Professional Documents
Culture Documents
Mechanical Behaviour of Flexible 3D Printed Gyroid Structures As A Tuneable
Mechanical Behaviour of Flexible 3D Printed Gyroid Structures As A Tuneable
Additive Manufacturing
journal homepage: www.elsevier.com/locate/addma
A R T I C L E I N F O A B S T R A C T
Keywords: Various areas of healthcare utilise custom foam cushioning to treat or mitigate conditions like pressure ulcers, or
Metamaterials to provide personalised support structures for patients with specific clinical needs. Polyurethane foams are often
Gyroid used; however, such materials require significant time and expertise to combine different foam types into a
Flexible filament
device that provides sufficient structural support in some areas, with soft pressure distribution in others. In this
Foam
Mechanical response
paper, flexible 3D printed gyroid based metamaterials are investigated as a tuneable replacement for poly
Fused filament fabrication (FFF) urethane foams. The impact of changing key gyroid structural characteristics on the material’s mechanical
Thermoplastic polyurethane (TPU) response is examined. Samples with six different unit cell geometries for each of two flexible TPU 3D printing
filaments (NinjaFlex and Flexion X60) were produced using fused filament fabrication, tested, and compared to
three types of conventional polyurethane rehabilitation foam. Compression tests were conducted focussing on
compressive stress-strain response, strain rate effect, print layer effect, and cyclic fatigue behaviour. In all tests it
was observed that gyroid samples of both filament types were able to produce compressive responses comparable
to the foams. Solid volume fraction was determined as the critical gyroid geometric parameter that influenced
compressive response, and solid volume fractions capable of reproducing the specific response of each of the
three rehab foams were determined. It is shown that 3D printed gyroid materials are a viable replacement for soft
polyurethane foams, and the direct control of material response possible with simple geometric changes means
such metamaterials may lead to improved optimisation of rehabilitation cushions.
1. Introduction foam cushions with layers of different density foam or shaping elements
are typically implemented to prevent pressure injuries [4]. The
Pressure ulcers, also known as decubitus ulcers; bed sores; pressure patient-specific nature of such a solution, however, requires significant
sores; or pressure injuries, are a significant contributor to patient time, cost, and expertise to produce. Temperature regulation and
discomfort and morbidity in the modern healthcare system. Circum cleaning of existing foam solutions is also challenging.
stances that may result in pressure ulcers typically involve small sus In many areas of healthcare, when patient specific solutions are
tained pressure, shear, and/or friction on a person’s skin [1,2]. This needed, additive manufacturing or “3D printing” has found wide
could result from extended confinement to bed due to immobility or ranging application. By connecting digital scanning technologies, com
illness, use of a wheelchair, or other circumstances where puter modelling, and cost-effective one-off manufacturing, additive
semi-permanent skin contact is present (like for prosthetics). To prevent manufacturing is having significant impact in the areas of implants [5,
pressure ulcers, specialised support structures are typically implemented 6], prosthetics [7], scaffolds [8–10], dentistry [11–13], and other areas
and often take the form of overlays that more evenly distribute associ of clinical importance [14,15]. Additive manufacturing technologies
ated contact loads [3]. In the case of wheelchairs, custom contoured have also been employed as a fabrication method for complex and
* Correspondence to: Queensland University of Technology, 2 George St, Brisbane, QLD 4000, Australia.
E-mail address: d.holmes@qut.edu.au (D.W. Holmes).
https://doi.org/10.1016/j.addma.2021.102555
Received 30 August 2021; Received in revised form 14 November 2021; Accepted 8 December 2021
Available online 11 December 2021
2214-8604/© 2021 Elsevier B.V. All rights reserved.
D.W. Holmes et al. Additive Manufacturing 50 (2022) 102555
flexible metamaterial structures, due to their ability to produce previ testing will look at compression, strain rate, layer orientation, and fa
ously impractical lattice geometries [16–19]. Some 3D printed meta tigue behaviour of the materials, with the aim of sufficiently charac
materials have been shown to mimic the behaviour of soft foams [20], terising the metamaterial foams to enable potential future use in medical
while others have demonstrated tunability of mechanical response via pressure offloading and other applications. In the following, Section 2
control of scaffold structural characteristics [21]. By utilising the details the structures and materials used, additive manufacturing pro
strength of additive manufacturing technologies for patient custom cesses employed, and testing methodology implemented. Section 3 then
isation, and developing soft materials with tailored mechanical confor details the test results across a range of studies, and associated discus
mity, it would become possible to develop devices that could perfectly sion. Section 4 finally presents conclusions and future work.
support a specific patient’s body contours. Many metamaterial struc
tures also have a more open structure than foam, with significant po 2. Materials and methods
tential benefits for cushion temperature regulation and cleaning. Beyond
healthcare, 3D printed flexible metamaterials would also have applica 2.1. Gyroid structure development
tion in areas like custom seating for motorsports, personalised shoes in
running sports, specialised helmets for impact protection, and a range of A range of sheet gyroid based TPMS geometries will be investigated
other areas. Significant further investigation is needed, however, to in this work to determine the impact of key structural characteristics on
determine the best material, lattice structure, and structural parameters the mechanical behaviour of prints from flexible TPU filament. Sheet-
that will most effectively enable controlled cushioning behaviours for based gyroids have been selected due to their suitability for 3D prin
such applications. ted structures and the simplicity of tuning the mechanical response
A growing range of elastomeric materials have been used within 3D through unit cell size and wall thickness specification [34]. The geom
printed lattice structures in recent times, including rubber-like UV cross- etries themselves are based on the theoretical gyroid surface following
linkable TangoBlack+ [22], polydimethylsiloxane (PDMS) silicon rub the mathematical definition
ber [20], tin-catalyzed silicone elastomers [23]1 and the thermoplastic
sin(x)cos(y) + sin(y)cos(z) + sin(z)cos(x) = 0 (1)
polyurethanes (TPU) SemiFlex [24], NinjaFlex [25] and most recently
Desmopan 3690 [26]. While the ability of such metamaterials to
where x, y and z are cartesian coordinates ranging between 0 and 2π,
reproduce soft foam-like behaviour has been shown in some of these
inclusively, and Eq. (1) defines a minimal surface in this space creating a
works, more work is required to determine repeatable structural features
single zero-thickness unit cell. Scaling this surface to the desired unit cell
and parameters that would enable controlled tuning of the compressive
side length, repeating the structure to a desired cell count in each di
response of such materials for custom cushion applications. Emerging
rection, and thickening the surface by a desired wall thickness defines
ultra-flexible polyester-based TPU filaments like the Flexion X60 [27]
the sheet gyroid solid for printing (see Fig. 1). All sample wall edges
are also yet to be characterised in the literature and show significant
were terminated parallel to the sample sides to minimise any possible
potential for this application.
boundary effects between the sample and test platens during testing.
While structural optimisation and tuning of flexible metamaterial
These geometries and the definition of the subsequent STL mesh ge
structures remains an open research question, investigations on rigid
ometry for printing were created using the MATLAB 2021a software
additively manufactured lattices have received significant attention in
(MathWorks, USA).
recent years. Spanning materials such as hard plastics (e.g., Polyamide
The specific structural options investigated in this work spanned unit
12, Acrylonitrile Butadiene Styrene (ABS)), and metals (e.g., Al-Si10-
cell side lengths, s, of 10, 15, and 20 mm, and wall thickness, t, of 0.6
Mg, Ti-6Al-4V, stainless steel), the most applicable of these in
and 1.2 mm. This resulted in 6 different sample structures for each of the
vestigations look at the mechanical performance of 3D printed lattices
two filament materials investigated. Each sample comprised repeating
based on shapes derived from differential geometry known as triply
cells sufficient to make a 60 × 60 × 60 mm total sample size. The
periodic minimal surfaces (TPMS) [28]. Such studies have been carried
effective solid volume fraction (SVF) resulting from the produced gyroid
out on TPMS variants such as Schwarz Primitive, Schoen IWP and
structures is defined following
Neovius [28–31], and especially the Gyroid TPMS [21,22,32–38]. The
studies generally vary lattice parameters such as strut/wall thickness, SVF =
Vwall
(2)
unit cell size, and corresponding solid volume fraction (net density) to Vwall + Vvoid
investigate their effect on mechanical response including compressive
strength, elastic moduli, energy absorption, and other critical measures. with Vwall and Vvoid the volumes of the solid and void respectively
A range of these studies have also looked at functionally graded struc (Fig. 1c). Here, theoretical Vwall was determined directly from the STL
tures where localised structural modifications enable local variation in files using MeshLab, version 2016.12 [39]. Vwall +Vvoid is equal to total
mechanical response [21,28,31,33,34,38]. The ability to easily control sample volume. An analysis of the target, and actual as-printed SVFs is
lattice mechanical response makes such geometries an ideal focus for the presented in Section 2.4.
present study.
In this paper we present a systematic investigation into the me 2.2. Flexible TPU filament selection
chanical behaviour of a flexible 3D printed metamaterial structure and
the material and morphological factors that influence its behaviour. We Filaments chosen for the study were the NinjaFlex filament from
focus the work on sheet gyroid-based geometries for a range of solid NinjaTek, and Flexion X60 from Diabase Engineering. These two fila
volume fractions, and test two different TPU filament materials (Nin ments represent some of the most flexible currently available with
jaFlex and Flexion X60) printed using Fused Filament Fabrication (FFF). filament properties provided in Table 1. Most notably, the Shore hard
The objective of this work is to investigate the mechanical tunability of ness of X60 of 60 A is stated by the manufacturer as being the “most
3D printed metamaterial foams and assess their capacity to replicate the flexible filament on the market” [27], but is yet to be characterised in the
behaviour of traditional elastomeric foams. Results will be compared literature. The two filament hardness values chosen ensure the results
with existing soft cushion foams to enable this comparison. A range of represent a range of material options.
2
D.W. Holmes et al. Additive Manufacturing 50 (2022) 102555
Fig. 1. Gyroid structural model showing (a) base surface model, (b) unit cell, and (c) volume element.
3
D.W. Holmes et al. Additive Manufacturing 50 (2022) 102555
Fig. 2. a) Prusa i3 MK3S printer used with Flexion extruder fitted, and b) close-up of Flexion extruder with E3D V6 HotEnd.
Fig. 3. 60 mm3 gyroid printed samples of NinjaFlex at different cell size, and wall thickness: a) 0.6 mm wall, 10 mm cell; b) 0.6 mm wall, 15 mm cell; c) 0.6 mm wall
20 mm cell; d) 1.2 mm wall, 10 mm cell; e) 1.2 mm wall, 15 mm cell; f) 1.2 mm wall 20 mm cell (length scale on zoomed photos shows half unit cell length).
Three categories of polyurethane foam commonly used in clinical 2.5. Testing methodology
rehabilitation where also tested for comparison with the NinjaFlex and
X60 gyroid structures. The foams chosen for testing were Green 2.5.1. Compression testing
EN36–90, Yellow EN40–230, and Pink MA35–600 foam manufactured Mechanical compression testing of the gyroid printed samples was
by Dunlop Foams, Australia, representing soft, medium, and firm performed to determine material loading behaviour. The testing was
response, and the manufacturer specified bulk densities of the foams carried out, and compression stress-strain characteristics determined,
were 36, 40 and 35 kg/m3, respectively. Effective SVFs of the foams following the international standards for Polymeric materials, cellular
were in the range of 3–4% (i.e., approximately comparable to the lowest flexible – Determination of stress-strain characteristic in compression – Part
SVF gyroids). As with the NinjaFlex and X60, three 60 × 60 × 60 mm 1: Low-density materials BS EN ISO 3386–1:1997 +A1:2010 and Part 2:
samples of each category were used for each test. High-density materials BS EN ISO 3386–2:1998 +A1:2010. The 60 mm3
sample size used met that required by the standard (minimum 50 mm
4
D.W. Holmes et al. Additive Manufacturing 50 (2022) 102555
Fig. 4. 60 mm3 gyroid printed samples of X60 at different cell size, and wall thickness: a) 0.6 mm wall, 10 mm cell; b) 0.6 mm wall, 15 mm cell; c) 0.6 mm wall
20 mm cell; d) 1.2 mm wall, 10 mm cell; e) 1.2 mm wall, 15 mm cell; f) 1.2 mm wall 20 mm cell (length scale on zoomed photos shows half unit cell length).
Table 2
Comparison of target (theoretical) vs actual measured solid volume fraction for the NinjaFlex and X60 printed samples. Actual mean SFV calculated based on filament
density and measured sample weights, averaged over the three samples for each cell size / wall thickness.
Material Cell Size Wall Thickness Target Actual Mean Bulk Density Actual Mean Relative Standard Deviation of SVF Variation:
(mm) (mm) SVF (kg/m3) SVF Actual SVF Actual − Target
Target
NinjaFlex 10 0.6 18.46% 172.79 14.52% 0.37% -21.34%
NinjaFlex 15 0.6 12.34% 117.16 9.85% 0.19% -20.22%
NinjaFlex 20 0.6 9.27% 90.85 7.63% 0.51% -17.60%
NinjaFlex 10 1.2 36.38% 343.46 28.86% 0.84% -20.66%
NinjaFlex 15 1.2 24.52% 238.44 20.04% 0.97% -18.29%
NinjaFlex 20 1.2 18.46% 188.47 15.84% 0.45% -14.21%
X60 10 0.6 18.46% 174.33 14.90% 0.58% -19.28%
X60 15 0.6 12.34% 118.7 10.15% 0.24% -17.79%
X60 20 0.6 9.27% 84.17 7.19% 0.65% -22.36%
X60 10 1.2 36.38% 303.28 25.92% 0.61% -28.74%
X60 15 1.2 24.52% 249.8 21.35% 1.49% -12.93%
X60 20 1.2 18.46% 166.39 14.22% 0.26% -22.97%
side), and prior to testing the test samples were conditioned undeflected m3). Due to this, and because application of the materials investigated in
and undistorted for at least 16 h at 23 ◦ C and 50% relative humidity. this work most closely align with soft or “low density” foams, 100 mm/
The gyroid printed samples for each structure and filament type, and min was determined to be most appropriate rate for all compression tests
the comparison rehabilitation foams, were tested in triplicate at a strain to ensure consistency throughout. The impact that decreasing the rate to
rate of 100 mm/min up to a strain of 80% (48 mm compressive 5 mm/min as specified in BS EN ISO 3368 Part 2 has also been inves
displacement) for five consecutive load-unload cycles using an Instron tigated in an additional study set for completeness (see Section 2.5.2).
5967 Universal Testing Machine (Illinois Tool Works Inc., USA) with
30 kN load cell. The corresponding compressive stress-strain behaviour 2.5.2. Strain rate evaluation
was recorded for the last of the five loading cycles as per the method Three further gyroid samples of each of the two filament materials
detailed in the standard. All the samples were subjected to a preload of (six samples in total), with 20 mm unit cell size and 1.2 mm wall
0.5 N to maintain initial surface contact between the force platen and thickness were printed to carry out a strain rate dependence study. These
the sample’s top surface throughout the five cycles. cell characteristics were chosen as a middle range SVF representation of
The strain rate for testing samples was chosen as 100 mm/min as per the whole sample set. The three samples of each material were com
Part 1 of BS EN ISO 3386, as opposed to 5 mm/min outlined in Part 2. pressed at a strain rate of 5 mm/min and compared to the three samples
While two out of the twelve gyroid classes exceed the specified 250 kg/ compressed at 100 mm/min. Aside from strain rate, all test procedures
m3 bulk density limit stated in the scope of Part 1 (i.e., the 10 mm cell, from Section 2.5.1 were upheld for the slower testing. Based on these
1.2 mm wall samples for both NinjaFlex and X60, see Table 2), the tests, the effect of strain rate on the sample compressive behaviour was
targeted rehab foam samples with comparable strength to the gyroids studied.
were all well within the Part 1 specified density range (i.e., 35–40 kg/
5
D.W. Holmes et al. Additive Manufacturing 50 (2022) 102555
3.1. Sample compressive stress repeatability of compressive behaviour of the 3D printed gyroid samples is equivalent
to the rebab foams, demonstrating their potential utility as an effective
The following presents an analysis of the compressive stress repeat substitute for rehab foam.
ability of 3D printed gyroid foams compared to the rehab foam samples,
to assess the consistency of using 3D printing for manufacture. Six
gyroid structures for each of the two filament types and three rehab 3.2. Sample large deformation compressive response
foam types were tested, with three different samples for each producing
45 total compression test sets. To analyse the variance between the three This section presents an analysis of the compressive strength and
samples for each class, and determine repeatability of the printed sam behaviour of each of the sample sets, analysing the effects of gyroid
ples particularly, the compressive stress value at 10% and 50% strain for structural characteristics and comparing this with the rehab foams.
each test are presented in Fig. 5, along with the associated relative Images of an indicative compression cycle for one of the NinjaFlex
standard deviation (RSD) for the set. The full compressive stress-strain gyroid tests is provided in Fig. 7. The compression of the gyroids were
plots for the three sets with highest sample variance are presented in characterised by an initial distributed elastic response, followed by a
Fig. 6. period of sequential layer-on-layer collapse, and a final densification
The results suggest that the 3D printed gyroid structures and the stage once all layers were closed on one another (Fig. 7 right). The
rehab foam samples have similar degrees of repeatability, with the RSD rehabilitation foams had a much smaller cell size, however similar
of compressive stress values under 10% for the majority of sample sets. indicative zones were observed in the stress strain curves, suggesting a
At 50% strain, one sample from each of the gyroid filament materials, broadly similar mechanism.
and two of the rehab foam sets had RSD less than 1%, while the softest of To compare between structural and material characteristics, an
the rehab foams had the greatest RSD of all samples at 26.82% (at 10% average of the three stress strain curves for each sample class was
strain) and 9.84% (at 50% strain). Fig. 6 shows that the spread between calculated. The resulting average compressive stress-strain curves for
stress curves remains quite consistent throughout the whole strain the six different cell and wall size options for NinjaFlex and six cell and
range, contributing to the apparent drop in relative standard deviation wall size options for X60, and the three classes of rehab foam, are all
observed between 10% and 50% strain. In all cases, the reproducibility presented in Fig. 8. For the same structural dimensions, NinjaFlex
6
D.W. Holmes et al. Additive Manufacturing 50 (2022) 102555
Fig. 6. Comparisons of the compressive stress results between three like samples showing the repeatability of results for each sample set: a) NinjaFlex at 20 mm cell,
0.6 mm wall, b) X60 at 15 mm cell, 1.2 mm wall, and c) Rehab Foam Green EN36-90. In each case, the results represent those with the greatest variation of all in that
material class. Red circles indicate the points captured in Fig. 5. (For interpretation of the references to colour in this figure legend, the reader is referred to the web
version of this article.)
Fig. 7. Compressive deformation images for NinjaFlex compression test with SVF 15.84%, 20 mm cell, 1.2 mm wall.
7
D.W. Holmes et al. Additive Manufacturing 50 (2022) 102555
Fig. 9. Comparison between rehab foam compressive stress strain plots and
candidate X60 and NinjaFlex topologies of similar strength (average result with
n = 3 for each line).
demonstrated a significantly stiffer compressive response; approxi 3.3. Loading rate analysis and comparison between samples
mately double that of X60. The three rehab foam classes corresponded to
the softest options of the gyroid foams. A comparison of the softest two An investigation into the influence of strain rate on compressive
NinjaFlex, and softest two X60 options is presented with the three rehab stress response for the gyroid foam samples was carried out on mid-
foam curves in Fig. 9 for direct comparison. range SVF samples with 20 mm cell and 1.2 mm wall for both Ninja
The compression response comparison in Fig. 9 shows that of the Flex and X60 filament. The results of these tests are shown in Fig. 11. As
gyroid material and unit cell options tested, the 20 mm cell, 0.6 mm wall expected, for each filament material, higher strain rate corresponds to a
X60 samples most closely match the response of the softest Green higher stress response. The difference between these responses is
EN36–90 rehab foam. Both the 15 mm cell, 0.6 mm wall X60, and the modest, with a 20× increase in strain rate corresponding to an increase
20 mm cell, 0.6 mm wall NinjaFlex samples matched well with the in stress of 1.24× for NinjaFlex and 1.46× for X60 (calculated at 10%
8
D.W. Holmes et al. Additive Manufacturing 50 (2022) 102555
Fig. 10. Comparative analysis of various mechanical responses against solid volume fraction for the investigated materials showing (a) compressive stress values at
10% strain, (b) compressive stress values 50% strain, (c) compressive modulus values, and (d) energy absorbed per unit volume over the first 50% strain. In each case,
both NinjaFlex and X60 results demonstrate a power law fit to the data. For the three rehab foam points in each plot, vertical point position follows Green EN36–90
lowest, Yellow EN40-230 middle, and Pink MA35–600 highest, and SVF was calculated based on the bulk density and an assumed raw polyurethane specific gravity
of 1 g/m3 for comparison purposes.
strain). The stress-strain curves are of comparable shape between the approaches the sample variation seen within a given sample class (see
strain rates indicating that rate shifts stress magnitude but maintains Section 3.1). The difference is slightly less for NinjaFlex than for X60 and
similar deformation behaviour throughout the testing. is consistent with NinjaFlex being the easier filament to print (i.e., prints
observed to be more consistent and have slightly fewer print artefacts
than X60). No damage or weakening of samples was observed upon re-
3.4. Effect of layer orientation on mechanical strength testing in the opposite layer orientation, and the consistency of the re
sults, particularly those observed for X60, indicate that layer orientation
A layer orientation study was performed to investigate the impact seems to have a small and deterministic impact on stress response.
that 3D print layer-based anisotropy has on compressive response. A
middle range SVF for each material was chosen with the smaller 0.6 mm
wall having the greatest likelihood of structural anisotropy due to the 3.5. Cyclic fatigue behaviour
greater challenge of printing thinner walls. Fig. 12 presents results for
the NinjaFlex and X60. Three out of the four gyroid samples (the Nin Cyclic testing was carried out on a representative sample from each
jaFlex sample that is first tested perpendicular to the test platens and filament type and one rehab foam class to provide preliminary insight
then horizontal to them, and both X60 samples) show a clear trend for into the fatigue-based behaviour of the gyroid materials, and to identify
horizontal orientation having a slightly higher stress response when any potential differences in cyclic degradation between them and con
compared to the perpendicular (i.e., upright) layer orientation. This is ventional rehab foam used for wheelchair seat applications. Peak stress
irrespective of whether the horizontal or perpendicular sample orien value against cycle number, and the full stress strain curves at 1, 100,
tation test is carried out first. The fourth sample (the NinjaFlex sample and 1000 cycles are presented for NinjaFlex, X60, and Yellow EN40–230
that is first tested with layers horizontal and then perpendicular) also rehab foam in Fig. 13. While 1000 cycles are below realistic fatigue
partially shows this trend except for a small region where the perpen requirements of cushion materials, the response observed within the
dicular test line surpasses the horizontal one (i.e., between 15% and first 1000 cycles is an indicator of longer term behaviour (see for
35% strain, Fig. 12 (a)). This can largely be attributed to minor variance example the longer term testing of Paul et al. [42] for similar rehab foam
within the sample. In each case, the difference in compressive stress materials), and the preliminary comparison provided is a valuable pre
between horizontal and perpendicular layer orientation is small and lude to more comprehensive fatigue testing to be published elsewhere.
9
D.W. Holmes et al. Additive Manufacturing 50 (2022) 102555
Fig. 12. Comparison of layer orientation tests for a) 10 mm cell, 0.6 mm wall NinjaFlex, and b) 10 mm cell, 0.6 mm wall X60, showing compressive stress vs strain
for samples with print layer in vertical and horizontal orientations relative to the test platens. Vertical and horizontal orientations were tested on the same sample
with vertical first for one, and horizontal first for the other to eliminate any potential history or damage influence from the previous test.
10
D.W. Holmes et al. Additive Manufacturing 50 (2022) 102555
Fig. 13. Cyclic testing for NinjaFlex 20 mm cell, 1.2 mm wall sample, X60 20 mm cell, 1.2 mm wall sample, and Rehab Foam Yellow EN40-230 sample showing (left
plots) peak compressive stress value reached for each of the 1000 cycles, and (right plots) the compressive stress vs strain plots of the 1st, 100th, and 1000th cycle
for each.
For the NinjaFlex sample, the more pronounced decrease in compressive X60 in particular, have a comparable resistance to fatigue-based
stress over the 1000 cycles (Fig. 13) may be attributable in part to some degradation to the rehab foam, and this result further supports the
temperature dependant change in material properties. The minor case for use of such metamaterials as a tuneable substitute.
change in temperature of the other two samples is believed to have had While a 15 N preload was needed to prevent separation between the
negligible influence on material response throughout the testing, and platens and the samples during the tests, it did have a differing effect on
both X60 and the rehab foam stress peaks are seen to asymptote after each sample. NinjaFlex and X60 samples were harder materials, with
approximately 300 cycles. The results indicate that the gyroid samples, 15 N representing only a small proportion of the force response,
11
D.W. Holmes et al. Additive Manufacturing 50 (2022) 102555
Fig. 14. Thermal imagery of cyclic compression testing on gyroid structures; a) NinjaFlex sample in fatigue apparatus, b) NinjaFlex thermal image at 1st compression
cycle, c) NinjaFlex thermal image at 1000th cycle, d) X60 sample in fatigue apparatus41, e) X60 thermal image at 1st compression cycle, f) X60 thermal image at
1000th cycle, g) Rehab Foam Yellow EN40–230 sample in fatigue apparatus, h) Rehab Foam Yellow EN40–230 thermal image at 1st compression cycle, i) Rehab
Foam Yellow EN40–230 thermal image at 1000th cycle32. Temperature reading on thermal imagery is for point at centre of crosshair.
12
D.W. Holmes et al. Additive Manufacturing 50 (2022) 102555
Table 3 Table 4
Summary of mechanical properties determined for the metamaterial and foam Summary of equivalent solid volume fractions for NinjaFlex and X60 to achieve
structures tested, including compressive modulus E, energy absorbed per unit comparable properties to three tested examples of rehabilitation foam.
volume over 50% strain, W@50%, and 2% offset compressive yield strength, Sy. Equivalent rehab foam Target SVF NinjaFlex Gyroid Target SVF X60 Gyroid
NinjaFlex
Green EN36–90 4.81% 6.93%
SVF 7.63% 9.85% 14.52% 15.84% 20.04% 28.86% Yellow EN40–230 6.91% 10.10%
E (kPa) 177.78 230.37 400.00 715.11 1020.28 1495.19 Pink MA35–600 8.42% 12.51%
W@50% (kJ/ 3.24 6.14 16.11 29.02 49.38 95.14
m3 )
Sy (kPa) 5.64 10.27 26.60 54.80 97.79 157.85 structure of the gyroids and associated advantages for comfort, skin
X60 health, and treatment or mitigation of pressure ulcers, needs to be
SVF 7.19% 10.15% 14.22% 14.90% 21.35% 25.92%
carefully investigated. Finally, the results, insights, and geometric de
E (kPa) 51.66 98.15 238.52 177.41 407.36 419.17
W@50% (kJ/ 1.10 2.87 10.41 7.79 24.17 28.44 tails presented in this work provide all necessary information to enable
m3 ) controlled production of flexible 3D printed structures to reproduce any
Sy (kPa) 1.85 5.01 20.87 13.79 51.83 45.90 required soft foam material response. Ultimately, the high tailorability
Rehab Foams of these gyroid foams may lead to improved optimisation of rehabili
Type Green EN36–90 Yellow EN40–230 Pink MA35–600
E (kPa) 30.00 86.67 190.00
tation cushions (such as wheelchair cushions) compared to traditional
W@50% (kJ/ 1.45 3.23 4.34 rehab foams and facilitate improved clinical outcomes for patients.
m3 )
Sy (kPa) 2.21 4.82 7.45 CRediT authorship contribution statement
mechanical properties, with repeatability measures similar to those Holmes David William: Writing – original draft, Visualization,
of the conventional rehab foams. Anisotropy due to layer orientation Validation, Software, Methodology, Investigation, Formal analysis, Data
of the printed gyroid structures has also been shown to be minimal curation, Conceptualization. Powell Sean K: Writing – review & edit
and within the range of sample repeatability. ing, Funding acquisition. Woodruff Maria A: Writing – review & edit
4. The 3D printed flexible gyroid structures have also been shown to ing, Supervision, Resources, Project administration, Funding
exhibit minimal strength degradation and show resistance to damage acquisition, Conceptualization. Pickering Edmund: Writing – review &
during preliminary cyclic fatigue testing. Again, when subjected to editing. Paxton Naomi C: Writing – review & editing, Supervision,
cyclic testing, the samples tested produced behaviour directly com Funding acquisition. Forrestal David: Writing – review & editing,
parable to that of the rehab foam. No issues with damage, print Investigation, Conceptualization, Funding acquisition. Slattery Peter:
quality, or de-lamination were observed at any stage of the testing, Writing – review & editing, Resources, Funding acquisition. Daley
including during fatigue testing. Ryan: Writing – review & editing, Methodology, Investigation,
5. In terms of flexible TPU filament, X60 was found to be more chal Conceptualization. Singh Dilpreet: Writing – original draft, Validation,
lenging to print accurately due to its highly flexible nature, with Methodology, Investigation, Data curation. Lamont Riki: Writing –
more artefacts visible in the final samples. The material response, original draft, Software, Methodology.
however, was found to be consistent across the samples, with vari
ances similar to NinjaFlex. X60 also exhibited approximately half of Declaration of Competing Interest
the stress result of NinjaFlex for the same structural characteristics,
making it a closer analog to the softest rehab foams for the tested The authors declare the following financial interests/personal re
sample sizes. lationships which may be considered as potential competing interests:
Maria A Woodruff reports financial support was provided by Royal
On the basis of all the tests presented, it is concluded that flexible Brisbane and Women’s Hospital. Peter Slattery reports a relationship
sheet gyroid structures present a viable alternative to existing poly with STARS, Metro North Hospital and Health Service, Qld Health that
urethane soft foams for rehab and wheelchair seat cushion applications includes: employment. David Forrestal reports a relationship with Metro
from a material response standpoint. While further tests are required to North Hospital and Health Service, Qld Health that includes: employ
verify this viability in more comprehensive fatigue and clinical envi ment. Ryan Daley reports a relationship with Metro North Hospital and
ronments, the results show that flexible gyroids can reproduce soft foam Health Service, Qld Health that includes: employment.
response and characteristics extremely closely. Based on the findings, it
is possible to determine a target gyroid SVF that can achieve the exact Acknowledgements
mechanical response of the Green EN36–90, Yellow EN40–230 and Pink
MA35–600 rehab foams tested here. These figures are presented in We acknowledge the Jamieson Trauma Institute, Royal Brisbane and
Table 4 for both NinjaFlex and X60 filaments. Women’s Hospital, Australia for seed funding, Advance Queensland,
In future, detailed fatigue testing and larger scale sample tests could Industry Research Fellowship, Australia funding for SKP (AQIRF2018)
further verify viability of the material, and optimised printing methods and NCP (AQIRF2020). We also acknowledge QUT Technical Services
could enable larger scale production. Testing in a clinical setting will for mechanical testing support.
also help identify other areas requiring optimisation, and more fully
evaluate the benefits and weaknesses of gyroid based foams for real References
wheelchair cushion and rehabilitation applications. For example, any
[1] S. Bhattacharya, R.K. Mishra, Pressure ulcers: Current understanding and newer
benefits that may be associated with the significantly more open cellular modalities of treatment, Indian J. Plast. Surg. 48 (01) (2015) 004–016.
[2] A. Goossens, R.H. Snijders, C.J. Holscher, T.G. Heerens, W.C. Holman, Shear stress
measured on beds and wheelchairs, Scand. J. Rehabil. Med. 29 (3) (1997)
131–136.
4
Note, white X60 filament was used for the fatigue testing to provide com [3] E. McInnes, A. Jammali-Blasi, S.E.M. Bell-Syer, J.C. Dumville, V. Middleton,
parable thermal imagery to the NinjaFlex. N. Cullum, Support surfaces for pressure ulcer prevention, Cochrane Database Syst.
3 Rev. 2015 (9) (2015).
Note, NinjaFlex and X60 thermal images recorded using FLIR T660 thermal [4] M.J. Rosenthal, R.M. Felton, D.L. Hileman, M. Lee, M. Friedman, J.H. Navach,
camera, while the Rehab Foam images were recorded using FLIR ONE PRO A wheelchair cushion designed to redistribute sites of sitting pressure, Arch. Phys.
camera. The difference was due to equipment availabilities during testing. Med. Rehabil. 77 (3) (1996) 278–282.
13
D.W. Holmes et al. Additive Manufacturing 50 (2022) 102555
[5] S.L. Sing, J. An, W.Y. Yeong, F.E. Wiria, Laser and electron-beam powder-bed [26] V. Beloshenko, et al., Mechanical properties of flexible tpu-based 3d printed lattice
additive manufacturing of metallic implants: A review on processes, materials and structures: role of lattice cut direction and architecture, Polymers 13 (17) (2021)
designs, J. Orthop. Res. 34 (3) (2016) 369–385. 2986.
[6] X. Wang, et al., Biomaterials topological design and additive manufacturing of [27] Diabase Engineering, Flexion X60 Ultra-Flexible Filament Material Information,
porous metals for bone scaffolds and orthopaedic implants: a review, Biomaterials 2019. 〈https://flexionextruder.com/shop/x60-ultra-flexible-filament-black/〉.
83 (2016) 127–141. [28] O. Al-Ketan, R.K. Abu Al-Rub,, Multifunctional mechanical metamaterials based on
[7] R.K. Chen, Y. Jin, J. Wensman, A. Shih, Additive manufacturing of custom orthoses triply periodic minimal surface lattices, Adv. Eng. Mater. 21 (10) (2019) 1900524.
and prostheses—a review, Addit. Manuf. 12 (2016) 77–89. [29] D.W. Abueidda, R.K. Abu Al-Rub, A.S. Dalaq, D.-W. Lee, K.A. Khan, I. Jasiuk,
[8] S. Bose, D. Ke, H. Sahasrabudhe, A. Bandyopadhyay, Additive manufacturing of Effective conductivities and elastic moduli of novel foams with triply periodic
biomaterials, Prog. Mater. Sci. 93 (2018) 45–111. minimal surfaces, Mech. Mater. 95 (2016) 102–115.
[9] Y. Yang, et al., Additive manufacturing of bone scaffolds, Int. J. Bioprint. 5 (1) [30] D.W. Abueidda, M. Bakir, R.K. Abu Al-Rub, J.S. Bergström, N.A. Sobh, I. Jasiuk,
(2018) 1–25. Mechanical properties of 3D printed polymeric cellular materials with triply
[10] P.S.P. Poh, D.W. Hutmacher, M.M. Stevens, M.A. Woodruff, Fabrication and in periodic minimal surface architectures, Mater. Des. 122 (2017) 255–267.
vitro characterization of bioactive glass composite scaffolds for bone regeneration, [31] J. Plocher, A. Panesar, Effect of density and unit cell size grading on the stiffness
Biofabrication 5 (4) (2013), 045005. and energy absorption of short fibre-reinforced functionally graded lattice
[11] A. Bhargav, V. Sanjairaj, V. Rosa, L.W. Feng, J. Fuh YH, Applications of additive structures, Addit. Manuf. 33 (2020), 101171.
manufacturing in dentistry: a review, J. Biomed. Mater. Res. Part B Appl. Biomater. [32] D. Ali, S. Sen, Finite element analysis of mechanical behavior, permeability and
106 (5) (2018) 2058–2064. fluid induced wall shear stress of high porosity scaffolds with gyroid and lattice-
[12] R. Galante, C.G. Figueiredo-Pina, A.P. Serro, Additive manufacturing of ceramics based architectures, J. Mech. Behav. Biomed. Mater. 75 (2017) 262–270.
for dental applications: a review, Dent. Mater. 35 (6) (2019) 825–846. [33] O. Al-Ketan, D.W. Lee, R. Rowshan, R.K. Abu Al-Rub, Functionally graded and
[13] M. Revilla-León, M. Özcan, Additive manufacturing technologies used for multi-morphology sheet TPMS lattices: Design, manufacturing, and mechanical
processing polymers: current status and potential application in prosthetic properties, J. Mech. Behav. Biomed. Mater. 102 (2020), 103520.
dentistry, J. Prosthodont. 28 (2) (2019) 146–158. [34] D. Li, W. Liao, N. Dai, Y.M. Xie, Comparison of Mechanical Properties and Energy
[14] M. Javaid, A. Haleem, Additive manufacturing applications in medical cases: a Absorption of Sheet-Based and Strut-Based Gyroid Cellular Structures with Graded
literature based review, Alex. J. Med. 54 (4) (2018) 411–422. Densities, Materials 12 (13) (2019) 2183.
[15] M. Ramola, V. Yadav, R. Jain, On the adoption of additive manufacturing in [35] T. Maconachie, et al., The compressive behaviour of ABS gyroid lattice structures
healthcare: a literature review, J. Manuf. Technol. Manag. 30 (1) (2019) 48–69. manufactured by fused deposition modelling, Int. J. Adv. Manuf. Technol. 107
[16] W. Tao ,M.C. Leu, Design of lattice structure for additive manufacturing, in: (11–12) (2020) 4449–4467.
Proceedings of the 2016 International Symposium on Flexible Automation (ISFA), [36] I. Maskery, N.T. Aboulkhair, A.O. Aremu, C.J. Tuck, I.A. Ashcroft, Compressive
2016, 325–332. failure modes and energy absorption in additively manufactured double gyroid
[17] K. Bertoldi, V. Vitelli, J. Christensen, M. van Hecke, Flexible mechanical lattices, Addit. Manuf. 16 (2017) 24–29.
metamaterials, Nat. Rev. Mater. 2 (11) (2017) 17066. [37] Y. Wang, et al., Numerical and experimental studies on compressive behavior of
[18] S.M. Montgomery, X. Kuang, C.D. Armstrong, H.J. Qi, Recent advances in additive Gyroid lattice cylindrical shells, Mater. Des. 186 (2020), 108340.
manufacturing of active mechanical metamaterials, Curr. Opin. Solid State Mater. [38] L. Yang, et al., Investigation on the orientation dependence of elastic response in
Sci. 24 (5) (2020), 100869. Gyroid cellular structures, J. Mech. Behav. Biomed. Mater. 90 (2019) 73–85.
[19] A. Alomarah, S.H. Masood, I. Sbarski, B. Faisal, Z. Gao, D. Ruan, Compressive [39] P. Cignoni, M. Callieri, M. Corsini, M. Dellepiane, F. Ganovelli, G. Ranzuglia,
properties of 3D printed auxetic structures: experimental and numerical studies, MeshLab: An open-source mesh processing tool, in: Proceedings of the Sixth
Virtual Phys. Prototyp. 15 (1) (2020) 1–21. Eurographics Italian Chapter Conference 2008, 129–136.
[20] X. Zhu, et al., Additive manufacturing of elastomeric foam with cell unit design for [40] NinjaTek, NinjaFlex® 3D Printing Filament: Flexible Polyurethane Material for
broadening compressive stress plateau, Rapid Prototyp. J. 24 (9) (2018) FDM Printers, 2016.
1579–1585. [41] D.W. Abueidda, M. Elhebeary, C.-S. (Andrew) Shiang, S. Pang, R.K. Abu Al-Rub, I.
[21] D. Li, W. Liao, N. Dai, G. Dong, Y. Tang, Y.M. Xie, Optimal design and modeling of M. Jasiuk, Mechanical properties of 3D printed polymeric Gyroid cellular
gyroid-based functionally graded cellular structures for additive manufacturing, structures: experimental and finite element study, Mater. Des. 165 (2019), 107597.
Comput. Des. 104 (2018) 87–99. [42] G. Paul, Y.J. Lee, P. Slattery, Modelling of multilayered foams for universal seat
[22] C. Pouya, et al., Characterization of a mechanically tunable gyroid photonic crystal design, Adv. Transdiscipl. Eng. 11 (2020) 237–248.
inspired by the butterfly parides sesostris, Adv. Opt. Mater. 4 (1) (2016) 99–105. [43] S. Koumlis, L. Lamberson, Strain rate dependent compressive response of open cell
[23] Y. Jiang, Q. Wang, Highly-stretchable 3D-architected mechanical metamaterials, polyurethane foam, Exp. Mech. 59 (7) (2019) 1087–1103.
Sci. Rep. 6 (2016) 1–11. [44] E. Flam, E. Isayeva, Y. Kipervas, V. Shklyarevsky, L. Raab, Skin temperature and
[24] J. Martínez, H. Song, J. Dumas, S. Lefebvre, Orthotropic k -nearest foams for moisture management with a low air loss surface, Ostomy/wound Manag., 41, 9,
additive manufacturing, ACM Trans. Graph. 36 (4) (2017) 1–12. 2902.
[25] F. Vanneste, O. Goury, J. Martinez, S. Lefebvre, H. Delingette, C. Duriez, [45] J. Kottner, J. Black, E. Call, A. Gefen, N. Santamaria, Microclimate: a critical
Anisotropic soft robots based on 3d printed meso-structured materials: design, review in the context of pressure ulcer prevention, Clin. Biomech. 59 (2018)
modeling by homogenization and simulation, IEEE Robot. Autom. Lett. 5 (2) 62–70.
(2020) 2380–2386.
14