5 4
2 4
1
2
5
Structural
Design
Manual
Purpose To list and homogenise the calculation methods and the
allowable values for the composite materials used at the
Aerospatiale Design Office.
This document belongs to AEROSPATIALE and cannot be given to third parties and/or be copied without
prior authorisation from AEROSPATIALE and its contents cannot be disclosed.
© AEROSPATIALE  1999
Composite stress manual
Foreword
SUMMARY OF CHAPTERS
SANDWICH
Effect of normal load Ny
N 4.2.1 page
1/2 number
4.2.1 . Effect of normal load Ny
Assuming that all layers are in a pure tension or compression condition, a
normal load Ny applied at the neutral line results in a constant elongation over
the whole crosssection. This elongation may be formulated as follows:
reference Ny
n3 ε =
of b (EMi ei + EMc ec + Ems es )
relation
This elongation this unduces:
 in the lower skin, a stress σi = Emi ε,
 in the core, a stress σc = Emc ε,
 in the upper skin, a stress σs = Ems ε.
The equivalent membrane modulus of the sandwich beam may be determined
by the relationship m14.
Ny
n4 ε ≈
b (EMi e i + Ems e s )
σs
Ny
Ems es Y
σc
Emc ec
Emi ei
b σi ε
SANDWICH
Example
N5 page
1/7 number
5 . EXAMPLE
Let a 10 mm wide sandwich beam be defined by the following stacking
sequence:
 an upper skin (carbon layers) of thickness es = 1.04 mm and of longitudinal
2
elasticity Es = 6000 daN/mm ,
 a core (honeycomb) of thickness ec = 10 mm and of longitudinal elasticity
2
modulus Ec = 15 daN/mm ,
 a lower skin (carbon cloths) of thickness ei = 0.9 mm and of longitudinal
2
elasticity modulus Ei = 4500 daN/mm .
We shall assume that the beam is subjected to the following two loads and
moment:
 Ny = 800 daN,
 Mx = 2000 daN mm,
Z
Tz = 250 daN
 Tz = 250 daN.
Mx = 2000 daN mm
1,04 Y
10
Ny = 800 daN
0,9
10
X
reference
1 step: to determine elongation ε induced by normal load Ny.
st
of
relation {n3}
ε= 800 = 7612 µd
10 ( 4500 0.9 + 15 10 + 6000 1.04 )
Z
ε = 7612 µd
X
DETAILED SUMMARY
N . SANDWICH  MEMBRANE/BENDING/SHEAR/ANALYSIS
1 . Notations
2 . Specificity
3 . Construction principle
4 . Design principle
4.1 . Sandwich plate
4.2 . Sandwich beam
4.2.1 . Effect of a normal load Ny
4.2.2 . Effect of a shear load Tx
4.2.3 . Effect of a shear load Tz  Honeycomb shear
4.2.4 . Effect of a bending moment Mx
4.2.5 . Effect of a bending moment Mz
4.2.6 . Equivalent properties
5 . Example
S . BONDED JOINTS
1 . Notations
2 . Bonded single lap joint
2.1 . Elastic behavior of materials and adhesive
2.1.1 . Highly flexible adhesive
2.1.2 . General case (without cleavage effect)
2.1.3 . General case (with cleavage effect)
2.1.4 . Scarf joint
2.2 . Elasticplastic behavior of adhesive and elastic behavior of materials
3 . Bonded double lap joint
4 . Bonded stepped joint
5 . Software
6 . Examples
T . BONDED REPAIRS
1 . Notations
2 . Introduction
3 . Analysis method
3.1 . Analytical method
3.2 . Digital method
4 . Example
U . BOLTED REPAIRS
1 . Notations
2 . Stiffness of fasteners
2.1 . Fastener in single shear
2.2 . Fastener in double shear
3 . Assumptions
4 . Geometrical characteristics
5 . Mechanical properties
© AEROSPATIALE  1999 MTS 006 Iss. A
Composite stress manual
V . THERMAL CALCULATIONS
1 . Notations
2 . Introduction
3 . Hooke  Duhamel law
4 . Behavior of unidirectional fibre
5 . Behavior of a free monolithic plate
5.1 . Calculation method
5.2 . Residual curing stresses
5.3 . Equivalent expansion coefficients
6 . Theory of the bimetallic strip
6.1 . Determining stresses of thermal origin
6.2 . Study of the link between two parts
6.2.1 . Bolted or riveted joints
6.2.1.1 . Force F taken by one fastener
6.2.1.2 . Force F taken by two fasteners
6.2.1.3 . Force F taken by three fasteners
6.2.1.4 . Force F taken by four or more fasteners
6.2.2 . Bonded joints
7 . Influence of temperature on aircraft structures
7.1 . General
7.2 . Temperature of ambient air
7.2.1 . Temperature envelope
7.2.2 . Variation of ambient air temperature
7.2.2.1 . Ambient temperature on ground
7.2.2.2 . Ambient temperature in flight
7.3 . Wall temperature
7.3.1 . Influence of solar radiation
7.3.1.1 . Maximum solar radiation
7.3.1.2 . Solar radiation during the day
7.3.2 . Influence of aircraft speed
7.3.3 . Temperature of structure
7.3.3.1 . Calculation method
7.3.3.2 . Thermal characteristics of the materials
7.3.3.3 . Temperatures of structure on ground
7.3.3.4 . Temperatures of structure in flight
7.4 . Recapitulative block diagram
8 . Computing softwares
9 . Examples
© AEROSPATIALE  1999 MTS 006 Iss. A
Composite stress manual
W . ENVIRONMENTAL EFFECT
1 . Temperature
2 . Aging
3 . Humidity
X . NEW TECHNOLOGIES
1 . R.T.M.
2 . Thermoplastic
2.1 . Shoft fibres
2.2 . Long fibres
3 . GlareArall
Y . STATISTICS
Z . MATERIAL PROPERTIES
1 . Prepreg unidirectional tapes
1.1 . First generation epoxy high strength carbon
1.2 . Second generation epoxy intermediate modulus carbon
1.3 . Epoxy R glass
1.4 . Bismaleimide carbon
2 . Fabrics
2.1 . Epoxy resin prepreg
2.1.1 . Carbon
2.1.2 . Glass
2.1.3 . Kevlar
2.1.4 . Hybrid
2.1.5 . Quartz polyester hybrid
2.2 . Phenolic resin prepreg
2.2.1 . Carbon
2.2.2 . Glass
2.2.3 . Kevlar
2.2.4 . Fiberglass carbon hybrid
2.2.5 . Quartz polyester hybrid
2.3 . Bismaleimide resin prepreg
2.3.1 . Carbon
2.4 . Wet layup epoxy (for repair)
2.4.1 . Carbon
2.4.2 . Glass
2.4.3 . Kevlar
2.4.4 . Fiberglass carbon hybrid
2.4.5 . Quartz polyester hybrid
3 . R.T.M.
3.1 . Epoxy resin
3.1.1 . Carbon
3.2 . Bismaleimide resin
3.3 . Phenolic resin
4 . Injection moulded thermoplastics
4.1 . Carbon
4.1.1 . PEEK
4.1.2 . PEI
4.1.3 . Polyamide
4.1.4 . PPS
4.1.5 . Polyarylamide
4.2 . Glass
© AEROSPATIALE  1999 MTS 006 Iss. A
Composite stress manual
4.2.1 . PEEK
4.2.2 . PEI
5 . Long fibre thermoplastics
5.1 . Carbon
5.1.1 . PEEK
5.1.2 . PEI
5.2 . Glass
6 . ArallGlare
7 . Metallic matrix composite materials (CMM)
8 . Adhesives
8.1 . Epoxy
8.2 . Phenolic
8.3 . Bismaleimide
8.4 . Thermoplastic
9 . Honeycomb
9.1 . Nomex
 Hexagonal cells
 OXCore
 FlexCore
9.2 . Fiberglass honeycomb
 Hexagonal cells
 OXCore
 FlexCore
9.3 . Aluminium honeycomb
10 . Foams
INTRODUCTION
General A 1
1 . INTRODUCTION  GENERAL
Each one of these two subjects shall make up one volume of the composite material
design manual.
Composite materials result from the association of at least two chemically and
geometrically different materials.
The bond between the reinforcing material and the matrix is created during the
preparation phase of the composite material and this bond shall have a fundamental
effect on the mechanical properties of the final material.
 wood,
 reinforced concrete,
In the aeronautical industry, the term "composite" is mainly associated with fibre
reinforced polymer resins.
INTRODUCTION
Composition  Fibres A 2.1
2.1 . Fibres
Their purpose is to ensure the mechanical function of the composite material. Fibres can
be of very different chemical and geometrical types, and the following properties shall be
specifically searched for:
 easy to use.
 low density.
 low cost.
They are made up of several thousand filaments (the number of filaments being indicated
by 3K: 3000 filaments, 6K: 6000 filaments or 12K: 12000 filaments) with a diameter
between 5 and 15 µm, and they are commercialised in two different forms:
 short fibres (a few centimeters long): they are felt, pylons (fabrics in which fibres are
laid out randomly) and injected short fibres,
 long fibres: they are cut during manufacture of the composite material, used as such
or woven,
INTRODUCTION
Composition  Matrices  Implementation A 2.2
3
1/3
2.2 . Matrices
 to provide a bond between the reinforcing fibres (cohesion of all fibres) while
maintaining a regular interval between them,
 resin matrices:
3 . PROCESSING METHOD
The reinforcing fibre/resin mix becomes a genuinely resistant composite material only
upon completion of the last manufacturing phase, i.e; curing of the matrix.
INTRODUCTION
Implementation A 3
2/3
This cycle is achieved following the chemical reaction between the various components 
this is the crosslinking phase.
The chemical reaction is initiated as soon as products are in contact, and it is often
accelerated by heat: the higher the temperature, the quicker and more explosive is the
reaction:
 the polyaddition reaction for epoxy resins where the weight of reactants is equal to
the weight of the compound,
 a gel level which allows getting a consistent temperature gradient throughout the
material before full gelation to limit internal stresses,
 a postcuring level which allows internal stresses to be relieved, and additional curing
for a better temperature resistance.
Note: the glass transition point is the temperature value at which all material properties
change. This important property must be measured, before and after wet aging.
INTRODUCTION
Implementation A 3
3/3
 Manufacturing facilities:
 Manufacturing processes:
• Single phase process: or "cocuring", skins are cured and bonded to the
substructure (ribs or honeycomb or stiffeners) in one single operation.
INTRODUCTION
Design A 4
 element geometry.
 element type.
Composite structures use the same types of design principles as metal ones:
INTRODUCTION
Assembly A 5
5 . ASSEMBLY
After being manufactured, the different composite (and metal) elements must be
connected to one another to allow load transfer.
The two most commonly used techniques are bonding and bolting (or riveting).
They are also difficult to control because even a sound adhesive film is a barrier to
ultrasounds.
INTRODUCTION
Advantages  Disadvantages A 6
 a weight gain which is reflected by fuel saving and, therefore, by a payload increase,
 the capacity to control stiffness and strength according to the areas of the structure,
thanks to the different types of layered materials. Composite materials naturally offer
membranebending coupling or plane coupling possibilities, which can have important
applications in the field of aeroelasticity,
 a good fatigue strength, which increases the life of aircraft parts concerned and
lightens the maintenance program considerably,
 humidity,
 temperature,
 the various aeronautical fluids such as Skydrol (hydraulic fluid), oils or solvents (MEK)
and fuels,
 radiation (ultraviolet).
On the other hand, the effects of lightning strikes (temperature rise, melting, impacts,
electronic damages) and shocks (delamination, separation, punctures) must be taken into
account in the design and justification of composite parts.
INTRODUCTION
Metal/composite material similitudes  System equilibrium A 7
7.1
1/4
Composite material and metal material structures obey the same basic rules of structural
mechanics.
On the other hand, composite material behavior laws are slightly different from those for
metals.
The purpose of this subchapter is to specify the similitudes between metal materials and
composite materials for the structural justification of structures.
Composite parts and metal parts have the same behavior with respect to:
 static equilibrium.
Whatever the type of system or element under study (metal, composite or combined), it is
subject to a set of external loads which may be of several types:
 Solid loads: distributed in the volume of the solid and of gravity (selfweight), dynamic
(inertial forces), electrical or magnetic origin.
 Areal loads: distributed over the external surface of the solid, such as normal
pressures due to a fluid or tangential loads due to friction phenomena.
INTRODUCTION
System equilibrium A 7.1
2/4
 Line loads: distributed over a line and which are, in fact, an idealized density of
surface load with a much smaller application width than length.
 Concentrated loads (P): acting in one point and which are, in fact, an idealized
density of surface load acting on a surface with smaller dimensions with respect to
the dimensions of the solid under study.
 Concentrated moments (M): acting in one point and which are, in fact, an idealized
concentrated moment.
ds
dl
P
dv
Z
To reach the equilibrium of the solid, all these external loads (C) must be equilibrated by
reactions at the bearing surfaces (R).
Σ (C) =  Σ (R)
INTRODUCTION
System equilibrium A 7.1
3/4
Let the solid be defined by its external loads and bearing surfaces:
external loads
+
bearing surfaces
a
A Z
B
Y
deformed system
external loads
+
reactions at bearing surfaces
Z
ra
Y
RA X
RB
INTRODUCTION
System equilibrium A 7.1
4/4
Σ (Mt/x) = 0
Σ (Mt/y) = 0
Σ (Mt/z) = 0
If the system is isostatic, the solving alone of these six equations allows all reactions at
the bearing surfaces to be found.
If the system is complex or if the degree of redundancy is high, only a point stress or a
matrix analysis makes it possible to find reactions at the bearing surfaces and the internal
loads they generate.
Whatever the case and whatever the type of structure (composite or metal), the three
following rules must always be applied before any stress and deformation calculation:
INTRODUCTION
Load distribution  Normal load N A 7.2.1
If a system made up of several parts which are connected together, is subject to a normal
load N, then, the load distribution within the different elements (whether metal or
composite) is as follows:
σ1
σ2
1 N
2
3
ε σ3
A.N.
we have:
å
3
N1 N2 N3 Nk N
k = 1
ε= = = = =
å å
3 3
E1 S1 E2 S 2 E3 S3 E Sk E Sk
k = 1 k k = 1 k
N Ei Si
a1 hence Ni =
å
3
E Sk
k =1 k
å
3
E Sk
k =1 k
a2 we may deduce Eeq. memb. (1 + 2 + 3) =
å
3
Sk
k =1
INTRODUCTION
Load distribution  Bending moment M A 7.2.2
A bending moment M applied to the neutral axis of the system is picked up in each layer
in proportion to its bending stiffness.
The moment M breaks down, in each layer (i), into a bending moment Mi and a normal
load Ni, so that:
εe σe
v1
1 M
2
3
σi
εi
A.N.
M Ei Si v i
a3 Ni =
å
3
E l
k =1 k k
a4 Mi =
M Ei ιi
å
3
E l
k =1 k k
å
3
E l
k =1 k k
a5 we may deduce E eq. flex. (1 + 2 + 3) =
å
3
l
k =1k
INTRODUCTION
Load distribution  Shear load T A 7.2.3
Assuming that layers 1, 2 and 3 are parallel and of the same height, a shear load T is
applied to each layer in proportion to its shear stiffness.
3
2 T
1
τm3
τm2
τm1
we have:
å
3
T1 T2 T3 T T
k = 1 k
γ= = = = =
å å
3 3
G1 S1 G 2 S 2 G 3 S3 Gk Sk Gk Sk
k = 1 k =1
T Gi Si
a6 hence Ti =
å
3
Gk Sk
k =1
å
3
Gk Sk
k =1
a7 we may deduce G eq. (1 + 2 + 3) =
å
3
Sk
k =1
INTRODUCTION
Material strength laws  Behavior laws A 7.3
∂2y M
The equation of the elastic line of a bent metal beam = becomes
∂x 2 EI
∂2y M
= for a composite structure.
∂x 2
å
n
Ek lk
k =1
Normal stress  normal load relationship: for a stressed or compressed metal beam, the
N N Ei
expression σ = becomes σi = for each layer of a composite beam.
å
n
S Ek Sk
k =1
TW
Shear stress  shear load relationship: for a sheared metal beam, τ = becomes
lb
T Ei w i
τi = for each layer of the composite beam.
å
n
Ek lk bk
k =1
INTRODUCTION
General instability A 7.4
For a beam, Euler's law which associates the general instability critical compression load
with the geometrical and mechanical properties of the beam remains valid, whatever the
material used (metal/isotropic or composite/orthotropic).
π2 E l
Fc = for metal beams,
l2
å
n
π2 E l
k = 1 k k
Fc = 2
for composite beams,
l
Regarding plates, the approach is more complex for composite materials, although bases
are identical.
The differential equation which governs composite plate instability is formulated in its
most general form:
∂4 w ∂4 w ∂4 w ∂2 w ∂2 w ∂2w
C11 + 2 (C + 2 C ) + C = N + N + 2 N
∂x 4 ∂x 2 ∂y 2 ∂y 4 ∂x 2 ∂y 2 ∂x ∂y
12 33 22 x y xy
where C11, C12, C33 and C22 are the temps of the matrix (Cij) binding the rotation tensor
and the bending load tensor (see chapter D).
E e3 æ ∂4 w ∂4 w ∂4 w ö ∂2w ∂2 w ∂2w
ç + + ÷ = N + N + 2 N
12 (1 − ν2 ) è ∂x 4 ∂x 2 ∂y 2 ∂y 4 ø ∂x 2 ∂y 2 ∂x ∂y
x y xy
INTRODUCTION
Metal/composite material differences A 8
1/3
These differences are actually covered by the composite material manual. A few
examples are given below:
y
F/S l
Isotropic material F F
2 1, 2, 3
1
3
x ∆l/l
0
Properties are independent from the
coordinate system direction
y
F/S l
Anisotropic material F F
1
2
2
1
3 3
x ∆l/l
0
Properties depend on the coordinate
system direction
INTRODUCTION
Metal/composite material differences A 8
2/3
 Failure criteria
 Effect of holes
Sizing of holes in composite materials not only takes into account the net cross
section coefficient (as for metal materials) due to material removal, but also a
decrease of the intrinsic material strength.
 Effect of bearing
 Damage tolerance
Corrosion and fatigue are the overriding factors of the limited life of metal structures.
Metal fatigue is controlled by the number of cycles required, on the one hand, to
initiate a crack and, on the other hand, bring it to its critical length (growth phase).
Influent factors of this phenomena are stress concentrations and tension loads.
INTRODUCTION
Différences composite/métal A 8
3/3
As a general rule, fatigue is not a design factor for composite elements of civil aircraft
with thin thicknesses and no structural irregularities. More specifically, mechanical
properties are such that static design requirements naturally "cover" fatigue design
requirements. Wohler curves are relatively flat and damaging loads are of the
compression type (R =  1).
Metal materials have an elastic range and a plastic range, in their behavior, which
lead to breaking, breaking occurs in carbon composite materials without plasticizing.
F/S l F/S l
F F F F
breaking
breaking
plastic zone
∆l/l ∆l/l
0 Plastic material 0 Brittle material
(metal) (composite)
INTRODUCTION
References A
BARRAU  LAROZE, Design of composite material structures, 1987
1 . NOTATIONS
2.1
MONOLITHIC PLATE  MEMBRANE
Definitions  Homogeneity  Isotropy  Coupling C 2.2.1
2.2.2
2 . GENERAL DEFINITIONS
 A material is socalled homogeneous when its properties are independent from the point
considered.
In the case of an orthotropic material, there is a “plane coupling” if the loading axis is not
coincident with one of its axes of orthotropy. In that case, normal loading (σ) generates
shear (γ) and shear loading (τ) generates elongation (ε).
N1 x N1
The laminate must be such that each layer has an identical symmetrical layer with relation
to the neutral plane.
3 . DESIGN METHOD
The design method for a flat plate consists in assessing stresses in each ply and in
determining the corresponding Hill’s criterion (see § G.3).
Let’s assume that all plies are made up of the same material, and that the laminate is
provided with the mirror symmetry property.
That is to say the central plane of the laminate (for example: (0°/45°/135°/90°) s =
(0°/45°/135°/90°/90°/135°/45°/0°). This property implies that there is no coupling between
the membrane effects and the bending effects.
Which means that the membrane flux tensor (Nx, Ny, Nxy) induces εx, εy, and γxy type
elongations only and that, on the other hand, the moment flux tensor (Mx, My, Mxy) induces
χx, χy and χxy type rotations only.
In other words, in the case of a laminate with the mirror symmetry property, the
relationship which binds loading and elongation may be formulated as follows:
Nx εx
Ny Aij 0 εy
Nxy γxy
=
Mx χx
My 0 Cij χy
Mxy χxy
A laminate (as well as the sign convention for membrane type load fluxes) may be
represented as follows:
z
y y
Ny > 0
Nxy > 0
x Nx > 0
3
θ2
θ1
x
1
θ3
1st step: Design of the stiffness matrix for the unidirectional layer in its own coordinate
system (l, t). This matrix shall be called (Ql, t).
El ν tl E l
σl 0 εl
1 − νlt ν tl 1 − ν lt ν tl
ν lt E t Et
σt = 0 εt
1 − ν lt ν tl 1 − ν lt ν tl
2nd step: Design of the stiffness matrix for the unidirectional layer in direction θ in the
reference coordinate system (x, y). This matrix shall be called (Qx, y,θ).
y
l
t
with:
Matrix (Tθ) corresponding to the basic transformation matrix for stress condition.
Matrix (T'θ) corresponding to the basic transformation matrix for elongation condition.
t y
ds
σt
l
τxy
τtl σx
θ
x
τyx
i
σy
σt
τtl
t y τlt
σl
σy
l
τyx σl
τlt
x
τyx
i σy
Expressions from which the matrix (Tθ) terms are easily taken.
Remark: the stiffness matrix (Qx, y, θ) also allows determination of the mechanical
properties of the unidirectional layer in direction θ in the reference coordinate
system (o, x, y). For the unidirectional layer, we have:
ν xy (θ) 1 µ yx
εy = − σy
E x (θ) Ey (θ) G xy (θ)
η x (θ) µ y (θ) 1
γxy τxy
E x (θ) E y (θ) G xy (θ)
where:
1
Ex(θ) =
c4
s 4
æ 1 ν ö
+ + c2 s2 ç − 2 tl ÷
El Et è Glt Et ø
1
Ey(θ) =
s4
c 4
æ 1 ν ö
+ + c2 s2 ç − 2 tl ÷
El Et è Glt Et ø
1
Gxy(θ) =
( )
2
æ 1 1 ν tl ö c 2 − s2
4c s ç +
2 2
+2 ÷ +
è El E t Et ø Glt
ν yx (θ) ν tl 4 æ1 1ö
E y (θ)
=
Et
( )
c + s4 − c 2 s 2 ç +
1
− ÷
è El E t Glt ø
E x (θ)
νxy(θ) = νyx(θ)
E y (θ)
3rd step: Knowing the stiffness matrix of each layer (Qx, y, θ) with relation to the reference
coordinate system (x, y), the laminate stiffness matrix can be calculated in this same
coordinate system: (Rx, y).
å å
n n
(Q x, y, θk ) ep (Q x, y, θk )
k =1 k =1
c3 (Rx, y) = or (Rx, y) =
n e
4th step: Determination of the laminate elongation tensor in the reference coordinate
system.
1
c4 (εx, y) = x (Rx, y)1 x (Nx, y)
e
εx Nx Nx εx
1
εy = (Rx, y)1 Ny or Ny = (A) εy
e
γ xy Nxy Nxy γ xy
where (A) is the laminate membrane stiffness matrix: (A) = e x (Rx, y).
Matrix (A) is the stiffness matrix which binds the stress flux tensor (N) with the elongation
tensor (ε).
Nx A 11 A 12 A 13 εx
Ny = A 21 A 22 A 23 x εy
Nxy A 31 A 32 A 33 γ xy
where
c6 Aij = å
n
k = 1
(Ε k
ij (z k − z k )
− 1)
ply No. k
thickness
ply No. 1
with
c ≡ cos(θ) where θ is the fibre direction in the reference coordinate system (o, x, y)
s ≡ sin(θ) where θ is the fibre direction in the reference coordinate system (o, x, y)
El
Εl =
1 − ν tl νlt
Et
Εt =
1 − ν tl νlt
σ lθ ε lθ
σ tθ = (Ql, t) ε tθ
τ lt θ γ lt θ
7th step: Assessment of Hill’s criterion in each fibre direction. Refer to chapter G (failure
criteria).
Equivalent membrane young's moduli are directly derived from the laminate stiffness
matrix (A):
1 ν yxmemb. equi.
− x
E xxmemb. equi. E yy memb. equi.
1 ν xymemb . equi. 1
c9 (A)1 = − x
e E xxmemb. equi. E yymemb . equi.
1
x x
Gxy memb. equi.
If reference axes (o, x, y) are coincident with the axes of orthotropy of the laminate, we
obtain:
A11 A 22 − (A 12 )2
E xxmemb. equi. =
e A 22
A 11 A 22 − (A12 )2
E yymemb. equi. =
e A11
A 66
Gxymemb . equi. =
e
A12
ν xy memb. equi. =
A 22
A 21
ν yxmemb . equi. =
A 11
5 . GRAPHS
Let a laminate be made up of plies in the same material and described as follows:
 overall thickness e,
If membrane fluxes Nx, Ny and Nxy, are applied to the laminate, so that Nx2 + Ny2 + Nxy2 = 1,
the design method outlined above allows loads inside each layer to be determined and the
overall plate margin (m) to be found (see § G "Failure criteria").
Ny
Nxy
m
Let's assume that the three fluxes are multiplied by the coefficient + 1.
100
In this case, the laminate subject to this new loading (Nx', Ny', Nxy') shall have a zero
margin.
Therefore, it is possible to associate each triplet (Nx, Ny, Nxy) with a flux triplet (Nx', Ny', Nxy')
so that the margin associated with it is zero.
If this operation is repeated for the set of points so that Nx2 + Ny2 + Nxy2 = 1 (sphere S with
radius 1), then, surface S' is obtained, corresponding to the set of points with a zero
margin. This is the material failure envelope.
Ny
S'
Nx
o
Nxy
It can be represented in a twodimensional space (Nx, Ny) in the form of graphs (each
curve corresponding to the intersection S' with an equation plane Nxy = Nxyi).
Ny
plane Nxy = 0
plane Nxyi
plane Nxyn
Nx
Nxy
If this set of curves is projected onto the plane (o, Nx, Ny), a network of curves is obtained
which constitutes the breaking graph of the laminate.
Ny
Nxyi = 0
Nxyi
Nxyn
Nx
o
This graph (corresponding to a given material and a specific layup) allows the laminate
margin (Hill's criterion) to be determined graphically.
Let a laminate be subject to fluxes Nxo, Nyo and Nxyo and the breaking graph associated
with it.
 Plot the straight line D crossing point o and point A of coordinates Nxo and Nyo.
 Perpendicular to this straight line, plot the value Nxyi segment corresponding to the
graph curve Nxyi. Repeat this operation for each graph curve.
 Plot curve C.
æo C ö
 The composite plate margin is equal to 100 ç − 1÷ .
èoB ø
Ny
N
xy
i
A C
Nyo N
xy
o
c
B
o Nx
Nxo
In practice, curves are represented in stress and not in flux values. This makes it possible
to group together some laminates per layup class (for example: 3/2/2/1 ≡ 6/4/4/2 ≡
9/6/6/3).
For a given material, a set of graphs may be created giving the mechanical properties
(strength and elasticity moduli) of an orthotropic laminate described by its percentages of
plies in each direction (see drawing below).
Gxy
% to 90°
Gxy
% to 45°
%
A number of those graphs associated with carbon T300/914 layer shall be found in
chapter Z “material properties”.
6 . EXAMPLE
0°: 6 plies
45°: 4 plies
135°: 4 plies
90°: 6 plies
The purpose of this example is to search for stresses applied to each ply (0°, 45°, 135°,
90°) knowing that the laminate is globally subject to the three following load fluxes in the
reference coordinate system (x, y):
Nx = 30.83 daN/mm
Ny =  2.22 daN/mm
Nxy = 44.92 daN/mm
These load fluxes being the continuation of the example covered in chapter K (Fastener
hole).
1st step: Design of stiffness matrix (Ml, t) for the unidirectional ply with relation to its own
coordinate system (l, t).
{c1}
0 0 465
13057 163 0
0 0 465
2nd step: Assessment of stiffness matrix for each unidirectional ply with relation to the
reference coordinate system (x, y).
{c2}
−1
1 0 0 13057 163 0 1 0 0
0 0 1 0 0 465 0 0 1
0 1 0 13057 163 0 0 1 0 −1
0 0 −1 0 0 465 0 0 −1
Thus, we find:
13057 163 0
0 0 465
467 163 0
0 0 465
3rd step: By applying the mixture law, the overall laminate stiffness matrix (Rx, y) is
formulated as follows.
{c3}
1
(Rx, y) = 12 x 163 + 8 x 2998 6 x 467 + 8 x 3928 + 6 x 13057 4 x 3148 − 4 x 3148
20
4 x 3148 − 4 x 3148 4 x 3148 − 4 x 3148 12 x 465 + 8 x 3299
5628 1297 0
0 0 1598
. xE − 4
188 − 4.32 x E − 5 − 6.11 x E − 20
4th step: Determination of the laminate strain tensor in the reference coordinate system (x,
y).
{c4}
εx . xE − 4
188 − 4.32 xE − 5 − 611
. xE − 20 3083
. 2262 x E − 6
1
εy = − 4.32 xE − 5 . xE − 4
188 . xE − 20
402 − 2.22 = − 673 x E − 6
2.6
γ xy − 611
. xE − 20 . xE − 20
402 6.25 x .E − 4 44.92 10807 x E − 6
{c7}
1 0 0 2262 x E − 6 2262 x E − 6
0 0 1 10807 x E − 6 10807 x E − 6
− 1 1 0 10807 x E − 6 − 2935 x E − 6
1 − 1 0 10807 x E − 6 2935 x E − 6
0 1 0 2262 x E − 6 − 673 x E − 6
0 0 −1 10807 x E − 6 − 10807 x E − 6
6th step: With the previous results, stresses in each ply are determined.
{c8}
7th step: In each direction, the corresponding Hill’s criterion is calculated (see chapter G),
which gives the following margins for each ply:
The ply at 135° is, therefore, the most brittle ply in this loading case.
8th step: The laminate margin may be found with the breaking graph corresponding to this
material (see chapter Z).
σx = 11.86 hb
σy =  0.85 hb ≈ 0 h
τxy = 17.28 hb
+ T = 22 HB
x T = 21 HB
Y T = 18 HB
+ T = 15 HB
x T = 12 HB
Y T = 9 HB
T = 6 HB
T = 3 HB
T = 0 HB
Scale: 1 cm ↔ 3.33 hb
æo C ö æ 72 ö
Marge = 100 ç − 1÷ = 100 ç − 1÷ ≈ 41 %
èoB ø è 51 ø
J. CHAIX, 436.127/91
1 . NOTATIONS
θ: fibre orientation
2 . INTRODUCTION
In chapter C, we examined the case of a laminate provided with mirror symmetry subject
to membrane type loading. In the paragraph below, we shall examine the case of a
laminate with the same properties but, this time, subject to pure bending type loads.
By convention, we shall consider that any positive moment compresses the laminate
upper fibre.
Let’s assume that bending moment flows Mx, My and Mxy generate εx, εy and γxy type
strains.
Let’s assume also (Kirchoff) that the neutral plane is coincident with the neutral line.
3 . DESIGN METHOD
y
My > 0
w
x, y
w
u, v wo
z
x, y
uo, vo
If the displacements from a point at position Z are defined as u, v and w in the coordinate
system (x, y, z), then we may write:
∂w o
u = uo  z
∂x
∂w o
v = vo  z
∂y
w = wo
where uo, vo et wo represent displacements from the neutral plane in the coordinate
system (x, y, z).
We deduce (by deriving with respect to coordinates) the corresponding nonzero strains:
∂2 w o
d1 εx = εox  z
∂x 2
∂2 w o
εy = εoy  z
∂y 2
∂2wo
γxy = γoxy  2 z
∂x ∂y
εx z
εox
neutral plan o
2
∂ w
tg(α) =
2
∂x
x
where εox, εoy and γoxy rerepresent strains at a point located on the neutral plane and εx, εy
and γxy represent strains at any point at position z.
h
From the general expression for the bending moment: M = ò
−
2
h
2
σ z dz , we obtain the
relationship between the bending load tensor (M) and the rotation tensor (χ):
∂2wO
Mx C11 C12 C13
∂x 2
∂2wO
My = C 21 C22 C 23
∂y 2
∂2 w O
M xy C 31 C32 C 33 2
∂x ∂y
where
æ k zk3 − zk3 − 1 ö
å
n
d3 Cij = çç Ε ij ÷÷
k = 1
è 3 ø
with
c ≡ cos(θ) where θ is the ply direction in the reference coordinate system (o, x, y)
s ≡ sin(θ) where θ is the ply direction in the reference coordinate system (o, x, y)
with
El
d5 Εl =
1 − ν tl νlt
Et
Εt =
1 − ν tl νlt
If the tensor of angles formed by the strain diagram in each direction is defined by (α):
(αx, αy, αxy) we may write in a simplified form the relationship:
d6 (χ) = tg (α)
By convention, we shall assume that (α) is negative when the upper fibre is in tension. We
have:
d7 (ε)z =  (χ) x z
z z
ply No. k
α
zk zk  1
h neutral plan σ ε
ply No. 1
This relationship makes it possible to determine each ply strain and, therefore, to find
(using chapter C) stresses applied to it.
Remark: The terms Cij must be determined with relation to the laminate neutral line
(Kirchoff’s assumption). In this case, the neutral plane shall also be used as a
reference for the overall load pattern.
Equivalent bending elasticity moduli are directly derived from the laminate stiffness matrix
(C):
1
x x
E xx bending equi.
12 1
d8 (C)1 = x x
e3 E yy bending equi.
1
x x
Gxy bending equi.
If reference axes (o, x, y) are coincident with the axes of orthotropy of the laminate, we
obtain:
C11 C 22 − (C12 )2
Exxbending equi. = 12
e3 C 22
C 11 C 22 − (C 12 ) 2
Eyybending equi. = 12
e 3 C 11
C 66
Gxybending equi. = 12
e3
5 . EXAMPLE
0°: 2 plies
45°: 2 plies
135°: 2 plies
90°: 2 plies
z8 = 0.52
k = 8 (0°)
z7 = 0.39
k = 7 (45°)
z6 = 0.26
k = 6 (135°)
z5 = 0.13
k = 5 (90°)
z4 = 0
k = 4 (90°)
k = 3 (135°)
k = 2 (45°)
k = 1 (0°)
El = 13000 hb
Et = 465 hb
νlt = 0.35
νtl = 0.0125
Glt = 465 hb
ep = 0.13 mm
The purpose of this example is to search for elongations at the laminate external surface,
knowing that the laminate is globally subject to the three following moment fluxes in the
reference coordinate system (x, y):
Mx = 10 daN
My = 0 daN/mm
Mxy =  5 daN/mm
z
Mxy =  5 daN
x Mx = 10 daN
{d5}
13000
Εl = = 13057 daN/mm2
1 − 0.35 0.0125
465
Εt = = 467 daN/mm2
1 − 0.35 0.0125
2nd step: For each ply, stiffness coefficients Εij expressed in daN/mm2 are calculated.
{d4}
ply at 0°
Ε11(0°) = 13057
Ε22(0°) = 467
Ε33(0°) = 465
Ε12(0°) = Ε21(0°) = 0.0125 x 13000 = 163
Ε13(0°) = Ε31(0°) = 0
Ε23(0°) = Ε32(0°) = 0
ply at 45°
Ε11(45°) = 0.7074 13057 + 0.7074 467 + 2 x 0.7072 0.7072 (0.0125 x 13057 + 2 x 465) = 3925
Ε22(45°) = 0.7074 13057 + 0.7074 467 + 2 x 0.7072 0.7072 (0.0125 x 13057 + 2 x 465) = 3925
Ε33(45°) = 0.7072 0.7072 (13057 + 467  2 x 0.0125 x 13057) = 3297
Ε12(45°) = Ε21(45°) = 0.7072 0.7072 (13057 + 467  4 x 465) + (0.7074 + 0.7074) x 0.0125 x 13057 = 2995
Ε13(45°) = Ε31(45°) = 0.707 x 0.707 {0.7072 13057  0.7072 467} = 3146
Ε23(45°) = Ε32(45°) = 0.707 x 0.707 {0.7072 13057  0.7072 467} = 3146
ply at 135°
Ε11(135°) = 3925
Ε22(135°) = 3925
Ε33(135°) = 3297
Ε12(135°) = Ε21(135°) = 2995
Ε13(135°) = Ε31(135°) =  3146
Ε23(135°) = Ε32(135°) =  3146
ply at 90°
Ε11(90°) = 467
Ε22(90°) = 13057
Ε33(90°) = 465
Ε12(90°) = Ε21(90°) = 163
Ε13(90°) = Ε31(90°) = 0
Ε23(90°) = Ε32(90°) = 0
3rd step: Calculation of laminate inertia matrix (C) coefficients Cij expressed in daN mm.
The laminate being provided with the mirror symmetry property, coefficients Cij shall be
calculated for the laminate upper half, then they shall be multiplied by 2.
{d3}
858 123 55
55 55 151
{d2}
∂ 2 wo
Mx 858 123 55
∂x 2
∂ 2 wo
My = 123 194 55 =
∂y 2
∂2 wo
Mxy 55 55 151 2
∂x ∂y
hence
∂ 2 wo
1287
. E−3 − 7.617 E − 4 − 1913
. E−4 Mx
∂x 2
∂ 2 wo
= − 7.617 E − 4 6.199 E − 3 − 198
. E−3 = My
∂y 2
∂2 wo
2 − 1913
. E−4 − 198
. E−3 7.414 E − 3 Mxy
∂x ∂y
∂2wo
1287
. E−3 − 7.617 E − 4 − 1913
. E−4 10
∂x 2
∂2wo
= − 7.617 E − 4 6.199 E − 3 − 198
. E−3 = 0
∂y 2
∂2 w o
2 − 1913
. E−4 − 198
. E−3 7.414 E − 3 −5
∂x ∂y
Thus, we find:
∂2wo
13.82 E − 3
∂x 2
∂2wo
= 2.283 E − 3
∂y 2
∂2 w o
2 − 38.98 E − 3
∂x ∂y
5th step: We now propose to calculate strains ε (0°) for the ply at 0° (at the external line of
the layer).
{d7}
∂2 w o h
εx(0°) =  x
∂x 2
2
∂2 w o h
εy(0°) =  x
∂y 2
2
∂2w o h
γxy(0°) =  2 x
∂x ∂y 2
hence:
J. CHAIX, 436.127/91
1 . NOTATIONS
θ: fibre orientation
2 . INTRODUCTION
We have seen in chapter C that there is a relationship which binds membrane strains and
loading of the same type.
We also saw in chapter D that there is a relationship which binds the curvature tensor and
the moment tensor.
If layup has the mirror symmetry property, then both phenomena are dissociated and
independent. In other words, the overall relationship which binds the set of strains and the
set of loadings may be formulated as follows:
Ny A 21 A 22 A 23 0 0 0 εy
Nxy A 31 A 32 A 33 0 0 0 γ xy
= ∂2wo
Mx 0 0 0 C11 C12 C13 ∂x 2
∂2wo
My 0 0 0 C21 C 22 C23 ∂y 2
∂2 wo
Mxy 0 0 0 C31 C 32 C33 2
∂x ∂y
3 . ANALYSIS METHOD
If layup is nonsymmetrical, then all zero terms of the previous matrix become nonzero
and there is a membrane/bending coupling. Both phenomena become dependent. The
relationship between loadings and strains is thus:
where
æ k zk2 − zk2 − 1 ö
å
n
e2 Bij =  çç Eij ÷÷
k = 1
è 2 ø
ply No. k
zk zk  1
neutral plane
ply No. 1
with
where
c ≡ cos(θ) where θ is the fibre direction in the reference coordinate system (o, x, y).
s ≡ sin(θ) where θ is the fibre direction in the reference coordinate system (o, x, y).
with
El
e4 Εl =
1 − ν tl νlt
Et
Εt =
1 − ν tl νlt
Remark: The terms Bij and Cij must be determined with relation to the laminate neutral
line (Kirchoff’s assumption). In this case, the neutral plane shall also be used as
a reference for the overall load pattern.
4 . EXAMPLE
0°: 1 ply
45°: 1 ply
135°: 1 ply
90°: 1 ply
z4 = 0.26
k = 4 (0°)
z3 = 0.13
k = 3 (45°) neutral plane
z2 = 0
k = 2 (135°)
z1 =  0.13
k = 1 (90°)
z0 =  0.26
El = 13000 hb
Et = 465 hb
νlt = 0.35
νtl = 0.0125
Glt = 465 hb
ep = 0.13 mm
The purpose of this example is to search for strains at the laminate internal and external
surfaces, knowing that the laminate is globally subject to the following fluxes in the
reference coordinate system (x, y):
Nx = 5 daN/mm
Ny = 0 daN/mm
Nxy = 0 daN/mm
Mx = 0 daN
æ mm daN ö
My =  0.15 daN ç ÷
è mm ø
Mxy = 0 daN
My =  0.15 daN
x Nx = 5 daN/mm
{e4}
13000
Εl = = 13057 daN/mm2
1 − 0.35 0.0125
465
Εt = = 467 daN/mm2
1 − 0.35 0.0125
2nd step: For each fibre direction, stiffness coefficients Εij expressed in daN/mm2, are
calculated.
{e3}
fibre at 0°
Ε11(0°) = 13057
Ε22(0°) = 467
Ε33(0°) = 465
Ε12(0°) = Ε21(0°) = 0.0125 x 13000 = 163
Ε13(0°) = Ε31(0°) = 0
Ε23(0°) = Ε32(0°) = 0
fibre at 45°
Ε11(45°) = 0.7074 13057 + 0.7074 467 + 2 x 0.7072 0.7072 (0.0125 x 13057 + 2 x 465) = 3925
Ε22(45°) = 0.7074 13057 + 0.7074 467 + 2 x 0.7072 0.7072 (0.0125 x 13057 + 2 x 465) = 3925
Ε33(45°) = 0.7072 0.7072 (13057 + 467  2 x 0.0125 x 13057) = 3297
Ε12(45°) = Ε21(45°) = 0.7072 0.7072 (13057 + 467  4 x 465) (0.7074 + 0.7074) x 0.0125 x 13057 = 2995
Ε13(45°) = Ε31(45°) = 0.707 x 0.707 {0.7072 13057  0.7072 467} = 3146
Ε23(45°) = Ε32(45°) = 0.707 x 0.707 {0.7072 13057  0.7072 467} = 3146
fibre at 135°
Ε11(135°) = 3925
Ε22(135°) = 3925
Ε33(135°) = 3297
Ε12(135°) = Ε21(135°) = 2995
Ε13(135°) = Ε31(135°) =  3146
Ε23(135°) = Ε32(135°) =  3146
fibre at 90°
Ε11(90°) = 467
Ε22(90°) = 13057
Ε33(90°) = 465
Ε12(90°) = Ε21(90°) = 163
Ε13(90°) = Ε31(90°) = 0
Ε23(90°) = Ε32(90°) = 0
3rd step: Calculation of laminate (membrane) stiffness matrix (A) coefficients Aij expressed
in daN/mm.
{c6}
hence
A11 = 2779
A12 = 821
A13 = 0
A21 = 821
A22 = 2779
A23 = 0
A31 = 0
A32 = 0
A33 = 978
4th step: Calculation of laminate (bending) inertia matrix (C) coefficients Cij expressed in
daN mm.
{d3}
C11 = 75.1
C12 = 6.06
C13 = 0
C21 = 6.06
C22 = 75.1
C23 = 0
C31 = 0
C32 = 0
C33 = 9.59
5th step: Calculation of membrane  bending coupling coefficients Bij expressed in daN.
{e2}
B11 =  319
B12 = 0
B13 =  53.2
B21 = 0
B22 = 319
B23 =  53.2
B31 =  53.2
B32 =  53.2
B33 = 0
{e1}
Nx A 11 A 12 A 13 B 11 B 12 B 13 εx
Ny A 21 A 22 A 23 B 21 B 22 B 23 εy
Nxy A 31 A 32 A 33 B 31 B 32 B 33 γ xy
= ∂2wo
Mx B 11 B 12 B 13 C 11 C 12 C 13 ∂x 2
∂2wo
My B 21 B 22 B 23 C 21 C 22 C 23 ∂y 2
∂2 wo
Mxy B 31 B 32 B 33 C 31 C 32 C 33 2
∂x ∂y
then
∂2w
o
= 5.00 x E − 3 − 2.0 x E − 3 2.33 x E − 3 3.57 x E − 2 7.23 x E − 3 1.67 x E − 2 M
x
∂x 2
2
∂ w
o
2.00 x E − 3 − 5.0 x E − 3 2.33 x E − 3 7.23 x E − 3 3.57 x E − 2 − 1.67 x E − 2 M
y
∂y 2
∂2w
o
2 3.61 x E − 3 3.61 x E − 3 0 1.67 x E − 2 − 1.67 x E − 2 1.44 x E − 1 M
∂x ∂y xy
By replacing loading by values quoted at the beginning of the example in the previous
relationship, we find:
εy − 174
. E − 3 mm / mm (− 1740 µd)
γ xy . E − 3 mm / mm
154 (1540 µd)
∂2wo =
∂x 2 2.38 E − 2 mm −1
∂2wo
∂y 2 4.57 E − 3 mm −1
∂2 wo
2 2.05 E − 2 mm −1
∂x ∂y
To do this, membrane strains (εx, εy, γxy) are added to strains resulting from the bending
æ ∂2wo h ∂ wo
2
h ∂2wo hö
effect ç x , x , 2 x ÷
è ∂x
2
2 ∂y 2
2 ∂x ∂y 2 ø
{d7}
∂2 w o h
εx(0°) = εx  x
∂x 2
2
∂2 w o h
εy(0°) = εy  x
∂y 2
2
∂2wo h
γxy(0°) = γxy  2 x
∂x ∂y 2
hence:
J. CHAIX, 436.127/91
1 . NOTATIONS
θ: fibre orientation
El: laminate overall inertia with relation to the (moduli weighted) neutral axis
E Wk: static moment with relation to the (moduli weighted) neutral axis of the set of plies k
to n
τ: shear stress
2 . INTRODUCTION
For simplification purposes, we shall assume that the laminate is made up of n identical
fibres but with different directions.
z
Tyz > 0
y
z
Txz > 0
x
y
θ
k=n
ep
k=1
Layer No. k in direction θ has the following longitudinal elasticity modulus with relation to
the reference coordinate system (o, x, y):
1
f1 Ek = see chapter C3.
c4
s 4
æ 1 ν ö
+ + c 2 s2 ç − 2 tl ÷
El Et è Glt Et ø
with
c ≡ cos(θ)
s ≡ sin(θ)
We shall assume that shear load Txz (direction z load shearing a plane perpendicular to x
axis) creates stress τxz and, based on the reciprocity principle, stress τzx.
Similarly, we shall assume that shear load Tyz (direction z load shearing a plane
perpendicular to yaxis) creates stress τyz and, based on the reciprocity principle, stress
τzy.
τzy
τzx
τyz
τxz
y
Tyz
Txz
3 . ANALYSIS METHOD
To calculate interlaminar stresses τxz (τzx) generated by shear load Txz (Tyz), use the
following methodology.
We shall only consider the case of a laminate subject to shear load Txz. The analysis
principle is the same for Tyz.
In this case, inertias (El) and static moments (E W k) are measured with relation to yaxis.
Elasticity moduli (Ek) are measured with relation to xaxis.
1st step: The position of the laminate neutral axis is determined. If the laminate lower fibre
is used as a reference, then the neutral axis is defined by dimension zg, so that:
å ( )
n
E k z k2 − z k2 − 1
k =1
f2 zg =
2å
n
(
Ek z k − zk − 1
k =1
)
z
ply No. n
ply No. k
zg
zk
zk  1
z1 ply No. 1
y
z0 = 0
2nd step: The (moduli weighted) bending stiffness of laminate El is determined with relation
to the layup neutral axis
(z )
3
− zk − 1 æ zk + zk − 1 ö
2
f3 El = å
n
k =1
Ek
k
12
+ å
n
k =1
(
Ek z k − z k − 1 ) ç
è 2
− zg ÷
ø
ply No. k
zk zk  1
zg
3rd step: Then the (elasticity moduli weighted) static moment E W k (of the material surface
located above the line where interlaminar stress is to be calculated), is determined. This
static moment shall be calculated with relation to the plate neutral axis.
If the line is a fibre interface surface (z = zk  1), then we have the following relationship:
æ zi + z i − 1 ö
f4 E Wk = å
n
i=k
(
Ei z i − z i − 1 ç
è
) 2
− zg ÷
ø
z
ply No. k
zk zk  1
zg
zk + zk − 1
If the line is situated at the centre of a fibre at z = , the relationship becomes:
2
z i + zi
f5 E Wk = å
n
i = k
(
Ei zi − zi − 1 ) æçè 2
− 1 ö
− zg ÷ −
ø
æ zk + z k − 1 ö æ zk + z k − 1 zk − 1 ö
Ek ç − zk − 1÷ ç + − zg ÷
è 2 ø è 4 2 ø
z
z +z
k k − 1
ply No. k
2
zg
zk zk  1
Txz . E Wk
f6 τxzk =
El
By using this analysis method for each ply interface (or at the center of each ply for
greater accuracy), it is possible to plot the interlaminar shear stress diagram over the
entire plate width.
The previous relationship shows that the shear stress is maximum when the static
moment is maximum as well, i.e. at the neutral axis (z = zg).
z z
τxzk
ply No. k
τzxk
zg
τxz y
Remark: The previous analysis is based on a shear load flux Txz applied to a section
perpendicular to xaxis.
In the case of any section forming an angle β in the coordinate system (o, x, y), the shear
load flux in this new section may be expressed as a function of Txz and Tyz.
T(β + π/2)
ds
T(β)
Txz
β
x
Tyz
As shown in the drawing above, the z equilibrium of the hatched material element leads to
the following relationship:
hence:
æ Tyz ö
It is easy to show that for β = Arctg ç ÷ , a modulus extremum T(β) (called main shear
è Txz ø
load flux) is reached that is equal to:
Example: if shear load fluxes Txz and Tyz are equal, then the maximum shear load flux is
located in the plane with a direction β = 45°. Its modulus equals 2 Txy (or 2 Tyz).
4 . EXAMPLE
0°: 1 ply
45°: 1 ply
135°: 1 ply
90°: 1 ply
The purpose of this example is to search for interlaminar shear stresses in the laminate,
knowing that it is subject to the following shear load flux:
Knowing the mechanical properties of the unidirectional fibre, elasticity moduli of each
fibre should be calculated in direction x.
{f1}
1
E2 = 4 4
0.707 0.707 æ 1 0.0125 ö
+ + 0.707 2 0.707 2 ç − 2 ÷
13000 465 è 465 13000 ø
{f2}
465 (0.13 2 − 0 2 ) + 925 (0.26 2 − 0.13 2 ) + 925 (0.39 2 − 0.26 2 ) + 13000 (0.522 − 0.39 2 )
zg =
2 ( 465 (0.13 − 0) + 925 (0.26 − 0.13) + 925 (0.39 − 0.26 ) + 13000 (0.52 − 0.39 ))
zg = 0.42 mm
z4 = 0.52
z3 = 0.39
z2 = 0.26
zg = 0.42
z1 = 0.13
z0 = 0
2nd step: Analysis of the laminate bending stiffness El with relation to the neutral axis
{f3}
2
æ 0.13 + 0 ö
465 (0.13 − 0) ç − 0.42 ÷ +
è 2 ø
2
æ 0.26 + 0.13 ö
925 (0.26 − 0.13) ç − 0.42 ÷ +
è 2 ø
2
æ 0.39 + 0.26 ö
925 (0.39 − 0.26) ç − 0.42 ÷ +
è 2 ø
2
æ 0.52 + 0.39 ö
13000 (0.52 − 0.39) ç − 0.42 ÷
è 2 ø
El = 19.67 daN.mm
3rd step: Analysis of static moments E W k (with relation to the neutral line) at the base and
center of each ply.
{f4}
E W 4 = 0 daN
z
0°
45°
135°
90° y
{f5}
æ 0.52 + 0.39 ö
E W 4 = 13000 (0.52 − 0.39 ) ç − 0.42 ÷ −
è 2 ø
E W 4 = 59.15  2.11
E W 4 = 57.04 daN
0°
{f4}
æ 0.52 + 0.39 ö
E W 4 = 13000 (0.52 − 0.39 ) ç − 0.42 ÷
è 2 ø
E W 4 = 59.15 daN
z
0°
{f5}
E W 3 = 55.4 daNp
45°
{f4}
E W 3 = 59.15  11.42
E W 3 = 47.73 daN
z
45°
{f5}
æ 0.26 + 0.13 ö
925 (0.26 − 0.13) ç − 0.42 ÷ −
è 2 ø
E W 2 = 35.35 daN
z
135°
y
{f4}
æ 0.26 + 0.13 ö
925 (0.26 − 0.13 ) ç − 0.42 ÷
è 2 ø
E W 2 = 19.87 daN
z
135°
y
{f5}
æ 0.13 + 0 ö æ 0.13 + 0 0 ö
465 ç − 0÷ ç + − 0.42 ÷
è 2 øè 4 2 ø
90°
y
{f4}
E W 1 = 0 daN
90° y
In the example given, it is located at the point where the static moment is maximum, i.e. at
the base of the ply at 0°. Its value equals at E W 0 = 59.15 daN, which gives stress τxz0:
{f6}
0.7 x 59.15
τxz0 = = 2.1 hb (21 MPa)
19.67
If these interlaminar shear stresses are analysed for each fibre, stresses are distributed
along the laminate thickness as follows:
0.7 E Wk
τxzk =
19.67
0.52
0.455
0.39
0.325
z (mm) 0.26
0.195
0.13
0.065
0
0 0.5 1 1.5 2 2.5
τ (hb)
FAILURE CRITERIA
Notations G 1
1 . NOTATIONS
FAILURE CRITERIA
Inventory G 2
The purpose of this chapter is to describe various failure criteria of the unidirectional fibre
within a laminate.
The following criteria shall be presented in chronological order (this is not an exhaustive
list):
 Puck's criterion
 Hill's criterion
 Norris's criterion
 Fischer's criterion
 Hoffman's criterion
For threedimensional criteria, we shall assume that the composite material is subjected to
the following stress tensor and strain tensor:
For twodimensional criteria, we shall assume that the unidirectional fibre is subjected to
the following stress tensor and strain tensor:
FAILURE CRITERIA
Maximum stress criterion G 2.1
for 1 ≤ i ≤ 6
For the twodimensional case, the failure envelope may be represented as follows:
σt
σl
τlt
FAILURE CRITERIA
Maximum strain criterion G 2.2
for 1 ≤ i ≤ 6
For the twodimensional case, the failure envelope may be represented as follows:
εt
εl
γlt
FAILURE CRITERIA
Norris and Mac Kinnon's criterion G 2.3
å
6
C i (σ i ) 2 = 1
1
σt
σl
τlt
FAILURE CRITERIA
Puck's criterion G 2.4
and
2 2 2
æ σ1 ö æσ ö æτ ö
g4 ç ÷ + ç 2 ÷ + ç 12 ÷ = 1
è X1 ø è X2 ø è X6 ø
σt
σl
τlt
FAILURE CRITERIA
Hill's criterion G 2.5
This criterion is valid for orthotropic materials or for slightly anisotropic materials only.
σt
σl
τlt
FAILURE CRITERIA
Norris's criterion G 2.6
and for 1 ≤ i ≤ 6
σt
σl
τlt
FAILURE CRITERIA
Fischer's criterion G 2.7
2 2 2
æ σl ö æσ ö σ σ æτ ö
g7 ç ÷ + ç t ÷ − K l t + ç lt ÷ = 1
è X1 ø è X2 ø X1 X 2 è X 6 ø
σt
σl
τlt
FAILURE CRITERIA
Hoffman's criterion FAILURE CRITERIA G 2.8
C1 (σ2  σ3)2 + C2 (σ3  σ1)2 + C3 (σ1  σ2)2 + C4 (σ4)2 + C6 (σ6)2 + C5 (σ5)2 + C'1 σ1 + C'2 σ2 +
C'3 σ3 = 1
Coefficients C1, C2, C3, C4, C5, C6, C'1, C'2 and C'3 depend on the material used.
σt
σl
τlt
FAILURE CRITERIA
Tsaï  Wu's criterion G 2.9
This criterion intends to be as general as possible and then, there is, a priori, no particular
hypothesis.
For 1 ≤ i ≤ 6
Σ Fi σi + Σ Fij σi σj + Σ Fijk σi σj σk + … = 1
g9 F1 σl + F2 σt + F6 τlt + F11 (σl)2 + F22 (σt)2 + F66 (τlt)2 + 2 F12 σl σt + 2 F26 σt τlt + 2 F16 σl τlt = 1
Coefficient F1, F2, F6, F11, F22, F66, F12, F26 and F16 depend on the material used.
FAILURE CRITERIA
Aerospatiale's criterion G 3
1/2
As seen previously, Hill's criterion is, in its general form, formulated as follows:
This noninteractive criterion is applicable at the elementary ply only. There is a laminate
failure when the most highly loaded layer is broken.
2 H = (1/Rl)2
2 2
æσ ö æσ ö æτ ö
2
æσ σ ö
g10 2
There is a failure if h = ç l ÷ + ç t÷ + ç lt ÷ − ç l 2 t÷ = 1
è Rl ø è Rt ø è Sø è Rl ø
FAILURE CRITERIA
Aerospatiale's criterion G 3
2/2
1 1
g11 RF = =
h 2
æ σt ö
2 2
æ σl σ t ö
æ σl ö æ τ lt ö
ç ÷ + ç ÷ + ç ÷ +ç ÷
è Rl ø è Rt ø è Sø 2
è Rl ø
This criterion is the one used by Aerospatiale. In order to avoid having a premature
theoretical failure in the resin, the transversal modulus Et was considerably reduced (by a
coefficient 2 approximately) with relation to the average values measured.
This "design" value is determined so that the transversal strain is greater than the
longitudinal one.
The allowable plane shear value S of the unidirectional fibre was determined for having, a
B good test/calculation correlation and significant tension and compression failures of
notched or unnotched laminates.
FAILURE CRITERIA
Example G 4
1/4
4 . EXAMPLE
Hill's criterion shall be applied to the example considered in the chapter "plain plate 
membrane". Stresses applied to fibres are calculated and presented in the corresponding
chapter (C.6) and quoted in the following pages.
0°: 6 plies
45°: 4 plies
135°: 4 plies
90°: 6 plies
The laminate is globally subjected to the three following load fluxes in the reference
coordinate system (x, y) (see chapter C.6) :
Nx = 30.83 daN/mm
Ny =  2.22 daN/mm
FAILURE CRITERIA
Example G 4
2/4
σl = 29.42 hb
σt = 0.06 hb
τlt = 5.03 hb
{g10}
2 2 2
æ 29.42 ö æ 0.06 ö æ 5.03 ö æ 29.42 x 0.06 ö
÷ +ç ÷ +ç ÷ −ç ÷ =1
2
h = ç
è 120 ø è 5 ø è 7 .5 ø è 120 2 ø
{g11}
1 1
Reserve Factor: R.F. = = = 14
.
h2 0.51
σl = 80.17 hb
σt =  1.14 hb
τlt =  1.36 hb
{f10}
2 2
æ − 114 æ − 136 æ 80.17 x (− 114
2
æ 80.17 ö . ö . ö . )ö
h2 = ç ÷ + ç ÷ + ç ÷ − ç ÷
è 120 ø è 12 ø è 7.5 ø è 120 2 ø
FAILURE CRITERIA
Example G 4
3/4
{g11}
1
Reserve Factor: R.F. = = 142
.
0.498
Margin ≈ 42 %
σl =  59.17 hb
σt = 2.14 hb
τlt = 1.36 hb
{g10}
2
æ − 59.17 ö æ − 59.17 x 2.14 ö
2 2
æ 2.14 ö æ 136
. ö
h2 = ç ÷ + ç ÷ + ç ÷ − ç ÷
è 100 ø è 5 ø è 7.5 ø è 100 2 ø
{g11}
1
Reserve Factor: R.F. = = 131
.
0.579
Margin ≈ 31 %
FAILURE CRITERIA
Example G 4
4/4
σl =  8.42 hb
σt = 0.95 hb
τlt =  5.03 hb
{g10}
2 2
æ − 8.42 ö æ − 5.03 ö æ − 8.42 x 0.95 ö
2
æ 0.95 ö
2
h = ç ÷ + ç ÷ + ç ÷ − ç ÷
è 100 ø è 5 ø è 7.5 ø è 100 2 ø
{g11}
1
Reserve Factor: R.F. = = 142
.
0.494
Margin ≈ 42 %
1 . NOTATIONS
2 . INTRODUCTION
The regulatory requirements in terms of structural justification concern, on the one hand,
the static strength JAR § 25.305 and, on the other hand, fatigue + damage tolerance JAR
§ 25.571. For the latter, three cases are to be considered:
For the static strength evaluation, Acceptable Means of Compliance ACJ 25.603 § 5.8
requests resistance to ultimate loads with "realistic" impact damage susceptible to be
produced in production and in service. This damage must be at the limit of the
detectability threshold defined by the selected inspection procedure. Also, static strength
must be demonstrated after application of mechanical fatigue (§ 5.2) and test specimens
must have minimum quality level, that is, containing the permissible manufacturing flaws
(§ 5.5) and "realistic" impact damage.
The static strength range is defined therefore for a detection threshold and by a "realistic"
cutoff energy leading to "realistic" impacts.
Detection threshold
(impact depth in mm)
Static strength
range
High thickness
Impact energy
2
MONOLITHIC PLATE  DAMAGE TOLERANCE
Introduction I 3
3.1
Distinction is made between the range above the detectability threshold where all damage
will be detectable and the range above the static cutoff energy and below the detectability
threshold where the damage will never be detected.
In this "Damage tolerance" section, we shall discuss both manufacturing defects and
impact damage for the static justification and the fatiguedamage tolerance justification.
Distinction is made between damage which may occur during manufacture and that which
occurs in service.
Damage, outside of the curing process, can occur a detail part or component level during
the assembly phases or during transport or on flight line before delivery to the customer.
For all composite parts, the acceptance/scrapping criteria must be defined by the Design
Office. Acceptable damage/flaws are incorporated into the ultimate load justification by
analysis and into the test specimens to demonstrate the tolerance of the structure to this
damage throughout the life of the aircraft.
This damage occurs in service in a random manner. Distinction is made between three
types of damage:
 fatigue,
 corrosion and environmental effects,
 accidental.
Composite materials are said to be insensitive to fatigue; more exactly, their mechanical
properties are such that the static dimensioning requirements naturally cover the fatigue
dimensioning requirements. This is valid for a laminate submitted to plane loads, less than
60 % of ultimate load. However, complex areas or areas with a sudden variation in rigidity
may favour the appearance of delaminations under triaxial loads.
Today, it is very difficult to (analytically or numerically) model the growth rate of a possible
flaw. This is why a "safelife" justification philosophy has been adopted. It is based on two
principles which must be underpinned by experimental results:
On account of the dispersion proper to composites and the form of the "Wohler" curves
associated with them (relatively flat curve with low gradient), the factor 5 normally used on
metallic structures for the number of lives to be simulated during a fatigue test, was
replaced by a load factor.
a) Corrosion
For information purposes, the table below shows various carbon/metal pairs over a scale
ranging from A to E.
We consider that type A and B couplings are correct and that those of types C, D and E
are not.
B Monel, inconel
b) Environmental effects
This is the most important type of damage and the damage most liable to call into
question the structural strength of the part. It can occur during the manufacture of the item
(drilling delamination) or in service (drop of a maintenance tool, hail or bird strikes).
4 . INSPECTION OF DAMAGE
This type of inspection is mainly conducted during production but can be used
exceptionally in service.
This type of inspection enables BVID (Barely Visible Impact Damage) to be detected.
4
MONOLITHIC PLATE  DAMAGE TOLERANCE
Inspection of damage I 4.1
4.2
Inspection  Walk Around Check (ref: Maintenance Review Board Document: MRB):
In general, the Walk Around check is considered as a general daily visual inspection.
These inspections are conducted with bulky facilities: ultrasonic, thermographic, Xrays,
etc. Minimum detectable sizes are related to the size of the U.S. probes and the accuracy
of the tools used, etc.
This type of damage is called BVID (Barely Visible Impact Damage). The geometrical
detectability criteria are as follows (cf. ref. 22S 002 10504):
This type of damage is called Minor VID (Minor Visible Impact Damage). The geometrical
detectability criteria are as follows (cf. ref. 22S 002 10504):
δ"
Depth of flaw "δ Size of perforation
2 mm
or thickness of structure 20 mm ∅
if < 2 mm
This type of damage is called Large VID (Large Visible Impact Damage). The geometrical
detectability criteria are not explicitly defined but the damage must be detectable without
ambiguities during a Walk Around Check.
The diagram below summarises these four detectability levels according the size of the
damage.
Special Detailed General
detailed visual visual
inspection inspection inspection Walk around
Size of
BVID Minor Large damage
VID VID
Depending on the type of visual inspection considered during the maintenance phases
(general or detailed), we will define three clearly separate detectability ranges:
These ranges are positioned as follows on the previously defined detectability scale:
Damage Inevitably
Undetectable susceptible to detectable
damage be detected damage
DVI WA
Damage Inevitably
Undetectable susceptible to detectable
damage be detected damage
GVI WA
4.5
MONOLITHIC PLATE  DAMAGE TOLERANCE
Influence of damage  Porosity I 5
5.1
5.1.1
Remark: Note that certain authors define the BVID notion according to the type of
inspection selected.
In this case, for a general inspection: MINOR VID ≡ BVID
In our document, we will conserve the initial definition related to the visual
detailed inspection.
5.1.1 . Porosity
→ Description
By "porosity", we mean a heterogeneity of the matrix leading, more often than not, to lack
of inter or intralayer cohesion, generally small in size, but distributed uniformly or almost
throughout the complete thickness of the laminate. Note that for unidirectional tapes the
porosities have a tendency to be located between the layers whereas, for fabrics, they are
more generally located where the weft and warp threads cross. The porosity ratio given is
a surface porosity ratio measured by the ultrasonic attenuation method. The permissible
absorption level is fixed at 12 dB irrespective of the thickness inspected (cf. note
440.241/90 issue 2  SIAM curve). All absorption areas above this limit will be considered
as a delamination and meet therefore the same criteria as a delamination.
However, only T300/N5208, more fluid than T300/BSL914 has a higher tendency to be
porous.
The test results were interpolated, for the V10F wing, on T300/N5208 with various
porosity ratios distributed in all interply areas to determine the influence on the mechanical
characteristics for a 3 % ratio considered as the acceptable limit. This ratio combined with
the fatigue, ageing and residual test effects at 80° C, led to the following losses in
mechanical characteristics:
3 % porosity
Loss of characteristics
T300/N5208 Loss of characteristics
after F + VC1 + 80° C
after F + VC1 + 80° C
BENDING  15 %  19 %
COMPRESSION  20 %  19 %
 that the spar boxes of the wings, movable surfaces or fin are subjected to very low
interlaminar stresses,
 that the 3 % porosity criterion distributed at all interply areas is today no longer
applied to primary structures. The permissible porosity ratio depends on the thickness
of the laminate.
5.1.2 . Delaminations
→ Description
A skin bottom delamination is a lack of cohesion between two welldefined plies. Natural
delaminations appear during manufacture (surface contamination). A foreign body left in
the laminate (separator) will be considered as a delamination.
For the V10F wing, a lack of interlayer cohesion up to 400 mm2 leads to a loss of
compression strength of around 10 % for the two materials (T300/N5208 and
T300/BSL914) tested in new condition at θ = 80° C. In aged/fatigue condition the drop in
strength is 20 % for T300/N5208 and 13 % for T300/BSL914 in relation to the new
state/80° C reference. Fatigue leads to no growth of the flaw.
Þ Fastener areas
→ Description
As for the skin bottom delaminations, the lack of cohesion in these areas occurs between
two welldefined plies, sometimes at several levels but generally adjacent. These flaws
come through to the bore. They are created during the drilling operations. The ultrasonic
inspections conducted after each test case showed no evolution of existing flaws.
damage ∅
The parameter representing the size of the damage is the number given by: φ =
fastener ∅
damage Ø
fastener Ø
Vb
The parameter representing the drop in characteristics is the number given by: ν =
Vc
where Vb represents the "B value" (see section Y) relevant to all tests characterising the
material and where Vc is the calculation value used. Provided that the calculation value is
lower than the "B value", the integrity of the item is ensured. For safety reasons, we will
impose a minimum margin of 10 % between the calculated value and the "b value".
Generally speaking, the graphs gives the values of ν for the flaw (delamination) but also
for repairs which may be made on it (injection of resin, NAS cup). They enable you to find
therefore:
For delaminated hole concentrations and irrespective of the size of the delaminations, the
flaw must be covered by a concession if:
 for aligned fasteners, more than 20 % of the holes are delaminated and/or two flaws
are less than 5 fastener pitches apart,
 for a delamination at a fastener or of another skin bottom area, they are less than
120 mm apart.
For delaminated hole concentrations and irrespective of the size of the delaminations, the
flaw must be covered by a concession if:
 for aligned fastener, more than 10 % of the holes are delaminated and/or two flaws
are less than 5 fastener pitches apart,
 for a delamination at a fastener or of another skin bottom area, they are less than
120 mm apart.
• if the flaws are located on several rows, they must be covered by a concession
if they are less than 175 mm apart.
The table below summarises the repair solutions to be applied when delaminations are
detected at fastener holes in materials T300/914, G803/914 and HTA/EH25 depending on
the loads and the damage ∅ ratio.
fastener ∅
 for a pure load, the repair or untreated delamination must resist ultimate loads under
the most severe environmental conditions,
 for a pure bearing stress test, the calculation value Vc is taken as reference. The
Vb
repair will not be acceptable if is lower than 1.
Vc
The validation range of the acceptable solutions given in the table below is damage ∅ ≤ 6.
fastener ∅
Without
Acceptable Unacceptable  Acceptable
bending
Bending Acceptable in
JOINT tensile Acceptable Unacceptable 
1000 µd "hollow"
Bending
Acceptable Unacceptable  Unacceptable
2500 µd
Without
Unacceptable Unacceptable  Acceptable
bending
JOINT Bending Acceptable in
Unacceptable Unacceptable 
compression 1000 µd "hollow"
Bending
Unacceptable Acceptable  Acceptable
2500 µd
❏ Stiffener runouts
Stiffener runouts represent a critical point for dimensioning. When these stiffener runouts
are made during moulding without later machining operations, these fairly tortured areas
may include lacks of cohesion either in the base, or in the stiffener itself.
Usection
Half core
Þ Crater
→ Description
This flaw is consecutive to too short a wedge which gives, after machining of the stiffener
runout, a crater at the end of the stiffener.
L
e
Loss of
Size of flaw Test conducted Conditions characteristics
due to flaw
Tensile
Compression New
ATR 72  28 %
(stiffener runouts θ = 20° C
T300/914 not protected)
L = 10 mm
l = 4 mm Tensile
e = 1 mm Compression New
0%
(stiffener runouts θ = 20° C
protected)
Compression
4%
(with reinforcement)
ATR 72 Aged
HTA/EH25 Compression θ = 70° C
(without  12 %
reinforcement)
For unprotected stiffener runouts (that is, when it was impossible to thicken the skin to
make structure relatively simple to manufacture), this flaw must be covered by a
concession. When it is located at protected stiffener runouts (that is with a significant skin
overthickness at stiffener runout), this flaw will be covered by a concession only if its size
is greater than the following values:
L = 10 mm l = 2 mm e = 0.5 mm
Þ Punching
→ Description
This flaw is due to an imperfect Mosite cut leading to flaws at stiffener ends.
L
e
Loss of
Size of flaw Test conducted Conditions characteristics
due to flaw
Tensile
Compression New
 20 %
ATR 72 (stiffener runouts θ = 20° C
T300/914 not protected)
Tensile
L = 10 mm Compression New
e = 1 mm 0%
(stiffener runouts θ = 20° C
protected)
L = 10 mm l = 2 mm e = 0.5 mm
→ Description
Flaw BC
A
B
C
Wedge
Usection
Half core
Loss of
Type of flaw Test conducted Conditions characteristics
due to flaw
V10F
Tensile
T300/N5208 New
(between wedge  17 %
200 mm2 θ = 20° C
and base skin)
(flaw B)
ATR 72 Tensile
Wet ageing
T300/914 (unprotected  20 %
θ = 50° C
(flaw BC) stiffener runouts)
ATR 72 Compression
Wet ageing
T300/914 (unprotected 0%
θ = 50° C
(flaw BC) stiffener runouts)
❏ Stiffener top
Lack of interlayer cohesion at top of stiffener between the Usection and the wedge does
not seem to modify the mechanical characteristics.
❏ Stiffener base
This flaw correspond to lack of cohesion between two welldefined plies in the web/flange
blendin radius.
Þ In standard areas: maximum local surface area between 2 ribs for a radius is 250
mm2, including delaminations and foreign bodies.
Þ In designated areas: maximum local surface area between 2 ribs for a radius is
150 mm2, including delaminations and foreign bodies.
Delamination
An acceptable flaw will however require a Hysol 9321 sealing operation on edge. Any
other flaws shall be covered by a concession.
Translaminar cracks have been detected on the ATR 72 outer wing spar box, the A340
aileron, the 2000 fin, the A300/A310 (cf. note 494.048/91); however there are none on the
flight V10F (cf. note 494.007/91).
These are elongated flaws due to the use of a corrosive stripper (MEK, Methyl Ethyl
Ketone). Currently, baltane is used. T300/914 and G803/914 have these flaws; the tests
conducted on IM7/9772 and HTA/EH25 showed no translaminar cracks (cf. note
494.056/91).
These cracks are detected by ultrasonic inspection in the fastener areas (the back surface
echo totally disappears). They concern all ply directions but do not touch between two
plies with different orientations. It is in the high crack density area that the ultrasonic signal
is totally damped. There a transition zone between this area and the healthy part of the
laminate where crack density decreases and the ultrasonic back surface echo reappears.
These cracks are parallel to the fibres leaving the holes. They first affect the plies at 0°,
then the plies at ± 45°. Some cracks are observed in the central plies at 90°. The axes of
these crack networks correspond approximately to the hole diameters.
They do not lead to a drop in the mechanical characteristics (cf. note 437.115/91).
The existence of flaws at fasteners can be masked by high density translaminar cracks.
Therefore, the threshold of the surface areas of the translaminar cracks which must be
plotted is coherent with the size of acceptable delaminations.
→ Description
An impact causes lack of interlayer cohesion at several levels depending on the energy of
the impact.
The drops in characteristics within the scope of the V10F programme are:
All points of the tests conducted on the V10F test specimens were plotted on the graph
below (the points of the static and fatigue test specimens are combined on this curve as it
has been demonstrated that the ageing effect is not significant for damage tolerance).
The curve used at Aerospatiale for the new states/residual test at ambient temperature
and aged/fatigue states/residual test at ambient temperature is shown on the curve below
by comparison at static test specimen and fatigue test specimen points.
 1500
 2000
Allongement de compression (µd)
 3500
COURBE ACTUELLE VALEURS DE CALCUL
Etat neuf/température ambiante ou
 4000 Etat vieilli/fatigue à 20° C/température ambiante
 4500 CES
CEF
 5000
The problem is generally posed as follows: we take a laminate consisting of a set of tapes
(or fabrics) that we will assume to be made of the same material, each one of them having
a specific orientation in relation to the reference frame (o, x, y).
The laminate is submitted to shear forces (of membrane type) Nx, Ny and Nxy. In the
presence of a delamination (without ply failure) in surface area Sd, what is the strength of
the plain composite plate?
Today, there are three methods for evaluating the residual strength in the presence of a
delamination (established from experimental results) which call on the stresses and/or
strains of the unidirectional fibre and not those of the laminate considered as a
homogeneous plate. Each fibre direction must therefore be justified.
1st method:
This first method consists in calculating a failure criterion determined from the strains of
each fibre in relation to their intrinsic frame (o, l, t).
By referring to the "plain plate  calculation method" section, it is possible to calculate the
strains in the various layers of the composite from the global flows Nx, Ny and Nxy applied
to the laminate and from the characteristics of the material used.
For layer "i" defined by its orientation α i, the strains of the fibre "i" in its own frame are
defined by the following strain tensor: (εli, εti, γlti).
We can define the following failure criterion C1 for each layer "i":
2 2
æ εl i ö æ γ ö
i1 C1 = ç ÷ + ç lt i ÷
çε ÷ çγ ÷
è adm ø è adm ø
where εadm and γadm are the permissible strains (longitudinal and shear) of the
unidirectional fibre (equivalent).
These values (obtained from the test results) depend on the material and the surface area
Sd of the delamination considered and the types of loads.
They are given in section Z (sheets giving calculation values and coefficients used).
This criterion was used for the dimensioning of the ATR 72 wing panels (dossier
22S00210460).
2nd method:
This second method consists in calculating a failure criterion C2 (Hill type criterion in which
the permissible stresses are reduced by coefficients κR and κS) calculated from the
stresses in each fibre in relation to their intrinsic frames (o, l, t).
By referring to the "plain plate  calculation method" section, it is possible to calculate the
strains in the various layers of the composite from the global flows Nx, Ny and Nxy applied
to the laminate and from the characteristics of the material used.
For layer "i" defined by its orientation α1, the stresses of the fibre "i" in its own frame are
defined by the following stress tensor: (σli, σti, τlti).
We can define the following failure criterion C2 for each layer "i":
2 2 2
æ σ li ö æσ ö æ τ ö σ li σ t i
i2 C2 = ç ÷ + ç t i ÷ + ç lt i ÷ −
çκ R ÷ çR ÷ ç κ S÷ (κ R R l )2
è R l ø è tø è s ø
where Rl, Rt and S are the permissible longitudinal, transverse and shear stresses of the
unidirectional fibre respectively (equivalent) and where κR and κs are the reduction
coefficients for these permissible stresses.
These coefficients depend on the material used and the surface area of the delamination
considered and are determined from the test results.
They are given in section V (sheets giving calculation values and coefficients used).
This criterion was used for the sizing of the A330/340 inboard and outboard aileron
panels.
3rd method:
This method consists in calculating a failure criterion C3 (similar to the one of method 1)
calculated from the strains of each fibre in relation to their intrinsic farmes (o, l, t).
For layer "i" defined by its orientation αi, the strains of the fibre "i" in its own frame are
defined by the following strain tensor: (εli, εti, γlti).
We can define the following failure criterion C3 for each layer "i" :
2 2
æ ε li ö æ γ ö ε ε
i3 C3 = ç ÷ + ç lt i ÷ + li t i
çε ÷ çγ ÷ (ε ab )2
è a ø è adm ø
where:
if 2 ôεadmô ≤ ôγadmô
1
i4 εa =
3 2
−
2 (ε adm ) (γ adm )2
2
1
i5 εab =
3 6
−
2 (ε adm ) (γ adm )2
2
else
εa = εadm
εab = + ∞
The particularity of this method is that it takes into account (in a significant manner) the
load transverse to the fibre.
Tests have shown that presence of a tensile force perpendicular to the fibre direction
compression increases the ultimate strength of the laminate.
Criterion C3 takes this phenomenon into account. Indeed, if εti is of tensile type and εli of
compression type, the third term of the criterion C3 becomes negative and tends to
increase the reserve factor and therefore the margin (RF = 1/C3).
Today, it is recommended to use this third finer method based on a high number of
experimental results.
→ Description
Sharp scratches are made by scalpels or cutting tools. Sharp scratches lead to drops in
tensile characteristics of around 15 %; for compression, we assume that there is no drop
in characteristics.
Any scratch concentrations must be covered by a concessions if the flaws are less than
20 mm apart.
Þ On A330/A340 inboard and outboard ailerons, if length of scratch is less than 100 mm
and if its depth is less than 0.15 mm for tapes and 0.3 mm for fabrics, sealing with
Hysol 9321 will be performed.
Þ On A330/A340, A320, A319, A321 nose landing gear doors (carbon fabrics G803/914),
5.2.1
MONOLITHIC PLATE  DAMAGE TOLERANCE
Indents  Scaling I 5.2.2
5.2.3
1/3
Þ On A330/A340, A320, A319, A321 main landing gear doors (carbon fabrics G803/914),
5.2.2 . Indents
"Indent" type flaws due, for instance, to abrasion of skin by a rototest are permissible if:
 only the 1st ply is totally damaged, that is 2nd ply visible.
Any flaw concentrations must be covered by a concession if two indents are less than
100 mm apart.
5.2.3 . Scaling
→ Description
By "scaling", we mean separation or removal of several fibres (locally) altering only the
first surface ply on monolith edge or on outgoing side of drilled holes.
For scaled hole concentrations, this flaw must be covered by a concession if, for aligned
fasteners, more than 20 % of the holes are scaled and/or two flaws are less than
5 fastener pitches apart.
Flaw 1 Flaw 2
For scaled hole concentrations, this flaw must be covered by a concession if:
 for aligned fasteners, more than 10 % of the holes are scaled and/or two flaws are
less than 5 fastener pitches apart,
Flaw 1 Flaw 2
175 mm
Flaw 1 For flaws 1 and 3:
to be covered by a
concession
Flaw 2
For flaws 1 and 2:
if S1 and S2 ≤ permissible
surface area permissible
Flaw 3
• for flaws on several rows; must be covered by a concession if they are less than
175 mm apart.
All scaled areas will be sealed with Hysol 9321 to restore flat surface and avoid scaling
developing during later operations.
Þ On A330/A340 inboard and outboard ailerons, scaling on 1 ply of skin will be sealed
with Hysol 9321. Permissible scaling flaws are defined as follows:
For flaw concentrations at fasteners, two flaws on same row must be separated by 9
fasteners.
For flaw concentrations, two flaws on a given row must be separated by 9 fasteners.
5.2.4 . Steps
→ Description
This is a fold of one or more skin plies which may occur between two (spar support)
blocks or on a sandwich skin during cocuring or in spar webs.
50 mm Filleralu
 Standard areas: steps on spar and rib passage areas are acceptable within a limit of
0.3 mm. This type of flaw will be compensated for by Filleralu over a width of 50 mm
on either side of the step.
→ On stiffeners
 standard areas: steps on stiffener flanges are acceptable within a height limit of
0.3 mm provided that:
Þ On A330/A340 inboard and outboard ailerons under spar and rib bearing surfaces,
steps lower than or equal to 0.2 mm and with a width lower than or equal to 3 mm will
be accepted, but:
Þ On A330 Pratt et Whitney thrust reverser sandwich skins mainly in areas with high
curvatures, steps with a height less than 0.5 mm are accepted in production. Steps
greater than 0.5 mm will be examined case by case.
7
MONOLITHIC PLATE  DAMAGE TOLERANCE
Justification of inservice damage I 7.1
7.1.1
As laid down in the regulations, any damage which cannot withstand the limit loads must
be readily detectable during any general visual inspection (50 flights) or obvious.
Þ Damage readily detectable within an interval of 50 flights must withstand 0.85 LL.
Þ Obvious damage (engine burst) which occurs in flight with crew being aware of it must
withstand 0.7 LL (gethome loads capability).
Þ Regulatory aspects
Þ General
For metallic structures, the two fundamental damage tolerance parameters are the
initiation of the damage and its growth before detection. Many tests have been conducted
therefore to evaluate the growth speed of the damage and the time required to reach its
critical size and therefore its residual strength (limit load).
εresidual
Repair
εU.L.
εL.L.
METALLIC
Growth
Time
Initiation
threshold
Inspection intervals
εresidual
εU.L.
εL.L.
COMPOSITE
Time
At time of impact
a) Semiprobabilistic methods
If the nogrowth concept of the flaw is demonstrated (by fatigue test), the size of the
damage no longer depends on an evolving phenomenon but on a random event
(accidental).
For the damage range between BVID and VID, the aim of the (analytical) justification will
be to determine an inspection interval so that the probability (Re) of simultaneously having
a flaw and a load greater than its residual load will be a highly improbable event
(probability per flight hour less than 109).
This probabilistic damage occurrence versus time aspect therefore replaces the
deterministic concept for metallic materials where the occurrence of a flaw depends either
on fatigue initiation, or, for certain areas, on an accidental impact; the effect of the latter
being a modification in the threshold.
The complete philosophy can be summarised by the curve below. It expresses the load
level to be demonstrated and the type of justification versus the damage range
considered.
The portion of the curve between the BVID and the VID depends on the results of the
probabilistic analysis. Probabilistic analysis
TOLDOM
εresidual
≥ L.L.
Re = E  9
εBVID ≥ U.L.
εU.L.
εVID
εL.L.
0.85 εL.L.
0.7 εL.L.
ULTIMATE
LOADS
In order not to design composite structures less reliable than metallic ones, an inspection
programme has been defined so that the probability of simultaneously having a flaw and a
load greater than its residual strength will be a highly improbable event (probability per
flight hour less than 109).
or again:
This condition involves several notions that we will specify in the following sections.
We defined, in § 4.2 to 4.4, the visual detection criteria for "A" value and "B" value
damage and the mean value for various types of inservice inspections. Knowing that the
"A" values correspond to a detection probability of 99 %, the "B" values to a probability of
90 % and the mean values to a probability of 50 %, we can deduce the curve below which
shows the probability of detection versus the depth of the indent and the type of
inspection.
Depth of indent
(mm)
*
2
0,5
**
0,3
Pdat
0 0.5 0.99 1 Detection probability
❑ Pat: probability of occurrence of an impact with given energy.
 projection of gravel,
 removal of the item,
 dropping of tools or removable items,
 shock with maintenance vehicle.
As for detection, we will define an impact source by a statistical distribution (in this case,
the Lognormal distribution).
We will therefore speak of the impact probability (or, more precisely, the impact energy
range) that we will call (Pat) and which will be characterised by mean energy Em and a
standard deviation (according to Sikorsky, the standard deviation σ has a constant value
equal to 0.217).
∞
We also obtain ò
0
f (E) x dE = 1
f(E)
Pat
E
Em 2J
The impact energy will generally be limited to 50 J (cutoff energy), except for THS root:
140 J corresponding to the energy of a tool box failing from the top of the fin.
Now that the impact has been defined, we must find the relation between the incident
energy (E), the size of the damage (Sd) and its indentation (f).
Generally, we have :
E
Sd = Ksd
e
3.3
æEö
f = Kf ç ÷
èeø
Test campaigns are however necessary to determine the coefficients Ksd and Kf which
depends on the types of materials, their thickness and the item bearing conditions.
❑ Prat: probability of having a loading case greater than the residual strength of the
impacted laminate.
The need to have three variables to characterise the number C (εl, εt, γlt ou σl, σt, τlt)
makes all theoretical exploitations of the item loads (or deformations) difficult.
ε admissible
We will therefore define a number εresidual = which represents the permissible
C
strain of damage of size Sd under a single compression load.
This residual deformation depends of course on the size of the damage Sd. The general
form of this relation can be represented by the following curve:
Sd
εresidual
εnominal
It is therefore possible to determine, for each point on the item studied, the probability of
occurrence of the load leading to the failure of the laminate with a delamination of size Sd
Knowing that the following gust occurrence probabilities are generally admitted:
We can plot the curve below associating a probability of occurrence Prat with all residual
strength levels (εresidual = k x εL.L.) such that:
5
2 x 10
9
10
εresidual
εL.L. εU.L.
This curve will in fact be compared to a lognormal type occurrence law (or a first
approximation linear law) for a larger deformation range.
Impact energy
range
Pat
Energy
Em
Detection probability mean
f f
Depth of
General visual inspection: * indent 3.3
Detailed visual inspection: ** æEö
Kf ç ÷
èeø
*
**
Pdat Energy
Pdat 1
Sd Sd
Delaminated
surface
E
Ksd
e
εresidual Energy
Prat
5
2 x 10
Prat
9
10
εresidual
εL.L. εU.L.
The inspection interval must be such that risk of failure in the interval:
Pat x (1  Pdat) x Prat ≤ 109/flight hours
Prat
Linear law
1
Lognormal law
0.5
5
2 x 10
9
10
εresidual
εmean εL.L. εU.L.
This curve, like all statistical distribution curves, is characterised by a mean value and a
standard deviation. A simple calculation enables us to obtain the following expressions:
To sum up, it is clear that by choosing a given impact energy range, the values of Pat, f,
Pdat, Sd, εresidual and Prat are implicitly determined.
The drawing above shows the links between these various quantities.
The calculation tool is based on the fundamental principle described above: all damage
susceptible to be detected during an inspection must have a probability of encountering a
load greater than its residual strength lower than 109 per flight hour (maximum value at
end of aircraft life or before last inspection).
❑ Prat: probability of occurrence of a loading case greater than the residual strength of
the impacted laminate.
The probability of having damage susceptible to encounter a load greater than its residual
strength is equivalent to the sum of the probabilities of having:
By discretizing the incident energy and therefore the type of the damage, each flaw range
can be dealt with independently of the others.
f(E)
We can therefore apply the fundamental principle to each energy interval then add the
results.
First of all we will consider an incident energy range between E and E+2 Joules.
If Pat is the probability of occurrence of the flaw per flight hour at time t1 (before first
inspection for instance) the probability of existence of the flaw is equal to: 1  (1  Pat)t1.
After the first inspection, the probability of occurrence of the flaw is therefore reduced to:
[1  (1  Pat)t1] (1  Pdat) then increases according to same curve as before but with a time
shift as initial probability is no longer zero. We repeat this operation up until the last
inspection.
The form of the function makes the calculations difficult; it is for this reason that we
compare the curve to its tangent: 1  (1  Pat)t ≈ t x Pat. This approximation remains valid
as long as the term t x Pat is small in comparison with 1.
Probability of
occurrence
of a flaw
1  (1  Pat) ^ t1
[1  (1  Pat) ^ t1] (1  Pdat)
t
IT1 t1 IT2 t2 IT3 t3 IT4 t4
Probability of
occurrence
of a flaw
IT x Pat IT x Pat
IT x Pat IT x Pat
ERL = n x IT
For constant inspection intervals, the mean probability of occurrence of the flaw is equal
to:
IT x Pat n − 1 n − i
2
+
i =1 n
å
x IT x Pat x (1 − Pdat )i
n −1
i8 Rd = IT x Pat + å IT x Pat x (1 − Pdat )
i=1
i
The mean probability of failure (Rr) of the flaw is therefore equal to:
ìïIT x Pat n − 1 n − i üï
Prat x í
ï 2
+
n
å
x IT x Pat x (1 − Pdat) i ý
ï
î i =1 þ
The maximum probability of failure (Rr) of the flaw is therefore equal to:
ìï n −1 üï
i9 Rr = Prat x íIT x Pat +
ï
å
IT x Pat x (1 − Pdat) i ý
ï
î i =1 þ
To find the mean overall risk per flight hour, all we need to do is to integrate this result into
all possible incident energy ranges.
E = 50 J ìïIT x Pat n − 1 n − i üï
å Prat x í
ï 2
+
n
å
x IT x Pat x (1 − Pdat) i ý
ï
E=0J î i=1 þ
The overall maximum risk per flight hour (Re) is equal to:
E = 50 J ìï n −1 ü
iï
i10 Re = å Prat x íIT x Pat +
ï
å IT x Pat x (1 − Pdat) ýï
E=0J î i =1 þ
The table below summarises (by giving the mathematical links between the various
variables) the method used to determine Re.
IT & ERL
02J
24J
46J i8
.
.
.
i9
.
.
.
.
.
.
.
44  46 J
46  48 J
48  50 J
i10
Re
The previous analysis can be substantiated by a static test with VID and a load level k.CL
(1 ≤ k ≤ 1.5).
❑ First method:
This method consists in initially evaluating the reduction coefficient α on the permissible
strengths of the material so that the final calculated risk Re is equal to 109 per flight hour
(this determination can only be done by successive approximations).
This means that we can suppose that the damage tolerance behaviour of the material is
degraded in relation to that really used, that is a material whose strength (under
compression loading after impact) will be equal to a certain percentage, called α, of that of
the real material.
In this case, the (εresidual; Sd) is submitted to a homothety in relation to the xaxis.
Sd
Basic curve
→ Re
Reduced curve
→ Re = 10 /fh
9
xα
εresidual
1
The number can therefore legitimately be compared to a reserve factor.
α
We will thus define a static test with VID (Visible Impact Damage) such that the margin in
relation to the residual strain ε (VID) of the flaw is the one defined above.
We obtain:
1 ε ( VID)
=
α K x εL.L.
hence:
ε ( VID)
i11 K=α = α x k ( VID)
ε L.L.
❑ Second method:
Another method would consist in directly considering the probability and load notions.
It is clear that for a static test, we can consider that the probabilities of occurrence of the
flaw (Pat) and the probabilities of detecting (Pat) and not detecting (1  Pdat) the flaw are
equal to 1 as we are sure that it is present in the item.
If we write the equivalence between the test and the maximum risk per flight hour from a
probabilistic viewpoint, we obtain: Re = Prat x PatVID x (1  PdatVID) = Prat.
The method consists therefore in determining a fictive ultimate load level such that the
probability of the flaw residual load level is equal to Re.
The drawing below shows that we must randomly subject the curve (strain level ε; Prat) to
1
a homothety with a factor so as to move point A to point B level. In this case, it appears
η
that the permissible load level of the VID has a probability of occurrence Re.
We see that this transformation also moves point A' to point B' which corresponds to the
fictive ultimate load level that must be applied to the structure.
Prat
1 Permissible
deformation of
VID
/η
5
2 x 10
9 B' A'
10
B A
Re
εresidual
εU.L. fictive εL.L. εU.L.
ε VID
By zooming in onto the part of the graph which concerns us and imposing a logarithmic
scale on the yaxis, we obtain the following representation:
 Log(Prat)
8.6 x K  3.9
B /η A
 Log(Re)
B' A'
9
4.7
ε residual
k =
1 K(VID) 1.5 ε L.L.
We obtain:
We can deduce the fictive ultimate load level to be demonstrated in the presence of VID
12 .9 x k ( VID )
i12 K =
− Log(Re ) + 3.9
The graph below represents the previous relation (the maximum risk Re per flight hour on
the xaxis and the load level K to be demonstrated on the yaxis). Each curve is relevant
to a residual load level of the flaw K.
ULTIMATE LOAD
1.5
1.4
K = 1.17 K = 1.7
1.3
k = 20
k = 1.9
Re = E15
k = 1.8
k = damage residual
k = 1.7
load level
1.2
k = 1.6
k = 1.5
k = 1.4
1.1 k = 1.3
k = 1.2
k = 1.1
LIMIT LOAD
1
1 . NOTATIONS
σ ∞x : stress to infinity
σx (y): stress along yaxis
σt (α): tangential stress around circular hole
L: plate width
∅: hole diameter
R: hole radius
2 . INTRODUCTION
The purpose of this chapter is to assess stresses at the edge of a hole without fastener on
an axially loaded composite plate and to anticipate failure of a notched laminate.
3 . GENERAL THEORY
From a theoretical point of view, the problem is formulated as follows: let an infinite plate
be subjected to stress flux σ ∞x and with the diameter hole: ∅.
The method is valid only if the xaxis is the laminate orthotropic axis. What is the stress σx
(y) distribution along the yaxis?
σx (y)
σ∞
x
σx (y = R)
∅ = 2R
This coefficient expresses hole edge stress concentration for the case of an infinitely large
plate. This is the hole coefficient.
For a composite plate, this term may be formulated as a function of the mechanical
properties of the laminate as follows:
æ E ö E
k2 K ∞T = 1 + 2 ç x
− ν xy ÷ + x
ç E ÷ G
è y ø xy
σ ∞x æ 2 4
æ 6 8
öö
k3 σx (y) = ç 2 + æç R ö÷ + 3 æç R ö÷ − K ∞ − 3 ç 5
( ) æ Rö
ç ÷
æ Rö
−7ç ÷ ÷÷
2 ç è yø è yø
T ç è yø è yø ÷÷
è è øø
2 + ξ2 + 3 ξ4
k4 σx (R + do) ≈ σ ∞x
2
R
k5 with: ξ =
R + do
σ∞ æ 2 4
æ 6 8 ö
ö
k6 σx (y) = β x ç 2 + æç R ö÷ + 3 æç R ö÷ − K ∞ − 3 ç 5 æç R ö÷ − 7 æç R ö÷ ÷ ÷
( )
2 ç è yø è yø
T ç è yø è y ø ÷ø ÷ø
è è
with:
3
æ ∅ö
2 + ç1 − ÷
è Lø
k7 β= as a first approximation
æ ∅ö
3 ç1 − ÷
è Lø
or
æ ∅ö
3 ç1 − ÷
è Lø 1 æ ∅ Mö
6
æ æ ∅ Mö
2
ö
β
1
=
æ ∅ö
3
+ ç
2è L ø
∞
(
÷ K T − 3 çç 1 −
è
) ç
è L ø
÷ ÷÷ as a second approximation
ø
2 + ç1 − ÷
è L ø
æ ∅ ö
ç 3 æç 1 − ö÷ ÷
è Lø
1− 8 ç − 1÷ −1
ç ∅
3 ÷
ç 2 + æç 1 − ö÷ ÷
è è Lø ø
in which: M2 = 2
æ ∅ö
2ç ÷
èLø
σx (y)
σ∞
x
σx (y = R)
L x
∅ = 2R
For an infinite plate, it expresses the ratio K ∞T between the loading stress to infinity σ ∞x
and the tangential normal stress at the edge σt (α) around the hole.
The position of the point is defined by the angle α with relation to the orthotropic
2
σ∞
x
σt (α)
α
x
φ
∅ = 2R
The first step consists in searching for the orthotropic axes (o, 1, 2) of the material. Angles
φ and α are thus determined (α being the angular coordinate of the point to be considered
with relation to the orthotropic coordinate system).
k8
K ∞T =
σt(α ) Eα
σx∞
=
E 1
{ [ ]
(− cos 2 φ + (k + n) sin 2 φ) k cos 2 α + (1 + n) cos 2 φ − k sin 2 φ sin 2 α −
with
E1
k9 k=
E2
Eα 1
k10 =
E1 E1 1æ E ö
sin 4 α + cos 4 α + ç 1 − 2 ν12 ÷ sin 2 2α
E2 4 è G12 ø
æE ö E
K11 n= 2 çç 1 − ν 12 ÷÷ + 1
è E2 ø G12
where E1, E2, G12 and ν12 are the mechanical properties of the laminate in the orthotropic
coordinate system (o, 1, 2).
σ t (α ) Eα
K ∞T =
σ ∞x
=
E1
{
− k cos 2 α + (1 + n) sin 2 α }
σ t (α )
K ∞T = = − cos 2 α + 3 sin 2 α
σ ∞x
For a nearlyisotropic layup and uniaxial loading, hole coefficients for 0°, 45°, 135° and
90° fibre directions are thus: 3, 1, 1 and  1.
1
1
1
1
σ∞
x
3
1
1
K∞
T
x
If the material is isotropic (or nearlyisotropic) and if the plate is infinitely large, then the
stress tensor may be formulated for any point P (identified by its coordinates r and α) on
the plate as follows:
σ ∞x æ R 2 ö σ ∞x æ 3 R4 R2 ö
k12 σr = ç 1 − ÷ + ç 1 + − 4 ÷ cos 2α
2 è r2 ø 2 è r4 r2 ø
σ ∞x æ R2 ö σ ∞ æ 3 R4 ö
σt = ç 1 + 2 ÷ − x ç 1 + 4 ÷ cos 2α
2 è r ø 2 è r ø
σ ∞x æ 3 R4 2 R2 ö
τrt = − ç 1 − 4 + 2 ÷ sin 2α
2 è r r ø
t
r
σ∞
x
P
r
α
x
∅ = 2R
This method is simple, fast but conservative. For more details, refer to chapter L
(MONOLITHIC PLATE  FASTENER HOLE) by considering the bearing load as zero.
σ∞
y
τ∞
xy
L x σ∞
x
The first step consists in calculating the principal stresses σ p∞ and σ p'∞ and in weighting
L
them with the net crosssection coefficient .
L−∅
Both stresses are then divided by coefficients Kt (K ct or K tt for direction p) and K't (K' ct or
K' tt for direction p').
These coefficients (smaller than 1) are a function of the material, the elasticity moduli in
the direction considered (p or p'), the hole diameter (∅) and the type of load (tension "t" or
compression "c"). They are found in the form of graphs (for carbon T300/914 layer in
particular) in chapter Z (sheets 3 and 4 T300/914).
σ Np
σ Fp =
Kt
σ Np'
σ Fp' =
K t'
Both stresses are expressed in the main coordinate system (o, p, p').
σN
p'
p'
σN
p
/Kt
σN
p'
/K't
L p
4.1 . Failure criterion associated with the "point stress" method (Whitney and
Nuismer)
At the composite material stress office of the Aerospatiale Design Office, one considers
("point stress" method) that there is a failure in the laminate when the longitudinal stress
of the most highly loaded fibre (located at the edge distance do) tangent to the hole is
greater than the longitudinal stress allowable for the fibre.
fibre at 0°
°
45
at
re b
fibre at 90° fi
fib
re
at
13
5°
do
σl σl
For complex loads, there is a software (PSH2 on mx4) which automatically models a finite
element mesh and finds loads in fibres that are tangent to the hole.
Longitudinal stress analysis is performed in a circle of elements, its center of gravity being
located at the hole edge distance do.
2 do
4.2 . Failure criterion associated with the "average stress" method (Whitney and
Nuismer)
This method consists in determining the average stress average σx average (ao) between
coordinate points (0, R) and (0, R + ao). It is assumed that the plate is infinitely large and
the loading uniaxial.
σx average (ao)
σ∞
x
(ao)
∅ = 2R
Based on the previous theory (see K 3.1), the following may be formulated as:
1 R + ao
σx average (ao) =
ao ò
R
σ x (y) dy
2 − ξ2 − ξ4
k14 σx average (ao) ≈ σ ∞x
2 (1 − ξ)
R
k15 with: ξ =
R + ao
This condition allows the "point stress" and the "average stress" method to become
equivalent.
The "average stress" method is rarely used at Aerospatiale, the same failure criterion as
for the "point stress" method may be applied: one considers that there is a failure in the
laminate when the longitudinal stress of the most highly loaded fibre tangent to the hole is
greater than the longitudinal stress allowable for the fibre.
After determining stresses σ Fp and σ Fp' , a smooth calculation must be performed (see
chapter C) in order to assess longitudinal stresses in fibres tangent to the hole.
B The Hill's failure criterion shall be used to each single ply (see chapter G3).
It may be noted that this method is relatively conservative because both coefficients Kt
and K't are assessed for different points, each one being the most critical with relation to
directions p and p'.
On the other hand, coefficient Kt and K't values were determined only for diameters
between ∅ 3.2 to ∅ 11.1. It is, therefore, necessary to use the theory for large diameters.
5 . Example
0°: 6 plies
45°: 4 plies
135°: 4 plies
90°: 2 plies
N ∞x = 10 daN/mm
N ∞y = 0 daN/mm
N ∞xy = 0 daN/mm
2
4
6
4
L = 120 x Nx = 10
∅ = 40
L = 120
Let's analyse, along the yaxis, the evolution of stress flux Nx (y).
The mechanical properties of the laminate in the reference coordinate system are the
following:
{k2}
æ 6256 ö 6256
K ∞T = 1 + 2ç − 0.4191÷ + = 3.28
ç 3410 ÷ 1882
è ø
This number represents the hole edge coefficient for the case of a plate of infinite width.
Since the plate does not have an infinite width L = 120 mm, we are led to calculate the
following number :
{k3}
3
æ 40 ö
2 + ç1 − ÷
β= è 120 ø = 1.148
æ 40 ö
3 ç1 − ÷
è 120 ø
We thus obtain the evolution of normal stress fluxes along the yaxis:
æ 2 4
æ 6 8 ö
ö
Nx (y) = 1.148
10 ç 2 + æç 20 ö÷ + 3 æç 20 ö÷ − (3.28 − 3) ç 5 æç 20 ö÷ − 7 æç 20 ö÷ ÷ ÷
ç ç ÷ ç ÷ ç
ç è y ø ÷ ç ÷ ÷
2
è è y ø è y ø è è y ø ÷ø ø
æ æ 20 ö
2
æ 20 ö
4
æ æ 20 ö 6 æ 20 ö ö÷ ö÷
8
ç ç
Nx (y) = 5.74 ç 2 + çç ÷÷ + 3 çç ÷÷ − 0.28 5 çç ÷÷ − 7 çç ÷÷ ÷
è y ø è y ø ç è y ø è y ø ÷ø ø
è è
40
37.65
35
30
25
Nx (y) 20
15
12.32
10 10
0
0 10 20 30 40 50 60
y
And we obtain, at the plate edge (y = 60) a flux of 12.32 daN/mm and at the hole edge
(y = 20) a flux of 37.65 daN/mm.
If one determines the flux at a hole edge distance do = 1 mm (see do in tension for the
T300/914), one gets: Nx (y = 20 + 1) = 32.47 daN/mm.
A smooth plate calculation (chapter C) with this flux makes it possible to determine the
longitudinal stress of the most highly loaded fibre (fibre at 0°): σl = 32.41 hb.
On the other hand, as the allowable longitudinal tension stress of the same fibre is equal
to Rl = 120 hb, based on the "point stress" failure criterion, we obtain:
æ 120 ö
Margin: ç − 1÷ 100 = 270 %
è 32.41 ø
At a hole edge distance do = 1 mm (see tension do for fibre T300/914 in chapter Z), flux Nx
is now only 32.47 daN/mm.
A smooth plate calculation makes it possible to find that fibres with a 0° direction are
subjected to a 32.41 hb longitudinal stress at this particular hole edge distance.
The longitudinal tensile strength of fibre T300/914 being 120 hb, the targeted margin is
thus:
æ 120 ö
Margin = 100 ç − 1÷ = 270 %
è 32 . 41 ø
0°: 6 plies
45°: 4 plies
135°: 4 plies
90°: 2 plies
N ∞x = 2.8 daN/mm
N ∞y =  7.8 daN/mm
N ∞xy = 5.3 daN/mm
Ny =  7.8
Nxy = 5.3
2
4
6
4
Nx = 2.8
x
∅ = 40
Let's determine the normal stress fluxes of the hole edge at point P (fibre at 0° tangent to
hole). To do this, we shall use the second method
First of all, (in order to eliminate the shear flux), let's be positioned in the main coordinate
system (o, p, p') which forms a 22.5° angle with the reference coordinate system (o, x, y).
Stress fluxes then become N p∞ = 5 daN/mm, N p'∞ =  10 daN/mm.
Orthotropic axes (o, 1, 2) are coincident with the reference coordinate system (o, x, y).
Np' =  10
p'
4
2
4 2
φ' = 112.5° y
6
α = 90°
P Np = 5
p
φ = 22.5°
∅ = 40
1
x
In the coordinate system (o, p, p'), the mechanical properties of the laminate are the
following:
In the reference coordinate system (o, x, y) and in the orthotropic coordinate system (o, 1,
2), the laminate properties are the following:
A first step shall consist in calculating the effect of the main flux N p∞ at point P as follows:
We have:
{k9}
6256
k= = 1.354
3410
{k10}
E 90° 1
=
E1 6256 1 æ 6256 ö
sin 4 90° + cos 4 90° + ç − 2 x 0.4191÷sin 2 2 x 90°
3410 4 è 1882 ø
E90° 1
= =1
E1 1
{k11}
æ 6256 ö 6256
n= 2ç − 0.4191÷ + = 2.48
ç 3410 ÷ 1882
è ø
{k1}
K ∞T =
σ t ( α = 90°)
=
6256
{
( − cos 2 22.5° + (1.354 + 2.48 ) sin 2 45°) 1.354 cos 2 90° +
σ p∞ 6256
((1 + 2.48 ) cos 2 22.5° − 1.354 sin 2 22.5°) sin 2 90° − 2.48 (1 + 1.354 + 2.48) sin 22.5° cos 22.5°
σ t (α = 90°)
K ∞T = = 2.773
σ p∞
p'
4
2
2
4 y
6
2.773
P Np = 5
p
1
x
A second step shall consist in calculating the effect of the main flux N p'∞ at point P.
{k1}
K' ∞T =
σ t ( α = 90°)
=
6256
{
( − cos 2 112.5° + (1.354 + 2.48 ) sin 2 45°) 1.354 cos 2 90° +
σ p∞' 6256
((1 + 2.48 ) cos 2 112.5° − 1.354 sin 2 112.5°) sin 2 90° − 2.48 (1 + 1.354 + 2.48 ) sin 112.5° cos 112.5°
σ t (α = 90°)
K' ∞T = = − 0.646
σ p∞'
Np' =  10
p'
4
2
2
4 y
6
 0.646
P
p
1
x
The deduction is that the normal stress flux tangent to the hole crossing point P is equal
to:
S.C. TAN, Finite width correction factors for anisotropic plate containing a central opening,
1988
J. Rocker, Composite material parts: Design methods at fastener holes 3 ≤ φ ≤ 100 mm.
B
Extrapolation to damage tolerance evaluation, 1998, 581.0162/98
W.L. KO, Stress concentration around a small circular hole in a composite plate, 1985,
NSA TM 86038
ERICKSON  DURELLI, Stress distribution around a circular hole in square plate, loaded
uniformly in the plane, on two opposite sides of the square, Journal of applied mechanics,
vol. 48, 1981
1 . NOTATIONS
F: bearing load
∅: fastener diameter
Sf: countersink surface of fastener
e: actual thickness of laminate
e*: thickness taken into account in bearing calculations
p: fastener pitch
N Bx
N By gross fluxes in panel
B
N xy
N Nx
N Ny net crosssection fluxes
N
N xy
N NM
N NM' net crosssection fluxes in the coordinate system specific to the bearing load
N NMM'
Nm
M additional flux due to the bearing load
NPN
net crosssection global fluxes in the main coordinate system
NPN'
α: main coordinate system angle with relation to the bearing load
N Fx
N Fy corrected final fluxes
F
N xy
2 . GENERAL/FAILURE MODES
The purpose of this chapter is to assess the structural strength of a notched and loaded
laminate fitted with fastener.
Depending on the loading level and the type of geometry, such a system may fail as per
several failure modes.
F
≥ σRm
∅e
e
∅ F ∅
F
≥ σxa
(b − ∅) e
b ∅ F
2.3
MONOLITHIC PLATE  FASTENER HOLE
General  Failure modes L 2.4
2.5
F
≥ τxya
2 (L − 0,35 ∅) e e
L
45°
F
F
≥ σya
æ ∅ö
çL − ÷ e
è 2ø L e
∅ F ∅
2F
σxa (b  ∅) + τxya L ≤
e
L e
b ∅ F ∅
4F
≥ τvisa
π ∅2
e
∅ F ∅
where:
The purpose of this subchapter is to outline the justification method of a hole with a
fastener to which is applied a bearing load in any direction, the laminate being subjected
to membrane type surrounding load fluxes and/or bending moment fluxes. The failure
mode associated with this method is a combined net crosssection failure mode in the
presence of bearing (see 2.1 and 2.2).
If the main loading is in the F1 direction, the pitch taken into account in the calculations
p1 + p2
shall be equal to: p = .
2
If the main loading is direction F2, the pitch (which is more commonly called edge
distance) taken into account in the designs shall be equal to: p = 2 p3.
For complex loading (or for simplification purposes), the following pitch value may be
used: p = mini (p1; p2; 2 p3).
It should be noted that for membrane or membrane and bending loading, pitch p is limited
to k ∅ where k depends on the material used. The value of k is generally between 4.5 and
5. For pure bending loading, this limitation does not apply.
F1
p1 p2
p3
F2
p1 + p2
p=
2
3.2.1 . Theory
l1 σ Nt + Km σm ≥ Kt σR
In the case of a membrane loaded single hole with fastener, the various justification
(broadly summed up by relationship I1) steps must be followed:
1st step: For load introduction zones (fittings, splices), the membrane gross flux NB to be
taken into account at fasteners is deduced from the constant flux to infinity N∞ by the
following relationship:
p
l2 NB = N∞ if p > 5 ∅ NB = N∞ if p ≤ 5 ∅
5∅
NB = N∞
N∞
B
N
5Ø
Flux
The drawing above shows the difference between the flux to infinity and the actual flux at
fasteners for a load introduction zone and highlights the existence of a working strip at
each fastener of a width equivalent to 5 Ø. This phenomenon is comparable to the one
described in chapter M.1.
2nd step: It consists in transforming pitch corrected gross fluxes (see previous step) into
net crosssection flux in the initial coordinate system:
F
β<0
x B p
Nx
∅
p
l3 N Nx = N Bx
Sf
p−∅−
e
p
N Ny = N By
Sf
p−∅−
e
p
N Nxy = N Bxy
Sf
p−∅−
e
y
N
Nx
F
x
β<0
Sf
∅' = ∅ +
e
Sf = b h = h2 tgθ
b
h
θ
∅'
3rd step: It consists in transforming the previously designed fluxes in the coordinate
system specific to the bearing load:
NNMM' sin β x cos β − sin β x cos β (cos β)2 − (sin β)2 NNxy
N
M' NM M
F
0
β<
x
Angle β is, in the trigonometric coordinate system, the angle leading from the Maxis
(bearing coordinate system) to the xaxis (reference coordinate system).
4th step: Once positioned in the bearing coordinate system, the flux due to the bearing
load N NM (reduced by coefficient Km), is added to (or subtracted from) flux N m
M .
Fe
l5 Nm
M =
∅ e*
The bearing height e* is voluntarily reduced for a large thickness to take into account
stress concentration at the element surface.
SINGLE SHEAR
DOUBLE SHEAR
NNM ± Nm
M Km
m
N ± Km N M
NM
l6 NNM' M' M
NNMM'
y
F
< 0
β
x
The values of Km depend on the type of loading, the bearing stress and the material used
(see chapter Z).
Three calculations shall be made with the following values:  Km; + Km; 0.
5th step: It consists in transferring fluxes so determined in their main coordinate system:
where:
1 æ 2 NNMM' ö
α= Arctg ç N ÷
è NM ± NM K m − NM' ø
m N
2
P
N
NP
M'
M
α>0
P' F
Angle α is, in the trigonometric coordinate system, the angle leading from the Maxis
(bearing coordinate system) to the Paxis (main coordinate system).
1
6th step: Fluxes are maximized by coefficient where Kt is the hole coefficient.
Kt
Kt values depend on the type of loading (tension or compression), the fastener diameter,
the mechanical properties and the material used (see chapter Z).
It should be noted that to each of both main fluxes is associated a hole coefficient which
may be different. This is why their notation differs from the sign*.
NPN
Kt
NPN'
l8
K *t
0 P
NPN
Kt
M'
M
α>0
P' F
7th step: Fluxes so maximized are recalculated in the initial coordinate system (o, x, y).
P
y
P'
F
F
Nx
αβ>0
x
NPN
NFx 2 2
(cos(β − α )) (sin(β − α )) 2 x sin(β − α ) x cos( β − α ) Kt
2 2 NPN'
l9 NFy = (sin(β − α )) (cos(β − α )) − 2 x sin(β − α ) x cos( β − α )
K *t
2 2
− sin(β − α ) x cos( β− α ) sin(β − α ) x cos( β − α ) (cos(β − α )) − (sin(β − α ))
NFxy 0
Angle (α  β) are, in the trigonometric coordinate system, the angle leading from the xaxis
(reference coordinate system) to the paxis (main coordinate system).
8th step: A smooth plate calculation is made with fluxes NF previously determined (see
chapter C) to obtain the margin.
B This software, which can be used on mx4 or PC, is simply the digital application of the
theory presented above, the eight steps being integrated into the calculation.
Let input data relating to the example covered further in this chapter be as follows.
CARACMF 1 3 4 1
2 4.8 21.6 4.91 30. 77. 40.
*3 T300 neuf
1 0.78 0.52 0.52 0.78
2 0. 45. 45. 90.
3 3.
MAT03 1 13000. 465. 465. .35 120.0 100. 5.
*2 12. 7.5 .13
*
The software gives the design margin for each value of Km, as well as all intermediate
results. To allow a quick check of loading, it represents the bearing load and main net
fluxes in the reference coordinate system.
^90
I
N2 = 20.19 I N1 = 22.19
* I *
* I *
* I *
* I * /
* I * /
* I * /
*I*/ FM = 77.
>0
I
I
I
I
I
I
I
If the notched plate is subjected to bending moment fluxes Mx, My and Mxy, follow the
additional steps described hereafter:
1st step: Determine stresses on the external and internal surfaces corresponding to
bending loads only.
external surface
σ Be
σB
l
internal surface
external surface σ Be
σB
l
internal surface
Thus, for each design direction (x, y and xy), the following stresses are obtained:
2nd step: From these stresses, "equivalent" membrane gross fluxes are evaluated.
∆neBx σ eBx
∆neBxy τ eBxy
l10 =e
∆niBx σiBx
∆niBxy τ iBxy
B
∆n e external surface
B
∆n i internal surface
3rd step: On the contrary of membrane analysis, no majoration between fluxes to infinity
N∞ and gross fluxes NB will be taken into account at load introduction areas.
NB = N∞
4th step: "Equivalent" membrane net fluxes are evaluated from "equivalent" membrane
gross fluxes.
p
l11 ∆ne Nx = ∆ne Bx
Sf
p−∅−
e
p
∆ne Ny = ∆ne By for external surface (with countersunk fastener head)
Sf
p−∅−
e
p
∆ne Nxy = ∆ne Bxy
Sf
p−∅−
e
p
∆ni Nx = ∆ni Bx
p−∅
p
∆ni Ny = ∆ni By for internal surface (no countersunk fastener head)
p−∅
p
∆ni Nxy = ∆ni Bxy
p−∅
N
∆n e external surface
N
∆n i internal surface
5th step: "Equivalent" membrane net fluxes are divided by the coefficient Kf (bending hole
coefficient) which depends on the material (in general Kf = 0.9).
∆n e Nx
l12 ∆ne Fx =
Kf
∆n e Ny
∆ne Fy = for external surface
Kf
∆n e Nxy
∆ne Fxy =
Kf
∆ni Nx
∆ni Fx =
Kf
∆niNy
∆ni Fy = for internal surface
Kf
∆n i Nxy
∆ni Fxy =
Kf
6th step: Final membrane fluxes from relation I9 are, then, added to fluxes calculated from
relation I12.
l13 N Fx + ∆ne Fx
N Fx + ∆ni Fx
The overall method for the membrane and bending analysis is summarized in the figure
here below.
N B p N B p N B p
Nx =Nx ∆n x = ∆n x Nx =Nx
Sf Sf Sf N B p N B p
p−∅− p−∅− p−∅− Nx =Nx ∆n x = ∆n x
e e e p − ∅ p − ∅
Net cross N B p N B p N B p
Ny =Ny ∆n y = ∆n y Ny =Ny N B p N B p
section Ny =Ny ∆n y = ∆n y
Sf Sf Sf
analysis p−∅− p−∅− p−∅− p − ∅ p − ∅
e e e
N B p N B p
N B p N B p N B p N xy = N xy ∆n xy = ∆n xy
N xy = N xy ∆n xy = ∆n xy N xy = N xy p − ∅ p − ∅
Sf Sf Sf
p−∅− p−∅− p−∅−
e e e
N N
NM NM
Rotation in
N N
↓ ↓ ↓
the load N M' N M'
coordinate
N N
system N MM' N MM'
β β
N m N m
N M ± Km N M N M ± Km N M
N N
↓ ↓ ↓
Addition of N M' N M'
bearing
N N
N MM' N MM'
β β
N N N
Rotation in NP NP NP
↓ ↓
the main
N N N
coordinate N P' N P' N P'
system
α αβ αβ
N N
F
∆n x F
∆n x
N ∆n x = N N ∆n x =
NP Kt NP NP Kt
Kt N Kt Kt N
Hole ∆n y ∆n y
F F
coefficient N ∆n y = N N ∆n y =
maximizing NP' Kt NP' NP' Kt
Kt N Kt Kt N
∆n xy ∆n xy
F F
α ∆n xy = αβ αβ ∆n xy =
Kt Kt
F F F
Nx Nx Nx
Rotation in
↓ ↓
the initial F F F
Ny Ny Ny
coordinate
system F F F
N xy N xy N xy
F F F F
N x + ∆n x N x + ∆n x
Addition of F F F F
N y + ∆n y N y + ∆n y
fluxes
F F F F
N xy + ∆n xy N xy + ∆n xy
3.4 . Justifications
Whatever the type of load (membrane or membrane + bending), make sure that:
 the plain monolithic plate subject to "equivalent" membrane load fluxes (NF + ∆neF) or
(NF + ∆niF) is acceptable from a structural strength point of view (refer to chapter C),
 the allowable bearing stress of material σm (which depends on the material, the
fastener diameter and the thickness to be clamped  see chapter Z) is greater or
equal to the bearing stress applied corresponding to a laminate thickness that is
smaller or equal to 1.3 ∅ for single shear or 2.6 ∅ for double shear (see subchapter
L.2.1  4th step):
This subchapter is directly related to concession processing. Here, simple rules are
outlined, that shall allow the stressman to assess the effect of a geometrical deviation,
such as a fastener diameter, its pitch or edge distance, on an initial margin.
The following paragraphs are valid only for a hole with fastener subject to membrane
fluxes.
However, for greater accuracy, it is recommended to redo the calculation or use the
B
software psg33.
Based on the theory we have just presented, any diameter change (∅ changes to ∅')
shall have an effect on:
 the net crosssection coefficient: the resulting reduction shall be equal to:
Sf'
p − ∅' −
l14 k= e
Sf
p−∅−
e
 the bearing stress: we shall assume that there is no effect on the bearing stress, even
if it tends to decrease (this assumption is conservative),
 the hole coefficient: if we assume that the hole coefficient value is in the most
unfavorable case Kt = 0.003684 ∅2  0.08806 ∅ + 0.886 (see corresponding curve in
chapter Z  material T300/914), the resulting reduction shall be equal to:
Sf'
p − ∅' −
∅ e
l15 RF' ≈ RF k k' ≈ RF
∅' p − ∅ − Sf
e
If loads are parallel to the free edge, no reduction is necessary on the reserve factor:
RF' ≈ RF
F2
p'
If loads are perpendicular to the free edge, the reduction on the reserve factor is equal to:
Sf
p' − ∅ −
l16 RF' ≈ RF e
Sf
p−∅−
e
F1
p'
If loads are parallel to the free edge, the reduction on the reserve factor is equal to:
Sf
p' − ∅ −
RF' ≈ RF e
Sf
p−∅−
e
p'
p
F2
If loads are perpendicular to the free edge, the reduction on the reserve factor is equal to:
0.54
æ p' ö
l17 RF' ≈ RF çç ÷÷ → (100 % ± 45°)
èpø
0.73
æ p' ö
RF' ≈ RF çç ÷÷ → (50 % 0; 50 % ± 45°)
èpø
1.65
æ p' ö
RF' ≈ RF çç ÷÷ → isotrope
èpø
F1
p'
p
Important remarks:
 These empirical relationships are valid only for low edge distance variations (2 ∅ ≤ p' ≤
2.5 ∅).
 For low edge distances, the fact that the failure mode described in subchapter L.2.3 is
not critical shall have to be demonstrated.
Procedure PSH2 allows the calculation of stresses in fibres around a circular hole with
fastener in a multilayer composite plate subjected to membrane type surrounding fluxes. It
is based on a finite element display of a drilled plate. Mapping calls for two separate parts:
The drilled hole is modeled by 8junction quadrangular elements and 6junction triangular
elements. The area adjacent to the hole is modeled by two rings of elements. The ring
nearest to the hole is thin and is not utilized directly on issued sheets. Issues are
presented on the second ring, the center of gravity of elements being at a design distance
from the hole corresponding to the point stress theory (do).
2 do
Contact elements between the plate and the bolt (which also simulate clearance between
the fastener and the edge distance) are of the variable stiffness type. Their stiffness is
very low when there is no contact with the plate, their stiffness is very high if there is a
contact.
3.6.2 . Justifications
 longitudinal stresses in fibres tangent to the hole edge distance (and located at a
distance do) are smaller than the longitudinal stress allowable for fibre Rl,
Fibre at 0°
°
45
at
re
Fibre at 90° Fib
Fi
br
e
at
13
5°
do
σl σl
 The allowable bearing stress of the material σm is greater or equal to the bearing
stress applied corresponding to a laminate thickness that is smaller or equal to 1.3 ∅
for single shear or 2.6 ∅ for double shear (see subchapter L.2.1  3rd step).
4 . MULTIPLE HOLES
The previous study allowed us to find the structural effect of a single hole with fastener (or
distant enough from others) on a monolithic plate subject to membrane or bending type
loads.
We shall now study the effect of several lined up holes. We shall assume that the plate is
subjected to a membrane type uniaxial load flux that is perpendicular to the row of
fasteners.
5∅ pas = 5 ∅ 5∅ 5∅
If the distance between two holes is smaller than 5 ∅, the net crosssection coefficient to
be used changes to:
5∅−d
l18
Sf
5∅−d−∅−
e
On the other hand, the hole coefficient in tension must also be modified. It changes to:
æ æ 5 ∅ − dö
2
æ5 ∅ − dö ö
l19 η Kt ≈ ç 0.065 ç ÷ − 0.65 ç ÷ + 2.625 ÷ k t (see values of η on next page)
ç è ∅ ø è ∅ ø ÷
B è ø
The hole coefficient in compression is unchanged (cf. note 440.197/84), but the
connection of the holes is ignored for the net section calculation.
These new values are to be taken into account in relationships l3 and l8.
pas = 5 ∅  d
When holes are very close to each other, the diameter ∅' envelope hole shall be
considered. The net crosssection coefficient then changes to:
pitch
l20
Sf
pitch − ∅' −
e
The hole coefficient is not modified by the number η but applies to diameter ∅'.
pitch = 4.25 ∅
∅'
1.5 ∅
Kt correction coefficient
2
E
1.9
P
P
1.8
O
1.7
L
1.6
E
V
η 1.5
N
1.4
E
1.3
U
1.2
O
R
1.1
T
1
1 1.5 2 2.5 3 3.5 4 4.5 5
5∅−d
∅
pitch pitch = 5 ∅  d
∅'
1.5 ∅
d d 5∅ pitch = 5∅ 5∅
0°: 6 plies
45°: 4 plies
135°: 4 plies
90°: 6 plies
It is subjected to the three following fluxes in the initial coordinate system (o; x; y):
N Bx = 8 daN/mm
N By =  6 daN/mm
N Bxy = 20 daN/mm
F = 185 daN
β =  30°
The fastener is a ∅ 4.8 mm countersunk head one (100° countersink angle, which
corresponds to a 4.91 mm2). The fastener pitch is 21.6 mm.
6
4
F = 185 daN
6
4 β =  30°
The purpose of the example is to determine the three final fluxes that shall be used for the
equivalent smooth plate design, which shall provide the hole margin looked for (this
calculation shall be covered in chapter C).
{l3}
21.6
NNx = 8 x = 11.59 daN / mm
4.91
21.6 − 4.8 −
2.6
21.6
NNy = ( − 6) x = − 8.69 daN / mm
4.91
21.6 − 4.8 −
2.6
21.6
NNxy = 20 x = 28.97 daN / mm
4.91
21.6 − 4.8 −
2.6
{l4}
N NM = 31.61 daN/mm
N NM' =  28.71 daN/mm
N NMM' = 5.7 daN/mm
β =  30°
185
The bearing stress is equal to: σm = = 14.82 hb
4. 8 x 2. 6
{l5}
Nm
M = 14.82 x 2.6 = 38.54 daN/mm
The flux in the load direction being a tension flux (+ 31.61 daN/mm), the value of K mt is
thus equal to 0.135 (see chapter Z  material T300/914  Sheet 2).
{l6}
K mt = 0.135
N NM = 31.61 + 0.135 x 38.54 = 36.81 daN/mm
N NM' =  28.71 daN/mm
N NMM' = 5.7 daN/mm
β =  30°
{l6}
K mt =  0.135
N NM = 31.61  0.135 x 38.54 = 26.4 daN/mm
N NM' =  28.71 daN/mm
N NMM' = 5.7 daN/mm
β =  30°
{l6}
K mt = 0
N NM = 31.61 + 0 = 31.61 daN/mm
N NM' =  28.71 daN/mm
N NMM' = 5.7 daN/mm
β =  30°
{l7}
N NP = 37.3 daN/mm
N PN' =  29.2 daN/mm
α = 4.9°
y P
M
α = 4.9°
β =  30°
x
N
N P = 37.3 daN/mm
N
N P' =  29.2 daN/mm
{l7}
N NP = 26.98 daN/mm
N PN' =  29.29 daN/mm
α = 5.8°
{l7}
N NP = 32.14 daN/mm
N PN' =  29.24 daN/mm
α = 5.4°
Angle α is the angle formed by the main coordinate system and the bearing coordinate
system.
Monolithic layup under study gives the following elasticity and shear moduli in the main
axes:
E
α + β =  34.9° E = 4470 G = 2078 = 2.151 K tt ≈ 0.6 K ct ≈ 0.87
G
E
α + β =  35.8° E = 4455 G = 2092 = 2.13 K tt ≈ 0.6 K ct ≈ 0.87
G
E
α + β =  35.4° E = 4461 G = 2086 = 2.139 K tt ≈ 0.6 K ct ≈ 0.87
G
The values are derived from chapter Z (T300/914 sheets 3 and 4).
Which gives the following new values for corrected main fluxes :
{l8}
37.3
N NP = = 62.17 daN/mm
0.6
− 29.2
N PN' = =  33.56 daN/mm
0.87
α  β = 34.9°
{l8}
26.98
N NP = = 44.97 daN/mm
0.6
− 29.29
N PN' = =  33.67 daN/mm
0.87
α  β = 35.8°
{l8}
32.14
N NP = = 53.57 daN/mm
0 .6
− 29.24
N PN' = =  33.61 daN/mm
0.87
α  β = 35.4°
{l9}
N Fx = 30.83 daN/mm
N Fy =  2.22 daN/mm
N Fxy = 44.92 daN/mm
y
N Fx = 30.83 daN/mm x
N Fy =  2.22 daN/mm
{l9}
N Fx = 18.06 daN/mm
N Fy =  6.76 daN/mm
N Fxy = 37.31 daN/mm
{l9}
N Fx = 24.32 daN/mm
N Fy =  4.36 daN/mm
N Fxy = 41.17 daN/mm
These fluxes are then used in a smooth plate design. Calculation shall be continued in
chapter C.6.
New, let's assume that, as a result of a defective drilling operation, the fastener diameter
had to be changed to a ∅ 6.35 mm with a 8.62 mm2 countersunk surface.
{l15}
8.62
21.6 − 6.35 −
4.8 2.6 = 0.92
RF' = 1.31 x x
6.35 4.91
21.6 − 4.8 −
2.6
Which corresponds to a  8 % margin, thus non allowable. However, a full manual analysis
B
(or using software PSG33) would have made it possible to find a 0 % margin.
If the calculation is conservative, it is due to the fact that the decrease of the bearing
stress corresponding to fastener oversizing was not taken into account (see chapter
L.3.5.1).
The preceding example shall also be fully covered in the composite material manual part
B
"Calculation programs" (PSG33 instructions).
Let's assume that three moment fluxes are superposed on membrane fluxes:
Mt Bx =  4 daN mm/mm
B
Mt y = 3 daN mm/mm
B
Mt xy = 5 daN mm/mm
y
β =  30°
F = 185 daN
Mt By = 3 daN
Mt Bxy = 5 daN
x Mt Bx =  4 daN
Assuming that a positive moment flux creates compression stresses on the external
surface, we obtain:
4
σe Bx = = 3.55 hb (35.5 MPa)
1.127
−3
σe By = =  2.66 hb ( 26.6 MPa)
1.127
−5
τe Bxy = =  4.44 hb ( 44.4 MPa)
1.127
−4
σi Bx = =  3.55 hb ( 35.5 MPa)
1.127
3
σi By = = 2.66 hb (26.6 MPa)
1.127
4
τi Bxy = = 4.44 hb (44.4 MPa)
1.127
EXTERNAL SURFACE
 4.44 hb
3.55 hb y
 2.66 hb x
4.44 hb
 3.55 hb
2.66 hb
INTERNAL SURFACE
The purpose of this example is to determine which bending type fluxes must be added to
membrane type fluxes for the fastener hole calculation.
The "equivalent" gross bending type fluxes necessary for the calculations thus have the
following value:
{l10}
The "equivalent" net bending type fluxes thus have the following value:
{l11}
21.6
∆ne Nx = 9.23 = 13.37 daN/mm
4.91
21.6 − 4.8 −
2.6
21.6
∆ne Ny =  6.92 =  10.02 daN/mm
4.91
21.6 − 4.8 −
2.6
21.6
∆ne Nxy =  11.54 =  16.72 daN/mm
4.91
21.6 −4.8 −
2.6
21.6
∆ni Nx =  9.23 =  11.87 daN/mm
21.6 − 4.8
21.6
∆ni Ny = 6.92 = 8.9 daN/mm
21.6 − 4.8
21.6
∆ni Nxy = 11.54 = 14.84 daN/mm
21.6 − 4.8
{l12}
13.37
∆ne Fx = = 14.86 daN/mm
0. 9
− 10.02
∆ne Fy = =  11.13 daN/mm
0 .9
− 16.72
∆ne Fxy = =  18.58 daN/mm
0 .9
− 11.87
∆ni Fx = =  13.19 daN/mm
0 .9
8. 9
∆ni Fy = = 9.89 daN/mm
0. 9
14.84
∆ni Fxy = = 16.49 daN/mm
0. 9
All prior calculations were made in the initial coordinate system (o; x; y). These
"equivalent" bending type fluxes are thus to be added to the membrane type fluxes found
in the first example (see summary table on next page).
This table summarizes the various steps of "equivalent" membrane flux calculation of the
previous example.
Data in the
8 9.23 8 8  9.23
initial
6  6.92 6 6 6.92
coordinate
20  11.54 20 20 11.54
system
Rotation in
the bearing 31.61 28.06
↓ ↓ ↓
 28.71  25.48
load 5.7 5.06
coordinate  30°  30°
system
+ Km  Km Km = 0 + Km  Km Km = 0
36.81 26.4 31.61 33.26 22.86 28.06
↓ ↓ ↓
Addition of
 28.71  28.71  28.71  25.48  25.48  25.48
bearing flux 5.7 5.7 5.7 5.06 5.06 5.06
 30°  30°  30°  30°  30°  30°
Rotation in
32.14 37.3 26.98 32.14 33.69 22.59 28.53
↓ ↓
the main
 29.24  29.2  29.29 29.24  25.91  26.01  25.95
coordinate 35.4° 34.9° 35.8° 35.4° 34.9° 36° 35.4°
system
Hole 53.57 14.86 62.17 44.97 53.57 56.15 37.65 47.55  13.19
coefficient  33.61  11.13  33.56 33.67  33.61  29.78  29.90 29.83 9.89
maximizing 35.4°  18.58 34.9° 35.8° 35.4° 34.9° 36° 35.4° 16.49
Rotation in
24.32 30.83 18.06 24.32 27.7 14.35 21.43
↓ ↓
the initial
 4.36  2.22  6.76  4.36  0.87  5.68  3.03
coordinate
41.17 44.92 37.31 41.17 38.82 30.81 35.12
system
LAFON, Carbon fibre structures: simplified rules for sizing at fastener holes, 1983, PL No.
139/83
LAFON, Justification of design methods used for carbon fibre structures  thin sheet
subject area, 1983, 440.156/83
LAFON, TROPIS, Structural strength of outer wing  justification of design values, 1989,
440.233/89
1 . NOTATIONS
2
SANDWICH  MEMBRANE / BENDING / SHEAR
Specificity  Construction principle  Design principle N 3
4
2 . SPECIFICITY
3 . CONSTRUCTION PRINCIPLE
The face sheets and core are assembled by bonding with synthetic adhesives. There are
several alternative manufacturing processes:
 multiple phase process: face sheets are cured separately, then bonding of face
sheets to the honeycomb is performed as a second operation,
 semicocuring process: the external face sheet is cured separately, the honeycomb
and the internal face sheet are then cocured on the external face sheet,
 single phase or "cocuring" process: face sheets and the honeycomb are cured in one
single operation.
4 . DESIGN PRINCIPLE
The design rules that shall be developed are derived from the classical elasticity (refer to
"distribution of load among several closely bound structural elements" in chapter A.7).
First of all, we shall consider that the three materials together are completely ordinary.
Then, we shall simplify the relationships obtained by considering that face sheets are thin
and stiff and that the sandwich core is thick and flexible.
Like monolithic metal or composite plates, sandwich plates are under the general plate
equation (see § A.7.4).
The determination of matrices (Aij), (Bij) and (Cij) which connect the strain tensor to the
load tensor is described in chapters C, D and E.
Here, we shall outline a short cut method applicable to sandwich beams. This method
does not take into account transversal loading, transversal effects socalled "Poisson"
effects and membranebending coupling. This simplification may lead to an error of
approximately 10 % on results obtained in cases of complex loading.
From the overall deformation point of view, sandwich plates obey the conventional
equations of classical elasticity theory. Stiffness equivalences (with isocrosssection) with
homogeneous beams are described by relationships n14 to n18.
 an upper skin of thickness es, of membrane elasticity modulus Ems and of equivalent
bending elasticity modulus Efs,
 a core thickness ec, of membrane elasticity modulus Emc and of equivalent bending
elasticity modulus Efc,
 a lower skin of thickness ei, of membrane elasticity modulus Emi and of equivalent
bending elasticity modulus Efi.
The bending modulus concept comes from the fact that lower and upper skins are
generally (in the case of honeycomb sandwiches) laminates with different membrane and
bending moduli (see chapters C and D). Its value depends on ply stacking. This concept
was extended to all three materials.
First of all, we shall develop the full sandwich beam theory while taking into account face
sheet thickness and bending stiffnesses, then we shall outline at the end of each sub
chapter, the simplified relationships in which face sheets shall supposedly be thin and
subject to membrane stress only.
ei 2 æ e ö æ e ö
Emi + Emc ec ç ei + c ÷ + Ems e s ç ei + ec + s ÷
2 è 2ø è 2ø
n1 zg =
Emi ei + Emc ec + Ems es
Remark: In the case of a beam in which Emc ec << Emi ei and Emc ec << Ems es, the
relationship becomes:
ei 2 æ e ö
Emi + Ems es ç ei + ec + s ÷
2 è 2ø
n2 zg =
Emi ei + Ems es
es
ec
ei
Ems Efs Gs
Emc Efc Gc
Emi Efi Gi
b
zg
We shall assume that the beam is subjected to the following overall load pattern at the
neutral axis:
Torsional moment My shall not be taken into account because it does not correspond to
any realistic loading.
The purpose of this chapter is to determine the stress and elongation diagram for each
one of these five loads.
We shall study the effects of Ny, Tx, Tz, Mx and Mz one by one.
Tz
Mx
y Mz
Ny
Tx
Assuming that all layers are in a pure tension or compression condition, a normal load Ny
applied at the neutral line results in a constant elongation over the whole crosssection.
This elongation may be formulated as follows:
Ny
n3 ε=
b (Emi ei + Emc ec + Ems es )
The equivalent membrane modulus of the sandwich beam may be determined by the
relationship n14.
Remark: In the case of a sandwich beam in which Emc ec << Emi ei and Emc ec << Ems
es, the relationship becomes:
Ny
n4 ε=
b (Emi ei + Ems es )
z
σs
Ems es y Ny
Emc ec σc
Emi ei
σi ε
b
x
By taking into account the remark assumptions of the previous page (ei << ec, es << ec,
Emc << Emi and Emc << Ems), it is possible to oversimplify load distribution in the different
sandwich layers.
We shall assume that load Ny applied at the beam neutral axis is fully picked up by two
membrane type normal loads (Fs and Fi) in both face sheets.
Emi ei
n5 Fi ≈ Ny
Emi ei + Ems es
Ems e s
Fs ≈ Ny
Emi ei + Ems e s
Fs
y Ny
Fi
Generally speaking, shear load Tx is distributed in each of the three materials in proportion
with their shear stiffness.
The maximum shear stress in each of the three layers may then be formulated as follows:
3 Tx Gs es
n6 τs =
2 b es Gs es + Gc ec + Gi ei
3 Tx Gc e c
τc =
2 b e c Gs es + Gc ec + Gi ei
3 Tx Gi ei
τi =
2 b ei Gs es + Gc ec + Gi ei
The equivalent shear modulus with relation to the xaxis may be determined by the
relationship n15.
τs
es Gs y
Tx
ec Gc τc
τi
ei Gi
b
x
where
2
b Efs es 3 æ e ö
n7 Σ El = + b Ems es ç ei + e c + s − z g ÷ +
12 è 2 ø
2
b Efc ec 3 æ e ö
+ b Emc e c ç ei + c − zg ÷ +
12 è 2 ø
2
b Efi ei3 æe ö
+ b Emi ei ç i − zg ÷
12 è2 ø
If we consider three critical points A, B and zg, moduli weighted static moments at these
points are equal to:
æ e ö
n8 EW A = b Ems es ç e i + e c + s − z g ÷
è 2 ø
2
æ e ö æe e zg ö
EW zg = b Ems es ç e i + e c + s − z g ÷ + b Em c ç i + c + ÷
è 2 ø è2 2 2ø
æ eö
EW B = b Emi ei ç z g − i ÷
è 2ø
Stress τzg corresponds to the maximum stress within the core. In the general case of a
honeycomb sandwich material, this stress is the maximum shear stress of the
honeycomb.
Stresses τA and τB correspond to shear of (adhesive bonding) interface between the core
and the skins (force sheets).
The equivalent shear modulus with relation to the zaxis may be determined by the
relationship n16.
z
τA Tz
τzg
τB
es y
ec
ei
b
x
Remark 1: In the case of a sandwich beam in which Emc ec << Emi ei and Emc ec << Ems
es, τA, τzg and τB take the following simplified from:
Tz
n10 τA ≈ τzg ≈ τB ≈
æ e e ö
b ç i + ec + s ÷
è2 2ø
z
τA Tz
τzg
τB
es y
ec
ei
b
x
Remark 2: It should be noted that the equivalent shear modulus of a thin face sheet
sandwich beam is on the same order of magnitude as the core for the
honeycomb it consists of, thus very low.
For example, for a sandwich beam simply supposed, loaded in its center, the deflected
shape due to the shear load may represent approximately 60 % of the overall deflection.
Let an aluminium beam and a sandwich beam with equivalent bending stiffness be, giving:
1 daN
300
f1
f2
P l3
f1: deflection due to the bending moment f1 =
48 E l
1.2 Pl
f2: deflection due to the shear load f 2 =
4GS
A bending moment Mx applied at the neutral line results in the creation of a linear
distribution of elongations along the crosssection. At the outer surfaces, we have:
Mx v s − Mx (ei + ec + es − z g )
n11 εs = =
Σ El Σ El
Mx v i Mx z g
εi = =
Σ El Σ El
with:
2
b Efs es 3 æ e ö
Σ El = + b Ems es ç ei + e c + s − z g ÷ +
12 è 2 ø
2
b Efc ec 3 æ e ö
+ b Emc e c ç ei + c − zg ÷ +
12 è 2 ø
2
b Efi ei3 æe ö
+ b Emi ei ç i − zg ÷
12 è2 ø
The equivalent bending modulus of the sandwich beam may be determined by the
relationship n17.
z
Mx
σs εs
y
Ems Efs es
Emc Efc ec
Emi Efi ei
σi εi
b
x
Remark: In the case of a sandwich beam in which ei << ec, es << ec, Emc << Emi and
Emc << Ems, self inertias of both face sheets and honeycomb stiffness may be
disregarded:
2 2
æ e ö æe ö
Σ El ≈ b Ems es ç ei + ec + s − zg ÷ + b Emi ei ç i − z g ÷
è 2 ø è 2 ø
We shall assume that moment Mx is fully picked up by two membrane type normal loads
(F's and F'i) in both face sheets.
Both loads have the same modulus but are opposite. Their value is equal to:
Mx
n12 F'i ≈  F's ≈
æ ei es ö
ç + ec + ÷
è2 2ø
F's
Mx
F'i
The maximum normal stress in each of the three materials may then be simply formulated
as follows:
6 Mz Ems e s
n13 σs = ±
b es Ems e s + Emc e c + Emi ei
2
6 Mz Emc e c
σc = ±
b ec Ems es + Emc e c + Emi ei
2
6 Mz Emi ei
σi = ±
b ei Ems e s + Emc e c + Emi ei
2
The equivalent bending modulus with relation to the zaxis is identical to the equivalent
membrane modulus with relation to the yaxis (see relationships n14 and n18).
σs ε
σc Mz
y
Ems es
Emc ec
Emi ei
 σi ε
b
x
For a sandwich beam, equivalences (with isocrosssection) with respect to typical loads
are the following:
e1 (3)
e2 (4)
e3
(5)
E1G1
E2G2 (1)
E3G3 (2)
å
3
Ek ek
k =1
n14 (1) E equivalent normal load =
å
3
ek
k =1
å
3
Gk ek
k =1
n15 (2) G equivalent shear load =
å
3
e
k =1 k
æ ek ö
å
3
ç ÷
k =1 G
1 è kø
n16 (3) =
å e
3
G equivalent shear load
k=1 k
å E l
3
k =1 k k
n17 (4) E equivalent bending moment =
å l
3
k =1 k
å E e
3
k =1 k k
n18 (5) E equivalent bending moment =
å e
3
k =1 k
æ e3 ö
lk: self inertia + "Steiner" inertia ç + e d2 ÷
è 12 ø
5 . EXAMPLE
We shall assume that the beam is subjected to the following two loads and moment:
 Ny = 800 daN
 Mx = 2000 daN mm
 Tz = 250 daN z
Tz = 250 daN
Mx = 2000 daN mm
y
1.04
10
Ny = 800 daN
0.9
10
The purpose of the first part of the example is to determine inner and outer surface
elongations of the beam subject to load Ny and moment Mx.
1st step: the neutral axis position has to be determined, this position being referenced with
relation to the inner surface.
{n1}
0. 9 2 æ 10 ö æ 1.04 ö
4500 + 1510 ç 0.9 + ÷ + 6000 1.04 ç 0.9 + 10 + ÷
Zg =
2 è 2 ø è 2 ø
4500 0.9 + 1510 + 6000 1.04
Zg = 7.09 mm
zg = 7.09
{n3}
800
ε= = 7612 µd (microstrain)
10 ( 4500 0.9 + 1510 + 6000 1.04 )
{n4}
800
ε= = 7774 µd the error is 2 %
10 ( 4500 0.9 + 6000 1.04 )
Ny = 800 daN
ε = 7612 µd
3rd step: To determine maximum elongations εi and εs induced by moment flux Mx.
{n11}
{n7}
2
10 6000 1.04 3 æ 1.04 ö
Σ El = + 10 6000 1.04 ç 0.9 + 10 + − 7.09 ÷ +
12 è 2 ø
2
10 15 10 3 æ 10 ö
+ 10 15 10 ç 0.9 + − 7.09 ÷ +
12 è 2 ø
2
10 4500 0.9 3 æ 0. 9 ö
+ 10 4500 0.9 ç − 7.09 ÷
12 è 2 ø
εs =  3256 µd
εi = 4761 µd
In fact, elongations (and stresses) are calculated at the center of each face sheet:
4.33
εs =  3256 =  2906 µd
4.85
6.64
εi = 4761 = 4459 µd
7.09
z
εs =  3256 µd
Mx = 2000 daN mm
y  2906 µd
4459 µd
εi = 4761 µd
{n12}
2000
F'i =  F's ≈ = 182.3 daN
0.9 1.04
+ 10 +
2 2
Which corresponds to average elongations in lower and upper face sheets equal to:
− 182.3
εs ≈ ≈  2921 µd the error is 0.5 %
10 1.04 6000
182.3
εi ≈ ≈ 4501 µd the error is 0.9 %
10 0.9 4500
z
εs = 4356 µd
Mx = 2000 daN mm
y
Ny = 800 daN
εi = 12373 µd
The second part of the example consists in calculating the evolution of shear stress due to
shear load Tz, at the neutral axis in particular, at point A (upper face sheet  honeycomb
interface) and at point B (lower face sheet  honeycomb interface).
1st step: To calculate the inertia of the elasticity moduli weighted beam.
{n7}
2 2
æ 1.04 ö æ 0 .9 ö
Σ El ≈ 10 6000 1.04 ç 0.9 + 10 + − 7.09 ÷ + 10 4500 0.9 ç − 7.09 ÷
è 2 ø è 2 ø
2nd step:
 To calculate the elasticity moduli weighted static moment EW zg (static moment with
relation to the neutral axis of part of the material located above it).
{n8}
2
æ 1.04 ö æ 0.9 10 1.04 ö
EW zg = 10 6000 1.04 ç 0.9 + 10 + − 7.09 ÷ + 10 15 ç + + ÷
è 2 ø è 2 2 2 ø
EW zg = 275538 daN mm
 To calculate the elasticity moduli weighted static moment EW A (static moment with
relation to the neutral axis at the upper face sheet).
æ 1.04 ö
EW A = 10 6000 1.04 ç 0.9 + 10 + − 7.09 ÷
è 2 ø
EW A = 270192 daN mm
 To calculate the elasticity moduli weighted static moment EW B (opposite of the static
moment with relation to the neutral axis at the lower face sheet).
æ 0.9 ö
EW B = 10 4500 0.9 ç − + 7.09 ÷
è 2 ø
EW B = 268920 daN mm
3rd step: to determine shear stresses at the neutral axis (shear stress in honeycomb), at
point A and point B.
{n9}
250 275538
τzg = = 2.31 hb (23.1 MPa)
10 2978541
250 270192
τA = = 2.26 hb (22.6 MPa)
10 2978541
250 268920
τB = = 2.25 hb (22.5 MPa)
10 2978541
It should be noted that, between point A and point B, the shear stress is practically
constant. It would be totally constant if the honeycomb elasticity modulus were zero
(which may be considered as such).
z
Tz = 250 daN
A
B y
τA = 2.26 hb
τzg = 2.31 hb
τB = 2.25 hb
{n10}
250
τA ≈ τzg ≈ τB ≈ = 2.28 hb (22.8 MPa)
æ 0.9 1.04 ö
10 ç + 10 + ÷
è 2 2 ø
The error is 2 %.
J. CHAIX, 436.127/91
BONDED JOINTS
BONDED JOINTS
Notations S 1
1/2
1 . NOTATIONS
λ: design constant
k: design constant
D: design constant
BONDED JOINTS
Notations S 1
2/2
F1i: normal load passing through material 1 (at center of step No. i)
F2i: normal load passing through material 2 (at center of step No. i)
∆Fi: normal load transferred by the adhesively bonded joint (in step No. i)
τmi: average shear stress in adhesively bonded joint (in step No. i)
τMi: maximum shear stress in adhesively bonded joint (in step No. i)
BONDED JOINTS
Bonded single joint  Highly flexible adhesive S 2.1.1
This technique consists in assembling two (or several) elements by molecular adhesion.
The adhesive must ensure load transmission.
 Single joints:
• highly flexible adhesive with respect to bonded laminates,
• general case (without cleavage effect),
• general case (with cleavage effect).
 Scarf joint.
e1 E1 Gc
ec
E2 e2
F
τ
F
τ
τm
x if E1 and E2 >> Gc
 l/2 l/2
BONDED JOINTS
General case  Without cleavage S 2.1.2
1/3
In the case of an adhesive with a very low stiffness as opposed to the stiffness of the
laminates to be assembled, shear stress may be considered as uniform and equal to:
F
s1 τm =
hxl
If τa is the allowable shear stress of the adhesive, the minimum length of the adhesively
bonded joint shall be equal to:
F
lm =
h x τ am
Fr = λ x τam x h
In practice, check that the average stress (which, in this case, is equal to the maximum
stress) is smaller or equal to τam.
F
τ
τ τM
x if E1 x e1 ≠ E2 x e2
 l/2 l/2
τ τM
x if E1 x e1 = E2 x e2
 l/2 l/2
BONDED JOINTS
General case  Without cleavage S 2.1.2
2/3
In the case of any bonded assembly (E1 x e1 > E2 x e2) (see drawing on previous page)
subjected to a normal load F, the shear stress in the adhesively bonded joint may be
formulated as follows (VOLKERSEN) :
λ x l æ cosh ( λ x x ) sinh (λ x x ) E x e1 − E 2 x e 2 ö
s2 τx = τm x çç + x 1 ÷÷
2 è sinh (λ x l / 2) cosh (λ x l / 2) E 1 x e1 + E 2 x e 2 ø
with:
G c E 1 x e1 + E 2 x e 2
s3 λ= x
e c E1 x e 1 x E 2 x e 2
and
F
τm =
hxl
In the case of an equilibrated joint, the maximum shear stress may be formulated as
follows:
λ xl æλ xlö
s4 τM = τm x x coth ç ÷
2 è 2 ø
with
2 x Gc
s5 λ=
E x e x ec
and
F
τm =
hxl
if λ x l << 0 then τM ≈ τm
λ xl
if λ x l >> 0 then τM ≈ τm x
2
BONDED JOINTS
General case  Without cleavage S 2.1.2
3/3
If τa is the allowable shear stress of the adhesive, the minimum length of the adhesively
bonded joint shall be equal to:
æ2 æ Fxλ ö ö
lm = Max ç x Arcth ç ÷; F ÷
çλ ç 2 x τa x h ÷ τa x h ÷
è è M ø m ø
æ æ λ x l ö 2 x τ aM x h ö
Fr = Min çç th ç ÷x ; l x τ am x h ÷÷
è è 2 ø λ ø
The latter relationship makes it possible to establish, for a bonded assembly, the concept
of optimum bonding length. Indeed, the function "th ( )" is asymptotically directed towards
1 when "λ x l/2" increases; now, value 1 is practically reached for a value of "λ x l/2" equal
to 2.7 (th (2.7) = 0.99).
Thus, we have:
λ x l = 2 x 2.7
hence:
5. 4 E x e x ec E x e x ec
l= = 5 .4 x = 3.82 x
λ 2 x Gc Gc
E x e x ec
s6 loptimal = 3.16 x
Gc
Fr
≈ 0.99.Fr
l
loptimal
BONDED JOINTS
General case  With cleavage S 2.1.3
F
M M
t F
Maximum shear and peeling stresses in the adhesive may, in that case, be formulated as
follows (Bruyne and Houwnik) :
λ xl æλ xl ö
s7 τM = τm x x 1 + 3 x k x coth ç x 1+ 3 x k ÷
2 è 2 ø
and
k E e F
s8 σM = σ x x 6x c x with σ =
2 E ec hxe
with
1
s9 k=
lxF l2 x F2
1+ +
2 x D 24 x D 2
and
E x t3
s10 D=
(
12 x 1 − n 2 )
BONDED JOINTS
Scarf joint S 2.1.4
1/2
2 2
æτ ö æσ ö
Check that τM ≤ τaM and that çç M ÷÷ + ç M ÷ ≤ 1.
è τaM ø è σa ø
ec; Gc e2
e1
F F
E1 α E2
In the case of an angle α, scarf joint, the average shear stress is equal to:
F x cos α
τm =
lxh
The maximum shear stress τM may be assessed using graphs on next page:
Gc æ 1 1 ö
as abscissa: λ x l with λ2 = x çç + ÷
e c è E1 x e1 E 2 x e 2 ÷ø
τm
as ordinate:
τM
BONDED JOINTS
Scarf joint S 2.1.4
2/2
E1 x e1
Each curve is representative of a value of ratio
E2 x e2
0.9
0.9
0.8
0.8
0.7
0.7
0.6
0.6
0.5
τm
0.5
τM
0.4
0.4
0.3
0.3
0.2
0.2
0.1
0.1
0.5 1 10 20 30
λl
The peeling stress in the adhesively bonded joint shall be considered as constant. It shall
be equal to the following value:
F x sin α
σm =
lxh
BONDED JOINTS
Elasticplastic behavior S 2.2
1/3
c
sti
rupture
ela
τr
τp plastic
Elastic
γ
γp γt
As long as maximum stresses at joint ends (τM) have not reached the critical value τp
(plasticizing stress of adhesive), the bonded joint behaves like a flexible joint and stress
evolution follows the rules defined in paragraph 1).
If the load increases, a plasticizing zone (with stress τp) is formed at the most highly
loaded end of the joint.
If loading is yet increased, the shear stress of the adhesive in this plasticizing zone
reaches the critical value τr (failure stress of adhesive), which causes the adhesively
bonded joint failure.
BONDED JOINTS
Elasticplastic behavior S 2.2
2/3
The drawing below illustrates, from a quality standpoint, the shear stress evolution in the
adhesively bonded joint as the bonding force increases.
τr: failure
x
l
Remark: There is no simple theory for the elasticplastic behavior of a bonded joint. A
finite element model only would allow justification of the structural strength of
such a system in this case.
BONDED JOINTS
Elasticplastic behavior S 2.2
3/3
However, in the case of an equilibrated joint and assuming that the adhesive has an
elasticplastic behavior such as described in the drawing below, it is possible to determine
(M.J. DAVIS, The development of an engineering standard for composite repairs, AGARD
SMP 1994) the length of plasticized adhesive and, of course, the length of adhesive in
elastic behavior.
τ
c
sti
ela
τp
elasticplastic
γ
γp
In the case of such behavior, the shear stress diagram in the adhesive is the following:
τ
Lp Le Lp
τp
x
L
F
Lp ≈
4 x h x τp
6 E x e x ec
Le ≈ ≈6
λ 2 x Gc
If the joint is equilibrated, the plasticized length is given by the following relationship:
æ 1 æ æ L ööö F 2 x Gc
τp ç L p − tanh ç Φ ç − + L p ÷ ÷ ÷ = with Φ=
è Φ è è 2 øøø 2 ec x E x e
BONDED JOINTS
Bonded double lap joint S 3
e1 E1 Gc
ec
x E2 2 x e2
e1 E1
For the case of a bonded double lap joint, shear stress distribution in the adhesive film is
given by the following formula (in replacement of relationship s2) :
ææ 1− β β ö ö
τx = τm x λ x l x çç çç + ÷÷ cosh (λ x x ) − (1 − β) sinh (λ x x ) ÷
÷
è è tanh (λ x l) sinh (λ x l) ø ø
−1
æ E x e2 ö
where β = çç1 + 2 ÷÷
è E1 x e1 ø
In the general case, the maximum shear stress at the joint ends is formulated as follows:
BONDED JOINTS
Bonded stepped joint S 4
1/3
When the laminates to be bonded are too thick or when the loads to be transmitted are
too high, the "stepping" or scarfing bonding technique is imperative.
The drawing below shows the general geometry of such a joint (the drawing shows a
threestepped joint (n = 3), a higher number may be considered).
The design method consists in determining, for each adhesively bonded joint portion, the
load fraction crossing it, then, in considering each step "i" as elementary.
This socalled "short cut" method is a strictly manual method which gives the order of
magnitude of average shear stresses per step. For greater accuracy, it is recommended
to use the computing software PSB2 (see § S4 and program PSB2 instructions).
Assumptions: Let's assume that transversal effects are insignificant (εy = 0 or Fy = 0). Let's
also assume that there is no secondary bending (offcentering from the neutral line shall
not be taken into account): joints below are considered as equivalent.
EQUIVALENCE
Material (2)
F Material (1)
l1 li ln
BONDED JOINTS
Bonded stepped joint S 4
2/3
1st step: Determination of loads (F1i et F2i) passing through both laminates (parent material
"1" and repair material "2") at the center of each step.
We shall assume that loads are distributed (at the center of each step) in proportion to the
rigidity of each material:
E1i x e1i
s11 F1i = F x F2i = F  F1i
E 1i x e1i + E 2i x e 2i
E2i, e2i
F2i F
F
F1i
E1i, e1i
We shall assume that the load evolution in material 1 (and consequently in material 2) is
linear by portions. Which leads to the following configuration:
F21
F
F2i
F F2n
F11
F1i
F1n
F2x
F2n
F2i
F2i
x
Evolution of the load transferred in the repair material
BONDED JOINTS
Bonded stepped joint S 4
3/3
2nd step: From the previously determined curve, the load (∆Fi) transferred by each step is
calculated.
l 2 x F1 + l1 x F2
s12 ∆F1 =
l1 + l 2
li + 1 x Fi + li x Fi + 1 li x Fi − 1 + li − 1 x Fi
∆Fi = − 2≤i≤n1
li + li + 1 li − + li
l n x Fn − 1 + ln − 1 x Fn
∆Fn = Fn −
ln − 1 + ln
å
n
We have also (∆Fi ) = F
i=1
∆F1
F
∆Fi
F ∆Fn
F2x
∆Fn
∆Fi
∆F1
x
0
We have:
∆Fi
s13 τmi = 1.05 x
h x li
BONDED JOINTS
EDP software S 5
where 1.05 is a fixed plus factor (according to the rule) allowing one to be conservative
with respect to results established by EDP software.
5 . EDP SOFTWARE
The EDP software PSB2 digitally processes problems with adhesively bonded stepped
joints and, therefore, with adhesively bonded single joints as well.
This computing program is based on a differential analysis of the adhesively bonded joint
and not on the "short cut" method outlined in chapter § S3.
 stresses in any point of a bonded stepped single or double joint (evolution of shear
stress and average stress per step),
For more information, refer to instructions for use or to the example in chapter § S6.2.
BONDED JOINTS
First example S 6.1
1/2
6 . EXAMPLES
h = 100 mm
2
Gc = 400 daN/mm
2
e1 = 2 mm E1 = 5000 daN/mm
ec = 0.1 mm 2
e2 = 2 mm
E2 = 5000 daN/mm
l = 50 mm
The allowable average shear value of the adhesive being: τam = 0.8 hb.
Assuming that the joint is subjected to load F = 1000 daN and that there is no cleavage
effect.
1000 1000
{s1} τm = = = 0.2 hb (2 MPa)
100 x 50 5000
2 x 400
{s5} λ= = 0.9
5000 x 2 x 0.1
0.9 x 50 æ 0.9 x 50 ö
{s4} τM = 0.2 x x coth ç ÷ = 4.5 hb (45 MPa)
2 è 2 ø
Check that the average stress τm is smaller than the allowable stress τam (0.8 hb; 8 MPa)
and that the maximum stress τM is smaller than τaM (8 hb; 80 MPa).
The margin thus obtained is equal to 77 % (RF = 1.77 = 8/4.5). Within the framework of
the previous example, let's calculate the optimum bonding length from which any increase
becomes useless over the decrease of maximum shear stress in the adhesive.
5000 x 2 x 0.1
{s6} loptimale = 3.16 x = 5 mm
400
BONDED JOINTS
First example S 6.1
2/2
This result proves that, concerning the maximum shear stress, a change in the bonding
length from 50 mm to 5 mm increases (after calculations) this stress by only 1 %. The gain
is thus insignificant.
1000
lm = = 12.5 mm
0.8 x 100
The drawing below shows the evolution of the actual stress (smooth curve) and the value
of the average stress (dotted curve) in the example quoted.
4.5
3.5
2.5
1.5
0.5
0
 25  20  15  10 5 0 5 10 15 20 25
BONDED JOINTS
Second example S 6.2
1/5
Let the following threestepped joint be defined by its geometry and mechanical
properties:
h = 10 mm
F = 100 daN
F = 100 daN
l1 = 15 mm l2 = 10 mm l3 = 15 mm
2
1
The allowable average shear value of the adhesive being: τam = 0.8 hb (8 MPa).
The allowable maximum shear value of the adhesive being: τaM = 8 hb (80 MPa).
We shall assume that the joint is subjected to load F = 100 daN and that there is no
cleavage effect.
The first stage consists in calculating, at the center of each step, loads passing through
each material.
Concerning the first step:
0.26 x 5250
{s11} F21 = 100 x = 20 daN
0.26 x 5250 + 0.78 x 7000
F11 = 100  20 = 80 daN
BONDED JOINTS
Second example S 6.2
2/5
0.52 x 5000
{s11} F22 = 100 x = 50 daN
0.52 x 5000 + 0.52 x 5000
0.78 x 7000
{s11} F23 = 100 x = 80 daN
0.26 x 5250 + 0.78 x 7000
The determination of these values allows the load evolution curve passing through
material 2, or repair material, to be plotted:
F21
F
F22
F F23
F11
F12
F13
F2x
F = 100 daN
F23 = 80 daN
F22 = 50 daN
F21 = 20 daN
0 x
The second stage consists in calculating, from the previous curve, loads transferred by
each step:
15 x 50 + 10 x 20
{s12} ∆F1 =  0 = 38 daN
15 + 10
15 x 50 + 10 x 80 15 x 50 + 10 x 20
{s12} ∆F2 = − = 24 daN
15 + 10 15 + 10
15 x 50 + 10 x 80
{s12} ∆F3 = 100  = 38 daN
15 + 10
BONDED JOINTS
Second example S 6.2
3/5
The drawing below represents the different loads ∆Fi transferred by each step.
∆F1
F
∆F2
F ∆F3
∆F3 = 38 daN
∆F2 = 24 daN
∆F1 = 38 daN
x
The third stage consists in determining for each step the average and maximum stresses
in the adhesively bonded joint, based on ∆Fi calculated previously.
steps 1 and 3 being equivalent for symmetry reasons, only the first two shall be justified.
{s13}
38
τm1 = 1.05 x = 0.266 hb (2.66 MPa)
10 x 15
{s13}
24
τm2 = 1.05 x = 0.252 hb (2.52 MPa)
10 x 10
and
BONDED JOINTS
Second example S 6.2
4/5
The fourth stage consists in checking that average stresses are smaller than τam.
F
F
Only a digital analysis (program PSB2) or a finite element analysis (program PSH14) shall
be able to determine with accuracy the shear stress evolution along each step.
BONDED JOINTS
Second example S 6.2
5/5
For information purposes, we present below the output file of PSB2 corresponding to the
previous example.
Basic data:
I 5) 3 3 10 1 3
F10) 10.0000000 10.0000000 ← loading
A 8)BONDED STEPPED JOINT SAMPLE
MAT1 MF 4 3
MAT1
MAT1 1001) 7000.00000 5000.00000 5250.00000 ← parent material
MAT1 2001) 3000.00000 3000.00000 3000.00000
MAT1 3001) 0. 0. 0.
MAT1 4001) .780000000 0.52000000 .260000000
MAT2P MF 4 3
MAT2P
MAT2P 1001) 5250.00000 5000.00000 7000.00000 ← repair material
MAT2P 2001) 3000.00000 3000.00000 3000.00000
MAT2P 3001) 0. 0. 0.
MAT2P 4001) .260000000 .520000000 .780000000
COLLE MF 2 3
COLLE
COLLE 1001) .050000000 .050000000 .050000000 ← adhesive
COLLE 2001) 300.000000 300.000000 300.000000
VF 3
1) 15.0000000 25.0000000 40.0000000 ← step dimensions
It may thus be observed that the short cut method provides (in this example), with respect
to the PSB2 method, a difference of:
BONDED JOINTS
References S
GAY, Composite materials, 1991
M.J. DAVIS, The development of an engineering standard for composite repairs, AGARD
SMP 1994
NASA CR 112235
NSA CR 112236
D.A. BIGWOOD A.D. CROCOMBE, Elastic analysis and engineering design formulae for
bonded joints
L.J. HART SMITH, Adhesively bonded joints for fibrous composite structures, Mc Donnell
Douglas Corporation
L.J. HART SMITH, The design of repairable advanced composite structures, Mc Donnell
Douglas Corporation 1985
L.J. HART SMITH, Adhesive bonded scraf and stepped lap joints, Mc Donnell Douglas
Corporation
J.W. VAN INGEN A. VLOT, Stress analysis of adhesively bonded single lap joint
S. MALL N.K. KOCHHAR, Criterion for mixed mode fracture in composite bonded joints,
University of MissourriRolla, 1986
M. DE NEEF, Study of composite material bonding with edge effects accounted for,
Alcatel Espace ; Août 1992
BONDED REPAIRS
BONDED REPAIRS
Notations T 1
1 . GENERAL NOTATIONS
β: angle
In principal direction p
τmi: average shear stress in adhesively bonded joint (in step No. i)
τMi: maximum shear stress in adhesively bonded joint (in step No. i)
τ'mi: average shear stress in adhesively bonded joint (in step No. i)
τ'Mi: maximum shear stress in adhesively bonded joint (in step No. i)
BONDED REPAIRS
Introduction T 2
2 . INTRODUCTION
When a panel undergoes a damage (hole, delamination, etc.), two types of repair may be
considered: a bolted repair (see chapter U) or a bonded repair.
Let the damaged (assuming that the damage is a hole) panel (monolithic skin) be
subjected to stress fluxes Nx∝, Ny∝, Nxy∝.
We shall assume that the repair is circular and of its stiffness close to that of the skin (no
increase of parent skin fluxes due to load transfer in a repair that is too stiff).
Ny∝
Nxy∝
Nx∝
BONDED REPAIRS
Analytical method T 3.1
1/6
3 . DESIGN METHOD
This is an extrapolation of the bonded joint method (see chapter S) and, therefore, it is not
suited for shear flux transfer. Thus, it is necessary to work within the principal coordinate
system in which stress fluxes are Np∝ and Np'∝ to return to the case of a single joint. This
method is conservative.
1st step: Calculation of principal fluxes Np and Np' and of main angle β.
N x ∞ + Ny ∞ 1
t1 Np∝ = + (Nx ∞ − Ny ∞ )2 + 4 Nxy ∞ 2
2 2
N x ∞ + Ny ∞ 1
t2 Np'∝ = − (N x ∞ − Ny ∞ )2 + 4 N xy ∞ 2
2 2
1 æ 2 N xy ∞ ö
t3 β= Arctg çç ÷
÷
2 è Nx∞ − Ny∞ ø
y
Np'∝
Np∝
p'
BONDED REPAIRS
Analytical method T 3.1
2/6
2nd step: For each calculation direction (p and p'), let's consider the repair as a 1 mm wide
material strip.
The drawing below shows that, based on a twodimensional repair (R), two one
dimensional bonded stepped joints (Jp) and (Jp') are determined (or isolated). Each one of
these elementary bonded joints must transfer a normal load Fp = 1 Np∝ and Fp' = 1 Np'∝.
For the determination of flux transfers from the parent material to the repair material, refer
to the design method for bonded stepped joints (see chapter S) or to the computing
software PSB2.
y
Np'∝
Np∝
J p'
Jp β
p' m
1m
p
BONDED REPAIRS
Analytical method T 3.1
3/6
3rd step:
From this analysis, the following results are extracted for each step and each direction (p
and p'):
 for direction p:
• average and maximum shear stresses in each step of the adhesively bonded joint: τmi
(average stress in step i) and τMi (maximum stress in step i),
• normal fluxes in the parent material for each step Nsi (step i),
• normal fluxes in the repair material for each step Nri (step i).
Jp
4
3
2
i=
1 p
τ Mi
τ mi
τ
p
Nsi
N Nri
Np
BONDED REPAIRS
Analytical method T 3.1
4/6
• average and maximum shear stresses in each step of the adhesively bonded joint: τ'mi
(average stress in step i) and τ'Mi (maximum stress in step i),
• normal fluxes in the parent material for each step N'si (step i),
• normal fluxes in the repair material for each step N'ri (step i).
p'
p'
J p' N' r i
N' s i
4 τ' m i
3 N p'
τ' M i
2
i=
N'
1
BONDED REPAIRS
Analytical method T 3.1
5/6
4th step: It consists of a combination of previously determined shear stresses and normal
fluxes.
 Average shear stresses in the adhesively bonded joint calculated for both directions p
and p' are vertorially combined (although points are different) and the resulting stress for
each step is compared with the allowable average shear value of the adhesive
considered.
t4 ( τmi )2 + (τ'mi )2 ≤ τ a m
τmi (p)
τ'mi (p')
adhesively bonded
joint step No. i
 Maximum shear stresses in the adhesively bonded joint calculated for both directions p
and p' are vectorially combined (although points are different) and the value found for
each step is compared with the allowable maximum shear value of the adhesive
considered.
t5 ( τMi )2 + (τ'Mi )2 ≤ τ a M
τMi (p)
τ'Mi (p')
adhesively bonded
joint step No. i
BONDED REPAIRS
Analytical method T 3.1
6/6
 In a plain plate calculation (see chapter C), normal stress fluxes Nsi and N'si for the
parent material are associated (although points are different).
This calculation shall be performed where fluxes are maximum (at the beginning of each
step).
N'si
Nsi
es
parent material
step No. i
 In a plain plate calculation (see chapter C), normal stress fluxes Nri and N'ri for the repair
material are associated (although points are different).
This calculation shall be performed where fluxes are maximum (at the end of each step).
N'ri
Nri
er
repair material
step No. i
BONDED REPAIRS
Digital method T 3.2
In the case of a highly loaded bonded repair or with complex loading, the use of finite
element modeling is preferable.
The software PSH14 has been developed for this purpose. It allows automatic modeling of
a circular bonded repair (see drawing below). This model is subjected to membrane stress
only and does not take cleavage effects into account.
The adhesively bonded joint is represented by type 29 volume elements (with elastic
plastic behavior), the panel and repair by type 80 and 83 elements.
Yaxis
507
407
same st
as
1
quadrant 307
207
Xaxis
details of 107
step
same as elements
st
1
same st
as quadrant
1
quadrant 16 7
26 17
36 27
BONDED REPAIRS
Example T 4
1/8
4 . EXAMPLE
Nxy∝ = 4 daN/mm
0°/90°
45°/135°
0°/90°
45°/135°
0°/90° 0°/90°
45°/135° 45°/135°
0°/90° 0°/90°
45°/135° 45°/135°
0°/90°
45°/135°
i=1 i=2
12 20
Let's assume (with a view to simplification) that both materials are nearlyisotropic (their
elasticity modulus being equal to 4417 daN/mm2 in all directions) for each step and that
steps are 12 and 20 mm long.
BONDED REPAIRS
Example T 4
2/8
We may deduce that the principal coordinate system has a 45° direction and that principal
fluxes are equal to Np∝ = 4 daN/mm and Np'∝ =  4 daN/mm.
{t1}
1
Np∝ = + 4 4 2 = 4 daN/mm
2
{t2}
1
Np'∝ =  4 4 2 =  4 daN/mm
2
{t3}
1 æ2x 4ö 1 1
β= Arctg ç ÷ = Arctg (∞ ) = 90 = 45°
2 è 0 ø 2 2
J
p' Jp β = 45°
p' p
BONDED REPAIRS
Example T 4
3/8
After running the software PSB2 (computation of a bonded stepped joint), the following
results are found for direction p (results may be multiplied by  1 for direction p'):
2.2 hb
τ
1.35 hb
1.21 hb
0.207 hb
0.076 hb
0
0 5 10 15 20 25 30 35
p
BONDED REPAIRS
Example T 4
4/8
3.5
2.5
Nsi 2
1.512 daN/mm
1.5
0.5
0
0 5 10 15 20 25 30 35
p
BONDED REPAIRS
Example T 4
5/8
Nri 3
2.472 daN/mm
0
0 5 10 15 20 25 30 35
p
BONDED REPAIRS
Example T 4
6/8
• The first check consists of a vectorial combination of average shear stresses for each
step. In this case, average shear stresses are the same (to the nearest sign) in both
directions p and p'.
The maximum value is equal to 0.207 daN/mm2 in each direction. We may deduce the
vectorial resultant stress:
{t4}
This value is to be compared to the allowable average shear value of the adhesively
bonded joint that is generally selected equal to 0.8 daN/mm2 (a 173 % margin is
obtained).
• The second check consists of a vectorial combination of maximum shear stresses for
each step. In this case, shear stresses are the same (to the nearest sign) in both
directions p and p'.
The maximum stress is reached at the beginning of the first step. The value reached is
equal to 2.20 daN/mm2.
{t5}
This value is to be compared to the allowable maximum shear value of the adhesively
bonded joint: 8 daN/mm2 (a 157 % margin is obtained).
BONDED REPAIRS
Example T 4
7/8
• The third check consists of a plain plate calculation of the parent material for each step
(where the flux is maximum: at the beginning of the step).
At the beginning of the first step, the flux in direction p is Ns1 = 4 daN/mm and the flux in
direction p' is N's1 =  4 daN/mm, which corresponds to a shear flux Nxys1 equal to
4 daN/mm in the reference coordinate system. At this location, the parent material is
made out of six fabrics (3 fabrics at 0°/90° + 3 fabrics at 45°/135°) G803/914 (supposed
new).
A running of the program PSB3 (plain plate computation) makes it possible to find a Hill's
criterion margin equal to 966 %.
At the beginning of the second step, the flux in direction p is Ns2 = 1.512 daN/mm and
the flux in direction p' is N's2 =  1.512 daN/mm, which corresponds to a shear flux Nxys2
equal to 1.512 daN/mm in the reference coordinate system. At this location, the parent
material is made out of two fabrics (1 fabric at 0°/90° + 1 fabric at 45°/135°) G803/914
(supposed new).
A running of program PSB3 (smooth plate computation) makes it possible to find a Hill's
criterion margin equal to 843 %.
• The fourth check consists of a smooth plate calculation of the repair material for each
step (where the flux is maximum: at the end of the step).
At the end of the first step, the flux in direction p is Nr1 = 2.472 daN/mm and the flux in
direction p' is N'r1 =  2.472 daN/mm, which corresponds to a shear flux Nxyr1 equal to
2.472 daN/mm in the reference coordinate system. At this location, the repair material is
made out of four fabrics (2 fabrics at 0°/90° + 2 fabrics at 45°/135°) G803/914 (supposed
new).
A running of the program PSB3 (plain plate computation) makes it possible to find a Hill's
criterion margin above 1000 %.
BONDED REPAIRS
Example T 4
8/8
At the end of the second step, the flux in direction p is Nr2 = 4 daN/mm and the flux in
direction p' is N'r2 =  4 daN/mm, which corresponds to a shear flux Nxyr2 equal to
4 daN/mm in the reference coordinate system. At this location, the repair material is made
out of eight fabrics (4 fabrics at 0°/90° + 4 fabrics at 45°/135°) G803/914 (supposed new).
A running of the program PSB3 (plain plate computation) makes it possible to find a Hill's
criterion margin above 1000 %.
B In conclusion, the minimum safety margin is assessed at 157 % under maximum stress in
the adhesively bonded joint.
BONDED REPAIRS
References T
BARRAU  LAROZE, Design of composite material structures, 1987
M. MAHE  D. GRIMALD, Implementation of a digital model for the finite element design of
bonded repairs on composite materials, 436.0086/95
BOLTED REPAIRS
BOLTED REPAIRS
Notations U 1
1/2
1 . NOTATIONS
∅: diameter of damage
Nx EF
meca. : flux Nx derived from E.F.s not influenced by the repair
Ny EF
meca. : flux Ny derived from E.F.s not influenced by the repair
Nxy EF
meca. : flux Nxy derived from E.F.s not influenced by the repair
∞
Nx or Nx meca . : mechanical origin flux Nx upstream of the repair
∞
Ny or Ny meca . : mechanical origin flux Ny upstream of the repair
∞
Nxy or Nxy meca . : mechanical origin flux Nxy upstream of the repair
Nxr meca. + therm.: mechanical and thermal origin flux Nx crossing the doubler
Nyr meca. + therm.: mechanical and thermal origin flux Ny crossing the doubler
Nxs meca.: mechanical origin flux Nx in the panel below the doubler
Nys meca.: mechanical origin flux Ny in the panel below the doubler
Nxys meca.: mechanical origin flux Nxy in the panel below the doubler
BOLTED REPAIRS
Notations U 1
2/2
lxi: position of row of fasteners "i" with relation to the axis of symmetry of the repair
lyj: position of row of fasteners "j" with relation to the axis of symmetry of the repair
Remark: Without an exponent, a notation looses its directional nature and thus becomes
general and applicable to x and yaxes.
BOLTED REPAIRS
Stiffness of a fastener in single shear U 2.1
2 . STIFFNESS OF FASTENERS
One of the most important parameters for the justification of a bolted repair is the stiffness
of fasteners which make it up. Their effect on load transfer is direct. The purpose of this
subchapter is to make an analytical assessment of the stiffness of a fastener.
Two cases are considered: single shear (the most common in our case) and double
shear.
Let a fastener of diameter D and longitudinal elasticity modulus E bind two parts of
thickness er and es and of elasticity moduli Er and Es.
Er
E
er Es
es
D
The stiffness of the fastener + parts system to be bound may be assessed by one of the
three relationships quoted below.
1 5 æ 1 1 ö
= + 0.8 ç + ÷ → Mac Donnel Douglas
r ED ç ÷
è Er er E s es ø
0.85 er 0.85 e s
1 2 D æ 1 3 ö 2 D æ 1 3 ö
= çç + ÷÷ + çç + ÷÷ → Boeing
r er E
è r 8 E ø es E
è s 8 E ø
2/ 3
1 æ e + er ö æ 1 1 1 1 ö
=ξç s ÷ ç + + + ÷ → Huth
r è 2D ø è Er e r E s e s 2 er E 2 e s E ø
with
ξ = 2.2 → rivet on metal joint
ξ=3 → screw on metal joint
ξ = 4.2 → carbon seal
BOLTED REPAIRS
Stiffness of a fastener in double shear  Assumptions U 2.2
3
1/2
In the software (Bolted Repairs), the stiffness of the fastener + parts system to be bound
is calculated by the Huth method:
2/3
1 æ e + er ö æ 1 1 1 1 ö
u1 = 4.2 çç s ÷÷ çç + + + ÷÷
r è 2D ø è Er er E s e s 2 er E 2 e s E ø
er
2
Er
E
Es
es
D
3 . ASSUMPTIONS
 the panel is subjected to membrane stress. The bending effects cannot be taken into
account in this method,
 the panel and doubler have a constant thickness and all fasteners are similar,
BOLTED REPAIRS
Assumptions U 3
2/2
 no overload on panel due to the presence of the doubler: to take it into account,
fluxes in the skin derived from finite element (N EF
meca. ) should be increased in
proportion to stiffnesses of the nondamaged skin alone and of the damaged skin with
reinforcing piece.
R ry + R sy E y e s hry
∞
Ny meca . = Ny EF
meca. with R sy ∅ = 0 = s
R sy ∅ = 0 Lys
∞ R rxy + R sxy
Nxy meca EF
. = Nxy meca. with R sxy∅ = 0 = Gxy s es
R sxy∅ = 0
α: x; y; xy
β: r; s
Generally speaking, a bolted repair attached to a panel subjected to three load fluxes
∞ ∞ ∞
Nx meca . , Ny meca. , Nxy meca. may be represented as follows:
∞
Ny meca.
∞
Nxy meca.
∞
Nx meca.
∞
Nx meca.
∞
Nxy meca. x
∞
Ny meca.
BOLTED REPAIRS
Geometrical data  Mechanical properties U 4
5
The justification method of such a repair shall first consist in calculating the three load
fluxes crossing the doubler (Nxr meca., Nyr meca. and Nxyr meca.) then in assessing loads applied to
the repair fasteners, based on these results. The set of fluxes at each fastener may be
determined on a unique basis.
Some geometrical and mechanical parameters of the structure shall be required for
conducting this study.
4 . GEOMETRICAL CHARACTERISTICS
Below are represented the general geometrical characteristics describing the repair. All
other geometrical characteristics appearing further in the document may be formulated
according to these characteristics.
y
2 hr
a*
x
2 hr
∅
y
er
es
x
a
5 . MECHANICAL PROPERTIES
For the panel: Longitudinal (direction x) and transversal (direction y) elasticity moduli,
shear modulus and thickness: Exs; Eys; Gxys; es.
For the doubler: Longitudinal (direction x) and transversal (direction y) elasticity moduli,
shear moduli and thickness: Exr; Eyr; Gxyr; er.
BOLTED REPAIRS
Distribution of flux Nx U 6.1
For fasteners: Number of loadcarrying fasteners (n) and elementary stiffness in shear (r)
of each fastener.
one the equivalent stiffness of the panel (Rs) and the equivalent stiffness of the doubler
(Rr) and in distributing the flux in proportion to those.
∞
If Nx meca . is the panel flux "far from the repair", the flux Nxr meca.crossing the doubler is equal
to:
∞ R rx
u2 Nxr meca. = Nx meca .
R rx + R sx
with
E xr e r hrx n r
4 Lxr
u3 R rx =
E xr e r hrx nr
+
Lxr 4
and
E x s es hsx
u4 R sx = without chamfer
Lxs
æ a* hx − a *ö
R sx = Exs es ç x + s x ÷ with chamfer
è Ls − a Ls ø
where 2 L rx is the distance between the centers of gravity of fasteners located on either
side of the damage (see shaded fasteners on drawing below). The number of columns
taken into account shall never exceed 3.
BOLTED REPAIRS
Distribution of flux Ny U 6.2
∅
hsx = hrx −
2 x
2 Ls
chamfer
x
hs
x
Lr x
center of gravity of fasteners
∞
If Ny meca . is the panel flux "far from the repair", the flux Nyr meca. crossing the doubler is equal
to:
∞ R ry
u5 Nyr meca. = Ny meca .
R ry + R sy
with
E yr e r hry n r
4 Lyr
u6 R ry =
E yr e r hry nr
+
Lyr 4
and
Ey s es hsy
u7 R sy = without chamfer
Lys
æ a hy − a ö
R sy = Eys es ç y + s y ÷ with chamfer
è Ls − a * Ls ø
BOLTED REPAIRS
Distribution of flux Nxy U 6.3
where 2 L ry is the distance between the centers of gravity of fasteners located on either
side of the damage (see shaded fasteners on drawing below). The number of columns
taken into account shall never exceed 3.
∅
hys = hry −
2
y
y Lr
2 Ls
x
chamfer
y
hs
∞
If Nxy meca . is the panel flux "far from the repair", the flux Nxyr meca. crossing the doubler is
equal to:
∞ R rxy
u8 Nxyr meca. = Nxy meca .
R rxy + R sxy
with
Gxy r er n r
u9 R rxy = 8
nr
Gxy r er +
8
and
Factor η makes it possible to take into account the effect of damage size ∅ on the panel
shear stiffness below the doubler.
BOLTED REPAIRS
Thermal origin loads in doubler U 7
V2 + W2 0.5
u11 η= 2 2
æ ö æ ö
ç ÷ ç ÷
ç V − 1+ 1 ÷ + ç W − 1+ 1 ÷
ç 1 ÷ ç 1 ÷
ç 1− ÷ ç 1− ÷
è W ø è V ø
with
2 Lxs
V=
∅
and
2 Lys
W=
∅
In the case of two geometrically different plates (1) and (2) bound by a system with
stiffness ℜ, the thermal loads applied to plate (1) are equal to:
∆θ (α 2 L 2 − α1 L1)
v9 F= (cf. § V6.1)
L1 L2 2
+ +
e1 b1 E1 e2 b 2 E2 ℜ
(1)
ℜ ℜ
(2)
L2
b1 L1 b2
By generalizing this relationship with the case of a bolted repair, we find thermal loads in
directions x and y which apply to the doubler:
BOLTED REPAIRS
Flux in panel  Loads per fastener due to Nx and Ny U 8
9
Fyr therm. =
(
2 ∆θ α y s Lys − α yr Lyr )
Lyr Lys 4
+ +
er hry E yr e s h sy Ey s nr
Fx r therm.
Nxr therm. =
2 hrx
Fyr therm.
Nyr therm. =
2 hry
Gross fluxes in the panel are deduced immediately form fluxes crossing the doubler:
∞
u13 Nxs meca. = Nx meca .  Nxr meca.  Nxr therm.
∞
Nys meca. = Ny meca .  Nyr meca.  Nyr therm.
∞
Nxys meca. = Nxy meca .  Nxyr meca.
Loads in fasteners are deduced from the geometry and from mechanical and thermal
fluxes crossing the doubler and calculated previously.
A half repair may be represented as follows. The analysis being similar for directions x
and y, indexes x and y have been removed to make the diagram as general as possible.
 straight edge,
 edge with chamfer.
BOLTED REPAIRS
Repair with 1 row of fasteners U 9.1
∞
2 hr N meca. + therm.
er Er
es Es
A a a a a a
Fr
F6 F5 F4 F3 F2 F1
F∞
Fs
∞
2 hr N meca. + therm.
er Er
es Es
A a a a a a
If the number of rows of fasteners is equal to 1, the load transmitted in the doubler is
equal to loads transmitted by all fasteners on the row (single). The load per fastener is
then deduced immediately by the relationship:
Fr 2 x Nr meca . + therm. x h r
f/fix. = =
number of fasteners number of fasteners
F1
F1
BOLTED REPAIRS
Repair with 2 rows of fasteners U 9.2
A a
F1 + F2
r2 F2 r1 F1
E2 S2 E1 S1
F
ri = ni x r
æ A 1 A ö æ A A ö FxA
F2 çç + + ÷ + F1 ç
÷ çE x S + E x S ÷÷ =
è E 2 ' x S 2 ' r2 E 2 x S 2 ø è 2' 2' 2 2 ø E2 x S2
æ − 1ö æ a a 1ö Fxa
F2 çç ÷÷ + F1 çç + + ÷÷ =
è r2 ø è E1' x S1' E 1 x S 1 r1 ø E1 x S1
BOLTED REPAIRS
Repair with 3 rows of fasteners U 9.3
A a a
F1 + F2  F3
r3 F3 r2 F2 r1 F1
E2 S2 E2 S2 E1 S1 F
æ A A 1ö æ A A ö
F3 çç + + ÷÷ + F2 çç + ÷÷ +
è E 2' x S 2' E 2 x S 2 r3 ø è E 2' x S 2' E 2 x S 2 ø
æ A A ö FxA
F1 çç + ÷÷ =
è E 2' x S 2' E 2 x S 2 ø E 2 x S 2
æ − 1ö æ a a 1ö æ a a ö Fxa
F3 çç ÷÷ + F2 çç + + ÷÷ + F1 çç + ÷÷ =
è r3 ø E
è 2' x S 2' E 2 x S 2 r 2 ø E
è 2' x S 2' E 2 x S 2 ø E 2 x S2
æ − 1ö æ a a 1ö Fxa
F2 çç ÷÷ + F1 çç + + ÷÷ =
è r2 ø è E1' x S1' E 1 x S 1 r1 ø E1 x S1
BOLTED REPAIRS
Repair with 4 rows of fasteners U 9.4
A a a a
F1 + F2 + F3 + F4
r4 F4 r3 F3 r2 F2 r1 F1
E2 S2 E2 S2 E2 S2 E1 S1 F
æ A A 1ö æ A A ö
F4 çç + + ÷÷ + F3 çç + ÷÷ +
è E 2' x S 2' E 2 x S 2 r4 ø è E 2' x S 2' E 2 x S 2 ø
æ A A ö æ A A ö FxA
F2 çç + ÷÷ + F1 çç + ÷÷ =
è E 2' x S 2 ' E 2 x S 2 ø è E 2' x S 2' E 2 x S 2 ø E 2 x S 2
æ − 1ö æ1 a a ö
F4 çç ÷÷ + F3 çç + + ÷÷ +
è r4 ø è r3 E 2' x S 2' E 2 x S 2 ø
æ a a ö æ a a ö Fxa
F2 çç + ÷÷ + F1 çç + ÷÷ =
è E 2' x S 2 ' E2 x S2 ø è E 2' x S 2' E 2 x S 2 ø E 2 x S 2
æ − 1ö æ a a 1ö æ a a ö Fxa
F3 çç ÷÷ + F2 çç + + ÷÷ + F1 çç + ÷÷ =
è r3 ø è E 2' x S 2' E 2 x S 2 r2 ø è E 2' x S 2' E 2 x S 2 ø E2 x S2
æ − 1ö æ a a 1ö Fxa
F2 çç ÷÷ + F1 çç + + ÷÷ =
è r2 ø è E1' x S1' E 1 x S 1 r1 ø E1 x S1
BOLTED REPAIRS
with a number of rows of fasteners greater than 4 U 9.5
We may easily find the previous equation system type for a number of rows of fasteners
greater than 4.
A a a a
Σ Fi
rn Fn r(n1) F(n1) r2 F2 r1 F1
E2 S2 E2 S2 E2 S2 E1 S1
F
However, we shall consider that, for a number of rows of fasteners greater than 6, the
rows greater or equal to 7 have an insignificant effect on load Fr transfer distribution (see
diagram below).
Fr
F8 = 0 F7 = 0 F6 F5 F4 F3 F2 F1
BOLTED REPAIRS
General resolution method for direction x U 9.6
Assessment of ratio k
2 x N xr meca . + therm .
x h rx
u14 k=
å
n
F
i =1 i
u15 Fxi = k x Fi
1.15 x Fxi
u16 fxi =
ni
2 x h rx
ni = (no chamfer)
a*
or
2 x h rx
ni = − 2 (with chamfer for the first row)
a*
BOLTED REPAIRS
Assessment of loads per fastener due to Nxy U 10
The fastener is identified by the letter "i" . "i" being the number of the row perpendicular to
the load. By definition, only full lines shall be considered and it shall be assumed that row
number 1 is located next to the free edge of the doubler.
Loads in the fasteners are deduced from the geometry and from the mechanical origin
fluxes crossing the doubler.
lx3
a
ly2
i=3
i=1
fy/xyij
j=2
fx/xyij y
a*
j=1
x
1≤j≤n=2 ∞
Nxy meca.
1≤i≤m=5
Loads on fasteners due to the transfer of Nxyr meca. flux are equal to:
Nxyr meca .
x a x ly j
u17 fx/xyij =
å
n
j =1
ly j
Nxyr meca .
x a * x lx i
fy/xyij =
å
m
i =1
lx i
The fastener is identified by letters "i" and "j". "i" being the number of the row parallel to
the yaxis and "j" being the number of the row parallel to the xaxis. By definition, only full
lines shall be considered and it shall be assumed that row number 1 is located next to the
free edge of the doubler.
BOLTED REPAIRS
Justifications U 11
1/2
11 . JUSTIFICATIONS
The repair justification consists of a notched and loaded plain plate calculation (for the
parent skin and for the repair) and a check of the behavior of the existing damage.
The most highly loaded fastener holes (of the four angles) must be justified by
superposing loads due to fluxes Nx meca. (fx/x), Ny meca. (fy/y) and to flux Nxy meca. (fx/xy and fy/xy).
This resulting load (f) must then be combined with fluxes N xs meca., Nys meca. and Nxys meca. (for
the initial skin) or Nxr meca., Nyr meca. and Nxyr meca. (for the repair) at the fastener considered.
Nyr Ny∝
Nxy∝
Nxyr  fx/x  fx/xy
y
fy/y  fy/xy
f
f
Nxr Nx∝
fy/xy  fy/y
x
fx/xy fx/x
On the other hand, the damage in the parent skin in the presence of fluxes Nxs meca., Nys meca.
and Nxys meca. at the repair center shall be justified.
General remark: It should be noted that there are two types of fastener arrangements: a
socalled "square" arrangement and a socalled "staggered"
arrangement. The "square" arrangement is preferred because the hole
coefficient is limited.
BOLTED REPAIRS
Justifications U 11
2/2
Both diagrams below show which value of "a" should be applied in the calculation of loads
per fastener in each of these cases.
a a
This theory was implemented with the desktop computing program "REPBOUL" (refer to
instructions).
BOLTED REPAIRS
Summary flowchart U 12
12 . SUMMARY FLOWCHART
step No. 1
INITIAL CALCULATE
DOUBLER THE STIFFNESS
LOADING
GEOMETRY OF FASTENERS
Nx∝ Ny∝ Nxy∝
r
step No. 6
ASSESS THE RESULTING
LOAD
meca. + therm.
fx/x fy/y fx/xy fy/xy
step No. 7
ASSESS THE
RESULTING
LOAD
meca. + therm.
f
BOLTED REPAIRS
Example U 13
1/11
13 . Example
Let the following repair be:
Mechanical characteristics original panel
Layup: 6/6/6/6
Exs = 4878 daN/mm2
Eys = 4878 daN/mm2
Gxys = 1882 daN/mm2
es = 3.12 mm
Doubler Mechanical characteristics
Layup: 2/4/4/2
Exr = 4008 daN/mm2
Eyr = 4008 daN/mm2
Gxyr = 2355 daN/mm2
er = 1.56 mm
Fastener characteristics
Total number n = 32
Stiffness r = 2000 daN/mm
∅ = 3.2 mm
Fastener Elasticity modulus = 7400 daN/mm2
Ny∝ =  32 daN/mm
Nxy∝ =  20 daN/mm
a = 18
A
x / xy
fA 1
y/y y / xy
fA 1 fA 1
= 108 mm
= 72 mm
A = 18
∅ = 20 mm
y
r
y
s
2L
2L
a* = 18
x
y
hs = 35 mm
x
2 L s = 72 mm
y
2 h r = 90 mm
BOLTED REPAIRS
Example U 13
2/11
Assessment of fluxes in the doubler and the panel below the doubler
This example does not include any thermal loads, we shall only cover mechanical fluxes.
Therefore, notations shall not have any index.
The first calculation step consists in determining both fluxes crossing the doubler (Nyr;
Nxyr) which entails assessing panel and doubler stiffnesses with respect to both of these
load types, first of all. The fact that flux Nx∝ is zero has the following consequence Nxr =
Nxs = 0.
{u6}
4008 x 1.56 x 45
30 x 2000
4 x 36
Rr =
y
= 5138 daN / mm
4008 x 1.56 x 45 30 x 2000
+
36 4
{u7}
4878 x 3.12 x 35
R sy = = 9864 daN / mm
54
{u9}
We have:
72
V= = 3.6
20
108
W= = 5.4
20
BOLTED REPAIRS
Example U 13
3/11
{u11}
3. 6 2 + 5. 4 2 0.5
η= 2 2
= 0.94
æ ö æ ö
ç ÷ ç ÷
ç 3.6 − 1 + 1 ÷ + ç 5.4 − 1 + 1 ÷
ç 1 ÷ ç 1 ÷
ç 1− ÷ ç 1− ÷
è 5.4 ø è 3.6 ø
{u10}
{u5}
5138
Nyr = Ny∝ = 0.34 Ny∝ = 0.34 ( 32) =  10.96 daN/mm
5138 + 9864
{u8}
2518
Nxyr = Ny∝ = 0.31 Nxy∝ = 0.31 ( 20) =  6.28 daN/mm
2518 + 5494
{u13}
BOLTED REPAIRS
Example U 13
4/11
Load due to Ny
Repair with 3 rows of fasteners: if F1, F2 and F3 corresponds to loads transmitted by the
rows of fasteners, the system displacement resolution leads to the three following
equations:
æ 18 18 1 ö
F3 çç + + ÷÷ +
è 4008 x 1.56 x 90 4878 x 3.12 x 90 5 x 2000 ø
æ 18 18 ö
F2 çç + ÷÷ +
è 4008 x 1.56 x 90 4878 x 3.12 x 90 ø
æ 18 18 ö F x 18
F1 çç + ÷÷ =
è 4008 x 1.56 x 90 4878 x 3.12 x 90 ø 4878 x 3.12 x 90
æ −1 ö æ 18 18 1 ö
F3 çç ÷÷ + F2 çç + + ÷÷ +
è 5 x 2000 ø è 4008 x 1.56 x 90 4878 x 3.12 x 90 5 x 2000 ø
æ 18 18 ö F x 18
F1 çç + ÷÷ =
è 4008 x 1 . 56 x 90 4878 x 3 .12 x 90 ø 4878 x 3.12 x 90
æ −1 ö æ 18 18 1 ö F x 18
F2 çç ÷÷ + F1 çç + + ÷÷ =
è 5 x 2000 ø è 4008 x 1.56 x 90 4878 x 3.12 x 90 5 x 2000 ø 4878. x 3.12 x 90
by assuming F = 1, loads F1, F2 and F3 per row "i" of fasteners are deduced from the
matrix resolution.
F1 = 0.1366 daN
F2 = 0.0669 daN
F3 = 0.0273 daN
BOLTED REPAIRS
Example U 13
5/11
{u14}
Assessment of ratio k
2 x ( − ) 10.96 x 45
k= = − 4275
0.1366 + 0.0669 + 0.0273
{u15]
{u16}
1.15 x − 584
fyi = = 1.15 x  116.8 =  134 daN
5
{u17}
− 6.28 x 18.54
fx/xyij = =  56.56 daN
54 + 36 + 18
− 6.28 x 18 x 36
fy/xyij = =  113.13 daN
36
(
The resultant shall be equal to 56.56 2 + (113.13 + 134 ) )
2 0.5
= 254 daN and the slope
angle of the load shall have the value Arctg ( 247.13/ 56.56) = 77° ( 180°) =  103°.
BOLTED REPAIRS
Example U 13
6/11
 on the panel
Nxs = 0 daN/mm
Nys =  32 daN/mm
Nxys =  20 daN/mm
 in the doubler
Nxr = 0 daN/mm
− 116.8
Nyr = =  6.49 daN/mm
18
fx/xy =  57 daN A
254 daN
y
BOLTED REPAIRS
Example U 13
7/11
Now, let's assume that the carbon reinforcing plate is replaced by an Aluminium plate of
thickness er = 0.84 mm (the thickness was selected so that the doubler stiffness may
remain constant: 0.84 x 7400 = 1.56 x 4008 = 6252 daN/mm) and of coefficient expansion
αr = 2.2 E5/°C.
The expansion factor of the parent skin (isotropic T300/314 laminate) is equal to: αs = 1.4
E6/°C (refer to chapter § V 4.2 for the calculation of equivalent coefficient of expansion of
a laminate).
We shall look for thermal loads for an absolute temperature of + 74° C, which corresponds
to a relative temperature with respect to the ambient temperature of ∆T = + 54° C (with a
view to simplicity, mechanical loads shall be considered as zero).
By applying the relationship {u12}, we find the thermal loads in the doubler in directions x
and y.
{u12}
In direction x :
In direction y :
2 x 54 (1.4 E − 6 x 54 − 2.2 E − 5 x 36 )
Fyr therm. =
36 54 4
+ +
0.84 x 45 x 7400 3.12 x 35 x 4878 30 x 2000
BOLTED REPAIRS
Example U 13
8/11
Thermal fluxes in the doubler are deduced by dividing the previous results by widths 2 h rx
and 2 h ry :
− 314
Nxr therm. = =  2.49 daN/mm
2 x 63
− 260
Nyr therm. = =  2.89 daN/mm
2 x 45
= 108 mm
= 72 mm
y
s
r
2L
2h
2L
= 53 mm
x
s
h
x
2 L r = 55.38 mm
x
2 L s = 72 mm
x
y
2 h r = 90 mm
BOLTED REPAIRS
Example U 13
9/11
Which allows calculation of thermal loads on the most highly loaded fasteners (those in
angles).
In direction x:
Repair with two rows of fasteners: if F1 and F2 correspond to loads transmitted by the rows
of fasteners, the system displacement resolution leads to the two following equations (see
chapter § U 9.2) :
æ 18 1 18 ö
F2 çç + + ÷÷ +
è 7400 x 126 x 0.84 6 x 2000 4878 x 126 x 3.12 ø
æ 18 18 ö F x 18
F1 çç + ÷÷ =
è 7400 x 126 x 0.84 4878 x 126 x 3.12 ø 4878 x 126 x 3.12
æ −1 ö æ 18 18 1 ö F x 18
F2 çç ÷÷ + F1 çç + + ÷÷ =
è 6 x 2000 ø è 7400 x 126 x 0.84 4878 x 126 x 3.12 7 x 2000 ø 4878 x 126 x 3.12
By assuming that F = 1, loads F1 and F2 per row "i" of fasteners are deduced from the
matrix resolution:
We find:
F1 = 0.1271 daN
F2 = 0.0455 daN
{u14}
Assessment of ratio k
2 x ( − 2.49 ) x 63
k= = − 1818
0.0455 + 0.1271
BOLTED REPAIRS
Example U 13
10/11
{u15}
{u16}
1.15 x − 231
fxi = = 1.15 x ( − 33 ) = − 37daN
7
In direction y:
For calculation in direction y, just use results for the transfer of mechanical origin fluxes
(distribution on rows of fasteners being independent from the load value  see § U 13 p.
5).
We know that:
mechanical loads:
therefore:
thermal loads:
− 2.89
Nyr therm. =  2.89 daN/mm → fyi = − 134 x = − 35 daN
− 10.96
BOLTED REPAIRS
Example U 13
11/11
fx therm. = 37 daN
fy therm. = 35 daN
Fresultant therm. = 51 daN
y
BOLTED REPAIRS
References U
BARRAU  LAROZE, Design of composite material structures, 1987
THERMAL CALCULATIONS
THERMAL CALCULATIONS
Notations V 1
1 . NOTATIONS
∆T, ∆θ: relative temperature (difference between effective and ambient temperatures)
Tather., θather.: athermane temperature
Tamb., θamb.: ambient temperature
Tstruc., θstruc.: structure temperature
L: plates length
f: force on fasteners
THERMAL CALCULATIONS
Introduction V 2
2 . INTRODUCTION
The aim of this chapter is to study the stresses and strains of thermal origin for
unidirectional fibers, composite plates, bimetallic strips and, lastly, aircraft structures,
submitted to regulation environmental conditions.
THERMAL CALCULATIONS
Hooke  Duhamel law V 3
If a material is submitted to a mechanical load, the stress  strain relation can be written in
its tensorial form:
where
éα l o où
ê ú
ê ú
(α) = ê o α ll oú
ê ú
ê ú
ëo o α lll û
If the material is submitted to a mechanical load, the stress  strain  temperature relation
can be written:
THERMAL CALCULATIONS
Behavior of unidirectional fiber V 4
where
Let us take a unidirectional fiber defined by its longitudinal direction (l) and by its
transverse direction (t), it is a transverse isotropic and orthotropic material.
t (3)
t (y)
l (x)
THERMAL CALCULATIONS
Behavior of a monolith plate V 5.1
1/3
y y
My > 0
Ny > 0
THERMAL CALCULATIONS
Behavior of a monolith plate V 5.1
2/3
Nx εx αEh x
Ny A B εy αEh y
Nxy γ xy αEh xy
v4 = ∂2wo  ∆t
Mx αEh2
∂x 2 x
∂2wo
My B C αEh2
∂y 2 y
∂2 wo
Mxy 2 αEh2
∂x ∂y xy
where the thermoelastic behavior of the laminate is described by vector (αEh) which terms
are equal to:
æ c 2 El (α l + ν tl α t ) + s 2 E t ( νlt α l + α t ) ö
å
n
v5 αEh x = ç k
( z − z k −1 ) ÷
k =1
è 1 − νlt ν tl ø
æ s 2 El (α l + ν tl α t ) + c 2 E t ( νlt α l + α t ) ö
å
n
αEh y = ç (z k − z k − 1) ÷
k =1
è 1 − νlt ν tl ø
æ c s E t ( ν lt α l + α t ) − c s E l (α l + ν tl α t ) ö
å
n
αEh xy = ç ( zk − zk − 1 ) ÷
k =1
è 1 − ν lt ν tl ø
æ z k2 − z k2 − 1 c 2 El (α l + ν tl α t ) + s 2 E t ( νlt α l + α t ) ö
å
n
αEh 2
=− ç ÷
k = 1ç 1 − νlt ν tl ÷
x
è 2 ø
THERMAL CALCULATIONS
Behavior of a monolith plate V 5.1
3/3
æ z k2 − z k2 − 1 s 2 E l (α l + ν tl α t ) + c 2 E t ( ν lt α l + α t ) ö
å
n
αEh 2
=− ç ÷
k = 1ç 1 − ν lt ν tl ÷
y
è 2 ø
æ z k2 − z 2 c s E ( ν + α + α ) − c s E (α + ν α ) ö
å ç t ÷
n
αEh =−
2 k −1 t lt l t l l tl
xy k = 1ç 2 1 − ν ν ÷
è lt tl ø
y
ply No. k ek
t l
zk zk  1
neutral plane
θ
ply No. 1 x
where:
c ≡ cos(θ) where θ is the orientation of the fiber in the basic reference frame (o, x, y)
s ≡ sin(θ) where θ is the orientation of the fiber in the basic reference frame (o, x, y)
THERMAL CALCULATIONS
Curing stresses  expansion coefficients V 5.2
5.3
For thermosetting resin composites, the plates are made by juxtaposing different layers
with different characteristics.
These layers are manufactured simultaneously to the plate. We assume that when the
resin cures, each layer is frozen in the state it is in at that time. Let Tp be the curing
temperature; the stresses in the plate can be considered as zero at this temperature.
To obtain the stresses at ambient temperature after cooling, apply relation v4.
If the plate does not have mirror symmetry, the coupling terms, (B) and αEh2 of v4 are
non zero; therefore, a uniform reduction in the temperature (from curing temperature to
ambient temperature) will create a strain in the plane and a curvature of the neutral plane.
In the same way, if there is inplane coupling (terms A16, A26 non zero), that is if plies are
not equal in + or  α, angular distortion will occur during cooling after curing, the plate will
be "parallelogram" shaped.
The equivalent expansion coefficient vector αequi. (αx equi., αy equi., αxy equi.) of an orthotropic
composite plate, with mirror symmetry without inplane coupling, can be determined by
the following relation:
where terms Aij of the laminate rigidity matrix (A) can be determined by relation c6 of
chapter C.3.
THERMAL CALCULATIONS
Bimetallic strip theory V 6.1
1/4
The aim of this subchapter is to study the mechanical influence of the temperature on a
system consisting of two beam elements with an infinitely rigid connection at their ends.
Let us take, therefore, two long plates (L >> b) with an infinitely rigid connection at their
ends and with the following mechanical characteristics:
Plate (1):
 length: L
 width: b
 thickness: e1
 modulus of elasticity: E1
 expansion coefficient: α1 (§ V 5.3)
Plate (2)
 length: L
 width: b
 thickness: e2
 modulus of elasticity: E2
 expansion coefficient: α2 (§ V 5.3)
THERMAL CALCULATIONS
Bimetallic strip theory V 6.1
2/4
(1)
(2)
L
b
Initially, let us assume that the two plates are free at their ends. If we submit each one of
them to a uniform relative temperature ∆T (in relation to the ambient temperature of the
setup), they will expand by the following lengths:
∆L1 = ∆T α1 L
where α1 ≠ α2
∆L2 = ∆T α2 L
(1) ∆L2
(2) ∆L1
Now, as these two plates are rigidly attached at their ends, they will deform by the same
length. Mechanical interaction forces of thermal origin (F1 and F2) are created:
THERMAL CALCULATIONS
Bimetallic strip theory V 6.1
3/4
(1) ∆L'2
F1 F1 F1
 F1 F2  F1
(2) ∆L'1
As the system is globally in equilibrium, the force applied to plate (1) is the same as the
one applied to plate (2) except for the sign: F1 =  F2.
F1 L
∆L'1 =
E1 e1 b1
F2 L − F1 L
∆L'2 = =
E2 e 2 L E2 e 2 b 2
∆L1
∆T (α 2 − α1)
v7 F1 =
Γ
∆T (α1 − α 2 )
F2 =
Γ
where
1 1
v8 Γ= +
E1 e1 b1 E2 e2 b 2
THERMAL CALCULATIONS
Bimetallic strip theory V 6.1
4/4
Remark: For two plates with totally different geometries attached at the ends by an
element (one or several fasteners for example) of global rigidity ℜ, the previous
relation is slightly modified.
(1)
ℜ ℜ
(2)
L2
b1 L1 b2
∆L1
2
∆L'1
∆L 2 ∆L' 2 2
2 2
∆θ (α 2 L 2 − α1 L1)
v9 F1 =
L1 L2 2
+ +
e1 b1 E1 e2 b 2 E2 ℜ
THERMAL CALCULATIONS
Bolted system V 6.2.1
6.2.1.1
1/2
6.2 . Study of the link between two parts
In practice, the link between two parts is ensured either by fasteners or by bonding.
We shall assume that the plates are sufficiently long so that the thermal expansion cannot
be absorbed by the play (and the rigidity) of the fasteners (see previous remark).
The thermal force, which will have been calculated previously by relation v4, will be
expressed herein by letter F.
Several hypotheses can be put forward concerning the number of fasteners (of rigidity r)
likely to take force F.
F
F
f
b
f=F
For information purpose, the table on the next page shows the various forces applied to
the structural elements for a splice and a doubler, these being submitted to tensile or
compression loads and at a positive or negative relative temperature.
THERMAL CALCULATIONS
Bolted system V 6.2.1.1
2/2
We will consider that the expansion coefficient of the upper material (aluminium for
instance) is higher than that of the lower material (isotropic carbon laminate for instance).
It is important to point out that the forces represented on the drawings are those applied to
the structure by the fasteners.
SPLICE DOUBLER
TENSION
COMPRESSION
0° 0°
∆T > 0
0° 0°
∆T < 0
Table V6.2.1.1: Transfer of mechanical and thermal forces to the splices and doublers
THERMAL CALCULATIONS
Bolted system V 6.2.1.2
6.2.1.3
F
F
r r
f2 f1
b
A+a
Γ+
Aar
v10 f1 = F
2A+a
Γ+
Aar
f2 = F  f1
A a a
F
F
r r r
f3 f2 f1
b
1 æA öæ 1 ö
+ A Γ + ç − 1÷ ç ÷
r èa ø è3r + a Γ r ø
2
v11 f1 = F
1 A æ1 ö
+ ç + a Γ÷
r a èr ø
1
v12 f2 = F
3+aΓr
f3 = F  f1  f2
THERMAL CALCULATIONS
Bolted system V 6.2.1.4
A a a
F
F
r r r r
f4 f3 f2 f1
b
1 æA öæ 1 ö
+ A Γ + ç − 1÷ ç 2÷
F r è a ø è3r + a Γ r ø
v13 f1 =
δ 1 A æ1 ö
+ ç + a Γ÷
r a r è ø
where
n−3
v14 δ=1+
10
THERMAL CALCULATIONS
Bonded joints V 6.2.2
1/2
For a bonded link, the (antisymmetrical) loading of the two plates is continuous from one
end of the system to the other; the main part of the loading is however at the start of the
bonded link.
L
Doubler
τMax.
 τMax.
L
2
F
τMax.
The maximum shear stress at the interface of the two plates can be written as follows:
Fλæ æ λ Lö æ λ L ö E1 e1 − E 2 e 2 ö
v15 τMax. = ç coth ç ÷ + tanh ç ÷ ÷
b è è 2 ø è 2 ø E1 e1 + E 2 e 2 ø
where
Gc E1 e1 + E2 e2
v16 λ=
e c E1 e1 E2 e2
THERMAL CALCULATIONS
Bonded joints V 6.2.2
2/2
This relation has been established by analogy with the bonded joint theory (see chapter S)
where the distribution of the shear stresses in the adhesive joint is of the symmetrical type
and where the value of this stress, although negligible, is not zero in the center.
+ +
τ≈C
+

τ=0
7.1
THERMAL CALCULATIONS
Influence of temperature V 7.2
7.2.1
7.1 . General
 thermal stresses induced by the different expansion coefficients per unit length of the
components of the composite and metallic structures (spars/skins/rib, etc.) and also
between the fibers and the resin,
 reduction of the mechanical properties especially the resin and the adhesives (certain
fibers are also sensitive to the temperature).
The demonstration of the resistance to the ultimate loads must be made in the most
penalizing association case of the ultimate temperatures of the structure combined with
selected designcritical mechanical loads.
We shall first of all define the various types of temperatures involved in the procedure
described in this chapter.
The static air temperature envelope to be considered on the ground and in flight are given
for each aircraft (DBD: Data Basis Design); they depend on changes in regulatory
requirements and aircraft operational limits.
For example, the maximum temperature to be considered on the ground was increased by
10° C between the A320 (45°) and the A340 (55°). The minimum temperature on the
ground is  54° C (see curves below).
THERMAL CALCULATIONS
Influence of temperature V 7.2.2
7.2.2.1
z
(fts)
45000
41100 ft
40000
35000
flight 30000
25000
20000
15000
12500 ft
10000
ground
5000
 80  60  54°C  40  20 0 20 40 60
 5000
OAT
(° C)
We must therefore determine the ultimate temperatures of the structure for all aircraft
flight phases (static and fatigue). This is dealt with in this subchapter.
The ambient temperature on the ground changes during the day. We will assume that its
variation is homothetic to the quantity of heat Qϕ received by the ground (see chapter §
V 7.3.1.2).
It therefore depends on the time of the day and the geographical location on earth
(latitude ζ/type of atmosphere).
The table and curves below show change of ambient temperature on ground for a tropical
atmosphere (55° C at 12 h ≡ ISA + 40° C) and for a polar atmosphere (regulatory lower
limit of  54° C).
THERMAL CALCULATIONS
Influence of temperature V 7.2.2.2
We consider however that heat builds up during the day which explains why the "night"
temperature (32° C for the tropical atmosphere) starts at 20 h and not at 18 h.
Time 6h 7h 8h 9h 10 h 11 h 12 h 13 h
Tamb. (° C) 32 38 43.5 48.3 51.9 54.2 55 54.5
Time 13 h 14 h 15 h 16 h 17 h 18 h 19 h 20 h
Tamb. (° C) 54.5 52.9 50.5 47.3 43.6 39.7 35.7 32
T amb. ground
0 t
6h 12 h 18 h 20 h 24 h
 54° C
POLAR
ATMOSPHERE
The ambient temperature in flight depends on the ambient temperature on the ground
(see previous chapter) and the altitude z. From 0 to 40000 fts (troposphere), we generally
consider that the temperature decreases on average 0.5° C for every 328 fts increase in
altitude with a lower limit of  54° C.
The diagram below gives the ambient temperature at a given altitude for all ambient
temperatures on the ground.
THERMAL CALCULATIONS
Wall temperature V 7.3
60
ISA + 40°
40
20
ISA
0
Tamb. z
(° C)
 20
 40
 54 ° C
 60
5 0 5 10 15 20 25
Altitude (x 1000 fts)
The combined effects of the solar radiation in flight (optional effect) and the speed of the
aircraft (Mach number M) lead to a significant increase in the wall temperatures when
compared with the temperature of the ambient air in flight.
THERMAL CALCULATIONS
Influence of solar radiation V 7.3.1.1
7.3.1.2
1/3
It is maximum outside the atmosphere (z ≥ 36000 fts) and is equal to 1360 w/m2. This
radiation is lower on earth due to the influence of the ozone layer, humidity and other
factors.
Between these two points, we assume that Qs varies in a linear manner as a function of
the altitude: Qsz ≈ 9.72 E3 x z + 1010 (see curve below).
1400
1360
1300
Qs z
2 1200
(w/m )
1100
1010
1000
0 5 10 15 20 25 30 35
Altitude (x 1000 fts)
Remark: These values are unchanged between ISA + 35° C and ISA + 40° C.
The quantity of heat Qϕ received by the ground depends on the quantity of heat Qs
emitted by the sun and passing through the atmosphere (Qs = 1010 w/m2) and the angle
of incidence ϕ between the light rays and the ground.
THERMAL CALCULATIONS
Influence of solar radiation V 7.3.1.2
2/3
Qs
ϕ
Qs
ϕ
Qϕ Qϕ
Ground
This angle of incidence ϕ itself depends on time t (represented by angle ω on the drawing)
and on the latitude ζ of the point under study.
ϕ = 90°
solar radiation
ζ
ω
The diagram below shows, between 6 h and 18 h the "theoretical" change in the quantity
of heat Qϕ that the ground receives for different types of atmospheres.
THERMAL CALCULATIONS
Influence of solar radiation V 7.3.1.2
3/3
Qϕ STANDARD TROPICAL
ATMOSPHERE 2 ATMOSPHERE
LATITUDE ζ ≈ 45° 1010 w/m LATITUDE ζ ≈ 0°
2
714 w/m
t
0 6h 12 h 18 h 24 h
POLAR
ATMOSPHERE
LATITUDE ζ ≈ 90°
Nevertheless, we will assume that during the night (from 18 h to 6 h), a certain quantity of
heat (≈ 280 W/m2 for a tropical atmosphere) is exchanged between the outside medium
and the structure ("night irradiation").
The table and curve below show the "regulatory" change in the quantity of heat Qϕ during
the day in tropical atmosphere and at sea level.
The curve Qϕ = Qs x cosϕ has therefore been (arbitrarily) offset at 280 w/m2 for 6 h and
18 h.
Time 7h 8h 9h 10 h 11 h 12 h
ϕ (w/m2)
Qϕ 636 838 929 980 1002 1010
Time 12 h 13 h 14 h 15 h 16 h 17 h
ϕ (w/m )
Qϕ 2
1010 1002 980 929 838 636
Qs
Q
2
1010 w/m
Qϕ
2
280 w/m Qs x cosϕ
t
0 6h 12 h 18 h 24 h
THERMAL CALCULATIONS
Influence of speed  Temperature of structure V 7.3.2
7.3.3
The effect of the speed of the aircraft (friction of the air) increases the ambient
temperature in flight to a level called the athermane temperature.
The athermane temperature (or friction temperature) is the temperature at which the
thermal flow exchange between the wall of the structure and the outside medium is zero.
To find the athermane temperature at structure stagnation point, the ambient temperature
at an altitude z must be multiplied by a coefficient which depends on the speed of the
aircraft:
æ γ −1 ö
Tath. z = Tam. z x ç 1 + Mach2 ÷
è γ ø
Cp
where γ = = 1.4 γ: ratio between molar heat capacities (perfect gas constant).
Cv
where Cp and Cv are the heat capacities of the gas (in this case, of the air at the altitude
concerned) at constant pressure and volume.
In the previous subchapters, we defined the various temperatures outside the structure
(ambient temperature on ground, ambient temperature in flight, wall temperature and
athermane temperature).
The aim of the next chapter is to determine the temperature of the various structural
elements.
THERMAL CALCULATIONS
Calculation method V 7.3.3.1
1/2
 the change of the athermane temperature which itself depends on the time, the
altitude, the speed of the aircraft and the type of atmosphere,
 the solar radiation at the altitude in question (generally not taken into account),
 the geometrical and thermal characteristics of the various elements comprising the
structure,
The calculation method consists in breaking down the structure into elements assumed to
be at a uniform temperature at time t and in writing the thermal equilibrium of each of
these elements assuming that at time t = 0 all the structure has a uniform temperature
equal to the temperature of the ambient air.
C x V x ∆T = Qa + Qc + Qi + (Qϕ  Qr)
where:
THERMAL CALCULATIONS
Calculation method  Thermal characteristics V 7.3.3.1
7.3.3.2
2/2
This method enables us to find at time t the temperature of each element and therefore to
deduce the forces and thermal stresses required for the fatigue and static justification.
Remark: The effects of the radiation of a section of the structure to another section need
not be taken into account in the calculations as this effect tends to make the
temperatures uniform.
 Conductivity
Carbon fiber
Light Tita
Nomex Drawing
alloy nium Transv.
25/25/25/25 50/20/20/10 10/20/20/50
B
 Specific heat
2 E6 2 E6
Specific heat
54 E3 2.6 E6 2.4 E6
(J/m3/° C) 22S.002.10502
440.092/92
THERMAL CALCULATIONS
Temperatures of structure on ground V 7.3.3.3
 Paints
To calculate the temperatures on the ground (or optionally in flight), the absorptivity and
emissivity coefficients of the paint must be defined.
The first step consists in determining, with software PST2, change in the temperature of
the structure on the ground during the day in order to evaluate the most critical initial flight
conditions.
The study is generally conducted over a complete day from 0 h to 24 h but can be
extended over two or three days in order to minimize the influence of the initil conditions
(ambient temperature at 0 h: 32° C at ISA + 40° C).
Several sections of the structure with different thermal and geometrical characteristics will
be modeled.
As the absoptivity coefficient corresponds to the color of the paint used and the ground
ambient temperature and insolation curves, we calculate the maximum temperature on
the ground for each structural item during a day (see drawing below).
THERMAL CALCULATIONS
Temperatures of structure in flight V 7.3.3.4
1/3
T struc. ground
TEMPERATURE OF
STRUCTURAL
ELEMENT
t
0 12 h 24 h
For this, we must define the aircraft operating scenarios. These scenarios depend on:
 distribution of the missions during the day (generally not taken into account for the
static justification),
 type of atmosphere,
 Typical mission
Several flight configurations or "missions" can be taken into account depending on the
way in which the aircraft is used.
THERMAL CALCULATIONS
Temperatures of structure in flight V 7.3.3.4
2/3
10.8
87.2
1.3
2.3
5.7
4.4
3.3
5
5
DIST.
664.9
(nm)
10.3
78.8
42.8
21.5
21.6
12.5
2.4
12
CRUISE 0.78 M
DESCENT 250 KT
HOLD 5 mm  240 kT
10000 fr
T.O. + CLIMB
5000 fr
LANDING
TAXIIN
1500 fr
A mean "typical" use is determined. For instance: for the A340, five 75' flights have been
considered distributed over 1 day as shown below:
0 400 1440
time (min)
Remark: For the static justification, we will not take into account the influence of the
previous flight on the initial conditions of the mission under study. Each flight will
be considered as isolated during the day. We shall therefore choose the most
penalizing time for the start of the mission (generally 12 h for positive
temperatures).
THERMAL CALCULATIONS
Temperatures of structure in flight V 7.3.3.4
3/3
 Definitions of atmospheres
 standard,
 tropical,
 polar.
Remark: For the fatigue analysis, it is sometimes necessary (if we do not want to be too
conservative) to use a random distribution of the type of atmosphere
encountered during one year in service. The "standard" operating time can for
instance be broken down as follows:
For the static justification, we shall choose the most penalizing atmosphere (tropical or
polar).
THERMAL CALCULATIONS
Typical  Standard
 Tropical
atmosphere  Polar
Tamb. ground Qϕ
Ambient temp. Solar radiation
Block diagram
on ground on ground
12 h 24 h 0h 12 h 24 h
 54° C
t mission Altitude
t z
Speed
M
Speed
x (1 + 0.18 Mach2)
M
t Qϕ z
Radiation in
flight
T Altitude
athermane z
T struc.
 Conduction
Daily Temperature T
 Convection
V
frequency of elements wall
 Radiation
0h 24 h t