Composite Structures 143 (2016) 189–201

Contents lists available at ScienceDirect

Composite Structures
journal homepage: www.elsevier.com/locate/compstruct

Design and manufacturing of an isogrid structure in composite

material: Numerical and experimental results
L. Sorrentino a,⇑, M. Marchetti b, C. Bellini a, A. Delfini b, M. Albano b
a
Department of Civil and Mechanical Engineering, University of Cassino and Southern Lazio, via G. Di Biasio 43, 03043 Cassino (FR), Italy
b
Department of Astronautic, Electrical and Energy Engineering, Sapienza University of Rome, via Eudossiana 18, 00184 Rome, Italy

a r t i c l e i n f o a b s t r a c t

Article history: Isogrid structures are constituted by a thin skin reinforced with a lattice structure. Such structures are
Available online 22 February 2016 adopted in aeronautical industry since they present both structural resistance and lightness. These struc-
tures were initially made of aluminium and were obtained by milling process. Nowadays, composite
Keywords: materials are used in order to improve the lightness peculiarities of such structures. The aim of this
Isogrid structures research work was to design an isogrid cylinder, made of composite material, fit to withstand a defined
Composite material axial load. The design was carried out in two step: in the former, the Vasiliev theory was used to define
Curing
the rib dimension of the lattice structure, while in the latter both material and thickness of the skin were
Complex shape mould
Compression behaviour
determined by FEM. Then, the manufacturing process was designed, paying particular attention to the
mould design and to the curing process. In fact, the curing tools and the thermal cycle strongly affect
the quality of the produced part. Finally, the designed part was produced and tested to assess the quality
of the manufacturing process and the correspondence to the design requirements.
Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction fuselage structures, inter-stage structures and payload adapters

[3]. Examples of the use of such structures can be space station
Nowadays the most advanced engineering applications require Skylab Orbital Workshop module or carriers of the family ‘‘Delta”.
the use of both advanced materials and design techniques. For Aeronautical large aircraft manufacturers such as Boeing have used
example, in the transport industry, fuel consumption and emis- such structures. Such structures are in fact designed to ensure the
sions can be reduced introducing very light parts that, however, structural integrity of thin-walled components subjected to buck-
must also be strong and withstand very high loads. An optimal ling failure modes caused by compression loads or high G loads.
solution to these problems consists in the adoption of composite Moreover, isogrid structures allow a considerable saving of mass
materials for the construction of isogrid structures. The former pre- [4,5].
sent high resistance associated with the lightness, while the latter Among the manufacture steps of polymer-matrix composite
are often used for aerospace and aeronautical structures, as they component, the most critical is the cure one. In fact, some defects
provide excellent performances for thin-walled components sub- may arise, such as low degree of cure, resin degradation and resid-
jected to buckling failure. These structures are constituted by a ual stress and deformation. The thermal conductivity of the resin is
thin cylindrical or conical wall and by an internal structure, con- so low that the heat reaching the core could be not sufficient to
sisting of circumferential and helical ribs that intersect each other cure the matrix, giving rise to low mechanical properties of the
in points called nodes. A distinctive parameter of isogrid structures component [6]. On the other hand, the heat developed from curing
is the c angle that the helical ribs form with the longitudinal axis; reaction could accumulate, provoking resin deterioration [7].
the final product is a very lightweight structure, which presents Moreover, the increase of the temperature in the inner layers
high mechanical performances. speeds up the cure, therefore the trend of the cure degree between
The isogrid structures have represented for many years now a the core and the external surface is different. This phenomenon lets
reality in the design of aircraft and space structures [1,2]. In aero- internal stresses arise since resin undergoes chemical shrinkage
nautic industry, large-scale lattice cylinders have been used for during cure, so the cure degree along the thickness of components
must be uniform to reduce curing stresses. However, residual
stresses also rise during the cooling phase (more than 50%) and
⇑ Corresponding author. are caused by CTE mismatch of fibre, matrix and mould material.
E-mail address: [email protected] (L. Sorrentino).

http://dx.doi.org/10.1016/j.compstruct.2016.02.043
0263-8223/Ó 2016 Elsevier Ltd. All rights reserved.
190 L. Sorrentino et al. / Composite Structures 143 (2016) 189–201

In fact, further stresses can arise during the cure due to differential 2.1. The Vasiliev model
strains between the part and the tool on which it is manufactured,
since aluminium or steel tools have a much higher CTE than com- The characteristic values, concurring in the Vasiliev model, of an
posite parts: they tend to stretch the parts during cooling, letting isogrid element are the curvature radius R and the height L, with
small shear stresses arise at the tool interface and causing tension the following quantities and size characteristics:
in the part [8,9].
Finally, it must be remembered that the resin viscosity dimin-
±u = 30°: the angle of the helical rib with the longitudinal axis
ishes and the matrix flows away during the cure process. There- of the structure.
fore, defects can arise, such as porosity, irregular fibres content
ac, ah: the normal distance between the circumferential rib (c)
and voids, that negatively affect the material mechanical strength, and the helical ones (h), with ac = ah = a, as will be shown later
so to wisely design the mould is very important to assure the right on.
degree of compaction.
bc, bh: thickness of the circumferential rib (c) and the helical
The aim of this research was to design an isogrid structure that ones (h), with bc = bh = b.
meets fixed resistance requirements, containing as much as possi-
Hc, Hb: height of the circumferential rib (c) and the helical ones
ble the weight. The forming technology, the necessary equipment (h), with Hc = Hb = H.
and the process parameters were also determined, because they
d: distance between the helical rib in the direction normal to
strongly affect the quality of the produced parts. In particular, the longitudinal axis of the structure.
the mould shape had to be carefully designed since the part pre-
sented a complex geometry, due to the presence of ribs. In fact, a The Vasiliev model executes the design by calculating only two
common defect that usually occurs is a bad compaction of the ribs, variables: H and b. The minimisation of the mass is obtained by
which involves porosity and low mechanical strength. The extrac- expressing the equations in terms of safety factors, each of which
tion of the part from the mould, after the cure process, must be must be greater than one to avoid the failure and instability of
guaranteed, therefore the mould shape must be wisely defined. the structure.
Furthermore, a good compaction and a high level of surface unifor-
mity must be ensured for the skin. The thermal cycle for cure pro- 2.1.1. Generality
cess must be defined in order to avoid the typical manufacturing Initially the radius R, the height L and the applied compressive
defects, such as residual stress and resin degradation due to load P must be fixed. Then the material to be used is chosen,
exothermic peak. Finally, some parts were manufactured and namely:
experimental tests were carried out in order to assess the reliabil-
ity of product/process design. At first, tests to value the degree of
Young’s modulus Eh of helical rib.
compaction and the degree of cure were effected on samples, then
Young’s modulus Ec of the circumferential rib.
a test regarding the strength of the whole structure was carried
Density qh of helical rib.
out.
Density qc of the circumferential rib.

The ultimate stress ru.
2. Isogrid structure design
The equation that describes the buckling of an isogrid structure
The structure to be designed is constituted by a cylindrical skin is:
reinforced with ribs, whose section dimensions has been designed qffiffipffiffiffiffiffiffiffiffiffi
according to the Vasiliev method [1], whose schematization is p
dH2 3
Eh Ec ¼ 1
shown in Fig. 1. Then, the numerical FEM design was used to
n0 P 2
ð1Þ

d ¼ b
choose the materials for the skin and to optimise the skin a
thickness. where n0 is the safety factor associated with the global buckling. As
regards the applied stress, it can be defined that:

3p
Dru H
d ¼ 1 ð2Þ
2ns P
where ns is the safety factor associated to the stress acting on the
structure.
The minimisation of the mass of the structure M is obtained
with ns = 1, therefore

2LP
Mopt ¼ ð2qh þ qc Þ ð3Þ
3r u
The equation that describes the local buckling in the plane tan-
gent to the point isogrid element is expressed as:

3kp3
Eh DH
d3 ¼ 1 ð4Þ
32nl P
where nl is the relative safety factor and k is the coefficient of buck-
ling, which is assumed equal to 4.
It is so found as follows:
rffiffiffi rffiffiffi
2 n0 P 1 3 Dru n0
Hopt ¼ p ffiffiffiffiffiffiffiffiffi ¼ pffiffiffiffiffiffiffiffiffi ð5Þ
3 p Eh Ec ðH
dÞopt 2 Eh Ec
Fig. 1. Skin + isogrid reinforced structure.
 L. Sorrentino et al. / Composite Structures 143 (2016) 189–201 191

pffiffiffiffiffiffiffiffiffiffiffiffi

d ¼ 2 pffiffiffi2Eh Ec P ð6Þ The mathematical expression that enables the calculation of a
3 3D2 r2u pn0 was so determined:
D
2
kP E2h Ec C ¼ 2pR ¼ 2p ¼ pD ¼ l  ne ð16Þ
nl n20 ¼ ð7Þ 2
4
54D r 5
u Therefore
in which no and nl can be easily determined by pre-assigning one of pD 2Dp
the two. l¼ ¼ ð17Þ
ne nh
The optimised value of b is:
where C is the circumference of the cylindrical element and D its
bopt ¼ a
dopt ð8Þ diameter, with:
pffiffiffi pffiffiffi pffiffiffi
pffiffiffi 3 pD 3 2Dp pD 3
3 a ¼ l  cos 30 ¼ ¼ ¼ ð18Þ
a¼ pD ð9Þ 2 ne 2 nh nh
nh
Due to its mechanical characteristics and lightweight, Carbon/
where nh is the total number of helical rib (right + left).
Epoxy material has been selected for the manufacturing of the iso-
In addition, the following relations are valid:
grid structure. This material selection has been limited to the skin
a ¼ l cos 30 ð10Þ material of the structure, and material properties considered for
Vasiliev calculation are reported in Table 1.
with ac = ah = a.
Given the radius R = 0.15 m and n (total number of rib) = 80, the
The final check is to verify that:
size of the isogrid structure then presents these features:
Hopt > bopt ð11Þ H = 5 mm, b = 2 mm, nc (circumferential ribs) = 5, l = 97.57 mm,
a = 84.5 mm, L = 0.338 m.
This condition is obtained by choosing a reasonable density of
reinforcements (nh and n).
2.2. Material selection for skin
Expressions of standard load (p) and the minimum mass nor-
malised (m) can be introduced
To determine the proper material for the skin, some numerical
4P analyse were performed varying the skin material, as denoted in
p¼ ð12Þ
pD4 Table 2.
Three materials have been selected as possible candidate for the
4M skin material. C/E skin material can allow the realisation of a
m¼ ð13Þ
pD2 L homogeneous structure with no variation in mechanical and ther-
mal characteristics. G/E skin could allow a cost saving of the struc-
So the: ‘‘universal” term of the minimised mass (miso) of an iso-
ture maintaining good mechanical characteristics. K/E skin is
grid lattice structure is:
characterised by very good mechanical characteristics but homo-
2p geneity will be lost and the final cost for the overall structure will
miso ¼ ð2qh þ qc Þ ð14Þ be higher.
3r u
All model just described is valid if and only if: 2.2.1. Mathematical model
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u As the body was symmetrical, it has been chosen to study a
u54D4 r5
P P Pt ¼ t u
ð15Þ quarter of the cylinder. 3D Cad model which represent the struc-
2
kEh Ec ture (skin and isogrid) has been meshed with the shape control cri-
teria. Rib section dimensions has been those ones calculated in
where Pt is a threshold value. paragraph 2.1.2, while a thickness of 1 mm was considered for
the skin to simplify the model. The skin thickness optimisation is
2.1.2. The calculus procedure described in paragraph 2.3. The obtained mesh is visible in Fig. 2.
For the sizing of a structural isogrid element (cylindrical) a com- Contact type between mesh elements was ‘‘bonded”, assuming
putational procedure was developed in MATLAB according to the an ideal adhesion between the skin and the isogrid structure.
method described in the previous paragraph. The procedure is Bonds have been applied along the quarter of cylinder length and
described in the following: were fixed support type, as denoted in Fig. 3. The applied load
has been expressed in Pascal and it has been applied uniformly

Load calculation of threshold Pt. on the skin surface (see Fig. 3).

The compression load applied P is set equal to that of threshold Material properties for skins have been taken from MAPTIS
(P = Pt). database. Their values are reported in Table 3. The structural

Calculation of: a, no, H, d, b, M in the optimised conditions in the
Vasiliev form.

Calculation of the normalised values of the load (p) and the Table 1
mass (m). Material properties of ribs.

Calculation of the ‘‘universal” mass minimised formula (miso).
Property Value Property Value

Calculation of the number of the circumferential rib (nc = L/a
+ 1) whose value is not an integer. Long. tensile strength [Mpa] 1823 Transv. tensile strength 50
[Mpa]

Calculation, by rounding, of the integer value of n. Long. Young modulus [GPa] 213 Transv. Young modulus 7.6

Calculation of the new value of the height L = a (n1) [GPa]
corresponding to the whole number of circumferential rib. Long. compression strength 1052 Density [kg/m3] 1480

Calculation of the change of the two values of the height L of the [Mpa]
Shear strength [Mpa] 81 Poisson coefficient 0.28
structure corresponding to n.
192 L. Sorrentino et al. / Composite Structures 143 (2016) 189–201

Table 2 risk of buckling (B), yield (Y) or failure (U) during operation. Loads
Materials of the case study. can be replaced by stresses if the load–stress relationship is linear.
Case Skin material Isogrid material The safety factor for composites is set at 1.25 as required from the
1 Carbon/epoxy Carbon/epoxy
ECSS-ST-32-10C. For minimum weight design, the MoS should be as
2 Glass/epoxy Carbon/epoxy small as practicable. The percentage utilisation of the failure criteria
3 Kevlar/epoxy Carbon/epoxy for composite materials according to some authors is summarised
in the Fig. 4 [10].
In this study the maximum stress criteria is adopted for the
properties have been considered identical for both longitudinal composite components and for brittle materials. For metallic com-
and transversal directions, since all the prepregs reinforcements ponents, like the springs, the Von Mises criteria is generally used
were balanced fabrics. [11–13].

2.2.2. Analysis method 2.2.3. Analysis of results

The numerical analysis has been carried out following the Max- The analysis of results were performed to choose the materials
imum stress criteria, according to the theory associated with Rank- of the skin. In the table reported below the results of the analysed
ine. According to this theory, failure will occur when the maximum cases with a load of 450 N are reported in Table 4.
principal stress in a system reaches the value of the maximum Note that the allowable stress considered for MoS calculation is
strength at elastic limit in simple tension. For the two dimensional relevant to the material that is used in the area where the maxi-
stress case this is obtained from the formula below. mum stress is concentrated.
All the analysed cases shows a good resistance of the structure
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 1 as all the MoS values are positive.
r1 ¼ ðrx  ry Þ þ ðrx  ry Þ2 þ 4s2xy ð19Þ
2 2 In most cases the central rib of the structure is more stressed
The design Factor of Safety (FoS) for the two dimensional case is and the deformation is out of plane as shown in the Fig. 5.
the ratio between the elastic limit from tensile test and the highest The isogrid lattice manufactured with C/E composite shows a
principal stress. Ply failure is predicted when the stress in one of good resistance if the skin is made of G/E as, although the deforma-
the material principal directions (longitudinal, transverse and tion is of 1 mm, stresses are limited. With a Kevlar skin, the stres-
in-plane shear) reaches the maximum allowable value. The ses would be too high for the overall structure. In the case of both
limitation of the maximum stress criteria is that no interaction is isogrid lattice and skin made of C/E composite, the structure
considered between stresses in different directions; in this weight would be low and there would be a uniform thermal and
way, the strength of the ply for multiaxial load states is electrical behaviour of the overall structure. Therefore, C/E com-
over-predicted; these relations are extensively used in the industry posite has been chosen for both rib and skin.
because of their simplicity:
2.3. Skin thickness optimisation
Fibre : r11 P ru11T or r11 6 ru11C
Matrix : r22 P ru22T or r22 6 ru22C ð20Þ For the skin optimisation, the model taken into account was the
Shear : =r12 = P ru12 overall cylinder structure. The mathematical model has been rea-
lised with the same criteria of the material selection study.
The analysis has been performed using safety margins called The applied load on the skin and isogrid is 50 kN, in axial direc-
Margin of Safety (MoS), that is defined in general as the ratio tion, as denoted in Fig. 6. The opposite side of the structure has
between allowable loads and actual loads multiplied by the appli- been fixed.
cable Safety Factor (SF) as per the following relationship: Four cases were analysed varying the skin thickness from 0.3 to
1.2 mm. These values match with the number of plies used (from 1
allowable load=stress
MoS ¼ 1 ð21Þ to 4).
applied load=stress  SF
As visible in Fig. 7, the skin is more stressed near the load appli-
where the allowable load is the allowable load under specific func- cation area. The compression stress of the skin is higher near the
tional conditions (e.g. yield, buckling, ultimate), the applied load is load application edge in the areas where there is not the isogrid
the computed or measured load under defined load conditions and reinforcement. Generally, the structure is more stressed on the skin
SF is the safety factor applicable to the specified functional condi- near the load application area; this can bring to the separation
tion including the specified load conditions (e.g. yield, ultimate, between the skin and the isogrid structure. As visible in Table 5,
buckling) plus additional SF if applicable. SF ensures the compliance from the analysed cases only the 4th one gives all positive MoS.
with the structural reliability objectives or ensures an acceptable In the cases 1 to 3 the structure would break, in terms of separation

Fig. 2. Mesh of the body.

 L. Sorrentino et al. / Composite Structures 143 (2016) 189–201 193

Fig. 3. Representation of boundary conditions and applied load.

Table 3 technology, the forming mould geometry and the polymerisation

Structural material properties. cycle suitable to guarantee the geometric/dimensional characteris-
Property Value tics and the mechanical properties of the final component.
Carbon/epoxy Glass/epoxy Kevlar/epoxy
To obtain a component that fulfils the geometric/dimensional
specifications, the appropriate choice of both the material and
Tensile strength [Mpa] 276 300 129
Young Modulus [GPa] 7.6 10 23.5
the geometry of the moulds is crucial. The more complex is the
Compression strength [Mpa] 207 250 91.7 geometry of the component to be produced, the more important
Shear strength [Mpa] 136 34 34 is this choice, mainly in the production of structural parts such
Density [kg/m3] 1550 1850 1380 as isogrid structures.
Poisson coefficient 0.3 0.3 0.3

3.1. Tool design

Due to the particular geometry of the structure, the mould must

be made of a material whose CTE is compatible with that of the
composite material. Therefore, traditional metals had not been
considered, preferring epoxy resin, which presented a thermal
behaviour compatible and it was also cost-effective for the small
quantity of pieces to be produced. The composite material forming
the ribs was laid on the mould by means of filament winding, a
technology that provides to wind impregnated fibres on a convex
mould. Therefore, the mould was constituted by a cylinder on
whose lateral surface grooves representing the negative of the rib
were milled. Since the component presented a closed reticular
structure with variable thickness ribs, the mould was constituted
Fig. 4. Usage of the failure criteria.
by a modular structure, which could be radially removed at the
end of the forming process. In particular, the mould was formed
of the isogrid structure from the skin, due to the shear stress. It has by five sectors. To facilitate the extraction of the mould from the
also to be considered that the skin could delaminate due to the component, the grooves presented draft angles k that were vari-
compression loads. ables as a function of the ribs thickness (12.75° in correspondence
of the minimum thickness and 18.50° in correspondence of the
3. Process design maximum one). In addition, the groove section was appropriately
designed near the nodal points, which is where the thickness vari-
Once the geometry of the component had been defined, the ation was located, as shown in Fig. 8. This arrangement was neces-
next step was to design the manufacturing process, defining the sary since a uniform compaction of the laminate had to be ensured,
194 L. Sorrentino et al. / Composite Structures 143 (2016) 189–201

Table 4
Structural analysis results for skin material selection.

Case Criteria Maximum stress area Stress (MPa) MoS Maximum deformation (mm)
1 Max principal Central C/E rib 0.34 0 1.25
C/E–C/E Min principal 0.29 0
Max Shear 0.17 0
2 Max principal Central C/E rib 0.31 0 1.00
G/E–C/E Min principal 0.26 0
Max Shear 0.16 0
3 Max principal Kevlar skin 0.49 0 0.48
K/E–C/E Min principal 0.25 0
Max Shear 0.25 0

Fig. 5. Example of numerical analysis results for the C/E–C/E case.

Fig. 6. Load applied on the designed structure.

Fig. 7. Structural analysis results for thickness optimisation.

 L. Sorrentino et al. / Composite Structures 143 (2016) 189–201 195

Table 5
Skin thickness optimisation.

Part Criteria Case 1 Case 2 Case 3 Case 4

0.3 mm 0.6 mm 0.9 mm 1.2 mm
Stress (MPa) MoS Stress (MPa) MoS Stress (MPa) MoS Stress (MPa) MoS
Skin Max principal 140 0.58 135 0.63 135 0.63 209 0.06
Min principal 412 0.6 230 0.28 236 0.29 84 0.97
Max Shear 228 0.53 153 0.29 130 0.16 104 0.038
Isogrid Max principal 29 6.61 31 6.12 30 6.36 142 0.55
Min principal 72 1.3 82 1.02 80 1.07 71 1.33
Max Shear 39 1.79 51 1.13 41 1.65 90 0.20

in such a way as to guarantee a constant volumetric fibre and dT

qc cp;c ¼ rðkc rTÞ þ qr V r Q_ ð22Þ
matrix percentage on the whole part, obtaining the surface shown dt
in Fig. 9. in which q represented material density, cp was the specific heat, t
Appropriate tools were made for the correct assembly of the was the time, T was the temperature, k was the anisotropic thermal
sectors and to ensure the stability of the mould structure during
conductivity coefficient of the composite material, Q_ was heat gen-
the phases of both fibres stratification and autoclave cure. Some
eration rate by chemical reaction, V was the volumetric percentage.
surface treatments were carried out on mould sectors to facilitate
The subscripts c, r and f referred respectively to composite, matrix
the extraction process after the polymerisation and to reduce the
and fibres properties. Composite density and specific heat were
porosity of the mould material. The finished mould is shown in
derived from the corresponding properties of raw materials, com-
Fig. 10.
bined through the rule of mixture, which was a weighted average
on volumetric and weight fractions respectively. As aforemen-
3.2. Curing numerical simulation
tioned, the thermal conductivity coefficient was anisotropic, due
to the nature itself of the composite material, in particular it was
The curing cycle design was carried out paying attention to the
transversely isotropic, i.e. it was symmetric about an axis that is
most critical conditions, which were located in higher thickness
normal to a plane of isotropy [16]. In-plane thermal conductivity
zones (about 7.2 mm) that were present in the nodal points, where
was calculated through a weighted average on volumetric fraction,
three ribs were coupled with skin. Finite element models, that
while the through thickness thermal conductivity was calculated
analyse the thermo-chemical properties of the whole system built
according to Springer-Tsai model [17]:
up by component and mould, are typically used to determine the " #
temperature and the cure degree inside the part, in order to ensure
that the cure develops in an optimal manner, avoiding the presence kt ¼ ð1  V 0:5
f Þ þ 1
1 þB  kr
V 0:5 2
ð23Þ
of the defects listed above. The analysis was performed considering  
f

the thermochemical characteristics of the part material and the B¼2 km

kf
1
thermal properties of the mould material, which had a high ther-
mal inertia and therefore affects the heat exchange between the Since the polymerisation reaction was exothermic, the heat
environment and the part to be cured. The energy balance was generation term was very important in the energy balance. The
the core of the numerical analysis of the cure process, according heat generation rate depended on the total heat of reaction Hr
to [14,15]: and on the cure rate (da/dt) according to [18]:
 
da
Q_ ¼ Hr ð24Þ
dt
in which Hr was a resin physical property while the cure rate was
expressed as [19,20]
 
da E
¼ Ac exp   am ð1  aÞn ð25Þ
dt RT
where E was activation energy, R was the universal gas constant, Ac
was frequency factor, n and m were reaction orders. All the con-
stants in these models were determined through laboratory tests,
such as differential scanning calorimetry (DSC) [21], or applying
inverse method on temperature data obtained from laminate during
cure process [22]. The material parameters used for cure simulation
are reported in Tables 6 and 7.
The geometry, on which the calculation mesh was prepared,
was derived from the component CAD. Specifically, the intersection
between the three ribs was considered, as it was the most critical
point for the cure process since in that area there was the maxi-
mum thickness, where the effects of the exothermic peak were
more important. Therefore, the Fig. 11 shows the geometry of the
three parts forming the calculation mesh: skin, isogrid and epoxy
mould. The three parts above introduced were discretized with
Fig. 8. Different thickness zones near the node of isogrid structure. four nodes tetrahedral elements, obtaining a total number of
196 L. Sorrentino et al. / Composite Structures 143 (2016) 189–201

Fig. 9. Grooves profile near the nodal points.

360,643 elements, while the total number of nodes was 68,493. As

a boundary condition, a convective heat flux was applied on both
the skin and the inside surface of the mould, while a condition of
thermal insulation was considered on the remaining surfaces.
Thermal cycle was formed by a heating ramp from room tempera-
ture to 130 °C, with a heating rate of 1 °C/min, and a dwell of
120 min at this temperature.
The above defined numerical model was implemented in
PAMRTM, a finite element software, and the results are reported
in the following. Fig. 12 shows the temperature in the skin/ribs/-

Fig. 10. Mould realised for isogrid structure. It consists of 5 sectors.

Table 6
Composite material parameter for numerical cure simulation.

Property Value Property Value

Hr [J/g] 250 q [kg/m3] 1500
Ac [1/s] 1.08  109 Cp [J/(kgK)] 950
E [J/mol] 79,856 Kt [W/(mK)] 0.54
m 0.89 Kl [W/(mK)] 1.38
n 0.81

Table 7
Mould material parameter for numerical cure simulation.

Property Value
q [kg/m3] 750
Cp [J/(kg K)] 1500
K [W/(mK)] 0.25
Fig. 11. Mesh used for numerical simulation.
 L. Sorrentino et al. / Composite Structures 143 (2016) 189–201 197

Fig. 12. Temperature trend in the skin/ribs/mould assembly.

mould assembly, describing the trend in time step of 30 min. The deformation. In order to evaluate it better, the temperature and
absence of exothermic peaks can be noted, therefore the risk of the cure degree were recorded for the two points indicated in
matrix degradation was eluded. Fig. 13 shows the results for the Fig. 14, which were the most critical. These trends are reported
cure degree of the assembly constituted only by skin and ribs, in Fig. 15, while Fig. 16 shows the same information in terms of
and a little polymerisation inhomogeneity is evident in the cross- the difference between the two considered points. In the latter
over region. This was an expected result since the thickness of this figure, a maximum difference of temperature of 1.85 °C is detected,
area was quite high, so this discrepancy was inevitable. However, if while as regards the cure degree it is 0.11. These values were rela-
it had been contained it would not cause residual stresses and tively low and so to redefine the thermal cycle was not necessary.

Fig. 13. Cure degree trend in the skin/ribs assembly.

198 L. Sorrentino et al. / Composite Structures 143 (2016) 189–201

Fig. 14. Locations where temperature and degree of cure are recorded.

Fig. 15. Temperature and degree of cure trend at P1 and P2 locations.

Fig. 16. Differences between P1 and P2 locations for temperature and degree of cure trend.
 L. Sorrentino et al. / Composite Structures 143 (2016) 189–201 199

Fig. 17. Process steps for isogrid + skin structure: (a) rib layering in the mould grooves; (b) skin layering; (c) covering with heat-shrinking tape; (d) introduction in the
vacuum bag.

4. Product realisation and experimental tests

Once the mould had been developed and the thermal cycle
had been defined, to produce the isogrid cylinder was possible.
The mould was mounted, by means of a special equipment, on
a turntable, and impregnated carbon fibres were wound around
it, filling the grooves of the rib, as shown in Fig. 17a. Once the ribs
deposition had been completed, four layers of prepreg carbon fab-
ric were stratified to make up the skin of the isogrid structure
(Fig. 17b), which was then covered with heat-shrinking tape
(Fig. 17c): the compression action exerted by this latter ensured
the right compaction throughout curing. Finally, the mould was
wrapped in the breather and inserted in the vacuum bag
(Fig. 17d). The produced structure is visible in Fig. 18: it was
manufactured through a pre-compaction vacuum bagging process
and a subsequent cure in autoclave according to the designed
thermal cycle.
In conclusion of the present study, experimental tests were con-
ducted to check the compliance with mechanical characteristics
and to assess the suitability of the manufacturing process. In par-
ticular, DSC tests and density determination/matrix digestion tests
were executed on some smaller specimen of ribs and skin. The aim
of these last tests was to determine if the right level of cure degree,
of compaction degree and of voids content had been obtained by
Fig. 18. The whole produced structure.
the designed process. Then, axial compression tests were carried
200 L. Sorrentino et al. / Composite Structures 143 (2016) 189–201

Fig. 19. Load [N] vs. Displacement [mm] and compression test set-up.

Table 8 the numerical simulation. The goal of the work, which is to achieve
Results of experimental tests on strength of designed structure. a component that was characterised by an axial compressive
Test Displacement [mm] Load [N] strength of at least 50,000 N, was reached. In fact, the test was
Average 0.14 52,299
interrupted with a load value of approximately 52,299 N and a
Standard deviation 0.010 1314.60 lowering detected at 0.14 mm, as visible in Fig. 19 and Table 8. It
Variation coefficient (%) 7.1 2.5 was then detected a difference between the theoretical and the
Numerical result 0.14 50,000 experimental value of 4.6% as regards the load, while there were
no differences between experimental and numerical values for
deformation.
out on the whole parts to determine the mechanical strength of the
same and so to verify the conformity with settled target. 5. Conclusion
As concern the DSC tests, some specimens were extracted from
the parts. In particular, five samples were taken out from ribs and An isogrid structure was designed, then a prototype of the
other five from skin for each part. It was discovered that the min- structure was manufactured and tested. Taking into account the
imum cure degree reached during the cure was 95%; therefore, a mechanical behaviour, in terms of deformations and stresses, the
good matrix quality was obtained. thermal and electrical properties and the weight of the material,
As regards the density determination/matrix digestion tests, the the carbon/epoxy composite was chosen for the structural ribs,
material samples were extracted in conformity with a scheme sim- that were dimensioned according to Vasiliev method. Then the
ilar to that of DSC tests. These tests found a good compaction material kind and the number of plies of the skin were selected
degree; in fact the mean fibres content registered was 58% by vol- and optimised on the basis of the numerical analysis. As result of
ume for skin and 60% by volume for ribs. Moreover, the mean void FEM analysis, carbon/epoxy composite material was selected for
content was 3.7% for the ribs and 4.9% for the skin. These values skin. On the basis of the mechanical behaviour, four plies were
were acceptable for the specifications of aeronautical industry, selected as the minimum number in order to guarantee the struc-
therefore the designed process was suitable for isogrid structures ture integrity.
manufacturing. A special mould and a proper curing cycle were designed to
The axial compression test was performed with a machine cap- manufacture the isogrid structure. The mould was formed by five
able of supporting traction/compression tests with a maximum sectors that could be easily assembled for prepreg stratification
load up to 100 kN, equipped with a load cell capable of constantly and disassembled for extraction from the cured structure. The
monitoring the test and an upper plate appropriate to the size of groove section was carefully designed to guarantee a good com-
the component. It was created a plate with four equal sides to pre- paction, especially near the nodal points. The curing cycle was sim-
vent problems of balance on the axis of the machine and an ade- ulated by FEM software, to ensure the correct temperature during
quate plate thickness was chosen (about 20 mm) to avoid the cure process.
machine punch deformation and to not negatively affect the test. The part was produced through autoclave vacuum bag and then
The test was performed by setting a fixed rate of displacement it was tested. The experiments consisted in DSC test and density
to machine head, determining then the load applied to the compo- determination/matrix digestion test carried out on sample of ribs
nent. The compression test lasted about 7 s to compress the com- and skin to evaluate the suitability of the designed process. More-
ponent up to 0.15 millimetres. The failure of the skin occurred at over, the conformity with settled targets was confirmed through
an average deformation value equal to about 0.14 mm (with a coef- the axial compression test executed on the whole structure.
ficient of variation of 7.1%, calculated on 10 samples realised); the A minimum cure degree of 95% was detected through DSC tests,
break did not occur on one of the two ends of the cylinder but in a mean volumetric fibres content of 58% and of 60% was measured
areas that were more central. Upon breaking, the measured load in the skin and in the ribs respectively and the mean void content
has averaged 52,299 N (with a variation coefficient of 2.5%, calcu- was 3.7% for the ribs and 4.9% for the skin.
lated on 10 samples realised), that is approximately 5331 kg. For the axial compression, the structure break mode generally
After the test and the results obtained, the next step was to expected was the delamination of the skin under compression load
compare the experimental data with the design specifications from and separation between the ribs and the skin. This second break
 L. Sorrentino et al. / Composite Structures 143 (2016) 189–201 201

mode underlines the importance of the adhesion between the [7] Antonucci V, Giordano M, Hsiao K-T, Advani SG. A methodology to reduce
thermal gradients due to the exothermic reactions in composites processing.
structure components. The goal of the work, which was to achieve
Int J Heat Mass Transf 2002;45:1675–84.
a component that was characterised by compression strength [8] Wisnom MR, Gigliotti M, Ersoy N, Campbell M, Potter KD. Mechanisms
applied at least of 50 kN, was reached. In fact, a load of 52.3 kN generating residual stresses and distortion during manufacture of polymer-
was experimentally detected, finding a difference between the the- matrix composite structures. Compos A Appl Sci Manuf 2006;37:522–9.
[9] Sorrentino L, Polini W, Bellini C. To design the cure process of thick composite
oretical and the experimental value of 4.6% as regards the load, parts: experimental and numerical results. Adv Compos Mater 2014;23:225–38.
while there were no differences between experimental and numer- [10] Sun CT, Quinn BJ, Tao J. Comparative Evaluation of Failure Analysis Methods
ical values for deformation, that was 0.14 mm. for Composite Laminates. U.S. Department of Transportation Federal Aviation
Administration Office of Aviation Research; 1996.
[11] Barbero EJ. Finite Element Analysis of Composite Materials. Boca Raton
Acknowledgements (USA): CRC Press; 2007.
[12] Camanho PP, Dávila CG, Pinho ST, Iannucci L, Robinson P. Prediction of in situ
strengths and matrix cracking in composites under transverse tension and in-
This work was conducted under the R&D project in implemen- plane shear. Compos A Appl Sci Manuf 2006;37:165–76.
tation of Asse I – Research, Innovation and Strengthening of the [13] Herakovich CT. Mechanics of Fibrous Composites. Wiley; 1997.
[14] Sorrentino L, Bellini C. Compaction influence on spring-in of thin composite
productive basis of the POR FERS Lazio 2007–2013 parts: Experimental and numerical results. J Compos Mater 2015;49:2149–58.
(CO-RESEARCH). Special thanks to ‘‘Tecnologie Avanzate s.r.l.” [15] Bellini C, Polini W, Sorrentino L. A new class of thin composite parts for small
and Eng. R. Aricò for their support in this work. batch productions. Adv Compos Lett 2014;23:111–6.
[16] Sorrentino L, Bellini C. Validation of a methodology for cure process
optimization of thick composite laminates. Polym Plast Technol Eng
References 2015;54:1803–11.
[17] Springer GS, Tsai SW. Thermal conductivities of unidirectional materials. J
[1] Regi M, Mancia F, Marchetti M, Sintoni F, Totaro G, De Nicola F, et al. Nano Compos Mater 1967;1:166–73.
structured composite materials and anisogrid lattice structures for aerospace [18] Carlone P, Aleksendrić D, Ćirović V, Palazzo GS. Meta-modeling of the curing
applications. In: International Astronautical Federation – 55th International process of thermoset matrix composites by means of a FEM–ANN approach.
Astronautical Congress. p. 1944–52. Compos B Eng 2014;67:441–8.
[2] Marchetti M, Regi M, Mancia F. Damage response of anisogrid lattice aerospace [19] Teplinsky S, Gutman EM. Computer simulation of process induced stress and
structures. In: ICOSAAR 2005, 9th International Conference on Structural strain development during cure of thick-section thermosetting composites.
Safety. Comput Mater Sci 1996;6:71–6.
[3] Zheng Q, Jiang D, Huang C, Shang X, Ju S. Analysis of failure loads and optimal [20] Joshi SC, Lam YC. Integrated approach for modelling cure and crystallization
design of composite lattice cylinder under axial compression. Compos Struct kinetics of different polymers in 3D pultrusion simulation. J Mater Process
2015;131:885–94. Technol 2006;174:178–82.
[4] Totaro G. Optimal design concepts for flat isogrid and anisogrid lattice panels [21] ASTM. E2070. Standard Test Method for Kinetic Parameters by Differential
longitudinally compressed. Compos Struct 2015;129:101–10. Scanning Calorimetry Using Isothermal Methods. West Conshohocken,
[5] Vasiliev VV, Razin AF. Anisogrid composite lattice structures for spacecraft and PA: ASTM International; 2013.
aircraft applications. Compos Struct 2006;76:182–9. [22] Farahani SD, Sefidgar M, Kowsary F. Estimation of kinetic parameters of
[6] Radford DW. Cure shrinkage-induced warpage in flat uniaxial composites. J composite materials during the cure process by using wavelet transform and
Compos Tech Res 1993;15:290–6. mollification method. Int Commun Heat Mass Transfer 2011;38:1305–11.