Results in Engineering 21 (2024) 101926

Contents lists available at ScienceDirect

Results in Engineering
journal homepage: [Link]/journal/results-in-engineering

Optimization of infill density, fiber angle, carbon fiber layer position in 3D

printed continuous carbon-fiber reinforced nylon composite
Basim El Essawi a, 1, Said Abdallah b, 1, Shafahat Ali c, 1, Abdallah Nassir Abdo Mohammed d,
Rahmat Agung Susantyoko d, Salman Pervaiz e, *
a
Department of Automatic Control, Lund University, P.O. Box 118, SE-221 00, Lund, Sweden
b
Abu Dhabi Vocational Education and Training Institute (ADVETI), Rabdan, Abu Dhabi, United Arab Emirates
c
School of Engineering, University of Guelph, Guelph, Canada
d
Dewa R&D Center, Dubai Electricity and Water Authority, P.O Box 564, Dubai, United Arab Emirates
e
Department of Mechanical and Industrial Engineering, Rochester Institute of Technology - Dubai Campus, P.O. Box 341055, Dubai, United Arab Emirates

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

Keywords: Composite materials have gained much attention in various industries, such as aerospace, automotive, sports,
3D printing marine, and construction, as these sectors rely on high-performance, durable, and cost-effective materials. Such
Onyx materials offer high strength, stiffness and heat resistance. However, the influence of printing parameters
Nylon
especially the position of carbon fiber layer on such material is rarely found in literature. The current study
Carbon fiber
Reinforced composites
focuses on optimizing different printing and testing parameters such as carbon fiber layer position, infill density,
fiber angle, and strain rate in 3D printed carbon-fiber reinforced nylon composite. The study also recommended
the optimal combination of these parameters for maximizing the mechanical strength and energy absorption of
related 3D printed parts. The investigation revealed that the most optimum condition was 80% infill density,
fiber angle of 0◦ , carbon fiber layer position of 12–13, and strain rate of 10 mm/min. It has been found in the
study that fiber angle was the most dominant input parameter with a contribution of 54.13%, whereas infill
density was the second dominant parameter with a contribution of 16.25%. The study also found that the po­
sition of the carbon fiber layer has comparatively less effect on the final mechanical properties of 3D printed
parts, with a contribution of 10.12%. To facilitate the optimization, the outcomes will be helpful for designing
and manufacturing 3D printed carbon-fiber reinforced nylon composite parts.

1. Introduction path in additive manufacturing stem from the need for materials with
higher mechanical strength, reinforcements of polymer materials, either
Additive manufacturing or 3D printing is a technique by which using shredded carbon fiber, or using it as matrix materials fusing with
simulated profiles and complex geometries are directly fabricated by Kevlar, glass fiber, or carbon fiber to produce mechanically superior
adding material layer by layer to form a 3D model of the computer- specimens [2,5,6]. Onyx is a filament material that is composed of
designed product [1]. With the prevalent increase in the popularity of partial nylon and partially chopped carbon fiber; the combination pro­
additive manufacturing techniques, it came the requirement for higher vides not only more tensile strength but also better flexural properties
availability of materials with greater mechanical properties to be used in compared to pristine nylon filament thermoplastics [7]. Onyx can also
additive manufacturing [2]. Whilst 3D printing takes away the incon­ be used as a matrix material to hold together continuous fiber samples; a
venience of forming complex geometries using advanced manufacturing combination of Onyx with continuous carbon fiber layers may provide a
methods, the drawback it brings is the lack of final component strength solution to the mechanical property demand from additive manufac­
relative to parts manufactured by conventional means [3]. This is tured (AM) parts [6].
because 3D printing often uses polymer or thermoplastics, which are Additionally, the mechanical properties of AM parts may vary
mechanically weaker than metals [4]. A growing interest and research greatly based on preprinting process parameters set by the user. These

* Corresponding author.
E-mail address: sxpcad@[Link] (S. Pervaiz).
1
Was previously affiliated with Rochester Institute of Technology – Dubai during the conduction of this study.

[Link]
Received 12 September 2023; Received in revised form 17 January 2024; Accepted 16 February 2024
Available online 8 March 2024
2590-1230/© 2024 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC license ([Link]
nc/4.0/).
B. El Essawi et al. Results in Engineering 21 (2024) 101926

could include shell thickness [8], bed and extrusion temperatures [9], of user-controllable factors like; infill orientation, layering, moisture,
infill patterns, printer velocity [10] and strain rate [10,11], layer and user-noncontrollable factors, such as filament spool-to-spool varia­
orientation, layer height, and feed rate, amongst many others. Infill tions, filament lot-to-lot variations, and printer-to-printer variations, on
density, which defines the amount of material that fills the shape of the the mechanical strength of 3D-printed Onyx. Their research showed a
material, synonymous with the volume density of the part, is one of the 50% higher tensile strength for 0◦ orientation as compared to 90◦ , as
most discussed process parameterrs [2,4,5,12] in literature, as well as well as significant differences in flexural and tensile properties when
having a great influence towards mechanical strength. Dave et al. [13] soaked in water, higher moisture levels included upwards of 40 MPa
researched the effects of infill patterns and densities on the tensile decrease in ultimate strength, and a 25% increase in elongation before
strengths of 3D-printed specimens. They modeled three densities, three fracture [21]. Traditional testing methods tend to be tedious as many
patterns, and three printing orientations; 60%, 80%, and 100% den­ iterations can be not only time-consuming but also cost ineffective.
sities; rectilinear, concentric, and Hilbert curve patterns. Their results Gonabadi et al. [22] experimented on the implication of FEM modeling
proved that increasing infill densities had a positive effect on tensile to investigate the effect of raster angles and infill densities using raster
strength increasing overall across all patterns. Similarly, Gunasekaran angles of 0◦ , 45◦ , and 90◦ for both flat and on edge printing orientations,
et al. [14] tested a wider range of infill densities between 25% and 100% which was then compared to realistic conducted experiments for vali­
whilst also concluding the positive correlation between infill density and dation, they discussed that the FEM model was only acceptably accurate
mechanical properties [14]. for an infill density of 100% and lower densities would not have an
Srinivasan et al. [15] also conducted tests to compare wear rates accurate representation in their modelling [22]. Verbeeten et al. [11]
between carbon fiber PLA specimens with varying infill densities and studied the effect of strain rate on the mechanical properties of
layer thickness, the experiment saw infill densities of 60%, 70%, and 3D-printed ABS as well as a carbon-fiber-reinforced version of the ABS;
80% and they concluded that with lower layer thickness and increasing in their findings they recorded an increase in strain at yield point as the
infill density did they achieve the lowest wear rate [15]. However, strain rate increases, with little to no effect on Young’s modulus. Their
increasing infill density costs more material. A trade-off between infill experiment also checked the strain rate dependency on the yield
density and mechanical properties needs to be achieved to extract the strength of the samples using an Erying-type flow equation to create
highest cost efficiency. Abdullah et al. [16] investigated the difference in regression models on the log-strain curves, concluding that the carbon
properties for a marginal reduction in infill density, and thus tested fiber reinforced samples had higher dependence on strain rate relatively
specimens of carbon-fiber reinforced ABS with infill densities of 70%, to the unfilled ABS [11].
80%, 90%, and 100% keeping all other conditions the same. The results Most of the literature presented currently discusses infill densities
of their experiment proved a large gap between 100% infill density and and patterns, raster angles, strain rates, and their effects on 3D-printed
90% infill density with over 20% loss in compressive strength; per polymers. The current literature addressing Onyx as a matrix material
contra, they discovered that the difference between 80% and 90% infill for carbon fiber has also addressed the aforementioned printing pa­
densities was negligible at around 3% difference in strength [16]. Raster rameters. Several studies were also conducted focusing on the process­
angle is also another noteworthy parameter, as it affects largely not only ing of natural fibers. Arunachalam et al. [23] focused the study on the
the tensile but also the flexural properties of the final workpiece [17]. processing of cotton/bamboo woven fabric reinforced epoxy compos­
This is due to some orientations simply being more susceptible to shear ites. Response surface analysis was conducted and lower thrust forces
strain as compared to others. Zhao et al. [18] conducted a study were found at lower diameter, higher speed and low feed rate. In
comparing the effect of various raster angles on the overall properties of another study Arunachalam et al. [24], focused on processing of Typha
3D-printed PLA; they discovered that with increasing angles between angustifolia fiber reinforced (TAFR) composites under unidirectional
0◦ and 90◦ the tensile properties almost always improved [18]. and bidirectional orientations. It was found that lower delamination was
Mishra et al. [19] investigated the effect of the mesostructure, which present under lower feeds and speed and higher drill diameter. Ahmed
is a function of infill density and pattern, on the impact strength of AM et al. [25] investigated the mechanical performance of natural com­
PLA parts by taking samples of varying infill densities and patterns and posite materials created using buffing waste and sugar fiber. The study
performing the standard Izod impact test. From their results, they revealed that optimal conditions increased the tensile strength and
determined that 85% infill density had the highest energy absorption; hardness by 61% and 37% respectively. The composite provided an
this was because, at lower infill densities, the gaps in the infill would alternate material for packaging and furnishing applications.
lead to stress concentrations, whilst at higher levels, the compact infill However, the effect of the position of the carbon fiber layers has not
would lead to easier crack propagation, and thus at 85% infill the been discussed in any prominent literature papers. This study aims to
highest energy absorption was recorded [19]. Suteja [20] conducted tackle the topic of carbon fiber placement as well as test the effect of
research relating the infill patterns, densities, and angles on the per­ various infill densities (30, 40, 50, 60, 80 %), raster angles (0◦ , 45◦ /-45◦ ,
formance of the 3D printing process, examining the possibilities of de­ 90◦ ), as well as strain rate dependency with strain rate trials of (1, 5, 10)
fects, filament length, and printing times. A positive correlation was s− 1. The cumulation of the various parameters may lead to a require­
claimed for densities on filament length, as expected, whilst 90◦ infill ment of over 100 test samples, with each being repeated three times for
angle was discovered to have the least filament length, and corre­ repeatability validation. Instead of performing the tedious and unnec­
spondingly least requirement for material [20]. Žmindák et al. [6] essarily costly experimentation, an alternate method of testing was
performed tensile test experiments to study the effect of carbon fiber performed using the Taguchi approach, shortening the total number of
orientations with 3D-printed Onyx as matrix material: three specimens required samples to 27. Furthermore, Grey relational analysis will be
of 3D-printed ASTM D 3039 standard rectangular profiles with com­ conducted to determine the homogenization and the weightage of each
posite tabs; each of the parts had orientations of CRFF layers at 90◦ , 45◦ , parameter.
and 0◦ respectively. Both FEA and tensile tests were conducted, and the
results were compared; the first specimen had the lowest tensile strength 2. Experimental design and methodology
and the greatest flexural capabilities, with a UTS of ~28 MPa and max
deformation of 38%, specimen 2 had a UTS of 75 MPa and max defor­ 2.1. Material and 3D printing of samples
mation of ~2.7%, while the third specimen had a UTS of ~190 MPa and
a max deformation of ~2.4%. Noteworthily, the 45◦ orientation speci­ For this experiment, the material used was the Markforged
mens were arranged symmetrically, alternating between 45/-45◦ ; this is composite-based material Onyx, a matrix material consisting of nylon
done in order to prevent torsional twisting [6]. and micro carbon fibers, noted for its relatively high flexural strength
Ma et al. [21] examined the relative effects on mechanical properties and toughness as an additive manufacturing filament. The sample

2
B. El Essawi et al. Results in Engineering 21 (2024) 101926

structures tested consisted of a simple sandwich structure with Onyx as densities of 30%, 40%, 50%, 60%, 70%, and 80% were used, raster
the matrix material and 2 layers of continuous carbon fiber. This study angles alternated between 0, 90, and ±45◦ , and carbon fibers were
aims to investigate the effect of preprinting parameters on the tensile located on either the 12th and 13th layers, the 18th and 19th layers, or
properties of continuous carbon fiber reinforced Onyx, specifically the the 24th and 25th layers, with other preprinting parameters such as
effect of infill density, fiber angle, and carbon fiber position. Infill layer height, printing temperatures and speeds etc. Kept constant. The

Fig. 1. (a) Markforged X7 3D printer used in this work (b) Layer composition includes floor, infill, fiber, and roof layers. (Example illustration for 30 % infill with
0◦ carbon fiber sample).

Fig. 2. Carbon fiber angle toolpath at (a) 0◦ , (b) ± 45◦ , (c) 90◦ .

Fig. 3. Infill density toolpath at (a) 30 %, (b) 40 %, (c) 50 %, (d) 60 %, (e) 70 %, (f) 80 %.

3
B. El Essawi et al. Results in Engineering 21 (2024) 101926

Table 1 samples were printed using the layer height of 0.125 mm, whereas the
Experimental taguchi design. printing temperature was controlled by the software itself without dis­
Exp. A: Infill B: C: Carbon fiber layer D: playing it. Fig. 1a and b shows a schematic illustration of the sample.
No Density (%) Fiber position Strain rate The Eiger slicing software produced toolpath with mandatory floor and
angle (◦ ) (mm/min) roof layers having fixed ±45◦ raster angle, infill layers as per user in­
1 30 0 12–13 1 puts, and fiber layers as per user inputs. The software automatically
2 30 45 18–19 5 inserts floor and roof layers to enclose the fiber layers; this means the
3 30 90 24–25 10 fiber layers do not directly connect to the infill layers. The testing pa­
4 40 0 12–13 5
5 40 45 18–19 10
rameters were also put into consideration, varying strain rates between
6 40 90 24–25 1 1, 5, and 10 mm/min; the tensile properties that were tested consist of
7 50 0 18–19 1 the Yield strength, Young’s Modulus (modulus of elasticity), ultimate
8 50 45 24–25 5 tensile strength, toughness, resilience, and the strain at the breaking
9 50 90 12–13 10
point or the maximum strain before fracture.
10 60 0 24–25 10
11 60 45 12–13 1 The samples were prepared using Fused Filament Fabrication (FFF)
12 60 90 18–19 5 and continuous fiber reinforcement (CFR) on a Markforged X7 3D
13 70 0 18–19 10 printer. The X7 is an industrial-grade 3D printer with excellent printing
14 70 45 24–25 1 quality and repeatability, specialized for printing Onyx-based materials.
15 70 90 12–13 5
16 80 0 24–25 5
The printer comes with two nozzles, one for the filament during fused
17 80 45 12–13 10 filament fabrication, while the second nozzle allows for the printing of
18 80 90 18–19 1 continuous fibers, such as Kevlar or carbon fiber. The samples produced

Table 2
The Taguchi experimental design and output results.
Sample Number Results

Ultimate Tensile Strength (MPa) Strain at break (%) Toughness (MPa) Young’s Modulus (GPa) Yield strength (MPa) Resilience (MPa)

1 59.814 6.907 205.264 1.133 35.197 0.548

2 31.003 9.349 134.659 0.793 23.000 0.341
3 25.408 7.996 117.169 0.683 19.657 0.283
4 54.451 6.461 172.487 1.267 38.053 0.612
5 34.009 11.088 246.080 0.767 23.467 0.360
6 22.308 8.584 111.017 0.680 16.550 0.201
7 57.615 6.215 172.878 1.107 36.130 0.600
8 29.930 7.146 112.247 0.753 23.000 0.351
9 27.809 10.995 195.512 0.580 16.160 0.225
10 55.276 8.038 221.590 0.747 42.037 1.192
11 29.060 8.075 144.532 0.970 23.080 0.275
12 31.033 11.270 235.093 0.760 22.690 0.341
13 59.448 7.043 198.074 0.977 35.753 0.676
14 29.782 7.483 121.930 0.953 24.480 0.316
15 30.536 12.647 259.977 0.837 22.533 0.304
16 55.058 5.129 120.854 1.497 37.607 0.476
17 32.269 12.287 286.339 0.780 23.310 0.350
18 37.094 10.309 231.473 0.933 26.110 0.369

Fig. 4. Specimen display, pre-test.

4
B. El Essawi et al. Results in Engineering 21 (2024) 101926

were ASTM D638 standard dog-bone specimens to be tested in uniaxial

tensile tests, as shown in Figs. 2 and 3. Fig. 2a, b, and 2c show the
toolpath of the continuous carbon fiber at 0◦ , 45◦ , and 90◦ , respectively.
Fig. 3 shows the infill density toolpath, with ±45◦ raster angle of the
infill layer. The mechanical tests were conducted on a Laryee’s Universal
testing machine (UTM). The input parameters entered include the length
of the specimen, and cross-sectional area, for which a rectangular solid
cell cross-section was taken at the narrow of the dog-bone specimen,
albeit the inconsistency due to printing methods and infill percentage.

2.2. Design of experiments

This study assesses the effect of four parameters consisting of: infill
density, fiber angle, continuous carbon fiber position, and strain rate. To
test every possible combination 162 runs of the experiment would be
Fig. 5. Graphed Results of experiments no 1, 2 and 3 as per Table 1.
required, along with 162 sample types and three specimens for each

Fig. 6. Mean effect plots (a) Ultimate Tensile strength (MPa), (b) Strain at break (%), (c) Toughness (MPa), (d) Young’s Modulus (GPa), (e) Yield Strength (MPa), (d)
Resilience (MPa).

5
B. El Essawi et al. Results in Engineering 21 (2024) 101926

Table 3
Normalized values.
Sample Number Normalized Values

Ultimate Tensile Strength (MPa) Strain at break (%) Toughness (MPa) Young’s Modulus (GPa) Yield strength (MPa) Resilience (MPa)

1 1.000 0.236 0.538 0.604 0.736 0.350

2 0.232 0.561 0.135 0.233 0.264 0.141
3 0.083 0.381 0.035 0.113 0.135 0.083
4 0.857 0.177 0.351 0.749 0.846 0.414
5 0.312 0.793 0.770 0.204 0.282 0.160
6 0.000 0.460 0.000 0.109 0.015 0.000
7 0.941 0.144 0.353 0.575 0.772 0.402
8 0.203 0.268 0.007 0.189 0.264 0.151
9 0.147 0.780 0.482 0.000 0.000 0.024
10 0.879 0.387 0.631 0.182 1.000 1.000
11 0.180 0.392 0.191 0.425 0.267 0.074
12 0.233 0.817 0.708 0.196 0.252 0.141
13 0.990 0.255 0.497 0.433 0.757 0.479
14 0.199 0.313 0.062 0.407 0.322 0.116
15 0.219 1.000 0.850 0.280 0.246 0.104
16 0.873 0.000 0.056 1.000 0.829 0.278
17 0.266 0.952 1.000 0.218 0.276 0.150
18 0.394 0.689 0.687 0.385 0.385 0.169

sample to conform with repeatability. Taguchi experiment design was fairness and to prevent slipping between the tapes adhesive surface and
used to perform the tests in cost effective and time efficient manner. the specimen. The goal of the study is to first acquire a reasonable un­
Taguchi’s method for the design of experiments is a statistical modelling derstanding of the effect of each preprinting parameter on the me­
tool, which adjusts certain control factors to reduce the number of inputs chanical properties of continuous-carbon-fiber-reinforced Onyx, and to
required in order to form unbiased estimation of results; this is done furthermore acquire the most optimal set of parameters and to verify the
with the help of signal-to-noise ratios and an optimal combination of all results. Fig. 5 shows the stress strain data of three experiments attained
input parameters [7]. For this experiment, as there are four parameters, from the experimentation. Using the data acquired from the tensile tests,
a four factor L18 Taguchi pattern was generated using the Minitab a Grey relational algorithm was used in order to unify the multi-response
software. Table 2 displays the experimental Taguchi design used in this outputs from the Taguchi model into a single normalized value, which
study. can be directly compared, this normalized value is known as the Grey
The specimens prepared according to the Taguchi design are shown relational co-efficient. For this process, none of the results were
in Fig. 4. All specimens were marked according to the order in the weighted giving each parameter equal relative prominence. The data
Taguchi design where Sample #1 corresponds to 30% infill density, 0- was optimized to maximize the values for ultimate tensile strength,
degree raster angle of the continuous carbon fiber, carbon fiber placed strain at fracture, toughness, resilience, yield strength, and Young’s
on layers 12 and 13, and the experiment conducted at a strain rate of 1 modulus. As the values were normalized and singled the optimal value
mm/min. might not contain the maximum value for all the parameters but is ex­
pected to have the highest overall ratio, and thus the best overall value.
The data acquired from the UTM values for ultimate tensile strength,
2.3. Tensile testing of samples yield strength, strain at fracture, toughness, and Young’s modulus were
deduced. Using the same data, the values for resilience were also
All specimens underwent destructive uniaxial tensile tests, the ex­ calculated using equation (1):
periments were conducted on a Universal Testing Machine, under con­
stant conditions of room temperature and 40% humidity. Before the
tests were conducted the taped tags were removed for the sake of

Table 4
Table of GRA results.
Sample Number Grey Relational Coefficient GRG Rank

UTS (MPa) Strain at break Toughness Young’s Modulus Yield strength Resilience

1 1.000 0.396 0.520 0.558 0.654 0.435 0.5936 3

2 0.394 0.533 0.366 0.395 0.405 0.368 0.4101 13
3 0.353 0.447 0.341 0.360 0.366 0.353 0.3701 17
4 0.778 0.378 0.435 0.666 0.765 0.460 0.5803 6
5 0.421 0.707 0.685 0.386 0.411 0.373 0.4971 9
6 0.333 0.481 0.333 0.359 0.337 0.333 0.3628 18
7 0.895 0.369 0.436 0.540 0.687 0.455 0.5637 7
8 0.386 0.406 0.335 0.381 0.405 0.371 0.3805 16
9 0.369 0.695 0.491 0.333 0.333 0.339 0.4268 12
10 0.805 0.449 0.575 0.379 1.000 1.000 0.7015 1
11 0.379 0.451 0.382 0.465 0.406 0.351 0.4056 14
12 0.395 0.732 0.631 0.384 0.401 0.368 0.4849 11
13 0.981 0.401 0.498 0.468 0.673 0.490 0.5854 4
14 0.384 0.421 0.348 0.458 0.424 0.361 0.3994 15
15 0.390 1.000 0.769 0.410 0.399 0.358 0.5543 8
16 0.798 0.333 0.346 1.000 0.745 0.409 0.6052 2
17 0.405 0.912 1.000 0.390 0.409 0.370 0.5811 5
18 0.452 0.616 0.615 0.449 0.448 0.376 0.4927 10

6
B. El Essawi et al. Results in Engineering 21 (2024) 101926

Fig. 7. Scanning electron micrograph of fractured region (a) Sample 10 (b) Sample 3.

Yield Strength2
Resillience = (1)
Table 5 2 ∗ Young′s Modulus
Response table for means. The results acquired are displayed in Table 2.
Level Infill Density Fiber Angle CF Layer Position Strain Rate Fig. 6 shows the mean effect plots for the output responses such as
1 0.458 0.605 0.524 0.470 ultimate tensile strength, strain at break, toughness, Young’s modulus,
2 0.480 0.446 0.506 0.503 yield strength and resilience. It can be seen from Fig. 6a that fiber angle
3 0.457 0.449 0.470 0.527 is the most influencing factor towards the ultimate tensile strength, yield
4 0.531 strength, strain at break, Young’s modulus and resilience. It is found
5 0.513
6 0.560
logical that the orientation of continuous carbon fiber controls the
Delta 0.103 0.159 0.054 0.057 strength and stiffness-related properties. The fiber orientation of 0◦ are
Rank 2 1 4 3 termed as longitudinal fibers as they appear parallel to the loading di­
rection. This alignment of fiber favors fiber to resist the force applied
along their length and as a result provide higher performance. Similarly,
the fiber angle of 90◦ is termed as traverse fibers as they appear

7
B. El Essawi et al. Results in Engineering 21 (2024) 101926

Fig. 8. Main Effect graph for GRG. Fig. 9. Stress-strain curve for optimized specimen (3- replications).

perpendicular to the loading direction. The orientation does not provide Step 1. : Normalize the data based on whether the desired values are
resistance to the loading and results in lower performance. However, the considered ‘better’ or more optimized if they are relatively higher or
orientation provides better compressive performance. The orientation of relatively lower, the terminology used here is higher-the-better or lower-
carbon fiber can be utilized to induce the desired strength and stiffness the-better using equations (2) and (3). Table 3 shows the results of the
in the 3D-printed part. Whereas the 45◦ fiber angle provides optimal normalized values, then proceed to acquire the deviation sequence as
combined tensile and compressive performance. Due to this importance per equation (4), by subtracting the normalized values from 1.
role of fiber orientation in tensile load bearing capability it has been

Absolute value − Minumum recorded value

Higher the better : Normalized value = (2)
Maximum recorded value − minimum recorded value

Maximum recorded value − Absoluted Value

Lower the better : Normalized Value = (3)
Maximum recorded value − Minimum recorded Value

ranked number one for most of the mechanical properties. The primary
role of carbon fiber layer position is to provide the reinforcement in the Deviation sequence : Δ = 1 − Normalized Value (4)
3D printed part. The customization of fiber layer position has a potential
to strategically tailor the energy absorption related properties such as
toughness and resilience. It can be seen as shown in Fig. 6c and f, where Step 2. Determine the Grey relational coefficient by using equation
carbon fiber layer position ranked first. Toughness variation with carbon (5). Where ξ is a constant number, the commonly used number is 0.5 this
fiber layer position can be linked with the associated load-bearing ca­ is to limit the effects of the delta maximum value not causing the data to
pacity of a 3D-printed part. The carbon fiber layer position can control become erroneous.
the resistance to crack formation and propagation.
In case of strength related properties, infill density was ranked sec­ Minimum Delta Value + (ξ ∗ Maximum Delta Value)
ond in most of the cases. Infill density is about how much material is GRC = (5)
Absolute delta Value + (ξ ∗ Maximum Delta Value)
filled inside the printed part. More material in the 3D printed part
generally results in higher yield strength and stiffness-based properties
due to the ability to withstand higher applied forces. The less dominant Step 3. : Determine the grey relational grade. The specialty of Grey
factor was found to be the strain rate that ranked fourth (lowest) among relational analysis is shown in this step, and it is also why it is so
other input parameters. compatible with the Taguchi method. The goal of the analysis method is
In order to homogenize as well as acquire the optimal preprinting to convert multi response systems into a single factor; the grey relational
parameters, Grey relational analysis (GRA) was performed. The process analysis does this with the help of a equations (6) and (7) as mentioned
of Grey relational analysis is as follows.

Table 6
Analysis of variance for GRG.
Source DF Seq SS Contribution Adj SS Adj MS F-Value P-Value

Infill Density 5 0.058 16.25% 0.058 0.012 1.360 0.356

Fiber Angle 2 0.195 54.13% 0.195 0.097 11.300 0.009
CF Layer Position 2 0.036 10.12% 0.036 0.018 2.110 0.202
Strain Rate 2 0.018 5.12% 0.018 0.009 1.070 0.401
Error 6 0.052 14.37% 0.052 0.009
Total 17 0.359 100.00%
R – Sq = 85.63%

8
B. El Essawi et al. Results in Engineering 21 (2024) 101926

Table 7
Confirmation test results.
Levels UTS Strain at break Toughness Young’s Modulus Yield strength Resilience Grey Relational Grade

Initial Condition A1B1C1D1 59.8136 6.9066 205.2641 1.13333 35.1966 0.54774 0.5679
Optimal Condition
Prediction A6B1C1D3 0.7160
Experimentation A6B1C1D3 60.1455 5.421 138.6829 1.49 51.825 0.9043 0.7221
Improvement in GRG ¼ 0.1542

below, and a weight factor, for the sake of this study the weight factor the same conditions. The results are as shown below in Fig. 9. Both the
amongst all parameters will be the same, 1. To acquire a weighted grey yield strength and Young’s modulus were shown to be amongst the
relational grade, further research and methodologies are required. highest amongst the tested parameters, whilst the strain before the
breaking point was the lowest, the ultimate tensile strength approached
1
Number∑
of factors 60 in most of the samples achieving close to the highest overall UTS.
GRG =
(Number of factors)
∗ Factor Weight ∗ GRC of factor A6B1C1D3 in Table 7 represents the average value of the optimized
samples; the data verifies the findings of the Grey relational analysis as
1

(6)
the optimized samples have a grey relational grade of 0.722 compared to
For this study it is used as shown below the initial condition grade of 0.5676, improvement in GRG was found to
be 21.4% better.
1 ∑6
GRG = ∗ 1 ∗ GRC (7)
6 1 3. Conclusion

Following conclusions can be concluded from the study.

Step 4. : Rank the Grey relational grades from 1 to 18, highest ranked
first. The grey relational grade represents how close a sample is to the • It has been found in the study that continuous fiber angle was the
optima values amongst all samples which in this case means how close a most dominant input parameter with the contribution of 54.13%,
sample is to the maximum mechanical properties amongst all recorded whereas infill density was the second dominant parameter with the
values, the results acquired are displayed in Table 4. It can be seen in contribution of 16.25%. The study also found that the position of the
Fig. 7 that highest rank showing closest optimum conditions were ach­ continuous carbon fiber layer has comparatively less effect on the
ieved for experiment 10 conditions (infill density = 60%, fiber angle = final tensile mechanical properties of 3D-printed parts, with the
0◦ , CF layer position = 24-25 and strain rate of 10 mm/ min). The contribution of 10.12%.
fractured samples of three dimensionally printed samples of carbon fiber • From the results, the desired values are 80% infill density, 0-degree
reinforced nylon composites were investigated using the scanning fiber orientation, carbon fibers in the 12th and 13th layers, and a
electron microscopy (SEM). To show the behavior of fracture two sam­ strain rate of 10 mm/min (A6B1C1D3). When the optimal condition
ples (2nd lowest rank and highest rank) have been reported in the study. was compared against the initial condition, improvement of 0.1542
The fracture in the pattern is linked with the matrix cracking mecha­ was found in GRG.
nisms that shows matrix fractured under the action of stress. Fiber • In this study, mean effect plots were generated for various output
fracture can also be seen in the two micrographs. Fracture of fiber is responses, including Ultimate Tensile Strength, Strain at Break,
linked with the poor fiber orientation angle that results in different stress Toughness, Young’s Modulus, Yield Strength, and Resilience. The
concentration in the object. The fiber angle of 0◦ provided the best results, shown in Fig. 6a, indicate that fiber angle is the most sig­
outcomes in sample 10 as it provided resistance to fracture and nificant factor impacting Ultimate Tensile Strength, Yield Strength,
enhanced the load carrying capability. It can be seen that the strain rate Strain at Break, Young’s Modulus, and Resilience. This finding is
and position of carbon fiber layer were the least affecting parameters logically consistent with the notion that the orientation of contin­
and are found common in both samples. On the other hand, higher infill uous carbon fibers has a controlling influence over the mechanical
density also contributed towards higher GRG. properties related to strength and stiffness. By carefully manipu­
lating the fiber orientation during the 3D printing process, it is
Step 5. : After acquiring the grey relational grade, to find the opti­
possible to produce parts with the desired levels of strength and
mized values, we use Minitab to get the response table for means and
stiffness.
graph values of the mean effects, as shown in Table 5 and Fig. 8. From
• In contrast, the position of the continuous carbon fiber layer was
the results, it can be determined that the desired values are 80% infill
found to have the most significant impact on toughness. The varia­
density, 0-degree fiber orientation, carbon fibers in the 12th and 13th
tion in toughness with the carbon fiber layer position can be attrib­
layers, and a strain rate of 10 mm/min (A6B1C1D3). The newly opti­
uted to the load-bearing capacity of the 3D-printed part. By
mized values need to be verified by reperforming the experiment with
controlling the position of the carbon fiber layer, it is possible to
the given parameters. Table 6 shows the analysis of variance (ANOVA)
regulate the resistance to crack formation and propagation, thus
performed on the GRG. It has been found in the study that fiber angle
enhancing the toughness of the part.
was the most dominant input parameter with the contribution of
• As observed during this study, position of carbon fiber layer can play
54.13%, whereas infill density was the second dominant parameter with
a critical role towards enhancing the toughness of 3D printed parts. It
the contribution of 16.25%. The study also found that the position of
can also be observed that strategic placement of carbon fiber layer
carbon fiber layer has comparatively less effect on the final mechanical
can regulate the crack formation and its propagation. Different in­
properties of 3D-printed parts, with contribution of 10.12%.
dustrial sectors such as aerospace, automotive, healthcare, and more
Step 6. Optimized Testing: As a result of the Grey relational analysis, can exploit this outcome for their benefit.
and the subsequent parameter optimization to acquire the highest
possible grey relational grade was A6B1C1D3. The combination was not CRediT authorship contribution statement
part of the experiments, so a specimen of the given parameters was
printed, and uniaxial tensile test was conducted on the specimen under Basim El Essawi: Data curation, Formal analysis, Investigation,

9
B. El Essawi et al. Results in Engineering 21 (2024) 101926

Validation, Writing – original draft. Said Abdallah: Investigation, [10] W.M.H. Verbeeten, R.J. Arnold-Bik, M. Lorenzo-Bañuelos, Print velocity effects on
strain-rate sensitivity of acrylonitrile-butadiene-styrene using material extrusion
Methodology, Writing – original draft, Formal analysis. Shafahat Ali:
additive manufacturing, Polymers 13 (1) (Jan. 2021) 1–20, [Link]
Formal analysis, Investigation, Methodology, Writing – original draft. 10.3390/polym13010149.
Abdallah Nassir Abdo Mohammed: Methodology, Validation, Data [11] W.M.H. Verbeeten, M. Lorenzo-Bañuelos, R. Saiz-Ortiz, R. González, Strain-rate-
curation. Rahmat Agung Susantyoko: Conceptualization, Investiga­ dependent properties of short carbon fiber-reinforced acrylonitrile-butadiene-
styrene using material extrusion additive manufacturing, Rapid Prototyp. J. 26
tion, Project administration, Resources, Supervision, Writing – review & (10) (Nov. 2020) 1701–1712, [Link]
editing. Salman Pervaiz: Conceptualization, Funding acquisition, [12] Saleh Atatreh, Mozah Saeed Alyammahi, Hayk Vasilyan, Tawaddod Alkindi,
Investigation, Project administration, Resources, Supervision, Writing – Rahmat Agung Susantyoko, Evaluation of the infill design on the tensile properties
of metal parts produced by fused filament fabrication, Results in Engineering 17
review & editing. (2023) 100954, [Link]
[13] H.K. Dave, N.H. Patadiya, A.R. Prajapati, S.R. Rajpurohit, Effect of infill pattern
Declaration of competing interest and infill density at varying part orientation on tensile properties of fused
deposition modeling-printed poly-lactic acid part, Proc. IME C J. Mech. Eng. Sci.
235 (10) (May 2021) 1811–1827, [Link]
The authors declare that they have no known competing financial [14] K.N. Gunasekaran, V. Aravinth, C.B.M. Kumaran, K. Madhankumar, S.P. Kumar,
interests or personal relationships that could have appeared to influence Investigation of mechanical properties of PLA printed materials under varying infill
density, Mater. Today: Proc. 45 (2021) 1849–1856, [Link]
the work reported in this paper. matpr.2020.09.041.
[15] R. Srinivasan, N. Aravindkumar, S. Aravind Krishna, S. Aadhishwaran, J. George,
Data availability Influence of fused deposition modelling process parameters on wear strength of
carbon fibre PLA, Mater. Today: Proc. 27 (Jan. 2020) 1794–1800, [Link]
10.1016/[Link].2020.03.738.
Data will be made available on request. [16] A.M. Abdullah, D. Mohamad, T.N.A.T. Rahim, H.M. Akil, Z.A. Rajion, Effect of
narrow infill density gap on the compressive properties of 3D printed carbon fibre
Acknowledgements reinforced acrylonitrile butadiene styrene, J. Mech. Sci. Technol. 33 (5) (May
2019) 2339–2343, [Link]
[17] W. Wu, P. Geng, G. Li, D. Zhao, H. Zhang, J. Zhao, Influence of layer thickness and
The authors would like to acknowledge the financial support of the raster angle on the mechanical properties of 3D-printed PEEK and a comparative
Dubai Electricity and Water Authority – R & D Center, Dubai, UAE. mechanical study between PEEK and ABS, Materials 8 (9) (2015) 5834–5846,
[Link]
[18] Y. Zhao, Y. Chen, Y. Zhou, Novel mechanical models of tensile strength and elastic
References property of FDM AM PLA materials: experimental and theoretical analyses, Mater.
Des. 181 (Nov. 2019), [Link]
[1] S.R. Piramanayagam, et al., Experimental investigation and statistical analysis of [19] P.K. Mishra, P. Senthil, S. Adarsh, M.S. Anoop, An investigation to study the
additively manufactured onyx-carbon fiber reinforced composites, J. Appl. Polym. combined effect of different infill pattern and infill density on the impact strength
Sci. 138 (May 2021) 18, [Link] of 3D printed polylactic acid parts, Compos. Commun. 24 (Apr. 2021), [Link]
[2] M. Nachtane, M. Tarfaoui, Y. Ledoux, S. Khammassi, E. Leneveu, J. Pelleter, org/10.1016/[Link].2020.100605.
Experimental investigation on the dynamic behavior of 3D printed CF-PEKK [20] J. Suteja, Effect of infill pattern, infill density, and infill angle on the printing time
composite under cyclic uniaxial compression, Compos. Struct. 247 (Sep. 2020), and filament length of 3D printing, Jurnal Rekayasa Mesin 12 (1) (May 2021) 145,
[Link] [Link]
[3] N. Li, Y. Li, S. Liu, Rapid prototyping of continuous carbon fiber reinforced [21] C. Ma, J. Faust, J.D. Roy-Mayhew, Drivers of mechanical performance variance in
polylactic acid composites by 3D printing, J. Mater. Process. Technol. 238 (Dec. 3D-printed fused filament fabrication parts: an Onyx FR case study, Polym.
2016) 218–225, [Link] Compos. 42 (9) (Sep. 2021) 4786–4794, [Link]
[4] S. Kannan, R. Vezhavendhan, S. Kishore, K.V. Kanumuru, Investigating the effect of [22] H. Gonabadi, Y. Chen, A. Yadav, S. Bull, Investigation of the effect of raster angle,
orientation, infill density with Triple raster pattern on the tensile properties for 3D build orientation, and infill density on the elastic response of 3D printed parts
Printed samples, IOP SciNotes 1 (2) (Sep. 2020) 024405, [Link] using finite element microstructural modeling and homogenization techniques, Int.
2633-1357/abb290. J. Adv. Manuf. Technol. (2021), [Link]
[5] P.S. Ramalingam, K. Mayandi, V. Balasubramanian, K. Chandrasekar, V.M. Stalany, [23] K. Aruchamy, S.K. Palaniappan, R. Lakshminarasimhan, B. Mylsamy, S.
A.A. Munaf, Effect of 3D printing process parameters on the impact strength of K. Dharmalingam, N.S. Ross, S. Pavayee Subramani, An experimental study on
onyx – glass fiber reinforced composites, Mater. Today: Proc. 45 (2020) drilling behavior of silane-treated cotton/bamboo woven hybrid fiber reinforced
6154–6159, [Link] epoxy polymer composites, Polymers 15 (14) (2023) 3075, [Link]
[6] M. Žmindák, P. Novák, J. Soukup, D. Milosavljevic, M. Kaco, Finite element 10.3390/polym15143075.
simulation of tensile test of composite materials manufactured by 3D printing, IOP [24] Saravanakumar Arunachalam, Rajeshkumar Lakshmi Narasimhan, Sathish Kumar
Conf. Ser. Mater. Sci. Eng. 776 (Apr. 2020) 1, [Link] Palaniappan, Nimel Sworna Ross, Arulmozhivarman Joseph Chandran, Sanjay
899X/776/1/012082. Mavinkere Rangappa, Suchart Siengchin, Machinability analysis of Typha
[7] Markforged. (n.d.). Composites V5.2 Datasheet. Retrieved from [Link] angustifolia natural fiber reinforced composites through experimental modeling –
[Link]/craft/materials/[Link]. influence of fiber orientation, Polym. Compos. 44 (Issue 7) (2023) 3808–3823,
[8] G. Ćwikła, C. Grabowik, K. Kalinowski, I. Paprocka, P. Ociepka, The influence of [Link]
printing parameters on selected mechanical properties of FDM/FFF 3D-printed [25] Hasan Ahmed, Md Ashikur Rahaman Noyon, MdElias Uddin, Mamun Jamal,
parts, IOP Conf. Ser. Mater. Sci. Eng. 227 (Aug. 2017) 1, [Link] Sathish Kumar Palaniappan, Biodegradable and flexible fiber-reinforced composite
1757-899X/227/1/012033. sheet from tannery solid wastes: an approach of waste minimization, Polym.
[9] H. Zakaria, S.F. Khan, M.F.C. Fee, M. Ibrahim, Printing temperature, printing speed Compos. 44 (Issue 11) (2023) 7545–7556, [Link]
and raster angle variation effect in fused filament fabrication, IOP Conf. Ser. Mater.
Sci. Eng. 670 (Dec. 2019) 1, [Link]

10