Structural Design of A Ulm Prandtlplane Wing System: Universit A Di Pisa

Universitá di Pisa
FACOLTÀ DI INGEGNERIA
Corso di Laurea Specialistica in Ingegneria Aerospaziale
Tesi di Laurea Specialistica
Structural design of a ULM

PrandtlPlane wing system
Relatori: Candidato:
Prof. Aldo Frediani Viti Alberto
Dott. Ing. Vittorio Cipolla
Anno Accademico 2010–2011

Contents
Contents I
List of Figures V
List of Tables IX
Listings X
1 Introduction 1
1.1 Best Wing System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 The PrandtlPlane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 Ultralight Aircraft . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.4 The purpose of the present thesis . . . . . . . . . . . . . . . . . . . . . 5
I ULM 6
2 The prototype 7
2.1 Regulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2 Geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.3 Wing Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.4 Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3 Flight loads 19
3.1 Data extrapolation process . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.2 Flight envelopes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
I
CONTENTS
II Design process 27
4 Structural model 28
4.1 Structural parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
4.2 Inertial data set calculation . . . . . . . . . . . . . . . . . . . . . . . . 29
4.3 Weight calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
4.4 Aerodynamic loads . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
4.5 Load distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
4.6 Stress model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
4.6.1 Sign conventions and notation . . . . . . . . . . . . . . . . . . . 35
4.6.2 Stress due to axial load . . . . . . . . . . . . . . . . . . . . . . . 37
4.6.3 Bending stresses . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.6.4 Stress due to torque . . . . . . . . . . . . . . . . . . . . . . . . 37
4.6.5 Stress due to shear forces . . . . . . . . . . . . . . . . . . . . . . 38
4.6.6 Equivalent stress . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.7 Sensibility the discrete division . . . . . . . . . . . . . . . . . . . . . . 40
5 Design procedure 46
5.1 Design process work-flow . . . . . . . . . . . . . . . . . . . . . . . . . . 46
5.2 Defining initial parameters . . . . . . . . . . . . . . . . . . . . . . . . . 48
5.2.1 Geometric parameters . . . . . . . . . . . . . . . . . . . . . . . 48
5.2.2 General parameters . . . . . . . . . . . . . . . . . . . . . . . . . 49
5.3 Defining initial distribution thickness . . . . . . . . . . . . . . . . . . . 50
5.3.1 Ribs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
5.3.2 Spars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
5.4 Calculating inertial properties . . . . . . . . . . . . . . . . . . . . . . . 51
5.5 Initialize loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
5.6 Loads calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
5.7 Load distribution and stress calculation . . . . . . . . . . . . . . . . . . 53
5.8 Stress constraint check and thickness definition . . . . . . . . . . . . . . 54
5.9 Global weight definition . . . . . . . . . . . . . . . . . . . . . . . . . . 54
6 Results 56
6.1 Design results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
6.2 Converting results into design . . . . . . . . . . . . . . . . . . . . . . . 60
II
CONTENTS
III F.E.M. Analysis 66
7 F.E.M. Modeling 67
7.1 Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
7.1.1 Main spars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
7.1.2 Auxiliary spar . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
7.1.3 Ribs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
7.1.4 Skin panels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
7.1.5 Winglet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
7.1.6 Laminate modeling . . . . . . . . . . . . . . . . . . . . . . . . . 73
7.1.7 Constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
7.1.8 Load introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 75
7.2 Definition of the flange thickness: 1st step . . . . . . . . . . . . . . . . 77
7.2.1 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
7.3 Definition of thickness: 2nd step . . . . . . . . . . . . . . . . . . . . . . 80
8 F.E.M. Results 82
8.1 Hoffman’s theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
8.2 Comparison between Prototype - New structure . . . . . . . . . . . . . 84
8.2.1 Spars model : nz = 4 . . . . . . . . . . . . . . . . . . . . . . . . 84
8.2.2 Full model : nz = 4 . . . . . . . . . . . . . . . . . . . . . . . . . 86
8.3 New structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
8.3.1 Spars model : nz = 4 . . . . . . . . . . . . . . . . . . . . . . . . 88
8.3.2 Full model : nz = 4 . . . . . . . . . . . . . . . . . . . . . . . . . 93
8.3.3 Full model : nz = −2 . . . . . . . . . . . . . . . . . . . . . . . . 98
8.4 Results of process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
8.5 Final layout of spars . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
9 Conclusions 108
9.1 Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
9.2 Later developments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
Appendix 112
A Material Properties 112
B Alternative optimization process 114
III
CONTENTS

R
C The Matlab code 116
C.1 Run the optimization tool . . . . . . . . . . . . . . . . . . . . . . . . . 116
C.2 The objective function . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
C.3 Structural weight . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
C.4 Results layout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
C.4.1 Substructure cds . . . . . . . . . . . . . . . . . . . . . . . . . . 133
C.4.2 Substructure geom . . . . . . . . . . . . . . . . . . . . . . . . . 135
C.4.3 Substructure iner . . . . . . . . . . . . . . . . . . . . . . . . . . 135
C.4.4 Substructure peso . . . . . . . . . . . . . . . . . . . . . . . . . . 136
C.4.5 Substructure discr . . . . . . . . . . . . . . . . . . . . . . . . . 137
C.4.6 Substructure carichi . . . . . . . . . . . . . . . . . . . . . . . . 137
C.4.7 Substructure distr_tens . . . . . . . . . . . . . . . . . . . . . . 138
C.4.8 Substructure distr_spess . . . . . . . . . . . . . . . . . . . . . . 138
C.5 Default values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
D Sensitivity Analysis 141

D.1 Analysis process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
D.2 Command window . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
Bibliography 147
IV
List of Figures
1.1 Best Wing System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.2 BWS lift distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Effect of h/b ratio on induced drag . . . . . . . . . . . . . . . . . . . . 3
1.4 Prandtl Plane Configuration . . . . . . . . . . . . . . . . . . . . . . . . 3
2.1 Final model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

2.2 Fuselage external dimensions . . . . . . . . . . . . . . . . . . . . . . . . 9
2.3 Wing system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.4 Wing-karman connection . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.5 Ribs layout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.6 Spar layout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.7 Front wing structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.8 Rear wing structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.1 wing split layout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

3.2 Front wing lift distribution . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.3 Rear wing lift distribution . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.4 Front wing drag distribution . . . . . . . . . . . . . . . . . . . . . . . . 21
3.5 Rear wing drag distribution . . . . . . . . . . . . . . . . . . . . . . . . 22
3.6 Front wing inertia distribution . . . . . . . . . . . . . . . . . . . . . . . 22
3.7 Rear wing inertia distribution . . . . . . . . . . . . . . . . . . . . . . . 23
3.8 Take off flight envelope . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.9 cruise flight envelope . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.10 Landing flight envelope . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
4.1 Ribs model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

4.2 Sections available . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
4.3 Cross section selection interface . . . . . . . . . . . . . . . . . . . . . . 30
V
LIST OF FIGURES
4.4 Section parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

4.5 Lift distributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
4.6 ETA reference system and conventions . . . . . . . . . . . . . . . . . . 33
4.7 Load distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
4.8 Convention about section element name . . . . . . . . . . . . . . . . . 36
4.9 Convention about loads . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
4.10 Equilibrium on a shell sector . . . . . . . . . . . . . . . . . . . . . . . . 38
4.11 Compared analysis:cds front wing . . . . . . . . . . . . . . . . . . . . . 41
4.12 Compared analysis:cds rear wing . . . . . . . . . . . . . . . . . . . . . 42
4.13 Compared analysis:cds winglet . . . . . . . . . . . . . . . . . . . . . . . 43
4.14 Time calculation against mesh refinement . . . . . . . . . . . . . . . . . 45
5.1 Design work-flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

5.2 Chordwise spar position . . . . . . . . . . . . . . . . . . . . . . . . . . 49
5.3 Spar positioning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
5.4 Geometry for initialize the optimization process . . . . . . . . . . . . . 52
5.5 Spar splitting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
5.6 Work-flow to define thickness . . . . . . . . . . . . . . . . . . . . . . . 55
6.1 Results:distribution of load characteristics . . . . . . . . . . . . . . . . 58

6.2 Results:distribution of thickness and related equivalent stresses . . . . . 59
6.3 Step width for upper and lower flanges . . . . . . . . . . . . . . . . . . 60
6.4 Thickness step functions . . . . . . . . . . . . . . . . . . . . . . . . . . 62
6.5 Comparison between prototype and optimized layout . . . . . . . . . . 65
7.1 Starting model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

7.2 Base model without skin and auxiliary spar . . . . . . . . . . . . . . . 68
7.3 Spar model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
7.4 Spar model (internal structures) . . . . . . . . . . . . . . . . . . . . . . 69
7.5 Element coordinates system . . . . . . . . . . . . . . . . . . . . . . . . 69
7.6 Nodes besides the bulkhead . . . . . . . . . . . . . . . . . . . . . . . . 70
7.7 Ribs model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
7.8 Skin model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
7.9 Winglet model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
7.10 Laminate builder tool . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
7.11 Material Coordinate system . . . . . . . . . . . . . . . . . . . . . . . . 73
7.12 stacking layout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
7.13 Spar-Karman constraints . . . . . . . . . . . . . . . . . . . . . . . . . . 74
7.14 Winglet spar connection . . . . . . . . . . . . . . . . . . . . . . . . . . 75
VI
LIST OF FIGURES
7.15 Winglet full connection . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

7.16 Load introduction through MPC . . . . . . . . . . . . . . . . . . . . . . 76
7.17 Entire load system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
7.18 Definition of wing trunks . . . . . . . . . . . . . . . . . . . . . . . . . . 77
7.19 Example of staking layout (front spar) . . . . . . . . . . . . . . . . . . 78
7.20 Example of staking layout (rear spar) . . . . . . . . . . . . . . . . . . . 78
7.21 Failure on webs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
7.22 Failure index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
7.23 Location of carbon pads on web . . . . . . . . . . . . . . . . . . . . . . 80
7.24 Layout of reinforced webs . . . . . . . . . . . . . . . . . . . . . . . . . 80
7.25 New staking layout (front spar) . . . . . . . . . . . . . . . . . . . . . . 81
7.26 New staking layout (rear spar) . . . . . . . . . . . . . . . . . . . . . . . 81
7.27 Layout of winglet webs and flanges . . . . . . . . . . . . . . . . . . . . 81
8.1 Failure index: legend . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82

8.2 Failure (front view) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
8.3 Failure (rear view) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
8.4 Failure (front view) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
8.5 Failure (rear view) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
8.6 Failure (Karman section) . . . . . . . . . . . . . . . . . . . . . . . . . . 88
8.7 Failure index (Webs) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
8.8 Failure index (Flanges) . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
8.9 Failure index (Bulk) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
8.10 Von Mises stress [MPa] . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
8.11 Failure (Karman section) . . . . . . . . . . . . . . . . . . . . . . . . . . 93
8.12 Failure index (Webs) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
8.13 Failure index (Flanges) . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
8.14 Failure index (Bulk) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
8.16 Failure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
8.17 Failure index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
8.19 Spars final layout: notations . . . . . . . . . . . . . . . . . . . . . . . 103
8.20 Spars final layout: upper flange . . . . . . . . . . . . . . . . . . . . . . 104
8.21 Spars final layout: upper flange . . . . . . . . . . . . . . . . . . . . . . 105
8.22 Spars final layout: lower flange . . . . . . . . . . . . . . . . . . . . . . . 106
8.23 Spars final layout: lower flange . . . . . . . . . . . . . . . . . . . . . . . 107
VII
LIST OF FIGURES

R
9.1 Example of wok-flow elaborated in Mode Frontier . . . . . . . . 110
A.1 Best Wing System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
B.1 Alternative optimization process . . . . . . . . . . . . . . . . . . . . . . 114
C.1 Start the tool . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117

C.2 Dialog command window used to enter the geometric parameters . . . . 118
C.3 Dialog command window used to enter the general parameters . . . . . 118
C.4 Legend . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
C.5 Structure of dati_out . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
C.6 Substructure cds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
C.7 Substructure geom . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
C.8 Substructure iner . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
C.9 Substructure peso . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
C.10 Substructure discr . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
C.11 Substructure carichi . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
C.12 Substructure distr_tens . . . . . . . . . . . . . . . . . . . . . . . . . . 138
C.13 Substructure distr_spess . . . . . . . . . . . . . . . . . . . . . . . . . . 138
D.1 Structural weight variation against parameters variation . . . . . . . . 143

D.2 Structural weight relative variation against parameters relative variation 144
VIII
List of Tables
2.1 RAI-V.EL: Load factors . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2.2 Regulations used in Europe . . . . . . . . . . . . . . . . . . . . . . . . 8
2.3 Wings geometric parameters . . . . . . . . . . . . . . . . . . . . . . . . 11
2.4 Main spars dimension set . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.5 Flange thickness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
4.1 Calculation time results . . . . . . . . . . . . . . . . . . . . . . . . . . 44

4.2 Calculation time results (percent) . . . . . . . . . . . . . . . . . . . . . 44
6.1 Flange spar thickness (front wing) . . . . . . . . . . . . . . . . . . . . . 63

6.2 Flange spar thickness (rear wing) . . . . . . . . . . . . . . . . . . . . . 64
8.1 Weights . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101

8.2 Displacements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
8.3 Thickness distribution (mm) . . . . . . . . . . . . . . . . . . . . . . . . 102
A.1 Elastic modules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113

A.2 Mechanical properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
A.3 Poisson ratios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
C.1 Default parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
D.1 Parameters used in sensitive analysis . . . . . . . . . . . . . . . . . . . 142
IX
Listings
C.1 Launch calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119

C.2 Objective function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
C.3 gen_conf routine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
C.4 Launch_calcolo_pesi 1st step . . . . . . . . . . . . . . . . . . . . . . . 123
C.5 Launch_calcolo_pesi 2nd step . . . . . . . . . . . . . . . . . . . . . . 124
C.6 Launch_calcolo_pesi 3rd step . . . . . . . . . . . . . . . . . . . . . . 124
C.7 A part of function distrib_portanza_generica . . . . . . . . . . . . . . 125
C.8 A part of function tens_ta . . . . . . . . . . . . . . . . . . . . . . . . . 125
C.9 A part of function distrib_resistenza_generica . . . . . . . . . . . . . 125
C.10 Launch_calcolo_pesi 4th step . . . . . . . . . . . . . . . . . . . . . . 126
C.11 The functionpar_conv . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
C.12 A part of function def sp . . . . . . . . . . . . . . . . . . . . . . . . . . 128
C.13 The function weight1long . . . . . . . . . . . . . . . . . . . . . . . . . 129
C.14 Launch_calcolo_pesi 5th step . . . . . . . . . . . . . . . . . . . . . . 130
X
Abstract
The task of the thesis is to optimize the wing structure of a non conventional very
light airplane starting from a PrandtlPlane prototype currently in phase of development.
This treatment is divided in several parts. In the first part an introduction about the
main features of a PrandtlPlane aircraft is given.
In particular, the first chapter explains what was the idea which led to this new
solution, its main advantages and the reasons which led to employ this configuration
in the branch of light aviation.
The second part depicts all the features of the prototype, in terms of regulation,
structures, materials used to build it, and loads that act on the structure. The third
part aims to the describe the design process and how it is implemented, starting from
a previous tool and showing how this process can be used in order to further develop
this configuration.
Finally, in the last part of this job the results of the design process will be verified
through a finite element analysis, in order to perform a comparison between the
prototype and the optimized structure.
XI
Chapter 1
Introduction
In the last years,the increasing cost of fuel and the need for a reduction in air traffic
pollution has led civil aviation to move its attention toward the construction of more
efficient aircraft. In order to achieve this goal, the aeronautic industries have usually
tried to improve aerodynamics and to build lighter structures but, recently several
solutions have been proposed in order to lead a radical change in the standard aircraft
configuration. Among them, the attention of the aeronautic community has been
focused on the Blended Wing Body and the multiplane configurations and currently,
NASA and Boeing are developing the first one.
This thesis aims to examine one of the multiplane configuration, which is called
PrandtlPlane . In particular the present work faces the problem of finding the minimum
weight solution for the main wing structures of a n ultralight PrandtlPlane . In the
last years, Pisa University Aerospace Engineering department has started this project
following the idea of Best wing System, an alternative aircraft configuration elaborated
by L.Prandtl in 1924.
1.1 Best Wing System

The Best Wing System idea was presented by L.Prandtl in the TN 182 NACA
report, in which Prandtl showed that once wingspan and total lift are given, as the
number of wings used increases, the induced drag decreases. Then, for instance a
biplane is more efficient than a monoplane, triplane is more efficient than biplane and so
on. In this respect the best result available is a box-shaped wing, shown in Figure 1.1,
then called Best Wing System. The main characteristics of this configuration, as shown
in [1], are an equal lift on upper and lower wing and a butterfly shaped (lateral) forces
with nil resultant on the vertical winglets (Figure 1.2). Even though such conditions
are necessary to minimize the induced drag, they aren’t suitable for practical use, since
1
1. Introduction 1.1. Best Wing System
Figure 1.1: Best Wing System
they could not verify the stability requirements related to the aircraft aeromechanics.
Moreover, the induced drag reduction, depends on the lift acting on the two wings
Figure 1.2: BWS lift distribution
and on the mutual induction between them, becoming more important as the distance
between the upper and lower wing decreases. The equation below represent the global
drag which affects a biplane wing system
Di = D11 + D22 + D12 + D21 (1.1.1)
where the first two terms represent the self induction, while the third and fourth terms
h
represent the mutual induction. In particular the last terms strongly depends on the b
ratio1 , in fact the drag decreases asymptotically when it increases. Thus, when there is
a sufficient gap between upper and lower wings, the lift is not affected by the induction
of each other, as shown Figure 1.3. These topics are fully illustrated and demonstrated
in [1]
1
Where h is the gap between upper and lower wing an b is the wing span
2
1. Introduction 1.2. The PrandtlPlane
Figure 1.3: Effect of h/b ratio on induced drag
1.2 The PrandtlPlane

The airplanes builded in the first years of aviation were biplanes since the aircraft
were made using wood and thus a monoplane solution didn’t allow to achieve an
adequate stiffness. Later on, tanks to the rising of light materials (like aluminium alloy)
the development of aircraft has been moved toward single wing solutions, that actually
are widely diffused in all the aviation branches. The Best Wing System solution, can
be transformed by moving the two wings in longitudinal way.This configuration, shown
in Figure 1.4, is called PrandtlPlane . In this case the wing and the horizontal tail
are integrated in a wing system able to produce the same total lift of a traditional
wing-tail solution and to satisfy the stability and controllability requirements as well.
Figure 1.4: Prandtl Plane Configuration
This kind of configuration is mainly suitable for commercial aircraft, in which
3
1. Introduction 1.3. Ultralight Aircraft
fuel consumption has a great influence on operating costs; for them, an induced drag
reduction leads to a higher probability of reduction of operating costs in terms of fuel
consumption. Moreover, this solution presents several advantages also in terms of
controllability. Indeed, the control surfaces are arranged on both wings (front and
rear), so that is possible to move them with opposite rotation allowing the pilot to
generate a longitudinal pure moment without affecting the total lift and, therefore,
increasing the accuracy of the command during the maneuvers.
Concerning structures, the over-constrained nature of the wing-box leads to a
greater stiffness of the overall system with advantages from the static and aeroelastic
standpoints(see [1]).
1.3 Ultralight Aircraft

The induced drag reduction is not one of the main goals of the "light aviation",
since fuel consumption has small influence on operating costs, if compared to a long
range aircraft. In fact,for this kind of airplane the parameters that affect the operating
costs are especially the landing fees and maintenance. However, there are other aspects,
especially related to safety, for which the use of PrandtlPlane in this category should
be considered.
Concerning the stall phenomenon, this kind of planes presents some advantages; in
fact, the downwash caused by front wing leads it to stall as first, so that he resulting
nose down moment keeps the aircraft out from such a situation.
Finally,unlike the typical ultralight airplane, the shape of this kind of aircraft
allows the designer to place the engine behind the cabin rather than inside aircraft
nose, allowing greater visibility for pilots and providing more distance between cabin
and engine compartment. Furthermore the rear engine solution provides an easier
installation of any cabin safety devices to protect the cabin against frontal impact.
To summarize, the characteristics that lead to prefer a PrandtlPlane configuration
rather than a traditional one can be summarized as follows:
• reduction of induced drag with a consequent fuel consumption;
• accuracy of longitudinal control system;
• smooth stall behavior;
• distance between pilots and dangerous components (fuel tank and engine)
4
1. Introduction 1.4. The purpose of the present thesis
1.4 The purpose of the present thesis

The purpose of this job is to design the wing structure of a very light PrandtlPlane
aircraft. This thesis starts from several previous works, in particular these are shown
in [2] and [3]. The first one defines the prototype, which is the object of this study and
shows the results of a finite element analysis performed on the initial configuration.
These results will be used to compare the behavior of the prototype with the one of
the new structure designed. The second, instead, provides a tool useful to build the
design code.
The prototype used as started model is ULM PrandtlPlane builded using as carbon
and several kind of wood. Mixed material instead has been used mainly because the
full carbon structure is not suitable for this purpose, concerning the requirements. In
fact the most part of very light aeroplanes are entirely built using wood, so that in the
years this solution has been classified as the most secure and reliable, in this concern a
full carbon structure could not found the agreement of authorities, because of the poor
backgrounds available.
The design is done through an analytical process, which aims to define the thickness
of the flanges in each point of the structures, and to verify wether the webs are suitable

R
to bear the loads. The first task is absolved by a code developed in Matlab language
while the verification is done using the F.E.M. model exposed in [2]. This model is

R
R
built using Patran as preprocessor and postprocessor and Nastran as solver.
The thickness of flanges is defined in order to verify the stress constraint and, even
though this constraint takes only the strength of material into account(no buckling
or other instability effect are considered in this paper), the ultimate stress imposed
accounts itself the compressive strength of material. Moreover, the main structure of
this kind of aircraft is composed of beams with box shaped cross-section, hence the
ratio between the width of the plate that compose the spars and them thickness is
quite low. In this respect, especially concerning flanges, the buckling effect should be
negligible. Nevertheless in order to verify this extended buckling analysis would be
requested. Furthermore the only condition of limit load factor will be investigated in
order to avoid non linear effect, which could lead to a not consistent results.
Finally, a brief exposition of a multiobjective optimization process will be given.
This extended optimization could be implemented in several optimization tools, and
can involved several sets of geometrical and technological parameters in order to find
the configuration which minimize both the induced and parasite drags and thus fuel
consumption.
5
Part I
ULM
6
Chapter 2
The prototype
This chapter aims at depicting the prototype of a PrandtlPlane ultralight aircraft

(ULM), currently under construction at SG Flyevolution in Udine. This work aims to
optimize the wing structures, with particular attention on spars. The features of such
prototype will be used as starting point to develop a numeric algorithm that aims to
define the minimum weight solution. In the next sections the main properties of the
airplane, regarding geometry, materials, structures and regulations, will be illustrated.
These topics are fully illustrated in [2].
2.1 Regulations
The present thesis is based on the requirements given by the Italian RAI-V.EL
regulation, which provides a conservative estimation about load factors compared to
European regulation CS-VLA, as sown in [2]. Unlike the "certified airplanes", the flight
envelope of very light aircraft is not obtained by superposition of gust and maneuver
diagrams, but only on maneuver and only at the zero altitude. Table 2.1 gives the
main specifications of the regulation used. The European regulation CS-VLA,requires
expensive certification methods, in terms of both times and certification costs. For
this reason, every country has developed his own regulation; in particular, the Italian
aviation, refers to RAI V.E.L, the United Kingdom refers to BCAR S and, finally,
Germany refers to BFU. As shown in Table 2.2, all these regulations are similar each
other, hence an aircraft builded in Italy has a good chances to fly also in the rest of
the Europe. This thesis makes reference to the Italian regulation.
7
2. The prototype 2.2. Geometry
RAI-V.EL
W to[Kg] 600
nzmax 4
nzmin −2
W
nzgust 4 for S
< 350 mN2
W
nzgust through graphical interpolation for S
> 350 mN2
Table 2.1: RAI-V.EL: Load factors
CS-VLA BCAR S RAI-V.EL BFU

(Europe) (United Kingdom) (Italy) (Germany)
MTOW[Kg] 750 450 600 450

Vmax land [Kts] 45 351 40.52 35.1
Climb speed[m/s] >2 >2 1.5 1
Landing speed[m/s] 1.3Vst 1.3Vs0 1.3Vst −−
Table 2.2: Regulations used in Europe
2.2 Geometry
The ULM external shape, is the result of a previous thesis ([4]),in which the
aerodynamic configuration was optimized. The resulting configuration is shown in
Figure 2.1. Figure 2.2 gives the main parameters of the fuselage, which is made of a
carbon fiber shell and is equipped with two appendices, called Karman, that are used
to connect the wings. Figure 2.3 gives the shape of wing system.
Table 2.3 gives the parameters used to define the aircraft
3
Corresponding to karman ribs
4
Corresponding to karman ribs
8
Figure 2.1: Final model
Figure 2.2: Fuselage external dimensions
9
(a) Top view
(b) Front view
Figure 2.3: Wing system
10
Parameter Size
General
Weng [HP ] 100

Vcruise [m/s] 69
Vland [m/s] 19
Cl cruise 0.13
Clm ax 1.35
W ing span [m] 8
2
Sref [m ] 14.37
M T OW [Kg] 500
hcr [m] 1000
Vcr [m/s] 69
Front wing
dihedral[deg] 7
Λ25 [deg] 24
Cr[m]3 0.98
Ct[m] 0.612
Airfoil GOE − 398
Rear wing
dihedral[deg] 0
Λ25 [deg] −11
4
Cr[m] 1.16
Ct[m] 0.57
Airfoil GOE − 398
Table 2.3: Wings geometric parameters
11
2. The prototype 2.3. Wing Structures
2.3 Wing Structures

The primary structure of each wing is a single spar, with a box shaped cross section.
Several materials are used, the spar flanges of both front and rear wings, are made of a
laminate composed by several layers of spruce and carbon fibers; the spar of winglet
is made of using a carbon fiber reinforced plastic laminate in order to simplify the
manufacturing process. A structure called karman,which is a part of the fuselage, joins
each wing to the fuselage. The spar is connected to the karman through a pair of
transversal pin, as shown in Figure 2.4. Moreover, in order to sustain the hinges of
control surfaces, there is an auxiliary floating spar on both wings. These spars are
not linked to the fuselage and thus they don’t bear loads. Furthermore unlike the
main spars the auxiliary ones have a "C-shaped" cross section. Table 2.4 gives the
(a) Spar connection
(b) Karman connection
Figure 2.4: Wing-karman connection
12
2. The prototype 2.3. Wing Structures
characteristics of the main spars at wing root and wing tip; this set of parameters will
be used, to initialize the optimization process.
Wing spar Height[m] Width[m]
Front @ root 0.1217 0.09

Front @ tip 0.0602 0.097
Rear @ root 0.1217 0.09
Rear @ tip 0.0602 0.097
Table 2.4: Main spars dimension set
Wing ribs are bonded on the spars and the volume of winglets is filled with a
polystyrene foam. Ribs provide the airfoil shape and transfer loads from skin to spars
while Bulkheads are stiffer than ribs.
There are 9 ribs on each wing; the bulkhead are positioned close to the karman, and
at wing tip to connect the winglet. Both of them are manufactured stacking twelve
plies of birch, each one 1mm thick. Both ribs and bulkheads are shown in Figure 2.5.
Moreover, in order to reinforce the zone of load transfer from ribs to spar, several pads
Figure 2.5: Ribs layout
composed by a laminate of birch are bonded inside spar, next to ribs location.
Finally, skin panels, made by stacking several plies of okumè, are connected to ribs.
13
2. The prototype 2.4. Materials
2.4 Materials
Several kind of material are used to build the prototype; in particular, carbon
laminates and wood. Carbon laminates are used to build the entire fuselage shell,
some layers of the upper flanges of both spars and the entire spar of winglet. Wood
is used in ribs, webs of main spars, skin and also in a great number of layer in the
flanges. In particular there are three types of wood: European Spruce, birch and okumè.
European Spruce is used to laminate the flanges of the main spars, birch is used in
ribs, reinforcing pads and also is the webs of the spars. Finally, okumè is used to build
the skin. An overview of material properties is available in appendix A.
In order to maximize strength and stiffness of the flanges against the bending
moment, the fibers are directed parallel to the beam axis. The spars of front an rear
wings are identical. Table 2.5 gives the longeron layout, where W means Wood, C
means Carbon and the number before them indicates the quantity of layer staked. The
carbon layers have a thickness of 1.2 mm and are arranged on the flange outer surface,
while the thickness of the spruce layers is 8 mm for each ply. Concerning the thickness
Section location[m] Upper flange[mm] Lower flange[mm]
0 ⇒ 0.45 26.4(3L, 2C) 24(3L)

0.45 ⇒ 0.95 34.4(4L, 2C) 24(3L)
0.95 ⇒ 1.2 26.4(3L, 2C) 24(3L)
1.2 ⇒ 1.6 26.4(3L, 2C) 16(2L)
1.6 ⇒ 2.32 18.4(2L, 2C) 16(2L)
2.32 ⇒ 3.72 9.2(1L, 1C) 8(1L)
Table 2.5: Flange thickness
distribution, as shown in the previous tables, it is assumed the minimum value at tip
and the maximum one at root because the regulation requires that the static test of
the structure is more confident without the winglet and, thus, both front and rear
wings are tested as cantilever beams and designed accordingly. Figure 2.6 gives the
spar layout
Finally, concerning the material of laminates, both carbon and wood was used for
the following reasons: first, a full carbon structure could be lighter than a mixed one,
due to the higher strength characteristic but when the thickness has to be large due to
buckling, a mixed solution using carbon and spruce could be most efficient, so even
though both strength and stiffness are greater than the wood ones, the weight resultant
could not be definitely lower. Finally, the main reason because the full carbon structure
14
is not optimum is concerned to the regulations. In fact most very light aeroplanes
are made entirely using wood, so that this solution has been classified as the most
safe and reliable, the strength of a fully carbon structure is assumed to depend on the
manufacturing process and large safety factors are considered.
Figures 2.7 and 2.8 gives the structures of entire front and rear wings.
15
Figure 2.6: Spar layout
16
Figure 2.7: Front wing structure
17
Figure 2.8: Rear wing structure
18
Chapter 3
Flight loads
In this chapter, the process used in [2] to define the flight loads, which are taken
into account for structural design, is summarized. Other loads, such as ground loads,
have been neglected.
3.1 Data extrapolation process

Aerodynamic loads are taken from CFD (Fluent) and VLM (AVL) results, carried
out in a previous research. The process consists in to splitting both wings in a number
of trunks, as shown in Figure 3.1, and calculating the product c · Cl in each trunk.
Then the lift force acting on each one is given by
li = ∆Xi · ci · Cli (3.1.1)
where ∆Xi is the width , ci is the mean chord and Cli is the lift coefficient of the
trunks. The sum of the lift forces acting on each strip equals the global lift on the wing.
The same process has been used to evaluate the drag distribution. Concerning vertical
winglet, although the global lift should be nil,(as said in section1.1), in this concern
this is not exactly verified, because this should implies, that the lift distribution on
upper and lower be the same.
Finally, inertial loads include the mass of the whole wing structures. In the next
section the distributions of lift, drag and inertial loads are shown under the condition
nz = 1, further conditions can be obtained scaling this load distributions with the
actual load factor raised from flight envelopes(see section ??). In particular Figures
3.2 and 3.3 highlight that the value of lift at the wing tip is different from zero, unlike
in the conventional wings. This effect is caused by the presence of the winglets which
provide a butterfly shaped lift distribution. Concerning drag(Figures 3.4 and 3.5), it is
19
3. Flight loads 3.1. Data extrapolation process
4
3
2
1
0
0
−1
−2
−3
−4
−5
−6
−7
Figure 3.1: wing split layout
correlated with lift as expected. Finally, a linear variation of inertial load is shown in
Figures 3.6 and 3.7.
150
100
[N]
50
0
0 0.5 1 1.5 2 2.5 3 3.5 4
Wing span [m]
Figure 3.2: Front wing lift distribution
20
Figure 3.3: Rear wing lift distribution
Figure 3.4: Front wing drag distribution
21
Figure 3.5: Rear wing drag distribution
Figure 3.6: Front wing inertia distribution
22
Figure 3.7: Rear wing inertia distribution
23
3. Flight loads 3.2. Flight envelopes
3.2 Flight envelopes

The flight envelopes has been defined in [2] starting from the following data
• Reference wing area= 14.37 [m2 ]
• mac= 1 [m]
• Wing span= 8 [m]
• MTOW= 500 [Kg]
• hcruise = 1000 [m]
• CLcruise = 0.13
• CD0 = 0.0213
• K = 0.0528
• Vcruise = 65 [m/s]
• Vland = 19 [m/s]
• CLmax = 1.35
• CLmaxHL = 1.5
• Engine Power= 74.6 [kW ] (100 HP )
• Take off distance= 90 [m]
The flight envelopes shown in Figures 3.8, 3.9 and 3.10 are created following the
requirements of the European regulation CS-VLA. In all cases, landing, cruise and take
off, the limit load factors are
• Maximum load factor: nz=3.8
• Minimum load factor: nz=-1.5
these load factor are then modified when the Italian regulation RAI v.e.l. is introduced,
in particular they become
• nzmax = 4
• nzmin = −2
which are the values used in this this work.
24
Altitude 0 m
Weight= 500 kg
VC = 62 m/ s
z
n
VD = 81.25 m/ s
Load Factor
VA = 39.61 m/ s
VH = 30.86 m/ s
VB = 28.77 m/ s
VF = 34.80 m/ s
VS1 = 20.32 m/ s
VS0 = 19.33 m/ s
Equivalent Air Speed [m/s]
Figure 3.8: Take off flight envelope
4.5
3.5
2.5
Altitude1000 m
2
Weight= 500 kg
z
n
1.5
VC = 65 m/ s
Load Factor
1
VD = 81.25 m/ s
0.5
VH VC VA = 39.61 m/ s
0
VS1 VB VA VD VH = 30.87 m/ s
−0.5
VB = 29.43 m/ s
−1
VS1 = 20.32 m/ s
−1.5
−2
−2.5
0 10 20 30 40 50 60 70 80 90 100
Figure 3.9: cruise flight envelope
25
Quot a 0 m
Weight= 500 kg
VC = 62 m/ s
nz
VD = 81.25 m/ s
Load Factor
VA = 39.61 m/ s
VH = 30.87 m/ s
VB = 28.77 m/ s
VF = 34.70 m/ s
VS1 = 20.32 m/ s
VS0 = 19.27 m/ s
Figure 3.10: Landing flight envelope
26
Part II
Design process
27
Chapter 4
Structural model
The design of the PrandtlPlane wing structures begins from a previous research
activity, in which a numerical code to dimension box shaped wing structures has been de-

R
veloped [3]. This code, called MDSA-2010 has been generated in Matlab language;
and requires as input the wing system geometry, the kind of structures(single spar,
double spar or wing box) and its main parameters (thickness,etc), and the material
properties (Elastic and poisson modules). The MDSA-2010 tool is able to calculate
the inertial properties of the wing system and, then, to solve the equilibrium equations
using the force method(M üller − Breslau equations) in order to define the loads
distribution on the wings. At the beginning, this code was developed to analyze just
the Prandtl Plane configurations but, with later updates, the function to compute also
a traditional wing-tail solution has been implemented.
4.1 Structural parameters

In addition to configuration data, another set of data is requested. Those parameters
are necessary to completely define the shape of spars and ribs and their inertial
properties. Finally, the rib model has been implemented as a frame (shown in Figure
4.1 and according to Figure 2.5); the additional parameters requested are (Figure 4.1)1 :
rib thickness at wing root, rib thickness at wing tip, frame thickness at wing root,
frame thickness at wing tip, density, elastic modulus, poisson ratio.
Concerning spars, the shape of the cross section must be chosen; in particular, the
code supports the three types shown in Figure 4.2, to be introduced through the user
interface shown in Figure 4.3,where:
Rettangolare piena button indicates a rectangular cross section,

1
this data set is requested for both front and rear wing
28
4. Structural model 4.2. Inertial data set calculation
Figure 4.1: Ribs model
Figure 4.2: Sections available
Rettangolare cava button indicates a box shaped cross section,
Doppia T button indicates a double T-shaped cross section,
The parameters required to define a box shaped cross section for a single spar
solution2 are shown in Figure 4.4 and must be defined both at wing root and wing
tip; moreover, density, elastic and poisson modules must be defined exposed in the
following list in according to Figure ??
This data set, as said above, is related to a single spar solution with a box shaped
cross section. In general, the structural data sets are different as the structural solutions
change; for a double spar configuration two data set are requested, one for each spar;
for a T-shaped cross section the thickness for both left and right webs is the same and
represents an equivalent thickness on left and right sides of the vertical symmetry axis.
Furthermore, even though the reference aircraft contains several kinds of composite
laminates, this code do not allows the laminate implementation so far, hence, the
material is considered as homogeneous isotropic.
4.2 Inertial data set calculation

The inertial data set is calculated as a vector where the elements represent the
cross section inertial properties, which are required to solve the equilibrium problem
through the M üller − Breslau’s equations. Since in the present paper,a box-shaped
cross section is considered, the equations below show the expressions of the inertial
2
the set is required for each wing
29
4. Structural model 4.3. Weight calculation
Figure 4.3: Cross section selection interface
properties.

A = 2 tl + ta h (4.2.1)
!
lt3 lth2
Ix = 2 + (4.2.2)
12 4
!
lt3 hta l2
Iy = 2 + (4.2.3)
12 4
Iz = Ix + Iy (4.2.4)
4.3 Weight calculation

The weight calculation is performed by considering the density of material and the
structural parameters listed in section 4.1. An example, where front wing, winglet and
rear wing are listed horizontally, from top to bottom, and in the columns from left to
right spars and ribs is shown are listed. The part of command window shown below
represent the weight matrix.
PESI: =
longheroni centine TOTALE
ala_ant 5.00904 1.00534 6.01438
paratia 1.79750 0.58215 2.37965
ala_post 5.33225 6.77128 12.10353
4.4 Aerodynamic loads

By default, the aerodynamic loads are calculated from the lift per unit length at root
and tip of both front and rear wing, correlated by an elliptic shaped lift distribution.
This method allowed to implement both traditional and PrandtlPlane configuration. In
fact, lift at tip can be set to zero in the first case and different from zero in the second
30
4. Structural model 4.5. Load distribution
Flange thickness
Upper flange
web thickness
Section height
Left Right
web web
Lower flange
Section width
Figure 4.4: Section parameters
one; moreover, starting from lift at the tips a linear butterfly shaped lift distribution is
generated on the vertical winglet. Figure 4.5 gives the lift distribution along the wing
span.
The drag force is calculated simply from the global efficiency of the aircraft and it
is proportional to the local lift ([2]). Finally, the aerodynamic moments are calculated
for each wing section using the tool Airfoil.
4.5 Load distribution

When inertial properties and loads are known, the next step is to calculate the load
distribution. The code ETA allows us to do it, using the force method, whatever is the
number of beams which compose the structure; as the first beam has been analyzed,
the forces and moments at the ends are converted into the force and moment to apply
at the start point of second one and so on. This is possible thanks to a rotation matrix
defined according to conventions shown in Figure 4.6 At the end of the loop on load
distribution, the routine calculate the unknowns X and then the reaction forces and
moments on the beam clamp. Now the distributions of forces and moments are defined
on all wing system, in particular we have
31
1000
Lift Ala Ant

500
0
0 0.5 1 1.5 2 2.5 3 3.5 4
400
Lift Ala Post
200
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
400
Lift Bulk
200
−200
0 0.5 1 1.5 2 2.5 3
Figure 4.5: Lift distributions
• (Mx ), X bending moment;
• (My ), Y bending moment;
• (Mz ), torque;
• (Tx ), X shear force;
• (Ty ), Y shear force;
• (Tz ), axial force
These forces and moments are used during the calculation of stress on the beam sections.
The topics discussed in this section are taken from [3]. The force method allows to
evaluate load distribution using the virtual works’ principle as follows
n
X
1 · ηi = ηi0 + Xi · ηij (4.5.1)
j=1
where
• ηi is the displacement of the point where the i-th unknown is applied, evaluated
in the actual system in the direction of the unknown itself
• ηi0 is the displacement of the point where the unknown of main system is applied
in Xi direction
32
Figure 4.6: ETA reference system and conventions
• ηij is the displacement of the point where the unknown of j-th system is applied
in Xi direction
• Xi are the unknowns to be evaluated
The factors ηij are evaluated using the virtual works’ principle as follows.
Z l
1 · ηij = Ti · i + Mi χi (4.5.2)
0
where and χ are the deformation related to forces and moments. Finally forces and
moment of effective system are evaluated as follows
6
X
Tx = T0x + Xj · Tjx (4.5.3)
j=1
6
X
Ty = T0y + Xj · Tjy (4.5.4)
j=1
6
X
Tz = T0z + Xj · Tjz (4.5.5)
j=1
6
X
Mx = M0x + Xj · Mjx (4.5.6)
j=1
6
X
My = M0y + Xj · Mjy (4.5.7)
j=1
6
X
Mz = M0z + Xj · Mjz (4.5.8)
j=1
In Figure 4.7, the loads calculated for the reference case are shown, where a single
spar solution (with a box shaped cross section)is considered.
33
Mz Twist Front wing [Nm] Tz Axial force [N]

100 300
80
200
60
40
100
20
0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4
Mx Bending [Nm] Tx Shear force [N]

4000 0
3000 −50
2000 −100
1000 −150
0 −200
−1000 −250
0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4
My Bending [Nm] Ty Shear force [N]

200 0
100
−500
0
−1000
−100
−1500
−200
−300 −2000
0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4
(a) Front wing

Mz Twist Rear wing [Nm] Tz Axial force [N]
80 0
−50
60
−100
40
−150
20
−200
0 −250
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5

1000 100
500 50
0 0
−500 −50
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5

0 500
−50
0
−100
−500
−150
−200 −1000
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
(b) Rear wing

Mz Twist Winglet [Nm] Tz Axial force [N]
0 400
−10 300
−20 200
−30 100
−40 0
0 0.5 1 1.5 2 2.5 3 0 0.5 1 1.5 2 2.5 3

400 250
200 200
0 150
−200 100
−400 50
−600 0
0 0.5 1 1.5 2 2.5 3 0 0.5 1 1.5 2 2.5 3

300 250
200
200
150
100
100
0
50
−100 0
0 0.5 1 1.5 2 2.5 3 0 0.5 1 1.5 2 2.5 3
(c) Bulk
Figure 4.7: Load distribution
34
4. Structural model 4.6. Stress model
4.6 Stress model

In order to calculate stresses, the thin walled theory, exposed in [5], is used. The
stress distributions are calculated considering the spars as thin-walled, box shaped
beams. In particular, the stresses are evaluated around the entire perimeter of cross
section in terms of
• normal stresses due to axial forces,
• normal stresses due to X and Y bending moments,
• tangential stresses due to twist moment,
• tangential stresses due to X and Y shear forces
according to St. Venant’s theory. As these stresses are calculated, an equivalent

stress is defined, using the Von Mises method. This process is repeated on each
cross-section. Since the thickness of webs are defined a priori and are not included in
the design process, the maximum equivalent stress is the maximum value among the
ones calculated on both upper and lower flanges.
4.6.1 Sign conventions and notation

In order to simplify the stress computation, the section has been divided in four
components, left and right webs and upper and lower flanges. The start point of section
curve abscissa is located in the upper right corner and runs in anti-clockwise sense,
the same anti-clockwise sense is used to name the elements of the section as shown in
Figure 4.8. Forces, moments and displacements are referred to an arbitrary system
of axes O x y z , of which Oz is parallel to the longitudinal axis of the beam and xy
are axes in the plane of the cross-section. The symbols M, S, P and T are assigned to
bending moment, shear force, axial or direct load and twisting moment, with suffixes
appropriate to indicate sense or direction; thus, M, is a bending moment about the
x axis, Sx is a shear force in the x direction and so on. Figure 4.9 shows positive
directions and senses for the loads and moments applied externally to a beam; positive
bending moments M, and My induce tension in the positive first quadrant of the beam
cross-section. The following paragraph explains how the single stresses are computed.
35
x1
x2
(1) Upper flange
(2) (4)
Left s Right
web web
(3) Lower flange
x4
x3
Figure 4.8: Convention about section element name
Figure 4.9: Convention about loads
36
4.6.2 Stress due to axial load

The normal stress acting on beam cross-section due to axial load is evaluated as
P
σza = (4.6.1)
As
where As is total area of cross-section.

4
X
As = Ai = A1 + A2 + A3 + A4 (4.6.2)
i=1
4.6.3 Bending stresses

As well as the axial load, also the bending moments affect the normal stress; as
is well known, while axial load produces a constant stress in the cross-section, the
bending stress is a butterfly-shaped in the section.
From the Navier’s Theory, the stress of Mx and My bending moment is
! !
My · Ixx − M x · Ixy M x · Ixx − M y · Ixy
σzb = 2
x+ 2
y (4.6.3)
Ixx · Iyy − Ixy Ixx · Iyy − Ixy
where: Z
Ixx = x2 dA (4.6.4)
ZA
Iyy = y 2 dA (4.6.5)
Z A
Ixy = y · x dA (4.6.6)
A
in particular, since a double symmetrical cross-section is considered, we have
Ixy = 0 (4.6.7)
and equation 4.6.3 becomes

My Mx
σzb = x+ y (4.6.8)
Iyy Ixx
4.6.4 Stress due to torque

Torque produces shear stresses and the closed section beams presents a better
behavior than open ones as shown in [5], hence we assume that the torsional moment
is fully borne by a closed shell which is composed by webs and a plates that include
10% of flange thickness (Figure 4.8). In fact, since as said in chapter 2, the flanges are
build using a carbon laminates on the outer surfaces, has been supposed that these,
37
together webs, works as torque box. This fraction of thickness will be used in the next
chapters in order to define a new optimized layout.
The shear flows due to torque moment are calculated according to the Bredt-Bato’s
theory, which in absence of axial constraints
I
T = p~ · ~q ds (4.6.9)
where T is the torque (Figure 4.9)and p~ is the vector which start in the origin of cross
section axis and it is perpendicular to the wall of the section in each point. Since for
hypothesis the flows are constant 3 we have
I
~q
p~ · ds = 2A (4.6.10)
|~q|
where A is the total area enclosed by the profile of cross-section, hence the flows are
calculated as follows
T
qt = (4.6.11)
2A
and thus the stress is
q
τt = (4.6.12)
t
The stress is different in each wall of section.
4.6.5 Stress due to shear forces

In order to analyze the stresses produced by shear forces we must solve the equilib-
rium equation on a sector of shell as depicted in Figure 4.10:
∂q ∂σz
+t =0 (4.6.13)
∂s ∂z
∂q ∂σz
+t =0 (4.6.14)
∂z ∂s
Figure 4.10: Equilibrium on a shell sector

3 ∂q
because it is assumed the absence of axial constraint and hence from equilibrium we has ∂s =0
38
Under the appropriate hypothesis, or rather there aren’t hoop stress and hence the
equation 4.10 can be neglected (see [5]), the equations above becomes
!Z !Z
s s
Sx · Ixx − Sy · Ixy Sy · Ixx − Sx · Ixy
qb = 2
t(s) · x ds + 2
t(s) · y ds
Ixx · Iyy − Ixy 0 Ixx · Iyy − Ixy 0
(4.6.15)
where qb is the basic flow and Sx and Sy are the shear forces (Figure 4.9). Equation
4.6.15 is valid just for open sections where s = 0 is boundary condition where q = 0.
Since the box is a closed section we must proceed as follows;
1. Operate a "cut" in a point of section to open it, the cut is supposed to be done
on the upper right corner.
2. Calculate the basic flow using equation 4.6.15.
3. Restore the compatibility as explained below
In order to restore the compatibility, since the torsion is treated separately4 , we

assume that loads act on the shear center and hence the torque is "zero"; this means
∂θ
that also the deformation ∂z
is nil, thus the compatibility condition becomes the
following I
∂θ qs
=0= ds (4.6.16)
∂z s Gt
where qs is the total flow obtained by
qs = qb + q0 (4.6.17)
and q0 is the unknown flows at the origin of s. Now q0 can be calculated as

H
qb ds
q0 = Hs (4.6.18)
s
ds
and the stress τ is evaluated through the following definition.
qs
τs = (4.6.19)
t
4
the torque can be also included in this treatment
39
4. Structural model 4.7. Sensibility the discrete division
4.6.6 Equivalent stress

The equivalent stress is the Von Mises stress; since just one axial stress and one
tangential stress are present we have
p
σeq = σz2 + 3τ 2 (4.6.20)
where σ is the sum of normal stresses caused by bending moment and axial forces,
while τ is the sum of stresses caused by torque and shear forces. As said before, as
the equivalent stress has been calculated above the entire cross section perimeter, the
maximum value observed on the flanges has been taken as value to compare with the
allowable stress.
4.7 Sensibility the discrete division

This section aims at analyzing the sensibility of the analytic model to the number
of intervals in which the beams are divided in order to calculate the stress distributions;
if the effect of interval division variation is sufficiently small, a coarser mesh could be
used in order to obtain a lower calculation time.
However, This analysis is done in order to extend the adopted design process, as
explained in appendix D, where hundreds of objective function evaluations occur. The
calculation time plays a leading role and it is strongly affected by the mesh refinement.
The following analysis is done using three degrees of mesh refinement, so defined
1. ∆ = 2.5% of beam length
2. ∆ = 5% of beam length
3. ∆ = 10% of beam length
where ∆ is the width, as fraction of beam length between two adjacent sections. Figures
4.11, 4.12, 4.13 gives the results of compared analysis for the three conditions exposed
above.
40
Mz Twist front wing [Nm] Tz Axial force [N]
100 300
∆=10% ∆=10%
∆=2,50% 200 ∆=2,50%
∆=5% ∆=5%
4. Structural model
50
100
0 0
0 1 2 3 4 0 1 2 3 4
Mx Bending [Nm] Tx Shear [N]

2000 300
∆=10% ∆=10%
0 ∆=2,50% 200 ∆=2,50%
41
∆=5% ∆=5%
−2000 100
−4000 0
0 1 2 3 4 0 1 2 3 4
My Bending [Nm] Ty Shear [N]

200 2000
∆=10% ∆=10%
Figure 4.11: Compared analysis:cds front wing

1500
0 ∆=2,50% ∆=2,50%
∆=5% 1000 ∆=5%
−200
500
−400 0
0 1 2 3 4 0 1 2 3 4
4.7. Sensibility the discrete division
Mz Twist rear wing [Nm] Tz Axial force [N]
80 0
∆=10% −50 ∆=10%
60 ∆=2,50% ∆=2,50%
−100
40 ∆=5% ∆=5%
4. Structural model
−150
20 −200
0 −250
0 1 2 3 4 5 0 1 2 3 4 5

1000 100
∆=10% ∆=10%
500 ∆=2,50% 50 ∆=2,50%
42
∆=5% ∆=5%
0 0
−500 −50
0 1 2 3 4 5 0 1 2 3 4 5

−100 1000
Figure 4.12: Compared analysis:cds rear wing

∆=10% ∆=10%
∆=2,50% 500 ∆=2,50%
−150 ∆=5% ∆=5%
0
−200 −500
0 1 2 3 4 5 0 1 2 3 4 5
Mz Twist Winglet[Nm] Tz Axial force [N]
0
300
∆=10% ∆=10%
−10
∆=2,50% 200 ∆=2,50%
−20 ∆=5% ∆=5%
4. Structural model
100
−30
−40 0
0 1 2 3 0 1 2 3

500 0
∆=10% ∆=10%
∆=2,50% −100 ∆=2,50%
43
0 ∆=5% ∆=5%
−200
−500 −300
0 1 2 3 0 1 2 3

300 0
Figure 4.13: Compared analysis:cds winglet

∆=10% −50 ∆=10%
200
∆=2,50% ∆=2,50%
−100
100 ∆=5% ∆=5%
−150
0 −200
−100 −250
0 1 2 3 0 1 2 3
4. Structural model 4.7. Sensibility the discrete division
As Figures 4.11, 4.12, 4.13 show, the difference between the three conditions are
sufficiently small to allow a coarse mesh; the calculation times are reduced according
to Table 4.15 , where
t_carichi is the time needed to define the distribution of aerodynamic and structural
loads in seconds;
t_cds is the time needed to define the load distribution in seconds;
t_tens is the time needed to evaluate the stress distribution in seconds;
t_tot is the time needed to complete a single iteration in seconds;
∆ t_carichi t_cds t_tens t_tot
2.5% 1.0353 3.2961 10.5314 14.8628

5% 0.705 2.2987 5.4798 8.4835
10% 0.55 1.9606 2.9635 5.4747
Table 4.1: Calculation time results
Table 4.2 gives the calculation times in percent of time needed to complete a
single iteration Figure 4.14 gives the trends of several calculation times against mesh
∆(%) t_carichi(%) t_cds(%) t_tens(%) t_tot(%)
2.5% 7 22.2 70.9 100

5% 8.3 27.1 64.6 100
10% 10.1 35.8 54.1 100
Table 4.2: Calculation time results (percent)
refinement.
5
these data refers to the algorithm presented in appendix B, however this treatment can be
generalized to the case examined in this job
44
Trend calculation time (t_carichi) Trend calculation time (t_cds)
1.2 3.5
3.29
1.1
1.03
3
1
4. Structural model
0.9
2.5
0.8 2.29
time(sec)
time(sec)
0.705
0.7 1.96
2
0.6 0.55
0.5 1.5
2 4 6 8 10 2 4 6 8 10
% discr % discr
45
Trend calculation time (t_tens) Trend calculation time (t_tot)
12 15
14.86
10.53
10
8
10
5.47
6
time(sec)
time(sec)
8.48
Figure 4.14: Time calculation against mesh refinement

4
2.96
5.47
2 5
2 4 6 8 10 2 4 6 8 10
% discr % discr
Chapter 5
Design procedure
The task of this chapter is to explain the process used to optimize the structures of
the wing system. Hereafter all features related to such a process will be discussed. In
particular the algorithms, the physical and mathematical models and the assumptions
upon the optimizer is based will be exposed. The structure of the optimization code is
explained in Appendix C.
5.1 Design process work-flow

The design algorithm is based on the idea of modifying the thickness distribution of
flanges until the stress levels are inside a fixed interval. This process is performed for
a set of points of the wing system and the stress levels at each point depends on the
other points; in fact, as the thickness changes, also stiffness and stresses change, and a
thickness variation affects also the weight of the structure which produce a variation
of the load applied, due to the mass relief. Figure 5.1 gives the entire work-flow used
at this step of the optimization. Another kind of process has been tested during the
development of this tool, obtaining similar results; this process and its work-flow are
better explained in appendix B.
In the following sections, any work flow step will be explained.
46
5. Design procedure 5.1. Design process work-flow
Launch
Front wing parameters

Define initial
nz, hcruise, Vcruise Rear wing parameters
parameters
Global parameters
Calculate
geometry
Define initial
Initialize Calculate inertial
thickness
loop properties
distribution
Calculate Calculate inertial

aerodynamic loads loads
Calculate inertial Calculate

properties load distrib.
Modify Calculate
thickness stress
Verify allowable
No
stress constraint
Yes
Define
results
Exit
Figure 5.1: Design work-flow
47
5. Design procedure 5.2. Defining initial parameters
5.2 Defining initial parameters
5.2.1 Geometric parameters

Input parameters:
ba is the wing span;
Sref is the reference Surface and includes the entire rear wing surface plus front wing
surface excluding the fuselage intersection;
Rs is the ratio between the whole rear wing surface and front wing surface where the
fuselage intersection is excluded;
Λa is the sweep angle of the front wing at 25% of chord;
Λp is the sweep angle of the rear wing at 25% of chord;
Γa is the dihedral angle of front wing;
Γp is the dihedral angle of rear wing;
λa is the taper ratio of the front wing;
λp is the taper ratio of the rear wing1 ;
l/b is the ratio between the horizontal distance of the two wings and wing span;
h/b is the ratio between the vertical distance of the two wings and wing span;
Tip and root chords for both wings are calculated as follows:
Si
Cti = ! (5.2.1)
1 ba −Dfi us
1+ λi
· 2
Ct
Cri = (5.2.2)
λi
where the superscript i means that the same equation is used for both wings and Df us
is the fuselage diameter at the intersection with front or rear wing, depends which is
considered. Furthermore the surfaces of the wings are evaluated by a subroutine, as
follows
1
in this case the chord at root matches with the chord at center line
48
5. Design procedure 5.2. Defining initial parameters
Sref
Sa = (5.2.3)
1 + Rs
Sp = Sa · Rs ; (5.2.4)
5.2.2 General parameters

The general parameters required to calculate the aerodynamics loads are
• nz, vertical load factor;
• V , cruise speed;
• h, altitude;
• σmax , operating value of stress;
• Emax , acceptable error on stress around the operating value where the solution is
valid; it is expressed as percent of σmax ;
These parameters, in addition to geometric parameters and lift vector, represent the
input database needed to start the design tool. At this step an other important
parameter must be defined, namely spar position
Spar position allows to move the spar in the chordwise direction; the default position
is set at the airfoil maximum thickness (as shown in Figure 5.2);
Figure 5.2: Chordwise spar position
Another purpose of this design step is to evaluate the length of the spars and their
space positioning, in order to compute the rotation matrix. Finally, at the end of this
step we can create a 3D-plot of the spar positioning, as shown in Figure 5.3. More
details about this topic are available in [3].
49
5. Design procedure 5.3. Defining initial distribution thickness
Disposizione geometrica dei longheroni
−1
4
2 6
4
0
2
−2 0
−4 −2
Figure 5.3: Spar positioning
5.3 Defining initial distribution thickness

The initial thickness distributions, for spars, skin and ribs, is defined in order to
have an initial configuration to start the design process. Skin thickness is set as constant
for the entire wing system, while for the other elements the thickness distribution is
defined as follows.
5.3.1 Ribs
Concerning ribs, a linear variation of parameters is possible along the wing span.
In fact, according to Figure 4.1, the input requested are just the values of frame
thickness and ribs thickness at the tip and root of each wings. These values are then
interpolated in order to define the entire distribution. The default values are set in
order to correspond with the ones of the prototype.
In this step, a definition of materials in terms of elastic modulus, poisson modulus
and density is also necessary .
50
5. Design procedure 5.4. Calculating inertial properties
5.3.2 Spars
About spars, as ribs, the same linear distribution of thickness is prescribed. The
spars components are webs and flanges, for both the thickness at the tip and root
of each wing are requested; since the initial data are set upon the aircraft prototype
values, webs have a constant thickness along span and the flanges are tapered at the
wing tips. At this step the definition of material properties is requested.
Moreover, since only the flanges are affected by the optimization process, the initial
web thickness remains unchanged during all process.
5.4 Calculating inertial properties

As the geometrical parameters for spars are defined, since at this step the code is
taken from the code MDSA-2010 , a selection of the shape of cross sections is necessary
for each wing. The types of section available are shown in Figure 4.2 and since the aim
of this job is to optimize the prototype exposed in chapter 2, the default section is the
box-shaped one. In this step, the width of spar cross section is required at wing tip
and wing root.
In order to have an initial stiffness distributions, used in the following to calculate
the initial stress distribution the calculation of inertial properties is possible in all
section. Figure 5.4 gives the solutions that can be obtained at the and of this step.
51
5. Design procedure 5.4. Calculating inertial properties
0.06
0.04 3.5
3
0.02
2.5
0
2
−0.02
1.5
−0.04
1
−0.06 0.5
0.04 0.02 0 −0.02 −0.04

0
(a) Box-shaped spar
3.5
2.5
0.06
2
0.04
0.02 1.5
−0.02 1
−0.04
0.5
−0.06
0.04 0.02 0 −0.02−0.04

0
(b) Rectangular spar
0.06
0.04
3.5
0.02 3
0 2.5
−0.02 2
1.5
−0.04
1
−0.06
0.5
0.04 0.02 0 −0.02−0.04

0
(c) double T-shaped spar
Figure 5.4: Geometry for initialize the optimization process
52
5. Design procedure 5.5. Initialize loop
5.5 Initialize loop

As the initial data set is completely defined, the optimization loop starts and it
works as shown in Figure 5.1. This loop ends when the convergence is verified or when
the maximum number of iteration is reached; hundred iteration are sufficient.
5.6 Loads calculation

As said before, loads acting on spars are affected also by the structural weight and,
then, the effect of the mass relief is evaluated at any iteration.The aerodynamic loads
are computed by the code MDSA-2010 as exposed in [3] and remain constant during
every iteration
A distinction between loads due to ribs and skin and loads due to spars is necessary.
The weight per unit length of ribs and skin is evaluated as
Wwing
P (z) = · c(z) · nz (5.6.1)
Swing
where Wwing is the total weight of skin an ribs, Swing is the wing reference surface
and c(x) is the law of chord variation; finally nz is the vertical load factor. Because
the quantities in 5.6.1 are given, these inertial loads don’t change during the iterative
process. Concerning spars, weight and thickness distributions are correlated. Each spar
has been split in a number of bays holding a constant pitch (Figure 5.5); volume and
the weight distribution has been computed, and the total weigh of spars is obtained
summing the contribution
wi = Ai · ∆zi · ρ (5.6.2)
Xn
Wspar = wi (5.6.3)
i=1
pi = −wi · nz (5.6.4)
where wi is the element of vector which represents the weight distribution, Wspar is
the total weight and pi is the element of vector which represents the load distribution
acting on the beam.
5.7 Load distribution and stress calculation

At this step, as shown in Figure 5.1,it is possible to evaluate the load distribution
using the M üller − Breslau equations as explained in section 4.5; thus, the stress
distribution can be obtained as depicted in section 4.6.
53
5. Design procedure 5.8. Stress constraint check and thickness definition
A(z)
Dz
Figure 5.5: Spar splitting
5.8 Stress constraint check and thickness definition

When the maximum equivalent stresses are known on each section of the beams, a
comparison between them and the allowable stress is performed.
σa − < σe < σa − (5.8.1)
where σa is the allowable stress, σe is the equivalent stress defined above and is an
error margin defined by user. When equation 5.8.1 is verified in all cross-section, a new
distribution of thickness is obtained. Three situation may occurs:
1. σe < σa −
2. σe > σa +
3. σa − < σe < σa −
In the first case, the equivalent stress is too small and the section considered is oversized,
thus, thickness is reduced in order to achieve the minimum weight solution. In the
second case the stress exceeds the limit and an increase of thickness is required. In the
third case, the stress is adequate and no changes of thickness occurs. The thickness
modification happens according to the equation below
ti+1 = ti ± ∆t (5.8.2)
where ∆t is an increment defined by user and the sign depends on the gap between
the actual and the allowable stress. As a new thickness is defined, a new computation
of inertial properties is performed in order to restart the loop from the point exposed
in paragraph 5.6. This process ends when the entire equivalent stress distribution is
enclosed in the range defined by equation 5.8.1.
5.9 Global weight definition

When the stress level is verified on the entire spar the final thickness distribution is
defined, and the evaluation of global weight is possible; the global weight includes the
54
5. Design procedure 5.9. Global weight definition
sigmae>sigmaa+err sigmaa-err<sigmae<sigmaa+err sigmae>sigmaa-err
ti+1=ti+ t ti+1=ti ti+1=ti- t
calculate new inertial properties
Figure 5.6: Work-flow to define thickness
contribution of skin, ribs and spars.
Wtot = Wskin + Wribs + Wspars (5.9.1)
The weight of spar is defined as follows

n Z
X l
Wspar = A(x) · ρ dx (5.9.2)
i=1 0
where n is the number of spars, A(x) is the distribution of area and ρ is the density of
material.
55
Chapter 6
Results
This chapter shows the results of the design process, concerning thickness and stress
distributions, and moreover explains how these are used in order to modify the aircraft
described in chapter 2.
6.1 Design results

The design code gives, finally, three main outputs, or rather a graph which plots the
load distributionfor each wing and another one which plots the thickness distribution
and the related maximum equivalent stress. Finally, the tool returns on command
windows the matrix shown below, which contains the weight of components and total
weight of structure.
matrice pesi =
longh cent skin tot
ant 5.54173 1.00534 5.09823 11.64530
bulk 2.92450 0.58215 3.84426 7.35091
post 4.39161 9.46982 6.93742 20.79886
TOT 0 0 0 39.79507
Since as said in the previous chapter, the model allows the user to implement only
an equivalent isotropic material, the weight so calculated might not exactly matches
with the F.E.M. models. In particular in this case each property has been evaluated
considering the features of material in the X direction of laminate according with the
equation 6.1.1 Pn
i:1 xi · ti
X= P n (6.1.1)
i:1 ti
where xi is the generic properties of the i − th ply, ti is the thickness of i − th ply,
n is the total number of plies and X is the generic properties of equivalent material.
Figures 6.1, 6.2 give load and thickness distributions respectively. The mini-
mum of thickness is not located at the tip, like for conventional airplanes. In the
56
6. Results 6.1. Design results
PrandtlPlane the load distribution and hence the thickness are affected by the over-
constrained nature of the structures. In this concern, the point of minimum stress
depends on load applied; in this case Figure 6.2 highlights that it is located about at
50% of span.
57
4 Front wing Winglet Rear wing
x 10
1 5000 5000
6. Results
0 0 0
My [Nm]
My [Nm]
My [Nm]
−1 −5000 −5000
0 2 4 6 8 10 12 0 2 4 6 8 10 12 0 2 4 6 8 10 12
ID ID ID
−182 56 −36
−184 54 −38
Mx [Nm]
Mx [Nm]
Mx [Nm]
−186 52 −40
0 2 4 6 8 10 12 0 2 4 6 8 10 12 0 2 4 6 8 10 12
ID ID ID
−500 1000 0
−1000 500 −1000
Mz [Nm]
Mz [Nm]
Mz [Nm]
−1500 0 −2000
0 2 4 6 8 10 12 0 2 4 6 8 10 12 0 2 4 6 8 10 12
ID ID ID
58
1000 1000 0
0 0 −200
Ty [N]
Ty [N]
Ty [N]
−400
−1000 −1000
0 2 4 6 8 10 12 0 2 4 6 8 10 12 0 2 4 6 8 10 12
ID ID ID
1578 136 −1560

1576 134 −1562
Tx [N]
Tx [N]
Tx [N]
1574 132 −1564

0 2 4 6 8 10 12 0 2 4 6 8 10 12 0 2 4 6 8 10 12
ID ID ID
Figure 6.1: Results:distribution of load characteristics

4
x 10
1 −1520 5000
0 −1522 0
Tz [N]
Tz [N]
Tz [N]
−1 −1524 −5000
0 2 4 6 8 10 12 0 2 4 6 8 10 12 0 2 4 6 8 10 12
ID ID ID
6.1. Design results
thickness distribution : front wing 7 stress distribution : front wing
x 10
30 4
6. Results
25 3
20 2
t[mm]
σeq[MPa]
15 1
10 0
1 2 3 4 5 6 7 8 9 10 11 1 2 3 4 5 6 7 8 9 10 11
ID ID
thickness distribution : winglet 7 stress distribution : winglet
x 10
25 4
20 3
15 2
t[mm]
59
σeq[MPa]
10 1
5 0
1 2 3 4 5 6 7 8 9 10 11 1 2 3 4 5 6 7 8 9 10 11
ID ID
thickness distribution : rear wing 7 stress distribution : rear wing
x 10
25 4
20
3
15
2
t[mm]
10
σeq[MPa]
1
5
Figure 6.2: Results:distribution of thickness and related equivalent stresses

0 0
1 2 3 4 5 6 7 8 9 10 11 1 2 3 4 5 6 7 8 9 10 11
ID ID
6.1. Design results
6. Results 6.2. Converting results into design
6.2 Converting results into design

The distribution of thickness obtained as the results of optimization must be
converted into an actual structural design. In fact the code gives the thickness
distribution as a continuous function along wing span, but such a solution can be
difficult to be manufactured and, also too much expensive. Therefore, the continuous
function of thickness has been transformed in a step function, where the step width is
defined properly; this condition is presented in Figure 6.3. The value of thickness is
upper flange division
section 1 section 2 section 3 section 4 section 5
1°step=0.45m
2°step=0.95m
3°step=1.6m
4°step=2.32m
lower flange division
section 1 section 2 section 3
1°step=1.2m
2°step=2.32m
Figure 6.3: Step width for upper and lower flanges
calculated as the average inside each division according to equation 6.2.1

R
t(x) dx
Tav = b (6.2.1)
b
where b is the step width, t(x) is the thickness function and Tav is the average value
of thickness; the results are shown in Figure 6.4. Since the flanges are built using a
laminate of wood and spruce, a separation between layers has been done imposing that
the carbon thickness is the 10% of total. The results of this process are exposed in
Tables 6.1, 6.2.
Finally, the comparison between the prototype layout and the analytic distribution,
(Figure 6.5) brings out the fact that, in general, the material has been moved from
the wing root to the wing tip and, in particular, a significant thickness variation has
60
6. Results 6.2. Converting results into design
occurred on the rear wing. Concerning vertical winglets, some partitioning is based
upon the default geometry of F.E.M. model defined in [2], but since this part of wing
system is built using a full carbon spar instead of material used to optimize the front
and rear wing, the results are not consistent. In this respect the new layout has been
defined just for the spars of main wings, while the winglet has been treated in a different
way(exposed in the following chapters).
61
thickness step func. :front wing (upper flange) thickness step func. :front wing (lower flange)
30 30
6. Results
step func. step func.

25 25 analytic func.
analytic func.
20 20
15 15
thickness [mm]
thickness [mm]
10 10
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
relative coordinate relative coordinate
thickness step func. :rear wing (upper flange) thickness step func. :rear wing (lower flange)
30 30
20 20
62
10 step func. 10
step func.
thickness [mm]
thickness [mm]
analytic func. analytic func.
0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Figure 6.4: Thickness step functions

thickness step func. :winglet (upper flange) thickness step func. :winglet (lower flange)
25 25
step func. step func.
20 20 analytic func.
analytic func.
15 15
10 10
thickness [mm]
thickness [mm]
5 5
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
6.2. Converting results into design
6. Results
Section location[m] Spruce ply thickness[mm] Carbon ply thickness[mm] Total thickness[mm]
Upper flange
0 ⇒ 0.45 22 2.5 24.54

0.45 ⇒ 0.95 19 2.2 21.62
0.95 ⇒ 1.6 15 1.6 16.29
1.6 ⇒ 2.32 11 1.2 12.13
63
2.32 ⇒ 3.72 15 1.7 16.52
Lower flange
0 ⇒ 1.2 20 2.2 21.99

1.2 ⇒ 2.32 12 1.3 13.18
2.32 ⇒ 3.72 15 1.7 16.53
Table 6.1: Flange spar thickness (front wing)
6. Results
Section location[m] Spruce ply thickness[mm] Carbon ply thickness[mm] Total thickness[mm]
Upper flange
0 ⇒ 0.45 9 1 9.88
0.45 ⇒ 0.95 9 1 9.60
0.95 ⇒ 1.6 7 0.7 7.40
1.6 ⇒ 2.32 5 0.7 5.76
64
2.32 ⇒ 3.72 15 1.7 17.06
Lower flange
0 ⇒ 1.2 9 0.9 9.46

1.2 ⇒ 2.32 6 0.6 6.12
2.32 ⇒ 3.72 15 1.7 17.06
Table 6.2: Flange spar thickness (rear wing)
comparison prot. -conf. opt. :front wing (upper flange) comparison prot. -conf. opt. :front wing (lower flange)
35 26
6. Results
24
30
22
25 20
18
20
16
thickness [mm]
thickness [mm]
15 14
12
10
prototype 10 prototype
analytic distribuiton analytic distribuiton
5 8
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
65
comparison prot. -conf. opt. :rear wing (upper flange) comparison prot. -conf. opt. :rear wing (lower flange)
35 25
30
20
25
15
20
15
10
thickness [mm]
thickness [mm]
10
5
Figure 6.5: Comparison between prototype and optimized layout

5 prototype prototype
analytic distribuiton analytic distribuiton
0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Part III
F.E.M. Analysis
66
Chapter 7
F.E.M. Modeling
In the previous chapters new flanges thickness distributions, have been defined. In
the following astatic F.E. analysis is shown on the modified structure This chapter is
divided into three steps
Modeling, description of finite element model
1st step, in which the new configuration is analyzed;
2nd step, some thickness of the new configuration are modified according to the
results of step 1
7.1 Modeling
The finite element model has been developed in [2], in which the structural solution
of the prototype has been analyzed; this solution has been used as a starting point.
Since the model implemented in the design code does not include the contributions
of skins and auxiliary spars, an analysis has been done using the structure presented in
Figure 7.2. In this case the presence of ribs can’t be avoided, since they introduce the
load on spars.
7.1.1 Main spars

Main spars are shown in Figure 7.3; they are modeled using CQUAD4 elements for
webs, flanges and reinforcing pads (Figure 7.4).
CQUAD4 are a four nodes shell element, to carry any kind of in-plane and out of
plane forces and moment, and allowing us to take both plate and membrane effect
into account. These elements is quadrangular with a linear strain field along the edges.
Figure 7.5 gives the element coordinate system, x and y axes are perpendicular each
67
7. F.E.M. Modeling 7.1. Modeling
Figure 7.1: Starting model
Figure 7.2: Base model without skin and auxiliary spar
other and lay on the plane, while z axis depends on the order in which the nodes
G1,G2,G3,G4 are assigned; z positive accords to the right hand rule. Furthermore, if
68
Figure 7.3: Spar model
Figure 7.4: Spar model (internal structures)
a non isotropic material is considered, it is possible to define his orientation through

the parameters θ, which is the angle between x axis of material coordinate system
and element x axis (Figure 7.5). Finally, the merging of nodes is obtained using the
command Equivalence to connect each other all the parts of the spar.
Figure 7.5: Element coordinates system
7.1.2 Auxiliary spar

Auxiliary spars are modeled using element CQUAD4. They are linked to all ribs
apart from the bulkheads located at the wing roots, in fact that nodes are excluded
from the merging process (Figure 7.6).
69
Figure 7.6: Nodes besides the bulkhead
7.1.3 Ribs
The ribs are modeled as a frame, and are implemented using CTRIA3 element1 ,
because CQUAD4 could lead excessive distortions on element. The ribs fem model is
shown in Figure 7.7
Figure 7.7: Ribs model
1
they have the same features of CQUAD4 but 3 nodes instead 4
70
7.1.4 Skin panels

The skin panels are entirely modeled using CQUAD4 element. The skin of front
wing is shown in Fig. 7.8
Figure 7.8: Skin model
7.1.5 Winglet
The winglet structures is realized with a box shaped spar modeled with CQUAD4
elements, while the ribs are modeled using CTRIA3 elements for the reasons explained
above. Figure 7.9 "a" gives the the internal structures of winglet while Figure 7.9 "b"
gives the complete winglet.
71
(a) Internal
(b) Complete
Figure 7.9: Winglet model
72
7.1.6 Laminate modeling

The laminates are modeled through the interface shown in Figure 7.10.
Figure 7.10: Laminate builder tool
Plies sequence origins from the z axis coordinate system, then the second is placed
upon the first and so on, as shown in Figure 7.11. The offset of laminate is such to
make correspond the external ply with the model surface. A check of the direction
of the z axis of elements is required, since the orientation affects the stacking layout
(Figure 7.12).
Figure 7.11: Material Coordinate system
73
reference surface
offset
layer 1
surface layer 2
z axis layer 3
layer 4
layer 5
layer 6
material
z axis
Figure 7.12: stacking layout
7.1.7 Constraints
The prismatic couple Spar-Karman has been simulated by considering that the x
displacement of the front web and the z displacement of the nodes of upper flanges
alongside the bulkhead are not permitted, because of the load applied; all the x,y and
z displacements of the nodes corresponding to the pin location are constrained (Figure
7.13).
Figure 7.13: Spar-Karman constraints
The links between wing tips and winglets are defined in order to provide the joint
of maximum stiffness. Then, the connection has been made using the MPC devices, in
particular the elements RBE2 and RJOINT. Elements RBE2 are defined through a
master node alongside the center of gravity of the spar section and several slave nodes
are placed on the section perimeter. This allows us to distribute the load acting on
74
the master node to the whole section. Finally, the joint is realized using the RJOINT
element which provides the load transfer between the master node of the two section to
connect. Two rigid joints are taken into account; in the first one the wing tip is linked
only to the winglet spar (Figure 7.14) while, in the second one, the entire winglet is
connected (Figure 7.15).
Figure 7.14: Connection of winglet spar with wing tip
Figure 7.15: Connection of full winglet with wing tip
7.1.8 Load introduction

The loads are introduced through the MPC elements called RBE3 , which allow
us to distribute the load from the master node to the entire rib, as shown in Figure
7.16.The master nodes are located in the centers of pressure acquired from [4]. Figure
7.17 gives the whole load system.
75
Figure 7.16: Load introduction through MPC
Figure 7.17: Entire load system
76
7. F.E.M. Modeling 7.2. Definition of the flange thickness: 1st step
7.2 Definition of the flange thickness: 1st step

The layout defined in Tables 6.1 and 6.2 is assumed as the first step in the opti-
mization process to design the flange thickness. Each section has assumed to be made
by two plies of wood and one ply of carbon considering that each ply of spruce has
about half of the total wood thickness. This is not necessarily the best way to build
these laminates, in fact, this layout has been assumed arbitrarily and hence it can be
optimized in order to improve the laminate properties. Figure 7.18 shows how the
laminates are divided along the span; in each trunk a different laminate is used, in
particular for the upper flanges of both wings five different stacking layout are used
while for the lower flanges three laminates are used. Some example of the laminate
used are shown in Figures 7.19, 7.20
upper flange division
section 1 section 2 section 3 section 4 section 5
1°step=0.45m
2°step=0.95m
3°step=1.6m
4°step=2.32m
lower flange division
section 1 section 2 section 3
1°step=1.2m
2°step=2.32m
Figure 7.18: Definition of wing trunks
7.2.1 Results
From the first analysis on the configuration depicted, three main topics arise:
• an overload occurs on the webs, especially in the regions close to the wing roots;
Figure 7.21 gives the failure index evaluated according to the Hoffmann’s criterion
(exposed in Appendix 8.1), in the purple colored regions, the value is bigger than
1 and, then, a failure occurs.
77
(a) lower flange: section 1 of 3 (thickness in m)
(b) Upper flange: section 1 of 5 (thickness in m)
Figure 7.19: Example of staking layout (front spar)
(a) lower flange: section 1 of 3 (thickness in m)
(b) Upper flange: section 1 of 5 (thickness in m)
Figure 7.20: Example of staking layout (rear spar)
• Stress and thus failure index are fairly low in the winglets (Figure 7.22). Hence,
in order to reduce the weight, thickness is reduced in this part of the structure.
• Close to the tips of both front and rear wings, stresses are quite low, especially
on the flanges (Figure 7.22).
78
(a) Rear view (b) Front view
Figure 7.21: Failure on webs
Figure 7.22: Failure index
79
7. F.E.M. Modeling 7.3. Definition of thickness: 2nd step
7.3 Definition of thickness: 2nd step

The structure has been modified. In particular, the webs are reinforced at wing
root, while at wing tips the thickness has been reduced. The web thickness passes from
a constant value of 3mm to 2.1mm along the beam length, apart from the region close
to the roots (Figure 7.23) where the webs are reinforced by two carbon plies. This
layout is shown in Figure 7.24.
Front Wing
1.6 m
rear web
root
front web
0.95 m
Rear Wing
1.6 m
rear web
root
front web
0.45 m
wood 1m
wood reinforced carbon
Figure 7.23: Location of carbon pads on web
Figure 7.24: Layout of reinforced webs
The flanges are initially oversized especially in the tip regions; the thickness of spars
is then reduced (see Figure 6.3) and the layouts of flanges become the ones shown in
Figures 7.25 and 7.26, where, since only the normal stresses are present in the flanges,
the fibers are oriented as the beam (angle=0), while the webs, to sustain shear stresses,
include 45 deg fibers.
Finally, the winglet is built using the layout shown in Figure 7.27. For both flanges
and webs , the results of this analysis are proposed in the next chapter.
80
7. F.E.M. Modeling 7.3. Definition of thickness: 2nd step
(a) lower flange: section 3 of 3
(b) Upper flange: section 5 of 5
Figure 7.25: New staking layout (front spar)
(a) lower flange: section 3 of 3
(b) Upper flange: section 5 of 5
Figure 7.26: New staking layout (rear spar)
Figure 7.27: Layout of winglet webs and flanges
81
Chapter 8
F.E.M. Results
This chapter shows the results obtained by the F.E.M. analysis. In particular, a
comparison between the actual structure of the prototype and optimized structure is
depicted at both maximum load factor (nz=4) and the minimum load factor (nz=-2).
The results are presented in terms of displacements, stresses and Hoffman’s Theory
failure index, exposed in section 8.1. Failure occurs when the failure index is bigger
than 1, according to Figure 8.1; the effects of skin and auxiliary spar will be evaluated.
NO FAILURE ZONE
Figure 8.1: Failure index: legend
82
8. F.E.M. Results 8.1. Hoffman’s theory
8.1 Hoffman’s theory

This section aims at giving a simple description of the Hoffman’s theory ([6], [7]),
a failure criterion used to evaluate the strength in composite materials.This criterion
permits to establish if some failure occurs in a specific ply; moreover, both tension and
compression failure strength are taken into account. In particular the failure occurs
when
F1 σx + F2 σy + F11 σx2 + F22 σy2 + F33 τ 2 + 2F12 σx σy ≥ 1 (8.1.1)
where
1 1
F1 = − (8.1.2)
Xt Xc
1 1
F2 = − (8.1.3)
Yt Yc
1
F11 = (8.1.4)
X t Xc
1
F22 = (8.1.5)
Yt Yc
1
F33 = 2 (8.1.6)
S
−1
F12 = (8.1.7)
2Xt Xc
(8.1.8)
where Xt ,Yt ,Xc ,Yc are the tensile and compressive strengths in longitudinal and transver-
sal ply direction, while S is the shear strength. The left-hand side expression in 8.1.1
is dented as the ply failure index (F.I.) and 8.1.1 is equivalent to
F.I. ≥ 1 (8.1.9)
83
8. F.E.M. Results 8.2. Comparison between Prototype - New structure
8.2 Comparison between Prototype - New structure
8.2.1 Spars model : nz = 4
(a) prototype.
(b) Optimized structure.
Figure 8.2: Failure (front view)
84
(a) prototype.
Figure 8.3: Failure (rear view)
85
8.2.2 Full model : nz = 4
(a) prototype.
Figure 8.4: Failure (front view)
86
(a) prototype.
Figure 8.5: Failure (rear view)
87
8. F.E.M. Results 8.3. New structure
8.3 New structure
8.3.1 Spars model : nz = 4
(a) Front wing.
(b) Rear wing.
Figure 8.6: Failure (Karman section)
88
(a) Front wing.
(b) Rear wing.
Figure 8.7: Failure index (Webs)
89
(a) Front wing.
(b) Rear wing.
Figure 8.8: Failure index (Flanges)
90
Figure 8.9: Failure index (Bulk)
91
(a) Carbon.
(b) Spruce.
Figure 8.10: Von Mises stress [MPa]
92
8.3.2 Full model : nz = 4
(a) Front wing.
(b) Rear wing.
Figure 8.11: Failure (Karman section)
93
(a) Front wing.
(b) Rear wing.
Figure 8.12: Failure index (Webs)
94
(a) Front wing.
(b) Rear wing.
Figure 8.13: Failure index (Flanges)
95
Figure 8.14: Failure index (Bulk)
96
(a) Carbon.
(b) Spruce.
97
8.3.3 Full model : nz = −2
(a) Front view.
(b) Rear view.
Figure 8.16: Failure
98
Figure 8.17: Failure index
99
(a) Carbon.
(b) Wood.
100
8. F.E.M. Results 8.4. Results of process
8.4 Results of process

The final results of the design process in terms of weight of spars are summarized in
Table 8.1 and it appears that the overall weight has been reduced from the 18.2 Kg of
the prototype to the 15.8 Kg of new design. The importance of this result is enhanced
by considering that an increase of stiffness of the wing system is associated to such
weight reduction. In fact, as shown in Table 8.2 the maximum vertical displacement has
been reduced from 0.14[m] to 0.11[m]. About the same trends can be observed in the
only spar model. Table 8.3 gives the comparison of the several thickness distribution
between the prototype and new structure1 ; such a table highlights that the a reduction
of thickness occurs on the wing roots (oversized in the prototype) while the contrary
occurs at the wing tip. This is due to over-constrained nature of the wing system, which
was not considered during the prototype development since the regulation imposed to
test both front and rear wings as cantilever beams. Concerning the effect of constraints,
Figure 8.11 shows that the peaks of stress occur where the spar structure is constrained;
in particular the more loaded areas are located around the holes of pins and bulkhead.
This effects are probably due to stress concentrations in these locations and, thus the
actual stress field could be determined by a more accurate static analysis. A similar
effect of stress concentration is detectable in the spar model (Figure 8.3b) on web,
where the carbon plate is interrupted (see Figure 7.23).
Parameter Prototype New structure
Front spar [Kg] 7.43 8.31

Rear spar [Kg] 7.37 5.56
Bulk spar [Kg] 3.38 1.97
Spars tot [Kg] 18.2 15.85
Table 8.1: Weights
Parameter Prototype New structure
spar model full model spar model full model
Vert. max disp [m] 0.193 0.14 0.16 0.11

−2 −2 −2
Hor. max disp [m] 5.93 · 10 4.16 · 10 4.78 · 10 2.68 · 10−2
Table 8.2: Displacements
1
in according to the beam division shown in Figure7.18
2
see Figure 7.23
101
8. F.E.M. Results 8.4. Results of process
Spar section Prototype New structure
Front wing spar (upper flange)
0 ⇒ 0.45[m] 26.4 24.5

0.45 ⇒ 0.95[m] 34.4 19.2
0.95 ⇒ 1.6[m] 26.4 16.6
1.6 ⇒ 2.32[m] 18.4 17.8
2.32 ⇒ 3.72[m] 9.2 13.7
Front wing spar (lower flange)
0 ⇒ 1.2 24 22.2
1.2 ⇒ 2.32 16 13.3
2.32 ⇒ 3.72 8 16.3
Rear wing spar (upper flange)
0 ⇒ 0.45[m] 26.4 10
0.45 ⇒ 0.95[m] 34.4 10
0.95 ⇒ 1.6[m] 26.4 7.7
1.6 ⇒ 2.32[m] 18.4 5.6
2.32 ⇒ 3.72[m] 9.2 16.7
Rear wing spar (lower flange)
0 ⇒ 1.2[m] 24 9.8
1.2 ⇒ 2.32[m] 16 6.6
2.32 ⇒ 3.72[m] 8 13.3
Spars (webs)
Wood sections[mm]2 3 3
Reinforced sections[mm] 2
Bulk spar (flanges)
[mm] 6 2.1
Bulk spar (webs)
[mm] 3.6 2.1

Table 8.3: Thickness distribution (mm)
102
8. F.E.M. Results 8.5. Final layout of spars
8.5 Final layout of spars

This section aims at show a comparison between the flanges layout of the prototype
and the new designed structure; Figure 8.19 gives the notation used to define the
several section of the beams. As said before, The upper flanges are divided in 5 sections
called from 1u to 5u from the root to tip, while the lower flanges are divided in 3
sections called from 1u to 3u. Figures 8.20, 8.21, 8.22, 8.23 gives the flanges layout of
front spar upper flange, rear spar upper flange, front spar lower flanges and rear spar
lower flange respectively.
3l
5u
2°step=2.32m
4°step=2.32m
2l
4u
upper flange
3°step=1.6m
lower flange
1°step=1.2m
2°step=0.95m
3u
1°step=0.45m
2u
1l
1u
Figure 8.19: Spars final layout: notations
103
NEW STRUCTURE PROTOTYPE

SEZ 1u (front spar)
1.2
CARBON
ply thickness [mm]
2.5
CARBON
fiber orientation
0 deg
ply thickness [mm]
1.2
fiber orientation
0 deg
SPRUCE 0 deg
8
11
SPRUCE 0 deg
SPRUCE 0 deg
8
11
SPRUCE 0 deg
SPRUCE 0 deg
8
SEZ 2u (front spar)
1.2
CARBON
fiber orientation
ply thickness [mm]
2.2
0 deg
1.2
CARBON
fiber orientation
0 deg ply thickness [mm]

SPRUCE 0 deg
SPRUCE 0 deg 8
9
SPRUCE 0 deg
8
SPRUCE 0 deg
8
SPRUCE 0 deg
8
SPRUCE 0 deg
8
SEZ 3u (front spar)

1.2
CARBON
ply thickness [mm]
ply thickness [mm]
1.6
fiber orientation
CARBON 0 deg
1.2
fiber orientation
0 deg
SPRUCE 0 deg
8
SPRUCE 0 deg
8
SPRUCE 0 deg
8
SPRUCE 0 deg
7
SPRUCE 0 deg
8
SEZ 4u (front spar)

ply thickness [mm]
1.8
CARBON
1.2
CARBON
fiber orientation
ply thickness [mm]
0 deg
fiber orientation
0 deg
1.2
SPRUCE 0 deg
8
SPRUCE 0 deg
8
SPRUCE 0 deg
8
SPRUCE 0 deg
8
SEZ 5u (front spar)

ply thickness [mm]
1.7
CARBON
fiber orientation
0 deg
SPRUCE 0 deg
6
SPRUCE 0 deg
6
Figure 8.20: Spars final layout: upper flange
104
NEW STRUCTURE PROTOTYPE

SEZ 1u (rear spar)
1.2
CARBON
ply thickness [mm]
CARBON
fiber orientation
0 deg
1.2
1
ply thickness [mm]

fiber orientation
0 deg
SPRUCE 0 deg
8
SPRUCE 0 deg
5
SPRUCE 0 deg
8
SPRUCE 0 deg
4
SPRUCE 0 deg
8
SEZ 2u (rear spar)
1.2
CARBON
fiber orientation
0 deg
1.2
ply thickness [mm]
CARBON
ply thickness [mm]
1
fiber orientation
0 deg 0 deg
SPRUCE
SPRUCE 0 deg 8
5
SPRUCE 0 deg
8
SPRUCE 0 deg
4
SPRUCE 0 deg
8
SPRUCE 0 deg
8
SEZ 3u (rear spar)

1.2
CARBON
fiber orientation
0 deg
1.2
ply thickness [mm]
SPRUCE 0 deg
8
SPRUCE 0 deg
8
SPRUCE 0 deg
8
SEZ 4u (rear spar)

1.2
CARBON
ply thickness [mm]
fiber orientation
0 deg
1.2
SPRUCE 0 deg
8
SPRUCE 0 deg
8
SEZ 5u (rear spar)

ply thickness [mm]
1.7
CARBON
fiber orientation
0 deg
SPRUCE 0 deg
8
SPRUCE 0 deg
7
Figure 8.21: Spars final layout: upper flange
105
NEW STRUCTURE SEZ 1l (front spar) PROTOTYPE
8
SPRUCE 0 deg
10
SPRUCE 0 deg
8. F.E.M. Results
8
0 deg
SPRUCE
10
0 deg
SPRUCE
fiber orientation
8
0 deg 0 deg
ply thickness [mm]
fiber orientation
SPRUCE
ply thickness [mm]

CARBON
2.2
SEZ 2l (front spar)
6
SPRUCE 0 deg
106
SPRUCE 0 deg
6
0 deg
SPRUCE
8
0 deg
SPRUCE
0 deg
fiber orientation
ply thickness [mm]

fiber orientation
CARBON
ply thickness [mm]
1.3
SEZ 3l (front spar)
Figure 8.22: Spars final layout: lower flange

6
SPRUCE 0 deg
8
0 deg
6
0 deg SPRUCE
SPRUCE
0 deg
fiber orientation
ply thickness [mm]

fiber orientation
CARBON
ply thickness [mm]
1.3
8.5. Final layout of spars
107
Figure 8.23: Spars final layout: lower flange
ply thickness [mm] ply thickness [mm] ply thickness [mm]
1.3 6 6 0.6 6 0.9 5 4
SPRUCE
CARBON
SPRUCE
SPRUCE
CARBON
SPRUCE
SPRUCE
CARBON
NEW STRUCTURE
0 deg
0 deg
0 deg
0 deg
0 deg
0 deg
0 deg
0 deg
fiber orientation fiber orientation fiber orientation

ply thickness [mm] ply thickness [mm] ply thickness [mm]
8 8 8 8 8 8
SEZ 3l (rear spar)

SEZ 2l (rear spar)
SEZ 1l (rear spar)
SPRUCE
SPRUCE
SPRUCE
SPRUCE
SPRUCE
SPRUCE
PROTOTYPE
0 deg
0 deg
0 deg
0 deg
0 deg
0 deg
fiber orientation fiber orientation fiber orientation
8.5. Final layout of spars 8. F.E.M. Results
Chapter 9
Conclusions
In this work a procedure to design the wing structure of a ULM PrandtlPlane has
been developed starting from the configuration of a prototype of ultra light aircraft
which was defined in [4] and analyzed in [2]. The design procedure has been applied to
define a new thickness distribution for the flanges of the wing system, in fact, since the
regulations impose that the validation tests must be done considering both front and
rear wing separately (without the winglet), the wing structures of such prototype was
defined as for a traditional aircraft.
Such procedure has been accomplished through the utilization of several software

R
R
as Matlab and Nastran . In the first part a numeric algorithm has been

R
implemented in Matlab language starting from the MDSA-2010 code defined in
[3]. This code use the virtual works’ principle to evaluate the load distribution and
then the stress level on the beams which compose the wing system. At the end of
process a new thickness distribution for flange is defined. The second part consist in
the validation of new structures through the utilization of fem tool, in particular in

R
R
this work Patran has been used as preprocessor while Nastran is the solver.
All the analysis developed concerned only static effects, no buckling phenomena has
been investigated.
9.1 Issues
The design process led to a structure stiffer and lighter than the prototype one, in
fact it has been observed a reduction of about 13% in weight on the spars structure,
to which a reduction in maximum vertical displacement of about 21.4% corresponds.
This result is the proof that, in this case, the over constrained structures allows the
designer to define a non traditional layout of spar, which provide a greater stiffness
and lower weight than traditional wings, leading advantages in terms of safety and fuel
108
9. Conclusions 9.2. Later developments
consumption.
Moreover the finite element analysis shows that regions of spars besides the wing
roots, especially web are overloaded in the prototype. The reinforcing carbon pads
are able to reduce this effect (as the fem analysis on the optimized structure gives).
Further effects of stress concentration can be due to constraints modeling.
9.2 Later developments

Even if the gains given above appears quite small, this process, applied to long
range aircrafts could lead significant reductions in terms of structural weight and hence
in fuel consumption. Furthermore, the effect of weight reduction could became more
important on this kind of aircraft, since in them, the main structure affects the overall
weight in higher percent than very light airplanes, in fact unlike the spars used in this
prototype, the wing box of a large airplane spreading for about 50% of chords.
Concerning the laminates, the number of plies of carbon an wood used in this job
is not definitely the best arrangement available, in fact several parameters affect the
numbers and thickness of each layer. In particular the stacking layout must be defined
in order to avoid a too much expensive and complex manufacturing process. In this
respect such features should be defined in according to the manufacturer company.
About the code, in this job we have developed a tool useful to optimize this kind
of structures, however this code can be used also to optimize bigger airplanes1 such
as cargo, transport and so on. In particular, the model used to evaluate the stress
distribution, or rather the thin-walled beams theory, should be more suitable for
this purpose, since in a wing box structure the ratio between the thickness and the
dimensions of cross-section is lower than in the single spar solution analyzed in this
thesis. Moreover, again concerning the code, this can be improved including a model
for materials which provide to analyze the composite laminates and also the second
order effects such as warping or shear lag. Such improvements could lead to a solution
closer to reality than the current.
Finally, as said in the previous chapters, the tool here developed represent the objec-
tive function, which can be used in an extended optimization process, where the weight
of structure and the parasite are evaluated simultaneously so that the configuration that
minimize the total drag, and hence the fuel consumption, can be founded. This topic
may be one of the most important research argument to develop in order to extend
the utilization of PrandtlPlane configuration to long range aircrafts.Several tools are

R
available to implement this analysis, an example is Mode Frontier , indeed this
1
for this purpose several modifications are required
109
9. Conclusions 9.2. Later developments
software allows to built a simple work-flow ad can be interfaced with the most common

R
R
R
codes as Matlab ,Nastran and so on. Furthermore, Mode Frontier allows
to perform several algorithm (both single and multi objective) to optimize the objective
function. Several kind of algorithms are depicted in [8]. An example of work-flow

R
elaborated in Mode Frontier is shown in Figure 9.1.
Figure 9.1: Example of wok-flow elaborated in Mode Frontier

R
110
Appendix
111
Appendix A
Material Properties
Figure A.1: Best Wing System
The following tables gives the main properties of materials used, in terms of elasticity,
strength and density. The meaning of superscripts and subscripts is defined as follows,
according to conventions shown in Figure A.1:
• l, longitudinal direction;
• t, tangential direction;
• r, radial direction;
• T, tensile (defined for strength);
• C, compressive (defined for strength);
112
A. Material Properties
Material El [GPa ] Et [GPa ] Er [GPa ] Glt [GPa ] Glr [GPa ] Grt [GPa ]
Carbon 135 9.24 — 6.28 — —

Spruce 9.9 0.42 0.77 0.6 0.63 0.03
Birch 13.9 6.95 1.08 0.945 1.02 0.236
Okumè 7.9 — — — — —
Table A.1: Elastic modules
Material ρ [Kg/m3 ] σlT [M Pa ] σtT [M Pa ] σlC [M Pa ] σtC [M Pa ] τlt [M Pa ]
Carbon 1730 1475 — 600 — 76.6

Spruce 360 65 27 35.7 3 6.7
Birch 620 114 35 56.3 6.7 13
Okumè 330 51 — 27.4 — 6.7
Table A.2: Mechanical properties
Material µlr µlt µrt µtr µrl µtl
Carbon — 0.32 — — — —
Spruce 0.37 0.47 0.44 0.24 0.04 0.02
Birch 0.426 0.451 0.692 0.426 0.043 0.024
Okumè — 51 — — — —
Table A.3: Poisson ratios
113
Appendix B
Alternative optimization process
Launch
Front wing parameters

Define initial
nz, hcruise , Vcruise Rear wing parameters
parameters
Global parameters
Calculate
geometry
Define initial
Initialize Calculate inertial
thickness
loop properties
distribution
Calculate Calculate inertial

aerodynanics loads loads
Calculate
load distrib.
Calculate inertial
properties
Calculate
stress
No
Modify No Yes Verify

Verify allowable
thickness convergence
stress constraint
Yes
Define
results
Exit
Figure B.1: Alternative optimization process
114
B. Alternative optimization process
In this this chapter an alternative procedure to design "optimized" structures

is proposed. This process, which work-flow is shown in Figure B.1, operate in a
different way concerning the stress calculation. In particular as the load distribution
are evaluated a second internal loop starts, this loop aims to find in each point of beam
system the value of thickness which it verifies the operating stress level holding the
same load distribution. When a new thickness distribution is defined the main loop
starts again to evaluate the new inertial properties, inertial loads and load distribution.
The process finishes when the thickness distribution remains constant, or rather the
differences are negligible, after two cycles of main loop.
This algorithm provides lower calculation time and results close to the one used in
this work, but has been discharged to improve the accuracy.
115
Appendix C
The Matlab
R
code

R
This chapter aims to describe how the Matlab code works. In particular it is
explained how to launch the tool and how to enter the main parameters. Moreover, it
is shown which the default settings are and how they can be modified them. Basically,
this chapter can be used as the optimizer reference manual. Finally, in order to run

this tool a Matlab R
2007 version or later is required.
C.1 Run the optimization tool

In order to launch this tool the file LAU N CH_opt.m, highlighted in Figure C.1
by the red arrow, must be run. The other sub-routines needed are in the folder opt
As the tool is launched, two dialog windows appear on the screen: the first, shown in
Figure C.2, is used to enter several geometric parameters. In particular, as explained
in subsection 5.2.1, these parameters are:
ba is the wing span;
Sref is the reference Surface and includes the entire rear wing surface plus front wing
surface excluding the fuselage intersection;
Rs is the ratio between the whole rear wing surface and front wing surface where the
fuselage intersection is excluded;
Λa is the sweep angle of the front wing at 25% of chord;
Λp is the sweep angle of the rear wing at 25% of chord;
Γa is the dihedral angle of front wing;
Γp is the dihedral angle of rear wing;
116

R
C. The Matlab code C.1. Run the optimization tool
Figure C.1: Start the tool
λa is the taper ratio of the front wing;
λp is the taper ratio of rear wing1 ;
l/b is the ratio between the horizontal distance of the two wings and wing span;
h/b is the ratio between the vertical distance of the two wings and wing span;
Df us is the fuselage diameter at the front wing intersection
As this dialog window is completed, another one appears on the screen (Figure C.3) to
define other parameters, in particular
nz that is the vertical load factor;
V that is the cruise speed;
h that is the cruise altitude;
σmax is the allowable value of stress
1
in this case the chord at root matches with the chord at center line
117

R
C. The Matlab code C.1. Run the optimization tool
Figure C.2: Dialog command window used to enter the geometric parameters
Figure C.3: Dialog command window used to enter the general parameters
118

R
C. The Matlab code C.2. The objective function
Furthermore, as well as the parameters discussed above, there are some choices to

R
do in order to customize the routine, these are included in the Matlab code using
a set of default parameters. For example, the part of the code of LAU N CH_opt.m
which launch the calculation is shown in listing C.1. In this part of code the field
scelta_struttura can be switched from 1 to 3 in order to analyze a wing box solution
instead a single spar one2 .

cd ( ’opt ’ )
% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% %%%% calcolo peso strutturale
tic
[ weight , dati_out , mchek_sing ] = peso ( in . b_a , in .S , in . Rs , ...

in . LA25_a , in . LA25_p , in . lambda_a , in . lambda_p , in . GAMMA_a , ...
in . GAMMA_p , ...
in . rap_h_b , in . rap_l_b , D_fus_a , sigma_ref , options , vec_in );
% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
plot_res ;
cd ../
elapsed_time = toc

Listing C.1: Launch calculations
C.2 The objective function

For the present work the objective function is the weight of the structures and it is
calculated by the routine peso.m. This routine is implemented as a function so that it
can be launched with several set of input parameters, furthermore this can be used as
objective function in another optimization algorithm which aims to find the optimal
data set using the parameters discussed above (see chapter D).
As shown in listingC.2, the function peso.m is composed of two main parts, in
the first part the routine defines the geometric parameters while in the second one it
launches the structural weight calculation. Moreover the first part is necessary in order
2
option 2 is not implemented
119

R
to convert the input parameters in the ones used by MDSA-2010 , which is integrated
in this code. The main parts of this routine are shown in listingC.3.

function [ weight , dati_out , mchek ]= peso ( b_a ,S , Rs , LA25_a , LA25_p , ...
lambda_a , lambda_p , GAMMA_a , GAMMA_p , rap_h_b , rap_l_b , D_fus_a , ...
Sigma_ref , options , vec_in , numero_el )
% ----- inizializzazione dati ------
D_fus_p =0;
GAMMA_a_rad = GAMMA_a * pi /180;

GAMMA_p_rad = GAMMA_p * pi /180;
scelta_struttura = options (1);

opt = options (2);
% ------------------------------------
[ Sa , Sp , Ar_a , AR_p , b_a , b_p ]= b_S ( opt , Rs ,S , b_a );
[ dati_ala_ant , dati_ala_post ]= gen_dati_conf ( D_fus_a , D_fus_p , ...

b_a , b_p , Sa , Sp , lambda_a , lambda_p , LA25_a , LA25_p , rap_h_b , rap_l_b );
dati_out . geom . ala_ant = dati_ala_ant ; % % geo ala ant
dati_out . geom . ala_post = dati_ala_post ; % % geo ala post
% ----- calcolo peso strutturale ------
Launch_calcolo_pesi ;
% ------------------------------------
weight = peso_totale ;
end

Listing C.2: Objective function

function [ dati_ala_ant , dati_ala_post ]= gen_dati_conf ( D_fus_a ...
, D_fus_p , b_a , b_p , Sa , Sp , lambda_a , lambda_p , LA25_a , ...
LA25_p , rap_h_b , rap_l_b )
120

R
semiDfus_a = D_fus_a /2;

semiDfus_p = D_fus_p /2;
bmed =( b_a * Sa + b_p * Sp )/( Sa + Sp );
h = rap_h_b * bmed ;
l = rap_l_b * bmed ;
semi_b_ant = b_a /2;

semi_b_post = b_a /2;
% parametri ala anteriore

% %%%%%% %%%% %%%% %%%% %%% %%%
% ------ definizione riferimento dell ’ ala ( si considera il punto

% al bordo d ’ attacco alla cr )
x_Wing_ant =0;
y_Wing_ant =0;
z_Wing_ant =0;
% ------------------------------------------------------------
% ----- definizione corde - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

Cta = Sa /((1+(1/ lambda_a ))*( b_a - D_fus_a )/2); % corda al tip
Cra = Cta / lambda_a ; % corda alla radice
CCla = corday ( Cra , Cta ,( semi_b_ant - semiDfus_a ) , semi_b_ant );
% ------------------------------------------------------------
[ tg_LA_le_ant , LA_le_ant , LA_le_r_ant ]= freccia_le ( LA25_a , ...

lambda_a , CCla , b_a );
dati_ala_ant (1)= Cra ;

dati_ala_ant (2)= Cta ;
dati_ala_ant (3)= CCla ;
dati_ala_ant (4)= x_Wing_ant ;
dati_ala_ant (5)= y_Wing_ant ;
dati_ala_ant (6)= z_Wing_ant ;
dati_ala_ant (7)= LA_le_r_ant ;
dati_ala_ant (8)= semi_b_ant ;
121

R
dati_ala_ant (9)= semiDfus_a ;

dati_ala_ant (10)= LA_le_ant ;
dati_ala_ant (11)= LA25_a ;
dati_ala_ant (12)= Sa ;
% parametri ala posteriore

% %%%%%% %%%% %%%% %%%% %%% %%%
% ------ definizione riferimento dell ’ ala ( si considera il punto
% di contatto anteriore delle semiali )
x_Wing_post = x_Wing_ant + l ;
y_Wing_post = y_Wing_ant +0;
z_Wing_post = z_Wing_ant + h ;
% -------------------------------------------------------------
% ----- definizione corde - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

Ctp = Sp /((1+(1/ lambda_p ))*( b_p - D_fus_p )/2); % corda al tip
Crp = Ctp / lambda_p ; % corda alla radice
CClp = corday ( Crp , Ctp ,( semi_b_post - semiDfus_p ) , semi_b_post );
% -------------------------------------------------------------
[ tg_LA_le_post , LA_le_post , LA_le_r_post ]= freccia_le ( LA25_p , ...

lambda_p , CClp , b_p );
dati_ala_post (1)= Crp ;

dati_ala_post (2)= Ctp ;
dati_ala_post (3)= CClp ;
dati_ala_post (4)= x_Wing_post ;
dati_ala_post (5)= y_Wing_post ;
dati_ala_post (6)= z_Wing_post ;
dati_ala_post (7)= LA_le_r_post ;
dati_ala_post (8)= semi_b_post ;
dati_ala_post (9)= semiDfus_p ;
dati_ala_post (10)= LA_le_post ;
dati_ala_post (11)= LA25_p ;
dati_ala_post (12)= Sp ;
end

Listing C.3: gen_conf routine
Furthermore in listing C.2,it is possible to modify the value of the fuselage diameter
at the intersection with rear wing (the default value is zero).
122

R
C. The Matlab code C.3. Structural weight
C.3 Structural weight

The computation of structural weight and optimum thickness distribution is done
by the routine Launch_calcolo_pesi launched in C.2. This routine is based upon the
MDSA-2010 code.
The first part of Launch_calcolo_pesi, which code is shown in listing C.4, aims to
gather the following additional data set:
par_disc is the size, expressed as a fraction of beam length, of the elements in which
the wing is divided. The default value is set on 0.1,this means that the size of
each element is the 10% of the beam lenght; a lower value increases the accuracy
as well as the calculation time, while a greater value has the opposite effect3 .
rho_skin is the density of material used to model the external skin. This parameter
can be defined for a wing box solution or for single spar one. Since the skin of
the ULM prototype is built using okumè, this value is set on 330Kg/m3
t_skin is the thickness used to model the external skin. For the ULM prototype the
default value is 3mm
ala this parameter allows to chose the kind of airplane, PrandtlPlane (enter 1) or
conventional(enter 2). By default it is set on 1

par_disc =0.1;
ro_aria = isasi ( h_cru )*[0;0;1;0;0;0]; % [ kg / m ^3] densità
% dell ’ aria in crociera
if scelta_struttura ==3 % % wing box
rho_skin =2700; % [ Kg / m ^3]
t_skin =0.001; % [ m ]
else
rho_skin =330; % [ Kg / m ^3] %% monolongherone
t_skin =0.003; % [ m ]
end
rad = pi /180;
ala =1; % menu ( ’ Scegli il tipo di ala : ’ , ’ chiusa ’ , ’ convenzionale ’);

Listing C.4: Launch_calcolo_pesi 1st step
The second part,shown in listing C.5, is taken from MDSA-2010 and provides the
evaluation of the airfoil coefficients and the geometry definition of spars. This part is
3
A treatment about this topic is exposed in chapter 4.7
123

R
better explained in [3]. Furthermore, the routine def _geometria includes the input
data file for the geometry of spar, wing box (if any) and ribs.

% script per la adefinizione della geometria dell ’ ala :
def_geometria ;
% calcolo coefficienti del profilo
Launch_Xfoil ;

Listing C.5: Launch_calcolo_pesi 2nd step
The third section (listing C.6) is into the iterative loop and its purpose is to evaluate
the inertial and aerodynamics loads, load distribution, and stress distribution. In this
part the following parameters are defined
iter is the maximum number of iteration;
aerodynamic loads that are the values of lift at wing root and wing tip of both
wings, they can be changed into the script distrib_portanza_generica, shown
in listing C.7;
metodo establish the criterion used to calculate the equivalent stress, and it is defined
by the function tens_ta, shown in listing C.8;
ta is the thickness of webs to set into the functiontens_ta;
E is the value of efficiency used as scale factor on lift in order to evaluate the drag
and it is implemented in distrib_resistenza_generica(see listing C.9)

% inizio ciclo
for iter =1:100
geo . nome = [{ ’ ala_ant ’} { ’ bulk ’} { ’ ala_post ’ }];

tic
iterazione = iter
% inserimento dei carichi

def_carichi ;
t_carichi = toc
% caratteristiche della sollecitazione ;
ETA_cos_direttori
124

R
save ( ’ wsp_monolongherone . mat ’ )

t_cds = toc
cd ( ’ calcolo tensioni ’ )
% %% calcolo tensioni
tens_ver ;
t_tens = toc

Listing C.6: Launch_calcolo_pesi 3rd step

% Distribuzione di portanza generica dati i valori alla radice
% ed all ’ estremità :
% ALA ANTERIORE
% disp ( ’ Carico sull ’ ’ ala anteriore : ’);
% L_r . ala_ant =655.7417;
Lz_r . ala_ant =650.5896* nz ; % = input ( ’ alla radice [ N / m ^2]: ’)
Lz_t . ala_ant =217.5021* nz ; % input ( ’ al tip [ N / m ^2]: ’)

Listing C.7: A part of function distrib_portanza_generica

% metodo = medodo di calcolo della tensione equivalente :
% 1 --> criterio di Beltrami
% 2 --> criterio di Tresca
% 3 --> criterio di Von Mises
...
n_p =10; % numero punti analizzati per ogni flangia
ta =0.003; % [ mm ]
if nargin <11
metodo =3;
end

Listing C.8: A part of function tens_ta

% Distribuzione di resistenza a partire dalla portanza e
% dall ’ efficienza :
E_ala . ala_ant =10;

E_ala . ala_post =10;
125

R
E_ala . bulk =10;
for k =1: length ( geo . nome )

nome = cell2mat ( geo . nome ( k ));
% E_ala . ala_post = L .( nome )/ R .( nome );
% for i =1:( n_cent )

Dy_ala .( nome ) = abs ( Lz_aer .( nome )/ E_ala .( nome ));
% end
end
% save d is t r ib _ re s i st e nz a _ ge n er i c a
save ( ’ distrib_resistenza_generica ’ , ’ Dy_ala ’ , ’s_a ’ , ’s_p ’ , ’s_b ’ )

Listing C.9: A part of function distrib_resistenza_generica
The fourth section of Launch_calcolo_pesi, shown in listing C.10 verifies if the

convergence occurs. In fact, after that the error margin and the allowable stress have
been imposed, the function par_conv analyzes all the elements of the stress vector,
composed of the stress distribution of each wing(listing C.11). If all point of this vector
are inside the range
σa − < σe < σa − (C.3.1)
where σa is the allowable value of stress which depends on material strength, σe is the
equivalent stress defined above and is the error margin, the function returns 1 and
the loop finishes, else par_conv returns 0 and the thickness distributions are modified
in order to re-start the loop

% %%% caclolo parametri di convergenza
% %% input
errore_max =10;
tens_ver1 =[ Sigma_ant , Sigma_bulk , Sigma_post ];
% %% calcolo parametro convergenza
Sig_ref_vett = zeros (1 , numel ( tens_ver1 ))+ Sigma_ref ;
[ count_glob ]= par_conv ( tens_ver1 , Sig_ref_vett , ...

errore_max , iterazione );
126

R
if count_glob ==1
break
end
% %% ricalcolo spessori
defsp ;
cd ../
end

Listing C.10: Launch_calcolo_pesi 4th step

function [ global_count , count ]= par_conv ( t_ver1 , t_ver , e_max , iter )
if nargin <4
iter =2;
end
if nargin <3
e_max =5;
end
% %% controllo errore minimo

if e_max <0
error ( ’ e_max deve essre maggiore di 0 ’ )
end
% ----------------------------------------
if iter ==1
global_count =0;
count = ’ non ␣ definito ’;
else
% %% controllo congruenza elementi

if numel ( t_ver1 )~= numel ( t_ver )
error ( ’i parametri t_ver1 e t_ver devono avere la ...
stessa lunghezza ’ )
end
127

R
% %% calcolo errore e definizione counts parziali

for cns =1: numel ( t_ver1 )
err = abs ( t_ver1 ( cns ) - t_ver ( cns ))/ t_ver ( cns )*100;
if err < e_max
count ( cns )=1;
else
count ( cns )=0;
end
end
% %% definizione count globale

count_nnz = nnz ( count );
if count_nnz == numel ( t_ver1 )
global_count =1;
else
global_count =0;
end
end
end

Listing C.11: The functionpar_conv
The thickness modification is the task of the script called def sp, shown in listing
C.12. This function works as explained in section 5.8 and the only parameter to set
is delta_sp, that represents the increment or decrement to assign to the thickness
distribution point by point. The default value is 0.2mm, a lower value improves the
accuracy but leads to a longer calculation time, a greater value does the opposite.

% %% definizione tolleranza
toll = Sigma_ref * errore_max /100;
% %% definizione incremento spessore
delta_sp =0.0002; % [ m ]
% -------------------------------------------------------------
% -------------------------------------------------------------
for i =1: numel ( s1 )
if Sigma_ant ( i ) >( Sigma_ref + toll )
t_ant ( i )= t_ant ( i )+ delta_sp ;
128

R
elseif Sigma_ant ( i ) <( Sigma_ref - toll )
t_ant ( i )= t_ant ( i ) - delta_sp ;

else
t_ant ( i )= t_ant ( i );
end
[ Ixa ( i ) , Iya ( i ) , Iza ( i ) , Aa ( i )]= CALCOLO_INERZIA ( ha ( i ) , la ( i ) ...

, t_ant ( i ))
end
tma_mm = t_ant *1000; % mm

Listing C.12: A part of function def sp
Finally the calculation of spar weight is done by the function weight1long shown in
listing C.13. Then, the last part of Launch_calcolo_pesi is used to save the output
data set. An example is shown in listing C.14.

function [ Peso , Area , Volume ]= Weight_1Long (s ,A , rho )
if nargin <3
rho =1;
warning ( ’ peso per unita di volume ’ )
end
% %% Calcolo area
% for i =1: numel ( s )
% A ( i )= l ( i )* h ( i ) -(( h ( i ) -2* t ( i ))*( l ( i ) -2* t ( i )));
% end
ss = linspace ( s (1) , s ( end ) ,10* numel ( s ));

AA = interp1 (s ,A , ss );
% %% Calcolo Volume
129

R
Volume = trapz ( ss , AA );
% %% Calcolo Peso
Peso = Volume * rho ;

Area = AA ;
end

Listing C.13: The function weight1long

% %% pesi
dati_out . peso . tot = peso_totale ;
dati_out . peso . long . ant = Peso_ant ;
dati_out . peso . long . post = Peso_post ;
dati_out . peso . long . bulk = Peso_bulk ;
dati_out . peso . cent . ant = peso_centine . ala_ant ;
dati_out . peso . cent . post = peso_centine . ala_post ;
dati_out . peso . cent . bulk = peso_centine . bulk ;
dati_out . peso . skin . ant = Peso_skin . ala_ant ;
dati_out . peso . skin . post = Peso_skin . ala_post ;
dati_out . peso . skin . bulk = Peso_skin . bulk ;
% %% spessori
dati_out . distr_spess . ant = tma_mm ;
dati_out . distr_spess . post = tmp_mm ;
dati_out . distr_spess . bulk = tmb_mm ;
dati_out . spessore_minimo = min ([ t_ant t_post tbulk ]);
% %% tensioni
dati_out . distr_tens . ant = Sigma_ant ;
dati_out . distr_tens . post = Sigma_post ;
dati_out . distr_tens . bulk = Sigma_bulk ;
dati_out . distr_tens . s_ref = Sigma_ref ;

Listing C.14: Launch_calcolo_pesi 5th step
130

R
C. The Matlab code C.4. Results layout
C.4 Results layout

The objective function peso gives two main results
1. the weight of entire structures;
2. an output structures which gather all the useful data computed.
In order to give an easier reading of results the tool returns on the command window
a matrix of weight that shows the weights of ribs, skin, spars and the sum of these
for each wing. Furthermore a plot of stress and thickness distribution on each wing
and a plot of load distributionfor each wing are given. The following figures gives
the structure of dati_out, which includes the database calculated by the optimizer,
according with the legend shown in Figure C.4.
LEGEND:
scalar
sub structure
matrix
vector
Figure C.4: Legend
131

R
discr
iterazione
carichi
spessore_minimo
distr_tens
geom distr_spess
tant_iter
DATI_OUT
tbulk_iter
cds tpost_iter
tens_ant_iter
peso
tens_post_iter
iner
tens_bulk_iter
Figure C.5: Structure of dati_out
132

R
Refers to Figure C.5 these are the meanings of the fields
iterazione is number of iterations required to obtain the convergence
spessore_minimo is the minimum value of thickness
geom contains the date related to the external geometry
cds contains the vectors of load distributionof each wing
peso contains the vectors of weight of each wing
iner contains the vectors of inertial data of each wing
discr contains the vectors of section location of each wing
carichi contains the vectors of loads of each wing
distr_tens contains the vectors of stress distribution of each wing
distr_spess contains the vectors of thickness distribution of each wing
tant_iter each row represents the vector of thickness distribution in a specific iteration
for front wing.The number of row represents the number of iteration.
tpost_iter each row represents the vector of thickness distribution in a specific

iteration for rear wing.
tbulk_iter each row represents the vector of thickness distribution in a specific

iteration for winglet.
tens_ant_iter each row represents the vector of stress distribution in a specific

iteration for front wing.The number of row represents the number of iteration.
tens_post_iter each row represents the vector of stress distribution in a specific

iteration for rear wing.
tens_bulk_iter each row represents the vector of stress distribution in a specific

iteration for winglet.
In the next paragraph are shown the various substructures defined in Figure C.5.
C.4.1 Substructure cds

This substructure contains the vectors of load distributiondistributions on each
wing and is shown in Figure C.6.
133

R
Tx
Ty
Tz
Mx
My
ant Mz
Tx
Ty
Tz
cds post
Mx
My
Mz
bulk Tx
Ty
Tz
Mx
My
Mz
Figure C.6: Substructure cds
134

R
C.4.2 Substructure geom

This substructure contains the output of the function gendaticonf and is shown in
Figure C.7.
dati_ala_ant
geom
dati_ala_post
Figure C.7: Substructure geom
C.4.3 Substructure iner

This substructure contains the vectors of inertial properties distributions on each
wing and is shown in Figure C.8.
A
ant
Ix post
Iner
bulk
Iy
Iz
Figure C.8: Substructure iner
135

R
C.4.4 Substructure peso

This substructure contains the global weight of the wing system and the one of each
component(Figure C.9), in particular the meaning of each field is explained below4 .
tot is the global weight of the wing system, hence include the weight of rib, skin and
spars of each wing;
long is the weight of the spars;
cent is the weight of the ribs;
skin is the weight of the skin panels;
ant refers to the front wing;
post refers to the rear wing;
bulk refers to the vertical winglet
tot
ant
longh post
peso
bulk
cent
skin
Figure C.9: Substructure peso
4
the same notations are used also in the following
136

R
C.4.5 Substructure discr

This substructure contains the vectors of the section locations on each wing. These
are results of partition used in order to analyze the beams, and is shown in Figure
C.10.
asc1
discr asc2
asc3
Figure C.10: Substructure discr
C.4.6 Substructure carichi

This substructure contains the vectors of load distributions which act on the beams
and is shown in Figure C.11.
portanza
ant
longheroni post
carichi
bulk
centine_skin
totale
Figure C.11: Substructure carichi
137

R
C.4.7 Substructure distr_tens

This substructure contains the vectors of stress distributions of each wing and is
shown in Figure C.12.
ant
post
distr_tens bulk
s_ref_int
s_ref
Figure C.12: Substructure distr_tens
C.4.8 Substructure distr_spess

This substructure contains the vectors of thickness distributions of each wing and
is shown in Figure C.13.
ant
distr_spess bulk
post
Figure C.13: Substructure distr_spess
138

R
C. The Matlab code C.5. Default values
C.5 Default values

The table C.1 gives the default value of each parameter and the file where they are
located
Parameter Default value File
ba [m] 8 LAUNCH_opt.m5
Sref [m2 ] 13.26 LAUNCH_opt.m
Sp/Sa 1.66 LAUNCH_opt.m
Λa [deg] 24.27 LAUNCH_opt.m
Λp [deg] −11.23 LAUNCH_opt.m
λa 0.59 LAUNCH_opt.m
λp 0.38 LAUNCH_opt.m
Γa [deg] 7 LAUNCH_opt.m
Γp [deg] 0 LAUNCH_opt.m
l/b 0.547 LAUNCH_opt.m
h/b 0.25 LAUNCH_opt.m
Df us [m] 1.24 LAUNCH_opt.m
nz 4 LAUNCH_opt.m
V [m/s] 69 LAUNCH_opt.m
h [m] 1000 LAUNCH_opt.m
σmax [M P a] 35 LAUNCH_opt.m
Load data
Lz_root.ala_ant [N/m2 ] 650.5896 distrib_portanza_generica.m

2
Lz_tip.ala_ant [N/m ] 217.5021 distrib_portanza_generica.m
2
Lz_root.ala_post [N/m ] 344.7744 distrib_portanza_generica.m
2
Lz_tip.ala_post [N/m ] 11.2189 distrib_portanza_generica.m
E 10 distrib_resistenza_generica.m
Ribs data
larghezza_centina [m] 0.012 dati_centina.m

spessore_centina_root.ala_ant [m] 0.0043 dati_centina.m
spessore_centina_tip.ala_ant [m] 0.0043 dati_centina.m
spessore_centina_root.ala_post [m] 0.0043 dati_centina.m
spessore_centina_tip.ala_post [m] 0.0043 dati_centina.m
spessore_centina_root.bulk [m] 0.0043 dati_centina.m

spessore_centina_tip.bulk [m] 0.0043 dati_centina.m
Skin data
rho_skin [Kg/m3 ] 330 Launch_calcolo_pesi.m

t_skin [m] 0.003 Launch_calcolo_pesi.m
Spars initial data
continued on next page
139

R
C. The Matlab code C.5. Default values
Parameter Default value File
spessore_anima [m] 0.003 dati_longherone.m

dati_ingr_longh.ala_ant [m] 0.081 dati_longherone.m
spessore_f langia.sup.root [m] 0.0264 dati_longherone.m
spessore_f langia.inf.root [m] 0.024 dati_longherone.m
spessore_f langia.sup.tip [m] 0.008 dati_longherone.m
spessore_f langia.inf.tip [m] 0.008 dati_longherone.m
3
ro_longh [Kg/m ] 495 dati_longherone.m
3
ro_paratia [Kg/m ] 495 dati_longherone.m
E [P a] 40e9 dati_longherone.m
ni 0.33 dati_longherone.m
Tension data
ta [m] 0.003 dati_tens_ta.m6

metodo 3 dati_tens_ta.m
Others data
scelta_struttura_ala 1 LAUNCH_opt.m
ala 1 Launch_calcolo_pesi.m
iter 100 Launch_calcolo_pesi.m
e_max 10 Launch_calcolo_pesi.m
delta_sp [m] 0.0002 Launch_calcolo_pesi.m
Table C.1: Default parameters
5
can be modified also through the dialog window
6
WARNING!:to change this parameter if web thickness changes
140
Appendix D
Sensitivity Analysis
As said in the previous chapters, this optimization process has been conceived to
find an optimum set of data which verifies just the static stress constraints, starting
from a given set of geometrical parameters. These parameters come from a previous
optimization (discussed in [4]) which aimed to define the best solution i.e the one with
minimum drag, takeing aerodynamics and flight mechanics into account. However, the
drag is also affected by the weight of airplane, hence a configuration that minimize the
parasite drag could lead to realize a too much heavy structure penalizing the global
drag. In this respect a multi objective optimization process is advisable in order to
minimize both parasite drag and structural weight and hence to find the parameters
which give the best trade-off solution.
Thus, in order to improve the efficiency of such process a sensitivity analysis is
required to determine how the solution, in this case the structural weight, is affected
by input parameters.
D.1 Analysis process

The process consist in running the objective function peso(described in chapter C)
using several configurations close to the starting one as input. Each configuration is
obtained changing the input parameters one by one assuming two value defined as plus
and minus 10% of the reference value, as shown in Table D.1.
As the analysis is complete, the graph shown in Figure D.1 is available. This shows
the variation of structural weight respect to the variation of each input parameter. In
order to better understand these trends, in Figure D.2 gives the relative variation of
1
the minmum and maximum value are arbitrary defined
141
D. Sensitivity Analysis D.1. Analysis process
Parameter Min Ref Max
General
ba [m] 7.2 8 8.8

Sref [m2 ] 12.82 14.37 15.67
Sp /Sa 1.21 1.35 1.48
h/b 0.22 0.25 0.27
l/b 0.66 0.74 0.81
Front wing
Γa [deg] 6.3 7 7.7

Λ25
a [deg] 20.88 23.2 25.52
Ct/Cr 0.45 0.50 0.55
Rear wing
Γa [deg]1 −3 0 3
Λ25
a [deg] −8.37 −9.3 −10.23
Ct/Cr 0.33 0.57 0.40
Table D.1: Parameters used in sensitive analysis
weight against the relative variation of each parameter, defined as follows
W i − W0
∆W = (D.1.1)
W0
X i − X0
∆X = (D.1.2)
X0
(D.1.3)
where W is total structural weight, X is the generic parameter and 0 is the subscript
related to the reference configuration. Finally, section D.2 gives the command window
returned by this analysis.
As shown in Figure D.2, the parameters which mainly affect the weight around the
reference configuration are the wing area, wingspan, sweep angles and l/b ratio. In
order to implement an extended optimization process (as said above) these parameters
can be used as input variables, while the others can be used as constants.
142
55 55 48 49
50 50 47.5 48
45 45 47 47
weight [Kg]
weight [Kg]
weight [Kg]
weight [Kg]
D. Sensitivity Analysis
40 40 46.5 46
7 8 9 12 14 16 1.4 1.6 20 25 30
’ba’ ’S’ ’Rs’ ’LA25a’
48 47.5 48 47.2
47.5 47.5
47 47.15
143
47 47
weight [Kg]
weight [Kg]
weight [Kg]
weight [Kg]
46.5 46.5 46.5 47.1

−12 −10 −8 0.4 0.6 0.8 0.350.40.45 6 7 8
’LA25p’ ’lambdaa’ ’lambdap’ ’GAMMAa’
47.25 47.25 55
47.2 47.2 50
47.15 47.15 45
weight [Kg]
weight [Kg]
weight [Kg]
47.1 47.1 40
−5 0 5 0.2 0.3 0.4 0.7 0.8 0.9
Figure D.1: Structural weight variation against parameters variation

’GAMMAp’ ’raphb’ ’raplb’
D.1. Analysis process
0.15 0.15 0.15 0.15
0.1 0.1 0.1 0.1
0.05 0.05 0.05 0.05
0 0 0 0
−0.05 −0.05 −0.05 −0.05
(W−W0)/W0
(W−W0)/W0
(W−W0)/W0
(W−W0)/W0
−0.1 −0.1 −0.1 −0.1
D. Sensitivity Analysis
−0.1 −0.05 0 0.05 0.1 0.15 −0.1 −0.05 0 0.05 0.1 0.15 −0.1 −0.05 0 0.05 0.1 0.15 −0.1 −0.05 0 0.05 0.1 0.15
(’ba’−’ba’0)/’ba’0 (’S’−’S’0)/’S’0 (’Rs’−’Rs’0)/’Rs’0 (’LA25a’−’LA25a’0)/’LA25a’0
0.15 0.15 0.15 0.15
0.1 0.1 0.1 0.1
0.05 0.05 0.05 0.05
0 0 0 0
144
−0.05 −0.05 −0.05 −0.05
(W−W0)/W0
(W−W0)/W0
(W−W0)/W0
(W−W0)/W0
−0.1 −0.1 −0.1 −0.1
−0.1 −0.05 0 0.05 0.1 0.15 −0.1 −0.05 0 0.05 0.1 0.15 −0.1 −0.05 0 0.05 0.1 0.15 −0.1 −0.05 0 0.05 0.1 0.15
(’LA25p’−’LA25p’0)/’LA25p’0 (’lambdaa’−’lambdaa’0)/’lambdaa’0 (’lambdap’−’lambdap’0)/’lambdap’0 (’GAMMAa’−’GAMMAa’0)/’GAMMAa’0
0.15 0.15 0.15
0.1 0.1 0.1
0.05 0.05 0.05
0 0 0
−0.05 −0.05 −0.05
(W−W0)/W0
(W−W0)/W0
(W−W0)/W0
−0.1 −0.1 −0.1
−0.1 −0.05 0 0.05 0.1 0.15 −0.1 −0.05 0 0.05 0.1 0.15 −0.1 −0.05 0 0.05 0.1 0.15
(’GAMMAp’−’GAMMAp’0)/’GAMMAp’0 (’raphb’−’raphb’0)/’raphb’0 (’raplb’−’raplb’0)/’raplb’0
Figure D.2: Structural weight relative variation against parameters relative variation
D.1. Analysis process
D. Sensitivity Analysis D.2. Command window
D.2 Command window

In the listings shown below, i indicates the parameters, the field incremento indicates
the increment related to the parameter, in particular 1 means minimum , 2 means
maximum and 0 is related to the the reference parameters. Moreover, iter indicates
the number of iteration needed and tmin is related to the minimum value of thickness.
Finally the others field gives the weight (in Kg) of component and the total on as
shown.
dati analisi di sensibilita =
i incremento peso_tot(Kg) long_ant long_post
1 0 0 47.13288 3.52154 7.56887
2 1.00000 1.00000 44.27420 2.69297 4.97740
3 1.00000 2.00000 52.51465 4.56915 11.20932
4 2.00000 1.00000 44.19675 3.61152 8.17000
5 2.00000 2.00000 50.53480 3.47834 7.10503
6 3.00000 1.00000 46.51620 3.62624 7.51409
7 3.00000 2.00000 47.73574 3.47167 7.59506
8 4.00000 1.00000 48.12617 3.64933 7.83953
9 4.00000 2.00000 46.17509 3.58633 7.22295
10 5.00000 1.00000 47.62842 3.53735 7.74476
11 5.00000 2.00000 46.62079 3.52879 7.36564
12 6.00000 1.00000 46.95004 3.50884 7.60216
13 6.00000 2.00000 47.29989 3.58568 7.51387
14 7.00000 1.00000 47.50488 3.54261 7.77214
15 7.00000 2.00000 46.80370 3.52681 7.37878
16 8.00000 1.00000 47.11936 3.52310 7.53052
17 8.00000 2.00000 47.14926 3.51390 7.60216
18 9.00000 1.00000 47.13025 3.38982 7.76010
19 9.00000 2.00000 47.22072 3.69672 7.41300
20 10.00000 1.00000 47.10849 3.48136 7.66439
21 10.00000 2.00000 47.22984 3.56830 7.49578
22 11.00000 1.00000 43.41074 3.76851 5.76256
23 11.00000 2.00000 51.25340 3.54920 9.17254
long_bulk cent_ant cent_post cent_bulk skin_ant

1 6.99662 1.50525 9.78253 0.61226 6.18259
2 4.85402 1.93703 12.07720 0.77273 6.18259
3 9.65546 1.20325 8.08474 0.49722 6.18259
4 7.34497 1.21925 7.92385 0.49593 5.56433
5 6.69199 1.82135 11.83687 0.74083 6.80085
6 6.89092 1.69432 8.91917 0.62635 6.55941
7 7.07886 1.34614 10.58570 0.60100 5.84672
8 7.59146 1.50525 9.78253 0.61226 6.18259
9 6.31995 1.50525 9.78253 0.61226 6.18259
10 7.30046 1.50525 9.78253 0.61226 6.18259
11 6.68051 1.50525 9.78253 0.61226 6.18259
12 6.95879 1.52447 9.78253 0.56877 6.18259
13 6.99532 1.48951 9.78253 0.65494 6.18259
14 7.18954 1.50525 9.92118 0.57002 6.18259
15 6.79614 1.50525 9.66237 0.65433 6.18259
16 7.01989 1.50525 9.78253 0.61226 6.18259
17 6.98735 1.50525 9.78253 0.61226 6.18259
18 6.93447 1.50525 9.78253 0.61226 6.18259
19 7.06514 1.50525 9.78253 0.61226 6.18259
145
D. Sensitivity Analysis D.2. Command window
20 6.91688 1.50525 9.78253 0.61226 6.18259

21 7.11991 1.50525 9.78253 0.61226 6.18259
22 4.83381 1.50525 9.78253 0.61226 6.18259
23 9.48581 1.50525 9.78253 0.61226 6.18259
skin_post skin_bulk iter tmin

1 7.02180 3.94142 4.00000 3.90000
2 6.87768 3.90259 4.00000 3.20000
3 7.13972 3.97320 5.00000 4.90000
4 6.31962 3.54728 5.00000 4.30000
5 7.72398 4.33556 4.00000 3.80000
6 6.70479 3.98091 4.00000 4.40000
7 7.30437 3.90623 4.00000 3.60000
8 7.02180 3.94142 4.00000 4.30000
9 7.02180 3.94142 4.00000 4.20000
10 7.02180 3.94142 4.00000 3.90000
11 7.02180 3.94142 4.00000 4.00000
12 7.02180 3.80009 4.00000 3.90000
13 7.02180 4.07364 4.00000 4.20000
14 7.02207 3.79948 4.00000 4.10000
15 7.02154 4.07590 4.00000 3.90000
16 7.02180 3.94142 4.00000 4.00000
17 7.02180 3.94142 4.00000 3.80000
18 7.02180 3.94142 4.00000 3.30000
19 7.02180 3.94142 4.00000 4.70000
20 7.02180 3.94142 4.00000 3.60000
21 7.02180 3.94142 4.00000 4.20000
22 7.02180 3.94142 4.00000 5.10000
23 7.02180 3.94142 5.00000 3.40000
146
Bibliography
[1] A. Frediani. Best wing sistem. Innovative Configurations and Advanced Concepts
for Future Civil Transport Aircraft, 2005.
[2] R. Tavella A. Clemente. Analisi agli elementi finiti di ala a box. Master’s thesis,
University of Pisa, 2010.
[3] L. Ruggeri. Modello parametrico per il dimensionamento strutturale di ali a sbalzo

e del tipo "boxwing". Master’s thesis, University of Pisa, 2010.
[4] S. Giuffrida V. Cipolla. Utilizzo di codici a pannelli nel progetto preliminare di

velivoli prandtlplane ultraleggeri; applicazione a nuove configurazioni. Master’s
thesis, University of Pisa, 2006.
[5] T.H.G. Megson. Aircraft Structures for engineering students. Butterworth Heine-
mann, third edition, 1999.
[6] R.M. Jones. MECHANICS OF COMPOSITE MATERIALS. Taylor & Francis,

second edition, 1999.
[7] Cranfield College of Aeronautics. Introduction to mechanics of frp composite

materials.
[8] E. Rizzo. Optimization Methods Applied to a innovative, non conventional aircraft

configuration. PhD thesis, University of Pisa, 2009.
[9] IDS S.p.A. ASD user’s manual, 2006.
[10] S. Marzano A. Sollazzo. Elementi di Meccanica dei Continui e Resistenza dei

Rateriali. Tipografia Sociale Torinese, 1988.
[11] MSC.Patran Reference Manual.
147

Structural Design of A Ulm Prandtlplane Wing System: Universit A Di Pisa

Uploaded by

Copyright:

Available Formats

Structural Design of A Ulm Prandtlplane Wing System: Universit A Di Pisa

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Structural Design of A Ulm Prandtlplane Wing System: Universit A Di Pisa

Uploaded by

Copyright:

Available Formats

Universitá di Pisa

Tesi di Laurea Specialistica

Structural design of a ULM

Anno Accademico 2010–2011

III F.E.M. Analysis 66

A Material Properties 112

B Alternative optimization process 114

D Sensitivity Analysis 141

1.1 Best Wing System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

2.1 Final model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

3.1 wing split layout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

4.1 Ribs model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

4.4 Section parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

5.1 Design work-flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

6.1 Results:distribution of load characteristics . . . . . . . . . . . . . . . . 58

7.1 Starting model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

7.15 Winglet full connection . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

8.1 Failure index: legend . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82

A.1 Best Wing System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112

B.1 Alternative optimization process . . . . . . . . . . . . . . . . . . . . . . 114

C.1 Start the tool . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117

D.1 Structural weight variation against parameters variation . . . . . . . . 143

2.1 RAI-V.EL: Load factors . . . . . . . . . . . . . . . . . . . . . . . . . . 8

4.1 Calculation time results . . . . . . . . . . . . . . . . . . . . . . . . . . 44

6.1 Flange spar thickness (front wing) . . . . . . . . . . . . . . . . . . . . . 63

8.1 Weights . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101

A.1 Elastic modules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113

C.1 Default parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140

D.1 Parameters used in sensitive analysis . . . . . . . . . . . . . . . . . . . 142

C.1 Launch calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119

1.1 Best Wing System

Figure 1.1: Best Wing System

Figure 1.2: BWS lift distribution

Di = D11 + D22 + D12 + D21 (1.1.1)

Figure 1.3: Effect of h/b ratio on induced drag

1.2 The PrandtlPlane

Figure 1.4: Prandtl Plane Configuration

This kind of configuration is mainly suitable for commercial aircraft, in which

1.3 Ultralight Aircraft

• reduction of induced drag with a consequent fuel consumption;

• accuracy of longitudinal control system;

• smooth stall behavior;

1.4 The purpose of the present thesis

This chapter aims at depicting the prototype of a PrandtlPlane ultralight aircraft

CS-VLA BCAR S RAI-V.EL BFU

MTOW[Kg] 750 450 600 450

Figure 2.1: Final model

Figure 2.2: Fuselage external dimensions

(a) Top view

(b) Front view

Figure 2.3: Wing system

Weng [HP ] 100

2.3 Wing Structures

(a) Spar connection

(b) Karman connection

Figure 2.4: Wing-karman connection

Wing spar Height[m] Width[m]

Front @ root 0.1217 0.09

Figure 2.5: Ribs layout