a computer application for parametric aircraft design

© 2016 F R Fraqueiro et al , published by De Gruyter Open This work is licensed under the Creative Commons Attribution NonCommercial NoDerivs 3 0 License Open Eng 2016; 6 432–440 ICEUBI 2015* Open Acc[.]

ICEUBI 2015* Open Access

Filipe R Fraqueiro*, Pedro F Albuquerque, and Pedro V Gamboa

A computer application for parametric aircraft

design

DOI 10.1515/eng-2016-0067

Received Mar 30, 2016; accepted Sep 05, 2016

Abstract: The present work describes the development

and final result of a graphical user interface tailored for

a mission-based parametric aircraft design optimization

code which targets the preliminary design phase of

un-manned aerial vehicles This development was built from

the XFLR5 open source platform and further benefits from

two-dimensional aerodynamic data obtained from XFOIL

For a better understanding, the most important graphical

windows are shown In order to demonstrate the

graphi-cal user interface interaction with the aircraft designer, the

results of a case study which maximizes payload are

pre-sented

Keywords: PARROT; graphical user interface; UAV;

para-metric design; aircraft design; Air Cargo Challenge

1 Introduction

The aircraft designer needs to have a comprehensive

knowledge on the mainstream disciplines of aircraft

de-sign This Includes aerodynamics, propulsion, structures,

stability and performance, among others However, the

most challenging part of designing an aircraft is to

syn-thesize the mutual interactions among these disciplines in

order to achieve enhanced design solutions at the earliest

stages of the design process These earliest stages are

typi-cally about a powerful and duly weighted mix of intuition

*Corresponding Author: Filipe R Fraqueiro: Department of

Aerospace Sciences, University of Beira Interior, Covilhã, 6201-001,

Portugal; Email: fraqueirofilipe@gmail.com

Pedro F Albuquerque: Department of Aerospace Sciences,

University of Beira Interior, Covilhã, 6201-001, Portugal; Email:

pffa@ubi.pt

Pedro V Gamboa: Department of Aerospace Sciences, University of

Beira Interior, Covilhã, 6201-001, Portugal; Email: pgamboa@ubi.pt

* International Conference on Engineering 2015 – 2–4 Dec 2015 –

Uni-versity of Beira Interior – Covilhã, Portugal

and knowledge However, the large number of disciplines, the complexity of the aircraft physics and the multiple cou-plings between those disciplines complicates this task Nevertheless, the development of comprehensive mul-tidisciplinary design codes is gradually contributing to a paradigm change, in the way these are expected to rev-olutionize the design process While the earlier concep-tual design phase decision making-process is commonly still based on the designers themselves, multidisciplinary design optimization methodologies have proven that they can be particularly worthwhile in saving time and re-sources while getting closer to the global optimum at a pre-liminary design stage [1]

Amongst the different multidisciplinary design pro-grams which include a graphical user interface, it is worth-while to mention some cornerstone developments in the context of aircraft disciplinary analysis and design opti-mization

One of the earliest such works was Advanced Aircraft Analysis (AAA) [2], a tool which enables aircraft design and optimization as it allows a wide spectrum of analy-sis, despite being a complex software and requiring a li-cense AAA is divided into ten independent modules such

as weight, aerodynamics, performance, stability and con-trols, among others Due to its multidisciplinarity, this software allows a comprehensive aircraft design analysis and optimization even though the latter is generally user guided through an informal process

Also complex, although freeware, CEASIOM (Comput-erised Environment for Aircraft Synthesis and Integrated Optimisation Methods) [3] has a geometry module which makes it possible to have a general view of the aircraft ge-ometry under analysis It also includes modules related to stability, controls and aerodynamics

It is also worthwhile to refer XFRL5 [4] developed from XFOIL [5] at the Massachusetts Institute of Technology, which is a widely known code to calculate airfoils’ aero-dynamic coefficients and to analyze aircraft wings, fuse-lages, and empennages Despite being an easy, accessible and widely used tool, it does not enable automatic opti-mization of airfoils, lifting surfaces and/or fuselages Con-versely, the analysis data can be used by the designer for

A computer application for parametric aircraft design | 433

optimization purposes, but the overall optimization

work-flow will be far more toilsome

A more recent example of a design optimization

pro-grams is SUAVE [6] – developed at Stanford University

– which is a comprehensive tool with four calculation

methods: Traditional Aircraft Design, Advanced

Configu-ration/Unconventional Technology Design, Optimization,

Aircraft/Discipline Analysis SUAVE is open source and

has been written with the Python language It can be

incor-porated using extensible interfaces and prototyped with a

top-level script that allows the creation of arbitrary

mis-sion profiles, unconventional propulmis-sion networks, and

right-fidelity at right-time discipline analyzes

The objective of the present work was to develop a

graphical user interface (GUI) to enable an easier

interac-tion between the designer and an in-house developed code

for the Parametric AiRcRaft OpTimization (PARROT) [7] of

Unmanned Aerial Vehicles (UAVs) at the preliminary

de-sign phase

This program is built into the XFLR5 freeware

Graph-ical User Interface (GUI) as a subpart called Aircraft

Opti-mization that is internally linked to the design

optimiza-tion code (PARROT) XFLR5’s open source code is written

in the C++ language while PARROT is written in Fortran In

order to demonstrate this tool’s capabilities, the results of

a case study are shown

PARROT is a comprehensive program, which main

as-sets include its mission tailored optimization

methodol-ogy and the multidisciplinarity of the physical models

used (e.g propulsion, aerodynamics, performance,

stabil-ity, etc.) It was thus found that it would be valuable to

de-velop a graphical interface to facilitate the user’s

interac-tion and widen the number of possible design problems

solved For that, it was necessary to develop this GUI in

such a way that the user can easily interact with all the

inputs and analysis’ outputs To make this research work

available for the community, its authors aim to have the

PARROT source code available online in the near future

2 Methodology

2.1 PARROT Program

A mission-based Parametric AiRcRaft design

OpTimiza-tion code (PARROT) has been developed [7] with the goal of

fostering a more efficient and effective preliminary aircraft

design process This code optimizes the wing size for one

of two different mission categories: surveillance mission or

maximum payload Whereas in the former the goal might

be to maximize the flight range or endurance, the latter’s objective is to maximize the useful payload lifted Con-straints may include specified performance criteria, like maximum take-off distance, climb rate, bank angle, cruise velocity and the like Internally the routine comprises sev-eral disciplinary subroutines, including low fidelity mod-els for the aerodynamics (based on XFOIL), for the propul-sion (with the possibility of choosing either a combustion engine or an electrical motor) and for the stability (where the horizontal and vertical empennages are sized) While the static stability is self-satisfied by the routine, which sizes the tail for meeting the user defined static margins and tail arm, the dynamic stability data is just an output The appropriate sizing of an aircraft is essential to pro-duce a high performance design Size and mass also have

a close correlation with costs The design methodology de-veloped is based on an extensive parametric study devel-oped in-house in a spreadsheet which has been converted into a Fortran code for the sake of efficiency, easiness of handling and modularity This methodology’s primary

de-sign parameters are the wing span (b) and the wing mean chord (c) Other design parameters, which may be used in the study, are the wing airfoil cruise lift coefficient (C l), the center of gravity (CG) position, the tail arm, the lifting sur-faces’ airfoils, the motor and the propeller size, among oth-ers

This code’s users have to choose between the two dif-ferent mission categories (surveillance or maximum pay-load) and to define the mission profile and the perfor-mance requirements at each mission phase The code will then generate several different wing geometries which can

be assessed against each other using parametric plots rep-resentation Therefore, the designer (user) can make more informed decisions at the preliminary design phase, which will significantly contribute to getting closer to the opti-mum solution in a fewer number of iterations

2.2 Graphical User Interface Development

The development of PARROT’s graphical user interface (GUI) was made using the open source XFLR5 GUI, which

is programmed in C++ The main reasons for using the XFLR5 framework were:it is an open source code, it is easy

to handle and it already has expedite methods for the aero-dynamic analysis of airfoils (using XFOIL)

After downloading the XFLR5 code, the “.pro” file was opened using the QtCreator [8] Firstly, a sub-menu called Aircraft Optimization (Figure 1) that enables selecting a new optimization module was created Then, it was nec-essary to think of a way to make the data handling task as

Figure 1: XFLR5 sub-menus.

light and straightforward as possible, as it is described in

section 2.3

In order to avoid building enormous, intricate, and

be-haviorally rich graphical user interfaces it is necessary to

capture a variety of aspects [9] Given the multiple

wid-gets (labels, text edits, push buttons, toggle buttons, lists,

tables, menus) that are possible to use in the making of

the GUI, it was important to choose and combine those

that are simpler for the task in question After becoming

aware of all the widgets and functions that QtCreator has,

a first look of the windows envisaged for the program was

sketched Once the windows were correctly defined the

next phase was the programming of the GUI

2.3 GUI Interaction

In order to develop a practical GUI it is necessary to make

it easy to understand To achieve that, PARROT is divided

into two main parts, inputs and outputs In the inputs

sec-tion, the user loads the mission profile definition and all

parameters concerning propulsion, aerodynamics,

perfor-manceand design variables ranges and increments

The first step was to create a new menu in XFLR5 called

Aircraft Optimization (Figure 1) This is made to

distin-guish the mission-based aircraft design optimization

per-formed with PARROT, the aerodynamic analysis and

de-sign of airfoils performed with XFOIL and lifting surfaces,

and airplane analysis performed with XFLR5 itself After

Figure 2: PARROT settings.

clicking on this new menu, it is possible to find a new one called Analysis in which the user can choose the PARROT program In the future, another option will be provided since this research aims at creating another aircraft opti-mization code which will make use of the same GUI Once the user has made the aforementioned selection,

it is possible to have a general view of the parametric de-sign code interface (Figure 2) The first options are related with general parameters Then it is possible to load the propulsion, systems, fuselage, aerodynamics and weight data as well as the intended mission profile performance targets

With the Aerodynamics Data button, it is possible to load the airfoils’ aerodynamic coefficients, which can be generated in XFLR5 in the menu XFoil Direct Analysis be-forehand For that, it is necessary to upload the airfoils coordinate files and then perform a Batch Analysis Fi-nally, in the Aerodynamics Data window, the user needs

A computer application for parametric aircraft design | 435

Figure 3: Aerodynamics Data Analysis fields.

Figure 4: Text files with Input data.

Table 1: Air Cargo Challenge 2015 regulations summary.

The Aircraft

Aircraft Size Limited to a 2.5 m side square (Figure 5)

When disassembled it must fit inside a 1.1

× 0.5 × 0.4 m box

Propeller APC 13”x7” Sport

Motor AXI Gold 2826/10

Battery Up to 3 cells in series and the product

of maximum continuous discharge rate

times the capacity has to be at least 45 A

The Competition (Goals)

Take-off Lifting maximum payload in

60 m (Figure 6) Equation (1)

Cruise As many 100 m legs in two

min-utes (Figure 6)

to write the airfoils’ names (according to the name used in

the Batch Analysis) in the respective fields (Figure 3)

The user can also load the Aerodynamics Data by

di-rectly clicking on Load Aerodynamics Data The developed

GUI will consecutively and respectively then ask for the

in-board and outin-board wing, horizontal tail, and vertical tail

airfoils’ aerodynamics data files (Figure 4) This last option

can be used provided that the files loaded follow the

aero-dynamics standard files layout, which will be described in

a user manual which shall soon be released together with

the PARROT code interface

As the number of input parameters is relatively large,

once the user has loaded all the data the first time, it is

pos-sible to generate a “.txt” file which will store all the project input data This file can be loaded in subsequent analysis, avoiding the tiresome and repetitive task of loading all the required data each time the PARROT routine is called It can be useful to load all the input data from the “.txt” file

if the user wants to rerun a previously saved analysis or if

it is only necessary to change a few inputs Therefore, the user can also load the general input parameters by click-ing on Load Data To make this possible, every time a new analysis is made, a file named “input_parrot.txt” is gener-ated, which can then be loaded in a forthcoming program run

Finally, and after clicking on the Analysis button – which will call PARROT’s executable file – it is possible

to visualize all the relevant outputs as functions of each flight phase’s start or end and wingspan versus wing mean chord combination (Figure 5) The user can also save this output data in a “.txt” in matrix form to enable an easy gen-eration of the respective parametric graphical representa-tions To have a more global view about all the inputs and outputs, it is also possible to save all the data in a specific folder with the project name selected

3 Case Study

3.1 Mission Definition

The case study described in this section is aimed at optimizing an aircraft for the Air Cargo Challenge 2015 (ACC’15) This competition was created in 2003 by stu-dents from IST1and it is an international biannual com-petition destined to the academic community with engi-neering background

Each team has the assignment of designing, building and flying a radio-controlled aircraft which main goal is to lift the highest useful payload possible in a 60 m runway Furthermore, each group has to provide written and oral support to its decisions The final score is a weighted sum

of the design report, technical drawings, oral presentation and flight score, with bonuses and penalties also being used The ACC’15 competition design specifications [10] are summarized in Table 1

The competition regulations establish that the objec-tive function will thus be a trade-off between the payload

mass lifted (m) and the number of legs flown in 120 s (l).

The flight competition score is calculated in accordance

1 Instituto Superior Técnico, University of Lisbon

Figure 5: Outputs menu.

Figure 6: Maximum dimensions allowed for the aircraft.

with Equation (1)

⎧

⎪

⎪

⎪

⎪

⎪

⎪

a = 2 for a valid start + invalid landing

a = 3 for a valid start + valid landing

d = 1 for a valid flight without crash

d = 0 for airplane losing parts or crashes or invalid

start

Figure 7: Take-off and cruise legs [10].

In addition, since the aircraft has to fit within a 2.5 × 2.5 m square, the maximum wingspan is limited to about

(b max = 3.5 m) As for the wing mean chord, it has been

limited to (c max = 0.45 m) because the wing planform shape is not dully optimized otherwise, which would im-pact the Oswald efficiency factor, and thereafter the wing performance The Oswald factor considered is 1.0 – an op-timized planform shape and twist distribution is assumed Furthermore, this limit allows the wing panels to fit the transportation box Finally, as the largest wingspans and wing mean chords were expected to deliver the best per-formances, it has been decided that the lower boundaries

of these two variables would stand on (c min= 0.30 m) and

(b min= 3.0 m)

The wing airfoil chosen for performing this optimiza-tion was the Selig 1223, which is the most widely used

air-A computer application for parametric aircraft design | 437

Figure 8: Top view of the flight path during the 120 s with the

depic-tion of the legs trajectories.

foil in former editions of the Air Cargo Challenge, because

of its high lift capabilities at low Reynolds numbers The

ACC’15 competition has also a speed requirement which

may hinder the Selig 1223 airfoil’s fitness to the task due

to its relative high drag coefficient at moderate lift

coef-ficients Nevertheless, it will be used for the sake of this

study The selected wing airfoil lift coefficient is (C l= 0.9),

because it is the lowest lift coefficient – highest velocity

-for which the airfoil per-formance is still not significantly

affected The airfoil chosen for the horizontal and vertical

stabilizers was the NACA 0009

The cruise stage, in which the aircraft is supposed to

perform as many 100 m legs as possible in 120 s is modeled

in two parts: a leveled straight flight and a leveled turn, as

it can be perceived from Figure 8 It has been considered

that one leg is composed of a leveled straight flight for 70 m

plus a leveled turn of 180∘at a bank angle (ϕ) of 45∘ The

balance of forces in the sustained turn are shown in

Fig-ure 9, let (L) be the lift force, (W) be the weight, (F c) be the

centripetal force, (F inertial ) be the inertial force and (R) be

the turn radius

After performing the vertical and horizontal balance

of forces of the leveled turn (Figure 9), it is possible to

ob-tain Equation (2), where (V) is the vehicle’s velocity, (R) is

the turn radius and (g) is the acceleration of gravity.

2

From Equation (2), it is possible to conclude that, for

the bank angle considered, a minimum velocity of about

Figure 9: Diagram of the forces acting on the airplane during each

turn.

10.2 m/s shall be required to make sure that the minimum

leg distance is performed as required (70 + 2R > 100).

In the end of the last leg, it is not required to perform a turn However, this effect will not be neglected to account for the increased length of the leveled flight to get to the

100 m line in the first and last leveled flight stages, as can

be seen in Figure 8

3.2 Results

The most important results are summarized in the plots

of Figures 10 through 14, where the variation of the most relevant performance metrics are plotted against the most important design variables (wing mean chord and wingspan)

Figure 10 shows how the structural weight varies with the wingspan versus wing mean chord combination As expected, the structural weight increases with the wing area Figure 11 shows that the wing layout that provides

the highest design weight is (c = 0.42 m; b = 3.5 m) This

is because the wings with the same wingspan and greater wing mean chord will not be able to meet the minimum rate of climb of 0.5 m/s specified for climbing, although they could lift more payload in the available 60 m run-way The same reasoning can be drawn to justify Figure 12, which features the payload weight This plot is the one that

is more closely related with the ACC’15 objective function

It should be noted that the competition’s objective function was to lift the highest payload and perform the maximum number of legs in two minutes (120 s) If one

Figure 10: Structural weight [N] as a function of wingspan and wing

mean chord.

Figure 11: Design take-off weight [N] as a function of wingspan and

wing mean chord.

fixes the airfoil lift coefficient of the cruise stage - as it

has been done to reduce the parasite drag coefficient of

the wing without putting the wing airfoil performance at

risk – the greater the vehicle’s wing loading (W/S), the

greater will be the velocity and therefore the number of

legs performed This means that the two objectives

(pay-load weight and number of legs) are slightly contradictory

because the higher wing loadings occur for the smaller

wings and the higher payloads tend to occur for the larger

wings Nevertheless, the variation of the number of legs

possible to perform within the range of wing spans and

Figure 12: Payload weight [N] as a function of wingspan and wing

mean chord.

Figure 13: Number of legs flown as a function of wingspan and wing

mean chord.

mean chords selected is almost negligible as seen in Fig-ure 13

At this point, it is worthwhile to mention that the com-putation of the number of legs has been made as if this was a continuous variable, which is not the case since only an integer discrete number of solutions is possible for scoring purposes Therefore, it is easy to conclude that all the analyzed wingspans and wing mean chord combina-tions allow the aircraft to perform a total of 12 legs Thus,

it is clear that the payload weight will determine the best wing layout from a scoring viewpoint The best wing

lay-A computer application for parametric aircraft design | 439

Figure 14: Flight score (points) as per Equation (1) as a function of

wingspan and wing mean chord.

out (c = 0.42 m; b = 3.5 m) can also be seen in

Fig-ure 14, which shows the total flight score as a function of

the payload mass lifted and of the integer number of legs

performed (12 for all the analyzed wings)

4 Conclusions and Future Work

A computational tool for parametric aircraft design was

developed This application is divided into two parts: the

first part consisted in the development of the analysis code

PARROT; the second part was the development of the GUI

The PARROT code can thus actively contribute to a more

ef-ficient and effective preliminary design optimization of

un-manned aerial vehicles, by synthesizing the interactions

between the core aeronautical design disciplines and

feed-ing the designer with mainstream performance figures,

while its GUI widens the spectrum of possible users while

making the data handling undertaking significantly easier

and straightforward

The results of an aircraft wing layout optimization for

the Air Cargo Challenge 2015 witness the usefulness of

the computational tool developed It is shown how the

parametric plots can help the user having optimized

es-timates for the most important design variables

Addition-ally, the user can easily understand the performance

im-pact of changing one or two of these most relevant design

variables, namely the wing mean chord and wingspan

Future work shall include the development of a

sim-ilar interface, embedded in the same GUI, for the

MulTi-level design OPtimization (MTOP) code [11] which is being developed within the same research project Once these two codes (PARROT and MTOP) are working with the pre-sented GUI, two user manuals will also be released to make sure that anyone can benefit from these design optimiza-tion codes

Acknowledgement: This work has been partially funded

by the European Community’s Seventh Framework Pro-gramme (FP7) under the Grant Agreement 314139 The CHANGE project (Combined morphing assessment soft-ware using flight envelope data and mission based mor-phing prototype wing development) is a Level 1 project funded under the topic AAT.2012.1.1-2 involving 9 part-ners The project started on August 1st 2012

Nomenclature

φ bank angle

a take-off and landing validity factor

b wingspan

b min , b max lower and upper wingspan bounds

c wing mean chord

c min , c max lower and upper mean wing chord bounds

CG centre of gravity position

Cl airfoil lift coefficient

d flight validity factor

F c centripetal force

F inertia centrifugal inertia force

g acceleration of gravity

l flown legs

m payload mass

R turn radius

S wing area

V flight velocity

W aircraft weight

References

[1] Martins J.R.R.A., Lambe A.B., Multidisciplinary Design Opti-mization: A Survey of Architectures, AIAA Journal, 2013, 51, 2049–2075.

[2] Advanced Aircraft Analysis: http://www.darcorp.com/Software /AAA/, last access 13/08/2015.

[3] CEASIOM website: http://www.ceasiom.com/, last access 13/ 08/2015.

[4] XFLR5 website: http://www.xflr5.com/xflr5.html, last access 28

/07/2015.

[5] XFOIL website: http://web.mit.edu/drela/Public/web/xfoil/,

last access 30/05/2015.

[6] Standford University:

http://adl.stanford.edu/papers/suave-open-source.pdf, last access 02/08/2015.

[7] Albuquerque P.F., Gamboa P.V., Silvestre M.A., Parametric

Air-craft Design Optimisation Using Span, Mean Chord and Wing

Airfoil Lift Coeflcient as Main Design Drivers, Advanced

Materi-als Research, 2014, 1016, 365–369.

[8] Qt Creator website: http://doc.qt.io/qtcreator/, last access 15/

07/2015.

[9] Mijailović Z., Milićev D., Empirical Analysis of GUI Programming Concerns, International Journal of Human–Computer Studies,

2014, 72 (10–11), 757–771.

[10] ACC’15: http://www.acc2015.com/inhalt/regulations/ACC2015 _Regulations_V1_00.pdf, last access 08/08/2015.

[11] Albuquerque P F., Gamboa P V., Silvestre M A., Multidisci-plinary and Multilevel Design Methodology of Unmanned Aerial Vehicles Using Enhanced Collaborative Optimization, Interna-tional Journal of Mechanical, Aerospace, Industrial and Mecha-tronics Engineering, 2015, 9 (4), 470–479.