Polymer Testing 80 (2019) 106148

Contents lists available at ScienceDirect

Polymer Testing
journal homepage: http://www.elsevier.com/locate/polytest

Material Properties

Development of Modified Arcan Fixture for biaxial loading response of

fiber-reinforced composites
P. Hao a, I. Ud Din b, S. Panier a, *
a
Laboratoire des Technologies Innovantes, LTI-EA 3899, Universit�e de Picardie Jules Verne, Amiens, 80025, France
b
Research Center for Modeling & Simulation (RCMS), National University of Sciences and Technology (NUST), Sector H-12, Islamabad, Pakistan

A R T I C L E I N F O A B S T R A C T

Keywords: A Modified Arcan Fixture (MAF) was developed to study the biaxial mechanical behavior of fiber reinforced
Woven composite composites. The MAF was calibrated compared with standard ASTM tests for an epoxy/twill woven-ply laminate.
Modified Arcan fixture Material properties obtained with ASTM tests are implemented in ABAQUS to find the optimal notch radius that
Multiaxial loading
creates the most uniform stress state in gauge area. The optimized configuration of Arcan specimen was adopted
FEM
to design the entire fixture numerically in preload and load steps to avoid plastic deformation. Stress concen­
Shear strength
tration factor was considered in Arcan specimen under tensile loading. Biaxial experimental data for woven-ply
composite in different loading angles was obtain with the MAF. Failure modes under biaxial loading show that
fibers rotated towards the principle tensile direction and failure was caused by excessive stress which reached
fiber limit. The ultimate shear strength corrected by considering the fiber rotation effect are in good agreement
with ASTM standard results.

1. Introduction Multiaxial loadings can be achieved by (1) off-axis specimen without

a special clamping system, (2) tubular and (3) cruciform specimens [4],
Mechanical properties for fiber reinforced polymers (FRPs) can be as shown in Fig. 1a-b. Internal multiaxiality achieved with off-axis
determined by standard tests such as American Society for Testing and specimens under uniaxial tension/compression loading were experi­
Materials (ASTM), Association Française de Normalisation (AFNOR), mentally investigated [5]. Tubular specimens [6] are capable to subject
International Organization for Standardization (ISO) and Japanese In­ combined tension/compression, internal pressure and/or torsional
dustrial Standards (JIS). Special cares as suggested by ASTM D6856 [1] loading while cruciform specimens [7] can subject biaxial
are made to characterize woven composites by refining the specimen tension-tension, tension-compression and compression-compression
dimension according to the size of Repeat Unit Cell (RUC). The me­ loading. The pressure vessels form of tubular specimens cannot repre­
chanical properties along and perpendicular to reinforcement direction sent typical laminate planar structures as they are manufactured by
has been fully understood. However, in applications of multiaxially filament winding yet the planar specimens are prepared in another
loaded laminates, intra strain occurs and shear phenomena limit the different processes. The machining of the cruciform specimens needs
material performance. Under biaxial loading conditions, the damage special care to the corner fillet to insure that failure occurs inside the
often evolves with coupled tension (compression)/tension and tension gauge section and reduction of the thickness. Furthermore, the tests
(compression)/shear. The World-Wide Failure Exercise (WWFE-I, II and require a sophisticated biaxial test machine.
III) [2,3] reported about benchmarking failure criteria in continuous As Iosipescu, V-notch rail shear and Arcan tests are gaining their
FRPs under two-dimensional (2-D) stresses and also gave details of the popularity in identifying in-plane shear properties for FRPs, improve­
input data for benchmarking triaxial failure criteria. It revealed that ment and modification in fixture have been made to acquire high quality
theoretical results should be analyzed and validated under multiaxial data [8,9]. Compared to Iosipescu configuration, the larger range of
loadings. Therefore, developing a simple and robust device to supply the shear dominated stress states can be achieved with Arcan specimen.
reliable material property data and to validate the multiaxial response Besides, a relatively simple manner to control specimen alignments and
becomes much more meaningful than ever. to conduct biaxial loading test is available although Arcan specimen was

* Corresponding author.
E-mail addresses: [email protected] (P. Hao), [email protected] (I.U. Din), [email protected] (S. Panier).

https://doi.org/10.1016/j.polymertesting.2019.106148
Received 9 August 2019; Received in revised form 1 September 2019; Accepted 8 October 2019
Available online 8 October 2019
0142-9418/© 2019 Elsevier Ltd. All rights reserved.
P. Hao et al. Polymer Testing 80 (2019) 106148

originally designed to study the shear response of polymers. It has been preload and tension steps. Von Mises stresses against yield stresses of
modified as a separated grip and a butterfly-shaped specimen version each components of fixture were investigated. The stress distribution
[10] used to investigate the failure of fiber reinforced material under between notch roots affected by notch radius was studied under biaxial
bi-axial loading [11]. Different types of Arcan fixture system have been loading. Lastly, an experiment investigation of the MAF under different
modified to ensure reliable parameters obtained under biaxial loads loading angles was reported. The mechanical parameters obtained with
[12–16]. A Modified Arcan Fixture (MAF) has been used to determine the MAF were compared to the ones with ASTM standard tests. The ef­
the tension/compression-shear failure envelope with a modified version fect of stress concentration on Ultimate Tensile Strength (UTS) under
of the butterfly specimen and the failure observed at the connecting pin loading angle 0� and fiber rotation on shear strength under loading angle
holes is a common issue [12]. Since there is no available standard for 90� was discussed. Results were corrected and compared with ASTM
Arcan test for biaxial loading, different specimen dimensions such as the D3039 [23] and ASTM D3518 [24].
notch angle and notch root radius under shear are generally adopted and
studied in published literature [13]. 2. The Modified Arcan fixture and specimen
The problems of engineering and mathematical physics can be solved
by a numerical method, the finite element method (FEM), with sub­ 2.1. The modified Arcan apparatus for biaxial tension(compression)/
dividing a large system into smaller, simpler parts. Typical problem shear loading
remains in many areas such as structural analysis, heat transfer, fluid
flow, electromagnetic potential, field of optics and biomedical engi­ Arcan specimen was originally designed to study the shear response
neering [8,13,17–20]. Many models were implemented to study of polymers. In this paper, a MAF was designed and fabricated to study
behavior of FRPs. Liu et al. [18] used mesoscopic voxel models to predict the mechanical response including damage accumulation and plasticity
linear and non-linear behavior of 3D woven composite. FEM is per­ of the textile reinforced composite under biaxial loads. The proposed
formed for the influence of parameters of FRPs during infusion process method using a MAF has several modifications.
[17]. Based on FEM, adhesive wear in uni-directional (UD) reinforced Fig. 2 shows the modified Arcan clamp system and the butterfly
carbon thermoplastic polyetherimide composite laminate was studied shaped specimen. Compared with the original one proposed by Arcan
[19]. A modified V-notch rail shear fixture was developed and evaluated et al. [25], The modified test fixture is equipped with guide units that
compared with the standardized one by using FEM [8]. The notch root increased the stiffness of the apparatus. Two guide columns were
and radius for Arcan specimen were studied numerically under shear considered to prevent the bending during the test to insure in-plane
and biaxial loading [13]. stress states. With the designed slideways, specimens can be mounted
The aim of the present paper is to re-design the Arcan fixture and without handling difficulties, besides, the misalignments were elimi­
optimize the specimen geometry to obtain a reliable experimental data nated during the tests. Low-friction slideways are suggested by using
with specimen rupture in the gauge zone. Several improvements have commercially available clearance-free guide units with a high accuracy
been achieved with the guide unit and additional clamping jaws. A full grade linear bushings SM16GU, and shaft diameter was chosen as
material card was obtained for epoxy/twill woven-ply laminate by using 16 mm. The modified fixture was made of aluminum alloy T7351 plates
this MAF with loading angles in pure tensile and shear. The material (thickness of 50 mm) for highly stressed structural parts within a circular
properties were corrected by considering stress concentration factor and platform (diameter of 162 mm).
fiber rotation effect. Results were highly in correspondence with Arcan specimens were attached on the inner platform which is co­
experimental data followed by standard ASTM tests. To authors axial with the loading head between two identical halves of the Arcan
knowledge, rare multiaxial experimental data for woven ply laminate is fixture with ten 8 mm diameter hex socket bolts in the region far from
available for model validation in the open literature. The design of an the gauge zone (see Fig. 3). The load can be transferred through hand-
advanced fixture for applying biaxial loading is still highly demanded. tightened bolts and gripping jaws from fixture to the butterfly spec­
In this paper, at first the modification for Arcan fixture was presented imen. In this configuration, tightening ten gripping bolts can decrease
as well as the optimal dimension of specimen. Secondly, the entire stresses around holes significantly, but inevitably, at a price of taking
fixture including the specimen was examined in FEM within commercial more time on assembling each specimen. Therefore, six bolts on the
software ABAQUS®. The stresses were evaluated for the specimen under outside ring overlapping with the ring of Arcan fixture halves only play a

Fig. 1. Specimens for biaxial loadings: (a) tubular specimen subjected combined tension and torsion [6], (b) typical cruciform specimen with four actuators under
biaxial tension-tension [7] and a modified Arcan specimen [26].

2
P. Hao et al. Polymer Testing 80 (2019) 106148

Fig. 2. Modified Arcan clamp system and specimen.

2.2. Arcan specimen configuration

The Arcan specimens made by epoxy/twill with thickness of 2.5 mm

were cut roughly by a waterjet machining system and then finely grin­
ded and drilled carefully to the final dimension by hand to avoid
delamination caused by the water pressure during machining. The
diameter L was chosen as 160 mm according to the circular platform on
the Arcan fixture. The width of specimen W is 90 mm and the gauge
length d between two notch roots is 25 mm. Ten holes with diameter
8 mm was drilled to match the gripping jaws and Arcan fixture halves.
Parameters in gauge and notch area were studied to specify Arcan
specimen dimension.

1. Gauge length and width

Fig. 3. Boundary conditions and contact pairs adopted in Arcan fixture model. Gauge length (between notch roots) and width (along 1-axis) are
borders of the red zone shown in Fig. 4. The gauge zone was determined
and fixed in this study. As suggested in the ASTM standard tensile tests
role of connecting gripping jaws and Arcan specimen without tightened.
for woven composites, the size of RUC is an essential factor needed to be
It becomes easier to disassemble and change loading direction of the pre-
considered in choosing the sample dimension. Therefore, two times of
charged specimen without inducing new damage to it. The adhesion
RUC width 8 mm was chosen for gauge width. The width here does not
coefficient of the clamping surface was increased by creating structural
affect the specimen configuration but was used to examine and ensure
surfaces as knurl pattern on gripping jaw part A and B so that specimens
the stress and strain uniformity in gauge zone. Furthermore, the gauge
with various thicknesses can be clamped firmly. The gripping jaws and
length plays a more important role in the specimen dimension and Ref
clamping bolts avoid the rotation of specimen effectively and transfer
[10] has given optimal values of the dimension parameters governed by
the loads evenly from tensile machine to specimen.
the ratios. Researchers have suggested different length. As the gauge
With holding a gap of 5 mm by a wedge (Fig. 2j), compression tests
length is too large, failure may not occur in the gauge zone and often
can be conducted without buckling thanks to the guide units. By
initiate in the area of clamp holes caused by the fiber rotation which
changing the angle of Arcan specimen, biaxial stress states can be ob­
leads to delamination of the laminate [26]. Therefore, gauge length was
tained, which leads that:
chosen as the larger value of 1) standard tension test specimen width
Ft ¼ F⋅cos α 25 mm and 2) two times of the size of RUC.

Fs ¼ F⋅sin α (1) 2. Notch angle and radius

where Ft and Fs are the tensile and shear loads decomposed from applied Optimal notch angle and radius producing plane stress were also
load F and α denotes the angle between x-axis and 2-axis shown as in suggested by Arcan for fiber reinforced material (FRM) specimen for
Fig. 2. shear in a ratio form [10].
d d
¼ pffiffi ¼ 5 (2)
w 2r

3
P. Hao et al. Polymer Testing 80 (2019) 106148

Fig. 4. Optimized dimension of Arcan specimen for the balanced twill composite.

where geometrical parameters d, w and r are length between the notch avoid the noise and inaccurate solutions introduced by sudden stress
roots, width of notch root and notch radius respectively. shock waves. The time required for a quasi-static solution with explicit
Notch angle of 90� was suggested in ASTM D5379 for Iosipescu type solver is estimated with the lowest mode frequency of corresponding
specimen. Ref [13] summarized the optimum notch angle values in the structure. Therefore, frequency procedure in Abaqus/Standard was used
published literatures. For transverse isotropic specimen as twill to approximate the natural frequencies of different parts of the MAF
balanced woven composites studied in this work, optimum notch angle with the lowest of 45 Hz in the guide columns, which corresponds to a
was chosen as 90� . Adams and Walrath [27] have also given notch angle period of 22 ms. To be certain a truly quasi-static solution, the period
of 90� when E11/E22 ¼ 1. Therefore, specimens with a notch angles, 90� was amplified 10 times. The history energies were studied to evaluate
have been used. the quasi-static response at the end of process.
Notch root radius has influence on the stress distribution in the gauge Fig. 3 shows the geometric model with prescribed boundary condi­
zone. Adams and Walrath [27] studied three different notch-root radii: tions. A uniaxial displacement-controlled loading was applied with the
0, 0.64, and 1.27 mm for Iosipescu shear specimen, and pointed out that other fixture half encastred. Part-independent method was used to
as the notch-root radius increases, the location of maximum shear stress discrete the model by tetrahedral element C3D4 for fixture halves and
tends to move away from the root of each notch toward the nearest inner hexahedral element C3D8 for Arcan specimen, bolts, clamping units and
loading point on Iosipescu shear specimen. K. W. Gan [12] has adopted guide columns. The element size has been matched within the contact
notch radius 5 mm among three (0.7, 5 and 10 mm) to encourage pre­ area. The total number of elements was 1,700,733 and Degrees Of
mature or unexpected failure. Five radii (0, 1.27, 2.38, 4.78 and Freedom (DOF) of the model was 1,121,058. In the mechanical contact
7.19 mm) with a notch angle 90� have been studied in Ref. [13]. For analysis, two contact properties were considered. A general contact
gauge length of 25 mm, radius was calculated as 3.54 mm according to interaction property refers to frictionless to the tangential motion and a
the ratios proposed by Arcan [10] and the stress uniformity was inves­ penalty-based contacts for the knurl surface of the clamping units so that
tigated with different radii in FEM section. the realistic contact between specimen and clamping jaws was well
simulated. The contact surfaces between connecting bolts and holes
3. Results and discussion located on fixture halves, specimen and clamping jaws were assumed as
frictionless.
3.1. Evaluation of the MAF with finite element analysis The linear elastic materials were used in the finite element analysis
because the irreversible plastic deformation is not considered in the
The FEM was adopted to investigate the suitability of the MAF. The whole MAF system. We adopted alloy 7075-T7351 to fabricate the MAF
stress analysis of the specimen and whole test fixture was done by thanks to its fair machinability and superior strength (Table 1). It is
imposing a fasten force on the clamping unit. Furthermore, the con­ widely utilized by the aircraft and ordnance industries as it offers the
necting holes on both the test fixture and clamping parts were evaluated highest strength of the common screw machine alloys. Socket cap screws
under tensile loading with normalized stress against yield stresses as and nuts M8 class 12.9 with a yield strength of 1100 MPa were used
well as the bolts. The stress distribution in the gauge area of Arcan (Table 2). In this section, instead of woven composite, a transverse-
specimen was examined according to various notch radius (0, 1.27, 2.38, isotropic carbon fiber epoxy laminate with high elastic modulus on 1-
3.54, 4.78 and 7 mm) with a fixed notch angle 90� . axis was chosen as the Arcan specimen (Table 3) to exploit the
maximum capability of the MAF.
3.1.1. Finite element model Because the connecting holes on the contact surfaces of fixture halves
Simulations were performed by using the finite element software surrounded by clamping jaws are not accessible during the entire test
ABAQUS/Explicit, which is established to analyze complex non-linear
contact problems. To ensure the solving of quasi static problems, some Table 1
special considerations are required to keep inertial forces insignificant in Isotropic mechanical parameters for aluminum alloy T7351 and steel.
a state of dynamic equilibrium in an accelerated process. In fast loading Material Density (g/ E ν Yield strength UTS
cases, loading rates are increased with remaining the quality of the cm3) (GPa) (MPa) (MPa)
quasi-static solution. Smooth amplitude curves were applied to ensure T7351 2.81 72 0.33 435 505
that changes in velocity and displacement are smooth. It also helps to Steel 7.82 206.8 0.29 350 420

4
P. Hao et al. Polymer Testing 80 (2019) 106148

Table 2 The contour of local stress along fiber direction σ11 over the Arcan
Mechanical Properties of Socket Cap Screw and nut of ISO 898-1. specimen is shown in Fig. 6 under tensile loading at the beginning of
Property Density E ν Proofload UTS Breaking load application and the end. The maximum σ11 on the specimen reaches
class (g/cm3) (GPa) Rp0.2 or Yield (MPa) torque 61.5 MPa around the clamping area with the torque of 18.8 Nm applied.
strength (Nm) In the gauge area of the specimen, the stress σ 11 is 36 MPa under
(MPa)
compression. The UTS in fiber direction of the composite implemented
12.9 7.82 206.8 0.29 1100 1200 39 in FEM is assumed as 1306 MPa, which is large enough for most common
(29280 N) used composite configuration. The distribution of stress in the gauge
area is quite normal with the optimized dimension. As increasing of the
process, the stress distribution on the connecting holes is important to applied tension, the maximum stress transfer to the tip of notch root.
evaluate whether the plastic deformation occurs. Thus, FEM gives an Distributions of stress show that the peak stress near tip reaches much
alternative to access and investigate whole stress distribution that larger magnitudes than the average stress over the gauge section. The
cannot be examined. To conduct a tensile test with the MAF, firstly, the section variation that causes the stress concentration is referred to as a
Arcan specimen should be assembled on the fixture halves with stress raiser. Theoretical stress concentration factor Kt in elastic range
clamping jaws at a desired angle. The connecting bolts supply a certain can be obtained as 2.38 in stress raiser form of opposite single V-shaped
fasten force to maintain the Arcan specimen firmly, which corresponds notches in finite-width plate [28] compared to 2.78, the value obtained
to a torque. A minimal applied torque introducing a sufficient numerically by FEM (max stress on normal stress).
compression prestress should be used, so that the slip of specimen form pffiffiffiffiffiffi
Kt ¼ C1 þ C2 Ktu þ C3 Ktu (3)
fixture halves can be prevented with bolts and knurl surfaces of
clamping jaws. On the other side, Ref [9] pointed out that the excessive � �2 � �3
2h 2h 2h
clamping force may induce detrimental stress concentrations in the Ktu ¼ C1u þ C2u þ C3u þ C4u (4)
D D D
specimen at the sides of the gauge section. Secondly, the specimen was
loaded until rupture occurs. In this step, the case of pure tensile on fiber where Ktu denotes U-shape raiser form and constants given by analogous
direction with loading angle 0� are chosen to examine whole system as it formulas were calculated C1u ¼ 5.7, C2u ¼ 7.2, C3u ¼ 3.9, C4u ¼ 1.38,
is the critical case for all the connecting holes in the highest stress state. C1 ¼ 1.44, C2 ¼ 2.22, C3 ¼ 0.13. h ¼ 19 mm and D ¼ 63 mm are the
dimensions of Arcan specimen.
3.1.2. Stress evaluation of preloaded Arcan specimen When σ11 reaches 1306 MPa, the simulation ceases. The normalized
Out of plane stress distribution around the clamping holes on the von Mises stresses against yield stresses of different parts of the fixture
specimen after torque applied is show in Fig. 5. The stress distribution of system were examined to ensure that obtained results are accurate with
whole fixture is presented in next section as the fasten stress induced by all parts working within an elastic phase to avoid irreversible plastic
applied torque can be neglected in tensile direction. The clamping tor­ deformation.
que was chosen as 11.1 Nm and 18.8 Nm. The out of plane stress σ 33 in
Arcan specimen reaches 21 MPa with higher applied torque especially 3.1.3. Stress evaluation of the entire fixture
around the connecting bolts, whereas σ33 in the area contacted with Two low-friction roller bearings are adopted in the guide units to
knurl surfaces is slightly higher than the one with lower applied torque. minimizing the frictional resistance. The clamping system including two
The knurl surfaces of clamping jaws disperse the stress distribution fixture halves, clamping parts and high stress bolts were studied with
effectively. normalized stresses against yield stresses. Fig. 7 presents the distribution

Table 3
Transverse-isotropic material properties of Carbon fiber-Epoxy (UD).
Areal density (g/cm2) E1 (GPa) E2 (GPa) E3 (GPa) G12 (GPa) G13 (GPa) G23 (GPa) ν12 ν13 ν23
1.54 137.0 8.2 8.2 5.5 5.5 3.1 0.277 0.277 0.32

Fig. 5. Out of plane normal stress σ 33 in the specimen clamping area with a uniform bolt torque of (a) 11.1 Nm and (b) 18.8 Nm.

5
P. Hao et al. Polymer Testing 80 (2019) 106148

Fig. 6. Local stress along fibers σ11 of the Arcan specimen: end of (a) preload and (b) tensile loading steps.

of normalized stresses in different parts. The plastic deformation is highly recommended as the indicator reach 0.93 with the proofload of
introduced as indicator reaching 1. For fixture halves and clamping 1100 MPa.
jaws, the UTS of 435 MPa for T7351 was considered. The maximum
indicator values are 0.88 and 0.37 around connecting holes for fixture 3.1.4. Influence of notch radius on specimen stress distribution
halves and clamping jaws respectively. Bolts in high class properties are The normalized stress profiles along and across the specimen and

Fig. 7. Normalized von Mises stress by using yield stresses of T7351 and bolts property class of 12 when stress σ 11 at notch of Arcan specimen reaches 1306 MPa.

6
P. Hao et al. Polymer Testing 80 (2019) 106148

along the specimen notch for shear test have been studied by Ref. [13]. 3.2.1. Mechanical response with different loading angles
In this paper, biaxial loading boundary conditions with loading angle Load - displacement curves extracted directly from tensile machine
45� were applied on the simplified model (zone between clamping jaws) for epoxy/twill composite with stacking [0/90]3s are shown in Fig. 9. As
of Arcan specimen with stacking sequence of [0/90]3s. Python script was the loading angle increased, maximum force decreased and the failure
used to generate Arcan specimens with different geometry dimensions. displacement increased. The main reason is that the combined stress
The butterfly-shaped specimens with six different radii namely 0, 1.27, state was introduced in biaxial loading. When it exceeds the material
2.38, 3.54, 4.78 and 7 mm and a fixed notch angle 90� were considered failure envelop, the material is expected to fail. The fibers in gauge area
in the FEM. Elastic properties shown in Table 4 obtained via ASTM tend to rotate along the principle tensile direction. In 0� pure tensile test,
standard tests were implemented into the implicit simulation. Shear there was no fiber rotation as all fibers were already aligned in the
stresses normalized with the average shear stress calculated by dividing tensile direction. In biaxial and pure shear tests, the fibers along 1-axis
the applied load Fs with cross section area on the corresponding path tend to lengthen, inducing a tensile stress as simple shear boundary
were examined. conditions were applied. As the fibers rotate, large displacement was
The normalized stresses along its notch are shown in Fig. 8. Notch needed to reach the rupture. Fig. 10 shows the failure modes captured by
radius affects the uniformity of stress distribution. All the curve shapes crack evolution and finished with tensile breakage of fibers on 1-axis.
presented are nearly symmetric along the path, the curves with radii
under 3.54 mm are concave like shapes, however, radii above 4.78 mm 3.2.2. The effect of fibers rotation on ultimate strength
are inverse. The normalized stresses with notch radius of 3.54- and 4.78- Herakovich et al. [29] predicted the tensile response of a [�45]s
mm show more uniformity with a plateau in the middle zone along the laminate by considering fiber rotation effect. The change of fiber
gauge length. The errors between the maximum and minimum orientation induced by deformation has a significant effect on the
normalized stresses regardless of the stress concentration in the zone non-linear behavior of IM7/5260 and AS4/PEEK composite laminates
from 5 mm to 20 mm from left notch root are 4.51% (0.00), 3.57% [30]. In this paper, the failure mechanism modes were examined, as
(1.27), 2.00% (2.38), 2.31% (3.54), 0.78% (4.78) and 4.36% (7.00). The shown in Fig. 11, for [0/90]3s woven ply laminates, it is indicated that
notch radius can be chosen between 3.54 mm and 4.78 mm according to plies with fibers along 2-axis underwent a transverse damage in matrix
the stress uniformity. Larger notch radius changes the position of stress and failed first. The plies with fibers along 1-axis rotated following with
concentration point. Therefore, notch radius of 3.54 mm was adopted. the clamp movement as fibers in these plies were clamped between the
fixture, which introduced rigid body rotation and resulted in simple
3.2. Biaxial mechanical response of a twill balanced composite shear stress state. In the final state, the constrained specimen also un­
dertook tension as well as pure shear. The specimen ultimately failed
The tests were performed on an electromechanical universal testing with fiber breakage due to the tensile stress reached the axial limit of
LR50K system with a 50 kN capacity load cell and a standard head fibers. Fig. 9 shows the simple shear effects occurred in all specimens
displacement rate of 2 mm/min at 23 � 1 � C of standard laboratory with fiber rupture under biaxial loading.
condition. Shear and normal strains were extracted via Digital Image The rigid rotation angle was calculated by the difference between
Correlation system (DIC) Vic2D™ at the center gauge area of each two angles (1) the one considering fiber movement against horizontal
specimen in ASTM and MAF tests. A series biaxial loading tests with axis x and (2) DIC detected shear deformation zone with additional rigid
angle step of 15-degree were conducted, and results show that rupture rotation. Four specimens under shear load with loading angle 90� were
was caused by fiber breakage along 1-axis for all specimens. The fiber investigated in the effects of fiber rotation. Fig. 12 shows rigid rotation
rotation in gauge area due to boundary condition of simple shear was angle changes in stress-strain curves of specimens labeled with AP1S25,
detected by DIC method and this effect provide us a critical strain where AP5S6, AP5S7 and AP5S9.
fiber align to the principle tensile direction to determine the ultimate The shear strain ε12 was selected as 2.5% corresponding to engi­
shear strength. The MAF was validated with ASTM standard tests in neering strain γ 12 5%. As the rupture was not appeared, S12, 5% can be
obtaining tensile and shear properties for an epoxy/twill woven-ply extracted according to ASTM D3518. However, the error between ASTM
laminate. tests and MAF listed in Table 4 is large, it demonstrated that the shear
strength derived from 5% engineering shear strain for ASTM standard
test is no more suitable for MAF. Furthermore, the fiber rotation
Table 4
Comparison of the material properties of twill balanced woven composites [0/
detected via DIC increased nonlinearly until it closed to shear strain of
90]3S using ASTM and MAF methods. 1% and a linear part appears. Less than shear strain ε12 of 1%, the me­
chanical response in gauge area is nearly pure shear and the shear
Material ASTM MAF Error
property
deformation continues. The shear deformation finished as all fibers in
Average Standard Average Standard % gauge zone aligned with the tensile principle direction. In this critical
value (c.v. deviation value (c.v. deviation
point, shear deformation initiated in the rest zone between gauge area
%) %)
and clamping parts, which is represented by linear part of the fiber
E1, E2 84.15 2.17 60.96 9.60 27.56 rotation curves. It can be also observed from the curly fibers outside of
(GPa) (2.58) (15.75)
G12 (GPa) 4.60 (3.38) 0.16 5.51 (5.40) 0.30 19.93
gauge zone along 1-axis in Fig. 11. Therefore, S12c derived from the
ν12 0.36 0.15 0.29 0.06 19.44 critical point should be adopted as shear strength when using MAF.
(40.27) (21.18)
XT, YT 847 (1.88) 15.95 395 (2.83) 11.16 53.39 3.2.3. Validation of MAF compared with ASTM standard tests
(MPa)
K. W. Gan [26] compared the uniaxial elastic modulus for glass/­
S12, 5% 44 (1.54) 0.68 57 (8.79) 5.07 30.81
(MPa) epoxy obtained from standard tests and a MAF. In this paper, Table 4
S12, 0.2% 24 (8.47) 2.08 39 (20.00) 7.86 60.10 shows the obtained experiment results of full material property card
(MPa) obtained for twill woven-ply composite. Stress concentration factor Kt in
S12c (MPa) – – 45 (11.71) 5.33 3.22 elastic range for on-axis tensile load was calculated as 2.38 for the
(S12,
dimension of Arcan specimen adopted. Nominal tensile stress of
5%)
ε12c (%) – – 1.00 ( ) – – 395 MPa was amplified to a peak stress 940 MPa compared to 847 MPa
ε1u (%) 0.84 (4.90) 0.04 0.70 (8.27) 0.06 16.05 obtained with ASTM D3039 unnotched tensile testing, with an error of
ε12u (%) 15.78 3.13 13.05 3.30 17.30 9.9%.
(19.82) (25.32)

7
P. Hao et al. Polymer Testing 80 (2019) 106148

Fig. 8. Arcan specimen with stacking [0/90]3s and notch angle 90� under biaxial loading angle of 45� : (a) normalized shear stresses along the gauge length with
various notch radius and (b) Shear strain ε12 extracted via DIC with notch radius 3.54 mm.

stress. S12, 0.2%, the offset shear strength and S12c, the critical shear stress
corresponding to the critical shear strain ε12c in MAF tests. The results
show that S12c has good correspondence with S12, 5% from ASTM tests
with an error of 3.22% compared to 30.81% in S12, 5% and 60.10% in S12,
0.2%.

4. Conclusion

This work was carried out to develop a novel Modified Arcan Fixture
(MAF) to conduct combined shear-tension/compression biaxial loading
for fiber reinforced composites and hence to contributes raw biaxial
experimental data for woven-ply composite in different biaxial loading
angles to be used to validate constitutive models. The latest version of
MAF and the specimen have been investigated in ABAQUS/Explicit with
normalized stresses against yield stresses of each components. It proved
that this solid structure has high resistance thanks to materials with
superior strength adopted. Specimen dimension was studied and opti­
mized with notch radius of 3.54 mm to obtain the most uniform stress
Fig. 9. Experimental load-displacement curves of different biaxial loading an­ state as possible under biaxial loading based on numerical studies.
gles for [0/90]3s Arcan specimens. The full material property card obtained with the MAF for twill
woven ply composite was validated with ASTM results. The post treat­
ment of experimental data for MAF was done by considering the fiber
σ max ¼ Kt ⋅σnom (5)
rotation detected via DIC. The fiber rotation effect during test was
where Kt denotes concentration factor for V-notch raiser form, σ max and examined in shear and data were corrected according to the fiber rotate
σ nom are the maximum and nominal stresses between the notch roots critical point. Results are agreed with the standard test. Biaxial response
respectively. of woven ply composite was obtained. Failure modes show that fiber
It can be noticed that three shear strength S12, 5% S12, 0.2% and S12c breakage is the main cause for all specimens with various loading angles.
were used. S12, 5% is calculated by maximum force at or below 5% en­ The new designed MAF has several advantages:
gineering shear strain suggested by ASTM D3518, as maximum shear

Fig. 10. Failure modes corresponding to the biaxial loading angles for [0/90]3s Arcan specimens in the gauge zone.

8
P. Hao et al. Polymer Testing 80 (2019) 106148

f) Loading angles can be changed easily thanks to the clamping system

and a series of tests can be conducted efficiently and effortless.
g) Equipped with guide units to insure the in-plane stress states during
quasi-static loading or fatigue loading and anti-buckling under
compression

Acknowledgements

The first author (Pei Hao) and the second author (Israr Ud Din) are
extremely grateful to their governments, China (China Scholarship
Council, CSC) and, Pakistan (Higher Education Commission, HEC, in
collaboration with Campus France) respectively. We acknowledge
sincerely the financial support so that this research can be undertaken.

References

[1] ASTM D 6856/D 6856M, Standard guide for testing fabric-reinforced, Text.
Compos. Mater. (2016).
[2] M. Hinton, A.S. Kaddour, P.D. Soden, Failure Criteria in Fibre Reinforced Polymer
Composites: the World-wide Failure Exercise, Elsevier, 2004.
[3] A. Kaddour, M. Hinton, Input data for test cases used in benchmarking triaxial
failure theories of composites, J. Compos. Mater. 46 (19–20) (2012) 2295–2312.
[4] A.S. Chen, F.L. Matthews, A review of multiaxial/biaxial loading tests for
composite materials, Composites 24 (5) (1993) 395–406.
[5] D. Cai, G. Zhou, X. Wang, C. Li, J. Deng, Experimental investigation on mechanical
properties of unidirectional and woven fabric glass/epoxy composites under off-
axis tensile loading, Polym. Test. 58 (2017) 142–152.
[6] S. Amijima, T. Fujii, M. Hamaguchi, Static and fatigue tests of a woven glass fabric
composite under biaxial tension-torsion loading, Composites 22 (4) (1991)
281–289.
[7] T. Shi, et al., Biaxial strength determination of woven fabric composite for airship
structural envelope based on novel specimens, Compos. Struct. 184 (2018)
Fig. 11. Pure shear, simple shear and rigid rotation of plies with 0-degree fibers 1126–1136.
in a simple shear specimen of [0/90]3s. [8] M. Gude, W. Hufenbach, M. Andrich, A. Mertel, R. Schirner, Modified V-notched
rail shear test fixture for shear characterisation of textile-reinforced composite
materials, Polym. Test. 43 (2015) 147–153.
[9] D.O. Adams, J.M. Moriarty, A.M. Gallegos, D.F. Adams, The V-notched rail shear
test, J. Compos. Mater. 41 (3) (2007) 281–297.
[10] M. Arcan, Z. Hashin, A. Voloshin, A method to produce uniform plane-stress states
with applications to fiber-reinforced materials, Exp. Mech. 18 (4) (1978) 141–146.
[11] A. Voloshin, M. Arcan, Failure of unidirectional fiber-reinforced materials—new
methodology and results, Exp. Mech. 20 (8) (1980) 280–284.
[12] K.W. Gan, T. Laux, S.T. Taher, J.M. Dulieu-Barton, O.T. Thomsen, A novel fixture
for determining the tension/compression-shear failure envelope of multidirectional
composite laminates, Compos. Struct. 184 (2018) 662–673.
[13] S.C. Hung, K.M. Liechti, Finite element analysis of the Arcan specimen for fiber
reinforced composites under pure shear and biaxial loading, J. Compos. Mater. 33
(14) (1999) 1288–1317.
[14] J.Y. Cognard, L. Sohier, P. Davies, A modified Arcan test to analyze the behavior of
composites and their assemblies under out-of-plane loadings, Compos. Part Appl.
Sci. Manuf. 42 (1) (2011) 111–121.
[15] M. Nikbakht, N. Choupani, Fracture toughness characterization of carbon-epoxy
composite using Arcan specimen, Int. J. Aerosp. Mech. Eng. 2 (2008) 247–253.
[16] S.T. Taher, O.T. Thomsen, J.M. Dulieu-Barton, S. Zhang, Determination of
mechanical properties of PVC foam using a modified Arcan fixture, Compos. Part
Appl. Sci. Manuf. 43 (10) (2012) 1698–1708.
[17] C. Li, A. Cantarel, X. Gong, Influence of structural parameters at microscale on the
fiber reinforcement, J. Compos. Mater. 53 (7) (2019) 863–872.
Fig. 12. Stress-strain curves of simple shear specimen of [0/90]3s with rigid [18] Y. Liu, I. Straumit, D. Vasiukov, S.V. Lomov, S. Panier, Prediction of linear and non-
rotation angle changes. linear behavior of 3D woven composite using mesoscopic voxel models
reconstructed from X-ray micro-tomography, Compos. Struct. 179 (2017) 568–579.
[19] I. Ud Din, S. Panier, P. Hao, G. Franz, J. Bijwe, L. Hui, Finite element modeling of
a) It is easier to obtain biaxial stress state compared to other tests such indentation and adhesive wear in sliding of carbon fiber reinforced thermoplastic
as tubular specimen which is difficult to fabricate and cruciform polymer against metallic counterpart, Tribol. Int. 135 (2019) 200–212.
[20] K. Ahmed, B.K. Paul, M.A. Jabin, B. Biswas, FEM analysis of birefringence,
specimen which needs sophisticate clamping system. Also, it is an dispersion and nonlinearity of graphene coated photonic crystal fiber, Ceram. Int.
alternative to the popular off-axis specimen. 45 (12) (2019) 15343–15347.
b) It is inspired by V-notch rail and iosipescu tests for shear properties [23] ASTM D 3039/D 3039M, Standard Test Method for Tensile Properties of Polymer
Matrix Composite Materials, 2017.
and takes advantages from both fixtures. [24] ASTM D 3518/D 3518M, Standard Test Method for In-Plane Shear Response of
c) It consists of two fixture halves connecting with the guide units Polymer Matrix Composite Materials by Tensile Test of a �45� Laminate, 2018.
which ensure initial alignment and 10 connect holes keep the Arcan [25] M. Arcan, The iosipescu shear test as applied to composite materials: Discussion,
Exp. Mech. 24 (1) (1984) 66–67.
specimen in the desired loading direction before starting tests. [26] T. Laux, K.W. Gan, J.M. Dulieu-Barton, O.T. Thomsen, A simple nonlinear
d) The clamping jaws prevented premature damage and shear out constitutive model based on non-associative plasticity for UD composites:
observed around connecting holes presented in literatures. development and calibration using a Modified Arcan Fixture, Int. J. Solids Struct.
162 (2019) 135–147.
e) Combined tension (compression)/shear monotonic and fatigue tests [27] D.F. Adams, D.E. Walrath, Further development of the losipescu shear test method,
can be conducted; The biaxial loadings can be applied easily, and Exp. Mech. 27 (2) (1987) 113–119.
angles of the applied loadings can be changed with a step of 15� .

9
P. Hao et al. Polymer Testing 80 (2019) 106148

[28] W.D. Pilkey, D.F. Pilkey, R.E. Peterson, Peterson’s Stress Concentration Factors, [30] C.T. Sun, C. Zhu, The effect of deformation-induced change of fiber orientation on
third ed., John Wiley, Hoboken, N.J, 2008. the non-linear behavior of polymeric composite laminates, Compos. Sci. Technol.
[29] C.T. Herakovich, R.D. Schroedter, A. Gasser, L. Guitard, Damage evolution in 60 (12) (2000) 2337–2345.
[�45]s laminates with fiber rotation, Compos. Sci. Technol. 60 (15) (2000)
2781–2789.

10