Polymer Testing 71 (2018) 318–328

Contents lists available at ScienceDirect

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

Test Method

Exploration of specimen geometry and tab configuration for tensile testing T

exploiting the potential of 3D printing freeform shape continuous carbon
fibre-reinforced nylon matrix composites
Lincy Pyl∗, Kalliopi-Artemi Kalteremidou, Danny Van Hemelrijck
Department of Mechanics of Materials and Constructions, Vrije Universiteit Brussel (VUB), BE-1050, Brussels, Belgium

ARTICLE INFO ABSTRACT

Keywords: Now that the design freedom of printing of continuous fibre-reinforced polymers has become available, the
3D printed polymer-matrix composites recommendations on the specimens’ geometry and tab configuration to experimentally determine the elastic
Tensile properties properties of conventionally manufactured composites are reviewed. To explore this design freedom, tensile tests
Design of test specimens for five types of specimen geometry and tabs were investigated. Continuous carbon fibre-reinforced Nylon
Experimental investigation
specimens were printed using a Mark Two commercial 3D printer. Dumbbell shape specimens according to
ASTM D638-14 with proposed radius dimension, 76 mm, and with enlarged radius, 244 mm to reduce stress
concentrations at the fillet and avoid crack initiation were tested. Rectangular specimens according to ASTM
D3039/3039M-14 outperform other specimen geometries. The effectiveness of printed end tabs was in-
vestigated. Even although 3D printing offers facilities for printing end tabs, there is no convincing evidence that
the performance of specimen gripping is better and the printing is more time consuming and expensive.
Rectangular specimens with paper end tabs work best and that is why they were used in the further char-
acterization. The effect of alternating Nylon layers was also investigated and shows drastic reduction in stiffness.
In a next step, the tensile properties of a set of eight rectangular specimens with 0° unidirectional layup were
characterized and compared with conventionally manufactured composites. The tensile properties for different
fibre orientations were also determined. The effect of fibre location and microstructure was studied bringing
important insights to the promising 3D printing but also revealing challenges to overcome (e.g. inhomogeneity in
fibre distribution) to be able to fully explore the design freedom.

1. Introduction – state-of-the-art
In today's technological evolution, 3D printing or Additive
Manufacturing (AM) is considered to be one of the main contributors to
what some claim will be the next industrial revolution [1]. According to
ASTM F42 and ISO TC261, AM is the process of joining materials to
make solid 3D objects in a layer by layer manner, starting from a 3D
digital design (CAD). The AM community has at its disposal a broad
panoply of 3D printing methods and materials with their own appli-
cations and individual challenges [2]. AM processes are categorised
according to the type of material used, the deposition technique or the
way the material is fused or solidified [3,4]. The focus in the work
leading to this publication is on the mechanical performance of 3D
printed continuous carbon fibre-reinforced Nylon polymer parts. The
extrusion and layer by layer deposition of a filament or so-called fused
deposition modelling (FDM) is used for the creation of the polymer
composites. This technique has received by far the most attention and
research interest. For the purpose of this study, only this technique will
be discussed. Thermoplastics like acrylonitrile butadiene styrene (ABS),
polylactic acid (PLA) and Nylon are mostly used in the literature. Only a
few studies report on thermoset materials, i.e. SiC/C-filled epoxy ink, in
3D printing [5].
The layer by layer deposition in FDM inherently results in poor
interlayer bonding. Inspired by nature, adding fibres to the 3D polymer
printing process can be exploited to increase the interlaminar fracture
toughness. One of the classical examples in nature is nacre, a mixture of
layers of brittle platelets of a form of calcium carbonate separated by
sheets of organic matrix composed of elastic biopolymer. To inhibit
transverse crack propagation, the use of alternating material interfaces
and microstructures has proven to be very efficient to increase the
fracture toughness [6]. The optimisation of the microstructure to
maximize the translaminar fracture toughness in 3D printed continuous

∗
Corresponding author.
E-mail addresses: Lincy.Pyl@vub.be (L. Pyl), Kalliopi-Artemi.Kalteremidou@vub.be (K.-A. Kalteremidou), Danny.Van.Hemelrijck@vub.be (D. Van Hemelrijck).

https://doi.org/10.1016/j.polymertesting.2018.09.022
Received 31 July 2018; Accepted 18 September 2018
Available online 19 September 2018
0142-9418/ © 2018 Elsevier Ltd. All rights reserved.
L. Pyl et al.
fibre-reinforced polymers has been studied by [7].
The addition of fibres also has a beneficial impact on the geometric
accuracy. The warping and shrinkage of polymers is substantially re-
duced by adding fibres [8].
Reinforcing the polymers with fibres strongly enhances the Young's
modulus and strength. 3D printing of fibre-reinforced polymer com-
posites is, however, in its infancy. A recent review article illustrates that
most studies on 3D printing of fibre-reinforced polymer composites date
from less than 3 years back [9]. Research on continuous fibre re-
inforcement is less common than the short fibre-reinforced version,
even although it does yield significantly better mechanical properties.
Several researchers worked on the development of micro- or nano-
composites for FDM applications ([10–12]). Other researchers suc-
cessfully explored mechanical property improvements using short fibre
reinforcements ([13–16], (with fibre lengths up to 600 μm)).
Current literature review will focus on the recent developments in
using long fibres in the production of thermoplastic parts with FDM for
improving the mechanical properties as this is the focus in the authors’
present study. Mori et al. [17] included carbon fibres in molten ABS and
showed a drastic improvement of the static and fatigue strength, pro-
vided that the thermal bonding between carbon fibres and plastic was
guaranteed. They used a dieless manufacturing technique where the
carbon fibres are sandwiched between lower and upper ABS plates
made by 3D printing. For the bonding of the fibres and the plates, the
specimens were placed in an oven for 15 min.
Three years ago, MarkForged [18] made its debut with the Mark
One 3D printer, the first and still the only commercial machine capable
of 3D printing reinforced composite parts with continuous fibres. The
latest machine, Mark Two, features faster and more reliable 3D printing
of fibre filament. The Mark One printer was used by van der Klift [19]
to prepare carbon fibre-reinforced Nylon samples. The results of tensile
tests of specimens with zero, two and six carbon fibre layers in between
pure Nylon layers were compared. A good comparison with the rule of
mixtures was obtained for the 2 carbon fibre layers but not for the 6
carbon fibre layers because of the larger void area. van der Klift also
noted that the locations of ending of fibres led to premature failure.
These findings trigger the need for further research. The Higher Tech-
nical School of Engineering (ETSI), University of Seville, recently
characterized Mark One 3D printed carbon and glass fibre-reinforced
Nylon samples in tension and compression [20] with the concluding
note that, although reinforcing remarkably improves the mechanical
properties, the 3D printed composites’ mechanical properties cannot
compete with conventionally manufactured pre-preg composites. Cal-
cination of the carbon fibre filament and glass fibre filament gave
≈40% and 60% fibre weight fraction. They also studied the potential of
improving the fracture resistance by 3D printing patterned interfaces in
bonded joints [21].
Melenka et al. [22] tested samples with Nylon filament and aramid
fibre reinforcement for different fibre volume fractions of 4%, 8% and
10% using the Mark One 3D printer. A volume averaging stiffness
method was developed, giving good comparison results with the ex-
periments for the higher volume fractions but also emphasising the
need for good dimensional accuracy of 3D printed parts. Typically,
sample dimensions differ ± 0.1 mm from the geometric input models.
Similar to van der Klift et al. [19], they observed sample failure at the
location where the fibre path begins. Fibre waviness occurred in the
aramid yarns, as they are not tensioned during 3D printing. This wa-
viness resulted in non-linear behaviour in the stress-strain curves and
negatively affected the mechanical properties.
Al Abadi et al. [23] performed tensile tests on 3D printed carbon,
kevlar and glass fibre-reinforced Nylon test pieces. The percentage of
fibre was varied between 40 and 100% by changing the number of
Nylon layers in between fibre layers. A reduction of 56% was observed
in axial modulus level in the carbon fibre coupons with only 40% fibre
level. The percentage of fibres equals the ratio of the thickness of the
fibre layers to the total thickness of the coupon. Note that this
319
Polymer Testing 71 (2018) 318–328
percentage does not include the fact that the fibre layers also contain
Nylon matrix. Results based on our own research studies showed a
percentage of fibre area in a fibre layer of only around 27% for the
carbon fibre filament of MarkForged's feedstock and printed with the
Mark Two 3D printing system. A 100% fibre composition in [23] thus
contains matrix between the fibres in a fibre layer. The drop in stiffness
by adding intermediate Nylon layers was confirmed. Al Abadi et al.
developed an analytical model based on the Volume Average Stiffness
method to predict elastic properties and a finite element model to
predict failure modes and damage. The present work overcomes the
limitations of the Mark One printer used in Al Abadi's study not being
capable of printing in the biaxial format for carbon fibre.
Matsuzaki et al. [24] modified commercially available FDM prin-
ters. A thermoplastic printing filament and continuous fibres were
supplied separately to the 3D printer. Polylactic acid (PLA) was used as
the matrix while the reinforcement fibres were either carbon fibres or
twisted yarns of natural jute fibres. The impregnation occurred inside
the heated printing nozzle during printing. They were able to double
the strength of their samples compared with the conventional 3D
printing process. They also claimed being able to expand the applic-
ability of the 3D printed parts to load-bearing structures. The nozzle
head seems to be the limiting factor for the fibre volume fraction as the
fibres have to pass the nozzle. The authors nevertheless think their
technique should be able to reach a fibre volume fraction of 40–50%.
This is reasonably high, knowing that carbon fibre composites in
aerospace have a fibre volume fraction of 60–67%.
Li et al. [25] introduced a novel extrusion nozzle for the continuous
extrusion of PLA with carbon fibre to print flat and curved structures
with higher mechanical strength. Surface modification of the carbon
fibre bundle led to increased interfacial strength. Their tensile and
bending results on the modified carbon fibre reinforced composites
improved 13.8% and 164% compared with the original carbon fibre
reinforced samples, highlighting the importance of good matrix-fibre
interfacial bond strength.
Bettini et al. [26] developed a technique to print aramid fibre re-
inforced PLA with fibre content of 9.5 wt % and compared tensile
modulus, tensile strength and shear modulus from micromechanical
models based on the rule of mixtures with experiments. Tensile strength
and tensile modulus transverse to the fibre direction correlated poorly.
The tensile strength from experiments was 30% less than the estimated
value and the authors attributed this to non-uniform stress distribution
over the fibres. They assumed that progressive fibre rupture occurred
because of sensitivity to boundary conditions, edge effects, manu-
facturing flaws and/or testing imperfections. As the experimental ten-
sile modulus transverse to the fibre direction was 60% less, they em-
phasised the need for increasing the adhesion between aramid fibres
and PLA. By adapting a commercial 3D printer, PLA reinforced with
aramid fibres was produced and mechanically tested [27].
Nakagawa et al. [28] manufactured continuous carbon fibre re-
inforced ABS by 3D printing using FDM of ABS layers and sandwiching
carbon fibres between the lower and upper layers. The tensile strength
improvement and increased maximum bending load by adding the
carbon fibres was experimentally demonstrated.
As evidenced by the abovementioned studies, the 3D printing
technology to manufacture continuous fibre-reinforced composites of-
fers great potential, as it can yield improved mechanical properties with
little changes to the basics of the process of polymer printing. The work
is in its infancy. Nearly all studies focused on tension and/or bending
properties. Having said that, the scientific community still lacks data on
the tensile properties of fibre-reinforced FDM composites, let alone
appropriate data for the proper selection of tensile specimen geometry
to promote failure in the gauge length. None of the dumbbell shaped
specimens studied by van der Klift et al. [19] and Melenka et al. [22]
failed in the gauge length where the cross section is the smallest. To
quote ASTM D3039/3039M-14 (paragraph 8.2 Sampling and Test
Specimens Geometry) “Design of mechanical test coupons, especially those
L. Pyl et al.
using end tabs, remains to a large extent an art rather than a science, with no
industry consensus on how to approach the engineering of the gripping in-
terface.” To fully exploit the potential of 3D printed polymers reinforced
with continuous fibres, characterization of the mechanical performance
is crucial, therefore it is important to establish geometry and tab re-
commendations for test specimens of fibre-reinforced FDM materials.
This article explores the test specimen shape and gripping to eval-
uate tensile performance. It forms part of an investigation in exploring
the design freedom of 3D printing. The authors explore in the present
study the geometry and tab recommendations and afterwards provide
the in-plane tensile properties of a 0° unidirectional, 0°/90°, quasi-iso-
tropic and 90° layup fibre orientation and the shear properties, obtained
from a ± 45° layup, taking into account these recommendations.
Ongoing research focuses on fibre deposition to reduce stress con-
centrations (e.g. around holes).
2. Experimental programme
2.1. Materials and 3D printing
Nylon and carbon fibre filaments were sourced from AMR Europe, a
distributor of MarkForged's feedstock and 3D printing systems [18]. All
specimens were printed using the Mark Two 3D printing system. A
STereoLithography (STL) 3D geometry of the tensile specimen was
created in Inventor 2017 [29] and imported into the MarkForged Eiger
software for slicing [30]. The tensile strength at break of the Nylon
according to ASTM D638-14 was measured to be 57 MPa (as compared
to 54 MPa in the material data sheet). The diameter of the carbon fibre
filament was 360 μm. At least one layer of solid plastic (Nylon in this
case), designated as floor and roof layer in the Eiger [30] slicing soft-
ware, was printed. The fibre layers were smoothly contoured by the
floor and roof layers and a wall layer. A part allowing easier removal
from the support plate is thus created by enclosing the fibre layers by
nylon. All specimens had a fully dense rectangular Nylon fill.
Differential Scanning Calorimetry (DSC) results showed a glass
transition temperature Tg = 39 °C and a melting temperature with onset
at Tm = 177 °C and a peak at 198 °C for the Nylon. The enthalpy was
equal to 34 J/g. The crystallization peak in the cooling down phase of
the DSC signal indicates that the MarkForged Nylon is a semi-crystalline
polymer. Dynamic Mechanical Analysis (DMA) confirmed the glass
transition temperature Tg = 34 °C corresponding to the second peak in
loss modulus. The carbon fibre filament consists of carbon fibre and
matrix in between the fibres. The high glass transition temperature of
the carbon fibre filament Tg = 125 °C points to aromatic polymer. No
melting temperature peak could be observed indicating an amorphous
polymer.
2.2. The rationale of the specimen types’ choice
Several avenues were open to the authors to fully explore the tensile
specimens’ design, thanks to the newly available design freedom in 3D
printing. The standards ASTM D638-14 [27] and ASTM D3039-3039M-
14 [31] address the test methods to determine the tensile properties of
plastics and of polymer matrix composite materials. ASTM D638-14 was
originally conceived for conventionally manufactured plastics, typically
demonstrating isotropic properties. Research studies have shown that
the mechanical performance of FDM processed parts is intrinsically
linked to the anisotropic properties. Croccolo et al. [32] defined tensile
specimen geometry requirements to reduce stress concentrations at the
fillet to avoid crack initiation and thus premature failure at the fillet of
FDM processed plastics.
Five types of specimen geometries were tested, as summarized in
Table 1: The standard dumbbell-shaped test specimens according to
ASTM D638-14 [27] and the variant suggested by Croccolo et al. [32]
as well as constant rectangular specimens for 0° unidirectional fibre
orientation according to ASTM D3039/D3039M-14. The performance
320
Polymer Testing 71 (2018) 318–328
of glued tabs and printed end tabs was investigated. The specimen
geometry and gripping are critically assessed in section 3.
Table 1 shows the carbon fibres (blue lines) and Nylon matrix
(white) arrangements for the five specimen geometries. The dumbbell
shape type I (ASTM D638-14) with four floor and four roof layers of
Nylon and eighteen layers of unidirectional carbon fibres from the Eiger
slicing software are shown with a top view for isotropic and concentric
fibre fill type and the layered structure through the thickness. The
Croccolo et al. specimen was printed with isotropic fibre fill type. The
top view and layered structure is shown. The dumbbell shape (ASTM
D638-14 type IV) with printed end tabs with reinforcement in the tabs,
overall continuous unidirectional fibres in the narrow section with al-
ternating carbon and nylon layers and a 3D view is also shown. The
rectangular specimen has a single floor and roof layer and eight carbon
fibre layers. A top view and layered structure through the thickness is
shown. The rectangular specimen with printed Nylon end tabs without
reinforcement in the tab is also shown with a 3D view to better re-
present the printed tabs and a layered structure through the thickness.
Careful preparation of all specimen geometries was crucial to avoid
damage initiation because of badly chosen fibre arrangements. Note
that the start/stop location of a fibre is in every layer located in the end
tabs to avoid failure (Fig. 1). The start/stop location was altered in each
layer to spread as much as possible potential damage initiation loca-
tions. However, where the dumbbell shape specimen widens out into
the grip width (the specimen's shoulder), additional start/stop locations
and infill Nylon matrix were added in failure prone areas where the
stresses are transferred from the grips to the narrow section.
Besides the specimen geometry and tab configuration, the effect of
the fibre and Nylon matrix layers, i.e. the alternating Nylon matrix
layers in between the fibre layers, plays an important role, as also
highlighted by Al Abadi et al. [23]. Therefore, the present study also
investigates the layer stacking influence.
Five clusters were defined for the 0° unidirectional fibre orientation.
The subdivision in clusters is indicated as follows in Table 2 and Fig. 2:
red cluster – twenty six layers, eighteen carbon layers and four roof and
floor layers (26-4Ne18Ce4N), dumbbell shape specimens with dif-
ferent radii and fibre fill type; grey cluster – twelve layers, eight carbon
layers and two roof and floor layers (12-2Ne8Ce2N), rectangular
shape specimen with printed end tabs with and without reinforcement
in the tabs; blue cluster – ten layers, eight carbon layers and a single
roof and floor layer (10-1Ne8Ce1N), comparing rectangular and
dumbbell shape specimens; green cluster – seventeen layers, eight al-
ternating carbon and nylon layers and a single roof and floor layer (17-
1Ne8C/Ne4N) and purple cluster – seventeen layers, five carbon layers
alternated with two layers Nylon in +45° and −45° direction and two
roof and floor layers (17-2Ne5C/+-45Ne2N).
As stated in literature [23], fibre orientation plays an important
role. So, now that the tensile performance of 0° unidirectional speci-
mens and the effect of the specimens’ geometry and tab configuration
were characterized, further mechanical characterization tests for dif-
ferent layups were carried out and the results of the 3D printed material
were compared with - conventionally manufactured composite mate-
rials. Rectangular specimens with paper end tabs were used as they
gave the best results, although with double thickness in accordance
with the recommendations [31]. A pair of [0/90°]4s specimens, a quasi-
isotropic [0/90/ ± 45°]2s specimen layup and a [ ± 45°]4s specimen
were tested. For these orientations, only a limited number of tests was
performed as it was only for an initial estimation of the mechanical
properties. More tests should be done for a full characterization.
2.3. Testing equipment
All tests were performed at a constant displacement rate of 1 mm/
min with an Instron 5885 H (Instron, Norwood,USA) tensile testing
machine. Strains were measured using an Instron Averaging
Extensometer type 2650-557 with a gauge length equal to 50 mm and a
L. Pyl et al. Polymer Testing 71 (2018) 318–328

maximum strain of about 5%. A DIC system (VIC-3D by Correlated
Solutions with two Stingray Cameras of 5 MP and 23 mm lenses) was
used in order to obtain the full-field strain maps during testing through
triangulation of the two cameras. All specimens were printed with the
same Mark Two default settings for the nozzle temperature and speed.
The printing settings are customized to the MarkForged feedstock and
cannot be changed, thus eliminating any potential use of materials from
other suppliers.
3. Assessment of the specimen geometry, tab configuration and
layup
3.1. Tensile test 0° unidirectional carbon fibre specimens
Fig. 2 shows the tensile stress versus strain for the different spe-
cimen types in Table 1. All samples were measured using a micrometer
and/or caliper. Only minor differences with the theoretical values have
been observed. The stress was calculated using the measured cross

Table 1
Specimen types. *Fig. 1 ASTM D638-14.

Type Standard Width of Width Length Thickness of Radius

narrow overall overall narrow of fillet
sec!on sec!on – tab
bevel
angle 1

[mm] [mm] [mm] [mm] [mm] -

[°]1

Dumbbell- ASTM D638-14 13 19 165 3.2 76

shape – type
I*

Isotropic fibre

Concentric fibre

Dumbbell- Croccolo et al. 13 19/25.4 240/300 3.2 244

shape [32]

Dumbbell ASTM D638-14 12 25.4 250 1.25/ 14 inner

shape – type – 25
IV* printed out er
tabs 10°

(continued on next page)

321
L. Pyl et al. Polymer Testing 71 (2018) 318–328

Table 1 (continued)

Rectangle ASTM 15 15 250 1.25/2.125 90°

glued tabs D3039/3039M

Rectangle ASTM 15 15 200/250 1.5 10°

printed tabs D3039/3039M

section area. Five clusters of test results were distinguished corre-
sponding to their geometry, tab configuration and layer stacking, and
thus percentage of carbon fibre filament versus the total volume of the
Nylon and carbon fibre filament layers.
Cluster in red: Fig. 3 shows the front views of the tensile failure of
the dumbbell shape specimens with eighteen layers of carbon and four
roof and floor layers. The concentric fibre layout results in slightly
lower tensile strength, strain to failure and modulus of elasticity
Table 2. From the results, it can be observed that the dumbbell shape
geometry plays a minor role in the mechanical properties under tensile
loading. The radius of the fillet equal to 76 mm according to ASTM 638-
14 versus 244 mm according to Croccolo et al. does not lead to
differences. Note nevertheless that, in both the ASTM 638-14 type I
with isotropic fibre fill type and in the Croccolo et al. design, thirteen
fibres evenly distributed over the width of the specimen were laid. The
stress (422 MPa versus 446 MPa) and stiffness (48.75 GPa versus
48.73 GPa) generated are very similar. The ASTM 638-14 type I with
concentric fibre fill type (DB-R76-CONC) has only twelve fibres evenly
distributed over the width of the specimen which results in a slightly
lower stress and stiffness (380 MPa–43.77 GPa). The Eiger slicing soft-
ware and the symmetry of the concentric fibre fill type doesn't allow an
odd number of fibres. Fig. 4 clearly shows that failure occurs in the neck
of the specimen where the fibres follow the curvature and infill Nylon is
used. Two contour fibre rings were introduced in the isotropic fibre

Fig. 1. Fibre arrangement (blue lines) and

Nylon matrix (with) with start/stop location
(red circle) in the end tabs of the ASTM 638-
14 type I (left) and Croccolo et al. (right)
dumbbell specimen. (For interpretation of
the references to colour in this figure le-
gend, the reader is referred to the Web
version of this article.)

322
L. Pyl et al. Polymer Testing 71 (2018) 318–328

Table 2 with fibre reinforcement, the volume of added carbon fibre filament,
Force, tensile strength, strain to failure and modulus of elasticity for the dif- which is the most expensive, is limited. However, it is not worthwhile
ferent specimens’ geometries and tabs and influence of the Nylon layers (rec- considering these printed end tabs because, as supports are printed for
tangular specimens show a higher maximum force (highlighted in bold) than this type of structure, the printing time doubles and no obvious better
dumbbell shape specimens).
performance is observed.
Force Stress Strain to failure Elas!city modulus Cluster in blue: To further explore the difference in specimen
x10³N MPa % GPa
geometry, a rectangular specimen was compared with a dumbbell
DB-R244-N-26-4N-18C-4N 446 0.91 48.73
DB-R76-ISO-26-4N-18C-4N 422 0.86 48.75 shape specimen. For the rectangular specimen's Glued End Tabs (GET),
DB-R76-CONC-26-4N-18C- carbon fibre epoxy tabs with a length of 55 mm, corresponding to the
380 1.10 43.77
4N grip length, were glued to the specimen using a two-component ad-
DB-R244-W-26-4N-18C-4N 451 0.97 46.08
R-PET-R-12-2N-8C-2N 15.39 684 1.43 47.75
hesive. Specimens with eight layers of carbon fibre and a single roof and
R-PET-12-2N-8C-2N 14.87 661 1.66 47.38 wall layer result in the highest stiffness (≈56 GPa), good tensile
R-GET-10-1N-8C-1N 12.66 654 1.21 57.09 strength (≈647 MPa) and strain to failure (≈1.29%). Fig. 5 shows the
DB-PET-10-1N-8C-1N 9.17 641 1.37 54.69
rectangular specimen with glued end tabs and the dumbbell shape
R-GET-17-1N-8C/N-1N 12.13 365 1.14 31.65
DB-PET-17-1N-8C/N-1N 10.74 422 1.34 31.25 specimen with printed end tabs. Proper failure although just outside the
R-GET-17-2N-5C/+-45N-2N 266 1.51 20.90 clamps is observed. However, the cost and printing time more than
tripled for the dumbbell shape – type IV specimen with printed tabs.
Cluster in green: For the cluster in green, the overall geometry is
type to enhance a smooth transition of the longitudinal fibres in the
narrow gauge length to the wider grips to avoid stress concentrations. the same as in the blue cluster with a rectangular and dumbbell shape
specimen but with seventeen layers instead of ten to study the influence
Nevertheless, damage starts in the infill region of the Nylon matrix and
rupture of the contour fibre rings was observed. The tensile force tore a of alternating intermediate Nylon layers. Although the overall number
of carbon fibre layers equal to eight is exactly the same as in the pre-
hole in the concentric fibre type fill specimen at the location where the
infill Nylon matrix was present. The multiaxial stress state in combi- vious tests, an important drop in modulus of elasticity can be observed.
The specimen with glued end tabs and eight layers of carbon stacked on
nation with the presence of the matrix rich regions and the poor shear
properties of 3D printed composites are the main reasons for this type of top of each other has a force at failure of 12.658 kN and a modulus of
elasticity equal to 57.09 GPa. The specimen with glued end tabs and
damage phenomenon. For the four dumbbell shape specimens, the
average and standard deviation of the tensile strength is equal to eight layers of carbon with each time a layer of Nylon in between has a
force at failure of 12.131 kN and a modulus of elasticity equal to
425 ± 28 MPa, the tensile strain to failure is equal to 0.96 ± 0.09%
and the modulus of elasticity is equal to 46.83 ± 2.07 GPa. 31.65 GPa. Based on the ratio between fibre volume and overall spe-
cimen volume (volume of fibre and Nylon matrix) according to [23],
Cluster in grey: 3D printing offers the possibility to print end tabs.
the fibre volume fraction is equal to 47%. This would lead to an axial
Two types of printed end tabs have been investigated for the rectan-
elastic modulus of 40 GPa obtained using the Volume Averaging Stiff-
gular specimens: end tabs with fibre reinforcement and end tabs with
ness (VAS) method. Discrepancies are attributable to differences in
pure Nylon (Fig. 4). Two floor and roof layers were printed. Both spe-
cimens have almost equal moduli of elasticity ≈47 GPa. The specimen adopted elastic properties of the printed materials in [23]: carbon fibre
strength equal to 850 MPa (corresponding ultimate tensile strength 0°
with the Nylon end tab has a slightly lower tensile strength (661 MPa
versus 684 MPa) and a slightly higher tensile strain to failure (1.66% in [33] for standard carbon fibre unidirectional composites is equal to
1500 MPa) and axial elastic modulus equal to 85 GPa, Nylon matrix
versus 1.43%). Note the sagging effect in the failed samples' outer
Nylon layers. This is attributed to dynamic effects when the specimen strength equal to 56 MPa and axial elastic modulus equal to 0.38 GPa.
According to the MarkForged material specifications, the carbon fibre
breaks (Newton's third law). The specimens failed near the clamps. The
extra cost is limited, in particular for the end tabs without reinforce- strength is 700 MPa (150 MPa less than in [23] which reports half of the
value in the data sheet [33]) and the modulus is 54 GPa (31 GPa less).
ment in the tabs. The volume of extra Nylon is limited and even for tabs

Fig. 2. Tensile stress versus strain for eighteen

layers of carbon and four roof and floor layers
(red), eight carbon layers and two roof and floor
layers (grey), eight carbon layers and a single
roof and floor layer (blue), eight alternating
carbon and Nylon layers and a single roof and
floor layer (green) and five carbon layers alter-
nated with two layers Nylon in +45° and −45°
direction and two roof and floor layers (purple).
(For interpretation of the references to colour in
this figure legend, the reader is referred to the
Web version of this article.)

323
L. Pyl et al. Polymer Testing 71 (2018) 318–328

Fig. 3. Tensile test after failure (from top to bottom) for the dumbbell-shape – type I with isometric and concentric fibres and dumbbell shape with radius of fillet
244 mm, tab width 19 mm and overall length 240 mm and tab width 25.4 mm and overall length 300 mm.

The Nylon matrix strength is 54 MPa (2 GPa less) and modulus is
0.94 GPa (0.56 GPa more). The authors’ experimental results for the
carbon fibre are in line with the MarkForged material specifications.
The pure Nylon matrix modulus by testing was 0.7 GPa (0.24 GPa less
than the MarkForged material specifications). The much lower axial
elastic modulus of the carbon fibre can never lead to the axial elastic
modulus obtained using VAS of 61 GPa in [23]. For the dumbbell shape
specimen, a modulus of elasticity equal to 54.69 GPa versus a modulus
of elasticity equal to 31.25 GPa in the case of alternating Nylon layers is
obtained.
Cluster in purple: Only 8.89 kN maximum force at failure was
obtained in the tensile test of the rectangular specimen with 5 carbon
layers and two alternating Nylon layers, which is to be expected. The
stiffness reduces even further with intermediate Nylon layers, which
was also confirmed in [23].
To sum up, the cluster in red shows a minor influence of the
dumbbell geometry and fibre fill type. The cluster in grey, showing the
results for rectangular specimens with printed end tabs, shows
negligible influence of adding reinforcement to the end tabs, but shows
that rectangular specimens outperform the dumbbell shape specimens.
The cluster in blue confirms the improved mechanical behaviour of the
rectangular specimen. The cluster, in blue, in green and in purple give
insight in the influence of the Nylon matrix layers in between the fibre
layers and clearly demonstrate a reduction in stiffness. The dumbbell
shape specimens can take a lower maximum force in comparison with
the rectangular specimens.
3.2. Extra exploration of use of end tabs for rectangular specimens
A seemingly appropriate specimen geometry is the rectangular
specimen. Further tests with paper end tabs and sand paper end tabs
instead of the glued end tabs were performed. The specimen geometry
was chosen consistent with the guidelines for 0° unidirectional layup
tensile specimens in ASTM D3039/D3039M-14 [29]. The length was
equal to 250 mm, the width was equal to 15 mm and the thickness was
equal to 1.25 mm, containing eight carbon layers of 0.125 mm each and

Fig. 4. Tensile test after failure for the rectangular specimens with printed carbon fibre reinforced end tabs (top) and with printed Nylon end tabs (bottom).

324
L. Pyl et al. Polymer Testing 71 (2018) 318–328

Fig. 5. Tensile test after failure for the rectangular specimen with glued end tabs (top) and the dumbbell shape – type IV specimen with printed tabs (bottom).

Fig. 6. Tensile test after failure for the rectangular specimens with paper end tabs and eight carbon layers; specimen 2 with Digital Image Correlation (DIC) and
extensometer (left) and specimen 8 with DIC and Acoustic Emission (AE).

Table 3 don't give information on the fibre volume in the printed parts. They
Strength, strain to failure and modulus of elasticity for the different specimens’ used bonded ± 45° pre-preg glass fibre tabs as they also considered
fibre orientations. printing of tabs too time consuming and costly. Conventionally manu-
Fibre direction Strength Strain to failure Modulus factured unidirectional carbon fibre epoxy composites outperform the
MarkForged 3D printed carbon fibre Nylon matrix composites. The
MPa % GPa mechanical properties were more than double those of 3D printed
materials with an axial tensile strength Xt = 1500 MPa and elastic
[0°]8 719 ± 46 1.26 ± 0.09 58.07 ± 1.86
[0/90°]4s 217 1.24 17.55 modulus E1 = 135 GPa [33]. Note, however, that the volume fraction
[0/90/ ± 45°]2s 133 1.20 10.89 Vf = 60% was also twice as much. A slightly lower strain to failure
[ ± 45°]4s 48 5 4 1.05% was given.
It is expected that similar trends will be observed in transverse and
in-plane shear directions. Justo et al. [20] determined the shear prop-
a single Nylon floor and roof layer. The sand paper did not perform erties for glass fibre composites only so comparison is not possible. That
better and thus paper end tabs were used. Eight specimens were tested. is why, in the next section, other fibre orientations were tested for the
The tensile stress was calculated based on the average gross cross sec- carbon fibre-reinforced Nylon matrix and compared with conventional
tion measured with a micrometer in three cross positions along the composites. As the rectangular specimens with paper end tabs gave the
gauge length. The cross section includes the floor and roof Nylon layer. best results, the tensile tests for the other fibre orientations were per-
The tensile strength was equal to 719 ± 46 MPa. The tensile strain to formed on this type of specimen.
failure was equal to 1.26 ± 0.09% and the modulus of elasticity
(average of the slopes of the trendlines) was equal to 4. Tensile test [ ± 45°]4s, [0/90°]4s, [0/90/ ± 45°]2s, 90° layup
58.07 ± 1.86 GPa. Proper failure in the gauge length was obtained in carbon fibre specimens
all samples (Fig. 6). Note that this type of specimen geometry and end
tabs gives the highest properties of all tested specimen types. Table 3 summarizes longitudinal strength, strain to failure and
Justo et al. [20] performed tests on Mark One 3D printed carbon modulus of elasticity for the abovementioned set of eight 0° unidirec-
fibre-reinforced parts. The magnitudes they obtained are in the same tional tensile tests, which are repeated for completeness for a pair of [0/
range with a tensile strength equal to 701 MPa ± 70 MPa and a mod- 90°]4s specimens, for a quasi-isotropic [0/90/ ± 45°]2s specimen layup
ulus of elasticity equal to 68.08 GPa ± 5.99 GPa. They only show an and the shear strength, strain to failure and shear modulus for a
approximate 40% carbon fibre weight for the carbon fibre filament but [ ± 45°]4s specimen. From test results of two [0/90°]4s specimens, a

325
L. Pyl et al. Polymer Testing 71 (2018) 318–328

Fig. 7. Trendline of the tensile stress versus strain for a 90° layup with eight layers of carbon fibre and 25 mm specimen width (blue line), eight alternating carbon
and Nylon layers and 15 mm specimen width (green line) and eight alternating carbon and Nylon layers and 25 mm specimen width (red line). All specimens have a
single roof and floor layer of Nylon and are tested with paper end tabs. (For interpretation of the references to colour in this figure legend, the reader is referred to the
Web version of this article.)

Fig. 8. Stereomicroscope image of the floor (left) and roof (right) layer of Nylon and underlying carbon fibres at 200 μm (right) scale.

tensile strength equal to 217 ± 10 MPa, a tensile strain to failure equal
to 1.24 ± 0.03% and a modulus of elasticity (average of the slopes of
the trendlines) equal to 17.55 ± 0.185 GPa was determined. The re-
sults for the quasi-isotropic layup were tensile strength equal to
133 MPa, strain to failure equal to 1.2% and modulus of elasticity equal
to 10.89 GPa.
The experimentally determined shear stress in the [ ± 45°]4s spe-
cimen was found to be 48 MPa. The number of carbon fibre layers was
twice that of the specimens with 0° unidirectional layup to comply with
the standard [31]. The test was stopped before failure as it kept
stretching without strength increase. A lower shear strength with a
difference of 22 MPa in comparison with the conventional carbon epoxy
composites with shear strength S = 70 MPa was obtained [33]. A shear
modulus of 4 GPa was determined, which is similar to that on con-
ventional composites (shear modulus G12 = 5 GPa, strain to failure
1.2%).
Fig. 7 shows the results for a 90° layup. The measurement data is
less smooth. The tensile stress versus strain for a 90° layup with eight
layers of carbon fibre and 25 mm specimen width (blue line) is the
highest. This is similar to the 0° results where the highest tensile
strength was obtained without adding intermediate Nylon layers. Eight
alternating carbon and Nylon layers and 25 mm specimen width (red
line) gave lower values. To check the influence of the width of the
sample, a specimen with a width of 15 mm was tested. It is probable
that not all fibres survive the fibre placement process, as it involves
severe radiuses of curvature as it is being laid down. In this case, it
seems that the fibres did not break despite the high and numerous
curvatures. The smaller width gives a lower strength.
The modulus of elasticity based on the slope of the trendline for the
90° layup with eight layers of carbon fibre and 25 mm specimen width
was equal to 1.46 GPa. No statistical meaning could be attributed to this
single test as only one of those 90° specimens with 8 was tested. As

326
L. Pyl et al.
Fig. 9. Scanning Electron Microscope (SEM) micrographs of a neatly polished cross section cut from a pristine specimen; a) left edge, b) middle and c) right edge (500
μm), d) 200 μm micrograph, e) 100 μm micrograph (inset of b), f) 20 μm micrograph, g) 10 μm micrograph and h) 5 μm micrograph.
expected, the performance in the transverse direction is dominated by
the plastic behaviour of the thermoplastic Nylon matrix, and compar-
ison with conventional 90° carbon epoxy composites is out of the
question. The transverse properties for the conventional composites
were [33]: tensile strength 90° Xt = 50 MPa, elastic modulus 90°
E2 = 10 GPa, strain to failure 0.5%.
5. Microstructural analysis
Fig. 8 shows the outer layers of Nylon and underlying carbon fibres.
The 45° printing direction of the outer layers can clearly be observed, as
can the underlying carbon layers. From the images, it is clear that fibre
waviness occurs in the specimens. This waviness did not result in non-
linear behaviour in the stress-strain curves in the unidirectional carbon
fibre specimens.
Fig. 9 shows Scanning Electron Microscope (SEM) micrographs of a
neatly polished cross section cut from a pristine specimen. The floor
Nylon layer is flatly spread on the printing bed with seams where the
Nylon fibre filaments touched during printing (a, b and c). The roof
Nylon layer is more bumpy (b). The outer wall, where a single Nylon
filament layer is printed shows the stacking, and a huge void can be
observed in the left edge of the specimen (a). The right edge shows that
not all carbon fibre layers are equally wide (c). Based on the 10 μm
micrograph (g), it can be concluded that the diameter of one carbon
fibre is approximately equal to 6.7 μm. In the carbon fibre layers,
clustering of carbon fibres can be observed with some regions without
fibres. Also, many voids are visible.
327
Polymer Testing 71 (2018) 318–328
The ImageJ 1.50e National Institutes of Health, USA, software in the
public domain was used to obtain the fibre volume fraction from a 10 μm
Scanning Electron Microscope (SEM) micrograph. Fibre volume fraction
estimations showed a percentage of fibre area to be almost equal to 27%.
This percentage of fibre area was obtained from a randomly chosen region
in a carbon fibre layer. Floor, mid and roof layers of Nylon matrix were not
considered. The abovementioned tensile properties of the 3D printed spe-
cimens show that, in comparison with the conventional composites, lower
strength and stiffness can be reached. This is in accordance with the find-
ings from the microstructural imaging showing a much lower fibre content,
a less homogeneous fibre distribution and a high void content in compar-
ison with the conventionally manufactured specimens.
6. Conclusions
The identification of an optimal tensile specimen geometry for the
mechanical characterization of continuous carbon fibre-reinforced
polymer matrix composites was possible. In combination with the
newly available potential in printing customized geometries and tabs,
several tests were performed in order to identify specimen geometries
with the location of failure in the gauge length and a proper force in-
troduction into the material. Flat rectangular specimens were shown to
perform best. Nonbonded paper tabs or a light abrasive between the
grip and a rectangular specimen were successfully used and are an ef-
ficient solution despite the newly available design freedom. The effect
of Nylon layers in between the fibre layers was investigated and showed
a reduction in stiffness by adding alternating layers of Nylon.
L. Pyl et al.
This article provides a broad overview of the level of development
reached currently in 3D printed polymer materials and their mechanical
characterization. The focus in this work was on the specimen geometry
and tabs. Having found an optimal specimen design, the characteriza-
tion of tensile properties of unidirectional 3D printed carbon fibre re-
inforced Nylon matrix specimens was performed with their elastic
tensile strength, strain to failure and modulus of elasticity found to be
equal to 719 ± 46 MPa, 1.26 ± 0.09% and 58.07 ± 1.86 GPa. The
shear modulus was acquired from ± 45° layup tensile test specimens.
The shear stress of 48 MPa and shear modulus of 4 GPa was experi-
mentally determined. The performance in tensile tests of 90°, 0/90° and
quasi-isotropic layup test specimens is also presented. Conventionally
manufactured fibre-reinforced polymer matrix composites show better
mechanical performance in comparison with 3D printed ones, which
was to be expected as the microstructural characterization shows con-
siderable inhomogeneity in fibre distribution and voids, and as the fibre
volume fraction is only around 27%.
Acknowledgement
The authors would like to thank the Research Group
Electrochemical and Surface Engineering of VUB for the assistance in
performing scanning electron microscopy and the Research Group
Physical Chemistry and Polymer Science for the assistance in per-
forming dynamic scanning calorimetry and dynamic mechanical ana-
lysis.
References
[1] C. Barnatt, 3D Printing: the Next Industrial Revolution, (2013).
[2] T.D. Ngo, A. Kashani, G. Imbalzano, K.T.Q. Nguyen, D. Hui, Additive manufacturing
(3D printing): a review of materials, methods, applications and challenges, Compos.
Part B Eng. 143 (December 2017) (2018) 172–196.
[3] AM Platform, Additive Manufacturing : Strategic Research Agenda, AM Platf., 2014,
pp. 1–64.
[4] J. Gausemeier, M. Wall, S. Peter, Thinking Ahead the Future of Additive
Manufacturing – Exploring the Research Landscape, Heinz Nixd. Inst, 2013, p. 143.
[5] B.G. Compton, J.A. Lewis, 3D-printing of lightweight cellular composites, Adv.
Mater. 26 (2014) 5930–5935.
[6] N. Wang, S. Xia, Cohesive fracture of elastically heterogeneous materials: an in-
tegrative modeling and experimental study, J. Mech. Phys. Solid. 98 (2017)
87–105.
[7] Y. Swolfs, et al., Designing and 3D printing continuous fibre-reinforced composites
with a high fracture toughness, in: B.D. Davidson, J.G. Ratcliffe, M.W. Czabaj (Eds.),
31st Technical Conference of the American Society for Composites, DESTech
Publications, Inc., Williamsburg, USA, 2016.
[8] L.J. Love, V. Kunc, O. Rios, C.E. Duty, A.M. Elliott, B.K. Post, R.J. Smith, C. a. Blue,
The importance of carbon fiber to polymer additive manufacturing, J. Mater. Res.
29 (2014) 1893–1898.
[9] X. Wang, M. Jiang, Z. Zhou, J. Gou, D. Hui, 3D printing of polymer matrix
328
Polymer Testing 71 (2018) 318–328
composites: a review and prospective, Compos. Part B Eng. 110 (2017) 442–458.
[10] X. Wei, D. Li, W. Jiang, Z. Gu, X. Wang, Z. Zhang, Z. Sun, 3D printable graphene
composite, Sci. Rep. 5 (2015) 11181.
[11] S. Dul, L. Fambri, A. Pegoretti, Fused deposition modelling with ABS-graphene
nanocomposites, Compos. Part A Appl. Sci. Manuf. 85 (2016) 181–191.
[12] S. Hwang, E.I. Reyes, K.S. Moon, R.C. Rumpf, N.S. Kim, Thermo-mechanical char-
acterization of metal/polymer composite filaments and printing parameter study
for fused deposition modeling in the 3D printing process, J. Electron. Mater. 44 (3)
(2015) 771–777.
[13] W. Zhong, F. Li, Z. Zhang, L. Song, Z. Li, Short fiber reinforced composites for fused
deposition modeling, Mater. Sci. Eng. A301 301 (2001) 125–130.
[14] F. Ning, W. Cong, J. Qiu, J. Wei, S. Wang, Additive manufacturing of carbon fiber
reinforced thermoplastic composites using fused deposition modeling, Compos. Part
B Eng. 80 (2015) 369–378.
[15] H.L. Tekinalp, V. Kunc, G.M. Velez-Garcia, C.E. Duty, L.J. Love, A.K. Naskar, C.
a. Blue, S. Ozcan, Highly oriented carbon fiber-polymer composites via additive
manufacturing, Compos. Sci. Technol. 105 (2014) 144–150.
[16] O.S. Carneiro, a. F. Silva, R. Gomes, Fused deposition modeling with polypropylene,
Mater. Des. 83 (2015) 768–776.
[17] K.-I. Mori, T. Maeno, Y. Nakagawa, Dieless forming of carbon fibre reinforced
plastic parts using 3D printer, Procedia Eng. 81 (October) (2014) 1595–1600.
[18] MarkForged, 2018. https://markforged.com/.
[19] F. van der Klift, Y. Koga, A. Todoroki, M. Ueda, Y. Hirano, 3D printing of continuous
carbon fibre reinforced thermo-plastic ( CFRTP ) tensile test specimens, Open J.
Compos. Mater. 6 (January) (2016) 18–27.
[20] J. Justo, L. Távara, L. García-Guzmán, F. París, Characterization of 3D printed long
fibre reinforced composites, Compos. Struct. 185 (July 2017) (2018) 537–548.
[21] L. García-Guzmán, L. Távara, J. Reinoso, J. Justo, F. París, Fracture resistance of 3D
printed adhesively bonded DCB composite specimens using structured interfaces:
experimental and theoretical study, Compos. Struct. 188 (December 2017) (2018)
173–184.
[22] G.W. Melenka, B.K.O. Cheung, J.S. Schofield, M.R. Dawson, J.P. Carey, Evaluation
and prediction of the tensile properties of continuous fiber-reinforced 3D printed
structures, Compos. Struct. 153 (2016) 866–875.
[23] H. Al Abadi, H.T. Thai, V. Paton-Cole, V.I. Patel, Elastic properties of 3D printed
fibre-reinforced structures, Compos. Struct. 193 (March) (2018) 8–18.
[24] R. Matsuzaki, M. Ueda, M. Namiki, T.-K. Jeong, H. Asahara, K. Horiguchi,
T. Nakamura, A. Todoroki, Y. Hirano, Three-dimensional printing of continuous-
fiber composites by in-nozzle impregnation, Sci. Rep. 6 (February) (2016) 23058.
[25] N. Li, Y. Li, S. Liu, Rapid prototyping of continuous carbon fiber reinforced poly-
lactic acid composites by 3D printing, J. Mater. Process. Technol. 238 (2016)
218–225.
[26] P. Bettini, G. Alitta, G. Sala, L. Di Landro, Fused deposition technique for continuous
fiber reinforced thermoplastic, J. Mater. Eng. Perform. 26 (2) (2016) 843–848.
[27] ASTM International, ASTM D638-14, Standard Test Method for Tensile Properties of
Plastics, ASTM International, 2014, pp. 1–17.
[28] Y. Nakagawa, K.I. Mori, T. Maeno, 3D printing of carbon fibre-reinforced plastic
parts, Int. J. Adv. Manuf. Technol. 91 (2017) 2811–2817.
[29] AUTODESK INC, Autodesk Inventor, (2017).
[30] MarkForged, 2018. https://www.eiger.io/signin.
[31] ASTM International, “ASTM D3039/D3039M-14, Standard Test Method for Tensile
Properties of Polymer Matrix Composite Materials, ASTM International, 2014, pp.
1–13.
[32] D. Croccolo, M. De Agostinis, G. Olmi, Experimental characterization and analytical
modelling of the mechanical behaviour of fused deposition processed parts made of
ABS-M30, Comput. Mater. Sci. 79 (2013) 506–518.
[33] A.C.P. Composites, “Mechanical Properties of Carbon Fiber Composite Materials,
Fiber/Epoxy Resin (120°C Cure), (2014) [Online]. Available www.acpsales.com ,
Accessed date: 28 July 2018.