Three-Dimensional Printing Limitations of Polymers Reinforced with Continuous Stainless Steel Fibres and Curvature Stiffness

Abstract

This study investigates the printability limitations of 3D-printed continuous 316L stainless steel fibre-reinforced polymer composites obtained using the Materials Extrusion (MEX) technique. The objective was to better understand the geometric printing limitations of composites fabricated using continuous steel fibres, based on a combination of bending stiffness testing and piezoresistive property studies. The 0.5 mm composite filaments used in this study were obtained by co-extruding polylactic acid (PLA), with a 316 L stainless steel fibre (SSF) bundle. The composite printability limitations were evaluated by the printing of a series of ’teardrop’ shaped geometries with angles in the range from 5° to 90° and radii between 2 and 20 mm. The morphology and dimensional measurements of the resulting PLA-SSF prints were evaluated using $\mu$CT scanning, optical microscopy, and calliper measurements. Sample sets were compared and statistically examined to evaluate the repeatability, turning ability, and geometrical print limitations, along with dimensional fluctuations between designed and as-printed structures. Comparisons of the curvature bending stiffness were made with the PLA-only polymer and with 3D-printed nylon-reinforced short and long carbon fibre composites. It was demonstrated that the stainless steel composites exhibited an increase in bending stiffness at smaller radii. The change in piezoresistance response of the PLA-SSF with load applied during the curvature bending stiffness testing demonstrated that the 3D-printed composites may have the potential for use as structural health monitoring sensors.

1. Introduction

Stainless steel fibres have been combined with polymer textiles for use in applications ranging from heated automotive seats to personal protective clothing for extremely low-temperature environments [1,2]. Another potential use of steel fibre-reinforced composites is in signal and information transfer, which takes account of their piezoresistivity response to an applied loading, for load sensing, smart impact textiles, and structural health monitoring [3,4]. Three-dimensional printed composite functionality can be adapted depending on their end-use application, i.e., electrical, structural, or impact applications [5]. This study evaluates the geometric printability limitations of 3D-printed continuous stainless steel fibre-reinforced polymer composites, focusing on cornering capabilities and the relationship between the polymer and SSF deposition through smaller angles and radii. An evaluation of individual semi-circle sections were carried out by developing a curvature bending stiffness test. The latter was evaluated in conjunction with the loading resistance response.

1.1. Printing of Continuous Fibre-Reinforced Composites

The most frequently used 3D printing technique for polymers and composites is material extrusion (MEX) [6,7]. Polymer printing involves feeding a thermoplastic polymer filament through a heated extruder nozzle (usually in the range of 150–260 °C) and depositing the material in layers onto the building platform [5]. For many engineering applications, it is necessary to reinforce the 3D-printed polymer with fibres to enhance their mechanical strength [8,9]. The fibre reinforcing of composites can be through the use of either continuous or short fibres [10,11]. Due to the fibre orientation, continuous fibre composites offer higher strength and stiffness than those of discontinuous fibre composites [8]. Amongst the continuous reinforcing fibres reported for use in 3D-printed polymer composites are glass, aramid, flax, jute, Nitinol, carbon, and basalt [9,12,13,14,15]. The thermoplastic polymer matrices most associated include polylactic acid (PLA), Polyamide (PA or nylon), Acrylonitrile butadiene styrene (ABS), polyurethane (PU), polyetheretherketone (PEEK), and Polycarbonate (PC) [4,9,16,17,18].

1.2. Geometric Print Designs

The wider adoption of robotics and multi-axis 3D printers for the fabrication of continuous fibre-reinforced components has led to the generation of more complex tooling paths and internal structures [8,19,20,21,22]. These more complex structural designs encompass a wide range of angles and curves, where more knowledge about how the continuous fibre affects the geometric print constraints (including cornering) is needed [23,24,25,26,27]. The industry demand for lightweight, mechanically strong structures drives research into 3D-printed continuous reinforced composite cores for sandwich structures. These core sandwich structures include truss (free-printing), unit cell combinations, and variations of repeating patterns, which use an array of angles and curves, typically joining top and bottom plates [28]. Knowledge of the material’s stiffness and critical force at different radii inform the structural design [6,20,29].

Geometrical limitations when printing with continuous fibres should be considered during the design phase; however, few studies report the deviation between designed dimensions and as-printed results. Precise fibre deposition in corners is difficult when 3D printing with continuous fibres at small radii circles or arcs. A further consideration is the materials relationship between the matrix and continuous fibre. Acute turning results in fibre migration (stringing), i.e., the movement of fibres out of the design path, which frequently occurs when printing smaller angles or radii, changing the resultant as-printed dimensions. Cornering was evaluated at radii from 2 to 20 mm for 3D printing ABS-continuous carbon fibre (cCF) by Matsuzaki et al. [30]. It was concluded that the cCF bundle size and radius of the turn affects the material deposition at small radii of <4 mm. It was also observed that a larger bundle size restricts deposition.

Three-dimensional printing smaller turns with continuous fibres also affects the exterior perimeter dimensional accuracy. The fabrication of sandwich core structures such as honeycomb, rhombus, rectangles, and circles cores, which combine unit cells consisting of angles or radii, results in the continuous printing paths making multiple turns [31]. Some core designs intend for the material of one cell to join another cell or plate [32,33,34]. Depending on the fibre reinforcement, the minimum turn angle or radius has limitations due to fibre fracture or dimensional restrictions between the design and printed structures [25].

An adopted approach to remove small radii cornering effects is fibre cutting and the use of a dual-head material printing system, such as that developed by Markforged [35]. In MEX, adjusting the print speed during 3D printing helps reduce fibre migration and also improves part strength and surface quality [36,37,38,39]. Yamamoto et al. [25] proposed an algorithm for generating a single-stoke tooling path of continuous carbon fibre MEX 3D printing, without the need for filament cutting between layers. The resultant geometry deviations, however, were found to leave structural gaps in positions where the material should join together, leading to unpredictable mechanical properties and structures.

Individual mechanical testing of continuous fibre components is primarily focused on aligned fibres, employing compression, tensile, or flexural techniques [14]. Several authors reported on the mechanical properties of 3D-printed fibre-reinforced sandwich structures exhibiting different grids or unit cell dimensions [10,28]. Most have tested structures incorporating an array of cells with angles or radii in the core structures printed in the horizontal or vertical plane [20,23,39]. Longer cell lengths, with larger radii turning paths, are associated with increased tensile or compressive properties.

A large radii curve bend test was developed by Zhang et al. [40] for 3D-printed leaf springs of PLA reinforced with continuous flax fibre, fabricated using a five-axis 3D printer. The part had the following dimensions: a 60 mm inner radius, a 15 mm width, a 2 mm height, and an 80 mm test span. The incorporation of the flax fibre was found to increase the stiffness by 115% compared with unreinforced PLA. While these studies evaluated the mechanical properties of 3D-printed sandwich composites, few have reported investigating the mechanical properties of individual arcs or cylinder sections at smaller radii.

1.3. Electrical Resistance

The incorporation of continuous conductive fibres into a composite has introduced the potential for their use as embedded sensors for structural health monitoring (SHM) [41,42]. Amongst the conductive filaments that can be used with these composites are copper, nickel–chromium, and carbon fibre [26,27,43].

Mechanical performance testing combined with piezoresistive responses can evaluate the correlation between resistance and load, strain, or tension. An example of a 3D-printed PLA-cCF SHM sensor was integrated into a composite finger studied by Yao et al. [24]. The authors reported that, during cyclical flexural testing, there was an increase in the fractional resistance as the strain fluctuations increased. A study on the tensile and piezoresistive properties of MEX 3D-printed copper wire-reinforced PLA and polyurethane (PU) for in-situ SHM sensing by Saleh et al. [4] concluded that correlations of the electrical resistance and mechanical strain of the embedded wire has no significant impact on the matrix material.

Partial fibre failure increased resistance peaks during cyclical dynamic and static testing of 3D-printed cCF cantilever beams due to fibre failure [27]. The fibre alignment of 3D-printed composites is directionally unrestricted, allowing the integration of individual or multi-layer sensors with the structure in various orientations. Galos et al. [44] investigated the electrical conductivity of cCF-nylon fabricated by MEX 3D printing, which oriented the filaments and laminate structures in various directions. A factor influencing a reduction in conductive performance was damage to the cCF at the exit of the print nozzle, which resulted in a 40% loss in longitudinal resistance.

The study reported in this paper evaluated the printability limitations of 3D-printed continuous 316 L stainless steel fibre-reinforced polymer composites based on a comparison between the design geometry and the properties of the printed parts. By identifying geometric 3D printing characteristics between the design and as-printed PLA-SSF parts, issues such as fibre migration and gaps can be addressed to define the appropriate dimensions.Many structures are mechanically tested in their complete form, but few studies evaluate individual struts or curvatures of continuous fibre-reinforced composites. This work also proposes a methodology to evaluate the mechanical properties of curvatures by loading sectioned cylinders to failure during curvature bending stiffness testing (CBS). Given the uses of conductive stainless steel fibres, piezoresistivity was investigated during CBS.

2. Theory and Calculation

The continuous fibres used in this study consisted of a 316 L stainless steel bundle, containing 90 fibres. These fibres were obtained from NV Bekaert SA (Zwevegem, Belgium), who fabricated them by ‘bundle wire drawing’ [45,46]. Each fibre had a diameter of 14 μm, a torsion per cm of 1, a linear density of 110 decitex (TEX), an overall bundle diameter of approx. 0.15 mm, and a breaking load of 20 N. Pellets of PLA were sourced from Nature Works, commercially called Ingeo^TM Biopolymer D4043D [47]. Other polymers and composites investigated for CBS testing included Markforged Onyx (Nylon—short carbon fibre) and Onyx reinforced with continuous carbon fibre (Onyx-cCF), obtained from Markforged [35]. These were printed using the conditions described in Section 2.2.

A 3devo laboratory-scale filament maker was used to fabricate both PLA and PLA-SSF filaments [48]. This was facilitated through the use of a custom-designed extrusion nozzle that incorporates the SSFs into the polymer at extrusion, as described previously [16]. The processing parameters for the PLA-SSF filaments, with a diameter of 0.5 mm (±0.6), are outlined in

Table 1. 3devo filament maker processing parameters, $T_1$ to $T_4$ related to extrusion barrel heater temperatures.

Filament	Filament Diameter (mm)	Cooling Fans (°C)	Extrude Speed (rpm)	${\mathit{T}}_1$ (°C)	${\mathit{T}}_2$ (°C)	${\mathit{T}}_3$ (°C)	${\mathit{T}}_4$ (°C)
PLA-SSF	0.5 (±0.8)	33.0	2.2	179	186	193	191

.

2.1. PLA-SSF Filament Fabrication and 3D Printing

‘Teardrop’ shapes were printed to evaluate the PLA-SSF materials deposition while cornering during MEX 3D printing. These typologies combine an angle, straight sections, and semi-circles, as shown in Figure 1. The angles investigated were 5°, 10°, 15°, 30°, 50°, 70°, and 90°, which were joined to straight sections that connected to semi-circles with inner radii of 2, 3, 4, 5, 10, 15, and 20 mm.

A mechanical test to determine the curvature bending stiffness (CBS) of the PLA-SSF composite used the same dimensions as the radii of the semi-circles teardrop sections. Oval cylinders were printed for sectioning with inner radii from 3 to 20 mm. The resistance response of the PLA-SSF composite was measured during CBS testing as the load deformed short sectioned cylinders.

In addition to the stainless steel composites, other materials were printed for comparison in the CBS study. These were neat PLA and nylon–short carbon fibre (Markforged, Onex), with the same geometric dimensions as the PLA-SSF semi-circles and Nylon with short and continuous carbon fibre (Markforged, Onex-cCF), with radii of 5, 10, 15, and 20 mm.

2.2. Three-Dimensional Printing Teardrop and Curvature Stiffness Samples

The polymer-SSF filaments were used to 3D print both teardrop and oval cylinder geometries designed using a computer-aided design (CAD) file (Solidworks 2021 software, version SP03, Dassault Systèmes SOLIDWORKS Corporation, Waltham, MA, USA) and then exported as a stereolithography (STL) file. The PLA and PLA-SSF material 3D printing g-code was generated using PrusaSlicer, an open-source slicing software. Anycubic i3 Mega polymer extrusion 3D printers facilitated the deposition of the PLA and, with modifications, the continuous PLA-SSF filament. These modifications included a printing nozzle designed to remove excess polymer during printing, an increased nozzle area on the lower face surface to increase the dwell time, and a soft wheel filament feeder [16].

The following 3D printing parameters were maintained at constant values during printing: the print head temperature of 220 °C, the print bed temperature of 65 °C, a layer height (h) of 0.22 mm, and a hatch spacing (w) of 0.4 mm. The printing speed was adjusted to address the requirements of varying angles and radii, as detailed in

Table 2. Printing parameters of the PLA-SSF.

Angle Radii (R), (mm)	90° 20.0	70° 15.0	50° 10.0	30° 5.0	20° 4.0	10° 3.0	5° 2.0
Speed, (mm/s)	8	8	4	2	1	1	1

. Note the reduction in printing speed as the radii and angles decreased. A slower speed increases the nozzle’s dwell time at a given location, allowing partial polymer solidification, thus reducing stringing. Polymer solidification helps retain the continuous SSF bundle in the structure, as the print head traverses the X-Y plane. The curvature test samples were oval cylinder designs to facilitate a continuous fibre toolpath and the same print speeds for the respective teardrop radii were adopted. All teardrop printing toolpaths were in an anticlockwise direction, starting at the corner of the angle.

PLA filaments with a diameter of 1.75 mm were also fabricated using the 3devo system; these were then used in an Anycubic printer to print parts. As outlined earlier, this investigation included comparison studies with Onyx and Onyx-cCF, printed using the Markforged Mark Two 3D printer [35]. The Onyx-cCF print study was limited to radii above 5 mm and a wall width above 2.8 mm, due to the printing limitations imposed by the Markforged slicing software version 3.170.0, Eiger version 3.19.46 (Markforged, Billerica, MA, USA). Hence, all Onyx and Onyx-cCF oval cylinders were designed with a wall width of 3 mm, replicating the PLA-SSF as-printed cylinders’ average wall width of 3.2 mm ($\pm 0.3$). The Onyx-cCF pattern is concentric, with a 0.125 mm layer height, a hatch packing of 0.25 mm, and a printing head temperature of 285 °C. Both the PLA and Onyx were printed in a concentric pattern. Based on the author’s previous study using the sample printing parameters, the typical volume fraction ($V_f$) was 30% $V_f$ and a porosity of 2% for the PLA-SSF composites was used [16].

2.3. Characterisation and Geometry Analysis

The as-printed PLA-SSF teardrop composite dimensions, printed structure, and morphology were recorded and compared with the design dimensions. The seven PLA-SSF teardrop geometries printed contain a minimum of three structures. The as-printed teardrop samples were examined using a GE Nanotom X-ray micro-computed tomography ($\mu$CT) scanner (GE Sensing & Inspection Technologies GmbH, Wunstorf, Germany), with an 8 μm resolution. The $\mu$CT scans of the teardrops were evaluated using the VG Studios software, version 3.5, (Volume Graphics GmbH, Heidelberg, Germany) The specimens’ dimensions were also recorded using digital vernier callipers (Digi Plus Line), an Olympus DSX1000 microscope (LECO Corporation, Tokyo, Japan), and a Dinolite AM7915MZT USB 2.0 microscope, combined with a lighting box.

Geometry differences between the as-printed teardrop parts and the CAD design dimensions were investigated. Figure 2 illustrates an example of the measurement locations around the teardrop geometry, with an angle of 70° and an inner radius of 15 mm. A particular focus was the structure in the semi-circle and the angle areas. All designs had a wall height of 4 mm and wall width of 1.2 mm (Figure 2a).

Table 3. CAD dimensions and key to sections of geometry measured.

Measurement Code	Description of Position on Geometry	90° R 20 mm	70° R 15 mm	50° R 10 mm	30° R 5 mm	20° R 4 mm	10° R 3 mm	5° R 2 mm
R_inner	Inner radius	20.00	15.00	10.00	5.00	4.00	3.00	2.00
R_outer	Outer radius	21.20	16.20	11.20	6.20	5.20	4.20	3.20
C_lenght	Corner material length	1.70	2.09	2.84	4.64	6.91	13.60	27.51
L_inner	Overall internal length	48.28	44.44	33.66	24.32	27.04	22.62	47.85

and Figure 2c detail the design dimensions and locations of each teardrop. In the angle area, the relationship between the polymer and SSF deposited was explored, relating to the length of the expected material at the corner of the angle (C_length), the overall internal length (L_inner), and outer radii (R_outer). All measurements were recorded at the base (B) and top (T) surfaces relative to the print bed. The statistical analysis of the measured dimensional data was executed using the SPSS software version 28 [49].

2.4. Testing Methods: Curvature Bending Testing

When a force is applied to a structure, the curved bending stiffness (CBS), or k (N/mm), is the ability of the material to resist elastic deformation. A test methodology was devised to determine the curvature bend stiffness (CBS) and the electrical resistance response of the stainless steel composites as force was applied. The CBS testing approach involved the further development of the methodology described by Zhang et al. [40], whereby the testing apparatus was redesigned to facilitate testing semi-circular composites with inner radii from 3 to 20 mm, as opposed to the 60 mm radius used by Zhang et al. The span for the CBS testing was equal to the inner radii due to the small radii, rather than shorter than the sample length.

A minimum of 5 semi-circular specimens were tested for each radius evaluated in all materials (PLA, PLA-SSF, Onyx, and Onyx-cCF). The semi-circular test samples were cut, using a fine-hacksaw blade, from the 3D-printed cylinders. The cut surface was ground down, using a water-cooled grinding wheel with silicon carbide grinding paper from 120 grit down to 1200 to achieve optimal contact with the test bed.

A testing jig was designed and fabricated to hold the semi-circle samples vertically as a force was applied, as shown schematically in Figure 3b. This jig consisted of two aluminium blocks with milled channels; three-dimensional printed PETG insulating sleeves were inserted in the channels separating two copper blocks to support the sample on each side of the curve. Two testing jigs were designed: one for larger radii (>5 mm) and the other for smaller radii samples (<5 mm). The large jig had two supporting walls and the smaller jig’s the inner walls were removed to prevent sample surface interference.

The CBS jig was then mounted onto a Zwick Rolle testing machine with a 15 mm radius PLA-SSF sample, as shown in the photograph of the test equipment in Figure 3c. Each sample was loaded into the appropriate design jigs and a compressive load was applied at testing at a speed of 1 mm/s. The end testing protocol was determined when the force decreased by 30% or when the maximum set travel was reached (the cross-sectional height of the sample). The curved bending stiffness k (N/mm) was calculated using Equation (1) [40]. CBS testing was carried out on the stainless steel composites, including neat PLA, Onex, and Onyx with continuous carbon fibre composites.

$$k={\displaystyle \frac{F}{\delta }},$$

(1)

where F denotes the force load applied to the sample (N) and $\delta$ is the deformation (mm) due to the force.

2.5. Resistance Response

Using the same CBS testing jig, the resistance response of the PLA-SSF was examined. The test schematic is illustrated in Figure 3b. A Keysight E4980AL Precision Inductance, Capacitance, and Resistance (KS-LCR) meter was used in conjunction with the Keysite BenchVue Software, version 2.5.0.18, (Keysite, Tipperary, Ireland) to record changes in resistance ($\Omega$) response during CBS testing for sample sets with radii of 4, 10, and 15 mm. The KS-LCR meter default measurement was a frequency of 1 kHz. The KS-LCR meter was calibrated using resistors of known values from 0.1 to 62 $\Omega$, with a tolerance between 1 and 5%. The measurements were compared with readings of the same resistors and a verification measurement was obtained using a Wayne Kerr Automatic LCR meter 4225. The calibration curve is available in the Supplementary Materials.

To gain an insight into the piezoresistive response of the PLA-SSF filament, a single 100 mm long PLA-SSF filament was attached to a plastic glove, as shown in Figure 3a. While wearing the glove, the KS-LCR meter was attached to either end of the SSF, and the resistance change was monitored between placing the glove finger flat and bent. A resistance of approximately 6.65 $\Omega$ and 7.25 $\Omega$ was recorded moving between straight or bent configurations, respectively.The results in Figure 3a demonstrate the reduction in resistance while the fibres were straight and its increase on bending the fibres.

The resistance response determination was based on a modified version of ASTM D4496-21 [50]. This involved connecting the steel fibre in the PLA-SSF semi-circle composite and KS-LCR meter. The semi-circle’s cross-sectional contact faces with the testing jig were ground with abrasive silica paper to 2000 grit, and then polished with a 3 μm diamond suspension, exposing the SSFs.

To accommodate the movement of the samples during testing, the KS-LCR gold crocodile clips were connected to a pair of copper strips and then seated into each copper supporting block (Figure 3c). An electrically conductive silver paste was added to each side of the semi-circle between the polished face and the copper strips. No significant difference in the fibre resistance values was obtained for this copper strips/silver paste connection arrangement when a comparison experiment was conducted using a 30 mm fibre bundle length, with crocodile clips, which were directly in contact with the steel. Given that the SSFs were coated using a polymer, they could be treated as insulated wires. The resulting structure could thus be considered as a compilation of individual fibres in a parallel circuit, for which the total resistance ($R_t$) could be calculated using Equation (2).

$${\displaystyle \frac{1}{R_t}}={\displaystyle \frac{1}{R_1}}+{\displaystyle \frac{1}{R_2}}+\cdots +{\displaystyle \frac{1}{R_n}}$$

(2)

where $R_t$ is the total resistance of the circuit, and $R_1$ to $R_n$ are the individual resistances of each SSF. The SSF bundles were electrically conductive with a reported resistance per meter of 65 $\Omega$/m, with each curvature containing 72 fibre bundles.

The resistance and time were recorded by the KS-LCR meter and exported as Excel-readable CSV files. Using Matlab 2020b, the results were analysed and plotted alongside the force against time from the CBS testing. Resistance against time and force against time were compared and interpolated.

3. Results

The objective of this study was to evaluate the performance and geometry limitations of 3D-printed continuous fibre-reinforced polymer-SSF components. Figure 4a shows the 3D-printed teardrop shapes, with angles ranging from 5° to 90° and radii from 2 to 20 mm, as detailed in Section 2.2. A comparison is made between the design geometry and the dimensions of the printed parts. We also evaluated the curvature bending stiffness of the semi-circles from the sectioned oval cylinder samples and the resistance resonance as force was applied, (Figure 4b).

3.1. Three-Dimensional Print Teardrop Geometry Repeatability

The accuracy of the printed ‘Teardrop’ geometry was assessed based on the print wall height, width, radius, angles, and polymer-SSF materials relationship, at nine measurement positions for each print.

3.1.1. Wall Width

The average as-printed wall width in the straight section was 2.6 mm ($\pm 0.9$), from b1 to b3 as described in Section 2.3, which was larger than the 1.2 mm designed wall width. The dimension difference results from the polymer-SSF filament volume were greater than the deposition volume. While some polymer was removed during printing through the print nozzle pressure vent, at the lower layer heights, excess polymer was also deposited on the outer surface of the part, thus increasing the wall thickness [16]. The remaining polymer created a section of unreinforced PLA around the outer edges of the teardrops. The CT scan images shown in Figure 5 were obtained for the composites printed with angles of 20° (a radius of 4 mm) and 70° (a radius of 15 mm). Both structures exhibited regions with unreinforced polymer (grey colour) around the inner and outer edges of the print, as just described, and the steel fibre-reinforced region (white colour).

The wall width was found to increase as the teardrop angle decreased. This was illustrated by comparing the a1 position thickness measurements for the teardrop of 20° (a radius of 4 mm) compared with the teardrop of 70° (a radius of 15 mm), as shown in Figure 5b. A similar observation was found at the top of the semi-circle (positions r₃ and b₁). The width dimensions were found to increase at the end of a turn due to fibre migration during more acute cornering.

3.1.2. Angle and Apex

A comparison was made between the design angle from the CAD drawing with the as-printed measured angle for each teardrop geometry. All angles acquired from the straight section of the as-printed teardrops (between $a_1$–$a_3$ and $b_1$–$b_3$ in Figure 2) statistically satisfied the design requirement. However, at the angle’s corner or apex, the deposited polymer or SSF failed to meet the designed apex points (Figure 6). The polymer migrated away from the external design position due to a strong bond between the matrix and fibres. This bond meant that as the fibre migrated, the polymer matrix also migrated, leaving an area with no deposited material, which diminished the external design dimension. This distortion was present at all angles, with an increased level of variation obtained for more acute toolpath cornering, particularly below 5°. Measurements were obtained between the outer apex to the PLA, the outer apex to the SSF outside, and the inner apex to the inner material, as described in Section 2.3.

Figure 6a,b help to compare the level of deviation observed for teardrop prints with angles of 20° and 70°, respectively. During cornering, the continuous SSF exhibited a curved shape, as opposed to the angular design angle specified. This curvature likely resulted from fibre migration due to the print head moving out of a linear toolpath. During turning, its rate of travel was faster than the rate of polymer solidification, which did not hold the SSF in its deposited position. The amount of fibre migration was reduced by decreasing the print speed for more acute toolpath cornering, as indicated in Section 2.2.

The difference between the design position of the outer and inner apex and the deposited polymer or SSF for all geometries is depicted in Figure 6c. The variation observed during cornering is a characteristic of 3D printing continuous fibre composites, as highlighted previously by Cheng et al. [51].

3.1.3. Semi-Circle Cornering

The accuracy of the semi-circle section of the teardrop composites is reviewed in this section. At radii of 5 mm and lower, it was observed that the fibre migration at the end of the turn increased; see measurement position ${\mathrm{r}}_3$ and ${\mathrm{b}}_1$ as described in Section 2.3. This change in the wall width at smaller radii is demonstrated in Figure 7a for a 4 mm radius and Figure 7b for a 15 mm radius. As the design radii reduced from 5 to 2 mm, the fibre migration initiated closer to the start of the cornering toolpath. The semi-circular section average wall width for all radii is plotted in Figure 7c. The increased wall width and data spread significantly increased for radii below 5 mm.

An examination of teardrop composites for all geometries with an inner radius above 4 mm demonstrated that the SSF was located at the position as defined using the CAD design. However, in all cases, a perimeter of unreinforced PLA was located on the outer surface of the composite as illustrated in Figure 7, and as previously described in Section 2.3. The average as-printed dimensions for the 20 mm radius resulted in an inner and outer radius of 14.2 mm (−0.8 mm) and 17.0 mm (+0.8 mm). The steel fibres were located in the positions specified during the printing set-up, with the outer fibre perimeters equivalent to the 16.2 mm design radii. In comparison, for the teardrop with a designed inner radius of 4 mm and outer radii of 5.2 mm, shown in Figure 7a, the deposited SSF outer radius was smaller at 4.9 mm (−0.3 mm). For the resulting as-printed average inner and outer radius of 2.8 mm (−1.3 mm), an outer as-printed radius of 6.0 mm (+0.8 mm) was achieved. All SSF inner radii deposition dimensions were below the design dimensions, as outlined in

Table 4. As-printed teardrop weight and average measured dimensions, where S.D. denotes the standard deviation, − is less, and + is more than the target dimensions.

Measurement	Angle	90°	70°	50°	30°	20°	10°	5°
Code	Radii (R), (mm)	20.0	15.0	10.0	5.0	4.0	3.0	2.0
${\mathrm{R\_inner}}_{PLA}$	Design difference	−0.8	−0.8	−1.1	−1.5	−1.3	−2.5	−1.4
${\mathrm{R\_inner}}_{SSF}$	Design difference	−0.1	−0.3	−0.2	−1.3	−1.2	−2.5	−1.2
${\mathrm{R\_outer}}_{SSF}$	Design difference	+0.1	0.0	+0.1	−0.1	−0.3	−1.9	−0.3
${\mathrm{R\_outer}}_{PLA}$	Design difference	+1.0	+0.8	+1.0	+0.9	+0.8	−0.4	+0.8
C_Length	Design difference	+1.0	+0.8	+1.0	+0.9	+0.8	−0.4	+0.8

.

3.1.4. Height

A statistical analysis was carried out using the Wilcoxon signed-rank test [49]. It demonstrated that in the case of the wall heights, the measured dimensions of all as-printed geometries yielded no significant difference from the design height of 4 mm. A comparison between the callipers and the CT scan techniques used to measure these heights yielded a probability (p-value) of less than 0.05, indicating that the measurements using these two approaches were statistically similar.

3.1.5. Deposition Effects

Figure 8 helps to illustrate the differences between the design and as-printed structures in the angle and semi-circular sections. Each dimension was measured at the bottom (B) and top (T) of the print to evaluate any significant variations in the geometry during layer deposition of the material. The Wilcoxon test indicated no significant difference between the dimensions of the bottom and top layers.

While the measurements on the top and bottom were broadly similar, the variations in the length of material deposited at the angle’s corner or apex varied. The composite length was measured between the outer and inner corner apex again for the base and top, denoted by C_B_length and C_T_length, respectively, as outlined in Figure 2. A plot of estimated marginal means vs. measurement positions in Figure 8a for the seven teardrop shapes. In this figure, the zero on the Y axis indicates no difference between the measured values and the CAD target measurement. For both C_B_length and C_T_length, differences existed between the measured values and CAD design measurements. There was no significant difference between the printed base and top measurements as they had the same overshot. As expected from the results detailed earlier, the variation increased at more acute angles or tighter cornering. However, the difference was very small for the geometry “90°, r = 20 mm” and significant for the geometry “5°, r = 2 mm”.

The inner and outer radius of the semi-circle (R_inner and R_outer) were found to exhibit no significant differences between the experimentally measured and the CAD set points. Figure 8b plots the design dimensions for each teardrop geometry and the resulting as-printed dimensions. As the angles and radii decreased, the geometry differences between the printed structure and the CAD design specifications increased.

Table 4 provides details on the average weight of the printed teardrops with varying angles/radii. All the test samples were printed with 18 filament layers, four concentric paths, and 72 SSF bundles. Despite variances in angle and radius, fabricated teardrop samples with angles and radii greater than 30° and 5 mm yielded as-printed parts with dimensions close to the design geometry. The accuracy of the prints decreased, however, as the acuteness of the turn increased. A potential route to address the unreinforced polymer observed in cornering regions would be to alter either the tooling path or the part design to minimise the offset distance issue observed in this study.

3.2. Curved Bending Test and Resistance Response

To gain an insight into the as-printed structures, the print radii test samples investigated in the geometry study were mechanically tested, while the conducive SSF resistance response was recorded during loading.

3.2.1. Curved Bending Test

The curve bend stiffness evaluation was carried out for PLA-SSF, PLA only, and with Nylon composites with short carbon fibres (Onyx), and short and continuous carbon fibres (Onyx-cCF), as detailed in Figure 9. As the radii of the test parts decreased, there was a broadly linear trend in the CBS for the range between 10 and 20 mm for the inner radii (for the larger test rig, as detailed in Section 2.4) and for 3 to 5 mm (the small test rig).

There are a number of potential explanations for the large CBS values obtained for the PLA-SSF as detailed in Figure 9 with lower CBS values for Onyx-cCF. A combination of both the material and composite structure issues along with printing techniques likely influenced the composite stiffness results. An examination of the as-printed cross-section microscope images of the PLA-SSF shown in Figure 9b with the Onyx-cCF shown in Figure 9c, it is clear that the continuous SSF reinforcing fibres were present at the top and bottom layers of the PLA-SSF. While the cCFs were present only in the middle of the Onyx-cCF composite. During the fabrication of the PLA-SSF composite, each line of deposited material simultaneously contained the matrix polymer and contentious fibre reinforcement. In contrast, the Onyx-cCF structure materials were deposited independently and only one line of CF was added per layer in the middle of the cross section. Another observation in Figure 9c of the Onyx and CCF was a void region on each side of the CCFs with no bonding between them. This void allowed the Onyx and CCFs to shear independently of each other and could contribute to the lack of fracture in the structure.

The first failure mode for all PLA and PLA-SSF samples was a tensile fracture, observed for most of the test samples at the loading contact point. For samples with radii below 4 mm, there was a second crushing action observed, creating an elevated maximum force over the displacement. All PLA samples failed by tensile fracture, while none of the Onyx samples fractured, but their shape was permanently deformed. A similar result was found from short beam testing by Caminero et al. [52].

3.2.2. Resistance Responses

During the curve bend testing, the conducive SSF resistance response was also recorded during loading. The plot of resistance against the applied force for the 10 mm radius sample set in Figure 10 demonstrates the decrease in resistance as loading was increased. Also, Figure 10 contains four photographs showing the effect of loading on the CBS sample 10e.1, at forces of 50, 200, 350 (first fracture), and 325 N (catastrophic failure). It was found that both the 5 and 15 mm sample sets followed a similar pattern, with a decrease in resistance as loading increased, as described by other authors [53]. For test samples where significant breakage of the conductive steel fibres occurred, there was an associated increase in the resistance (tending to infinity). Where complete failure occurred, the resistance $R_t$ was zero; thus, in this situation, the resistance approaches infinity.

The MEX 3D printing process adopted caused minimum damage to individual stainless steel fibres in the bundle. A comparison of the resistance measurements between raw SSF, the PLA-SSF filament, and a single 3D-printed deposited line of the composite resulted in a similar resistance of 8.9 $\Omega$ for a 120 mm length of SSF. The results show there was a low level of resistance fluctuation during testing until, approaching failure. The piezoresistivity of the SSF indicated the relative absence of fibre damage in longitudinal resistance associated with damage to the cCFs, as detailed in Section 1.3.

The resistance measurement at a set loading can vary. Several reasons may account for this, including sample stiffness, preparation methods, internal fibre connections, and movement of the connecting pads during testing. In the future, a potential application of the 3D-printed polymer SSFs is in the fabrication of embedded strain gauges due to the piezoresistance response of the composite to loading. Initially, each sensor would need to be individually calibrated as the material properties can be unique to the components, such as changes in the deposition of the fibres, the density, and fibre alignment during the manufacturing process. Polymer-SSF filaments can be 3D printed for after-market sensitisation onto the surface of pre-existing structures utilising robotic technologies, thus informing designers and engineers for part planning drawings and print parameters.

4. Conclusions

This study evaluates the geometry limitations in the 3D printing of continuous polymer-SSF parts and curvature bending stiffness. Teardrop-shaped components incorporated an angle, straight sections, and a semi-circle dimensional evaluation. To evaluate the dimensional repeatability, a statistical analysis was carried out using a $\mu$CT scan and callipers data obtained from the 3D-printed composite samples with print angles between 5° to 90°, and radii ranging from 2 to 20 mm. It was concluded that there was no significant difference between the radii (5 mm) and the geometric dimensions estimated based on marginal means analysis, indicating a decrease in the dimensional accuracy at the apex and curved sections. At smaller angles (≤5°) and radii (≤4 mm), the acute toolpath cornering towards the end of the turn gave rise to a higher level of distortion. The fibre migration led to some stringing in a number of the print geometries with lower radii semi-circles. This was likely associated with the slow solidification of the polymer as the print head changed direction. As a result, the SSFs moved out of the targeted print track position. Controlling the speed at which the printer head travels around corners was found to be efficient in successfully reducing the level of stringing for the smaller radii print structures. A curvature bending stiffness test methodology was developed to establish a relationship between the curve size (<20 mm) and stiffness. It was demonstrated that there was an increase in the curvature bend stiffness as the radii decreased for all materials tested. Comparison materials included PLA, Onyx (Nylon with short carbon fibres), and Onyx-cCF (Nylon with short and continuous carbon fibre prints). There was almost a linear increase in the curvature bending stiffness as the print radii were reduced in the 3 to 20 mm range investigated. The PLA-SSF composite exhibited the most significant stiffness increase at radii below 5 mm. This study helps to demonstrate the geometric 3D printing limits during cornering SSF-reinforced polymers and how the resultant print gaps obtained due to print geometry limitations can potentially be accounted for by modifying the design dimensions. The piezoresistance response of PLA-SSFs with the applied load during curve bend stiffness testing demonstrated their potential for use as sensors and in structural health monitoring.