Rapid Prototyping Journal

The impact of process parameters on mechanical properties of parts fabricated in PLA with an open-
source 3-D printer
Antonio Lanzotti Marzio Grasso Gabriele Staiano Massimo Martorelli
Article information:
To cite this document:
Antonio Lanzotti Marzio Grasso Gabriele Staiano Massimo Martorelli , (2015),"The impact of process parameters on
mechanical properties of parts fabricated in PLA with an open-source 3-D printer", Rapid Prototyping Journal, Vol. 21 Iss 5
pp. -
Permanent link to this document:
http://dx.doi.org/10.1108/RPJ-09-2014-0135
Downloaded on: 07 August 2015, At: 22:23 (PT)
References: this document contains references to 0 other documents.
To copy this document: permissions@emeraldinsight.com
The fulltext of this document has been downloaded 14 times since 2015*
Users who downloaded this article also downloaded:
Downloaded by UNIVERSITY OF EXETER At 22:23 07 August 2015 (PT)

Garrett W Melenka, Jonathon S. Schofield, Michael R. Dawson, Jason P. Carey, (2015),"Evaluation of dimensional accuracy
and material properties of the makerbot 3D desktop printer", Rapid Prototyping Journal, Vol. 21 Iss 5 pp. -
Eujin Pei, (2015),"Entry level additive manufacturing: the next frontier", Rapid Prototyping Journal, Vol. 21 Iss 5 pp. -
Gerald Anzalone, Bas Wijnen, Joshua M. Pearce, (2015),"Multi-material additive and subtractive prosumer digital fabrication
with a free and open-source convertible delta RepRap 3-D printer", Rapid Prototyping Journal, Vol. 21 Iss 5 pp. -

Access to this document was granted through an Emerald subscription provided by emerald-srm:463575 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for Authors service
information about how to choose which publication to write for and submission guidelines are available for all. Please
visit www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company manages a portfolio of
more than 290 journals and over 2,350 books and book series volumes, as well as providing an extensive range of online
products and additional customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee on Publication
Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive preservation.

*Related content and download information correct at time of download.

 The Impact of Process Parameters on Mechanical Properties
of Parts Fabricated in PLA with an Open-source 3-D Printer

1. INTRODUCTION
Rapid Prototyping (RP), Additive Manufacturing (AM) and 3D Printing (3DP) are three
synonyms used to describe a range of processes which makes it possible to fabricate
parts, by various materials, through an additive process, layer upon layer, starting directly
from a CAD model. The first term which was coined in the mid-80s to identify these
processes was Rapid prototyping [1,2]; Additive Manufacturing is the term that scientific
and technical communities prefer to use as the official standard term, according to ASTM
Downloaded by UNIVERSITY OF EXETER At 22:23 07 August 2015 (PT)

F42 and ISO TC261 committees [3,4]; 3D printing is by far the most popular term which is
being used today. As a matter of fact, Google shows about 23 million hits for 3D printing,
compared to 19,5 million for Additive Manufacturing and 5,93 million for Rapid Prototyping
[5].
From the first patent registered on March 11th 1986 [6] up to the present day, many things
have changed. In fact, many economists define the adoption and use of 3D printing today
as the “third industrial revolution”, following mechanization in the 19th century and
assembly-line mass production in the 20th century [7-9].
In particular, the turning point was marked by the expiration in 2009 of the Fused
Deposition Modeling (FDM) patent, a technique developed by S. Scott Crump (June, 9th
1992), and the widespread open-source movement which created significant cost
reductions for these new 3D printers.
Nowadays, there are many low cost 3D printers available on the market (<2000 €). They
fall into three categories: not-assembled “DIY” (Do It Yourself) open-source, fully
assembled open-source, and commercial systems with proprietary software.
One of the best-known open-source 3D printer project was developed in 2005 by Dr.
Adrian Bowyer, of the University of Bath (UK) and is known as the Rep-Rap (Replicating
Rapid prototyper) Project [10].
These self-replicating 3D printers can replicate a significant number of their own
components and produce objects using the fused-filament fabrication (FFF) typically in
PLA (Polylactic acid) or ABS (Acrylonitrile butadiene styrene) polymers. These systems
consist of a combination of printed components, stepper motors for 3D motion and
extrusion, and a hot-end for melting and depositing sequential layers of material, which are
controlled by an open-source microcontroller, such as the Arduino [11]. Different official 3D
printing machines have been released in the course of the Rep-Rap Project and nowadays
there are hundreds of variations.
Rep-Rap 3D printers demonstrated their use in a wide range of applications for
conventional prototyping and engineering [12], customizing scientific equipment [13] and
appropriate technology-related product manufacturing for sustainable development [14].
However, no detailed information is currently available about the mechanical properties of
parts printed by low-cost 3D printers.
Several studies were performed to investigate the modification of process parameters in
order to improve the mechanical properties of parts created by using the original Stratasys
FDM technology [15]. In [16] a study on the tensile strength of polycarbonate parts made
by Stratasys FDM with varying process parameters such as air gap, raster width and
raster angle was made. The main result was the experimental identification of the tensile
 strength for parts made from FDM and its comparison with the tensile strength of the
moulded and extruded PC parts. In [17] the experimental testing of ABS materials carried
out by different research teams to determine the strength of FDM is reported and
discussed giving a complete view of the effects produced by the different geometrical
parameters. In [18] the attention is focused on the determination of the tensile strength,
yield strength and modulus of elasticity for different values of build orientations of
polycarbonate materials, which is, together with ABS, widely used for this kind of
application. In [19] the identification of optimum values for the main geometrical
parameters was related to the manufacturing time in order to achieve minimum
maintenance costs. Research results suggest that orientation has a more significant
influence than the raster angle on the surface roughness and the mechanical behaviour of
the resulting fused deposition part. However, by examining literature, it can be noted that
the investigation of mechanical properties of parts processed by open-source low-cost 3D
printers is not currently receiving much attention. In particular, this applies to the material
Downloaded by UNIVERSITY OF EXETER At 22:23 07 August 2015 (PT)

used in this study, the polylactic acid (PLA) which, unlike ABS, has not been extensively
used in this kind of experiments [20].

In [21] the characteristics of mechanical properties of PLA components made using

different desktop open-source RepRap 3D printers are presented. The results indicated
that 3D printed components by the above 3D printers can be compared for what concerns
the tensile strength and the elastic modulus to parts printed by commercial 3-D printing
systems. However, it should be pointed out that a study should examine the process
parameters settings in relation to the mechanical properties.

Fig. 1 - Rep-Rap Prusa I3

The present study, which was carried out at the Fraunhofer Joint Lab IDEAS-CREAMI
(Interactive Design and Simulation - Center of Reverse Engineering and Additive
Manufacturing Innovation) of the University of Naples Federico II, by using an open-source
Rep-Rap 3-D printer, Prusa Iteration 3 (Fig. 1), reports on the impact of three process
parameters - layer thickness, infill orientation and the number of shell perimeters - on the
mechanical properties of parts fabricated in PLA. The purpose is to improve our
knowledge about optimal settings and assist users in the correct selection of process
parameters.

2. MATERIALS AND METHODS

The open-source RepRap Prusa I3 3D printer (Fig. 1) with 0.35 mm nozzle diameter was
used to fabricate the tensile test specimen (Fig. 2b) in 2.85 mm PLA material. Typical
values of main mechanical properties are reported in Table 1.

Table 1 – PLA mechanical properties

The 3D printer was assembled in two days at JL IDEAS-CREAMI and calibrated with an
accuracy of ±10 µm using the Mitutoyo 2046-08 (Mitutoyo, Japan), magnetic base dial
indicator.
The Marlin firmware open-source software and the Simplify 3D slicing software were used
to generate G-code files and to command and control the 3D printer for the fabrication of
the desired parts.
 The tensile testing was performed by a JJ Instruments - T5002 test machine type with
crosshead speed of 2.4 mm/min (which is the minimum speed allowed by the test
machine) and a load cell of 1100 N. Three specimens per experimental run i.e. a total of
60 parts were fabricated.

Fig. 2 a) - Specimen positioned and ready to be tested – b) Test specimen dimensions [mm]

Each specimen was placed in the grips of the electromechanical testing machine (Fig. 2a)
at a specified grip separation and loaded along the longitudinal axis until failure. Each
manual grip clamps the end of the specimen allowing the application of the testing load
through frictional contacts with the surfaces of the specimen.
The relative angular position between the loading line and the infill is highlighted in Fig. 3.
Downloaded by UNIVERSITY OF EXETER At 22:23 07 August 2015 (PT)

Fig. 3 - Boundary conditions and relative angle between infill and loading line

In order to determine the failure of the specimen at the minimum cross section, the
geometry of the specimen conforming to the ASTM:D638 has been modified. For such a
purpose, a parabolic profile for the fillet was used in combination with a curvature radius
equal to 1000 mm for the opposite faces of the central portion. This modification allows the
reduction of the stress concentration due to the curvature variation and ensures the failure
of the specimen at the minimum cross section.

2.1 Experimental plan

The experimental characterization of FDM parts made of ABS was performed considering
the tensile load. Due to the surface roughness and the manufacturing process the tensile
behaviour is likely to be more critical in comparison with the compressive behaviour.
Moreover, the key factor in the design process of each component is the identification of
the ultimate tensile strength (UTS) by which the admissible stress is derived when a
material shows a brittle behavior, as in the case which is being discussed here. For such a
reason, UTS values will be identified and discussed in the following sections.
On the basis of literature review about consolidated FDM (Fused Deposition Modelling)
processes [21-26] and through a focus group with open-source 3D printers experts, the
main control factors considered for the tensile strength of Rep Rap processed parts are
defined as follows:

- layer thickness [mm], the thickness of each slice of the building part. It is well-
known [1, 2] that the lower the layer thickness the better the accuracy of the part.
This is strictly related to the diameter of the nozzle used.
- Infill orientation [degree], the pattern angles for each layer and it can take values
ranging from 0° to ±180°.
- Number of shell perimeters, is the number of shells to use for the exterior skin of the
part. This one ranges from a minimum of 1 to a maximum equal to the number of
filaments extruded, having a diameter equal to the diameter of the nozzle, which
can be included within the perimeter of the layer.

The detailed factors were set constant, as shown in Table 2:

- Flow rate [%], the flow of the material that is extruded from the hot-end and is
expressed as a percentage of the number of the motor revolutions of the extruder,
to extrude 1 mm filament.
 - X-Y and Z deposition speed [mm/min], the hot-end speed. The lower the deposition
speed, the better the accuracy of the part and the longer the fabrication time.
- Fill Density [%], the quantity of material inside the part. The greater the percentage
of fill the better the mechanical properties of the part but the longer the printing time
and the amount of material to be used.
- Bed Temperature [°C], the temperature of the printing bed. This parameter depends
on the material and its correct setting increases as the adhesion on the bed
prevents warping phenomena.
- Printing Temperature [°C], the extrusion temperature of the material.
- Outline Overlap [%] it is a parameter which removes gaps between solid infill and
outline shells.
Table 2 - Constant factors

In order to build an empirical model for tensile strength, a sequential approach to

Downloaded by UNIVERSITY OF EXETER At 22:23 07 August 2015 (PT)

experimentation, based on central composite design (CCD), was adopted aiming at

characterizing and optimizing the process [27-29].
A CCD is an experimental design which is often used in the Response Surface
Methodology (RSM) to build a second order (quadratic) model for the response variable,
without the need of a complete three-level factorial experimental plan [27]. It contains an
embedded factorial or fractional factorial design with center points which are augmented
with a group of ’star points’ that allows an estimation of the curvature [27]. The values of
axial or star points should be selected in consideration of the rotatability of a CCD.
The CCD can fit second order polynomial. In order to reduce the experiment run, the half
factorial 2k design (k factors each at two levels) is considered. Maximum and minimum
value of each factor are coded into +1 and -1 respectively. The center point is coded into 0
and ±α level (star points) of each factor are also included. In the present work, in order to
get a reasonable estimate of the experimental error, six center points are considered.
Moreover, each center point was determined as the mean value over three repetitions [22].
The type of CCD used for the study is a rotatable design which uses the factor setting as
the star points and creates a full factorial design within those limits. The value of α is set as
భ
ߙ = [2௞ ]ర . It depends on the number of factors and is chosen to maintain the rotatability
property. In our case k=3 factors which gives α=1.682.
The Ultimate Tensile Stress (UTS) and the nominal strain at break (εf) were considered as
objective functions. UTS parameters were computed as the maximum value of the stress
reached during the test, whilst the εf is the strain attained at the UTS value.
All control factors and condition sets for the experimental treatments are listed in Tab. 3.

Table 3 - Control Factors and their levels

3. RESULTS
The experimental results obtained from the CCD runs related to the 60 samples are shown
in Table 4.

Table 4 – Experimental data obtained from the CCD runs

Experimental data, obtained from CCD design runs, were analyzed through the MINITAB
software using the following full quadratic response surface model:
 ௞ ௞

‫ߚ = ݕ‬଴ + ෍ ߚ௜ ܺ௜ + ෍ ߚ௜௜ ܺ௜ ܺ௜ + ෍ ෍ ߚ௜௝ ܺ௜ ܺ௝ + ߝ

௜ୀଵ ௜ୀଵ ௜ழ௝

where y is the response, Xi is the factor, k is total number of factors.

The ANalysis Of VAriance (ANOVA) of the response surface regression of UTS versus
layer thickness, infill orientation and the number of shell perimeters is shown in Table 5.

Table 5 – Response surface regression of UTS versus layer thickness,

infill orientation and number of shell perimeters

Table 6 shows the ANOVA results of the response surface regression of εf versus layer
Downloaded by UNIVERSITY OF EXETER At 22:23 07 August 2015 (PT)

thickness, infill orientation and number of shell perimeters.

Table 6 – Response surface regression of εf versus layer thickness,

infill orientation and, number of shell perimeters

The single p-values given in the ANOVA tables (Tabs. 5 and 6) support the influence on
the response variable UTS of X1 (layer thickness), X2 (infill orientation) and X3 (the number
of shell perimeters) and the advantage of a second order term for X3. The proposed model
is therefore:

‫ݕ‬௎்ௌ = ߚ଴ + ߚଵ ܺଵ + ߚଶ ܺଶ + ߚଷ ܺଷ + ߚଷଷ ܺଷ ଶ + ߚଶଷ ܺଶ ܺଷ + ߝ

Fig. 4 shows the predicted optimal combination of parameters for UTS maximization.

Fig. 4 Predicted optimal combination to maximize UTS

Fig. 5 shows the predicted optimal combination of parameters for Elastic modulus
maximization.

Fig. 5 Predicted optimal combination to maximize E

In Fig. 6 the contour plots of the predicted optimal combination to maximize UTS (left) and
Elastic modulus (right) are shown.

Fig. 6 Contour plots of the predicted optimal combination to maximize UTS (left) and E (right)

As shown in Figure 4 and 6 (left), the combination of parameters that maximize the UTS
value, by means of the response surfaces, is: layer thickness 0.2, infill orientation 0°
degrees and shell parameters 3. The predicted value, of max UTS, was 55.6 MPa.
As shown in Figure 5 and 6 (right), the combination of parameters which maximize the
Elastic modulus value, by means of the response surfaces, is: layer thickness 0.2, infill
orientation 0° degrees and shell parameters 3. The predicted value, of max Elastic
modulus, was 3736 MPa.
 4. DISCUSSION
In the present paper the UTS was evaluated together with the nominal strain at break and
the Elastic Modulus of FD printed specimens according to the variation of three process
parameters.
It is well known that parts printed by a Rep-Rap process show an anisotropy behaviour
and are very sensitive to the processing parameters affecting the meso-structure and the
fibre-to-fibre bond strength [23]. The printing process consists of deposition of semi-melted
filaments in a directional way, which implies that the mechanical behaviour of FD
components can be investigated using the same approach adopted for composite
materials. Rodriguez [23, 24] and Li [30] proposed analytical formulations to establish the
constitutive models and determine the elastic constants of FD components. The common
approach is that to evaluate the elastic constants using the mixture rule and introduce
corrective factors by which the fibre-to-fibre bonding strength can be taken into account.
Downloaded by UNIVERSITY OF EXETER At 22:23 07 August 2015 (PT)

However, in most applications the filament deposition in each layer is not along a single
direction. In particular, the specimens produced for the present work have different
numbers of filaments deposed along the perimeter, which are aligned along the
longitudinal direction. The remaining part of the cross section is filled by filaments deposed
along a variable angle with respect to the longitudinal direction. The deposition sequence
makes the identification of a single direction in each layer impossible, in this way limiting
the applicability of the analytical approach previously mentioned.
The number of perimeters is essentially a different view of the infill orientation. The effects
on the strength are related to the fact that they are always oriented along the longitudinal
direction of the specimen and thus contribute to withstand the axial load. Taking into
account these general considerations, the results of tests carried out on the 60 specimens
are being discussed in this paragraph.
First of all, it is suggested to modify the geometry of the specimen in order to overcome
anomalous failures occurred in the fillet during some preliminary tests. This problem is
common in literature [21, 25, 30] and it seems to be related to the approximation of the
continuous curvature with the stepped geometry produced by the filaments bonded in
sequence. The proposed modifications, which allow for testing all the 60 specimens
always having the failure at the minimum cross section (Fig. 7), consist of a modification of
the fillet geometry using a parabolic profile which is tangent to the middle part of the
specimen.
Fig. 7 Failure of all 60 specimens always in correspondence of the minimum cross section
The comparison of the overall results, in terms of mean values, shows variation between
different specimen sets which fall in the range 42.28 ÷ 53.59 MPa for the UTS, 2799.43 ÷
3497.63 MPa for E (Elastic modulus) and 0.01511 ÷ 0.01932 for the εf.. These results
confirm those reported in literature for the PLA [21] case. Moreover, it could be shown that
the general trend can also be compared with that observed for similar printed parts [23,
24]. In Tabs. 7÷9, the mean values and the standard deviation of all the results are
reported as a function of the three process parameters, each with five levels.

Table 7 – Mean effects of infill orientation

Table 8 – Mean effects of Layer thickness

Table 9 – Mean effects of the number of shells

 In Tabs 7 ÷ 9 a larger standard deviation for the Elastic modulus and UTS values at some
parameters values was observed. Moreover, the high variability of both mechanical
properties occurred at the same values of the parameters. This occurrence can be
explained, in accordance with [21, 22, 23], by considering that the fabrication process
accuracy of about 0.125 mm, in combination with a positive air gap between fibres, affects
the bonding among fibres in each layer. During the deposition process, the semi-melted
filament is deposed and a discontinuous bonding occurs. The nature of such phenomenon
is random and is due to this variability, different values of the stiffness and failure stress in
each test are observed. The effects of such variability which are reported in the present
paper are in accordance with the results discussed in a previous experimental work [21].

Fig. 8 Contour plot of the UTS in function of the three process parameters
Downloaded by UNIVERSITY OF EXETER At 22:23 07 August 2015 (PT)

The statistical analysis, for UTS, shows that all linear terms are significant while only the
number of shell perimeters square terms is significant. Interactions between infill
orientation and number of shell perimeters are significant. In this case R-sq(adj) is 68%.
The combined effects of process parameters on UTS values are reported in the graphs in
Fig. 8.
In particular, the strength increases as the infill orientation decreases with a rate which is
as greater as the layer thickness increases. Similar results were observed considering the
combined effects of the infill orientation and the number of perimeters. Increasing the
number of perimeters, which are oriented along the longitudinal direction, the number of
fibres, which withstand the tension load increase until the infill orientation is set to 0° and
the number of perimeters is ineffective, since all fibres are lying along the longitudinal
direction. Moreover, the strength increases as the layer thickness and the number of
perimeters increase. As reported by [22], this effect can be explained considering that by
increasing the layer thickness, a lower number of layers are needed for a given total
thickness and distortion effects are minimized with an increase in strength. By reducing the
number of perimeters, the tensile load is taken by the bonding surfaces between fibres and
the effects of air gap become significant. The reduction becomes more and more
significant as the infill orientation approaches 90°.

Fig. 9 a) - UTS and Elastic modulus vs Infill orientation – b) Normalized UTS vs infill orientation

In Fig 9 a the UTS and the Elastic modulus values are reported as a function of the infill
orientation. The trends are in accordance with those theoretically and experimentally
derived in [23, 24, 26, 30]. Moreover, in order to better validate these findings, the results
of the presented research were compared to those reported in literature. In particular,
Rodriguez [23, 24] produced many experimental data by testing specimens made of ABS
P400. The data produced and those reported in literature were normalized for comparison
purposes (see Fig. 9b). In Fig. 9b it can be noted that experimental values lay on the same
curve. It can be then stated that the trend followed by the elastic modulus in function of the
infill orientation does not depend on the material. Moreover, a linear fitting can be defined
considering the mean values of both the elastic modulus and UTS values, grouped by infill
orientation (see Fig. 10).

Fig. 10 Elastic modulus vs UTS values and linear regression line

 The combined effects of layer thickness and the number of perimeters on the values
achieved by the elastic modulus are reported in Fig. 11. The maximum value of the elastic
modulus is reached at the minimum value of the layer thickness and four perimeters. On
the other hand, the maximum value is reached at the minimum value of the infill orientation
and the maximum value of shell perimeters. In other words, the maximum value of the
Elastic modulus is reached when all fibres are oriented along the loading line. In this
condition, the specimen shows the highest stiffness since each fibre takes the load and the
effects of the fibre-to-fibre bonding is minimized. Moreover, the minimum value is reached
at the minimum value of the layer thickness and at the maximum and minimum value, in
the range investigated in the present work, of the number of perimeters. As for UTS
values, lower values of the layer thickness promote a stronger bond between adjacent
fibres in each layer and among layers, improving the strength. However, the competitive
effects of the layer thickness and the infill orientation reduce the stiffness. This result is in
accordance with the previous considerations since low values of the layer thickness or the
Downloaded by UNIVERSITY OF EXETER At 22:23 07 August 2015 (PT)

infill orientation result in a reduced specimens deformability and, as a consequence, in an

increase in stiffness [21].

Fig. 11 Contour plot of the Elastic modulus in function of the three process parameters

In Fig. 12, a microscopic view of the fracture surfaces of three tested specimens highlights
the occurrence of the phenomena previously described. There are shear planes at a
discontinuous bonding between fibres (Fig. 12a). Ductile failure occurs on different planes
parallel to the fracture surfaces being offset from each-other. In (Fig. 12b) fibres are pulled
until yielding with necking and material separation on plane which is normal to the loading
direction. Infill orientation close to 90° reduces strength and stiffness because part of the
tensile load is taken by the bonding surfaces among fibres, which are weaker and more
prone to fail (Fig. 12c).

Fig. 12 SEM images reporting details of the fracture surfaces

According to [24], the UTS value first decreases and then increases as the layer thickness
increases due to thermal diffusion properties. At the first increase in layer thickness, the
effect of the thermal diffusion is to make the fibre-to-fibre bonding stronger by reducing the
deformability of each layer. For a given value of the total thickness, an increase in the
layer thickness results in a reduction of the number of layers, in a decrease of the
deformability and consequently of the strength. However, a further increase in layer
thickness allows semi-melted material to flow between the increased space among fibres
and results in an increase in strength.
Fig. 13 shows the increase in strength with the number of perimeters. Moreover, the
number of filaments aligned along the loading direction increases as the number of
perimeters increase. The nonlinear trend of the UTS values with the number of perimeter
can be explained considering that the specimens were built using a positive gap, which
increases the possibility of discontinuous bonding surfaces between adjacent fibres.
Therefore, stress concentration at the local bonding promotes the premature failure of the
specimen.

Fig.13 - a UTS vs Number of perimeters – b UTS vs layer thickness

 There is a lack in literature regarding the effects of process parameters on the strain. The
statistical analysis, for εf, show that only a number of shell perimeters linear terms is
significant, whereas only layer thickness square terms is significant. The interaction
between infill orientation and the number of shell perimeters is significant. In this case R-
sq(adj) is 54%. In Fig. 14 the strain at failure in function of the layer thickness and the
number of perimeters are reported. An interesting effect is related to the layer thickness,
since the strain value is maximum at 0.15 mm and decreases as it approaches 0.2 mm.
The minimum value is reached at 0.1 mm.

Fig. 14 - a Strain at failure vs Number of perimeters – b strain at failure vs layer thickness

In order to further analyse interactions between UTS and εf, Fig. 15 shows the scatter plot
between UTS values and the nominal strain, when results are grouped using the number
Downloaded by UNIVERSITY OF EXETER At 22:23 07 August 2015 (PT)

of shell perimeters.

Fig. 15 Stress strain mean values estimated at the number of shell perimeters

The effects of the number of perimeters and the infill orientation on the strain at failure, the
UTS and the elastic modulus values should also be discussed considering the graphs
reported in Fig. 15, in combination with the two behaviours reported in the σ-ε curves (Fig.
16). These two parameters also affect the material behaviour passing from an essential
brittle fracture when the filaments are all oriented along the longitudinal direction of the
specimen, to a ductile behaviour as the number of perimeters is reduced and the infill
orientation approaches 90° degrees.

Fig. 16 σ-ε curves

Different failures were also observed depending on different materials behaviour. The
ductile fracture (Fig. 17a) corresponds to an inclined plane of fracture whilst for the brittle
fracture (Fig. 17b) the separation of the two parts of the specimen occurs on the same
plane of the minimum cross section.

Fig. 17 Details of the two types of failures – a ductile failure – b brittle failure

As previously stated, the strength is also affected by the ratio ρ between the portion of the
cross section filled with respect to the voids [30]. Fig. 18 shows pictures of the cross
sections of specimens having different ρ values. Note that the different layers are not all in
contact with each-other. During the loading phase, the interaction among the different
perimeters is not uniform over the thickness of specimens as well as those between
perimeters and the infill material. The effect is a stress concentration at some layers that
affect the specimen strength.

Fig. 18 - Images of the cross sections of specimens (20x) – a infill orientation 0° - b infill orientation 72 °
This consideration can justify some discrepancies in the experimental results presented in
this paper up to a certain degree only. In addition, there may be possible parameters (e.g.,
 micro and macro-geometrical variability, humidity and temperature) which are not being
investigated in this study, but could have affected the results.
In order to validate the statistical model adopted, two replications of the confirmation
experiment were carried out by setting parameters on their optimal set for UTS (layer
thickness 0.2, infill orientation 0° degrees and shell parameters 3). The mean UTS value
obtained from the confirmation experiment (52.3 MPa) is only 6% lower than the predicted
value by the mean response surface (55.6 MPa).
The Elastic modulus value predicted by means of response surfaces, which is
characterized by the following values: layer thickness 0.2, infill orientation 0° degrees and
shell parameters 3, is equal to 3736 MPa. The mean Elastic modulus value obtained from
tests was 3326,77 MPa, which is 10,9% lower than the predicted value.
Considering the high results variability, which was previously discussed, related to several
parameters some of which are not examined in the present paper, the predicted value is in
good accordance with the experimental one.
Downloaded by UNIVERSITY OF EXETER At 22:23 07 August 2015 (PT)

5. CONCLUSIONS
In this paper a Rep-Rap Prusa I3 open-source 3D printer was studied. By using the
Simplify 3D slicing software, it was possible to vary the common process parameters and
understand the impact on the mechanical properties of specimens in PLA.
The second order response surface model was used to derive the required relationship
among process parameters and the tensile strength.
The analysis of the experimental results made it possible to understand the impact of
control factors on the mechanical properties of specimens produced using the Rep - Rap
method. With regard to UTS values, it is possible to observe a decrease in strength as the
infill orientation approaches 90° degrees and an increase as the perimeters increase. An
initial increase is evident as the layer thickness approaches 0.18 mm. Beyond this value, a
reduction in strength values occurs. An interesting effect is related to the layer thickness,
since the strain value reaches its maximum at 0.15 mm and decreases as it approaches
0.2 mm. Considering the combined effects, further correlations were observed between
UTS and εf, when results were grouped with the infill orientation and the number of
perimeters. The reliability of the statistical model was validated by comparing the predicted
maximum UTS value with that established experimentally for a given parameter set.
The lack of data in literature and the high variability in experimental results, as well as the
effects of other factors, such as micro and macro-geometrical variability, humidity and
temperature suggest that further investigations are needed in order to improve the
knowledge about the mechanical behaviour of printed components using PLA.
The experimental results can be translated into practical suggestions for the settings of
process parameters with a view to improve the performance of 3D printers in relation to
mechanical properties.
The methodology utilised in this study can be applied for future analyses on other low cost
3D printers.
ACKNOWLEDGEMENTS

We would like to express our sincere gratitude to the anonymous referees for their
constructive comments on the paper, these have helped to improve the quality of the
paper significantly. We gratefully acknowledge Proff. Biagio Palumbo and Antonio Lepore
(University of Naples Federico II) for their suggestions and useful insights about statistical
analysis. We also thank Luciano Cortese (Istituto di Ricerche sulla Combustione, CNR,
Napoli, Italy) for the SEM images of the samples.
 REFERENCES
[1] Jacobs P.F., Rapid Prototyping & Manufacturing: Fundamentals of StereoLithography,
Society of Manufacturing Engineers (SME), 1992, pp. 434, ISBN: 0-87263-425-6.
[2] Jacobs P.F.,StereoLithography and other RP&M Technologies: from Rapid
Prototyping to Rapid Tooling, Society of Manufacturing Engineers (SME), 1996, pp.
450, ISBN:0-87263-467-1.
[3] ISO/ASTM 52915: 2013 E, Specification for Additive Manufacturing File Format (AMG)
Version 1.1.
[4] ISO/ASTM 52921: 2013 E, Terminology for Additive Manufacturing—Coordinate
Systems and Test Methodologies.
[5] Homepage, http://www.google.com.Accessed September, 4 2014.
[6] Hull C., Apparatus for Production of Three-Dimensional Objects by Stereolithography,
U.S. Patent 4,575,330, March 11, 1986.
[7] Huang S.H., Liu P., MokasdarA.,Hou L., Additive manufacturing and its societal
Downloaded by UNIVERSITY OF EXETER At 22:23 07 August 2015 (PT)

impact:a literature review, International Journal of Advanced Manufacturing

Technology,2013, 67:1191–1203.
[8] Council A., Petch M., 3D Printing: Rise of the Third Industrial Revolution, Gyges 3D,
2014, pp.116.
[9] A third industrial revolution, The Economist, 21 April 2012.
http://www.economist.com/node/21552901.
[10] Jones R., Haufe P., Sells E., Iravani P., Olliver V., Palmer C., Bowyer A., RepRap-the
replicating rapid prototyper. Robotica, 2011, Vol. 29, Issue 1, pp. 177-191
doi:10.1017/S026357471000069X.
[11] Kentzer J, Koch B, Thiim M, Jones RW, Villumsen E., An open source
hardwarebasedmechatronics project: the replicating rapid 3-D printer, Proc. of the 4th
International Conference on Mecatronics (ICOM), 2011. p. 1–8, 978-1-61284-435-0.
[12] Pearce J.M., Building research equipment with free open-source hardware,Science
2012, 337(6100):1303–4.
[13] Sells E, Bailard S, Smith Z, Bowyer A. Reprap: the replicating rapid prototype:
maximizing curstomizability by breeding the means of production. In: Piller FT, Tseng
MM, editors. Handbook of research in mass customization and personalization:
strategies and concepts, vol. 1. World Scientific; 2010. pp. 568–80.
[14] Pearce J.M., Morris Blair C., Laciak K.J., Andrews R., Nosrat A., Zelenika-Zovko I., 3-
D Printing of open source appropriate technologies for self-directed sustainable
development. J Sustain Dev 2010;3(4):17–29.
[15] Masood, S.H., Advances in Fused Deposition Modeling, In: Advances in Rapid
Manufacturing and Tooling, Masood, S.H. (Ed.)., Comprehensive Materials
Processing, 2014, Elsevier Science Direct, UK, pp. 69-91.
[16] Masood, S.H., Mau, K., Song, W.Q., Tensile properties of processed FDM
polycarbonate material, Materials Science Forum, 2010, 654-656, pp. 2556-2559.
[17] Novakova-Marcincinova, L., Novak-Marcincin, J., Testing of ABS material tensile
strength for fused deposition modeling rapid prototyping method, Advanced Materials
Research, 2014, 912-914, pp. 370-373.
[18] Smith, W.C., Dean, R.W., Structural characteristics of fused deposition modeling
polycarbonate material, Polymer Testing, 2013, 32 (8), pp. 1306-1312.
[19] Durgun, I., Ertan, R., Experimental investigation of FDM process for improvement of
mechanical properties and production cost, Rapid Prototyping
Journal, 2014, 20 (3), pp. 228-235.
[20] Stephen A.O., Dalgarno K.W., Munguia J., Quality assurance and process monitoring
of fused deposition modelling made parts, Advanced Research in Virtual and Rapid
Prototyping, 2014, Taylor & Francis Group, London, ISBN 978-1-138-00137-4.
 [21] Tymrak B.M., Kreiger M., Pearce J.M., Mechanical properties of components
fabricated with open-source 3-D printers under realistic environmental conditions,
Materials and Design, 2014, 58:242–246.
[22] Sood, Anoop Kumar, Ohdar R. K., Mahapatra S.S., Parametric appraisal of
mechanical property of fused deposition modelling, Material and Design, 2010, 31, 287
– 295.
[23] Rodriguez, Jose F., Thomas, James P, Renaud, John E, Mechanical behavior of
acrylonitrile butadiene styrene fused deposition materials modeling, Rapid Prototyping
Journal, 2003, 9, 4, 219 – 230.
[24] Rodriguez, Jose F., Thomas, James P, Renaud, John E, Design of fused-deposition
ABS components for stiffness and strength , Mechanical Design Journal, 2003, 125,
545 – 551.
[25] Vega, V., Clements, J., Lam, T., Abad, A., Fritz, B., Ula, N. and Es-Said, O.S., The
effect of layer orientation on the mechanical properties and microstructure of a
Downloaded by UNIVERSITY OF EXETER At 22:23 07 August 2015 (PT)

polymer, Journal of Materials Engineering and Performance, 2011, 978 – 988.

[26] Ahn, Sung-Hoon, Montero, Michael, Odell, Dan, Roundy, Shad and Wright, Paul K.,
Anisotropic material properties of fused deposition modeling ABS, Rapid Prototyping
Journal, 2002, 8, 4, 248-257.
[27] Coleman D.E., Montgomery D.C., A systematic approach to planning for a designed
industrial experiment. Techno metrics. 35(1), 1–12, 1993.
[28] Lanzotti A., Robust design of car packaging in virtual environment, International
Journal on Interactive Design and Manufacturing, 2008, 2:39–46.
[29] Palumbo, B., De Chiara, G., Sansone, F., Marrone, R. Technological scenarios of
variation transmission in multistage machining processes, 2011 Quality and Reliability
Engineering International 27 (5) PP. 651 – 658.
[30] Li, L, Sun, Q., Bellehumeur, C., Gu, P., Composite modeling and analysis for
fabrication of FDM prototypes with locally controlled properties, Journal of
Manufacturing Processing, 2002, 4, 2, 129 – 141.
 List of Figures

Fig. 1 - Rep-Rap Prusa I3

Fig. 2 a) - Specimen positioned and ready to be tested – b) Test specimen dimensions [mm]
Fig. 3 - Boundary conditions and relative angle between infill and loading line
Fig. 4 Predicted optimal combination to maximize UTS
Fig. 5 Predicted optimal combination to maximize E
Fig. 6 Contour plots of the predicted optimal combination to maximize UTS (left) and E (right)
Fig. 7 Failure of all 60 specimens always in correspondence of the minimum cross section
Fig. 8 Contour plot of the UTS in function of the three process parameters
Fig. 9a) -UTS and Elastic modulus vs Infill orientation – b) Normalized UTS vs infill orientation
Downloaded by UNIVERSITY OF EXETER At 22:23 07 August 2015 (PT)

Fig. 10 Elastic modulus vs UTS values and linear regression line

Fig. 11 Contour plot of the Elastic modulus in function of the three process parameters
Fig. 12 SEM images reporting details of the fracture surfaces
Fig.13 - a UTS vs Number of perimeters – b UTS vs layer thickness
Fig. 14 - a Strain to failure vs Number of perimeters – b strain to failure vs layer thickness
Fig. 15 Stress strain mean values estimated at the number of shell perimeters
Fig. 16 σ-ε curves
Fig. 17 Details of the two types of failures – a ductile failure – b brittle failure
Fig. 18 Images of the cross sections of specimens (20x) – a infill orientation 0° - b infill orientation 72 °
Downloaded by UNIVERSITY OF EXETER At 22:23 07 August 2015 (PT)

Fig. 1 - Rep-Rap Prusa I3

Fig. 2 a) - Specimen positioned and ready to be tested – b) Test specimen dimensions [mm]
Downloaded by UNIVERSITY OF EXETER At 22:23 07 August 2015 (PT)

Fig. 3 - Boundary conditions and relative angle between infill and loading line

UTS [MPa]

Fig. 4 Predicted optimal combination to maximize UTS

Fig. 5 Predicted optimal combination to maximize E

Downloaded by UNIVERSITY OF EXETER At 22:23 07 August 2015 (PT)

Fig. 6 Contour plots of the predicted optimal combination to maximize UTS (left) and E (right).

Fig. 7 Failure of all 60 specimens always in correspondence of the minimum cross section
Downloaded by UNIVERSITY OF EXETER At 22:23 07 August 2015 (PT)

Fig. 8 Contour plot of the UTS in function of the three process parameters.

Fig. 9 a) - UTS and Elastic modulus vs Infill orientation – b) Normalized UTS vs infill orientation
 Fig. 10 Elastic modulus vs UTS values and linear regression line.
Downloaded by UNIVERSITY OF EXETER At 22:23 07 August 2015 (PT)

Fig. 11 Contour plot of the Elastic modulus in function of the three process parameters.

Fig. 12 SEM images reporting details of the fracture surfaces.

 Fig.13 - a UTS vs Number of perimeters – b UTS vs layer thickness.
Downloaded by UNIVERSITY OF EXETER At 22:23 07 August 2015 (PT)

Fig. 14 - a Strain to failure vs Number of perimeters – b strain to failure vs layer thickness.

Fig. 15 Stress strain mean values estimated at the number of shell perimeters
Downloaded by UNIVERSITY OF EXETER At 22:23 07 August 2015 (PT)

Fig. 16 σ-ε curves

Fig. 17 Details of the two types of failures – a ductile failure – b brittle failure.
Downloaded by UNIVERSITY OF EXETER At 22:23 07 August 2015 (PT)

Fig. 18 - Images of the cross sections of specimens (20x) – a infill orientation 0° - b infill orientation 72 °
 List of Tables

Table 1 – PLA mechanical properties

Table 2 - Constant factors
Table 3 - Control Factors and their levels
Table 4 – Experimental data obtained from the CCD runs
Table 5 – Response surface regression of UTS versus layer thickness, infill orientation and number of shell
perimeters
Table 6 – Response surface regression of εf versus layer thickness, infill orientation and, number of shell
perimeters
Table 7 – Mean effects of infill orientation
Table 8 – Mean effects of Layer thickness
Downloaded by UNIVERSITY OF EXETER At 22:23 07 August 2015 (PT)

Table 9 – Mean effects of the number of shells

 Table 1 – PLA mechanical properties
Physical Nominal Value
Specific Gravity (23°C) 1,24 to 1,26 g/cm³
Melt Mass-Flow Rate (MFR)
210°C/2.16 kg 6,0 to 78 g/10 min
190°C/2.16 kg 1,5 to 36 g/10 min
Molding Shrinkage
Flow : 73°F 3,7E-3 to 4,1E-3 mm/mm
73°F 0,30 to 1,1 %
Mechanical Nominal Value
Tensile Modulus (23°C) 2020 to 3550 MPa
Tensile Strength Yield, (23°C) 15,5 to 72 MPa
Downloaded by UNIVERSITY OF EXETER At 22:23 07 August 2015 (PT)

Tensile Strength Break, (23°C) 14 to 70 MPa

Tensile Elongation Yield, (23°C) 9,8 to 10 %
Tensile Elongation Break, (23°C) 0,50 to 9,2 %
Flexural Modulus (23°C) 2392 to 4930 MPa
Flexural Strength (23°C) 48 to 110 MPa

Table 2 - Constant factors

Constant factors Value

Flow rate (%) 100
x-y deposition speed (mm/min) 3600
z deposition speed (mm/min) 1000
Fill Density (%) 100
Bed Temperature (°C) 60
Printing Temperature (°C) 205
Outline Overlap (%) 15

Table 3 - Control Factors and their levels

FACTOR

Layer thickness [mm] Infill orientation [degree] Number of shell perimeters

-1,682 0,1 0 2
-1 0,12 18 3
LEVEL

0 0,15 45 4
1 0,18 72 5
1,682 0,2 90 6
 Table 4 – Experimental data obtained from the CCD runs

FACTOR UTS [Mpa] εf

Number of Layer Infill Number of
Treatment Standard Standard
Replications Thickness Orientation Shell Mean Mean
Perimeters
Deviation Deviation
[mm] [deg]
1 3 0,12 18 3 47,65 0,82 0,017 0,0007
2 3 0,18 18 3 52,63 1,19 0,016 0,0002
3 3 0,12 72 3 37,62 1,15 0,015 0,0001
4 3 0,18 72 3 41,40 0,72 0,015 0,0004
5 3 0,12 18 5 50,33 1,43 0,017 0,0004
6 3 0,18 18 5 51,42 0,86 0,017 0,0003
7 3 0,12 72 5 46,71 0,30 0,018 0,0003
8 3 0,18 72 5 50,36 0,14 0,017 0,0003
9 3 0,1 45 4 49,29 2,16 0,015 0,0011
10 3 0,2 45 4 48,49 0,97 0,016 0,0004
Downloaded by UNIVERSITY OF EXETER At 22:23 07 August 2015 (PT)

11 3 0,15 0 4 53,59 0,77 0,017 0,0003

12 3 0,15 90 4 43,39 1,73 0,017 0,0006
13 3 0,15 45 2 42,28 2,31 0,016 0,0009
14 3 0,15 45 6 50,67 1,16 0,019 0,0002
15 18 0,15 45 4 48,37 3.61 0,017 0,0010

Table 5 – Response surface regression of UTS versus layer thickness,

infill orientation and number of shell perimeters

Source DF Adj SS Adj MS F-value p-value

Model 9 827,56 91,951 14,64 0
Linear 3 688,9 229,632 36,57 0
Layer Thickness [mm] 1 32,41 32,413 5,16 0,027
Infill Orientation [deg] 1 407,95 407,95 64,97 0
Number of Shell Perimeters 1 248,53 248,532 39,58 0
Square 3 28,93 9,644 1,54 0,217
Layer Thickness [mm]*Layer Thickness [mm] 1 0,09 0,087 0,01 0,907
Infill Orientation [deg]*Infill Orientation [deg] 1 0,39 0,391 0,06 0,804
Number of Shell Perimeters*Number of Shell Perimeters 1 28,22 28,218 4,49 0,039
2-Way Interaction 3 109,73 36,578 5,83 0,002
Layer Thickness [mm]*Infill Orientation [deg] 1 0,69 0,685 0,11 0,742
Layer Thickness [mm]*Number of Shell Perimeters 1 6,06 6,061 0,97 0,331
Infill Orientation [deg]*Number of Shell Perimeters 1 102,99 102,988 16,4 0
Error 50 313,97 6,279
Lack-of-Fit 5 46,81 9,362 1,58
Pure Error 45 267,16 5,937
Total 59 1141,54
Model Summary
S R-sq R-sq(adj) R-sq(pred)
250589 72,50% 67,54% 65,26%
 Table 6 – Response surface regression of εf versus layer thickness,
infill orientation and, number of shell perimeters

Source DF Adj SS Adj MS F-value p-value

Model 9 0,000053 0,000006 8,62 0
Linear 3 0,000023 0,000008 11,36 0
Layer Thickness [mm] 1 0 0 0,26 0,616
Infill Orientation [deg] 1 0 0 0,37 0,548
Number of Shell Perimeters 1 0,000023 0,000023 33,46 0
Square 3 0,000023 0,000008 11,02 0
Layer Thickness [mm]*Layer Thickness [mm] 1 0,000022 0,000022 32,4 0
Infill Orientation [deg]*Infill Orientation [deg] 1 0,000001 0,000001 1,89 0,176
Number of Shell Perimeters*Number of Shell Perimeters 1 0 0 0,43 0,514
Downloaded by UNIVERSITY OF EXETER At 22:23 07 August 2015 (PT)

2-Way Interaction 3 0,000007 0,000002 3,47 0,023

Layer Thickness [mm]*Infill Orientation [deg] 1 0 0 0,09 0,764
Layer Thickness [mm]*Number of Shell Perimeters 1 0 0 0,14 0,714
Infill Orientation [deg]*Number of Shell Perimeters 1 0,000007 0,000007 10,17 0,002
Error 50 0,000034 0,000001
Lack-of-Fit 5 0,000009 0,000002 3,28
Pure Error 45 0,000025 0,000001

59 0,000087
Total
Model Summary
S R-sq R-sq(adj) R-sq(pred)
0,000827 60,80% 53,74% 46,98%

Table 7 – Mean effects of infill orientation.

Εf UTS E
Infill
Orient Standard Standard Standard
Mean Mean Mean
Deviation Deviation Deviation
0,00 0,01694 0,00032 53,59 0,77 3388,57 148,00
18,00 0,01665 0,00044 50,51 2,14 3227,89 176,85
45,00 0,01711 0,00146 48,09 3,59 3087,20 298,77
72,00 0,01619 0,00113 44,02 5,13 2963,35 248,13
90,00 0,01739 0,00057 43,39 1,73 2799,43 85,50

Table 8 – Mean effects of Layer thickness.

Εf UTS E
layer
thickness Standard Standard Standard
Mean Mean Mean
Deviation Deviation Deviation
0,10 0,01511 0,00107 49,29 2,16 3497,63 136,12
0,12 0,01653 0,00108 45,58 5,07 2962,15 207,59
0,15 0,01741 0,00121 48,01 4,24 3030,56 289,56
0,18 0,01631 0,00062 48,95 4,68 3229,09 221,48
0,20 0,01615 0,00040 48,49 0,97 3256,77 75,92
 Table 9 – Mean effects of the number of shells.

Εf UTS E
Number
of shell Standard Standard Standard
Mean Mean Mean
Deviation Deviation Deviation
2,00 0,01552 0,00088 42,28 2,31 2951,83 111,50
3,00 0,01594 0,00092 44,82 6,07 3002,23 286,04
4,00 0,01706 0,00116 48,49 3,71 3125,03 314,28
5,00 0,01690 0,00049 49,71 2,00 3189,01 173,68
6,00 0,01932 0,00021 50,67 1,16 2857,80 107,22
Downloaded by UNIVERSITY OF EXETER At 22:23 07 August 2015 (PT)