
Figure 4.15 - DITT values and approximate 6
Observing Figure 4.14 and 4.15, it can be easily seen that the
the 310085 engine are similar to those displayed by the fleet, generally stable profiles with some small natural
increases over time. Also in terms of mean values, the discussed engine presents normal results, not reflecting
the trend registered for the DFF parameter. With
APPENDIX I and through the experience of the engineering and maintenance
that this trend meant that this engine had to be closely and frequently monitored. On the 23
phenomenon of burn-through took place in a few blades belonging to the high pressure turbine of engine
310085, like the one described in the next Sub
was detected shortly after it took place avoided further damage, thanks to the up close monitoring of the engine
which was being conducted. This particular case is yet another example of how important ECM is, not only for
cost-reduction or maintenance-optimization purposes, but al
0
0,5
1
1,5
2
2,5
23-12-2009
12-01-2010
01-02-2010
DeltaN2(%)
-10
-9
-8
-7
-6
-5
-4
-3
-2
-1
0
23-12-2009
12-01-2010
01-02-2010
DeltaITT(%)
DN2 values and approximate 6th degree polynomial trend, for both engines of CS
DITT values and approximate 6th degree polynomial trend, for both engines of CS
Observing Figure 4.14 and 4.15, it can be easily seen that the DN2 and DITT parameter evolutions for
are similar to those displayed by the fleet, generally stable profiles with some small natural
the trend registered for the DFF parameter. With the reference of the Trend Guideline Chart displayed in
APPENDIX I and through the experience of the engineering and maintenance (E&M) department, PGA
that this trend meant that this engine had to be closely and frequently monitored. On the 23
through took place in a few blades belonging to the high pressure turbine of engine
310085, like the one described in the next Sub-Section. Although not a serious problem, the fact that the event
r it took place avoided further damage, thanks to the up close monitoring of the engine
optimization purposes, but also to increase flight safety.
01-02-2010
21-02-2010
13-03-2010
02-04-2010
22-04-2010
12-05-2010
01-06-2010
21-06-2010
11-07-2010
31-07-2010
Date
Fleet Average
310085
311077
Poly(Fleet Avg)
Poly(310085)
Poly(311077)
CS-TPJ Delta N2
01-02-2010
21-02-2010
13-03-2010
02-04-2010
22-04-2010
12-05-2010
01-06-2010
21-06-2010
11-07-2010
31-07-2010
Date
Fleet Average
310085
311077
Poly(Fleet Avg)
Poly(310085)
Poly(311077)
CS-TPJ Delta ITT
53
trend, for both engines of CS-TPJ
degree polynomial trend, for both engines of CS-TPJ
parameter evolutions for
are similar to those displayed by the fleet, generally stable profiles with some small natural
the reference of the Trend Guideline Chart displayed in
department, PGA decided
that this trend meant that this engine had to be closely and frequently monitored. On the 23rd
September 2010 a
through took place in a few blades belonging to the high pressure turbine of engine
Section. Although not a serious problem, the fact that the event
r it took place avoided further damage, thanks to the up close monitoring of the engine
Fleet Average
Poly(Fleet Avg)
Poly(310085)
Poly(311077)
Fleet Average
Poly(Fleet Avg)
Poly(310085)
Poly(311077)

54
4.4.2 – ECM as a Problem Identification Tool – Drop in ITT Margin
In this Sub-section and in following Sub-sections 4.4.3 and 4.4.4, and also in Sub-Section 5.3.3, some
important examples will be presented to show the various real day-to-day situations where Engine Condition
Monitoring has proven to be invaluable, namely in identifying and analysing detected problems, in predicting
and preventing future malfunctions, and by quantifying the effects of maintenance actions like compressor
washes and therefore analysing whether those maintenance actions are cost-efficient or not.
In this particular Sub-section, let us focus on the ECM’s capability of detecting non-conformities or an
unplanned event, of quantifying it and finding possible causes for that non-conformity. The main reason why the
ECM should be conducted in the most regular basis possible is because it constitutes a tool which can detect
small parameter variations, sometimes not noticeable to the flight crew, but which may represent the beginning
of the propagation of a serious, expensive and potentially dangerous problem. After quantifying the detected
variations, through prediction models and experience, it is possible to point possible causes for these variations
and proceed with the necessary maintenance actions.
In order to better assess engine condition and to verify whether the existing safety margins are sufficient
to guarantee airworthiness during any mechanical malfunction or similar event, COMPASS software determines
other variations of the referred Key Parameters that are better suited to give the user a more clear grasp of the
impact of such events in engine condition in terms of safety. Parameter ‘nominal’ values are calculated by
COMPASS based on the engine model and on the supplied aircraft and engine input parameters such as airspeed,
TAT, altitude, N1, bleed status and Anti-Ice status. Because the particular event studied in this Sub-Section was
mostly noticed during Take-off, this analysis will focus on the Take-off data produced by the Embraer’s CMC.
As was already discussed in Section 4.2, Delta Parameters are calculated as the difference between the actual
measurements and the 'nominal' expected values. In general, Parameter Margins can be defined by the difference
between the red-line parameter value, a representation of a safety threshold, and the parameter value of the
engine at maximum take-off rating at a declared flight condition. Parameter Margins are used to calculate output
parameters which can be monitored to ensure operation of the engine is within certified limits. Margins are
calculated at sea-level worst case conditions and at a condition representing absolute engine worst case
operation. The Take-off SLOATL, Sea Level Outside Air Temperature Limit, is a measure of the ambient
temperature at which the engine would have zero take-off margin if maximum take-off rating were used. These
definitions are necessary to better understand the diagram below, which illustrates the margin calculation output
parameters, together with an interpretation guide to the SLOATL parameter output.

55
Figure 4.16 - Margin parameters; interpretation guide to the engine limitation depending on SLOATL
ITT and shaft speed margin should be monitored to ensure adequate margin. An engine which has zero
margin as indicated by COMPASS, will only be limited if maximum take-off rating is used at the flight
condition specified for the margin calculation. The smoothed parameter margin trends should be monitored in
order to predict when such a limitation might occur.
After an introduction to the parameters that were used in this Sub-Section’s analysis, let us present the
actual event being currently studied. When PGA was routinely conducting ECM to their Embraer fleet in late
August 2009, an unusual behaviour was detected in the #1 engine of CS-TPH, a few days before, on the 27th
August. This event was defined by a sudden and significant drop of more than 35ºC in the Sea Level ITT
Margin, accompanied by smaller variations in other parameters such as an increase in Sea Level High Pressure
Speed Margin. These variations are represented in graphical form in Figure 4.17 and 4.18.
Figure 4.17 - Variation of smoothed SL N2 margin parameter for Embraer CS-TPJ
00,511,522,53
03-08-2009
04-08-2009
06-08-2009
08-08-2009
09-08-2009
12-08-2009
13-08-2009
14-08-2009
16-08-2009
18-08-2009
19-08-2009
21-08-2009
23-08-2009
25-08-2009
27-08-2009
28-08-2009
30-08-2009
01-09-2009
03-09-2009
N2Margin(%)
Date
Smoothed Sea Level HP Speed Margin
311088: N2__MS (I) 310057: N2__MS (I)
311088: N2__MSL (I) 310057: N2__MSL (I)

Figure 4.18 - Variation of smoothed SL ITT margin parameter for Embraer CS
Analysing the plots above, the event is clearly visible, consis
Margin at Take-off of approximately 35.7ºC
was a sudden increase in SL N2 Margin of about 0.5%
311088 engine, while engine 310057
with the help of the Trend Guideline Chart
Guide[6] supplied by the manufacturer
trim plug failure, ITT thermocouple
Figure 4.19 - Boroscope inspection to hpt blades in the SN311088 engine
After close inspection with a boroscope (Figure 4.1
some high pressure turbine blades, a
kind of damage doesn’t usually compromise flight safety, another engine was placed on wing and the engine in
question was sent for repair, where the worn parts were replaced.
020406080100120140
03-08-2009
04-08-2009
06-08-2009
ITTMargin(ºC)
ariation of smoothed SL ITT margin parameter for Embraer CS
Analysing the plots above, the event is clearly visible, consisting of a sudden decrease of
approximately 35.7ºC in total and about 25.8ºC in a single flight. Si
N2 Margin of about 0.5%. It is also clear that this variation only takes place for the
311088 engine, while engine 310057 maintains its normal operation. Through the company’s experience and
the Trend Guideline Chart presented in Appendix I and the Engine Trend Plot Interpretation
supplied by the manufacturer, it was found that the most likely causes for these variations were
ouple failure or some kind of damage in the high pressure system.
Boroscope inspection to hpt blades in the SN311088 engine
After close inspection with a boroscope (Figure 4.19), significant wear was detected in the casing of
some high pressure turbine blades, as can be visualized in Figure 4.20 in the right bottom corner.
question was sent for repair, where the worn parts were replaced.
y = -0,0778x + 107,32
06-08-2009
08-08-2009
09-08-2009
12-08-2009
13-08-2009
14-08-2009
16-08-2009
18-08-2009
19-08-2009
21-08-2009
23-08-2009
25-08-2009
27-08-2009
28-08-2009
30-08-2009
01-09-2009
03-09-2009
Date
Smoothed Sea Level ITT Margin
311088: ITT_MS (I) 310057: ITT_MS (I)
311088: ITT_MSL (I) 310057: ITT_MSL (I)
Linear (310057: ITT_MSL (I))
56
ariation of smoothed SL ITT margin parameter for Embraer CS-TPJ
ting of a sudden decrease of SL ITT
single flight. Simultaneously there
It is also clear that this variation only takes place for the
maintains its normal operation. Through the company’s experience and
in Appendix I and the Engine Trend Plot Interpretation
causes for these variations were ITT
d of damage in the high pressure system.
Boroscope inspection to hpt blades in the SN311088 engine
wear was detected in the casing of
in the right bottom corner. Even if this

57
Figure 4.20 - Boroscope inspection to hpt blades in the SN311088 engine (detail)
Analysing Figure 4.18’s plot further, it can be noted that, aside from the previously described event,
both engines show a constant and slow decrease in the Sea Level ITT Margin, which happens because of the
natural degradation of the engines. Because the ITT Margin should always be positive and as large as possible,
the rate at which the margin decreases has to be continuously monitored. Furthermore, it is even possible to
predict how many flights the engine can endure until it runs out of margin and has to be repaired or refurbished.
The evolution of the SL ITT Margin for engine 310057 was found to have an approximate linear behaviour,
represented by the yellow line in the plot in question. This linear function corresponds to the following
equation: yሺxሻ ൌ െ0,0778x ൅ 107,32, which means that in the beginning of May 2009 this engine had a SL ITT
Margin of 107.32ºC and that it was losing that margin by 0.0778ºC in every flight. The engine endured 182
flights during the represented one month period, losing 14.16ºC of SL ITT Margin. Therefore to calculate how
many flights could be done before the engine reaches zero margin, it was necessary to simply divide the margin
at the end of the studied period (approximately 93.16ºC) by 0.0778, and divide the result by the average number
of daily flights:
Days to Zero Margin ൌ
Current Margin
Rate of Degradation ൈ Daily Usage
ൌ
93,16
0,0778 ൈ 5,5
ൌ 217,76 ൎ 217 days
Equation 4.1
Assuming the aircraft flies on average 5.5 times per day, this means that the engine will run out of
margin in 217 days. The ability to predict when the engine will reach a point where maintenance action is needed
is valuable to help operators to better organize their maintenance actions and personnel, having the aircraft flying
as much as possible, without the risk of compromising safety, as long as the engine is continually monitored.
4.4.3 – ECM as a Pre-emptive Tool - Vibrations High Pressure Shaft
As aircraft engines evolved and vibration monitoring was introduced, the purpose of vibration
monitoring was not so much to provide detailed mechanical diagnostics to reduce maintenance costs as it was to
provide very basic protective functions for failures that might be encountered during a flight. In recent years,
vibration monitoring has gained importance as a mechanical diagnostics tool, apart from maintaining its original

58
basic protective function discussed above. Nowadays, fierce competition within the commercial aviation market
has increased the level of demand, not only to maintain air safety, but to do it cheaper.
ECM and particularly vibration monitoring have become important tools to optimize engine
performance, thus reducing costs, predicting deviations to normal behaviour, finding possible causes for such
behaviour and eventually enforcing maintenance actions in optimized timings. Also in terms of flight safety,
there is now a more conscious and heavy regulated aviation industry, which makes an effort in identifying and
avoiding structural faults that can compromise airworthiness, do additional damage to the engine or even
degrade flight performance, and vibration monitoring is a key aspect to identify such faults preventively.
Unusual wear or mechanical faults or cracks can be identified after the vibration monitoring indicates steps or
spikes in vibration parameters. An example of a vibration shaft analysis conducted to an Embraer 145 aircraft is
shown in Figure 4.21 and Figure 4.22 and the obtained results are discussed below.
Figure 4.21 - HP shaft vibration values and approximate linear trend during Takeoff in CS-TPM
Figure 4.22 - HP shaft vibration values and approximate linear trend during Cruise in CS-TPM
In Figure 4.21 and Figure 4.22, evolutions of a vibration parameter corresponding to the high-pressure
shaft are represented, for the Takeoff and Cruise flight phases respectively, and for both engines of the Embraer
00,20,40,60,811,2
01-01-2010
24-01-2010
18-02-2010
26-02-2010
07-03-2010
15-03-2010
01-04-2010
11-04-2010
21-04-2010
29-04-2010
19-05-2010
27-05-2010
04-06-2010
14-06-2010
22-06-2010
30-06-2010
07-07-2010
14-07-2010
22-07-2010
30-07-2010
08-08-2010
HPShaftVibration(IPS)
Date
HP Shaft Vibration Takeoff
310008: VB2 (I) 310037: VB2 (I)
Linear (310008: VB2 (I)) Linear (310037: VB2 (I))
00,10,20,30,40,50,6
01-01-2010
10-01-2010
30-01-2010
19-02-2010
01-03-2010
07-03-2010
14-03-2010
23-03-2010
05-04-2010
12-04-2010
22-04-2010
30-04-2010
19-05-2010
24-05-2010
30-05-2010
09-06-2010
17-06-2010
22-06-2010
30-06-2010
07-07-2010
13-07-2010
20-07-2010
26-07-2010
02-08-2010
09-08-2010
HPShaftVibration(IPS)
Date
HP Shaft Vibration Cruise
310008: VB2 (I) 310037: VB2 (I)
Linear (310008: VB2 (I)) Linear (310037: VB2 (I))

59
ERJ-145 CS-TPM aircraft. This particular aircraft and period were more attentively analysed because, in the
representation of the high pressure shaft vibration for the SN310037 engine during the cruise phase, some of the
higher values triggered an imposed warning, exceedances represented as big spikes in Figure 4.22. It was
therefore necessary to analyse the behaviour of this parameter on this engine to find and correct possible causes.
A linear trend approximation has been attributed to each case, in order to easily quantify the variations of
average values, as a whole. These linear functions allow the observer to see the parameter’s evolution, beyond
the already discussed spikes, and quantify the average variation of a determined parameter.
Firstly, it should be noted that engine SN310008 presents a roughly stabilized profile with low
amplitude variations, therefore it is not suspected of having any significant vibration problems, and consequently
will be used in this analysis as a reference for near-normal behaviour in terms of engine high pressure shaft
vibration.
Comparing the vibration evolutions for the SN310037 engine for both flight phases, it becomes evident
that both present behaviours that can be causes for concern, although in different ways. In Figure 4.21, the
variation in vibration represented by the linear function is quite large, of about 0.4 ips, resulting in final takeoff
vibrations of about 0.7 ips, much higher than those presented by the reference engine, a behaviour that should
definitely be further analysed. However, the vibration crescendo is continuous without big steps or spikes, which
points to gradual structural wear, although faster than usual. In opposition, the linear function enables analysts to
see that the vibration in the SN310037 engine during the cruise phase only increases in about 0.1 on average, still
staying comfortably below the other engine, while the represented spikes are the real cause of concern,
suggesting an unidentified, more localized and more specific structural problem than wear.
The large spikes are the result of some underlying problem during more demanding flights namely in
terms of weight. Obviously an increase of weight will result in an increase of demand of thrust and thus a higher
rotation speed which results in more vibration. The same happens in a smaller degree with the outside
temperature, when it gets hotter the engine becomes less efficient and naturally will need to maintain thrust
through an increase in speed rotation, which again will increase shaft vibration. This would be its theoretically
predicted behaviour, which for example in the case of the reference engine isn’t verified. After this brief
analysis, it’s already possible to conclude that some mechanical fault or wear mechanism exists in one of the
aircraft’s engines, and that it can be mainly described by a continuous increase in vibration since the beginning
of the year, accompanied with some alarming spikes that in fact aren’t as worrying as the steady climb. Vibration
monitoring detected a problem, maybe an accelerated wear problem, assessed that it its effects are growing and
that this engine and the problem should be followed more closely, but that at the time there is no need to call a
maintenance action for this engine. The vibration monitoring would continue in a regular basis and if the
vibration values would become higher that the imposed safety limits, the engine would be inspected and
probably removed, but the constant monitoring of the situation also allows the maintenance team to better predict
when maintenance action will be necessary and schedule that action to the most convenient moment.

60
4.4.4 – Overhaul or Midlife Influence in Engine Performance
The overhaul of an engine provides an engine with a completely new life cycle and the engine goes out
into service embodied with improved design details, thus making it more reliable than before. The overhaul
procedure consists of stripping, washing, inspecting, building and testing the engine.
Due to high thermal and mechanical stresses, together with extremely tight operating clearances, the gas
turbine rotor is the component which requires the most upkeep and maintenance. The blades and rotor have
calculated life expectancies and need to be refurbished, repaired, and/or replaced after a certain number of
operating hours. Some parts which are critical to ensure flight safety, which were already mentioned and
described in Section 3.1, have to be replaced after a determinate number of cycles, usually during an overhaul or
mid-life process. Usually whole modules are replaced, to ensure that all related critical parts are in the same
condition at a certain moment, to facilitate behaviour prevision.
Following the replacement of critical parts/modules within the engine and the thorough disassembly,
cleaning, repairing and reassembly processes, an overhauled engine should present greatly improved
characteristics and performance, comparing to long-installed engines which haven’t been recently refurbished.
Through the Engine Condition Monitoring, utilizing the COMPASS software’s tools it’s possible to roughly
quantify the improvement in the Key Parameters’ results for any given engine that has gone through an overhaul,
midlife or any other maintenance process. It is yet another example of what can be gained when using ECM,
inspection and maintenance as one, with the objective of optimizing the operations of the company cost-wise,
while increasing the confidence and flight safety levels.
Figure 4.23, Figure 4.24 and Figure 4.25 contain graphical representations of the evolution of the DFF,
DN2 and DTGT parameters from the 1st
April 2010 to the 20th
May 2010, to illustrate the change between two
AE3007 engines in the CS-TPI Embraer 145 PGA aircraft. The engine SN17353 was removed from the right
position of the aircraft on the 29th
April and the engine recently-refurbished SN17205 was assigned to its place.
Engine SN17228 was kept in the first position during the referred period and can be considered as a reference in
terms of outside, flight and operational conditions.

61
Figure 4.23 – DFF variation with engine change in CS-TPE in 29/04/2010
Figure 4.24 – DN2 variation with engine change in CS-TPE in 29/04/2010
Figure 4.25 - DTGT variation with engine change in CS-TPE in 29/04/2010
On a previous note, it is important to notice that all the parameters plotted on the above pictures are in
fact Smoothed Delta Key Parameters as described in Section 4.3, defined in Representative Cruise Points.
Furthermore, generally comparing the three plots, it is clear that DFF and DTGT present constant, short-period
-4-20246
01-04-2010
03-04-2010
05-04-2010
07-04-2010
12-04-2010
15-04-2010
18-04-2010
27-04-2010
30-04-2010
12-05-2010
14-05-2010
15-05-2010
18-05-2010
19-05-2010
DeltaFuelFlow(%)
Date
Engine Change - Delta Fuel Flow
17228: DFF___L (I) 17353: DFF___L (I) 17205: DFF___L (I)
-2-1012
01-04-2010
03-04-2010
05-04-2010
07-04-2010
12-04-2010
15-04-2010
18-04-2010
27-04-2010
30-04-2010
12-05-2010
14-05-2010
15-05-2010
18-05-2010
19-05-2010
DeltaN2(%)
Date
Engine Change - Delta N2
17228: DN2___L (I) 17353: DN2___L (I) 17205: DN2___L (I)
-1-0,500,511,52
01-04-2010
02-04-2010
03-04-2010
04-04-2010
05-04-2010
06-04-2010
07-04-2010
11-04-2010
12-04-2010
14-04-2010
15-04-2010
16-04-2010
18-04-2010
21-04-2010
27-04-2010
29-04-2010
30-04-2010
01-05-2010
12-05-2010
13-05-2010
14-05-2010
15-05-2010
15-05-2010
17-05-2010
18-05-2010
18-05-2010
19-05-2010
20-05-2010
DeltaTGT(%)
Date
Engine Change - Delta TGT
17228: DTGT__L (I) 17353: DTGT__L (I) 17205: DTGT___L (I)

62
variations, which makes the visualization of possible trends more difficult, a problem which was addressed in
Section 4.3, and thus will be overlooked in this and following analyses.
Observing Figures 4.23 through 4.25, it is clear that the refurbished engine presents values
approximately averaging zero, which means that this engine’s operational results are very similar to those
predicted by the engine’s theoretical models. This proximity between the practical and theoretical results is a
very good indication of how well the engine was refurbished, because a good proximity means that the used
engine was roughly returned to its as-new condition, with minimal losses due to wear or degradation. In the
limit, one could consider at this point that any deviations from the theoretical model, in this case represented by
the horizontal axis, exist solely due to the variations of flight conditions and changes to the operational profiles.
Regarding the actual engine permute, the first noticeable aspect is that the engine change produces
significant and graphically obvious results, as the parameters that describe the condition of the right side engine,
initially represented by the blue lines (engine SN17353), are replaced by the green lines which correspond to the
SN17205 engine. For the DFF parameter for example, the on-wing engine before the change presented average
values of DFF of roughly 2.1% and after the permute the (new) equipped engine averaged values of DFF of
about -0.5%, which corresponds to a reduction of 2.6% in engine fuel flow. This improvement will definitely
have consequences in terms of the aircraft consumption as a whole and reduce fuel costs in some degree.
Also for the DTGT parameter plotted in Figure 4.25 the improvement is from about 1.5% to 2%, which
is very significant in terms of engine temperatures if one bears in mind that it is equivalent to a reduction of
roughly 12ºC to 18ºC in the turbine average temperature, which after many flights makes a terrible difference in
terms of wear caused by thermal low-cycle fatigue.
Regarding the DN2 parameter, the high-pressure shaft rotation speed of the pre-change engine
presented values closer to those theoretically predicted, so the improvement wasn’t as significant, with a
reduction of about 1% in terms of average values. The DN2 parameter for the SN17353 was in fact closer to
predictions than for the SN17228, which means that in this aspect the latter would benefit more from a
refurbishment than the first, although the engines are often replaced not for being more degraded but because of
the arise of a non-predicted maintenance operation which implies the removal of the engine. Because the aircraft
cannot be grounded more time than absolutely necessary, a spare engine is the solution to guarantee safety
without cancelling flights, which is always the worst situation for an airline company.
A relevant question can be now proposed: after an engine overhaul, for how long does the engine
present better conditions relatively to the fleet’s average? In other words, it would be valuable to have an idea on
how the engine’s condition degrades, in order to better predict when new refurbishment would be necessary, to
coordinate that process with other maintenance programmed actions and therefore facilitate fleet and resource
management. Furthermore, can engine trend monitoring assist on such behaviour understanding and prediction?

There are few publications indicating the rate of degradation on gas turbines, howev
many instances, the initial degradation on a new engine is seen as more rapid than the degradation after several
thousand hours of operation. One design feature that can explain this behaviour as it slows down degradation is
the effort to thermally match stationary and rotating parts of the engine. This means that the thermal growth of
components is matched so that the running clearance remains constant during thermal cycles. Dat
Veer (2004) [5], indicate that both recovera
a large rate of initial degradation that is reduced gradually. As an example, the nonrecoverable degradation of the
power output of his data indicated a loss of power of 3.5% in the fir
the next 5000 hours.
Figure 4.26 – DN1 values before and after overhaul of engine SN17317
Figure 4.27 - TGT values for critical N1
In Figure 4.26 and Figure
representations of two important parameters when engine con
15 engines SN17317 and SN17392
respectively. The plots represent the evolutions of N1
the last years, and clearly show that there is a logarithmic
varying deterioration rate of the engine. In
80
81
82
83
84
85
86
87
C160205 C020505
N1(%)
680,0
690,0
700,0
710,0
720,0
730,0
2006
TGT(ºC)
N2 Climb
There are few publications indicating the rate of degradation on gas turbines, howev
to thermally match stationary and rotating parts of the engine. This means that the thermal growth of
components is matched so that the running clearance remains constant during thermal cycles. Dat
, indicate that both recoverable and nonrecoverable degradation follow a logarithmic pattern, i.e.,
power output of his data indicated a loss of power of 3.5% in the first 5000 hours, followed by only 0.5% over
DN1 values before and after overhaul of engine SN17317
TGT values for critical N1 and approximate logarithmic trend
and Figure 4.27, this logarithmic tendency is illustrated through the graphi
arameters when engine condition assessment is concerned.
SN17392 were submitted to overhauls at the end of 2005 and beginning of 2
The plots represent the evolutions of N1 and TGT respectively for the SN17392 engine
the last years, and clearly show that there is a logarithmic-like trend in these parameters, which reflect the
the engine. In both graphs, the points represent averages of critical points already
y = 2,7905ln(x) + 80,136
C020505 B090206 B150406 B150706 B141106 B130207 B180407 B140807
Aircraft and Date
N1 Evolution with OVH
Post OVH Pre OVH LOG(Post OVH)
y = 9,9475ln(x) + 707,73
y = 7,7709ln(x) + 690,16
2006 2007 2008 2009
Year
TGT with Max N2
N2 Climb N2 Takeoff LOG(N2 Climb) LOG(N2 Takeoff)
63
There are few publications indicating the rate of degradation on gas turbines, however it seems that, in
to thermally match stationary and rotating parts of the engine. This means that the thermal growth of
components is matched so that the running clearance remains constant during thermal cycles. Data compiled by
ble and nonrecoverable degradation follow a logarithmic pattern, i.e.,
st 5000 hours, followed by only 0.5% over
DN1 values before and after overhaul of engine SN17317
and approximate logarithmic trend
rithmic tendency is illustrated through the graphical
dition assessment is concerned. Engines TAY650-
the end of 2005 and beginning of 2006
for the SN17392 engine, throughout
like trend in these parameters, which reflect the
averages of critical points already

64
utilized to determine the flight profile in Chapter 3, in the Figure 4.26 case the presented parameter is the low-
pressure rotor speed N1 (maximum values) and in the Figure 4.27 the parameter in question is the TGT as the N2
reached its maximum values. These points were used in this analysis solely because there wasn’t enough data for
the last years to determine cruise or takeoff Representative Points as they were defined in Section 4.2.
Nevertheless, the conditions used to determine the flight profile are also acceptable to assess engine condition as
long as they are consistently used (like in the flight profile determination), even if the Representative Points
would give more accurate results. The achieved representations therefore show truthfully, at least in a qualitative
manner, the already referred tendencies that reflect the degradation of TAY650-15 engines SN17317 and
SN17392, in Figure 4.26 and 4.27 respectively. They also constitute another example how engine condition
monitoring can predict engine behaviour, through theory models and practical experience, analysing possible
deviations to the expected results.
Proper maintenance and operating practices can significantly affect the level of performance
degradation and thus the time between repairs and overhauls of a gas turbine. To accurately quantify the
relationship between them is a challenge because it is subject to a variety of operational and design factors that
typically cannot be entirely controlled. Proactive condition monitoring will allow the operator to make intelligent
service decisions based on the actual condition of the fleet’s engines rather than on fixed and calendar based
maintenance intervals. Maintenance and overhaul decisions are ultimately based on economic and safety
considerations. Understanding performance degradation, as well as factors that influence degradation can help in
these decisions.

65
5 – Fuel Monitoring
5.1 - Monitoring as a Route to Savings
The recent world economical and financial crisis has taken a toll in all fields of industry but
especially on those related with leisure, tourism, IT and machinery like the automotive industry. Commercial
aviation was also affected by this crisis, with reducing profit margins and several bankruptcies. It is therefore
natural that companies look for opportunities to optimize their operation, taking steps to reduce costs wherever
possible.
The analysis conducted in Chapter 3 had the objective to look for a reduction in costs associated with
engine critical parts. In Chapter 4, an ECM tool was used to better comprehend the behaviour of the engines,
allowing the operator to balance engine use and optimize the planning of necessary maintenance actions. In this
Chapter, the objective is to understand how constant monitoring of fuel consumption throughout both PGA fleets
can provide the information needed to optimize flight operations and detect cases where excessive fuel is being
consumed, pointing possible causes and corrective measures. Fuel monitoring, associated with other data such as
TOW, number of passengers and cargo, flight duration and route, can be used in a broad spectrum of analysis
namely:
Comparing the efficiency of different fleets and of different aircraft within a fleet
Determining the profitability of each route throughout the year
Quantifying each route’s ground efficiency, GE ൌ Block Hours Flight Hours ሺ%ሻ⁄
Characterizing aircraft performance in terms of fuel flow throughout the year
Comparing operational results of the current year with the previous year
Figure 5.1 - Evolution of kerosene prices, in US cents per US gallon [7]
Optimization is sought in every aspect of operation these days, but because of the impact that fuel has
in the total expenses of airline companies, it is understandable that continuous efforts are made to improve fuel
consumption efficiency. These efforts have been reinforced in the last few years, as the recurrent global

66
economical instability and a constant rise in the price of the oil barrel have threatened the viability of many
airline companies. In Figure 5.1, the constant rise of kerosene jet fuel over the last decade is represented.
Small airline companies like PGA, operating a medium size fleet of regional jet (RJ) aircraft on short
and medium-haul routes, are confronted with some operational factors that penalize the flights’ efficiency,
making PGA’s necessity to optimize operation even more critical. PGA’s aircraft:
Fly shorter stage lengths than large aircraft spending more time at airports, taxiing and manoeuvring,
therefore in non-optimum, non-cruise stages of flight
Require similar runway length to large aircraft, not taking advantage of the available secondary cheaper
and less congested urban airports which turboprops use
Fly shorter routes, and because they have the same optimum FL as larger aircraft, they’ll spend much of
their flight time in non-optimum conditions
Have high ratio of cycles per flight hour, which increases maintenance costs in non-time related
maintenance inspections which must be carried out much more often.
Correct flight operations and line maintenance have significant impact on trip fuel burn. Flight and
maintenance crew technical training information is used as a guideline for their day-to-day routine. Further
investments in this area would be the easiest method to raise awareness and reduce operational costs, via fuel-
related efforts. This investments would quickly pay-off, because of the importance of fuel related costs in global
company costs. To demonstrate this, a simple example will be now presented of a small fuel savings and its
yearly result cost-wise. Considering the ERJ145 model, and using FAA and EASA domestic reserves, the typical
performance for a 400 NM trip are:
Typical Operational Profile
Embraer ERJ145
FAA
Reserves
EASA
Reserves
PGA Fleet
Average LIS-BIO
Takeoff Weight
kg 18.970 18.908 17.801
lb 41.821 41.685 39.243
Trip Fuel
kg 1.260 1.258 1.487
lb 2.778 2.773 3.278
Trip Time min 70 70 73
Fuel Savings of 1%
per Flight Leg
kg 13 15
lb 29 33
Annual Savings
for 8 Aircraft Fleet
kg 188.400 219.000
lb 416.000 482.804
Annual Savings
€ 144.649 170.820
$ 198.082 233.921
Table 5.1 - Typical operational profile and PGA example for the Embraer 145 aircraft; savings analysis
In Table 5.1, note that apparently small amounts of fuel burn reduction lead to significant annual
savings proportionally to the fleet size. In the FAA/EASA reference values example, a 13 kg (29 lb) fuel
consumption reduction on each leg achieved through adequate operational and maintenance practices, reflects

67
itself in the company’s annual savings of 144.649€, almost 0.2 million dollars. Using the same method, based on
the 390NM route LIS-BIO and assuming that this route represents all PGA operations, it is possible to predict
PGA’s yearly savings for their Embraer 145 fleet. A fuel savings of about 15 kg per flight leg, with an average of
5 legs per day would result in annual savings of 219.000 kg of fuel, corresponding to 170.820 €.
An increasingly demanding and competitive economic situation, associated with growing environmental
awareness and regulation, force the airline companies to optimize their operations and reduce maintenance,
operational and fuel-related costs as much as possible. Because of the constant and seemingly never-ending
climb of oil price in the international markets, and fuel costs being a very significant part of airliners’ total
expenses, optimizing fuel consumption is crucial and fuel trend monitoring will prove to be an indispensable tool
to achieve such goal.
5.2 – A Fuel Monitoring Tool
In order to proceed with the monitoring of fuel consumption trends, it was necessary to develop some
kind of tool that would easily compile all the required information, process it according to the analysis the user
wanted to conduct and quickly represent the requested parameter evolutions in the graphical form, for easier
visualization of trends.
A small program was then created, conceived with the Visual Basic for Applications programming
language based on Microsoft Office Access, whose functions and capabilities will be shortly described. Firstly,
the user must open the “Fuel Monitoring.mdb” file and double-click on Form1. The form represented in Figure
5.2 a) will then appear and the user must fill the required boxes according to the analysis in question and user
preferences.
Figure 5.2 a) and b) – Empty and filled form, part of the fuel monitoring tool
In Figure 5.2 b) is an example of a set of criteria selected by the user, who can choose: the fleet;
whether he wants to compare the acquired results with those from the same period in the previous year; whether
he wants to study one or more particular aircraft or the whole fleet; the period he desires from the beginning of

68
2006 to a few days prior to the analysis; if he wants to study a specific route; if he would like the departure and
arrival locations to be interchangeable between them; the departure airport IATA code; the arrival airport IATA
code; the number of days he wants average points to be based on, a type of mesh refinement; the full path of the
file where the results from the analysis will be recorded.
After filling all the boxes and pressing “Calculate”, the program will process for some time, depending
on the number of aircraft selected, the desired time-frame and whether the analyst chose to compare the results
with those from the previous year. The Fuel Monitoring tool would then access an existing query which
compiles all the necessary PGA operational data, extracting to an Access table the relevant information
necessary to proceed with a particular study, within the time frame chosen by the analyst.
When the processing finishes, open the selected output file and all the relevant data will be available
like so: in the first Microsoft Excel sheet, important data for all flights for the requested routes and within the
selected period will be presented, separated by aircraft; in the second Excel sheet the flights are averaged by the
chosen day periods, to facilitate trend visualization by eliminating unnecessary scatter and some metrics that will
be discussed in Sub-Section 5.3.1 are calculated; in the third and fourth Excel sheets, the data referring to the
previous year is presented, if this option has been selected previously; finally in the fifth Excel sheet there will
be the graphical plots of the referred metrics, which the analyst can consult directly. This prevents the user from
having the trouble to find the desired data, compile it, run the calculations and represent the results in a clear
plot. The fifth Excel sheet is displayed in Figure 5.3, corresponding to the criteria introduced before and shown
in Figure 5.2 b).
Figure 5.3 – Example of output produced by the fuel monitoring tool

69
From these graphical representations, many conclusions can be withdrawn by an experienced eye,
giving important information about the fleet’s status not only in terms of fuel consumption but also in terms of
usual TOWs and aircraft/route occupancy.
5.3 – Acquired Results
5.3.1 – Comparison between fleets/routes
In order to better proceed with the various fuel related analysis, some new parameters had to be defined
to compare fuel consumption between flights with different weights, flying in diverse routes corresponding to
completely different distances and flight durations. Two of the metrics utilized here, TFC and DFC were first
defined by Pereira, 2009 [1] and although simple they are the right parameters to proceed with the following fuel
evolution studies. Two more parameters were defined, one similar to the TFC but including ground time and
another defined to calculate fuel per passenger, but in the present work these were not as utilized given the
specific necessities of the conducted analysis.
The TFC or Time-Based Fuel Consumption is a metric conveniently created to allow a direct
comparison between two flights with different flight durations, with the objective of comparing their efficiency
in terms of fuel burnt per hour. This parameter is also corrected for the TOW, enabling comparisons between
flights with different number of passengers, cargo and FOB. It takes into account the total burnt fuel in the whole
flight and the total flight time (FH), thus excluding on-ground time.
TFC ൌ
Total Fuel ሺkgሻ
FH ሺhሻ
ൈ
MAXTOW ሺkgሻ
TOW ሺkgሻ
Equation 5.1
For the Fokker 100 and Embraer 145 maximum TOW, refer to APPENDIX II – Fokker F28 Mk 100
Specifications and APPENDIX III – Embraer ERJ145 Specifications, respectively. The dimensional analysis of
this quantity renders kg/h as the final dimension. Because this parameter uses the Total Fuel, therefore including
the fuel spent on ground, the flight’s ground efficiency has been considered, even if the FH only comprises the
time between takeoff and landing.
The DFC or Distance-Based Fuel Consumption was defined to compare two flights covering different
routes, in order to get an “automobile-like” mileage of kg/NM. It’s based on the ground orthodromic distance
separating the departure and destination airports and it can also be a good parameter to compare the flights’
efficiency.
DFC ൌ
Total Fuel ሺkgሻ
Distance ሺhሻ
ൈ
MAXTOW ሺkgሻ
TOW ሺkgሻ
Equation 5.2

70
Both parameters shall be used to measure flight efficiency, even if the TFC provides more information
and a more accurate comparison between flights for it considers ground efficiency. Furthermore, DFC is limited
because it uses fixed orthodromic distances, opposing to the desirable real covered distance, which is
unfortunately difficult to correctly calculate. The fixed distance between airports is limited because it doesn’t
take the flight plan into account, nor the effect of predominant winds or even re-routing or hovering procedures.
However, because it doesn’t depend on as many factors, it is also more filtered, and therefore can provide a
direct comparison in terms of fuel consumption between two flights in the same route for instance, and the
resulting conclusions can be then compared with those achieved by the TFC, namely the impact of FH variation
in the TFC parameter and flight efficiency.
Let us now use the defined parameters to obtain a brief idea of the type of analysis that can be
performed whilst monitoring fuel and the kind of conclusions that can be withdrawn. In order to actually save
money through optimized fuel consumption, some aspects like predictability, stability and viability play key
factors. The first one can be assured by fuel monitoring experience and prediction models. Stability and viability
will be the focus of the two following analysis.
A uniform behaviour throughout a fleet makes it much easier to properly monitor overall fuel
consumption and to define strategies to optimize and reduce costs. If aircraft display very different trends
concerning fuel, one can immediately conclude that, firstly, the fleet isn’t acting in unison as a fleet, and
therefore causes must be found for such behaviour; secondly, that such result scatter will probably make most
optimization efforts fruitless. It is then valuable to compare the behaviour of both PGA fleets, to assess their
uniformity and stability.
Figure 5.4 - Normal distributions of the DFC metric in the LIS-OPO route from January to July 2010
0,0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
8,0 9,0 10,0 11,0 12,0 13,0 14,0 15,0
ProbabilityDensity
DFC (kg/NM)
DFC Normal Distribution LIS-OPO - Fokker Fleet
TPA
TPB
TPC
TPD
TPE
TPF

71
Figure 5.5 - Normal distributions of the DFC metric in the LIS-OPO route from January to July 2010
Consider the normal distributions of the DFC metric for the LIS-OPO route, throughout the Fokker 100
and Embraer 145 fleets, represented in Figure 5.4 and 5.5 respectively, information collected from the first seven
months of 2010. LIS-OPO was the selected route, for it represents a significant part of PGA flights and an
example of frequently flown short-routes. The Embraer fleet presents higher probability densities and smaller
standard deviations than the Fokker fleet, which means that the Embraer aircraft are generally more reliable, for
they produce consistently more results within an acceptable band, seldom stepping out of said band. On the other
hand, the Fokker aircraft all have very similar mean values, whereas the Embraer aircraft present differences
between mean values of almost 10%. This uniformity makes the Fokker fleet easier to work with in terms
defining fuel consumption optimization strategies. In terms of the absolute DFC mean value, the Embraer fleet
has much lower DFC than the Fokker fleet and is therefore more efficient in a distance-based perspective. This
difference can be explained by the Embraer’s younger, more modern fleet, by the size of the Embraer model, and
in general because of its better performance and efficiency, especially on this kind of short-routes.
Figure 5.6 - Normal distributions of the DFC metric, in the OPO-BRU route from January to July 2010
Even if the DFC metric has taken the distance between departure and arrival into account, other aspects
haven’t been considered, namely that aircraft in shorter flights spend much of the flight time climbing, therefore
in non-optimum operation, which makes shorter flights more inefficient. To see the differences, let us compare
0,0
0,2
0,4
0,6
0,8
1,0
1,2
1,4
4,0 4,5 5,0 5,5 6,0 6,5 7,0 7,5 8,0 8,5
ProbabilityDensity
DFC (kg/NM)
DFC Normal Distribution LIS-OPO - Embraer Fleet
TPG
TPH
TPI
TPJ
TPK
TPL
TPM
TPN
0
1
2
3
4
5
6
7
2,7 2,9 3,1 3,3 3,5 3,7 3,9 4,1
ProbabilityDensity
DFC (kg/NM)
DFC Normal Distribution OPO-BRU - Embraer Fleet
TPG
TPH
TPI
TPJ
TPK
TPL
TPM
TPN

72
the previous DFC normal distribution for the LIS-OPO route, with the same type of distribution for the OPO-
BRU longer route, displayed in Figure 5.6. It is immediately clear that the time based flight efficiency has almost
doubled, thanks to the increase of cruise length, an understandable result if one bears in mind that in the
approximately 50-minute LIS-OPO flight only five to 15 minutes are made in cruise, depending on the attributed
FL, severely damaging its efficiency. Also easily noticeable is the decrease in standard deviation values and the
related increase in the probability density. The reduction of variance was expected, because while small
variations due to various reasons have a large impact in the short-haul flights, in longer flights those variations
become diluted and loose significance. Finally, after attentive observation of both plots, it can be seen that the
relative positions of the Gaussian distributions are very similar, which reflects the differences between the
condition of the engines and airframes of each aircraft and how they contribute as a whole to increase or reduce
fuel efficiency. For example, it can be concluded with some degree of confidence that the CS-TPL and CS-TPN
aircraft are the most inefficient within the Embraer fleet, in terms of fuel consumption.
Figure 5.7 - Passenger efficiency with a 15-day period per average point for both fleets
Another possible use of the fuel monitoring tool is to measure “passenger efficiency”, or in other words,
the fuel spent per passenger in kilograms per flight. Studying passenger efficiency is very important for airline
companies in the present days, because passengers are the only source of revenue for operators and therefore the
rate of occupancy of each flight is ultimately a measure of its economical viability. The passenger efficiency
metric is defined by: PE ൌ Total Fuel Number of passengers⁄ , so it does not include an explicit reference to the
TOW, however the weight is somewhat implicit in the number of passengers, although there is no relation with
the flight length or duration.
0
40
80
120
160
23-12-2009 11-02-2010 02-04-2010 22-05-2010 11-07-2010
FuelperPassenger(kg/pax)
Date
Passenger Efficiency
Emb145 LIS-OPO Fokk100 LIS-OPO Emb145 ALL
Fokk100 ALL Emb145 LIS-BCN Fokk100 LIS-BCN
Fokk OPO-FCO Emb145 OPO-MXPFokk100 OPO-FCO

73
Figure 5.8 – Time-based fuel efficiency for several routes for the Fokker 100 aircraft
Analysing Figure 5.7, one can observe that the Embraer 145 model is more efficient globally in the
flights that it performs (green line), compared to the Fokker’s global operation (purple line). The same happens
to the particular case of LIS-OPO, for the already appointed reasons. Regarding the LIS-BCN city pair, the
passenger efficiency isn’t very different, with the Fokker model taking a slight advantage due to the increase of
range and thanks to high occupancy rates. For longer and busier routes, the aircraft must fly heavier, and the
Embraer model is limited in terms of TOW, becoming much more inefficient when TOW comes closer to
MTOW. The OPO-MXP route is a good example of the Embraer’s poor efficiency in the medium-haul flights.
Finally, for all that was said above, the OPO-FCO route should be efficient in terms of fuel consumption but that
isn’t what the plot shows. The reason is that the flight probably does present good efficiency from a consumption
per time or distance point of view, but not passenger efficient due to possible low occupancy rates, especially
until mid-March, coinciding with the end of winter. The good TFC efficiency can be confirmed in Figure 5.8.
This is one route that should be closely monitored, to assess whether it is still economically viable. Varying
levels of improvement have been registered as the year goes on, due to an increase of occupation rates,
especially in summer.
Many other analyses could be conducted and in a much more detailed manner with the assistance of a
fuel monitoring tool, an indispensible asset for any airline company seeking success through fuel cost reduction.
5.3.2 – Engine Offline Washing
The majority of engine degradation is recoverable, however the level of effort or costs that have to be
assumed to correct certain types of degradation is what determines what is in fact worth recovering or not. The
degradation recovery by any means of washing is one of the most usual and simple recoverable degradation
methods and it can be done in different ways, namely through online washing and offline washing. In this Sub-
section, offline engine washing will be briefly presented as a means of recovering some of the engine
degradation and its impact on fuel consumption and on the engine’s working conditions will be lightly analysed.
2000
2050
2100
2150
2200
2250
23-12-2009 11-02-2010 02-04-2010 22-05-2010 11-07-2010
TítulodoEixo
Date
TFC - Fokker 100
Fokk100 LIS-OPO Fokk100 ALL
Fokk100 LIS-BCN Fokk OPO-FCO

74
The importance of constant engine condition monitoring will be, also for this type of procedure, once again
demonstrated.
Gas turbine compressors consume approximately 60% of the overall cycle energy during operation.
This cycle consumes very large quantities of air and although this air is filtered, small quantities of dust, aerosols
and water pass through the filters and deposit on the blades. These deposits decrease the air flow of the
compressor and the overall performance of the gas turbine. Compressor cleanliness can be maintained using a
routine program of water washing. There are two water wash manoeuvres performed on gas turbines: offline and
online. An offline manoeuvre is conducted with the gas turbine in a cooled state using cranking speed, while an
online manoeuvre is conducted with the machine at operating temperature and uses water only. Both operations
use highly atomized water spray patterns designed to completely enter the compressor core. The online washing
cleans the early stages and maximizes the time period between needed offline washing to provide peak
availability and the offline cleans the entire core and recovers lost performance. For obvious reasons, online
washing isn’t a viable solution to recover jet turbofan engines, not only because the washing can’t be done while
the engine is at usual, operating conditions, but also because that would imply extra fuel to be burnt during the
cleaning and also the fact that this type of washing should be done very frequently, which is not practical or cost-
efficient. Therefore we will focus solely in the compressor-washing from an aviation perspective and evaluate
offline washing as an engine performance optimization tool.
Previous investigations have been performed on the effectiveness of offline cleaning on the
performance recovery of gas turbine compressors that had been deteriorated namely due to dirt ingestion. In
modern gas turbines, redeposition of dirt at the later stages during offline washing can be easily avoidable, as
long as the water fuel flow is sufficient, for an increased water-to-air ratio reduces the level of redeposit.
However too much water and large liquid droplet sizes will lead to surface erosion and possible aerodynamically
induced high cycle fatigue if operated for extended periods. Also, although some manufacturers claim that
specialized liquids will improve the cleaning efficiency and reduce fouling redeposition, previous test results are
not conclusive to demonstrate whether such products do in fact significantly affect either one. Clean, deionised
water can work as well as any liquid without introducing undesirable deposits.
Figure 5.9 - Example of an action of offline engine washing
In offline washing, the gas turbine is run at sub-idle shaft speeds while a cleaning solution is injected
into the engine. This method is well-proven and effective in removing deposits not only in the axial compressor

75
but also on the interior surfaces of the entire gas path. Before an offline wash, the gas turbine unit must be shut
down and cooled to avoid excessive thermal loading of the internal gas turbine components. This causes loss of
availability, for the aircraft has to be grounded during the cooling, the procedure and for a short period after the
washing, which usually in PGA’s fleet case doesn’t result in any additional costs or loss in terms of revenue
because the total down time isn’t very long and therefore an offline washing can be done overnight for example
without affecting operations. The costs associated with regular offline washing procedures will be discussed at
the end of this Sub-section.
Manufacturers and operators of gas turbines have tried to quantify the effect of offline water wash
systems with respect to reduced degradation, increased power output and reduced fuel consumption. However,
field testing is challenging, and the degradation rates are continuously changing due to environmental conditions,
over which the operator has limited influence. Hence, the test results are often difficult to compare and provide
limited conclusions to the overall phenomena of deterioration mechanisms and performance restoration.
Figure 5.10 - Dust removal from pre-wash (on the left) to post-wash (on the right) from turbine blades
In the particular case currently here presented, there is an increased difficulty in interpreting the
obtained results and reaching definite conclusions about the effect of compressor washing in engine performance
and degradation. PGA has no tradition and therefore no experience in engine washing, and is at the moment
interested in assessing if such action would be beneficial for both fleets, if implemented on a regular basis. This
study is composed of three stages, the first consisting in the washing of a sample of TAY650-15 engines, while
continuously monitoring the evolution of engine condition, before and after the wash.
The results of this first stage are the ones presented in this Sub-Section, namely in Table 5.1 and Figure
5.11 and 5.12, and from them conclusions should be drawn, whether there is an interest in implementing this
kind of procedure for more TAY650-15 engines (or even for all) during a relatively short period of time – second
stage – to more safely and thoroughly assess if compressor washing would reduce costs taking into account
PGA’s operational profile. In this second stage a greater number of engines would be analysed to exactly
quantify the gains and losses associated with implementation of compressor washing. The obtained results can
also justify or not the investment that has to be made in the purchase of a compressor-washing machine, staff
training or machine and crew rental. If the results are in general satisfying across the fleet, the regular
compressor wash could be enforced for all Fokker 100 aircraft – third stage, and a similar analysis process
should be initiated for the AE3007 engines of the Embraer 145 model.

76
The first of the above referred stages was originally meant to comprise the washing and analysis of five
TAY650-15 engines, but it was impossible to carry them all out at the predicted time interval due to high staff
workload. Consequently, only three engines were washed and this small sample associated with some gaps in
terms of engine condition data, made this analysis somewhat less representative of the effect of compressor
offline washing in engine condition than would be desirable. The obtained results for the three washed engines
are displayed in Table 5.1, through the Delta Key Parameters defined in Chapter 4 and divided into a Pre-Wash
and a Post-Wash period.
The washes were undertaken from on the 17th
and 19th
in March 2010 in TAY650-15 engines with the
serial numbers 17277, 17276 and 17318, the first two belonging to the CS-TPA aircraft and the latter to the CS-
TPB aircraft. The 17317 engine also belongs to the CS-TPB aircraft but wasn’t washed, so it will be considered
a baseline that can be used to give an idea of the key parameters’ evolution due to operational or flight
conditions, not related with the offline washing degradation recovery. Pre-Wash and Post-Wash periods have the
duration of approximately fifteen days. This was the chosen period length because it was necessary a minimum
period of time to ensure that there was enough data to guarantee the representativeness of the results. On the
other hand the considered period couldn’t be too long because engine conditions change with the variation in
outside conditions throughout the year and also the engine continues to degrade after the washing, which could
influence the results.
The COMPASS software tool was used to process the representative points which correspond to every
flight each engine endured during the referred periods and from these points, simple average and deviation
calculations were performed for each engine, concerning each parameter of interest and the results are displayed
in Table 5.2.
Engine SN
Delta FF (%) Delta N2 (%) Delta TGT (%)
Mean Deviation Mean Deviation Mean Deviation
Pre-Wash
(PreWD)
17277 3,2210 3,3823 1,3409 0,2832 1,9945 0,9995
17276 0,7345 4,4547 -0,0580 0,3214 1,0520 0,8282
17318 2,5511 1,7729 1,2338 0,2163 0,3346 0,6626
17317 3,5979 1,8337 0,4095 0,2530 2,6977 0,6748
Post-Wash
(PostWD)
17277 3,0326 1,6468 1,3469 0,2497 1,5527 0,8113
17276 1,0749 3,3512 -0,1792 0,3309 0,8470 0,6743
17318 2,1199 1,3277 1,2085 0,2361 0,0820 0,5365
17317 4,0382 1,6225 0,5622 0,3434 2,7723 0,6589
Absolute
Improvement in Delta
ሺ‫۷ۯ‬ ൌ ‫۲܅܍ܚ۾‬ െ ‫۲܅ܜܛܗ۾‬ሻ
17277 0,1884 1,7355 -0,0060 0,0335 0,4418 0,1882
17276 -0,3404 1,1035 0,1212 -0,0095 0,2050 0,1539
17318 0,4312 0,4452 0,0253 -0,0198 0,2526 0,1261
17317 -0,4403 0,2112 -0,1527 -0,0904 -0,0745 0,0159
Table 5.2 - Normal Distributions results before and after wash

77
For the analysis of Table 5.2, let us focus on Absolute Improvement parameter, whose definition can be
found in the same table, and that represent the variation of the Delta or Smoothed Delta Key Parameters between
the Pre-Wash period and the Post-Wash period, in absolute Delta variation. The AI parameter informs the
analyst about how much absolute variation there was in a Delta Parameter, or in other words, how much the
results have become closer to those predicted in the theoretical model. Comparing the DFF, DN2 and DTGT
parameters, it becomes evident that the improvements are always more significant for the DFF and DTGT
parameters than for DN2, which isn’t as affected by the offline engine washing.
Regarding the engines, three conclusions can be withdrawn from the data presented in Table 5.1. Firstly
that engine 17276 confirms the strange behaviour announced in Chapter 4, with an apparent degradation in terms
of DFF and an improvement in DTGT, which is unusual and once again points to insufficient filtering or other
sensor related issue. Secondly, that the rest of the washed engines, 17277 and 17218, present consistently DFF
and DTGT reductions of 0.20-0.45% in terms of mean value and also elevated reductions in terms of the results’
deviation, both good indicators of improvement in engine condition due to compressor washing. Thirdly, it can
also be noticed that the reference engine 17317 presents a slight worsening of mean values for all parameters,
consistent with natural engine degradation, with an especially accentuated increase in DFF, which is reaching
values that will probably damage the engine’s fuel consumption.
In this study, a stronger focus is applied to the mean parameter values, because they represent the
average operation of the engine and therefore the average variations are more important and should be taken
more seriously than isolated spikes, because while the first usually imply actual mechanical problems, the latter
can be often explained by sensor errors and other data capturing malfunctions. However if the deviations are
small, the engine’s operation is more uniform and it spends more time in the average or optimal working
conditions, also reducing thermal fatigue, etc, which is also important and the reason both parameters were
considered in the previous table.
Figure 5.11 - Effect of engine wash on DTGT for TAY650-15 engines
-20246
01-03-2010
06-03-2010
09-03-2010
12-03-2010
14-03-2010
16-03-2010
21-03-2010
24-03-2010
26-03-2010
28-03-2010
02-04-2010
05-04-2010
DeltaTGT(%)
Date
Delta TGT with Engine Wash
17277 Pre 17276 Pre 17277 Post 17276 Post

78
Some of the results resumed in the table 5.2 are represented graphically in Figure 5.11 and 5.12, namely
the evolutions of Delta TGT and Smoothed Delta FF parameters respectively, through longer time frames. In
Figure 5.11, it is noticeable that, due to the offline engine wash, there is a slight reduction in the values of DTGT
for both 17277 and 17276 engines, with AI of about 0.44% and 0.20% respectively. Also in terms of controlling
the scatter of results some improvement has been achieved, which results in a more uniform profile.
Figure 5.12 - Effect of engine wash on smoothed DFF for some TAY650-15
In Figure 5.12 all the engines considered and studied in Table 5.1 are represented and the graphical
representations confirm and illustrate the already referred tendencies. It is easy to visualize the non-linear
behaviour presented by engine 17276, with big and well-pronounced variations before and after the wash. The
only reason these variations aren’t very problematic and haven’t triggered any alarms is because the mean values
for all parameters in engine 17276 are significantly lower than the other engines in study, which enable the
values to stay within the acceptable limits. It is also possible to see that for the 17277 and 17318 engines there is
a reduction in the DFF parameter, given by the AI, of about 0.18% and 0.43% in terms of mean value and also it
is clear that these engines’ profiles become more stable and more uniform after the wash. Finally one can also
observe that the evolution of the DFF parameter in the 17317 unwashed engine is very coherent throughout the
studied time-frame with no significant variations.
The COMPASS software can monitor gas turbine performance through continuous import and analysis
of thermodynamic data from each engine’s control system: fuel flow, air ambient condition, compressor outlet
and exhaust gas temperatures, pressure ratio and outdoor air pressure. Warnings can be issued when the system
detects changes in the relative performance over time. Furthermore, based on the historic development and on
experience, engine trend monitoring through COMPASS makes it possible to extrapolate a forecast of future
performance degradation, enabling the staff to plan short term cyclic maintenance actions at the most
economically feasible time. This minimizes production loss and avoids increased fuel consumption due to
deteriorated performance. Online condition and performance monitoring are therefore crucial to efficiently
implement a compressor-washing routine for optimizing the operational conditions of PGA’s fleets.
-5-3-11357
01-03-2010
07-03-2010
09-03-2010
13-03-2010
15-03-2010
17-03-2010
22-03-2010
25-03-2010
27-03-2010
29-03-2010
03-04-2010
05-04-2010
SmoothedDeltaFF(%)
Date
Delta Fuel Flow with Engine Wash
17277 Pre___L 17277 Post___L 17317 Pre___L
17318 Pre___L 17317 Post___L 17318 Post___L

79
After analysing the effect of the engine wash in the parameters that represent the condition of the
engine, and through that the impact of this degradation recovering procedure in terms of structures, maintenance
and airworthiness, it becomes necessary to measure the washing effect on fuel consumption, for this is where the
most savings could be accomplished. The Fuel Monitoring Tool revealed itself as very useful to facilitate this
study and produced the graphical results displayed in Figure 5.13 and 5.14.
Due to the fact that this analysis also pretends to determine the evolution of fuel consumption, as did the
one conducted at the beginning of the present chapter, the same problems hamper the following study – the
existence of too many variables makes it difficult to make out a tendency, and in this case the impact of the
engine wash on such tendency. To proceed with the assessment of the effect induced in fuel consumption by the
washing procedure, the TFC and DFC metrics defined previously will be used, in order to detect fuel
consumption variation per flight hour or per distance, throughout the first seven months of 2010. The referred
metrics’ evolutions are represented in Figure 5.13 and Figure 5.14, regarding both aircraft whose engines were
washed and the fleet average for the same period, as a reference for variations induced by weather or operational
conditions.
Figure 5.13 - Variation of TFC with engine wash
Figure 5.14 - Variation of DFC with engine wash

80
When analysing the two graphical representations in Figure 5.13 and Figure 5.14, one can immediately
notice that both metrics for both aircraft present similar behaviours to the ones displayed by the respective fleet
average evolutions. Naturally associating the average fleet variations with factors external to the actual engines
or aircraft, it becomes once again evident the impact of flight conditions and operational profiles in fuel
consumption.
In a previous note, it should be reminded that although values presented henceforth for the TFC and
DFC have a dimension of kg/h and kg/NM, the values don’t represent the quantitative savings, because of the
MAXTOW/TOW ratio which is also part of the metrics. They only provide a qualitative approach to fuel
savings.
As for the effects of the wash procedure, it can be noted in both plots that there is a significant
reduction in TFC and DFC from the 9th
to the 19th
March period, until the 20th
to the 30th
March period with a
reduction of approximately 470 kg/h and 0.8 kg/NM for CS-TPA and 520 kg/h and 0.7 kg/NM for CS-TPB. The
fleet average evolutions also present a negative variation in that period, although smaller than the ones of CS-
TPA and CS-TPB, situation that can raise some questions. Given the small size of the Fokker 100 fleet, are these
aircraft’s results responsible for the drop in the fleet overall results or is this drop a consequence of other factors,
which are also in part responsible for the reductions in the two considered aircraft? It is hard to be sure without a
thorough investigation around that variation, but it is probable that a part of the fleet’s improvement is due to
external factors indeed and a smaller part is due to the impact that the washing of the engines had on aircraft CS-
TPA and CS-TPB and consequently on the average of the fleet. The aircraft parameters will decrease in the same
amount as the fleet due to external factors and the difference between that and the total reduction will be due to
the engine washing. Taking this fact into account, the improvement due to the engine washings would be
approximately of 100 kg/h and 150 kg/h regarding the TFC for CS-TPA and CS-TPB respectively, and roughly
0.3 kg/NM for the DFC, for both aircraft. These reductions, assuming the mean values of about 3000 kg/h and
11 kg/NM, would represent improvements of 3.33% and 4.98% for TFC and 2,7% for DFC. During 2010
operations, considering the two referred aircraft, the average TOW/MAXTOW ratio was 0.802, therefore, by
multiplying the TFC and DFC results by 0.802, the actual savings in fuel can be predicted.
After the referred period when the engines are washed, the TFC and DFC parameters don’t stay
stagnated, presenting variations again due to reasons not engine-related, although in general these variations
become slightly weaker and there is a definite shift downwards for the mean values from that period on.
Some lack of crucial data, particularly for the CS-TPA aircraft for the months of April and May, allied
with the input problems previously pinpointed for the Fokker 100 fleet, especially for the fuel flow parameter,
made the analysis of quantifying the effect of engine offline washing on engine condition and fuel consumption
much harder than it should be. In conclusion there are good indications that the engine washing procedure may
in fact recover some of the engine incurred degradation, improving engine’s operation between 0,18% and
0,44% and reducing fuel consumption per flight hour by 3.33% to 4.98% and fuel per nautical mile by 2.77%.
Better data filtration and systematic downloading, along with more washes in more aircraft will be indispensable

81
to confirm these results and further study the possible implementation of regular offline compressor washing in
PGA’s fleets and to give the confidence and representation needed to conclude without hesitation if this would
be cost-efficient and a further step into optimization.

82
6 - Conclusions/Future Work
It is said that the worse times bring out the best in people. Well the same happens to airline companies.
In the current scenario of international crisis, increasing fuel prices, building awareness to the environmental
issues and fierce open-market competition, airline companies are giving the best of themselves in order to keep
flying. Optimization is the word of order, operationally, regarding maintenance practices and fuel consumption,
and it’s this will to optimize in response to adversity, that fully justifies the development of the present work and
many others like this.
The objective of the work presented in Chapter 3 was to revise PGA’s Fokker 100 flight profile,
regarding life-limited mandatory engine parts, and assess whether this profile could be improved. A simple tool
was created to process FDR files, to compile and analyse a year’s worth of data, detecting for each flight the
Critical Points, necessary to determine the flight profiles, and finally averaging the results to obtain annual
results that could be globally analysed. It was concluded that PGA would maintain its flight profile “B” rating
for the moment, which in itself doesn’t represent an improvement, but the results have nevertheless given PGA
the confidence to use the developed program to begin assessing the evolution of their flight profile each quarter,
with the objective of closely monitoring its status while studying possible strategies to improve it.
Furthermore, an eventual change in the engine’s selected thrust mode was suggested, during the climb
phase, whose benefits and consequences were discussed. The very promising results achieved require a further,
more detailed study, based in more flight data, to confirm the benefits in changing the thrust mode, namely the
desirable improvement of the fleet’s flight profile.
The analysis of ECM possibilities and its capability to serve as a valuable tool for a company with
optimization as its motto was one of the main focus points of the present work, which was described in Chapter
4. ECM and the utilized trend monitoring software COMPASS were briefly introduced, and through them,
several analyses were conducted and presented, in order to demonstrate how operators can use ECM from
different perspectives and with different goals. ECM was used to compare fleets, to detect aircraft with lower
efficiencies, to quantify the gain in performance achieved through maintenance actions and to serve as both a
prognostic and a diagnostic tool, to great effect.
The use of ECM was fully justified both in terms of increasing safety and as a crucial contributor to
significant savings in fuel and maintenance costs. Future work in this domain would consist of improving the
filtering of flight data for the Fokker 100 fleet, in order to remove existent variations in fuel flow and other key
parameters and also to further assess the influence of the Autothrottle system in said variations.
In Chapter 5, a simple fuel monitoring tool is presented, which was developed by the author to serve as
a link between maintenance and flight operations, by associating maintenance procedures with actual fuel
consumption. This was verified through a specific maintenance action, which was the offline wash of three

83
engines of PGA’s Fokker 100 fleet, whose effect was quantified both in terms of engine health through the ECM
tool and in reduced fuel consumption through the fuel monitoring tool. In the future, more engine washes and a
more thorough study should be done to confirm the good indications given by this brief analysis, regarding
savings due to engine offline washing.
The developed program is also able to detect which aircraft are being less efficient in overall operation
or in certain routes, which serves as an indication to the maintenance department that that particular aircraft is
inefficient, and therefore causes should be identified and corrected. It also delivered results that show that the
Embraer 145 fleet is generally more efficient and present a more stable profile in their current operation than the
Fokker 100 fleet.
The tool possesses other capabilities more related with flight operations, and a couple of examples were
given, namely the comparison of both fleets’ consumption per mile or per hour in their global operation and on
specific routes, and both fleets’ passenger efficiency throughout the first seven months of 2010, which confirmed
that the Embraer fleet is more efficient in terms of passengers for the short domestic and Iberian routes, and that
for the medium-haul the Fokker is the most indicated model for most cases.
In a nearby future, the capabilities of this simple fuel tool should be amplified, for its utility and value
have been already verified. Furthermore, there should be an effort to directly relate this tool and the fuel analysis
with the ECM, to facilitate the understanding of processes, optimizing operation and reducing costs even further.

84
References
[1] Pereira Pedro M. F., Regional Airline’s Operational Performance Study and Appropriate
Enhancement Techniques PGA – Portugália Airlines as a case study, 2009
[2] UK NATS, Aeronautical Information Publication - General Rules and Procedures,2010
[3] Fishbeyn B. D. and Pervyshin N. V., Determination of the Effect of Atmospheric Humidity on the
Characteristics of a Turbofan Engine, 1987
[4] Brooks D. B. and Garlock E. A., The Effect of Humidity on Engine Power at Altitude, 2005
[5] Veer Timot, Measured Data Correction for Improved Fouling and Degradation Analysis of Gas
Turbines, 2004
[6] Rolls-Royce, Engine Trend Plot Interpretation Guide, 2005

85
Bibliography
Airbus, Getting to grips with Aircraft Performance Monitoring, 2002
BREDERODE, Vasco de; Fundamentos de Aerodinâmica Incompressível; Author’s edition, 1997
Clifton David, Condition Monitoring of Gas-Turbine Engines, 2006
Cranfield University, Fuel and Air Transport – A report for the European Commission, 2008
Elisabet Syverud, Axial Compressor Performance Deterioration and Recovery through Online
Washing, 2007
Embraer, Aircraft Maintenance Manual 45-45-00 Central Maintenance System, 1998
Embraer, Fuel Conservation, 2008
Fokker, FLYFokker Supplement to IATA Guidance Material and Best Practices for Fuel and
Environmental Management, 2010
Kurz Rainer and Brun Klaus, Gas Turbine Tutorial – Maintenance and Operating Practices Effects on
Degradation and Life, 2008
IATA, IATA Guidance Material and Best Practices for Fuel and Environmental Management, 2007
Intergovernmental Panel on Climate Change, Aviation and the Global Atmosphere, Cambridge
University Press, Cambridge, 1999.R.C. Miake-Lye, Aerodyne Research, Inc.; Advancing the
Understanding of Aviation’s Global Impacts, 2005
Propulsão I: Acetatos, AEIST Secção de Folhas
ROSKAM, Jan; LAN, C. T.; Airplane Aerodynamics and Performance; DARCorporation, 1997

86
Appendix I – ECM Trend Guideline Chart

87
Appendix II – Embraer ERJ145 and AE3007 Specifications

88
Appendix III – Fokker 100 and TAY650-15 Specifications

89
Appendix IV – IATA and ICAO Codes of Relevant Airports
IATA Code ICAO Code City Airport
AMS EHAM Amsterdam Amsterdam Airport Schiphol
BCN LEBL Barcelona Barcelona Airport
BRU EBBR Brussels Brussels Airport
CDG LFPG Paris Paris - Charles de Gaulle Airport
CMN GMMN Casablanca Mohammed V International Airport
FAO LPFR Faro Faro International Airport
FCO LIFR Rome Leonardo da Vinci – Fiumicino Airport
FNC LPMA Funchal Funchal Madeira Airport
GVA LSGG Geneva Geneva Cointrin International Airport
LIS LPPT Lisboa Lisbon Portela Airport
LGW EGKK London London Gatwick Airport
LUX ELLX Luxembourg Findel Airport
LYS LFLL Lyon Lyon – Saint Exupéry Airport
MAD LEMD Madrid Madrid Barajas Airport
MRS LFML Marseille Marseille Provence Airport
MXP LIMC Milan Milano Malpensa Airport
NCE LFMN Nice Nice Côte d’Azur Airport
OPO LPPR Porto
Francisco Sá Carneiro – Porto
International Airport
PXO LPPS Porto Santo Porto Santo Airport
SVQ LEZL Seville Sevilla – San Pablo Airport
TLS LFBO Toulouse Toulouse Blagnac Airport
ZRH LSZH Zurich Zurich Airport

Condition monitoring case study

More Related Content

Viewers also liked (20)

Recently uploaded (20)

Condition monitoring case study