Modeling the Life Cycle Cost of Jet Engine Maintenance

Reglerentwurf zur Synchronisation einer hybridenAktuatorkonfiguration

Diplomarbeit

Modeling the Life Cycle Costof Jet Engine Maintenance

Students Research Project

by

Ralf [email protected]

HamburgOctober 2010

Abstract

Cost incurred by aircraft maintenance, repair and overhaul (MRO) make up a considerableproportion of the total life cycle cost of an aircraft. For an evaluation of the economic efficiencyof commercial aircraft, it is therefore crucial to estimate its MRO cost. The largest share of thesecost are incurred by the maintenance of the aircraft engines. Engine maintenance is performedon-condition in dedicated workshops mainly independent from the regular maintenance checkevents of the remainder aircraft. For a consideration of aircraft life cycle cost, it is hence alsonecessary to predict the intervals of these engine shop visits.

The present study discusses an approach for estimating the life cycle cost of aircraft enginemaintenance. It provides an extensive review on basic concepts of jet engine MRO as well as on theprimary factors that affect the engine maintenance cost and intervals. Based on these resources,a database was assembled from historic maintenance data provided by the aviation magazine“Aircraft Commerce”. Through linear regression analysis of the database, cost estimatingrelationships (CERs) describing the correspondence between maintenance cost/intervals andbasic engine specifications were derived. These CERs are complemented by a series of adjustmentfactors that were developed in order to reflect additional influential effects, such as operationalseverity or engine age. The resulting model demonstrates that reasonable figures for the engineshop visit intervals and cost can be estimated by considering the engine take-off thrust, enginedry weight, engine maturity, average flight length, applied derate and environmental conditions asprimary influence factors. Since the assembled database contains only maintenance informationof the currently mature engine generation, the validity of the developed model is limited to thecurrent engine generation. However, it is assumed that the basic maintenance characteristicsremain unchanged with the next engine generation. Plausibility tests, which compare the modelresults with estimates for the maintenance cost of the next engine generation, indicate that themore advanced engines can be represented by the developed model through the use of technologyfactors.

The developed model is intended to complement the aircraft life cycle cost simulation tool(LCC-tool), which is being developed at the Institute of Air Transportation Systems at HamburgUniversity of Technology. The LCC-tool uses Matlab as programming environment and enablesthe evaluation of technologies under the incorporation of expertise in form of technology factors.Therefore, the resulting model was implemented into the structure of the existing Matlabprogramme sequence.

I

Contents

List of Abbreviations V

List of Symbols VII

1 Introduction 11.1 Thesis Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.2 Thesis Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

2 Literature Review 32.1 Aircraft Gas Turbine Engines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2.1.1 Basic Principles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32.1.2 The Turbofan Engine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

2.2 Aircraft Engine Maintenance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112.2.1 On-Wing Engine Maintenance . . . . . . . . . . . . . . . . . . . . . . . . 112.2.2 Engine Overhaul - Shop Visit . . . . . . . . . . . . . . . . . . . . . . . . . 132.2.3 Engine Time On-Wing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192.2.4 Engine Maintenance Costs . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

2.3 Modeling of Engine Maintenance Cost . . . . . . . . . . . . . . . . . . . . . . . . 262.3.1 Reflection of EMC in DOC methods . . . . . . . . . . . . . . . . . . . . . 272.3.2 Parametric Cost Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . 27

3 Development of Cost Estimating Relationships 323.1 Database Assembly . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

3.1.1 Establishing the Hypothesis . . . . . . . . . . . . . . . . . . . . . . . . . . 323.1.2 Review of Data Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333.1.3 Data Collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353.1.4 Data Normalization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363.1.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

3.2 Data Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403.2.1 Candidate Relationship Screening . . . . . . . . . . . . . . . . . . . . . . 403.2.2 Regression Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

3.3 Results of the Parametric Cost Modeling . . . . . . . . . . . . . . . . . . . . . . . 433.3.1 Shop Visit Interval CERs . . . . . . . . . . . . . . . . . . . . . . . . . . . 433.3.2 Shop Visit Cost CERs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

II

Contents III

4 Modeling of Engine Maintenance Cost 464.1 Model Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

4.1.1 CER-Module . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 484.1.2 Effect-Module . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 494.1.3 Spare Engine Charges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

4.2 Example Application . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 524.2.1 Base Costs and Intervals from CERs . . . . . . . . . . . . . . . . . . . . . 524.2.2 Adjustment of Intervals . . . . . . . . . . . . . . . . . . . . . . . . . . . . 534.2.3 Final Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

4.3 Model Plausibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 554.3.1 Model Results vs. Original Database . . . . . . . . . . . . . . . . . . . . . 554.3.2 Model Results vs. Additional Data Sources . . . . . . . . . . . . . . . . . 584.3.3 Summary of the Plausibility Tests . . . . . . . . . . . . . . . . . . . . . . 60

4.4 Sensitivity Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 604.4.1 Sensitivity of Model Output . . . . . . . . . . . . . . . . . . . . . . . . . . 614.4.2 Sensitivity of Life Cycle SVC . . . . . . . . . . . . . . . . . . . . . . . . . 63

5 Implementation into existing LCC-Tool 685.1 Function Definition and Input Modification . . . . . . . . . . . . . . . . . . . . . 685.2 Estimating the Shop Visits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 705.3 Processing the Predefined Shop Visits . . . . . . . . . . . . . . . . . . . . . . . . 725.4 Consideration of Spare Engine Costs . . . . . . . . . . . . . . . . . . . . . . . . . 725.5 Estimation of Required Shop Visit Number . . . . . . . . . . . . . . . . . . . . . 725.6 Definition of Last Shop Visits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 735.7 Output Generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

6 Summary and Conclusion 75

Bibliography 77

List of Figures i

List of Tables ii

Appendix ii

A Maintenance Costs iiiA.1 Engine MRO Cost Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiiA.2 Shop Visit Cost Driver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii

B Database ivB.1 Aircraft Commerce Shop Visit Reserves & Intervals Example Table . . . . . . . . iv

Contents IV

B.2 Classification of Aircraft Engines . . . . . . . . . . . . . . . . . . . . . . . . . . . ivB.3 Core Database . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v

C Regression Analysis viC.1 First Interval SH Engines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viC.2 Mature Interval SH Engines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiC.3 First Interval MLH Engines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiiC.4 Mature Interval MLH Engines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ixC.5 First Shop Visit Restoration Costs . . . . . . . . . . . . . . . . . . . . . . . . . . xC.6 Mature Shop Visit Restoration Costs . . . . . . . . . . . . . . . . . . . . . . . . . xiC.7 LLP Cost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xii

D Model Parameters xiiiD.1 Averaged Short-Haul-Engine Severity Curve . . . . . . . . . . . . . . . . . . . . . xiiiD.2 Averaged Medium-Long-Haul-Engine Severity Curve . . . . . . . . . . . . . . . . xivD.3 Time & Material Factor Curves . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv

E Model Analysis xvi

List of Abbreviations

AC . . . . . . . . . . . . . . . . . AircraftACA . . . . . . . . . . . . . . . Aircraft Commerce articlesAD . . . . . . . . . . . . . . . . . Airworthiness directiveBLS . . . . . . . . . . . . . . . . Bureau of Labor StatisticsBPR . . . . . . . . . . . . . . . . Bypass ratioCER . . . . . . . . . . . . . . . . Cost estimating relationshipDB . . . . . . . . . . . . . . . . . Data baseDMC . . . . . . . . . . . . . . . Direct maintenance costDOC . . . . . . . . . . . . . . . Direct operating costsECI . . . . . . . . . . . . . . . . Employment cost indexECM . . . . . . . . . . . . . . . Engine condition monitoringEF . . . . . . . . . . . . . . . . . Environment factorEFC . . . . . . . . . . . . . . . . Engine flight cycleEFH . . . . . . . . . . . . . . . . Engine flight hourEGT . . . . . . . . . . . . . . . Exhaust gas temperatureEGTM . . . . . . . . . . . . . Exhaust gas temperature marginEMC . . . . . . . . . . . . . . . Engine maintenance costsEPR . . . . . . . . . . . . . . . . Engine pressure ratioFADEC . . . . . . . . . . . . . Full authority digital engine controlFOD . . . . . . . . . . . . . . . . Foreign object damageFR . . . . . . . . . . . . . . . . . First-runHPC . . . . . . . . . . . . . . . . High pressure compressorHPT . . . . . . . . . . . . . . . . High pressure turbineIMC . . . . . . . . . . . . . . . . Indirect operating costsIOC . . . . . . . . . . . . . . . . Indirect operating costsIPC . . . . . . . . . . . . . . . . Intermediate pressure compressorLCFH . . . . . . . . . . . . . . Life cycle flight hoursLLP . . . . . . . . . . . . . . . . Life limited partLM . . . . . . . . . . . . . . . . . Line maintenanceLPC . . . . . . . . . . . . . . . . Low pressure compressorLPT . . . . . . . . . . . . . . . . Low pressure turbineMIF . . . . . . . . . . . . . . . . Maintenance inflation factorMLH . . . . . . . . . . . . . . . Medium-long-haulMR . . . . . . . . . . . . . . . . . Mature-run

V

Contents VI

MRO . . . . . . . . . . . . . . . Maintenance, Repair and OverhaulOAT . . . . . . . . . . . . . . . . Outside air temperatureOEM . . . . . . . . . . . . . . . Original equipment manufacturerPM . . . . . . . . . . . . . . . . . Parts manufacturer approvalPPI . . . . . . . . . . . . . . . . . Producer price indexSEC . . . . . . . . . . . . . . . . Spare engine costsSF . . . . . . . . . . . . . . . . . . Severity factorSH . . . . . . . . . . . . . . . . . Short-haulSLOATL . . . . . . . . . . . Sea level outside air temperature limitSV . . . . . . . . . . . . . . . . . Shop visitSVC . . . . . . . . . . . . . . . . Shop visit costsSVR . . . . . . . . . . . . . . . . Shop visit rateSVRC . . . . . . . . . . . . . . Shop visit restoration costsTE . . . . . . . . . . . . . . . . . Temperate environmentTIT . . . . . . . . . . . . . . . . Turbine inlet temperatureTMF . . . . . . . . . . . . . . . Time & material factorTOC . . . . . . . . . . . . . . . Total operating costsTOW . . . . . . . . . . . . . . . Time On-WingTOW . . . . . . . . . . . . . . . Time on-wingTSF . . . . . . . . . . . . . . . . Three-spool factorTSFC . . . . . . . . . . . . . . Thrust specific fuel consumptionTWR . . . . . . . . . . . . . . . Thrust-weight ratioUSD . . . . . . . . . . . . . . . . US DollarWPG . . . . . . . . . . . . . . . Workscope planning guide

List of Symbols

Symbol Unit Description

BaseInterval EFH Base interval between shop visits determinedthrough CER

BaseSV RCEFHUSDEFH Base SV restoration costs per EFH determined

through CER

BaseSV R USDEFH Base shop visit rate determined through

BaseInterval

EFHLC EFH Accumulated engine flight hours throughout thelife cycle

Interval EFH Interval length between shop visits

LLPCost USDEFC Costs per EFC incurred by LLP replacement

LLPCostEFHUSDEFH Costs per EFH incurred by LLP replacement

SEC USD Costs incurred by providing spare engines

SV C USD Total costs for a shop visit

SV CEFHUSDEFH Total SVC per EFH

SV CLC USD Accumulated engine shop visit costs throughoutthe life cycle

SV R SV s1000EFH Number of shop visits per 1000 EFH

SV RC USD Total costs for the engine’s performance restora-tion

SV RCEFHUSDEFH SV restoration costs per EFH

Util ann EFH Annual utilization of an engine

Y earsLC years Number of years of the engine life cycle

VII

1 Introduction

The global market for passenger and freight air transportation has tremendously grown overthe past decades and it is expected to keep expanding at a high pace. At the same time, theairlines see themselves in a more competitive market environment, especially with the emergingnumber of low-cost-carriers that has marked a turning point in the market structure. In orderto stay competitive, airlines need to constantly seek cost saving potentials. This ambition isclosely linked to evaluating new technologies and their possible contribution to reducing thelong term costs for owning and operating the entire aircraft system throughout its life cycle. Aconsiderable share of these life cycle costs (LCC) are expenditures for maintenance, repair andoverhaul (MRO) of the individual aircraft systems. The biggest proportion of the aircraft MROcosts is incurred by the engine (fig. 1.1).

Engine MRO

35%

Components MRO

21%

Line Maintenance

21%

Airframe Heavy

MRO

14%

Modifications

9%

Figure 1.1: Aircraft MRO cost overview [Jet08]

Most aircraft engines used in today’s air transportation industry are gas turbine engines.The mechanical complexity of these engines results in considerable labour costs required forMRO related tasks such as disassembly, inspection, reassembly and test. In addition, the enginedesign requires highly tensile and thermo resistant materials, which results in high material costsfor repair and replacement of worn parts. Therefore, engine MRO is considered as cost driverand it is in the interest of aircraft operators to estimate the life cycle costs caused by enginemaintenance, when making decisions regarding their engine fleet.

1

1.1 Thesis Objectives 2

1.1 Thesis Objectives

The objective of this thesis is to develop a model that is capable of predicting the MRO costsof commercial jet engines, using the method of parametric cost modeling based on availablehistoric data. The focus of the model is supposed to lie on the engine MRO that is performedin regular intervals off-wing in dedicated engine workshops. Therefore, it should also enablethe estimation of engine maintenance intervals. This work is intended to complement that ofSchilling [Sch09], which established a method for considering the various maintenance events ofan aircraft life cycle as part of a LCC simulation tool realized in Matlab. The newly developedmodel is supposed to elaborate the existing consideration of engine maintenance as part ofthe LCC maintenance module. The LCC-tool enables the evaluation of technologies under theincorporation of expertise in form of technology factors. Therefore, the model is not aimed toforecast accurate figures of engine MRO costs and intervals. Rather, it is intended to qualitativelyreflect the general influence factors of engine maintenance and estimate reasonable cost andinterval figures accordingly.

In support of the development of the engine MRO model, this thesis is meant to serve as a reviewon basic concepts and relationships in jet engine maintenance. The aim of this review is a betterunderstanding of the decisive characteristics that affect the maintenance costs of aircraft engines.

1.2 Thesis Structure

To meet the specific objectives, this thesis is structured into three major sections.

First Section The first section (chapter 2) includes a brief introduction to gas turbine engines ingeneral followed by an extensive literature review on engine maintenance and parametric costmodeling.

Second Section The second part (chapters 3 and 4) describes the process of developing costestimating relationships using a methodical approach introduced in the first section. Therefore,an adequate database is assembled based on available data sources. The resulting database is thenstatistically analyzed in order to establish valid cost estimating relationships. Building on theserelationships and based on the conclusions of the literature review, the considerations that led tothe final engine maintenance model are subsequently described including a demonstration of anexample application. Finally the model is checked for its plausibility using additionally availableindependent data and its sensitivity on changes of different input parameters is illustrated.

Third Section The third section (chapter 5) revolves around implementing the developed modelinto the existing LCC simulation tool. The model is implemented in an independent sub modulethat is integrated in the existing programme sequence. However, the implementation required afew minor adjustments to the input file as well as to the tool structure itself. All changes arerespectively documented.

2 Literature Review

This chapter is an summary of the reviewed literature and provides a theoretical background forthe present study. It is structured in three sections. First an overview of the working principleand the composition of aircraft gas turbine engines is given, followed by an analysis of the variousaspects in engine maintenance. The last section reviews relevant concepts for the modeling ofengine maintenance costs.

2.1 Aircraft Gas Turbine Engines

The first successful application of gas turbines engines for powering an aircraft in 1939 waspreceded by a long development time. Simple gas turbines have been used as windmills sinceancient times. However, it was not before the industrial revolution in the 19th century that firstattempts were made to use gas turbines for propulsion. As a result of this early gas turbineengines were developed, shortly after the first successful engine driven flight in the beginning ofthe 20th century. Since then, the development grew rapidly. Bill Gunston [Gun95] describes indetail the evolution of the turbine aero engine from the early prototypes over the first applicationsin second world war to modern civil and military aircraft engines. Today, there are severaldifferent kinds of gas turbine engines, all sharing the same basic engine core principle [Tew07]. Abrief description of this engine core and its basic principles will be addressed in the next sectionsalong with a closer look at the turbofan engine, which is by far the most common engine used intoday’s civil aviation. Since the cost estimation model as result of this work is based on costinformation from turbofan maintenance, it is important to have a rough picture of their layoutand composition.

2.1.1 Basic Principles

The gas turbine engine is an internal combustion engine based on the following process: acontinuous flow of air is sucked in an inlet, densified by a compressor, heated up by burning fuelin the combustion chamber and eventually leaves through a turbine. The compressor and theturbine are placed on one shaft and is sometimes referred to as a spool. Therefore the compressoris powered by using a part of the kinetic energy from the hot compressed air, that escapesthrough the turbine. Gas turbine engines also handle the working fluid in a smooth continuousflow and each part of the working cycle takes place simultaneously in a different part of theengine, unlike in piston engines. This basic configuration consisting of a compressor, a combus-tion chamber as well as a turbine is the engine core, which all gas turbine engines have in common.

3

2.1 Aircraft Gas Turbine Engines 4

compressor turbine

12 3

4

fuel

combustion

chamberair exhaust

W

The Brayton Cycle

volume temperature

pressure

pressure

1 4

2 3

1

2 3

4

expansion

compressio

n

q q

OAT [C°]

EGT [C°]

EGT Margin

Redline EGT

Corner Point SLOATL

Take-Off EGT

OAT [C°]

EGT [C°] Redline EGT The engine becomes less

efficient, due to wear of

compressor/turbine blades

The loss of efficiency has

to be compensated by

an increased fuel burn

The increase in fuel burn

results in a higher EGT

New Engine

Deteriorated Engine

EGT Margin

Engine Flight Cycles

EGTM Erosion [C°]

Installation

Loss

Restoration

in Shop Visit

Engine Time On-Wing

$/ESV

$/EFH

Engine DMC [$/EFH]

High Cost

due to low

utilization

Target

TOW

Engine Time On-Wing

Increasing cost

due to extended

workscopes

EGT Limit

Figure 2.1: Engine core of gas turbine engines

The working cycle of such gas turbines is called the Brayton cycle. The cycle efficiency dependson the achieved temperature ratio T3/T1 as well as on the given pressure ratio p2/p1. A closerlook on the thermal and cycle efficiencies of gas turbine engines is given in [Cum97].

compressor turbine

12 3

4

fuel

combustion

chamberair exhaust

W

The Brayton Cycle

volume temperature

pressure

pressure

1 4

2 3

1

2 3

4

expansion

compressio

n

q q

OAT [C°]

EGT [C°]

EGT Margin

Redline EGT

Corner Point SLOATL

Take-Off EGT

OAT [C°]









New Engine

Deteriorated Engine

EGT Margin


EGTM Erosion [C°]

Installation

Loss

Restoration

in Shop Visit

Engine Time On-Wing

$/ESV

$/EFH

Engine DMC [$/EFH]

High Cost

due to low

utilization

Target

TOW

Engine Time On-Wing

Increasing cost

due to extended

workscopes

EGT Limit

Figure 2.2: Working cycle of a gas turbine engine

There are two ways of generating thrust from this working principle. On the one hand, thehot high pressure airflow leaving the turbine can be accelerated to high speed by a nozzle behindthe turbine. This is called jet propulsion. The other approach is to utilize the kinetic energy ofthe air flow mainly for providing shaft power to drive a propeller. In such a set up, there is stillan exhaust air flow that can contribute to the over all thrust. However, the main share of thepropulsion is generated by a turbine driven propeller or fan. Whereas jet propulsion is based onaccelerating a relatively small mass of air at high speed, the propeller accelerates a large mass ofair at much lower speed [Bur97]. Several gas turbine engine types have developed out of thisworking principle. These types are briefly outlines in the following:

Turbojet The turbojet is the earliest and simplest type of all gas turbine engines [Gun95]. Itconsists of an air-intake an engine core as described above and a nozzle to accelerate the exhaustair flow. The turbojet is a pure jet propulsion engine, thus the turbine is supposed to extractjust enough energy from the gas flow to drive the compressor, so that as much energy as possibleis left in the flow to form the propulsive jet. The turbojet engine provides a great amount of


thrust at high speed and high altitude, but has the disadvantage of low thrust at low forwardspeeds (i.e. take-off).

Turboprop Turboprop engines consist of the engine core like the turbojet but with the additionof a propeller output reduction gear and a second propeller shaft, which makes this engine typemore complicated and heavier than other gas turbine engines. Unlike turbojets it is the aim toextract all the energy from the gas flow in the turbine and convert it into shaft power for drivinga propeller. Hence the turboprop generates only a small amount of jet propulsion. Turbopropengines are characterized by a high propulsive efficiency at low airspeeds. The engine is thereforeable to develop very high thrust at take-off. However, together with the propulsive efficiency thethrust falls rapidly at speeds above 800 km/h.

Turbofan Like the other gas turbine engines, the heart of the turbofan engine is the core turbine.In addition, it has a duct-enclosed fan which is usually mounted on the front of the engine. Theair entering the engine passes through the fan and splits into two separated air streams. The corestream provides the working fluid for the combustion cycle, whereas the second stream bypassesthe engine core, hence its name, bypass airflow. In the following section, the turbofan engine isdiscussed in some detail.

Propfan A recent development of gas turbine engines is a combination of the Turbofan and theTurboprop. The propfan, also known as ultra-high-bypass- or open rotor jet engine, is featuredby an unducted propeller of radically different design to conventional propellers [OCE91]. Thereare two types of propfans, one with the propeller module in the front of the engine and one atthe rear of the turbine module. This design is said to result in a very low fuel consumption athigh sub-sonic speeds.

2.1.2 The Turbofan Engine

As mentioned earlier, the turbofan is the main engine used in today’s commercial aircraft. Thedevelopment of these engines began early, almost simultaneously with the first turbojet engines inthe 1930’s [Gun95]. In order to improve the low efficiency of the turbojet at take-off and subsonicspeeds, it was proposed to use extra turbine shaft power to drive a bigger compressor (fan), thatcould accelerate a way bigger airflow at lower speeds. This low speed airflow is bypassed aroundthe engine core and usually expands in a separate nozzle at the outlet. The result of this conceptis the turbofan or bypass engine. It is usually realized in a multiple-spool configuration. Thisallows the big fan to rotate independently from the compressor stages of the engine core. Theturbofan can be seen as a compromise between turbojet and turboprop [FAA04]. The turbofanis featured by an increased thrust at low forward speeds, similar to a turboprop. However, itsthrust is not penalized with increasing airspeed up to about Mach 1. The thrust specific fuelconsumption (TSFC) as well as the specific weight of the turbofan engine fall between turbojetand turbofan. Noteworthy is also the considerable lower noise level of turbofans due to the


low exhaust velocity of the bypassed airflow. The sum of these features makes the turbofanapplicable for a wider range of aircraft in comparison to turbojet and turboprop engines. Amore detailed comparison of the three gas turbine engines and their characteristics along withdescriptive diagrams that show the correlation between net thrust as well as TSFC and theairspeed at sea level and in high altitude is given in [Tre79].

source: GE

[modified]

bypass airflow

core airflow

Engine Core

[HPC - Combustor - HPT]LPC LPT NozzleFanIntake

Figure 2.3: GEnx-2B - high bypass twin-spool turbofan

The amount of air that is bypassed in relation to the airflow going through the engine coreis called bypass-ratio (BPR). Turbofans can be distinguished in low and high-bypass engines.The former have a BRP in the range of 0.2:1 to 1:1 and can be found in super-sonic combataircraft due to their fuel economy at high speeds [Hue03]. Engines with a BPR of 5:1 andmore are termed high bypass-ratio engines. Today they practically make up all engines in highsub-sonic military and civil aircraft. Similar to a turboprop, most of the total thrust of highbypass turbofan engines is produced by the bypass air accelerated in the fan stage, whereas theengine core primarily acts as gas generator providing the power to drive the turbines. Generallyspeaking, a higher BPR leads to a reduced TSFC. However with increasing BPR also the sizeand weight of the engine rise. As a result of this, the BPR is somewhat limited by factors likeavailable ground clearance under the wing or tolerable stress in the wing structure.

The design of conventional turbofan engines can also be distinguished between two-spool andthree-spool configurations. In the more common two-spool turbofan, the low-pressure compressor(LPC) stages and the fan stages are mounted on one shaft together with the low-pressure turbine(LPT). The second shaft is hollow and contains the high-pressure compressor (HPC) as well asthe high-pressure turbine (HPT). In order to reach higher bypass-ratios in an effort to reducethe fuel consumption, fan diameters of turbofan engines have been steadily increased over thelast decades. However, the tip speed of fan blades is limited to less than supersonic, due tomaterial constraints. Since the fan and LPC of two-spool turbofans are places on the same shaft,


the rotational speed of the LPC is limited to the tolerable revolutions per minute (RPMs) ofthe fan. Therefore, the LPC and HPC need a relatively high number of stages to achieve highcompression ratios. The three-spool concept was developed to overcome this issue. The fan andthe first compressor module in the core engine are mounted on different shafts so that they canturn at different RPMs. The result of this is an intermediate-pressure compressor (IPC) thatcan turn faster then the LPC of a two-spool engine. Since the compressors turn at higher RPMs,they also require less stages, which leads to a shorter and lighter design and generally to moredurability than two-spool engines of the same size. The configuration with the fan, IPC andHPC each placed on different shafts and driven by a dedicated turbine also makes it possiblethat each spool can turn at optimized velocity. The disadvantage of the three-spool design is thecomplexity of its construction [Air08a].

LPC + Fan

HPC

Turbines IPCHPC

Turbines

Fan

two-spool three-spool

Figure 2.4: Comparison: two- and three-spool configuration

2.1.2.1 Layout and Module Characteristics

The design of today’s turbofan engines follows a modular concept. This modular design essentiallyreflects maintenance aspects. Each of the modules has its own identity, service history and specificinspection schedules. During a shop visit, any of the individual modules can be removed from theengine as an entire unit without disassembling it into its piece parts. Figure 2.5 illustrates themodular structure of a typical two-spool turbofan engine (IAE V2500-A5). A short descriptionof each of this main modules is given below.

1 Introduction

1.1 Engine Systems in General

A turbine engine consists of its main components, which change the state

of the gas flow in the sequence of the thermodynamic working cycle. The

design of modern turbofan engines follows a modular concept. Thus a

typical twin-spool turbofan engine, like the V2500-A5 shown in Fig. 1.1,

is composed of the following main modules:

• Fan module

• Low pressure compressor module

• Core engine or gas generator

• Low pressure turbine module

• Accessory gearbox module

The core engine consists of the high pressure compressor, the combus-

tion section and the high pressure turbine. This modular design of the en-

gine mainly reflects maintenance aspects. During engine disassembly each

Fig. 1.1 The main modules of a V2500-A5

Fan Module Low Pressure Compressor Module

Core Engine Module

Low Pressure Turbine Module

Accessory Gearbox Module

Figure 2.5: The main modules of a V2500-A5 [Lin08]


Fan The fan is simply a specialized type of a compressor and usually contains one stage. Thefan draws air into the engine, compressing the bypass airflow to produce most of the enginesthrust and supplying air to the gas turbine core. The fan module consists of the fan disk withfan blades mounted to the low-pressure shaft. Today’s fan blades and disks are made of titaniumalloys. However, more and more blades of newer generation models are also made of carbon fibrereinforced plastics (CFRP) [Rol07].

Low-Pressure Compressor The main purpose of the LPC is to increase the pressure of the airthrough the gas turbine core. In this example, the LPC module contains not only the low-pressure compressor case and stages, but also the fan case. Large civil gas turbine engines thatare considered in this paper have axial-flow compressors. That means the air is compressed in adirection parallel to the engine axis. An axial-flow compressor is made up of alternating stagesof rotating blades and static vanes. In order to achieve a high pressure rise, the compression isspread over a number of stages. Today’s LPC blades and vanes are generally made of aluminumalloys [Cum97].

Core Engine The core engine module consists of the inner casings, a high-pressure shaft, ahigh-pressure compressor, the combustion system as well as the high-pressure turbine. The HPCis used in conjunction with the LPC and also contains alternating stages of rotor blades andstator vanes, which further compress the air before it is supplied to the combustor. It is especiallythe later stages of a HPC that handle an airflow at considerable higher temperature and pressure,which is why the blades and vanes are made of more temperature resisting titanium and nickelalloys. In the combustion system, fuel is burnt with the air received from the compressor modules,sending hot gas downstream to the HPT. It consists of a combustion chamber, a fuel injector, anigniter and nozzle guide vanes. The following HPT is made up of one or more turbine rotors aswell as a set of stationary nozzle guide vanes. The HPT converts part of the energy stored withinthe hot gas into kinetic energy to drive the HPC and the accessory gearbox. Both combustorand HPT are exposed to the maximum temperatures that occur in the engine therefore, coolingair and ceramic coated nickel alloys are used to increase component lives. Generally, in a runningengine, it is the core engine module that is subjected to the most compelling conditions in termsof temperature, pressure and rotational velocity. Thus, it will be the module that suffers thefastest deterioration of performance [Ack10].

Low-Pressure Turbine The LPT module is located in the rear of the engine downstream of theHPT module. It is an assembly of disks with turbine blades that are attached to the low pressureshaft, nozzle guide vanes and a rear frame. The LPT removes the remaining energy from thecombustion gases to power the LPC module.

Accessory Gearbox The accessory gearbox is attached to the bottom or side of the engine.The aircraft engine not only provides thrust, but it also supplies power for engine and aircraftaccessories. This includes starters, fuel and oil pumps as well as hydraulic pumps and generators


for cabin power. The accessory gearbox is where all this mechanical-driven components aremounted to the engine.

This compound of main engine modules can be referred to as basic engine. However, this basicengine by itself is not operable and cannot serve all necessary functions. In addition to its maincomponents the engine needs various systems to become operable. These engine-related systemsinclude amongst others an air cooling and sealing system, a lubrication system, a fuel distributionsystem, an exhaust and thrust reverser system as well as an air inlet and a nozzle [Lin08].

2.1.2.2 Engine Operating Parameters

Modern aircraft are equipped with a multitude of gauges to provide the flight crew with feedbackinformation about the engine condition. The main operating parameters contain the speeds ofthe engine spools and the engine pressure ratio (EPR) for performance monitoring, as well asthe temperatures of the turbine gases for health monitoring. A brief description of these keyoperating parameters is given below:

N1 and N2/N3 speeds In a jet engine, every main revolving section has a separate gauge tomonitor its RPMs. Depending on the engine type, the N1-gauge keeps track of the LPC and/orfan speed. The core section is monitored by the N2-gauge, whereas a three-spool engine hasan additional N3-gauge. Due to the high revolving velocities, the RPMs of the engine spoolsare displayed as percentage of the design RPM rather than actual RPM. The N1-speed is theprimary indication of thrust on most turbofans [FAA04].

Engine Pressure Ratio (EPR) The EPR is the total pressure ratio across the engine and is definedas the ratio of the pressure at turbine exit (exhaust) to the pressure at the intake. On someturbofans, it serves as primary thrust indication gauge.

Turbine Inlet Temperature (TIT) The TIT is the gas temperature from the combustor exit as itenters the first HPT stage. As the highest temperature inside a gas turbine engine, the TIT isone of the limiting factors for the power output of an engine. However, it is difficult to measuretherefore, the exhaust gas temperature (EGT) is usually the parameter measured.

Exhaust Gas Temperature (EGT) The EGT is the temperature of the exhaust gases as they enterthe tail pipe, after passing through the LPT. It is expressed in degrees centigrade and can beseen as one of the most important health monitoring parameters. The engine gas temperatureshave to be closely monitored, as exceeding temperature limits may lead to serious heat damageto the turbine components [FAA04]. In addition, the EGT is a measure of the engine’s efficiencyin producing its design level of thrust. A high EGT may indicate that the engine has sufferedsignificant hardware deterioration during service. Generally, the EGT reaches its maximumduring take-off or right after lift-off, as the engine operates here at its peak.


EGT Margin (EGTM) In order to measure the level of an engine’s performance degradation,the so called EGT margin has been introduced. The EGT margin of an engine is the differencebetween the maximum tolerable EGT (Redline EGT) and the peak EGT during take-off. Thisredline EGT is the absolute temperature limit, which cannot be exceeded without damaging theengine [Bra04]. Therefore, the EGT margin is a measure for how well below this limit the engineoperates in times of maximum power output at take-off. As the EGT of an engine increases overtime, due to hardware deterioration, the EGT margin decreases. Theoretically, an engine canremain on wing until its EGT margin has become zero. It is normally at its highest level whenthe engine is new or has just been refurbished.The EGT margin is furthermore highly influenced by the present outside air temperature (OAT).For a given thrust setting, the EGT rises at a constant rate as the OAT increases. Figure 2.6shows the relationship between take-off EGT and OAT.

compressor turbine

12 3

4

fuel

combustion

chamberair exhaust

W

The Brayton Cycle

volume temperature

pressure

pressure

1 4

2 3

1

2 3

4

expansion

compressio

n

q q

OAT [C°]

EGT [C°]

EGT Margin

Redline EGT

Corner Point SLOATL

Take-Off EGT

OAT [C°]









New Engine

Deteriorated Engine

EGT Margin


EGTM Erosion [C°]

Installation

Loss

Restoration

in Shop Visit

Engine Time On-Wing

$/ESV

$/EFH

Engine DMC [$/EFH]

High Cost

due to low

utilization

Target

TOW

Engine Time On-Wing

Increasing cost

due to extended

workscopes

EGT Limit

Figure 2.6: Correlation between Take-Off EGT and OAT [Air06b]

The pictured curve is a result of the power management schedule of the digital engine controller(FADEC). It is programmed to provide constant maximum thrust with increasing OAT. As theOAT rises, the air density decreases. Therefore, the throttle has to be increased in order tomaintain constant thrust, which results in an increase in EGT. However, constant maximumthrust is only maintained up to a certain OAT (corner point). The FADEC is then programmedto keep the EGT constant for OATs higher then the corner point temperature. This powermanagement setting is called flat rating and makes sure that the engine operates with enoughEGT margin also at high OATs. The constant EGT is maintained by reducing the engine thrustas the OAT rises beyond the corner point [Air06b]. Without flat rating, the EGT would continueto rise with increasing OAT as the dashed line in fig. 2.6 indicates. The OAT at which the EGTwould reach the redline EGT, if maximum take-off thrust was maintained is termed sea leveloutside air temperature limit (SLOATL). The actual highest permitted thrust setting for a givenOAT can be determined by calculating the SLOATL.Since the EGT margin is the main indicator for an engine’s health status, it is normally expressedindependently from the OAT. That means the EGT margin is given as the difference between

2.2 Aircraft Engine Maintenance 11

redline EGT and the actual EGT at maximum thrust at the corner point OAT (fig.2.6).

2.2 Aircraft Engine Maintenance

This section is supposed to give an overview over the different aspects and in engine maintenance.The aircraft engine as a major airplane component, in terms of investment, operating cost aswell as its complexity, follows its own maintenance schedule mainly independently from theregular maintenance check events of the remainder aircraft. For a more detailed discussion of themaintenance programme of an aircraft, it is referred to [Sch09]. Engine maintenance is a broadfield that includes not only engineering but also various complex planning and managementproblems. As such, it is often also referred to as engine management.Modern engine maintenance is based on the so called on-condition method. After this method,engines removals and overhauls only take place when the engine condition demands it [Rup00].Whereas in the past, engines were removed and maintained after a fixed time interval (hard-timeinterval), which had the disadvantage of engines being removed, even in case of a safe operatingengine. Similar to the maintenance of the remainder aircraft, engine maintenance can be dividedinto on-wing and off-wing maintenance or overhaul, in the following referred to as shop visit(SV). The next two subsections will discuss both maintenance components. Subsequently, enginetime on-wing and maintenance cost are discussed separately, due to their importance for theproposed model.

2.2.1 On-Wing Engine Maintenance

On-wing engine maintenance, also known as engine line maintenance, includes all maintenanceand inspection activities that can be done without engine removal and disassembly on the flightline. As such it is generally included into the line maintenance schedule of the aircraft operation.As as result of the on-condition maintenance concept, a great share of on-wing maintenanceactivities involves Engine Condition Monitoring. The aim is to monitor and analyze the mainoperating parameters as well as the internal physical condition of the engine, in order to identifypotential problems before they become serious and to provide data that can be used to determinethe most economic times for engine shop visits. A second group of tasks can be summarizedas On-Wing Repair and Replacement. In the recent past, more and more actions have beendeveloped to access the site of engine damage directly on-wing and without complete disassembly.As a result of this, more engine problems can be fixed on-wing, which significantly extends thetime on-wing (TOW) of the engine [Bur10]. The following is a more detailed discussion on bothkinds of on-wing maintenance.

2.2.1.1 Engine Condition Monitoring (ECM)

Today’s ECM systems evolved as a result of aviation authorities requiring flight crews to monitorbasic engine performance parameters from the flight deck instruments. The recorded data wasthen used by the engineering departments of the airlines to determine the maintenance programme


for the engine. Therefore, ECM data was historically recorded manually and only during take-offand once in the cruise. On modern aircraft, ECM information is gathered automatically inhigher number and quality and can even be recorded and transmitted to a ground station inreal-time [Air05a]. The engine performance parameters that are measured can be divided intotwo categories. The first consists of parameters that are not heavily influenced by flight conditionsand engine thrust, like engine vibrations as well as oil temperature and pressure. The second typeof parameters comprise those that are affected by flight conditions and thrust. These parametersinclude the gas path temperatures like the EGT, EPR, fuel flow as well as the N1 and N2 speeds.In order to also provide data for indications of the present flight conditions, parameters likealtitude Mach number and air temperature are measured and recorded as well [Air05a]. Thekey objective of ECM is to plot the performance trend data, so that it can be compared to amodel of how the engine is expected to behave under the experienced flight conditions. Shifts inperformance indicate hardware deterioration or operational problems. Combinations of specificparameter changes are known to be indications for specific deviations in the engine. The datacan be further interpreted to find out which part of the engine is inducing the problems. Thisanalysis of the recorded data is undertaken by specialized ECM software usually provided bythe original equipment manufactures (OEMs). It is expected that future ECM systems willcapture more accurate data and have more elaborated data interpretation capabilities than thecurrent generation [Air06c]. In addition to the recording and analysis of engine performancedata, ECM also includes monitoring the physical condition of internal engine parts with the helpof inspection borescopes. An inspection borescope is an optical diagnosis tool comprising of along flexible tube and an optical lense, that gives an magnified and illuminated view of hardlyaccessible areas inside the engine. It allows to inspect internal engine parts for defects such ascracks, stress fractures and corrosion.To sum up, ECM allows the concept of on-condition maintenance of aircraft engines. It helpsto manage the timing of both scheduled and unscheduled shop visits and it prevents excessivehardware deterioration and it provides initial alerts that allow engines to be fixed on-wing[Air05a].

2.2.1.2 On-Wing Repair and Replacement

Aircraft engines usually have a design life that exceeds the achieved actual shop visit intervals byfar. This is due to part failures and unexpected damages. For instance, the fan and LPT modulesare often the first areas to suffer environmental damage due to their exposure to birds anddebris [Bur10]. The ECM systems described above, are able to detect such problems and provideinformation that help the maintenance engineers to decide if an on-wing repair or replacementshould be conducted. Together with ECM systems, the on-wing repair capabilities are gettingmore and more sophisticated. Today, on-wing maintenance includes repairs that historically havebeen high-cost shop repairs [Bur10]. However, as a result of progressive hardware deterioration,an engine overhaul is eventually unavoidable. On-wing repair though, contributes to extendthe engine’s time on-wing as close to its design life as possible, despite unexpected failures or


damages. Also falling into this category is the replacement of line replaceable units (LRUs).These are parts that are designed to be quickly replaced on the flight line. They are usuallysealed units like, sensors, pumps, filters or tanks and can be replaced independently from theirsurroundings.On-wing repair and replacement not only has the benefit that it saves the time and moneyfor engine removal and complete disassembly, but also that there is no need for a spare enginein order to keep the aircraft in service. In addition, it can be included into the aircraft’s linemaintenance schedule. GE’s On Wing Support for instance, performs flight line repairs likeborescope blending of compressor blades, fan module and gearbox workscopes as well as top casecompressor repairs [GE 08]. Another technique that falls into this category is the so called enginewater wash. It can be done without requiring additional ground time and involves spraying abouta hundred liters of water repeatedly into the front of the turning but not burning engine, wherethe engine cleans itself. This procedure reduces fuel consumption by improving the EGT marginand therefore extends the on-wing intervals [KLM07].

2.2.2 Engine Overhaul - Shop Visit

Despite better on-wing maintenance technology, eventually every engine has to be removedfrom wing and disassembled in order to get more extensive maintenance. Airlines usually haveaccess to a pool of spare engines. Therefore, the removed engine is immediately replaced by aspare engine for the time the engine is being refurbished. Depending on the engine type, thereplacement can be performed by 3-4 mechanics in a full 8-hour shift. Thus, the engine removalcan be conducted on the flight line. The overhaul is performed in a dedicated engine workshop,hence its name engine shop visit (ESV). This subsection discusses the main causes that force anengine to be removed and overhauled followed by a brief description of the shop visit processand shop visit management considerations.

2.2.2.1 Main Causes of Engine Removals

As previously mentioned, a modern aircraft engine’s condition is constantly monitored. Thisallows one to predict the time when the engine has deteriorated to a level where an engineoverhaul becomes necessary. That means an ESV is generally a scheduled event that repeats inregular predictable intervals. However, especially in situations where an unexpected part failureor damage cant be fixed by on-wing maintenance efforts, an engine has to be removed and sentto the workshop prematurely. The primary engine removal causes can be categorized in fourgroups: EGT margin degradation, expiry of Life Limited Parts (LLPs), hardware deteriorationand other unscheduled removal causes. [Air00]. The causes of engine removals depend heavily onthe type of aircraft operation. Engines operating on short-haul routes show a higher percentageof removals caused by EGT margin degradation and LLP expiry, while medium- and long-hauloperating engines tend to have a higher share of removals due to hardware deterioration andEGTM degradation. The distribution of the engine removals on the removal causes dependingon the aircraft operation and the engine age status is illustrated in figure 2.7. In the following


parapraphs the primary removal causes are briefly discussed.

Figure 2.7: Removal causes depending on aircraft operation [Ack10]

EGT Margin Degradation The degradation of an engine’s EGTM is one of the dominatingremoval driver. EGT margin degradation is a result of the gradual wear of the compressor andturbine blades. This leads to a rising clearance between the blade tips and the surroundingshrouds and thus, to an increasing leakage of the working fluid in the compressor and turbinestages [Air08c]. Such leakage causes a decrease in the overall engine efficiency and performance.Therefore, a deteriorated engine has to burn more fuel than a new engine in order to achieve thesame required thrust level. As the engine wear continues, the EGT will ultimately rise until theEGT margin is so little that severe damage to the turbine components cannot be excluded. Inthis case, the engine needs to get a performance restoration in order to remain operable. Figure2.8 illustrates the effects of engine wear on the EGT margin.

compressor turbine

12 3

4

fuel

combustion

chamberair exhaust

W

The Brayton Cycle

volume temperature

pressure

pressure

1 4

2 3

1

2 3

4

expansion

compressio

n

q q

OAT [C°]

EGT [C°]

EGT Margin

Redline EGT

Corner Point SLOATL

Take-Off EGT

OAT [C°]









New Engine

Deteriorated Engine

EGT Margin


EGTM Erosion [C°]

Installation

Loss

Restoration

in Shop Visit

Engine Time On-Wing

$/ESV

$/EFH

Engine DMC [$/EFH]

High Cost

due to low

utilization

Target

TOW

Engine Time On-Wing

Increasing cost

due to extended

workscopes

EGT Limit

Figure 2.8: Effects of engine wear on the EGT Margin [Ack10]

The rate of EGT margin erosion normally depends on the thrust rating as well as on how theengine is operated. A more detailed discussion on the factors that influence the EGT marginerosion rate is given in the following section 2.2.3. Generally, the rates of EGT margin degradationare highest during the first 1000 - 2000 engine flight cycles (EFC) after installation. This iscalled Installation Loss. The erosion rate stabilizes thereafter and reaches a fairly constantlevel until the the engine is removed [Air05b]. In the following engine shop visit, the EGTmargin is restored. However, it is normally not possible to fully restore the initial EGT margin


of a new engine. Overhauled engines typically achieve 70%-80% of the initial EGT marginlevel. This contributes to the fact that an engine’s first TOW is usually the longest achievedduring its life cycle. Figure 2.9 shows the qualitative relationship between EGT margin erosionand accumulated EFC’s. Since an engine’s EGTM is constantly monitored, it is fairly easy todetermine its actual stabilized EGTM degradation rates. Therefore, it is possible to predict thepoint where the EGT margin becomes zero. In other words, EGT margin degradation generallyleads to a scheduled ESV.

compressor turbine

12 3

4

fuel

combustion

chamberair exhaust

W

The Brayton Cycle

volume temperature

pressure

pressure

1 4

2 3

1

2 3

4

expansion

compressio

n

q q

OAT [C°]

EGT [C°]

EGT Margin

Redline EGT

Corner Point SLOATL

Take-Off EGT

OAT [C°]EGT [C°] Redline EGT The engine becomes less








New Engine

Deteriorated Engine

EGT Margin


EGTM Erosion [C°]

Installation

Loss

Restoration

in Shop Visit

Engine Time On-Wing

$/ESV

$/EFH

Engine DMC [$/EFH]

High Cost

due to low

utilization

Target

TOW

Engine Time On-Wing

Increasing cost

due to extended

workscopes

EGT Limit

Figure 2.9: Trend of EGT margin erosion rates over accumulated EFC

Life Limited Part Expiry Life limited parts (LLPs) are defined as engine rotor and major staticstructural parts whose failure could result in hazardous engine effects. Such effects include forinstance uncontrolled fire or complete inability to shut down the engine. The Advisory Circular(AC) 33.70 [FAA09], issued by the FAA, regulates the standards for the design and testing ofengine LLPs. Life limited rotational parts include disks, spools, spacers and shafts, whereasstatic structural parts generally include high-pressure cases and non-redundant engine mountcomponents. For each LLP, an operating limitation or life limit must be established in order toensure that no hazardous effect occurs. The life limit specifies the maximum number of finiteflights or engine flight cycles (EFC) a LLP is allowed to be in service. The life limit for rotatingparts is for example equal to the minimum number of EFC, that is required to trigger a crack ofabout 7 mm in length by 3.5 mm in depth [FAA09]. The definition of maximum permissible lifetimes for certain engine parts is an exception from the general on-condition maintenance conceptfor aircraft engines. The life limit of LLPs is defined by the engine manufactures and typicallyranges between 15,000-30,000 EFC. However, some individual parts can have restricted lives,due to technical issues and imposed airworthiness directives (AD) [Air04a]. Once an engine hasaccumulated as many EFC as the shortest life limit of all equipped LLPs, it has to be removedand sent to a workshop in order to replace the used up LLPs. Hence, LLP replacement can bescheduled in coordination with the expected point of full EGT margin erosion. LLP replacementis a major cost driver in engine maintenance and as such it is subjected to several cost savingmeasures.


Hardware Deterioration The third main removal cause is the deterioration of the engine hardware.All engine components are exposed to different kinds of deterioration mechanisms. These includeamongst others, low and high cycle fatigue, thermo-mechanical fatigue as well as corrosion[MM10]. These mechanisms lead to a degradation of the part lives or in worst case to a partfailure as well as to a loss of engine performance. In contradiction to LLPs, the remainder engineparts are replaced on an on-condition basis. A safely operating engine therefore relies on a wellfunctioning ECM system to detect problems related to hardware deterioration. The engine’score module, being exposed to the highest temperatures and revolving velocities within theengine, suffers in particular from the mentioned deterioration mechanisms. These engine failuresare practically not predictable but modern ECM systems are capable of detecting them soonafter they arise and therefore allow to prevent more severe damage. However, if such an enginehardware deterioration problem cannot be fixed on-wing, it forces the engine to get a prematureunscheduled shop visit.

Other removal causes The last group includes unscheduled removal causes from foreign objectdamage (FOD), engine system failures and engine vibration. FOD is engine damage resultingfrom ingestion of foreign objects. Foreign objects include birds, ice or ash as well as a runwaydebris [Ack10]. Especially the ingestion of larger objects like birds can lead to significant damageof the fan- and the LPC blades. However, such an incident usually does not affect the safeoutcome of a flight and may not even be noticed by the flight crew [Tur04]. But the ingestion offoreign objects poses a risk of latent effects, like minor cracks that can propagate by progressiveengine wear. Hence, it is important that the ECM is able to detect the occurrence of FOD. Alsofalling into this category is engine system failures. Especially lubrication system problems, suchas leaks or oil pump malfunctions can result in severe engine damage if they are not fixed.

2.2.2.2 Engine Shop Visit Process

Engine shop visits can generally be categorized by the extend of the conducted workscopes andthe number of modules on which work is performed. The level of the shop visit has a strongimpact on the following removal interval. An extensive shop visit results in a significantly longertime on-wing before the next shop visit becomes due. As mentioned in 2.1.2.1, the individualengine modules are considered independently during a shop visit. This is necessary becausethey have different rates of deterioration. Therefore each module normally requires differentworkscopes at each shop visit [Air09]. The workscopes are typically performed in dedicated shopdepartments, that are specialized on a certain module. It is also not unusual that engine shopsoutsource the overhaul of individual modules or parts to shops with more capabilities in thisparticular field. Shop visits that include work on all modules can last up to 50-90 days [ACT08].A shop visit process following the general incoming inspection and the definition of the objectedworkscopes is illustrated in figure 2.10. A detailed analysis of the different shop visit stages foreach of the main engine modules is given in [Air09].


Parts

Inspection

Repair &

Replacement

Certification &

Reassembly

Engine

Testing & Release

• Testing of the engine

on-wing or in test cell

• Updating of engine

documents

Engine

Disassembly

• Disassembly into

engine modules

• Full disassembly of

modules into piece

parts if necessaryGEnx-2B

Folie 4

Vortrag > Autor > Dokumentname > Datum

Incoming

Inspection

• Inspection of engine

exterior and interior

• Analysis of engines

service history

• Workscope definiton

Inspection

• Cleaning of parts

• non-destructive tests

for detecting cracks

• Dimensional checks

on blades and vanes

Replacement

• Part repair in special

workshops

• Replacement of non-

repairable parts

Reassembly

• Certification of

refurbished parts

• Balancing of rotables

• Reassembly of

module units

Figure 2.10: Engine shop visit process

2.2.2.3 Shop Visit Management

The primary objective of shop visit management is to minimize the long-term engine directmaintenance cost (DMC), expressed in cost per flight hour (USD/EFH) [Ack10]. A more detailedbreakdown of the composition of the DMC of an engine follows under 2.2.4. A dominant factorthat influences the engine maintenance cost is the time on-wing between shop visits. An increasingTOW will result in increased engine deterioration [Air07a]. Hence, the engine modules requiremore extensive maintenance at each shop visit. This leads to rising shop visit cost as the TOWincreases. Generally, the raised shop visit costs are compensated by the extended removal interval,due to the increased TOW. As a result of this, the overall maintenance costs per flight hourdecrease. However, the increase in engine deterioration accelerates after a certain TOW, withthe result that the required shop visit workscopes extend so severely that the shop visit costare raised to a level, where the overall USD/EFH begin to increase again [Air07a]. Figure 2.11illustrates this relationship between the engine’s TOW and the DMC per flight hour.

compressor turbine

12 3

4

fuel

combustion

chamberair exhaust

W

The Brayton Cycle

volume temperature

pressure

pressure

1 4

2 3

1

2 3

4

expansion

compressio

n

q q

OAT [C°]

EGT [C°]

EGT Margin

Redline EGT

Corner Point SLOATL

Take-Off EGT

OAT [C°]









New Engine

Deteriorated Engine

EGT Margin


EGTM Erosion [C°]

Installation

Loss

Restoration

in Shop Visit

Engine Time On-Wing

$/ESV

$/EFH

Engine DMC [$/EFH]

High Cost

due to low

utilization

Target

TOW

Engine Time On-Wing

Increasing cost

due to extended

workscopes

EGT Limit

Figure 2.11: Influences of the TOW on the DMC of an engine [Eng10]


The aim is to find a balance between shop visit cost and time on-wing, so that the lowestcost per EFH are achieved. This is a challenging task, which requires thoughtful shop visitmanagement. Four key considerations concerning the shop visit management are subsequentlydiscussed:

Workscope Planning The level of workscope to be performed on a module or an individualitem is proposed by the workscope planning guide (WPG) , issued by the engine manufacturer[Air09]. However, it also heavily depends on the result of the module inspection, the removalcause, the time on-wing (TOW) since the last shop visit, as well as the extend of previousshop visit workscopes [Air07a]. According to the set operational and economic engine buildtargets, the workscope plan of a shop visit can individually deviate from the shop visit manual[Jet08]. That means the performed workscopes can be adjusted to meet certain target on-wingtimes or target shop visit costs. In general, there are three levels of workscopes: MinimumLevel, Performance Level and Full Overhaul. A full overhaul on a module involves a completedisassembly to piece-part level and an inspection as well as repair or replacement of all parts.Lighter workscopes, like performance restorations, usually require just partial disassembly andrepair works only on certain items [Ack10].

LLP Management The management of the LLPs is essential in minimising maintenance cost,particularly for engines used on short- and medium-haul operations [Air04b]. In long-haul engines,LLPs account for a smaller share of total maintenance cost. This is because LLPs have fixedlives defined in engine flight cycles (EFC). Therefore they can last for many years in long-hauloperating engines, due to the low number of flight cycles (FC) these engines accumulate per year.Short-haul engines on the other hand, accumulate a considerably more FC each year. Hence,their LLPs have to be replaced every few years [Air04a]. As stated in 2.2.2.1 LLP life expiry is amain removal cause that forces an engine into a shop visit. The task of LLP management is tocoordinate the remaining lives of the equipped LLPs with other criteria for shop visit timing.Ideally, the LLP replacement would coincide with the optimal TOW as illustrated in Fig. 2.11as well as with the date of full EGT deterioration. However, these events may occur at a timewhen some LLPs still have a few thousand EFC left. If not replaced, this remaining life, alsocalled “stub life”, would limit the subsequent removal interval. In order to prevent an early nextshop visit, these LLPs have to be replaced and scrapped without utilizing all their available life.However, wasting remaining LLP life raises the average cost per flight hour [Air04b]. It becomesapparent that a compromise between utilization of the LLP lives and optimal time on-wing has tobe found in order to achieve the lowest long-term maintenance cost. In short-haul engines, whereLLP expiry is the main removal cause and their frequent replacement is a major cost driver, itis common that the shop visit workscopes are tailored around the LLP lives, so that the shopvisits coincide with LLP replacement [Air99]. Another key consideration is that LLPs shouldideally be replaced during a heavier shop visit, when the engine has gone through a high level ofdisassembly [Air04a]. This is because replacing LLPs also requires a high degree of disassemblyand reassembly. Man-hours for assembly works account for a large percentage of shop visit


cost. LLP replacement during a light shop visit would increase the necessary workscope andtherefore the cost. The lowest maintenance cost per EFH is accomplished when a heavy shopvisit coincides with full LLP utilization [Air04b].

PMA Parts PMA stands for Parts Manufacturer Approval and is a combined design andproduction approval for modification and replacement parts. A manufacturer who holds thePMA is allowed to produce and sell FAA approved parts that are eligible for installation on typecertificated aircraft [FAA10]. Aerospace original equipment manufacturers (OEMs) usually havea strong monopoly position on replacement parts [Ben08]. This results in generally high pricesfor original parts in the aftermarket. The OEMs legitimate the high margins on their parts withthe high level of investment in research and development, required to market a modern aircraftengine. Parts from approved manufacturers are considerably cheaper and the savings potentialcan range between 45%-75% compared to OEM pricing [Air06b]. With the OEMs continuingto impose yearly price increases of five to six percent for spare parts [Ben08], the PMA marketis expected to carry on growing. However, the engine manufacturers are striving to strengthentheir position, by restricting technical support and warranty coverage when PMA parts are usedin their engines. Yet, the legal status of PMA parts is clear. According to FAA regulations, PMAparts can be seen as equivalent to parts from the original equipment manufacturers [Hol08].

Parts Repair The increasing development of repairs for engine parts has considerably contributedto the reduction of the overall engine maintenance cost. The two key objectives of parts repair areto maintain the on-wing life and at the same time, reduce the shop visit costs [Air07a]. Repaircosts have to be considered against the costs of used parts on the surplus market and new partseither from PMA or from the original parts manufacturer. Most repairs have been developed forairfoils such as blades and vanes, since they have the largest economic impact. But repairs arealso developed for other parts that prove to be a major maintenance cost driver as the engineages. In general, it takes some time and investment for repairs to be developed for new enginemodels. Hence, repairs are rather available for mature engines [Air07a]. Another considerationis that repaired parts tend to lead to shorter subsequent on-wing life in comparison to engineswhere parts were replaced. Generally, the reduction in costs makes up for the shorter followingshop visit interval and so the overall cost per EFH are lower when utilizing repairs. Repairingparts costs five to 10 times less than replacing them [Air99]. However, the quality of repairs andtheir effect on the remainder engine has to be considered, since poor quality repairs may result inunscheduled removals. On most repairable parts, two repairs can be performed before they arescrapped at the third removal. In doing so, the costs for the second repair will be higher thanthe first. This also has to be considered when planning the parts repair strategy for a shop visit.

2.2.3 Engine Time On-Wing

The achieved time on-wing is an important parameter in engine maintenance. Until the enginereaches its target TOW, a longer TOW generally leads to reduced over all shop visit DMC per


EFH (see figure 2.11). However, engines are often forced into shop visits before reaching thistarget time on-wing. This happens not only because of unexpected engine damages but also dueto accelerated performance deterioration. The previous section demonstrated how shop visitmanagement influences the engine TOW. This section discusses the influence factors on thehardware & EGTM deterioration rates. These factors resulting from the engine built and theoperation conditions also heavily influence the engine TOW. They are summarized as follows:

• Engine Thrust Rating

• Operational Severity

• Engine Age

2.2.3.1 Engine Thrust Rating

Normally, there are several thrust ratings for a given engine model. The CFM56-7B for instance,comes in six different engine variants, all rated at different thrusts. The basic engine build is thesame, the rating is the result of the power control setting of an engine. The engine variants witha higher thrust1 level generate higher gas path temperatures [Air05b]. This results in a lowerEGT margin and normally also in a more severe EGT and hardware deterioration, due to theincreased thermal stress. Low initial EGT margin and high EGT margin erosion rates will leadquickly to complete EGT margin degradation. That means high rated engines are more likely tobe forced to get a shop visit because of full EGT margin deterioration than lower rated ones.Table 2.1 sums up the different engine variants and their EGT related parameters.

Engine Variant -7B18 -7B20 -7B22 -7B24 -7B26 -7B27

Thrust [lbs] 19,500 20,600 22,700 24,200 26,300 27,300

Initial EGT mar-gin [degC]

125-130 125-130 105 100 60 55

EGTM erosion[degC/1000EFC]

2.5-4.0 2.5-4.0 2.5-4.0 4.0-6.0 4.0-6.0 4.0-6.0

Table 2.1: Initial EGTM and mature EGT erosion rates for CFM56-7B variants [Air08c]

As a result of the more severe hardware deterioration, higher rated engines generally also tendto achieve shorter times on-wing than engines variants with low thrust ratings.

2.2.3.2 Operational Severity

Furthermore an engine’s time on-wing is heavily influenced by its operating conditions. Moredemanding conditions will result in greater stress on the engine and therefore increase the wearof the engine hardware. The major parameters of operating severity include:

1 thrust as engine specification generally means the maximum take-off thrust


• Average Flight Time

• Take-Off Derate

• Outside Air Temperature

• Environment

Average Flight Time The measurement of time length an engine is operating on-wing can bequantified in both engine flight hours (EFH) and engine flight cycles (EFC) . A flight hourrepresents one hour of flight, whereas a flight cycle represents one sequence of take-off, cruise andlanding. Both measurements can be frequently found in the literature. However, the number ofaccumulated EFC is generally the more appropriate measure for engines operating on short-haulroutes, while the time on-wing of engines that are operated on medium and long flight timesshould be considered in terms of EFH [Air08b]. During one flight cycle, it is the take-off andclimb phase, where the engine is exposed to the greatest thermal stress and engine wear. Theengine hardware deteriorates considerably less during the following cruise and landing. Therefore,the number of accumulated EFH is not a representative time measure for an engine operated onshort cycle lengths. Rather it is the number EFC in service, that is an indicator for the enginewear. On the other hand, for medium and long-haul operating engines the accumulated EFH hasmade its way as a common time on-wing measure in terms of maintenance. Figure 2.12 comparesthe flight profiles of a short-haul and a medium-haul aircraft. The flight profile of an aircraft canbe expressed by the flight hour to flight cycle ratio (FH:FC), also known as the flight leg lengthor flight time. The average FC:FH is an important parameter for the operational severity of anengine. In general, an engine that operates on a short average flight time will suffer a more rapidperformance deterioration and therefore has shorter shop visit intervals and higher DMC perflight hour. Conversely, as the the FC:FH increases the engine is exposed to less wear and canremain longer on wing with reduced USD/EFH. The mean time on-wing between shop visitsis often represented by the shop visit rate (SVR) . This characteristic is analogue to the directmaintenance cost per flight hour (USD/EFH) in terms of operational severity. It is defined as

OAT [C°]

EGT [C°]

EGT Margin

Cruise

Corner Point SLOATL

Take-Off EGT

OAT [C°]









New Engine

Deteriorated Engine

EGT Margin


EGTM Erosion [C°]

Installation

Loss

Restoration

in Shop Visit

Engine Time On-Wing

$/ESV

$/EFH

Take-off & Climb

Redline EGT

Descent & Landing

Take-off & Climb

Descen

t & Landing

Cruise

1 flight hour

Medium-Haul Operation FH:FC = 3.0

Short-Haul Operation FH:FC = 1.0

1 flight hour 1 flight hour

3 flight hours

SVR & DMC [$/EFH]

EFH:EFC [h]short

cycle time

long

cycle time

Severity Factor

EFH:EFC [h]

5%

10%

15%

1.0 1.5 2.0 2.5 3.0

1.0

0.8

1.2

1.4

1.6

Increasing Derate = Lower Severity

Derate Base Point

Figure 2.12: Two example flight profiles [Ack10]


the number of shop visits per 1000 EFH for a given engine [SV/1000EFH]. A mean TOW of17000 EFH is for instance equal to a SVR of 1000/17000 = 0.0588. Equation (2.1) describes theconversion from shop visit intervals into SVRs.

SV R = 1000ShopV isitInterval

(2.1)

More severe operation conditions will lead to an increase of both, SVR and DMC. The qualitativerelationship between EFH:EFC and the SVR as well as the DMC is illustrated in figure 2.13.

OAT [C°]

EGT [C°]

EGT Margin

Cruise

Corner Point SLOATL

Take-Off EGT

OAT [C°]









New Engine

Deteriorated Engine

EGT Margin


EGTM Erosion [C°]

Installation

Loss

Restoration

in Shop Visit

Engine Time On-Wing

$/ESV

$/EFH

Take-off & Climb

Redline EGT

Descent & Landing

Take-off & Climb

Descen

t & Landing

Cruise

1 flight hour




3 flight hours

SVR & DMC [$/EFH]

EFH:EFC [h]short

cycle time

long

cycle time

Severity Factor

EFH:EFC [h]

5%

10%

15%

1.0 1.5 2.0 2.5 3.0

1.0

0.8

1.2

1.4

1.6


Derate Base Point

Figure 2.13: Shop visit rate and DMC in relation to the flight hour flight cycle ratio

Take-Off Derate Another issue that influences the operational severity is the manual reductionof the maximum thrust at take-off. This is usually referred to as take-off derate. Derating theengine during take-off is in the discretion of the pilot. It ranges between 0-20%, and typicallyfalls between 10-15% [Air04c]. Derating is used when the take-off weight is below the maximumtake-off weight of the aircraft, a long runway is available or the ambient temperatures duringtake-off are relatively low [Air06c]. The result is a lower EGT at take-off and thus a reducedrate of engine deterioration and prolonged time on-wing. In general, engines that operate onshort average flight times benefit more from take-off derate than those operated on long-haulroutes. Also, it is generally that the first 5% of derate have a bigger impact in terms of reducingthe operational severity than following derate steps of 5% [Air06d].

Since both the average flight time and the level of derate heavily affect the engine’s deterioration,engine manufacturers develop severity curves to illustrate the combined influence of both param-eters on the severity of an engine’s operation [Air06d]. These curves are collected from statisticaldistributions and allow the operators to perform benchmarking and sensitivity studies in orderto achieve the lowest cost of ownership. Severity curves are often developed separately for eachengine variant of an engine model [Ack10]. Figure 2.14 illustrates an example severity curve.


OAT [C°]

EGT [C°]

EGT Margin

Cruise

Corner Point SLOATL

Take-Off EGT

OAT [C°]









New Engine

Deteriorated Engine

EGT Margin


EGTM Erosion [C°]

Installation

Loss

Restoration

in Shop Visit

Engine Time On-Wing

$/ESV

$/EFH

Take-off & Climb

Redline EGT

Descent & Landing

Take-off & Climb

Descen

t & Landing

Cruise

1 flight hour




3 flight hours

SVR & DMC [$/EFH]

EFH:EFC [h]short

cycle time

long

cycle time

Severity Factor

EFH:EFC [h]

5%

10%

15%

1.0 1.5 2.0 2.5 3.0

1.0

0.8

1.2

1.4

1.6


Derate Base Point

Figure 2.14: Example severity curve

Each of the multiple curves in the graph represents one derate level as stated at the right.These multiple severity curves are often also expressed in form of a matrix. The output of aseverity curve is a certain severity factor (SF) , that is used to adjust the SVR or the maintenancecost according to the operational severity. The following calculation based on the severity curveillustrated above will demonstrate this concept.

An example engine is operated at an average flight time of EFC:EFH = 2 and an averagederate of 5%. Under this base conditions the SVR is known to be 0.05 [SVs/1000EFH] andthe maintenance cost per flight hour are 100 [USD/EFH]. To estimate how the SVR and themaintenance cost change when the same engine is operated at a flight time of EFC:EFH = 3.0and a derate of 10%, the operator can determine the severity factor for these two parametersand subsequently adjust the base cost and SVR.

Base engine severity: EFH : EFC = 2.0 [h], Derate = 5%

Base data → SV R = 0.05 SV s

1000 EFHand Cost = 100 USD

EFH

Adjustment to deviating severity: EFH : EFC = 3.0 [h], Derate = 10%

1. Determination of severity factor via severity curve → SF = 0.8

2. Multiplication of SF with base SVR and Cost

results in → SV R = 0.8 · 0.05 = 0.04 SV s

1000 EFHand Cost = 0.8 · 100 = 80 USD

EFH

Therefore the operator can expect 20% reduced SVR and cost, if the engine is operated underless severe conditions as proposed in the example.

Outside Air Temperature As demonstrated in figure 2.6, the EGT during take-off is directlyinfluenced by the ambient air temperatures. In order to prevent the engine from operating atEGTs that could result in severe damage, the digital engine control keeps the EGT and EGTmargin constant at all OATs above the corner point temperature, by reducing the engine’s thrust.


However, at OATs below the corner point, the thrust is kept constant and the available EGTmargin increases as the OAT decreases [Air07b]. In other words, low ambient air temperaturesresult in low gas path temperatures, which reduces the thermal stress on the engine’s hardwareand thus prolongs engine time on-wing.

Environment Also contributing to the severity of an engines operation are environmentalconditions. Particulate matter that results from air pollution, such as dust, sand, industryemissions or volcanic ash can erode compressor and turbine blades as well as block coolingholes. Salty environments in coastal areas will accelerate corrosion and oxidation of the enginecomponents [Jet08]. These environmental conditions can have a severe impact on the engine’shardware deterioration and thus, on the time on-wing.

2.2.3.3 Engine Age

A general observation from analyzing engine maintenance data is that older engines remainon-wing shorter and cost more to maintain than newer engines. In terms of maintenance, enginescan therefore be distinguished in first-run and mature-run phases. There is no clear definitionwhen an engine’s mature phase starts. Maturity may begin as early as after the first shop visit,depending on the engine model. In general, first-run engines will achieve considerably longertimes on-wing than subsequent runs, as a result of increasing rates of hardware deterioration asthe engine ages. However, once the engine reaches maturity, the shop visits intervals and coststabilize to a relatively steady state [Ack10]. The engine phase also has a significant influence onthe cause of engine removals as previously seen in figure 2.7.

2.2.4 Engine Maintenance Costs

Prior to this section, several references to the influence factors on engine maintenance costs(EMC) have been made. This subsection shall further discuss the EMC as part of the totaloperating costs (TOC) of an aircraft including a cost breakdown structure as it is considered inthe scope of this paper. Aircraft TOC are generally divided into direct operating costs (DOC) andindirect operating costs (IOC) [Ros90]. DOC comprise of all costs, that can be clearly allocatedto the aircraft operation including fuel costs, crew costs and maintenance, while IOC consist of allgeneral costs indirectly related to the aircraft operation, like costs for planning and organizationas well as marketing and ticket sales. A more detailed cost breakdown structure of aircraft TOCcan be found in [Pet08]. Engine maintenance costs are generally considered to be part of thedirect operating costs. Modeling the DOC of an aircraft has historically been an important toolto evaluate the economics of an aircraft design. Thus, there are various different DOC methodsthat include EMC as part of the total maintenance costs. Many DOC methods are based onthe ATA 1967 DOC method [Air67]. An example is the DOC method after Roskam [Ros90],where the EMC per flight hour comprise three components: labour, material and maintenanceburden costs. The division of maintenance cost in labour- and material costs can be found inmost cost estimation publications reviewed in the scope of this project. The maintenance burden


are allowances to reflect general charges for facilities, spare engines and engine leasing, trainingof the staff or general engineering and administrative services related to engine maintenance.Therefore, some DOC methods consider these charges as indirect operating costs and do notinclude maintenance burden, as in [Sch98] or [AEG00]. In this context, engine maintenance costsare often also divided into direct maintenance cost (DMC) and indirect maintenance costs (IMC).The focus lies in general on the estimation of the DMC since they are the more meaningfulbenchmark for comparing two engine designs and they are directly influenced by the aircraftoperation [CFM09]. Section 2.3.1 analyzes the reflection of EMC within common DOC modelsin more detail.

2.2.4.1 Cost Breakdown Structure of Engine Maintenance

The cost breakdown is the foundation of a cost estimating model. The objective is to provide areliable structure that includes all elements the cost estimate will cover [NAS08]. The followingstructure is based on the DMC breakdown in Ackert [Ack10], expanded to also include indirectmaintenance costs.

Engine DMCEngine DMCEngine DMCEngine DMC

Shop Visit DMCShop Visit DMCShop Visit DMCShop Visit DMC

• Labour:Labour:Labour:Labour:assembly/disassemblycleaning, inspection,

• Material: Material: Material: Material: replacementof parts and material

• Repair of partsRepair of partsRepair of partsRepair of parts

• Fees, testing, Fees, testing, Fees, testing, Fees, testing, logisticslogisticslogisticslogistics

Line DMCLine DMCLine DMCLine DMC

• Line Labour:Line Labour:Line Labour:Line Labour:Line inspectionTroubleshootingLRU replacement

• Line Material:Line Material:Line Material:Line Material:consumables

Input of Shop Visit DMCInput of Shop Visit DMCInput of Shop Visit DMCInput of Shop Visit DMC

Engine Maintenance CostsEngine Maintenance CostsEngine Maintenance CostsEngine Maintenance Costs

• Providing Facilities:Providing Facilities:Providing Facilities:Providing Facilities:offices, workshops

• Staff member trainingStaff member trainingStaff member trainingStaff member training

• Administration:Administration:Administration:Administration:planning, engineering

Engine IMCEngine IMCEngine IMCEngine IMC

Maintenance BurdenMaintenance BurdenMaintenance BurdenMaintenance Burden

Spare EnginesSpare EnginesSpare EnginesSpare Engines

• access to spare engines,leased or owned

Folie 5


Shop Visit DMC [$/EFH] = Shop Visit Cost (SVC)/TOW

SVC [$] = Restoration Cost + LLP Cost

Input of Shop Visit DMCInput of Shop Visit DMCInput of Shop Visit DMCInput of Shop Visit DMC

Figure 2.15: Engine maintenance cost breakdown structure

While engine line maintenance incurs costs continuously as it is performed in relatively shortintervals on the flight line, shop maintenance takes place after comparable long intervals thuscauseing costs only during shop visits. Due to the different character of line maintenance andshop visit, they are reflected separately within the engine DMC. The engine line maintenanceis included in the aircraft maintenance planning document (MPD). Thus, it has already beenimplemented into the existing LCC-tool. The IMC include maintenance burden also known as

2.3 Modeling of Engine Maintenance Cost 26

overhead costs and charges for spare engines.

Figure 2.15 also illustrates an alternative way of accounting for the shop visit DMC. In contrastto the traditional separation into material and labour costs, it is also common to divide theshop visit costs (SVC) into restoration costs and LLP costs [Ack10][Bec09]. Restoration costsare charges for labour and material related to restoring the engine’s performance, while LLPcosts reflect expenditures for the LLP replacement. The shop visit DMC per flight hour arethen calculated by dividing the SVC by the mean time on-wing. This cost breakdown correlateswith the concept of adjusting maintenance cost via severity factors. It splits up the shop visitcost in one component that depends on the severity and one that is mainly independent fromthe operational conditions. Only the restoration cost are escalated according to the operationalseverity. The LLPs are replaced after a hard time independently from the severity of the flightconditions.

2.2.4.2 Common forms of engine maintenance contracts

Whereas many airlines historically performed the overhaul of their engines in their own workshops,today the engine overhauls of most airlines are contracted to external MRO service providers.The commissioning of engine overhauls to external shops is based on contracts that containall services to be performed. There are generally three basic types of payment methods thatare typically arranged in the contracts [Rup00]. In the so called Time and Material contracts,the MRO provider issues for each engine a detailed invoice with the required man hours andmaterials. The customer pays the bill according to the arranged labour rates and material costs.In return, the contracted shop guarantees a certain minimum subsequent time on-wing of theengine. In contrast to this are Fly-By-Hour arrangements, where the airline pays a fixed amountof money per engine flight hour to the MRO provider. The contracted shop has to finance allcoming shop visits from the received advance payment. This results in a good predictabilityof the maintenance cost for the airlines. The third payment method is based on Fixed Pricesfor certain shop services to be performed. The arranged workscopes vary from full overhauls tolimited overhauls on certain modules. There are also diverse hybrid forms possible. The airlinesgenerally try to arrange a customized contract that fits the requirements of the airline and resultsin low cost and reliable cost forecasting.

2.3 Modeling of Engine Maintenance Cost

An overview of concepts for modeling the maintenance costs of aircraft engines is given inthe following. Starting with a summary of the reflection of EMC in common DOC methods.Subsequently the basics of cost estimation with focus on parametric cost modeling are alsoreviewed as background for the development for cost estimating relationships (CERs).


2.3.1 Reflection of EMC in DOC methods

The analysis of the consideration of the EMC within different DOC methods is a first referenceon how engine maintenance costs maybe modeled. Therefore, table 2.2 summarizes the reflectionof engine maintenance costs in different DOC models available. For a more explicit comparisonof several DOC methods in general it is referred to Mildt [Car00].

DOCMethod Cost Breakdown Labour C lab Material C mat Input IMC

Roskam[Ros90]

C EMC = C lab +C mat + C burden

C lab =[(0.718 + 0.0317 · TT O

1000

)·

1100MT BR

+ 0.10]

· labRmhr

C mat =[

5.43 · 10−5 · EP ·

Esppf − 0.47]

· 1KHem

TT O : take-off thrust,labRmhr : labour rate,MTBR, EP: engineprice, Esppf: sparepart price factor,Khem:MTBR factor

3

AAE[AAE04]

C EMC = C lab +C mat + C burden

C lab =[

0.645 ·( 0.05·TT O

104

)·(

0.566 + 0.434F T

)]· F T ·

labRmhr

C mat =[

25 ·( 0.05·TT O

104

)·(

0.62 + 0.38F T

)]· F T

TT O : take-off thrust,labRmhr : labour rate,FT: flight time

3

AEG[AEG00] C EMC = C lab+C mat

C lab =3.26 · 10−5 · TT O · labRmhr

C mat =[(3.63 + 0.91 · TT O

1000

)· K2 + 5.07

]· ECM

TT O : take-off thrust,labRmhr : labour rate,K2: material coeffi-cient, ECM: economicefficient

7

LH[TB07]

C EMC =C LM,lab + C LM,mat +C SV,lab + C SV,mat +

C LLP + C NAC

C LM,lab =(a + b

F T+ c

F Tp,day

)·

labRmhr · Elab

C SV,lab = a · fetops ·(T b

T O · BP Rc · (year − 1970)d)

·labRmhr · Elab · SF

C LM,Mat =(a + b · TT O + c · T 2

T O

)·

Emat

C SV,mat = a · fetops ·(T b

T O · BP Rc · (year − 1970)d)

·Emat · SF

TT O : take-off thrust,labRmhr : labour rate,a,b,c,d:regression co-efficients, Elab,mat:cost escalation factor,fetops: cost factor,SF: severity factor,FT:flight time

7

Table 2.2: Comparison of EMC consideration in different DOC methods

The structure of the DOC methods is quite similar. All consist of cost components for labourand material adjusted by several regression coefficients and cost factors. However, the LH methodsticks clearly out, due to a higher level of cost breakdown and more complex estimation functions.It is the only method that considers the costs for line maintenance (LM) and shop visit (SV)maintenance separately. In addition, there are explicit cost estimation functions for the LLPs(CLLP ) and the engine nacelle (CNAC). All models include the take-off thrust as major inputparameter, while the LH method also considers severity factors for engine de-rate and averageflight time and even the Bypass-Ratio (BPR) is reflected. The factor (year − 1970)d models theinfluence of the age of the engine design.

2.3.2 Parametric Cost Estimation

Parametric cost estimation is based on historical data and mathematical expressions, that relatecost as a dependent variable to selected cost-driving independent variables. The result are socalled cost estimating relationships (CERs), which are defined as:

Cost Estimating Relationships (CERs) are mathematical expressions relating cost asthe dependent variable to one or more independent cost driving variables. [Bru96]


A typical CER for instance is estimating the manufacturing cost of a product by using theproduct weight [Lon00]. The implicit assumption is that the future cost of a product are affectedby the same forces that affected the cost in the past. This approach is generally applied, whenonly a few key parameters are known. Therefore, it suits the present problem of estimatingthe maintenance cost of aircraft engines with limited access to detailed primary data sources.However, the key parameters of an engine, such as weight and thrust, as well as to a certainextend the overall maintenance costs are openly available and can be utilized in a parametriccost analysis. A major advantage is that once the CERs are established, the cost estimates canbe conducted quickly and easily replicated [NAS08], which is necessary for an implementationinto a variable software tool. The problem is that the collection of the necessary data and thesubsequent determination of CERs is a complex and time consuming process. The scope ofthis study, however, is limited on engine maintenance, which makes it possible to apply theparametric estimation approach. The NASA cost estimating handbook [NAS08] provides furtherresources on the applicability of different cost estimating methods including a summary of theirpros and cons. Figure 2.16 shows the methodological procedure of parametric cost estimating.This procedure is the foundation of the data analysis in chapter 3. Therefore, the key stages arebriefly discussed in the subsequent paragraphs.

Co

st E

stim

atin

gC

os

t Ris

kE

co

no

mic

& S

up

po

rting

An

aly

sis

Care

er D

ev

elo

pm

en

t Gu

ide

Refe

ren

ce

Kn

ow

led

ge M

an

ag

em

en

tC

ost E

stim

atin

gC

os

t Ris

kE

co

no

mic

& S

up

po

rting

An

aly

sis

Care

er D

ev

elo

pm

en

t Gu

ide

Refe

ren

ce

Kn

ow

led

ge M

an

ag

em

en

t

2008 NASA Cost Estimating Handbook Section 4. Cost Estimating Process

Volume 1 Page 1-27

4.2.2 Task 5: Select Cost Estimating Methodology

The goal of this task is to select the best cost estimating methodology (or combination of

methodologies) for the data available to develop the most accurate cost estimate possible.

Based upon the phase that the project/system is entering and the data available to conduct the

estimate, follow the quick reference chart shown in Table 1-3 to select the cost estimating

methodology (or methodologies).

Table 1-3. Cost Estimating Methodology Selection Chart

Pre-Phase A Phase A Phase B Phase C/D Phase E

Parametric 4 4 2 2 0

Analogy 4 2 2 2 0

Engineering Build Up 2 2 4 4 4

Legend: 4 Primary 2 Applicable 0 Not Applicable

Parametric Cost Estimating

Estimates created using a parametric approach are based on historical data and mathematical

expressions relating cost as the dependent variable to selected, independent, cost-driving

variables through regression analysis. Generally, an estimator selects parametric cost estimating

when only a few key pieces of data are known, such as weight and volume. The implicit

assumption of parametric cost estimating is that the same forces that affected cost in the past will

affect cost in the future. For example, NASA cost estimates are frequently of space systems or

software. The data that relates to estimates of these are weight characteristics and design

complexity respectively. The major advantage of using a parametric methodology is that the

estimate can usually be conducted quickly and be easily replicated. Figure 1-12 shows the steps

associated with parametric cost estimating.

Define Estimating

“Hypothesis”Collect

“Relationship”Data

Evaluate & Normalize

DataAnalyze Data for Candidate Relationships

Perform Statistical

(Regression) Analysis

Test Relationships Select Cost

Estimating Relationship

Figure 1-12. Parametric Cost Estimating Process Steps Figure 2.16: Parametric Cost Estimating process steps [NAS08]

2.3.2.1 Estimating Hypothesis Definition

The objective of defining an estimating hypothesis is to identify potential cost driving variablesand to propose logical cost relationships. This demands a good understanding of the technicalcharacter and the requirements of the examined project. The result is a hypothesis of a forecastingmodel necessary to develop CERs [Bru96].


2.3.2.2 Data Collection and Evaluation

The assembly of a database is essential when deriving cost estimating relationships. A lack ofvalid CERs is often the result of an inappropriate database [Bru96]. The first step in buildingup a good database is the data collection. There are generally two types of data, cost andnon-cost data. Non-cost data includes technical information coming from drawings, specifications,certification documents or direct measurement as well as schedule and programmatic informationthat can be obtained from operations departments [BJ82]. Cost data comprises labour hoursor direct cost information extracted from accounting systems or through interviews. The datatypically comes from many different sources. It is important that the estimator judges the qualityof the data origin and identifies the best source [Gal08]. Data can be obtained from internalsources, such as accounting or workshop databases, programme recaps or engineering notes,as well as from external sources like professional articles or public record informations. Thedisadvantage of external sources is that the user has no knowledge of the procedures used tocollect and process the data. It is further distinguished between primary data that is directlyobtained from the original source, and secondary data which is derived and possibly “sanitized”from primary data. Hence, primary data is generally considered best in quality and reliability[Gal08].

2.3.2.3 Data Normalization

When establishing a database, it is often discovered that the collected raw data turns out tobe irregular and inconsistent or partly in the wrong format for analytical purposes. Therefore,adjustments to the raw data have to be made to ensure a comparable and consistent database[BJ82]. For instance the normalization of raw data adjusts inconsistencies in currencies, mea-surement units and the scope of the data. Historical data should furthermore be adjusted foranomalies, improvement in technology as well as inflation. Any kind of adjustment or judgmentsused in processing historical data should be fully documented. The data collection, evaluationand normalization is a fundamental step in generating a parametric cost estimating model. Thus,a considerable amount of time is devoted to assembling a database [Bru96].

2.3.2.4 Data Analysis

The first stage of the data analysis is screening the database for candidate relationships betweenthe dependent and independent variables. This process is built on the hypothesis. However, theremay be additional relationships that were not foreseen during establishing hypothesis. Once thecandidate relationships have been established, one can perform a regression analysis to modelthe CERs. The objective of regression analysis is to determine the parameters for the functionthat fits the set of data best. The data is fit using techniques like:

• Linear Regression: unknown parameters are estimated from the data using linear functions• Nonlinear Regression: applied for data that is not essentially linear


For CERs, the dependent variable is always the cost to be estimated and the independent variablewill be the cost driver [NAS08]. The dependent variable responds to changes of the cost driveraccording to the determined regression function. An example would be the hypothesis thatthe cost of a product development maybe driven by the weight of the final product. With thisassumption one could plot the historical data on cost over weight, with the possible result of thechart in figure 2.17.

weight

cost

r(x) = ax + b

data

linear regression

Figure 2.17: Example data points for cost-weight dependency

In this case, a linear regression has already been performed with the aim to fit a straight lineto the data points. The result is an equation that describes the line, expressed by r(x) = ax + b.In this CER, x represents the weight and r(x) equals the estimated costs. Often, there aremore than one independent variable, that have an effect on the cost. Multivariate regression iscapable of observing and analyzing the effect of multiple independent variables on the dependentvariable, through the addition of possible explanatory coefficients. Usually a computer softwareis used to assist in determining the regression coefficients. For a closer look on the mathematicalbackground of the different regression methods, in context with parametric cost modeling it isreferred to the parametric cost estimating handbook from the US department of defense [Bru96].

2.3.2.5 Testing the Relationship Results

After the determination of a CER through regression analysis, it is crucial to evaluate and testthe regression results. Therefore, it is necessary to have a look on more than just one criteria[Chu08]. Only the consideration of a multitude of factors will give the whole picture of the qualityof the CER. Table 2.3 summarizes the key criteria that should be evaluated when reviewing thequality of a CER. As for the regression analysis, a computer software is widely used to conduct aquick and reliable determination of the statistical criteria. For further information on the theoryof probability and statistics, it is referred to the statistics ebook from the UCLA [UCL].


Symbol Description Reference Evaluation/Test

X,Y data observationCheck data range, number ofdata points and especially out-liers

are there enough data points for a represen-tative CER? can the outliers be explained orcorrected?

R2 coefficient ofdetermination

measure for the strength of therelationship

ranges from 0-1, while 1 as the maximum rep-resents the best overall fit of the model to thedata

R2adjR2 adjusted for

degrees of freedom

measure for the contributionof additional explanatory coef-ficients

similar to R2, however it only increases if anadded explanatory coefficient considerably im-proves the model

RMSEroot mean square

errormeasures the accuracy of therelationship

check if the actual mean deviation of the datapoints to the model is acceptable

p probability value measure for the significance ofthe hypothesis

the lower the p-value, the more significant isthe hypothesis, a hypothesis is generally re-jected when the p-value is higher than the sig-nificance level α, which is often equal to 0.05

t t-ratiomeasure for the validity ofadding a particular cost drivervariable

generally, a t-ratio above 2 leads to the accep-tance of the hypothesis, that a cost term addspredictive value to the CER

Table 2.3: Criteria for the evaluation of regression results

3 Development of Cost Estimating Relationships

The purpose of this chapter is to document the process of developing the cost estimatingrelationships needed for the engine maintenance model. The objective is to generate cost-to-non-cost CERs for the two dependent variables:

• Shop Visit Cost per Flight Hour• Shop Visit Interval

The development procedure follows the methodological approach described in the previous chapterunder 2.3.2. This approach is not a fixed single sequence of steps to be conducted. Rather,it is as an iteration loop as illustrated in figure 2.16 on page 28. The outcome of each step isevaluated to determine if the next step can follow or if one has to go back a few stages to startall over again.

3.1 Database Assembly

As mentioned before, the database (DB) assembly is crucial for the success in developing CERs.Therefore, most time was spent on collecting and processing the data. The following describesthe proposal of the hypothesis as a starting point for the data collection. Subsequently, differentdata sources are reviewed and the process of data normalization is reported.

3.1.1 Establishing the Hypothesis

As a result of the literature review, the technical background regarding aircraft engine maintenancehas been worked out. From this information, the cost estimating hypothesis can be derived. Bothcost and interval length are heavily influenced by operating parameters, such as average flighttime, derate, OAT and environmental conditions. However these parameters are no independentvariables that can be directly allocated to the engine built. Since these operational parametersstill have a great impact on the dependent variables, they are subjected to a normalization ofthe collected data. Another influence factor is the thrust rating of an engine. The thrust ratingcan be represented by its thrust-weight ratio (TWR) . The weight1 reflects the constant builtof an engine model. The more thrust2 is generated from the hardware, the higher the TWR,which leads to higher shop visit rate and also higher cost within one engine model range. Theassumption is made that the TWR can also serve as variable to compare the thrust ratings of two

1 within the framework of this study, the “weight” as engine specification always means the dry weight2 “thrust” as engine specification always refers to the take-off thrust

32

3.1 Database Assembly 33

different engine models. Therefore, the TWR is introduced as first independent variable. Fromthe DOC methods, it can be derived that the engine take-off thrust TTO is also an importantcost driver. In addition, it is assumed that the engine weight as general measure for the enginesize is directly related particularly to the engine maintenance costs. Table 3.1 sums up thesethree independent variables and the corresponding proposed relationships.

Summary of CER Hypothesis

TWR higher TWR ⇒ higher SVR and cost

Take-off Thrust higher thrust ⇒ higher cost

Dry Weight higher weight ⇒ higher cost

Table 3.1: Summary of cost estimating relationship hypothesis

3.1.2 Review of Data Sources

With the hypothesis established, available data sources are reviewed to find the necessary datafor the proposed relationships. The review of the data sources is separated into technical- andcost data sources.

3.1.2.1 Technical Data

Standard technical specifications of aircraft engines, like take-off thrust and dry weight, aregenerally no sensitive data. Hence, they can be be obtained directly from the engine OEMs(website or specification sheets). This is a primary source and it can be considered as very reliable.It is also possible to find engine specifications in public databases. This has the advantage thatthe data for a wide range of engine models and variants is concentrated in one single source. Twosuch single sources are the Database Handbook for Turbofan and Turbojet Engines from ÉlodieRoux [Rou07] and the Jet Engine Specification Database from Nathan Meier [Mei05]. Eventhough these databases are strictly speaking no primary sources, it can be assumed that thedata is still reliable, since it is simply a summary of the primary source without any deviation.This assumption was also confirmed by a few random comparison checks. Both sources provide ahuge database for a wide range of engine models and variants and their specifications, includingspecifications beyond the engine dry weight and take-off thrust.

3.1.2.2 Cost and Interval Data

In contrast to technical engine data, cost and removal interval information are highly sensitiveand well protected by the MRO providers and airlines. Hence, it was not possible to makeprimary sources accessible. However, there is a range of secondary data sources, that wereavailable in the framework of this study. These sources are briefly described in the followingparagraphs:


Form 41 Databases The US department of transportation1 maintains databases for aircrafttraffic, capacity data and other operational data for air carriers operating to and from the UnitedStates. The database includes monthly data of engine maintenance containing labour, repair andmaterial costs. However, all costs are given only for the different aircraft models. There is noindication of the engine model version. Therefore, this database is not adequate for collectingdata for specific engine models and variants [Bec09].

MRO Prospector Aviation Week is a weekly magazine reporting on the aerospace industry. Partof their portfolio is the MRO Prospector2, a online database for fleet data and contract details.In contrast to the Form 41 database, it provides comprehensive tables with cost data for a widerange of specific engine models. The problem is that, there is no indication of the engine variant.However, the thrust rating of one engine model can vary considerably. Since thrust and TWRare key independent variables, it is necessary to have more detailed maintenance cost informationon engine variant level. Furthermore, there are no informations given about the operationalconditions the data is based.

Aircraft Commerce Archive Aircraft Commerce3 is an aviation magazine published every twomonths. It provides intelligence on fleet planning, maintenance costs and aircraft leasing forthe commercial aircraft industry. In regular intervals, it publishes detailed operator & ownerguides dedicated to specific engine models. These articles give comprehensive information aboutthe engine’s shop visit planning, removal causes, hardware degradation, LLP management andthe influence of the operational severity. They also summarize data about shop visit intervalsand costs in clearly represented tables. In appendix B.1 an example table from the magazine isillustrated. Usually, there are distinct information about each variant within an engine modelrange. The maintenance costs are generally expressed in estimated reserves per EFH or EFC, inwhich the reserves are distinguished between restoration reserves and LLP reserves. It is alsoindicated how the reserves change as the engine ages, by showing distinct reserves for first, secondand third or mature shop visits. The articles are fairly consistent in their structure throughoutthe years. The magazine maintains an online archive with articles of the past ten years. Thisenables to collect and summarize the data.The aircraft commerce archive is clearly the best available source for building up the database,since it provides information for specific engine variants including indications about the operatingconditions. The disadvantage is that all the articles have to be collected and particularly readin order to get all necessary information. Another issue is the fact that it is mainly unknownhow exactly the data was collected and the reserves estimates were established. On request,the editorial office of the magazine stated, that the data is gained directly from maintenancefacilities. Also, it was possible to clarify further questions about the data collection through

1 http://www.bts.gov/data_and_statistics/2 http://mrop.aviationweek.com/3 http://www.aircraft-commerce.com/


direct correspondence with the editorial staff of Aircraft Commerce. Therefore, it is consideredan adequate secondary source for the database assembly.

AeroStrategy AeroStrategy1 is an aerospace consulting firm providing strategic consultingservices to aerospace clients. Amongst others it provides market estimates on engine SV intervalsand SV restoration costs. Costs for LLPs are unfortunately not included. The respective datatables are not to be disclosed in public. Therefore, they are not displayed in the framework ofthis study. The AeroStrategy data distinguishes between first and mature removals and it alsoshows distinct estimates on certain engine variants depending on the aircraft they are equippedto. Even though it is not clearly stated what flight conditions the estimates are based on, theinformation on which aircraft the engines are applied to enables assumptions on the averageflight time the estimation relates to. In sum, the AeroStrategy data tables are less appropriatefor the database assembly. However, they form an adequate independent data source for testingthe plausibility of the targeted model.

3.1.3 Data Collection

The collection of the raw data was done in one single excel table. The cost and interval datafrom the Aircraft Commerce Articles (ACA) was collected first, since it is only available forcertain engine models and variants. This data was separated in first-run and mature-run data.Thus, it was assumed that each engine reaches maturity after the first shop visit. This is a validassumption, since the available data indicated that the cost and intervals reached a fairly steadylevel already after the first removal. For each engine it was noted what year and month therespective articles were published. This is necessary for a subsequent normalization of inflation.In addition, the average flight time, on which the cost and interval estimation of the engines isbased on, was collected. This made it possible to collect the interval data as both, EFC and EFHwith the average flight time as conversion factor. Direct information about the environmentalconditions and the derate could not be extracted. However, from studying the articles, theassumption was made that the average standard derate for estimating the cost and interval dataequals 10 %. Since the articles also provide detailed information about the LLP managementand the EGT degradation rates of the engines it was considered to collect this data as well. Theidea was that especially the LLP cost, LLP lives and number of LLPs in an engine may have adetectable relationship with the dependent variables. However, it turned out that the articles donot report this information consistently. As a result of this, the focus laid on the collection ofthe cost and interval data. The LLP reserves were generally collected in USD/EFC, whereasthe restoration reserves were collected in USD/EFH. After the collection of the data from theACA, the rows were filled up with the respective engine specifications. The collection of thetechnical data was extended to additionally include available engine specifications like pressureratio, BPR, fan diameter, engine length and the number of stages in each turbine and compressor.

1 http://www.aerostrategy.com/


The objective was to determine if one of these additional independent variables can significantlycontribute to the CER development. Table 3.2 displays the basic structure of the colltected rawdata table.

Engine Specifiations First Removal Mature RemovalThrust[lbf]

Weight[lb]

EFC:EFH[h] ... Interval

[EFC]Interval[EFH]

RestorationReserves

LLPReserves

Interval[EFC]

Interval[EFH]

RestorationReserves

LLPReserves

. . . . . . . . . . . .

. . . . . . . . . . . .

Table 3.2: Structure of the collected Data

From the previous chapter, it is known that engines designed for short-haul (SH) aircraft havedifferent maintenance characteristics than medium-long-haul (MLH) operating engines. Becauseof this, the DB has been arranged in a way that both engine types are listed in separate groups.This enables both the combined and separate analysis of the two engine types. Appendix B.2explains what considerations led to the classification of the database engines into SH and MLH.

3.1.4 Data Normalization

General data inconsistencies caused by the varying presentation of the information in the ACAwere adjusted simultaneously with the collection of the data. However, adjustments of the datafor instance due, to inflation and operational severity have to be done subsequently to the datacollection.

3.1.4.1 Inflation

The reviewed ACA were published over a time span of eight years, which makes fluctuating labourrates and material prices an issue when comparing cost data. Thus, the cost data is normalizedby adjusting it through inflation factors. In general, material and repair & replacement coststend to exhibit a higher price fluctuation. This is mainly because of the increasing applicationof more advanced and expensive materials and the generally greater imbalances in supply anddemand for these materials. To account for this, two separate economic indices correlating toboth labour and material are utilized to determine the overall inflation factor for each engine.These indices are the Employment Cost Index (ECI) for aircraft manufacturing wages & salariesand the Producer Price Index (PPI) for industrial commodities. The proportion of total enginemaintenance costs is in the order of 35% labour and 65% material [Ack10]. The escalationyear and month were set to May 2010. The economic indices for the escalation month and therespective base month of the engine data were obtained from the website of the US Bureau ofLabor Statistics1 (BLS) . The method used to calcualte the maintenance inflation factor (MIF)is expressed in the following formula:

1 http://data.bls.gov/cgi-bin/srgate


MIF = 0.35 · ECIescECIbase

+ 0.65 · PPIescPPIbase

ECIesc = mean ECI of the 3 months prior to escalation month

ECIbase = mean ECI of the 3 months prior to base month of collected engine cost data

PPIesc = mean PPI of the 3 months prior to escalation month

PPIbase = mean PPI of the 3 months prior to base month of collected engine cost data

To balance short term fluctuation, each index is averaged over the three months prior to theactual month. For each engine of the DB, the corresponding MIF was calculated and the LLPcost and restoration costs were adjusted accordingly.

3.1.4.2 Flight Time

As discussed in 2.2.3.2, both the shop visit DMC and the SVR/mean TOW are influenced bythe average flight time. This effect has to be normalized, if possible. The database contains theaverage flight time for all cost and interval estimates. The objective is to normalize this datato a standard flight time level. Theoretically, this is possible if for each engine of the DB thecorresponding severity curve was available (see figure 2.14). In this case, one could predefine astandard flight time and calculate for all data points the severity factor that would adjust the costand interval data to the level of the standard flight time. The problem is that, each engine modeland even each engine variant has a distinct severity curve. These curves are sensitive informationthat could not be obtained from the engine manufacturers. However, it was possible to getan example curve for a short-haul operating engine (A320) as well as for a medium-long-haulaircraft engine (Boeing 777). Together with the scattered information on severity factors fromthe ACA, it was succeeded in assembling averaged severity curves for both SH engines and forMLH engines, based on the two example curves. The assumption is made that the entire rangeof distinct SH severity curves can be adequately approximated by one averaged severity curve. Itis assumed that the same applies for the group of MLH aircraft engines. This is an assumptionmade by the author of this thesis. It results from general observations made while studyingthe aircraft commerce archive. The two determined average severity curves are subsequentlydisplayed in shape of a table for a derate of 10%.

EFH:EFC 0.5 1.0 1.5 1.9 2.5 3.0 4.0 5.0 6.0Severity Factor 2.40 1.75 1.30 1.00 0.86 0.78 0.706 0.66 0.63

Table 3.3: Determined average SH severity curve for a derate of 10%

EFH:EFC 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0Severity Factor 2.20 1.70 1.40 1.23 1.08 1.00 0.93 0.88 0.86 0.84 0.82 0.80

Table 3.4: Determined average MLH severity curve for a derate of 10%

Since the LLP replacement is mainly independent from the operational severity and the LLP


reserves were collected in USD/EFC, they were excluded from the flight time normalization.Therefore, only the restoration cost reserves and removal intervals of the database were adjustedwith these severity factors. Both engine categories have been escalated to the base flight time ofthe respective average severity curve (Severity Factor = 1.0). That means the SH engines werenormalized to a flight time of 1.9, whereas the MLH engines were normalized to a flight time of6.0.

3.1.4.3 LLP Reserves

The deviation of the LLP reserves between the first shop visit and subsequent ones is minor. Inaddition, there is no trend detectable. Therefore, the LLP reserves of all available data havebeen averaged for each engine. This average LLP reserves serve as basis for the cost analysis.That means the LLP reserves are not divided into first-run and mature-run data.

3.1.4.4 Remaining Anomalies

According to the previous chapter, there are several other effects which have a significant impacton the maintenance cost and the time on-wing. Ideally, these effects would be also normalized toa standard level. The following summarizes these effects and discusses how they were consideredwhile assembling the database.

Derate The ACA do not clearly state on what derate the estimates are based on. However,there are often information about the average derate the engines are operated with. Most enginesoperate on average with a derate of 10%. Thus, it is assumed that all estimations were based onthis derate. Therefore an adjustment of the data according to the derate level does not apply.

Environment and OAT Since the information about the environment and the OAT are veryscarce, it was not possible to utilize this as foundation for a data normalization. In general, theestimates reflect a worldwide average and thus it is assumed that all values were establishedbased on a temperate environment (TE) .

Engine Age The effects of the engine age are already included in the database. The aircraftcommerce guides publish estimated cost reserves and intervals for first and mature shop visits.This data is represented separately in the database. Therefore it is possible to do a distinctanalysis of first SVs and mature SVs. As mentioned before, this is based on the assumption thatthe engines reach maturity in terms of maintenance after the first shop visit. This may varyespecially for engines that have only a few shop visits during their life cycle. However, for theseengines, it is also applicable that the first interval is at most times considerably longer thansubsequent ones.

Improving Technology and Learning Curve When building up a database from historic data, it isalso an issue to consider effects from improved technology and developing know-how. Over the


past two decades, the trend in engine maintenance went to longer intervals and thus reducedmaintenance costs per EFH. This was a result of improvements throughout the maintenanceprocess such as better materials, ECM, on-wing repairs and also design for durability. Theseeffects were clearly visible in the reviewed data sources. However, it was not possible to collectenough data from older engine models for establishing reliable escalation factors. Therefore, thedatabase only contains maintenance data for newer generation engine designs based on similartechnology1. However, neither does the DB contain engines from the latest generation, sincemost of these engines have not even been through their first shop visit and thus there is nohistoric maintenance data available.For normalizing learning curve effects, it applies the same as for the technology improvement. Fora few engines, there is data accessible indicating that an early engine built has raised SVRs andcosts compared to later revisions of the same engine model. This is due to improvements in bothdesign and maintenance as the engine model is in service. However, since available informationwas limited, a normalization could not be performed. In order to make sure that these twoeffects do not influence the data analysis, data of older engine generations and data from newlymarketed engines was marked and excluded from the general data analysis. However, the existingdata can be utilized as input for a determination of technology factors for the maintenance model.

Number of Spools In section 2.1.2, it was discussed that the spool configuration of an engineinfluences the achievable shop visit intervals. The database contains only six engines with athree-spool configuration, all from the same OEM. As with the learning curve and the improvingtechnology, it was not possible to collect enough information to adjust the intervals to the level oftwo spool engines. However, the accessible data confirmed this effect. Therefore, the three-shaftengines were also marked and excluded in the interval analysis.

3.1.5 Summary

A lot of findings and decisions made while assembling the database were the result of first dataanalysis procedures, which were necessary to evaluate the adequacy of the DB. This process isnot reported into detail, in order to keep this documentation clearly presented. The result of thedatabase assembly is the foundation for the subsequent extensive data analysis. However, onehas to keep in mind that the collected data is based on only one single secondary external source.As a result of the inconsistency of the presentation of the shop visit data in the ACA and the factthat the exact manner the data was collected is unknown, a lot of assumptions had to be madeto fit the data to the defined database structure. The DB is divided column by column in enginespecifications, first-run, mature-run and normalized intervals and cost reserves. The rows arefurthermore split after the engine type. Short-haul and medium-long-haul operating engines arerepresented separately. The same applies for three-spool engines and older generation engines.Noteworthy is also that some engine variants appear repeatedly with varying average flight times.

1 most engines of the DB entered the market between 1990-2000

3.2 Data Analysis 40

In appendix B.3, the assembled database is displayed with reduced number of columns. Sinceseveral influential effects have been normalized during the DB assembly, it is important to reportthe conditions the final DB is based on. Table 3.5 sums up these base conditions for short-hauland medium-long-haul engines.

Parameter SH MLHEFH:EFC 1.9 6.0Derate 10% 10%No.Spools 2 2Environm. TE TE

Table 3.5: Summary of DB base conditions

3.2 Data Analysis

The data analysis is divided into determination of candidate relationships and regression analysisof the found relationships. Both stages were aided by the extensive use of statistical computersoftware1. The dependent variables shop visit interval and shop visit cost per EFH were

Shop Visit Interval Shop Visit Cost1 2 3 4 5

First Interval Mature IntervalFirst

RestorationReserves

MatureRestorationReserves

LLP Reserves

Table 3.6: Summary of the preliminary dependent variables to be analyzed

split according to the structure of the database. The SV intervals consist of intervals for first-and mature removals, while the SV costs consist of restoration reserves for first- and matureremovals and average LLP reserves for all removals. A further differentiation according to theengine type is also evaluated. As seen in table 3.6, the minimum number of CERs is thereforefive.

3.2.1 Candidate Relationship Screening

The first step is to determine candidate relationships. This includes the evaluation of a possiblefurther separation of the dependent variables. The procedure was to do a database screening forall five dependent variables (see table 3.6). The applied statistic software provides a screeningfunction, that assists in finding independent variables that significantly contribute to modelingthe analyzed dependent variable. This enables an interpretation of the proposed CERs from thehypothesis simultaneously to an analysis of the additional engine specifications. Figure illustratesthe results of such a data screening.

1 JMP 8.0 from SAS

3.2 Data Analysis 41Short_cost_Analysis- Screening of 3rd cost adj 2 Page 1 of 1

thrust

HPC Stages

BPR

OPR

length

weight

LPC Stages

Airflow

SFC

thrust*thrust

thrust*HPC Stages

HPC Stages*HPC Stages

thrust*BPR

HPC Stages*BPR

BPR*BPR

thrust*OPR

HPC Stages*OPR

BPR*OPR

OPR*OPR

Term

78,6617

-7,4830

-9,4701

-13,2651

5,5572

-7,9594

-6,2270

-1,0819

-0,6221

4,2789

-0,1146

4,0165

-1,2031

-0,8808

0,4734

1,0578

-1,2074

-0,8278

-3,4446

Contrast

*

*

*

*

*

*

*

*

*

*

43,43

-4,13

-5,23

-7,32

3,07

-4,39

-3,44

-0,60

-0,34

2,36

-0,06

2,22

-0,66

-0,49

0,26

0,58

-0,67

-0,46

-1,90

Lenth

t-Ratio

<,0001*

0,0043*

0,0015*

0,0004*

0,0169*

0,0032*

0,0110*

0,5760

0,7512

0,0341*

0,9522

0,0428*

0,5332

0,6505

0,8099

0,5839

0,5315

0,6706

0,0719

Individual

p-Value

<,0001*

0,0506

0,0192*

0,0035*

0,1589

0,0386*

0,1016

1,0000

1,0000

0,3368

1,0000

0,4114

1,0000

1,0000

1,0000

1,0000

1,0000

1,0000

0,6043

Simultaneous

p-Value

Contrasts

-10

0

10

20

30

40

50

60

70

80

90

Ab

so

lute

Co

ntr

ast

BPRweightHPC StagesLPC Stageslengththrust*thrustHPC Stages*HPC StagesOPR*OPR

thrust

OPR

0,0 0,5 1,0 1,5 2,0 2,5

Half Normal Quantile

Half Normal Plot

Lenth PSE=1,81115

Asterisked terms were forced orthogonal. Analysis is order dependent.

P-Values derived from a simulation of 10000 Lenth t ratios.

Make Model Run Model

Screening for 3rd cost adj

Figure 3.1: JMP screening function

For each of the independent variables, the respective t-ratio’s and p-values are displayed.This is a first hint of what independent variable could contribute to the model of the analyzeddependent variable. In conjunction with the screening, there is the possibility to conduct aquick regression analysis with the highlighted variables as input. This first regression analysisenables an evaluation of the capability of certain variable combinations to model the dependentvariable. The result of this initial screening was, that all additional engine specifications do notprovide any valuable contribution to modeling the CERs. Another important result was, thatthe interval analysis turned out to be more complicated. It was not possible to find acceptableregression results for an interval analysis of the entire engine range of the database. Therefore,the separation of the database in SH and MLH engines was utilized. The shop visit intervals werefurther divided in first and mature-run intervals for SH engines and MLH engines. While theengines with three spool configuration were excluded from the interval analysis, they turned outto be eligible for the cost analysis. Older generation engines or engine data from newly marketedengines have been excluded entirely from the data analysis. The initial screening confirmed theassumption that this data could not be modeled adequately together with the remainder datapoints. As visible on figure 3.1, the software also automatically evaluates the significance ofvariable combinations. This is an useful function, since it is hard to predict how such variablecombination could contribute to the model.

Shop Visit Interval Shop Visit Cost1 2 3 4 5 6 7

Short-HaulFirst Interval

Medium-Long-Haul FirstInterval

Short-HaulMatureInterval

Medium-Long-Haul Mature

Interval

FirstRestorationReserves

MatureRestorationReserves

LLP Reserves

Table 3.7: Summary of the final dependent variables to be analyzed

Table 3.7 summarizes the seven dependent variables for which the respective cost estimatingrelationships are developed in the following section.

3.2 Data Analysis 42

3.2.2 Regression Analysis

With all CERs established, the next step was to find the best fit of valid combinations ofindependent variables to the analyzed dependent variable. This stage was also conducted throughapplying the JMP 8.0 software. The software does the statistical data analysis independently.That means it determines for a predefined set of independent variables the prediction functionthat results in the best least square fit to the analyzed dependent variable. Promising sets ofindependent variables can be derived from the data screening. The best combination of variableswas then simply determined by evaluating and comparing the regression results of differentvariable sets. An example output of the regression results established by the used computersoftware is shown in figure 3.2. From the initial data screening, the possible independent variableshave been narrowed down to the main engine specifications: take-off thrust and dry weight aswell as the ratio of both. This simplified the process of finding the combinations that result inthe best overall fit.

LLP- Fit Least Squares Page 1 of 2

0

100

200

300

400

500

600

700

800

900

LL

P R

ese

rve

s

Actu

al

0 100 300 500 700 900

LLP Reserves Predicted

P<.0001 RSq=0,95 RMSE=44,13

Actual by Predicted Plot

RSquare

RSquare Adj

Root Mean Square Error

Mean of Response

Observations (or Sum Wgts)

0,952884

0,950528

44,13013

254,4868

64

Summary of Fit

Model

Error

C. Total

Source

3

60

63

DF

2363135,5

116848,1

2479983,6

Sum of

Squares

787712

1947

Mean Square

404,4799

F Ratio

<,0001*

Prob > F

Analysis of Variance

Intercept

weight

thrust

(weight-8608,78)*(weight-8608,78)

Term

-115,3133

0,0194512

0,0031206

2,6924e-6

Estimate

13,6202

0,007095

0,001069

3,188e-7

Std Error

-8,47

2,74

2,92

8,44

t Ratio

<,0001*

0,0080*

0,0049*

<,0001*

Prob>|t|

Parameter Estimates

-150

-100

-50

0

50

100

150

200

250

LL

P R

ese

rve

s

Re

sid

ua

l

0 100 300 500 700 900


Residual by Predicted Plot

Whole Model

Response LLP Reserves

Figure 3.2: JMP regression results example output

3.3 Results of the Parametric Cost Modeling 43

3.3 Results of the Parametric Cost Modeling

This sections sums up the results in shape of the determined prediction functions separated ininterval and cost CERs. The developed CERs have at most three different components as input.This also includes combinations of the three remaining independent variables: thrust, weight

and TWR. The units of measurement of these independent variables were adopted from thedatabase. Their definition is summarized in table 3.8. A regression summary as illustrated infigure 3.2 for each of the seven CERs is displayed in appendix C.

Input Symbol Unit

thrust lbf pound-force

weight lb pound

TWRlbflb

pound-force per pound

Table 3.8: Defined standard units for the input parameters of the CERs

3.3.1 Shop Visit Interval CERs

As previously established, the SV interval as a major dependent variable was divided in foursub-variables, in order to account for occurring anomalies in the database. The correspondingCERs are subsequently summarized. From the hypothesis, it was expected that the intervallength mainly depends on the TWR of an engine. This was not confirmed by all four developedCERs. The first intervals of both, SH and MLH engines are only marginally under the directinfluence of the TWR. However, these two CERs also heavily depend on the input of both, theengine thrust and the engine weight, so that the hypothesis is still represented. The developmentof the interval CERs was based on intervals given in EFH, thus the output of the following CERsis also defined in EFH as standard interval measurement unit.

3.3.1.1 First Interval for SH-Engines

The first removal intervals for short-haul operating engines can be expressed as a function ofTWR, weight and weight2:

IntervalFR,SH = 68466 − 8267.81904 · TWR − 1.00444 · weight

+ (weight− 5407) · [(weight− 5407) · 0.00012125] (3.1)

3.3.1.2 Mature Interval for SH-Engines

For mature removal intervals the regression analysis showed that the CER is just a function ofthe thrust-weight ratio (TWR):

IntervalMR,SH = 40684 − 5022.8116 · TWR (3.2)


3.3.1.3 First Interval for MLH-Engines

The determined CER for the first removal interval of medium-long-haul operating engines includesthe dependent variables weight, thrust and weight2:

IntervalFR,MLH = 22539 + 1.4329 · weight − 0.3147 · thrust

+ (thrust− 76305) · [(thrust− 76305) · 0.0000034421] (3.3)

3.3.1.4 Mature Interval for MLH-Engines

Mature intervals for medium-long-haul operating engines are, similar to equation (3.1) reflectedas a function of TWR, weight and weight2:

IntervalMR,MLH = 34415 − 2759.25322 · TWR − 0.36625 · weight

+ (weight− 12072) · [(weight− 12072) · 0.000101795] (3.4)

3.3.2 Shop Visit Cost CERs

As a result of the data source structure, the shop visit costs as second major dependent variable,have been divided into restoration cost and LLP cost. Since the LLP reserves in the databaseare represented in USD/EFC, the LLP CER generates LLP costs given in USD/EFC. The LLPcosts turned out to be fairly stable as the engine ages. Therefore, only the restoration costs havebeen furthermore divided in first-run and mature-run costs. The restoration cost CERs weredeveloped based on shop visit restoration costs given in USD/EFH. Hence, the output of therestoration costs CERs is also defined in USD/EFH. The developed CERs largely reflect thepredictions resulting from the hypothesis. The three corresponding CERs are presented in thefollowing.

3.3.2.1 First-Run Restoration Costs

The restoration cost per EFH for first shop visits have been modeled as a function of the thrust.This CER is valid for the entire range of engines:

SV RCFR,EFH = 7 + 0.002361887 · thrust (3.5)

3.3.2.2 Mature-Run Restoration Costs

The mature shop visit restoration cost were also modeled with the thrust as only input component:

SV RCMR,EFH = 46 + 0.002886118 · thrust (3.6)


3.3.2.3 LLP Costs

Eventually the CER estimating the LLP costs of all engines of the database is a function ofweight, thrust and weight2:

LLPCost =− 115 + 0.0194512 · weight + 0.0031206 · thrust

+ (weight− 8608.78125) · [(weight− 8608.78125) · 2.69234 · 10−6] (3.7)

4 Modeling of Engine Maintenance Cost

As mentioned earlier, the foundation of a cost estimating model is the cost breakdown, thatincludes all relevant costs. The cost breakdown structure from the perspective of this study hasbeen established in figure 2.15 on page 25. Since the engine line maintenance is already includedin the existing LLC-Tool, the focus here lies on modeling the shop visit costs. Maintenancecosts are normally divided in direct and indirect costs. Therefore, the SV costs can be furtherdifferentiated in SV DMC and SV IMC. Since the collected data reflects calculated prices fromMRO providers, it is assumed that charges for maintenance burden are already included in thecollected cost data. However, the costs for maintaining a spare engine pool or leasing spareengines for the duration of a shop visit are generally not included. Therefore, spare enginecosts as part of the indirect maintenance costs are considered separately. The objective is todevelop a qualitative maintenance cost model that focuses on estimating the shop visit DMC andintervals. Charges for spare engines are optionally added based on average leasing rates. Theassembled database and the resulting cost estimating relationships developed in the previouschapter, predetermine parts of the DMC model structure. However, since the CERs are based ona normalized database, they do not reflect the impact of major influence factors such as, flighttime, derate, number of spools or environment. These effects have to be modeled subsequentto the CERs. This chapter summarizes the considerations that led to the final cost estimatingmodel structure.

4.1 Model Structure

The first step in establishing the model structure is to determine what the input and whatthe output parameters are. The objective of the engine maintenance model is to estimate SVintervals and SV costs. The developed CERs distinguish between first-run and mature-run shopvisits. Thus, there are four output parameters: SV interval and SV costs for each engine phase.The input parameters depend first of all on the necessary input for the developed CERs.These parameters are, the engine thrust and the engine weight. Since the interval CERs arefurther divided in short-haul and medium/long-haul engines, an additional input parameter thatdetermines what CER is applied, has to be introduced. This additional parameter was termedengine application and is considered as an engine specification, since it is a static parameterlinked to the engine variant. All of these input parameters can be derived from the developedCERs. However, as mentioned above, there are important effects that are not modeled in theCERs. Therefore, there are more necessary input parameters. These include the number ofengine spools as well as operational factors like flight time, derate and information about the

46

4.1 Model Structure 47

severity of the environment.

AC Engine

Maintenance

Cost Model

Weight

FR Removal

Interval

EFH:EFC FR Shop Visit

Cost

MR Removal

Interval

Thrust

No. Spools

Application

Derate

Engine Utilization

Engine Specifications

SV Interval

SV Costs

Folie 3


*FR: First-Run

*MR: Mature-Run

MR Shop Visit

Cost

Derate

Environment

Engine Utilization

Figure 4.1: Black box of maintenance model

Figure 4.1 illustrates the maintenance model as black box with a summary of all input andoutput parameters. With the model input and output established, the inner structure of the blackbox can be generated. In order to match the output parameters, the model contains in generaltwo separate lines: a cost-line and an interval-line. The results of the CERs are normalizedvalues for costs and intervals, based on the engine’s weight and thrust. Thus, they do not modelany operational severity effects. These effects are modeled in conjunction with the normalizedvalues from the CERs. Therefore, the inner structure of the model has been split into two serialmodules. The first module reflects the developed CERs, while the second represents all additionaleffects influencing the shop visit costs and intervals. The two modules are thus termed as follows:

• CER-Module• Effect-Module

The CER-Module determines normalized base values for the shop visit costs and intervals. Thesebase values are then adjusted in the Effect-Module with a series of adjustment factors. Theadjustment factors are determined in correspondence to the respective input parameters. Theentire model structure is illustrated in figure 4.2. The two modules are described in some moredetail in the following subsections.


Time&Material Factor Three-Spool Factor

EffectEffectEffectEffect----ModuleModuleModuleModule

FR Interval Func.SMHE

MR Interval Func.SMHE

FR Interval Func.MLHE

MR Interval Func.MLHE

ShortShortShortShort----Medium HaulMedium HaulMedium HaulMedium Haul MediumMediumMediumMedium----Long HaulLong HaulLong HaulLong Haul

Base Shop Visit IntervalBase Shop Visit IntervalBase Shop Visit IntervalBase Shop Visit Interval

FR Cost Func.

MR Cost Func.

Restoration CostRestoration CostRestoration CostRestoration Cost

LLP Cost Func.

Base Shop Visit CostBase Shop Visit CostBase Shop Visit CostBase Shop Visit Cost

FR Rest.Cost [$/EFH]

MR Rest.Cost [$/EFH]

FR Base Interval [EFH]

MR Base Interval [EFH]LLP Cost [$/EFC]

CERCERCERCER----ModuleModuleModuleModule

Folie 2


FR Interval [EFH]

MR Interval [EFH]

FR Cost [$/EFH]

MR Cost [$/EFH]

EFC:EFH RatioSeverity Factor

Environment Factor

Figure 4.2: Inner structure of the cost estimating model

4.1.1 CER-Module

The CER-Module basically comprises of nothing more than the seven CERs developed in theprevious chapter (see blue frames in figure 4.2). With the input of the engine weight and thrustplus the information if it is a SH or MLH engine, the CER-Module generates five outputs:

• LLP Costs [USD/EFC]• FR Restoration Costs [USD/EFH]• MR Restoration Costs [USD/EFH]

• FR Base Interval [EFH]• MR Base Interval [EFH]

These base outputs are valid only for the normalized conditions on which the CER developmentwas based (see table 3.5). The adjustment to the operational severity is performed in the followingeffect-module. For this adjustment, the determined shop visit intervals as output of the intervalCERs have to be converted into a shop visit rate (SVR). In this instance, the SV interval andthe SVC are represented in analogue measurements (both relating to [1/EFH]). This enables theadjustment of both values with the same factors.


4.1.2 Effect-Module

The Effect-Module, generates the factors necessary to adjust the base costs and intervals from theCERs according to the input of the operational severity and the number of engine spools. Thereare five factors (red frames in figure 4.2), which are subsequently discussed. The output of theeffect-module are adjusted SV intervals and SVC per EFH divided in first-run and mature-runshop visits. Since the effect-module merges the LLP costs and the restoration costs from theCER-module, it generates four outputs:

• FR Shop Visit Costs [USD/EFH]• MR Shop Visit Costs [USD/EFH]

• FR Shop Visit Interval [EFH]• MR Shop Visit Interval [EFH]

With this output, the absolute shop visit costs can be calculated through multiplying the SVCper EFH with the respective SV interval.

4.1.2.1 Severity Factor

The concept of severity factors extracted from severity curves has been introduced in 2.2.3.Severity factors adjusts restoration costs and shop visit intervals corresponding to the averageflight time and derate under which the engine was operated. It was not possible to obtain theseverity curve of each engine in the DB. Therefore, two average severity curves that approximatethe severity curves of a range of engines, have been developed (see 3.1.4.2). These averagecurves were already applied to normalize the flight time of the DB engines (see tables 3.3 and3.4). This normalization did not include an adjustment of the derate, since the DB entries wereassumed to have a constant derate. However, the average severity curves were developed to alsoinclude multiple curves for each of the common derate levels. The two developed severity curvesare fully illustrated in appendix D. These average curves are now the basis for modeling theeffects of flight time and derate on the restoration costs and shop visit intervals as part of theeffect-module. With the flight time and the derate as input, the severity curve simply gives outthe corresponding severity factor which is then multiplied with the restoration costs and theinterval (as seen in the example calculation on page 23).

4.1.2.2 Time & Material Factor

The time & material factor (TMF) has been introduced to account for the effect, that theabsolute shop visit restoration costs (SVRC) generally increase with increasing TOW. When theseverity factor is applied alone, the absolute SVRC remain constant regardless of the flight timeor derate. This is because the SF adjusts both the interval and the restoration costs per EFHsimultaneously. However, the increased TOW due to raised derate and flight time should resultin increasing SVRC. The time & material factor models this effect. Therefore, one could expectthat the TMF can be expressed similar to the severity factor via multiple curves, only inverted


so that the factor increases with decreasing flight time and derate. Due to lack of data it was notachieved to develop such multiple curves. However, it was possible to obtain a single examplecurve from a contact person in the engine maintenance industry. This curve does not reflectthe impact of the derate. Since the influence of the derate on the TOW is generally less severeand the accessible data is limited, the contribution of different derate levels was neglected. Theavailable single curve was considered as basis for developing T&M curves valid for all derates.As with the severity curves, two curves have been developed. One for all SH engines and one forall MLH engines. The two curves are subsequently illustrated in shape of a table. A graphicalillustration of the time & material curves can be found in appendix D.

EFH:EFC 0.5 1.0 1.5 1.9 2.5 3.0 4.0 5.0 6.0T&M Factor 0.90 0.95 0.98 1.00 1.02 1.03 1.04 1.05 1.06

Table 4.1: SH Engine Time & Material factor with respect to the flight time

EFH:EFC 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0T&M Factor 0.85 0.91 0.94 0.96 0.98 1.00 1.03 1.05 1.07 1.09 1.10 1.11

Table 4.2: MLH Engine Time & Material factor with respect to the flight time

4.1.2.3 Three-Spool Factor

The three-spool factor (TSF) models the extended TOW of engines with a three-spool config-uration compared to the more common two-spool engines. In general, there was no detailedadditional information on the impact of the three-spool configuration on the achievable SVintervals accessible. However, since the DB indicates that three-spool engines achieve significantlonger SV intervals, the available data from the assembled DB was enabled to determine a simpleconstant factor that models this effect. This factor was determined through averaging the offsetof the original three-spool data points over the generated intervals from the two-spool CERswith the respective three-spool engine specifications as input. However, it has to be noted thatall three-spool engines of the DB are MLH engines. It is assumed that SH engine are influencedin a similar manner. The result of the analysis was the following offset factor:

• TSF = 0.7 for three-spool engines• TSF = 1.0 for two-spool engines

In case the input indicates that the proposed engine is a three-spool engine, the SVRs generatedfrom the interval CERs are simply multiplied with the TSF = 0.7 to account for the expectedlonger TOW of a three-spool configuration. The TSF simply equals 1.0, in case of a standardtwo-spool engine. For a qualitative consideration, this simple approach is sufficient to model theinfluence of the number of spools.

4.1.2.4 Environment Factor

The environment factor (EF) reflects the impact of the present environmental conditions includingthe outside air temperature on engine maintenance. Studying the ACA indicated that the


environment influences the SV intervals and costs considerably. This was also confirmed throughthe correspondence with different professionals in the aircraft engine maintenance field. However,it was difficult to locate clear data on this topic. As guideline for modeling the environmentalimpact served a paper from Ackert [Ack10]. Ackert defines three gradual levels of environmentalseverity and relates each level to a certain escalation factor. These environment levels and theircorrelating EFs are listed in table 4.3. The respective EF is then multiplied with the overall SVRand SVRC in order to adjust the intervals and costs to the present environmental severity. Table4.3 also indicates typical regions for each environment level.

Environment EF Typical Regions

Temperate 1.0 North America, Europe, Australia

Hot/Dry 1.1 Middle East, North Africa

Erosive 1.2 Coastal China, SE Asia, India

Table 4.3: Environment factors for different environmental conditions

4.1.2.5 EFC:EFH Ratio

Strictly speaking the EFC:EFH ratio is not a factor that is intended to model a certain influentialeffect on engine maintenance. The EFC:EFH ratio, as the reciprocal value of the flight time(EFH:EFC), translates costs represented in USD/EFC into USD/EFH. The effect module sumsup the LLP costs and the restoration costs in order to obtain one measure for the overall shopvisit costs per EFH. However, the LLP costs are generally given in USD/EFC, whereas theintervals and the restoration costs are based on EFH. Therefore, the LLP costs have to beconverted into USD/EFH in order to enable the summation of LLP costs and restoration costs.

4.1.3 Spare Engine Charges

Aircraft engines that are removed and sent to the workshop are normally replaced with spareengines, in order to keep the aircraft in service while its original engines are overhauled. Costsincurred by either owning or leasing spare engines are generally considered as a cost driver inengine maintenance. Therefore, the targeted model is supposed to enable the considerationof spare engine charges. For a qualitative estimation model, it is not practical to model theexpenditures for spare engines into detail. That would require comprehensive information onthe airline’s engine fleet situation and access to spare engine pools as well as on current leasingmarket developments. However, in correspondence with one of the product managers of LufthansaTechnik1, it was established that a reasonable estimation of spare engine costs can be achievedthrough current leasing rates. Leasing rates can vary between 2000-5000 USD/day. This deviationis not only related to the engine type, but it also heavily relates to the current supply anddemand situation for the respective engine. Therefore, there were no CERs developed that reflectspare engine leasing rates. In the framework of this study, the spare engine costs are simply

1 http://www.lufthansa-technik.com

4.2 Example Application 52

estimated with a predefined leasing rate and the information of the duration of the shop visit.Current engine leasing rates are not sensitive information and can be obtained on request fromengine lessors. In case there are no leasing rates accessible it is proposed to assume an averageleasing rate of 3500 USD/day.A reasonable average shop visit duration is 80 days. Since the developed SV DMC model considersthe workload of each shop visit as equal, this average of 80 days is established as standard shopvisit duration. Therefore, the total spare engine costs (SEC) for an average shop visit yield to:

SEC = LeasingRate · SV duration ≈ 3500 · 80 = 280,000 USD (4.1)

If more detailed information is available, the two parameters of this simple approach can beadjusted at all times. The proposed values here give an idea of the dimension of the costs andwill serve as default values of the model.

4.2 Example Application

This section illustrates the functionality of the cost estimating model through an examplecalculation that includes all equations necessary to generate the output of the model. Theexample is based on the following input parameters:

Engine Specifications

Parameter Input

Thrust [lbf] 79900

Weight [lb] 14545

No. Spools 2

Application MLH

Table 4.4: Input engine specifications

Engine Utilization

Parameter Input

EFH:EFC [h] 8.0

Derate 10%

Environment Temperate

Table 4.5: Input engine utilization

This input relates to a Pratt & Whitney 4077 operated on a long-haul route typical for awide-body aircraft like the Boeing 777-200. With this input, the CER-module first determines thebase intervals and costs, which are subsequently adjusted through a series of adjustment factors.Since the input parameters are already given in the required measurement unit, a conversion ofthe units does not apply.

4.2.1 Base Costs and Intervals from CERs

The PW 4077 is classified as MLH engine. Therefore, the equations (3.3) and (3.4) are appliedto determine the base interval lengths:


BaseIntervalFR = 22539 + 1.4329 · 14545 − 0.3147 · 79900

+(79900− 76305) · [(79900− 76305) · 0.0000034421] (4.2)

= 18500 EFH

BaseIntervalMR = 34415 − 2759.25322 · 7990014545 − 0.36625 · 14545

+(14545− 12072) · [(14545− 12072) · 0.000101795] (4.3)

= 14700 EFH

The restoration cost for FR and MR shop visits are calculated using the eqs. (3.5) and (3.6):

BaseSV RCFR,EFH = 7 + 0.002361887 · 79900 = 194 USD

EFH(4.4)

BaseSV RCMR,EFH = 46 + 0.002886118 · 79900 = 275 USD

EFH(4.5)

The LLP cost are eventually determined through equation (3.7):

LLPCost = −115 + 0.0194512 · 14545 + 0.0031206 · 79900

+(14545− 8608.78125) · [(14545− 8608.78125) · 2.69234 · 10−6] (4.6)

= 509 USD

EFC

All intermediate results generated through the CERs are summarized in table 4.6.

IntervalF R [EFH] IntervalMR [EFH] SV RCF R, EF H

[USDEF H

]SV RCMR, EF H

[USDEF H

]LLPCost

[USDEF C

]18500 14700 194 275 509

Table 4.6: Summary of CER results for the example input parameters

4.2.2 Adjustment of Intervals

The adjustment factors of the effect-module are defined so that they relate to the elapsed EFH.Therefore, the determined intervals of the CERs have to be translated into SVRs. This enablesan analogue application of the adjustment factors for both the SV intervals and SV restorationcosts. With equation (2.1) the two determined intervals from (4.2) and (4.3) are converted to:

BaseSV RFR = 1000BaseIntervalFR

= 100018500 = 0.054 SV s

1000 EFH(4.7)

BaseSV RMR = 1000BaseIntervalMR

= 100014700 = 0.068 SV s

1000 EFH(4.8)


In order to obtain the final SVR, the base SVRs are multiplied with the three-spool factor (TSF),the severity factor (SF) and the environment factor (EF) as illustrated in figure 4.2.

SV R = BaseSV R · TSF · SF · EF (4.9)

The input indicates that both the TSF and EF equal 1.0, since the example engine has two spoolsand is operated in a temperate environment. The SF is obtained from the average severity curvefor MLH engines (appendix D.2). For a flight time of 8.0 hours and a derate of 10% the severityfactors yields to SF = 0.88. Therefore, the actual first-run and mature-run SVRs result in:

SV RFR = 0.054 · 1.0 · 0.88 · 1.0 = 0.048 =̂ 20800 EFH (4.10)

SV RMR = 0.068 · 1.0 · 0.88 · 1.0 = 0.060 =̂ 16700 EFH (4.11)

These final SVRs can be converted back into an interval expressed in EFH, as performed above.

4.2.2.1 Adjustment of Costs

The total shop visit costs per EFH consist of LLP costs per EFH (LLPCostEFH) and theadjusted restoration costs (SV RCEFH).

SV CEFH = LLPCostEFH + SV RCEFH (4.12)

The LLP costs are not adjusted by any effect factors. However, since the LLP costs are given inUSD/EFC, they have to be converted into USD/EFH:

LLPCostEFH = LLPCost · EFC

EFH= 509 · 1

8.0 = 64 USD

EFH(4.13)

The base restoration costs from eqs. (4.5) and (4.5) have to be multiplied with the time &material factor, the severity factor and the environment factor in order to get the final restorationcosts.

SV RC = BaseSV RCEFH · TMF · SF · EF (4.14)

The TMF is determined with the respective time & material curve for MLH engines (appendixD.3). With a flight time of 8.0, it yields to TMF = 1.05. The severity factor and environmentfactor are known from before. Thus, the FR and MR restoration costs result in:

SV RCFR,EFH = 194 · 1.05 · 0.88 · 1.0 = 179 USD

EFH(4.15)

SV RCMR,EFH = 275 · 1.05 · 0.88 · 1.0 = 254 USD

EFH(4.16)

With the results from the eqs. (4.15) and (4.15), the total shop visit cost per EFH are thencalculated through eq. (4.12):

4.3 Model Plausibility 55

SV CFR,EFH = LLPCostEFH + SV RCMR,EFH = 64 + 179 = 243 USD

EFH(4.17)

SV CMR,EFH = LLPCostEFH + SV RCMR,EFH = 64 + 254 = 318 USD

EFH(4.18)

4.2.3 Final Results

Eventually the total shop visit costs can be calculated through the multiplication of the SVC perEFH and the respective shop visit interval or time on wing.

SV C = SV CEFH · Interval (4.19)

SV CFR = SV CFR,EFH · IntervalFR = 243 · 20800 = 5.1 mil USD (4.20)

SV CMR = SV CMR,EFH · IntervalMR = 318 · 16700 = 5.3 mil USD (4.21)

The final output of the cost estimation are first-run and mature-run shop visit intervals and shopvisit costs. The results for the proposed example are summarized in table 4.7.

IntervalF R [EFH] IntervalMR [EFH] SV CF R [USD million] SV CF R [USD million]

20800 16700 5.1 5.3

Table 4.7: Final output for the example input parameters

4.3 Model Plausibility

The plausibility of the model was continuously monitored while developing the CERs and creatingthe model structure. This intermediate plausibility tests significantly contributed to the decisionsmade throughout the development process. This section illustrates the examination of thecredibility of the final model. In general, it is important to avoid using the same data that wasapplied to develop the model for subsequent plausibility tests. It can be expected that the modelreflects the collected data of the database. However, since the database has been normalized andthe final model structure includes not only the developed CERs but also a series of adjustmentfactors, it is first analyzed how well the final model reflects the original data points, prior to theflight time normalization. Subsequently, the model results are compared to additional availabledata sources.

4.3.1 Model Results vs. Original Database

As indicated, it is established that the developed CERs reflect the normalized data points quitewell (see regression results in appendix C). However, the objective of the following analysisis to illustrate how the combination of the different CERs and the subsequent effect-modulereflect the original data before the flight time normalization. Therefore, each of the primary


output parameters of the model is plotted over the respective actual original data points fromthe database. Ideally, the resulting points would lead to a graph that equals the standard linearcurve f(x) = x. In this case every predicted value would be equal to the respective actual value.This ideal linear curve is plotted as a blue dotted line. However, it can be expected that theplotted data points do not lie perfectly on this line. Furthermore it is possible that the idealcurve does not even represent the trend line of the data points. Therefore, a linear regressionline that fits the data points is developed and additionally plotted as a red continuous line.Coinciding red and blue dotted lines indicate that the data points can be fitted by the ideal curvethrough linear regression. Clearly crossing lines would indicate opposing trends and thus refer toa bad reflection of the actual data through the model. In the following tables, the resulting plotsillustrating the comparison of the model output with the original database are displayed. Eachengine application is considered separately. In addition, each plot displays the root mean squareerror (RMSE) between data points and ideal curve.

0.5 1 1.5 2 2.5 3 3.5

x 104

0.5

1

1.5

2

2.5

3

3.5x 10

4 FR Interval SH-Engines

Predicted Interval [EFH]

ActualIn

terval[E

FH]

RMSE: 3088EFH

Data PointsRegression of DataIdeal Curve

0.5 1 1.5 2 2.5

x 104

0.5

1

1.5

2

2.5x 10

4 MR Interval SH-Engines


ActualIn

terval[E

FH]

RMSE: 2094.6EFH


0.5 1 1.5 2 2.5 3

x 104

0.5

1

1.5

2

2.5

3x 10

4 FR Interval MLH-Engines


ActualIn

terval[E

FH]

RMSE: 1588EFH

2 Shaft Data3 Shaft DataRegression of DataIdeal Curve

0.5 1 1.5 2 2.5

x 104

0.5

1

1.5

2

2.5x 10

4 MR Interval MLH-Engines


ActualIn

terval[E

FH]

RMSE: 1042.1EFH


Intervals The plots show that the model generally represents the original database intervalswell. In the MLH plots, the three-spool engines have been highlighted. It can be seen thatthe three-spool data points follow the trend of the remaining data quite well. This is achieved


through the three-spool adjustment factor. In addition, the RMSE values of this analysis relate tothe RMSEs of the regression analysis that was performed to develop the interval CERs (appendixC).

100 200 300 400 500 600 700 800

100

200

300

400

500

600

700

800FR SVC per EFH SH-Engines

Predicted Cost [USD/EFH]

ActualCost

[USD/EFH]

RMSE: 30.7 USDEFH


100 200 300 400 500 600 700 800

100

200

300

400

500

600

700

800MR SVC per EFH SH-Engines


ActualCost

[USD/EFH]

RMSE: 35.3 USDEFH


100 200 300 400 500 600 700 800 900100

200

300

400

500

600

700

800

900FR SVC per EFH MLH-Engines


ActualCost

[USD/EFH]

RMSE: 32.4 USDEFH


100 200 300 400 500 600 700 800 900100

200

300

400

500

600

700

800

900MR SVC per EFH MLH-Engines


ActualCost

[USD/EFH]

RMSE: 52.7 USDEFH


Shop Visit Costs per EFH These plots generally indicate a weaker reflection of the original datathrough the developed model. This probably results from the fact that the SVC per EFH aremade up of two independently modeled cost components (SV RC and LLPCost). Especially theSH engine comparison reveals a clearly visible deviation between model and database. The datapoints here are clustered around a low cost level, while only a few data points reach higher costdimensions. The explanation for this is that the database of SH engines mainly consists of smallengines with a thrust level of about 20,000-30,000 lbf. The few data points that stick out aremade up of the CF6-80C2A series engines. These are the only bigger size short-haul engines ofthe database. Therefore, the credibility of the model in this region is somewhat limited. It seemsthe model tends to predict generally higher costs for such big SH engines. For MLH engines, thepicture is slightly more favorable. The data points are not as clustered around a certain costlevel. The two data points that stick out stand for the PW4074/77 operated on a short haulroute. The model is capable of reflecting this engine sufficiently however, one has to be careful


again, since there are only two data points that confirm the displayed trend in the higher costlevel of short-haul operated MLH engines.

4.3.2 Model Results vs. Additional Data Sources

It is crucial to also compare the developed model with additional available data. However,adequate data sources are limited as it was pointed out in 3.1.2. Solely, the AeroStrategydatabase proved to be appropriate for a plausibility test. However, since this database doesnot provide all information that are compulsory for running the model, a few assumptionshad to be made. The average derate is again assumed to be 10%, while the environment isconsidered as temperate. In addition it was necessary to define a legitimate flight time for eachdata point of the AeroStrategy table. As mentioned earlier, the AeroStrategy table indicatesfor most engines the respective aircraft, on which the estimation is based. Together with theACA aircraft operator & owner guides it was determined what the global average flight timefor these aircraft is. This average flight times were assigned to each data point and served asinput for the model. Subsequently, the AeroStrategy database was further extended to alsoinclude the respective weight, thrust, number of spools and the engine application of each datapoint. Hence, all necessary input parameters have been defined to perform a comparison of themodel with the addtional data. The plausibility based on the AeroStrategy database is testedfor six output parameters. On the cost side, the SV restoration costs per EFH and the total SVrestoration costs are compared separately for first and mature shop visits. Since the databasedoes not provide any cost figures for LLP charges, this cost component could not be checkedfor its credibility. In addition, the first and mature removal intervals are compared betweenmodel and AeroStrategy estimates. The presentation of the plausibility analysis correlates to theprevious plots. As before, the displayed RMSE values refer to the error between the data pointsand the dotted blue line (ideal curve).

0.5 1 1.5 2 2.5 3

x 104

0.5

1

1.5

2

2.5

3x 10

4 FR Interval -Model vs. AeroStrategy


ActualIn

terval[E

FH]

RMSE: 3169.6EFH


0.5 1 1.5 2

x 104

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

2.2x 10

4 MR Interval -Model vs. AeroStrategy


ActualIn

terval[E

FH]

RMSE: 1745.2EFH


Intervals At first glance it becomes apparent, that the predicted intervals generally tend to belonger than the intervals of the AeroStrategy data tables. However, the model clearly reflects the


prevailing trend of the reference data table. The resulting RMSE values are in the same scalelike the RMSEs of the other interval analysis plots, which indicates that model relates similarlyto both the Aircraft Commerce and AeroStrategy database.

50 100 150 200 250

50

100

150

200

250

FR SVRC per EFH - Model vs. AeroStrategy

Predicted SVRC [USD/EFH]

ActualSVRC

[USD/EFH]

RMSE: 32.4 USDEFH


100 150 200 250 300 350 400

100

150

200

250

300

350

400

MR SVRC per EFH - Model vs. AeroStrategy

Predicted SVRC [USD/EFH]

ActualSVRC

[USD/EFH]

RMSE: 41.2 USDEFH


SV Restoration Costs per EFH Comparing the SVRC per EFH of the developed model withAeroStrategy also reveals a satisfying picture. The model represents the trend of reference datapoints very well and the displayed RMSEs are in an acceptable scale.

0 1 2 3 4 5 60

1

2

3

4

5

6FR SVRC - Model vs. AeroStrategy

Predicted SVRC [USD million]

ActualSVRC

[USD

million]

RMSE: 0.6 milUSD


0 1 2 3 4 5 60

1

2

3

4

5

6MR SVRC - Model vs. AeroStrategy

Predicted SVRC [USD million]

ActualSVRC

[USD

million]

RMSE: 0.6 milUSD


Total SV Restoration Costs The plots that are relating to the comparison of the total SVRCgenerally show a bigger deviation between model and AeroStrategy. This is somewhat expected,since now the combined results of intervals and costs per EFH in shape of the total restorationcosts are compared. It becomes apparent that the model generally predicts higher costs thanestimated by AeroStrategy for engines that require high investment for an overhaul. Theseestimations mainly belong to the newest generation engines of widebody aircraft that were notpart of the assembled database. Hence, it can be assumed that new generation engines generallyachieve longer intervals with reduced cost per EFH. This is a trend that has been confirmedthrough comparing the older engine generation of the database with the current generation

4.4 Sensitivity Analysis 60

engines that formed the foundation of the database. It is likely that this trend continues now,leading to even longer intervals and lower shop visit costs per EFH for the newest generationengines. However, the overall deviation is considered as acceptable and these results are stillregarded as confirmation of the developed model.

4.3.3 Summary of the Plausibility Tests

Considering that the objective was to develop a qualitative model that is capable of predictingrealistic SV intervals and costs with respect to basic relationships concerning engine maintenance,the model relates to the expectations. Undeniably, there are considerable deviations betweenmodel and available databases. However, the model does not attempt to give an exact forecastof shop visits costs. It was proven that the model qualitatively reflects the general correlationsthat define shop visit intervals and costs. In addition, the predicted absolute values lie in theexpected dimension. The present variations are considered admissible, granting that the exactforecast of SV intervals and costs is to hard to achieve.

4.4 Sensitivity Analysis

This section presents the results of a basic sensitivity analysis on the developed model. Excludingthe engine application, the model relates to six input parameters (see fig. 4.1).

Parameter SH base values MLH base values

Thrust [lbf] 27000 78000

Weight [lbs] 5139 14545

EFH:EFC [h] 1.9 6.0

Derate [%] 10 10

EF 1.1 1.1

TSF 1.0 1.0

Table 4.8: Input parameters and their base values for the sensitivity analysis

Based on these parameters the analysis was split into two parts. First, it was examined how theisolated output of the model reacts to changes on one single input parameter when the remainingparameters are kept constant. The six analyzed parameters and the values they are held constantto are summarized in tab. 4.8. The base SH engine values relate to the IAE V2500-A5 (A320family) and the MLH values are derived from the PW 4077 (Boeing 777). These are two verycommon engines in their respective field of application. The environment was assumed to behot and dry (EF = 1.1) and the two engines have two spools (TSF = 1.0). The impact of theenvironment and the number of spools was only modeled in rough discrete steps, while the modelenables continuous changes of the remainder input parameters. As a result of this, the analysisof the four continuous parameters - thrust, weight, flight time and derate - was grouped together,while the impact of the number of spools and the environment is illustrated separately. Thisensures a consistent presentation of the results. The second part is a sensitivity analysis on the


impact of the six input parameters on the entire SV life cycle cost of an aircraft engine.

4.4.1 Sensitivity of Model Output

The impact of the continuous parameters is presented in two tables. The tables consist of fourplots, each plot relating to one changing input parameter. Each table on the other hand relatesto one certain output parameter. The results here are presented only for the range of SH engines.The MLH sensitivity analysis generally produced similar results. These results can be found inappendix E.

4.4.1.1 Continuous Parameters

2.2 2.4 2.6 2.8 3 3.2 3.4

x 104

0.8

1

1.2

1.4

1.6

1.8

2

2.2

2.4

2.6

2.8x 10

4

Thrust [lbf ]

Interv

al[E

FH]

Thrust - Interval Sensitivity

EFH:EFC = 1.9[h]Derate = 10%Weight = 5139[lbs]

First-Run

Mature-Run

4000 4500 5000 5500 6000 65000.8

1

1.2

1.4

1.6

1.8

2

2.2

2.4

2.6

2.8x 10

4

Weight [lbs]

Interv

al[E

FH]

Weight - Interval Sensitivity

EFH:EFC = 1.9[h]Derate = 10%Thrust = 27000[lbf ]

First-Run

Mature-Run

1 1.5 2 2.5 3 3.5 40.8

1

1.2

1.4

1.6

1.8

2

2.2

2.4

2.6

2.8x 10

4

EFH:EFC [h]

Interval[E

FH]

EFH:EFC - Interval Sensitivity

Derate = 10%Thrust = 27000[lbf ]Weight = 5139[lbs]

First-Run

Mature-Run

0 5 10 15 20

1.2

1.4

1.6

1.8

2

2.2

2.4x 10

4

Derate [%]

Interv

al[E

FH]

Derate - Interval Sensitivity

EFH:EFC = 1.9[h]Thrust = 27000[lbf ]Weight = 5139[lbs]

First-Run

Mature-Run

Table 4.9: SH Engines - Impact of the continuous parameters on the SV Interval

Intervals The results of the interval sensitivity reflect the theory very well. An increase in thrustwould lead to a rapid drop of the achievable SV intervals, while an increase of the weight hasan opposed effect (upper two diagramms in tab. 4.9). This represents the expectation that ahigher thrust rating generally leads to shorter removal intervals. The influence of the derate andthe flight time directly relates to the implemented severity curves (see D). In addition, mature


SV intervals are generally shorter than first SV intervals, which has also been predicted by theoutcome of the literature review.

2.2 2.4 2.6 2.8 3 3.2 3.4

x 104

100

120

140

160

180

200

220

Thrust [lbf ]

SVC

per

EFH

[$/EFH]

Thrust - SVC per EFH Sensitivity


First-Run

Mature-Run

4000 4500 5000 5500 6000 6500100

110

120

130

140

150

160

170

180

190

200

Weight [lbs]

SVC

per

EFH

[$/EFH]

Weight - SVC per EFH Sensitivity


First-Run

Mature-Run

1 1.5 2 2.5 3 3.5 450

100

150

200

250

300

350

EFH:EFC [h]

SVC

per

EFH

[$/EFH]

EFH:EFC - SVC per EFH Sensitivity


First-Run

Mature-Run

0 5 10 15 20100

110

120

130

140

150

160

170

180

190

200

210

Derate [%]

SVC

per

EFH

[$/EFH]

Derate - SVC per EFH Sensitivity


First-Run

Mature-Run

Table 4.10: SH Engines - Impact of the continuous parameters on the SV costs per EFH

Shop Visit Costs per EFH The sensitivity of the shop visit costs per EFH is generally expectedas well. The flight time and the derate affect the SVC per EFH as defined in the severity curves.An increase of the thrust yields to a considerable linear increase of the costs per EFH, while theweight affects the costs only to a minor extent. The little effect the weight has on the SVC perEFH results mainly from the fact that the CER for the restoration costs per EFH only dependon the thrust (see eqs. (3.5),(3.6)).

4.4.1.2 Discrete Parameters

The remaining input parameters are implemented as discrete variables. Therefore, their impactis illustrated in bar plots. Apart from this, the presentation is analogue as seen before. Eachcouple of plots relates to one output parameter and each single plot shows the influence of onechanging input parameter, while the remaining parameters are held constant according to tab.4.8. The following plots illustrate the results for MLH engines using the example of the PW4077.The effect of the EF and the TSF on the SH engine intervals and costs are practically identical.


Thus, they are not illustrated explicitly. The results largely reflect the expectations and directlyrelate to the implementation of the environment and three-spool factors.

Environment

Interval[E

FH]

Environment - Interval Sensitivity

temperate hot&dry erosive0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

2.2x 10

4

First-RunMature-Run

No. Spools

Interval[E

FH]

No. Spools - Interval Sensitivity

3 Spools 2 Spools0

0.5

1

1.5

2

2.5

3x 10

4

First-RunMature-Run

Environment

SVC

perEFH

[$/EFH]

Environment - SVC per EFH Sensitivity


50

100

150

200

250

300

350

400

450

500

550First-RunMature-Run

No. Spools

SVC

perEFH

[$/EFH]

No. Spools - SVC per EFH Sensitivity

3 Spools 2 Spools0

50

100

150

200

250

300

350

400

450First-RunMature-Run

Table 4.11: MLH Engines - Impact of the discrete parameters on the direct model output

4.4.2 Sensitivity of Life Cycle SVC

The developed model is dedicated to serve as a module for a more complex aircraft life cyclecost tool. Therefore, it is now analyzed on how changes to the input parameters affect theaccumulated shop visit costs of the entire life cycle of an engine. The life cycle shop visit costsSV CLC based on the developed model can be estimated as follows:

SV CLC = IntervalFR · SV CFR,EFH + [EFHLC − IntervalFR] · SV CMR,EFH (4.22)

The remaining EFH after the first removal are multiplied with the mature SVC per EFH toaccount for all mature SVs. The accumulated engine flight hours of the life cycle (EFHLC) arecalculated with the number of years in service Y earsLC and the annual utilization Util ann.

EFHLC = Y earsLC · Util ann (4.23)

The annual utilization of the two example engines has been determined from the aircraft owner& operator guides of the aircraft commerce archive. According to [Air08d] a Boeing 777 on a


long-haul route of 6.0 EFH:EFC equipped with a PW4077, typically achieves an annual utilizationof 4500 EFH. An IAE V2500-A5 mounted to an A320 that is flying on 1.9 EFH:EFC short-haulroute is likely to achieve around 2800 EFH [Air06a]. With an estimated life cycle duration ofY earsLC = 20 years, the total number of EFH in service for the two example engines yields to:

EFHLC, SH = 20 · 2800 = 56,000 EFH (4.24)

EFHLC,MLH = 20 · 4600 = 90,000 EFH (4.25)

The sensitivity analysis of the total life cycle SVC is analogue to the previous analysis of the SVintervals and SVC per EFH. The following tables show the effect of the four continuous and thetwo discrete parameters. Subsequently it is illustrated through tornado charts what parametershave the most significant impact. The applied base values relate to table 4.8.

4.4.2.1 Continuous Parameters

The sensitivity of the influence of the continuous parameters is illustrated for each engineapplication separately.

2.2 2.4 2.6 2.8 3 3.2 3.4

x 104

7.5

8

8.5

9

9.5

10

10.5

11

11.5

12x 10

6

Thrust [lbf ]

LifeCycleSVC

[USD]

Thrust - LCC Sensitivity - SH


4000 4500 5000 5500 6000 65009.3

9.4

9.5

9.6

9.7

9.8

9.9

10

10.1x 10

6

Weight [lbs]

LifeCycleSVC

[USD]

Weight - LCC Sensitivity - SH


1 1.5 2 2.5 3 3.5 40.4

0.6

0.8

1

1.2

1.4

1.6

1.8x 10

7

EFH:EFC [h]

LifeCycleSVC

[USD]

EFH:EFC - LCC Sensitivity - SH


0 5 10 15 200.85

0.9

0.95

1

1.05

1.1

1.15x 10

7

Derate [%]

LifeCycleSVC

[USD]

Derate - LCC Sensitivity - SH


Table 4.12: SH Engines - Impact of the continuous parameters on the life cycle SVC


Short-Haul Engines The results for the range of SH engines largely relate to the sensitivityanalysis of the SVC per EFH in tab. 4.10. However, the reduced life cycle SVC with increasingengine weight seem to be slightly odd on the first glance. Generally, one would expect thatincreasing weight leads to increased total SVC. However, the shown curve is a result of the mainlyconstant SVC per EFH and significantly prolonged first intervals with increasing engine weight(compare with tab. 4.10). From these four continuous input parameters, only the flight time andthe derate can actually adopt a wide range of values depending on the operation of the aircraft.The engine thrust and weight are depending on the aspired performance level somewhat limitedby design constraints. When observing the plots, it becomes apparent that especially the averageflight time defines the resulting life cycle SVC. For the given example, it ranges from 6 mil USDin case of EFH:EFC = 4.0 to about 17 mil USD for a short-haul operation with EFH:EFC = 1.0.

7.4 7.6 7.8 8 8.2 8.4 8.6

x 104

3.15

3.2

3.25

3.3

3.35

3.4

3.45

3.5

3.55x 10

7

Thrust [lbf ]

LifeCycleSVC

[USD]

Thrust - LCC Sensitivity - MLH

EFH:EFC = 6[h]Derate = 10%Weight = 14545[lbs]

1.2 1.25 1.3 1.35 1.4 1.45 1.5 1.55 1.6

x 104

3.15

3.2

3.25

3.3

3.35

3.4

3.45x 10

7

Weight [lbs]

LifeCycleSVC

[USD]

Weight - LCC Sensitivity - MLH

EFH:EFC = 6[h]Derate = 10%Thrust = 78000[lbf ]

3 4 5 6 7 8 9 10 11 122.5

3

3.5

4

4.5

5x 10

7

EFH:EFC [h]

LifeCycleSVC

[USD]

EFH:EFC - LCC Sensitivity - MLH


0 5 10 15 203

3.1

3.2

3.3

3.4

3.5

3.6

3.7x 10

7

Derate [%]

LifeCycleSVC

[USD]

Derate - LCC Sensitivity - MLH

EFH:EFC = 6[h]Thrust = 78000[lbf ]Weight = 14545[lbs]

Table 4.13: MLH Engines - Impact of the continuous parameters on the life cycle SVC

Medium-Long-Haul Engines For a better understanding of the sensitivity of the life cycle SVCfor MLH engines, it is additionally referred to the respective tables in appendix E illustratingthe sensitivity of the isolated model output for MLH engines. In general, the results resemblethe previous SH engine plots. Solely, the impact of the weight is opposed. For MLH engines


the dependency of the life cycle SVC on the engine weight relates to the general expectation.This results from the fact that unlike with the SH engines, the SVC per EFH of MLH enginesconsiderably increase with increasing engine weight. Again the life cycle SVC considerably rangewith the average flight time.

4.4.2.2 Discrete Parameters

As with the analysis of the isolated model output, the influence of the discrete parameters isillustrated only for MLH engines. The resulting plots for SH engines would show the exact sametendencies just in a different scale.

Environment

LifeCycleSVC

[USD]

Environment - LCC Sensitivity - MLH


0.5

1

1.5

2

2.5

3

3.5

4x 10

7

No. Spools

LifeCycleSVC

[USD]

No. Spools - LCC Sensitivity - MLH

3 Spools 2 Spools0

0.5

1

1.5

2

2.5

3

3.5

4x 10

7

Table 4.14: MLH Engines - Impact of the discrete parameters on the life cycle SVC

4.4.2.3 Sensitivity Analysis via Tornado Charts

Tornado charts are often used to illustrate the sensitivity of a target parameter with respect tochanges on all input variables simultaneously. The typical tornado shape results from arrangingthe input parameters in descending order according to the significance of their impact on theoutput. In order to show what parameters influence the life cycle SVC most considerably, tornadocharts were generated for both SH and MLH engines. Therefore, it was determined how adeviation of -10% to +10% around the base values from tab. 4.8 affect the life cycle SVC. Thethree-spool factor was excluded from this consideration, since it only provides two discrete steps.The three discrete steps of the environment factor on the other hand, happen to fit in the chosenpattern. The EF for hot&dry environments serves as base value, while the two remaining stepsroughly relate to the ±10% deviation, that was applied for the continuous input parameters.The results are illustrated in the following tornado charts.


SH Engines

1000

Base Result: 12.657

Name Low High Delta +

Derate 12.882 12.522 0.360

Weight 12.891 12.518 0.373

EF 11.755 13.558 1.803

Thrust 11.527 13.799 2.272

EFH:EFC 13.897 11.336 2.561

- 10% + 10%

11.0 11.5 12.0 12.5 13.0 13.5 14.0

Derate

Weight

EF

Thrust

EFH:EFC

LCC - Sensitivity - SH Engines

+ 10%

- 10%

11.0 11.5 12.0 12.5 13.0 13.5 14.0

Derate

Weight

EF

Thrust

EFH:EFC

Life Cycle Shop Visit Costs [USD million]

LCC - Sensitivity - SH Engines

+ 10%

- 10%

Figure 4.3: Tornado chart on the sensitivity of the life cycle SVC of SH engines

Short-Haul Engines For SH engines the model indicates that especially the average flight time, thethrust and the environment factor have a significant impact on the accumulated SVC throughoutthe life cycle, while the weight and the derate play a comparable minor role.

Base Result: 30.670

Name Low High Delta

Derate 31.021 30.469 0.552

Weight 29.870 31.629 1.759

EFH:EFC 32.248 29.255 2.993

EF 28.388 32.952 4.564

Thrust 28.130 33.195 5.065

28,0 29,0 30,0 31,0 32,0 33,0 34,0

Derate

Weight

EFH:EFC

EF

Thrust


LCC - Sensitivity - MLH Engines

+ 10%

- 10%

28,0 29,0 30,0 31,0 32,0 33,0 34,0

Derate

Weight

EFH:EFC

EF

Thrust


LCC - Sensitivity - MLH Engines

+ 10%

- 10%

Figure 4.4: Tornado chart on the sensitivity of the life cycle SVC of MLH engines

Medium-Long-Haul Engines The tornado chart for the MLH engines reveals a generally similarpicture. However, the effect of the average flight time is considerably less severe, while changeson the engine weight gained in significance compared to SH engines. The explanation for this isthat the chosen base flight time of EFH:EFC = 6.0 is rather a long-hong-haul route. In long-hauloperations, slight changes on the average flight time generally have a less significant impact onthe engine maintenance (see severity curve in appendix D.2).

5 Implementation into existing LCC-Tool

For the implementation of the model, the engine maintenance was sourced out into a dedicatedfunction that is called in the main executive m-file lccmain.m of the LCC-tool. Up to thispoint, the developed LCC-tool required a fixed predefined number of checks. These checksincluded the engine shop visits, while it was assumed that engines generally have three shopvisits in their life cycle. As a result of this thesis, it became obvious that this approach notnecessarily reflects the reality. The number of shop visits can range significantly dependingon the achieved shop visit intervals and on the total flight hours of the proposed life cycle.Therefore, the programme structure has to be modified to enable a flexible number of checks.The existing structure dictates that the number of total maintenance checks has to be definedprior to the utilization module lccmaintutil. Since the generated engine maintenance modulelccmaintengine determines the expected intervals and thus the required number of SVs, it hasto be executed before the call of lccmaintutil. The outsourcing into a dedicated function wasdone to concentrate the contribution of this thesis to the existing LLC-tool in one central place.The objective was to change the existing surrounding structure as little as possible. One mainrequirement for the implementation is that the new engine maintenance module enables both theestimation of shop visits according to the developed model and the predefinition of shop visitsinformation extracted from available sources. In the following sections, the implementation ofthe key functionalities of the developed maintenance module is briefly described. However, itshould be noted that a complete understanding of the subsequent explanations requires basicknowledge about the existing programme sequence. In case of unclarity, it is also referred to thecommented programme code.

5.1 Function Definition and Input Modification

The generated function as it is implemented in the global executive m-file is defined as follows:

[Maint] = lccmaintengine(Aircraft,...

Routes,...

Maint,...

General,...

File,...

CostTechFactor);

The various input parameter necessary for the this function are subsequently briefly described:

• Aircraft Holds aircraft specifications as they are defined in lcc_frame_in_xxx.xml

including key engine specifications

68

5.1 Function Definition and Input Modification 69

• Routes Contains all route informations, including flight time, derate and environment

• Maint Maint includes all maintenance data for the aircraft and engine. Thus, it contains allengine specifications as well as ESV intervals and costs, as they are set in lcc_maintineng.

xml. It holds the key parameters that are applied to account for all maintenance events inthe following programme sequence.

• General Contains general information, like life cycle length and basic utilization parameterssuch as, number of curfew hours or flight days per week.

• File Includes the information of the loaded xml files and controls what files are loaded.

• CostTechFactor Incldes all cost technology factors.

The existing global programme structure accesses the Maint struct for the following reflectionof the aircraft maintenance as part of the AC life cycle. Hence, the existing Maint struct,which is also an input of the developed function, is modified and represents the only output.The performed modifications result from the outcome of the implemented engine maintenancemodel. The model implementation is based on the assumption that the utilization of the aircraftis constant throughout the life cycle. That means all input parameter that are defined in theroutes branch of the lcc_frame_in_xxx.xml are assumed to be constant for the entire lifecycle.The engine maintenance function provides different control settings, which allow the user toinfluence the processing of the functions. These settings were implemented with simple true-false queries. They are summarized as follows:

Variable Name true “1” false “0” struct xml file

Maint.ctrl.engMroType

engine maintenance modelestimates SV intervals andcosts according to input pa-rameters

predefined SV intervals andcosts from engineMroFileare applied for life cyclemodeling

Maint lcc_frame_in_

File.input.engInputSource

engine specificationsfor maintenance estima-tion are extracted fromengineMroFile

engine specificationsfor maintenance estima-tion are extracted fromlcc_frame_in_xxx

File lcc_frame_in_

Maint.ctrl.spareEngineCostspare engine costs are con-sidered and included in themaintenance costs

spare engine costs are NOTconsidered Maint lcc_frame_in_

Maint.ctrl.engLastSvTypelast shop visit is performedfor a targeted remainingTOW until end of life cycle

last shop visit is fully per-formed regardless of ex-pected end of life cycle

Maint lcc_frame_in_

Table 5.1: Summary of the main control settings of lccmaintengine

All these control setting variables have been added to the main xml input file lcc_frame_in_

xxx. In addition, several input parameters that are required for executing the developed enginemaintenance model were not yet defined and respectively had to be added to the input xml files.This includes, the the derate, environment, engine application, number of spools as well as shopvisit duration, spare engine leasing rate an a few technology factors. Since the developed functionallows to extract the required engine specifications from two different input xml files, both the

5.2 Estimating the Shop Visits 70

engineMroFile and the lcc_frame_in_xxx have been modified. All additional modificationson each xml file are summarized in the tables 5.2 and 5.3.

Added Parameters in lcc_frame_in_xxx.xml

Parameter Description xml branch Input into LCC-Tool

deRate [%] average derate flown, canrange between [0.0-0.2] Routes Routes.deRate

environment accounts for the environmentcondition, [1.0 1.1 1.2] Routes Routes.environment

nSpools number of engine spools, [2 3] Aircraft Aircraft.engine.nSpools

range accounts for the aircraft’s de-sign application [’SH’ ’MLH’] Aircraft Aircraft.engine.range

leasingRate[USD/day]

expected leasing rate forspare engines Aircraft Aircraft.engine.leasingRate

sVduration[days]

expected average shop visitduration Aircraft Aircraft.engine.sVduration

Table 5.2: Summary of the lcc_frame_in_xxx.xml modification

Added Parameters in engineMroFile.xml

Parameter Description xml branch Input into LCC-Tool

range accounts for the aircraft’s de-sign application [’SH’ ’MLH’] range Maint.engineMaint.range

leasingRate[USD/day]

expected leasing rate forspare engines leasingRate Maint.engineMaint.leasingRate

sVduration[days]

expected average shop visitduration sVduration Maint.engineMaint.sVduration

Table 5.3: Summary of the engineMroFile modification

5.2 Estimating the Shop Visits

The programme code is generally separated into two main branches. Depending on the settingof Maint.ctrl.engMroType either one of them is active. Most of the executed operationsare dedicated to their respective branch. Thus, they appear twice in slightly modified fromin the programme code, while only the operations of the active branch are executed. Incase Maint.ctrl.engMroType is set to ’0’, the function estimates the shop visit costs andintervals according the developed engine maintenance model. Thus, this first branch requires theimplementation of the maintenance model structure as developed in 4.1. Therefore, each of theseven CERs as presented in 3.3 was simply implemented as a dedicated function. The LLP costCER is for instance carried out by the follwoing function:

function [LLP_cost] = LLP_function(thrust,weight)

%% WHAT DOES IT DO?

%this function generates the Base LLP costs [dollar/EFC] of the engine

% INPUT: thrust and weight of analysed engine

% OUTPUT: LLP costs [dollar/EFC]

5.2 Estimating the Shop Visits 71

% Author: Ralf Seemann

% Date: 11.08.2010

%% PREDICTION FUNCTION

LLP_cost = −115.31326 ...

+0.0194512.*weight ...

+0.0031206.*thrust ...

+(weight−8608.78125).*((weight−8608.78125).*2.69234e−6);

Analogue to the developed CERs, the severity and time & material curves are also reflected bydedicated functions. While the TMF curves can be each implemented by one single vector (seetables 4.1 and 4.2), the severity curves have to be represented by matrices. According to the inputparameters the output value is then interpolated between the values defined in the curve vectoror multiple curve matrix (see appendix D). The interpolation between two values simultaneouslyas required for the severity matrices was achieved by the matlab function interp2. An exampleprogramme code for the implementation of the severity curves is given below:

function [severity_factor]= SM_severity_value(FcToFH,DeRate)

%% WHAT DOES IT DO?

%this function generates the severity matrix for short−haul engines

%and gives out the respective severity factor according to the input

% INPUT: FC:FH Ratio, De−Rate% OUTPUT: severity_factor

% Author: Ralf Seemann

% Date: 11.08.2010

% Severity Matrix

%rows = [0.5 1 1.5 1.9 2.5 3 4 5 6 ]−−−> FC:FH

%columns = [0% 5% 10% 15% 20%] −−−> De−Rateseverity_matrix = [2.8 2.6 2.4 2.28 2.16

2.10 1.93 1.75 1.65 1.54

1.62 1.47 1.32 1.23 1.14

1.20 1.10 1.00 0.94 0.88

1.06 1.11 0.88 0.83 0.77

0.96 0.88 0.80 0.75 0.71

0.85 0.78 0.71 0.66 0.62

0.78 0.72 0.66 0.62 0.59

0.74 0.69 0.63 0.60 0.56];

% defining derate and FcToFH steps

derate_vec = [0 5 10 15 20];

FcToFH_vec = [0.5 1 1.5 1.9 2.5 3 4 5 6];

%% interpolation using interp2

severity_factor = interp2(derate_vec,FcToFH_vec,severity_matrix,DeRate,FcToFH);

5.3 Processing the Predefined Shop Visits 72

With the four effect curve functions the total number of generated functions sums up to eleven.The remaining effects do not have to be reflected by dedicated functions, since they result directlyfrom the input. The model branch that is active when the SV intervals and costs are supposedto be estimated, calls these functions and performs all necessary conversions as dictated by themodel structure seen in figure 4.2. The implementation of the developed model structure is alsorepresented by the example calculation in 4.2. The input parameters for the CER and effectfunctions are extracted either from lcc_frame_in_xxx.xml or from engineMroFile, dependingon the setting of File.input.engInputSource.

5.3 Processing the Predefined Shop Visits

The second main branch of the programme code is active when Maint.ctrl.engMroType is setto ’1’. In this case, the shop visits informations are extracted from the engineMroFile, wherethey have been predefined. The following estimation of the required number of shop visits, theconsideration of spare engine costs as well as the definition of the last shop visits and the outputgeneration are performed in each branch separately. For estimating the number of requiredshop visits, the average mature shop visit interval is calculated based on the defined shop visitintervals extracted from engineMroFile. Therefore, the intervals of all shop visits following thefirst one are averaged. This relates to the general assumption that engines reach maturity aftertheir first shop visit. The following programme parts are defined in each of the two branches.

5.4 Consideration of Spare Engine Costs

Depending on the setting of Maint.ctrl.spareEngineCost, this programm part either estimatesthe costs for spare engines according to eq. (4.1) or it sets them to zero. The spare engine costsare then added to the shop visits costs during the output generation.

5.5 Estimation of Required Shop Visit Number

The objective was to implement a flexible total number shop visits. Therefore, the number ofnecessary shop visits throughout the life cycle has to be estimated. In order to achieve this, onehas to estimate the expected total flight hours during the life cycle. The exact total life cycleFH are not known before the utilization function lccmaintutil has been executed. However,the number of shop visits has to be fixed for running the utilization function. This problem wassolved by simplifying the calculation of the total life cycle FH (LCFH) . Therefore, the routinefor calculating the maximum possible FH during the LC was adopted from the lccmaintutil

function. The reduction of this maximum number of flight hours due to the various maintenanceevents was neglected. This calculation is one of the few programme parts that is executed globally,since it is applied for both programme lines.

5.6 Definition of Last Shop Visits 73

%% ESTIMATION OF YEARLY UTILIZATION PRIOR TO THE UTILIZATION MAINT MODULE

% this is necessary for the estimation of how many SV are necessary

opsHoursWeek_h = General.opsDaysWeek_d*(24−General.curfewHoursDay_h); % [h]

weeklyFC = (Routes.relativefrequency(:) .* opsHoursWeek_h) ./ Routes.cycleTime_h(:);

weeklyFH = weeklyFC.* fcToFh'; %fcToFh = flight_time

weeklyFC = sum(weeklyFC); %number of FCs per week

weeklyFH = sum(weeklyFH); %number of FH per week

%total flight hours of engine life with the factor 0.98 for reducing total

%flight hours due to maintenance events (reduced available FHs)

lcFH = 52*weeklyFH*yearsInService*0.98;

The following estimation of the shop visit number is dedicated to the respective branch. Dependingon the defined or estimated first and mature SV intervals, the number of required SVs is calculatedthrough a while loop. Noteworthy is that this loop only calculates the number of full shop visits.This is explained by the following example. The first SV interval is assumed to be 15,000EFH,mature intervals equal 10,000EFH and the LCFH equals 48,000EFH. In this calculation, thethird shop visit would take place after:

3rdSVEFH = IntervalFR + IntervalMR + IntervalMR

= 15,000 + 10,000 + 10,000 = 35,000 EFH (5.1)

In this pattern the fourth SV would take place after an accumulated flight time of 45.000EFH.However, since the life cycle ends already after 48,000EFH, it is not necessary to perform afull fourth shop visit that would enable the engine to remain on-wing for another 10,000EFHresulting in accumulated 55,000EFH. In this case, the loop gives out that the engine requires ’3’full shop visits during the proposed life cycle. The last shop visit is considered separately in adistinct programme part.

5.6 Definition of Last Shop Visits

The consideration of the last shop visit relates to the setting of Maint.ctrl.engLastSvType. Ifit is set to ’1’ the programme determines the remaining TOW between last shop visit and end oflife cycle. In the example above, this relates to:

remainingTOW = LCFH − 4thSVEFH = 48,000− 45,000 = 3,000 EFH

Thus, the last shop visit would have to restore the engine to a level that it can remain on-wingfor another 3.000EFH. Therefore, the last shop visit is considered as targeted SV with reducedworkscopes. The incurred costs for this targeted SV are calculated with the mature shop visitcost per EFH (see eq. (4.18)) multiplied with the remainingTOW . If it is assumed that amature SV in the previous example costs 3milUSD, then the matre SVC per EFH yield to:

5.7 Output Generation 74

SV CMR,EFH = SV CMR

IntervalMR= 3,000,000

10,000 = 300 USD

EFH

Hence, the cost for the last shop visit as indicated in the example result in:

SV C last = SV CMR,EFH · remainingTOW = 300 · 3,000 = 900,000 USD

This calculation is performed in both branches with the respective interval and cost data resultingeither from the implemented engine maintenance model or from the predefined shop visit data.In case Maint.ctrl.engLastSvType is set to ’0’, the programme handles the last shop visit likeall previous SVs as full shop visit regardless of the expected remaining TOW until the end of thelife cycle. This setting should be preferred when selling the engine on the surplus market afterthe end of the aircraft’s life cycle is considered.

5.7 Output Generation

In the existing programme structure, the shop visit intervals and costs are read from theengineMroFile xml file and then written in Maint.engineMaint.shopVisit according to thexml structure. This data is then applied to determine the utilization and to define the maintenanceevent costs. The objective was to keep this structure. Therefore, the programmed enginemaintenance module simply overwrites the Maint.engineMaint.shopVisit entries according tothe outcome of the engine maintenance function. Hence, the output has the exact same structureas it is defined in in the engineMroFile xml file, with the addition of a field for the spare enginecosts. This enables an uncomplicated implementation of the developed function into the globalprogramme sequence.The following maintenance cost function lccmaintcost classifies all check expenses into fourcategories. The cost elements from the modified Maint.engineMaint.shopVisit variable areallocated as follows:

Allocation of SVC on CheckExpenses

manhours -

materialcost LLP costs

fixcost restoration costs + spare engine costs

laborcost -

Table 5.4: Reflection of the engine maintenance costs in the lccmaintcost function

This allocation enables a differentiated adjustment of the restoration cost and the LLP costthrough the existing technology factors for each CheckExpenses category.

6 Summary and Conclusion

The objectives of this work were to review the literature on maintenance of commercial jet enginesand based on that, to develop a qualitative model that estimates engine shop visit costs andintervals depending on the major influence factors on engine maintenance. Furthermore, it wasintended to implement this model into the existing maintenance module of the LCC simulationtool.

After building up a comprehensive review on the prevailing concepts of engine maintenance, costestimating relationships (CER) regarding the engine shop visit costs and intervals were developedusing the methodological approach described in the NASA Cost Estimating Handbook [NAS08].Therefore, a database that contains numerous current engine model variants and their shop visitintervals and costs was assembled from an extensive review of the operator & owner guides ofthe Aircraft Commerce magazine archive. However, important effects like the environmentalconditions and the operational severity have been normalized for the database assembly, whichled to CERs that do not reflect the influence of these factors. Therefore, the developed CERswere complemented by a subsequent effect-module that adjusts the results of the CERs accordingto the severity of the engine’s operation. The final model relates to six different input parameters:engine thrust, dry weight, number of spools, average flight length, applied derate and the presentenvironment. Since the literature review and the assembled database indicated that short-hauloperated engines generally exhibit different maintenance characteristics than engines that areoperated on medium-long-haul routes, the model was split into two separate paths, each dedicatedto one of these engine applications. In addition the model distinguishes between first-run andmature-run shop visits to account for the generally longer intervals and lower maintenance costper EFH of new engines compared to engines that reached maturity.

The resulting model was then tested for its plausibility by comparing the model results withavailable cost and interval estimations from AeroStrategy. The conclusion of these plausibilitytests were that the general trend of the developed model and the Aerostrategy estimationscoincide. However, the AeroStrategy estimates for new generation engines tend to lie below thepredicted values of the model. This was not unexpected, since the past has shown that newergenerations engines generally achieve longer intervals and lower maintenance costs per EFH thanthe previous generation. Since the database assembly was limited to engines that have been inoperation for the last two decades, the developed model reflects the current generation enginesbest. The problem is that there is no reliable data on the average intervals and costs for thenewest engine generation. However, with these information available one could determine atechnology factor that adjusts the model results and enables a better forecast also for the these

75

6 Summary and Conclusion 76

new engines. The basic engine maintenance characteristics are assumed to remain constant alsowith newer generation engines. Therefore, the applied CERs and adjustment parameters couldbe also replaced with newly developed relationships that are based on available data for thenewer generation engines, while the rest of the model structure could remain unchanged.

The developed model was subsequently implemented into the LCC simulation tool as an inde-pendent module. Therefore, the engine maintenance was excluded from the existing maintenancemodule. This ensures that the contribution of this thesis to the LCC-tool is clearly separated andit enabled the consideration of a flexible number of shop visits, while the previous maintenancemodule relied on a fixed number of shop visits. Therefore, the new engine maintenance moduleestimates the anticipated number of shop visits of the life cycle depending on the utilizationinput and the estimated shop visit intervals. Since the developed engine MRO model requires afew new input parameters that have not been included in the original input files, the xml inputfiles have been modified accordingly. The existing global structure of the LLC simulation toolremained unchanged. The new module keeps the functionality that the shop visit intervals andcosts can be predefined if known from reliable data sources. The problem is that the actual shopvisit costs and intervals heavily depend on the engine’s operational severity. That means, theapplied predefined shop visit estimations have to relate to the utilization defined in the input file.This applies especially for the flight time as major influence factor. Alternatively, the availablemaintenance data could also be adjusted with the average severity curves established in theframework of this thesis.

Bibliography

[AAE04] AAE 451 - Aircraft Design: Method for Calculating Direct Operat-ing Cost (2004), URL http://www.soton.ac.uk/~jps7/Aircraft%20Design%

20Resources/Cost%20data/aircraft%20operating%20cost%20equations.pdf

[Ack10] Ackert. S: Engine Maintenance Concepts for Financiers: Elements of Engine ShopMaintenance Costs (2010)

[ACT08] ACT: Keep on running - engine maintenance and cost reduction. AIRCRAFT TECH-NOLOGY (Jun/Jul2008), (94): S. 52–60

[AEG00] AEG Working Group: Aircraft Evaluation Guidelines: Ref:V62/CMP/SSA/16700C, issue 2, rev. 2 Aufl. (August 2000)

[Air67] Air Transport Association of America: Standard Methode of Estimatingcomparative direct Operating Cost of Turbine Powered Transport Airplanes (1967)

[Air99] Aircraft Commerce: Fine tuning engine maintenance costs. AIRCRAFT COM-MERCE (Jul/Aug 1999), (6): S. 42–48

[Air00] Aircraft Commerce: Ageing CF6-50 delivers acceptable maintenance costs: Main-tenance & Engineering. AIRCRAFT COMMERCE (Nov/Dec 2000), (14): S. 18–24

[Air04a] Aircraft Commerce: LLP management for widebody engines. AIRCRAFT COM-MERCE (2004), (39): S. 29–36

[Air04b] Aircraft Commerce: LLP management for short-haul engines. AIRCRAFT COM-MERCE (Apr/May 2004), (34): S. 36–44

[Air04c] Aircraft Commerce: CFM56-3B1/B2/C1 maintenance cost analysis. AIRCRAFTCOMMERCE (Feb/Mar 2004), (33): S. 27–34

[Air05a] Aircraft Commerce: Using ECM to reduce engine maintenance costs. AIRCRAFTCOMMERCE (Aug/Sep 2005), (42): S. 36–41

[Air05b] Aircraft Commerce: CFM56-7B maintenance costs & reserves analysis. AIRCRAFTCOMMERCE (Feb/Mar 2005), (39)

[Air06a] A320 Family Maintenance Analysis & Budget. AIRCRAFT COMMERCE (2006),(44): S. 18–31

[Air06b] Aircraft Commerce: Owner’s & Operator’s Guide: CFM56-3 Series. AIRCRAFTCOMMERCE (Apr/May 2006), (45): S. 9–31

[Air06c] Aircraft Commerce: Systems for Engine health monitoring. AIRCRAFT COM-MERCE (Feb/Mar 2006), (44): S. 53–57

77

Bibliography 78

[Air06d] Aircraft Commerce: CF6-80C2 maintenance analysis & budget. AIRCRAFTCOMMERCE (Oct/Nov 2006), (48): S. 18–30

[Air07a] Aircraft Commerce: The economics of repairing engine parts. AIRCRAFT COM-MERCE (Aug/Sep 2007), (30): S. 33–37

[Air07b] Aircraft Commerce: CFM56-5A/5B series specifications. AIRCRAFT COM-MERCE (Feb/Mar 2007), (50): S. 6–9

[Air08a] Aircraft Commerce: Big engine in-service performance & maintenance. AIRCRAFTCOMMERCE (Aug/Sep 2008), (59): S. 41–53

[Air08b] Aircraft Commerce: V2500 maintenance analysis & budget (Feb/Mar 2008), (56): S.17–31

[Air08c] Aircraft Commerce: CFM56-7B maintenance analysis & budget. AIRCRAFTCOMMERCE (Jun/Jul 2008), (58): S. 18–28

[Air08d] Aircraft Commerce: 777 family maintenance analysis & budget. AIRCRAFTCOMMERCE (Oct/Nov 2008), (60): S. 12–23

[Air09] Aircraft Commerce: Analysing the engine shop visit process. AIRCRAFT COM-MERCE (Feb/Mar 2009), (62): S. 45–52

[Bec09] Beck A.: Engine Maintenance Cost Monitoring (29.10.2009)

[Ben08] Bennett J.: PMA parts penetration and acceptability. THE ENGINE YEARBOOK(2008): S. 74–79

[BJ82] Birkler J.L., Garfinkle J. B.: Development and Production Cost Estimating Rela-tionships for Aircraft Turbine Engines, RAND (1982)

[Bra04] Braeunling, W.: Flugzeugtriebwerke, Springer, Berlin Heidelberg, 2 Aufl. (2004)

[Bru96] Brundick B.: Parametric Cost Estimating Handbook (1996)

[Bur97] Burton M.: Piston Engines Propellers Gas Turbine Engines: The commercial pilot’sstudy manual series (1997)

[Bur10] Burchell B.: Wanted: More On-Wing Wizardry. AVIATION WEEK (Apr 2010)

[Car00] Carsten Mildt: Entwicklung einer Methode zur Abschätzung der Kostenfür die Instandhaltung der Flugzeugzellen und -systeme im kommerziellenFlugverkehr, TU Berlin (2000), URL http://www.dfld.de/Downloads/20000725_

Wartungskosten.pdf

[CFM09] CFM: CFM56 Cost of Ownership (October 2009)

[Chu08] Churchwell P.: Fundamentals of CER Development: Defense Acquisition University(June 2008)

[Cum97] Cumpsty N.A.: Jet Propulsion: A simple guide to the aerodynamic and thermodynamicdesign and performance of jet engines, Cambridge University Press (1997)

Bibliography 79

[Eng10] Engine Yearbook: Labour qualitiy, not labour rates drives maintenance cost opti-mization. THE ENGINE YEARBOOK (2010): S. 52–56

[FAA04] FAA - Federal Aviation Administration: Airplane Flying Handbook: FAA-H-8083-3A (2004)

[FAA09] FAA - Federal Aviation Administration: GUIDANCE MATERIAL FOR AIR-CRAFT ENGINE LIFE-LIMITED PARTS REQUIREMENTS (2009)

[FAA10] FAA - Federal Aviation Administration: Aircraft Certification: Parts Manufac-turer Approval (PMA) (2010), URL http://www.faa.gov/aircraft/air_cert/

design_approvals/pma/

[Gal08] Galorath D.: Parametric Estimating Handbook: 4th Edition (2008)

[GE 08] GE Aviation: Engine Education: Vocabulary (2008), URL http://www.geae.com/

education/vocabulary.html

[Gun95] Gunston B.: The Development of Jet and Turbine Aero Engines, Patrick StephensLimited (1995)

[Hol08] Holland J.: End of the PMA legitimacy debate. AIRCRAFT TECHNOLOGY(Oct/Nov 2008), (96): S. 64–71

[Hue03] Huenecke K.: Jet Engines: Fundamentals of Theory, Design and Operation, AirlifePublishing Ltd (2003)

[Jet08] JetEngine Consulting: Our Consultancy Philosophy (2008), URL http://www.

jetengineconsulting.com/pdf/JEC_Consulting_Philosophy.pdf

[KLM07] KLM: Engine Water Wash at KLM Engineering & Maintenance (2007), URLhttp://www.klm.com/travel/csr_en/images/Engine%20Water%20Wash%20EN_

tcm256-100165.pdf

[Lin08] Linke-Diesinger A.: Systems of Commercial Turbofan Engines: An Introduction toSystems Functions, Springer Verlag, Berlin Heidelberg (2008)

[Lon00] Long J.: Parametric Cost Estimating in the New Millenium (2000)

[Mei05] Meier N.: Jet Engine Specification Database: Civil Turbojet/Turbofan Specifications(2005), URL http://www.jet-engine.net/civtfspec.html

[MM10] Mueller M., Staudacher S.: Probalisitsche Modellierung des Einflusses variierenderUmwelt- und Betriebsbedingungen auf die Triebwerksinstandhaltung (2010)

[NAS08] NASA: Cost Estimating Handbook, http://www.ceh.nasa.gov (2008)

[OCE91] Otis C. E., Vosbury P. A.: Aircraft Gas Turbine Engines of the World: Dictonary ofthe Gas Turbine, IAP, Inc. (1991)

[Pet08] Peter Nolte: Bewertung des Zusammenhangs zwischen globalen Szenarien und demökonomisch getriebenen Flugzeugentwurf: Diplomarbeit (2008)

Bibliography 80

[Rol07] Rolls-Royce Ltd.: Gas turbine technology: Introduction to a jet engine (2007),URL http://www.rolls-royce.com/Images/gasturbines_tcm92-4977.pdf

[Ros90] Roskam J.: Alirplane Design: Part VIII: Airplane Cost Estimation: Design, Develop-ment, Manufacturing and Operating, Lawrence, Kansas (1990)

[Rou07] Roux E.: Turbofan and Turbojet Engines: Database Handbook, Elodie ROUX (2007)

[Rup00] Rupp O.: Instandhaltungskosten bei zivilen Strahltriebwerken (2000)

[Sch98] Scholz D.: DOCsys - A Method to Evaluate Aircraft Systems, Bonn (1998)

[Sch09] Schilling T.: Methoden zur Modellierung von Flugzeuginstandhaltung - Diplomarbeit(2009), URL http://www.tu-harburg.de/ilt/

[TB07] TU-Berlin: DOC-Abschätzungsmethoden (2007), URL http://www.ilr.tu-berlin.

de/LB/

[Tew07] Tewari A.: Atmospheric and Space Flight Dynamics: Modeling and Simulation withMatlab and Simulink, Birkhäuser, Boston Basel Berlin (2007)

[Tre79] Treager I.E.: Aircraft gas Turbine Engine Technology: Second Edition, McGraw-HillBook Company (1979)

[Tur04] Turso J.A.: A Foreign Object Damage Event Detector Data Fusion System forTurbofan Engines: NASA/TM—2004-213192 (2004)

[UCL] UCLA: Probability and statistics, Dissertation, URL http://wiki.stat.ucla.edu/

socr/index.php/Probability_and_statistics_EBook#Two-Way_ANOVA

List of Figures

1.1 Aircraft MRO cost overview [Jet08] . . . . . . . . . . . . . . . . . . . . . . . . . . 1

2.1 Engine core of gas turbine engines . . . . . . . . . . . . . . . . . . . . . . . . . . 42.2 Working cycle of a gas turbine engine . . . . . . . . . . . . . . . . . . . . . . . . 42.3 GEnx-2B - high bypass twin-spool turbofan . . . . . . . . . . . . . . . . . . . . . 62.4 Comparison: two- and three-spool configuration . . . . . . . . . . . . . . . . . . . 72.5 The main modules of a V2500-A5 [Lin08] . . . . . . . . . . . . . . . . . . . . . . 72.6 Correlation between Take-Off EGT and OAT [Air06b] . . . . . . . . . . . . . . . 102.7 Removal causes depending on aircraft operation [Ack10] . . . . . . . . . . . . . . 142.8 Effects of engine wear on the EGT Margin [Ack10] . . . . . . . . . . . . . . . . . 142.9 Trend of EGT margin erosion rates over accumulated EFC . . . . . . . . . . . . 152.10 Engine shop visit process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172.11 Influences of the TOW on the DMC of an engine [Eng10] . . . . . . . . . . . . . 172.12 Two example flight profiles [Ack10] . . . . . . . . . . . . . . . . . . . . . . . . . . 212.13 Shop visit rate and DMC in relation to the flight hour flight cycle ratio . . . . . 222.14 Example severity curve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232.15 Engine maintenance cost breakdown structure . . . . . . . . . . . . . . . . . . . 252.16 Parametric Cost Estimating process steps [NAS08] . . . . . . . . . . . . . . . . . 282.17 Example data points for cost-weight dependency . . . . . . . . . . . . . . . . . . 30

3.1 JMP screening function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413.2 JMP regression results example output . . . . . . . . . . . . . . . . . . . . . . . . 42

4.1 Black box of maintenance model . . . . . . . . . . . . . . . . . . . . . . . . . . . 474.2 Inner structure of the cost estimating model . . . . . . . . . . . . . . . . . . . . . 484.3 Tornado chart on the sensitivity of the life cycle SVC of SH engines . . . . . . . 674.4 Tornado chart on the sensitivity of the life cycle SVC of MLH engines . . . . . . 67

i

List of Tables

2.1 Initial EGTM and mature EGT erosion rates for CFM56-7B variants [Air08c] . . 202.2 Comparison of EMC consideration in different DOC methods . . . . . . . . . . . 272.3 Criteria for the evaluation of regression results . . . . . . . . . . . . . . . . . . . 31

3.1 Summary of cost estimating relationship hypothesis . . . . . . . . . . . . . . . . 333.2 Structure of the collected Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363.3 Determined average SH severity curve for a derate of 10% . . . . . . . . . . . . . 373.4 Determined average MLH severity curve for a derate of 10% . . . . . . . . . . . . 373.5 Summary of DB base conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . 403.6 Summary of the preliminary dependent variables to be analyzed . . . . . . . . . 403.7 Summary of the final dependent variables to be analyzed . . . . . . . . . . . . . 413.8 Defined standard units for the input parameters of the CERs . . . . . . . . . . . 43

4.1 SH Engine Time & Material factor with respect to the flight time . . . . . . . . . 504.2 MLH Engine Time & Material factor with respect to the flight time . . . . . . . 504.3 Environment factors for different environmental conditions . . . . . . . . . . . . . 514.4 Input engine specifications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 524.5 Input engine utilization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 524.6 Summary of CER results for the example input parameters . . . . . . . . . . . . 534.7 Final output for the example input parameters . . . . . . . . . . . . . . . . . . . 554.8 Input parameters and their base values for the sensitivity analysis . . . . . . . . 604.9 SH Engines - Impact of the continuous parameters on the SV Interval . . . . . 614.10 SH Engines - Impact of the continuous parameters on the SV costs per EFH . 624.11 MLH Engines - Impact of the discrete parameters on the direct model output . 634.12 SH Engines - Impact of the continuous parameters on the life cycle SVC . . . . 644.13 MLH Engines - Impact of the continuous parameters on the life cycle SVC . . 654.14 MLH Engines - Impact of the discrete parameters on the life cycle SVC . . . . 66

5.1 Summary of the main control settings of lccmaintengine . . . . . . . . . . . . . 695.2 Summary of the lcc_frame_in_xxx.xml modification . . . . . . . . . . . . . . . 705.3 Summary of the engineMroFile modification . . . . . . . . . . . . . . . . . . . . 705.4 Reflection of the engine maintenance costs in the lccmaintcost function . . . . 74

ii

A Maintenance Costs

A.1 Engine MRO Cost Analysis

24 November 2006 4

��

New Material66%

Used Material

12%

PMA Parts

3%Life Limited Parts

(LLP)19%

HPT aifoils21%

Other airfoils21%

Stationary parts15%Combustor

6%

FAn5%

Life Limited Parts (LLP)4%

Other28%

60%

Material

25%

Parts Repair

12% DAT*

3% Fees

*) Disassembly, Assembly, TestSource: 2007 AeroStrategy Aeroengine Parts Repair & Material Forecast

Operational CostEngine MRO Cost Analysis

US$ 9.3 billion

US$ 3.9 billion

Material

Repairs

A.2 Shop Visit Cost Driver

Überholung führen kann.

Ein weiteres Beispiel für den Einfluß, den dieFluggesellschaften auf die Instandhaltungskostenund insbesondere die On-wing Zeit eines Trieb-werks haben können, betrifft die Reinigungs-prozeduren des Triebwerks. Zur Entfernung vonAblagerungen auf den Verdichter-Schaufeln unddamit zur (teilweisen) Wiederherstellung derTriebwerksleistung kann im Leerlaufbetrieb desTriebwerks Wasser oder auch ein Kohlepulvervorne in das Triebwerk eingespritzt werden.Dieses wäscht einen Teil der Verunreinigungenab, verbessert den Wirkungsgrad des Verdichtersund vermindert durch das so wieder reduzierteTemperaturniveau im Triebwerk auch die Be-lastung im Heißteil.

BILD 3. Einfluß von Schubminderung unddurchschnittlicher Fluglänge auf dieInstandhaltungskosten.

Das Ergebnis einer konsequenten Durchführungsolcher On-wing Maßnahmen ist beispielhaft inBILD 3 dargestellt. Im vorliegenden Fall führteeine Fluggesellschaft ab Anfang 1999 eine interneBestimmung ein, die eine regelmäßige Reinigungder Triebwerke mit Kohlepulver (Coke Cleaning)vorsah. Wie im Bild zu erkennen, führte dies zueiner deutlich erhöhten On-wing Zeit der Trieb-werke. Gleichzeitig veränderten sich die Kostenbei der anschließend erfolgten Überholung abernur unwesentlich.

Sicherlich ist die in BILD 3 dargestellte Verbesse-rung der On-wing Zeit ein extremes Beispiel -begünstigt durch die Tatsache, dass bei dieserFluggesellschaft der Hauptgrund für einen Trieb-werksausbau eine geringe Leistung der Trieb-werke ist. Dennoch bleibt festzuhalten, dass dieLine Maintenance Philosophie einer Fluggesell-schaft die Instandhaltungskosten je Flugstundemaßgeblich beeinflußt.

6. HAUPTKOSTENTREIBER

Die für die Instandhaltung eines Triebwerks

anfallenden Kosten können im wesentlichen indrei Bereiche aufgeteilt werden: Kosten für dieMontage/Demontage des Triebwerks, Reparatur-kosten für Einzelteile und Materialkosten für denErsatz von nicht mehr reparierbaren Bauteilen. DieVerteilung der Kosten auf bestimmte Bauteile undBereiche ist von großer Bedeutung, da sie Rück-schlüsse auf potentiell erreichbare Kostenreduzie-rungen für die einzelnen Bauteile - beispielsweisedurch die Entwicklung neuer Reparaturen - zuläßt.

BILD 4 zeigt beispielhaft die Verteilung derInstandhaltungskosten bei einer Überholung aufdie Bauteile, unterteilt nach Materialkosten (blaugekennzeichnet) und Arbeitskosten (rot bzw. gelbgekennzeichnet). Ausdrücklich sei hier daraufhingewiesen, dass lebensdauerbegrenzte Teile(LLP's) hier nicht mit berücksichtigt sind.

BILD 4. Kostentreiber bei der Überholung vonTriebwerken.

Auffallend ist der große Kostenanteil für Airfoils -insgesamt fast 50% der Überholungskosten.Primär entstehen diese Kosten durch den Ersatzvon nicht mehr reparablen Airfoils aus demBereich der Hochdruckturbine. Dies sind sicherlichmit die am höchsten belasteten Bauteile imgesamten Triebwerk, wobei erschwerend hinzu-kommt, dass beispielsweise Laufschaufeln invielen Fällen laut Vorgabe vom Hersteller auch nureinmal komplett repariert werden können, bevorsie beim nächsten Überholungsereignis dann ggf.durch Neuteile ersetzt werden müssen. BeiNeuteil-Kosten von fast. US-$500000,- für einenSatz HDT Stufe 1 Laufschaufeln (CF6-80C2) istdie hohe Quote dieser Bauteile an den Gesamt-kosten gut nachvollziehbar.

Insgesamt machen die Materialkosten mehr als50% der Überholungskosten aus, ein Faktor, dergerade Überholungsdienstleister dazu anregt,auch für komplizierte Bauteile Reparaturen zuentwickeln und somit die Gesamtkosten für einÜberholungsereignis zu reduzieren. Hintergrundfür die hohen Materialkosten ist auch die Tat-sache, dass die Triebwerkshersteller einen Groß-teil der eigenen Entwicklungskosten inzwischen

-4000

-2000

0

2000

4000

6000

8000

10000

12000

Jul Jan Jul 98 Jan 99 Jul 99

Jan JulRemoval Date

∆-

M

Prior Coke Cleaning

Introduction

After Coke Cleaning

Introduction

∆-MTBSV

Improvement of MTBSV

after Introduction of

regular Coke Cleaning

Disassembly

Assembly 15 %

Airfoils 14 %

Accessories 5 %

Stationary Parts 8 %

Rotating Parts 4 %Combustor 2 %Seals 2 %

others 10 %

Combustor 2,5 %

Cases 5 %

Bearings 2,5 %

Airfoils 30 %

*LLP cost not includedBlue: material costRed: labour cost

[Rup00]

iii

B Database

B.1 Aircraft Commerce Shop Visit Reserves & Intervals Example Table

may be a complete overhaul of allmodules.

“The -7B18/20/22 will have a corerefurbishment without LLP replacement,plus fan and booster refurbishment withLLP replacement,” explains Crawford.This would be at a total accumulatedtime close to 30,000EFC.

The medium-rated -7B24 will have asimilar second shop visit.

Higher rated -7B26 engines that didnot have work done on the LPT at thefirst shop visit would have all majormodules worked on during the secondshop visit, as well as replacement of LPTand fan/LPC LLPs. This would be at atotal accumulated time of 25,000EFC.

“The highest rated -7B27s wouldhave another core refurbishment andworkscope on the LPT, with LLPs beingreplaced in all these modules,” saysCrawford. This raises the issue of thefan/LPC module. Total accumulated timeon-wing at this stage will be 17,000-20,000EFC, and so there would be up to13,000EFC remaining until LLPs in thefan/LPC needed replacing. The fan/LPCmodule could thus be worked on at thethird shop visit.

Engine management These approximate on-wing intervals

and LLP lives strongly influence enginemanagement and shop visit workscopes.Total time to the second shop visit will beclose to 30,000EFC for low-ratedengines, which at an average EFC time of2.0EFH will be near to 60,000EFH. Thiswill be equal to about 20 years ofoperation for most airlines.

The higher-rated -7B26 will have atotal accumulated time at the secondshop visit of 25,000EFC, or about

45,000-50,000EFH. This will be equal to15-17 years of operation for mostairlines’ operations. The highest thrustrated -7B27 will have accumulated about20,000EFC and 38,000EFH at the secondshop visit, after a total time of about 13-15 years’ service.

This indicates that the second shopvisits will not occur for about anotherseven years for the oldest and highestrated engines. It is therefore too early toestimate the intervals to the third shopvisits and their subsequent workscopes.

Shop visit inputs Inputs for shop visits in terms of man-

hours (MH), materials and parts and costof sub-contract repairs for the -7B, can beestimated on the basis of inputs for the -3series.

“Man-hour inputs are expected to besimilar to the -3 for similar shop visitworkscopes, perhaps slightly lower,” saysBeale. “Material costs for a -7B areexpected to be 20-35% more than a -3and the -7B’s cost of sub-contractedrepairs will be 5-15% more than thoseexperienced by the -3.”

Sekinger estimates a corerefurbishment plus some work on theLPT will use 2,700 man-hours (MH).Charged at a labour rate of $70 per MHthis will take total cost to about$140,000. Cost of materials and partswill be about $700,000-800,000, on thebasis of average part scrap rates. “If workis required on the LPT to repair rails onthe inner wall another $200,000 can beadded to the total cost of the shop visit.The cost of sub-contract repairs will be$300,000-350,000. This would take totalcost for the shop visit to $1.15-1.30million, not including the cost of

replacing LLPs. Crawford estimates that a heavier

shop visit that included work on the LPTmodule would add up to another 500MH, $200,000 for materials and partsand another $100,000 for sub-contractrepairs. The same labour rate of $70 perMH would take the total cost for theshop visit to $1.5-1.6 million.

Crawford estimates that a completeoverhaul would use about 3,000MH, upto $1.2 million in materials and parts,and about $500,000 for sub-contractrepairs. This would take total cost for theshop visit to about $1.7 million. Someengine shops estimate that the MH inputsfor these heavier shop visits may be ashigh as 4,000.

LLP amortisation LLP amortisation has to consider

probable intervals to the third shop visit.This is because LLPs replaced at the firstshop visit will be replaced at the thirdshop visit. Their cost should thus beamortised over the combined second andthird removal intervals. Although somemodules have LLPs replaced togetherduring the first or second shop visit, theymay then get LLPs replaced for a secondtime at different shop visits. LLPs for thefan/LPC have a list price of $305,000,LLPs in the HP system a list price of$785,000 and LLPs in the LPT a list priceof $426,000.

It has been assumed here that lowrated -7B18/20/22 engines will have theircore and LPT LLPs replaced at the firstshop visit, after about 18,000EFC. Asecond interval of 11,000EFC and thirdpossible interval of 9,000EFC means theywill be replaced again after about20,000EFC at the third shop visit. The

AIRCRAFT COMMERCE ISSUE NO. 39 • FEBRUARY/MARCH 2005

44 I MAINTENANCE & ENGINEERING

POSSIBLE MANAGEMENT, SHOP VISIT PATTERN & LLP REPLACEMENT TIMING OF CFM56-7B SERIES ENGINES

Removal Interval Accumulated Workscope Cost-$ $/EFC LLP LLP cost LLP Total Total

EFC EFC content replacement $ $/EFC $/EFC $/EFH

-7B27

1st 10,000 10,000 Core 1,200,000 120 - 79 199 111

2nd 7,000-8,000 17,000-18,000 Core & LPT 1,500,000 200 Core & LPT 1,211,000 79 279 155

-7B26

1st 13,000 13,000 Core 1,250,000 96 Core 785,000 91 190 106

2nd 12,000 25,000 fan/LPC & LPT 1,700,000 142 Fan/LPC & LPT 731,000 91 233 130

-7B24

1st 16,000 16,000 Core & LPT 1,550,000 97 Core & LPT 1,211.000 87 184 102

2nd 14,000 30,000 Core & fan/LPC 1,700,000 121 Fan/LPC 305,000 71 192 107

-7B18/20/22

1st 17,000-18,000 17,000-18,000 Core & LPT 1,600,000 92 Core & LPT 1,211,000 78 170 95

2nd 12,000-13,000 28,000-30,000 Core & fan/LPC 1,700,000 136 Fan/LPC 305,000 73 209 116

B.2 Classification of Aircraft Engines

Putting aircraft engines into categories according to their application is rather subjective. Aregional airline may classify a certain route as middle-haul, while an intercontinental operatingairline could classify the same route as short distance. The following table reflects the perspectiveof the author of this thesis. In the framework of this thesis, the engines of the database havebeen divided in short-haul (SH) and medium/long-haul (MLH) engines. The classification isnot based on the absolute range of the aircraft, on which the engines are mounted. Rather itcorresponds to the rough average flight time these aircraft and respectively the engines weredesigned for. However, engines can by all means operate outside their design range. A goodguideline for classifying the engines for the developed model is the common distinction betweenshort-haul aircraft (e.g. A320, Boeing 737, Embraer E-Jets) and wide body aircraft (e.g. A330,A340, Boeing 777/747). From the perspective of the developed model, wide-body aircraft arepowered by typical MLH engines, short-haul aircraft respectively by SH engines.

Short-Haul Medium-Haul Long-Haul< 3FH 3− 6FH > 6FH

iv

B.3 Core Database v

B.3 Core Database

OEM Model

T_TO

[lbf]

Weight

[lb] TWR

EFH:

EFC

MIF

removal

Interval

EFC

Removal

Interval

EFH

LLP

Reserves

[$/EFC]

SV

Reserves

[$EFH]

removal

Interval

EFC

removal

Interval

EFH

LLP

Reserves

[$/EFC]

SV

Reserves

[$EFH] SF

1st

Interval

SVR Adj

1st

cost

Adj

mature

interval

SVR adj

mature

cost

Adj

Mean

LLP Cost

CFM CFM56-3-xxx 18500 4276 4,33 1,40 1,10 18000 25200 88,00 53,90 10000 14000 83,05 96,25 1,30 32760 37,69 18200 74,04 85,53

CFM CFM56-3B1 20000 4276 4,68 1,40 1,10 16000 22400 97,90 60,50 8000 11200 72,60 119,90 1,30 29120 42,31 14560 92,23 85,25

CFM CFM56-3B2 22000 4301 5,12 1,40 1,10 10000 14000 80,30 68,20 6500 9100 80,30 130,90 1,30 18200 47,69 11830 100,69 80,30

CFM CFM56-3C1 23500 4301 5,46 1,40 1,10 7500 10500 93,50 90,20 5000 7000 84,70 141,90 1,30 13650 63,08 9100 109,15 89,10

CFM CFM56-5A1 25000 4995 5,01 1,20 1,09 8000 9600 100,28 178,76 1,46 14016 122,44 100,28

CFM CFM56-5A5 23500 4975 4,72 1,20 1,09 8000 9600 100,28 178,76 1,46 14016 122,44 100,28

CFM CFM56-5B3 32000 5250 6,10 1,50 1,09 7000 10500 119,90 95,92 5000 7500 119,90 171,13 1,22 12810 72,13 9150 140,27 119,90

CFM CFM56-5B5 22000 5250 4,19 1,80 1,09 15000 27000 113,36 80,66 10000 18000 94,83 109,00 1,00 27000 74,00 18000 109,00 104,10

CFM CFM56-5B6 23500 5250 4,48 1,80 1,09 15000 27000 113,36 80,66 10000 18000 94,83 109,00 1,00 27000 74,00 18000 109,00 104,10

CFM CFM56-5B7 27000 5250 5,14 1,80 1,09 14000 25200 113,36 80,66 9000 16200 94,83 109,00 1,00 25200 74,00 16200 109,00 104,10

CFM CFM56-7B18 18500 5216 3,55 1,80 1,18 18000 32400 92,04 71,98 13000 23400 86,14 101,48 1,00 32400 61,00 23400 101,48 89,09

CFM CFM56-7B20 20600 5216 3,95 1,80 1,18 17500 31500 92,04 71,98 13000 23400 86,14 101,48 1,00 31500 61,00 23400 101,48 89,09

CFM CFM56-7B22 22000 5216 4,22 1,80 1,18 17000 30600 92,04 71,98 13000 23400 86,14 101,48 1,00 30600 61,00 23400 101,48 89,09

CFM CFM56-7B24 24000 5216 4,60 1,80 1,18 16000 28800 93,22 75,52 12000 21600 93,22 90,86 1,00 28800 64,00 21600 90,86 93,22

CFM CFM56-7B26 26400 5216 5,06 1,80 1,18 13000 23400 107,38 75,52 11000 19800 107,38 105,02 1,00 23400 64,00 19800 105,02 107,38

CFM CFM56-7B27 27300 5216 5,23 1,80 1,18 10000 18000 93,22 90,86 8000 14400 93,22 142,78 1,00 18000 77,00 14400 142,78 93,22

GE CF34-3B1 9220 1670 5,52 1,16 1,10 15000 17400 50,05 51,70 8000 9280 72,60 103,40 1,50 26100 31,33 13920 68,93 61,33

GE CF34-8E5 13800 2470 5,59 1,28 1,07 8984 11500 70,62 65,27 5859 7500 70,62 142,31 1,39 15984 43,88 10424 102,38 70,62

GE CF34-8E5A1 15000 2470 6,07 1,28 1,07 7422 9500 70,62 79,18 5078 6500 70,62 164,78 1,39 13205 53,24 9035 118,55 70,62

GE CF34-10E5 18285 3700 4,94 1,28 1,07 14062 17999 64,20 50,29 7812 9999 64,20 139,10 1,39 25019 33,81 13899 100,07 64,20

GE CF34-10E6 19000 3700 5,14 1,28 1,07 13281 17000 69,55 53,50 7031 9000 69,55 148,73 1,39 23630 35,97 12510 107,00 69,55

GE CF34-10E7 20300 3700 5,49 1,28 1,07 10156 13000 73,83 57,78 4688 6001 73,83 169,06 1,39 18070 38,85 8341 121,63 73,83

IAE V2522-A5 22000 5230 4,21 1,90 1,01 11500 21850 112,11 67,67 9500 18050 112,11 136,35 1,00 21850 67,00 18050 136,35 112,11

IAE V2524-A5 24000 5139 4,67 1,90 1,01 11000 20900 112,11 70,70 9000 17100 112,11 143,42 1,00 20900 70,00 17100 143,42 112,11

IAE V2527-A5 27000 5139 5,25 1,90 1,01 8750 16625 136,35 94,94 7750 14725 136,35 164,63 1,00 16625 94,00 14725 164,63 136,35

IAE V2530-A5 30000 5139 5,84 1,90 1,01 7400 14060 121,20 104,03 6000 11400 121,20 168,67 1,00 14060 103,00 11400 168,67 121,20

IAE V2533-A5 33000 5139 6,42 1,90 1,01 7000 13300 121,20 109,08 5750 10925 121,20 175,74 1,00 13300 108,00 10925 175,74 121,20

GE CF6-80C2A5 61300 9389 6,53 1,00 1,10 5000 5000 198,00 550,00 1,63 8125 338,46 198,00

GE CF6-80C2A5 61300 9389 6,53 2,00 1,10 4500 9000 220,00 278,30 0,98 8820 283,98 220,00

GE CF6-80C2A3 60200 9360 6,43 3,00 1,10 4000 12000 209,00 199,10 0,80 9624 248,25 209,00

GE CF6-80C2A5 61300 9389 6,53 4,00 1,10 3500 14000 222,20 174,90 0,71 9884 247,73 222,20

GE CF6-80C2A2 53500 9360 5,72 3,00 1,10 6000 18000 198,00 135,30 5000 15000 198,00 176,00 0,80 14436 153,37 12030 219,45 132,00

PW PW4074 77440 14545 5,32 1,50 1,00 10000 15000 514,00 267,00 7500 11250 514,00 441,00 1,22 18285 219,03 13714 361,77 342,67

PW PW4077 79960 14545 5,50 1,50 1,00 9333 14000 514,00 286,00 6800 10200 514,00 410,00 1,22 17065 234,62 12434 336,34 342,67

PW PW4158 58000 9213 6,30 1,50 1,05 3300 4950 242,55 420,00 1,22 6034 344,54 242,55

GE CF6-80C2B6 60800 9670 6,29 6,00 1,10 2500 15000 253,00 163,90 1,00 15000 163,90 253,00

GE CF6-80C2B1F 58090 9499 6,12 7,00 1,10 2400 16800 209,00 146,30 0,93 15540 158,16 209,00

GE CF6-80C2D1F 61960 9850 6,29 7,00 1,10 2200 15400 213,40 168,30 0,93 14245 181,95 213,40

PW PW4052 52200 9213 5,67 3,00 1,05 4250 12750 157,50 159,60 1,40 17850 114,00 157,50

PW PW4056 56750 9213 6,16 8,00 1,05 1900 15200 197,40 138,60 0,88 13376 157,50 197,40

PW PW4060 60000 9213 6,51 7,00 1,05 2100 14700 252,00 139,65 0,93 13598 150,97 252,00

PW PW4062 62000 9213 6,73 7,00 1,05 2100 14700 252,00 139,65 0,93 13598 150,97 252,00

PW PW4168 68600 12400 5,53 7,00 1,00 2857 19999 418,00 135,00 2150 15050 418,00 247,00 0,93 18499 145,95 13921 267,03 278,67

First Removal Mature Removal Normalized DataSpecifications


PW PW4090 91790 15584 5,89 7,00 1,00 2643 18501 661,00 243,00 2000 14000 661,00 366,00 0,93 17113 262,70 12950 395,68 440,67

PW PW4098 99040 16500 6,00 7,00 1,00 2643 18501 661,00 243,00 2000 14000 661,00 366,00 0,93 17113 262,70 12950 395,68 440,67

PW PW2037 36600 7185 5,09 3,00 1,00 6667 20001 200,00 100,00 5167 15501 200,00 181,00 1,30 26001 76,92 20151 139,23 133,33

PW PW2040 40100 7185 5,58 3,00 1,00 6667 20001 200,00 110,00 4667 14001 200,00 200,00 1,30 26001 84,62 18201 153,85 133,33

GE CF6-80E1A2 67500 11162 6,05 3,00 1,00 4500 13500 370,00 170,00 3400 10200 370,00 284,00 1,40 18900 121,43 14280 202,86 246,67

GE CF6-80E1A2 67500 11162 6,05 6,00 1,00 2833 16998 343,00 147,00 2167 13002 343,00 235,00 1,00 16998 147,00 13002 235,00 228,67

GE CF6-80E1A2 67500 11162 6,05 8,00 1,00 2250 18000 343,00 172,00 1812 14496 343,00 217,00 0,88 15840 195,45 12756 246,59 228,67

GE GE90-85B 84700 15596 5,43 6,00 1,00 3000 18000 700,00 211,00 2417 14502 700,00 297,00 1,00 18000 211,00 14502 297,00 466,67

GE GE90-110B 110000 18260 6,02 8,00 1,00 2450 19600 807,00 215,00 2100 16800 807,00 278,00 0,88 17248 244,32 14784 315,91 538,00

GE GE90-110B 110000 18260 6,02 10,00 1,00 2150 21500 807,00 200,00 1850 18500 807,00 240,00 0,84 18060 238,10 15540 285,71 538,00

PW PW4168 68600 12400 5,53 6,00 1,00 3000 18000 375,00 140,00 2333 13998 375,00 243,00 1,00 18000 140,00 13998 243,00 250,00

PW PW4168 68600 12400 5,53 3,00 1,00 4500 13500 400,00 178,00 3500 10500 400,00 292,00 1,40 18900 127,14 14700 208,57 266,67

PW PW4168 68600 12400 5,53 8,00 1,00 2750 22000 375,00 123,00 2100 16800 375,00 233,00 0,88 19360 139,77 14784 264,77 250,00

PW PW4077 79960 14545 5,50 8,00 1,00 2750 22000 389,00 173,00 2250 18000 389,00 244,00 0,88 19360 196,59 15840 277,27 259,33

PW PW4090 91790 15584 5,89 8,00 1,00 2500 20000 556,00 200,00 2125 17000 556,00 279,00 0,88 17600 227,27 14960 317,05 370,67

RR RB.211-535E4 40100 7264 5,52 3,00 1,00 6000 18000 185,00 190,00 1,40 25200 135,71 185,00

RR Trent 772-60 71100 10550 6,74 3,00 1,00 5333 15999 330,00 250,00 4000 12000 330,00 367,00 1,40 22399 178,57 16800 262,14 220,00

RR Trent 772-60 71100 10550 6,74 6,00 1,00 4000 24000 330,00 215,00 2667 16002 330,00 275,00 1,00 24000 215,00 16002 275,00 220,00

RR Trent 772-60 71100 10550 6,74 8,00 1,00 3750 30000 330,00 191,00 2500 20000 330,00 220,00 0,88 26400 217,05 17600 250,00 220,00

RR Trent 884-17 86910 13100 6,63 8,00 1,00 3000 24000 429,00 196,00 2375 19000 429,00 255,00 0,88 21120 222,73 16720 289,77 286,00

RR Trent 895-17 95000 13186 7,20 8,00 1,00 3000 24000 698,00 200,00 2375 19000 698,00 263,00 0,88 21120 227,27 16720 298,86 465,33

PW JT9D-7R4E 50000 8905 5,61 3,50 1,29 1857 6500 144,48 366,36 144,48

PW JT9D-7J 50000 8850 5,65 7,00 1,29 1350 9450 144,48 237,36 144,48

PW JT9D-7Q 53000 9295 5,70 5,00 1,29 1600 8000 144,48 335,40 144,48

PW JT9D-7Q 53000 9295 5,70 8,00 1,29 1150 9200 199,95 245,10 199,95

PW JT9D-7R4G2 54750 9135 5,99 6,00 1,29 1500 9000 201,24 279,93 201,24

PW JT9D-7R4G2 54750 9135 5,99 8,00 1,29 1250 10000 216,72 251,55 216,72

GE CF6-50C2 52500 8731 6,01 5,00 1,29 1500 7500 185,76 258,00 185,76

GE CF6-50E2 52500 8768 5,99 6,00 1,29 1500 9000 135,45 236,07 135,45

GE CF6-50C2 52500 8731 6,01 2,00 1,29 2500 5000 180,60 451,50 180,60

PW PW4158 58000 9213 6,30 3,00 1,29 3300 9900 189,63 238,65 189,63

PW PW4060 60000 9213 6,51 6,50 1,29 1850 12025 243,81 180,60 243,81

PW PW4062 62000 9213 6,73 7,00 1,29 1700 11900 261,87 180,60 261,87

PW PW4056 56750 9213 6,16 8,00 1,29 1750 14000 175,44 161,25 175,44

GE CF6-80C2A2 53500 9360 5,72 1,00 1,29 5000 5000 258,00 490,20 258,00

GE CF6-80C2A3 60200 9360 6,43 3,00 1,29 2500 7500 220,59 309,60 220,59

GE CF6-80C2B1F 58090 9499 6,12 7,50 1,29 1700 12750 223,17 168,99 223,17

GE CF6-80C2D1F 61960 9850 6,29 7,00 1,29 1500 10500 216,72 196,08 216,72

Short-Haul Three-Spool

Medium-Long-Haul Old and Newly marketed Engines

C Regression Analysis

C.1 First Interval SH EnginesSE_Data- Fit Least Squares Page 1 of 2

15000

20000

25000

30000

35000

1st

Inte

rva

l

ad

j A

ctu

al

10000 20000 25000 30000

1st Interval adj Predicted

P<.0001 RSq=0,77 RMSE=3247,8


RSquare

RSquare Adj


Mean of Response


0,769685

0,740895

3247,839

22162,07

28

Summary of Fit

Model

Error

C. Total

Source

3

24

27

DF

846039131

253162938

1099202069

Sum of

Squares

282013044

10548456

Mean Square

26,7350

F Ratio

<,0001*

Prob > F


Intercept

TWR

weight

(weight-5407)*(weight-5407)

Term

68466,325

-8267,819

-1,004437

0,0001212

Estimate

5575,065

942,4547

0,435151

5,961e-5

Std Error

12,28

-8,77

-2,31

2,03

t Ratio

<,0001*

<,0001*

0,0299*

0,0531

Prob>|t|

Parameter Estimates

TWR

weight

weight*weight

Source

1

1

1

Nparm

1

1

1

DF

811801410

56202317

43648191

Sum of

Squares

76,9593

5,3280

4,1379

F Ratio

<,0001*

0,0299*

0,0531

Prob > F

Effect Tests

-8000

-6000

-4000

-2000

0

2000

4000

6000

1st

Inte

rva

l

ad

j R

esid

ua

l

10000 20000 25000 30000

1st Interval adj Predicted


Whole Model

Response 1st Interval adj

Prediction FunctionInterval first,SH = 68466.325284 − 8267.81904 · TWR − 1.00444 · weight

+ (weight− 5407) · [(weight− 5407) · 0.00012125]

vi

C.2 Mature Interval SH Engines vii

C.2 Mature Interval SH EnginesSE_Data- Fit Least Squares Page 1 of 1

5000

10000

15000

20000

25000

ma

ture

In

terv

al

ad

j A

ctu

al

5000 10000 15000 20000 25000

mature Interval adj Predicted

P<.0001 RSq=0,79 RMSE=2112,5


RSquare

RSquare Adj


Mean of Response


0,790264

0,783909

2112,513

14305,73

35

Summary of Fit

Model

Error

C. Total

Source

1

33

34

DF

554897670

147269489

702167159

Sum of

Squares

554897670

4462711,8

Mean Square

124,3409

F Ratio

<,0001*

Prob > F


Lack Of Fit

Pure Error

Total Error

Source

31

2

33

DF

146461838

807651

147269489

Sum of

Squares

4724575

403825

Mean Square 11,6996

F Ratio

0,0817

Prob > F

0,9988

Max RSq

Lack Of Fit

Intercept

TWR

Term

40684,376

-5022,812

Estimate

2392,421

450,443

Std Error

17,01

-11,15

t Ratio

<,0001*

<,0001*

Prob>|t|

Parameter Estimates

TWR

Source

1

Nparm

1

DF

554897670

Sum of

Squares

124,3409

F Ratio

<,0001*

Prob > F

Effect Tests

-5000

-4000

-3000

-2000

-1000

0

1000

2000

3000

4000

5000

ma

ture

In

terv

al

ad

j R

esid

ua

l

5000 10000 15000 20000 25000

mature Interval adj Predicted


Whole Model

5000

10000

15000

20000

25000

ma

ture

In

terv

al a

dj

Le

ve

rag

e R

esid

ua

ls

3,5 4,0 4,5 5,0 5,5 6,0 6,5 7,0

TWR Leverage, P<,0001

Leverage Plot

TWR

Response mature Interval adj

Prediction FunctionIntervalmature,SH = 40684.37633 − 5022.8116 · TWR

C.3 First Interval MLH Engines viii

C.3 First Interval MLH EnginesBE_Data_wo_3s- Fit Least Squares Page 1 of 2

14000

16000

18000

20000

22000

24000

26000

28000

1st

inte

rva

l

ad

j A

ctu

al

14000 18000 22000 26000

1st interval adj Predicted

P<.0001 RSq=0,91 RMSE=969,3


RSquare

RSquare Adj


Mean of Response


0,911592

0,88949

969,2997

18937,16

16

Summary of Fit

Model

Error

C. Total

Source

3

12

15

DF

116253879

11274503

127528381

Sum of

Squares

38751293

939541,89

Mean Square

41,2449

F Ratio

<,0001*

Prob > F


Lack Of Fit

Pure Error

Total Error

Source

5

7

12

DF

5046845

6227658

11274503

Sum of

Squares

1009369

889665

Mean Square 1,1345

F Ratio

0,4233

Prob > F

0,9512

Max RSq

Lack Of Fit

Intercept

thrust

weight

(thrust-76305)*(thrust-76305)

Term

22539,976

-0,314694

1,4329074

3,4421e-6

Estimate

1210,754

0,078371

0,496822

4,789e-7

Std Error

18,62

-4,02

2,88

7,19

t Ratio

<,0001*

0,0017*

0,0137*

<,0001*

Prob>|t|

Parameter Estimates

thrust

weight

thrust*thrust

Source

1

1

1

Nparm

1

1

1

DF

15149153

7815403

48533396

Sum of

Squares

16,1240

8,3183

51,6564

F Ratio

0,0017*

0,0137*

<,0001*

Prob > F

Effect Tests


Whole Model

Response 1st interval adj

Prediction FunctionInterval first,MLH = 22539.9757 + 1.4329 · weight − 0.3147 · thrust

+ (thrust− 76305) · [(thrust− 76305) · 0.0000034421]

C.4 Mature Interval MLH Engines ix

C.4 Mature Interval MLH EnginesBE_Data_wo_3s- Fit Least Squares Page 1 of 2

12000

13000

14000

15000

16000

17000

18000

19000

20000

21000

3rd

in

terv

al

ad

j A

ctu

al

12000 15000 17000 19000

3rd interval adj Predicted

P<.0001 RSq=0,80 RMSE=880,27


RSquare

RSquare Adj


Mean of Response


0,798364

0,766527

880,272

14805,49

23

Summary of Fit

Model

Error

C. Total

Source

3

19

22

DF

58293477

14722696

73016173

Sum of

Squares

19431159

774878,73

Mean Square

25,0764

F Ratio

<,0001*

Prob > F


Lack Of Fit

Pure Error

Total Error

Source

12

7

19

DF

10460068

4262628

14722696

Sum of

Squares

871672

608947

Mean Square 1,4314

F Ratio

0,3263

Prob > F

0,9416

Max RSq

Lack Of Fit

Intercept

TWR

weight

(weight-12072)*(weight-12072)

Term

34415,709

-2759,253

-0,366246

0,0001018

Estimate

3013,422

485,7675

0,06245

1,816e-5

Std Error

11,42

-5,68

-5,86

5,61

t Ratio

<,0001*

<,0001*

<,0001*

<,0001*

Prob>|t|

Parameter Estimates

TWR

weight

weight*weight

Source

1

1

1

Nparm

1

1

1

DF

25001151

26650979

24347698

Sum of

Squares

32,2646

34,3937

31,4213

F Ratio

<,0001*

<,0001*

<,0001*

Prob > F

Effect Tests


Whole Model

Response 3rd interval adj

Prediction FunctionIntervalmature,MLH = 34415.70939 − 2759.25322 · TWR − 0.36625 · weight

+ (weight− 12072) · [(weight− 12072) · 0.000101795]

C.5 First Shop Visit Restoration Costs x

C.5 First Shop Visit Restoration CostsBase_Data- Fit Least Squares Page 1 of 1

50

100

150

200

250

1st

co

st

ad

j A

ctu

al

50 100 150 200 250

1st cost adj Predicted

P<.0001 RSq=0,93 RMSE=20,538


RSquare

RSquare Adj


Mean of Response


0,925128

0,923535

20,53848

122,1495

49

Summary of Fit

Model

Error

C. Total

Source

1

47

48

DF

244971,87

19825,98

264797,85

Sum of

Squares

244972

422

Mean Square

580,7369

F Ratio

<,0001*

Prob > F


Lack Of Fit

Pure Error

Total Error

Source

30

17

47

DF

14101,817

5724,161

19825,979

Sum of

Squares

470,061

336,715

Mean Square 1,3960

F Ratio

0,2371

Prob > F

0,9784

Max RSq

Lack Of Fit

Intercept

thrust

Term

7,1451068

0,0023619

Estimate

5,602079

0,000098

Std Error

1,28

24,10

t Ratio

0,2084

<,0001*

Prob>|t|

Parameter Estimates

thrust

Source

1

Nparm

1

DF

244971,87

Sum of

Squares

580,7369

F Ratio

<,0001*

Prob > F

Effect Tests

-50

-40

-30

-20

-10

0

10

20

30

40

50

1st

co

st

ad

j R

esid

ua

l

50 100 150 200 250

1st cost adj Predicted


Whole Model

50

100

150

200

250

1st

co

st

ad

j

Le

ve

rag

e R

esid

ua

ls

0 20000 50000 80000 110000

thrust Leverage, P<,0001

Leverage Plot

thrust

Response 1st cost adj

Prediction FunctionRCostfirst = 7.14511 + 0.002361887 · thrust

C.6 Mature Shop Visit Restoration Costs xi

C.6 Mature Shop Visit Restoration CostsBase_Data- Fit Least Squares 2 Page 1 of 1

50

100

150

200

250

300

350

400

ma

ture

co

st

ad

j A

ctu

al

50 100 150 200 250 300 350 400

mature cost adj Predicted

P<.0001 RSq=0,81 RMSE=38,591


RSquare

RSquare Adj


Mean of Response


0,807809

0,804709

38,59068

190,3113

64

Summary of Fit

Model

Error

C. Total

Source

1

62

63

DF

388089,29

92332,90

480422,19

Sum of

Squares

388089

1489

Mean Square

260,5955

F Ratio

<,0001*

Prob > F


Lack Of Fit

Pure Error

Total Error

Source

41

21

62

DF

78144,940

14187,961

92332,901

Sum of

Squares

1905,97

675,62

Mean Square 2,8211

F Ratio

0,0065*

Prob > F

0,9705

Max RSq

Lack Of Fit

Intercept

thrust

Term

46,528678

0,0028861

Estimate

10,12921

0,000179

Std Error

4,59

16,14

t Ratio

<,0001*

<,0001*

Prob>|t|

Parameter Estimates

thrust

Source

1

Nparm

1

DF

388089,29

Sum of

Squares

260,5955

F Ratio

<,0001*

Prob > F

Effect Tests

-100

-50

0

50

100

ma

ture

co

st

ad

j R

esid

ua

l

50 100 150 200 250 300 350 400

mature cost adj Predicted


Whole Model

50

100

150

200

250

300

350

400

ma

ture

co

st

ad

j

Le

ve

rag

e R

esid

ua

ls

0 20000 50000 80000 110000

thrust Leverage, P<,0001

Leverage Plot

thrust

Response mature cost adj

Prediction FunctionRCostmature = 46.52868 + 0.002886118 · thrust

C.7 LLP Cost xii

C.7 LLP Cost

LLP- Fit Least Squares Page 1 of 2

0

100

200

300

400

500

600

700

800

900L

LP

Re

se

rve

s

Actu

al

0 100 300 500 700 900


P<.0001 RSq=0,95 RMSE=44,13


RSquare

RSquare Adj


Mean of Response


0,952884

0,950528

44,13013

254,4868

64

Summary of Fit

Model

Error

C. Total

Source

3

60

63

DF

2363135,5

116848,1

2479983,6

Sum of

Squares

787712

1947

Mean Square

404,4799

F Ratio

<,0001*

Prob > F


Lack Of Fit

Pure Error

Total Error

Source

48

12

60

DF

101360,93

15487,16

116848,09

Sum of

Squares

2111,69

1290,60

Mean Square 1,6362

F Ratio

0,1780

Prob > F

0,9938

Max RSq

Lack Of Fit

Intercept

weight

thrust

(weight-8608,78)*(weight-8608,78)

Term

-115,3133

0,0194512

0,0031206

2,6924e-6

Estimate

13,6202

0,007095

0,001069

3,188e-7

Std Error

-8,47

2,74

2,92

8,44

t Ratio

<,0001*

0,0080*

0,0049*

<,0001*

Prob>|t|

Parameter Estimates

weight

thrust

weight*weight

Source

1

1

1

Nparm

1

1

1

DF

14635,69

16609,48

138881,53

Sum of

Squares

7,5152

8,5288

71,3139

F Ratio

0,0080*

0,0049*

<,0001*

Prob > F

Effect Tests


Whole Model

Response LLP Reserves

Prediction FunctionLLPCost =− 115.31326 + 0.0194512 · weight + 0.0031206 · thrust

+ (weight− 8608.78125) · ((weight− 8608.78125) · 2.69234 · 10−6)

D Model Parameters

D.1 Averaged Short-Haul-Engine Severity Curve

0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 60

0.5

1

1.5

2

2.5

3Averaged SH-Engine Severity Curve

Cycle Time [h]

Sev

erity

Factor

Derate: 0%

Derate: 5%

Derate: 10%

Derate: 15%

Derate: 20%

Flight Time [h] 0% 5% 10% 15% 20%

0.5 2.800 2.600 2.400 2.280 2.1601.0 2.100 1.925 1.750 1.645 1.5401.5 1.600 1.450 1.300 1.210 1.1201.9 1.240 1.120 1.000 0.940 0.8802.5 1.000 0.910 0.860 0.792 0.7443.0 0.920 0.840 0.780 0.738 0.6964.0 0.826 0.766 0.706 0.670 0.6345.0 0.770 0.715 0.660 0.627 0.5946.0 0.740 0.685 0.630 0.597 0.564

xiii

D.2 Averaged Medium-Long-Haul-Engine Severity Curve xiv

D.2 Averaged Medium-Long-Haul-Engine Severity Curve

1 2 3 4 5 6 7 8 9 10 11 120

0.5

1

1.5

2

2.5

3Averaged MLH-Engine Severity Curve

Cycle Time [h]

Sev

erity

Factor

Derate: 0%

Derate: 5%

Derate: 10%

Derate: 15%

Derate: 20%

Flight Time [h] 0% 5% 10% 15% 20%

1.0 2.800 2.500 2.200 2.020 1.9002.0 2.000 1.850 1.700 1.610 1.5203.0 1.600 1.500 1.400 1.340 1.2804.0 1.405 1.315 1.225 1.171 1.1175.0 1.260 1.180 1.100 1.052 1.0046.0 1.140 1.070 1.000 0.958 0.9167.0 1.045 0.985 0.925 0.889 0.8538.0 0.990 0.935 0.880 0.847 0.8149.0 0.970 0.915 0.860 0.827 0.79410.0 0.940 0.890 0.840 0.810 0.78011.0 0.920 0.870 0.820 0.790 0.76012.0 0.890 0.845 0.800 0.773 0.746

D.3 Time & Material Factor Curves xv

D.3 Time & Material Factor Curves

0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 60.7

0.75

0.8

0.85

0.9

0.95

1

1.05

1.1

1.15

SH Time&Material Factor Curve

Cycle Time [h]

T&M

Factor

EFH:EFC 0.5 1.0 1.5 1.9 2.5 3.0 4.0 5.0 6.0T&M Factor 0.90 0.95 0.98 1.00 1.02 1.03 1.04 1.05 1.06

1 2 3 4 5 6 7 8 9 10 11 120.7

0.75

0.8

0.85

0.9

0.95

1

1.05

1.1

1.15

MLH Time&Material Factor Curve

Cycle Time [h]

T&M

Factor

EFH:EFC 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0T&M Factor 0.85 0.91 0.94 0.96 0.98 1.00 1.03 1.05 1.07 1.09 1.10 1.11

E Model Analysis

Sensitivity Analysis for MLH engines

7.4 7.6 7.8 8 8.2 8.4 8.6

x 104

1.3

1.4

1.5

1.6

1.7

1.8

1.9

2x 10

4

Thrust [lbf ]

Interval[E

FH]

Thrust - Interval Sensitivity


First-Run

Mature-Run

1.2 1.25 1.3 1.35 1.4 1.45 1.5 1.55 1.6

x 104

1.2

1.3

1.4

1.5

1.6

1.7

1.8

1.9

2

2.1x 10

4

Weight [lbs]

Interval[E

FH]

Weight - Interval Sensitivity


First-Run

Mature-Run

3 4 5 6 7 8 9 10 11 121

1.2

1.4

1.6

1.8

2

2.2

2.4x 10

4

EFH:EFC [h]

Interval[E

FH]

EFH:EFC - Interval Sensitivity


First-Run

Mature-Run

0 5 10 15 20

1.3

1.4

1.5

1.6

1.7

1.8

1.9

2

2.1x 10

4

Derate [%]

Interval[E

FH]

Derate - Interval Sensitivity


First-Run

Mature-Run

xvi

E Model Analysis xvii

7.4 7.6 7.8 8 8.2 8.4 8.6

x 104

260

280

300

320

340

360

380

Thrust [lbf ]

SVC

per

EFH

[$/EFH]

Thrust - SVC per EFH — Sensitivity


First-Run

Mature-Run

1.2 1.25 1.3 1.35 1.4 1.45 1.5 1.55 1.6

x 104

240

260

280

300

320

340

360

380

Weight [lbs]

SVC

per

EFH

[$/EFH]

Weight - SVC per EFH Sensitivity


First-Run

Mature-Run

3 4 5 6 7 8 9 10 11 12

200

250

300

350

400

450

500

550

EFH:EFC [h]

SVC

per

EFH

[$/EFH]

EFH:EFC - SVC per EFH Sensitivity


First-Run

Mature-Run

0 5 10 15 20

250

300

350

400

Derate [%]

SVC

per

EFH

[$/EFH]

Derate - SVC per EFH Sensitivity


First-Run

Mature-Run

Modeling the Life Cycle Cost of Jet Engine Maintenance

Documents