What s next for the airliner? Historical Analysis and Future Predictions of Aircraft Architecture and Performance.

Transcription

1 What s next for the airliner? Historical Analysis and Future Predictions of Aircraft Architecture and Performance by Demetrios Kellari M.Eng., Imperial College London (2014) Submitted to the Institute for Data, Systems, and Society in partial fulfillment of the requirements for the degree of Master of Science in Technology and Policy at the MASSACHUSETTS INSTITUTE OF TECHNOLOGY June 2016 c Massachusetts Institute of Technology All rights reserved. Author Institute for Data, Systems, and Society May 6, 2016 Certified by Bruce G. Cameron Director, System Architecture Lab Thesis Supervisor Certified by Edward F. Crawley Ford Professor of Engineering Thesis Supervisor Accepted by Munther Daleh William Coolidge Professor of EECS Director, Institute for Data, Systems, and Society Acting Director, Technology and Policy Program

2 2

3 What s next for the airliner? Historical Analysis and Future Predictions of Aircraft Architecture and Performance by Demetrios Kellari Submitted to the Institute for Data, Systems, and Society on May 6, 2016, in partial fulfillment of the requirements for the degree of Master of Science in Technology and Policy Abstract Air travel has advanced greatly since the inception of the first commercially viable airliner in the 1930s, the Douglas DC-3. In the 80 years since then, the number of annual air passengers in the US has increased from 6000 to over 800 million. In order to accommodate this demand growth, civil passenger aircraft architectures have changed over time, investment in aircraft technologies has rapidly increased, and aircraft performance has improved. The historical context of aircraft architectures are analyzed in order to establish trends and precedents for architectural change. This historical analysis is carried out based on a database of 157 architectures from the DC-3 to the Boeing 787, and concludes that historical aircraft architecture changes are mainly driven by selection of engine type. More recently aircraft performance has experienced diminishing returns in terms of efficiency, on the order of 1% reduction in fuel consumption annually since Meanwhile, according to projections by Airbus and Boeing, air passenger traffic is expected to increase % per annum. ICAO has recommended that overall energy efficiency be reduced by 2% annually. The rate of increase in demand and decrease in fuel consumption, raises the question of how this goal can be met. Much prior work has been done to optimize design variables within the context of a single aircraft architecture for maximum performance. Additionally optimization of point designs at the corners of the architecture design space has also been extensively examined. Despite the abundance of work in this domain, there has been limited work done to understand the potential trajectories that would cause current architectures to evolve into these potential future architectures. Therefore, the goal of this thesis is to analyze the conditions that could break the current architecture of commercial aircraft. and identify the stringency of policies that could invoke such a disruption. To answer this research question, the major drivers of increasing engine performance are identified including increasing bypass ratio, increasing overall pressure ratio and turbine inlet temperature, and increasing component efficiency. A hybrid 3

4 analytical-empirical model is presented optimizing the airframe-engine interactions. This model enables us to quantitatively forecast the impact of four engine technology scenarios on the mission block fuel. Results indicate that for existing airframes, namely the 737 and A320, performance is expected to increase by 6-38% relative to the 737MAX and A320neo within the next years, depending on engine technology development. For a new aircraft with the dominant architecture and unconstrained geometry, we expect a performance increase of 17-40% versus an optimized aircraft with current technology. The maximum performance is expected to occur in the next years, suggesting that a break in the dominant architecture will occur in this timeframe. Finally, two policy scenarios based on ICAO and IATA targets are shown to incentivize technology development, reduce the uncertainty in performance and architecture predictions, and reduce the time to an expected break in architecture. Thesis Supervisor: Bruce G. Cameron Title: Director, System Architecture Lab Thesis Supervisor: Edward F. Crawley Title: Ford Professor of Engineering 4

5 Acknowledgments This thesis is the culmination of two years of research in the System Architecture Lab. It also marks the completion of my seven years as a university student, and it has been an incredible journey. The mining town of Kitwe in the Copperbelt in Zambia, where I spent my first decade, and the ancient divided city of Nicosia in Cyprus, where I spent the subsequent decade, are a far-cry from the academic might of Cambridge. The journey has been long, with many obstacles overcome, to have the privilege of calling those three letters home - MIT. My time here has been lifechanging. This is in no small part due to the people that I have met here, and those that got me here in the first place. I would like to acknowledge the major role they have had in this success. It s hard to know where to start, but I think it apt to firstly thank my two advisors Bruce and Ed. Bruce I do not know where I would be without your guidance, insights and encouragement. Ed I am grateful for your support and advice, despite you living many timezones away in Russia. You both gave me the flexibility to pursue my interests - I could not have asked for anything more. To my friends in , in spite of the state of our office, I have enjoyed our countless conversations and musings on the world. Marc, you have been a role model in the lab, your intellect never ceases to amaze me. Narek, I do not have enough space to write all the things you have taught me - your advice, support, and expertise have been irreplaceable and I am proud to call you a friend. Sydney we have struggled together, in research, writing, and at the gym - I look forward to seeing you on the ISS soon. Inigo, Chris, Andrew, Veronica, Sam I have learned a great deal about many disparate things from each one of you and I will miss our conversations. My TPP buddies Chetan, Sergio, Seb, John, Varun, and everyone else, have been a fundamental part of this experience. I cannot imagine what it would have been like without you guys. It s been an absolute pleasure and we won t lose touch even though we are going our separate ways. There are few specific moments I can pin point that have had as much of a profound effect on my life as when I met Willada. You have had to put up with a lot from me during my long hours of work and those grumpy moments. It is so rare that someone comes along who not only humbles you but inspires you too. Willada I am thankful for your tolerance, patience, and unyielding optimism throughout. You have taught me alot about life along the way. Stav and Christiana, you have helped me as much as anyone else to get here. We have grown up together and no one knows me better than you guys - thanks for always being there. Last but not least, mum & dad, I did it! One more box checked off the list. I will forever be thankful for your sacrifices - I hope this work shows my appreciation for all that you have done for me. From Zambia, to Cyprus, to England, to the US, you have always been there guiding me in every decision, carrying me in every difficult moment, and celebrating every success. Your unconditional support is more than I could have ever asked. The only thing I know is I know nothing. To the end of an era. 5

6 6

7 Contents 1 Introduction Motivation Context Generic thesis objectives Literature review System architecture Aircraft architecture Architecture evolution Aircraft architecture evolution Aircraft performance metrics and trends Optimization of point designs Specific Objectives Thesis Overview Historical analysis of commercial aircraft architectures Introduction Problem formulation Scope Aircraft architecture definition Aircraft performance metric Architecture decision subset selection Analysis Architecture variation over time Architecture performance Architectural decision space Conclusion Dominant architecture airframe-engine model Introduction Trends in civil aircraft technologies First principle analysis of technology drivers Summary of review Forecasting airframe and engine technologies Bypass ratio trends Component efficiency trends Turbine inlet temperature and overall pressure ratio trends Overview of model Inputs Mission profile inputs Technology and design parameter inputs Model description

8 3.6.1 Initial airframe-engine geometry & weight estimation Optimization of the aircraft design parameters Outputs Verification of the model Engine model verification Verification of specific fuel consumption Airframe model validation Verification of airframe geometry Disrupting the current dominant aircraft architecture Introduction Performance of existing aircraft given engine technology improvements Limits of dominant architecture given engine technology improvements Model inputs Scenario 1: Bypass ratio increase with baseline technology Scenario 2: Bypass ratio increase with increase in component efficiency Scenario 3:Bypass ratio increase with increase in turbine inlet temperature and overall pressure ratio Scenario 4: All engine technologies improve Summary and discussion of results Aviation policy: implication for architectural changes Introduction and motivation Background A brief history of aviation emissions standards Gaps in current standards Goals Policy options and recommendation Engine certification standards Aircraft certification standards Market-based options Recommendation Effect of recommended policy on aircraft architecture Policy Scenario 1: 2% reduction in fuel burn Policy Scenario 2: 70% fuel burn reduction by 2050 relative to Discussion Conclusions Conclusions 131 A Architecture decisions 145 8

9 B Airframe-engine model 147 B.1 Operating conditions calculation B.1.1 Function overview B.1.2 Inputs B.1.3 Process B.1.4 Outputs B.2 Initial Sizing calculation B.2.1 Function overview B.2.2 Inputs B.2.3 Processes B.2.4 Outputs B.3 Thrust-to-weight ratio and wing loading B.3.1 Function overview B.3.2 Inputs B.3.3 Processes B.3.4 Outputs B.4 Geometry definition B.4.1 Inputs B.4.2 Process B.4.3 Outputs B.5 Refined aircraft sizing B.5.1 Inputs B.5.2 Process B.5.3 Outputs B.6 Landing gear tire sizing B.6.1 Function overview B.6.2 Inputs B.6.3 Processes B.6.4 Outputs B.7 Engine Analysis B.7.1 Function overview B.7.2 Inputs B.7.3 Process B.7.4 Outputs B.8 Aerodynamic calculations B.8.1 Inputs B.8.2 Process B.8.3 Outputs B.9 Component weight calculation B.9.1 Function overview B.9.2 Inputs B.9.3 Processes B.9.4 Outputs B.10 Center of gravity calculation B.10.1 Function overview

10 B.10.2 Inputs B.10.3 Processes B.10.4 Outputs B.11 Longitudinal static stability calculation B.11.1 Function overview B.11.2 Inputs B.11.3 Processes B.11.4 Outputs B.12 Engine and landing gear optimization B.12.1 Function overview B.12.2 Inputs B.12.3 Processes B.12.4 Outputs B.13 Landing gear weight calculation B.13.1 Function overview B.13.2 Inputs B.13.3 Processes B.13.4 Outputs B.14 Engine weight calculation B.14.1 Function overview B.14.2 Inputs B.14.3 Processes B.14.4 Outputs C Model verification method: Error Analysis 191 D Technology scenario 4: Results 193 E Turbine inlet temperature trends 195 F ICAO timeline of emissions regulations 197 G Engine constraints

11 List of Figures 2-1 Graph showing the frequency of different options for the decisions of how to lift the payload. These decisions include the wing configuration, the wing shape, the wing structural configuration, and the wing vertical location Graph of performance against time, showing aircraft architectures as blue dots and the year-averaged performance over time, including the standard deviation in this number. Key aircraft architectures are highlighted on the graph (Left) Graph of frequency of distinct architectures and distinct aircraft per year. (Right) Plot of the ratio of distinct architectures to distinct aircraft over time overlain with yearly averaged performance over time Plot of the ratio of distinct architectures to distinct aircraft over time overlaid with yearly averaged performance over time, showing individual architectures. Key moments in the timeline of aircraft evolution are shown including the introduction of various technologies as well as regulations Graphs of the yearly averaged performance of aircraft showing the performance bounds excluding outliers, and the mean performance for each set. The whole data set is shown in the top left graph, and the data has been stratified for regional aircraft (top right), narrow-body aircraft (bottom left) and wide-body aircraft (bottom right). Regional aircraft are classified as those with a capacity of fewer than 100 passengers, narrow-body those between 100 and 250 and wide body above 250, considering a single class economy layout A graph of the decision space showing interconnections between the decisions. The arrows show the direction of influence of one decision over another Decision space view of sensitivity of the decisions to the performance metric against connectivity of each decision. This matrix separates the decisions by priority according to the se two dimensions, as shown Boxplot graphs of the performance of each option for four different decisions, showing the mean as a red line, the interquartile range as a blue box, the range excluding outliers as dotted lines and the outliers as red dots. The four decisions shown are Engine Type (top left), wing shape (top right), number of engines (bottom left) and wing vertical location (bottom right) Evolution of aircraft architecture over time including future forecast of breaking point of the current dominant architecture. The architecture space has been collapsed onto one dimension on the y axis

12 3-2 Trend of structural weight as a percentage of maximum takeoff weight over time Variation of normalized block fuel weight for three different scenarios of improving technologies. It is shown that all else held constant, SFC has the largest potential for impacting aircraft performance Bypass ratio variation over time, for turbofan engines for aircraft with a maximum takeoff weight above 100 tons. Linear and power trendlines are fit to the data, with 85% confidence intervals shown. Real engine data is shown as blue circles while engine concept data from literature is shown as green triangles Bypass ratio variation over time, for turbofan engines for single-aisle aircraft. Linear and power trendlines are fit to the data, with 85% confidence intervals shown. Real engine data is shown as blue circles while engine concept data from literature is shown as green triangles High level overview of the airframe-engine model, including inputs and outputs Typical mission profile for air passenger aircraft showing the various mission segments Block diagram of the architecture of the airframe-engine model showing the flow of variables in between specific functions A schematic diagram of the idealized turbofan engine used in this model A front-view diagram (left) and plan-view diagram (right) of the output aircraft geometry from the model used in this model. This shows all the major components and dimensions, including location of the center of gravity in green and neutral point in red Verification plots for fan diameter for 25 different airframe-engine combinations Verification plots for specific fuel consumption for 25 different engines present on aircraft in the database Verification plots for maximum takeoff weight for 25 different airframeengine combinations Verification plots for empty weight for 25 different airframe-engine combinations Front view drawings of the (a) /900 NG and (b) 737-7/8/9 MAX showing the difference in minimum ground clearance due to the new engine Diagram showing the ground clearance requirement for the engines in the dominant architecture Variation of fan diameter with bypass ratio for the Boeing 737-7/8/9 showing the 95% confidence bounds in the model estimations and the location of existing designs. The geometrical limit for this aircraft geometry is shown, which has been estimated using the method described in this thesis

13 4-4 Variation of performance parameters with bypass ratio for turbofan engines powering the /900 airframe Variation of mass of fuel required to perform the design mission with increasing bypass ratio for the /900 airframe Graphs showing the variation in fan diameter with increasing bypass ratio, showing the effects of changing engine technology and the geometric limit of the /900 airframe. Shown on the graph as a blue square is the NG and as the red circle is the MAX Graphs showing the variation in required fuel weight with increasing bypass ratio for the /900 airframe under 3 technology scenarios. The range of possible bypass ratios corresponding to the geometric limitations imposed by ground clearance constraints are shown Forecast of expected value of BPR using the whole dataset, overlain with geometry limits dictated by BPR for the 737 for the four different scenarios. The 85% confidence intervals both in time and BPR are shown Forecast of expected value of BPR using only single-aisle aircraft, overlain with geometry limits dictated by BPR for the 737 for the four different scenarios. The 85% confidence intervals both in time and BPR are shown Front view drawings of the (a) A and (b) A320 neo showing the difference in minimum ground clearance due to the new engine Variation of fan diameter with bypass ratio for the A320 showing the 95% confidence bounds in the model estimations and the location of existing designs. The geometrical limit for this aircraft geometry is shown, which has been estimated using the method described in this thesis Variation of performance parameters with bypass ratio for turbofan engines powering the A320 airframe Variation of mass of fuel required to perform the design mission with increasing bypass ratio for the A320 airframe, showing the range of possible bypass ratios falling within the 95% confidence interval for ground clearance requirements Graphs showing the variation in fan diameter with increasing bypass ratio, showing the effects of changing engine technology and the geometric limit of the A320 airframe. Shown on the graph as a blue square is the A NG and as the red circle is the A320 neo Graphs showing the variation in required fuel weight with increasing bypass ratio for the A320 airframe for 3 technology scenarios. The range of possible bypass ratios corresponding to the geometric limitations imposed by ground clearance constraints are shown Forecast of expected value of BPR using the whole dataset, overlain with geometry limits dictated by BPR for the A320 for the four different scenarios. The 85% confidence intervals both in time and BPR are shown

14 4-17 Forecast of expected value of BPR using only single-aisle aircraft, overlain with geometry limits dictated by BPR for the A320 for the four different scenarios. The 85% confidence intervals both in time and BPR are shown The evolution of aircraft design parameters as BPR increases in scenario 1. On the diagram the trend is clearly highlighted with the different colors Variation of aircraft design parameters as BPR increases for scenario Variation of aircraft performance variables with bypass ratio for scenario Aircraft performance against bypass ratio, quantified as the block fuel weight, for scenario 1. Shown on the graph are 95% confidence bounds the expected value, as well as current maximum BPR range and the maximum possible performance for this scenario Graph of BPR against year showing both projection models based on different data sets for scenario The evolution of aircraft design parameters as BPR increases in scenario 2. On the diagram the trend is clearly highlighted with the different colors Variation of aircraft design parameters as BPR increases for scenario Variation of aircraft performance variables with bypass ratio for scenario Aircraft performance against bypass ratio, quantified as the block fuel weight, for scenario 2. Shown on the graph are 95% confidence bounds the expected value, as well as current maximum BPR range and the maximum possible performance for this scenario Graph of BPR against year showing both projection models based on different data sets for scenario The evolution of aircraft design parameters as BPR increases in scenario 3. On the diagram the trend is highlighted with the different colors Variation of aircraft design parameters as BPR increases for scenario Variation of aircraft performance variables with bypass ratio for scenario Aircraft performance against bypass ratio, quantified as the block fuel weight, for scenario 3. Shown on the graph are 95% confidence bounds the expected value, as well as current maximum BPR range and the maximum possible performance for this scenario Graph of BPR against year showing both projection models based on different data sets The evolution of aircraft design parameters as BPR increases in scenario 3. On the diagram the trend is highlighted with the different colors

15 4-34 Aircraft performance against bypass ratio, quantified as the block fuel weight, for scenario 3. Shown on the graph are 95% confidence bounds the expected value, as well as current maximum BPR range and the maximum possible performance for this scenario Comparison of the maximum performance increase for the four technology scenarios for each of the aircraft Comparison of the architecture break point for the four technology scenarios for each of the aircraft Contributions to radiative forcing through aviation emissions and their projections over time A comparison of the expected year for the architecture break given 3 different policy scenarios, for 4 different technology scenarios and for 3 different aircraft cases A-1 Trends in architectural decision-options over time plotted in performancetime space. Four decisions are visualized here corresponding to engine type (top left), wing shape (top right), number of engines (bottom left) and wing vertical location (bottom right) B-1 Block diagram of operating conditions function showing inputs and outputs B-2 Block diagram of initial sizing function showing inputs and outputs B-3 Block diagram of thrust-to-weight and wing loading calculation function showing inputs and outputs B-4 Block diagram of geometry definition function showing inputs and outputs B-5 Block diagram of refined sizing function showing inputs and outputs. 157 B-6 Block diagram of turbofan engine analysis function showing inputs and outputs B-7 Block diagram of lift calculation function showing inputs and outputs. 166 B-8 Block diagram of drag calculation function showing inputs and outputs.167 B-9 Block diagram of center of gravity calculations showing inputs and outputs B-10 Block diagram of longitudinal static stability calculation function showing inputs and outputs B-11 Block diagram of landing gear and engine optimization function showing inputs and outputs B-12 Model of the (a) main landing gear and (b) nose landing gear used for this analysis B-13 Idealization of the landing gear for structural analysis in (a) yield failure and (b) buckling failure D-1 Variation of aircraft design parameters as BPR increases for scenario D-2 Variation of aircraft performance variables with bypass ratio for scenario

16 D-3 Graph of BPR against year showing both projection models based on different data sets for scenario E-1 Turbine inlet temperature against year including civil and military aircraft as well as the effect of cooling technologies E-2 Graph of specific thrust against turbine inlet temperature showing theoretical stoichiometric limit F-1 A timeline of emissions standards developed by ICAO [89] G-1 Engine constraints

17 List of Tables 2.1 Decision options corresponding to the functional decomposition at the third level. These decisions comprised the initial set prior to subset selection Comparison between component efficiencies today and the maximum possible efficiencies from literature The dominant aircraft architecture Mission profile variables used as input parameters to the model Engine technology input variables for the engine model Airframe technology input variables for the airframe model Output parameters from the airframe-engine model Assumed input values into the model due to a lack of available information. These were taken from aircraft design books as well as comparable aircraft Statistical comparison of fan diameter from real data and model outputs Statistical comparison of specific fuel consumption from real data and model outputs Statistical comparison of maximum takeoff weight and empty weight from real data and model outputs Statistical comparison of aircraft geometry parameters from real data and model output data for the whole data set Statistical comparison of aircraft geometry parameters from real data and model output data for the single-aisle aircraft from the dataset Comparison of two generations of turbofan engines for the The upper and lower 95% confidence bounds as well as the expected values (EV) for bypass ratio, SFC, engine weight and weight of fuel, which are attainable given the ground clearance requirements of the current 737 airframe. These values are calculated for a fan diameter constraint of 1.87m The upper and lower 95% confidence bounds as well as the expected values for aircraft performance for the Summary of the expected values for the year of architecture break, for both BPR forecast models Comparison of two generations of turbofan engines for the A The upper and lower 95% confidence bounds as well as the expected values (EV) for bypass ratio, SFC, engine weight and weight of fuel, which are attainable given the ground clearance requirements of the current A320 airframe. These values are calculated for a fan diameter constraint of 2.12m

18 4.7 The upper and lower 95% confidence bounds as well as the expected values of aircraft performance for the A Summary of the expected values for the year of architecture break, for both BPR forecast models Mission profile variables used as input parameters for the analysis Comparison of geometry and performance variables for scenario 1, for today s performance versus the maximum performance Expected values for architecture break year including confidence bounds for two BPR forecast models for scenario Comparison of geometry and performance variables for scenario 2, for today s performance versus the maximum performance Expected values for architecture break year including confidence bounds for two BPR forecast models for scenario Comparison of geometry and performance variables for scenario 3, for today s performance versus the maximum performance Expected values for architecture break year including confidence bounds for two BPR forecast models for scenario Comparison of geometry and performance variables for scenario 4, for today s performance versus the maximum performance Expected values for architecture break year including confidence bounds for two BPR forecast models for scenario Expected year of break of architecture for the 737, A320, and a new optimized aircraft with the dominant architecture, for four technology scenarios under policy scenario Expected year of break of architecture for the 737, A320, and a new optimized aircraft with the dominant architecture, for four technology scenarios under policy scenario

19 Chapter 1 Introduction Civil passenger aircraft have a short commercial history, relative to other modes of transportation. Although the first scheduled commercial airline flight took place on 1st January 1914 from St. Petersburg, FL to Tampa, FL, [61] it was not until the inception of the Douglas DC-3 that this opened up to a larger commercial market [82]. Airlines such as Imperial Airways in Great Britain and Pan American World Airways in the US started international flights in the 1930s. Imperial airlines carried 23,900 passengers in 1930, its longest route being its London-Cape Town service [95], while the entire US airline industry accounted for approximately 6000 passengers in the same year [3]. Fast forward to 85 years later, the number of air passengers in the US has grown to over 800 million in 2015 [88]. This contrast shows that despite its relatively short life air transport has developed immensely. To enable this immense growth air travel has had to become faster, more comfortable, and cheaper for passengers. Although vast amounts of capital have been deployed to develop the air transport system infrastructure, the most important driver of this passenger traffic trend has been the development of economically efficient passenger aircraft [46]. Due to its significance, this thesis places the aircraft at center stage, analyzing the most important decisions in the conceptual stage of design, defined as the aircraft architecture. The evolution of civil passenger aircraft architectures and performance over the past 80 years is mapped, extracting lessons and extrapolating trends, in order to predict potential future trajectories. 1.1 Motivation Over the last eighty years of commercial aviation, the airliner has improved immensely in terms of traditional metrics of aircraft performance, such as fuel efficiency and load factor [75]. These gains have been driven by tremendous innovation in technologies such as engines, materials, and control systems [91]. This period is associated with a reduction in the variation of architectures and the emergence of a dominant design. In the early years of commercial aviation, there were substantial variations in aircraft architecture [82]. As time has progressed many architectural options, such as engine location above the wing, have died out leading to a consolidation of aircraft architecture. Given the above background, there are three main factors which motivate this thesis work: Aircraft performance is experiencing diminishing returns due to the trend of 19

20 incremental improvements in aircraft technologies. Marginal improvements are being made in order to reduce operating costs for airlines through increasing reliability, decreasing maintenance costs, and increasing fuel efficiency [75]. It is projected that fuel burn will decrease at an annual rate of % until 2050, given various technology and operational scenarios [90]. The number of passengers is expected to increase dramatically in the future putting a strain on infrastructure, existing aircraft and the environment. It is projected that there will be 7 billion passengers by 2034, up 112% from 3.3 billion in 2014 [14]. Airbus predicts an annual growth rate of 4.6% over the next 20 years [5], while Boeing expect between % annual growth in the next decade [24]. The ink is still drying on the Paris agreement to keep global average temperature increase to below 2 C since pre-industrial levels. Following this, ICAO s Committee on Aviation Environmental Protection (CAEP) have introduced a recommendation for increasing energy efficiency by 2% per year, as a CO 2 efficiency standard for new aircraft [90]. It can be seen that the diminishing improvements in aircraft performance combined with the increase in passenger traffic demand are in direct conflict with the environmental goals of the next few decades. This raises the important question of how the current architectural trends might lead to this conflict being resolved in the future Context The factors mentioned in Section 1.1 raise the question of, given the trend of consolidation of aircraft architecture, what potential future trajectories will enable the future demand to be met while realizing the ambitious environmental goals. There has been an abundance of work done in this general domain including: the optimization of existing aircraft architectures for maximizing aircraft performance or minimizing environmental impact; examining potential future architectures which have superior performance over the current dominant design; and, extrapolating performance trends in order to predict future aircraft performance. There has been little work analyzing how the current state of aircraft architecture may evolve over time, continuing to increase performance to meet the goals of a future society. Due to the uncertainty associated with the future of passenger air transportation, there are many potential trajectories in aircraft performance and architecture. While many have postulated what a future air transport system may look like [13], this thesis extracts lessons and trends from the past mapping specific trajectories that may lead to such a future. 20

21 1.2 Generic thesis objectives Section 1.1 and identified the need for evaluating historical and predicting future aircraft architectures and performance. The overarching objective of this thesis is to use historical insights to highlight potential evolution pathways of the current dominant design and predict the conditions that may lead to a break in this architecture. This translates into a need to develop a method to quantify aircraft architectures, measure historical performance, identify the most important trends and predict future architecture and performance under uncertainty More specifically, this entails: 1) developing a framework to quantify aircraft architectures, defined as the most important decisions in the conceptual design of aircraft, 2) devising a metric which accurately measures architecture performance, 3) assembling a database of civil passenger aircraft to use for the historical analysis, 4) analyzing and extrapolating trends in aircraft architecture and technologies, identifying th major drivers of performance, 5) developing a model to consider these trends in the context of future aircraft architecture, 6) discussing potential policy options to allow us to meet the future demands of passenger air travel. 1.3 Literature review Before commencing with the main body of the thesis, a review of existing literature will be carried out. This is aimed at giving a succinct overview of academic knowledge in the fields of system architecture, particularly pertaining to passenger aircraft, the optimization of aircraft design, the metrics by which aircraft performance is measured, and the trends in architecture and performance over time System architecture The field of System Architecture is often considered as one of the first steps in the System Engineering process, which considers a system as a set of interacting elements arranged in such a way as to fulfill a specific purpose. It is often used to support decision-making under uncertainty in the early stages of design of a complex system. Crawley defines system architecture as the embodiment of a concept: the allocation of physical/informational function to elements of form and definition of interfaces among them and with the surrounding context [32]. It can be thought of as the set of decisions that define a system at its highest level of design [105]. According to Crawley et al. [30] good system architecture enables complex systems to meet stakeholder needs and deliver value, while integrating easily, evolving flexibly, and operating simply and reliably. They have assembled a list of 26 Principles of System Architecture which enable system architects to create good system architecture. This framing of good system architecture raises the question of how does one determine whether one architecture is superior to another. This is often done by measuring the value of a particular architecture, where value is defined as the benefit at a given cost [30]. The system benefit can be defined in many different ways, depending on 21

22 the major functions the system must deliver, the stakeholders in the system and the system context. Therefore it is up to the architect to define metrics which adequately define this system benefit. Meanwhile costs are usually measured using life cycle cost which includes up-front investment as well as, manufacturing, operating and disposal costs Aircraft architecture In aircraft design, the conceptual phase consists of aggregating the design requirements and available technology, culminating in a concept sketch [98, 100]. Usually this also includes an initial sizing whereby domain knowledge from various experts is utilized to roughly approximate the sizes of the various major components. Raymer [98] describes this process as including a combination of customer requirements, new concept ideas and available technologies; therefore the decisions are not explicitly stated but rather are reliant on legacy designs and expert knowledge. Torenbeek describes the aircraft architecture or configuration as the general layout, the external shape, dimensions and other relevant characteristics, thus exclude minor decisions such as the layout of the high lift devices [111]. According to Sadraey [101], in the conceptual phase an aircraft architect must decide on the architecture, which consists of lifting surface arrangement, control surface location, propulsion system selection, payload storage, landing gear and subsystem configuration. Howe [60] defines several decisions as being static, namely a cantilever monoplane wing, separate vertical and horizontal tail surfaces, a discrete fuselage and retractable tricycle landing gear. According to Howe the conceptual decisions are therefore the number of engines and their location, the vertical position of the wing and the configuration of the empennage. Meanwhile Torenbeeck [111] details the initial baseline design as comprising of decisions pertaining to wing, fuselage, handling qualities, structural qualities, systems design, powerplant integration. These decisions are motivated by assessment of customer and airworthiness requirements. Although there has been much work done in the realm of optimizing design parameters such as sizes and dimensions [71, 39, 79, 72, 76], which will be discussed in Section 1.3.6, there has been little research done to analyze and quantify the effects of the initial configuration decisions that constitute the aircraft architecture Architecture evolution According to Henderson and Clark, architectural changes are distinct from incremental changes or modular changes in that they involve reconfiguration of components within a system without necessarily changing the components themselves [56]. In this context architectural innovation involves a significant change in the linkages between entities of a system without a changing the entities themselves. For example, two major architectural changes in commercial aircraft were the introduction of the metallic monoplane and the introduction of the transonic jet aircraft, which were separated by a period of incremental innovation. Architectures may be enumerated within a tradespace to enable groups of architectures to be compared in terms of performance, 22

23 cost and architectural design. Creating a network within such a tradespace allows one to track the evolution of an architecture over time, correlating this to the architecture cost and performance. Koo et al. [70] and Arney [11] analyze incremental architectural changes from an initial baseline in order to identify architectures on the Pareto front in the design space. In the domain of software architecture, Nakamura and Basili [85] trace the evolution of software from the beta until the final version, in order to analyze the cost of various trajectories and the influence of architecture on this system characteristic. Both Silver et al. [106] and Davison et al. [37, 36] utilized a similar methodology to track architecture development over time taking into account ease of evolution through switching costs. Their motivation was to enhance the flexibility of architecture development pathways and mitigate the uncertainty associated with long lead times Aircraft architecture evolution There has been documentation of historical aircraft design in the form of anecdotal rather than analytical analysis. Gardiner [46] documents the major design trajectories in civil aircraft from the 1930s to the 1980s linking these to the major economic climate of each decade. In his research, rather than considering an overall aircraft architecture, the designs forming these trajectories are characterized by the trends in state of the art of various technologies. These trends include introduction and development of gas turbine engines; the development of new materials such as plastics, fibers and titanium; and, microelectronic control systems. Gardiner notes that for any design trajectory there are one-offs that violate the trend, such as the Concorde or flying boats. Miller and Sawers [82] highlight several key advances, which changed the course the technical development of aircraft from the 1920s to the 1970s, including the introduction of the metallic monoplane architecture with the DC-3 and the use of the jet engine for civil aircraft. Similarly, Green [48] documents the potential of new technologies, for example contra-rotating engine technology, and aircraft architectures such as the blended-wing body, concluding that substantial reduction in CO2 emission will require radical changes to aircraft design in particular deviation away from the dominant swept-wing architecture Aircraft performance metrics and trends In the analysis of multiple architectures, performance metrics are required in order to track their evolution over time [30]. In the domain of civil aviation, there is an abundance of work tracking the performance of modern aircraft in terms of technical, economic, environmental and operational factors. Aircraft technical performance is typically characterized by payload-range graphs, take-off and landing field lengths, climb rate/angle, and the cruise performance [98, 100, 111] which uses metrics such as cruise speed, specific fuel consumption, and lift-to-drag ratio [7, 8]. Additionally there is an abundance of metrics to characterize aircraft productivity. These include aircraft productivity index [81], utilization per day, average stage-length, load factor, available seat mile and revenue passenger mile [115]. In terms of operational efficiency, airlines 23

24 typically use total aircraft block hours, daily airborne hours, number of departures per aircraft day, and other similar metrics [94]. While the above are conventionally used in industry, researchers often devise their own metrics to assess the particular field of interest. For instance, Hileman and Katz [57] define a productivity metric to measure the energy required to move a given payload a certain distance, defined as the payload fuel energy efficiency. This metric is chosen due to its ability to incorporate cargo as well as passenger flights in a productivity measure, as well as compare alternative fuels since the cost function is the energy of fuel. Lee et al. [75] and Babikian et al. [15] use an energy intensity metric in statistical and analytical models to examine the influence of aircraft performance on cost from the 1960s until the early 2000s. Lee finds that an annual decline in air transport energy intensity of 1.2%-2.2% is not sufficient to offset the increase of 4%-6% in passenger air travel, therefore emissions are expected to increase. By contrast, Dallara et al. [35] have devised a metric known as average temperature response, to quantify the lifetime global mean temperature change caused by aircraft operations. Antoine [9] and Schwartz [104] use this metric to analyze the impact of different aircraft designs on global climate finding that a 30% reduction in global warming impacts is attainable by changing aircraft operating conditions. While the aircraft design is varied in terms of design parameters, the aircraft architecture remains the same in this analysis Optimization of point designs There has been much prior research dedicated to optimization of a single point in the aircraft architectural space. This is usually carried out with the use of multidisciplinary design optimization (MDO) for the optimization of design variables for a single architecture, with a specific objective [75, 71, 9, 10]. For example, Bower et al. use a MDO methodology to design a single aisle aircraft to minimize direct operating costs (DOC), CO 2 emissions and NO x emissions [26]. They find a Pareto frontier on which these three objectives are traded in the context of the dominant architecture, with CO 2 and DOC being very correlated. Similarly Antoine et al. [9] developed a design tool using a MDO approach to quantify tradeoffs between noise performance, DOC, and emission performance. This was carried out by modifying design variables within the dominant architecture, showing that modification of the design parameters can only go so far in order to increase performance, without using abatement technologies. They cite that any such design tool is heavily reliant on verification and validation since there is inherent uncertainty in the model. Raymer [97] evaluates multiple MDO methods in the context of four different architectures including a commercial airliner. He demonstrates how optimization improves the weight and cost of an such design concepts by approximately 2-10% through tweaking design variables. Although, within the context of a single architecture, this method generally demonstrates increase in performance compared to traditional design methods, it is not able to optimize across the architectural decision space. MDO has also been used to optimize individual aircraft components. Obayashi has used this methodology to optimize wing planforms, with the objective of minimizing drag, or minimizing weight for a given fuel payload, showing that it is possible to generate feasible results 24

25 during conceptual design [87]. In addition to the so-called conventional or dominant architecture, optimization has been carried out for other point designs in the architectural space, representing potential future architectures. A often cited way to increase the lift-to-drag ratio of passenger aircraft is to shift to a blended-wing body (BWB) or flying wing architecture [77]. According to a study by Cranfield University a BWB architecture could increase lift-to-drag ratio by 15% resulting in a 17% reduction in fuel burn per unit of payload-range [107]. In addition to this, such architectures provide other benefits, such as enabling alternative propulsion systems. Green carried out studies of BWB and flying wing architectures with unducted fan engines, demonstrating a projected fuel burn per unit payload-range of 50% lower than current passenger aircraft [109]. Similarly NASA have carried out studies of two potential future architectures for the time period, one defined as a double-bubble architecture (the D-8 Series), and the other defined as a hybrid wing body architecture (the H-3 Series) [84]. The D-8 demonstrates an reduction in fuel burn of 70%, reduction in noise of 60dB, and a reduction in NO x emissions of 87% relative to today s baseline technology. Although there is a clear outline of the technology requirements to reach these performance improvements, there is a gap in how the architecture will evolve from the baseline dominant architecture to this future double-bubble architecture. While there has been much research dedicated to the design and optimization of these singular point designs, there has been significantly less work analyzing the effect of architectural decisions. Furthermore there has been little research dedicated to examining the path that may take us from the current civil passenger aircraft architectures to these potential future architectures. Despite the clear advantages of such future architectures, the apparent architectural lock-in of the current dominant design is a major barrier for the industry to overcome and there is a lack of research and understanding of trajectories that could lead to something such as the D-8 or H-3. For example, if the current dominant design continues to improve on performance for the next 50 years, and given the risk of developing a new architecture, what incentive is there to re-open the architectural design space. 1.4 Specific Objectives Given the overarching thesis objectives presented in Section 1.2, a gap in the literature has been identified, which has led to the development of the following specific thesis objective. To quantify the trends in civil passenger aircraft performance due to the evolution of aircraft architecture over time, and prescribe policy options to meet the demands of a future society by: 1. Understanding the progress in historical aircraft architecture and performance, and identifying the major driving forces in architectural change. 2. Identifying and extrapolating the major trends in aircraft technologies that are likely to improve aircraft performance in the future. 3. Predicting future improvements in aircraft performance, and associated implications on aircraft architecture, based on possible technology scenarios. 25

26 Using an airframe-engine model that sizes the geometry and evaluates the performance of the dominant aircraft architecture, given a particular level of technological advancement. 1.5 Thesis Overview The structure of the rest of this thesis is as follows: Chapter 2 begins by presenting a framework for defining aircraft architectures, and a method for computing architecture performance. These two methods are then applied to a database of 157 civil passenger aircraft from the Douglas DC-3 to the Boeing 787, mapping the evolution of architectures and performance over time, and identifying the key factors in architectural change. In Chapter 3 a first principle analysis of high-level aircraft performance is carried out using the Breguet range equation and the major technological drivers of aircraft performance are identified. Subsequently the semi-analytical, semi-empirical airframe-engine model is described in detail followed by a verification analysis based on an in-depth dataset of 27 aircraft exhibiting the dominant architecture. Chapter 4 is devoted to the results of the two main analyses of this thesis. The first analysis applies four future technology scenarios to two existing airframes, namely the A320 and 737, in order to predict when their maximum performance will be reached given geometric constraints. The second analysis applies the same four technology scenarios to flexible aircraft geometry within the context of the dominant architecture, predicting when these might reach their maximum performance. These results include confidence bounds to account for the uncertainty in the model quantified through verification using the database of existing aircraft. Chapter 5 discusses several policy options that could be used in order to meet the demands of a future society in terms of aircraft performance within the assumptions of the presented model. Finally Chapter 6 summarizes the contributions of this thesis, discusses the limitations of the work that has been presented and ends by proposing potential improvements and future work. 26

27 Chapter 2 Historical analysis of commercial aircraft architectures 2.1 Introduction In the design of large-scale engineering systems, technical decisions made in the early stages have a large bearing on the final system performance [30, 31, 22]. Said otherwise, revisiting early architectural decisions can force substantial rework due to cascading changes. An example of this issue occurred during the re-design of the U.S. Navy F/A-18 to the Swiss F/A-18 configuration, which increased the cost per aircraft significantly due to unanticipated propagation of changes throughout the system [38]. With the inception of new technologies such as the geared turbofan, we hypothesize that these could force reconsideration of the current dominant commercial aircraft architecture. That is, when incremental improvements of the current architecture are exhausted, technological innovation could drive architectural disruption if there is an associated increase in system performance. This chapter develops a framework for consideration of commercial aircraft architectures as decisions, analyzing the trends in these architectural decisions over time. The conceptual stages of aircraft design comprise high-level decisions pertaining to the aircraft configuration or architecture. An example of such a decision is the vertical location of the wing, taking the options high wing or low wing. The objective of the work in this chapter is to chart the historical evolution of civil aircraft from the perspective of system architecture. This involves identifying and prioritizing the most important architectural decisions and mapping these to past and present passenger aircraft. As a consequence we are able to study which architecture decisions have evolved over time and which have remained stable. By defining a performance metric, based on aircraft efficiency, performance and market value, we are able to track the performance trends alongside the evolution of architectures. Data obtained from aircraft and mission specifications is used to calculate this performance, which is visualized and compared for regional, narrow-body and wide-body aircraft classes. Further, we compute the main effects and interactions of the architectural decisions to analyze the relative impact of each decision. 27

28 2.2 Problem formulation Scope In order to reasonably compare architectures, consistency in aircraft mission is required. As such this paper focuses on airliners, and excludes military aircraft, solepurpose cargo aircraft, and rotorcraft. To focus on aircraft serving a consistent set of needs, aircraft with a capacity of less than 30 passengers have been excluded, since design drivers for such aircraft (usually business aircraft or light aircraft) differ from commercial passenger aircraft. A further constraint that is imposed on the set of aircraft is that the minimum number manufactured is 10. This is done to exclude experimental aircraft, which may exhibit aircraft architectures that are not commercially viable, which is a proxy for commercially feasible production and value in the market. It is widely believed that the age of air passenger travel began with the inception of the DC-3 [82]; hence the scope of this analysis will begin from the 1930s, the decade when this aircraft was first produced. Note that dual functionality aircraft such as civil & cargo aircraft are included within the scope of this analysis Aircraft architecture definition During the concept creation process, the aircraft architect makes many decisions related to the architecture of the aircraft. A concept, within a given context, is the allocation of function to form. In the case of commercial aircraft the choice between single-aisle or twin-aisle aircraft is often viewed as an architectural decision. The framing of architecture decisions, as defined by Crawley et. al [30], encompasses elements of form which enable these functions to be carried out. The choice between twin-aisle or single-aisle, is a design parameter related to the performance of the aircraft, rather than an architectural decision that enables a new function to be fulfilled. In this paper, the set of architectural decisions are those that enable the most important functions to be fulfilled, such as lifting payload, storing payload, propelling payload, taxiing payload and maintaining stability. The process of defining aircraft architectures began from the first principles of theoretical system architecture. That is, a functional decomposition was carried out in order to determine the top-level decisions. For each decision a set of options was created representing the allocation of form to function. That is, an architecture, i I, consists of set of decision options defined by, i = {{d ab }} = {{d 1b }, {d 2b },, {d Nb }} = {{d 11, d12,, d 1m1 }, {d 21, d 22 {d 2m2 }},, {d N1, d N2,, d Nm2 }}, (2.1) where a = 1, 2,, N represents the architecture decisions and b = 1, 2,, m a represents the respective options that each of these decisions can take. An individual architecture is enumerated by down-selecting a single option, b, for each decision, a, 28

29 from each of the decision sets {d ab }; therefore {d ab } gives the set of discrete option values that a given architecture takes. This is distinct from setting the values of design variables such as wing planform area and wingspan, but rather encompasses the decisions with respect to the aircraft configuration at the top level of design. An example of such a decision is the vertical location of the wing, with three distinct options low wing, mid wing and high wing. In this instance, {d 1b = wing vertical location } ={d 11 = low wing, d 12 = mid wing, d 13 = high wing }, (2.2) whereby only one of the values in the set can be assigned to this decision. Before explicitly enumerating the architectures, we need to define the measure of performance, which will enable us to refine the architecture model Aircraft performance metric Aircraft performance can be split into multiple components, which include technical performance, operational performance and economic performance. Traditionally performance is measured using metrics such as range, take-off weight, cruise speed, fuel efficiency, lift-to-drag ratio, climb performance, direct operating costs, revenue passenger mile, aircraft price, time out-of-service etc. [100, 7, 8]. In order to capture the major performance measures at the overall aircraft level, an aggregated metric was devised based on previous work and the availability of historical data. The performance metric, M i, for a given architecture, i, can be decomposed into contributions from passenger carrying efficiency (PCE), aircraft technical performance (thrust-to-weight T ratio,, and maximum cruise velocity, V) and market value (list price, P). W [ ] [ ( P CEi P CE T ) min M i =w 1 + w 2 ( ) ] T W i W ( min P CE max P CE T ) min ( ) + T W max W min [ ] [ ] (2.3) Vi V min Pi P min w 3 + w 4 V max V min P max P min This is a linear weighted sum of four performance characteristics. The first, passenger carrying efficiency is defined as, P CE i = N paxi R i W 0i W ei W paxi N paxi (2.4) where, N paxi denotes the passenger capacity of aircraft i, R i denotes the range with maximum payload, W 0i the maximum take-off weight, W 1 ei the mass of the aircraft excluding any fuel or payload (operating weight empty ) and W paxi is the 1 Note that OWE includes all crew, all fluids, all equipment, but excludes payload (passengers in this case) and fuel weights. 29

30 weight assigned to a passenger (75kg person + 25kg baggage). This component can be thought of as the available seat kilometer per kilogram of fuel carried, which is an energy efficiency metric similar to the energy intensity metric devised in [75] or the miles-per-gallon metric for automobiles. That is this measures the ability of an aircraft to carry a specific number of passengers a given distance with respect to the mass of fuel required to do so. The payload fuel energy efficiency metric defined by Hileman and Katz as the payload moved a given distance per unit of energy, is similar to this but uses fuel energy rather than weight in order to compare alternative fuels, and it utilizes a different measure for range [57]. Despite its similarity, this was not utilized here in order to disaggregate the contributions of technical performance and operational performance of aircraft. The second component, the thrust to weight ratio, is defined as, ( ) T = T i N E (2.5) W i W 0i where T i denotes the thrust per engine and N E denotes the number of engines. This is a common measure of aircraft performance and often serves as a proxy for maneuverability [44]. Additionally in cruise conditions this is equal to the inverse of the lift-to-drag ratio, which is another commonly used performance metric. As mentioned above, the third and fourth components of this metric correspond to maximum cruise velocity and list price respectively. The maximum cruise velocity is another measure for aircraft performance and is architecturally distinguishing, whereas the list price penalizes more expensive architectures since in reality budget constraints are present. Each of these contributing factors are normalized and form a linear weighted sum, where w 1 = 0.3, w 2 = 0.3, w 3 = 0.3 and w 4 = 0.1. This is due to the fact that in this architecture analysis the metrics related to technical specifications are prioritized, however a measure of monetary value is necessary to capture economical constraints. As noted by Gardiner [46], a metric based on two or three dimensions trying to characterize the performance of such a complex engineering system, naturally only depicts part of the picture. Depicting each and every tradeoff that goes into the design of an aircraft would be nearly impossible, therefore the metric used in this paper is a realistic alternative for high-level performance capture Architecture decision subset selection In this context the aircraft architecture design space consists of the set of conceptual design decisions, which are most impactful on the performance metrics and most connected to other downstream decisions. Given this, it was necessary to analyze the initial set of decision-options, in order to evaluate which of them are the most important. The initial design space consisted of 28 decisions, each with multiple options yielding a design space consisting of over potential architectures. The decisions were aggregated from a variety of sources, many of which have been highlighted in the literature review section. Several methods for conceptual aircraft design are described in [98, 100, 111, 60], as detailed in the literature review. While these sources do not explicitly express architectures as discrete decisions in a well- 30

31 formulated architecting problem, they do detail the major considerations faced by an architect when attempting to satisfy stakeholder requirements through the design process. In Table 2.1, the set of high level functions and architectural decision-options associated with these can be viewed. The downselection to the subset of decisions that are most impactful on performance metrics consisted of reviewing existing literature and data analysis. Furthermore analysis of the frequency of options for each decision is carried out, and decisions with only one option over the whole data set are excluded as they dont display any variability. Therefore, by analyzing the frequency of occurrence of each option over time and the sensitivity on the performance metric, the decisions were winnowed. To carry this out, an architecture was described for each of the 157 aircraft in the database. It is worth noting that the options for each decision encompass a larger space than is present in historic aircraft, therefore the first step was to analyze which options have not been realized in commercial aircraft within the scope of this study. All the decisions in this sample of architectures were examined in order to observe the frequency of occurrence of each option over time. The goal of this was to determine which decisions only take the value of one option throughout the sample, thus implying that a dominant design exists and this decision should not be considered in our analysis. Having a decision with a single option cannot be defined as an architectural decision in the formulation described above, for the purposes of a historical analysis. Several of these decisions are shown in Figure 2-1 for the lifting payload function. It can be seen that the structural configuration decision takes only one option, standalone cantilever as opposed to wire or strut-braced structural integration. Likewise the wing configuration takes a single value of monoplane for the whole sample. These two single-option-value decisions contrast with the wing shape and wing vertical location decisions, which take different values through the sample. Therefore the former two decisions are removed from the set since they are not architecturally distinguishing. This analysis was repeated for all the decision-options, the entire set as well as the architectural decision subset can be seen in Table 2.1. The second method employed to refine the decision subset selection was the impact of the decisions on the performance metric. One might imagine that this could be evaluated solely by computing the sensitivity of the performance metric to each of the decisions. Since decisions are often coupled, the sensitivity analysis may be misleading. That is, one could envision a case whereby a non-influential decision takes a different value for a single poor performing architecture leading to a large computed sensitivity. The correlation between poor performance and this decision in such a case would not be causal. A better approach requires examination of existing literature and domain knowledge to establish a causal link between the performance metric and a given decision. Decisions such as the longitudinal location of horizontal control surfaces, the wing structural configuration and the number of wheels per landing gear are therefore excluded from the architecture definition. While the wing structural integration may dictate local subsystem performance, the overall effect on architectural performance is minimal and usually dictated by a higher priority decision such as engine type. The architecture definition must be detailed enough such that they are distinguishable in 31

32 Figure 2-1: Graph showing the frequency of different options for the decisions of how to lift the payload. These decisions include the wing configuration, the wing shape, the wing structural configuration, and the wing vertical location. the design space; however the decisions need to encompass the allocation of form to function that enables differentiation in the metric space. A subset of the original decisions emerged as a result, which are highlighted in Table 2.1, making the design space consist of over 20 million potential architectures, the vast majority of which have never been produced. Note that because of the formulation of the problem in the context of historical architectures all past concepts is not included, including concepts such as blended-wing-body aircraft [77], the MIT D-8, double-bubble concept [40] or a fully morphing aircraft concept [83]. 2.3 Analysis Data was collected from a multitude of sources including the annual Janes All the World s Aircraft dating back to 1911 [65], other resource books, Aviation Week and Flight Global archives as well as manufacturer and airline datasheets and archives. The database of airliners used in this analysis includes 157 distinct aircraft and 45 distinct architectures. Preliminary analysis of this dataset yields some interesting trends over time, which can be seen in the figures below. From Figure 2-2 it can be observed that over the past 80 years the performance has more than doubled. Compared to domains such as electronics, where there is exponential growth according to Moores Law, this may seem like a small improvement; however this is an immense improvement when one considers that automobiles have experienced approximately the same twofold improvement over the past 120 years [103]. Moreover aircraft safety has improved immensely, decreasing from over 50 incidents per million departures in the 1960s to less than 2 today [6]. Safety has not been incorporated as part of the performance metric mainly due to a dearth of available data from the early years of commercial aviation. 32

33 Option 1 Option 2 Option 3 Option 4 Option 5 Option 6 Option 7 Function 1: Lifting payload Configuration Monoplane Biplane Triplane Box Wing C-wing Annular Wing Wing Vertical Location High Wing Mid Wing Low Wing Parasol Wing Shoulder Wing Wing Shape Rectangular Tapered Delta Swept Back Swept Forward Structure Cantilever Strut-braced Wire-Braced Passive Control Shape Dihedral Anhedral Straight Gull-wing Polyhedral LE devices None LE Flap Slat Kruger Flap Leading Edge Slot Elliptical TE devices None Plain Flap Split Flap Slotted Flap Kruger Flap Double slotted flap Tip devices Winglets Wing Fence Downlets Split winglet Function 2: Storing payload Number of Fuselages BWB Flying wing Structure Monocoque Semimonocoque Shape Cylindrical Airfoilshaped Wing Integration Wing Box carrythrough Truss Box-shaped Geodisc Blended Ring Frames Bending Beam Function 3: Accelerating payload Engine Type Piston Prop Electric Turboprop Turbofan Turbojet Ramjet Number of Engines Engine Location Inside Vertical Tail Side of fuselage aft of wing Above/in fuselage Configuration Horizontal V-shape Tailless Horizontal location Aft of wing Canard Three surface Vertical location Fuselage (Inverted-T) Vertical Tail (cruciform) Vertical Tail (T-Tail) Behind fuselage Function 4: Maintaining stability Pitch Control Shape Swept back Tapered Straight Elliptical Angle Anhedral Dihedral Straight Polyhedral Configuration 1 Standalone surface 2 Standalone surfaces Attachment Location Fuselage On horizontal tail Mechanism Fixed Retractable Enclosed Landing gear type Wheels Wheels/Skid 3 Standalone surfaces Triple-tail Retractable not enclosed Wheels/Floats Yaw Control V-shaped Function 5: Taxiing payload Tandem Variable Sweep Triple Slotted Flap Under Wing Above Wing In Wing etc. Landing gear Arrangement Single Main Tail Dragger Bicycle Tricycle Quadricycle Tricycle w/ triple body gear No. wheels nose gear 1 wheel bogie 2 wheel bogie No. wheels per body gear Location of stowed landing gear In the Wing Wing Podded Fuselage Podded In the Fuselage Wing- Fuselage Option appears in historical database Option does not appear in historical database Decision included in final architecture set In Nacelle Table 2.1: Decision options corresponding to the functional decomposition at the third level. These decisions comprised the initial set prior to subset selection. 33

34 Figure 2-2: Graph of performance against time, showing aircraft architectures as blue dots and the year-averaged performance over time, including the standard deviation in this number. Key aircraft architectures are highlighted on the graph. The highest performing architecture according to the defined metric is that ofthe Tupolev Tu-144, a supersonic jet, which was in fact grounded after a mere 55 passenger flights and 102 commercial flights due to economic and safety reasons [69]. This highlights one of the limitations of examining the architecture performance using this view. Meanwhile, the Airbus A319 can be seen to be the highest performing of the rest of the architectures. Including cost factors as well as safety factors in the analysis is not within the scope of this paper, however it is a topic for further research Architecture variation over time The history of passenger aircraft development is filled with considerable nuances, involving geopolitics, government subsidies, various stakeholder interests and many other effects, which could not be captured in this analysis. It is commonly known that before the 1930s aircraft varied immensely in their architecture and technology, however tracking these is beyond the scope of this paper. Given the extensive data collected to create the database of architectures, it has been possible to track the number of distinct over time. Figure 2-3 shows the trend in the number of distinct aircraft being produced and the number of distinct architectures among these aircraft on the left. On the right, the ratio of these two at a given time in history and the yearly average performance for the architectures, are shown. This ratio gives us the variation in architecture at a given point in time by telling us what proportion of the aircraft manufactured at a given point in time have different architectures. 34

35 Figure 2-3: (Left) Graph of frequency of distinct architectures and distinct aircraft per year. (Right) Plot of the ratio of distinct architectures to distinct aircraft over time overlain with yearly averaged performance over time. It is evident from the figure that the number of distinct aircraft being produced has increased from approximately 3 to over 50 in the 80 year timeframe, for those aircraft that fall within the scope of the analysis. Intriguingly it can be seen that the number of distinct architectures increased in parallel with this from 1936 to the late 1950s at which point these two graphs diverge. The number of architectures remains fairly constant for the next 40 years until approximately 2000 where it can be seen that the variation in architectures is in fact decreasing. The solid line in Figure 2-3 clearly displays the decrease in variation of architectures over this period of time. This graph can be further analyzed using Figure 2-4. It is evident from graph of number of architectures over time that up until the early 1960s architectural innovation is occurring, since the number of distinct architectures continually increased in this period. Additionally, this timeframe is associated with a number of new technology introductions, particularly in engine technology, as can be seen in Figure 2-4. In the post-1960 era, the number of architectures remains relatively constant until the early 1990s. The ratio of number of architectures to number of aircraft decreases in this period, demonstrating that more aircraft of the same architecture were being produced. This suggests the end of the architecture exploration stage and a period of incremental innovation. This point is reinforced by the steady increase in performance during this period, which is indicative of incremental product innovations. From 1990 to the present day there are indications of architectural consolidation in civil airliners. The number of distinct architectures during this period is steadily decreasing, indicating the emergence of a dominant design. It is notable that the largest decrease in the ratio of architectures to aircraft occurs immediately after the EPA released environmental standards for hydrocarbon and other emissions in civil aviation. One might hypothesize that architectures which could not possibly comply with such standards begun to be phased out at this point in time. Along with architecture performance, such external driving forces may have caused this emergence of a dominant design. 35

36 Figure 2-4: Plot of the ratio of distinct architectures to distinct aircraft over time overlaid with yearly averaged performance over time, showing individual architectures. Key moments in the timeline of aircraft evolution are shown including the introduction of various technologies as well as regulations. A major driving force in the emergence of dominant designs in aircraft architecture is the inheritance of type certification for aircrafts and the development of aircraft families. According to Howe, in many successful aircraft, the design is a direct development of an earlier type. The reasons being that to be commercially viable a new aircraft must be a technical improvement upon its predecessors and risks are mitigated through utilizing existing knowledge and legacy designs. Howe states that therefore a lack of experience, uncertainty of design data, and customer reservations usually eliminate unconventional configurations. This phenomenon will not be analyzed in depth in this paper, however it is one of the many factors that has shaped these architectural trends Architecture performance In order to control for the effects of scale, it is possible to stratify the data set according to mission type. That is regional, narrow body and wide body aircraft can be separated to enable us to dive deeper into these trends. For each aircraft architecture the performance metric described in Section was evaluated to measure the ability to efficiently transport passengers a given distance as fast as possible, given the price and technical characteristics of the aircraft. Through stratification of the aforementioned different aircraft classes, we further ensure that architectures are comparable based on the more homogenous mission profiles. Included on the four graphs of Figure 2-5 are boundaries around the range of performance at a given time, with the exception of the two supersonic passenger jets, 36

37 Figure 2-5: Graphs of the yearly averaged performance of aircraft showing the performance bounds excluding outliers, and the mean performance for each set. The whole data set is shown in the top left graph, and the data has been stratified for regional aircraft (top right), narrow-body aircraft (bottom left) and wide-body aircraft (bottom right). Regional aircraft are classified as those with a capacity of fewer than 100 passengers, narrow-body those between 100 and 250 and wide body above 250, considering a single class economy layout. and a trend line of the average performance per year. As previously mentioned, an amalgamation of the many tradeoffs that go into the design of an aircraft into a single performance metric cannot possibly capture the whole picture. Each individual architecture has been optimized for its specific mission purposes and thus the variance is accounted for by this phenomenon. The motivation for stratification can be observed in this figure where the differences in the mean performance of each set is evident. As expected the effects of scale can be seen, that is, the larger aircraft types are generally higher performing due to the efficiencies captured at this scale. The overall performance trend can be seen to follow a curve similar to that of a technology S-curve [113, 114, 108, 28], starting with a period of constant performance, followed by a period of performance increases and engine with a decline in the rate of these improvements. Examining the trends of the different aircraft types in terms of the overall group rather than the yearly average, it is noticeable that they too follow a similar trajectory. As mentioned previously, the majority of architectural exploration occurred in non-commercial passenger aircraft prior to the 1930s, therefore one would expect a period of constant performance during the very early years of aviation, which would emphasize this S-curve shape. In the period from the 37

38 early 1990s until the present day the progress of the previous years slowed down as architecture performance matured and approached a ceiling. This phenomenon is an innate characteristic of the aforementioned architectural consolidation, since incremental or modular innovations eventually lead to diminishing returns in performance [108]. From Figure 2-5 it can immediately be seen that the dates of entry into service between different aircraft types are staggered, that is, as time progressed larger aircraft became more widely available. There are common architectures across these stratified groups and variations in architecture within them, which is shown in Appendix A. Therefore this trend in size can be attributed to technological advances, such as in structures and materials with the inception topological optimization and composites. Comparing these graphs with that of Figure 2-4 one can make observations about these technological innovations in the context of the various aircraft types. The start of narrow-body aircraft coincides with the introduction of turboprop aircraft and likewise the beginning of wide-body aircraft coincides with the introduction of turbofan aircraft. On an aircraft systems level, the advantages of each of these technologies is well-known [68]; however there has been little research done from the perspective of architectures in a tradespace. This work shows the influence of engine technology on aircraft architecture. With new engine technologies on the horizon, such as the geared turbofan, it is possible that we will see a new age of aircraft architectural exploration driven by engine innovation. An interesting observation for wide-body aircraft is the increase in performance for these has been very marginal over the past forty to fifty years. This is in stark contrast to the regional and narrow body types, which have experienced more than doubling in performance. Considering that architectures such as that of the Boeing 747 have persisted for over forty years, this observation is hardly surprising; in fact the variation of architectures within this group will be shown to be fairly small, which may contribute to this effect. Additionally it has been well documented that the beneficial performance effects of scaling aircraft diminish beyond a certain size [118, 81]. Hence one could conclude that, given architectural consolidation and the dominance of incremental innovation, these aircraft lie at the limit of diminishing marginal returns on performance Architectural decision space With a large database of architectures, one can utilize data mining techniques in order to extract significant insights into the implications of architectural decisions on performance. In order to do so we compute the sensitivity of each decision in the metric space, and the connectivity or degree of coupling of these decisions. Each of these is explained below. Sensitivity The sensitivity of a decision in the metric space is a measure of the degree of influence of that decision on that particular metric [30]. In dealing with decisions with more 38

39 than binary variables, the sensitivity is calculated for a group of architectures i I, with possible architectural decisions D j that have option values given by z Z. N 1,z is the number of architectures for which the value of decision D j is set to option z and N 0,z is the number of architectures where D j is not set to z. In this way we are averaging the difference in the performance metric M(x) with the decision taking the value of one option over not taking the value of that option. S(D j, M) = 1 1 Z M(i) 1 M(i) N 1,z N 0,k (2.6) z Z {i i Dj =z} {i i Dj z} Connectivity Since architectural decisions are often coupled, a measure is required to capture this interdependence, which influences change propagation in the design process. A design structure matrix (DSM) was used with the constraints representing relationships between the various decisions forming the rows and columns of the matrix. It is evident that there is an element of subjectivity in this formulation, firstly because it is dependent on the formulation of the architectural decisions [19] and secondly it is dependent on a priori knowledge in which the definition of a relationship is fuzzy and may be subjective. Despite these shortcomings the DSM is a commonly used tool in system architecture due to the ability to capture and visualize complexity in engineering systems. The connectivity using this method is simply the number of connections between one decision and the others, which are captured in aircraft design books such as Roskam [100] and Raymer [98]. The resulting relationships between decisions are shown in Figure 2-6 in the form of a graph, rather than a DSM in order to visualize these better, where a directional arrow indicates either a strong or a weak influence of the source decision over the sink decision. Decision space view Plotting the sensitivity against the connectivity for each of the architectural decisions, architectural decisions can be prioritized. The results for the whole dataset of architectures can be seen in Figure 2-7. The architectural decisions can be prioritized by categorizing the space into a 2-by-2 matrix of sensitivity against a normalized connectivity, with the highest priority decisions being located in the top right quadrant. It can be observed for this formulation of the aircraft architectures, the decisions pertaining to engine type and wing shape are those that take priority. It is worth noting that by making a given decision, the design space becomes restricted by any constraints that the value of the decision imposes. For example, by selecting Engine Type as a Turboprop, the Wing Vertical Location decision is usually going to be High-wing, due to the physical constraints imposed by having an engine with a large propeller. In other words, these two decisions are closely coupled; therefore by ranking the decisions in order of importance and keeping track 39

40 Figure 2-6: A graph of the decision space showing interconnections between the decisions. The arrows show the direction of influence of one decision over another. of cascading constraints during architecture enumeration, any potential conflicts can be minimized. The decisions in the upper left quadrant are made second these are decisions that have a large effect on performance of the architecture depending on the value they take, but are not closely coupled with many other decisions. In this context it is important to note that the correlation between a decision and the aircraft performance does not imply causation. For example, the horizontal tail shape is often closely coupled to the shape of the wing and the engine type, therefore the high sensitivity is as a result of this coupling rather than as a direct effect of the value taken by this decision in isolation. Interestingly, the location of the landing gear and the landing gear configuration decisions seemingly have a large effect on the aircraft performance. The reason for this is that higher performing architectures are associated with larger aircraft in general, and larger aircraft require different landing gear configuration and landing gear location than smaller aircraft. Once again in this instance the correlation can be explained by the connectivity between the decisions, rather than a direct causation of these decisions. In addition, upstream effects such as the aircraft mission profile which influence the aircraft decisions and scale also have an impact on the architecture performance. It is possible that in further work mission decisions could be formulated as architectural decisions and included in a similar analysis. In the bottom left quadrant are located the lowest priority decisions, one of these being the number of engines on the aircraft. This result suggests that selecting the number of engines does not change the performance of the aircraft significantly for 2 engines versus 3 or 4 engines, for example. Considering the performance metric defi- 40

41 Figure 2-7: Decision space view of sensitivity of the decisions to the performance metric against connectivity of each decision. This matrix separates the decisions by priority according to the se two dimensions, as shown. nition, factors such as direct operating costs including maintenance are not captured in this view, which is one of the major benefits in a fewer number of engines. Relationship between architectural decisions and performance Given this dataset, it is possible to visualize the relationship between architectural decision-options and performance. Plots of four decisions showing the mean, interquartile range, range and perceived outliers are shown in Figure 2-8. Examining the Engine Type decision it can be seen that turbofan engines are associated with architectures of higher performance. One must note here that the correlation does not imply that a high performing architecture is caused by any given decision, since there are interaction effects and coupling which are not displayed in this view of the architectures. Similarly delta wing shapes are the highest performing followed by swept-back wings, due to the fact that this architectural decision is associated with supersonic passenger aircraft. The correlation between number of engines and performance is interesting, since it can be seen that increasing the option from two to four engines corresponds to a decrease in performance on average. The time variation of these decision-options can be seen on the graphs in Appendix???. In the case of the Number of Engines, clusters of each option can be seen in the time-performance space. It can be seen that on average, over time, architectures have shifted from a four-engine architecture to a two-engine architecture, in parallel with an increase in architecture performance. 41

42 An improvement in engine performance and reliability enabled this switch, which improves passenger carrying efficiency and decreases maintenance costs for twin-engine aircraft. Figure 2-8: Boxplot graphs of the performance of each option for four different decisions, showing the mean as a red line, the interquartile range as a blue box, the range excluding outliers as dotted lines and the outliers as red dots. The four decisions shown are Engine Type (top left), wing shape (top right), number of engines (bottom left) and wing vertical location (bottom right). 2.4 Conclusion In this chapter a method for considering architectural decisions in historical conceptual aircraft design has been developed. Consequently the trends in historical aircraft architecture and performance have been analyzed and relationships between decision-options and performance have been examined for the whole data set and each aircraft type. The sensitivity and connectivity of these decisions to performance metrics has been studied enabling the prioritization of architectural decisions. The benefit of this analysis in aircraft conceptual design is in applying a structure, which makes explicit the key decisions and their implications on architectural performance. While previously aircraft conceptual design was based on domain knowledge and prior experience of the architect, the results here have provided a quantitative approach using historical architectures to prioritize the major architectural decisions and their 42

43 relationship to performance. It has been shown that over the past eighty years, the variation in architectural decision-options has decreased and a dominant architectural design has emerged. In parallel with this the performance of aircraft on average has increased by over two times. The architecture performance over this period has followed a trajectory similar to that of a technology S-curve. These trends implies that passenger aircraft have gone through a period of architectural innovation followed by incremental and modular innovation mainly in propulsion and materials technologies. Therefore there are signs to suggest that a dominant design has emerged during this period, which may have implications on future aircraft. The method developed can be expanded to include more decision-options, in order to accommodate architectures that have not been historically produced. For example it has been shown that the blended-wing body aircraft could offer significant performance advantages over current architectures, according to Liebeck [77]. Another main conclusion from this analysis that motivates the following chapters of this thesis is the close coupling between engine technologies and aircraft architectures. It has been shown that, historically, innovations in engine technology have led to aircraft architectural innovation. Since over the past years it has been shown that there has been consolidation to a dominant architecture, an interesting question which motivates this research is when might we expect this dominant architecture to be broken, that is when might we expect another period of aircraft architectural innovation. Extrapolating the historical trend of aircraft architecture and engine technology coupling, one might assume that engine technology advances in the future would be the major driving force leading to the dominant architecture being broken. Therefore assuming the current dominant architecture and forecasting engine technology advances, over the next few chapters we shall examine: 1) the extent to which current airframes such as the 737 and A320 can continue viably increasing their performance 2) the extent to which future airframes with same dominant architecture but variable airframe design parameters can continue viably increasing their performance. In order to carry out this analysis an airframe-engine model has been created enabling the comparison of aircraft performance, given certain engine technological trends. 43

44 44

45 Chapter 3 Dominant architecture airframe-engine model 3.1 Introduction In Chapter 2 the coupling between engine technologies and aircraft architecture was shown through the historical evolution of aircraft architecture and the high sensitivity and connectivity of architectural decisions related to the engine. The consolidation of aircraft architecture to a dominant design has been highlighted. The question was posed as to when we might expect this dominant architecture to be broken, entering a new period of architecture exploration. The influence of engine technology on the design vector of the aircraft architecture is expected to continue in the future, therefore the breaking of the dominant architecture is expected to come as a result of advances in engine technology. There has been previous work done in predicting point solutions of future aircraft architectures at the corners of the design envelope, such as the blended-wing-body, double-bubble, flying wing or many others [40, 77]. While there has been some work done on the conditions that may lead to the current dominant architecture being broken [109], there has been a little research done into the trajectory that may lead us to that point, as visualized in Figure 3-1. Looking at the current aircraft architecture, for how much longer can we expect large enough improvements in performance such that searching other areas of the architectural design space will not be necessary. To phrase this in another way, given current technological trends within the dominant architecture, will the expected returns in performance diminish to a point in which we will require architectural innovation in order to see further improvements. As mentioned in the previous chapter two motivating questions emerge from this: 1. To what extent can current airframes continue to increase their performance, and when will the trend of installing new engines no longer be viable? 2. To what extent can the dominant architecture continue to increase in performance, without constraints on the airframe geometry? In order to answer such questions a model in which to analyze airframe and engine geometry and performance is required. The required inputs to such a model should include aircraft mission parameters, airframe technology, and engine technology, such that any changes in these can be analyzed in terms of the outputs including aircraft performance and geometry. Before describing a potential model it is necessary to analyze the major trends in aircraft performance in terms of airframe and engine 45

46 Figure 3-1: Evolution of aircraft architecture over time including future forecast of breaking point of the current dominant architecture. The architecture space has been collapsed onto one dimension on the y axis. technology. This is necessary in order to design a model which is capable of handling this range of technology inputs producing accurate performance results. In this context the technological trends can be categorized into those associated with aircraft propulsion, aerodynamics, structures, materials and control systems. It has been shown in Chapter 2 of this thesis that historically, engine technology has been the major driver of aircraft architectural changes. That is, in order to gain the full performance benefits of an improvement in engine technology or a change in engine architecture, the aircraft architecture has had to evolve over time. For example the transition from a High-wing, T-tail architecture associated with propeller-driven engine architectures, to a Low-wing, Inverted T-tail architecture associated with turbofan engines, due to the mutual advantages of these couplings. Based on this analysis, it was put forth that future aircraft architectural changes may be expected to follow the same trend. Hence the major improvements in the near future, for the current dominant civil passenger aircraft architecture, are expected to occur in propulsion [12]. For example the inception of the geared turbofan engine, which can be considered a subset of turbofan engines for the purpose of the architectural decisions, is set to cut specific fuel consumption according to Rolls-Royce [110]. Therefore by extrapolating the trends in current turbofan or geared turbofan engines, it is possible to examine how the current dominant architecture design parameters and architectural performance will change. Over the next two sections the extrapolation of such trends will be examined before describing the model to be used in the analysis. Before presenting this we will give a brief summary of the resources used to develop the method used in the model. There is an abundance of literature on the aircraft preliminary design stage for the dominant aircraft architecture. Commonly used methods and resources in academia and industry include those presented by Raymer [98], Roskam [100], Torenbeeck [111], Howe [60], Sadraey [101], Anderson [7] and many more. The methods presented in 46

47 these books tend to use empirical formulae based on statistical analysis of existing aircraft for high level design, given a specific aircraft architecture. For example, the fuel weight fraction method which is used to compute the fuel requirements and aircraft takeoff weight, for a specific mission, is based on knowledge of weight fractions for particular segments of a mission such as take-off or landing. Most of these methods for designing civil passenger aircraft are based upon the dominant design, defined in Table 3.2. A review of the literature pertaining to optimization of aircraft design, given a particular architecture, is given in Section 1.3 and therefore will not be repeated here. The airframe model described in this section is based mainly around the methods of Raymer, however it utilizes analytical methods where possible and draws upon methods from the other resources where necessary. The optimization is based on a semi-analytical, semi-empirical formulation of the problem and constraints. The methods used will be described further in the sections of this chapter, however they are also described in explicit detail in Appendix B. 3.2 Trends in civil aircraft technologies Before embarking on formulation of a model, we shall carry out a brief review of technology forecasts which form the basis for the analysis in this section. There have been many national and international efforts to commit resources to research and development for the future of aviation, in particular aircraft propulsion. In the US, NASA s Environmentally Responsible Aviation (ERA) project put forth 5 main research areas including advanced ultra-high bypass ratio (UHBR) engine designs (BP R > 15) for specific fuel consumption reduction and noise reduction, as well as advanced combustor design for low NO x emissions [20]. Meanwhile in the EU, the Clean Sky Initiative, a joint technology initiative between European aerospace stakeholders, have dedicated resources to six projects, one of which is the sustainable and green engines initiative which has the same objective as the NASA project [50]. Similar projects exist for the airframe and architecture exploration, indicating the drive by governments and the aerospace industry to make aviation more sustainable. Over the next few sections we will try to predict some of these engine technology trends and their implications on the airframe. The drivers for the technology advances are to make aircraft more environmentally friendly, more reliable, safer, and more fuel efficient. These goals can be aggregated into a single main aim, which is to lower the operating costs for airlines. According to Birch [21] and Green [109], in terms of engine technologies the major technology trends that will result in these objectives are: 1. Increasing bypass ratio (BPR) which increases propulsive efficiency and thus decreases specific fuel consumption. 2. Increasing turbine inlet temperature (TIT) and overall pressure ratio (OPR), which increases thermal efficiency. 3. Increasing component efficiency, that is the polytropic efficiencies of fans, compressor stages, and turbines. 47

48 4. Decreasing structural weight through higher material specific strength, decreasing specific fuel consumption. NASA has set goals for dramatically improving noise, emissions and performance as part of their Subsonic Fixed Wing Project in the N+3 timeline ( ) [51]. These technology goals, relative to 2005 technology, include reducing noise by 71dB, NO x emissions by 80% and fuel consumption by 60%. Benzakein [20] has listed key technologies to be developed in order to meet these goals in the required time frame. The engine technologies enable high efficiency, high OPR gas generators, including better hot section materials such as Ceramic Matrix Composites (CMCs), tip end and wall aerodynamics and controls for better efficiencies, low NO x combustors, and highly loaded front block compressors. According to this research, these technologies will enable bypass ratios to be pushed from their current levels around to over 20, by minimizing the core size as will be discussed in Section 3.3. In order to better demonstrate the effects of these technology improvements on aircraft performance, the following section will run through a brief analysis from first principles, showing the key parameters involved. The data and trends will be showcased for these key parameters since they will subsequently be used as inputs for the model First principle analysis of technology drivers Reducing specific fuel consumption To begin with we recall the aircraft performance metric devised in Chapter 2, [ ] [ ( P CEi P CE T ) min M i =w 1 + w 2 ( ) ] T W i W ( min P CE max P CE T ) min ( ) + T W max W min [ ] [ ] Vi V min Pi P min w 3 + w 4. V max V min P max P min (3.1) Considering this metric, for modern passenger aircraft V and ( ) T W are essentially constant due to mission specifications being similar. Furthermore predicting the price is beyond the scope of this analysis which is more focused on technology aspects; therefore the focus is on maximizing the passenger carrying efficiency (PCE=the ratio of passenger miles to block fuel). In order to increase the PCE term the number of passengers miles (range multiplied by pax) for the amount of fuel used must be increased. According to Peeters [92] who uses a similar energy intensity metric, this depends on aerodynamic efficiency, weight efficiency, cabin density, and engine efficiency. The Breguet range equation captures all of these terms and is given in Equation 3.7 which can be rearranged for fuel weight to give, [ ( ) ] R SF C W f = exp V ( ) 1 [W L e + W p ]. (3.2) D 48

49 This will be described in further detail in the description of the optimization process of the model in Section From this equation, if a mission is defined, that is the range R, velocity V, and payload weight W p are specified, there are three parameters that can be altered to try to decrease the block fuel required for the mission: 1. ( ) L D 2. SF C 3. W empty. 1. Increasing lift-to-drag ratio: This can be done by either increasing the lift or decreasing the drag, simultaneously or seperately. The maximum lift-to-drag ratio is given by [49], ( ) L D max π = b, (3.3) 4kS D0 where b is the wing span, k is the vortex drag factor and S D0 is the wetted area at zero lift. According to Marec drag has a large implication on direct operating costs, directly contributing approximately 22% [78]. In order to increase this ratio, we could increase span, or decrease vortex drag factor or wetted area at zero lift. Vortex drag factor for the swept-wing in the dominant architecture is fairly constant at 1.2 and there is not much that can be done to change it [49]. Increasing span with constant S D0 will increase ( L D) proportionally. However in reality S D0 will also increase; moreover increasing span increases structural weight and is close to optimum for current material properties. In addition to this airport gate requirements are a constraint on aircraft wing span, since the aircraft must operate within the bounds of these constraints. Decreasing wetted area could lead to a reduction in drag. Given a specific mission profile, within the bounds of the dominant architecture this is not practically feasible. That is, reducing wetted area implies reducing the geometry of the aircraft; given that passengers expect a certain amount of cabin space, this option is not considered a realistic change within the scope of this thesis. Boundary layers over the whole aircraft are generally turbulent, therefore technologies that promote laminar flow act to reduce drag caused by skin friction, such as natural laminar flow of hybrid laminar flow control technologies [49, 119]. 2. Decreasing specific fuel consumption: The specific fuel consumption can be represented as, SF C = ṁf T = f ṁ c T, (3.4) 49

50 where, T is the engine thrust, ṁ f is the mass flow of fuel into the engine, ṁ c is the mass flow of air in the engine core and f is the ratio of mass flow of fuel to mass flow of air in the core of the engine. The equation can be restated as, SF C V η th η prop Q R (3.5) where Q R is the calorific value of fuel and V is the flight velocity. Hence the focus of this analysis is on increasing the thermal and propulsive efficiencies η th and η prop, which intuitively makes sense. Thermal efficiency, as defined in Appendix B, involves increasing the efficiency of the engine core, which is done by increasing overall pressure ratio (OPR), temperatures, component efficiencies and minimizing losses. Improvements happen as the engine gets hotter, which is limited by the turbine inlet temperature (TIT) and bounded by the stoichiometric limit of the hydrocarbon fuel. Current materials are limited due to their creep resistance properties, creep being the tendancy for a material to permanently deform under long exposure to high temperatures and stresses. Thus improvements in material creep characteristics (increasing TIT) and component efficiencies are a possibility for improving aircraft performance. Increasing OPR and TIT increases NO x emissions [49], therefore this would require advanced combustor design to minimize this impact. This is an example of where environmental impact and fuel efficiency are competing objectives. The thermal efficiency of an engine has a theoretical maximum for a given OPR, and exhibits diminishing returns as OPR is increased [21]. Considering a Brayton cycle for gas turbines, thermal efficiency can be written, given 100% efficient components. Increasing OPR increases efficiency but at a given OPR the maximum theoretical thermal efficiency is constrained. as η th = 1 (OP R) 1 γ γ The propulsive efficiency, as defined in Appendix B, is increased by increasing the propulsor, therefore increasing the bypass ratio. Although increasing bypass ratio is usually associated with reductions in specific thrust, it is not necessarily always so. There is a coupling between BPR and OPR/TIT; by increasing the latter two one could fit an engine with a smaller and hotter core, while mainlining constant thrust and overall mass flow, resulting in no change in efficiency [21]. In such a case, increases in aircraft performance result from decreases in engine weight and drag from a smaller engine. It is worth noting that increasing bypass ratio without reducing the core size has the added advantage of reducing noise for the current turbofan architecture. Propulsive efficiency attempts to minimize the difference between the velocity of the jet and that of cruise. Thus propulsive efficiency theoretically can reach 50

51 100% however that would require the jet velocity to be equivalent to the velocity of travel of the aircraft which is not possible in practice. To gain sufficient thrust one could either quickly move a small mass of air, or slowly move a large mass of air with the latter having a higher propulsive efficiency. To increase the mass of air moved one needs to increase the bypass flow. Hence as bypass ratio tends to infinity propulsive efficiency tends to 100%. There are other potential future options which involve changing engine architecture. The following analysis is focused on the dominant architecture therefore these will not be considered, however distributed propulsion, or unducted fan engines [109, 47] have been proposed due to their potential to increase propulsive efficiency. Furthermore alternative fuels or sources of energy such as electrically powered aircraft have also been considered as future options [99]. 3. Decreasing the structural weight: Since fuel is expended carrying the aircraft structure as well as the payload, it is beneficial for aircraft performance to maximize the weight of payload carried per unit weight of aircraft structure required. The trend of structural weight for civil aircraft can be seen in Figure 3-2. This figure was created from a database of aircraft with the ( ) W 0 W e is dominant architecture. Since there is a lack of available data for fuel weight, ( W used as a proxy for initial W final ). This term is derived from the Breguet range equation 3.7, whereby a higher value increases performance. It can be seen that structural efficiency as defined by this metric has not changed much over time. With the inception of carbon fiber reinforced polymer (CFRP) materials for the majority of the structure in the Boeing 787 and Airbus A350, one should expect an increase in this value. Poll highlights a similar trend stating that despite the ever increasing use of lightweight structural materials, the empty weight of new aircraft designed for a given payload-range has not changed significantly over the past 40 years [96]. He states that the potential weight reduction has been traded for improved passenger facilities and increases in lift to drag ratio through higher wing-spans and in propulsion efficiency through higher bypass ratios. Although this is not indicative of technical aircraft performance it is a consequence of the requirements of today s society. Therefore although minor improvements in structural efficiency could result in increases in aircraft performance, in reality these tend not to happen due to operating procedures. In order to limit the scope of the analysis decreasing structural weight has not been included as part of this analysis of aircraft performance Summary of review Taking the above drivers of performance a rough first principles analysis can be carried out. Assuming that SFC can be reduced by 80% [51], structural weight can be reduced by 10% [117], and lift-to-drag ratio can be increased by 20% [77] the trends in Figure 3-3 is derived. Taking baseline level of technology for a single-aisle aircraft such as a 737, and applying these improvements in SFC, W e, and ( L D), the normalized reduction 51

52 Figure 3-2: Trend of structural weight as a percentage of maximum takeoff weight over time. in block fuel weight is visualized, with everything else remaining constant. It can be seen that SFC and W e improvements offer the largest potential for fuel reduction. Based on the fact that reductions in structural weight are often offset by the operator as mentioned above, the focus of this thesis has been narrowed to SFC. The above review of aircraft performance from first principles motivates the analysis in this chapter. It is evident for the current dominant architecture that the majority of performance increase will come from increasing engine performance. This follows the recent trend of aircraft manufacturers involving re-engining their existing airframes, as mentioned previously. Several engine technologies have been highlighted above in order to increase the thermal and propulsive efficiencies of a turbofan engine. The levels of technology development over time needs to be forecasted in order to include the temporal dimension in the results of this thesis. While it would be valid to use the absolute limits of each of these parameters for the current dominant architecture, valuable information can be obtained from forecasting when these changes are expected to occur. This will be carried out over the following sections using trends predicted in existing literature where possible and our own predictions where there are gaps in the literature. 3.3 Forecasting airframe and engine technologies Bypass ratio trends In the early 2000s it was widely considered that the bypass ratios were at the economic optimum for the available technology [21]. This was believed to be so since, any gain in propulsive efficiency by a reduction in specific thrust, would be offset by increasing nacelle weight and drag due to increasing fan diameter. In addition to this, stringent noise requirements lead to higher bypass ratios, which in turn increase engine size 52

53 Figure 3-3: Variation of normalized block fuel weight for three different scenarios of improving technologies. It is shown that all else held constant, SFC has the largest potential for impacting aircraft performance. adding weight and installation drag, and thus offsetting any performance gains [21]. That being said, according to Bower [26], all designs in a MDO study of single aisle aircraft show benefits in operating costs, and CO 2 and NO x emissions, from using higher bypass ratio engines than the current single aisle commercial transports. It is desirable to increase bypass ratio (BPR) since it enables increases in propulsive efficiency, resulting in reductions in specific thrust. According to analysis done by Birch, bypass ratio increased from 1 to 2 in the 1960s, reaching 5-6 in the 1970s and 1980s, increasing further to 7-9 in the 1990s [21]. More recent turbofan engines have changed architecture slightly to incorporate a gear, thus enabling turbine and fan speeds to differ, resulting in higher bypass ratios using current material and structural technologies. An example of this is the CFM LEAP 1A engine which will push bypass ratios to around 11 by the end of 2016 [63]. NASA in collaboration with MIT and others, have carried out several studies to predict future trends in airframe and propulsion technologies, including their N+2 and N+3 studies for the D8 and H3 concepts [51, 66]. In addition, Rolls-Royce predicted engine performance based on business as usual scenarios, inclduing a direct drive turbofan (DD) and geared turbofan (GTF) [110]. These predictions indicate a 20% increase in efficiency for the DD (BPR=11) and a 25% for the GTF (BPR=15) by 2025, relative to technology from the mid 1990s. The dispursal project [64] included predictions for a ultra-high bypass engine study, driven by the European Commission s Flightpath 2050 challenge. All of the above predictions have been plotted on the graph in Figure 3-4, along with data from an engine database in order to predict the trends in bypass ratio. Additionally data solely from single-aisle aircraft is used for Figure 3-5, since this gives a smaller confidence interval, therefore a narrower prediction. The data in the database has been extracted from a variety of sources including Jane s [65], and includes turbofan engines used to power aircraft 53

54 above a maximum takeoff weight of 100 tons. A linear and power-law trendline have both been fitter to the data. Although Ballal and Zelina [16] have a linear trendline for bypass ratio against time, if one plots proposed concepts on the graph it is clear that a power law fit is more appropriate. This of course will not go on increasing so one would expect the trend to follow the technology s-curve shape as the benefits of increasing bypass ratio experience diminishing returns. Since there is uncertainty as to when this may take place, that is it could be in 20, 30 or 50 years, the best fit curve remains a power-law curve for the purposes of this thesis. Since there is much uncertainty in this prediction, 85% confidence bounds have been included on the graph to account for this. It is unclear about the limitations of the current turbofan architecture, however as mentioned before there are many other potential engine architectures worth exploring for bypass ratios above 30, such as the unducted fan [121]. Figure 3-4: Bypass ratio variation over time, for turbofan engines for aircraft with a maximum takeoff weight above 100 tons. Linear and power trendlines are fit to the data, with 85% confidence intervals shown. Real engine data is shown as blue circles while engine concept data from literature is shown as green triangles Component efficiency trends Each turbofan component has an efficiency associated with it, which are inputs to the model. In [16] Ballal & Zelina, the improvements in the components for aeroengines are discussed in terms of the technological advances that have made these possible. That is compressor, fan and turbine efficiencies have increased over time due to computational fluid dynamic (CFD) analysis optimizing blade geometry, the 54

55 Figure 3-5: Bypass ratio variation over time, for turbofan engines for single-aisle aircraft. Linear and power trendlines are fit to the data, with 85% confidence intervals shown. Real engine data is shown as blue circles while engine concept data from literature is shown as green triangles. use of advanced materials such as titanium aluminide compressor blades, and more precise manufacturing processes minimizing of losses through leakage and optimizing stall margins. There are still potential improvements to be made in terms of material properties and manufacturing processes, such as the use of additive manufacturing to create more complex geometries with more efficient cooling systems [59]. However the data in Table 3.1, shows that these improvements will not yield very significant changes in component efficiency, since they are already very high today. Although maximum possible values of component efficiency differ depending on the resource used, the variation is fairly small and will not have a large effect on the results. In the analysis that follows, the maximum possible component efficiencies are assumed to have been reached in the time frames that are being analyzed. That is, it is assumed that the advances in manufacturing processes and material science will have been developed in this period. This assumes that the possible improvements in component efficiencies are not the limiting factor in the increase in performance, since there is not much scope to increase these beyond their current level. It assumes that improvements in technologies related to increasing bypass ratio are more of a limiting factor. 55

56 Component Today [%] Maximum predicted [%] Fan [116] Compressor [55] Turbine [54] Combustor [69] Diffuser [102] Nozzle [102] Table 3.1: Comparison between component efficiencies today and the maximum possible efficiencies from literature Turbine inlet temperature and overall pressure ratio trends Previously it has been described how turbine inlet temperature (TIT) limits the hotness of the core, which in turn limits thermal efficiency. Ballal & Zelina show the trend in TIT from 960K in the early 1940s to approximately 1700K today, or 2000K if one includes cooling technologies [16]. Since the stoichiometric turbine inlet temperature is close to 2590K, there is considerable room for improvement. Several methods have been proposed to increase this temperature, however they rely on considerable research and development investments. These include developing new materials such as Ceramic Matrix Composites (CMC) or carbon-carbon composites, developing new thermal barrier coatings, or developing new manfacturing technologies such as additive manufacturing (AM) enabling more efficient cooling systems to be designed and manufactured [16, 69, 20]. Koff shows that the effect of advanced cooling combined with single crystalline materials would enable temperatures close to the stoichiometric temperature to be reached [69]. Meanwhile Benzakain proposed that technologies to be developed on a timeline should include a 1900K high pressure turbine blade and vane [20]. Graphs of TIT trends against time and specific thrust are shown in Appendix E. According to Birch the increases in TIT will only result in increasing thermal efficiency, if increasing together with overall pressure ratio (OPR) at a constant component efficiency [21]. The maximum possible OPR increases as TIT increases thus allowing for the benefits of these changes to be realized. Since Birch has predicted a TIT of 1800K 1850K by 2025, it is assumed that the BPR trend is enabled by this technology trend. 3.4 Overview of model Given the above trends, it is necessary to create a model for the dominant architecture, taking as inputs aircraft design parameters and engine technology parameters, and computing a feasible aircraft geometry and the architectural performance. Over the next chapters, the term dominant architecture refers to the architecture detailed in Table 3.2, based on the decision-options in Table

57 Decision Option number Option Wing Vertical Location 3 Low Wing Wing Shape 4 Swept Back Wing Passive Control Shape 1 Dihedral Engine Type 4 Turbofan Number of Engines 2 Two Engine Location 5 Under-wing Pitch Stabilizer Vertical Location 1 Fuselage (Inverted T) Pitch Stabilizer Shape 1 Swept Back Landing Gear Arrangement 3 Tricycle Location of Stowed Landing Gear 5 Wing-Fuselage Table 3.2: The dominant aircraft architecture. Given that engine architecture is a major driver in the vector input for aircraft architecture, it is reasonable to believe that future developments in aircraft propulsion may lead to the breaking of the current dominant aircraft architecture. In order to examine this potential scenario it is necessary to model the aircraft design and performance, the engine design and performance and the interactions between these two. For a civil passenger aircraft displaying the dominant architecture, given a vector of inputs for mission specification and initial aircraft definition parameters, x, the model, F sizes the aircraft in terms of weight, computes the aircraft drag, optimizes for engine and landing gear integration while maintaining stability, and outputs aircraft and engine geometry and performance, y, that is y = F (x). A high-level schematic of the overall model can be seen in Figure 3-6. Each of the blocks in the diagram is further explained in Appendix B. 3.5 Inputs The model takes as inputs details about the aircraft mission; as was mentioned in Chapter 2, each aircraft is optimized for a specific mission which has a major bearing on the final geometry and performance. Additionally technology trends, particularly those pertaining to the engine, are used as inputs in order to forecast future aircraft performance Mission profile inputs A design mission profile for the aircraft is defined through various mission segments. A typical mission profile can be observed in Figure 3-7, where segments are labeled in between the number markers. It is worth noting that in real operations the aircraft will rarely follow the exact design mission due to specific airline operations as well as external factors such as air traffic management. The mission profile is often 57

58 Figure 3-6: High level overview of the airframe-engine model, including inputs and outputs determined by market analysis of currently operational aircraft and identifying gaps, or upgrading existing aircraft. The two main factors in this analysis usually are the aircraft payload and range; other factors such as cruise altitude and Mach number are also considered however with the current dominant architecture these tend to be fairly constant across all civil airliners. Figure 3-7: Typical mission profile for air passenger aircraft showing the various mission segments. The specific parameters used to define a mission profile are showing in Table 3.3. These include parameters defined by the aircraft manufacturer based on market analysis or customer requirements, and the requirements imposed by the Federal Aviation Administration (FAA) in the Federal Aviation Regulations Part 25 (FAR 25), governing airworthiness standards for transport category airplanes [1]. 58

59 Input parameter Variable Description Units Typical value Range r cruise The design cruise range of the aircraft km 5000 Range alternate r alternate The extra range required to fly to an alternate airport km 1000 Loiter time t loiter The time spent in loiter, waiting for permission to land hr 1 Pax N pax The number of passengers 180 Crew N crew The number of crew as dictated by FAR 25 6 Pax/Crew mass w pax The mass of the passengers/crew & luggage kg 100 Cruise altitude h cruise The altitude flown during cruise segment m Cruise Mach number M cruise The Mach number flown during cruise 0.82 Loiter Mach number M loiter The Mach number flown during loiter 0.3 Takeoff altitude h to The design altitude of takeoff m 1000 Takeoff field length T OF L The FAR 25 takeoff length requirement including m 1500 safety features to account for engine failure situations Approach speed V approach The speed at which an aircraft approaches a runway for landing. FAR category C for airliners. ms 1 75 Minimum climb gradient that is 100 θ climbmin The minimum rate of climb for a given takeoff scenario, % 1.2 vertical Obstacle clearance limit S a horizontal Clearance requirements for airliner after takeoff. Dictated by FAR 25. m 330 Table 3.3: Mission profile variables used as input parameters to the model Technology and design parameter inputs As well as inputs for the specific mission, the model requires input parameters for technologies within the scope of the analysis. These can be segmented into engine technology inputs and airframe technology inputs, highlighted in the following sections. Engine inputs The engine technology inputs are highlighted in Table 3.4. These pertain to the efficiencies of the major turbofan components, the performance of particular components such as pressure ratios, as well as design parameters such as bypass ratio. Airframe inputs The airframe technology inputs and design parameter inputs for the model can be seen in Table 3.5. The design parameter inputs are initial assumptions usually based on legacy aircraft. 3.6 Model description Initial airframe-engine geometry & weight estimation Previously, a brief overview of the model was presented. In the following section this will be elaborated on in more detail. The methods utilized by the model and the various assumptions will be described while the specific mathematical details of each function are described in Appendix B. Before describing the model it is worth noting 59

60 Figure 3-8: Block diagram of the architecture of the airframe-engine model showing the flow of variables in between specific functions. 60

61 Input parameter Variable Description Units Typical value Compressor pressure P rc The ratio of pressure at the front fact of the compressor 30 ratio to that of the back face Fan pressure ratio P rf The ratio of the pressure at the front face of the fan 1.4 to that of the back face Turbine inlet temperature T ti The temperature of the flow at the turbine front face K 1700 Overall pressure ratio P ro The combined pressure ratio across the fan and compressor 50 Bypass ratio BP R The ratio of mass flow through the bypass region to 9 the mass flow through the core of the engine Diffuser efficiency η d Efficency of the diffuser in compressing flow 0.97 Fan efficiency η f Efficency of the fan in compressing flow 0.9 Compressor efficiency η c Efficency of the compressor in compressing flow 0.9 Fan nozzle efficiency η fn Efficency of the fan nozzle in accelerating flow 0.97 Burner efficiency η b Efficency of the combustor in converting chemical energy 0.98 of the fuel-air mix to kinetic energy of the flow Turbine efficiency η t Efficency of the turbine in extracting energy from the 0.98 flow Nozzle efficiency η n Efficency of the core nozzle in accelerating flow 0.98 Combustor pressure ratio η n Pressure ratio across the combustor 1 Table 3.4: Engine technology input variables for the engine model. Input parameter Variable Description Units Typical Lift-to-drag ratio ( ) value L The maximum ratio of the aircraft lift to the aircraft 18 D max drag Zero-lift drag coefficient C D0 The coefficient for aircraft parasitic drag which is skinfriction drag plus pressure drag Airfoil lift coefficient ( C lmax The maximum coefficient of lift of the airfoil 1.5 Thickness-to-chord t ) The ratio of the maximum thickness of the airfoil 0.14 c max i ratio to the chord length for component i, where i = {wing, horizontal tail, vertical tail} Aspect ratio A wing The ratio of the square of the wingspan divided by the 8 wing area Taper ratio λ The ratio of the chord at the tip to the chord at the 0.3 Quarter chord sweep angle Λ c 4 i root The angle of sweep at the line one quarter of the chord aft of the leading edge for component i, where i = {wing, horizontal tail, vertical tail} Wing dihedral angle δ wing The vertical angle of the wing with respect to the horizontal for passive stability Fuselage fineness ratio fr fus The ratio of the length of the fuselage to the maximum width Airplane maximum C Lmax The maximum lift coefficient for the airplane without clean lift coefficient any high-lift devices deployed Airplane maximum C LmaxHL The maximum lift coefficient for the airplane with high-lift lift coefficient high-lift devices deployed Oswald s efficiency e Factor to incorporate difference between ideal wing factor and 3D wing effects Table 3.5: Airframe technology input variables for the airframe model. that the values chosen for the assumed design parameters are all trade-offs themselves between factors such as complexity, manufacturability, operating costs and performance. Hence for the purpose of the analysis in this thesis utilizing existing 61

62 values from legacy aircraft for these is deemed appropriate, given a similar architecture and mission profile. It would be infeasible to perform trade studies on each of these parameters given the hugely interdependent nature of this engineering system. A similar method based on legacy designs is employed in many aircraft conceptual design books [98, 100, 101, 111, 60, 7]. These in turn are used as the basis for most of the model described below. Operating conditions calculation: The first step of the model is to calculate the operating conditions at each segment of the mission profile taking as inputs altitude and Mach number, and calculating air density, velocity and dynamic pressure. Initial sizing of aircraft: Secondly an initial estimate of the aircraft empty mass and the fuel mass is computed using the weight fraction method [98] based on assumptions of a weight fraction for each segment of the mission. Thrust-to-weight and wing loading calculation: Assuming a value for the Lift-to-drag ratio (( L ) D cruise) of the aircraft based on existing aircraft, the thrust to weight ratio (( T )) during cruise and takeoff are calculated. This value and the aircraft operating W conditions are used to determine the wing loading ( W ) for each of the mission segments. The lowest wing loading value of these is selected, since this gives the limiting S condition for wing area, i.e. the maximum wing area. Airframe geometry calculation: The airframe geometry can then be determined utilizing this value of wing loading, and assumptions for geometrical parameters from legacy aircraft such as aspect ratio, taper ratio, fineness ratio etc. This method which incorporates assumptions for certain parameters, shown in Appendix B, is valid within the context of the dominant architecture, and is widely employed in industry and academia for high-level aircraft design [98, 100]. Refined aircraft sizing: Given the major geometrical elements have now been defined, the mass of the aircraft can be refined based on a similar weight fraction method utilizing more details specific to this particular aircraft. Initial engine analysis: Consequently the initial engine geometry and performance can be determined based on the turbofan engine model in Figure 3-9. This turbofan engine model takes as inputs the aircraft thrust requirement as well as engine technology parameters highlighted in Section It utilizes an approach derived from first principles, based on 1D fluid flows, to determine the engine size and performance such as specific fuel consumption. This method is based on well-known analytical equations for turbofan analysis [33] and is powerful due its ability to reliably predict performance for a wide range of inputs. 62

63 Figure 3-9: A schematic diagram of the idealized turbofan engine used in this model Landing gear loads & tire computation: In parallel with this the loads on the landing gear and the tire sizes are computed, based on the worst case landing scenarios. Component weight calculation: Next the weights of all the aircraft high-level structural components such as the wing, fuselage, landing gear etc. are computed using a statistical method based on the aircraft takeoff weight [100]. Center of gravity calculation: As a result, the aircraft center of gravity can be initially estimated using assumptions for the locations of certain components. It is worth noting that in modern aircraft advanced control systems enable fuel to be moved between tanks to ensure that the center of gravity is in the desired location for aircraft stability. Additionally this analysis is based purely on the airframe structure and the engines; therefore by adding the aircraft systems the center of gravity could be moved, if desired by the manufacturer. For the purposes of this analysis the location of the center of gravity is desired in order to perform stability analysis. Aerodynamics computation: Given the airframe and engine geometry the initial aerodynamic performance of the aircraft can be computed. In the conceptual design stages aircraft manufacturers tend to have their own tools to rapidly calculate the lift and drag of the aircraft at different phases of operation. These tools are not publicly available and methods such as computational fluid dynamics are not appropriate at such an early stage where the aircraft geometry is only known at a high level and not fully defined. In order to approximate the aircraft lift the method described in Raymer is used [98]. This involves utilizing semi-empirical formula to estimate the lift curve slope, the maximum lift with and without high lift devices, the angle of attack for maximum lift, and the zero lift angle of attack. These formula utilize aircraft geometry, specifically the wing geometry as well as airfoil data, which is obtained by 63

64 selecting an airfoil from a database which meets the requirements of the aircraft mission. On the other hand, for drag approximations, the drag build-up method is used to estimate parasitic drag and the Oswald span efficiency method derived from classical wing theory is used to estimate the lift-induced drag [98]. This method requires aircraft geometry inputs including wetted area estimations in order to approximate skin friction coefficients and form factors for each of the components; these along with assumptions for interference drag are used to calculate the overall parasitic drag of the aircraft (see Appendix B for more details). Once again, given the consistent aircraft architecture these semi-empirical methods are applicable. Longitudinal static stability computation: The next step in the process is to use longitudinal static stability equations to calculate the neutral point of the aircraft. An iterative method is used to position the aircraft components such that the conditions for longitudinal static stability are met. This method requires re-calculation of the center of gravity of the aircraft for each iteration due to the relocation of components; this iteration continues until the design has converged on component locations which satisfy the criteria for longitudinal static stability. Since lateral stability requires sizing of control surfaces, it is assumed for this analysis that the combined action of aileron and spoiler control surfaces, coupled with advanced control systems, would enable lateral stability to be maintained. Similarly, it is assumed that in the sizing of the vertical tail using the volume coefficient method (see Appendix B), the yaw stability requirements have been accounted for. For the purposes of this thesis further stability analysis is not required since there is negligible effect on the aircraft geometry and our performance metric. At this point we have an estimate of the aircraft geometry, mass, center of gravity, aerodynamic performance, fuel consumption, thrust to weight ratio, lift to drag ratio, etc. at different segments of the design mission, where valid. These are all high-level estimates using mainly semi-empirical models for the given architecture and mission requirements, and a few analytical models, such as that of the engine Optimization of the aircraft design parameters In order to analyze the effects of the engine, and advances in engine technology, on the aircraft design parameters and architecture performance, it is necessary to capture the interactions between the engine and airframe. The engine location in this dominant architecture, as seen in the previous section, is under the wing. The significance of this is that there are interactions between the engine, the wing, the landing gear and the fuselage. Capturing these interactions is required in order to optimize the aircraft design for increased performance. Optimization in this context uses as an objective the maximization of the aircraft performance metric as defined in Chapter 2. This can be formulated as follows, 64

65 max x ac,x eng s.t. M ( ( ) ) T P CE, V,, P W mission profile geometric constraints architecture constraints physics constraints technology constraints, (3.6) where x ac and x eng represent the aircraft and engine design parameter vectors respectively, and M( ) is the performance metric to be maximized. As mentioned earlier, for modern passenger aircraft V and ( ) T W are essentially constant due to mission specifications being similar, and predicting price is beyond the scope of this analysis which is more focused on technology aspects. Therefore this optimization is essentially maximizing P CE = R Npax W f. For a given design mission we have a constant R and N pax, therefore the optimization translates into minimizing the mission fuel required, otherwise known as the block fuel, W f. Block fuel can be related to the airframe and engine technologies, which govern the respective efficiencies, through the Breguet range equation, R = ( ) L V D SF C ln ( Winitial W final ), (3.7) where ( ) ( ) L D represents the aerodynamic efficiency, W initial W final represents structural efficiency including block fuel, SF C represents engine efficiency, and R, V are constants for a given design mission. Since W initial = W final + W f and W final = W e + W p this equation can be rewritten as, [ ( ) ] R SF C W f = exp V ( ) 1 [W L e + W p ]. (3.8) D Based on this, the previous optimization can be reformulated with greater granularity as, (( ) ) L min W f, SF C, W e x ac,x eng D s.t. mission profile geometric constraints architecture constraints physics constraints technology constraints. (3.9) It can be seen from equation 3.8 that in order to reduce W f, one would need to either decrease W e, D, SF C or increase L, or a combination of any of these. In practicality these are often interrelated as one alters x ac or x eng ; hence, as was mentioned before, 65

66 it is of vital importance that these interactions and their effects of the performance variables are captured in the model. Given the focus on engine technology advances, an explanation of the interactions between the engine and the wing, landing gear and fuselage are described below, as well as the optimization being described in detail. Engine-fuselage interaction: In terms of the engine-fuselage interaction, this mainly takes the form of interference drag between the two [73]. The interference drag is captured in the model in the interference drag factor used in the drag build-up method when calculating the aircraft aerodynamics. Data from Kroo is used to place constraints on the engine location with respect to the fuselage in order to create negligible interference drag, and maximize performance [73]. The data used can be seen in Appendix G. Engine-wing interaction: The engine-wing interaction comprises a combination of aerodynamic and structural interactions. The engine generates a moment about the wing root joint; on the ground this moment adds to that of the structural weight of the wing and that of the fuel in the tanks and is therefore not a desirable moment generation. On the other hand, during flight, the wing experiences loads due to lift and the engine provides a counter-moment to balance the moment created by the lift force. It can be seen that there is a trade-off between placing the engine closer to the fuselage to minimize bending moments on the ground and placing the engine further outboard to offset bending moments in the air. Structural analysis can be carried out here in order to determine the maximum root bending moment on the wing. At a high level this involves idealizing the wingbox structure as a cantilever beam; however it requires being able to accurately predict structural weight from high level design parameters. Such a method may yield highly inaccurate results, but even if it does not there is no available data to verify this. Hence from inspection of existing aircraft it can be assumed that the engines are placed as inboard as possible, since it is intuitive that this would decrease wing structural weight, since the limiting structural consideration would be full tanks in ground operations. Based on this, the main contribution of this interaction to aircraft performance is through changes in the overall weight and changes in drag. It is assumed that the interference drag between the engine and the wing is captured in the drag build-up method and remains constant since the longitudinal position of the engine with respect to the wing remains constant during the optimization process. The wing weight is calculated empirically based on the maximum takeoff weight, therefore changes in the engine weight which lead to changes in the takeoff weight will cascade down to the wing weight. Engine-landing gear interaction: The engine and the main landing gear are closely coupled in this aircraft architecture, due to their proximity under the wing. The landing gear has a multitude of design constraints which in turn constrain the engine location; meanwhile the engine diameter has a major effect on the length and thus the weight of the landing gear. The main landing gear must be in a longitudinal position which enables stable aircraft tipping during takeoff and landing, without scraping the aft fuselage. Meanwhile the location of the nose gear must be such that it bears enough weight to be able to steer the aircraft, yet not too much so as to increase its weight unnecessarily. The main gear track, that is the distance between the two main gear in this tricycle architecture, should be wide enough to prevent the aircraft 66

67 from overturning yet is constrained by the engine location. Furthermore ground clearance constraints of the under-wing engine act to increase the length of the landing gear while the stowage requirement and performance maximization through weight minimization act to decrease the length of the landing gear. These constraints are mathematically formulated in Appendix B, with some further details given later in this section. It is clear that this is an optimization problem with many interactions and constraints among the variables for size and location of the wing, landing gear and engine. The objective is to maximize the performance through minimization of weight, drag and specific fuel consumption, as shown in the formulation in 3.9. The way this optimization is carried out in the model is to discretize the space under the wing into a cloud of points. Each of these points represents a potential location of the center of gravity of the engine and the landing gear. They are each subject to the aforementioned constraints dictated by the wing, engine, fuselage and landing gear interactions. A full-factorial enumeration of the locations, within the constraints of each of the components, including their individual geometries, is carried out, with each being evaluated for performance. In this optimization process the overall aircraft weight and geometry are constantly changing, which have a large change propagation within the model. Hence in each optimization loop, the aircraft takeoff weight is changing, which in turn causes changes in the component weights and geometries, the required thrust and thus the engine performance and geometry, the landing gear loads and geometries and the aircraft aerodynamic performance. These effect on the aircraft center of gravity and longitudinal stability which in turn are fed back into the beginning of the loop. In this way it can be seen that iteration is required until there is convergence of a given criterion. This model continues this iteration of trying to maximize performance until this has converged, which is measured through convergence of the maximum takeoff weight. The complex interdependencies in the high-level aircraft design is clear from this one example, however there are many other such trade studies which are beyond the scope of this analysis. 3.7 Outputs The outputs from the model include aircraft geometry and performance, as shown in Table 3.6. Additionally a visualization of the aircraft geometry is presented in order to enable the user to view the design, an example is shown in Figure Verification of the model From the original database of aircraft architectures, 25 were selected which exhibit the dominant architecture, to be used to validate the model. Additional data was required in order to use as inputs into the model according to Table 3.3, and to use to verify outputs from the model as in Table 3.6. The inputs to the model include actual data from the database, as well as assumed values based on empirical values from aircraft 67

68 Figure 3-10: A front-view diagram (left) and plan-view diagram (right) of the output aircraft geometry from the model used in this model. This shows all the major components and dimensions, including location of the center of gravity in green and neutral point in red. design methodology. These assumed parameters, x i, were taken from existing singleaisle aircraft, therefore the model was fine tuned for this type of aircraft. These input values for the verification process were assumed due to lack of available information and are shown in Table 3.7. Meanwhile the input values taken from the database of actual aircraft data comprise the remainder of the inputs from Section 3.5, which are not included in this table. The outputs of the model are compared with the actual outputs and the results of this verification analysis are described in the sections below. The analysis of the model error is used to create confidence intervals for each of the model outputs. The major assumption in this analysis is that the model error, e i = ŷ i y i follows a normal distribution with mean, µ e, and standard deviation σ e, that is e N(µ e, σ 2 e). (3.10) The method used is described further in Appendix C Engine model verification Verification of engine fan diameter The diameter of the engine is a major driver of this analysis, since it is driven by technology improvements such as the engine bypass ratio. The diameter of the engine is predominantly driven by the engine fan diameter, and is a variable in which there is available data to test. Therefore the model data was compared with actual data for 25 aircraft-engine combinations, from the original dataset for aircraft with the dominant architecture. The results, presented in Table 3.8 and Figure 3-11 show that the mean difference between the two is 11.88% with standard deviation 8.17%, which for the purposes of a high level architectural model is within the bounds of acceptability. 68

69 Output parameter Variable Description Units Maximum takeoff W to The maximum weight of the aircraft at takeoff kg weight Empty weight W e The weight of the aircraft minus the payload and fuel weight kg Fuel weight W f The weight of the aircraft mission fuel including reserves kg Component weight W c The weight of aircraft component c, where c = kg wing, fuselage, HT, VT, engine, main LG, nose LG lbm Specific fuel consumption SF C i The fuel consumption per unit thrust of the aircraft lbf hr ( ) Maximum thrust-toweight ratio mission profile T The maximum value of thrust divided by weight across the W max Breguet range r Breguet The range of the aircraft as calculated using the Breguet range equation km Lift-to-drag ratio ( L )max The maximum ratio of the aircraft lift to the aircraft drag D Wing area S The planform area of the wing m 2 Wing span b The span of the wing m Wing m.a.c. c The mean aerodynamic chord of the wing m Fuselage length l The length of the fuselage from nose to tip m Fuselage diameter d max The maximum fuselage diameter, taken as the fuselage height m Fuselage width d width The lateral width of the fuselage m HT area S HT The planform area of the horizontal tail m 2 HT span b HT The span of the horizontal tail m HT moment arm l HT The length from the aerodynamic center of the horizontal tail m to the aircraft center of gravity VT area S V T The planform area of the vertical tail m 2 VT span b V T The span of the vertical tail m VT moment arm l V T The length from the aerodynamic center of the vertical tail to m the aircraft center of gravity Engine fan diameter d fan The diameter of the fan at the turbofan engine inlet m Engine diameter d engine The diameter of the engine including the nacelle m Landing gear track l LGtrack The distance between the two main landing gear m Landing gear wheelbase l LGwb The distance between the main landing gear and the nose m landing gear Main landing gear tire d main The diameter of the main landing gear tire m diameter Main landing gear tire width d main The width of the main landing gear tire m Table 3.6: Output parameters from the airframe-engine model. From the table it can be seen that there is a positive bias, highlighting that that model slightly underestimates the fan diameter on average, making the estimates a little more conservative. Since there are only 25 sample points for architectures being verified, the expected values which rely on the law of large numbers, are not very accurate. This will be accounted for in the results by using confidence intervals in order to bound the possible values. 69

70 Input parameter Variable Description Units Typical value Range alternate r alternate The extra range required to fly to an alternate airport km 1000 Loiter time t loiter The time spent in loiter, waiting for permission to land hr 1 Pax/Crew mass w pax The mass of the passengers/crew & luggage kg 100 Loiter Mach number M loiter The Mach number flown during loiter 0.3 Takeoff altitude h to The design altitude of takeoff m 1000 Minimum climb gradient θ climbmin The minimum rate of climb for a given takeoff scenario, that is the vertical distance (in m) over the horizontal distance (in km) mkm 1 35 Obstacle clearance S a Clearance requirements for airliner after takeoff. Dictated m 330 limit by FAR 25. Compressor pressure P rc The ratio of pressure across the compressor ratio Fan pressure ratio P rf The ratio of pressure across the fan Turbine inlet temperature T ti The temperature of fluid flow at the turbine entry K 1700 point Bypass ratio BP R The ratio of mass flow through the bypass region to 9 the mass flow through the core of the engine Diffuser efficiency η d Efficency of the diffuser in compressing flow 0.97 Fan efficiency η f Efficency of the fan in compressing flow 0.90 Compressor efficiency η c Efficency of the compressor in compressing flow 0.90 Fan nozzle efficiency η fn Efficency of the fan nozzle 0.97 Burner efficiency η b Efficency of the combustor 0.98 Turbine efficiency η t Efficency of the turbine in extracting energy from the flow 0.90 Nozzle efficiency ( η n Efficency of the core nozzle in accelerating flow 0.98 Thickness to chord t ) The ratio of the maximum thickness of the airfoil to 0.12 c max ratio the chord length. This is selected for the wing and the 0.15 Oswald s efficiency factor e horizontal and vertical stabilizers. Factor to incorporate difference between ideal wing and 3D wing effects 0.8 Table 3.7: Assumed input values into the model due to a lack of available information. These were taken from aircraft design books as well as comparable aircraft. (a) Model fan diameter compared with actual(b) Plot of the frequency of error for fan diameter. fan diameter Figure 3-11: Verification plots for fan diameter for 25 different airframe-engine combinations. 70

71 Parameter ē ē [%] bias\ē ē [%] RMSE σ σ [%] Fan diameter [m] Table 3.8: Statistical comparison of fan diameter from real data and model outputs Verification of specific fuel consumption The engine specific fuel consumption (SFC) is a very important variable in determining architecture performance, since it is present in the range equation. Additionally it is important in determining the geometry of both the engine and the airframe. The results of comparing the model outputs to the 25 engines from the database are presented in Table 3.9 and Figure It can be seen that there is one anomalous value where the model has produced a difference of 33.60% from the actual value for the Interestingly, the engine for this aircraft is one of the first generation of turbofans, with a very low bypass ratio of 1.08; the engine model in this thesis is geared towards more advanced engine analysis, which accounts for this difference. Nevertheless, when this value is excluded, the engine model replicates real engine performance with a very high level of accuracy, with a difference of less than 5%. The bias can be seen to be slightly positive meaning that the model underestimates the SFC on average. The value of the bias is very small however it tends to make the engine seem more fuel efficient than it would be in reality by a very small amount. The bias is essentially negligible but is expected in such an analysis when only 25 sample points are used. Even so, this will be taken into account by using confidence intervals in the results section in order to bound the possible values. (a) Model SFC compared with actual SFC. (b) Plot of the frequency of error for SFC. Figure 3-12: Verification plots for specific fuel consumption for 25 different engines present on aircraft in the database. 71

72 Parameter [ ] ē ē [%] bias\ē ē [%] RMSE σ σ [%] SFC lbm lbf h [ ] SFC (single-aisle) lbm lbf h Table 3.9: Statistical comparison of specific fuel consumption from real data and model outputs Airframe model validation Verification of aircraft weights Two of the main outputs of the model which are used to determine aircraft performance are the maximum takeoff weight and the empty weight. The model attempts to optimize the aircraft geometry and performance subject to the mission specifications, with one of the objectives being minimization of the aircraft weight in order to maximize performance. The output of the analysis can be seen in Figure 3-13 and Figure 3-14 and the results of the verification are in Table This shows a mean difference of 24.81% and a standard deviation of 23.67% for maximum takeoff weight and a mean difference of 21.46% and standard deviation of 21.43% for empty weight. These values are higher than the previous ones for the engine outputs, however within the context of a high level aircraft design model, these are acceptable. From Figure 3-13 and Figure 3-14, it can be seen that 3 of the 25 architectures in question ( , A , ER) have large differences between the actual values and model values. These anomalous values were removed from the data sample and verification of the model was carried out again, showing a significant improvement in accuracy as seen in Table The data points removed correspond to aircraft designed for a very long range such as the ER, in which the objective for optimization would have been slightly different to the maximization of aircraft performance as defined in this model. For the purposes of this thesis the model is predominantly designed to maximize the passenger carrying efficiency for each mission, whereas in reality this is most likely not the case since there are many more considerations. It can be further seen from Figure 3-13 that there is a fairly large negative bias for maximum takeoff weight and from Figure and Figure 3-14 a much lower negative bias for empty weight. This can be accounted for by the mission profile inputs; the model assumes certain inputs for the loiter time and the cruise to alternate variables which significantly affect these weights. The actual design mission profile inputs for existing aircraft are unknown, therefore these must be assumed. The fact that maximum takeoff weight is more biased than empty weight confirms this, since it shows that the weight of mission fuel estimated accounts for the main discrepancy. These parameters could be further fine-tuned in order to decrease this error however this would be for specific aircraft causing over-fitting of the model and a loss of generalizability. Since the results we are interested in pertain to the relative differences in performance 72

73 as technology improves, the absolute values of these weights are not as important, therefore a mean error of the order of 15% is acceptable. In any case, the results section will include estimation bounds in order to capture the uncertainty in the model. Finally, based on the above insights, it is evident that the model is limited to inputs of aircraft design range below 10,000km in order to produce accurate results. The data for the aircraft which do not fall within this limit have been kept in the data set for the rest of the verification, so as to present the reader with the limitations of the model. The analysis has been done in each table for accuracy with these anomalous results included and without them included for comparison. (a) Model maximum takeoff weight compared(b) Plot of the frequency of error for maximum takeoff with actual maximum takeoff weight. weight. Figure 3-13: Verification plots for maximum takeoff weight for 25 different airframeengine combinations. Parameter ē ē [%] bias\ē ē [%] RMSE σ σ [%] MTOW [kg] 24, , ,620 27, MTOW (single-aisle) [kg] 18, , ,500 19, Empty weight [kg] 9, , ,860 12, Empty weight (single-aisle) [kg] 8, , ,792 10, Table 3.10: Statistical comparison of maximum takeoff weight and empty weight from real data and model outputs. 73

74 (a) Model empty weight compared with actual(b) Plot of the frequency of error for empty empty weight. weight. Figure 3-14: Verification plots for empty weight for 25 different airframe-engine combinations Verification of airframe geometry Verification of the airframe geometry was carried out and is summarized in Table 3.11 for the whole dataset and Table 3.12 for single-aisle aircraft only. It can be seen that on average across all the geometric parameters for the whole dataset the absolute percentage error is 16.60%, the percentage error is 10.48%, and the standard deviation is 18.90%. While for the single aisle data the mean of the absolute percentage error is 12.18%, the mean percentage error is 6.96%, and the standard deviation is 12.48%. For a high-level model these values are relatively low and therefore the model has been deemed acceptable from a verification standpoint. Additionally the robustness of the model was tested and it has been shown that below an input limit of 10,000km range the model proves to be accurate to approximately 10%. Since several of the design mission profile parameters for the actual aircraft are unknown and have been assumed in the model, it is difficult to produce a model robust enough to produce results within 5% of the real data without overfitting ocurring. An individual aircraft design could be replicated to within a very small percentage by fine tuning the assumed design parameters, however the goal of the model is to be robust enough to handle aircraft with varying mission profiles and requirements. For this purpose specific model accuracy has been traded to increase the generalization of the model, therefore generalization error is reduced however individual architecture error has slightly increased. For similar reasons to those mentioned above there is bias inherent in the model. Firstly, the mission profile inputs have been assumed, such as cruise range to alternate and loiter time, which have a large bearing on take-off weight, and thus cascade onto the aircraft design parameters such as wing area and span. This is also a result of the empirical formula being used which have been derived from examining legacy aircraft [98] and thus have a certain error associated with them. Interestingly some values of the bias, act to decrease error in the model, such as the coupling of the horizontal tail area and the horizontal tail moment arm. The former has been underestimated, whereas the latter has been overestimated. This tends to decrease the error in the 74

75 neutral point of the aircraft since the error in the moment, which is a multiplication of these two, tends to decrease. Moreover since we are interested in trends, that is the relative differences in these values, the bias in the model is eliminated on average. Parameter ē ē [%] bias\ē ē [%] RMSE σ σ [%] Wing Area, S, [m 2 ] Wing Span, b, [m] Wing m.a.c., c, [m] Fuselage length, l, [m] Fuselage width, d max, [m] HT area, S HT, [m 2 ] VT area, S V T, [m 2 ] HT moment arm, l HT, [m] VT moment arm, l V T, [m] HT span, b HT, [m] VT span, b V T, [m] LG track, l LGtrack, [m] Absolute Mean Table 3.11: Statistical comparison of aircraft geometry parameters from real data and model output data for the whole data set. Parameter ē ē [%] bias\ē ē [%] RMSE σ σ [%] Wing Area, S, [m 2 ] Wing Span, b, [m] Wing m.a.c. c, [m] Fuselage length, l, [m] Fuselage width, d max, [m] HT area, S HT, [m 2 ] VT area, S V T, [m 2 ] HT moment arm, l HT, [m] VT moment arm l V T, [m] HT span, b HT, [m] VT span, b V T, [m] LG track l LGtrack, [m] Mean Table 3.12: Statistical comparison of aircraft geometry parameters from real data and model output data for the single-aisle aircraft from the dataset. 75

76 76

77 Chapter 4 Disrupting the current dominant aircraft architecture 4.1 Introduction There have been several proposed aircraft architectures which promise increased aircraft performance, which have been previously mentioned, such as the unducted fan pusher, double bubble, blended wing body and flying wing. These architectures are corner points on the envelope of the architectural design space. If one considers the evolution of architectures as a pathway or trajectory, the historical trajectory has been mapped in Chapter 2, and is possible to conceive of many such potential future trajectories. There are many possible ways that aircraft architecture could change, when one considers the entire space of possible architectural decisions. Rather than propose new architectures and show the potential benefits, the goal of this thesis is to analyze the conditions that could trigger a break of the current dominant architecture. Thus it is to propose possible initial conditions for a new trajectory in the architectural design space, from a technological perspective, leading to architectural evolution. This background tees up the main research questions which are to be answered using the model described in Chapter 3. The recent trend in civil passenger aircraft is installation of new engines on existing airframes, present in the Airbus new engine option (neo) family and the Boeing 737 MAX aircraft family. In the context of this trend, the questions of interest can be stated in the following way: 1. Given the current geometric limitations of the airframe and advances in turbofan engine technology, what is the limit of improvement in aircraft performance? 2. Secondly, following on from this, given the current architecture, allowing for changes in airframe design parameters, and given trends in turbofan engine design and performance, what are the performance limits of the current dominant architecture? That is, when can the current dominant architecture be expected to break, given trends in engine technology, where architectural changes occur due to limits in performance of the current architecture? 3. Lastly, what other influencing factors act as a forcing function for architectural change, and may contribute to this in the future. 77

78 4.2 Performance of existing aircraft given engine technology improvements As previously mentioned, the most recent trend in civil passenger airlines is to reengine existing aircraft, with the A320neo and the 737MAX leading the way. The competitive threat of Bombardier, who was looking to enter the same market as these two aircraft provoked this trend. In response to this threat Airbus and Boeing decided to re-engine these aircraft rather than design new aircraft from scratch [45]. This reengining paradigm assumes that airframe design parameters remain constant, with the exception of perhaps reinforcing structures within the existing geometry to handle any increased loads due to the new engine. This trend has taken off (pun intended) since the development of a completely new aircraft is associated with high costs and a long lead time, therefore in an uncertain market post-2008 aircraft manufacturers opted for this less risky strategic choice. The analysis in this section involves aircraft geometry taken from existing aircraft, coupled with the engine model described in the previous chapter, to determine the limits of existing designs. The focus of the analysis will be on short-to-medium range, single-aisle aircraft which are more constrained in terms of geometry, hence the 737 and the A320 were selected. In terms of engine technology, four separate scenarios are analyzed for each of these families corresponding to the major trends highlighted in the preceding section. These scenarios are as follows: 1. Baseline scenario: Technologies related to increasing BPR are developed while component efficiencies, TIT, and OPR remain at today s level of progress 2. Component efficiency increase: BPR increases along with improvements in component efficiency, at constant TIT and OPR. 3. TIT and OPR increase: BPR increases along with improvements in TIT and OPR, at constant component efficiencies. 4. All technologies advance: All engine technologies improve. diagrams for 3 scenarios would do wonders These scenarios are used in the analysis throughout this section. The effect of improving various technologies are considered separately in order to segregate the effect of each on the overall performance. The parallel to this in the real world is the allocation of limited R&D funds by manufacturers to improve a given technology. Boeing 737 family The 737-MAX family is the latest in Boeing s short-to-medium range aircraft, and is essentially a re-engined version of the 737-Next-Generation (737NG) family, with almost exactly the same airframe geometry, as shown in Figure 4-1 [23]. Along with new engines these aircraft were also given minor changes to the wing tips for improved aerodynamic efficiency. 78

79 Figure 4-1: Front view drawings of the (a) /900 NG and (b) 737-7/8/9 MAX showing the difference in minimum ground clearance due to the new engine. Parameter /900 NG [53] 737-7/8/9 MAX [63] Difference [%] Engine CFM56-7B CFM LEAP-1B - Bypass Ratio Thrust [lbs] 24,000 23,000-28, Cruise SFC [ lbm lbf hr ] Fan diameter [m] Table 4.1: Comparison of two generations of turbofan engines for the 737. Figure 4-1 shows how the ground clearance limits have been reduced with the inception of new engine technology. A comparison of the engine design parameters and performance is shown in Table 4.1 for these two iterations of the 737. It can be seen from this table that there is a substantial increase in bypass ratio associated with technological innovations in materials and structures, enabling larger fan diameters without harming performance. As these engine technological trends continue one expects that the ground clearance limits will be reached. Using these drawings and simple trigonometry the geometric limits for ground clearance can be determined. This can then be used in conjunction with the engine model to determine at what point this particular aircraft design will no longer benefit from incremental engine improvements. As mentioned previously, there is a trend of improving SFC with increasing bypass ratio, provided that the fan pressure ratio is optimized for each bypass ratio [121]. In the engine model, which is detailed in Chapter 3, the fan and compressor pressure ratios are varied, enabling the SFC to be optimized for. This formulation assumes that problems related to fan surge and stall limits are overcome using variable fan geometry and other such technologies, which is a big assumption. The engine ground clearance constraints can be seen in Figure 4-2 with detailed geometry presented on the right of the figure. Although this varies depending on the reference cited, according to Roskam [100], the requirement for engine ground clearance is the angle subtended between the line from a 6inch buffer from the nacelle boundary to the landing gear, and the horizontal, φ 5, hence when x = 0.172m. Using simple trigonometry it can be calculated that, given z = 1.97m, the minimum 79

80 Figure 4-2: Diagram showing the ground clearance requirement for the engines in the dominant architecture. ground clearance which satisfies the above requirement is y = x = = 0.325m. Given that the current ground clearance is 0.43m, there is still room to upgrade the current engines if desired, to a maximum fan diameter of 1.87m. Since the airframe geometry in this analysis is constant, the engine model was fine-tuned for turbofan engines in the single-aisle aircraft class, using the dataset. The following analysis was run with engine design requirements for a , keeping all engine input variables constant, while varying the bypass ratio. The results from scenario 1 are shown in Figure 4-3. The engine geometrical limits for this airframe geometry has been estimated using the method described above. This estimate assumes that the top of the nacelle is aligned with the upper wing surface, which causes the bottom of the nacelle to lie approximately one fan diameter below the underside of the wing (see Appendix G). Therefore the fan diameter has been assumed to dictate the geometric constraint, along with bounds to account for any errors in the model. Note that since there were biases in the model, these have been accounted for in the results section; therefore shown on the graph is the expected estimate and the expected true value including the model bias. From Figure 4-3 it can be seen that for the given geometric constraint requirements, with all engine parameters constant apart from the bypass ratio, we can say with 95% confidence that a bypass ratio between 9 and 14 would be the maximum permissible with a true expected value of approximately 11, assuming normally distributed model error. That is the maximum possible fan diameter would be reached and the ground clearance criteria violated due to an increase in the fan diameter. In terms of performance, the model outputs results for specific fuel consumption (SFC) and engine weight, which are traded off as bypass ratio increases since SFC decreases but engine weight increases. These interactions need to be captured in order to gauge whether it is even beneficial to continue with this trend of increasing bypass ratio. As an aside, the engine weight is a tough parameter to verify since each data source accounts for weight in a different way. For example some sources include certain parts of the fuel system as part of the engine weight, some include the nacelle while others only count the dry engine weight. Here the method in Appendix B of Guha et al. is used [52]. Considering these points, the estimate for engine weight has a mean error of 9.13% and a standard deviation of 8.31%. The graphs showing the variation in SFC and engine weight as bypass ratio in- 80

81 Figure 4-3: Variation of fan diameter with bypass ratio for the Boeing 737-7/8/9 showing the 95% confidence bounds in the model estimations and the location of existing designs. The geometrical limit for this aircraft geometry is shown, which has been estimated using the method described in this thesis. creases can be seen in Figure 4-4. Clearly as bypass ratio increases there is a conflict as engine weight contributes to decreasing aircraft performance and SFC contributes to increasing aircraft performance. The architectural performance as defined in Chapter 2 consists of 4 contributing factors: passenger carrying efficiency, thrust-to-weight ratio, velocity and aircraft price. For the purposes of this analysis the latter three of these can be assumed to be constant, since thrust-to-weight ratio and velocity are design parameters and aircraft price is difficult to forecast. As mentioned in Section maximizing this performance metric corresponds to minimizing the block fuel for the mission. Based on the data from Figure 4-4, using equation 3.8, and fuel weight fractions, the weight of fuel was calculated for the design mission (taxi, takeoff, climb, cruise, descent, landing) and can be seen in Figure 4-5. Note that the segments cruise to alternate and loiter were not included since real data on these design mission variables is not available. Since these mission output values such as block fuel will only be compared with values from the same mission, this is not an issue. The range of feasible maximum bypass ratios is shown on this graph, computed from the estimated fan diameter and ground clearance requirement. The true expected value of bypass ratio of 11 results in an expected decrease in required fuel for the mission of approximately 1395 kg from the 737MAX to the 737 with a maximum possible bypass ratio turbofan and 3766kg from the 737NG. This corresponds to an expected decrease in block fuel weight of 8.8% versus the 737MAX and a decrease of 20.6% versus the 737NG, corresponding to a increase in aircraft performance of 9.7% and 25.9% respectively for the design mission. These results are highlighted in Table 4.2 and 4.3. Hence, given current airframes, there is still some room for improvement in per- 81

82 (a) Variation of specific fuel consumption with increasing bypass ratio. (b) Variation of engine weight with increasing bypass ratio. Figure 4-4: Variation of performance parameters with bypass ratio for turbofan engines powering the /900 airframe. Figure 4-5: Variation of mass of fuel required to perform the design mission with increasing bypass ratio for the /900 airframe. formance assuming that engine technologies can advance to produce bypass ratios in this range. Realistically there will be further losses due to unaccounted for 3D effects, such as weight increases due to cooling systems, which are not accounted for in the engine model; therefore 9.7% would be an upper bound on the true expected value of performance increase. Furthermore as mentioned previously, production of NO x has a minimum point as overall pressure ratio increases, and any regulations on this emission could decrease the potential gains in performance as defined in this thesis. A major assumption that is included in this analysis, enabling OPR and TIT to increase is therefore the development of low NO x combustors. In addition to increasing bypass ratios, other engine technologies could be improved, as described in the four technology scenarios. The effects of these on the engine fan diameter are shown in Figure 4-6. These changes in engine technology contribute to SFC, engine weight and as a result affect the weight of fuel required for 82

83 the design mission as shown in Figure 4-7. (a) Variation of fan diameter with bypass ratio for scenario 2. (b) Variation of fan diameter with bypass ratio for scenario 3. (c) Variation of fan diameter with bypass ratio for scenario 4. Figure 4-6: Graphs showing the variation in fan diameter with increasing bypass ratio, showing the effects of changing engine technology and the geometric limit of the /900 airframe. Shown on the graph as a blue square is the NG and as the red circle is the MAX. The analysis is carried out for four technology scenarios, with results shown in Table 4.2 and Table 4.3. The baseline technology scenario assumes a change in bypass ratio, without any changes in component efficiencies or turbine inlet temperature (T ti ); that is, using technology for the core that is similar to today s technology, using component efficiencies in 3.1 and T IT = 1700K. It is worth noting that increases in bypass ratio are associated with improvements in structures and materials, and any deterioration in performance due to these advances not being realized are not considered in the model. The second analysis involves the case of increasing bypass ratio with an increase in component efficiency, according to the levels forecast in Table 3.1, while keeping T ti and OPR constant at the baseline level. The next analysis involves increasing bypass ratio and an increase in T ti from 1700K to 1800K, allowing the OPR to vary by increasing the component pressure ratio limits, while keeping component efficiencies constant at the baseline levels. Finally, the last analysis involves increasing all of the above technologies simultaneously to the same projected technology level. From the above analyses it is clear that within the limits of the current

84 (a) Variation of weight of fuel required for the design mission against bypass ratio, for scenario 2. (b) Variation of weight of fuel required for the design mission against bypass ratio, for scenario 3. (c) Variation of weight of fuel required for the design mission against bypass ratio, for scenario 4. Figure 4-7: Graphs showing the variation in required fuel weight with increasing bypass ratio for the /900 airframe under 3 technology scenarios. The range of possible bypass ratios corresponding to the geometric limitations imposed by ground clearance constraints are shown. airframe there is capacity for further improvement in terms of substituting turbofan engines with incremental increases in performance. The expected values of increase in performance for scenarios 1-4 are 9.7%, 30.5%, 20.3%, and 38.1% respectively with the upper and lower bounds of the 95% confidence interval shown in the table. These are enabled by accommodating, within the geometrical constraints, turbofan engines with expected bypass ratios of 11, 15, 14, and 17 respectively for each of the four technology scenarios. Although performance improvements are probable, it can be observed that the lower bound of the maximum feasible bypass ratio for this airframe is 9, which happens to correspond to the CFM LEAP 1B engine of the 737-7/8/9 MAX, as seen in Table 4.1. That is to say there is a small possibility that the fan diameter and hence the bypass ratio limits have already been reached for current engine technology. Taking the baseline technology level for component efficiency and turbine inlet temperature, if fan blade materials and structures were advanced enough we would expect a bypass ratio engine of approximately 11 to be the maximum possible for the current airframe, 84

85 Technology scenario Bypass ratio [ ] SFC lbm Engine Weight Mission Fuel Fuel Weight lbf hr [kg] Weight [kg] Decrease vs. 737MAX [%] Lower Upper EV Lower Upper EV Lower Upper EV Lower Upper EV Lower Upper EV 1. Baseline technology ,157 3,808 3,368 15,012 13,491 14, Component efficiency increase 3. Turbine inlet temperature increase 4. All technologies improve ,199 3,716 3,520 12,934 11,810 12, ,424 4,064 3,787 14,154 12,748 13, ,368 3,834 3,657 12,135 11,200 11, Table 4.2: The upper and lower 95% confidence bounds as well as the expected values (EV) for bypass ratio, SFC, engine weight and weight of fuel, which are attainable given the ground clearance requirements of the current 737 airframe. These values are calculated for a fan diameter constraint of 1.87m. Technology scenario Performance increase vs. 737MAX Lower Upper Expected Value 1. Baseline technology Component efficiency increase Turbine inlet temperature increase All technologies improve Table 4.3: The upper and lower 95% confidence bounds as well as the expected values for aircraft performance for the 737. which corresponds to an increase of 9.7% from the performance of the 737-MAX. The results that follow the baseline technology analysis, namely increases in component efficiency and TIT & OPR, serve to show the impact of other possible improvements in engine technology. As mentioned in Section 3.2, these technology trends are considered to be the most likely to be seen in turbofan engines in the near future. The results of improvements along these two dimensions are done independently based on the forecasted values. It can be seen that improvements in both of these independently tend to decrease the fan diameter, therefore enabling possible increases in performance for the given geometric constraints. Note that another technological innovation which has not been included explicitly in this analysis is the inception of a gear to power the fan. The use of geared turbofan has been allowed in the model by through setting the range of possible fan pressure ratios and optimizing despite any mismatch in turbine and fan speeds. In the weighting module this has not been considered, however the addition of a gear causes the low pressure turbine to decrease in size which therefore tends to decrease any error in weight estimation. Considering the time dimension of these technological advances, based on the forecasts in Section 3.2, it is possible to predict an expected value for when the current 737 airframe will no longer be able to increase in performance due to the geometric constraints associated with the under-wing engine architecture. The three technology scenarios are shown on Figure 4-8 showing the expected value of BPR over time including 85% confidence bounds. The BPR forecast is based on fitting 85

86 Figure 4-8: Forecast of expected value of BPR using the whole dataset, overlain with geometry limits dictated by BPR for the 737 for the four different scenarios. The 85% confidence intervals both in time and BPR are shown. Figure 4-9: Forecast of expected value of BPR using only single-aisle aircraft, overlain with geometry limits dictated by BPR for the 737 for the four different scenarios. The 85% confidence intervals both in time and BPR are shown. 86

87 a trend to historical data and extrapolating; therefore it assumes that technological advances enable such a bypass ratio, that is, the advances in component efficiency and turbine inlet temperature are assumed in this trend. One could possibly develop different bypass ratio trends based on different technology scenarios, however there is too much uncertainty and a lack of fidelity at the component level in the available historical data to do so. The results of this future forecast are shown in Table 4.4. The expected values from both models are included in the table, model 1 being a forecast based on all engine data, and model 2 based on single-aisle engine data only. The expected value of the predicted year for the geometric constraint to be hit are highlighted and they differ between the two models. Each of these values has an upper bound highlighted on the graphs, and a lower bound, which statistically makes sense however not necessarily in practice (e.g. a lower bound of a date prior to today). Based on this data it can be seen that depending on the technology scenario the geometric constraint for the 737 could be hit as early as this year, 2016, all the way up to An average of the two models was taken reducing this range to These predictions enable the reader to see the range of possible years that an airframe such as the 737 can continue increasing in performance simply by changing engine. The expected value of BPR which will infringe on ground clearance requirements varies from depending on the technology scenario, it is clear that there is uncertainty in this model. That being said it is highly likely that in the next 12 years this event will occur. Scenario Predicted year EV model 1 EV model 2 Mean EV 1: Baseline : Component efficiency : TIT & OPR : All technologies Table 4.4: Summary of the expected values for the year of architecture break, for both BPR forecast models. Airbus A320 family The A320 new engine option (neo), similar to the 737MAX is Airbus s latest iteration of the single aisle aircraft, consisting of the essentially same airframe geometry as the original A320, with new engines installed. The ground clearance constraints for two variants of this aircraft are shown in Figure A comparison of the two engines can be seen in Table 4.5. This table and the dimensions in the figure can be compared with the the details of the 737 in Table 4.1 and Figure 4-1. It can be seen that the larger ground clearance of the A320 enables engines with larger fan diameters and higher bypass ratios. Using the same geometry for engine ground clearance as shown in Figure 4-2, the same analysis is carried out for the A320 airframe. The model inputs were tweaked 87

88 Figure 4-10: Front view drawings of the (a) A and (b) A320 neo showing the difference in minimum ground clearance due to the new engine. Parameter A [53] A320neo [4] Difference [%] Engine CFM56-5B6 CFM LEAP-1A - Bypass Ratio Thrust [lbs] 23,500 24,500-32, Cruise SFC [ lbm lbf hr ] Fan diameter [m] Table 4.5: Comparison of two generations of turbofan engines for the A320. and fine-tuned to minimize the estimation error, in the same way as with the 737. Using the same method as before, with y = 0.46 and z = 1.955, the geometric limit of the engine, subject to the ground clearance requirements is calculated, resulting in x = 0.171m and a maximum fan diameter of d fan = ( ) = 2.12m. This constraint has been overlain on the graph of fan diameter against bypass ratio in Figure 4-11 in order to obtain bounds on the potential bypass ratio given ground clearance requirements. It can be seen that, varying technology related to increasing bypass ratio, while keeping other technology at its baseline levels, the range of possible maximum bypass ratios, which are on the geometric limit of the ground clearance requirement for the A320 airframe are between 10 and 16, with an expected value of just under 13. The corresponding specific fuel consumption and engine weight variation with increasing bypass ratio can be seen in Figure 4-12 for the A320 airframe. These result in a mission fuel weight variation with bypass ratio shown in Figure 4-13, using the same method as with the 737. The expected value for bypass ratio results in an increase in aircraft performance of 5.7% versus the A320neo and 20.4% versus the A This shows that, similarly to the 737, the A320 still has a margin for improvement by installing improved engines given constant airframe design parameters. This increase in performance is associated with a 95% confidence bound, due to the inaccuracies of the model. Using such a high fidelity model to make absolute statements about definitive numbers would not make sense in this context therefore, an upper and lower bound have been included based on the assumption of a normal distribution of error. These results are shown in Table?? and Table 4.7, 88

89 Figure 4-11: Variation of fan diameter with bypass ratio for the A320 showing the 95% confidence bounds in the model estimations and the location of existing designs. The geometrical limit for this aircraft geometry is shown, which has been estimated using the method described in this thesis. for the baseline technology. If other engine technologies were improved simultaneously, one could expect further improvements in performance using the current airframe. As before, improvements in component efficiency and TIT & OPR are examples of technologies in which substantial research has been devoted in terms of improvements in structural design and creep-resistant materials [20]. Improvements in these lead to a smaller ad lighter engine which enables further aircraft performance increase within the given geometric boundaries. The effects of increasing these are shown in Figure 4-14, on a graph of fan diameter against bypass ratio. For scenario 2, the range of permissible maximum bypass ratios increases to between 12 and 20, with an expected value of approximately 16 and for scenario 3 the range is with an expected value of 15. For scenario 4 the range of BPR is with a true expected value of 18. The resulting performance in terms of fuel weight reduction with respect to independent improvements in each of these technologies can be seen in Figure The decrease in fuel weight is a result of the decrease in specific fuel consumption enabled by the improved technology, despite the increase in engine weight. It is worth noting that in this analysis the increase in engine weight due to new technologies has not been explicitly modeled, since the engine weight model depends on the fan diameter and associated increase in nacelle weight, as in [52]. With this in mind it is possible that the estimations for performance increase may be slightly over-estimated. A compilation of these analyses can be seen in Table?? and Table 4.7. An additional point to note is that the engine analysis could be carried out for the engine at different phases of operation, such as takeoff, climb, cruise etc. For the purposes of this analysis it has been carried out at static sea level conditions since this tends to 89

90 (a) Variation of specific fuel consumption with increasing bypass ratio. (b) Variation of engine weight with increasing bypass ratio. Figure 4-12: Variation of performance parameters with bypass ratio for turbofan engines powering the A320 airframe. determine the maximum diameter of the engine, and at cruise conditions since this has the largest effect on performance. Slightly different results would be expected if this analysis were to be carried out at other phases of operation. It is clear from the data that, based on the level of technological advances, the four independent scenarios estimate an expected value for decrease in block fuel weight of 5.5%, 19.1%, 12.6%, and 23.3% respectively. These correspond to performance increases of 5.8%, 23.6%, 14.4%, and 30.4%, since performance is a function of the inverse of fuel weight, as shown in Table 4.7. These are associated with increasing th BPR from its current value of 11 to 12, 16, 15, and 18 for each of the four scenarios respectively, in order to fully utilize the available aircraft geometry within the constraints of the system. [ ] Technology scenario Bypass ratio SFC lbm Engine Weight Mission Fuel Performance vs. lbf hr [kg] Weight [kg] A320neo [%] Estimate Lower Upper EV Lower Upper EV Lower Upper EV Lower Upper EV Lower Upper EV 1. Baseline technology ,270 4,178 3,691 13,837 12,057 12, Component efficiency increase 3. Turbine inlet temperature increase 4. All technologies improve ,313 4,132 3,750 11,831 10,513 11, ,546 4,357 4,033 12,916 11,444 11, ,488 4,146 3,881 11,124 10,128 10, Table 4.6: The upper and lower 95% confidence bounds as well as the expected values (EV) for bypass ratio, SFC, engine weight and weight of fuel, which are attainable given the ground clearance requirements of the current A320 airframe. These values are calculated for a fan diameter constraint of 2.12m. As with the 737, the geometric constraints of this airframe considering engine technology advances are projected using the expected values of BPR over time. This is done using the whole dataset in Figure 4-16 and just single-aisle data in Figure These graphs of BPR over time assume that advances in technology occur; however does not segregate different technology scenarios due to a lack of available 90

91 Figure 4-13: Variation of mass of fuel required to perform the design mission with increasing bypass ratio for the A320 airframe, showing the range of possible bypass ratios falling within the 95% confidence interval for ground clearance requirements. Technology scenario Performance increase vs. A320NEO Lower Upper Expected Value 1. Baseline technology Component efficiency increase Turbine inlet temperature increase All technologies improve Table 4.7: The upper and lower 95% confidence bounds as well as the expected values of aircraft performance for the A320. data to make such a prediction. The four scenarios are shown on the graphs along with 85% confidence bounds, with the results of expected value highlighted in Table 4.8. These results show that in the baseline scenario, the geometry constraints of the dominant architecture will be reached by The expected values for scenarios 2 and 3 are 2027 and 2029 respectively. Hence given these, it is clear that in the next years further performance advances of the A320 will not be possible due to the geometric limitations of having under-wing engines. These results show that, depending on the technology scenario, the geometric constraints of the A320 airframe are expected to be hit by , when the results from the two models are averaged. Given the time-frames required to improve upon current turbofan technologies, it is expected that by 2030 the A320 airframe will no longer be able to improve upon its performance. The assumptions inherent in this forecast, are that the technology trends in Section 3.2 hold. Note that the results for scenarios 1, 2, and 3 are based on a BPR trend that is purely based on empirical data. Thus this trend implicitly assumes that the aforementioned technology trends 91

92 (a) Variation of fan diameter with bypass ratio for scenario 2. (b) Variation of fan diameter with bypass ratio for scenario 3. (c) Variation of fan diameter with bypass ratio for scenario 4. Figure 4-14: Graphs showing the variation in fan diameter with increasing bypass ratio, showing the effects of changing engine technology and the geometric limit of the A320 airframe. Shown on the graph as a blue square is the A NG and as the red circle is the A320 neo. are occurring simultaneously. Given this, it is likely that the expected forecasts are on the conservative side, meaning these effects could take place sooner. These predictions depend on many different factors, which explain the large confidence bounds on the expected values. Conclusion To summarize the analysis carried out in the previous two sections, the Boeing 737 and Airbus A320 airframes were simultaneously analyzed with the current trend of upgrading aircraft by installing improved engine technology. The geometric constraints pertaining to ground clearance requirements of the dominant architecture were computed, in order to bound the permissible engine technology. Meanwhile the engine model was used to analyze the performance of turbofan engines which fit within the bounds of these geometrical constraints. This analysis was carried out while extrapolating the main engine technology trend of increasing bypass ratio, as well as other technology trends such as increasing component efficiency and increasing turbine inlet temperature. The forecast of these engine technologies has been presented, using 92

93 (a) Variation of mission block fuel against bypass ratio for scenario 2. (b) Variation of mission block fuel against bypass ratio for scenario 3. (c) Variation of mission block fuel against bypass ratio for scenario 4. Figure 4-15: Graphs showing the variation in required fuel weight with increasing bypass ratio for the A320 airframe for 3 technology scenarios. The range of possible bypass ratios corresponding to the geometric limitations imposed by ground clearance constraints are shown. predictions from literature and from our own analysis. The results include expected values as well as upper and lower bounds for design variables and engine performance variables, to account for the inherent error in such a high level model. It was shown that given the current airframe geometric constraints, there is still potential to increase the performance of the latest iteration of these two classic aircraft, solely by installing new engines. The 737MAX represents a 14.9% increase in performance over the 737NG, as measured by the increase in PCE for the design mission, solely by following this above trend of re-engining. Four technology scenarios for engine technology have been extrapolated and analyzed within the context of a 737 airframe. The purpose of using separate technology scenarios is to highlight that different advances in technology lead to different outcomes; although the costs of these advances have not been considered, manufacturers may use a similar analysis to try to prioritize partnerships and R&D spending. The first scenario involves keeping engine technology at baseline levels except for increasing engine bypass ratio, which results in an approximate expected value of 9.7% increase in performance versus the 737MAX. The second scenario involves the 93

94 Figure 4-16: Forecast of expected value of BPR using the whole dataset, overlain with geometry limits dictated by BPR for the A320 for the four different scenarios. The 85% confidence intervals both in time and BPR are shown. Figure 4-17: Forecast of expected value of BPR using only single-aisle aircraft, overlain with geometry limits dictated by BPR for the A320 for the four different scenarios. The 85% confidence intervals both in time and BPR are shown. 94

95 Scenario Predicted year EV model 1 EV model 2 Mean EV 1: Baseline : Component efficiency : TIT & OPR : All technologies Table 4.8: Summary of the expected values for the year of architecture break, for both BPR forecast models. same analysis as the first but allowing component efficiencies to increase, as described in Section 3.2, resulting in an approximate expected value of performance increase of 30.5%. Scenario 3 allows the turbine inlet temperature to increase at constant component efficiency giving an expected value of performance increase versus the 737MAX of 20.3%. Finally, allowing all the engine technologies to increase, the associated performance increase is expected to be 38.1%. Likewise the A320neo represents a 13.6% increase in performance over the A , due to replacing the engines. The same four scenarios have been computed for the A320 airframe as the 737. The first baseline technology scenario has an expected performance increase of approximately 5.8% over the A320neo, the second scenario is associated with a 23.6% improvement in performance, the third scenario has a 14.4% expected performance improvement, and the forth scenario improves on the A320neo by an expected value of 30.4%. Comparing these values with those of the 737 is not a completely fair comparison. They start from a different baseline and their design missions are not exactly the same, therefore the magnitude of potential improvement in performance will likely be different. The above results do not highlight a superior performance of one over the other since the input values into the model are not the same. As was shown, if the forecasts for technological advances in engine technology are accurate, we can expect that within the next 12 years the 737 will reach its absolute performance limit, and the A320 within the next 14 years. Further improvements in aerodynamics, structures and materials for the airframe are possible and may enable improvements beyond those predicted in this section. Nevertheless, since SFC has the largest potential for performance improvements, as described in Section and since the engine is the main driver of aircraft architecture, as presented in Chapter 2, the performance improvements due to the engine are considered to be the most significant. This analysis serves to highlight that in reality, this would be the time frame in which we would expect the architecture to break. Since this dominant architecture has persisted for many years using essentially the same airframes for these two families of aircraft, this breaking point for these particular airframes would offer a great opportunity for manufacturers to reopen the architectural design space and explore superior designs in terms of aircraft performance. Returning to the analysis of the above section, the aforementioned improvements 95

96 in engine technology can not be taken for granted. They require substantial amounts of time and money to be spent on research and development, which may not be in the economic interest of the manufacturer. Furthermore more components increase the complexity which could have an effect on engine reliability and direct operating costs. Considering time, cost and complexity, further advances in these technologies yield diminishing returns in terms of performance. That is, at some point the economics of a marginal improvement in bypass ratio, component efficiency, turbine inlet temperature, or a geared turbofan may be substantially less than their associated marginal cost in development and operations. The analysis in the next section will enable the design parameters of the dominant architecture to vary. Given a similar design mission profile to the 737 and A320 the projected limits of the dominant architecture will be forecasted rather than a particular airframe as in this section. 4.3 Limits of dominant architecture given engine technology improvements The analysis in the previous section took the airframe geometry as a constraint, to quantify the possible performance improvements of current single-aisle aircraft, namely the Boeing /900 and Airbus A320. The purpose was to use real world constraints to predict when the current trend of re-engining an existing airframe will no longer be possible for performance increases due to geometric constraints of the dominant architecture within the context of these airframes. This next section will take the analysis one step further, removing the constraint of a constant airframe. This will thus relax the design variables for the airframe, remaining within the bounds of the dominant architecture, but unconstrained by the previous geometrical limits. The purpose of this is to analyze, when the dominant architecture could potentially break. That is, examining trends in turbofan engine technologies, at what point will aircraft performance no longer increase, therefore potentially requiring architectural exploration. As has been mentioned previously, the major trends in turbofan technology are increasing bypass ratio, increasing component efficiency, and increasing turbine inlet temperature along with overall pressure ratio. These tend to result in the fan diameter of the turbofan increasing, and therefore an engine which requires more space underneath the wing for installation. As the ground clearance requirements of the current airframes are approaching their limits, there are several design parameters which can be modified in order to accommodate larger engine diameters. Within the confines of the dominant architecture, these include: 1. Increase the landing gear length 2. Increase the distance of the engine from the centerline 3. Increase the landing gear track 96

97 4. Increase the wing dihedral 5. Reducing the engine diameter through engine core technology advances 6. Reducing the nacelle thickness. Alternative options such as changing the engine architecture, relocating the engine, or relocating the wing for example to a high-wing configuration, would constitute a change in architecture and are therefore not included as options in this analysis. Such changes would be considered breaking the architecture, whereas this analysis is considering the options before this point. Of the design parameter changes above, all are considered in the model, apart from numbers 4 and 6. The reason that it may not be desirable to increase the wing dihedral is because this would create an aircraft which would be too laterally stable and thus unable to maneuver. Additionally, lateral stability has not been explicitly included in the model since the effects on aircraft geometry and architectural performance are considered to be negligible. Reducing the nacelle thickness has not been explicitly included as part of the analysis, since it is already assumed to be at the lower bound of thickness in the model, therefore any further change would be negligible. The other four possible design variable changes, are highly connected to each other, and to other design variables, therefore changing each one has a large change propagation within the design of the system [38]. These highly non-linear relationships are difficult to predict and therefore are included in a brute force discrete optimization of the airframe-engine interaction, as described in Section Note that airframe trends such as increasing specific strength of materials for the landing gear for example are not explicitly accounted for in this analysis in order to bound the scope on engine technologies as the driver for performance improvements. Nevertheless there are potential airframe technology improvements, such as those mentioned in Section 3.2, which could potential provide incremental improvement in aircraft performance. The same technology scenarios to those presented in the previous section will be used in this one. As a recap these are: 1. Baseline scenario: BPR increases, with constant component efficiencies, TIT and OPR 2. Component Efficiency increase: BPR and component efficiency increase at constant TIT and OPR 3. TIT & OPR increase: BPR and TIT & OPR increase, at constant component efficiency 4. All increase: BPR, component efficiency and TIT & OPR increase. The level of these technology improvements are associated with the time frames presented in Section 3.2. These scenarios will be analyzed in the sections below, allowing the aforementioned hypotheses to be tested. 97

98 4.3.1 Model inputs The aircraft analyzed in this section has a similar mission profile to a typical singleaisle aircraft such as the 737 or A320. The mission profile inputs into the model are shown in Figure 4.9. The other inputs are the same as those detailed in Section 3.5 for airframe technology and Section 3.2 for engine technology. Input parameter Value Unit Range r cruise 5000 km Range alternate r alternate 1000 km Loiter time t loiter 1 hr Pax N pax 180 Crew N crew 6 Pax/Crew mass w pax 100 kg Cruise altitude h cruise m Cruise Mach number M cruise 0.82 Loiter Mach number M loiter 0.3 Takeoff altitude h to 0 m Takeoff field length T OF L 1500 m Approach speed V approach 75 ms 1 Minimum climb gradient θ climbmin 1.2 % Obstacle clearance limit S a 330 m Table 4.9: Mission profile variables used as input parameters for the analysis. Each of the scenarios is compared to the state of technology today in order to measure the potential improvements. The current state of technology assumes the same airframe and mission inputs, and baseline engine technology inputs. The BPR of current turbofan technology is assumed to be 10 for the purposes of this analysis since this is similar to the most advanced 737 and A320 aircraft Scenario 1: Bypass ratio increase with baseline technology This scenario assumes that only the technologies related to increasing the bypass ratio of the turbofan engine will improve, with all other technologies remaining at the baseline level. As in the previous section, aircraft performance is measured using the block fuel required to complete the mission, given using equation 3.8, which is composed of the SFC, lift-to-drag ratio and the empty weight, for a constant payload weight, aircraft range and velocity. The changes in design parameters to accommodate this scenario of engine technology improvement can be seen in Figure It can be seen that three of the design parameter modifications mentioned in the previous section are occurring, that is increase in landing gear length, landing gear track and distance of the engine from the 98

99 centerline. The engine diameter can be seen to be increasing since BPR is increasing without any core technologies being improved. Figure 4-18: The evolution of aircraft design parameters as BPR increases in scenario 1. On the diagram the trend is clearly highlighted with the different colors. The aforementioned constituents of aircraft performance including SFC, lift-todrag ratio, and aircraft empty weight (or final mission aircraft weight) change with increasing bypass ratio. The SFC is directly affected by this increase in bypass ratio, however the high-level trends in lift-to-drag ratio and aircraft weight are driven by aircraft design parameters, namely the engine diameter and weight, and the landing gear length and weight. That is, as BPR increases engine diameter increases, therefore drag and empty aircraft weight tend to increase. As landing gear length increases, the landing gear weight tends to increase, leading to an increase in empty aircraft weight. The trends in aircraft design parameters and aircraft performance constituents can be seen in Figure 4-19 and Figure On these graphs the estimate expected value is adjusted for potential model bias, producing the true expected value along with 95% confidence bounds. Note that the graphs for landing gear length and weight do not have a true expected value since there was a lack of data to verify these parameters. In order to calculate the 95% confidence interval for these values the average verification values for the entire model were used. The trends of increasing size and weight of the engine and landing gear components with increasing BPR are clearly shown. It can be seen for this scenario that, as BPR increases, the specific fuel consumption decreases at a decreasing rate. This is the expected result if one examines the relationship between turbofan efficiency (or SFC) and BPR presented in Appendix B. In addition to this it can be seen from Figure 4-20b that ( L D) remains essentially constant as BPR increases. This means that the increase in drag associated with larger diameter turbofan engines is negligible when observed relative to the overall aircraft drag. To quantify this, the decrease in ( L D) from a BPR of 1 to 24 is approximately 0.25%. The majority of the contribution to decreasing aircraft performance is the increase in the final mission aircraft weight or aircraft empty weight (W final = W e +W p ), composed of increases in engine, landing gear and wing weights. The overall effect of these effects on the aircraft performance is shown in Figure It can be seen that in this baseline scenario, the weight of fuel required to complete the given mission decreases as bypass ratio increases until a BPR of approximately 99

100 (a) Engine fan diameter against BPR showing expected values and 95% confidence bounds. (b) Engine weight against BPR showing expected values and 95% confidence bounds. (c) Main landing gear length against BPR showing expected values and 95% confidence bounds. (d) Total landing gear weight against BPR showing expected values and 95% confidence bounds. Figure 4-19: Variation of aircraft design parameters as BPR increases for scenario It is noticeable that the curve of aircraft performance follows a similar trend to that of the SFC, which is expected given that the engine has been shown in previous sections to be the most important driver of aircraft architecture and performance. Beyond a BPR of 22 it can be seen that the SFC and hence the block fuel weight begins to increase. At this point the power requirement of the increased diameter fan is too large to be driven by the turbine within the given technology scenario. That is the turbine is not able to extract enough power from the core flow to drive the fan. Therefore there OPR drops suddenly as BPR increases beyond 22, leading to suboptimal engine performance and eventually a breakdown of the model as flow in the engine begins to reverse. The ultimate performance limit within the constraints of this technology scenario is given by the block fuel weight at this minimum point. As a general comment is clear that the gains produced by increasing BPR are marginal, since we are currently located at the kink in the curve and further BPR improvements in the future will yield diminishing returns in performance. This is particularly pertinent when one considers that the cost for each additional unit of BPR (or each additional unit of performance increase) is likely to be greater than the last, as is common with complex systems such as turbofan engines. A comparison of the current state of technology to the state of technology in scenario 1 that gives the maximum possible aircraft performance is presented in Table 100

101 (a) SFC against BPR showing expected values and 95% confidence bounds. (b) Lift-to-drag ratio against BPR showing expected values and 95% confidence bounds. (c) Aircraft mission final weight against BPR showing expected values and 95% confidence bounds. Figure 4-20: Variation of aircraft performance variables with bypass ratio for scenario The confidence interval for the performance increase is calculated using the method in Appendix C for the quotient of two normal random variables. A single standard deviation is used to calculate the confidence bound for the performance increase. This is used because a 95% bound on these values is extremely large given the uncertainty in the model. In order to produce meaningful results a single standard deviation was used, which gives an approximate 68% confidence interval rather than a 95% confidence interval. Note that for the values where a standard deviation is not known the model average of 12.48% is used. Note that the BPR values are taken from current state of technology and the one that gives maximum performance, therefore there are no associated confidence bounds with these. Additionally the bounds on certain values do not give much value, such as the lift-to-drag ratio, which utilizes the average standard deviation of the model. It can be seen that the expected value is projected to remain essentially constant as BPR increases, while the 95% confidence bounds are fairly large. While this makes sense statistically, physically the confidence bounds do not have much value but are presented for the sake of completeness. The data shows that if technology trends enabled engine BPR to increase, with all other technology levels held constant, then we can expect a performance increase 101

102 Figure 4-21: Aircraft performance against bypass ratio, quantified as the block fuel weight, for scenario 1. Shown on the graph are 95% confidence bounds the expected value, as well as current maximum BPR range and the maximum possible performance for this scenario. of 17.1%. This is associated with the geometrical trends presented in Figure Previously it was shown that the 737 and A320 would experience approximately 10% increase in performance in the baseline scenario due to the geometric constraints. Comparing these previous values with the performance increase in this case is not an accurate comparison. The percentage improvement calculated in this instance assumes a initial value which is calculated using the model, whereas in the previous case the initial value was taken as the from existing aircraft. The computed initial values differ in that they are optimized for current technology based on a specified design mission. The existing aircraft were designed over 30 years ago, including some flexibility for potential upgrades, with a largely unknown design mission, therefore the comparison is not valid. Based on the values calculated for this scenario it is possible to use the BPR trends presented in Figure 3-4 and Figure 3-5 to estimate when this maximum performance will occur. It is assumed that at this point in time no further aircraft performance improvements will be possible based on the engine technology assumptions of this scenario, therefore requiring a break in the dominant architecture. The results of this projection are presented in Figure 4-22, including 85% confidence bounds. In such a model there is a large amount of uncertainty, as can be seen by the large bounds in this graph. Typically a single standard deviation is used to bound the values in such problems to give more meaningful results. These results are compiled in Table This shows that the expected value in technology scenario 1 for the year in which performance will stop increasing is 2040 using model 1 and 2030 using model 2 with respective single standard deviation confidence intervals of [???] and [???]. Taking the average of both models this is expected to be in Given that it is expected that this architecture will no longer be able to produce performance increases, this is 102

103 Parameter Current technology Best case baseline scenario Difference [%] Lower Upper EV Lower Upper EV Lower Upper EV BPR Fan diameter [m] Engine weight [kg] 3,382 4,362 3,872 5,241 6,221 5, Landing gear length [m] Landing gear ] weight [kg] 3,010 4,859 3,934 3,342 5,397 4, SFC [ lbf lbm hr Lift-to-drag ratio Final aircraft weight W final 47,118 78,478 62,798 51,815 83,175 67, Block fuel weight [kg] 13,379 21,608 17,493 11,458 18,505 14, Performance Table 4.10: Comparison of geometry and performance variables for scenario 1, for today s performance versus the maximum performance. assumed to be when the current dominant architecture will break, given the assumptions of this scenario. Figure 4-22: Graph of BPR against year showing both projection models based on different data sets for scenario Scenario 2: Bypass ratio increase with increase in component efficiency This scenario assumes that the technologies related to increasing the bypass ratio of the turbofan engine will improve as well as the component efficiencies, as defined in Section 3.2. The evolution of aircraft geometry associated with these changes can be seen in Figure As in scenario 1, the landing gear length, landing gear track and distance of the engine from the centerline are all increasing with increasing BPR. Similar trends to those in scenario 1 are seen, but to a greater extent, since component efficiency has increased as well as the BPR. The results presented in the 103

104 85% confidence 68% confidence Lower Upper EV Lower Upper EV Model Model Mean Table 4.11: Expected values for architecture break year including confidence bounds for two BPR forecast models for scenario 1. Figure 4-23: The evolution of aircraft design parameters as BPR increases in scenario 2. On the diagram the trend is clearly highlighted with the different colors. graphs in this section assume the maximum component efficiency for every value of BPR, without consideration of time frame. Therefore it is necessary to compare the values from these graphs with the baseline scenario at a BPR of 11, which is representative of the current state-of-the-art technology. This is done in for the results in Table The component geometry and weight trends with increasing BPR can be seen in Figure The estimate expected value, the true expected value corrected for bias, and the 95% confidence bounds are plotted for scenario 2 and the baseline scenario. This technology scenario can be compared with the baseline technology scenario. Firstly it can be seen that by improving component efficiency a larger bypass ratio is enabled than the baseline scenario. This is due to the fact that there are fewer energy losses due to the more efficient components; hence the power output of the turbine is larger because there is more energy in the core flow which enables a larger fan to be driven. This larger fan therefore increases the bypass ratio for the same mass flow through the core. This scenario assumes that problems such as mismatch of rotational speeds of the fan and turbine, the structural integrity of the fan blades, and aerodynamics issues of separation at the fan tip are solved. Solving these issues is not trivial. Not only does it technically challenging in terms of development of technology, but it is challenging in terms of creating a more complex system which is required to be as reliable and economically viable as its less complex predecessor. For the purposes of this analysis it is assumed that these issues are implicitly solved in the forecasted trends from Section 3.2. That is, in order for these trends to be valid it has 104

105 (a) Engine fan diameter against BPR showing expected values and 95% confidence bounds. (b) Engine weight against BPR showing expected values and 95% confidence bounds. (c) Main landing gear length against BPR showing expected values and 95% confidence bounds. (d) Total landing gear weight against BPR showing expected values and 95% confidence bounds. Figure 4-24: Variation of aircraft design parameters as BPR increases for scenario 2. been assumed in previous literature that the obstacles in the way of improving engine technology are overcome. It can be seen that as component efficiency increases, with BPR held constant a smaller fan diameter is required, simply by inverting the logic presented in the previous paragraph. This decrease in fan diameter with increasing component efficiency has the cascading effect of causing a decrease in landing gear length and hence engine weight and landing gear weight both decrease. The trends in the three components affecting the aircraft performance, namely SFC, aircraft weight, and lift-to-drag ratio are shown in Figure 4-25, against increasing BPR. As expected, the SFC decreases with increasing BPR; moreover as BPR is held constant, SFC decreases as component efficiency increases. As mentioned previously the decrease in SFC is compounded by the fact that not only does it decrease at a constant BPR as component efficiency increases, but a larger bypass ratio is possible. Meanwhile it can be seen that increasing component efficiency at constant BPR causes a decrease in final aircraft weight, due to the fact that landing gear weight and engine weight are decreasing. In addition to this we see the intuitive trend of final weight increasing as BPR increases. Finally the lift-to-drag ratio remains essentially constant as BPR increases, showing that the increasing drag of a larger engine nacelle has minimal effect at an overall aircraft level, according to the drag buildup method of 105

106 this model. Note that there is not true expected value included for this graph since there was no dataset for verification of this parameter, therefore the average standard deviation was used to calculate confidence bounds. (a) SFC against BPR showing expected values and 95% confidence bounds. (b) Lift-to-drag ratio against BPR showing expected values and 95% confidence bounds. (c) Aircraft mission final weight against BPR showing expected values and 95% confidence bounds. Figure 4-25: Variation of aircraft performance variables with bypass ratio for scenario 2. The aircraft performance associated with these trends is shown in Figure 4-26, in terms of the block fuel weight. It can be seen that the maximum performance occurs at a BPR of approximately 28. At bypass ratios above this it can be seen that performance deteriorates due to the fact that the power requirements of a bigger fan are too large for the turbine to supply sufficient power from the given flow. Eventually this leads to an infeasible turbofan engine beyond a BPR of 30. The associated performance changes can be seen in Table The time frame predicted for this scenario are projected using both of the BPR trend models from Section 3.2, with the results displayed in Figure 4-27 and Table It can be seen that the expected value for the maximum performance increase is 2049 using model 1 and 2037 using model 2, with respective single standard deviation confidence intervals of [] and []. The mean of these values is 2043, which is the expected year in which the dominant architecture will need to change in order to further increase aircraft performance, given the assumptions of the model and technology scenario

107 Figure 4-26: Aircraft performance against bypass ratio, quantified as the block fuel weight, for scenario 2. Shown on the graph are 95% confidence bounds the expected value, as well as current maximum BPR range and the maximum possible performance for this scenario Scenario 3:Bypass ratio increase with increase in turbine inlet temperature and overall pressure ratio This scenario assumes that the technologies related to increasing the bypass ratio of the turbofan engine will improve as well as the turbine inlet temperature, as defined in Section 3.2. The overall pressure ratio constraints are relaxed since as shown in [21] in order to fully obtain the benefits of increasing TIT the OPR must also increase. The assumed TIT increase is from 1700K to 1800K, although this could possibly increase further this is done based on the predictions presented from previous research. The evolution of aircraft geometry associated with these changes can be seen in Figure As in the previous two scenarios, the landing gear length, landing gear track and distance of the engine from the centerline are all increasing with increasing BPR. The geometry and weight changes of these components are shown in Figure As in scenario 2, increasing TIT & OPR at a constant BPR leads to a smaller fan diameter, shorter landing gear and therefore lower engine and landing gear weight. Once again the values at the BPR which gives maximum performance are compared with the baseline scenario at a BPR of 11, representative of the current state-of-theart technology. Included on these graphs are the estimate expected values, the true expected values corrected for bias, and 95% confidence bounds, for both scenario 3 and the baseline scenario. As in scenario 2, it can be seen that by increasing TIT a larger bypass ratio is enabled than the baseline scenario. This is due to the fact that the energy added to the system is now higher since the limiting temperature of the engine core has now increased. This further energizes the core flow allowing more power to be extracted therefore enabling a larger fan to be driven by the turbine. Since more fuel is being 107

108 Parameter Current technology Best case baseline scenario Difference [%] Lower Upper EV Lower Upper EV Lower Upper EV BPR Fan diameter [m] Engine weight [kg] 3,382 4,362 3,872 5,302 6,281 5, Landing gear length [m] Landing gear ] weight [kg] 3,010 4,859 3,934 3,351 5,413 4, SFC [ lbf lbm hr Lift-to-drag ratio Final aircraft weight W final 47,118 78,478 62,798 51,967 83,327 67, Block fuel weight [kg] 13,379 21,608 17,493 9,288 17,128 13, Performance Table 4.12: Comparison of geometry and performance variables for scenario 2, for today s performance versus the maximum performance. Figure 4-27: Graph of BPR against year showing both projection models based on different data sets for scenario 2. added to increase the energy, increasing the OPR is required to reduce SFC. In order for this scenario to be viable new materials for low pressure turbines with high creep resistance will have to be developed. In addition the increase in OPR will increase NO x emissions, therefore low NO x combustors will have to be developed since emissions regulations will require this. Both of these developments of technologies should not be taken for granted since they require substantial research and development efforts. Having said this, these advances are implicit in the trends of Section 3.2. The resulting changes in SFC, aircraft final weight and lift-to-drag ratio with increasing BPR can be seen in Figure It is noticeable that these trends are the same as those in technology scenario 2 with a slightly lower magnitude of change. The aircraft performance associated with these trends is shown in Figure 4-31, in terms of the block fuel weight. The minimum value of block fuel weight can be seen to occur at a BPR of 24. The values at this BPR are compared with the baseline values for current technology in Table

109 85% confidence 68% confidence Lower Upper EV Lower Upper EV Model Model Mean Table 4.13: Expected values for architecture break year including confidence bounds for two BPR forecast models for scenario 2. Figure 4-28: The evolution of aircraft design parameters as BPR increases in scenario 3. On the diagram the trend is highlighted with the different colors. The two BPR projection models are used to estimate a time frame for this technology scenario, with the results presented in Figure 4-32 and Table It can be seen that the expected value for the maximum performance increase is 2044 using model 1 and 2034 using model 2, with respective single standard deviation confidence intervals of [] and []. The mean of these values is 2039, which is the expected year in which the dominant architecture will need to change in order to further increase aircraft performance, given the assumptions of the model and technology scenario Scenario 4: All engine technologies improve This scenario assumes that all the turbofan technologies mentioned in Section 3.2 improve. The evolution of aircraft geometry associated with these changes can be seen in Figure 4-33, and follows the same trends as the previous two scenario. Naturally this scenario follows similar trends to those of the previous 2 scenarios with increasing BPR; therefore the graphs of these trends have been included in Appendix D. The graph of block fuel weight can be seen in Figure A comparison with the baseline at a BPR of 11, in terms of geometry, weight and performance, can be seen in Table As with the previous two scenarios, these results are subject to the assumptions of the model and assuming that obstacles associated with the development of these technologies can be overcome. In this scenario an even larger bypass ratios are made possible due to the more efficient extraction of energy from the core flow. The range of BPR extends to ap- 109

110 (a) Engine fan diameter against BPR showing expected values and 95% confidence bounds. (b) Engine weight against BPR showing expected values and 95% confidence bounds. (c) Main landing gear length against BPR showing expected values and 95% confidence bounds. (d) Total landing gear weight against BPR showing expected values and 95% confidence bounds. Figure 4-29: Variation of aircraft design parameters as BPR increases for scenario 3. proximately 34 in this scenario with maximum performance occurring at a BPR of 30. Once again the two BPR projection models are used to estimate a time frame for this technology scenario. The graph is presented in Appendix D, with the results summarized in Table 4.17 It can be seen that the expected value for the maximum performance increase is 2052 using model 1 and 2039 using model 2, with respective single standard deviation confidence intervals of [] and []. The mean of the values from both models is This is the year in which it is expected that the dominant architecture will reach its peak performance under the conditions and assumptions of scenario 4. It is worth noting the significance of the value of a BPR of 30 for a turbofan engine. According to Ciepluch obtaining the full benefits of bypass ratio maximization requires ratios of the order of 30-50, which requires an unducted fan or open rotor configuration [29]. Meanwhile Czech cites an effective BPR of 25 or above to be favorable to an open rotor architecture over a ultra-high BPR turbofan [34]. The prediction of the above analysis that at a BPR of 30 an aircraft architectural change may be required in order to further increase performance, aligns with the 110

111 (a) SFC against BPR showing expected values and 95% confidence bounds. (b) Lift-to-drag ratio against BPR showing expected values and 95% confidence bounds. (c) Aircraft mission final weight against BPR showing expected values and 95% confidence bounds. Figure 4-30: Variation of aircraft performance variables with bypass ratio for scenario 3. results from literature. We predict a break of the current architecture subject to the given assumptions of this scenario at a BPR of 30 due to maximization of performance of this architecture. The results from literature show that beyond a BPR of approximately 25-30, an unducted fan is required to fully take advantage of BPR maximization. Should one switch to an unducted fan, this constitutes a break in the current dominant architecture due to a change in Engine Type. It is likely that this change will be associated with changes in the airframe architecture in order to fully exploit the potential benefits of a new engine architecture. Interestingly, there is no reason that these two approaches should coincide in terms of their results, however they both come to the same conclusion. We take this one step further by predicting that this is expected to occur in 2045 given current trends. One would assume that, given the long development cycle of a civil passenger aircraft, on the order of 10 years, this is more likely to be around 2035 in this scenario Summary and discussion of results The reason that these two analyses have been carried out separately is to analyze two possible situations. The first analysis corresponds to the potential decision of manufacturers to continue the current trend of simply installing new engines on existing 111

112 Figure 4-31: Aircraft performance against bypass ratio, quantified as the block fuel weight, for scenario 3. Shown on the graph are 95% confidence bounds the expected value, as well as current maximum BPR range and the maximum possible performance for this scenario. airframes. The second analysis corresponds to the potential decision of a manufacturer to develop a new aircraft with the dominant architecture. These two potential decisions encompass the possible trajectories which lead to the current architecture remaining dominant. Given that it has been shown that engine technology and architecture is the major driver for aircraft architecture, the engine has been the focus of these analysis. Consequently, taking current trends in engine technology and examining when the aircraft performance will reach is maximum gives an idea of when we might expect the current dominant architecture to break. Engine technology trends have been identified from literature including increases in bypass ratio, component efficiencies, turbine inlet temperature and overall pressure ratio. Trends in these over time have been extrapolated from literature and models have been taken or created to forecast these, particularly the main trend of increasing bypass ratio. Four technology scenarios were chosen from these prevailing engine technology trends to represent realistic improvements within the given time frames. For example scenario 2 involves improvements in technologies enabling larger bypass ratios and higher component efficiencies, which represents a scenario in which a manufacturer chooses to focus its funding on research towards improving these. Although costs of technology improvements are not explicitly considered int his analysis, examining these trends separately is a proxy for this. Additionally this enables us to isolate the extent to which each of these trends drives aircraft performance, changes in geometry of the dominant architecture and architecture disruption. The first half of Chapter 4 focused on examining the trend of substituting more efficient engines onto existing airframes. Four technology scenarios were compared, for both the 737 and A320 airframes. It was shown that depending on the scenario a further 10-38% improvement in performance for the 737 and a further 6-30% for 112

113 Parameter Current technology Best case baseline scenario Difference [%] Lower Upper EV Lower Upper EV Lower Upper EV BPR Fan diameter [m] Engine weight [kg] 3,382 4,362 3,872 5,027 6,007 5, Landing gear length [m] Landing gear ] weight [kg] 3,010 4,859 3,934 3,309 5,344 4, SFC [ lbf lbm hr Lift-to-drag ratio Final aircraft weight W final 47,118 78,478 62,798 51,281 82,641 66, Block fuel weight [kg] 13,379 21,608 17,493 10,704 18,544 14, Performance Table 4.14: Comparison of geometry and performance variables for scenario 3, for today s performance versus the maximum performance. Figure 4-32: Graph of BPR against year showing both projection models based on different data sets. the A320 are possible. This is expected to happen in the next years given the current technology trends, at which point it is expected that a new architecture may be pursued in order to further increase performance. The second half of Chapter 4 analyzes the same trend within the context of the dominant architecture, without constraints on the geometry of the airframe. Hence technology trends are used to forecast the maximum possible performance given the same four technology scenarios. To summarize these results, it is shown that in scenario 1 we can expect a performance increase of 17.1% by 2035; in scenario 2 the expected value of performance improvement is 32.4% by 2043; in scenario 3 the performance increase is expected to be 19.6% by 2039; and finally in scenario 4 the performance is expected to increase by 39.9% by Note that the baseline for improvement in the first analysis is existing aircraft, which are not optimized for the most state-of-the-art technology. In the second analysis the baseline for measuring the performance improvement is optimized for current 113

114 85% confidence 68% confidence Lower Upper EV Lower Upper EV Model Model Mean Table 4.15: Expected values for architecture break year including confidence bounds for two BPR forecast models for scenario 3. Figure 4-33: The evolution of aircraft design parameters as BPR increases in scenario 3. On the diagram the trend is highlighted with the different colors. technology resulting in a higher performance; therefore it is not fair to compare the values of existing airframes to the values in this second analysis. Furthermore the assumptions of each of the technology scenarios must be taken into account since they assume a certain level of technology development in order to reach these quoted maximum performances. Despite this the results are summarized in Figure 4-35 for performance increase and 4-36 for architecture break year. While comparisons between performance improvements are not comparable across aircraft it is easy to compare the results across the four scenarios for each aircraft case. Nevertheless it is possible to compare the architecture break point across the aircraft. This shows us that the A320 architecture will break later than the 737 which is expected due to its longer landing gear and therefore larger area for engine placement under the wing. Furthermore allowing new aircraft to be build by varying the geometry of the dominant architecture naturally allows this architecture to persist for longer than existing airframes. The main result from this graph is that, for a given technology scenario, it is possible to see the expected value of architecture break for two cases: 1) if the manufacturer does not develop a new aircraft and keeps the existing airframe and 2) if the manufacturer optimizes the geometry for the largest possible engine for a given technology scenario. The results of this analysis can be compared with existing literature. Mentioned in scenario 4 of the second analysis are results from Czech [34] and Ciepluch [29], which have predicted that beyond a BPR of an unducted fan is more beneficial than a turbofan or geared turbofan architecture. This coincides with the predictions 114

115 Figure 4-34: Aircraft performance against bypass ratio, quantified as the block fuel weight, for scenario 3. Shown on the graph are 95% confidence bounds the expected value, as well as current maximum BPR range and the maximum possible performance for this scenario. of these results that an architectural change will be necessary to improve aircraft performance beyond a BPR of depending on the technology scenario. It is well documented that aircraft development requires long lead times on the order of a decade [43]. When this is taken into account it is evident that if for example maximum aircraft performance is expected to occur in 2035, a manufacturer would have to begin the development process around When an aircraft in a given class requires replacing, the manufacturer must decide whether to modularly innovate by simply installing new engines and other small add-ons, incrementally innovate by designing a new aircraft with the same architecture and small improvements, or architecturally innovate. They must decide whether a potential performance improvement is worth the associated cost. This means that even though an incremental improvement may be possible, it may be more effective to change the architecture in terms of the associated costs and benefits. The analysis in this section assumes that a new aircraft is produced for each value of bypass ratio, since the design is re-optimized in each case to give the maximum performance for that given level of technology. In reality implementing such changes is not so easy due to the aforementioned long lead time development cycle. This lead time to launch a new aircraft means that new technologies at the start of the development tend to be old by the time the aircraft enters service. Including these real world issues in this analysis would change the results, however it is beyond the scope of this research. Instead it is assumed that as long as the performance can be improved within the current architecture, this will dominate any decision to change architecture. Therefore the time frames associated with maximum performance within the four technology scenarios is assumed to be the approximate date of a break in architecture. If the dominant architecture is to reach its performance limit in the suggested 115

116 Figure 4-35: Comparison of the maximum performance increase for the four technology scenarios for each of the aircraft. Figure 4-36: Comparison of the architecture break point for the four technology scenarios for each of the aircraft. 116

117 Parameter Current technology Best case baseline scenario Difference [%] Lower Upper EV Lower Upper EV Lower Upper EV BPR Fan diameter [m] Engine weight [kg] 3,382 4,362 3,872 4,979 5,959 5, Landing gear length [m] Landing gear ] weight [kg] 3,010 4,859 3,934 3,301 5,331 4, SFC [ lbf lbm hr Lift-to-drag ratio Final aircraft weight W final 47,118 78,478 62,798 51,162 82,522 66, Block fuel weight [kg] 13,379 21,608 17,493 8,584 16,424 12, Performance Table 4.16: Comparison of geometry and performance variables for scenario 4, for today s performance versus the maximum performance. 85% confidence 68% confidence Lower Upper EV Lower Upper EV Model Model Mean Table 4.17: Expected values for architecture break year including confidence bounds for two BPR forecast models for scenario 4. time frame then a period of architectural exploration will ensue. In Section 3.2 it was argued that for current aircraft performance as defined by the Breguet range, SFC could have the maximum possible impact, with aircraft weight providing less of an impact, and lift-to-drag ratio providing essentially no impact. These were verified in the results section where it can be seen that SFC dominates the other parameters. If aircraft architecture were to be changed, there would be an opportunity to increase lift-to-drag ratio and decrease aircraft empty weight, relative to the payload weight (i.e. the structural efficiency of the aircraft). Concepts such as the flying wing or blended-wing body could result in this occurring. Having said this, architectural changes such as moving the engine location are more likely due to a smaller change propagation, therefore a lower risk for manufacturers. The effect on lift-to-drag ratio and structural efficiency would not be as significant in this case. Hence once again with an architectural change the main driver of improvement in performance would be a decrease in SFC of the engines. This will be associated with a change in engine architecture, since the turbofan architecture as we know it would no longer provide an increase in performance. It is worth noting that the aircraft operates within an air transport system, therefore there are other systemic factors which need to be considered when viewing this problem. In this thesis, the focus is primarily on the technological aspect affecting the performance of a given passenger aircraft. For this reason the main driver of aircraft performance that has been used is the passenger carrying efficiency, a measure of the operating efficiency of a given aircraft. Additionally the focus on the aircraft 117

118 fuel consumption relative to the number of passengers and aircraft range is a proxy for the economic viability of an particular aircraft. Therefore in the results it is assumed that the mass of fuel consumed is the major driver for aircraft improvement. While this is generally a valid assumption it has been shown that when environmental effects such as contrail formation and NO x emissions are considered as part of the aircraft performance, decreasing fuel burn does not necessarily translate into increasing performance [49]. In bounding the problem, architecture and performance drivers pertaining to environmental impact, aircraft maintainability and reliability, air traffic management, airline operations, environmental regulations, etc. have not been explicitly considered. Additionally aircraft airworthiness for certification is a major consideration for manufacturers, due to the high costs associated with this process. Certification generally takes a long time, which for manufacturers delays the period by which they may begin making returns on their billions of dollars of development costs. This is one of the main reasons for the risk aversion for manufacturers, since with every new aircraft they are essentially betting the existence of their company [58]. In this example it is evident that policies such as airworthiness standards for certification play a major role in the system. These are both good because they check for aircraft safety and reliability, however they carry a financial burden for the industry. Policies such as these could be used to create incentives for a desired behavior, such as a lower environmental impact. One of the original motivations of this thesis, was the Paris Agreement and reducing the effect of aviation on anthropogenic global average temperature rise. The following chapter will examine how various policies could be implemented to stimulate technological innovation for reducing the environmental impact of aviation, and what effects such policies could have on aircraft architecture. 118

119 Chapter 5 Aviation policy: implication for architectural changes 5.1 Introduction and motivation Nearly half of the global population, over 3 billion people, traveled by air in 2015 [17]. Air transportation directly contributes to increased quality of life by enabling economic activity to be carried out more efficiently, as well as enabling travel for pleasure across the globe. As well as these intangible economic benefits, the air transport industry is responsible for the direct and indirect employment of 56 million people worldwide [41]. The FAA, ICAO, IATA and many more civil aviation authorities have all identified the need for aviation to develop to meet the needs of an expanding population and a growing global economy, in an environmentally sustainable manner. This is stated in the FAA Aviation Environment and Energy Policy Statement [2], which cites the guiding principles of 1) limiting and reducing future aviation environmental impacts to levels that protect public health and welfare and 2) ensuring energy availability and sustainability. One of the motivating factors in this research is the Paris Agreement and the global commitment to keep average global temperature rise to well below 2 Celsius [18]. As is well established, one of the major contributors to the rise in average global temperatures is the emission of man-made greenhouse gases (GHG). According to the US Environmental Protection Agency (EPA), transportation accounts for 26% of total GHG, with aviation responsible for 12% of these, and approximately 3% of the total emissions [93]. However given that these emissions are usually at an more sensitive altitude, the contribution of aviation to radiative forcing is higher than this, estimated to be 1.5 times that at ground level [93]. Recalling the main literature review, Lee et al projected a 1.1% annual increase in fuel efficiency for aircraft [75]. A recent report by the firm Transport & Environment estimated this to be between % depending on regulatory pressure and industry efforts [67]. Meanwhile demand is expected to increase to over 7 billion passengers by 2034 [14], an annual rate of 4.6%, according to Airbus [5]. Given that the decrease in fuel efficiency will occur at less than a quarter of the rate of increase in passenger traffic, we can expect that the impact of aviation on the environment is set to increase. Owen et al have projected a increase in carbon dioxide emissions from aviation from 2000 to 2050, and a factor increase in the oxides of nitrogen emissions, based on various scenarios [91]. Given the fraction of global emissions due to aviations and the goals of the Paris Agreement, it is clear that the trend in aviation emissions will have a greater impact on the environment in the future, if something is not done. 119

120 In Chapter 4 of this thesis, the effects on aircraft performance of technology options were examined, where aircraft performance was quantified in terms of the weight of fuel required to carry out a particular mission. Reducing the amount of fuel required was seen as an increase in aircraft performance due to the increase in economic efficiency of the aircraft. Therefore, commercially, there is a drive to improve technologies since fuel consumption accounts for approximately 30-40% of the direct operating costs for airlines, depending on their operating model [112]. In addition to measuring economic efficiency, fuel burn is often viewed as a proxy for environmental impact, therefore it would seem at first glance that economic and environmental interests are aligned in this instance. This is shown not to be entirely true by Green, who cites that while fuel burn is directly correlated to carbon dioxide emissions, the production of oxides of nitrogen, another GHG, are not necessarily correlated to fuel burn [48]. In particular, he states that there is a conflict between the goals of reducing fuel burn through greater thermal efficiency and of reducing NO x emissions. It is evident that a decrease in specific fuel consumption results in lower operating costs for a given price of fuel. However, from the review of alternative aircraft architectures in Section 1.3, it is clear that more fuel efficient architectures such as the BWB [77], are not available in the market. We would usually assume that because the demand of the market for passenger aircraft is for the most fuel efficient product, the supply side would provide this solution. For a number of reasons the market has failed in this respect. These reasons may include: architectural lock-in of the passenger air transport system including all infrastructure; the trade-off between upfront capital expenditure to develop a new aircraft versus incremental improvement of current aircraft; risk aversion of the aerospace industry including the cost of developing and certifying a new aircraft, as well as deploying new, unproven technologies; and, the incentive to continue improving aircraft reliability, to reduce maintenance costs and maximize time-in-service for airlines. Given the realities of this industry, it is the goal of a policy-maker to implement solutions to fix these market failures, by incentivizing a desired outcome. Policy solutions to all of the above issues is a rather large undertaking. In order to narrow on scope, the main theme of this thesis, focusing on aircraft architecture and performance will be continued; therefore the focus will be on reducing fuel burn with the goal of reducing CO 2 emissions. Before doing so we will embark on a brief background of aviation emissions and the current policies and regulations, in order to identify where additional policies or amendments to existing policies could result in the desired outcome. 5.2 Background Firstly, a commonly used metric for measuring the drivers of climate change is radiative forcing, which is defined by Ramaswamy et al. as the change in net (down minus up) irradiance (solar plus longwave; in W m 2 ) at the tropopause after allowing for stratospheric temperatures to readjust to radiative equilibrium, but with surface and tropospheric temperatures and state held fixed at the unperturbed values. It is commonly used to characterize the effects of aviation emissions on global climate. As 120

121 can be seen in Figure 5-1a, the aviation emissions related to anthropogenic radiative forcing consist of three main elements: 1. Carbon Dioxide: The effects of carbon dioxide on climate change is well understood [93]. CO 2 acts as a GHG through absorption and reemission of infrared radiation [25]. An increase in CO 2 atmospheric concentration causes a warming of the troposphere and a cooling of the stratosphere. Thus, the atmospheric concentration of CO2 is one of the most important factors in climate change [93]. 2. Oxides of Nitrogen: The influence of NO x is important in the chemistry of the troposphere and the stratosphere as well as in ozone production and destruction processes [93]. In these atmospheric regions NO x emissions tend to increase ozone levels, which then acts as a GHG. NO x emissions are also known to contribute to reducing the atmospheric lifetime of another GHG, methane and be offset by H 2 O and sulphur emissions, although these processes are not well understood yet [93]. 3. Condensation trails (contrails): Contrails are formed from water vapour emitted by aircraft, depending on exhaust and atmospheric conditions. They are thin cirrus clouds, which reflect solar radiation and trap outgoing longwave radiation, resulting in positive radiative forcing. The level of scientific understanding of contrail radiative forcing is low since there are many uncertainties in global values [93]. According to the IPCC, the contributions of these and other emissions to radiative forcing, and the projections over time given different scenarios, can be seen in Figure 5-1. It is evident that aviation emissions amount to more than just CO 2 emissions, which constitute approximately 35% of the total contribution of aviation to climate change [48]. Nevertheless due to its direct correlation to fuel burn which is in itself an economic driver, and the much greater understanding of its effects on climate change, the focus of this chapter will be on curbing CO 2 emissions in particular. 5.3 A brief history of aviation emissions standards The United Nations International Civil Aviation Organization (ICAO) Committee on Aviation Environmental Protection (CAEP) has been developing standards and recommended practices for environmental protection since the 1980s [62]. The standards and practices adopted in countries such as the US are often based on these, in order to align with the international community. A timeline of ICAO s emissions standards can be seen in Appendix??. It can be seen from this figure that ICAO provides standards for emissions certification applying to turbojet and turbofan engines and their derivatives, where relevant certificating authorities have the power to grant this certification. The emissions that are currently controlled for are: 121

122 (a) Radiative forcing of aviation emissions in 1992, according to the IPCC. (b) IPCC projections of aircraft contributions to radiative forcing for different scenarios. Figure 5-1: Contributions to radiative forcing through aviation emissions and their projections over time. 1. Smoke 2. Gaseous emissions: (a) Unburned hydrocarbons (UHC) (b) Carbon monoxide (CO) (c) Oxides of Nitrogen (NO x ) This standards applies only to the landing and take-off (LTO) cycle, and are measured and reported in grams, for particular thrust settings and time in operating mode, at ISA conditions and at sea level. In the US, the responsibility for promulgating these standards lies with the Environmental Protection Agency (EPA) under the Clean Air Act (CAA), while the Federal Aviation Authority (FAA) is responsible for enforcing them [42]. The EPA has most recently adopted Tier 8 standard for engine models certified after 1 January 2014, aligning with ICAO CAEP/8 standard, and representing a 15% reduction in NO x emissions from the previous Tier 6 standard. This standard applies to all newly certified engines above a thrust of 26.7kN, therefore applying to many civil passenger aircraft Gaps in current standards The above enforced standards apply solely to: 1. LTO cycles which cover an altitude below 3000 feet 2. NO x emissions of the main GHG contributors identified in this section 122

123 3. Aircraft engines Therefore there is a gap in covering the whole operating envelope, other GHG emissions such as CO 2, and standards for the entire aircraft. The focus on LTO cycles is to regulate local air quality in areas surrounding airports. More recently, as a reaction to the Paris Agreement, ICAO has proposed fuel efficiency standards for new aircraft, implying a proposed standard on CO 2 emissions. This new environmental measure was unanimously recommended by the 170 international experts on ICAOs Committee on Aviation Environmental Protection (CAEP), pertaining to new aircraft beyond 2020 and new deliveries beyond 2023 [86]. Initial references cite that this goal will be to reduce CO 2 emissions by 2% per annum [25]. Given that there are plans to introduce standards for CO 2 it is the goal of this chapter to propose how such a standard might work and analyze the effects of this on aircraft architecture and performance. 5.4 Goals The goal of the policy analysis in this thesis is to encourage the air transport industry to develop new technologies or architectures to curb aviation emissions, and to adopt these in current and future aircraft. The emphasis on technology development focuses the analysis on aircraft and engine manufacturers, rather than on operations of aircraft and airline operators. Hence the goals of this policy section are: 1. Identify and recommend potential policies to incentivize manufacturers to increase aircraft fuel efficiency 2. Analyze how such a policy will affect aircraft architecture 5.5 Policy options and recommendation Engine certification standards Similar to the standards on smoke and gaseous emissions, a standard for engine certification based on CO 2 emissions throughout the possible engine operating conditions could be adopted. This is a way to directly regulate emissions, that fits in well with the current certification process. Developing such a standard will require specification of a metric for measuring CO 2 emissions against an engine parameter, such as thrust. This is required in order to normalize CO 2 emissions for varying aircraft sizes. The specific fuel consumption under various operating conditions is one such parameter that could be used. Additionally it would be necessary to examine the extra cost and time for certification so as not to place undue burden on manufacturers. One aspect of this will be to use a more in-depth analysis to the one presented in this thesis, with more specific technology forecasts to predict possible improvements in fuel efficiency. Not that the standards should be stringent enough to force technology development and stimulate innovation, however not so stringent as to place huge financial burdens 123

124 on the industry, and compromise aircraft safety. Much additional research beyond the scope of this research would be required to analyze such trade-offs Aircraft certification standards The current standards focus on engine certification, however as was mentioned in Section 3.2, other levers to increase fuel efficiency include increasing ( L D) and decreasing W e. One could conceive of potential scenarios in which there is a reduction in SF C that is offset by these two overall aircraft parameters. Hence there is a need to conceive of a certification standard, to promote reduction in overall aircraft system level CO 2 emissions. In order to create incentives for airframe manufacturers to innovate, an aggregated metric to set standards for aircraft certification based on fuel burn is required. This would require measuring lift, drag, aircraft weight, fuel burn and normalizing these to account for differing missions, that is number of passengers and range. Such a standard would require more time and costs in the certification phase of new aircraft. Measuring such parameters would place a larger burden on manufacturers to prove that the airframe and engines meet the required standard, hence this system must be carefully designed. Nevertheless, setting a threshold for overall aircraft fuel burn would then force new technologies or architectures to be developed and adopted. Often parameters such as lift, drag, weight, and SFC are traded for each other. Hence a standard beyond just a subsystem level is required to account for this. Developing a metric and methods to measure such a standard will be required. One could conceive of a aggregated performance metric similar to the passenger carrying efficiency metric used in this thesis. Further research and analysis would be required to devise a more in-depth measurement since, in reality, the Breguet range equation is not entirely accurate due to its assumption of cruise-climb Market-based options There are methods, other than a direct standard on performance, to try to incentivize more fuel efficient aircraft, via improved technologies or architectures. Market-based methods rely on market forces to create a demand-side need for more fuel efficient aircraft. This in turn creates incentives for more fuel efficient aircraft to be supplied to the market by manufacturers. Two of the most promising market-based solutions, as cited by Green [48], are presented below. Voluntary agreements Voluntary agreements on fuel efficiency levels seems an obvious choice, since it is in the interest of most of the stakeholders in the system to increase fuel efficiency. For manufacturers, greater fuel efficiency means their products are more attractive to their customers. This option may seem redundant since manufacturers aim to do this regardless of any voluntary agreement; moreover the fact that it is voluntary means they are not required to do so anyway. The advantage in voluntary agreements is the fact that they could be implemented much quicker, and they are less likely to 124

125 face strong opposition [48]. Additionally working with industry to set targets is itself a useful activity. In contrast to other market-based options such as levies or taxation, this does not place a direct financial burden on the industry, and does not require enforcing by a regulatory body. Adherence to such a voluntary agreement could qualify manufacturers for increased levels of capital allowances to be spent on research and development. One such voluntary agreement could include speeding up the phase-out in production of older less fuel efficient aircraft. Emissions trading It is clear that voluntary agreements do not guarantee that fuel efficiency will increase. An often cited market-based policy is a carbon cap and trade scheme, either within the aviation industry or as part of a wider scheme involving other industries. Traditionally there are issues with such a system for airlines, since there are many hundreds of airlines across the globe. Simply designing a system that all airlines would buy into seems like an impossible task, never-mind the cost and mechanism to enforce it. With the recent Paris Agreement, the opposition to such a system would be weakened however it is still practically difficult to implement. On option that has been suggested is to enforce a cap and trade scheme at the manufacturer level [48]. This would involving capping the number of permits for the level of annual CO 2 emissions for airframes and engines sold by the manufacturers. Such a system would grandfather in existing operational aircraft, but would create incentives for the manufacturers to pursue more fuel efficient options. The value of each permit is left to the market to decide through the trading of permits. This system directly controls the level of CO 2 emissions going forward, however restricts the growth of the aviation industry to that achievable by increases in fuel efficiency. To avoid this restriction one might consider including the manufacturers within a larger cap and trade system. That being said, the proposed closed system provides more incentive for developing high fuel efficiency technologies and architectures. Since the vast majority of the single aisle and wide-body aircraft are manufactured by either Airbus or Boeing, and there are three major engine manufacturers (GE Aviation, Pratt & Whitney, Rolls-Royce), the number of separate entities to regulate is immediately reduced from hundreds to less than 10. The additional burden on manufacturers would be passed onto airlines and eventually passengers, thereby spreading the costs of this system. There are many issues which require additional research in order to implement such a system. For example, the level at which to cap the CO 2 emissions and how this should change over time requires extensive research. An additional question is how to distribute the permits among players in the industry. These issues are non-trivial and must be answered in order to implement a cap and trade system Recommendation On of the goals of this chapter was to identify and recommend a policy that will directly create incentives to reduce fuel burn through the development of new architectures and technologies for airframe and engine manufacturers. Given this context, 125

126 the policy option that provides the largest incentive for a reduction in overall aircraft fuel burn, and hence CO 2 emissions, through innovation in aircraft systems is enforcing an aircraft certification standard. This is not a new proposition, for example it has been suggested by Bonnefoy et al. in their assessment of CO 2 emissions in aviation [25]. This option has been selected above the other options since it directly addresses the goal of innovation in technologies, as well as providing a flexible standard that allows trade-offs to be performed between the various levers for fuel burn reduction. Currently emissions certification is done at the engine level, however we recommend an aircraft level certification standard, to account for other factors contributing to fuel burn. Bonnefoy et al. recommend a similar policy, citing that there is no strong correlation between aircraft and engine level efficiency and for a given engine fuel efficiency the aircraft fuel efficiency can vary by a factor of 3. This option is favored above market-based options, since it is more direct than voluntary agreements, and it fits within the current structure easier than an emissions trading system. As mentioned above, introducing any certification standard will increase the time and costs of certification. This is a fine balance between stimulating innovation and causing manufacturers excessive financial burden, therefore the standards must be carefully planned and implemented. Since this in is the interest of the general public and in line with the goals of the Paris Agreement, one way to solve this problem initially would be to provide financial assistance for research and development as well as to ease the initial certification costs. In addition to this, existing aircraft will be grandfathered in to such a standard. In order to prevent manufacturers from simply creating derivatives of these aircraft in the future incentives for phasing out existing models could be provided, including increased capital allowances to spend on developing new aircraft. The effect of such a standard on aircraft architecture will be examined in the following section. 5.6 Effect of recommended policy on aircraft architecture It has been shown that fuel burn and CO 2 emissions are directly correlated. In fact it can be calculated that approximately 3kg of CO 2 is produced per kg of jet fuel burned directly [27]. Therefore the metric used previously, namely block fuel weight for the design mission, is an adequate proxy metric for CO 2 emissions. At this high level utilizing the Breguet equation in order to calculate the block fuel is a reasonable assumption, despite the fact that it assumes the aircraft continually climbs during cruise to make the most economical use of fuel. Since ICAO have proposed normalizing the fuel burn by the number of kilometers flown, a fuel burn per flight kilometer metric will be used in the analyses below. Two specific cases will be analyzed in this section, the first based on ICAO s goal for fuel burn reduction [25] and the second based on IATA s Vision 2050 goals [13]: 1. Under the current ICAO recommendation of 2% reduction in fuel burn per year, when can we expect the dominant architecture to break. 126

127 2. If we wish to cut fuel emissions to 70% relative to their 2010 levels by 2050, when can we expect the dominant architecture to break Policy Scenario 1: 2% reduction in fuel burn A 2% annual reduction in fuel burn, in this context is taken to mean using 2% less fuel per flight kilometer each year. The maximum reduction in fuel burn for each of the four technology scenarios has previously been calculated, for both existing and future aircraft exhibiting the dominant architecture. Previously the estimate of the time for the dominant architecture to be disrupted was based on forecasts of engine technology improvements. In this case, it is assumed that technologies at the airframe or engine level will be developed and adopted in order to reach this mandated goal of a 2% reduction in fuel burn per year. Therefore the driver has changed to a policy-push rather than a technology-pull scenario. The fuel burn per flight kilometer for each of the aircraft under each of the scenarios, is computed. Compounding a 2% reduction in fuel burn per year until the maximum possible fuel reduction for each scenario is reached, it is possible to forecast the year in which the architecture is expected to break. For number of years n, this is done using the following formula, (0.98) n = 1 κ ij, (5.1) where κ is the maximum possible decrease in fuel burn per flight kilometer for each aircraft i and scenario j as calculated in Chapter 4. In this analysis the 737-8MAX and A320neo are taken to be the baseline in 2016 for their respective analyses. The baseline for the new aircraft is taken to be an aircraft optimized for current technology and it is assumed that in each subsequent year the aircraft is re-optimized for the new standard. These results are presented in Table 5.1. Scenario Year of architecture break 737 airframe A320 airframe New aircraft 1. Baseline Scenario Component efficiency TIT & OPR All technologies Table 5.1: Expected year of break of architecture for the 737, A320, and a new optimized aircraft with the dominant architecture, for four technology scenarios under policy scenario Policy Scenario 2: 70% fuel burn reduction by 2050 relative to 2010 In this situation it is assumed that a 70% reduction in overall fuel burn translates to 2.97% reduction per year as compared to the previous year. It is possible that such 127

128 fuel reductions could occur as a result of improvements in technology or operations. For the purposes of this discussion it is assumed that this is expected to come solely from technological or architectural innovation. Therefore this policy scenario is more stringent than the first scenario. The estimated years for a break of architecture are shown in the table below. Scenario Year of architecture break 737 airframe A320 airframe New aircraft 1. Baseline Scenario Component efficiency TIT & OPR All technologies Table 5.2: Expected year of break of architecture for the 737, A320, and a new optimized aircraft with the dominant architecture, for four technology scenarios under policy scenario Discussion The two policy scenarios analyzed in this section represent situations in which a certification standard for aircraft based on fuel burn is imposed. Rather than having a staged decrease in the maximum acceptable fuel burn over time, a continuous standard is used, which becomes increasingly stringent every year based on a compound percent decrease. Comparing these two scenarios with the expected values from the scenarios in which there is no policy forcing function, it is possible to analyze the potential efficacy of such a policy. In Chapter 4 forecasts of technology improvement are used to forecast the architecture break point. Since these forecasts are based on expected technology improvement they can be viewed as the natural progression of the industry, in which case a policy that does not stimulate innovation more than this is assumed to be ineffective. One might be tempted to view the previous technology forecasts as the industry limit. Since these are based on extrapolation of current technologies, a policy that has the goal of forcing innovation, is meant to cause more than just incremental improvements or business as usual, and push the boundaries of architectural innovation. This comparison can be seen in Figure 5-2. Comparing across the three policy scenarios, as expected it can be seen that in the cases where there is an emission standard, the architecture breaks sooner. Additionally, the more stringent standard causes a break in architecture before the less stringent standard for all technology scenarios and aircraft. Policy scenario 1 was adapted from the ICAO goal of 2% reduction in fuel burn per year, and policy scenario 2 was adapted from the IATA Vision 2050 goal of 70% reduction in fuel burn by 2050 relative to The goal of these policies was to stimulate innovation and clearly in these situations they have. However in order to implement such a policy, there are associated costs which could be very high. These costs have not been taken into consideration in this analysis, but these would have to be weighed against the benefits 128

129 Figure 5-2: A comparison of the expected year for the architecture break given 3 different policy scenarios, for 4 different technology scenarios and for 3 different aircraft cases. of the reduction in fuel burn. The efficacy of these policies in stimulating innovation in the aviation industry is debatable, since the architecture break point in most of the scenarios has not changed by many years. Having said this it could be worth spending money to implement such a system, despite small expected improvements in fuel burn. What has not been shown on the figure is the uncertainty involved in these estimates. It has been shown in Chapter 4 that the uncertainties in the estimates based on a technology forecast are fairly large. By contrast, an emissions standard significantly reduces this uncertainty. In fact, given that manufacturers are required to comply with this standard for new aircraft and derivatives of existing aircraft, we are certain that we will reach the desired fuel burn targets. A carefully designed emission standard which does not impose excessive financial burden on the industry reduces this uncertainty. Although the year of architecture break of the 129

130 scenario with no standard is similar to the scenarios with a standard, the variance in this expected value is much higher. Therefore the value in implementing an aircraft emissions certification standard for CO 2 is to decrease the uncertainty in this desired income, in this case through technological innovation. 5.7 Conclusions The following can be concluded from this chapter: 1. There exists a gap in current policies to regulate the emissions of the air transportation industry. Currently, of the major GHG emissions, only NO x is regulated for the landing and takeoff cycles, through the use of an engine certification standard. 2. In order to reduce CO 2 emissions over time for the entire operating envelope of an aircraft, several policy options are presented. An overall aircraft CO 2 emissions certification standard is recommended due to its advantages over subsystem certification and other policy options. 3. In order to implement an aircraft emissions certification standard, there are many things to be considered. It has been shown that this can be effective in obtaining the desired reduction in CO 2 emissions and fuel burn over time through creating incentives for innovation. Three policy scenarios are analyzed across the four technology scenarios for each of the aircraft in Chapter 4. It is shown that the ICAO and IATA recommendations for fuel burn result in the architecture break point occurring earlier in time. Additionally these policies are shown to be effective in reducing the uncertainty over time, which is a benefit for all stakeholders in the system. 4. Before such a policy is implemented there are many obstacles that need to be overcome. The stakeholders in the system must be included in its design so as to not face opposition in its implementation. The extra financial burden to the industry must be considered, and adequate support provided so as to not encounter a situation where progress is not possible due to excessive demands. 5. Finally, the recommended policy must operate within a complex system, which includes other regulations and standards, particularly NO x emissions regulations, noise regulations and airworthiness requirements or safety standards. The interactions with these standards must be considered since usually there are trade-offs involved. A poorly designed certification standard could give rise to emergent properties of the system which compromise safety - this must be very carefully considered. 130

131 Chapter 6 Conclusions Since aviation was first commercialized for passenger travel, it has rapidly grown to be one of the most technologically advanced and safe modes of transportation. This development has been associated with advances in the passenger air transport system of systems, particularly developments in aircraft architecture and technology, which have driven continuous improvements in system performance. The significance of aircraft architecture in the past and future development of passenger air transportation has been a central theme of this thesis. The following points summarize the main results: 1. A method to consider aircraft architecture in the conceptual stages of design has been presented. Analysis of 157 historical aircraft architectures has shown a consolidation over time to a dominant design. Aircraft performance, as defined by the metric in this thesis, has followed an S-curve trajectory over the same period. 2. There is a tight coupling between engine technologies/architecture and aircraft architecture. It has been shown that, historically, innovations in engine technology have led to aircraft architectural innovation to fully exploit the benefits of the given engine. 3. A first principles analysis of the dominant aircraft architecture was carried out showing that the most important driver for performance improvement is the specific fuel consumption of the engine. The major drivers for reduction in SFC were identified as increasing bypass ratio, increasing overall pressure ratio & turbine inlet temperature, and increasing component efficiency. These constituents were analyzed and forecast using empirical models and a review of literature, producing four engine technology scenarios. 4. A hybrid analytical-empirical model to analyze the interactions between airframes and engines for the dominant architecture was developed. It was verified with a database of 25 dominant architecture aircraft, resulting in mean error on the order of 5%. 5. The dominant architecture was analyzed in the context of existing aircraft. Using geometrical constraints, the 737 and A320 airframes were examined for four turbofan engine technology scenarios. It was found that there is still potential to increase performance of these aircraft solely by re-engining. For the four technology scenarios, the expected improvements in performance for the 737 are 9.7%, 30.5%, 20.3%, and 38.1% compared with the baseline 737-8MAX. 131

132 For the A320 these expected values are 5.8%, 23.6%, 14.4%, and 30.4% compared with the A320neo. It is shown that the expected date for architecture disruption for the 737 is depending on the scenario. The equivalent years for the A320 are expected to be These are the expected years under the technology scenarios that the under-wing turbofan engine infringes on the geometric constraints of the airframe. It is not realistic to assume that in 2017 or 2018 there will be a break in architecture since the 737MAX and A320neo are new aircraft and this does not account for development time of a new aircraft.however this lower bound serves to highlight that the current airframes are reaching their limit and at this point either a new aircraft with the dominant architecture is needed or the architecture will break. 6. In the second main analysis the dominant architecture was examined without constraints on the airframe geometry, for the same four engine technology scenarios. It is shown that the expected performance increase is 17.1%, 32.4%, 19.6%, and 39.9% for scenarios 1-4 respectively. These maximum performance increases are expected by 2035, 2043, 2039 and 2045 respectively. 7. These two main studies represent the options faced by manufactures due to the geometric limitations of the dominant architecture, given engine technology trends. The first analysis represents the case in which a manufacturer decides to continually upgrade the existing airframe. The second analysis represents the case in which a manufacturer decides to develop a new aircraft with the dominant architecture. The mission profile and performance baseline for the aircraft in the two analyses are different therefore the increase in performance cannot be compared. Nevertheless the year of architecture break can be compared and it is shown that the 737 and A320 will likely break in the next years, whereas the option for a newly designed aircraft would lead to an architecture break in years. The alternative to these two options would be to develop a new architecture, however this is assumed to occur only after the current architecture reaches its performance limits. 8. Finally, policy options for curbing CO 2 emissions are presented and their effect on aircraft architecture is analyzed. It is recommended that an aircraft certification standard is adopted in order to create incentives for the civil passenger aircraft manufacturing industry to develop and adopt more fuel efficient technologies. It is shown that ICAO s recommendation of 2% reduction in fuel burn per year and IATA s Vision 2050 target of 70% lower fuel burn, both stimulate technological innovation more than in a situation where this is left to the market. Furthermore the scenario with no policy is based on a technology forecast with much uncertainty. Implementing certification standards has the desired effect of reducing this uncertainty. The above results and conclusions must be examined within the context of the original motivating questions. Diminishing improvements in aircraft performance, combined with increase in passenger demand are in direct conflict with environmental 132

133 goals. This raised the question of how this conflict could be resolved without demandside restrictions. Given this context, historical aircraft architecture and performance were examined and their trajectories charted. A consolidation of architecture to a dominant design was observed associated with the aforementioned diminishing returns in performance. The importance of engine technology and architecture on aircraft architecture and performance was a major finding. Extrapolating the most important engine technological trends within the context of this dominant architecture, trajectories for the future of civil passenger aircraft architecture were charted. Four engine technology scenarios were examined for two different cases: existing airframes including the 737 and A320; and a new aircraft with the dominant architecture. Given the dominant architecture, the trajectory of future aircraft architecture was assumed to be a continuation of this same trend until a break point in the architecture - an assumption extracted form the historical analysis which shows a gradual incremental development rather than radical disruptions in the aviation industry. This break point in the dominant architecture is assumed to occur when it is no longer possible to improve upon the performance of the aircraft for the given level of technological development in each of the four scenarios. Finally, analysis of policy options to curb fuel burn was carried out. Two policy scenarios were formulated based on real proposals and examined to observe the effects relative to a situation in which there is no implemented policy. It is seen that smart policies can be used to stimulate innovation in this industry and reduce the uncertainty of future emissions, if designed to not place undue financial burden on the industry. The rule was used for identifying the technologies to forecast in each of the four scenarios. That is, there exist other levers which could be used to improve the performance of the aircraft which have been presented in the first principles analysis in Section 3.2. This serves to highlight that the results presented in this thesis must be examined by taking into account the assumptions in the inputs, in the model and the constraints at the boundaries of the analyses. In this thesis, a new way to view civil passenger aircraft has been presented. As far as this author knows, there has been no previous comprehensive and quantitative analysis of aircraft architecture trajectories, past and future. This work serves to give the reader a different perspective of the industry, which tends to follow a law of architectural continuity. In order for passenger air transportation to advance technologically, it is not enough to suggest a superior point design in the architectural space. Lessons from the historical analysis have taught us that incentives for a disruptive change in aircraft architecture are so low that technology advances within the current paradigm would have to be exhausted before architectural exploration would be considered. This thesis does not analyze the entire spectrum of possible technological advances, therefore there is much scope for future work. Firstly technological advances other than those within the engine could be examined, such as new structural materials for the airframe, laminar flow technology and many more. Secondly the influence of external factors on aircraft architecture could be examined, such as airport infrastructure. It has been mentioned that airworthiness certification requirements lead to risk aversion in the industry; therefore these requirements could be examined in more detail and policies could be suggested to incentivize technological innovation within this context. Finally, in this thesis the trajectories of aircraft architecture and 133

134 performance have been analyzed from the 1930s until potential breaking points in the dominant architecture. This could be extended to beyond this breaking point; that is, one could link the postulated future architectures to this break point by analyzing them in the context of the entire system including technology development, airworthiness certification, emissions standards, manufacturer risk aversion, and of course their potential costs and benefits to society as a whole. 134

135 Bibliography [1] Federal Aviation Administration. Faa federal aviation regulations. AC [2] Federal Aviation Administration. Aviation environmental and energy policy statement. Technical report, Department of Transportation, [3] Simthsonian National Air and Space Museum. America by air: Airline expansion and innovation [4] Airbus. A320 aircraft characteristics airport and maintenance planning, [5] Airbus. Airbus Global Market Forecast Technical report, [6] Commercial Airplanes. Statistical summary of commercial jet airplane accidents. Worldwide Operations, 2008, [7] John David Anderson. Aircraft performance and design, volume 1. WCB/McGraw-Hill Boston, [8] Mr Gerrod Andresen and Mr Zachary Williams. Metrics, key performance indicators, and modeling of long range aircraft availability and readiness. NATO. Research and Technology Organization (Ed.), Enhanced aircraft platform availability through advanced maintenance concepts and technologies, RTO Technical Report, NATO, Research and Technology Organisation, Neuilly-sur-Seine CEDEX, France, pages 7 1, [9] Nicolas E Antoine and Ilan M Kroo. Optimizing aircraft and operations for minimum noise. AIAA Paper, 5868:1 3, [10] Nicolas E Antoine and Ilan M Kroo. Framework for aircraft conceptual design and environmental performance studies. AIAA journal, 43(10): , [11] Dale Curtis Arney. Rule-based graph theory to enable exploration of the space system architecture design space. PhD thesis, Georgia Institute of Technology, [12] Scott W Ashcraft, Andres S Padron, Kyle A Pascioni, Gary W Stout Jr, and Dennis L Huff. Review of propulsion technologies for n+ 3 subsonic vehicle concepts [13] International Air Transport Association et al. Vision Available in: www. iata. org, [14] International Air Transportation Association. Iata air passenger forecast shows dip in long-term demand. 135

136 [15] Raffi Babikian, Stephen P Lukachko, and Ian A Waitz. The historical fuel efficiency characteristics of regional aircraft from technological, operational, and cost perspectives. Journal of Air Transport Management, 8(6): , [16] Dilip R Ballal and Joseph Zelina. Progress in aeroengine technology ( ). Journal of aircraft, 41(1):43 50, [17] World Bank. Air transport, passengers carried. [18] Scott Barrett, Carlo Carraro, and Jaime De Melo. Towards a workable and effective climate regime. Technical report, [19] Jonathan A Battat, Bruce Cameron, Alexander Rudat, and Edward F Crawley. Technology decisions under architectural uncertainty: informing investment decisions through tradespace exploration. Journal of Spacecraft and Rockets, 51(2): , [20] Meyer J Benzakein. What does the future bring? a look at technologies for commercial aircraft in the years Propulsion and Power Research, 3(4): , [21] NT Birch vision: the prospects for large civil aircraft propulsion. Aeronautical Journal, 104(1038): , [22] Benjamin S Blanchard, Wolter J Fabrycky, and Walter J Fabrycky. Systems engineering and analysis, volume 4. Prentice Hall New Jersey;, [23] Boeing. Airport compatibility brochure 737 max, [24] Boeing. Boeing long term market: Current market outlook Technical report, [25] Philippe A Bonnefoy, Graham Burdette, R Thomas Carroll, John Hansman, James Hileman, Steven Isley, Michelle Kirby, Dongwook Lim, Dimitri Mavris, Aleksandra Mozdzanowska, et al. Assessment of co2 emission metrics for a commercial aircraft certification requirement. PARTNER Project, 30, [26] Geoffrey C Bower and Ilan M Kroo. Multi-objective aircraft optimization for minimum cost and emissions over specific route networks. In 26th international congress of the aeronautical sciences, [27] Guy P Brasseur, Mohan Gupta, Bruce E Anderson, Sathya Balasubramanian, Steven Barrett, David Duda, Gregg Fleming, Piers M Forster, Jan Fuglestvedt, Andrew Gettelman, et al. Impact of aviation on climate: Faas aviation climate change research initiative (accri) phase ii. Bulletin of the American Meteorological Society, (2015), [28] Rick Brown. Managing the s curves of innovation. Journal of Consumer Marketing, 9(1):61 72,

137 [29] Carl C Ciepluch, Donald Y Davis, and David E Gray. Results of nasa s energy efficient engine program. Journal of Propulsion and Power, 3(6): , [30] Edward Crawley, Bruce Cameron, and Daniel Selva. System Architecture: Strategies and product development for complex systems. Pearson, 3 edition, [31] Edward Crawley, Olivier De Weck, Christopher Magee, Joel Moses, Warren Seering, Joel Schindall, David Wallace, Daniel Whitney, et al. The influence of architecture in engineering systems (monograph [32] Edward F. Crawley and Olivier deweck. System architecture: An overview and agenda, [33] Nicholas Cumpsty and Andrew Heyes. Jet propulsion. Cambridge University Press, [34] Michael J Czech and Russell H Thomas. Experimental studies of open rotor installation effects. AIAA Paper, 4047, [35] Emily Schwartz Dallara, Ilan M Kroo, and Ian A Waitz. Metric for comparing lifetime average climate impact of aircraft. AIAA journal, 49(8): , [36] Peter Davison, Demetrios Kellari, Edward F Crawley, and Bruce G Cameron. Communications satellites: Time expanded graph exploration of a tradespace of architectures. Acta Astronautica, 115: , [37] Peter Leslie Davison. Tradespace exploration for space system architectures: a weighted graph framework. PhD thesis, Massachusetts Institute of Technology, [38] Olivier de Weck. Strategic engineering: Designing systems for an uncertain future [39] Adam John Diedrich. The multidisciplinary design and organization of an unconventional, extremely quiet transport aircraft. PhD thesis, Massachusetts Institute of Technology, [40] Mark Drela. Development of the d8 transport configuration. AIAA Paper, 3970:2011, [41] Oxford Economics. Aviation benefits beyond borders, [42] US EPA. Regulatory announcement: New emission standards for new commercial aircraft engines. Technical report, [43] Henry L Eskew. Aircraft cost growth and development program length: Some augustinian propositions revisited. Technical report, DTIC Document,

138 [44] John P Fielding. Introduction to aircraft design, volume 11. Cambridge University Press, [45] Flightglobal. Can new entrants take on airbus and boeing, [46] J P Gardiner. Design trajectories for airplanes and automobiles during the past fifty years. Design, Innovation and Long Cycles in Economic Development, Francis Printer, London, pages , [47] Amir S Gohardani, Georgios Doulgeris, and Riti Singh. Challenges of future aircraft propulsion: A review of distributed propulsion technology and its potential application for the all electric commercial aircraft. Progress in Aerospace Sciences, 47(5): , [48] JE Green. Greener by design. In Proc. AAC-Conference, Friedrichshafen, Germany, 30 June, pages , [49] JE Green. Air travel-greener by design mitigating the environmental impact of aviation: Opportunities and priorities-report of the air travel-greener by design science and technology sub group london. Aeronautical Journal, 109(1099): , [50] Royal Aeronautical Society Greener by design group. Greener by design annual report Technical report, [51] Edward M Greitzer, PA Bonnefoy, E DelaRosaBlanco, CS Dorbian, M Drela, DK Hall, RJ Hansman, JI Hileman, RH Liebeck, J Lovegren, et al. N+ 3 aircraft concept designs and trade studies. volume 2; appendices-design methodologies for aerodynamics, structures, weight, and thermodynamic cycles [52] Abhijit Guha, D Boylan, and P Gallagher. Determination of optimum specific thrust for civil aero gas turbine engines: a multidisciplinary design synthesis and optimisation. Proceedings of the Institution of Mechanical Engineers, Part G: Journal of Aerospace Engineering, page , [53] Bill Gunston. Jane s Aero-engines. Jane s Information Group, [54] David Kenneth Hall. Performance limits of axial turbomachine stages. PhD thesis, Massachusetts Institute of Technology, [55] DK Hall, EM Greitzer, and CS Tan. Performance limits of axial compressor stages. In ASME Turbo Expo 2012: Turbine Technical Conference and Exposition, pages American Society of Mechanical Engineers, [56] Rebecca M Henderson and Kim B Clark. Architectural innovation: The reconfiguration of existing product technologies and the failure of established firms. Administrative science quarterly, pages 9 30,

139 [57] James I Hileman, Jeremy B Katz, Jose Mantilla, and Gregg Fleming. Payload fuel energy efficiency as a metric for aviation environmental performance. In Proceedings of the 26th International Congress of the Aeronautical Sciences, pages 1 11, [58] Peter Hollingsworth. An investigation of value modelling for commercial aircraft. In Air Transport and Operations Symposium, volume 2012, pages , [59] Timothy J Horn and Ola LA Harrysson. Overview of current additive manufacturing technologies and selected applications. Science progress, 95(3): , [60] Denis Howe. Aircraft conceptual design synthesis, volume 5. Wiley, [61] IATA. 100 years of commercial flight. [62] ICAO. Annex 16 to the convention on international civil aviation: Environmental protection [63] CFM International. Cfm leap brochure, [64] AT Isikveren. Distributed propulsion and ultra-high by-pass rotor study at aircraft level (dispursal). FP7. AAT, [65] Paul A Jackson, Kenneth Munson, Lindsay Peacock, and Susan Bushell. Jane s All the World s Aircraft Jane s, [66] Brian K Kestner, Jeff S Schutte, Jonathan C Gladin, and Dimitri N Mavris. Ultra high bypass ratio engine sizing and cycle selection study for a subsonic commercial aircraft in the n+ 2 timeframe. In ASME 2011 Turbo Expo: Turbine Technical Conference and Exposition, pages American Society of Mechanical Engineers, [67] Anastasia Kharina and Daniel Rutherford. Fuel efficiency trends for new commercial jet aircraft: 1960 to [68] Bernard L Koff and Bernard L Koff. Gas turbine technology evolution: A designers perspective. Journal of propulsion and power, 20(4): , [69] Bernard L Koff and Steven G Koff. Engine design and challenges for the high mach transport. International Journal of Turbo and Jet Engines, 26(4): , [70] Benjamin HY Koo, Willard L Simmons, and Edward F Crawley. Algebra of systems: a metalanguage for model synthesis and evaluation. Systems, Man and Cybernetics, Part A: Systems and Humans, IEEE Transactions on, 39(3): ,

140 [71] Ilan Kroo. Distributed multidisciplinary design and collaborative optimization. VKI lecture series on optimization methods and tools for multicriteria/multidisciplinary design, [72] Ilan Kroo, Steve Altus, Robert Braun, Peter Gage, and Ian Sobieski. Multidisciplinary optimization methods for aircraft preliminary design. AIAA paper, 4325:1994, [73] Ilan Kroo and Richard Shevell. Aircraft design: Synthesis and analysis. Desktop Aeronautics Inc., Textbook Version 0.99, [74] Ajoy Kumar Kundu. Aircraft design, volume 27. Cambridge University Press, [75] Joosung J. Lee, Stephen P. Lukachko, Ian A. Waitz, and Andreas Schafer. Historical and future trends in aircraft performance, cost, and emissions, [76] Leifur T Leifsson, William H Mason, Joseph A Schetz, Raphael T Haftka, and Bernard Grossman. Multidisciplinary design optimization of low-airframe-noise transport aircraft. AIAA Paper, 230:2006, [77] Robert H Liebeck. Design of the blended wing body subsonic transport. Journal of aircraft, 41(1):10 25, [78] J-P Marec. Drag reduction: a major task for research. In Aerodynamic Drag Reduction Technologies, pages Springer, [79] Joaquim RRA Martins and Andrew B Lambe. Multidisciplinary design optimization: a survey of architectures. AIAA journal, 51(9): , [80] William Mason. Aircraft design information sources: Landing gear design, [81] John H McMasters and Russell M Cummings. Airplane design-past, present, and future. Journal of Aircraft, 39(1):10 17, [82] Ronald E Miller and David Sawers. The technical development of modern aviation. Praeger Publishers, [83] D Moorhouse, B Sanders, M Von Spakovsky, and J Butt. Benefits and design challenges of adaptive structures for morphing aircraft. Aeronautical Journal, 110(1105): , [84] John Morgenstern, Nicole Norstrud, Marc Stelmack, and Craig Skoch. Final report for the advanced concept studies for supersonic commercial transports entering service in the 2030 to 2035 period, n+ 3 supersonic program [85] Taiga Nakamura and Victor R Basili. Metrics of software architecture changes based on structural distance. In Software Metrics, th IEEE International Symposium, pages IEEE,

141 [86] ICAO newsroom statement. New icao aircraft co2 standard one step closer to final adoption. [87] Shigeru Obayashi. Multidisciplinary design optimization of aircraft wing planform based on evolutionary algorithms. In IEEE International Conference on Systems Man and Cybernetics, volume 4, pages , [88] Bureau of Transportation Statistics. U.s. air carrier traffic statistics through january [89] ICAO Committee on Aviation Environmental Protection Working Group 3. Key areas of activity and description of work. [90] International Civil Aviation Organization. Technical report. [91] Bethan Owen, David S Lee, and Ling Lim. Flying into the future: aviation emissions scenarios to Environmental science & technology, 44(7): , [92] Paul Peeters, J Middel, and A Hoolhorst. Fuel efficiency of commercial aircraft. An overview of historical, [93] Joyce E Penner. Aviation and the global atmosphere: a special report of ipcc working groups i and iii in collaboration with the scientific assessment panel to the montreal protocol on substances that deplete the ozone layer. Technical report, [94] Almarin Phillips. Technology and market structure: A study of the aircraft industry. Heath Lexington Books, [95] Gordon Pirie. Passenger traffic in the 1930s on British imperial air routes: refinement and revision. The Journal of Transport History, 25(1):63 83, March [96] DIA Poll. A first order method for the determination of the leading mass characteristics of civil transport aircraft. Aeronautical Journal, 115(1167): , [97] Daniel Raymer. Enhancing aircraft conceptual design using multidisciplinary optimization [98] Daniel P Raymer. Aircraft design: A conceptual approach. American Institute of Aeronautics and Astronautics, 4 edition, [99] JA Rosero, JA Ortega, E Aldabas, and LARL Romeral. Moving towards a more electric aircraft. Aerospace and Electronic Systems Magazine, IEEE, 22(3):3 9, [100] Jan Roskam. Airplane design: Parts I-VII. Darcorporation, 2 edition,

142 [101] Mohammad H Sadraey. Aircraft Design: A Systems Engineering Approach. John Wiley & Sons, [102] Herbert Ian Howard Saravanamuttoo, Gordon Frederick Crichton Rogers, and Henry Cohen. Gas turbine theory. Pearson Education, [103] Robert R Schaller. Moore s law: past, present and future. Spectrum, IEEE, 34(6):52 59, [104] Emily Schwartz and Ilan M Kroo. Aircraft design: Trading cost and climate impact. AIAA 2009, 1261, [105] Daniel Selva. Rule-based system architecting of Earth observation satellite systems. PhD thesis, MIT, [106] Matthew R Silver and Olivier L De Weck. Time-expanded decision networks: A framework for designing evolvable complex systems. Systems Engineering, 10(2): , [107] H Smith. College of aeronautics blended wing body development programme. In Icas 2000 congress, [108] Ashish Sood and Gerard J Tellis. Technological evolution and radical innovation. Journal of Marketing, 69(3): , [109] Greener by Design Technology Sub-Group. Air travel: the technology challenge. Technical report, [110] Mark Thomas. Better power for a changing world: Next generation engines. Technical report, [111] Egbert Torenbeek. Synthesis of subsonic airplane design: an introduction to the preliminary design of subsonic general aviation and transport aircraft, with emphasis on layout, aerodynamic design, propulsion and performance. Springer Science & Business Media, [112] Gerassimos Tsoukalas, Peter Belobaba, and William Swelbar. Cost convergence in the us airline industry: An analysis of unit costs Journal of Air Transport Management, 14(4): , [113] James M Utterback. Innovation in industry and the diffusion of technology. Science, 183(4125): , [114] James M Utterback, Bengt-Arne Vedin, Eduardo Alvarez, Sten Ekman, Susan Walsh Sanderson, Bruce Tether, and Roberto Verganti. Design-inspired innovation. World Scientific, [115] Bijan Vasigh, Ken Fleming, and Thomas Tacker. Introduction to air transport economics: from theory to applications. Ashgate Publishing, Ltd.,

143 [116] Philip P Walsh and Paul Fletcher. Gas turbine performance. John Wiley & Sons, [117] Tony Washburn. Airframe drag/weight reduction technologies. Green Aviation Summit-Fuel Burn Reduction, NASA Ames Research Centre, [118] Wenbin Wei and Mark Hansen. Cost economics of aircraft size. Journal of transport Economics and Policy, pages , [119] TM Young, B Humphreys, and JP Fielding. Investigation of hybrid laminar flow control (hlfc) surfaces. Aircraft Design, 4(2): , [120] Y. Zhu. Fatigue stress (north carolina state university lecture notes), [121] RA Zimbrick and JL Colehour. Investigation of very high bypass ratio engines for subsonic transports. Journal of Propulsion and Power, 6(4): ,

144 144

145 Appendix A Architecture decisions Figure A-1: Trends in architectural decision-options over time plotted in performancetime space. Four decisions are visualized here corresponding to engine type (top left), wing shape (top right), number of engines (bottom left) and wing vertical location (bottom right). 145

146 146

147 Appendix B Airframe-engine model B.1 Operating conditions calculation B.1.1 Function overview This function calculates the operating conditions for each mission segment. Figure B-1: Block diagram of operating conditions function showing inputs and outputs. B.1.2 Inputs Parameter M h h i Description The Mach number at a given altitude, set by mission specifications The altitude at a given mission segment, i B.1.3 Process A database of atmospheric air conditions is used to obtain air properties at different altitudes. The velocity of the aircraft is calculated using B.1 and the dynamic pressure is given by B.2. v i = M i a h (B.1) q i = 1 2 ρ hv 2 i (B.2) 147

148 B.1.4 Outputs Parameter v i a h ρ h T h q i σ h Description The velocity of the aircraft at given of air at a given mission segment i The speed of sound at a given altitude h The density of air at a given altitude h The temperature of the atmosphere at a given altitude h The dynamic pressure at given mission segment i The ratio of air density at altitude h to the air density at sea level B.2 Initial Sizing calculation B.2.1 Function overview This function uses the weight fraction method of Raymer to estimate the weight of the empty aircraft and the fuel weight [98]. This method involves assumptions as well as first principles calculation of weight fractions for each segment of the mission, followed by an iterative process to converge on a takeoff weight for the aircraft. Figure B-2: Block diagram of initial sizing function showing inputs and outputs. 148

149 B.2.2 Inputs Parameter Description r cruise The cruise range of the aircraft r alternate The range to an alternate airport t l oiter The time spent in loiter above an airport W pax The weight of a passenger and their luggage N pax The maximum number of passengers W crew The weight of a crew member and their luggage N crew The maximum number of crew members ( L )max The maximum lift-to-drag ratio of the aircraft D SF C cruise The specific fuel consumption during cruise. This is approximated from existing engine data as a first estimate. SF C loiter The specific fuel consumption during loiter. This is approximated from existing engine data as a first estimate. B.2.3 Processes The total aircraft weight can be expressed as, W to = W crew + W p + W f + W e, (B.3) which can be re-written in the following form, W to = W crew + W p. 1 W f W to We W to (B.4) The initial sizing method uses the expression in B.4 to iteratively solve for the takeoff weight. The fuel weight fraction W f W to is estimated using the fuel weight fraction method described below. Meanwhile the empirical formula in B.5 is used to estimate the empty weight fraction We W to, obtained from Raymer [98]. W e = 0.97Wto 0.06 W to (B.5) The following weight fractions of fuel are assumed for the specified segments of the mission profile. Segment Description Weight fraction W j W j Start, warm-up & taxi, takeoff & accelerate Climb Descent, land & taxi For the cruise segments, 4-5 and 7-8, the Breguet range equation is used to determine the weight fraction. This is a commonly used equation derived from first 149

150 principles, and can be rearranged as in B.6 to give the mission fuel weight fraction. [ ] W i 1 r cruise SF C cruise = exp W i V ( L ) (B.6) D cruise It is assumed here that ( L ) D cruise = 0.866( L ) D max, which is proven in Raymer [98]. For the loiter segments, 5-6, the Breguet endurance equation B.7 is used, which is a slightly modified version of the previous equation. W i 1 W i = exp [ t loiter SF C loiter ( L D ) loiter ] (B.7) For the purposes of this sizing it is assumed that there is no rejected takeoff segment, and the aircraft would simply be told from its loiter operation to cruise to an alternate airport. Given the above weight fractions the total mission weight fraction is calculated using B.8, where r f is the percentage of reserve fuel, necessary for contingency and safety and is equal to W f = r f W to i=1 W i 1 W i (B.8) Meanwhile the weight of the payload is calculated using W p = W pax N pax, since the only payload assumed for the purposes of this analysis is passengers. Similarly the weight of the crew is calculated. B.2.4 Outputs It is worth noting that often the words mass and weight are used interchangeably in the aerospace industry. both of these refer to mass, which is measured in kilograms, unless explicitly stated. Parameter W to/w 0 W e W p W f W j W j 1 Description The maximum takeoff weight of the aircraft The empty weight of the aircraft, the takeoff weight excluding fuel and payload The weight of the payload The weight of fuel The weight fraction of a segment in between segment start and end points j 1 and j for mission segment i ( L D ) cruise The lift-to-drag ratio of the aircraft during cruise B.3 Thrust-to-weight ratio and wing loading B.3.1 Function overview This function calculates the maximum thrust-to-weight ratio for the aircraft based on its size and the mission specification, and uses this to determine the wing-loadings 150

151 the aircraft will experience during each segment. The minimum value of ( W S ) i, for the various mission segments, i, is the one which is used to determine the wing area, S. A note on indexing, the mission segment 1 corresponds to that between numbers 1 and 2 in Figure 3-7, therefore mission segment i corresponds to the segment length rather than segment start and end points. Figure B-3: Block diagram of thrust-to-weight and wing loading calculation function showing inputs and outputs. B.3.2 Inputs Parameter Description C D0 The zero-lift drag coefficient of the aircraft. This is initially estimated based on legacy aircraft and then updated when the values are obtained for the current aircraft. A The wing aspect ratio. (same method of estimation as above) e Oswald s efficiency factor to account for 3D wing effects compared to ideal infinite wings. This is approximated from existing data. C LmaxHL The maximum coefficient of lift of the aircraft with high-lift devices deployed. This is assumed based on legacy aircraft and then updated when the values for the current design are available. θ climbmin The minimum climb gradient of the aircraft after takeoff. This is given by FAR 25 standards usually or based on legacy aircraft. T OP The takeoff parameter, an empirical factor used to obtain wing loading for takeoff. This is taken from Figure 5.4 in [98] ( L )max The maximum ratio of lift to drag for the aircraft. This is initially estimated based on D legacy aircraft and then updated when the values are obtained for the current aircraft. ρ i The density of air at given mission segment i v i The velocity of travel at given mission segment i q i The dynamic pressure of air at given mission segment i S a The obstacle clearance height after takeoff, given by FAR 25 standards for airliners W j 1 The fuel weight fractions for each mission segment, where j 1 and j correspond to the W j numbers at the end and beginning of each segment i C lmax The maximum coefficient of lift for the selected airfoil W to The aircraft maximum takeoff weight 151

152 B.3.3 Processes The first step in this function is to calculate the maximum thrust-to-weight ratio. The first estimate is made for cruise operations, given by B.9. ( ) T 1 = ( W L (B.9) D) cruise cruise This is used to calculate the thrust-to-weight ratio at takeoff using B.10, ( ) ( ) ( ) ( ) T T Wcruise Tto =, (B.10) W W ( where ( ) W cruise W to is calculated using the weight fractions, ) T to T cruise to cruise is between 4 and 5 and ( Wj 1 W j ), from the initial sizing function. W to T cruise The climb thrust-to-weight ratio is calculated using B.11. ( ) T 1 = ( W L + θ climbmin (B.11) D) climb Another method used to estimate the maximum thrust-to-weight ratio, based on empirical data is given in B.12 [98]. This is compared with the previously calculated values in B.13 and the maximum value is selected as the performance requirement for the aircraft. ( ) T = 0.267Mcruise (B.12) ( T W W 0 ) max max [( ) ] T = max W i (B.13) With ( ) T, the wing loadings for various operating conditions can be calculated W max in the following way. For stall speed the wing loading can be computed using, ( ) W = 1 S 2 ρv2 stallc LmaxHL, (B.14) where C LmaxHL is estimated using, stall [ ] S flapped S unflapped C LmaxHL = 0.9 (C lmax ) flapped + (C lmax ) S unflapped. (B.15) ref S ref In this analysis C lmaxhl is the estimated maximum coefficient of lift for the airfoil with high lift devices (= 2.4 for flaps and slats), V stall = V approach, and S flapped 1.3 S ref is the ratio of flapped area to wing reference area, taken as a first approximation to equal

153 The takeoff distance wing loading is calculated using, ( ) ( ) W T = (T OP )σc Lto S to W to (B.16) where C Lto = C Lmax HL The landing distance wing loading is given using equation B.17 for landing distance taking into account safety factors. The landing distance is estimated by the empirical formula S landing = 0.3v 2 approach, [98] ( ) ( ) W 1 S landing = S landing σc Lmax + S a (B.17) Cruise wing loading is given by, and loiter wing loading is given by, ( ) W S ( ) W πaecd0 = q cruise S cruise 3 loiter Meanwhile the wing loading required for climb is given by, ( ) W = S climb (B.18) = q loiter πaecd0. (B.19) { [( T ) θ ] [( W climb climb min + T ) θ ] 2 4C W climb climb min D0 πae 2 q climb πae }. (B.20) Each of the wing loadings must be converted to takeoff conditions using B.21, in order to make them comparable, that is, such that the W being used is the takeoff weight. ( ) ( ) W W = S i,to S i ( Tto Finally the limiting wing loading is given by, ( ) [ (W ) W = min S S T i i,to ). (B.21) ]. (B.22) 153

154 B.3.4 Outputs Parameter Description ( W S ) i The wing loading for given mission segment, i ( T W ) i The thrust-to-weight ratio for given mission segment, i B.4 Geometry definition This function defines the major geometrical parameters of the preliminary design of the aircraft, in order to meet the performance requirements of the mission specification. Figure B-4: Block diagram of geometry definition function showing inputs and outputs. 154

155 B.4.1 Inputs Parameter ( ) W S min W to W f A\A wing A HT A V T λ\λ wing λ HT λ V T Λ c 4 Λ c 4 HT Λ c 4 V T ( t c ) max C HT C V T fr fus q ( cruise ) W S Description The aircraft wing loading required at the limit of performance The aircraft maximum takeoff weight The aircraft mission fuel weight The wing aspect ratio - the ratio of the square of the wingspan divided by the wing area The horizontal tail aspect ratio The vertical tail aspect ratio The wing taper ratio - the ratio of the wing tip chord to the wing root chord The horizontal tail taper ratio The vertical tail taper ratio The sweep angle at the wing quarter chord position The sweep angle at the horizontal tail quarter chord position The sweep angle at the vertical tail quarter chord position The maximum thickness to chord ratio of the airfoil The horizontal tail volume coefficient, which is an empirical value used for sizing the horizontal tail area The vertical tail volume coefficient, which is an empirical value used for sizing the vertical tail The fineness ratio of the fuselage, the ratio of the length of the fuselage to the maximum width The cruise dynamic pressure The wing loading at the performance limit B.4.2 Process The following are taken from the design books [98, 100, 111, 101], assuming a trapezoidal swept wing, as is required by the dominant architecture. The wing area, S, is calculated using B.23. S = ( W W ) (B.23) S The wing space is computed using B.24, the wing root chord using B.25 and the wing tip chord using??. b = A S (B.24) c root = 2S b(1 + λ) c tip = c root λ The wing mean chord is computed using B.27, c = 2 3 c 1 + λ + λ 2 root 1 + λ, (B.25) (B.26) (B.27) while the spanwise position along the wing of the mean chord (the y-direction) is given by, ȳ = b 1 + 2λ λ. (B.28) 155

156 The wing leading edge sweep is calculated using B.29, assuming a trapezoidal wing, tan(λ LE ) = tan(λ c ) + 1 λ 4 A(1 + λ). (B.29) The horizontal and vertical tail are assumed to have the same trapezoidal shape, therefore the geometries of these are calculated in the same way as the wing geometry replacing wing values with the tail values. The tail areas are required to carry this out, which are computed using the tail volume coefficient method. Tail volume coefficient values are assumed, that is c HT = 1 and c V T = 0.09, which are based on legacy aircraft [98] and used in B.30 and B.31 to find the respective areas. We have, c HT = l HT S HT c w S w (B.30) c V T = l V T S V T (B.31) b w S w where l HT, the distance from the quarter chord point of the wing to that of the horizontal tail, is assumed as a first approximation to be 0.5 to 0.55 of the fuselage length l. Likewise l V T is assumed to be between the same two values as a first approximation. Consequently the fuselage geometry can be approximated by assuming a fuselage fineness ratio, fr fus = l d max, between 8 and 10, based on legacy aircraft, and utilizing the empirical formula in B.32. l = 0.287Wto 0.43 (B.32) ( ) l d max = l (B.33) d max The fuselage lateral width, as opposed to d max which refers to the fuselage height, is assumed from existing similar aircraft to be, d width = 0.9d max. (B.34) Given the main aircraft geometry has been defined, a first order viability check can be performed by rapidly estimating the fuel tank volume and ensuring that the mass of fuel required for the mission can be stored. The fuel tank is assumed to lie integrated into the wing, inside a 3D tapered wing-box. The wing box is assumed to run from the aircraft centerline to the wing tip, from the bottom wing surface to the top surface, and in between the front and rear spars. The front and rear spars are assumed to be at 0.25 of the chord and 0.75 of the chord length respectively. Given the airfoil shape, z ( ) ( x c for the upper and lower surfaces, along the chordline x ) c, the volume of the wingbox is computed by, V tank = 2 ( ) b z ( x ) ( x ) d. (B.35) c c

157 The weight of the fuel required is then compared with W ftank = ρ f V tank, where ρ f = 0.68[kg/l] for kerosene, and a warning message appears if W f > W ftank. B.4.3 Outputs Note that wing parameters wither have a w subscript or no subscript. Parameter Description S k The area of component k, where k = {wing, HT, V T }, including the non-exposed area, from tip to the aircraft centerline b k The span of component k where k = {wing, HT, V T } c rootk The root chord length of component k where k = {wing, HT, V T }, and the root lies along the aircraft centerline c tipk The tip chord length of component k where k = {wing, HT, V T } c k The mean chord length of component k where k = {wing, HT, V T } ȳ k The lateral location from the aircraft centerline of the mean chord of component k where k = {wing, HT, V T } Λ LEk The leading edge sweep angle of component k where k = {wing, HT, V T } l The fuselage length d max The maximum fuselage diameter, which usually corresponds to the fuselage height d width The fuselage lateral or horizontal width V tank The volume of the fuel tanks B.5 Refined aircraft sizing Based on the calculated geometry, a more refined estimate of the aircraft weight can be computed using a similar mission fuel weight fraction method as the initial sizing function. Figure B-5: Block diagram of refined sizing function showing inputs and outputs. 157

158 B.5.1 Inputs Parameter r cruise r alternate t l oiter W to W p W c ( ) Wj 1 W j ( L D Description The cruise range of the aircraft The range to an alternate airport The time spent in loiter above an airport The maximum takeoff weight The payload weight The crew weight The weight fractions for mission segments between segment points j and j 1 )max The maximum lift-to-drag ratio of the aircraft SF C cruise The specific fuel consumption during cruise. This is approximated from existing engine data as a first estimate. SF C loiter The specific fuel consumption during loiter. This is approximated from existing engine data as a first estimate. q cruise The dynamic pressure during cruise q loiter The dynamic pressure during loiter C D0 The coefficient of zero lift drag. This is approximated from existing data. e Oswald s efficiency factor to account for 3D wing effects compared to ideal infinite wings. This is approximated from existing data. A Aspect ratio of the wing, the ratio of the square of the wing span to the wing area, approximated from legacy aircraft ( S ) The wing area, including non-exposed area under the fuselage T The thrust-to-weight ratio at given mission segment i ( W ) i W The wing loading for given mission segment i or at a specific operating condition such as S i stall B.5.2 Process The weight fractions for segment 1-3 and 8-10 given in?? remain the same. The weight fraction for climb is now given by, W 3 W 4 = M. (B.36) The cruise fuel weight fraction and the cruise to alternate fuel weight fraction are calculated in the following way, using the newly computed aircraft geometry for more accurate estimates. The equations B.37 and B.38 are used with cruise and cruise to alternate values in order to calculate the respective fuel weight fractions. ( ) L D cruise ( Wi 1 W i 1 = qc D0 + ( ) W ( W S ) S cruise cruise ) [ ] r cruise SF C cruise = exp ( v L ) cruise D Meanwhile for loiter B.39 and B.40 are used. ( ) L 1 = qc D D0 loiter + ( ) W ( W S ) S loiter 158 cruise loiter 1 qπae 1 qπae (B.37) (B.38) (B.39)

159 ( Wi 1 W i ) [ ] t loiter SF C loiter = exp ( L D) loiter (B.40) These equations follow from first principle evaluation of lift-to-drag ratio, and utilizing the Breguet equations. With these new weight fractions an iterative method is used to calculate the takeoff weight. An initial value for the takeoff weight is assumed, which is usually the value calculated from the initial sizing. In each iterative loop, the total mission fuel weight is calculated using, ( W f = r f 1 ( ) ) Wj 1 W to. (B.41) W j j Simultaneously the empty weight fraction, statistical formula taken from [98], ( We W to ) ( T = Wto 0.13 A 0.3 W ( ) W e W to ) 0.6 ( Wto is computed using the following S ) 0.05 M 0.05 cruise, (B.42) where ( ) T W is taken to be the maximum thrust-to-weight ratio calculated previously, A is the assumed aspect ratio, and ( W to ) S is the takeoff weight divided by the wing area. Finally W to is given by, ( ) We W to = W p + W f + W to, (B.43) and this loop is continued with the new W to as the initial assumption, until the difference between successive values is negligible. As can be seen S is used to calculate the takeoff weight but it itself was calculated from the previous takeoff weight calculation. Therefore this could be further iterated to converge on a better estimation of S. However as a first order approximation for the purposes of this analysis, since the optimization of S is not critical, this was not carried out.( try to do this ) W to B.5.3 Outputs Parameter W to W e W f ( ) Wj 1 W j Description The maximum takeoff weight The aircraft empty weight, the takeoff weight minus the fuel and payload weights The fuel weight The weight fractions for mission segments between segment points j and j 1 159

160 B.6 Landing gear tire sizing B.6.1 Function overview This function calculates the loads on each of the landing gear and selects the appropriate tires from a database in order to meet the aircraft requirements. B.6.2 Inputs Parameter x CG x CGL G m x CGL G n W to Tire database Description The x coordinate of the CG of the entire aircraft The x coordinate of the CG of the main LG The x coordinate of the CG of the nose LG The aircraft maximum takeoff weight A database of aircraft tires including details on loads, weights and geometries, from the Goodrich catalog B.6.3 Processes Firstly the loads on each of the landing gear in the tricycle configuration are calculated. It is assumed that the nose landing gear bears from 8-15% of the maximum load and the two main landing gear the remaining 85-92% [98]. The exact number is calculated in another function which positions the landing gear optimally, however assuming that the proportion of the load on the main landing gear is given by p m, the load on each main gear is calculated using, L main = 1 2 (p mw to ), (B.44) and the load on the nose landing gear is simply L nose = W to 2L main. Assuming that each landing gear comprises 2 wheels, the loadings are divided by 2 in order to obtain the loading per wheel. Using the loading per wheel a search is performed on the database for the smallest diameter tires which meet the loading requirement. Note that it is possible to carry this process out for the main tires and assume the nose tires to be anywhere between % the size of the main tire [98]. 160

161 B.6.4 Outputs Parameter d main d nose d rimm d rimn width main width nose W tm W tn L main L nose Description The diameter of the main LG tires The diameter of the nose LG tires The rim (inner) diameter of the main LG tire The rim (inner) diameter of the nose LG tire The width of the main LG tires The width of the nose LG tires The weight of the main landing gear tires The weight of the nose landing gear tires The load on each of the main landing gear during ground operations at maximum takeoff weight The load on the nose landing gear during ground operations at maximum takeoff weight B.7 Engine Analysis B.7.1 Function overview This function utilizes a physics-based method from first principle analysis of 1D engine flows, to model a two-spool turbofan engine. Hence the following equations have been derived from first principles, however the full derivations have been left out of this text. Figure B-6: Block diagram of turbofan engine analysis function showing inputs and outputs. 161

162 B.7.2 Inputs Parameter Description η c The efficiency of component c, where c = {diffuser, fan, compressur, burner, turbine, nozzle, fan nozzle} γ c The ratio of specific heat at constant pressure to specific heat at constant volume of air, for component c R The molar gas constant ( Q R ) The energy density of kerosene T The takeoff thrust-to-weight ratio requirement for the aircraft W to BP R The bypass ratio of the engine P r c The compressor pressure ratio P r f The fan pressure ratio T ti The turbine inlet temperature M The Mach number for which the analysis is to be performed T a Ambient temperature at the altitude of the analysis B.7.3 Process Given a specified level of advancement of engine technology, quantified in this case by the bypass ratio and the turbine inlet temperature, the engine is optimized for minimum specific fuel consumption while varying the pressure ratios of the compressor and the fan and meeting the thrust requirement. A full factorial evaluation of a specified discretized range of possible values for these pressure ratios is carried out with the minimum SF C design being selected. The assumed turbofan model is shown in Figure 3-9 in the main body. For each pair of P r c and P r f, each section of the engine is analyzed in the following way. Note that subscript s denotes the isentropic value which is used to take into account the component efficiencies for the real turbofan analysis [33] and subscript 0 denotes stagnation values for the given property. T c and P c denote the temperature and pressure respectively at a given point in the engine as defined by Figure 3-9. The required thrust per engine is given by, T = 1 ( ) T W to. (B.45) 2 W max In the scenario of an engine out during takeoff, the thrust requirements for a single engine is assumed to be double this value. Analysis from point a to point d, that is across the diffuser is as follows, ( ) ) γ 1 T 0d = T a (1 + M 2 (B.46) 2 T 0ds = T a + η d (T 0d T a ) P 0d P a = ( T0d T a ) γ d γ d 1 (B.47) (B.48) 162

163 P 0d = ( P0d P a ) P a. (B.49) The properties of the flow through the engine from d to f, i.e. across the fan, can now be calculated. P 0f = P 0d P r f (B.50) T 0fs T 0d = P r ( ) γf 1 γ f f (B.51) T 0fs = T 0fs T 0d T 0d T 0f = T 0d + 1 η f (T 0fs T 0d ) (B.52) (B.53) Consequently analysis across the compressor from station f to c is given by, P 0c = P 0f P r c (B.54) T 0cs T 0f = P r ( γc 1 γc ) c (B.55) T 0cs = T 0cs T 0f T 0f T 0c = T 0f + 1 η c (T 0cs T 0f ) (B.56) (B.57) Burner analysis is done to calculate the mass flow of fuel required to raise the temperature of the flow from the compressor exit as much as possible, limited by the turbine inlet temperature. The ratio of mass flow of fuel to mass flow of air in the engine core, that is f = ṁf ṁ c, is given by, f = C p η b Q R (T ti T 0c ). (B.58) The analysis from station b to station t, across the turbine is done using the following. The pressure at the burner exit is the same as the pressure at the burner inlet with some losses accounted for by r b, P 0b = P 0c r b. (B.59) Assuming the energy extracted by the turbine is used to drive the compressor and the fan, the following is derived from a simple energy balance equation T 0t = T 0b (T 0c T 0f ) BP R (T 0f T 0d ). (B.60) 163

164 Then the next few equations are used to give the flow properties at the turbine outlet. T 0ts = T 0b 1 η t (T 0b T 0t ) (B.61) P 0t P 0b = ( T0ts T 0b ) γ t γ t 1 (B.62) P 0t = P 0t P 0b P 0b (B.63) The final components to be analyzed are the core nozzle and the fan nozzle. The flow is expanded optimally and the velocity of the jet is calculated. For the core nozzle we have, [ ( ) )] 1 γn U e = P0t γn 2ηn C pn T 0t (1 (B.64) P a T 0t T e = ( P0t P a ) γn 1 γn (B.65) and for the fan nozzle we have, U ef = 2η fn T e = ( T0t T e ) 1 T 0t, (B.66) C pfn T 0f 1 ( P0f P a ) 1 γ fn γ fn (B.67) ( T 0f P0f = T ef P a ) γ fn 1 γ fn (B.68) T ef = ( T0f T ef ) 1 T 0f. (B.69) From this analysis the engine performance can now be calculated. The velocity of flow into the engine is calculated using U 0 = Ma, where a is the sped of sound at the altitude of the analysis. The specific thrust is given by, T sp = U e + BP R U ef (BP R + 1) U 0, (B.70) and the core and bypass mass flow are given by, and ṁ c = T T sp ṁ b = BP R ṁ c. 164 (B.71) (B.72)

165 The specific fuel consumption and engine efficiencies can now be calculated. The propulsive efficiency is defined as [33], Work done by thrust per unit time η prop = rate of production of the propellant kinetic energy T U 0 = ( (1 + f) U 2 e + BP R Uef 2 (1 + BP R) U ), 0 2 (B.73) and the thermal efficiency is defined as, η th = = Rate of addition of kinetic energy to propellant energy consumption rate ] ṁ c [(1 + f) U e 2 2 U ṁ f Q R (B.74) The overall engine efficiency is simply η o = η th η prop. (B.75) The final performance parameter to be calculated is the specific fuel consumption which is given by, SF C = f (ṁ c + ṁ f ). (B.76) T The last step of the turbofan analysis is to compute the geometries, namely the nozzle diameters and the inlet diameter. The nozzle areas are calculated using, A e = ṁc ρ e T e, (B.77) and A ef = ṁf ρ ef T ef, (B.78) where ρ e = Pa R nt e and ρ ef = Pa R fn T ef, are the densities of the core jet and bypass flow jet respectively. Assuming the turbofan is cylindrical, the diameter of the core nozzle is computed using, Ae d e = 2 π. (B.79) The fan nozzle diameter is more complicated to calculate and not necessary for the pruposes of this analysis. The inlet on the other hand is the largest diameter of the turbofan and dictates the nacelle size. Assuming a Mach number at the inlet of as described in [121], the inlet geometry is calculated with, A i = ṁc + ṁ b M i a i ρ i. (B.80) 165

166 d i = 2 Ai π. (B.81) Finally the nacelle diameter is assumed from legacy aircraft to be equal to 10% more than the inlet diameter (estimated using database of engines and [74]). For the purposes of this thesis this analysis is carried out at static sea-level conditions, that is M = 0 and altitude = 0, for the takeoff thrust-to-weight ratio. This is selected since it tends to be the limiting condition of the engine geometry, that is it determines the maximum size of the engine diameter. The calculated SF C is that at takeoff conditions, therefore the cruise SF C is calculated using SF C to = 0.6SF C cruise, which is a formula obtained from a statistical analysis of engine data. B.7.4 Outputs Parameter d nac SF C to SF C cruise P r cmin P r fmin P r o η prop η th η o Description The diameter of the nacelle of the engine The specific fuel consumption at takeoff The specific fuel consumption during cruise The compressor pressure ratio at the minimum SF C The fan pressure ratio at the minimum SF C The overall pressure ratio, the product of the fan and compressor pressure ratios The engine propulsive efficiency (defined above) The engine thermal efficiency (defined above) The overall engine efficiency (defined above) B.8 Aerodynamic calculations This function uses the method defined in Raymer [98] to calculate the lift and drag properties of an aircraft with a dominant design architecture. Figure B-7: Block diagram of lift calculation function showing inputs and outputs. 166

167 Figure B-8: Block diagram of drag calculation function showing inputs and outputs. B.8.1 Inputs Parameter C lmax α 0l ( t c ) max ( x ) c max,t l d max l eng A c Λ ( c 4 ) c c c S c e M i v i ρ i k Description The maximum lift coefficient of the airfoil The zero lift angle of attack of the airfoil The maximum thickness-to-chord ratio of the airfoil for the wing, horizontal tail and vertical tail The chordwise location of the maximum airfoil thickness for wing, horizontal tail and vertical tail The fuselage length The fuselage maximum diameter The engine length, this is assumed from empirical data The aspect ratio of component c, where c = {wing, horizontal tail, vertical tail} The quarter chord sweep angle of component c The mean chord for component c The area of component c where c = {wing, horizontal tail, vertical tail} Oswald s efficiency factor to account for 3D wing effects compared to ideal infinite wings. This is approximated from existing data. The Mach number for mission segment i The velocity for mission segment i The density of air for mission segment i The surface roughness B.8.2 Process Lift calculations The first order approximation of lift of an aircraft is characterized by the graph of lift versus angle of attack. For this several components are required, the maximum lift, the lift curve slope and the zero-lift angle of attack for the clean case (without high lift devices) and the case with high lift devices. To calculate the curve of this slope, the wing effective aspect ratio is required, which takes into account wing tip devices, that is A eff = 1.2A. The lift curve slope 167

168 for the subsonic regime is given by, C Lα = 1 + 2πA eff ( ) 4 + A2 eff β2 1 + tan2 Λ max,t η 2 β 2 ( Sexposed S ) F, (B.82) where F = 1.07 ( ) 2, 1 + dmax b β 2 = 1 Mcruise 2 and η = C lα. The lift curve slope of 2π\β the airfoil, C lα is obtained from data of the selected airfoil, and Λ max,t is the angle of sweep at the location of maximum thickness of the wing, which is calculated using the location of maximum thickness along the chord of the selected airfoil. One will note that for this calculation the value of S exposed is required; this is the area of the wing or tail which is not covered by the fuselage and is calculated using the wing, tail and fuselage geometries. The maximum lift coefficient, C Lmax, and the angle of attack at the maximum lift coefficient, α CLmax are calculated using, C Lmax = 0.9C lmax cos ( ) Λ c (B.83) 4 and α CLmax = C L max C Lα + α 0L + α CLmax. (B.84) Note that the wing zero lift angle of attack is approximated to be the same as the airfoil zero lift angle of attack, ie α 0L α 0l. The incremental angle of attack at the maximum lift, α CLmax is taken from a table lookup derived from Figure in Raymer [98]. With all this information the entire lift profile at various angles of attack for the aircraft is known. The same procedure is carried out to determine the lift of the horizontal tail for longitudinal stability calculations and the lift produced by the vertical tail for yaw stability and engine out safety (see later functions). The lift with high lift devices must also be calculated. Assuming a high-lift configuration, which in this case is assumed to be single slotted flaps and slats as is common with most short-medium range airliners, the incremental lift and angle of attack are given by, ( ) Sflapped C Lmax = 0.9 C lmax cos (Λ HL ) (B.85) S and ( ) Sflapped α 0L = ( α 0l ) airfoil cos (Λ HL ), S (B.86) where S flapped is the area of the wing ( in between ) the start and end points of the flap. Sflapped For the purposes of this analysis is assumed to be C S lmax, the change in maximum lift coefficient of the airfoil due to high lift devices, is assumed based on statistical data from Raymer [98] and Roskam [100], and ( α 0l ) is taken to be 10 for takeoff settings, and 15 for landing, from the same source. These increments are added to the clean values calculated previously to give the high lift values for landing and takeoff. 168

169 The values of C Lmax and C LmaxHL which were previously estimated based on legacy aircraft, are now updated to reflect these newly calculated values. Drag calculations Drag can be broken down into segments based on the cause. The two major contributors are Parasitic or zero-lift drag, C D0, and Lift-induced drag, C Di. Parasitic drag is composed of skin friction drag, pressure drag and wave drag, and lift-induced drag is composed of the component of lift which contributes to drag. In this method parasitic drag is estimated via the component drag buildup method and lift induced drag from a physics based approximation. The component drag buildup method revolves around the following equation, c (C D0 ) subsonic = (C fcf F c Q c S wetc ) + C Dmisc + C DL&P, (B.87) S ref where subscript c denotes the aircraft components, that is c = {wing, fuselage, HT, VT, nacelle }, C f denotes the skin friction drag, F F denotes the form factor, Q denotes the interference factor, and S wet denotes the wetted area. Meanwhile C DL&P is the leakage and protuberance drag and C Dmisc accounts for other miscellaneous forms of drag. Each of these is explained further below and in Raymer [98]. The skin friction coefficients are calculated in the following way, assuming fully turbulent flow across all components, which is realistic given the current state of aircraft technology (this can be modified to include technology upgrades such as the laminar wing). For this, the Reynolds number and a Reynolds cutoff number for each component must be computed utilizing, R = ρvlc and R µ cutoff = ( l ck ) M 1.16, where the parameters are taken at cruise conditions. In this formulation l c is the characteristic length (see Raymer) of the component, k is the surface roughness of the component (taken to be ) m) and µ is the kinematic viscosity of air at cruise conditions. This can be done for all mission segments but for the purposes of this thesis it is done solely for cruise conditions since the final aircraft performance will be done for this mission segment which accounts for the overwhelming majority of the contribution to the performance metric. Given these two values for R the lower of the two is used in the following formula, for skin friction drag, for each component, C fturbulent = (log 10 R) 2.58 ( M 2 ) (B.88) The form factor estimates the pressure drag due to viscous separation. For the wing and tail we have, [ F F wing,ht,v T = 1 + ( 0.6 x ) c max,t ( ) t c max ( ) ] 4 t [1.34M 0.18 (cosλ 0.28] max,t), c max (B.89) where ( ) x is the chordwise location of the airfoil maximum thickness, ( ) t the c max,t c max 169

170 maximum thickness-to-chord ratio of the airfoil and Λ max,t is the sweep angle of the maximum thickness line of the wing, HT or VT. For the fuselage, and the nacelle, F F fus = f + f 3 400, (B.90) F F fus = f, (B.91) where f = lc d, l c being the length of the components and d the maximum diameter. The interference factors are based on statistical data as a first order approximation, given in the table below. Component Description Value Q nac Interference between nacelle and fuselage, wing Q w Interference between low-wing and fuselage 1.0 Q HT,V T Interference between horizontal tail, vertical tail and fuselage 1.05 Q fus Fuselage interference factor 1.0 The wetted areas of the components are simply the areas which are exposed to the flow of air. These are approximated in a separate fucntion based on the geometries calculated previously and several simplifying assumptions. The miscellaneous drag is assumed to be zero, since the landing gear are retractable and there are no other miscellaneous sources of drag. The leakage and protuberance drag component is estimated by multiplying the other components in the main equation by 2 5%. Wave drag is assumed to be negligible for such an airliner since it is operating at Mach numbers below the drag divergence Mach number. The lift-induced drag for cruise is calculated using, C Di = KC 2 L, (B.92) where K = 1. This equation is a the physical approximation based on classical πae wing theory. For takeoff and landing conditions, the incremental drag is calculated using, C Di = kf 2 ( C LHL ) 2 cos ( ) Λ c 4, (B.93) where k f = for high-lift devices approximately spanning 75% of the wing. The major drag components of the aircraft have been calculated and now the total aircraft drag during cruise and takeoff are given by summing these components, C D = C D0 + C Di. (B.94) 170

171 B.8.3 Outputs Parameter C D C D0 C Di C Lmax C LmaxHL C Lα C LmaxHT,V T C LαHT,V T α CLmax α 0L Description The total aircraft drag coefficient The aircraft zero-lift drag coefficient The aircraft lift-induced drag coefficient The maximum lift coefficient of the wing (clean) The maximum lift coefficient of the wing (with high-lift devices) The lift curve slope of the wing The maximum lift coefficient of the horizontal tail, vertical tail The lift curve slope of the the horizontal tail, vertical tail The angle of attack of the wing for maximum lift The zero-lift angle of attack of the wing B.9 Component weight calculation B.9.1 Function overview This function utilizes a statistical method based on that in Roskam to determine the weight of each of the components [100]. B.9.2 Inputs Parameter W to Description The aircraft maximum takeoff weight B.9.3 Processes For the purposes of this analysis, the component weights are based on a statistical method for airliners of a similar size, such as the Boeing 737. These are simply based on a proportion of the maximum takeoff weight using the following weightings, Component (c) Weight (w ic ) Aircraft Structure 0.27 Engine Fixed equipment Wing Fuselage Empennage Landing gear 0.05 Nacelle These are used as first order approximations for component weights, using, W c = w ic W to. (B.95) 171

172 Figure B-9: Block diagram of center of gravity calculations showing inputs and outputs. The horizontal and vertical tail weights are estimated based on the empennage weight and the areas of the respective components, where W HT = S HT S HT +S V T W emp and W V T = S V T S HT +S V T W emp. Furthermore the total landing gear weight is split into the nose and main landing gear weights in the proportion 1 : 9 respectively [98]. These are not extremely accurate estimates, however for the purposes of this high-level model these approximations are sufficient. Further in depth analysis of the weights of the engine and landing gear are carried out later in the model. B.9.4 Outputs Parameter Description W c The weight of aircraft component c, where c = {wing, fuselage, horizontal tail, vertical tail, nose landing gear, main landing gear, engine, nacelle} B.10 Center of gravity calculation B.10.1 Function overview This function is used to calculate the center of gravity of the aircraft based on positions of the center of gravity of each of the components. This function is used throughout the airframe-engine model, and where positions of components are not yet fixed assumptions for their respective CG are made. 172

173 B.10.2 Inputs Parameter Description W c The weight of aircraft component c, where c = {wing, fuselage, horizontal tail, vertical tail, nose landing gear, main landing gear, engine, nacelle} x LE The x-location of the leading edge of the wing at the aircraft centerline c c The mean aerodynamic chord of aircraft component c, where c = {wing, horizontal tail, vertical tail} ȳ c The y-location of the mean aerodynamic chord of aircraft component c, where c = {wing, horizontal tail, vertical tail} Λ LEc The leading edge sweep angle of aircraft component c, where c = {wing, horizontal tail, vertical tail} l The length of the fuselage d max The maximum fuselage diameter d width The lateral fuselage diameter c rootc The chord length of the root of component c, where c = {wing, horizontal tail, vertical tail} c tipc The chord length of the tip of component c, where c = {wing, horizontal tail, vertical tail} b c The span of component c, where c = {wing, horizontal tail, vertical tail} d engine The engine diameter l engine The engine length y engine The lateral location of the engine CG. This is originally assumed based on existing aircraft, then calculated during geometry optimization width LGm The main landing gear tire width width LGn The nose landing gear tire width d LGm The main landing gear tire diameter d LGn The nose landing gear tire diameter ( δ w The wing dihedral angle t ) The maximum thickness-to-chord ratio of the selected airfoil for the wing c max B.10.3 Processes The aircraft axis used in this analysis has its origin at the aircraft nose on the centerline, at the bottom of the cylindrical-like fuselage. The x-direction runs along the fuselage length, the y-direction runs laterally outwards in the direction of the wing tip, and the z-direction vertically upwards. Since the y-axis runs perpendicular to the centerline and in the same plane as the plan of symmetry the y-coordinate of the center of gravity is zero, on the aircraft centerline. The x-coordinate and z-coordinate will be calculated separately, using the method presented below. Longitudinal center of gravity The horizontal and vertical tail are assumed to be as far back along the fuselage as possible, in order to maximize the moment arm about the aircraft center of gravity. The center of gravity of these components is assumed to be 40% of the way along their respective mean aerodynamic chord [98], that is for the horizontal tail, and the vertical tail, x CGHT = l c rootht + ȳ HT tan (Λ LEHT ) c HT (B.96) x CGV T = l c rootv T + ȳ V T tan (Λ LEV T ) c V T. (B.97) 173

174 The wing center of gravity is also assumed to be 40% of the way along the mean aerodynamic chord, given by, x CGwing = x LE + ȳtan (Λ LE ) c, (B.98) where the leading edge of the wing x LE, is either assumed based on statistical values if it has not been yet optimized, or it is taken as the optimized value. The x-coordinate of the CG of the fuselage is assumed to be 45-50% of the length of the fuselage from the nose [100], that is, x CGfus = 0.45l (B.99) The engine center of gravity is given by, x CGengine = x LE + l engine, (B.100) 2 assuming the center of gravity of the engine to be half the way along the length and assuming the engine leading edge to be aligned with the location where the wing leading edge meets the aircraft centerline. The latter assumption is based on empirical analysis of existing aircraft. These assumptions are made before the engine location is optimized in the optimization function; once optimized the new location is used in lieu of these assumptions. Prior to geometry optimization, the center of gravity of the main landing gear is assumed based on the location of the engine (and wing), and the nose landing gear is assumed based on statistical aircraft values to be l. The main landing gear CG is assumed to be a specified distance inboard of the engine and close to the wing trailing edge for stowage pruposes. This is given by, [ x CGLGm = x LE + y engine d ] Engine width LGm tan (Λ LE ) + c wlg (B.101) 2 where the factor of 0.3m assumes a buffer region between the outermost LG tire and the innermost location of the engine. This formulation assumes a main landing gear design of two tires in parallel, of width width LGm and a space of approximately equal to this width separating the two tires. The parameter c wlg is the calculated chord length at the y-position of the assumed landing gear location. Finally the center of gravity of the entire aircraft is calculated simply using, c x CG = W cx CGc c W, (B.102) c for each component c. Note that c W c W e since the former term accounts only for aircraft structure, landing gear and the engines, whereas the second term includes aircraft systems such as hydraulics and fixed equipment such as cabin furnishings. This is worth noting, since the calculation of center of gravity may seem very approximate; however in reality this is difficult to predict at such an early stage and can 174

175 actually be positioned with much flexibility in later design stages. Vertical center of gravity Similar to before the center of gravity in the z direction can be calculated as follows. The z-coordinate of the CG of the fuselage is assumed to be half way up the height, that is, z CGfus = 1 2 d max. (B.103) The horizontal tail center of gravity is approximately at the fuselage height, therefore z CGHT d max, whereas the vertical tail CG is estimated to be at the location of the mean aerodynamic chord, that is, z CGV T = ȳ V T + d max. (B.104) The wing CG is approximately at the location of the mean aerodynamic chord in the middle of the wing thickness, which is given by, z CGwing = ȳtan (δ w ) + 1 ( ) t c. (B.105) 2 c Meanwhile the engine CG, assumed to be in the center of a cylindrical-shaped nacelle, is given by, z CGengine = y engine tan (δ w ) d engine, (B.106) 2 where it can be seen that the engine uppermost surface is assumed to be at the location of the wing lower surface, which is a realistic assumption based on existing aircraft. To calculate the z-location of the CG of the landing gear, the landing gear tipping constraint for stable takeoff conditions and the loading proportion constraint, are used [100]. The former constraint is the angle subtended by the vertical from the x-location of the main landing gear, and the line from this point to the location of the aircraft enter of gravity, θ tip, should be between 15 and 20 degrees [100]. The latter is that the weight should be distributed among the nose and main landing gear, whereby the main landing gear bears between 85-92% of the total weight. This is given by, max z CGLG = z CG x CG LGm x CG tanθ tip. (B.107) It can be seen that this requires knowing the z location of the entire aircraft CG. To solve this issue, this is substituted into the main equation for calculating the center of gravity, resulting in the following equation, 1 z CG = x CGLGm xcg (W c z CGc ) W LG (B.108) c c LG (W c ) w LG tan (θ c tip ) c LG 175

176 B.10.4 Outputs Parameter Description x CG The x-coordinate of the center of gravity of the entire aircraft z CG The z-coordinate of the center of gravity of the entire aircraft x CGc The x-coordinate of the center of gravity of aircraft component c, where c = {wing, fuselage, horizontal tail, vertical tail, nose landing gear, main landing gear, engine, nacelle} z CGc The z-coordinate of the center of gravity of aircraft component c, where c = {wing, fuselage, horizontal tail, vertical tail, nose landing gear, main landing gear, engine, nacelle} B.11 Longitudinal static stability calculation B.11.1 Function overview This function uses longitudinal static stability criteria to position the aircraft components, with several assumptions being made. The goal of longitudinal static stability is to ensure that the center of gravity (CG) of the aircraft is between 0.1 c and 0.2 c ahead of the neutral point (NP) of the aircraft. The neutral point is the location in which the aircraft is neutrally stable, that is, if the center of gravity were to move further aft the aircraft would become statically unstable. This result comes from the fact that the horizontal tail produces a negative lift which tends to create a stabilizing moment about the aircraft center of gravity. The horizontal and vertical tail are assumed to be as far back along the fuselage as possible in order to maximize the moment arm around the aircraft center of gravity, therefore their location is fixed along the fuselage. As a first approximation the location of the nose and main landing gear are assumed to be a certain proportion along the length of the fuselage according to empirical data [100]. The nacelles and engines are assumed to be fixed a certain distance in front of the leading edge of the wing, based on statistical data as a first approximation. The neutral point location is dependent on the aerodynamic centers of the wing and the horizontal tail. Therefore it is clear that by moving the wing and nacelles, the neutral point and the center of gravity both change location. Given this, an iterative method is utilized, whereby the wing and nacelle location are changed, and the NP and CG are computed until the aforementioned stability criteria is met. 176

177 Figure B-10: Block diagram of longitudinal static stability calculation function showing inputs and outputs. B.11.2 Inputs Parameter Description x CG The x-location of the center of gravity of the entire aircraft x CGc The x-location of the center of gravity for each component c, where c = {wing, nacelle, fuselage, HT, VT, nose LG, main LG} x ACc The aerodynamic center of component c, where c = {wing, HT} C Lαc The lift curve slope of component c, where c = {wing, HT} x LE The x-location of the leading edge of the wing c c The mean aerodynamic chord length of component c, where c = {wing, HT} ȳ c The spanwise location (y-location) of the mean aerodynamic chord of component c, where c = {wing, HT} S The wing planform area S HT The horizontal tail planform area B.11.3 Processes As mentioned previously, an iterative method is used to locate the wing and nacelles until the longitudinal static stability criterion is satisfied [100], that is, 0.1 c x NP x CG 0.2 c. (B.109) Assuming an initial wing location, the center of gravity of the aircraft is calculated using the aforementioned function. This is then used to calculate the neutral point of the aircraft in the following way. The coefficient of moment about the center of gravity of the aircraft is given by, C mcg = C L ( XCG X ACw ) +Cmw +C mfus η HT S HT S C L HT ( XCG X ACHT ), (B.110) ignoring the effects of flap deflections, and engine thrust offset effects. In this formulation a bar above the variable indicates that it has been expressed as a fraction of 177

178 the wing mean aerodynamic chord for simplicity, e.g. XCG = x CG c. Meanwhile C m denotes the coefficient of moment for various components of the aircraft, and η HT is the ratio of the dynamic pressure at the tail to the free stream dynamic pressure, assumed to be 0.9 as a typical value [98]. For static stability any change in the angle of attack, α, should make the aircraft tend back to its original position - mathematically C mα Cm CG is negative. Therefore differentiating the above equation with respect to α, setting C mα equal to zero α and solving for X CG, gives the location of the center of gravity which corresponds to the neutrally stable point, or the NP. Therefore the equation for the NP is given by, X NP = C S L α XACw C mfus + η HT HT S C Lα + η HT S HT S C L αht α HT α C L αht α HT α X ACHT. (B.111) The iteration of changing x LE causes X ACw and x CG to change until the static margin, SM = x NP x CG reaches the convergence criteria. In this equation C mfus = 0 since in cruise the angle of attack of the fuselage is assumed to be zero so it generates no lift. The rate of change of the angle of attack at the horizontal tail with respect to the angle of attack at the wing, α HT is given by, α where ɛ α 2C Lα πa. α HT α = 1 ɛ α (B.112) B.11.4 Outputs Parameter x CG x NP x LE x CGw x CGnac x ACw Description The x-location of the center of gravity of the entire aircraft The x-location of the neutral point of the aircraft The x-location of the leading edge of the wing, at the centerline of the aircraft The x-location of the center of gravity for the wing The x-location of the center of gravity for the nacelle The aerodynamic center of the wing B.12 Engine and landing gear optimization B.12.1 Function overview This function optimizes the location and geometry of the landing gear and the location of the engine, which are tightly coupled. The objective of this optimization is to maximize the aircraft performance which directly translates to a minimization of the weight of the landing gear. The optimization is done by discretization of the space under the aircraft wing into possible locations for the engine and landing gear CG, evaluating possible combinations for aircraft performance, and selecting the one which maximizes this performance. The space under the wing is constrained by 178

179 ground clearance constraints, landing gear retraction constraints and aircraft loading and stability constraints. Meanwhile a proxy for the performance is the landing gear weight (drag increases are minimal due to retractable LG), which is a function of the landing gear length, and the engine drag, mainly quantified by skin friction and interference drag. Figure B-11: Block diagram of landing gear and engine optimization function showing inputs and outputs. B.12.2 Inputs Parameter Description ( W to ) The maximum takeoff weight of the aircraft T The takeoff thrust-to-weight ratio W to d engine The maximum diameter of the engine nacelle d width The lateral diameter of the fuselage x CG The x-location of the center of gravity of the aircraft z CG The z-location of the center of gravity of the aircraft x CGc The x-location of the center of gravity of component c, where c = {wing, HT, VT, fuselge, nose LG, main LG, engine} x LE The x-location of the leading edge of the wing at the aircraft centerline c c The mean aerodynamic chord of component c, where c = {wing, HT, VT} c root The chord length of the root of the wing c tip The chord length of the tip of the wing ȳ c The y-location of the mean aerodynamic chord of component c, where c = {wing, HT, VT} Λ LEc The leading edge sweep angle of component c, where c = {wing, HT, VT} b The wing span of the aircraft S c The planform area of component c, where c = {wing, HT, VT} S exposedc The exposed planform area of component c, where c = {wing, HT, VT} d main The diameter of the main landing gear tire d nose The diameter of the nose landing gear tire M to The takeoff Mach number q to The dynamic pressure at takeoff conditions Λ c 4 V T The quarter chord sweep angle of the vertical tail C lmaxv T The maximum coefficient of lift for the airfoil selected for the vertical tail L main The load on each of the main landing gear experienced at the maximum takeoff weight L nose The load on the nose landing gear experienced at the maximum takeoff weight 179

180 B.12.3 Processes The objective of this function is to maximize the aircraft performance by minimizing the weight of the landing gear and hence its length subject to the main landing gear CG location p LG = (x LG, y LG, z LG ) and the engine CG location p eng = (x eng, y eng, z eng ). This can be roughly formulated in the following way, min p LG s.t. l LG ground clearance engine location stowage aircraft safety (B.113) where the constraints are mathematically formulated and explained in more detail below. The first step in this process is to discretize the space under the wing into a cloud of points. The space in the y direction, is discretized into n y points from the edge of the fuselage to the wing tip, giving points y j where, y j = b 2 d width n y. (B.114) Similarly the x space is discretized into n x points along the length of the chord at each spanwise y-location, x ij = x LE j x T Ej n x, (B.115) where x LEj corresponds to the x-coordinate of the leading edge of the wing at spanwise location y j. Although the engine protrudes beyond the leading edge there is no need to include this space since the main focus is on constructing constraints for the interactions between the engine, landing gear and wing. The z space is treated as continuous for this analysis since it is the most important in terms of minimizing landing gear length. In order to bound the optimization space, two engine location constraints are used. The first involves calculating the furthest outboard location possible for the engine, in the event of the limiting situation of an engine out during takeoff. Physically, losing thrust from one engine during takeoff requires the other engine to produce enough thrust to enable the aircraft to takeoff and land safely. Meanwhile the vertical tail must produce enough of a moment about the center of gravity to counteract the moment produced by the single engine. Therefore the maximum lift produced by the vertical tail is calculated, which bounds the outermost position of the engine. The maximum vertical tail lift is calculated in the same way as previously described in the aerodynamics function, using ( ) C LmaxV T = 0.9C lmaxv T cos Λ c, (B.116) 4 V T 180

181 and hence lift at takeoff is given by, L maxv T = C LmaxV T q to. (B.117) Given this the maximum outboard position of the engine is given by, y engmax = (x AC V T x CG ) L ( maxv T T ), (B.118) W to W to which bounds the engine location. The second engine constraint bounds the engine location on the inboard part of the wing. This constraint pertains to interference drag between the engine and the fuselage. According to [73] the interference drag becomes negligible when the engine is approximately 2 nacelle diameters away from the fuselage, therefore, y engmin = d width + 2d engine. (B.119) 2 The optimization occurs iteratively by evaluating the minimum length meeting all constraints required for the landing gear at each potential engine location. The potential engine locations include all positions along the span of the wing, y j in between the minimum and maximum y-locations calculated previously. This corresponds to the position of the engine CG. The x-position of the engine is not a constraint since it is assumed that the engine is place in front of the wing in line with the point where the wing leading edge meets the centerline. Meanwhile, the z-location of the engine is assumed to be half of the engine diameter below the wing lower surface. For each of these engine locations the LG constraints are as follows. Firstly the landing gear must be located on the inboard side of the engine in order to meet ground clearance constraints and enable retraction and stowage in the wing and fuselage, or mathematically, for each engine location j, the set of potential landing gear locations i is given by, {y LGi } < y engj d engine. 2 The second constraint is a ground clearance constraint stating that an angle of 5 between the ground and the underside of the engine subtended where the landing gear meets the ground is required in addition to a 6 inch buffer. Therefore an exclusion zone for potential landing gear positions is created by defining a boundary from each potential engine location corresponding to this condition, above which any landing gear location is ruled out. Mathematically this is formulated for potential landing gear locations i at each engine location, as [ z LGi tan(5)y LGi + z engj 1 ] 2 d ( ) engine tan(5)y engj. (B.120) cos(5) (include figures!!) The next step involves constraining the set of potential landing gear by checking whether it is possible for the tires to retract into the fuselage, that is the landing gear must be long enough for this however short enough to not cross the aircraft centerline. This is not as simple as it may seem since the landing gear includes an oleo-pneumatic strut (OLEO), which is essentially a shock absorber used to dampen 181

182 the load on landing. Associated with OLEOs is a stroke which must be taken into account when considering the retraction of the landing gear. The stroke length is calculated using the following method [98]. The energy required to be absorbed on aircraft landing is given by, E l = 1 2 W tov 2 sink, (B.121) and the stroke deflection is given by, [ 1 St = 1.1 η sh ( E 2P m l f )]. (B.122) In this context, V sink is the maximum sink speed usually taken to be 10 ft/s, η sh is th shock efficiency which is usually 0.8 and l f is the load factor taken to be 1.5 usually. Meanwhile P m is the maximum static load on the gear which is taken to be L main multiplied by an FAR safety factor of The gear length for retraction is therefore taken to be the gear length during maximum deflection plus the stroke length. The gear length during maximum deflection is given by, l LGm = {y LGi tan(δ w ) [( t c ) c yi ]} {z LGi St}, (B.123) max where the first term represents the wing hinge point of the landing gear, assumed to be in the middle of the wing, and c yi is the chord length at spanwise location y LGi. The second term is the location of the bottom of the landing gear, which is subtracted due to the coordinate system (the z-location of the landing gear bottom is always negative). Given the length, the retraction constraint is given by, y LGi d width d main 2 l LGm y LGi d main 2. (B.124) The next constraint is for the tipping angle of the aircraft, that is the angle subtended by the vertical from the landing gear and the line from the landing gear running through the aircraft CG, should be [100]. This assumes the minimum possible landing gear length for each potential spanwise location of the landing gear, for a given engine location, as given by the above constraints. At each possible landing gear y-location and z-location the set of x-locations within the confines of the discretization are analyzed to see whether this constraint is met, resulting in a set of feasible locations in 3D space for the landing gear. Mathematically this constraint can be formulated as, z CG ( ) z LGi + tan (90 15) x LGij x LGi z CG ( ) z LGi + tan (90 25) x LGij. tan (90 15) tan (90 25) (B.125) This essentially extends boundaries from each landing gear location corresponding to the 15 and 25 constraints and if the aircraft CG lies within this boundary, the 182

183 location is feasible. The next constraint pertains to the aircraft overturn criteria, which is the tendency to overturn when taxiing around a sharp corner. In order for safe ground operations without the aircraft overturning, a minimum track length, the distance between the main landing gears, is required. This is measured by the angle from the CG to the main wheel when the main wheel and nose wheel are aligned, viewing from behind the aircraft. This angle must be less than 63 [100], mathematically formulated as, tan 1 z LGi z CG ( 63 (x CG x LGn ) sin (tan (B.126) 1 y LGi x LGi xlg n )) where x LGn is the x-location of the CG of the nose landing gear, and subscript i denotes a potential main landing gear location. Finally the load distribution of each potential landing gear meeting the above constraints is checked. According to [98], the nose landing gear should bear between 8-15% of the total load experienced during ground operations in order to have enough load to be able to steer while not being over-loaded. For each feasible main landing gear, this is checked using, 0.85 x CG x LGn x LGi x LGn (B.127) In instances where this criteria is not met, the nose landing gear location constraint is relaxed and adjusted in order to meet this criteria provided no other constraints are violated. The landing gear length is subsequently calculated for each of the potential locations in the discretized space which fall within the boundaries set by the above constraints. For each potential engine location, a set of feasible main landing gear locations and lengths have been enumerated. Using the minimization described in B.113, the optimized engine-landing gear according to the above process is found. It is worth noting that the overall takeoff weight will change, therefore wing weight and engine parameters will change as a result. With this in mind the engine-landing gear optimization function is included in an iterative loop, to converge on the overall takeoff weight and ultimately an optimized solution. 183

184 B.12.4 Outputs Parameter x CGeng y CGeng z CGeng x CGLGm y CGLGm z CGLGm x CGLGn y CGLGn z CGLGn l LGm l LGn Description The x-coordinate of the location of the engine CG The y-coordinate of the location of the engine CG The z-coordinate of the location of the engine CG The x-coordinate of the location of the main LG CG The y-coordinate of the location of the main LG CG The z-coordinate of the location of the main LG CG The x-coordinate of the location of the nose LG CG The y-coordinate of the location of the nose LG CG The z-coordinate of the location of the nose LG CG The length of the main landing gear The length of the nose landing gear B.13 Landing gear weight calculation B.13.1 Function overview This function calculates the weight of the landing gear with a first principles structural analysis approach based on simple models of the main and nose landing gear. It utilizes the worst-case loading scenario for the possible modes of failure including buckling, yielding and fatigue. The landing gear is idealized as a main strut with a single side strut for the main gear, with a back strut replacing the side strut in the nose gear, as can be seen in Figure B-12. For the yield and fatigue analysis the idealization is of a beam loaded at one end and pin-jointed at the other with constraints on its displacement due to the side strut as in Figure B-13. For the buckling analysis the landing gear is assumed to be a constrained beam, pin-jointed at one end and loaded at the other end. The main strut and side strut are assumed to be hollow cylinders, and the diameter of the side strut is assumed to be a certain percentage of the diameter of the main strut to simplify the analysis. 184

185 Figure B-12: Model of the (a) main landing gear and (b) nose landing gear used for this analysis. B.13.2 Inputs Parameter x CGLGm y CGLGm z CGLGm x CGLGn y CGLGn z CGLGn l LGm l LGn ρ 300M ρ Al6061 E 300M E Al6061 σ y300m σ yal6061 L main L nose W to d main d nose d rimm d rimn width main width nose W tm W tn St Description The x-coordinate of the location of the main LG CG The y-coordinate of the location of the main LG CG The z-coordinate of the location of the main LG CG The x-coordinate of the location of the nose LG CG The y-coordinate of the location of the nose LG CG The z-coordinate of the location of the nose LG CG The length of the main landing gear The length of the nose landing gear The density of 300M steel The density of 6061 aluminium The elastic modulus of 300M steel The elastic modulus of 6061 aluminium The yield strength of 300M steel The yield strength of 6061 aluminium The load on the main landing gear during ground operations at maximum takeoff weight The load on the nose landing gear during ground operations at maximum takeoff weight The maximum takeoff weight The diameter of a main landing gear tire The diameter of a nose landing gear tire The diameter of a main landing gear rim The diameter of a nose landing gear rim The width of a main landing gear tire The width of a nose landing gear tire The weight of a main landing gear tire The weight of a nose landing gear tire Stroke length of the main gear OLEO B.13.3 Processes Yield analysis Using a first principles structural analysis approach the normal forces (N ab and N bc ), shear forces (V ab and V bc ) and bending moments (M ab and M bc ) in the beam are cal- 185

186 Figure B-13: Idealization of the landing gear for structural analysis in (a) yield failure and (b) buckling failure. culated. For the main landing gear, these are calculated using the worst-case loading scenario, arising from a hard landing on a single strut at the maximum takeoff weight. Assuming a maximum aircraft sink speed V sink = 10ft\s and constant deceleration of the OLEO strut over the stroke length St to absorb the kinetic energy of the aircraft, a first-principles approach yields the following equation for the maximum dynamic loading on a single strut, sink L d1 = W to [ g + V 2 2St Meanwhile the dynamic loading of the nose gear is given by [80], ]. (B.128) L dn = (x CG m x CGn ) W to (z CG z CGm ). (B.129) (x CGm x CGn ) + (z CG z CGm ) The internal forces and moments are used to calculate the Von Mises stress, σ vm, in the beam which is used to size the strut according to the design criterion [80], σ y σ vm 1.5. (B.130) In this equation σ y corresponds to the yield stress of the chosen material which for the landing gear struts tends to be 300M steel. The Von Mises stress is calculated using, σ vm = τzz 2 + 3τzx 2 (B.131) 186

187 where τ zz is the normal stress given by, τ zz = N k A k + M k I yy x (B.132) and τ zx is the shear stress given by, τ zx = 4V k 3A k [ d 2 o + d o d i + d 2 i d 2 o + d 2 i ]. (B.133) In these equations subscript k refers to the location along the beam, that is either ab or bc, d i and d o correspond to the inner and outer diameters of the cylinder respectively, I yy = π(d4 o d4 i) is the moment of inertia of a cylinder, and A = π(d2 o d2 i) is the area of 64 4 the cylinder. The inner diameter of the cylinder is assumed to be equal to the diameter of the piston d pi, required for the OLEO. The piston is sized according to the mean of the two values using methods from [80] and [100]. The first of these two methods is, d pi = 4Lmain g πp max, (B.134) where P max is the maximum allowable internal pressure of the OLEO taken to be 1500psi. The second is an empirical method using imperial units given by, d pi = L 1 2 main. (B.135) Assuming that d i = d pi, the previous equations for the design criterion can be rearranged and solved fro d o, hence giving the size of the strut that meets the safety conditions for failure in yield. This is done for both the main and nose landing gears using the worst-case loading scenarios, with the internal forces and moments calculated using the dynamic loading condition. Buckling analysis The buckling analysis assumes the worst case landing scenario with a safety factor of 1.5 since this is usually enforce in industry. This is simply done using an Euler buckling analysis for a cylinder, that is the outer diameter is given by, [ 64Ldi l LGi d o = π 3 E 300M + d 4 i ] 1 4, (B.136) where i denotes either the main or nose landing gear, therefore Ld i is the maximum dynamic loading on each of these, and the chosen material is 300M steel. 187

188 Fatigue analysis In practice fatigue requires experimental analysis for accurate results since it is dependent on geometry, material, manufacturing process etc. A crude estimation is carried out here assuming that 300M steel has a fatigue stress, σ f = 700MP a [120]. Assuming a loading cycle from zero to the maximum dynamic loading and a maximum 1 allowable stress of of the fatigue stress limit the diameter is given by 1.5 ( ) 1 Ldi d o = π 4 d2 i. (B.137) The maximum of the three calculated outer diameters of the cylinder is selected for the landing gear geometry. The mass of the main strut for each gear is then calculated using, ( W mainstrut = l LG 1 ) [ ] π (d 2 2 d o d 2 i ) LG m ρ 300M (B.138) 4 Other parts that require sizing include the side strut, axel and wheel rims as well as the fluid inside the OLEO strut. As mentioned previously the side strut diameter is assumed to be a percentage of the diameter main strut, that is, d oss = 0.5d o for the side strut of the main gear and back strut of the nose gear. The axle is assumed to run in between the tires with slightly lower diameter than the main strut and a length twice the distance of the main strut diameter. The piston is assumed to be a cylinder the length of the stroke length and the diameter calculated previously and the remaining internal space within the mains strut is assumed to be filled with fluid. Lastly, the wheel rims are assumed to take up the space in the middle of the tire, given from the tire database. The mass of the side strut is given by, [ ] π0.25 (d 2 W sidestrut = (l ss ) o d 2 i ) ρ 300M (B.139) 4 where the length of the side strut, L ss is calculated as the distance from the bottom of the main strut above the tires to the fuselage. The remainder of the masses are calculated in a similar manner from the same material with the exception of the rims being made from 6061 aluminium, and the fluid being a standard hydraulic fluid for aircraft pneumatic systems. The weight of the landing gear is then the summation of the weights of the respective components. σ f 1.5 B.13.4 Outputs Parameter W LG W LGm W LGn Description The total weight of the landing gear The weight of each of the main landing gear The weight of the nose landing gear 188

189 B.14 Engine weight calculation B.14.1 Function overview This function uses an empirical model from Guha et al. [52] to estimate the weight of a turbofan engine and a method from [98] to estimate the nacelle weight. B.14.2 Inputs Parameter d fan l engine d engine S wetnac Description The diameter of the engine fan The length of each of the engine The diameter of the engine The wetted area of the nacelle B.14.3 Processes The bare engine weight is estimated using, W engine,bare = ( 1.81d 3 fan 19.8dfan) (B.140) where d f an is in inches. Meanwhile the total nacelle weight is given by, W nac = K ng l 0.1 nacd enginenz ( 2.331W enginek p K tr ) N en S wet nac, (B.141) where K ng = captures the pylon mounted configuration, N z = 4.5 to account for the limit load, K p = 1.0 is a factor for turbofan engines, K tr = 1.18 assuming the engine has reverse thrusters, N en = 2 is the number of engines, and l nac is the length of the nacelle assumed to be the same as the engine length. The total weight of the engine is simply the addition of the weight of the bare engine and the total nacelle weight divided by the number of engines, W engine = W engine,bare + W nac 2. (B.142) B.14.4 Outputs Parameter W engine W nac W engine,bare Description The total weight of the engine The weight of the nacelle The weight of the bare engine (without a nacelle) 189

190 190

191 Appendix C Model verification method: Analysis Error In order to verify the model, outputs from the model were compared with values from real aircraft from the collected database. The results from this analysis are in the main body, however the method used for error analysis will be described in more detail here. A major assumption of the model is that the error is normally distributed. This assumption enables the use of confidence bounds on the outputs from the model in the results section. This assumption that the error is a random variable following a normal distribution is difficult to verify given that there is a very sparse dataset for verification. Methods such as bootstrapping, k-fold validation are useful for partitioning this sparse dataset and performing multiple verification and testing loops, however they do not help in this case with verifying the assumption of normal error. Since the true distribution of the error will remain unknown the assumption of normality is made, since it is a likely scenario and helps in simplifying the problem of bounding the error. The model error, e i for each aircraft i in the population n is given as the difference between the real value, y and the output from the model, ŷ, e i = ŷ y (C.1) with e i distributed as a normal distribution with mean µ and standard deviation σ, e 1, e 2,, e n N(µ, σ 2 ). (C.2) The sample mean is also a random variable distributed in the following way, ē N(µ, σ2 n ). (C.3) One would expect it to be the case that µ = 0, since the error is a randomly distributed variable; however it can be seen from the analysis in the main body that this is not the case. That is, there is assumed to be a bias in the model, where the bias is given by, bias [ŷ] = E [ŷ] y (C.4) which is related to the model error since µ = E [e i ] = E [ŷ] y. Given the above, confidence intervals for each the outputs of the model can be calculated. A two-sided 100(1 α)% confidence interval for a quantity ŷ with a normally distributed error 191

192 and bias β can be expressed in the following way, (ŷ β) ± z α σ. (C.5) 2 Therefore using the values for β and σ calculated in the verification of the model, and the appropriate z-values, z α, a confidence interval for the output ŷ can be found. 2 In order to compute the confidence intervals for the ratio of two normal random variables, x 1 and x 2, the following is used, [ ] [ x1 Var µ2 x 1 σ 2 x1 2 Cov(x ] 1, x 2 ) + σ2 x 2, (C.6) x 2 µ 2 x 2 µ 2 x 1 µ x1 µ x2 µ 2 x 2 where in the analysis of this thesis Cov(x 1, x 2 ) is assumed to be zero. Using the fact that V ar(x) = σ 2 x the above equation for the confidence interval can be applied in this instance. 192

193 Appendix D Technology scenario 4: Results (a) Engine fan diameter against BPR showing(b) Engine weight against BPR showing expected values and 95% confidence expected values and 95% confidence bounds. bounds. (c) Main landing gear length against BPR(d) Total landing gear weight against BPR showing expected values and 95% confidenceshowing expected values and 95% confidence bounds. bounds. Figure D-1: Variation of aircraft design parameters as BPR increases for scenario

194 (a) SFC against BPR showing expected values(b) Lift-to-drag ratio against BPR showing and 95% confidence bounds. expected values and 95% confidence bounds. (c) Aircraft mission final weight against BPR showing expected values and 95% confidence bounds. Figure D-2: Variation of aircraft performance variables with bypass ratio for scenario 4. Figure D-3: Graph of BPR against year showing both projection models based on different data sets for scenario

195 Appendix E Turbine inlet temperature trends Figure E-1: Turbine inlet temperature against year including civil and military aircraft as well as the effect of cooling technologies. Figure E-2: Graph of specific thrust against turbine inlet temperature showing theoretical stoichiometric limit. 195

196 196

197 Appendix F ICAO timeline of emissions regulations Figure F-1: A timeline of emissions standards developed by ICAO [89]. 197