1. Introduction
Carbon fibre-reinforced polymers (CFRP) are a strong competitor to the conventional steels used for tensile structural elements and are becoming increasingly attractive in the construction industry due to their outstanding mechanical performance, lower weight, durability, and sustainability [
1]. CFRP has the potential to replace steel ropes and cables, as they are prone to corrosion from environmental exposure that compromises their overall performance and causes substantial expenditure for corrosion protection measures and periodical renewal [
2]. Several structures that implement CFRP tensile elements have been constructed so far, a review of which is presented by Liu et al. [
3]. A recent, world-first example of a large bridge (127 m span) that fully relies on CFRP hangers is the network arch light rail bridge installed in May 2020 over the A8 motorway in Stuttgart, Germany [
4], the deck of which is entirely supported by 72 pin-loaded, unidirectional CFRP strap elements. Another rail bridge with a similar configuration (network arch with CFRP strap hangers, 130 m span) is currently under construction at the Küstrin-Kietz rail crossing over the Oder river at the German–Polish border and is expected to be completed in 2023 [
5]. Due to the low weight of the CFRP hangers, no cranes or supporting pillars were required in the installation process, since they can be easily lifted by hand and installed by two workers from a mobile lifting platform. Moreover, the construction was both more economical and more sustainable than an analogous network arch bridge with flat steel hangers [
4]. This project was groundbreaking in terms of using CFRP as the sole material for the tensile elements supporting the bridge deck, particularly since hangers in such bridges are subjected to high tensile fatigue loads. This motivated the Swiss Federal Laboratories for Material Science and Technology (Empa) previously to investigate the fatigue behaviour of pin-loaded CFRP straps, with a specific focus on the fretting phenomena that are present in this application due to the constant relative motion between the loading pin’s surface and the curved portion of the CFRP strap at the connection points (
Figure 1). Previous experimental studies concerned strap specimens that were laminated using the same materials and a scaled-down geometry from those used in the above-mentioned network arch bridge [
6]. These focused on the fatigue performance of the CFRP straps that were fretted against titanium and CFRP pins at room temperature [
7] and elevated service temperatures [
8], as well as on their thermomechanical behaviour at high temperatures that are representative of accidental load cases (i.e., fire) [
9]. In the study presented herein, a new method of fabrication of CFRP straps based on 3D printing is explored in order to investigate its potential and durability limits against the conventional tape laying and out-of-autoclave lamination techniques that are currently the norm in composite manufacturing. In particular, the fretting fatigue behaviour of the novel straps anchored with titanium pins was studied with a comprehensive series of tensile fatigue experiments on small-scale specimens performed at 23 °C in air. An elastic stress analysis after Schürmann [
10] was confirmed to be an appropriate design tool to capture the triaxial stress state in the vertex area at the onset of the strap’s curvature for the pin-loaded element.
1.1. 3D-Printing of Continuous Fibre-Reinforced Polymer Composites
Several types of CFRP straps are possible, which can be conventionally produced with either lamination, pultrusion, tape-winding, or pull-winding [
3]. In this project, the tension straps were manufactured by 3D-printed continuous CFRP filaments that were subsequently stretched and compacted in a mould before being tested for quasi static tensile strength and under tensile fatigue loading. Three-dimensional printing of continuous CFRP is a relatively new approach to composite manufacturing and has been a topic in research and development in the past ten years. However, based on a recent literature review published by Sanei and Popescu [
11], this technique needs to be thoroughly investigated, especially with respect to fatigue resistance. By implementing a 3D printing process for CFRP, one increases the production flexibility by continuously changing the direction of the fibres through the part with a precise and repeatable fibre deposition without needing a complex mould to produce a preform. There are many different approaches to the 3D printing of CFRP parts with a thermoplastic polymer matrix; these can be categorised by the type of reinforcing fibres (chopped fibres versus continuous tows) and the process of introducing the polymer matrix in the printed composite material (e.g., fusion of pre-impregnated carbon fibre filaments, versus impregnation of dry tows within the extrusion head or on the printing platform itself) [
12,
13,
14]. After the manufacturing of the part by 3D printing, post-processing of the preform is required in the 3D printing methods developed by the composites industry in the last decade [
14,
15,
16,
17]. In particular, it is crucial to perform compression of the printed preform at an elevated temperature to achieve the final geometry and to increase the fibre volume fraction while minimizing porosity and therewith increasing the mechanical properties of the part to a level acceptable for a structural application [
18]. In the opinion of the authors, this additional and necessary post-processing step is a serious limitation of today’s 3D printing FRP composite technologies because it adds considerable cost and component production time.
1.2. Fatigue Behaviour of CFRP Straps
Fatigue crack propagation and damage modes of CFRP are considerably different and more complex than those in isotropic materials, due to factors such as the anisotropic behaviour of the carbon fibres, the viscoelastic nature of the matrix, the fibre–matrix interaction, the layup sequence, and stress concentrations at the load introduction (anchorage) areas. Regardless of the numerous factors influencing the fatigue behaviour, Reifsnider [
19] distinguished three different stages in the fatigue damage of multiaxial fibre-reinforced composites (FRPs). In the first stage, the laminate experiences severe matrix cracking in the off-axis fibre orientation, generally within the first 10–15% of the laminate’s life. At the end of the first stage, intralaminar matrix cracking reaches a uniform saturation spacing. The second stage incorporates up to 80% of the fatigue life, and the damage progression continues, however, at a much slower rate. Stage three is reached when the damage propagation increases for the second time and ends with the failure of the composite. There has been scant research in recent years on the tensile fatigue behaviour of 3D-printed continuous fibre-reinforced thermoplastic composites [
20,
21]. In [
20], the upper stress levels of continuous carbon, glass, and aramid fibre-reinforced nylon were studied in the time domain of the SN curve and reached a maximum of a rather modest 130 MPa for a unidirectional (UD, i.e., 0°) carbon fibre arrangement loaded in the fibre direction failing at 80,000 load cycles. The loading stress ratio was set at R = 0.1 in tensile–tensile load condition. On the contrary, promising initial results were presented in [
21] on the tensile fatigue behaviour (at R = 0 and with a loading frequency of 2 Hz), for flat UD carbon fibre-reinforced polyamide strips printed with a device from [
17]. The obtained SN curve (with a Ps = 50% probability of survival) showed relatively high maximum stress values of 718 MPa achieved for a fatigue life of 293,000 load cycles. Serious limitations in this work were the very low amount of tested fatigue specimens (only five strips at different load levels) and the very low specimen cross-section area (3.9 mm
2) and thickness of 0.6 mm.
Although the fatigue behaviour of FRP composites is complex on its own, in the case of pin-loaded straps it becomes even more complex due to the presence of fretting problems at the pin-to-strap interfaces. Fretting fatigue is a result of wear due to frictional contact between two components that are subjected to cyclic displacement relative to each other [
22]. Friedrich et al. [
23,
24,
25] conducted pioneering investigations on the fretting wear phenomena and fatigue life of carbon fibre-reinforced epoxy laminates, studying the different damage mechanisms for different fretting materials and laminate orientations. Cirino et al. [
26] later also showed that fibre orientation and sliding direction have a strong influence on the abrasive wear behaviour of polymer composite materials and that the optimum wear resistance occurs when the sliding direction is normal to the fibre orientation, whereas material removal is greater when the fibres are oriented in the plane of the sliding surface.
The recent Empa studies of Baschnagel et al. [
6,
7] investigated the fatigue performance of thermoset CFRP straps, in which the curved parts of the UD laminates were fretting under tensile fatigue loading against the anchoring pins. Scaled-down specimen models were used in these studies and were compared against three full-scale strap specimens identical to the actual bridge hangers described in [
4]; the fatigue tests were performed at a frequency of 10 Hz and stress ratio R of 0.1. Microscopic investigations of the small-and full-scale strap specimens revealed carbon fibre thinning and fibre–matrix debris agglomerating in the vertex area of the straps after failure [
7]. The observed fretting products on the pins and straps included mostly short broken fibres, and carbon and resin particles that were attached to their surfaces. The reported ultimate failure mode was delamination that initiated at the end of the straps’ overlap and progressed towards the curved (pin) area, followed by fibre fracture. Overall, the fatigue behaviour of the straps was excellent, the endurance limit being at 750 MPa for straps sustaining a minimum of 3 × 10
6 load cycles. This allowed the team led by Meier [
4] to be granted the “structural design type approval” by the relevant German authorities, which was necessary for the construction of the world-first bridge fully relying on CFRP strap hangers in over the A8 motorway Stuttgart in 2020.
1.3. Aim and Scope
This research study investigates the efficiency of 3D printing fabrication of continuous, unidirectional CFRP straps with thermoplastic matrix using a Fused Filament Fabrication (FFF) technology [
16] and followed by a post-printing compaction process, when compared with the conventional strap fabrication process using out-of-autoclave thermoset prepregs. It seeks to answer whether the purported precision and production efficiency in fabricating composite parts using this robotic fabrication technology offers improved mechanical performance for the case of pin-loaded CFRP straps.
This was addressed by setting up a small-scale production process for a feasibility study using a bench-top 3D printer setup [
16] and by developing a post-printing compaction process (i.e., a stretching jig and compaction mould) at elevated temperature. To investigate the effects of 3D printing with respect to mechanical behaviour, the quasistatic and tensile fatigue performance of the 3D-printed straps was examined. This was done on strap specimens with the same geometric proportions as for the conventional thermoset-matrix straps investigated previously in [
6,
7] (and by following the same scaling down principles from the full-scale bridge straps in [
4]).
4. Elastic Analysis
An analytical estimation of the degree of exploitation of the carbon fibres in the 3D-printed/compacted CFRP strap (meant as the ratio of the tensile strength of the strap compared to the average tensile strength of the UD coupons) was performed by an elastic analysis following [
10]. This analysis is summarized here, and is based on the model of a thick-walled pipe subjected to internal pressure [
34], with the consideration of side supports (ring-shaped webs of the thimbles) made of titanium, as shown in
Figure 11.
We are interested in the three-dimensional stress distribution
in the looped area of the strap supported by the pin and sideways by the ring-shaped webs of the thimble. The curved area of the ‘thick’ CFRP strap is modelled as a cylindrical pipe with carbon fibres in the hoop direction subjected to an internal pressure
. This analysis is performed following an elasto-statics approach ([
10], pp. 485–496), which leads to a closed-form solution for the above stress components in the curved part of the unidirectionally reinforced strap. The following assumptions are made in this strap model: The stresses over the width of the strap (parallel to pin’s axis z) are evenly distributed (i.e., constant). Due to the rotation symmetry, stress differences in the strap cannot arise in the hoop direction (t), but only radially (r). Finally, friction of the strap over the thimble and pin is neglected.
The solution for the radial stress—that varies over the radius coordinate—at the apex (top) of the strap is given by:
The maximum radial stress is located at the inner radius
and corresponds to the haunch pressure
. Knowing the solution (3) for the radial stress distribution
, the tangential (hoop) stress distribution can be determined via equilibrium on the infinitesimal strap element to:
This stress component corresponds to
(average normal stress in fibre direction) in the UD ply and is therefore decisive for the strength analysis of the pin-loaded strap and considers the stiffness ratio
:
This corresponds to the square root of the longitudinal to the transverse stiffness coefficients of the unidirectional CFRP ply in its local coordinate system, which is defined by the fibre direction (
-axis) and the fibre perpendicular axis of the UD ply (⊥).
The above-cited haunch pressure
caused by the total force (tension)
loading the pin is given per equilibrium by:
When supporting sideways with the ring-shaped webs of the titanium thimble (or in the test setup shown in
Figure 8B with the side surfaces of the loading adapters), we obtain a three-dimensional stress state. The maximum normal (compressive) stress in the axial direction
appears at the inner radius of the strap. It is computed after [
10]—out of
and
at the radial coordinate
:
The main result of this elasto-static stress analysis is that at the strap inner radius
(
Figure 11 shows the coordinate system), we obtain pronounced stress peaks. We now adopt the as called “refined” Puck fibre failure criteria for the unidirectional ply [
35], describing the tensile failure of the CFRP strap due to fibre tensile fracture:
With being the stress exposure value for fibre failure (i.e., corresponds to fibre fracture in tension or compression) and corresponding to the tensile strength of the UD coupon ( would be its compressive strength, not relevant for the pin-loaded strap under tension).
An iterative solution of Equation (9) for fibre tensile failure, i.e.,
at
for the vertex of the 3D-printed and compacted straps investigated leads to a theoretical strap tensile failure load of F = 52.3 kN
. The CFRP strap’s geometric and material properties considered are given in
Table 5.
The tensile experiments at room temperature gave an average tensile strength of the CFRP straps of 1314 MPa (
Table 4), which is only 70.7% of the above-estimated theoretical strap tensile strength of 1857 MPa (corresponding to F = 52.3 kN).
Reasons for this theoretical overestimation are to be found in the assumptions and idealisations made in Schürmann’s analysis: In [
10], an ideal UD ply with perfect UD fibre alignment and fibre–matrix composite action is assumed while the straps investigated show fibre waviness, see
Appendix B, and zones with low impregnation of the carbon fibres particularly at the critical curvatures of the straps,
Appendix C. In addition, the presence of friction between the pin and the UD strap ply in the experiments leads to stress concentrations along the curvature [
36]. One also needs to consider that the tensile strength
of the CFPA12 coupons is probably higher than the strength of the shaft of the strap (due to better compaction in the thickness direction and fibre alignment in the coupon) and that Schürmann’s model does not consider residual stresses due to differential thermal fibre/matrix expansions in the FFF process nor local bending effects at the strap’s vertex: As discussed in
Section 3.3, the tensile load transfer at the strap end curvatures leads to local bending stresses in the vertex area that are superimposed to the above
-stress peak at
[
10]. This has been analysed via finite element modelling in [
6] for geometrically very similar CFRP straps made of IMS60 carbon fibres with an epoxy matrix. This work showed a local
increase of 30% leading to premature strap failure at the onset of its curvature (the as-called “strap vertex”, see
Figure 9).
5. Discussion and Conclusions
The results presented in this study show a good potential for further research on 3D-printed CFRP straps under the assumption of an appropriate compaction procedure. In standard tensile tests at room temperature, the average strap strength was 1314 MPa (standard deviation of six specimens 149 MPa). This corresponds to 70.7% of the theoretical tensile strength of the strap analysed with Schürmann’s elasto-static analysis [
10] considering refined Puck fibre fracture criteria [
35]. The deviation is explained by model assumptions and UD ply imperfections.
Without a post-processing step consisting of a longitudinal stretching followed by a transverse consolidation, the average strap strength was only 264 MPa (standard deviation of five specimens 56 MPa). This is only 14.2% of the theoretical value according to Schürmann and 20% of the tensile strength of a consolidated strap. This highlights the need for appropriate post processing.
The SN curve obtained in the tensile fatigue of 28 post-processed strap specimens at a loading frequency of 10 Hz (R = 0.1) and using Ti64 pins to anchor the straps is similar to that obtained for OAA laminated straps based on unidirectional carbon fibre epoxy prepregs and investigated previously by the corresponding author’s Empa laboratory [
6]. A fatigue endurance limit for the thermoplastic matrix straps of 500 MPa could be determined. Two respective four samples loaded with a maximum tensile stress of 500 MPa were able to withstand
and 9
fatigue load cycles, and their remnant tensile strengths (1222 MPa after 3 million load cycles and 1244 MPa after 9 million load cycles) were only slightly lower than that of pristine straps. The fatigue endurance limit of 500 MPa at R = 0.1 and 10 Hz would correspond to approximately 38% of the CFRP straps’ ultimate tensile strength.
As a comparison, straps made with a stronger IMS60 carbon fibre that were conventionally produced with an out-of-autoclave process (epoxy matrix) have a fatigue endurance limit of 750 MPa, which corresponds to 46% of their ultimate tensile strength [
6].
The main drawback of the processing method presented in this paper is the necessary post processing after 3D printing (FFF) of the CFRP straps. This includes an axial stretching followed by a transversal post compaction (i.e., a 6 MPa compression of the strap at 200 °C in the width direction) in a complex and expensive steel mould. This tool needs to be designed and produced for each strap geometry in a practical application, e.g., for network arch bridge straps with a diameter of 33 mm and lengths of several metres [
4]. This rather demanding stretching and compaction, which is necessary to exploit the carbon fibre strength in the looped tensile element, is a clear limitation of the 3D printing technique by FFF of thermoplastic matrix straps as it makes the strap production expensive and inflexible. It therefore greatly compromises the advantages (geometric freedom, fast production) of printing CFRP laminates with a thermoplastic matrix. Novel research from Japan [
37] is trying to avoid this additional consolidation post-processing step by integrating compaction in the 3D printing head with an advanced additive manufacturing device.
The presented results on the tensile fatigue performance of 3D-printed, axially stretched, and transversally compressed CFRP straps look promising, and further research will focus on the improvement of the manufacturing process of the strap and mould design in order to improve fibre impregnation (
Appendix C) and to reduce fibre waviness (
Appendix B). With further optimization of the post-compacting mould design, there is potential to further enhance the UD strap’s quality and therefore the tensile and fatigue performance of the CFRP strap.
For a better exploitation of the geometrical flexibility of the presented FFF 3D printing process, further developments focussing on the topological optimization of the curved end areas of tension straps should be performed in the future with the aim to reduce the discussed stress concentrations at the strap vertex.