University of Southern California Dissertations and Theses

Modeling and analysis of propulsion systems and components for electrified commercial aircraft

(USC Thesis Other)

Modeling and analysis of propulsion systems and components for electrified commercial aircraft

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content MODELING AND ANALYSIS OF PROPULSION SYSTEMS AND COMPONENTS FOR ELECTRIFIED COMMERCIAL AIRCRAFT by Saakar Byahut A Dissertation Presented to the FACULTY OF THE USC GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulllment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (AEROSPACE ENGINEERING) May 2021 Copyright 2021 Saakar Byahut Acknowledgements This section of my dissertation was the most dicult for me to write, because I knew from the start that (1) it would be personal and I am not very good at writing things of a personal nature, (2) I could not possibly give everyone the recognition they deserve, and (3) writing this represents the end of a chapter in my life, and it hasn’t sunk in yet. Regardless, I’ll attempt to acknowledge everyone who has supported me through this journey. On the academic side, I would like to rst and foremost thank Prof. Alejandra Uranga, who gave me a chance to purse a PhD in my eld of interest at a time where I was not fully convinced I wanted to start a PhD program a second time. She knew my background was not in aerospace engineering, yet she took me in, recognized my passion for the aerospace and airline industry, and most importantly, pushed me and supported me every step of the way. Despite our research group starting with my PhD cohort, she was an excellent and understanding mentor. I could not have asked for a better adviser. To the rest of my committee members, generally, thank you for providing valuable feedback through my qualifying exam and for being there as part of the USC community whenever I needed advice. More specically, to Prof. GeoSpedding: thank you for introducing me to the world of ight through AME 105. That class has been one of my favorites, and I feel a little jealous that my undergraduate education (as excellent as it was) did not oer something similar. Thank you for allowing me to take part in Glider Days over the years, for the work environment in the Wind Tunnel, and for always having your door open when I needed something. To Prof. Mihailo Jovanovic, whom I have known since my University of Minnesota days: ii thank you for your support and friendliness over the years. It has always been great catching up with you when we meet. ToProf. MartyBradley: thank you for the AIAA course on Electric Aircraft Design that helped shaped the direction of my dissertation after the NASA LEARN3 project was over. I learned a lot during that course, and of course, I learned a lot sitting in on your AME 481 lectures, and presentations through AIAA and USC. ToProf.CharlesRadovich: thank you for letting me TA for you last summer, in a lab setting which I hadn’t done since my undergrad days, for letting me be part of the AME community through Water Rocket Days, and for allowing me to be a part of the USC Aero Design Team. Even though I could not attend the weekly meetings this year, I appreciate and enjoyed the ones I did in previous years. ToProf.IvánBermejo-Moreno: thank you for stepping in at the last minute during my qualifying exams and for your thoughtful comments and questions. Thank you for the opportunities to TA for you in AME 309 and AME 511. I would also like to acknowledgeProfs. PaulNewton,CarlosPantano-Rubino, andMitulLuhar for giving me opportunities to TA for them. Thank you for giving me the freedom to do things my way, and most importantly, for allowing me a glimpse into the way all of you teach. You helped me realize my passion for teaching and I learned something dierent from each of you. I know I will carry my experiences and observations forward whenever I teach in the future. My time at USC would not have been as amazing if not for the awesome sta members at the AME department. Thank you to Melissa, Irice,Ken,Natalie,Juli,Brian, andChrissy for allowing me to pester you with questions and concerns. Thank you to Chelsea, Amanda, Silvana, and Kim for being so helpful whenever I needed to arrange something. Since the pandemic-related lockdown, it feels strange not being able to go into the AME oce for a water rell and bits of conversation; they added bits of joy to my days. I would not have been here without my family’s continuous and unwavering support. My parents, Sitaram andShantiByahut, have given me a degree of independence most sixteen-year-olds would not dream of having when I left for Wales, and they have been there every step of the way since. Words iii are inadequate to describe my appreciation, but I hope that I have made you proud. This lockdown has also made me aware of how important a support network is, and to all my extended family members on our weekly Zoom calls, thank you for your support and love. In particular, I am grateful to have so many family members in the US who have been there through my undergrad and PhD days. ToJyajya andMaiju Shankar andUjalaManandhar, and to my cousinsAbhilasha andAastha, thank you for making it feel like home every time I visit. To my Dijjus, SmritiManandhar andSusmitaAdhikariJoshi, and your families: Rameet andSushobhan dai, Sushrut, Nilisha, and nowRonesh, thank you for bringing joy into my life and for taking such good care of me whenever I visit. To my amazing host families:Beau and Tim in Sharon, VT; andMechthild andJo in Berlin, Germany: thank you for making me feel like a part of your families. Honestly, at the very least, all of you have fueled my passion for ying, and I could not be more grateful for that. My USC experience would not have been complete without amazing labmates, colleagues, and friends here. I say without hyperbole that I was fortunate enough to have the best labmates ever. First, toMichael Kruger, who taught me so much about aircraft design, thank you. I could not have asked for a better lab- mate to work with on the NASA LEARN3 project. Your work ethic, drive, and passion have always been sources of motivation for me. Thanks for putting up with me wherever we ew somewhere together, for making me feel comfortable socializing with the numerous people we met at conferences and presenta- tions, and for always easing my nerves before major presentations. ToJamesCroughan: we met before we both started here, and I couldn’t have imagined a better labmate at the time. Thanks for listening to me ramble and rant, for struggling with me on the homeworks for classes we took together, and for always inviting me to social events. To Arturo Cajal: I enjoyed working with you on SHINE. Thanks also for your support at the beginning of Covid, which I really appreciated, as some weeks, that would be the only longish conversation I would have with anyone. To Raye (Tianbo) Xie: thank you for always oering great comments on my work whether as a TA or as a colleague. To Andrew Dorsey, who joined our iv group a bit later and t right in, thanks for teaching me so much about the airline/aerospace industry, and for keeping me up to date on aerospace news. ToTanmayMukherjee: thank you for all your help integrating our models, and for making mine run a lot faster – I fully appreciated the eort you put in as I was generating results for this dissertation. To Prof. Spedding’s students, whom I consider labmates as well: Lelanie Smith, you helped me more than you probably realize when I rst started at USC; I will be forever grateful for that. ToJoeTank: you were a fantastic TA and gave me useful comments on my research as well, but most of all, thank you for getting me through the screening exam – your meticulous notes were a huge help. ToBradleyMcLaughlin, for your friendship and for the (mis)adventures we had playing Pokemon Go! To Yohanna Hanna, I don’t even know what to say in words here; there are so many things to thank you for. Thanks for geeking out with me about aviation stu, for indulging me in aviation-related and other events, for always being a friend day after day, for going out of your way to help me out and make me feel included, for long thoughtful conversations, for being a role model through your community service, and a whole lot more. To Trystan, Chris (Chan-Ye), and Emma: thanks for being part of our lunch group. A pandemic makes you realize how much you cherish something as routine as eating lunch together in a group. To the extended AME and USC students:Shilpa, it was a pleasure taking classes and TA’ing with you; Mark,Edwin, andSam: thanks for getting me through control theory;AndrewC.,Christoph,Vamsi, ChelseaA. andKarina: thank you for taking the time to critique my presentations. It was a pleasure to be able to interact with the multitude of undergraduates who worked in our lab (particularlyAra,Patrick and Steven); thank you for making my workdays a little bit brighter. I am also particularly grateful for the chance to mentor over my time at USC. My amazing SHINE students Manaeha, Alex, Achintya, and Katelyn: thank you for showing me what driven people can do even from a young age, and for your thoughtful questions, eagerness to help, and for going above and beyond the SHINE requirements to help out in our lab. To my graduate menteesBryan,Huaiyu(Bryce),Fares, andAditya, thank you for v allowing me to introduce you to USC, and for your eorts at not letting the mentor-mentee relationship fade after your rst semesters here. To two amazing roommatesBaibhav andSushant: I could not have asked for better roommates. ToNirakardai,Ashrant,Sudha, and the rest of the Nepali gang: thanks for accepting me as I am, for the wonderful times, and the delicious momos. And nally, to everyone else I’ve forgotten to mention here, I appreciate all of you. During the lock- down, I found myself going through old pictures and through old memories, which made me incredibly thankful for all the dierent ways you have touched my life. To all my friends, teachers, sta, faculty, co-workers, colleagues from Rato Bangala School, Ruthin School, Dartmouth College, University of Min- nesota, and here at USC, thank you for being a part of my life, whether in the past or present. I end this with the African proverb that I truly applies here: “It takes a village to raise a child”. You have all raised me and supported me throughout this, and I would not be here without you. vi TableofContents Acknowledgements ii ListofTables ix ListofFigures x Abstract xiii Chapter1: Introduction 1 1.1 Motivation: Electried Propulsion for Aircraft . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.1.1 Promises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.1.2 Challenges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.2 Terminology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 1.3 Previous Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 1.4 Unied Propulsion System and Electrical Components . . . . . . . . . . . . . . . . . . . . 17 1.5 Performance Metric . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 1.6 Thesis Scope and Organization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 Chapter2: ElectriedPropulsionTechnology 25 2.1 Technology Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 2.1.1 Batteries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 2.1.2 Electric Machines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 2.1.3 Power Electronics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 2.1.4 Wiring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 2.1.5 Thermal Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 2.2 Technology Predictions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 2.2.1 Batteries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 2.2.2 Electric Machines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 2.2.3 Power Electronics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 2.3 Technology Scenarios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 Chapter3: Low-FidelityAnalysis 39 3.1 Models and Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 3.1.1 Propulsion System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 3.2 Results: Electric Component Technology Levels . . . . . . . . . . . . . . . . . . . . . . . . 46 3.2.1 Eects of Battery Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 3.2.2 Eects of Component Specic Power . . . . . . . . . . . . . . . . . . . . . . . . . . 49 vii 3.2.3 Technology Level Analysis Summary . . . . . . . . . . . . . . . . . . . . . . . . . . 51 3.2.4 Limiting Cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 3.2.4.1 Battery Specic Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 3.2.4.2 Electrical Component Specic Powers . . . . . . . . . . . . . . . . . . . . 54 3.3 Usefulness and Limitations of the Low-Fidelity Approach . . . . . . . . . . . . . . . . . . . 55 3.4 Low-Fidelity Study Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 Chapter4: Higher-FidelityAnalysis 58 4.1 Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 4.1.1 Battery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 4.1.2 Converter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 4.1.3 Power Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 4.1.4 Cables and Wiring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 4.1.5 Motor Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69 4.1.6 Propulsor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 4.1.7 Thermal Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 4.2 Integration and Sizing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86 4.2.1 Propulsion System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86 4.2.2 Aircraft and Mission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89 4.2.2.1 Reference Aircraft . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89 4.2.2.2 Mission Power Prole . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 4.2.2.3 Aircraft Sizing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93 4.3 Results: Subsystem-Level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94 4.3.1 Comparison of Low- and Higher-Fidelity Models . . . . . . . . . . . . . . . . . . . 94 4.3.1.1 Cruise-Only Mission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94 4.3.1.2 Climb-Cruise-Approach Mission . . . . . . . . . . . . . . . . . . . . . . . 97 4.3.2 Component Eciencies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 4.3.3 Power Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 4.3.4 Wiring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104 4.4 Results: Aircraft-Level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 4.4.1 System Mass Breakdown . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 4.4.2 Distributed Propulsion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 4.5 Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 Chapter5: Conclusions 113 Bibliography 117 viii ListofTables 1.1 Propulsion system architectures represented by the unied view and their dening parameters: load and source electrication factors,f S andf L . . . . . . . . . . . . . . . . . 19 2.1 Technology scenarios: assumptions values for electrical component parameters. . . . . . . 38 3.1 Comparison of all-electric with limiting BSE vs conventional . . . . . . . . . . . . . . . . . 53 3.2 Comparison of turbo-electric with limiting specic power vs conventional . . . . . . . . . 54 4.1 Motor core power loss coecients for the dierent motor segments and loss types. . . . . 78 4.2 Parameters for the SRM. The values with an asterisk are initial guesses and allowed to be optimized; all other values are selected. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 4.3 Baseline aircraft and mission specications [63] . . . . . . . . . . . . . . . . . . . . . . . . 89 4.4 Comparison between the low-delity and higher-delity component parameters for a cruise-only mission with low-delity eciencies from the intermediate 2035 technology scenario. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 4.5 Comparison between the low-delity and higher-delity component parameters for a cruise-only mission with low-delity eciencies set to those obtained from higher-delity models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 4.6 Comparison between the low-delity and higher-delity component parameters for a climb-cruise-approach mission with low-delity eciencies set to those obtained from higher-delity models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99 4.7 Safe operating voltages for dierent aircraft classes at dierent safety factors. . . . . . . . 104 4.8 Wiring resistance, mass, and power for copper and aluminum conductors for 2 and 20 propulsors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 ix ListofFigures 1.1 Illustration of weight reduction benets of distributed propulsion [11]. . . . . . . . . . . . 4 1.2 Illustration of the aerodynamic benet of boundary layer ingestion [12]. . . . . . . . . . . 5 1.3 Various electried propulsion system architectures [4]. . . . . . . . . . . . . . . . . . . . . 8 1.4 Various current, in-development, and conceptual aircraft with dierent propulsion system architectures. Images ©Viking Air, Boeing, Airbus, NASA, ESAero, Eviation, Pipistrel respectively. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 1.5 Unied propulsion system model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 2.1 Theoretical and cell-level BSE for several battery chemistries. . . . . . . . . . . . . . . . . 27 2.2 Cell BSE as a percentage of theoretical BSE for dierent battery chemistries in use. . . . . 28 2.3 Ragone diagram showing the relationship between BSE and BSP, adapted from [36]. . . . . 30 2.4 Non-dimensional Ragone diagram showing the relationship between BSE and BSP. . . . . 31 2.5 Summary of cell-level BSE projections in previous studies . . . . . . . . . . . . . . . . . . . 34 2.6 Survey of electric machine specic power versus continuous power; existing machines and future design projections. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 2.7 Summary of power electronics specic power versus continuous power; existing designs and future design projections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 3.1 Framework overview: modules and their interactions [11]. . . . . . . . . . . . . . . . . . . 39 3.2 Eect of battery technology on PSEC with DP for 100 nmi all-electric (f S = 1, f L = 1) commuter aircraft assuming optimistic 2035 technology for other components ([P=m] mot = 16 kW/kg, [P=m] conv = 19 kW/kg). . . . . . . . . . . . . . . . . . . . . . . . . 47 x 3.3 Eect of battery technology onPSEC with DP and BLI for 100 nmi all-electric (f S = 1, f L = 1) commuter aircraft assuming optimistic 2035 technology for other components ([P=m] mot = 16 kW/kg, [P=m] conv = 19 kW/kg). . . . . . . . . . . . . . . . . . . . . . . . . 48 3.4 Eect of component specic powers on PSEC with DP for 100 nmi all-electric (f S = 1,f L = 1) commuter aircraft assuming an optimistic 2035 battery technology at BSE = 900 Wh/kg. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 3.5 Eect of component specic powers onPSEC with DP and BLI for 100 nmi all-electric (f S = 1,f L = 1) commuter aircraft assuming an optimistic 2035 battery technology at BSE = 900 Wh/kg. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 3.6 Eect of improving component specic powers on PSEC for a medium-haul turbo-electric (f L = 0:48, f BLI M = 0:2) aircraft with N fan M = 2, d fan M = 1:14 m, f BLI E = 0:5, N fan E = 308,d fan E = 0:104 m. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 4.1 Discharge prole of the cell used in the batteries of the all-electric NASA X-57 Maxwell [45]. 59 4.2 Schematic of (a) ideal DC-DC converter, (b) equivalent circuit for DC-DC transformer representation of the converter, and (c) current waveforms over a cycle. . . . . . . . . . . . 63 4.3 Non-ideal converter: (left) schematic with switch, diode, and inductance, and (right) equivalent circuit model for a full cycle of operation. . . . . . . . . . . . . . . . . . . . . . 63 4.4 Paschen’s curve for air. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 4.5 (Left) Switched Reluctance Motor (SRM) cross-section schematic, and (right) illustration of ux linkages versus current curves for an SRM. . . . . . . . . . . . . . . . . . . . . . . . 71 4.6 (Left) Flux paths when the rotor and stator pole pairs are aligned, and (right) magnetic equivalent circuit for ux path FP1 (blue paths in the left image). . . . . . . . . . . . . . . . 72 4.7 Algorithm to calculate inductances for the motor (adapted from [56]). . . . . . . . . . . . . 76 4.8 Heat exchanger model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82 4.9 Propulsion system schematic. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87 4.10 Twin Otter cross-section schematic to calulate wiring length. . . . . . . . . . . . . . . . . . 90 4.11 Twin-Otter climb-cruise-approach mission: (left) altitude and (right) power prole. . . . . 93 4.12 Battery mass and energy dierences for the low- and higher-delity models for the Twin Otter mission. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 4.13 Eciency of motor and converter over the ight segments for (left) 2 propulsors and (right) 20 propulsors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 xi 4.14 (a) Variation of breakdown voltage for uninsulated conductors, and (b) variation of safe operating voltage (SOV) for insulated conductors at dierent pressures and conductor spacing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 4.15 (a) Variation of safe operating voltage (SOV) with insulation thickness, and (b) with insulation material at dierent pressures. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103 4.16 Propulsion system mass breakdown by component for (left) two propulsors and (right) 20 propulsors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 4.17 Propulsion system mass changes with distributed propulsion; (left) showing all the components, and (right) showing more detail of the smaller components. . . . . . . . . . . 108 4.18 Propulsion system distribution current changes with distributed propulsion. . . . . . . . . 109 4.19 Eciency changes with distributed propulsion; (left) for the converter and motor, and (right) showing the overall system eciency. . . . . . . . . . . . . . . . . . . . . . . . . . . 110 xii Abstract Commercial aircraft currently rely on hydrocarbons as the sole source of energy for propulsion. The variability of fuel prices, the growing emphasis on environmental sustainability, and the increased demand for air transportation have led to enhanced interest in improving fuel eciency and reducing emissions for transport aircraft. Electrication of the aircraft propulsion system has the potential to achieve both goals. It can also better leverage the benets of distributed propulsion (DP) and boundary layer ingestion (BLI) to further enhance energy eciency. Electrication, however, poses major challenges. Batteries provide substantially less energy per unit mass than hydrocarbon fuel, which means battery-powered aircraft will weigh more than hydrocarbon-fueled aircraft for the same mission. Replacing conventional propulsion systems with electried ones is thus not expected to be benecial due to added weight and complexities. Despite the challenges, there are scenarios for which electrication could be benecial. This work focuses on the modeling of electried propulsion system components. The use of a uni- ed propulsion system that can handle all architectures from conventional to hybrid-electric to all-electric enables comparison between dierent architectures and is used to put together the models of various com- ponents. A technology overview captures trends in the development of electrical component technologies, based on predictions of how the masses and eciencies of these components will improve in the far term (2035), and thus help reduce the electried propulsion system weight. A low-delity framework was de- veloped to capture the major trade-os of electrication at cruise condition, as well as the eects of DP and BLI. Results from this framework applied to a 20-passenger commuter aircraft ying 100 nmi showed that xiii all-electric aircraft become feasible and benecial over conventional after substantial advances in battery technology. Other components like the motor and the converter play a less meaningful role, but improve- ments to both also yield energy savings, but the benets plateau with technology advances beyond the 2035 scenarios, simply because the battery makes up a much larger fraction of the propulsion system mass than the other components. Both DP and BLI improve the feasibility and energy eciency of the all-electric aircraft considered. Limitations of the low-delity framework with regard to the modeling of electrical components led to interest in analyzing the behavior of these components under more representative operational loads. Higher-delity models of batteries, electrical machines, power electronics, wiring, and thermal manage- ment systems (TMS) are developed. These components are integrated into the unied propulsion system model to better evaluate an all-electric architectures and gain a better understanding of how the compo- nents behave under ight loads. The results are used to better model their masses for a more accurate estimate of the weight of the electried propulsion systems as well as their eciencies. For the repre- sentative mission of a 20-passenger aircraft ying 100 nmi, the electrical components are sized for the highest-power climb segment, and operated at lower-power cruise and approach segments. This higher- delity model predicts about a 120% higher energy requirement than the previous low-delity approach. Motor eciency drops substantially at lower-power segments from 95% at climb to 77% at cruise for a 2- propulsor aircraft, and the resulting ineciency requires both a larger battery and TMS. A safe operating voltage of 3 kV is determined based on the maximum service ceiling of the baseline aircraft and then used to set the distribution current and wire gauge. In terms of masses, the battery dominates, and DP helps lower the overall system mass by 27% and also improves the system eciency at o-design power loads, from 30% to 45%, both for a 20-propulsor aircraft compared to a 2-propulsor aircraft. For the missions considered, all-electric aircraft are found to have double the source-to-load power conversion eciency than conventional aircraft, with the eciency increasing by up to 5% with more DP. xiv Chapter1 Introduction 1.1 Motivation: ElectriedPropulsionforAircraft In 2019, about 4 billion people traveled by air. The International Air Transport Association (IATA) expects this number to nearly double to 7.8 billion within the next 20 years [1]. Aircraft manufacturer Airbus expects the world annual trac of air passengers to double every fteen years, or grow at 4.4% per an- num, with the trend resilient to shocks such as the 2008 global nancial crisis and the 2020 Covid-19 pandemic [2]. Similarly, Boeing expects air trac to grow about 4.7% every year for the next 20 years in its 2018 market forecast [3]. A 2016 report by the National Academies of Sciences, Engineering, and Medicine (NAE) found that although aviation accounts for only 2–2.5% of global carbon dioxide (CO 2 ) emissions today, reducing emissions from air transportation is crucial for three reasons. First,reductions in emissions may some day be mandated by legislation. Second, there is a time lag between the devel- opment of new technology and its introduction into the global aviation eet. Finally, reducing emissions in the future mitigates the impact of current, ongoing emissions [4]. The volatility of fuel prices, coupled with the growing demand for air travel and an increased emphasis on environmental sustainability opens up the eld of aircraft design for novel architectures and congurations that could meet these future needs. 1 Several concepts have been proposed for future aircraft, including high-aspect ratio truss-braced wing aircraft, blended wing-body aircraft (both from Boeing [5]); and double-bubble aircraft ingesting the fuse- lage boundary layer through aft-mounted propulsors (MIT [6]). There has also been growing interest on the promise shown by electried aircraft. Some of these include a turbo-electric design which uses a gener- ator to convert turbine shaft power into electrical power to drive an aft-fuselage boundary-layer-ingesting propulsor (NASA [7]) and a hybrid-electric truss-braced wing concept with battery pods mounted on the wings (Boeing [5]). Airbus has developed technology demonstrators such as the all-electric two-seater E-Fan and under-development turbo-electric E-Fan X [8]. Budding aircraft manufacturers are developing small all-electric aircraft, including the Eviation Alice, which seats nine passengers for ights up to 440 nautical miles (nmi) and is projected to enter into service in the 2020s [9]. Previous studies have shown benets of electrication for transport aircraft by considering point- designs for a specic aircraft and propulsion architecture. However, there is a lack of a broader approach and framework that could be used to explore the broad design space, and identify areas where electried aircraft are feasible and benecial with a relatively high degree of condence. The present work aims to develop propulsion system models to address the challenges of electrication and highlight areas with potential. Using component models, technology levels, a unied view of propulsion system architectures, and a set of aircraft congurations and missions, the ultimate goal is to identify the areas favorable to electrication and the potential performance benets. A low-delity model identies areas where electri- ed aircraft makes sense, and a higher-delity model explores these areas deeper with detailed electrical component models and ight segments. Aircraft propulsion system electrication has the potential to provide higher component eciencies as well as lower emissions, oering some inherent benets over conventional propulsion systems. Electri- cation, however, poses major challenges. The major promises and challenges are discussed next. 2 1.1.1 Promises Source-to-LoadConversionEciency Electrical components such as power converters and motors are highly ecient at converting power from the source to the load. Eciencies reach about 95% today and are expected to grow to over 99% in the next 20 years [4]. For an all-electric propulsion system, a conversion chain from battery (energy/power source) to converter to motor to propeller (prime mover) yields an overall eciency of about 70%. For a conventional turboprop system, the conversion from hydrocarbon fuel to mechanical shaft to propeller yields an overall eciency of about 40%, largely as a result of the roughly 50% eciency of the thermody- namic cycle associated with combustion [10]. An electrical system thus converts power more eciently than a hydrocarbon-based conventional system, reducing the on-board energy required to produce the useful work necessary for ight. DistributedPropulsion(DP) Other benets of electried propulsion stem from the benecial technologies that it enables. Distributed propulsion (DP), or the use of multiple smaller propulsors rather than the one or two propulsors in con- ventional aircraft, can provide a weight reduction benet. This can be seen by considering the relation between the propulsor mass and the mass ow through it, the latter being directly related to the thrust. The mass of a propulsor, m prop , to rst order, can be assumed to scale with its volume and hence with the cube of its characteristic length,`, while the propulsor mass ow, _ m prop , scales with its frontal area, represented by its characteristic length squared. Thus,`m 1=3 prop and` _ m 1=2 prop , such that the mass of a propulsor and its mass ow rate are linked by a cube-squared relationship: m prop _ m 3=2 prop : (1.1) 3 Single propulsor Distributed Propulsion: Same fan area, half the weight Distributed Propulsion: Same weight, 1:6 more fan area Figure 1.1: Illustration of weight reduction benets of distributed propulsion [11]. The thrust-to-weight ratio of a propulsion system, which scales like _ m prop =m prop , thus decreases as _ m 1=2 prop increases. As an example, consider the use of four small DP units instead of a large one as illustrated in Fig. 1.1 used in [11] to illustrate this point. If the single large propulsor (shown under the wing in gray) is replaced by a four-unit DP system (shown under the wing in hashed blue) with the requirement of same total thrust, then for the same total fan area, the weight would be reduced by half (mass scales with square-root of the number of propulsors). If instead, the requirement is to maintain the same propulsion system weight, the four-unit DP system (red circles above the wing) would provide 1.6 times more total fan face area (mass ow scales with cubic root of the number of propulsors), enabling reduced fan pressure ratio and increased propulsive eciency. To rst order, the use of a distributed propulsion system thus has the advantage of being lighter or more ecient, and the larger the number of propulsors the better—at least in principle. However, operational or other design constraints, such as requiring the same level of smooth operability, maintenance costs and work hours as two propulsors, and structural strengthening limit too-large a number (hundreds) of distributed propulsors even if the massive distribution yields energy savings. 4 BoundaryLayerIngestion(BLI) In a conventional engine installation, the propulsors are mounted away from the airframe. They ingest uniform freestream ow and their jets counteract the momentum defect in the airframe wake. At cruise, the jets and wakes combine to a zero net momentum (thrust equals drag), but both the airframe wake and the propulsor jet represent wasted kinetic energy, as illustrated in Fig. 1.2. A more ecient alternative is to place the propulsors in the boundary layer of the airframe, thus re- energizing the slower-moving ow, which otherwise forms the wake. The resulting combined wake and jet has lower kinetic energy and hence lower losses than a conventional propulsor. The process of having at least part of the airframe boundary layer ingested by the propulsion system is called boundary layer ingestion (BLI), which is known to increase the overall eciency of the aircraft. The level of benet that BLI provides relative to a conventional engine installation is a function of the amount of boundary layer ingested [13]. The optimal case is when all of the boundary layer of the body is ingested and accelerated exactly to freestream conditions, but achieving full ingestion requires that the propulsor system inlets cover all of the airframe trailing edges (including fuselage, wings, and tails). This can be dicult to realize in practice, especially if only a small number of large propulsors are to be used. Electrication and distribution can facilitate BLI, since propulsors that are distributed over surfaces can wake, or “draft” Wasted Kinetic Energy Zero Net Momentum combined wake and jet propulsor jet + + + + + + + - - - Figure 1.2: Illustration of the aerodynamic benet of boundary layer ingestion [12]. 5 ingest more of the boundary layers. Thus, with electrication, the aerodynamic benet of BLI can be better exploited. 1.1.2 Challenges Electried propulsion also has a number of inherent disadvantages, the major one being the weight of batteries. The weight of energy stored in batteries is two orders of magnitude higher than if it were stored as fuel. The specic energy (energy per unit mass) of fuel is around 13 000 Wh/kg, whereas aerospace- grade batteries like the ones used in the all-electric two seater Airbus E-Fan achieve a specic energy of only 175 Wh/kg at pack level today [8]. Even with revolutionary breakthroughs in battery chemistry, it is unlikely that batteries will achieve specic energies as high as those of hydrocarbon fuels. Novel designs like Lithium-air have theoretical capacities of about 3500 Wh/kg; however, these values will be lower at pack level, due to the ineciencies associated with packaging and thermal management. Furthermore, such novel chemistries have not been commercialized yet. While battery specic energy will never reach values close to that of hydrocarbons, improvements will lead to aircraft in certain classes becoming viable for a variety of missions. Battery technology will be discussed in detail in Sec 2.1. As a result of battery weight, directly replacing a conventional internal-combustion engine on an exist- ing aircraft with an electric propulsion system is not expected to be benecial due to the added complexity and weight [14]. To take advantage of electrication, the full aircraft needs to be recongured, which may oer the potential for novel aircraft designs that, for example, use distributed propulsion with signicant boundary layer ingestion. These designs, however, also lead to modeling complexities due to the increased coupling between aircraft aerodynamics, structures, and propulsion. There are also operational challenges related to electrical propulsion that must be addressed. Current civil aviation infrastructure is built around a hydrocarbon-based energy supply system at airports. This system would need to be modied in order to recharge batteries on the ground, or to overhaul other 6 electrical components like motors and converters. Furthermore, certication requirements for electrical propulsion systems are not yet in place. Battery chemistries like Lithium-ion are prone to thermal runaway and pose a re hazard [15]: their safety in aviation applications needs to be addressed. Structures to contain thermal runaway damage add weight to the battery packs, reducing their pack-level specic energy further. In addition, electrical propulsion systems are currently at low technology readiness levels (TRL) com- pared to the well-established turboprop and jet engines. Batteries have relatively small life cycles before their energy capacities diminish signicantly, while aircraft engines currently achieve thousands of cy- cles (takeos and landings) before needing major overhaul. Due to the relative immaturity of electrical propulsion system components, more work needs to be done to make them as reliable as engines. The simultaneous requirements of safety, reliability, and cost are additional challenges that need to be resolved before electried propulsion becomes commercially viable. Despite these challenges, there are promising developments in electried propulsion. There are al- ready small two-seater all-electric aircraft in development or under production, such as the Airbus E-Fan, Pipistrel Alpha Electro, and Pipistrel Taurus Electro. As battery specic energy increases and electrical component technology improves, the feasible space for electried aircraft expands. The question that re- mains – and towards whose answer the present work aims to contribute – is whether the promises of electrication can be leveraged to overcome its challenges, and if so, for which missions, with which air- craft congurations, with which propulsion system architectures, and with which technology. 1.2 Terminology The terminology used in this work to describe electried propulsion systems is derived from work done by NASA, as illustrated in Fig. 1.3. There are various types of electried propulsion systems, but before dening them all, it is important to dene conventional propulsion. The termconventional is used to refer 7 Figure 1.3: Various electried propulsion system architectures [4]. to a propulsion system architecture that uses hydrocarbon fuel as the sole source of energy and employs no electrical components for propulsion. Examples of conventional aircraft currently ying today include commuter aircraft like the 19-seater Viking Air Twin Otter to mid-size workhorse oerings like the 150- passenger Boeing 737 and Airbus A320, all the way up to intercontinental transports like the 350-passenger Boeing 777 and 550-passenger Airbus A380. Aturbo-electric architecture refers to a propulsion system that retains the hydrocarbon fuel as the sole energy source, but employs electrical components in the conversion from source to load: one or more gas turbines generate power that is distributed to one or more fans through a component chain of a generator, converter, and motor. In a fully turbo-electric architecture, all the fans are electrically driven (i.e. through electric motors), whereas in a partial turbo-electric design, some fans are electrically driven while others are mechanically driven (i.e. through a shaft powered by the gas turbine). Examples of conceptual turbo- electric aircraft include the single-aisle ECO-150, the hybrid wing body N3-X (both fully turbo-electric) and the single-aisle NASA STARC-ABL (partial turbo-electric). 8 Ahybrid-electric architecture relies on both batteries and hydrocarbon fuel to store energy for propul- sion. Hybrid-electric architectures could be further classied as either series or parallel. In series, the propulsors receive only electrical power from the turbo-generator and the battery, whereas in parallel, the mechanical fans receive additional power from a battery-powered motor mounted on the same shaft as the turbine. Boeing’s conceptual 154-passenger SUGAR Volt employs a parallel hybrid-electric architecture. Finally, in anall-electric architecture all the energy needed for propulsion is stored in batteries. Exam- ples of all-electric aircraft include the Airbus E-Fan, Pipistrel Alpha Electro, and Pipistrel Taurus Electro–all two-seaters– and the Eviation Alice, a nine-seater. The term “electried” is used to refer to a propulsion system that uses electrical components to generate thrust. It therefore encompasses turbo-electric, hybrid-electric and all-electric architectures. Note that this denition means that an electried propulsion system may not necessarily have electrical energy stored onboard in batteries; instead, electrical power may be generated by burning hydrocarbons which power generators. Figure 1.4 shows some of the various conceptual, in-development, and currently ying electried and conventional aircraft. 1.3 PreviousStudies With the increased interest in electried propulsion, several researchers have approached designs in dif- ferent ways. Some of these designs replace or augment conventional propulsion systems with electried counterparts, and thus retain the overall shape and conguration of the reference aircraft. In other words, such electried aircraft look very similar to the conventional aircraft ying today. Other designs open up the design space, taking advantage of benecial technologies like distributed propulsion (DP) and bound- ary layer ingestion (BLI), enabled by electrication. These aircraft look novel. 9 Conventional Aircraft (a) Twin Otter (19 pax) (b) Boeing 737 (150 pax) (c) Airbus A320 (150 pax) Turbo-Electric Aircraft (a) STARC-ABL (150 pax): partial turbo-electric (b) ECO-150 (150 pax): fully turbo-electric Hybrid-Electric Aircraft (a) Boeing SUGAR Volt (150 pax): parallel hybrid All-Electric Aircraft (a) Airbus E-Fan (2 pax) (b) Eviation Alice (9 pax) Figure 1.4: Various current, in-development, and conceptual aircraft with dierent propulsion system ar- chitectures. Images©Viking Air, Boeing, Airbus, NASA, ESAero, Eviation, Pipistrel respectively. Moore and Fredericks [14] present four misconceptions that exist about electried aircraft. The rst is that design of electried aircraft is no dierent than existing aircraft. Existing commercial aircraft largely use two underwing podded turbofan engines or wing-mounted turboprop engines. Merely swapping out these engines to place electried propulsion systems will not yield benets over conventional designs due 10 to added weight and complexity. Instead, to leverage the benets that DP and BLI oer, designs need to consider the cross-coupling between the aerodynamics and propulsion. The second misconception stems from trying to compare propulsion systems on an isolated basis. Taking into account the benets provided by the synergistic coupling between aerodynamics, weight, and propulsion in electried designs with DP is essential. The third misconception relates to the idea that electried aircraft, like electric cars, will not be feasible nancially. The authors argue that there are business models like shared-use high-utilization scenarios, which could compensate for potentially higher acquisition costs and limited battery life. Dierent usage cycles and recharge cycles based on o- peak times could also lower electricity costs, while electricity itself could be comparatively cheaper in the future, and hydrocarbon fuels more and more expensive. Coupled with lower maintenance of electrical components, electried aircraft may be able to compete on overall life cycle costs. The nal misconception addresses the issue of energy storage. Current batteries have two orders of magnitude lower specic energy (energy per unit mass) than hydrocarbon fuels, rendering batteries infeasible to achieve the long ranges of larger conventional aircraft. Here, the authors argue that modest improvements in battery specic energy are enough to make smaller hybrid-electric commercial aircraft feasible. Electried aircraft will start becoming prevalent for smaller missions, like the training aircraft and experimental aircraft that are already in production or under development, and larger aircraft will become viable as battery technology improves. While batteries will never achieve the specic energy numbers of hydrocarbon fuels, innovative electried aircraft can still capture the substantial market for reduced-range mission while providing energy and cost benets over their conventional competitors. To summarize, addressing these misconceptions means that electried aircraft will look dierent than today’s aircraft. They will start out small and grow bigger in terms of payload and missions as technology improves. Comparing conventional and electried aircraft must be done at the system level, not just on an isolated propulsion system basis, and across the aircraft life cycle. 11 Nevertheless, there are several aircraft studies that compare conventional and electried aircraft on a propulsion system basis. Antcli et al. at NASA [16] looked at a small 50-passenger aircraft (similar to an ATR-42) and swapped out the conventional propulsion system with a parallel hybrid-electric system. They found cost benets of the hybrid over conventional at dierent levels of hybridization, battery technologies, and electricity costs, with the benets increasing with advanced technology. However, at low battery specic energy values, the hybrid-electric was less attractive than an advanced conventional aircraft. They also concluded that the hybrid-electric case is sensitive to range, performing better than the conventional at shorter ranges. Boeing also considered a hybrid-electric concept – dubbed the SUGAR Volt – in the Boeing 737 class (150 pax, 3500 nmi) [5]. The SUGAR Volt used a parallel hybrid- electric propulsion system with two un- derwing turbofans similar to those in use today, except with a high-aspect ratio high-wing conguration. It was able to achieve a 60% fuel burn reduction and 54% energy over current conventional aircraft on a 900 nmi mission. However, compared to an advanced conventional design, it consumed more energy. If electricity prices improved over current gures, then the Boeing SUGAR Volt also performed favor- ably against an advanced conventional design. Friedrich and Robertson [17, 18] found similar results for their hybrid-electric designs of the same class: hybrid-electrics burned up to 10% less fuel than competing conventional designs, but consumed more energy when accounting for the energy in the fuel and in the battery. They also looked at a single-seater microlight aircraft and found that the maximum energy saving was 30% for the hybrid-electric design versus its conventional counterpart. For even smaller aircraft in the unmanned aerial vehicle (UAV) class, they found still higher energy savings – as much as 47%. Welstead and Felder at NASA [7] considered a bigger aircraft in the Boeing 737/ Airbus A320 class of 150 passengers ying a design mission of 3500 nmi. This aircraft, called the STARC-ABL, is a turbo- electric design (without batteries; all onboard energy stored as fuel) and leverages a boundary layer in- gesting propulsor on the aft fuselage to help deliver a 12% reduction in fuel burn over a conventional 12 aircraft. Incorporating the aft BLI propulsor also allowed weight savings due to the smaller underwing engines required and also compensated for the additional weight of electrical components in the turbo- electric architecture. The STARC-ABL concept is an ongoing project at NASA with model renements and improvements [19]. Kim, Felder et al. [20] at NASA also considered a 350-passenger hybrid-wing body (HWB) aircraft (Boeing 777 class) called the N3-X, which used a turbo-electric propulsion system. Two large turboshaft engines drove generators that provided electrical power to the array of 14 motor-driven fans near the trailing edge of the upper surface of the HWB that also ingested the boundary layer. This design was found to achieve an energy reduction of 70–72% over the reference Boeing 777 aircraft. The HWB used lightweight, highly ecient superconducting motors, generators, and transmission lines. The study also points out that apart from fuel savings, turbo-electric propulsion systems oer a great amount of design and operational exibility because power production is separated from power consumption. This enables distributed propulsion to reduce total propulsor weight as well as synergize with boundary layer ingestion, further increasing propulsive eciency. These benets are not obtainable with discrete turbofans like those today or mechanical power distribution through gearboxes and shafts. Vegh et al. [21] designed a lithium-air battery-powered regional (114 passengers) aircraft. Their ad- vanced batteries assumed specic energy as high as 2000 Wh/kg, and due to the nature of lithium-air, gained weight as they discharged. The resulting designs matched the weight of the conventional aircraft when design ranges were restricted to 3000 km, but had a much higher aircraft weight at typical mis- sion ranges of 4300 km. These weight increases proved infeasible (the sizing loop did not converge) with low battery specic energy assumptions. The authors concluded that merely replacing the conventional propulsion system with an electric one did not result in a feasible aircraft. However, for smaller-range mis- sions, the electric designs weighed about the same as the conventional reference. Unconventional concepts like blended wing-body (BWB) and designs taking advantage of distributed propulsion and boundary layer 13 ingestion could yield benets but also mass gains. They also identied operational issues that need to be addressed, for instance: should the battery be fast charging or easily replaceable so that it can be swapped out between ights? Furthermore, lithium-air batteries gain mass as they discharge, which aects the aircraft center of gravity as well. Stückletal. [22] studied an all-electric 68-passenger aircraft, dubbed Voltair, that also used high specic- energy lithium-air batteries and superconducting motors. The aft-fuselage propulsor also used boundary layer ingestion. It also incorporated improvements like a laminar ow wing, winglets, and a fuselage with low slenderness ratio to achieve a potential 25% improvement in energy eciency over the reference con- ventional aircraft. However, Voltair weighed roughly twice as much as its conventional counterpart for the same design mission. In addition, since the mass of the propulsion increased exponentially, a mass limit was identied beyond which the aircraft became infeasible. Pornet and Isikveren [23] at Bauhaus Luftfahrt designed a conceptual hybrid-electric aircraft carrying 180 passengers. The aircraft had two underwing geared turbofan engines supplemented by two underwing motor-powered electrical fans. They found that increasing hybridization would reduce fuel burn, in spite of an increase weight, while resulting in an optimized design that ies at lower altitudes (31,000–34,000 ft compared to the baseline at 35,000 ft) and slower (Mach 0.67-0.70 compared to 0.76 for the baseline) depend- ing on the degree of hybridization for minimum fuel burn. The hybrid-electric designs performed better at low-range missions, and their benets diminished with increasing range. Hybrid-electrics also performed better with substantially improved electrical components, including high-specic energy batteries, and superconducting motors and wires. Isikveren et al. [24] also studied a larger aircraft (Boeing 787 class) that utilized high-specic energy batteries and superconducting motors to achieve zero emissions during ight. This aircraft, called the Ce- Liner had a takeo mass 50% higher than the reference conventional aircraft, due to the large weight of the batteries compared to conventional fuel and due to structural changes required to support and lift that 14 excess weight. Furthermore, its range was signicantly reduced compared to the reference aircraft. The authors estimated parity in cash operating costs for the Ce-Liner in some energy cost scenarios, however, they concluded that due to operational costs like airport landing fees being calculated based on aircraft weight, the overall cost to operate the Ce-Liner was higher compared to an advanced conventional ref- erence. While achieving zero emissions, the costs to operate the Ce-Liner did not compare favorably and the authors suggested using the results as a foundation of renements in component and advanced design studies. Steiner et al [25] looked at the eects of distributed propulsion (DP), enabled by electrication for a 180-passenger all-electric aircraft ying a 900 nmi mission. They found that increasing the number of propulsors from two to four, six, eight, and ten reduced the overall weight of the aircraft, with the reduc- tions diminishing as the number of propulsors increased from six to eight. All of the DP designs achieved a range increase over the baseline with two propulsors. The range benet also decreased with increasing DP, due to the additional interference drag from the nacelles. Their hybrid-electric designs were com- pared against an advanced conventional aircraft that factored in potential future advances in propulsion system eciency and lighter materials. In another paper, Steiner et al [26] looked at various boundary layer ingesting (BLI) congurations that distributed propulsors along the upper aft fuselage, along a split wing, circumferentially along the aft fuselage, or embedded along the wing trailing edge. Their best con- guration, the one with circumferential aft fuselage propulsors, led to a range increase (proxy for energy savings) of about 10% with maximum BLI and about 3% with less BLI. Several of the aforementioned aircraft studies reach similar conclusions regarding electrication and electried aircraft, which can be summarized as: 1. Electried aircraft could outperform conventional aircraft at low ranges once electrical component technologies improve. Their benets diminish with increasing range, or with more conservative technology assumptions. 15 2. Electried aircraft could achieve lower fuel burn than their conventional counterparts at design ranges for larger aircraft, but consume more on-board energy, when accounting for the energy stored in the fuel and in the battery. 3. Electrication can take advantage of synergies between aerodynamics and propulsion through dis- tributed propulsion and boundary layer ingestion, more research into synergistic benets is needed. All of the aforementioned studies look at point designs of electried aircraft. In addition, the electried aircraft themselves are very dierent in terms of propulsion system architecture, even for similar missions. On their own, the point designs oer valuable insight: which aspects of their design allow for an energy consumption benet over the reference conventional aircraft and what challenges need to be addressed in order to develop these aircraft from concepts to in-service products. However, the choice of architecture, for the most part, relies heavily on the assumptions for technological parameters. Optimistic estimates of battery specic energy enable quite large all-electric aircraft. More conservative battery numbers mean all-electric aircraft are not considered and hybrid- or turbo-electric designs are used. Due to the fact that these are point designs, the propulsion architectures are chosen as opposed to optimized. The design space of dierent propulsion system architectures oers much more. Exploration of the design space can identify broadly what architectures would lead to more benecial electried designs in dierent missions, and narrow down the design space before moving onto more detailed point designs. In order to traverse this broad design space, a propulsion system that can model all architectures, from conventional, to turbo-electric, to hybrid-electric, to all-electric needs to be developed. With this unied propulsion system, the range of architectures can be analyzed to determine the best architecture for each mission, and identify and analyze potential sweet spots for electrication. Which type of electried aircraft oers more benets and by how much with dierent technology scenarios? How can the dierent designs exploit the benets from distributed propulsion and boundary layer ingestion? If the design space 16 is explored with a unied propulsion system, potential sweet spots for electrication can be identied and why these sweet spots exist can be examined. 1.4 UniedPropulsionSystemandElectricalComponents Figure 1.5 shows the unied propulsion system model that we developed as part of a NASA LEARN3 project ∗ carried out from 2016 to 2018 [11]. This unied model represents all propulsion system archi- tectures, from a conventional one with gas turbines powering fans mechanically through shafts, to an all-electric one with the battery powering fans through a chain of converter and motor. The level of electrication is represented through two parameters: one based on power split at the source and another based on power split at the load. The source electrication factor, f S , quanties the ∗ NASA/MIT collaborative agreement NNX16AK25A. Mechanically-Powered Propulsors Gas Turbines N fan E N fan M N turb bat Electrically-Powered Propulsors turb fan M P K M P K E Mechanical Source P turb P bat Electrical Source Battery System f S = P bat P bat +P turb f L = P K E P K E +P K M Source (Consumed Power) Load (Useful Power) Mechanical Load Electrical Load gen= conv mot P inv P mot P fan E mot inv fan E P fan M P gen P conv Link P link Figure 1.5: Unied propulsion system model 17 fraction of power supplied by batteries (electrical source),P bat , to that of the total power from both batteries and hydrocarbon fuel (mechanical source),P turb . It is dened as f S = P bat P bat +P turb : (1.2) Conventional and turbo-electric aircraft havef S = 0, hybrid-electrics havef S between 0 and 1, and all- electrics havef S = 1. Useful load power is quantied by the mechanical ow power delivered by the propulsors,P K , as de- ned in Drela’s power balance method [27]. Electrication of the load relates the ow power delivered via mechanically-driven propulsors (mechanical load),P K M , and via electrically-driven propulsors (electrical load),P K E . The load electrication factor,f L , is dened as f L = P K E P K E +P K M ; (1.3) where the denominator represents total power required by the propulsors. A conventional aircraft has f L = 0, partial turbo-electric and hybrid aircraft havef L between 0 and 1, while fully turbo-electric and all-electric aircraft havef L = 1. Load electrication thus distinguishes between partial and fully turbo- electric, as well as between hybrid- and all-electric architectures. The entire design space of electried propulsion architectures can thus be described by the two parametersf S andf L , each of which is set to a value between 0 and 1, thus providing a unied view of both conventional and electried propulsion systems. In the unied view of the propulsion system illustration of Fig. 1.5, the leftmost part shows the energy sources: hydrocarbon fuel (powering gas turbines) and battery system. From these, power ows to the load on the right end of the gure, comprised of mechanically- and/or electrically-driven propulsors (typically ducted fans). When a fan is powered by a turbine via a shaft, as in the top part of the gure, the propulsor 18 Table 1.1: Propulsion system architectures represented by the unied view and their dening parameters: load and source electrication factors,f S andf L . f L # f S ! 0 (0; 1) 1 0 Conventional Parallel Hybrid All-electric (0; 1) Partial Turbo-electric Series/Parallel Partial Hybrid 1 Fully Turbo-electric Series Hybrid is referred to as mechanically powered. When a fan is powered by a motor (either via a battery or a turbine+generator), as in the bottom part of the gure, the propulsor is electrically powered. When a zero value is used for one of the factors, the corresponding component is massless and thus removed from the system. For instance, iff S = 0, no batteries are carried on board. Iff L = 0, there are no electrically-driven fans. A combinedf S =0 andf L =0 represents a conventional aircraft. The specic values of (f S ;f L ) that dene each particular architecture are given in Table 1.1. This unied propulsion system model has mechanical and electrical sub-systems, shown on the top and bottom portions of Fig. 1.5 respectively, optionally linked via a generator. The mechanical part consists of sets made of one gas turbine connected to one mechanically-powered propulsor via a shaft. There can be a number N fan M of these sets. For a conventional aircraft, this is the entire propulsion system, and usually,N fan M is one or two. The electrical part consists of a battery system (essentially a large battery pack) connected to a converter that provides power toN fan E motors, which in turn driving one fan each. For an all-electric architecture (f S =f L = 1), the top half is removed and there is no generator to function as a link. The level of distribution for a distributed propulsion system is characterized by the total number of fans,N fan E +N fan M . A turbine can be used to send power via an electro-mechanical conversion link to the electrical part, for either recharging the batteries or distributing power to a range of electrical motors+fans. In this scenario, the electrical machine in the link functions as a generator and power ows from top to bottom sub-parts in the gure. Conversely, a battery can be used to augment the power to a mechanically-driven fan, in which 19 case power ows from bottom (electrical) to top (mechanical) parts, and the link functions as a motor. Thus, the sub-systems are connected by an electrical machine that can transfer power in both directions. For turbo-electric architectures, gas turbines power both mechanically-driven propulsors and either the generator (partial turbo-electric) or just the generator (fully turbo-electric). In either case, the generator drives motors and electrically-driven propulsors, and there is no battery (f S = 0). Since there are no mechanically-driven propulsors in a fully turbo-electric architecture,f L =1. For hybrid-electric architectures, both the gas turbine and the battery provide energy. All parts of the model are activated, andf S andf L both take on values between 0 and 1. Not shown in Fig 1.5 are the wiring and the thermal management system (TMS). The wiring is conguration- specic, and depends on how far the electrical components are placed from each other on the aircraft. For example, in the case of the conceptual Zunum hybrid-electric aircraft, the battery packs are integrated into the wing, while the propulsors are mounted onto the aft fuselage [28]. So, the wiring would have to run the length of the aft half of the aircraft to deliver power to the propulsors. In addition, electried designs use components like batteries and motors, which release heat under operation. A suitable thermal management system (TMS) needs to be considered for the electrical part of the system to account for this waste heat. Including the TMS adds additional mass to an electried propulsion system, and takes up volume in the aircraft, both of which also need to be accounted for. In contrast, any thermal management required for a conventional system is assumed to be lumped with the gas turbine itself, both in terms of mass and volume. 1.5 PerformanceMetric A suitable metric is needed to compare various aircraft with dierent propulsion system architectures. The energy use in the present work is dened in terms of on-board energy storage, but another possibility 20 considered would have been to expand our control volume for the energy usage to include the fuel needed, in a ground based power generation, to fully charge the battery. This would result in a disadvantage for battery-powered aircraft if the batteries were recharged with the existing power grid, which derives a portion of its energy from hydrocarbon sources like coal and natural gas. In 2017, 63% of the electric- ity generated in the US was produced from fossil fuels (coal, petroleum, and natural gas), and only 17% from renewable sources (hydro, geothermal, wind, and solar) [29]. When taking into account the energy required to recharge batteries on the ground, or to produce and later recycle them, using electried propul- sion systems may not have a net lifetime emissions benet over conventional propulsion. Encouragingly, trends in the US show that the percentage of electricity generated from renewable sources is growing, from 9% in 2000 to 17% in 2017 [29]. Assuming this trend continues, recharging batteries might produce less emissions in the future. In addition, the electric grid and its greenness also varies between geographical regions, which further complicates the emissions analysis. For example, a ight between Philadelphia, PA and Vancouver, BC would have dierent emissions from recharging the battery at Philadelphia versus recharging at Vancouver on the return leg. However, electrical power generation is not necessarily related to fuel usage, and as the power grid gets greener (e.g. with energy sourced from wind or solar), energy required to recharge batteries will con- tribute less emissions than what would result from producing and burning hydrocarbon fuel. In the most optimistic scenario, fossil fuels are no longer used, and the energy cost to produce synthetic hydrocarbons would be counted against the fuel energy in the larger system control volume. Considerations related to production of electrical energy to charge the batteries on the ground and the corresponding chain for hydrocarbon fuel delivery are outside the scope of this work. In other words, a “well-to-wake” analysis is not pursued for this work. Instead, the metric chosen is on-board energy and the battery at the start of ight is taken to be fully charged. 21 We will compare performance for dierent missions by using the productivity-specic energy con- sumption (PSEC) as a performance metric. It is dened as the on-board mission energy per payload mass per range PSEC = m fuel h fuel +m bat BSE m PL R ; (1.4) wherem fuel ,m bat , andm PL are the masses of the fuel, battery, and the payload respectively;h fuel =43 MJ/kg is the specic energy of hydrocarbon fuel; BSE is the pack battery specic energy; andR is the mission range. The product m fuel h fuel represents the energy stored in fuel, and m bat BSE the energy stored in batteries. Thus, PSEC is a measure of the on-board energy required to bring passengers from point A to point B, and accounts for how eciently the energy is used during ight to perform that mission. 1.6 ThesisScopeandOrganization While there are electried aircraft point designs studied by teams that oer benets over conventional aircraft, a tool that could evaluate a multitude of electried architectures in terms of feasibility and quantify potential energy savings. It would allow studying the trends and trade-os of the various design choices, to extensively explore the design space, and identify promising regions for electrication. The development of propulsion component models that, put together into a framework, can simulate the performance of these components over variable-power mission segments is the goal of this doctoral thesis. Electried aircraft are, of course, highly dependent on the electrical components used to build them. Therefore, it becomes important to model these too. Simple power-to-mass based models oer insight in terms of the general trends, but more detailed models are required to capture electrical component behavior across varying power demands during dierent ight phases, while taking physical, operational, and safety considerations into account. To reect these behaviors, modeling beyond power-to-mass based sizing methods becomes necessary. Detailed electric component models could then be fed back into a mission 22 analysis scheme to evaluate electried aircraft across the design space while oering higher-accuracy quantication of the potential benets. Thus, the overall scope of this work is to build a unied propulsion system model that represents the breadth of electried architectures and whose level of delity is sucient to capture key aspects of electrical component behavior under dierent ight phases, and determine to a high level of certainty whether electrication provides a net benet in energy consumption. It also becomes important to understand the trends in technology levels in order to see what electrical components could look like in the future, around the time where the projected aircraft are envisioned to enter into service. Additionally, it is useful to determine the technology level that is needed in order for some missions to be feasible and benecial. Chapter 2 looks at data from a variety of sources: aircraft studies, ongoing research, planned developments, and current aircraft in an attempt to quantify the major parameters that characterize electrical components: specic energy, specic power, and eciency. Cur- rent values for these are tabulated, and predictions for 2035 are made under conservative and optimistic assumptions. In essence, with technology level, the forward question of what a certain level of technology enables, as well as the inverse question of what technology level is needed to make electried aircraft benecial are addressed here. A rough analysis was done using a low-delity framework, and is presented in Chapter 3. This was part of the NASA LEARN3 project[11], a collaboration between teams at USC, MIT, and Aurora Flight Sciences (now a Boeing company), which set about exploring the feasibility of electried aircraft. Dierent aircraft were modeled across a variety of missions (payload and range) for a projected entry-into-service in 2035. Conventional aircraft of dierent classes, from a 20-passenger, 500 nmi-range thin haul commuter aircraft to a 350-passenger, 6000 nmi long range aircraft were compared with comparable electried aircraft, which consisted of turbo-electric, hybrid-electric, and all-electric designs. A range of technology scenarios were also considered, including current, as well as conservative and optimistic 2035 predictions. The eects of 23 technologies like distributed propulsion (DP) and boundary layer ingestion (BLI) were examined and their benets quantied. For the analysis of this broad tradespace of aircraft congurations, the LEARN3 work involved mod- eling aircraft components at low delity and in cruise-only operation. Sizing correlations for structural components were used, while simple energy-to-mass and/or power-to-mass ratios were used to size the electrical components. This level of delity provided insight into major trends, but there were several el- ements of electried propulsion systems not captured in the low delity analysis. Considerations such as the nonlinear natures of battery discharge, converter and motor operations are needed for a more accurate picture of how electried propulsion systems and their components behave under loads. The higher-delity models of Chapter 4 aim to address such considerations. An all-electric propul- sion system is modeled, consisting of a chain of battery, converter, power distribution system, and motor, which powers a propulsor. Higher delity models of batteries, converter, power distribution, wiring, and motors, beyond simple energy-to-mass and power-to mass scalings address the aforementioned behavior of these components under simulated ight electrical loads. These models are subsequently integrated into the propulsion system. In addition, the thermal management system (TMS) is also modeled. The overall electried propulsion system is then simulate and the results analyzed to learn about how the operational behavior of these components changes as the power demands vary during dierent ight segments. The ineciency of these components, i.e. the dierence between unity and their eciency, also plays a major role in sizing the battery and the TMS. Therefore, the eciencies at o-design ight-segment power loads are also analyzed. 24 Chapter2 ElectriedPropulsionTechnology 2.1 TechnologyParameters A major performance driver for electried aircraft is the mass of the electrical components. The electrical part of the propulsion system is modeled as a chain of components starting with the energy source (battery) and going to the electrical load (fans). For each electrical fan, an inverter converts the direct current from the battery to alternating current required to power the motor, which drives the fan to propel the aircraft. For the approaches used in Chapters 3 and 4, components masses are calculated based on the energy or power that needs to be supported by the components while in operation. The battery, which works both as an energy and power source, is characterized by both a specic energy and a specic power. The inverter and motor are dened by their specic power and eciency. The eciency is also used in sizing the thermal management system (TMS). The components commonly in use today have low specic energy and specic power values. They are not specically targeted for aerospace applications, but the trends in the future look encouraging [30]. Additionally, for the higher-delity approach in Chapter 4, operational considerations are taken into account. For example, battery discharge is nonlinear, and as more power is drawn from the battery, it drains faster. Dierent ight segments like takeo and climb have a higher power requirement than cruise and descent. The higher-delity models capture these variable power draws across dierent ight segments as 25 well as their impact on battery discharge, and consequently, the required battery capacity. Similarly, the eciency of electric machines (motors) and power electronics (converters) depends on their power loads as well, and changes when the power diers from their rated power. The higher-delity approach models this behavior as well. Furthermore, other considerations like TMS and wiring are handled to greater detail. To determine the eect of technology on electrication, three technology levels are considered: cur- rent, conservative 2035, and optimistic 2035. In the subsequent sections, a rationale and values for the specic energy, power, and eciencies of electrical components used throughout this work for these tech- nology levels are presented. 2.1.1 Batteries Battery specic energy (BSE) is dened as the energy per unit mass stored in a battery. The theoretical BSE, BSE th , is set by the electrochemical reactions that generate the energy, and thus is based on the mass of the reactants only. In practice, batteries are made up of several cells. Cell manufacturers quote the nominal capacity,Q nom nominal voltage,V nom and mass of the cell,m cell , from which the BSE at cell-level is calculated as BSE cell = V nom Q nom m cell = cell BSE th : (2.1) This cell-level BSE includes the mass of electrodes (usually graphite), current collectors, electrolytes, sep- arator, and binders, in addition to the mass of the reactants. In a lithium-ion cell, the lithium only makes up about 3% of the mass, whereas the cathode and anode together make up between 75–85% of the cell mass [31]. Including these essential non-reactant parts to produce a useful cell from the chemical reaction meansBSE cell is lower thanBSE th , and can be expressed as an eciency factor, cell , (less than unity) multiplyingBSE th . The cell-level BSE is usually what is quoted in the literature. Figure 2.1 shows this dierence between theoretical and cell-level BSE in some battery chemistries in use today [32]. 26 [mature] [mature] [mature] [Boeing 787 APU] [Airbus E-Fan] [Tesla Model S] Figure 2.1: Theoretical and cell-level BSE for several battery chemistries. A battery, made up of several cells arranged in a pack, has additional mass due to the packaging struc- ture and the battery management system. Current high-energy-density battery chemistries that include lithium-ion are susceptible to thermal runaway, caused by dendrite formation across the separator, poten- tially leading to a short circuit and subsequently a large heat release, further causing thermal runaway in neighboring cells, with the potential for a re. To prevent this, a battery management system (BMS) is needed. The BMS monitors each cell for dendrite formation and isolates damaged cells to prevent thermal runaway, so its inclusion adds mass to a battery, thus lowering the BSE further at pack level. The additional mass is included in a BSE dened at the pack level, which is the relevant parameter for electried aircraft system level considerations, and is dened as BSE pack = pack cell BSE th = pack BSE cell = N cell m cell m bat BSE cell ) BSE pack = N cell V nom Q nom m bat ; (2.2) 27 [mature] [mature] [mature] [Boeing 787 APU] [Airbus E-Fan] [Tesla Model S] Figure 2.2: Cell BSE as a percentage of theoretical BSE for dierent battery chemistries in use. wherem bat is the battery mass,N cell the number of cells in the battery pack, and pack the packing e- ciency. Mature batteries such as nickel-cadmium and nickel-metal-hydride achieve a cell eciency of less than 40%, as seen in Fig. 2.2 [32]. Lithium-ion chemistries, which are less mature, exhibit cell eciencies of around 30%. The Li-ion batteries used in the Boeing 787 and the Tesla Model S have cell eciencies of 28% and 36% respectively; their pack eciencies are 78% and 59% respectively [33, 15]. The all-electric Airbus E-Fan aircraft achieves a higher pack eciency of 84% [8]. Pack eciencies are expected to improve as cell manufacturers focus on improving not just cell chemistries of batteries, but also packaging materials. Given mature-battery cell eciencies of 40% and pack eciencies of around 80%, a combined pack cell of more than 33% seems unlikely to be achieved. Therefore, current lithium-ion chemistries are unlikely to yield a pack-level BSE of more than 250 Wh/kg in the 2035 timeline, down from the 740 Wh/kg theoretical value. Novel lithium-ion chemistries, like lithium sulfur (Li-S) and lithium-air (Li-air), with theoretical BSE values of 2600 Wh/kg and 3500 Wh/kg respectively, can potentially provide a substantially higher BSE. 28 Li-air cells have been demonstrated to reach a BSE of 778 Wh/kg [34], which translates to a pack value of 540 Wh/kg assuming a pack eciency of 70% (representative of a less-mature design). In terms of power, a battery is characterized by the battery specic power (BSP), dened as the power available per unit mass. Dierent batteries have varying discharge proles depending on how they are used. As a result, it is dicult to place a constant value on BSP, which varies nonlinearly with loads. Mathematically, the relationship between the energy and power in a battery can be described via the Ragone relation [35] as: P load = V 2 0 R i E load E tot 1 E load E tot ; (2.3) whereV 0 is the open-circuit voltage,R i is the internal resistance,E load andP load are the energy available and the power delivered to the load respectively, andE tot is the total energy stored in the battery. The energy delivered to the loadE load is always less than the total energy storedE tot since some energy is lost to overcome internal resistance during discharge, and because batteries cannot be fully discharged due to life-cycle issues. Dividing each of the energy and power terms by mass expresses the equation in terms of BSE and BSP. The relationship between BSE and BSP as the battery discharges can be illustrated through a Ragone diagram, as shown in Fig. 2.3. For low-power applications, the battery specic energy is highest, but as the the power draw increases, the BSE decreases drastically. For high-power applications, BSE is several orders of magnitude lower than that for low-power applications; hence, a much larger battery is needed. Therefore, for aircraft applications, where the discharge proles (based on power draws) vary during dif- ferent stages of ight, BSE and BSP cannot simply be taken as constants. A way to model this nonlinear discharge behavior is by non-dimensionalizing the Ragone relation, as presented by Kuhn and Sizmann in [35]: 29 1000 100 Specific Energy (Wh/kg) Specific Power (W/kg) 10 1 10 0 1h 10h Lead-Acid Ni-MH Capacitors Li-ion Fuel Cells 100h 0.1h 36s 3.6s 10 1 10 2 10 3 10 4 Source: Product data sheets Range Acceleration 2 4 6 2 4 6 2 4 6 IC Engine Figure 2.3: Ragone diagram showing the relationship between BSE and BSP, adapted from [36]. load = 4 load (1 load ) ; (2.4) where the energy ratio, load =E load /E tot , is a measure of the energy eciency of the battery. The power ratio, load = P load /P max , is the ratio of the power delivered to the load versus the maximum power that can be delivered by the battery. This non-dimensional relationship is shown in Fig. 2.4. The battery can deliver energy more eciently when the power drawn from the load is minimal: load approaches 1 as load goes to 0. At maximum power however ( load ! 1), the battery is only 50% energy ecient ( load ! 0:5). This represents a drastic change in the BSE and BSP as the load on the battery varies. The higher-delity battery model in Sec. 4.1 takes into account this change in BSE and BSP, and calculated the BSE based on the total energy required to deliver power at various loads throughout the ight. 30 0 1 0.5 1 ψ η Figure 2.4: Non-dimensional Ragone diagram showing the relationship between BSE and BSP. 2.1.2 ElectricMachines Electric machines, specically motors and generators, are rated based on power. Motors convert energy into shaft power, while generators convert shaft power into energy. Important considerations for aerospace applications include parameters like power, mass (and hence specic power), and eciency at low delity. For higher delity models, considerations like angular speed (rpm), voltages, currents, and physical di- mensions must also be taken into account. Motors are rated based on maximum power and continuous power. In theory, a motor can run indenitely at its continuous power setting as long as it is supplied with energy. The motor eciency is usually quoted at this continuous power. At maximum power however, the motor runs at a lower eciency. Running at maximum power for extended periods is also detrimental to the life of the motor and its components. There are various classes of motors, the two major ones being direct current (DC) and alternating current (AC). DC motors are simpler, cheaper, and technologically mature. AC motors are lighter, can run faster, and require less maintenance. Automotive applications use both DC and AC motors. While there are electried aircraft currently ying, these are small, and only a subset of them use custom-designed motors. Existing motors do not have the high power and low weight required for electried commercial aviation, but eorts to develop suitable motors are underway, as will be discussed in Section 2.2. 31 2.1.3 PowerElectronics The term power electronics encompasses converters and inverters, which are also sometimes referred to as motor controllers. Inverters convert direct current (DC) to alternating current (AC), and converters can switch from DC to AC as well as AC to DC. Batteries supply DC, while motors usually require AC, so a converter is required between the battery and the motor. Additionally, if mechanical energy is being converted to electrical energy through a generator, a converter (in this case, also called a rectier) is used to convert AC to DC in order to recharge the battery, for example. Important parameters to size power electronics for aerospace applications include power and mass (and hence specic power) and the conversion eciency at the very least. Higher-delity models need to take into account considerations related to voltages, currents, and physical dimensions. As is the case for motors, advances in the specic power of power electronics are needed for aerospace applications. Advancements in materials as well as in switching topologies in converters will facilitate aerospace-grade power electronics, and work is underway to develop such power electronics, to be discussed in Section 2.2. 2.1.4 Wiring In electried systems, wires connect various components and deliver power through them. Important parameters for wiring include the mass, voltages, and currents. The length of the wiring required is highly conguration-specic. For an aircraft design like the NASA STARC-ABL, with the underwing turbofan generators providing power to the aft propulsor through a motor, the wiring has to run through the aft- half of the aircraft. For an aircraft like the NASA X-57 Maxwell, where the aircraft itself is smaller and has batteries closer to the wing-distributed motor and propulsor array, the wiring only has to run through the wing. 32 In low-delity models, the wiring can be lumped within each electrical component. In high-delity models, the length, cross-sectional area, and eciency become important, as does the wire mass per unit length or mass per unit volume (density). 2.1.5 ThermalManagement Each of the electrical components discussed previously has an eciency associated with power conversion or transmission. The wasted energy is released as heat. Managing the heat and maintaining each compo- nent at operating temperatures requires a thermal management system (TMS). Important parameters for TMS include the heat energy to be rejected per unit time (units of power) and the mass (hence the specic power). High-delity models also need to take into account allowable temperatures, allowable speeds for favorable laminar ow regimes, and physical dimensions. The heat exchanger adds to the aircraft mass, just as the wiring does. 2.2 TechnologyPredictions 2.2.1 Batteries A literature survey of BSE values used in conceptual hybrid-electric and all-electric designs is illustrated in Fig. 2.5, which shows cell-level BSE predictions plotted against the envisioned entry-into-service (EIS) year of the proposed aircraft. Conceptual designs predict cell-level BSE values between 500 and 2000 Wh/kg in the 2030s decade [5, 8, 14, 16, 17, 18, 23, 24, 25, 33, 37, 38]. Although in most cases, the authors do not mention the specic battery chemistries, it is assumed these predictions are for novel lithium chemistries. Tesla Motors [33] predicts a growth rate of between 5% to 8% per year, which, starting from a 2012 value of 265 Wh/kg for the Model S car, leads to a value ranging from 700 to 1300 Wh/kg in 2035. Friedrich and Robertson [18] predict an improvement of 5.5% each year, leading to a value of around 600 Wh/kg based 33 Lab demonstrated for Li-air 40% of theoretical for Li-ion 40% of theoretical for Li-air Figure 2.5: Summary of cell-level BSE projections in previous studies on a reference of 200 Wh/kg in 2014. Mostly, the values quoted in the literature are at cell-level, and thus lack the pack eciency factor, pack , needed when considering batteries at system-level. Using a theoretical value for BSE of 740 Wh/kg for current lithium-ion chemistries (like the LiNiCoAlO 2 in the Boeing 787), and an overall eciency ( pack cell ) of 33%, the pack BSE can be expected to reach about 250 Wh/kg in 2035. Using the same process for novel lithium chemistries, like Li-S with a maximum the- oretical BSE of 2600 Wh/kg, the pack BSE could be as high as 900 Wh/kg in 2035. These are the values that are assumed for the conservative and optimistic 2035 technology scenarios. 2.2.2 ElectricMachines Figure 2.6 shows a range of electric machine (motor) designs as represented by their continuous power plotted versus specic power on a semi-logarithmic scale. Commercial-o-the-shelf (COTS) motors have specic powers of around 2 kW/kg and are rated in the 100 kW range. Current electried aircraft generally 34 use COTS motors, and only rarely custom-made. Motors for mid-size and larger electried commercial aircraft need to be in the megawatt class, if they are to simply replace conventional propulsion systems. Aircraft studies predict a large range of motor specic powers across motors of various sizes: some of the studies utilize distributed propulsion with a higher number of smaller motors, whereas others augment or replace a convention propulsion system with the same number of motors. A conservative 2035 estimate by the National Academies of Sciences, Medicine, and Engineering (NAE) report predicts motor specic powers of 9 kW/kg, rated at powers of 2 MW [4]. However, NASA is cur- rently funding motor research for aerospace applications with motors rated between 1.0–2.5 MW at specic powers of up to 16 kW/kg [30], some of which are superconducting but self-cooled (i.e. the cooling system is an integral part of the motor). These levels are expected to be achieved on test-beds in the near term. These numbers are used as the conservative and optimistic 2035 technology estimates respectively. Superconducting motor designs are predicted to reach both higher specic powers, as shown by the boxed data points in Fig. 2.6, and higher eciencies. However, the needed cryocooler adds complexity, and reduces the specic power at the overall system level. As a result, superconducting designs which require cryocoolers have not been considered here. Finally, with respect to eciency, current motors are about 95% ecient, and this eciency is projected to grow to 98–99% by 2035 [4, 30]. 2.2.3 PowerElectronics Similar extrapolations have been made for power electronics (converters and inverters), as shown in Fig. 2.7. Existing products have power and specic power levels of 200 kW and 2.2 kW/kg respectively, and are not suitable for aerospace applications. Aircraft studies assume much higher powers and specic powers than what is available today, with megawatt class converters at specic powers of 16 to 25 kW/kg. Some of these designs are superconducting but do not take into account the mass of the cryocooler, so 35 Figure 2.6: Survey of electric machine specic power versus continuous power; existing machines and future design projections. the specic powers will be lower once the cryocooler is accounted for. Just like with electric machines, superconducting power electronics designs are not considered for this work. Conservative power electronics parameter values are predicted by the NAE report to be 500 kW power and 9 kW/kg specic power by 2035. The NAE report expects improvements to come through advances in component materials, switching components and topologies, and passive components like transformers, packaging, and thermal management systems [4]. However, there are ongoing projects that promise better specic powers. Based on the projects currently funded by NASA [30], optimistic estimates have been set at 19 kW/kg rated at 1 MW. These values have been demonstrated at testbed level, and can be reasonably expected to achieve the predicted numbers at system-level in 2035. Finally,current power electronics are about 95% ecient, and are predicted to reach 98–99% eciency by 2035 [4, 30]. 36 Figure 2.7: Summary of power electronics specic power versus continuous power; existing designs and future design projections 2.3 TechnologyScenarios Table 2.1 summarizes the current and predicted 2035 technology scenarios that are used in the present work to assess the eect of technology level on the feasibility of electried aircraft. The current battery specic energy (BSE) was taken from the all-electric two-seater Airbus E-Fan. Conservative 2035 BSE assumes no new breakthroughs in battery technology, whereas the optimistic value assumes novel battery chemistries that are made rechargeable and commercialized by that time, as discussed in Section 2.2. In all three cases, the battery specic power (BSP) is assumed to scale with the BSE based on the BSE/BSP ratio for the NASA X-57 Maxwell [39]. For the other electrical components, the current and conservative specic powers and eciencies are taken from the NAE report [4]. Ongoing NASA-funded research has demonstrated much higher numbers at testbed level [30], and it is reasonable to expect that these numbers will be achieved at the system 37 Table 2.1: Technology scenarios: assumptions values for electrical component parameters. Parameter Units Current Conservative 2035 Intermediate 2035 Optimistic 2035 Pack BSE Wh/kg 175 250 575 900 Pack BSP W/kg 520 745 1720 2700 Motor Specic Power kW/kg 2 9 12 16 Converter Specic Power kW/kg 2.2 9 14 19 Eciency [–] 0.95 0.96 0.98 0.99 level in 2035: these numbers form the basis of the optimistic 2035 assumptions. To form what we call the intermediate 2035 assumptions, we take the average of the conservative and optimistic 2035 assumptions. 38 Chapter3 Low-FidelityAnalysis 3.1 ModelsandMethodology This portion of the work was done as a part of a NASA LEARN3 collaborative project between USC, MIT, and Aurora Flight Sciences (now a Boeing company). A modular approach is taken to size and integrate an aircraft and evaluate its performance over a given mission. The aircraft model consists of four modules, as shown in Fig. 3.1. Each module “builds up” a part of the aircraft based on the mission requirements and constraints. The propulsion system takes into account whether the architecture is conventional or electried, and the amount of distribution, to determine the mass and size of its components. Components are sized based on constant specic energy and/or specic power values. The mission integration module takes the mission requirements of payload, range, and cruise speed, and uses a generalized range equation to calculate takeo mass, power required,etc. The airframe mass is calculated from the mass buildup in the aero-structure module. Sizing correlations are obtained from the textbook by Raymer [40] and from the Propulsion System Aero-Propulsive Performance Aero-Structural Sizing Mission Integration Payload Range Cruise Speed Fuel flow Battery power draw Propulsion system mass Source electrification, fS Load electrification, fL Airframe drag buildup Amount of BLI Lift-to-drag ratio Fuselage geometry Wing loading Propulsive power Propulsor mass flow Airframe mass Figure 3.1: Framework overview: modules and their interactions [11]. 39 TASOPT software developed by Drela at MIT [41]. The aero-propulsive performance module determines the drag and includes the eects of boundary layer ingestion (BLI). A more detailed description of the modules can be found in the LEARN3 Final Report [11] and the conference paper published at the end of the project [42]. The bi-directionality of the arrows in Fig. 3.1 indicates that information ows both ways between modules. The modules are integrated into a design framework using the GPKit geometric programming optimization tool [43], with the chosen metric, the productivity-specic energy consumption (PSEC), as the objective function. Using GPKit allows specication of the problem as a set of variables and constraints. GPKit then solves the problem by minimizing PSEC, returning the values for the variables of the optimal design and their sensitivities. For a given technology level, mission requirements, amount of BLI and distribution, the framework thus produces an aircraft that minimizes the onboard energy as quantied by PSEC. 3.1.1 PropulsionSystem At the aircraft level, the propulsion system is characterized by its power and its mass. A thermal manage- ment system is included to handle the heat rejected by each component. Power The power levels at the individual junctions of the unied propulsion system are labeled in Fig. 1.5, repro- duced on the next page for easier reference. and the power analysis is done based on the power balance method [27]. The total mechanical power delivered to the ow by the propulsion system is P K =P K M +P K E ; (3.1) 40 whereP K M andP K E are the mechanical powers delivered to the ow by the mechanically- and electrically- driven propulsors respectively. Denoting byN fan E andN fan M the number electrical and mechanical propul- sors, respectively, the ow powers are given by P K E =N fan E fan P fan E (3.2) P K M =N fan M fan P fan M ; (3.3) whereP fan E andP fan M are the per-propulsor shaft powers, and fan is the fan eciency, assumed to be the same for all fans. Mechanically-Powered Propulsors Gas Turbines N fan E N fan M N turb bat Electrically-Powered Propulsors turb fan M P K M P K E Mechanical Source P turb P bat Electrical Source Battery System f S = P bat P bat +P turb f L = P K E P K E +P K M Source (Consumed Power) Load (Useful Power) Mechanical Load Electrical Load gen= conv mot P inv P mot P fan E mot inv fan E P fan M P gen P conv Link P link 41 Down the electrical chain, each electrical propulsor is assumed to be driven by a motor of eciency mot , and an inverter ∗ of eciency inv . We denote byP mot the power delivered to the motor by the inverter, andP inv the power delivered to the inverter, such that P fan E = mot P mot (3.4) P mot = inv P inv : (3.5) The power that reaches the inverters comes from the battery system and possibly from the mechanical source via the electro-mechanical energy-conversion link, i.e., N fan E P inv =P bat +P link : (3.6) On the electrical source side, the non-dimensional relationship between battery eciency, bat , and power,P bat , is given by the Ragone relation, as outlined in Section 2.1: P bat P max = 4 bat (1 bat ); (3.7) whereP max is the maximum power that the battery can deliver. This eciency represents the losses inside the battery and has the eect that the usable amount of energy eectively decreases as the battery is operated at higher power levels. When E tot and P max are divided by the battery mass, battery specic energy (BSE) and battery specic power (BSP) are obtained. ∗ In this work, any converter (inverter or rectier) is assumed to include a controller. 42 On the mechanical side, hydrocarbon fuel is burned at a rate _ m fuel to driveN turb gas turbines. Denoting byP turb the power output of each turbine, th their thermal eciency, andh fuel = 43 MJ=kg the jet fuel specic energy, the total power out of the mechanical source system is N turb P turb = th h fuel _ m fuel : (3.8) It is assumed that there are as many turbines as there are mechanically-driven fans, i.e.,N turb =N fan M . Finally, power may be directed from the mechanical source towards the electrical load, such that at the top junction of the electro-mechanical energy-conversion link of Fig. 1.5, we have N turb P turb = P gen + N turb P fan M ; (3.9) whereP gen is the power sent down to the generator. The amount of powerP link that exits the link, and which may be used to recharge the batteries in addition or instead of driving the electrical fans, depends on the converter eciency conv and the generator eciency gen , and is such that P link =N turb conv P conv (3.10) P conv = gen P gen : (3.11) Note that when power ows upwards from the electrical part towards the mechanical part, the powers owsP link ,P conv ,P gen are reversed and the eciencies inverted. 43 ThermalManagement Each electrical component is assumed to dissipate heat at a rate _ Q, which must be removed via the thermal management system (TMS). The total dissipated heat is _ Q = N fan M _ Q gen + _ Q conv + N fan E _ Q inv + _ Q mot + _ Q bat ; (3.12) where the heat dissipation of each component is determined by its power throughput and eciency as _ Q () = 1 () P () : (3.13) For instance, a 90% ecient component loses 10% of its power as heat. The gas turbines are assumed to include their own thermal management system, which is accounted for in the turbine mass. Mass The overall propulsion system mass is equal to the sum of its component masses, namely m prop =N fan M m turb +m gen +m conv +m fan M +m nace M (3.14) + N fan E (m inv +m mot +m fan E +m nace E ) + m TMS : (3.15) The masses of the turbines and fans (mechanically- and electrically-driven) are calculated from their re- spective mass ows following the cube-squared law, namely m turb =K turb _ m 1:2 turb ; m fan =K fan _ m 1:2 fan : (3.16) In the low-delity framework, values ofK turb = 45:6 andK fan = 1:3 are used (derived from TASOPT results[41]), with the units for the coecients as required to give dimensions of mass in the product. 44 Initially, a cube-squared scaling was used; however, an exponent of 1.2 was found to give a better t to existing TASOPT data. Nacelle masses are assumed to scale linearly with the propulsor mass ow as m nace M=E =K nace _ m; (3.17) where a value ofK nace = 4:56s (derived from TASOPT results[41]) was used in this framework. The generator, converter, inverters, and motors are assumed to have constant power densities, so their masses are determined from an assumed power-to-mass ratio P m as m () =P () P m 1 () : (3.18) The thermal management system (TMS) size scales with the heat ow, again via a power-to-mass ratio, so its mass is m TMS = _ Q P m 1 TMS : (3.19) The specic power values are set based on the technology level as given in Table 2.1. The mass of the wires are not explicitly included, but are assumed to be accounted for in the masses of the individual components. Power distribution wiring strongly depends on the aircraft conguration, and the placement of the dierent components within the airframe. Such level of detail is beyond the scope of the low-delty framework and trade-space analysis. It will be shown in Ch. 4 that wiring mass is a very small factor of the overall aircraft weight (0.5%2.5% depending on level of distribution), and therefore, neglecting it is a good approximation. 45 3.2 Results: ElectricComponentTechnologyLevels The framework developed was then applied to explore the design space with the goal of determining the particular missions where electrication has the most potential. In this section, the eects of technology level on the feasibility and performance of electried aircraft are examined, in order to quantify the ben- ets as technology improves. All-electric aircraft were shown to be feasible only at reduced ranges for all baseline cased considered [42, 11]. Therefore, to illustrate the eects of electried propulsion for all ar- chitectures, the performance of a reduced-range commuter aircraft is assessed as a function of technology parameters. The baseline conventional aircraft is modeled on a Viking Air Twin Otter, which carries 20 passengers over a range of 100 nmi. It is powered by two mechanically driven fans without BLI (f S = 0,f L = 0, and f BLI M = 0). The all-electric aircraft (f S = 1 andf L = 1) ies the same mission. Battery eects (BSE and BSP) and other component eects (specic powers of motors and converters) are considered separately. When one set of parameters is varied, the other technology parameters are set at the optimistic 2035 values from Section 2.3. 3.2.1 EectsofBatteryTechnology For this mission, the all-electric aircraft is not feasible with current and conservative 2035 battery technol- ogy levels. Figure 3.2 shows the eect of increasing BSE (and with it, BSP) on PSEC, with other component technology xed at optimistic 2035 levels. It should be noted here that xing the technology levels of other components at the intermediate 2035 values would not change the trends, as the system mass is much more sensitive to BSE than to other components’ specic powers. The conventional aircraft has a constant PSEC as BSE varies because it carries no batteries. In terms of conguration, the closest all-electric aircraft has two electric fans and no BLI, and is shown on the red dotted curve. When the BSE is under 350 Wh/kg, this all-electric aircraft is infeasible. Between 350–400 Wh/kg, the all-electric aircraft becomes feasible, but it 46 0 250 500 750 1000 1250 1500 1750 2000 Battery Specific Energy, BSE [W h/kg] 0 1 2 3 4 5 6 7 8 PSEC [kJ/kg km] Optimistic 2035 Current Tech Conservative 2035 Conventional, 2 fans, no BLI All-electric, no BLI 2 fans 100 20 Intermediate 2035 Figure 3.2: Eect of battery technology on PSEC with DP for 100 nmi all-electric (f S = 1, f L = 1) commuter aircraft assuming optimistic 2035 technology for other components ([P=m] mot = 16 kW/kg, [P=m] conv = 19 kW/kg). requires more energy (larger PSEC) than the conventional aircraft. As BSE increases further, the battery mass to carry the mission energy decreases, leading to a sharp drop in the PSEC. For the optimistic 2035 battery technology assumptions, the all-electric aircraft consumes about 37% less energy than the conven- tional aircraft. At higher BSE values, the PSEC curve attens out, since the battery mass becomes a smaller fraction of the aircraft takeo mass, and further increases in BSE provide diminishing benets in energy consumption. The all-electric aircraft becomes more benecial with a greater number of smaller-diameter fans. With distributed propulsion, the all-electric aircraft with more fans becomes feasible at lower BSE values. It also provides a larger PSEC reduction at a given BSE value. For the optimistic 2035 battery technology assumptions, the 20-fan all-electric aircraft provides a PSEC benet of about 40% over the conventional, up from the 37% with two fans. As with increasing BSE, however, increasing DP has diminishing returns: going from 20 fans to 100 fans provides less benets than going from two to 20. 47 0 250 500 750 1000 1250 1500 1750 2000 Battery Specific Energy, BSE [W h/kg] 0 1 2 3 4 5 6 7 8 PSEC [kJ/kg km] Optimistic 2035 Current Tech Conservative 2035 Conventional, 2 fans, no BLI All-electric, 2 fans, no BLI All-electric, 100 fans,50% BLI Intermediate 2035 Figure 3.3: Eect of battery technology onPSEC with DP and BLI for 100 nmi all-electric (f S =1,f L = 1) commuter aircraft assuming optimistic 2035 technology for other components ([P=m] mot = 16 kW/kg, [P=m] conv = 19 kW/kg). Figure 3.3 shows the range of propulsor congurations available for the all-electric aircraft: from no DP and no BLI to a case with 100 fans and 50% BLI. At optimistic 2035 battery technology, the all-electric aircraft with 2 fans and no BLI consumes about 60% less energy than the conventional baseline. When the design space is opened up to include massive distribution of fans and BLI, the benets are twofold: (i) the aircraft becomes feasible at smaller BSE values, and (ii) it oers even greaterPSEC reduction at a given BSE value. Furthermore, all-electric aircraft are feasible at reduced ranges with BSE levels only slightly above the conservative 2035 estimates. In addition, they provide a PSEC benet over conventional aircraft, and this benet increases with DP and BLI. Thus, DP and BLI facilitate electrication. Note that conventional aircraft with such high DP and BLI levels are not considered here because of the diculty in achieving those levels with only mechanical systems. 48 0 5 10 15 20 25 Component Specific Powers [kW/kg] 0 1 2 3 4 5 6 7 8 PSEC [kJ/kg km] Current Tech Optimistic 2035 Conservative 2035 All-electric,no BLI Conventional, 2 fans, no BLI 2 fans 100 20 Intermediate 2035 Figure 3.4: Eect of component specic powers onPSEC with DP for 100 nmi all-electric (f S = 1,f L = 1) commuter aircraft assuming an optimistic 2035 battery technology atBSE = 900 Wh/kg. 3.2.2 EectsofComponentSpecicPower Figure 3.4 shows the eects of increasing component (motors and inverter) specic powers on energy us- age. All-electric aircraft are feasible and benecial over conventional aircraft with current technology (but with optimistic battery technology). Again, the conventional aircraft has a constant PSEC as component specic powers improve, since it does not carry any converters or motors. For the all-electric aircraft, PSEC improves as specic powers increase. With 2 electric fans and optimistic battery technology, current component technology already provides a benet of about 25% over conventional. At conservative and optimistic 2035 values, this benet increases to 28% and 30% respectively. With DP enabled, the all-electric aircraft provides even greater PSEC benets, however, with dimin- ishing returns. Going from 2 fans to 20 provides a PSEC benet of about 3%, whereas going from 20 fans to 100 fans only provides a 2% further improvement for conservative 2035 numbers. 49 0 5 10 15 20 25 Component Specific Powers [kW/kg] 0 1 2 3 4 5 6 7 8 PSEC [kJ/kg km] Current Tech Optimistic 2035 Conservative 2035 All-electric, 2 fans, no BLI All-electric, 100 fans, 50% BLI Conventional, 2 fans, no BLI Intermediate 2035 Figure 3.5: Eect of component specic powers onPSEC with DP and BLI for 100 nmi all-electric (f S = 1, f L = 1) commuter aircraft assuming an optimistic 2035 battery technology atBSE = 900 Wh/kg. Figure 3.5 also shows the eect of DP and BLI on PSEC. At conservative 2035 values, an all-electric aircraft with 2 fans and no BLI oers a PSEC reduction of about 42% over conventional, which increases to 48% with 100 fans and 50% BLI. Little benet is obtained for values higher than 8 kW/kg since the components’ mass make up an increasingly smaller fraction of the aircraft takeo mass. The attened PSEC curve also suggests that the metric is less sensitive to component specic powers than it is to battery technology, This indicates that the obstacles for a feasible all-electric aircraft lie with battery technology, rather than with motors and converters. Following this conclusion, a turbo-electric architecture (no batteries) could be feasible even with cur- rent technology. Figure 3.6 demonstrates this, here for a larger medium-haul aircraft designed to carry 180 passengers over 3000 nmi, similar to a Boeing 737. The conventional aircraft has two mechanically driven fans with no BLI and a PSEC of about 4.2 kJ/kgkm. The turbo-electric aircraft has 308 electrically distributed fans, ingesting 20% of the total boundary layer over the fuselage and 50% over the wing. 50 0 5 10 15 20 25 Component Specific Powers [kW/kg] 0 1 2 3 4 5 6 7 8 PSEC [kJ/kg km] Optimistic 2035 Current Tech Conservative 2035 Conventional, 2 fans, no BLI Turboelectric Figure 3.6: Eect of improving component specic powers on PSEC for a medium-haul turbo-electric (f L = 0:48,f BLI M = 0:2) aircraft withN fan M = 2,d fan M = 1:14 m,f BLI E = 0:5,N fan E = 308,d fan E = 0:104 m. Even with current technology, the turbo-electric aircraft has a PSEC benet of 7% over the conventional aircraft. This advantage increases to 16% with conservative 2035 technology and to 19% with optimistic 2035 technology. Thus, while all-electric aircraft may be infeasible for longer missions, turbo-electrics are feasible for longer missions. This was demonstrated here for the medium-haul, but was found to be true for all classes, including commuter (like a Twin Otter), regional (like an Embraer E-175), and long-haul (like a Boeing 777) aircraft. 3.2.3 TechnologyLevelAnalysisSummary For the dierent mission proles, analysis of technology levels shows that current battery specic energy and specic power are too low to enable all-electric propulsions for all missions. Motor and converter specic powers are high enough to allow turbo-electric aircraft. However, these aircraft have little (< 51 5%) to no benet over the corresponding conventional aircraft. Benets are seen by adding distributed propulsion (DP) and boundary layer ingestion (BLI) enabled by electrication. At conservative 2035 technology levels, battery technology is still too low to render all-electric aircraft feasible, even for the 100 nmi commuter aircraft. BSE and BSP must increase signicantly for all-electric aircraft, even for the smallest class and shortest missions: for instance, the commuter aircraft becomes feasible for BSE values above 350 Wh/kg. However, conservative 2035 technology more than adequate to enable turbo-electric aircraft across all missions and provide PSEC benets over the conventional cases. With optimistic 2035 technology, the results for turbo-electric aircraft stand with greater PSEC reduction of as much as 17%, and battery technology also improves enough to enable hybrid- and all-electric commuter aircraft, albeit at lower ranges of up to 300 nmi for the commuter versus a design range of 500 nmi. Looking at the results from another angle, optimistic 2035 technology allows for all-electric commuter aircraft at a reduced range of 100 nmi with an energy consumption benet of 50% over the conventional case. The design mission of 500 nmi, however, requires a BSE 1.6 times higher than the optimistic 2035 numbers. For a design regional mission (80 passengers, 1500 nmi), the BSE would have to be twice the op- timistic prediction for 2035 and reach 1800 Wh/kg, so it is reasonable to assume that such an aircraft will not take shape in the future. For all-electric aircraft, battery specic energy is more important than spe- cic power of other components, so battery technology needs to improve substantially before commercial missions with such aircraft are possible. If the battery is not in consideration, as with turbo-electric aircraft, design commuter, regional, and medium-haul missions are feasible with energy benets of up to 6% within conservative 2035 specic powers. Overall, improvements in electrical component technology make electried aircraft feasible, and enable lower energy consumption than the current conventional aircraft. 52 3.2.4 LimitingCases 3.2.4.1 BatterySpecicEnergy In the previous section, the feasibility and eciencies of electried aircraft were shown to improve with advancements in technology parameters. Analyses were done based on dierent predicted technology levels. This section approaches the matter a little dierently – how would electried aircraft perform with the “best of the best” technology numbers? In other words, if the battery is considered, there is a drop in its specic energy and specic power at the aircraft system level compared to the theoretical values. How would the aircraft performance change if the theoretical numbers were used? For novel lithium-ion chemistries, the lithium-air battery has the highest theoretical specic energy of 3500 Wh/kg. In the literature, values as high as 11 000 Wh/kg are quoted; however, those numbers are misleading as they only take into account the mass of lithium in the reaction. The other reactant, oxygen, is drawn by the battery from its environment and accumulates, adding to the battery mass as it discharges. When the mass of the oxygen is accounted for, the lower 3500 Wh/kg value is obtained. Using this value, scaling the battery specic power appropriately, and keeping all other parameters at the optimistic 2035 level, each class of all-electric aircraft was own on its design mission and the resulting productivity- specic energy consumption (PSEC) was compared with the respective conventional aircraft, as shown in Table 3.1. Table 3.1: Comparison of all-electric with limiting BSE vs conventional Mission ConventionalPSEC[kJ/kgkm] All-electricPSEC[kJ/kgkm] Percentbenet Commuter 6.593 2.816 57.3% Regional 5.764 3.080 46.6% Medium haul 4.147 2.713 34.6% Long haul 8.247 – – 53 From the table, where the aircraft is feasible, the results show a substantial reduction in the specic energy consumption over the smaller mission. The energy benet decreases as the mission grows in pay- load and range. At the Li-ion theoretical BSE of 3500 Wh/kg, the specic energy is still smaller compared to that of hydrocarbon fuel, so as more energy is required on-board, the lower the benet. On the other hand, even this high BSE is not enough to facilitate all-electric aircraft for long haul missions. For those missions, the design will have to be hybrid-electric (less energy stored in the battery) or turbo-electric (no battery). 3.2.4.2 ElectricalComponentSpecicPowers In Section 3.2.3, it was noted that the electried designs were a lot more sensitive to battery specic energy (BSE) than to component specic powers. To remove the eects of BSE, only turbo-electric designs were considered in this section, in order to isolate the eects of drastically improving specic powers. In section 3.2.2, it was also observed that the PSEC curves attened out for increasing specic powers, leading to PSEC benets with diminishing returns. This section compares the conventional aircraft for each mission with the respective turbo-electric aircraft for the same mission, but with the component specic powers set to 100 kW/kg. The results are shown in Table 3.2. Table 3.2: Comparison of turbo-electric with limiting specic power vs conventional Mission ConventionalPSEC[kJ/kgkm] Turbo-electricPSEC[kJ/kgkm] Percentbenet Commuter 6.593 4.860 26.3% Regional 5.764 4.898 15.0% Medium haul 4.147 3.467 16.4% Long haul 8.247 5.757 30.2% Across all missions, the high component specic powers enable a reduction in the on-board energy consumed. The PSEC benets are higher for commuter and long haul compared to regional and medium haul missions. The varying level of benets can be attributed to the dierent baseline turbo-electric designs used in each case. It was found that for smaller classes, fully turbo-electric designs provided the most 54 benet, whereas for larger classes, partially turbo-electric designs were optimal, as seen in ??. In the absence of a minimum fan diameter constraint in the optimizer, the results when the number of electric fans was allowed to oat led to a large number (thousands) of very very small (less than micrometer- scale) fans, which was deemed impractical. However, improving component specic powers for turbo- electric designs results in smaller PSEC benets compared to improving BSE for all-electric designs, so this drastic improvement in component specic powers is unlikely to yield substantial PSEC benets over the optimistic 2035 predictions. 3.3 UsefulnessandLimitationsoftheLow-FidelityApproach Although the unied propulsion system architecture described in Section 1.4 is valid at any level of - delity, the analysis framework described in this section is of low delity. The main limitation is that it approximates the mission solely as a cruise segment. As a result, electrication eects, as well as other aircraft sizing considerations related to take-o, climb, and descent are not considered. This simplication is likely to more signicantly impact the results for short missions, for which take-o and descent make up a signicant fraction of the ight time. The various segments also have variable power requirements, instead of the constant power assumed by this cruise-only approach. Work is underway at USC to develop a framework that models the dierent ight segments, including takeo, climb, and descent. In the future, this model will help provide a better comparison between electried aircraft and the conventional aircraft ying today. The technology parameter values chosen are based on literature surveys, technical reports, and ongo- ing NASA-funded research eorts. It is, however, dicult to predict future developments, and the tech- nology level numbers carry a signicant uncertainty. The technology assumptions have a particularly large impact on the feasibility of electried aircraft that employ batteries (all-electric and hybrid-electric). 55 In addition, due to the nonlinear nature of battery discharge, and eciencies of electrical components at o-design operations, the low-delity energy-to-mass or power-to-mass ratio models do not capture the behavior of electrical components fully. The thermal management system (TMS) was also sized based on a simple cooling loop. This power-to-mass model simplies a TMS considerably and does not capture the operational considerations fully. The higher-delity propulsion system model to be developed and discussed in the next chapter aims to address these limitations. Electrical components will be modeled in more detail, allowing for consid- erations like variable power draws, varying eciencies based on the power ow, potential high voltage and temperature converns, and so on. These, along with the loads on electrical components, as well as battery discharge proles based on dierent ight segments, will also be modeled and incorporated into the detailed ight model under development at USC. Finally, at the level of delity at which the low-delity analysis was carried out, conguration-specic layout considerations were not taken into account, in particular those related to the placement of the propulsors and the associated wiring length requirements. The masses of associated wiring (including those for distribution) were lumped together with the masses of each electrical component. The same was also done for mechanical shafts in partial turbo-electric designs. The higher-delity model also aims to address placement concerns in terms of mass and volume of components, as well as the wiring distribution based on conguration, accounting for its mass and volume as well. 3.4 Low-FidelityStudyConclusions The goal of this section was to carry out a broad (though approximate) trade-space exploration and narrow down the design space to regions where aircraft electrication shows the most potential. As such, the 56 development of the framework focused on capturing the most fundamental trends and trade-os, which it did for the most part. The results about the feasibility of electried aircraft over conventional, that 1. they are benecial at shorter ranges, 2. they start becoming more feasible and also more benecial as technology improves, and 3. distributed propulsion and boundary layer propulsion contribute substantially to their feasiblility, are valid results from this analysis. However, the exact benets and the exact points where electried air- craft start becoming viable could be pinpointed with more accuracy. Those “sweet spots” of the tradespace could be explored in more detail, with a higher delity approach that better models ight conditions and the variable requirements on electrical components. The next chapter aims to address some of these issues. It will focus on the electried propulsion system with detailed models of components such as the battery, electrical machines, and power electronics. These components are then integrated into a propulsion system model, and a basic aircraft sizer to look at the eects of their performance across the various ight segments. 57 Chapter4 Higher-FidelityAnalysis 4.1 Models This section presents a set of higher-delity models for the electrical components illustrated in Fig. 1.5, namely, ordered from source to load, the battery, converter, power distribution, wiring, motors, propulsors, and thermal management that capture operational behavior of these components under dierent power loads, so the components can be sized accordingly. They are of higher-delity than the models in Ch. 3, but without going into the detailed analyses of the properties of materials used to manufacture these components. Then, these models are integrated into the propulsion system and aircraft sizing framework that simulates the behavior of these components over a representative mission for a commuter aircraft carrying 20 passengers over 100 nmi. 4.1.1 Battery Batteries have dierent discharge proles based on a variety of factors, including electrical loads on the battery, cell chemistry, operating temperature, and age. Dierent batteries also use various forms of pack- aging, resulting in a knockdown from cell-level specic energy and power values to pack-level. While it 58 is dicult to nd detailed specications for battery packs in commercial use, most packs consist of stan- dard cells whose manufacturer specications are generally available and can be used to build up a battery model. The discharge of batteries is modeled here, with the goal of assembling a model of an all-electric propul- sion system that depletes the battery through the mission. The charging behavior would be dierent than just the inverse of a discharge, and is left to be modeled in future work, if needed. The nearly-linear dis- charge model of [44] is used here. It takes into account the nonlinear nature of discharge by performing a curve t on discharge data from manufacturer cell specication sheets to obtain an equation for capacity discharge. Cell capacity, denotedQ, is the amount of available charge in units of Ampere-hours (Ah). A discharge curve plots the capacity against the voltage,V . Figure 4.1 shows the discharge prole of the Samsung INR 18650-30Q cells used in the NASA X-57 Maxwell [45]. The voltage drop during discharge is nonlinear around the upper and lower extremes of capacity, and close to linear in the middle. The dierent currents show how the capacity would discharge under dierent loads, with faster depletion at higher discharge currents. It is important to note that the entire capacity of the cell cannot be used. Dipping into the upper extremes of capacity reduces the lifetime of the cells and degrades performance through faster discharging. Figure 4.1: Discharge prole of the cell used in the batteries of the all-electric NASA X-57 Maxwell [45]. 59 Frequent charging to 100% capacity stresses the cells and thus also shortens battery life. Consequently, only the middle portion of the capacity is to be used in practice. For this analysis, it is therefore assumed that the lower 10% and upper 20% of the capacity are unavailable following [44]. Once we limit ourselves to the capacity range of 10% to 80%, the discharge curve can be approximated as linear. A simple, nearly-linear equation [44] models this region of discharge as V = V 0 KQRIGIQ; (4.1) whereV is the voltage under the electrical currentI,V 0 is the open source (no load) voltage that can be found in the cell datasheet,Q is the total capacity discharged up to the present instant in time, andR is the internal resistance of the cell. The two modeling constants areK, which sets the primary dependency between voltage and capacity, andG, which represents the change in the slope of the discharge curve due to current. Fitting the discharge model Equation (4.1) to the manufacturer’s discharge curves allows us to deter- mine the four unknowns in the above equation: K;R;V 0 ; andG. The values ofK andG can be found from the slopes of two discharge lines (a system of two linear equations), whileV 0 andR (if not given by the manufacturer) can be obtained from the y-intercepts. Using this approach, the parameters for the cells of the X-57 are found to be:K =0:371 V=Ah,G=0:00520 V=A 2 h,V 0 =4:16 V, andR=0:0265 . The above approach assumes constant current throughout discharges and only traverses one current curve at a time. However, for aircraft applications, the cell does not discharge at a constant current; instead, it discharges at (piecewise) constant power based on utilization along ight segments. Since power is the product of current and voltage for DC circuits, the previous equation can be rewritten in terms of power, P , as V =V 0 KQRP=VGQP=V : (4.2) 60 In order to easily compute the energy consumption, this expression is linearized as follows. Introduce the rst-order Taylor series expansion of 1=V andQ=V about a point (Q n ;V n ) on the curve, 1 V 1 V n 1 V 2 n (VV n ) ; (4.3) Q V Q n V n Q n V 2 n (VV n ) + 1 V n (QQ n ) : (4.4) into Equation (4.2), and rearrange to solve for the voltage, thus obtaining the linear expression V =V n ~ K (QQ n ); (4.5) where ~ K K +GP=V n 1RP=U 2 n GPQ n =U 2 n ; (4.6) and V n 1 2 h (V 0 KQ n ) + p (V 0 KQ n ) 2 4(RP +GPQ n ) i : (4.7) The energy E provided during a discharge period is the integral of the voltage with respect to charge: E = R dE = R V dQ. This integral can be evaluated as the area of the trapezoid under theV vs. Q line, namely E = 1 2 (V i +V f ) (Q f Q i ); (4.8) in which (V i ;Q i ) and (V f ;Q f ) are the points at the start and end of the discharge, respectively. Substitut- ing expressions forV i andV f from the linearized model in Equation (4.5) and simplifying yields the energy provided by the battery: E = V n ~ K Q i +Q f 2 Q n (Q f Q i ) : (4.9) 61 In the above equation, the constant power appears explicitly in the expression (4.6) for ~ K. Additionally, if the energy is thought of as the time integral of power, then the energy delivered over a short time t is simply the product E =P t. While simple, this model captures the constant-power dynamics of battery cells, useful for modeling ight segments with dierent power requirements. The constant power energy delivery over small incre- ments of time can be piecewise integrated to oer a picture of what the energy consumption would be over the entire mission while accounting for the eect of power level on energy consumption. Furthermore, this models allows analysis of cells that are already in the market. Novel battery cells can be incorporated via their discharge proles, and one can then evaluate whether those cells could form building blocks of batteries to benet powered aircraft. 4.1.2 Converter The converter transforms the power from the battery into appropriate voltages and currents suitable for distribution to the motor-propulsor arrays in the propulsion system. If we consider N prop propulsors, then the converter, through the distribution system, has to deliver power toN prop motors, each powering a propulsor. The motor modeled here, a switched reluctance motor (SRM) – also used by Boeing on the SUGAR Volt [5] – requires a converter to switch power delivery among its phases. Over one switching cycle, each phase of the motor needs to have the associated windings energized so that the rotor poles align with them sequentially, and the motor completes revolutions to generate mechanical energy. SRMs run on DC power since the switching of current (turning current on and o) sequentially between the phases is enough to drive the motor. The appropriate converter can then be modeled as a non-ideal DC- DC transformer, or equivalently, an ideal DC-DC transformer plus losses [46]. The ideal converter, illustrated in Fig. 4.2(a), is represented as the DC-DC transformer equivalent circuit of Fig. 4.2(b). The output voltageV conv;out , resistanceR load , and currentI represent the connection to the 62 Switching DC-DC Converter Power Input Power Output Control Input + V conv;in R load + V conv;out I Transformer k d R T I I P t k d I (a) (b) (c) Figure 4.2: Schematic of (a) ideal DC-DC converter, (b) equivalent circuit for DC-DC transformer repre- sentation of the converter, and (c) current waveforms over a cycle. + V conv;in R load + V conv;out + R L L C Diode Switch + + R L + + k d R on k 0 d R D k 0 d V D k 0 d V conv;out k 0 d I Switch Diode V conv;in V conv;out R load + V L I C Figure 4.3: Non-ideal converter: (left) schematic with switch, diode, and inductance, and (right) equivalent circuit model for a full cycle of operation. motor, while the input represents the power delivered by the battery. A control input called the duty cycle, k d , determines what fraction of the cycle has current on and what fraction has current o out of the cycle length (period). The duty cycle changes based on the switching requirements of the motor. The current time-evolution through the circuit, illustrated in 4.2(c), has a peak value I P with ripples of magnitude I assumed small compared toI P . The model shows an error of about 0:3% for I=I P = 0:1, so the DC transformer approach works well to represent the converter [46]. The output voltage,V conv;out , is a function of the input voltage,V conv;in , and we need to construct a model for their relationship. The non-ideal or practical converter, shown in Fig. 4.3, consists of switches and diodes, which have an associated resistance and a voltage drop when in operation that make up the switching losses. The resistance of the wiring is also taken into account, and results in copper losses similar to those of the motor. When the current is high (I =I P i fromt = 0 tot = k d ), the switch is on and the diode is reverse-biased, leading to an open circuit where the diode is located. Using Kircho’s voltage and current 63 laws (KVL and KCL), we can write expressions for the voltage drop across the inductor and for the current through the capacitor respectively as for 0tk d ; V L (t) =V conv;in IR L IR on (4.10) and I C (t) = V conv;out R load ; (4.11) whereR L is the resistance of the inductor andR on is the on-resistance of the switch. When the current is low, the switch turns o, but the diode is forward biased by the inductor current. Using KVL and KCL again for this case gives for k d t ; V L (t) =V conv;in IR L V D IR D V out ; (4.12) and I C (t) =I V conv;out R load : (4.13) The average of these over one time period,, are calculated as <V L > = 1 Z 0 V L (t) dt; (4.14) <I C > = 1 Z 0 I C (t) dt: (4.15) Setting both these average voltage and current to zero allows us to determine the output voltage V conv;out = 1 k 0 d V conv;in k 0 d V D k 02 d R load k 02 d R load +R L +k d R on +k 0 d R D ; (4.16) which is a function of the duty cyclek d , its complementk 0 d = 1k d , the input voltageV conv;in , the voltage dropV D across the diode, as well as the respective resistancesR on ,R D andR L of the switch-on, diode, 64 and inductor. It also depends on what the load is, in this case a motor with resistanceR load through which a currentI ows. We now have all the variables needed to calculate the input and output powers for the converter, P conv;in =V conv;in I ; (4.17) P conv;out =V conv;out k 0 d I ; (4.18) as well as the converter eciency conv = P conv;out P conv;in : (4.19) For the subsequent analysis, we use the representative values for the on-resistance of diodes,R D =0:5 , and the voltage drop across diodes,V D =3:75 V, from [47], and take the value on-resistance of switches, R on =5 from [48]. The resistance of the inductor,R L , is assumed to be negligible, and the load resis- tance is calculated in the motor model, as the motors represent the loads on the converter. All of these sub-components are assumed to be able to operate in high-voltage, high-current environments, as required for the large amounts of power consumed over aircraft ight. 4.1.3 PowerDistribution The power distribution system is a combination of a converter that takes power from the battery, trans- forms it to the voltage optimal for transmission, and distributes it to the motor-propulsor arrays. For a propulsion system withN prop propulsors, there is an equal number of motor-propulsor arrays; each of those receive power from the converter. In an earlier version of the work, documented in [49], it was thought that the array consisted of converters, motors, and propulsors, but the results indicated system 65 voltages about two orders of magnitude beyond the breakdown voltage set by Paschen’s Law [50]. Operat- ing at voltages beyond this limit compromises the safety and functionality of the electrical components and power distribution due to the risk of arcing and discharge through the air. Therefore, such high voltages are not representative of a practical system. These voltages require safe distribution of power, as well as safety limits in case the voltages are beyond any limits prescribed by certication requirements, physical and operational constraints. Therefore, in this work, instead of there being N prop converters, there is only one converter that forms the integral part of the power distribution system. This converter steps up the battery voltage to a voltage suitable for safe transmission, and thus transfers power to the motor-propulsor arrays. In this sense, this modeling approach assumes that each motor includes an appropriate controller. To set limits for power distribution, it is important to look at the voltage limits set by physics and by operational constraints. Paschen’s Law [51] gives the breakdown voltage, or the voltage required to create a discharge or an electric arc between two electrodes in a gas in terms of the pressurep and the separation between the electrodesd, V bd = Bpd ln " Apd ln 1+ 1 # ; (4.20) where for air, the constantsA = 11:25 Pa 1 m 1 ,B = 273:75 V Pa 1 m 1 can be used, and = 0:011 is the secondary ionization coecient. Figure 4.4 plots the breakdown voltage on a logarithmic scale as a function of the product of pressure and gap lengthpd. It can be seen that this curve shows a minimum breakdown voltage ofV bd = 327 V atpd = 0:760 Pam, indicating that this voltage is the absolute worst case for any product of pressure and distance. Specically for aircraft, system voltages less than 327 V, such as the current standard of 270 V used in military aircraft [52], or the270 VDC in the Boeing 787 [50], should never have arcing or partial discharge issues. 66 10 -1 10 0 10 1 10 2 10 3 p.d [Pa.m] 10 2 10 3 10 4 10 5 10 6 Breakdown Voltage, V bd [V] Figure 4.4: Paschen’s curve for air. However, since the curve is plotted as a function of the productpd, a minimum breakdown voltage of 327 V corresponds to a spacing of only 7.5m at sea-level pressure of 1 atm, and a spacing of about 0.05 mm at a pressure of 0.16 atm at the service ceiling of 43,000 ft [53] for large, long-range aircraft like the Boeing 777 or Airbus A350. Such separation distances are too small to be practical. Furthermore, with increased separation, the breakdown voltage increases beyond the absolute minimum of 327 V. In addition, Eqn. (4.20) applies to two uninsulated conductors, whereas, for most wiring, the conductors are sheathed using insulation that is usually made up of plastic materials. The fraction of the wire voltage across the air gap for insulated conductors [50] is given by f V = d d + t i "r ; (4.21) wheret i and" r are the thickness and dielectric constant of the insulation material, respectively. Using this fraction, a safe operating voltage, SOV, can be calculated as SOV = V bd f V : (4.22) 67 With insulation and an increased separation between conductors, the SOV can be much higher than the absolute lowest limit from Paschen’s Law. On the other hand, applying a safety factor of 1.2 as is common for power transmission [54] or of 1.5 as normally used in aerospace yields a lower SOV. However, this limit can be used with more condence in order to account for transient eects and surges. In this work, a safety factor of 1.5 is applied at all aircraft-level results. 4.1.4 CablesandWiring To mitigate the eects of high voltages, cables and wiring design also have to be considered. A safe operat- ing voltage for the aircraft is calculated based on the maximum operating altitude, the separation between wires, the wiring thickness, and the insulation material. Since the wires primarily supply power to the mo- tor from the power distribution system, they are sized based on the maximum power they need to handle, and the SOV. The standard electrical power equation, P = IV ; (4.23) is used with P = P mot;in the input power to each motor, V = SOV, and I the current through the wires. Wires can be selected from standard diameters from the American Wire Gauge (AWG), based on the maximum current that needs to ow through them. A direct current (DC) system, in line with previously-developed models for the motor and the con- verter [49], typically uses four wires, with one wire each carrying the positive and negative voltages and two ground wires [50]. Depending on the material of the wires (typically copper or aluminum), its resis- tance can be calculated from its dimensions and the conductivity of the material as R cable = L cable A cond ; (4.24) 68 where A cond = D 2 cond = 4 is the cross-sectional area of the conductor with diameter D cond , is the conductivity of the material, andL cable is the total length of the cables used. The cable length depends strongly on the conguration, i.e., where in the aircraft the motors and propulsors are placed in relation to the battery that supplies power to them. The method used to calculate the total wiring length for the reference aircraft is outlined in Sec. 4.2.2.1. The power dissipated in the cable can be calculated as P cable = I 2 R cable : (4.25) This dissipated power, and thus the wasted energy, is accounted for when sizing the battery and the thermal management system. Each cable is composed of a conductor, assumed to be a solid cylinder, and the cylindrical shell in- sulation surrounding it. The mass of each part is then a product of the volume and the density of the material. The mass of the wiring can thus be calculated as the sum of the masses of the conductor and insulation, namely m wiring = 1:2 (m cond + m ins ) = 1:2 1 4 cond D 2 cond L cable + 1 4 ins h (2t i +D cond ) 2 D 2 cond i L cable ; (4.26) where the factor of 1.2 accounts for the mass of wire clamps, mounts, and other xing components [55]. 4.1.5 MotorModel The motor model is derived using a rst-principles approach that considers the ux paths through the stator and rotor, including the stator/rotor poles and yokes, the air gap, and the rotor shaft. A Switched ReluctanceMotor (SRM) is chosen for its suitability in high-speed as well as variable-speed operations [56] 69 — needed to deliver the dierent power levels across dierent ight segments. SRMs oer a high power density and a compact size. These motors oer high reliability as well, as the motor phases are electrically independent with negligible mutual coupling. As a result, even if one phase develops a fault, the motor can still operate, although at reduced power. In addition, SRM design is simpler since the machine has windings in the stator only, negating the complexity of having windings in the moving rotor. Having windings in the stator also allows for better manufacturability, greater accessibility in terms of maintenance, and easier cooling. The Boeing SUGAR Volt concept is an example of a study that opted for an SRM. [5]. Drawbacks include high torque ripple, which can be controlled with careful design, and acoustic noise, whose reduction is the topic of ongoing research [56]. An SRM, whose cross-section is schematically illustrated in Fig. 4.5, consists of an outer portion that houses the stator yoke and the stator poles, and an inner portion that includes the rotor poles, the rotor yoke, and the rotor shaft, with an air gap separating the two portions. The stator poles have windings (coils) around them, through which current ows to induce a magnetic eld. In the position shown in Fig. 4.5, diametrically opposite rotor polesR 2 andR 2 0 are aligned with stator polesS B andS B 0 respectively. When current ows in the windings around diametrically opposite stator poles (S A andS A 0), the resulting magnetic eld attracts the unaligned rotor pole pairsR 1 andR 1 0 towardsS A andS A 0, respectively. Once these rotor poles and stator poles are aligned, the current in the windings around stator polesS A andS A 0 is turned o, and the current in the windings around stator polesS B andS B 0 is turned on. This switching of current attractsR 1 andR 1 0 towardsS B andS B 0, respectively. If the current is switched in the sequence ABC the rotor rotates clockwise, whereas a sequenceACB produces a counterclockwise rotation. Aq- phase SRM has a numberN SP of stator poles and a numberN RP of rotor poles in the ratioq :q 1. The motor shown in Fig. 4.5 has 6 stator poles and 4 rotor poles, making it a 3-phase machine. 70 S A S A 0 S B S B 0 S C 0 S C R 1 R 1 0 R 2 0 R 2 Stator yoke (SY) Stator windings Stator pole (SP) Air gap Rotor shaft Rotor pole (RP) Rotor yoke (RY) Aligned Unaligned Flux Linkage, [V.s] Current, I [A] I P W m Figure 4.5: (Left) Switched Reluctance Motor (SRM) cross-section schematic, and (right) illustration of ux linkages versus current curves for an SRM. The method from [56] is adopted for sizing the SRM based on power output, which is as follows. When the rotor pole pairs are aligned or unaligned with the stator pole pairs, ux linkages are present because of the ux paths induced by the magnetic eld. Flux linkage,, is dened as = L(;I)I ; (4.27) where L is the inductance, is the rotor position, and I is the operating current. Figure 4.5 shows a representative ux linkage versus operating current plot for the aligned and the unaligned cases. The area between the aligned and unaligned curves represents the incremental mechanical work done per stroke, W m , of the machine operating at a peak currentI p . The average torque can then be calculated as T avg = W m N SP N RP 2 : (4.28) The mechanical power that the motor outputs is then the product of its average torque and the operating rotational speed, P out;mot =T avg ! mot : (4.29) 71 Thus, calculating the output power of the motor involves generating the ux linkage plots for the aligned and unaligned cases, then calculating the area between them. Since the ux linkage is a product of the inductanceL and the currentI, it is sucient to calculate the inductances for the aligned and unaligned cases. The left side of Fig. 4.6 shows the ux paths when a rotor pole pair is aligned with a stator pole pair. The ux paths can be divided into two sub-paths: FP1, shown as blue dashed lines, and FP2, shown as solid red lines. Each sub-path is modeled dierently. For the ux sub-path FP1, a magnetic equivalent circuit, as seen in Fig. 4.6, is constructed using all the segments of the motor that the ux passes through (stator yoke, stator poles, air gap, rotor poles, and rotor yoke). Each segment has an associated reluctance, analogous to a resistance in an electrical circuit. The magnetic ux,, through each segment is the “current”, and the magnetomotive force (mmf),F , is analogous to the electromotive force. By Ampere’s circuital equation, the applied mmf has to equal the sum of the mmf’s across each seg- ment, such that F 1 = X paths H` = 2 (H SP 1 ` SP 1 +H RP 1 ` RP 1 +H G 1 ` G 1 ) + 1 2 (H SY 1 ` SY 1 +H RY 1 ` RY 1 ) ; (4.30) R SY1 2R SP1 R RY1 R RY1 2R RP1 2R G1 F a 2 a a 2 + FP2 FP1 R SY1 Figure 4.6: (Left) Flux paths when the rotor and stator pole pairs are aligned, and (right) magnetic equiva- lent circuit for ux path FP1 (blue paths in the left image). 72 whereH represents the magnetic eld intensity and`, the average length of the ux path in a segment. The subscriptsSP andRP respectively indicate stator and rotor poles; the subscriptsSY andRY respectively indicate stator and rotor yoke; and the subscriptG refers to the air gap. The magnetic ux in the aligned scenario is the product of the magnetic ux density,B, and the area of the segment in the ux path,A. For the stator pole, it is then a =B SP 1 A SP 1 ; (4.31) with similar expressions for the remaining segments. To calculate the inductance and ultimately the output motor power, a ux density in the stator poles, B SP 1 , is assumed, and the ux densities in the other segments calculated per the corresponding area relative to that of the stator pole, namely B G 1 = a A G 1 =B SP 1 A SP 1 A G 1 ; (4.32) B RY 1 = a 2A RY 1 = B SP 1 2 A SP 1 A RY 1 ; (4.33) B SY 1 = a 2A SY1 = B SP 1 2 A SP 1 A SY 1 ; (4.34) B RP 1 = a 2A RP1 = B SP 1 2 A SP 1 A RP 1 : (4.35) The mmf for each segment is then calculated by looking up the magnetic eld intensity (H) values from a B-H curve of the lamination material, assumed to be electrical-grade M19 steel [57]. The productsH` of dierent segments are summed up to obtain an estimate ofF 1 . Since the applied mmf should be equal to F 1 = (N coil =q)I ; (4.36) 73 where (N coil =q) is the number of turns of winding per phase andI the current through the motor, we iterate on the initial guess forB SP 1 until X paths H` = (N coil =q)I : (4.37) With the converged values of theH’s, the reluctances for each segment can now be calculated as R SP 1 = H SP1 ` SP 1 B SP 1 A SP 1 ; (4.38) R G 1 = ` G 1 0 A G 1 ; (4.39) R RP 1 = H RP1 ` RP 1 B RP 1 A RP 1 ; (4.40) R SY 1 = H SY1 ` SY 1 B SY 1 A SY 1 ; (4.41) R RY 1 = H RY1 ` RY 1 B RY 1 A RY 1 ; (4.42) where 0 is the permeability of free space. Finally, the inductance can be computed as L a 1 = (N coil =q) a I = (N coil =q) 2 R EQ 1 ; (4.43) whereR EQ 1 is the total reluctance of the magnetic equivalent circuit, found by replacing the reluctance network in Fig. 4.6 by an equivalent reluctance, in a manner similar to how a resistor network is replaced by an equivalent resistor for electrical circuits. 74 A similar iterative procedure is used to calculate the inductance due to ux path FP2, which only includes the stator poles and the stator yoke. Only three-quarters of the stator mmf accounts for the ux in FP2, but there are four paths, so the inductance due to FP2 in the aligned case is L a 2 = 4 (N coil =q) 3 4 a I ! = 3 (N coil =q) 2 R EQ 2 : (4.44) whereR EQ 2 is the total reluctance of the magnetic equivalent circuit of FP2. The total aligned inductance is then the sum of the inductances due to the two parts L a =L a 1 +L a 2 : (4.45) The process for calculating the inductances for the unaligned case is more complex, since the ux paths do not ow as cleanly from the stator poles to the unaligned rotor poles. Instead, they break away from the stator poles at angles and enter rotor poles at angles, requiring the use to trigonometry to estimate average path lengths and segment areas covered. Instead of two ux sub-paths, the ux lines must be divided into more sub-paths, each requiring its own analysis of equivalent magnetic circuit, Ampere’s law equation, etc. The iterative process of calculating the magnetic ux density for the unaligned case in each segment and thus the inductance is nevertheless similar to the aligned case. The method is adopted from and described extensively in [56], so will not be reproduced here. Figure 4.7 presents this method as an algorithm. After all the inductances are calculated, the process is repeated for a range of operating currents from low current to the peak current of the machine. The ux linkage ( =LI) is then computed and plotted against current. The area between the aligned and unaligned ux linkage is used to calculate torque, and hence the output power, as seen in Fig. 4.5. 75 The power output by the motor is smaller than the power that it receives due to the presence of losses in the system, which can be divided broadly into two types: copper losses, or losses in the windings around the stator eld as current ows through them; and core losses, or losses at the core parts of the motor which is typically made of a ferromagnetic material. Copper losses are due to the resistance of the wires making up the coil, and the power loss can be calculated as P Cu =qI 2 (R coil =q); (4.46) Start Assume an initial value for stator pole ux density,B SP . Find the stator pole ux, SP . Calculate the ux in path k for all the machine segments. Calculate ux densities (B) in each segment. Use a B-H curve to nd H-values for the corresponding B-values. Compute mmf for each segment (H`). Derive the equivalent magnetic circuit for the ux path. Compute error in the mmf (F k =F k P H`). Compute reluctances from nal B SP values. Compute inductance due to the ux path k (L k ). Sum all inductances to get the total inductance of the winding. Stop Calculate the areas of cross-sections (A) of the segments through which the ux passes. Calculate the mean length (`) of ux path k in each segment. Write Ampere's circuital equation. IsjFj< error bound? B SP =B SP B SP Repeat for all ux paths Figure 4.7: Algorithm to calculate inductances for the motor (adapted from [56]). 76 whereq is the number of phases in the motor,I is the operating current, and (R coil =q) is the resistance of the stator winding per phase. This resistance can be calculated using the standard equation for the resistance of a wire, (R coil =q) = ` coil (N coil =q) A coil ; (4.47) where is the specic conductivity of the conductor (usually copper), ` coil is the mean length of wire in one turn of the coil, (N coil =q) is the number of turns per phase, andA coil is the cross-sectional area of the conductor. Core losses are determined using power loss coecients tabulated according to the motor switching frequency and operating speed. Core losses can be subdivided into three types: hysteresis, eddy-current, and excess losses. Hysteresis losses arise from changes in the ux densityB and magnetic eld intensityH of the core. Eddy-current losses are due to parasitic currents through the core. Excess losses are produced by the movement of the magnetic domain walls, as well as the rotation of the domain dampened by eddy currents [58]. For a switched-reluctance motor operating across a range of speeds, specic power loss coecientsP (power lost per unit mass) were calculated for each segment (stator yoke, stator poles, rotor yoke, and rotor poles) and for each core loss type in [58]. The results revealed that the loss coecients are relatively insensitive to operating speed. Therefore, for this analysis, loss coecients are assumed to be constant and given the values in Table 4.1. The total core losses are then calculated as a sum of the core losses of all segments P core = X k P k m k = X k (P hys;k +P eddy;k +P exs;k ) k V k ; (4.48) where k ; m k ;V k are the density, total mass, and total volume respectively of segmentk. That is, forN RP rotor poles,m RP includes the mass of all of them. 77 Table 4.1: Motor core power loss coecients for the dierent motor segments and loss types. Segment Core Loss Coecients [W/kg] Hysteresis Eddy-current Excess P hys P eddy P exs Stator pole 5 13 1.4 Rotor pole 8 32 1.8 Stator yoke 3.5 19 0.95 Rotor yoke 9.5 20 1.3 The input power to the motor is then the sum of the output power and the losses P mot;in =P mot;out +P Cu +P core ; (4.49) and the motor eciency is given by mot = P mot;out P mot;in : (4.50) Motor sizing is based on the output equation for an SRM [56], P out;mot = mot k d k 1 k 2 BA s D 2 L! mot ; (4.51) wherek 1 =4,B is the motor’s magnetic ux density,D its bore diameter (or the inner stator diameter), L its length in the direction perpendicular to the plane of the cross-section shown in Fig. 4.5, and! mot its rotational speed. The factorsk d andk 2 will be dened later. Note thatD is the diameter of the machine when the rotor is taken out and all that is left is the stator yoke and the stator poles with windings. As such,D incorporates the air gap` G between the stator poles and rotor poles in the aligned position. To size the motor, it is common design practice to express its length as a fraction of the diameter, namelyL=kD, such that the output equation becomes P out;mot = mot kk d k 1 k 2 BA s D 3 ! mot : (4.52) 78 This expresses the output power as a function of the rotor bore diameter:P out;mot /D 3 , and provides a starting point to size the motor. The duty cycle k d = i qN RP 2 is the fraction of a time period over which the current in a phase is high, and is calculated from the number of phases,q, the number of rotor poles,N RP , and the current conduction angle for a rising induction prole i = 1 2 [ RP ( s + r )] ; where RP = 2 N RP is the rotor pole pitch (in radians), and s and r are the stator and rotor pole arcs respectively (also in radians). The constantk 2 is determined by the unaligned inductanceL u and the slopeL s a of the line joining the aligned inductanceL a to the origin in the ux linkage versus current plot of Fig. 4.5, as k 2 = 1 L u L s a Finally, A s = 2 (N coils =q)Im D : is the specic electric loading, calculated using the number of turns per phaseN coils =q, the currentI, and the number of phases that conduct simultaneouslym. 79 Table 4.2: Parameters for the SRM. The values with an asterisk are initial guesses and allowed to be opti- mized; all other values are selected. Parameter Variable Units Value Number of phases q [] 3 Number of stator poles N SP [] 6 Number of rotor poles N RP [] 4 Air gap length ` G mm 0.5 Number of turns per phase (N coils =q) [] 500 Stator pole arc s rad 0.418 Rotor pole arc r rad 0.628 Cross-sectional area of conductor A coil mm 2 1.588 Ratio of length to bore diameter* k [] 0.74 Duty cycle* k d [] 1 —* k 2 [] 0.7 Magnetic ux density* B T 1.7 Linear current density* A s A.m 1 60000 Initial values are assumed for all the constants that depend on the motor dimensions themselves, and the diameter is calculated for the required output power. Some of the motor dimensions have to be deter- minedapriori, so parameters like phase, number of rotor and stator poles, have to be set rst. This analysis assumes a 3-phase motor with 6 stator poles and 4 rotor poles, so we takeq = 3,N SP = 6,N RP = 4, and m = 1. These parameters in turn allow either calculating or setting limits on other geometric dimensions like s , r , and RP . The rest of the dimensions are calculated using the geometry of the switched reluctance motor. An air gap length` G =0:5 mm is assumed, in line with currently available industrial motors [56]. Table 4.2 lists the chosen values and initial guesses for the parameters. Through initial guesses of some parameters and calculations of others, the motor performance is ana- lyzed using the model presented in Sec. 4.1.5. The motor output power developed by the initial design is then checked against the required output power before iteratively re-designing the motor until the require- ment is met. Once the design closes, the total input power to the motor,P mot;in , is calculated by adding the output power and the losses. This motor input power becomes the output power for the converter (P conv;out =P mot;in ), and provides the starting point for converter sizing. 80 4.1.6 Propulsor Since this work focuses on commuter aircraft, it is assumed that the propulsor is a propeller, in order to provide a relevant comparison with conventional aircraft like the Twin Otter. The propulsor model is then incorporated as a scaling law [59] that sets the mass (in kg) of each propeller as m prop = k prop D prop P TO p B prop 0:782 ; (4.53) whereD prop is the propeller diameter in meters,P TO is the takeo power per propulsor in horsepower, B prop is the number of blades per propeller, andk prop = 0:124 is a scaling constant with appropriate units. The nacelle or motor fairing is also sized using a scaling law [59] based on the equivalent shaft horse- power, ESHP, of the engine in the conventional case and the motor shaft power for all-electric aircraft. ESHP is the measure used to rate turboprop engines, where the shaft horsepower (SHP) is the power delivered to the propeller shaft from a turboprop engine, and ESHP also includes the contribution from residual jet thrust from the turbine exhaust. The nacelle mass is then computed as m nace = k nace ESHP; (4.54) wherek nace = 0:0635 kg/hp is a scaling constant. 4.1.7 ThermalManagement Each component in the propulsion system operates at a certain eciency,, less than unity. The energy wasted is assumed to be dissipated as heat, necessitating a thermal management system that handles the wasted heat. The TMS is a heat exchanger that circulates wasted heat from all the electrical components, including the battery, electrical bus, converter, motor, and wiring, and cools these components during 81 operations. The product of ineciency of each component (dened as 1) and the input power to each component determines the wasted heat energy per unit time _ Q () = (1 () )P ();in (4.55) where () is any of the converter, motor, and wiring. The sum of the _ Q values from all the components then sizes the TMS. The heat exchanger is modeled as having a hot side and a cold side separated by a wall, as shown in Fig. 4.8. The specic type modeled here is a compact double-pipe crossow heat exchanger, where the hot side carrying waste heat from the electrical components in the propulsion system forms the inner pipe with the coolant owing through it, and the cold side using air from the freestream forms the outer pipe, separated by a wall. Each part has an associated thermal resistance,R. The following procedure to size the heat exchanger is derived using the number of thermal units (NTU) method outlined in [60], [61], and [62]. Hot side Cold side Wall Waste heat from components Coolant to components Electrical components Cold air Hot air R h R c R w T h;in T c;in T c;out T h;out Cross section for double pipe heat exchanger d i D i d o Figure 4.8: Heat exchanger model. 82 The basic equations that relate the heat transfer rate _ Q on each side are given as _ Q = _ m h c p;h (T h;in T h;out ); (4.56) _ Q = _ m c c p;c (T c;out T c;in ); (4.57) where the subscripts ‘h’ and ‘c’ denote the hot and cold sides, _ m is the mass ow rate,c p is the specic heat at constant pressure, andT is the temperature. The heat transfer rate can also be calculated as _ Q = U 0 A 0 T `m ; (4.58) whereA 0 = d o L ex is the heat transfer surface area computed from the lengthL ex of the heat exchanger and the outer diameterd o . The log mean temperature dierence is given by T `m = T 1 T 2 ln (T 1 =T 2 ) ; T 1 = T h;in T c;out ; T 2 = T h;out T c;in ; The overall heat transfer coecient,U 0 , is calculated in terms of the thermal resistancesR of the hot side and the cold side (assuming wall thermal resistance to be negligible) as U 0 = (R h +R c ) 1 ; (4.59) where R = 1 hA 0 ; (4.60) 83 where h is the convection coecient of heat transfer, which depends on the Nusselt number Nu, the thermal conductivityk, and the hydraulic diameterD h as h = (Nu)k D h : (4.61) To size the heat exchanger for maximum waste heat rejection, it is assumed that the ow is turbulent, so that the Nusselt number can be calculated using the Dittus-Boelter equation: Nusselt number: Nu = 0:023 Re 0:8 Pr 0:4 ; (4.62) Reynolds number: Re = D h G ; (4.63) Prandtl number: Pr = c p k ; (4.64) where is the dynamic viscosity,G = _ m=A c the ow mass velocity, andA c the cross-sectional area. To complete the set of equations, the pressure loss is needed, which can be calculated as p = 2 fG 2 L pipes D h ; (4.65) where is the density of the uid,L ex is the length of the heat exchanger, and f is the friction factor, which for turbulent ow can be estimated as f = 0:046 Re 0:02 : (4.66) The power needed to overcome this pressure loss can be calculated as P pres = _ m p pump ; (4.67) 84 where pump is the eciency of the pump, assumed to be 0.8 for this work. This power is assumed to be supplied from the battery. The steps from Eq. (4.60) to (4.67) must be repeated for both the hot and the cold sides. Finally, the heat exchange between the hot side and the cold side is calculated in terms of the number of thermal units (NTU) as NTU = U 0 A 0 C min ; (4.68) where C min = min _ m h c p;h ; _ m c c p;c is the minimum of the heat capacity rates; in this case, between the coolant and the air. Similarly, deningC max = max _ m h c p;h ; _ m c c p;c , the heat capacity ratio can be written as C r = C min C max ; (4.69) which can then be used to calculate the heat exchanger eectiveness = 1 exp 1 C r NTU 0:22 exp (C r ) NTU 0:78 1 : (4.70) Then, the actual heat transfer rate is calculated as _ Q = C min (T h;in T c;in ) : (4.71) This system of equations is solved to size the TMS in terms of the geometry (d i ,d o ,D i , andL ex ) based on the ambient conditions, as well as limits on acceptable temperatures and pressure losses. The diameter of the inner pipe,d i , which carries the coolant to and from the electrical components is also used to size the 85 pipes needed for transport. The pipe network is assumed to consist of cylindrical pipes of inner diameter d i and thicknesst pipes , such that the pipe mass can be calculated as m pipes = 1:2 1 4 pipes h d i + 2t pipes 2 d 2 i i L pipes + 1 4 coolant d 2 i L pipes ; (4.72) whereL pipes is the total length of the pipe network and is the density of the pipe material. As with the wiring, the factor 1.2 accounts for the mass of mounts and xes. 4.2 IntegrationandSizing 4.2.1 PropulsionSystem The all-electric propulsion system, illustrated in Fig. 4.9, consists of the power delivery chain starting at the battery, connected to a converter that distributes power to a set of motor, each driving a propulsor fan. Power (shown in blue) ows from the battery through the converter and power distribution system, then is distributed to theN prop propulsors through an array ofN prop motors. For each component, the power it outputs to the next component is the product of its input power and eciency. Whatever power it does not output is wasted as heat (shown in red), and rejected using the thermal management system (TMS). With each propulsor having an eciency fan , the total mechanical ow power added to the ow by the propulsion system is P K =N fan fan P shaft ; (4.73) whereP shaft is the power delivered to each fan by its driving motor. A constant propulsor fan eciency of fan =0:8 is assumed, following [42]. 86 Motor Battery Propulsor P mot;out P shaft P mot;in prop mot conv : : N prop Pconv;out Nprop P shaft Motor N prop propP shaft P ow = : : Thermal Management Power Flows Heat Flows Propulsor Converter + Power Distribution P conv;in Pconv;out Nprop . . . . . P TMS N prop Figure 4.9: Propulsion system schematic. Each propulsor is powered by a motor, which delivers its power P shaft =P mot,out = mot P mot,in ; (4.74) where mot is the eciency of the motor. Similarly, each motor is driven by power from the converter through the wiring, such that P mot,in = wiring P conv,out N prop = wiring conv P conv,in N prop : (4.75) The battery supplies power to the converter and the thermal management system, P bat = P conv,in + P TMS : (4.76) The heat ow lines (in red) connect each component ultimately to the TMS, which is a heat exchanger and thus cools the components. Note that the placement of the heat ow arrows are illustrative only, and the optimality of heat exchanger ow paths are beyond the scope of this work. In addition, wiring and TMS piping are not shown in the schematic; however, the mass of both are also taken into account in the analysis. 87 Components are sized based on the power ows that they handle. In general, the mass of each com- ponent is calculated as follows: m comp =P comp = SP comp ; (4.77) whereP comp is the output power of the component, and SP comp is the specic power, or power per unit mass, of the component, obtained from technology scenarios expected in the 2035 timeline, as discussed in Chapter 2. For the TMS, numbers quoted in literature sometimes use the inverse of specic power [7], or the mass per rejected heat energy per unit time. Where appropriate, these alternative numbers have been used in place of specic power to size components. The mass of the propulsion powertrain is calculated by adding the masses of its components, m power =m bat +m conv +m TMS +m wiring +N fan (m propulsor +m nacelle +m mot ); (4.78) and the component masses are calculated from their specic energy or specic power; for example, m bat = E bat =BSE (4.79) m conv = P conv,in =SP conv (4.80) m mot = P mot,in =SP mot : (4.81) The wiring and TMS piping masses are calculated directly from the volume sizing and the materials used in both. 88 4.2.2 AircraftandMission 4.2.2.1 ReferenceAircraft Results from Chapter 3 showed that all-electric power system architectures will only be feasible for small aircraft ying short missions in the near future. Therefore, the representative mission studied here is that of a commuter aircraft (modeled on the Viking Air Twin Otter), with the relevant mission parameters given in Table 4.3. These parameters were used to construct a mission power prole for the Twin Otter, as well as to calibrate parameters related to distributed propulsion. For example, for a case of 10 propulsors versus the two on the Twin Otter, the total fan area was kept constant. The smaller diameters for the 10 propellers were calculated to meet this requirement. For conguration-specic calculations like the length of the wiring, the Twin Otter fuselage was mod- eled as having an ellipsoidal cross-section, as seen in Fig. 4.10. The battery and power distribution system are assumed to be placed below the passenger cabin at the bottom of the fuselage. From there, the wires Table 4.3: Baseline aircraft and mission specications [63] Twin Otter Number of passengers N pax 19 Payload m pay 1 842 kg Mission range R 100 nmi (185 km) Cruise speed V cruise 94 m/s Takeo mass m TO 5 670 kg Fuselage height h fuse 2.09 m Fuselage width w fuse 1.75 m Propeller diameter D prop 2.6 m Number of blades per propeller B prop 3 Cruise lift-to-drag ratio (L=D) cruise 12 Cruise altitude h cruise 10,000 ft (3,050 m) Service ceiling h SC 27,600 ft (8,138 m) Max rate of climb _ h max 490 m/min Climb gradient dh=dR 107 m/km 89 y clear;fuse y clear;prop D prop Wiring Propulsor Wing h fuse w fuse Fuselage Battery Figure 4.10: Twin Otter cross-section schematic to calulate wiring length. travel along the fuselage sidewalls and the wing to transfer power to the motor-propulsor arrays dis- tributed over the wingspan. The wiring is assumed to be contained in the same plane as the cross-section shown in Fig. 4.10, i.e. the distances that arise from the distribution system not being in the same plane as the motors are neglected. For a system withN prop propulsors spread symmetrically throughout the wingspan, the length of the wiring can be calulated as L cable = 4 2 4 N prop C fuse +y clear;fuse + 2 1 2 Nprop X k=1 k 1 2 D prop + (k 1)y clear;prop 3 5 ; (4.82) where the factor of 4 accounts for there being 4 wires (+V=2,V=2, and a ground for each) for a DC power distribution system. The distance between the blade span of two adjacent propellers isy clear;prop , and the clearance between the fuselage and the propulsor isy clear;fuse , set to half the fuselage width plusy clear;prop . The diameter of each propulsor isD prop . The circumference of the fuselage cross-section, assumed to be an ellipse, is calculated as [64] C fuse = (a +b) 1 + 3h 10 + p 4 3h ; (4.83) with h = (ab) 2 (a +b) 2 ; 90 wherea = 1 2 h fuse , andb = 1 2 w fuse are the semi-major and semi-minor axes of the elliptical fuselage cross-section, respectively. To calculate the overall length of the pipe network that recirculates coolant between the thermal man- agement system and the motors, the same approach is used, except only two pipelines are needed – to transport coolant to and from the motors. As a result, the length of the pipes needed isL pipes = L cable =2, since the total length of the cables assumes four cables per line. 4.2.2.2 MissionPowerProle For this mission prole, it is assumed that the Twin Otter climbs at constant power to reach its cruise altitude at the maximum climb rate and climb gradient. For approach, the descent rate is set to be half the magnitude of climb rate, while the descent gradient is xed at the allowed maximum of 3 for an instrument landing system (ILS) approach. The remaining distance to complete the mission is assumed to covered during cruise. What follows presents the method for determining the actual power prole. The aircraft is assumed to climb continuously at constant power to reach its cruising altitude from sea level. Since the acceleration is assumed to be zero, the equations of motion for the aircraft are [40] T climb D climb W sin ( ) = 0 (4.84) L climb W cos ( ) = 0; (4.85) whereT climb ,D climb ,L climb , andW are the four forces – thrust, drag, lift, and weight – on the aircraft during climb, and = arctan (dh=dR) is the climb angle set by the height-to-range climb gradient dh=dR. For all-electric aircraft, no fuel is burned and the weight is constant throughout the mission. Drag can be 91 expressed as a function of lift using a reduced lift-to-drag ratio, in this case assumed to be 2=3 of the cruise L=D, namely D climb = L climb (L=D) climb = L climb 2 3 (L=D) cruise = W cos( ) 2 3 (L=D) cruise : (4.86) The climb speed can be estimated using the climb rate _ h and the climb angle as V climb = _ h= sin ( ): (4.87) Then, the power required to climb is P climb =T climb V climb ; (4.88) whereT climb is calculated from Eq. (4.84) given Eq. (4.86). The time it takes to climb up to cruise level is determined by the cruise altitude and the rate of climb, while the energy consumed during climb is then the product of the climb power and climb time, namely E climb =P climb t climb : (4.89) For cruise, lift balances weight and thrust balances drag, and thus thrust can be calculated as T cruise =D cruise = L cruise (L=D) cruise = W (L=D) cruise : (4.90) The cruise power and energy usage are then set by the cruise speed,V cruise , and cruise time durationt cruise , namely P cruise = T cruise V cruise (4.91) E cruise = P cruise t cruise : (4.92) 92 Note that the cruising time is equal to the cruise range divided by cruise speed, which is assumed constant. Both are mission parameters. The approach segment is similar to the climb, except that the weight contribution in the thrust/drag/weight balance Equation (4.84) has the same sign as thrust since the angle is negative indicating a decent. We set the approach angle to =3 per the standard ILS approach. The thrust and power requirement for descent is then of course lower than for climbing. In addition, we use the cruise lift-to-drag value for descent, such that D approach = L approach (L=D) approach = L approach (L=D) cruise = W cos( ) (L=D) cruise : (4.93) Figure 4.11 shows the altitude and power prole for our representative climb-cruise-approach mission. These power values are used to simulate and size the propulsion system components for the commuter aircraft mission. The power needed for climb is over double that for cruise, whereas approach requires very little power. 4.2.2.3 AircraftSizing The all-electric aircraft is sized as follows. Starting with the same ratio of operating empty mass minus the engine mass to the takeo mass (r = m rest =m TO ) as the conventional aircraft of Table 4.3, the all-electric 0 5 10 15 20 25 30 35 40 45 Time, t [min] 0 500 1000 1500 2000 2500 3000 3500 Altitude, h [m] Climb Cruise Approach 0 5 10 15 20 25 30 35 40 Time, t [min] 0 100 200 300 400 500 600 700 800 900 1000 Average Power Needed, P [kW] Climb Cruise Approach Figure 4.11: Twin-Otter climb-cruise-approach mission: (left) altitude and (right) power prole. 93 propulsion system is sized based on Sec. 4.2.1, which gives the mass of the electric propulsion system (m EPS ). Then, the new takeo mass is calculated as: m TO = m rest + m pay + m EPS (4.94) from the payload mass,m pay and the mass of the rest of the aircraft (operating empty mass minus the engine mass),m rest . Note that the electric propulsion system mass already includes the mass of the battery i.e., the mass of the energy required for the mission. The bookkeeping is dierent from the conventional case, which split the mass of the energy (fuel) from the mass of the propulsion system (engines). Expressing m rest as a ratio of the takeo mass, Eq. 4.94 can be rewritten as m TO = m pay + m EPS 1r : (4.95) Keepingr constant to the same value as that for the conventional baseline, the propulsion system mass m EPS can be recalculated iteratively, updating the takeo mass until it converges and the aircraft sizing is complete. 4.3 Results: Subsystem-Level 4.3.1 ComparisonofLow-andHigher-FidelityModels 4.3.1.1 Cruise-OnlyMission The focus of this chapter is to understand the dierences in operational behavior of components at vari- ous power loads during dierent power segments. However, when comparing between the higher-delity models from this chapter with the low-delity models from Chapter 3, it becomes important to distinguish how much of the dierences are due to the cruise-only assumptions of the mission in Chapter 3 versus 94 Table 4.4: Comparison between the low-delity and higher-delity component parameters for a cruise- only mission with low-delity eciencies from the intermediate 2035 technology scenario. Parameter Units Low Fidelity Higher Fidelity Dierence Propulsion system mass [kg] 1180 2770 135% Battery mass [kg] 930 2050 120% Battery energy [kWh] 530 1180 120% Converter mass [kg] 47 87 84% Converter eciency [–] 0.99 0.84 -15% Motor mass [kg] 54 85 57% Motor eciency [–] 0.98 0.95 -3% Wiring mass [kg] — 16 — Wiring eciency [–] — 0.998 — TMS mass [kg] 34 366 991% the dierences in the delity level of the electrical component models. Therefore, rst, a comparison be- tween the two levels of delity is presented here for the reference aircraft ying a cruise-only mission. The cruise power is calculated for both approaches according to Sec. 4.2.2.2, and the model is run assum- ing this constant cruise power over the entire ight. For these comparisons, the low-delity propulsion system component models are those from Chapter 3, with constant assumed values for eciency set to the intermediate 2035 technology scenario from Sec. 2.3. The higher-delity component models are from those those at the beginning of this chapter, and are sized to provide power for cruise only. The eciencies of the higher-delity components are calculated from the models and not set by technology scenarios like the low-delity case. Both these low- and higher-delity propulsion systems are incorporated into the aircraft sizing method from Sec. 4.2.2.3 to compare the eects of delity level on the propulsion system components. Table 4.4 shows the dierences in the propulsion system components. The higher-delity model pre- dicts a 135% increase in propulsion system mass, with a corresponding increase of 120% in battery mass and energy for the same mission. The dierences arises mainly from the assumed eciency values in the low-delity models versus the calculated eciency values from operational behavior in the higher-delity models. The converter, which sees the motor as a load, has its overall eciency drop to 84% compared to 95 Table 4.5: Comparison between the low-delity and higher-delity component parameters for a cruise- only mission with low-delity eciencies set to those obtained from higher-delity models. Parameter Units Low Fidelity Higher Fidelity Dierence Propulsion system mass [kg] 2640 2770 5% Battery mass [kg] 1930 2050 6% Battery energy [kWh] 1110 1180 6% Converter mass [kg] 84 87 3% Converter eciency [–] 0.84 0.84 0% Motor mass [kg] 82 85 3% Motor eciency [–] 0.95 0.95 0% Wiring mass [kg] — 16 — Wiring eciency [–] — 0.998 — TMS mass [kg] 350 366 5% the constant 99% assumed eciency in the low delity case. The motor has an eciency of about 95%, close to the assumed eciency of 98% in the low delity analysis, but it still produces more heat. Together, the drop in eciencies of the motor and the converter means that the TMS is sized to be almost 10 times larger for the higher-delity approach. The battery, in turn, and the rest of the propulsion system as well as the aircraft have to be sized larger to support this larger TMS driving up the battery mass and energy required substantially. If the eciencies of the motor and the converter are instead xed to the values obtained from the higher-delity models and the low-delity analysis is re-run, then the results are much closer. Table 4.5 shows that now the propulsion system is only 5% heavier, with a 6% larger battery The dierence comes from the added wiring model in the higher-delity case. The ineciency of the wiring model, however small, makes the TMS larger, and the incremental mass associated with the wiring and the 5% larger TMS also forces the battery and the rest of the components, as well as the aircraft, to be sized larger. This result suggests that for this case, the using the low-delity approach for quicker modeling would need a BSE penalty of 6%, but to quantify this penalty, higher-delity modeling was necessary. 96 4.3.1.2 Climb-Cruise-ApproachMission Next, the mission consisting of climb, cruise, and approach segments as illustrated in Sec. 4.2.2.2 is consid- ered. Figure 4.12 shows the mass of the battery and the energy required for each ight segment, as well as the entire mission, for the all-electric aircraft. Note that in both low- and higher-delity models the battery is assumed to only operate between 10% and 80% charge level, as explained in Section 4.1.1. The biggest numerical dierence between the results predicted by the two delity levels is in the climb segment, for which the high-delity approach predicts 156% more energy usage. With the low-delity model, climb amounts to 40% of the total mission energy, while climb consumes 46% of the mission energy when using the higher-delity models. The energy consumed during approach diers the most percentage- wise between the two delity level. Most of these dierences can be attributed to the motor not operating at peak eciency in lower-power segments (91% at cruise and 77% at approach) compared to climb (95%), which it was sized for. As a result, the battery is sized to be bigger to account for the more wasted energy, making the overall aircraft much larger in the higher-delity approach compared to the low-delity one. In fact, the higher-delity models predict about 65% higher takeo mass than the low-delity did. In terms Climb Cruise Approach Total 0 500 1000 1500 2000 2500 3000 Battery mass [kg] Low Fidelity Higher Fidelity 0 200 400 600 800 1000 1200 1400 1600 Energy Required [kW.h] +156% +78% +207% + 123% Figure 4.12: Battery mass and energy dierences for the low- and higher-delity models for the Twin Otter mission. 97 of the overall mission, the higher-delity approach results in roughly 123% higher energy consumption than the low-delity model value. The low-delity model does not capture the high-power battery discharge dynamics, while the higher- delity model does account for the loss in battery eciency for high-power demands and hence predicts the need for a larger battery as a result. This distinction reveals the importance of accounting for the eect of power level on energy consumption at the battery level (basically the Ragone relation), and its impact on mission energy eciency. Another distinction between the two delity levels is that the higher-delity model predicts a higher energy needed for climb rather than cruise, despite climb only lasting 1/3 of the time that cruise does. This is due to the dynamics of battery discharge: it is more constraining to require a very high power level during a short time, than it is to require a lower power level for a longer time. As a result, the battery ends up being sized mostly to meet the demands of the climb segment. If instead, as in the previous section, the low-delity eciencies of the motor and converter are re- placed with those values from the higher-delity cases, then the results become much closer here too. For these results, the higher-delity model resulted in the motor having dierent eciencies in dierent ight segments: 95% at climb, 90% at cruise, and 77% at approach. The converter eciency stayed around 99% in all segments. The reasoning behind these numbers are explained in the next section, but for this section, these results were used to calculate an overall (mission) eciency for the motor and converter — weighted averages incorporating the duration of each segment, as shown in Table 4.6. Rerunning the low-delity model with these averaged eciency numbers for the motor and converter resulted in a higher-delity propulsion system that is 11% more heavy, compared to the previous result of 126%. The higher-delity motor and converter are within 3% of their low-delity counterparts, but the TMS is sized 55% smaller. The dierence is due to the the climb eciency of the motor calculated to be 95% for higher-delity, thereby wasting less energy as heat, compared to the 87% for low-delity, resulting in more heat loss and necessitating a larger TMS for the low-delity case. It is interesting to note here that 98 Table 4.6: Comparison between the low-delity and higher-delity component parameters for a climb- cruise-approach mission with low-delity eciencies set to those obtained from higher-delity models. Parameter Units Low Fidelity Higher Fidelity Dierence Propulsion system mass [kg] 3940 4370 11% Battery mass [kg] 2370 2800 18% Battery energy [kWh] 1360 1610 18% Converter mass [kg] 228 223 -2% Overall converter eciency [–] 0.99 0.99 0% Motor mass [kg] 264 257 -3% Overall motor eciency [–] 0.87 0.87 0% Wiring mass [kg] — 347 — Wiring eciency [–] — 0.998 — TMS mass [kg] 653 295 -55% the wiring is an important factor of the propulsion system in terms of mass, due to the thick, 35-mm wires needed to safely carry the system currents of around 450 A. This is not reected in the low-delity approach at all, which does not model wiring separately and instead lumps it with each component. So, propulsion system mass dierence from the two delity levels are close not because of the closeness of modeling, but due to the larger low-delity TMS mass (plus no wiring) being comparable with the smaller higher-delity TMS mass plus wiring. To gain more meaningful results, the wiring would need to be modeled at low delity as well, and increasing the delity level to incorporate currents and voltages, something that the low-delity model avoided. In addition, it is not as straightforward in the climb-cruise-approach mission to quantify the dierences as a penalty to BSE consistently, since some components have a larger mass and other have a smaller mass at dierent levels of delity. The dierences arise due to the operational behavior of the dierent components at varying power levels, where the nuances cannot be accurately portrayed with a simple overall mission eciency number. These operational behaviors are discussed next. 99 4.3.2 ComponentEciencies Figure 4.13 shows the eciency of the motors and the converters across the dierent ight segments for an aircraft with two and 20 propulsors (representing distributed propulsion). Both components are sized for the maximum power load they have to deal with, which occurs during climb. The graphs show that the motor operates most eciently at its design point (max power) in both cases. The eciency drops in o-design cases such as approach, where the same motors sized for substantially higher power are running at lower power. Note that the alternative would be to idle some motors during approach so the non-idle ones can function more eciently. But such operational strategies are not considered here in this work. The motor losses account for these changes in eciency. Core losses scale with motor weight, and since running the motor at lower power does not change its weight, the core losses stay relatively constant in absolute terms, but in relative to the output power, the lost power is proportionally higher for low power- output operation. Winding (copper) losses scale with motor power: copper losses are lower at low power, but not low enough to oset the higher proportion of core losses. The overall eect is a drop in motor eciency at lower powers. It should be noted that the motor eciency never drops below 75% in both cases, meaning that an electric powertrain may well have an overall eciency advantage over conventional system even if running at sub-optimal electrical component eciencies. The converter eciency, on the other hand, behaves opposite to the motor eciency. The switching losses are assumed to be constant, since the diodes and the transistor switches have the same on-resistance and voltage drop over a wide range of voltages [48, 47]. These make up a smaller portion of the converter losses, and even more so at high power, where the copper losses dominate. As a result, at high power, the converter eciency drops due to higher copper losses, but improves at lower power. Distributed propulsion improves the eciency of the motor across all segments, as seen on the right chart in Fig. 4.13. Since there are 20 motors each powering a propulsor, the dierence between the max- imum power at climb and the lower power at cruise and approach are smaller proportionally than they 100 would be with two motors. As a result, the motor components benet both from being sized for lower power individually, and from the power drop from climb to cruise, resulting in an increase in eciency over the lower-power segments. Signicantly, in the 20-propulsor case, the motor still operates at over 90% eciency in all segments. The converter, however, takes a slight penalty in eciency with distribution, although its eciency remains at over 95% for all segments. The motor resistance, which appears as a load resistance to the converter, does not scale linearly with power, so with multiple motors, the total resistance is larger than for fewer motors providing the same power. This eect is not seen in the motors themselves, because with smaller motors, the core losses, which contribute to the motor eciency, scale with the power, providing some balance compared to the resistive (copper) losses. 4.3.3 PowerDistribution Equation (4.20) in Sec. 4.1.3 describes the relationship between the breakdown voltage and the product of air pressure and separation between electrodes. This productpd does not provide an intuitive way of understanding the eects that pressure and separation individually have on the breakdown voltage. Fig- ure 4.14(a) shows the eect when pressure and separation are decoupled. This result is valid for uninsulated Motor Converter 0.7 0.75 0.8 0.85 0.9 0.95 1 Component Efficiency, 2 Climb Cruise Approach Motor Converter 0.7 0.75 0.8 0.85 0.9 0.95 1 Component Efficiency, 2 Figure 4.13: Eciency of motor and converter over the ight segments for (left) 2 propulsors and (right) 20 propulsors. 101 0 0.2 0.4 0.6 0.8 1 Pressure, p [atm] 10 1 10 2 10 3 10 4 10 5 10 6 Breakdown Voltage, V bd [V] d=7.5 m d=0.01mm d=0.1mm d=1mm d=10mm d=100mm 43,000 ft 27,600 ft 0 0.2 0.4 0.6 0.8 1 Pressure, p [atm] 10 1 10 2 10 3 10 4 10 5 10 6 Safe Operating Voltage, SOV [V] d=7.5 m d=0.01mm d=0.1mm d=1mm d=10mm d=100mm 43,000 ft 27,600 ft Figure 4.14: (a) Variation of breakdown voltage for uninsulated conductors, and (b) variation of safe oper- ating voltage (SOV) for insulated conductors at dierent pressures and conductor spacing. conductors given by Eqn. (4.20). As seen for very small spacings (d 0:1 mm), the breakdown voltage rst drops sharply, reaching a minimum, then increases gradually with pressure. For larger gap lengths (d 1 mm), this eect is present but cannot be seen in the plots, as the lowest value pressure used was that corresponding to the US Standard Atmosphere at an altitude of 25 km, where the isothermal layer ends in the stratosphere, and well beyond the 43,000 ft (13.1 km) maximum altitude current commercial aircraft y at. At 43,000 ft, for example, the breakdown voltage is 1.2 kV for a separation of 1 mm, about four times the 327 V minimum breakdown voltage for any combination ofp timesd. At the 27,600-ft service ceiling of the reference Twin Otter aircraft, the breakdown voltage is 2.1 kV atd = 1 mm. When wires are modeled with insulation, as in real-world applications, the safe operating voltage is considered instead of the breakdown voltage, as indicated in Eqns.(4.21) and (4.22). Figure 4.14(b) shows how the safe operating voltage (SOV) as the separation between the conductors varies when the conductors are insulated. The trends are similar to the uninsulated case, but the safe operating voltages are higher than the corresponding breakdown voltages. For a separation of 1 mm, the safe operating voltage increases to 2.4 kV at 43,000 ft and to 4.3 kV at 27,600 ft. 102 0 0.2 0.4 0.6 0.8 1 Pressure, p [atm] 10 1 10 2 10 3 10 4 10 5 10 6 Safe Operating Voltage, SOV [V] t i =0.1mm t i =1.0mm t i =2.0mm t i =3.0mm t i =4.0mm 43,000 ft 27,600 ft 0 0.2 0.4 0.6 0.8 1 Pressure, p [atm] 10 1 10 2 10 3 10 4 10 5 10 6 Safe Operating Voltage, SOV [V] r =2.0 r =2.5 r =3.0 r =3.5 r =4.0 43,000 ft 27,600 ft Figure 4.15: (a) Variation of safe operating voltage (SOV) with insulation thickness, and (b) with insulation material at dierent pressures. For the previous studies, the insulation thickness and the material were kept constant. Figure 4.15 shows the eects of varying the thicknesst i (a), and insulation material, represented by dielectric con- stant" r (b). The ranges fort i were determined from the appropriate thicknesses based on the American Wire Standard wire gauges as presented in [50], and those for" r were based on the dierent insulating materials used for commercial conducting wires[65]. Across the ranges considered, the SOV increases with increasing pressure while varying botht i and" r . To increase SOV, the insulation thickness must be increased and a material with a lower dielectric constant must be used. The eects of both, however, pale in comparison to the eect of increasing the separation between the conductors. In addition, the types and thicknesses of insulation used in standard wiring are not expected to change drastically, so any gains for SOV must come from separation between conductors only, assuming electried aircraft are expected to be certied to the same service ceiling (hence, pressure/altitude) as current aircraft. Consider some point analyses for current aircraft. As a reasonable case, the wires are assumed to be insulated with polyethylene (" r = 2:28) with an insulation thickness oft i = 1 mm and a conductor spacing of at leastd = 2 mm. Table 4.7 shows the results of safe operating voltage calculations for dierent aircraft classes. For a long range aircraft with a service ceiling of 43,000 ft, the SOV is 2.47 kV. Applying a 103 safety factor of 1:5 reduces this to 1.65 kV. For the commuter Twin Otter-like aircraft with a service ceiling of 27,600 ft, the SOV is 4.52 kV and is reduced to 3.01 kV with a 1:5 safety factor. This analysis shows that SOV for aircraft can be much higher than the minimum predicted by Paschen’s Law after accounting for insulation and conductor spacing, even at the altitudes aircraft y at. All of these values fall under the 4.5 kV high-voltage architectures supported by the NASA Electric Aircraft Testbed (NEAT) [30]. For the all-electric commuter aircraft analyzed in this work, a SOV of 3 kV will be used. Table 4.7: Safe operating voltages for dierent aircraft classes at dierent safety factors. Class of Service Safe Operating Aircraft Example Ceiling Voltage (SOV) SOV [kV] SOV [kV] [ft] [kV] Safety factor 1.2 Safety factor 1.5 Long-range 777, A350 43,000 2.47 2.06 1.65 Commuter Twin Otter 27,600 4.52 3.77 3.01 4.3.4 Wiring Using the power required by each motor-propulsor array, the electrical conductors for a four-wire DC transmission system are sized. The length of the wiring required for a Twin Otter-like commuter aircraft is calculated using the method in Sec. 4.2.2.1. The wire gauge required, as well as the resistance, power loss, and mass are all calculated using the equations in Sec. 4.1.4. Table 4.8 shows point design studies for wiring. For 2 propulsors, the wiring length is considerably shorter compared to that for 20 propulsors, but the wires themselves are thicker to handle the larger currents. As a result, despite the large dierence in wiring lengths for the two cases, the total wiring mass is smaller for 20 propulsors, since the diameter of the wires used are about an order of magnitude smaller. However, compared to the aircraft takeo mass of about 12 000 kg, the wiring mass is negligible, even for 2 propulsors. In terms of power losses, all four cases considered result in a wiring eciency of over 99%, i.e. the power lost in the cables is negligible compared to the climb power required. The material used 104 also makes little dierence: using copper results in lower power losses at the penalty of being heavier; however, compared to the masses and power of other components in the propulsion system, the choice of material for wiring makes little dierence. Hence, copper wires are chosen for all subsequent analyses. Table 4.8: Wiring resistance, mass, and power for copper and aluminum conductors for 2 and 20 propulsors. Number of Current Conductor Diameter Wiring Length Material Resistance Total Mass Power Loss Propulsors I [A] D cond [mm] L cable [m] R cable [ ] m cable [kg] P cable [kW] 2 520 31.1 35 Cu 0.02 288 0.65 Al 0.04 90 1.03 20 44 2.6 627 Cu 6.22 43 12.04 Al 9.84 19 19.04 It should be noted here that these analyses were done using the 3 kV safe operating voltage identied at the beginning of this section. Lower voltages could be used for power distribution, with the current subsequently increasing, but this would result in higher masses and power losses in the wiring. The mass is proportional to the square of the conductor diameter, which is sized by the maximum current rating. Power losses scale with the square of the current. Increased power losses also require a larger battery to supply extra power to overcome those losses. Therefore, for ecient, lightweight power distribution, high voltage and low current provide the best results. An alternative would be to use superconducting cables with very low resistances, allowing much higher currents and low voltages. However, a cryocooler would be needed to cool the conductors to the extremely low temperatures required. Superconductivity would add a layer of complexity to the modeling and is beyond the scope of this work. 4.4 Results: Aircraft-Level 4.4.1 SystemMassBreakdown Figure 4.16 shows the mass breakdown of the propulsion system for dierent numbers of propulsors: two propulsors on the left and 20 on the right. For both cases, the battery accounts for over 60% of the 105 system mass, with the next largest contribution coming from the propulsors. This result indicates that the battery technology aects aircraft-level performance the most, and so there is most to gain from from technological advances in specic energy and specic power. The other components aect aircraft-level performance less, and benet relatively less from advances in specic power; however, their eciencies aect the battery sizing as well. Therefore, it is benecial at aircraft-level to improve component eciency, even at the cost of added mass, since the batter mass is an order of magnitude larger than that of the other components. The motors and converter each account for just over 5% of the propulsion system mass. For 20 propul- sors, the fraction of the propulsor mass goes up; however, the actual total mass of the propulsors goes down compared to 2 propulsors. This is due to the benets of the cube-squared law of mass to propulsive power scaling, as discussed in Sec. 1.1.1. The mass scales with the volume of the propulsor (volume characteristic length cubed) whereas the power, from the product of thrust and ight speed, scales with the fan area (area length squared). For the same total fan area as prescribed in this analysis, 20 smaller propulsors collectively weigh less than the two larger ones. In addition, for 20 propulsors, even though the battery represents a slightly larger fraction of the system mass, the actual mass of the battery goes down, System Mass Breakdown Nprop --- 2 20 Battery mass_battery kg 2797 2115 Converter mass_converter kg 223 184 Motors mass_motors kg 257 205 Propulsors mass_propulsors kg 456 389 Wiring mass_wiring kg 347 44 TMS mass_TMS kg 295 301 total_mass_propsys kg 4.37E+03 3.62E+03 Battery 64% Converter 5% Motors 6% Propulsors 10% Wiring 8% TMS 7% 2 Propulsors Battery 65% Converter 6% Motors 6% Propulsors 12% Wiring 2% TMS 9% 20 Propulsors Figure 4.16: Propulsion system mass breakdown by component for (left) two propulsors and (right) 20 propulsors. 106 in part due to the mass of the propulsors going down. This dierence in percentages is primarily due to the changes in wiring mass. For 2 propulsors, the conductors have a diameter of 31.1 mm to distribute large power to only two motors. For 20 propulsors, a conductor diameter of 2.6 mm is sucient to handle power distribution since each motor requires substantially less power. Since the wiring mass scales with its volume, the square of the diameter has a more meaningful impact than the length for cylindrical wires. Hence, despite the 20-propulsor case having about 18 times the wiring length of the 2-propulsor case, the reduction in the diameter by a factor of 12 means the wiring is 87% lighter. The mass of the motors and converters decrease only slightly with distribution, but contribute to the propulsion system mass in roughly the same proportion. The TMS increases in both actual mass and as a fraction of the total mass with distribution, due to the need for cooling an increased number of motors. In terms of actual mass, the propulsion system with 20 propulsors weighs 3,155 kg, about 27% less than the propulsion system with 2 propulsors. The mass of the propulsors going down directly contributes to the mass of the overall propulsion system going down with increased distribution. The next section looks at the eects of distribution in more detail. 4.4.2 DistributedPropulsion Distributed propulsion has the potential to decrease propulsion system mass due to the cube-squared rela- tionship between thrust and propulsion weight discussed above. Figure 4.17 tracks the component masses as the number of propulsors is varied from 2 to 20. The plot on the left shows the masses of all com- ponents, including the total propulsion system mass in black, and the plot on the right zooms in on the smaller components to provide more detail. The overall mass of the propulsion system decreases with more distribution by about 27% going from 2 to 20 propulsors. It can be seen that going from 2 to 4 propulsors provides the most dramatic reduction in mass, with the trend leveling out and even increasing slightly with more distribution. From 2 to 4 propulsors, the benets are most profound in the wiring and the propulsors. 107 2 4 6 8 10 12 14 16 18 20 Number of Propulsors 0 500 1000 1500 2000 2500 3000 3500 4000 4500 Mass [kg] Total Battery Motors Converter Propulsors Wiring Thermal Mgmt (TMS) 2 4 6 8 10 12 14 16 18 20 Number of Propulsors 0 50 100 150 200 250 300 350 400 450 500 Mass [kg] Motors Converter Propulsors Wiring Thermal Mgmt (TMS) Figure 4.17: Propulsion system mass changes with distributed propulsion; (left) showing all the compo- nents, and (right) showing more detail of the smaller components. The wiring weight drops substantially because the maximum distribution current in the system decreases from 513 A to 211 A, requiring a conductor of diameter 9.3 mm compared to 31.1 mm based on the standard American Wire Gauge (AWG). This eect is shown in Fig. 4.18. The propulsor mass also drops due to the aforementioned cube-squared relationship. Together, these reductions mean the battery is sized for less mission energy, reducing its mass too, and having favorable eect for the mass of all the other components and of the full aircraft. Going from 4 to 20 propulsors on Fig. 4.17, the motor and converter masses stay roughly at, whereas the thermal management system mass increases with more distribution. The invariance in converter mass can be explained by the fact that there is one converter that supplies power to the motors and acts as the power distribution system; therefore, as long as the total ow power needed remains roughly constant irrespective of the number of propulsors, the size of the converter also does not change. In addition, both the motor and the converter are optimized in each case. In particular, the motor eciency remains roughly constant whether the motor is sized for half of the ow power needed (2 propulsors) or for 1/20 th (20 propulsors). 108 0 5 10 15 20 Number of Propulsors 0 100 200 300 400 500 600 Maximum Distribution Current [A] Figure 4.18: Propulsion system distribution current changes with distributed propulsion. In contrast, distribution lowers the mass of the propulsors more signicantly and the overall system mass decreases with more distribution. However, as the number of propulsors is increased further, the eect diminishes, largely due to the increased mass of thermal management system (TMS) required to deliver cooling to the increased number of motors. For 20 smaller motors versus 2 larger ones, the iron losses in the motor remain constant while the resistive losses decrease; as a result, the power loss of each motor is a larger fraction of the output power. Collectively across 20 motors, this means more total heat has to be rejected, requiring a larger TMS. Therefore, with 20 propulsors, the weight savings from the smaller propulsors fail to provide overall system weight benets compared with 10 propulsors. Figure 4.19 shows the eect of distributed propulsion on component eciency. On the left, the ef- ciency of the motor increases across all ight segments. Smaller motors have lower copper (resistive) losses while their core losses per unit power stays constant, reducing overall losses and hence increasing eciency. The motor is sized to meet climb power requirements, so it operates most eciently there, but its o-design performance improves with increased distribution. With more motors, the fraction of peak power to o-peak power in cruise and approach decreases, thus increasing eciency by reducing copper 109 0 5 10 15 20 Number of Propulsors 0.75 0.8 0.85 0.9 0.95 1 Component Efficiency [-] Converter, Climb Converter, Cruise Converter, Approach Motor, Climb Motor, Cruise Motor, Approach 0 5 10 15 20 Number of Propulsors 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Overall Efficiency [-] Climb Cruise Approach Figure 4.19: Eciency changes with distributed propulsion; (left) for the converter and motor, and (right) showing the overall system eciency. losses while core losses stay constant. The converter eciency decreases with distribution, as its switch- ing losses (due to the on-resistance of the diodes and switching elements) remain constant, but at lower power, have a larger eect than the decreased resistive losses. Note that the converter is still sized smaller for 20 propulsors than for 2, but its eciency takes a minor hit. With decreased load resistance from the motors, the converter operates more eciently at o-peak power segments. These components all operate at over 75% eciency at all segments, and with increased distribution, going beyond 4 propulsors yields eciencies of over 85% across the board. The right side of Fig. 4.19 shows how the overall system eciency changes with distribution. Note that they-axis on this plot has a lower bound of zero compared to the left plot. First of all, the system eciencies across all the segments are substantially lower than the component eciencies of the left plot. Primarily, this is due to the propulsor eciency of 0:8, which would be of equivalent concern on a conventional aircraft as well. Further, power delivery chain also includes power to the TMS supplied by the battery, thus requiring additional power that does not get delivered to the ow. For all the cases plotted on this gure, the wiring had an eciency of over 99.5%, meaning its losses are negligible compared to those of the other components. 110 For climb, the system eciency drops with increasing distribution due to the TMS being larger. On the other hand, cruise and approach eciencies improve with distribution, largely due to the motor eciency improvement. Approach eciency is the lowest because this is where the motor, sized for the much higher climb power, is the least ecient. However, even in approach, the system eciencies of the all-electric aircraft model here are comparable to, or higher than the 40% achieved by current turboprops at design power [10]. This result conrms that all-electric aircraft have higher system eciencies compared to conventional aircraft for most of the mission, except for a lower system eciency at approach, which grows with increased distribution. 4.5 Limitations The major limitation of this work is that it looks at the propulsion system for an all-electric aircraft in isolation without considering the other aspects of aircraft sizing — fuselage, wings, etc. — with the same level of detail. The power prole of a commuter aircraft is used to size the electrical components needed to provide that power at intermediate 2035 technology levels. The sizing of electrical components then iteratively resizes the aircraft, albeit in a very basic manner. Integrating this propulsion system framework into a more detailed aircraft sizing tool to better understand aircraft-level results should be the primary goal of any future work. There is ongoing work at USC to develop the Library for Unied Conceptual Aircraft Synthesis (LUCAS) framework that models the sizing and mission at higher delity. The models from this work will eventually be integrated into the LUCAS framework. Additionally, this work did not consider the volume constraints of the components, specically the wiring and the thermal management with its network of pipes. These two details are rather conguration- specic and was outside the scope of this work, given the focus on the performance of electrical compo- nents during ight. 111 Other limitations include the focus on commuter aircraft like the Twin Otter with a payload of 19 passengers, and the focus on an all-electric propulsion system. The results from the LEARN3 work in Chapter 3 and [42, 11] showed promise for hybrid- and turbo-electric aircraft for larger classes ying longer missions, and to provide the same level of analysis for other aircraft classes and propulsion system architectures (including turbo-electric and hybrid-electric), this framework could be expanded to cover those as well. 112 Chapter5 Conclusions This dissertation focused on the modeling of electrical components in the propulsion system for electried aircraft, namely, the battery, converter (power electronics), motor (electric machine), wiring, and thermal management. The parameters important to model these components at low-delity — specic energy, specic power, and eciency — were tabulated across a variety of sources. The values of these parameters were predicted for a 2035 timeline in three scenarios: a conservative estimate suggesting only modest im- provements over the start-of-the-art today, an optimistic estimate assuming commercialization of various technologies that are currently at testbed level, and an intermediate estimate taking the average of the rst two scenarios. These numbers were also compared with what other researchers and industry sources predict for a similar timeline. For battery technology to improve signicantly, it was found that novel chemistries must be commercialized. The current lithium-ion chemistry does not improve battery specic energy (BSE) more than 1.5 times over what is seen today. Additionally, the losses in BSE going from chem- ical reaction-based theoretical values to cell-level values quoted in literature to pack-level values suitable for aircraft designs were quantied. Looking at the eects of technology level for a 20-passenger all-electric aircraft ying 100 nmi using a low-delity approach, it was found that improving technology made all-electric aircraft feasible and benecial with intermediate 2035 technology, and the energy consumption improved signicantly with 113 optimistic 2035 technology. Increasing BSE was found to have much larger eects than increasing specic powers of other components, because the battery, carrying all of the on-board energy, was signicantly heavier than other components. Improvement in component specic powers also yielded PSEC benets, but with diminishing returns as technology evolved beyond optimistic 2035 levels. Both DP and BLI im- proved the feasibility and energy consumption of all-electric aircraft, providing additional benets over those from increased power delivery chain eciency. Based on the promises seen in low-delity design space exploration, a targeted study of the behavior of electrical components of the propulsion system was conducted at higher delity, to see how they would behave at dierent ight segments with varying power loads. Operational models of the battery, converter, motor, wiring, propulsor, and thermal management system (TMS) were developed, and their performance analyzed for a commuter aircraft mission carrying 19 passengers over 100 nmi, representative of a Twin Otter. Components were sized for the highest-power segment, and subsequently simulated to observe their performance at lower-power segments. The motor, most ecient at its design power, sees its eciency drop at lower power due to the constant core losses that scale as a function of its size despite the reduc- tion in copper (resisitive) losses. The converter was more ecient at lower power, due to the lower load resistance from the motors. As a result of the drop in motor eciency, the higher-delity model predicted about 120% higher energy requirements for the same mission as the low-delity model, since the battery had to be sized to provide enough power to overcome the lower eciency in o-design segments. Cruise, despite lasting longer than climb time-wise, needed 13% less energy than climb, and about half as much power as climb. Power losses in the wiring and its weight were found to be negligible compared to the takeo mass of the aircraft, with negligible dierences from using copper versus aluminum wires. For the representative commuter aircraft with a service ceiling of 27,600 ft, the safe operating voltage (SOV) was found to be about 3 kV after applying a safety factor of 1.5 and assuming insulated conductors spaced at least 2 mm apart. This SOV is an order of magnitude larger than the 327 V set by Paschen’s law to 114 prevent electrical arcing. This result was used to size the distribution wiring based on what current would be needed to distribute power through the system. For larger aircraft like the Boeing 777 and the Airbus A350 with a service ceiling of 43,000 ft, the SOV is about half as much at 1.65 kV. The losses in wiring scale with the square of current, so enabling lower operating currents through high system voltages reduces the losses in the system, resulting in a smaller battery, TMS, and wiring, and ultimately bringing down the aircraft mass. In terms of the mass breakdown, the battery dominated the propulsion system mass, making up slightly inder two-thirds of the total. Propulsors constituted the next largest fraction. Distributed propulsion lowered the total mass of the system, with the largest benet going from 2 to 4 propulsors, largely due to decrease in propulsor and wiring masses. Beyond 4 propulsors, the system mass decreased, plateauing at 8 propulsors, then increasing slighty for more propulsors, largely due to the diminished reductions in wiring and propulsor mass combined with a larger TMS to reject heat losses from multiple motors. Distribution also lowered the maximum current used for power distribution. Finally, the motors became more ecient with more distribution across all segments while the propulsors became slightly less ecient. The overall system eciency at climb dropped slightly going from 2 to 20 propulsors, due to the larger TMS, but cruise and approach eciencies improved to a greater degree, due to the improvement in motor eciency. While the system eciency at approach for 26 propulsors dropped below that for a conventional hydrocarbon- based propulsion system, the system eciency matched or signicantly bettered that of the conventional across all other mission segments at all the numbers of propulsors considered. This result showed that all-electric aircraft do oer a power delivery chain eciency advantage over conventional aircraft, even in a majority of the low-power segments for the scenario considered here. To summarize, the results form this work show that for a commuter aircraft ying 19 passengers over 100 nmi, an all-electric architecture with 6—8 propulsors results in the lowest takeo mass. It is important to keep the distribution current low by increasing the system voltages in order to avoid the resistive losses 115 that scale with current squared. Since the battery accounts for about 2/3 of the mass of the propulsion system, advances in improving the BSE lead to substantial improvements in the aircraft energy eciency, much more so than improving the specic powers or the eciencies of other components. 116 Bibliography [1] International Air Transport Association (IATA). 2036 Forecast Reveals Air Passengers Will Nearly Double to 7.8 Billion. https://www.iata.org/pressroom/pr/Pages/2017-10-24-01.aspx. Accessed: 08/02/2018. 2018. [2] E. Shulz. Global Networks, Global Citizens: Global Market Forecast 2018-2037. https://www.airbus.com/content/dam/corporate-topics/publications/media-day/Presentation- Eric-Schulz-GMF-2018.pdf. Accessed: 08/02/2018. 2018. [3] Randy Tinseth. Commercial Market Outlook 2018-2037. https://www.boeing.com/resources/boeingdotcom/commercial/market/commercial-market- outlook/assets/downloads/2018-cmo-09-11.pdf. Accessed: 02/02/2019. 2018. [4] The National Academies of Sciences, Engineering, and Medicine. Commercial Aircraft Propulsion and Energy Systems Research: Reducing Global Carbon Emissions. NAE, 2016. [5] M.K Bradley and C.K Droney. “Subsonic Ultra Green Aircraft Research: Phase II – Volume II – Hybrid Electric Design Exploration”. In: (2015). [6] Mark Drela. “Development of the D8 Transport Conguration”. In: 29th AIAA Applied Aerodynamics Conference (2011).doi: 10.2514/6.2011-3970. [7] J.R Welstead and J.L Felder. “Conceptual Design of a Single-Aisle Turboelectric Commercial Transport with Fuselage Boundary Layer Ingestion”. In: 54th AIAA Aerospace Sciences Meeting (2016). [8] Airbus Group. E-Fan The New Way to Fly. http: //www.airbusgroup.com/service/mediacenter/download/uuid=48b1bd2c-a428-4c65-82e5-ed3e923bd142. Accessed: 07/26/2016. 2015. [9] Eviation. Eviation Alice. https://www.eviation.co/aircraft/. Accessed: 02/15/2021. 2021. [10] Martin Hepperle. “Electric Flight-Potential and Limitations”. In: (2012). 117 [11] D.K. Hall, E.M. Greitzer, A.P. Dowdle, J.J. Gonzalez, W.W. Hoburg, J.H. Lang, J.S. Sabnis, Z.S. Spakovszky, B. Yutko, C. Courtin, W. Thalheimer, L. Trollinger, J. Tylko, N. Varney, A. Uranga, S. Byahut, and M. Kruger.FeasibilityofElectriedPropulsionforUltra-EcientCommercialAircraft, Final Report. 2018. [12] A. Uranga, M. Drela, E.M. Greitzer, D. Hall, N.A. Titchener, M.K. Lieu, N.M. Siu, A.C. Huang, G.M. Gatlin, and J.A. Hannon. “Boundary Layer Ingestion Benet of the D8 Aircraft”. In: AIAA Journal 55.11 (2017), pp. 3693–3708.doi: 10.2514/1.J055755. [13] Alejandra Uranga, Mark Drela, David Hall, and Edward M. Greitzer. “Boundary Layer Ingestion Benet of the D8 Aircraft”. In: AIAA Journal 56.11 (2018), pp. 4271–4281.doi: 10.2514/1.J056781. [14] Mark D Moore and Bill Fredericks. “Misconceptions of electric propulsion aircraft and their emergent aviation markets”. In: American Institute of Aeronautics and Astronautics. 2014. [15] National Transportation Safety Board (NTSB). Auxiliary Power Unit Battery Fire Japan Airlines Boeing 787-8, JA829J Boston, Massachusetts January 7, 2013 (Incident Report). https://www.ntsb.gov/investigations/AccidentReports/Reports/AIR1401.pdf. Accessed: 08/02/2016. 2014. [16] K.R Antcli, M.D Guynn, Marien T.V, D.P Wells, S.J Schneider, and M.T Tong. “Mission Analysis and Aircraft Sizing of a Hybrid-Electric Regional Aircraft”. In: SciTech 2016 (2016). [17] C. Friedrich and P.A. Robertson. “Hybrid-Electric Propulsion for Aircraft”. In: Journal of Aircraft 52.1 (2015). [18] C. Friedrich and P.A. Robertson. “Hybrid-Electric Propulsion for Automotive and Aviation Applications”. In: CEAS Aeronautical Journal 6.2 (2015), pp. 279–290. [19] Anil Yildirim, Justin S. Gray, Charles A. Mader, and Joaquim R. R. A. Martins. “Performance Analysis of Optimized STARC-ABL Designs Across the Entire Mission Prole”. In: 2021.doi: 10.2514/6.2021-0891. [20] Hyun Dae Kim, James L. Felder, Michael T. Tong, and Michael Armstrong. “Revolutionary Aeropropulsion Concept for Sustainable Aviation: Turboelectric Distributed Propulsion”. In: 21st International Society for Air Breathing Engines (2013). [21] J. Michael Vegh, Juan J. Alonso, Tarik H. Orra, and Carlos R. Ilrario de Silva. “Flight Path and Wing Optimization of Lithium-Air Battery Powered Passenger Aircraft”. In: 2015.doi: 10.2514/6.2015-1674. [22] S. Stueckl, J. van Toor, and H. Lobentanzer. “Voltair - The All-Electric Propulsion Concept Platform – A Vision for Atmospheric Friendly Flight”. In: 28th International Congress of the Aeronatical Sciences (ICAS) (2012). [23] C Pornet and A.T Isikveren. “Conceptual design of hybrid-electric transport aircraft”. In: Progress in Aerospace Sciences 79 (2015), pp. 114–135. 118 [24] Askin Isikveren, Arne Seitz, Patrick Vratny, Clement Pornet, and Kay Ploetner. “Conceptual Studies of Universally-Electric Systems Architectures Suitable for Transport Aircraft”. In: Sept. 2012. [25] H. Steiner, P.C. Vratny, C. Gologan, K. Wieczorek, A.T. Isikveren, and M. Hornung. “Optimum Number of Engines for Transport Aircraft Employing Electrically Powered Distributed Propulsion”. In: CEAS Aeronaut Journal 5 (2014), pp. 157–170. [26] Hans-Joerg Steiner, Arne Seitz, Kerstin Wieczorek, Kay Ploetner, Askin T. Isikveren, and Mirko Hornung. “Multi-Disciplinary Design and Feasiblility Study of Distributed Propulsion Systems”. In: 28th International Congress of the Aeronatical Sciences (ICAS) (2012). [27] M. Drela. “Power Balance in Aerodynamic Flows”. In: AIAA Journal 47.7 (2009), pp. 1761–1771. [28] Zunum Aero. Zunum Aero Unveils Details on Launch Hybrid-Electric Aircraft Oering Unprecedented Economics. https://www.prweb.com/releases/2017/10/prweb14750814.htm. Accessed: 09/26/2018. 2017. [29] US Energy Information Administration (EIA). Electricity in the United States. https://www.eia.gov/energyexplained/index.php?page=electricity_in_the_united_states. Accessed: 09/16/2018. 2018. [30] Ralph H. Jansen, Cheryl Bowman, Amy Jankovsky, Roger Dyson, and James Felder. “Overview of NASA Electried Aircraft Propulsion Research for Large Subsonic Transports”. In: 53rd AIAA/ SAE/ ASEE Joint Propulsion Conference (2017). [31] X. Wang, C. Babbitt, G. Gaustad, and M. Ganter. “Economic and environmental characterization of an evolving Li-ion battery waste stream”. In: Journal of Environmental Management (2014). [32] Ralph J. Brodd, ed. Batteries for Sustainability: Selected Entries from the Encyclopedia of Sustainability Science and Technology. Springer, 2013. [33] J.B. Straubel. Energy Storage, EVs and the Grid (Presentation). https://www.eia.gov/conference/2015/pdf/presentations/straubel.pdf. Accessed: 08/02/2016. 2015. [34] Tao Zhang, Nobuyuki Imanishi, Yuta Shimonishi, Atsushi Hirano, and Yasuo Takeda. “A novel high energy density rechargeable lithium/air battery”. In: Chemical Communications 46 (2010), pp. 1661–1663. [35] H Kuhn and A Sizmann. Fundamental prerequisites for electric ying. Deutsche Gesellschaft für Luft-und Raumfahrt-Lilienthal-Oberth eV, 2012. [36] Venkat Srinivasan, David Hafemeister, B Levi, M Levine, and P Schwartz. “Batteries for vehicular applications”. In: AIP conference proceedings. Vol. 1044. 1. AIP. 2008, pp. 283–296. [37] G. Abbe and H. Smith. “Technological Development Trends in Solar-Powered Aircraft Systems”. In: Renewable and Sustainable Energy Reviews 60 (2016), pp. 770–783. 119 [38] Clément Pornet, Corin Gologan, Patrick C Vratny, Arne Seitz, Oliver Schmitz, Askin T Isikveren, and Mirko Hornung. “Methodology for sizing and performance assessment of hybrid energy aircraft”. In: Journal of Aircraft 52.1 (2014), pp. 341–352. [39] Y. Gibbs. NASA Armstrong Fact Sheet: NASA X-57 Maxwell. https://www.nasa.gov/centers/armstrong/news/FactSheets/FS-109.html. Accessed: 05/07/2018. 2017. [40] D.P. Raymer. Aircraft Design: A Conceptual Approach, Fifth Edition. American Institute of Aeronautics and Astronautics (AIAA), 2012. [41] Mark Drela. TASOPT 2.00 Transport Aircraft System OPTimization: Technical Description. http://web.mit.edu/drela/Public/N+3/TASOPT_doc.pdf. Accessed: 05/08/2018. 2010. [42] Michael Kruger, Saakar Byahut, Alejandra Uranga, Aidan Dowdle, Jonas Gonzalez, and David K. Hall. “Electried Aircraft Trade-Space Exploration”. In: 2018 Aviation Technology, Integration, and Operations Conference (2018).doi: 10.2514/6.2018-4227. [43] Edward Burnell and Warren Hoburg. GPkit software for geometric programming. https://github.com/convexengineering/gpkit. Version 0.7.0. 2018. [44] Brian J. German. Introduction to Batteries and Cell Discharge Modeling for Electric Aircraft Design (Lecture and Presentation). 2018. [45] NASA. Scalable Convergent Electric Propulsion Technology and Operations Research (SCEPTOR) Critical Design Review Day 1 Package. https://www.nasa.gov/sites/default/files/atoms/files/sceptor_cdr_day_1_package.pdf. 2016. [46] R. W. Erickson and D. Maksimovic. Fundamentals of Power Electronics, Second Edition. Kluwer Academic Publishers, 2004. [47] S. Sundaresan, C. Sturdevant, M. Marripelly, E. Lieser, and R. Singh. “12.9 kV SiC PiN diodes with Low On-state Drops and High Carrier Lifetimes”. In: Materials Science Forum 717-720 (2012). [48] Q. Zhang, M. Das, J. Sumakeris, R. Callanan, and A. Agarwal. “12-kV p-Channel IGBTs with Low On-Resistance in 4H-SiC”. In: IEEE Electron Device Letters 29.9 (2008). [49] Saakar Byahut and Alejandra Uranga. “Propulsion Powertrain Component Modeling for an All-Electric Aircraft Mission”. In: AIAA 2020-0015, 2020 AIAA SciTech, Orlando, FL, 6–10 January. 2020.doi: 10.2514/6.2020-0015. [50] I. Christou, A. Nelms, I. Cotton, and M. Husband. “Choice of optimal voltage for more electric aircraft wiring systems”. In: IET Electrical Systems in Transportation 1.1 (Mar. 2011), pp. 24–30.doi: 10.1049/iet-est.2010.0021. [51] E. Husain and R. S. Nema. “Analysis of Paschen Curves for air, N2 and SF6 Using the Townsend Breakdown Equation”. In: IEEE Transactions on Electrical Insulation EI-17.4 (Aug. 1982), pp. 350–353.doi: 10.1109/TEI.1982.298506. 120 [52] Michael J. Armstrong, Mark Blackwelder, Andrew Bollman, Christine Ross, Angela Campbell, Catherine Jones, and Patrick Norman. “Architecture, Voltage, and Components for a Turboelectric Distributed Propulsion Electric Grid (Final Report)”. In: Accessed: 07/08/2020. NASA/CR—2015-218440. 2015. [53] European Union Aviation Safety Agency (EASA). Airbus A350 Type-Certicate Data Sheet. https://www.easa.europa.eu/sites/default/files/dfu/TCDS_EASA%20_A_151_Airbus350_Iss22-2019- 11-27.pdf. Accessed: 07/08/2020. 2019. [54] H. H. Farr. Transmission Line Design Manual. United States Department of the Interior Water and Power Resources Service, 1980. [55] Robert Doering and Frank Thielecke. “An Evaluation of the Potential of Local Electric Power Packages on Aircraft Fleet Level”. In: AIAA 2019-2804, 2019 AIAA Aviation Forum, Dallas, TX, 17–21 January. 2019.doi: 10.2514/6.2019-2804. [56] R. Krishnan. Switched Reluctance Motor Drives. CRC Press, 2001. [57] R. A. Kisner, A. M. Melin, A. Drira, F. K. Reed, and D. Fugate. “Embedded Sensors and Controls to Improve Component Performance and Reliability – Bench-scale Testbed Design Report”. In: (2015). doi: 10.13140/RG.2.1.2554.9525. [58] M. Torrent, P. Andrada, B. Blanque, E. Martinez, J. I. Perat, and J. A. Sanchez. “Method for estimating core losses in switched reluctance motors”. In: European Transactions on Electrical Power (2010).doi: 10.1002/etep.475. [59] Egbert Torenbeek. Advanced aircraft design: Conceptual design, technology and optimization of subsonic civil airplanes. John Wiley & Sons, 2013. [60] Jeryes W. Chapman, Sydney L. Schnulo, and Michael P. Nitzsche. “Development of a Thermal Management System for Electried Aircraft”. In: AIAA 2020-0015, 2020 AIAA SciTech, Orlando, FL, 6–10 January. 2020.doi: 10.2514/6.2020-0545. [61] J.E. Hesselgreaves. Compact Heat Exchangers: Selection, Design, and Operation. Elsevier Science & Technology Books, 2001. [62] H.S. Lee. Thermal Design: Heat Sinks, Thermoelectrics, Heat Pipes, Compact Heat Exchangers, and Solar Cells. John Wiley & Sons, Inc., 2010.isbn: 9781118004708. [63] Viking Air Ltd. Viking Twin Otter Series 400 Technical Specications and Standard Equipment List. 2016. [64] S. Ramanujan, G.H. Hardy, P.V.S. Aiyar, and B.M. Wilson. Collected Papers of Srinivasa Ramanujan. Cambridge University Press, 2015.isbn: 9781107536517. [65] Bruce S. Bernstein. Electrical Properties of Cable Insulation Materials (Presentation). https://www.pesicc.org/iccWebSite/subcommittees/E/E04/2001/f2001_bernstein.pdf. Accessed: 07/08/2020. 2001. 121

Abstract (if available)

Abstract Commercial aircraft currently rely on hydrocarbons as the sole source of energy for propulsion. The variability of fuel prices, the growing emphasis on environmental sustainability, and the increased demand for air transportation have led to enhanced interest in improving fuel efficiency and reducing emissions for transport aircraft. Electrification of the aircraft propulsion system has the potential to achieve both goals. It can also better leverage the benefits of distributed propulsion (DP) and boundary layer ingestion (BLI) to further enhance energy efficiency. Electrification, however, poses major challenges. Batteries provide substantially less energy per unit mass than hydrocarbon fuel, which means battery-powered aircraft will weigh more than hydrocarbon-fueled aircraft for the same mission. Replacing conventional propulsion systems with electrified ones is thus not expected to be beneficial due to added weight and complexities. Despite the challenges, there are scenarios for which electrification could be beneficial. ❧ This work focuses on the modeling of electrified propulsion system components. The use of a unified propulsion system that can handle all architectures from conventional to hybrid-electric to all-electric enables comparison between different architectures and is used to put together the models of various components. A technology overview captures trends in the development of electrical component technologies, based on predictions of how the masses and efficiencies of these components will improve in the far term (2035), and thus help reduce the electrified propulsion system weight. A low-fidelity framework was developed to capture the major trade-offs of electrification at cruise condition, as well as the effects of DP and BLI. Results from this framework applied to a 20-passenger commuter aircraft flying 100 nmi showed that all-electric aircraft become feasible and beneficial over conventional after substantial advances in battery technology. Other components like the motor and the converter play a less meaningful role, but improvements to both also yield energy savings, but the benefits plateau with technology advances beyond the 2035 scenarios, simply because the battery makes up a much larger fraction of the propulsion system mass than the other components. Both DP and BLI improve the feasibility and energy efficiency of the all-electric aircraft considered. ❧ Limitations of the low-fidelity framework with regard to the modeling of electrical components led to interest in analyzing the behavior of these components under more representative operational loads. Higher-fidelity models of batteries, electrical machines, power electronics, wiring, and thermal management systems (TMS) are developed. These components are integrated into the unified propulsion system model to better evaluate an all-electric architectures and gain a better understanding of how the components behave under flight loads. The results are used to better model their masses for a more accurate estimate of the weight of the electrified propulsion systems as well as their efficiencies. For the representative mission of a 20-passenger aircraft flying 100 nmi, the electrical components are sized for the highest-power climb segment, and operated at lower-power cruise and approach segments. This higher-fidelity model predicts about a 120% higher energy requirement than the previous low-fidelity approach. Motor efficiency drops substantially at lower-power segments from 95% at climb to 77% at cruise for a 2-propulsor aircraft, and the resulting inefficiency requires both a larger battery and TMS. A safe operating voltage of 3 kV is determined based on the maximum service ceiling of the baseline aircraft and then used to set the distribution current and wire gauge. In terms of masses, the battery dominates, and DP helps lower the overall system mass by 27% and also improves the system efficiency at off-design power loads, from 30% to 45%, both for a 20-propulsor aircraft compared to a 2-propulsor aircraft. For the missions considered, all-electric aircraft are found to have double the source-to-load power conversion efficiency than conventional aircraft, with the efficiency increasing by up to 5% with more DP.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

The challenges and potential benefits of electrified propulsion for aircraft

PDF

Development and applications of a body-force propulsor model for high-fidelity CFD

PDF

High temperature latent heat thermal energy storage to augment solar thermal propulsion for microsatellites

PDF

Energy-efficient shutdown of circuit components and computing systems

PDF

A joint framework of design, control, and applications of energy generation and energy storage systems

PDF

Modeling and analysis of parallel and spatially-evolving wall-bounded shear flows

PDF

Design and analysis of reduced complexity transceivers for massive MIMO and UWB systems

PDF

Efficient inverse analysis with dynamic and stochastic reductions for large-scale models of multi-component systems

PDF

Energy proportional computing for multi-core and many-core servers

PDF

Nano-engineered devices for display and analog computing

PDF

Multidimensional characterization of propagation channels for next-generation wireless and localization systems

PDF

Modeling and simulation of multicomponent mass transfer in tight dual-porosity systems (unconventional)

Asset Metadata

Creator Byahut, Saakar (author)

Core Title Modeling and analysis of propulsion systems and components for electrified commercial aircraft

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Aerospace Engineering

Publication Date 04/30/2021

Defense Date 03/10/2021

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag aircraft design,all-electric,battery,converter,electric aircraft,electric propulsion,hybrid-electric,motor,OAI-PMH Harvest,propulsion system components,turbo-electric

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Uranga, Alejandra (committee chair), Bradley, Marty (committee member), Jovanovic, Mihailo (committee member), Spedding, Geoff (committee member)

Creator Email byahut@usc.edu,saakar.byahut@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-457649

Unique identifier UC11669111

Identifier etd-ByahutSaak-9574.pdf (filename),usctheses-c89-457649 (legacy record id)

Legacy Identifier etd-ByahutSaak-9574.pdf

Dmrecord 457649

Document Type Dissertation

Rights Byahut, Saakar

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA