You are on page 1of 8

Mechanics of Materials 148 (2020) 103533

Contents lists available at ScienceDirect

Mechanics of Materials
journal homepage: www.elsevier.com/locate/mechmat

Research paper

Hybrid composites: Experimental, numerical and analytical assessment T


aided by online software

Eduardo A.W. de Menezesa, , Frederico Eggersb, Rogério J. Marczaka, Ignacio Iturrioza,
Sandro C. Amicoa
a
Graduate Program in Mechanical Engineering, Federal University of Rio Grande do Sul, Porto Alegre/RS, Brazil
b
SENAI Institute of Innovation in Polymers Engineering, São Leopoldo/RS, Brazil

A R T I C LE I N FO A B S T R A C T

Keywords: Hybrid composites are commonly applied to obtain tailor-made properties due to additive or synergistic effect
Hybrid composites between matrix and fibers, yielding a wider range of properties than the use of a single fiber could achieve. The
Online software combined use of two or more reinforcements makes the mechanical behavior of hybrid composites more com-
Micromechanics plex, demanding validation of classical micromechanical approaches for the estimate of their properties. The aim
Finite element
of this work is to investigate the mechanical properties of a unidirectional hybrid fibrous composite, at different
constituent volumetric fractions, through micromechanical analytical models implemented in the software
MECH-Gcomp and through a finite element-based homogenization analysis. An experimental campaign was
carried out by producing epoxy composites with glass and carbon fibers via filament winding processing. The
samples were submitted to tensile (longitudinal and transverse directions) and in-plane shear tests to obtain
elastic moduli, Poisson's ratios and shear moduli of the hybrid composites. Good correlation was found between
experimental, numerical and analytical approaches considering the adopted assumptions.

1. Introduction strain since compressive stresses are created in the carbon fibers during
manufacturing due to the distinct thermal expansion coefficients
A hybrid composite (HC) is defined as a composite material con- (Manders and Bader, 1981). A synergistic effect is also credited to fibers
taining at least two distinct types of matrix or reinforcement. The case size and dispersion, where a minimum reinforcement content is re-
regarding multiple fibers is much more common, and commonly aims at quired in order to produce the multifracture phenomenon, yielding
achieving an additive effect between them (Kalantari et al., 2016), i.e. elongation enhancements of up to 100% (Aveston and Kelly, 1980). In a
the final properties are approximately a weighted sum of the individual recent review (Swolfs et al., 2014), where the ultimate strain of carbon/
components’ properties (Mishra et al., 2003). glass HC of 14 different researches were compared, the relative en-
Hybridization may offer a very cost-effective way of meeting design hancement was in the 10–50% range. However, due to surface damage
requirements, allowing composites to reach a tailor-made set of prop- from the co-mingling process, the benefit may significantly decrease.
erties that may include thermal stability, strength and stiffness, Other examples of the use of hybrids are available in the literature.
toughness, and fatigue and impact resistance (Ferrante et al., 2015). Glass-fiber reinforced polymers (GFRP) can reach environmental fa-
There are different types of hybrids. The intraply (or intimately mixed) tigue life of up to 107 cycles by combining them with carbon fibers
category stands out since it can significantly reduce mismatches be- (Shan et al., 2002). In Sun (Sun et al., 2018), insertion of basalt fiber in
tween components in comparison with interply hybridization a carbon/epoxy composite resulted in a toughening effect, preventing
(Shan et al., 2002). cracks from quickly propagating through the thickness and increasing
A synergistic effect is sometimes achieved in hybrid composites. For flexural strength, even though it lowered tensile strength. The combi-
instance, in carbon/glass HC, a synergistic effect can improve ultimate nation of glass with natural fibers was extensively studied over the past

Abbreviations: dmm, Direct micromechanics method; FMM, feairect micromechanics method; FEA, Finite element analysis; gfrp, Glass fiber reinforced polymer; hc,
Hybrid composite; he, High elongation; im, Intermingled; ip, Intraply; LE, Low elongation; PC, Primary composite; ROM, Rule of mixtures; SC, Secondary composite;
RVE, Representative volume element

Corresponding author:
E-mail addresses: eduardo.menezes@ufrgs.br (E.A.W. de Menezes), fredericoeggers@yahoo.com.br (F. Eggers), rato@mecanica.ufrgs.br (R.J. Marczak),
ignacio.iturrioz@ufrgs.br (I. Iturrioz), amico@ufrgs.br (S.C. Amico).

https://doi.org/10.1016/j.mechmat.2020.103533
Received 9 March 2020; Received in revised form 31 May 2020; Accepted 6 July 2020
Available online 08 July 2020
0167-6636/ © 2020 Elsevier Ltd. All rights reserved.
E.A.W. de Menezes, et al. Mechanics of Materials 148 (2020) 103533

few years (Almeida et al., 2013; Jawaid and Abdul Khalil, 2011; where subscripts m, f1 and f2 designate matrix, and primary and sec-
Nunna et al., 2012). The advantages include improvement in mechan- ondary fibers, respectively.
ical properties, impact resistance, durability and reduction in property For the other engineering constants, Banerjee (Banerjee and
variability. Even with the addition of a small amount of glass fibers, one Sankar, 2014) extended the semi-empirical models proposed by Halpin
can significantly decrease the water absorption and improve tensile and (J. C. Halpin and Kardos, 1976) as:
flexural strength of natural fiber composites (Mishra et al., 2003).
E1, f 1
These references are examples of how large can be the number of −1
⎡ 1 + ξ (ηf 1 Vf 1 + ηf 2 Vf 2) ⎤ Em
combinations of materials, in different volume fractions, in order to E2 = E3 = Em ⎢ ; η = ;
1 − ηf 1 Vf 1 − ηf 2 Vf 2 ⎥ f 1 E1, f 1

adjust one or more physical properties of the final composite. In such ⎣ ⎦ Em
context, mathematical methodologies become an important tool to E1, f 2
−1
Em
narrow the choices before compromising manufacture and testing. ηf 2 = E1, f 2
Due to the typical anisotropic behavior of composite materials, Em
+ξ (3)
usually assumed as transversally isotropic (for fibers uniformly dis-
tributed in the cross section (Barbero, 2010; Daniel and Ishai, 2006)),
G12, f 1
five engineering constants and five strengths need to be experimentally ⎡ 1 + ξ (ηf 1 Vf 1 + ηf 2 Vf 2) ⎤ Gm
−1
determined to fully characterize their mechanical behavior, making the G12 = G13 = Gm ⎢ ; η = ;
1 − ηf 1 Vf 1 − ηf 2 Vf 2 ⎥ f 1 G12, f 1

process expensive and time-consuming (Kaw, 2005). Finite element ⎣ ⎦ Gm
analysis (FEA) and closed-form analytical solutions are two effective G12, f 2
Gm
−1
ways to avoid long experimental campaigns, allowing the estimate of ηf 2 = G12, f 2
mechanical properties and the extrapolation of results for different fiber Gm
+ξ (4)
contents, with reasonable accuracy compared to experimental data
(Barbero, 2013; Kaw, 2005). where E2 is transverse Young Modulus, G12 is in-plane shear modulus, V
The numerical approach consists in building a representative vo- is volume fraction, and ξ is a parameter associated with fiber geometry
lume element (RVE) able to statistically represent the overall material. (for circular fibers, it is 1.165 in Eq. (3) and 1.01 in Eq. (4)). Eq. (4) is
Prescribed loads or displacements are applied at the RVE, and the re- also applied to evaluate G23, for ξ equal to 0.9.
sulting stresses and strains are correlated with the whole structure at To estimate longitudinal tensile strength (σ1T), it is assumed that the
the macroscale by means of the averaging principle (Yin and Zhao. Y., composite fails when one of the constituents reaches its ultimate strain,
2016). However, 2D RVE may not give accurate results for composite named ε1 (Banerjee and Sankar, 2014), as given in Eq. (5):
materials (Kuksenko et al., 2018), especially for out-of-plane properties,
demanding more laborious and time-consuming 3D models σ1T = E1 ε1 (5)
(Nazarenko et al., 2018). As an alternative, analytical models are used
for composite homogenization, usually being derived from strength of Alternatively, as proposed by Chamis (Chamis and Sinclair, 1980),
materials or semi-empirical relations. Numerical and analytical models HC properties can be estimated by splitting it on primary composite
usually share important assumptions in the homogenization process, (PC) and secondary composite (SC), where the first is reinforced by fiber
namely, neglecting fiber/matrix interface and voids, homogeneous fi- 1 and the second by fiber 2, with the matrix proportionally distributed.
bers and matrixes and perfect fiber alignment. Also, in the evaluation PC and SC properties are calculated analogously to single-fiber com-
process, the properties of the individual constituents are required, and posites (see Chamis (Chamis, 1989)), as depicted below. Then, ROM is
these may be hard to obtain, especially for fibers. applied to compute E2 from E2,PC and E2,SC. This same methodology is
This work aims at gathering relevant analytical models with closed- adopted by Chamis [20] to compute G12 and G23. The ν23 can be
form solutions for predicting elastic constants and strengths for hybrid computed through Eq. (7), under the transversally isotropic con-
composites. These solutions were incorporated into a composite soft- sideration.
ware MECH-Gcomp (“MECH-Gcomp, http://www.ufrgs.br/
Em Em
mechgcomp/,” 2018). The assumptions and simplifications intrinsic E2, PC = ; E2, SC =
E E
to these models are evaluated through numerical models implemented 1 − Vf , PC ⎛⎜1 − m ⎟⎞ 1 − Vf , SC ⎜⎛1 − m ⎟⎞
E2, f 1 E2, f 2
⎝ ⎠ ⎝ ⎠
in a FE platform under two different approaches, considering the matrix E2 = E2, PC VPC + E2, SC VSC (6)
separately reinforced by carbon and glass fibers (commonly referred as
intermingled hybridization (Yu et al., 2015)), or taking into con-
sideration that the HC was produced from towpregs (glass/epoxy and G23
ν23 = −1
carbon/epoxy) (commonly referred as intraply hybridization (Yu et al., 2E2 (7)
2015)). Experimental tests for distinct relative fiber volume fractions
are used for comparison. Also, this work aims at extending the meth- and the strengths can be estimated through:
odology presented in Banerjee and Sankar (2014), introducing more
analytical models and exploring a different strategy for modeling RVE's σ1,TPC = Vf , PC σ1,Tf 1; σ1,TSC = Vf , SC σ1,Tf 2; σ1T = σ1,TPC VPC + σ1,TSC VSC (8)
of intraply hybrid composites.

2. Analytical models σ2,TPC = σmT ⎡1 − ( Vf , PC − Vf , PC ) 1 −



( ) ⎤⎦ σ Em
E2, f 1
T
2, SC

) (1 − ) ⎤⎦;
Em
Regarding longitudinal Young modulus (E1) and in-plane Poisson = σmT ⎡1 − ( Vf , SC − Vf , SC E2, f 2

ratio (ν12), the rule of mixtures (ROM) is able to provide accurate re-
σ2T = σ2,TPC VPC + σ2,TSC VSC (9)
sults, even for hybrid composites (Banerjee and Sankar, 2014;
Chamis and Sinclair, 1980; Chou, 1992). The equations are a simple
extension of those applied for unidirectional single-fiber composites. where σ2T and σmT are the composite and matrix transverse tensile
strength. Substituting the superscript T for C in Equations (8) and (9),
E1 = E1, f 1 Vf 1 + E1, f 2 Vf 2 + Em Vm (1) one can extend those equations to obtain longitudinal and transverse
compressive strength, respectively. And the HC shear strength is eval-
ν12 = ν13 = ν12, f 1 Vf 1 + ν12, f 2 Vf 2 + νm Vm (2) uated according to:

2
E.A.W. de Menezes, et al. Mechanics of Materials 148 (2020) 103533

min) and using analogical extensometers. Seven specimens for each


τ12, PC = τm ⎡1 − ( Vf , PC − Vf , PC ) 1 −

( Gm
G12, f 1 ) ⎤⎦ composite were tested to obtain elastic moduli and Poisson's ratio. For
transverse tensile tests, length of the samples was 27 mm shorter than
) (1 − ) ⎤⎦;
Gm
τ12, SC = τm ⎡1 − ( Vf , SC − Vf , SC G12, f 2 suggested by the standard due to the limited dimensions of the plates.

The shear test was carried out based on ASTM D7078 at a crosshead
τ12 = τ12, PC VPC + τ12, SC VSC (10) speed of 2 mm/min using four samples for each composite. Strain gages
The model presented by Chou (Chou, 1992) applies the same divi- of the rosette type KFG-5–120-D17–11 from KYOWA were used to en-
sion and equations proposed by Chamis (Chamis and Sinclair, 1980) to able calculation of in-plane shear modulus. All tests were performed in
compute PC and SC properties. However, the final properties are an Instron Universal machine model 3382 with a 100 kN load cell.
evaluated using an inverse ROM as:
4. Numerical Models
E2, PC G12, PC
E2 = ; G12 =
1 + VSC ( E2, PC
E2, SC
−1 ) 1 + VSC ( G12, PC
G12, SC )
−1
(11) The RVE was modelled using the FEA method considering two dif-
ferent strategies. In the first approach carbon and glass fibers were
In order to predict the composite longitudinal tensile strength, individually modeled, and assumed to have a circular cross-section and
Aveston (Aveston and Kelly, 1980) model proposes a synergistic effect to be distributed in a square array, as depicted in Fig. 2, representing an
when the amount of fiber with higher ultimate strain overcomes a intermingled (IM) hybridization. The adopted global coordinate system,
minimum value (Vmin) defined in Eq. (12): XYZ, is also shown in Fig. 2, along with the RVE dimensions. A square
T
array was adopted instead of a hexagonal array to follow the assump-
σ1, LE
Vmin = T T T tions of Chamis (Chamis and Sinclair, 1980) and Chou (Chou, 1992)
σ1, HE + σ1, LE − E1, HE ε1, LE
models. Although this approach is a simple extension of the single-fiber
σ1T = σ1,TLE VLE + ε1,TLE E1, HE VHE (12) composite RVE proposed by Barbero (2013) to a HC RVE, under linear
elasticity assumptions, more recent models are available in the litera-
where the subscripts LE and HE are used for low elongation and high
ture the account for material nonlinearities, fiber twisting (Storm et al.,
elongation fibers, respectively. On the other hand, if VHE < Vmin,
2018) and fillers of different geometries and orientations (Tian et al.,
strength is given by Eq. (5).
2019).
The relations presented above are well accepted in the literature for
Aiming at using a more representative model based on actual mi-
hybrid composites (Swolfs et al., 2014). However, many new models
crographs obtained for the HC, shown in Fig. 3(a), which is closer to
with greater complexity were developed, avoiding assumptions related
intraply (IP) hybridization, a second RVE was proposed. This RVE
to long fibers (Henry and Pimenta, 2017) and straight fibers (Escalante-
considers each towpreg as a homogeneous media, and the HC is formed
Solís et al., 2015), or focusing on other reinforcement geometries, such
by a set of towpregs with different geometries. The great advantage of
as particles or nanotubes (Hassanzadeh-Aghdam et al., 2018). The semi-
this approach is related to the required input data, i.e. towpreg prop-
analytical model proposed by Aboudi (Aboudi, 1995) was also applied
erties, while the latter demands constituents’ individual properties,
for HC (Babu et al., 2005), yielding good mechanical predictions.
which are usually hard to find, especially for the fibers. Three models
Nevertheless, the present work focused on models with closed-form
were created for each type of hybrid to better capture the variability
solutions that do not require detailed fiber geometry data to estimate
along the cross section of the specimen, which was sectioned into
homogenized properties.
smaller regions (Fig. 3(b)) of approximately the same constituent vo-
lume fractions. After building the RVEs with adjusted volume fractions,
3. Experimental depicted in Fig. 3(c), the mechanical properties were evaluated using
the FE commercial software Abaqus ("Abaqus 2016, SIMULIA Academic
Hybrid and non-hybrid composites were produced using glass/ Research, Dassault Systèmes Americas Corp," 2016). An 8-node linear
epoxy and carbon/epoxy towpregs from TCR Composites (Toray 158B- hexahedral element (C3D8R), with one integration point and hourglass
AB-450 and T700–12K-50C, respectively), both pre-impregnated with control, was applied in the numerical models.
UF3369 epoxy resin system, and with a fiber weight fraction (Wf) of For the computation of engineering constants, the same procedure
0.751 and 0.703, respectively. The flat composites were manufactured was adopted for both RVEs. For an orthotropic material, the con-
by dry filament winding using a KUKA 140 L100 robot integrated with stitutive and the compliance matrices are given in Eq. (13), considering
peripheral control systems from MF TECH. The equipment delivery eye normal (σ) and shear (τ) stresses and normal (ε) and shear (γ) strains.
is able to process up to four towpregs simultaneously allowing the The nine independent constants of the constitutive matrix can be used
manufacturing of intraply hybrid composites of distinct compositions. to obtain the engineering constants, as shown in Eq. (14).
The composite design was performed in the CadWind filament winding
software and the simulation is processed and converted into a robot σ ⎡ C11 C12 C13 0 0 0 ⎤ ε1
⎡ σ1 ⎤ ⎢C12 C22 C23 0 0 0 ⎥⎡ ⎤
programming language.
⎢ 2⎥ ⎢ ⎢ ε2 ⎥
Hoop flat composites were produced from the deposition of three ⎥ ε
⎢ σ3 ⎥ = ⎢C13 C23 C33 0 0 0 ⎢ 3⎥

layers of towpregs on a stainless-steel mandrel (327 × 228 × 12 mm³), ⎢ τ23 ⎥ ⎢ 0 0 0 C44 0 0 ⎥ ⎢ γ23 ⎥
⎢ τ12 ⎥ ⎢ 0 ⎢ ⎥
as shown in Fig. 1. The material was then cured in a hot hydraulic press ⎢ τ13 ⎥ ⎢ 0 0 0 C55 0 ⎥ ⎢ γ12 ⎥
⎣ ⎦ ⎥ ⎣ γ13 ⎦
under 6 ton at 120 °C for 4 h and later the system was cooled down to ⎣ 0 0 0 0 0 C66 ⎦
room temperature and the mandrel unscrewed to extract the flat com- 1 −ν12 −ν13
⎡ E1 0 0 0 ⎤
posite. ⎢ −ν
E1 E1

1 −ν 23
Five composites were produced, two single fiber, glass or carbon, ε ⎢ 12 0 0 0 ⎥ σ1
⎡ 1 ⎤ ⎢ E1 E2 E2
⎥⎡ σ ⎤
and three HC. Table 1 shows the amount of tows used, the final ε2
⎢ ⎥ ⎢ −ν13 −ν 23 1
0 0 0 ⎥ ⎢ σ2 ⎥
thickness of the composites, and the composite identification based on ⎢ ε3 ⎥ ⎢ E1 E2 E3 ⎥⎢ 3 ⎥
⎢ γ23 ⎥ = ⎢
the final carbon and glass fiber volume fractions in each case. ⎢γ ⎥ ⎢ 0 0 0 1
0 0 ⎥ ⎢ τ23 ⎥
G 23 ⎥ ⎢ τ12 ⎥
The unidirectional composites were cut using a CNC router ma- ⎢ 12 ⎥ ⎢
⎣ γ13 ⎦ ⎢ 0 0 0 0 1
0 ⎥⎢ τ13 ⎥
chine, longitudinally and transversely to the fiber direction. The spe- G12 ⎥⎣ ⎦
cimens were then sanded and the dimensions measured. Tensile tests ⎢ 0 0 0 0 0 1 ⎥

⎣ G13 ⎥
⎦ (13)
were performed according to ASTM D3039 at constant speed (2 mm/

3
E.A.W. de Menezes, et al. Mechanics of Materials 148 (2020) 103533

Fig. 1. (a) Manufacturing of a flat composite by filament winding and (b) arrangement with four simultaneous towpregs.

Table 1 E2
E2 = E3 ; ν12 = ν13 ; G12 = G13 ; G23 =
Specification of the manufactured composites. 2(1 + ν12) (15)
Composite ID Tows Mean thickness (mm) Although in the present case the HC was assumed as transversally
Carbon/epoxy Glass/epoxy
isotropic, this hypothesis is further investigated.
C0:G64 0 2 2.01 As in the previous analytical expressions, the engineering constants
C19:G44 1 3 2.25 are expressed in local coordinates, with the fibers being parallel to di-
C36:G30 1 1 2.00 rection 1. However, in the FE platform, displacements are applied based
C49:G13 3 1 2.27 on global coordinates system (XYZ), depicted in Fig. 2.
C61:G0 2 0 2.06
Five different simulations were carried out to evaluate the elastic
constants. In the first one, the RVE was submitted to a longitudinal
displacement applied at the frontal face lying in the XY plane, equal to
the RVE length, while the opposite face was clamped. Since the dis-
placements in Y and X directions were constrained in all faces, the
whole body experienced only a unit longitudinal strain. Given that, the
average principle was applied to compute the C terms with Eq. (13),
where stresses in the RVE of volume V and n elements are evaluated at
each i th element (with one integration point each) and weighted by
their size, yielding Eq. (16):

C11 =
1
V
∫ σ1 dV = V1 ∑in= 1 (σ1)i Vi
V

C12 =
1
V
∫ σ2 dV = V1 ∑in= 1 (σ2)i Vi , for ε1 = 1
V

C13 =
1
V
∫ σ3 dV = V1 ∑in= 1 (σ3)i Vi
V (16)
Similarly, a displacement in the transverse directions (Y and X)
equal to the RVE width (squared cross-section) can be applied, yielding
a unit transverse strain state and allowing the computation of C22 and
C23, according to Equations (17-18):
Fig. 2. RVE applied for the hybrid composite with glass (white) and carbon
(black) fibers (diameters: 4.37 mm and 4.79 mm, respectively) (Composite ID C22 =
1
V
∫ σ2 dV = V1 ∑in= 1 (σ2)i Vi
C36G30). V
, for ε2 = 1
C23 =
1
V
∫ σ3 dV = V1 ∑in= 1 (σ3)i Vi
2 2 ](C − C )
V (17)
2C12 [C11 (C22 + C23) − 2C12 22 23
E1 = C11 − C22 + C23
; E2 = 2
C11 C22 − C12
2 ](C − C )
[C11 (C33 + C23) − 2C13
C33 =
1
V
∫ σ3 dV = V1 ∑in= 1 (σ3)i Vi
33 23
E3 = V
2
C11 C33 − C13 , for ε3 = 1
2
C11 C23 − C12
C23 =
1
V
∫ σ2 dV = V1 ∑in= 1 (σ2)i Vi
C12 C13 (18)
ν12 = C22 + C23
; ν13 = C33 + C23
; ν23 = 2
C11 C22 − C12
V

G23 = C44 ; G12 = C55 ; G13 = C66 (14) The C55 can be evaluated through the application of a displacement
equal to the RVE length at the top surface (plane XZ), while the bottom
For the particular case where fibers are uniformly distributed in the surface is clamped. Relative to the faces lying in the plane YZ, dis-
cross-section, the material is usually considered as transversally iso- placements along the X axis were constrained, and faces in plane YX
tropic. In that case, C12 = C13 , C22 = C33 , C44 = were coupled, in order to keep a constant distance (RVE's length)
1
(C22 − C23 ) and C55 = C66 , and the following relations can be obtained: throughout the deformation process, producing a unit shear strain state.
2
The C66 is evaluated with this same procedure, applying the displace-
ment in the YZ plane, while C44 is evaluated through the relations

4
E.A.W. de Menezes, et al. Mechanics of Materials 148 (2020) 103533

Fig. 3. Extended micrograph of the HC (C36G30 sample) showing carbon/epoxy (darker color) and glass/epoxy (lighter color) regions/tows (a), zoomed area of the
HC cross section (b), and its adjusted equivalent RVE (c).

Table 2 5. Results and discussion


Properties of fibers and resin of the towpregs.
Property Glass fiber 158B-AB- Carbon fiber Epoxy resin UF3369–100
Eqs. (1)-(12) were used to evaluate the composites depicted in
450 T700 Table 1. Fig. 4 and Fig. 5 present the analytical results from the mi-
cromechanical models and the numerical results from intermingled
E1 (GPa) 72.4a 230a 3.10a (IM) and intraply (IP) RVEs along with the experimental data. Towpreg
E2 (GPa) 72.4a 15.0a 3.10a
ν12 0.200b 0.200b 0.350b
properties applied in the intraply RVE were computed from the inter-
G12 (GPa) 3.08b 14.7b 1.24b mingled RVE, for the C0:G64 and C61:G0 configurations. The relative
G23 (GPa) 3.08b 0.64b 1.24b deviations of the analytical and numerical results in comparison with
σT (MPa) 2410a 4900a 92.4a the experimental values are displayed in Table 3.
σC (MPa) 1450b 2180b 92.4a
ROM and FEA accurately predicted E1 of the composites, including
τ12 (MPa) — — 112.6b
ρ (kg/m³) 2580a 1800a 1180a the HC samples. The larger deviations for high glass fiber content
suggests a mismatch between its actual Young modulus and that used as
a
From manufacturer datasheet;. input in the simulations (extracted from the literature - see Table 3).
b
Extracted from (Hinton et al., 2004). Also, the Banerjee (Banerjee and Sankar, 2014), Chamis (Chamis and
Sinclair, 1980) and FEA models were able to predict with relatively
depicted in Eq. (15). good accuracy the measured E2 and G12 data. The large difference be-
tween longitudinal and transverse Young moduli justifies why deviation
C55 =
1
V
∫ τ12 dV = V1 ∑in= 1 (τ12)i Vi , for γ12 = 1 for the Banerjee model, based on an isotropic assumption, increases as
V the carbon fiber content increases. Furthermore, since the Chou model
C66 =
1
V
∫ τ13 dV = V1 ∑in= 1 (τ13)i Vi , for γ13 = 1 (Chou, 1992) relies on the application of the inverse ROM (i.e. with
V (19)
“−1″ exponent in the fiber volume fraction), it underestimates the HC
properties, yielding values similar to the composite with the lowest
Finally, Eqs. (16-19) are substituted into Eq. (14), and the en-
modulus. Regarding ν12, ROM and FEA could not adequately fit the
gineering constants can be evaluated, as described in details by Barbero
measured values. This is perhaps due to inadequate (generic) input
(Barbero, 2013). After that, the Direct Micromechanics Method (DMM)
values adopted for fibers and resin.
(Zhu et al., 1998) was applied to evaluate HC strengths, where the
The overall performance of numerical models (both IM and IP) was
stress concentrations factors for transverse and shear strengths were
superior to that of the micromechanical models in comparison with
computed according to Gibson (R. F. Gibson, 1994).
experimental data for HC, with FEA (IP) being slightly superior to FEA
Fibers and resin mechanical properties and density (ρ) are compiled
(IM). Small differences between the two approaches were expected,
in Table 2. The data shown in this table was obtained from the litera-
since large stress concentrations are present in the IM due to the sig-
ture (Hinton et al., 2004) when not provided by the manufacturer.
nificant differences between matrix and fiber properties. These stress
concentrations have a direct impact on the engineering constants when
computing the average stress and applying Eqs. (16–19) but are less

Fig. 4. E1 (a) and E2 (b) values for the various composite configurations.

5
E.A.W. de Menezes, et al. Mechanics of Materials 148 (2020) 103533

Fig. 5. Poisson's ratio ν12 (a) and shear modulus G12 (b) values for the various composite configurations.

Table 3 not hold for RVEs modelled under interlaminated conditions, depicted
Deviation among experimental data and the models. in Fig. 6(a), a hybrid configuration usually adopted to prevent dela-
Deviation Composition 0C:64G 19C:44G 36C:30G 49C:13G 61C:0G
mination (Yu et al., 2015). As shown in Fig. 6(b), stress concentrations
among glass fibers increased the average stress in the X direction (di-
E1 (%) ROM 13.6 2.4 6.9 8.2 1.8 rection 2 in the local coordinate system) when applying a unit strain in
FEA (IM) 14.1 2.2 6.5 7.8 1.4 that direction. Indeed, E3, ν13 and G13 deviated less than 0.5% from the
FEA (IP) — 5.61 4.86 3.07 —
ones obtained through the RVE of Fig. 2, whereas E2, ν12 and G12 de-
E2 (%) Banerjee 13.7 2.4 22.4 28.1 50.9
Chamis 8.7 10.5 4.4 9.2 4.3 viated 3.1%, 0.3% and 1.6%, respectively.
Chou 44.8 43.9 35.9 28.2 12.6 Out-of-plane properties were also analytically and numerically
FEA (IM) 10.8 2.2 0.1 7.2 2.0 analyzed for the various configurations and the values are plotted in
FEA (IP) — 5.43 2.50 2.97 —
Fig. 7. While G23 was again underestimated by the Chou model
ν12 (%) ROM 27.0 29.95 28.72 31.79 18.89
FEA (IM) 21.7 25.38 24.62 28.72 16.59 (Chou, 1992), the Banerjee (Banerjee and Sankar, 2014), Chamis
FEA (IP) — 24.42 26.75 28.24 — (Chamis and Sinclair, 1980) and FEA models yielded more similar re-
G12 (%) Banerjee 1.0 4.06 18.14 4.87 8.92 sults. Also, since ν23 is obtained from E2 and G23 (Eq. (7)), its accuracy
Chamis 11.7 6.92 30.98 17.44 2.11 is linked to these values, and the isotropy consideration of Banerjee
Chou 32.7 34.40 17.63 22.56 29.81
results in clearly overestimated values.
FEA (IM) 8.7 3.91 27.93 10.05 5.63
FEA (IP) — 1.92 0.76 7.95 — Fig. 8 shows strength predictions via analytical and numerical
models in the longitudinal direction. Since Chamis model Chamis and
*Deviations below 15% have been colored for easier visualization. Sinclair, 1980) applies ROM to evaluate strength, as in Eq. (8), it under-
estimates the effect of loss in strength due to the addition of a more
important in the IP configuration due to the relatively smaller differ- brittle constituent, as can be seen in Fig. 8 for the HC with low carbon
ences between towpreg properties. Standard deviations obtained for the content. On the other hand, Banerjee (Banerjee and Sankar, 2014) and
FEA (IP) are also shown in Fig. 4 and Fig. 5 and, except for the E2 value Aveston (Aveston and Kelly, 1980) models (Eqs. (5) and ((12), re-
of the C36:G30 configuration, negligible values were found indicating spectively) yielded similar results in comparison to FEA (DMM method,
that the chosen regions are representative of the whole composite. The considering the constituents lowest ultimate strain) for all five com-
RVE shown in Fig. 2 was also analyzed without the transversely iso- posites.
tropic assumption. The deviations for E3, ν13 and G13, were less than Relative to transverse and shear strength estimates, shown in Fig. 9,
0.1% in relation to E2, ν12 and G12, respectively. This, however, does the analytical model yielded larger values compared to the numerical

Fig. 6. RVE for the interlaminated hybrid configuration (a), and stresses (kPa) in the X direction (direction 2 in the local coordinate system) when submitted to a unit
transverse strain state (ε2 = 1, as depicted in Eq. (17)) (b).

6
E.A.W. de Menezes, et al. Mechanics of Materials 148 (2020) 103533

Fig. 7. Poisson's ratio ν23 (a) and shear modulus G23 (b) values for the various composite configurations.

Fig. 8. Longitudinal tensile strength for the various composite configurations.

Fig. 9. Transverse tensile strength (a) and shear strength (b) for the various composite configurations.

models. Due to the insertion of a stress concentration factor propor- configurations and two numerical models were implemented using
tional to the ratio between fiber and matrix properties, FEA tends to FEA.
give more conservative estimates. Differences between FEA (IM) and Considering the related assumptions and simplifications, and the
FEA (IP) are directly related to the different computations of en- difficulty in obtaining some materials properties to feed the models,
gineering constants, which are applied to compute strengths through analytical and numerical results were in general good agreement with
DMM. the experimental data. Relative to E1, E2 and G12, the micromechanical
models showed a deviation of about 12% in comparison with experi-
mental data, except for the Chou model that clearly underestimated the
6. Conclusions
hybridization effect. And the numerical models resulted in about 6%
deviation. As for the Poisson´s ratio, it could not be accurately predicted
Micromechanical models for hybrid composite materials obtained in
by any model. Analytical and numerical estimates for G23 showed close
the literature were implemented on an online software (MECH-
results. Regarding strength properties, FEA and micromechanics were
GComp), allowing the quick evaluation of their mechanical properties
in good agreement for longitudinal tensile strength, but considerably
based on fibers and matrix properties and their volume fractions. In
differed for transverse and shear strengths. It is important to highlight
order to verify the models’ accuracy relative to the engineering con-
that the second RVE, modelled based on carbon/epoxy and glass/epoxy
stants, experimental tests were performed for five different

7
E.A.W. de Menezes, et al. Mechanics of Materials 148 (2020) 103533

towpregs, does not require individual constituent properties (fibers and estimation of biaxial thermomechanical responses of hybrid fiber-reinforced metal
resin properties), simplifying its use in practical cases. This proposal matrix nanocomposites containing carbon nanotubes. Mech. Mater. 119, 1–15.
https://doi.org/10.1016/j.mechmat.2018.01.002. https://doi.org/.
may have significant benefit in the design of hybrid composites with Henry, J., Pimenta, S., 2017. Modelling hybrid effects on the stiffness of aligned dis-
tailored characteristics which usually demand recursive optimization continuous composites with hybrid fibre-types. Compos. Sci. Technol. 152, 275–289.
procedures with many simulations. https://doi.org/10.1016/j.compscitech.2017.08.017. https://doi.org/.
Hinton, M.J., Kaddour, A.S., Soden, P.D., 2004. Failure Criteria in Fibre-Reinforced-
In all, the present work showed that closed-form analytical solutions Polymer Composites. Elsevierhttps://doi.org/10.1016/B978-0-080-44475-8.X5000-
and RVEs simulated through FEA can be satisfactorily applied to predict 8. https://doi.org/.
hybrid composites properties as first estimates. Since both of them Halpin, J.C., Kardos, J.L., 1976. The Halpin-Tsai equations: a review. Polym. Eng. Sci. 16,
344–352. https://doi.org/10.1002/pen.760160512. https://doi.org/.
showed similar performance, the former may be preferred since it can Jawaid, M., Abdul Khalil, H.P.S., 2011. Cellulosic/synthetic fibre reinforced polymer
be readily applied especially for estimating engineering constants. hybrid composites: a review. Carbohydr. Polym. 86, 1–18. https://doi.org/10.1016/j.
carbpol.2011.04.043. https://doi.org/.
Kalantari, M., Dong, C., Davies, I.J., 2016. Multi-objective analysis for optimal and robust
Declaration of Competing Interest
design of unidirectional glass/carbon fibre reinforced hybrid epoxy composites under
flexural loading. Compos. Part B Eng. 84, 130–139. https://doi.org/10.1016/j.
None. compositesb.2015.08.050. https://doi.org/.
Kaw, A.K., 2005. Mechanics of Composite Materials. Mechanical and Aerospace
Engineering Series. CRC Press.
Acknowledgements Kuksenko, D., Böhm, H.J., Drach, B., 2018. Effect of micromechanical parameters of
composites with wavy fibers on their effective response under large deformations.
Adv. Eng. Softw. 121, 206–222. https://doi.org/10.1016/j.advengsoft.2018.04.013.
The authors gratefully acknowledge the support of National Council https://doi.org/.
for Scientific and Technological Development (CNPq) and CAPES, and Manders, P.W., Bader, M.G., 1981. The strength of hybrid glass/carbon fibre composites -
also UFRGS for the undergraduate scholarships. Part 2 A statistical model. J. Mater. Sci. 16, 2246–2256. https://doi.org/10.1007/
BF00542387. https://doi.org/.
MECH-Gcomp, http://www.ufrgs.br/mechgcomp/, 2018.
References Mishra, S., Mohanty, A..., Drzal, L..., Misra, M., Parija, S., Nayak, S..., Tripathy, S..., 2003.
Studies on mechanical performance of biofibre/glass reinforced polyester hybrid
composites. Compos. Sci. Technol. 63, 1377–1385. https://doi.org/10.1016/S0266-
Abaqus 2016, 2016. SIMULIA Academic Research. Dassault Systèmes Americas Corp.
3538(03)00084-8. https://doi.org/.
Aboudi, J., 1995. Micromechanical analysis of thermo-inelastic multiphase short-fiber
Nazarenko, L., Stolarski, H., Altenbach, H., 2018. Thermo-elastic properties of random
composites. Compos. Eng. 5, 839–850. https://doi.org/10.1016/0961-9526(95)
composites with unidirectional anisotropic short-fibers and interphases. Eur. J. Mech.
93123-D. https://doi.org/.
- A/Solids 70, 249–266. https://doi.org/10.1016/j.euromechsol.2018.01.002.
Almeida, J.H.S., Amico, S.C., Botelho, E.C., Amado, F.D.R., 2013. Hybridization effect on
https://doi.org/.
the mechanical properties of curaua/glass fiber composites. Compos. Part B Eng. 55,
Nunna, S., Chandra, P.R., Shrivastava, S., Jalan, A.K., 2012. A review on mechanical
492–497. https://doi.org/10.1016/j.compositesb.2013.07.014. https://doi.org/.
behavior of natural fiber based hybrid composites. J. Reinf. Plast. Compos. 31,
Aveston, J., Kelly, A., 1980. Tensile first cracking strain and strength of hybrid composites
759–769. https://doi.org/10.1177/0731684412444325. https://doi.org/.
and laminates294, 519–534. https://doi.org/10.1098/rsta.1980.0061.
Gibson, R.F., 1994. Principles of Composite Material Mechanics, 1st ed. Mc Graw Hill,
Babu, P.E.J., Savithri, S., Pillai, U.T.S., Pai, B.C., 2005. Micromechanical modeling of
New York.
hybrid composites. Polymer (Guildf) 46, 7478–7484. https://doi.org/10.1016/j.
Shan, Y., Lai, K.-.F., Wan, K.-.T., Liao, K., 2002. Static and Dynamic Fatigue of
polymer.2005.05.120. https://doi.org/.
Glass–Carbon Hybrid Composites in Fluid Environment. J. Compos. Mater. 36,
Banerjee, S., Sankar, B.V., 2014. Mechanical properties of hybrid composites using finite
159–172. https://doi.org/10.1177/0021998302036002555. https://doi.org/.
element method based micromechanics. Compos. Part B Eng. 58, 318–327. https://
Storm, J., Götze, T., Hickmann, R., Cherif, C., Wießner, S., Kaliske, M., 2018.
doi.org/10.1016/j.compositesb.2013.10.065. https://doi.org/.
Homogenisation by cylindrical RVEs with twisted-periodic boundary conditions for
Barbero, E.J., 2013. Finite Element Analysis of Composite Materials using AbaqusTM.
hybrid-yarn reinforced elastomers. Int. J. Solids Struct. 139–140, 283–301. https://
Composite Materials. CRC Press.
doi.org/10.1016/j.ijsolstr.2018.02.006. https://doi.org/.
Barbero, E.J., 2010. Introduction to Composite Materials Design, Second Edition.
Sun, G., Tong, S., Chen, D., Gong, Z., Li, Q., 2018. Mechanical properties of hybrid
Composite Materials. Taylor & Francis.
composites reinforced by carbon and basalt fibers. Int. J. Mech. Sci. 148, 636–651.
Chamis, C.C., 1989. Mechanics of composite materials: past, present, and future. J.
https://doi.org/10.1016/j.ijmecsci.2018.08.007. https://doi.org/.
Compos. Technol. Res. 11, 3–14.
Swolfs, Y., Gorbatikh, L., Verpoest, I., 2014. Fibre hybridisation in polymer composites: a
Chamis, C.C., Sinclair, J.H., 1980. Mechanics of intraply hybrid composites—Properties,
review. Compos. Part A Appl. Sci. Manuf. 67, 181–200. https://doi.org/10.1016/j.
analysis, and design. Polym. Compos. 1, 7–13. https://doi.org/10.1002/pc.
compositesa.2014.08.027. https://doi.org/.
750010104. https://doi.org/.
Tian, W., Qi, L., Chao, X., Liang, J., Fu, M., 2019. Periodic boundary condition and its
Chou, T.-.W., 1992. Microstructural Design of Fiber Composites. Cambridge University
numerical implementation algorithm for the evaluation of effective mechanical
Press, Cambridge. https://doi.org/10.1017/CBO9780511600272. https://doi.org/.
properties of the composites with complicated micro-structures. Compos. Part B Eng.
Daniel, I.M., Ishai, O., 2006. Engineering Mechanics of Composite Materials, 2nd ed.
162, 1–10. https://doi.org/10.1016/j.compositesb.2018.10.053. https://doi.org/.
Oxford University Press, New York.
Yin, H., Zhao, Y., 2016. Introduction to the Micromechanics of Composite Materials.
Escalante-Solís, M.A., Valadez-González, A., Herrera-Franco, P.J., 2015. A note on the
Introduction to the Micromechanics of Composite Materials, 1st ed. CRC Press, Boca
effect of the fiber curvature on the micromechanical behavior of natural fiber re-
Raton https://doi.org/10.1201/b19685.
inforced thermoplastic composites. Express Polym. Lett. 9, 1119–1132. https://doi.
Yu, H., Longana, M.L., Jalalvand, M., Wisnom, M.R., Potter, K.D., 2015. Pseudo-ductility
org/10.3144/expresspolymlett.2015.100. https://doi.org/.
in intermingled carbon/glass hybrid composites with highly aligned discontinuous
Ferrante, L., Tirillò, J., Sarasini, F., Touchard, F., Ecault, R., Vidal Urriza, M.A., Chocinski-
fibres. Compos. Part A Appl. Sci. Manuf. 73, 35–44. https://doi.org/10.1016/j.
Arnault, L., Mellier, D., 2015. Behaviour of woven hybrid basalt-carbon/epoxy
compositesa.2015.02.014. https://doi.org/.
composites subjected to laser shock wave testing: preliminary results. Compos. Part B
Zhu, H., Sankar, B.V., Marrey, R.V., 1998. Evaluation of failure criteria for fiber com-
Eng. 78, 162–173. https://doi.org/10.1016/j.compositesb.2015.03.084. https://doi.
posites using finite element micromechanics. J. Compos. Mater. 32, 766–782.
org/.
https://doi.org/10.1177/002199839803200804. https://doi.org/.
Hassanzadeh-Aghdam, M.K., Ansari, R., Mahmoodi, M.J., 2018. Micromechanical

You might also like