Rapid, iterative design approach for preliminary configuration of a High Altitude, Long Endurance aircraft

This paper presents a rapid iterative design approach, used to develop a preliminary configuration of an Unmanned Aerial Vehicle (UAV), which is to perform a High-Altitude, Long Endurance (HALE) mission. First, based on a set of requirements, an initial aircraft concept is developed. The rapid iterative design approach revolves around an energy-based flight profile simulation, which is updated with an improved aerodynamic model of the aircraft during every iteration. An aircraft configuration is defined by a set of constant parameters. A corresponding flight profile consists of an altitude and velocity profile, as a function of time. The behaviour of the aircraft is simulated based on the specified flight profile. For sizing the aircraft components exposed to the airstream, the aerodynamic design phase aims to create an airframe which achieves the performance outlined in the flight performance simulation, utilizing the obtained aircraft parameters. Using the geometric data, a 3D-model of the aircraft is created, which is essential for the aircraft’s weight and balance calculations and for determining the airframe shape exposed to the airflow. A Computational Fluid Dynamics (CFD) simulation is conducted to obtain the actual aerodynamic characteristics of the airframe. If the performance requirement of the flight performance simulation is not met, another iteration of the aerodynamic design phase is required. The design framework stands out due to its rapid iteration capability, accessibility, and straightforward application, making it particularly advantageous for educational settings in which students engage with aircraft design. The approach aims to offer satisfactory results for an initial aircraft conceptualization. This design framework emerged from the DLR Design Challenge 2023, an annual competition hosted by the German Center for Aerospace Research, where student teams develop preliminary aircraft designs. This years challenge tasked the teams with designing an aircraft system to restore internet coverage over a disaster-affected region. The described approach was used to develop the Sentinel System of the DHBW Ravensburg team. However, the simulation and development process is not limited to applications within this challenge.


Introduction
This paper introduces an iterative design process used to develop a preliminary configuration of an Unmanned Aerial Vehicle (UAV).
The following set of requirements stems from the competition hosted by the DLR.For other aircraft design tasks, a different set of requirements is to be used.The Sentinel aircraft is used as an example, however, the methodology is not limited to this aircraft in particular.The main point of focus is the development approach [1] [2]: • Operational altitude of 18000 m • Flight time of 48 hours • Flying at an airspeed of 90 m s for two hours • A payload with a mass of 43 kg and a power draw of 3965 W is to be carried and operated during the entire flight

State of the art
Depending on the amount of information available on the aircraft, different methods of flight simulation are available.If, for instance, the complete state space of an aircraft is known, a 6dof-simulation of the aircraft might be conducted, by solving the full set of equations of motion [3].Depending on how well explored the aerodynamic properties of the aircraft are, non-linearities of the system should be considered.However, during the early concept phase of an aircraft development project, the system matrices are still unknown, requiring a lower-fidelity methodology to obtain flight endurance and range.Many approaches use the Breguet equation to make an initial guess at the range and flight time of an aircraft [4,5].The following basic parameters are needed for the equation: lift-to-drag ratio L D , the mass at take-off m 0 , end mass m 1 , the gravitational acceleration g and the specific fuel consumption SF C, speed V as well as the propulsion system efficiency η prop , the equation of flight time for a propeller driven aircraft is: However, the Breguet equation is only valid for flight at a constant velocity and coefficient of lift [6].This means that the aircraft gains altitude, as mass is constantly lost due to fuel consumption.Different trajectories of constant altitude, during which engine power or velocity are varied instead, can not sufficiently modeled using the Breguet equation.Furthermore, other changes in the energy state of an aircraft, such as payload or on-board power requirements, are not modeled.

Details of methodologies
Upon investigating the applications of both methodologies of aircraft performance simulation, it appears that both approaches are unsuitable for the task at hand.The Brequet range equation for example only considers aerodynamic drag.Other changes in the energy state of the aircraft, such as changes in altitude, are disregarded.Due to the requirement of climbing to 18000 m, changes in potential energy are considered to be a significant factor for the aircraft performance calculation.This makes the Breguet equation unsuitable to sufficiently determine aircraft performance for the design mission.
Although the known state-space system of an aircraft allows for very accurate simulation, the derivatives to populate the system matrices have to be determined in the first place.This requires extensive data from simulations or tests, meaning that the state space system only becomes known in later stages of the design process, once the initial concept design is completed.This is not the case for the design task at hand.Instead, the initial concept design is to be conducted, which allows for the formulation of the state-space system of an aircraft later on.In conclusion, both presented design -and simulation-approaches are unsuitable for the given task, meaning that a custom design method needs to be developed, populating a middle-ground between the methods presented.

Aircraft Design
Once a suitable aircraft configuration has been found via the flight profile simulation, an airframe has to be designed which is able to achieve the required performance.This is done via initial sizing methods.Once the airframe is complete, a CFD simulation is conducted to obtain a new aerodynamic model for the next iteration of the flight profile simulation.

Propulsion system
The propulsion system is defined by a certain type of engine and therefore efficiency η, gravimetric energy density of the propellant ϕ and behaviour with respect to altitude and velocity.
The performance of different propulsion systems changes, if the airspeed varies.On the one hand, the power output of jet engines remains approximately the same at different air speeds.On the other hand, a piston engine powering a propeller outputs more power if the airspeed decreases [5].This effect has to be considered when choosing a propulsion system for a given mission profile.Combined with the engine efficiency and the specific energy of the fuel, the mass flow of propellant is calculated.The example aircraft is propelled by a Rotax 914UL piston engine, equipped with a tripleturbocharger to facilitate operation at 18000 m [7].The engine is to be fueled by AvGas with ϕ = 44.5 MJ kg [1].The entire power train achieves an efficiency of η = 0.28 [1].

Airfoil
The NFL1015 asymmetric laminar-flow airfoil is chosen, as it offers very high lift-to-drag ratios of up to L D = 165, when flown at RE = 10 6 .The downsides are a low stall angle of α stall = 5 • and a limited range of Reynolds-numbers to achieve high lift-to-drag ratios.The airfoil was originally developed by NASA for a proposed research aircraft with a similar mission profile to the example aircraft [8], being flown in loitering patterns at high altitude.The limited angle of attack and sensitivity to changes of the Reynolds number are considered as acceptable for a loitering mission at high altitude.

Wing platform
During the design process of the main wing, care must be taken to achieve a balance between aerodynamic efficiency and practicality.High aspect ratios AR increase the efficiency of the wing, expressed as the Oswald -factor ϵ.However, gust tolerance and ground handling capabilities of the aircraft are reduced.Following a further investigation of the requirements of the example aircraft, an aspect ratio of AR = 15 is selected.An initial conservative estimate of ϵ = 0.8, following comparisons with similar aircraft, is made.

Iterative aircraft design process
The following chapter gives an overview of the design process.It should be noted that a purely energetic simulation of the aircraft configuration is performed.No aerodynamic design or optimization is conducted within the algorithm presented.Instead, a basic aerodynamic model of the aircraft is handed as a set of input parameters to the algorithm, with an initial guess being required as a starting point.This might be taken from an existing aircraft design with a similar airframe design.The flight profile simulation is used to rapidly iterate aircraft parameters as well as the flight profile, to find the optimal aircraft configuration for the design mission.Although the lightest, most efficient aircraft is desirable for an economic as well as ecologic design, sufficient margins are to be achieved for all flight states.Once a suitable aircraft configuration has been found, the obtained parameters are used to drive a linked 3D-model.Miscellaneous parts, such as gear fairings, the wing root or the wing tips are modelled manually.All components are positioned by a dynamic center of mass calculation.The main considerations of the airframe design are: • Fitting the airframe into the available mass budget, as obtained by the flight profile simulation • Achieving the required aerodynamic performance • Achieving compatibility with the design requirements A CFD-simulation using SimCenter StarCCM+ of the resulting geometry is conducted to verify the aerodynamic performance of the airframe or obtain a new aerodynamic model for another iteration of the algorithm.This is repeated, until the required aerodynamic performance is achieved.In case this is not possible due to physical limitations, the flight profile simulation has to be revisited and an aircraft configuration with more conservative aerodynamic characteristics has to be found.

Flight profile simulation
During each iteration of the algorithm, the energetic behaviour of an aircraft configuration is investigated.In order to achieve this, multiple sources of power during each flight state are considered [9].These are the following: • Power to overcome aerodynamic drag • Power due to altitude changes • Miscellaneous on-board power, such as flight controls, avionics and the payload To rapidly evaluate the performance of different aircraft configurations over a given flight profile, certain flight parameters are of interest, such as the angle of attack, or the engine power.The assumed flight profile of the Sentinel -aircraft is rather simple: (i) Takeoff & ascent to 18000 m (ii) Cruise flight at 18000 m (iii) Dash at 90 m s true air speed for two hours (iv) Cruise flight at 18000 m until a time of flight of 48 hours is reached

Simulation approach
The goal of the flight profile simulation is to approximate the behaviour of a certain aircraft configuration over a given flight profile.The aircraft configuration is to be specified by fixed parameters such as empty and fuel mass, engine efficiency or aerodynamic coefficients.An aerodynamic optimization is conducted outside of the presented simulation algorithm.The flight profile is defined by two functions describing the aircraft velocity v(t) as well as the aircraft altitude h(t) as a function of time.Furthermore, each flight state is assumed to be horizontal, without changes in altitude and velocity.The energy required for changes in altitude is accounted for separately.The mass of fuel burned in the time-step between two points on the flight profile is calculated to obtain the new aircraft mass for the next time-step.

Aerodynamic performance
An initial guess at the aerodynamic performance of the aircraft is required.As a more detailed airframe model is designed on the basis of each simulation, numerous iterations are needed until the aerodynamic performance of the airframe matches the assumed coefficients of the flight profile simulation.The final aerodynamic model of the entire airframe as well as the approximations that are implemented in the flight profile simulation are shown in figures 1 and 2 .

Simulation algorithm
In the following chapter, the simulation algorithm for the time t n is explained, and results in the aircraft parameters for the time t n+1 = t n + ∆t.
• The total aircraft mass m tot at time t n is calculated, being the sum of the empty mass m empty , the payload mass m pld as well as the fuel mass m f uel (t n ): • As a first step, the air density ρ(t) is calculated from the altitude profile h(t).
• Making use of the parameters of the aircraft configuration, the angle of attack α is calculated for the flight state, as by the assumptions that the lift L equals the gravitational force m tot •g.α 0 is the angle of attack at which C A equals zero: • As the drag and drag coefficient are a function of α, they can be calculated for the time step, with regard to the drag coefficient at zero lift, C D0 : • As the drag for the flight state has been obtained, the power to overcome the aerodynamic drag can be calculated for the time step, making use of the velocity, as specified for the time t n by the velocity profile: • Before calculating the engine power for the flight state with regard to η n , all other power requirements of the aircraft need to be considered: -Power requirements to conduct altitude changes p pot -Miscellaneous power of on-board systems p misc The power requirement for altitude changes is calculated from the altitude profile as well as the aircraft mass at t n , m tot (t n ): The total net power P tot for the flight state therefore equates to: • With the specified energy density of the fuel, ϕ, the amount of mass lost is calculated for the time t n and the time step ∆t: • Finally, a new total aircraft mass for the next loop at t n+1 = t n + ∆t is calculated:

Validation of the flight profile
To ensure that the implemented flight profile simulation algorithm results in sufficiently accurate results, it has been verified by simulating operational aircraft and their assumed flight profiles.As a reference, the U-2 aircraft was implemented based on publicly available data and was simulated over an approximate flight profile.The predicted time of flight is 15% higher than the stated performance of the U-2.This might be due to some inaccurate estimations of the aerodynamic characteristics of the aircraft or an unsuitable recreation of the U-2 's flight profile.However, considering the simplicity of the simulation algorithm, an accuracy of ±15% to calculate the operational endurance is considered to be acceptable.Therefore, the presented design process can be characterized as being very accessible, while still offering sufficient accuracy for the task at hand, therefore being a good solution for university design projects.For aircraft design projects entering a more detailed design phase, methods of higher fidelity are required.

Table 2 .
Requirement Check9.SummaryBy making use of basic approaches to the problems of aircraft design, a robust and technically feasible aircraft configuration has been found.All tools with the exception of Star-CCM+ are available to most engineering students.Other programming environments aside from MatLab can be substituted, should no license be available.Software like OpenFoam may be a feasible alternative to a full-fidelity CFD software.