Engineering Structures 272 (2022) 114922

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Cross-sectional behaviour of wire arc additively manufactured

tubular beams
Cheng Huang *, Xin Meng , Leroy Gardner
Department of Civil and Environmental Engineering, Imperial College London, UK

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

Keywords: Wire arc additive manufacturing (WAAM) is a promising metal 3D printing technique in the construction in­
Digital image correlation (DIC) dustry for its ability to produce large and complex-shaped elements, with reasonable printing accuracy, time and
Irregular hollow section costs. There is currently, however, a lack of fundamental test data on the structural performance of WAAM el­
Metal 3D printing
ements. To address this, an experimental study into the cross-sectional behaviour of WAAM tubular beams has
Rectangular hollow section (RHS)
Square hollow section (SHS)
been conducted and is presented herein. A total of 14 stainless steel square, rectangular and irregular hollow
Wire arc additive manufacturing (WAAM) sections, spanning over all cross-section classes of EN 1993-1-4 and AISC 370, were tested in four-point bending.
3D laser scanning, silicone casting and Archimedes’ measurements were employed to collectively determine the
as-built geometry and local geometric imperfections of the test specimens, while digital image correlation (DIC)
was used to monitor the deformation responses of the specimens during testing. The full moment–curvature
histories and key experimental results are presented and discussed. Similar cross-sectional behaviour to that of
equivalent, conventionally manufactured sections was observed, with the more slender cross-sections showing
increased susceptibility to local buckling. However, owing to the inherent geometric variability of WAAM, the
tested 3D printed beams exhibited more variable flexural capacities between the repeat specimens than is
generally displayed by conventionally produced stainless steel sections. Finally, the test results were used to
assess the applicability of current cross-section design provisions in the European (EN 1993-1-4) and American
(AISC 370) structural design specifications, as well as the continuous strength method (CSM), to WAAM stainless
steel tubular beams.

1. Introduction
Utilising incremental layer-by-layer deposition of material to build
components, additive manufacturing (AM) has gained considerable in­
dustry and academic attention in recent decades, primarily due to its
potential for automation and its flexibility in fabricating complex-
shaped structures [1,2]. With advances in technology, AM systems can
now be applied to a variety of materials, including polymers, ceramics,
concrete and metals [3], and are being increasingly implemented across
multiple industries, such as aerospace, biomedicine and automotive [4].
In particular, the recent emergence of metal additive manufacturing is
poised to have a substantial impact on the construction industry [5–7].
The main types of metal AM set out in ISO/ASTM 52900 [8] include
sheet lamination, powder bed fusion (PBF) and directed energy depo­
sition (DED); the latter can be further classified into powder-based and
wire-based DED according to the form of the feedstock material. Among
these metal AM techniques, sheet lamination is unlikely to be able to
fabricate the geometric complexity or scale of parts required for con­
struction [9]; PBF and powder-based DED techniques have maximum
part size limitations with lengthy build times and high costs, despite the
ability to print high complexity and high quality parts [10]. In contrast,
wire-based DED techniques provide a higher building speed and
significantly expand the maximum part size that can be built, though
with compromises on dimensional accuracy and surface quality. Wire
arc additively manufacturing (WAAM), a wire-based DED technique that
uses metal wire feedstock and a welding arc to build up parts in layers, is
deemed to be well suited to structural engineering applications in terms
of build scale, speed and cost.
Although WAAM has great potential for the construction sector,
there are numerous technical challenges to be resolved [11]. Thus far, a
number of studies have been carried out to investigate the performance
of WAAM elements, aimed at addressing issues such as variability of
material properties, residual stresses, thermal distortion, porosity, detail
resolution and surface finish [12–17]. Emphasis has been placed on the

* Corresponding author.
E-mail address: cheng.huang118@imperial.ac.uk (C. Huang).

https://doi.org/10.1016/j.engstruct.2022.114922
Received 8 June 2022; Received in revised form 2 August 2022; Accepted 29 August 2022
Available online 10 October 2022
0141-0296/© 2022 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
C. Huang et al.
Fig. 1. Overall view of the MX3D bridge.
Fig. 2. As-built and machined WAAM plates and extraction directions of ten­
sile coupons.
mechanical response, metallurgy, deposition process control and opti­
mization of WAAM parts, while investigations into the structural per­
formance of WAAM elements at the cross-section, member or structure
levels are still relatively scarce [18–22].
WAAM is still at the nascent stage and its use in the construction
sector has only emerged in recent years. A landmark demonstrator of
WAAM for structural use is the MX3D bridge shown in Fig. 1, which is a
10.5 m span 3D printed stainless steel footbridge and the first of its kind
[20]. This bridge, featuring complex as-built geometry and strongly
anisotropic material properties, is beyond the scope of current structural
design specifications; hence, a comprehensive series of experiments,
comprising tensile coupon testing [23], cross-section compressive
testing [18,21] and full-scale structural testing [20] and modelling [24],
has been undertaken to verify the structural safety of the bridge and also
to understand better the performance of WAAM structures.
In this paper, four-point bending tests on 14 WAAM stainless steel
beams with square, rectangular and irregular hollow section profiles,
printed using the same feedstock material and process parameters as
used for the MX3D bridge, are presented. The present study is among the
first to investigate the buckling behaviour of WAAM cross-sections, in
which a series of novel measuring methods, including 3D laser scanning,
silicone casting and digital image correlation (DIC), were used to
2
Engineering Structures 272 (2022) 114922
Fig. 3. Typical measured stress–strain curves of as-built and machined (θ =
90◦ ) WAAM stainless steel.
capture the as-built geometries and deformation responses of the test
specimens. The aim of this research is to address the lack of cross-section
buckling test data on WAAM beams, as well as to support the safety
verification of the MX3D bridge. Key mechanical properties determined
from previous tensile coupon tests [23] on the examined WAAM mate­
rial are first summarised. The manufacture, geometric measurements
and four-point bending tests of the WAAM beams are then described,
followed by a discussion of the obtained test results. Lastly, the experi­
mental results are used to assess the existing slenderness limits and the
cross-section resistance predictions from EN1993-1–4 [25], AISC 370
[26] and the continuous strength method (CSM) [27].
2. Material tests
To determine the stress–strain response of the WAAM material,
tensile tests on coupons extracted from stainless steel plates printed in
the same manner as the studied beams were conducted [23]. Coupons
extracted from both as-built and machined WAAM plates at 0◦ , 45◦ and
90◦ relative to the deposition direction (see Fig. 2) were tested, to
investigate the material anisotropy and the influence of the geometric
variability on the resulting mechanical properties. The detailed testing
procedure has been described by Kyvelou et al. [23], while the key test
results are summarised in this section. Typical measured stress–strain
curves for both as-built and machined coupons of two nominal thick­
nesses tnom are plotted in Fig. 3, and the obtained key material properties
are presented in Table 1, including the Young’s modulus E, 0.2% and
1.0% proof stresses, σ 0.2 and σ 1.0, ultimate strength σu and ultimate
strain εu, as well as the strain hardening exponents n, m1.0 and mu of the
compound Ramberg-Osgood material model [28–32]. Note that only the
material properties of the θ = 90◦ coupons (as defined in Fig. 2), which
correspond to loading in the longitudinal direction of the beams, are
presented herein; the mechanical properties obtained from the material
tests are used in Sections 5 and 6 for the analysis of the test results of the
studied WAAM beams.
3. Manufacture of test specimens
The test specimens were wire arc additively manufactured by MX3D
[33] using their proprietary WAAM system, featuring a six-axis ABB
robot and a metal inert gas (MIG) welder, as shown in Fig. 4(a). 3D CAD
models of the specimens were created and sliced into a series of layers to
define the deposition paths. Following the planned build paths, the
robot deposited successive layers of material to build up the tubular
specimens. In this fashion, tubular sections of 3.5 mm and 8.0 mm
nominal thickness tnom were printed, using Grade 308LSi austenitic
C. Huang et al. Engineering Structures 272 (2022) 114922

Table 1
Average measured material properties of as-built and machined θ = 90◦ coupons.
Coupon tnom (mm) E (MPa) σ0.2 (MPa) σ1.0 (MPa) σu (MPa) εu n m1.0 mu

As-built 3.5 90,200 261 319 448 0.119 6.5 2.5 2.6
8.0 109,100 271 326 423 0.103 5.5 2.6 2.5
Machined 3.5/8.0 139,600 338 381 554 0.297 6.8 2.3 2.7

ABB robot

MIG power source

with wire feeder
CAD model Shielding gas

MIG torch

Build
direction

Computer
Base plate
(a) Robot Controller

(b)
Fig. 4. (a) Illustration of the WAAM process; (b) WAAM specimens of square, rectangular and irregular hollow sections.

stainless steel feedstock wire with diameters of 1.0 mm and 1.2 mm,
Table 2
respectively. The chemical and mechanical properties of the feedstock
Process parameters for tested WAAM beams.
material have been reported in [23], while a summary of the key WAAM
Deposition parameters Details process parameters, including the travel speed, wire feed speed, arc
tnom = 3.5 mm tnom = 8.0 mm current, arc voltage, and shielding gas composition and flow rate, is
Wire diameter (mm) 1.0 1.2 provided in Table 2.
Travel speed (mm/s) 15–30 13 Following printing, each WAAM tube was detached from the sub­
Wire feed speed (m/min) 4–8 5.7 strate plate using a plasma arc cutter, and was cut using a rotary
Welding voltage (V) 18–21 18–21 hacksaw to a length of about five times the maximum cross-sectional
Welding current (A) 100–140 100–140
dimension, such that the obtained specimens were sufficiently long to
Shielding gas 98% Ar + 2% CO2 98% Ar + 2% CO2
Gas flow rate (L/min) 10–20 10–20 include representative residual stress and local imperfection patterns
Dwell time (s) 30 30 and to allow the development of local buckling without any significant
boundary effects [34]. The specimens, which included square,

3
C. Huang et al. Engineering Structures 272 (2022) 114922

Fig. 5. Employed methods for geometric measurements of WAAM beams: (a) Archimedes’ measurements; (b) 3D laser scanning.

Fig. 6. Production procedure of silicone casts for internal geometry acquisition of WAAM tubular beams: (a) preparing silicone mixture; (b) degassing; (c) casting
silicone and curing; (d) extracting a silicone cast; (e) a silicone cast and its parent specimen.

rectangular and irregular hollow sections, are shown after fabrication in
Fig. 4(b). The adopted specimen labelling system begins with the cross-
section shape (‘S’, ‘R’ and ‘IR’ for square, rectangular and irregular
hollow sections, respectively), followed by the nominal cross-sectional
dimensions (height × breadth of compression flange × wall thickness)
and member length in mm; the number at the end of a specimen label
denotes the test ID, omitted if there was no repeat test.
4. Geometric measurements
The geometry and local imperfections of the WAAM specimens were
examined before testing. 3D laser scanning, in conjunction with silicone
casting, was used to capture both the external and internal surface
profiles of the tubular specimens, with the scan data verified against
volume measurements based on Archimedes’ principle. In parallel with
geometric characterization, the local imperfections of the specimens
were also determined from the 3D scan data, as described in this section.
4.1. Geometric properties
Archimedes’ measurements, based on the principle that a submerged
object is subject to a buoyancy force equal to the weight of the displaced
water, were taken to determine the volume of the specimens; a typical
specimen being measured underwater is shown in Fig. 5(a). Upon
determination of the volume of each specimen, the average cross-
sectional area AArch was derived by dividing the volume by the mem­
ber length L, measured using callipers.
In order to comprehensively capture the geometric details of the
WAAM specimens, 3D laser scanning was employed. As shown in Fig. 5
(b), scanning was performed using a Faro ScanARM, which is capable of
capturing up to 600,000 points per second, with a reported resolution of
0.075 mm and an accuracy of 0.1 mm. The outer surface of each spec­
imen was fully scanned, while direct scanning of the complete inner
surface was precluded by the size of scanning arm head. A scannable
replica of the interior geometry was therefore produced by means of
silicone casting. The silicone casts were made of SUPERSIL 25, a two-
component (base and catalyst) silicone elastomer, and were produced
following the procedure illustrated in Fig. 6, including (a) preparing the
silicone mixture, (b) degassing in a vacuum chamber, (c) casting the
silicone within the tubular specimens and curing for at least 24 h, and
finally, (d) extracting the silicone casts from within the tubes. Note that
a central insert was placed within each specimen, as shown in Fig. 6(c),

4
C. Huang et al.
Fig. 7. Typical specimen, its merged laser scan model and cross-sectional contours with grid points.
Fig. 8. Results of alignment between outer and inner scans of Specimen
IR130×100×3.5-500.
and was extracted first to facilitate easier removal of the silicone casts
without surface damage. A typical extracted silicone cast, as well as its
parent tubular specimen, is presented in Fig. 6(e).
Upon extraction from within their parent specimens, the silicone
casts were laser scanned; the resulting scans were merged with the outer
scans of the tubes using Geomagic Wrap [35] to create complete 3D scan
models of the test specimens. The models were then converted into
polygon objects and imported into Rhino 3D [36], whereby a series of
operations was performed for geometric analysis, including aligning the
models with the global coordinate system, contouring the models along
the length direction and generating grid points from the contours. A
schematic illustration of a typical specimen processed in Rhino 3D is
shown in Fig. 7, where only a limited number of cross-sectional contours
and grid points are presented for illustrative purposes. For each spec­
imen, the centroid positions of both the outer and inner contours were
derived from the scan data and plotted against the member length to
evaluate the alignment of the outer and inner scans along the y and z
5
Engineering Structures 272 (2022) 114922
Fig. 9. Typical distributions of measured wall thicknesses for SHS and
RHS specimens.
directions (defined in Fig. 7). Good alignment was achieved for all
specimens, and typical results are shown in Fig. 8, where the outer and
inner contour centroids generally follow the same trend along the
specimen length, and the linear regression lines of the centroid de­
viations are essentially parallel to the length direction. After confirma­
tion of alignment, the contours were used to calculate the cross-sectional
areas, and grid points along the contours were generated to determine
the wall thickness t and other cross-sectional dimensions. Typical dis­
tributions of the wall thickness measurements for specimens of the two
nominal thickness values (i.e. tnom = 3.5 mm and 8.0 mm), together with
their respective normal distribution fits, are depicted in Fig. 9. The
histograms of the measured thicknesses clearly indicate greater varia­
tion than seen with conventionally formed hollow sections, while,
compared to the normal distributions, the measured thickness distri­
butions exhibit a skewness close to 0 and a kurtosis greater than 3.
A summary of the cross-sectional dimensions derived from the laser
scans, along with the member length L, is presented in Table 3, where A,
Amax, Amin are the average, maximum and minimum cross-sectional
areas respectively, t and tsd are the mean and standard deviation of
the thickness respectively, and H, B, R, r are the height, breadth, outer
corner radius and inner corner radius of the square and rectangular
hollow sections (SHS and RHS) respectively. For the four irregular
hollow sections, the outer cross-sectional dimensions are given in
Fig. 10. The average cross-sectional areas AArch, as obtained based on
Archimedes’ principle, are also reported in Table 3; the obtained AArch
and A are shown to be consistently within 3%, confirming the reliability
C. Huang et al. Engineering Structures 272 (2022) 114922

Table 3
Average measured geometric properties of WAAM tubular beams.
Specimen H B R r t tsd A AArch L w0
(mm) (mm) (mm) (mm) (mm) (mm) (mm2) (mm2) (mm) (mm)

S150×150×3.5-750 149.7 149.4 8.6 4.7 4.13 0.53 2360.0 2385.6 749.8 2.39
S120×120×3.5-600 119.9 119.7 8.5 4.4 4.00 0.50 1795.0 1846.3 600.3 1.13
S110×110×3.5-550 110.0 109.8 8.5 4.3 3.84 0.46 1638.2 1616.9 549.7 0.73
S105×105×3.5-525 104.8 104.8 8.3 4.2 4.00 0.49 1554.3 1600.3 525.3 1.22
S100×100×3.5-500-1 99.8 99.6 8.1 4.3 4.01 0.44 1493.1 1527.5 500.0 0.58
S100×100×3.5-500-2 99.6 99.8 8.5 4.8 3.95 0.47 1476.0 1493.1 500.1 1.95
S70×70×3.5-340-1 70.2 70.0 8.3 4.4 4.00 0.54 1032.4 1047.7 340.2 0.60
S70×70×3.5-340-2 70.0 70.0 8.3 4.4 3.98 0.49 1008.8 1037.8 340.0 0.97
R80×120×3.5-590 80.2 120.7 9.2 5.4 4.05 0.46 1501.7 1535.2 590.4 1.07
R75×110×8.0-550 75.9 115.8 9.1 5.2 6.12 0.53 2113.6 2131.4 550.3 0.84
IR55×95×3.5-500 See Fig. 10 for cross-sectional shapes and dimensions 3.66 0.68 1245.0 1245.7 500.0 1.55
IR130×100×3.5-500 See Fig. 10 for cross-sectional shapes and dimensions 3.60 0.66 1577.0 1578.6 500.2 1.59
IR45×75×3.5-500 See Fig. 10 for cross-sectional shapes and dimensions 3.71 0.67 906.0 923.3 499.6 1.43
IR90×100×3.5-500 See Fig. 10 for cross-sectional shapes and dimensions 3.80 0.63 1695.0 1670.3 500.0 1.34

Fig. 10. Measured outer cross-sectional dimensions (in mm) of irregular hollow section specimens.

Grid points along the centrelines used for

local imperfection measurements
Mid-point of face 1
Face 1

Mid-point Mid-point
of face 4 of face 2
Face 4
Mid-point
of face 3

Face 3

Fig. 11. Determination of local geometric imperfections [21].

of the employed measurement methods.
4.2. Local geometric imperfections
The determination of the geometric imperfections is important for
future numerical analyses of the WAAM specimens. Only local geometric
imperfections were considered herein, since the closed nature of the
tubular sections and their relatively short beam lengths precluded
lateral torsional buckling. As illustrated in Fig. 11, the local imperfec­
tions were determined using the grid points along the centreline of the
specimen outer faces, in line with the approach proposed by Kyvelou
et al. [21]. For each face, the local imperfection was defined as the de­
viation of each grid point from a straight line fitted to all the points along
the member length using least squares regression, and the maximum
deviation was taken as the local imperfection amplitude. This definition
is deemed appropriate because the derived deviation corresponds to the
out-of-flatness along the longitudinal axis of the structural elements; this
out-of-flatness triggers and amplifies local plate buckling and thus
governs the ultimate cross-section strength. However, the presence of
some particularly prominent surface undulations and weld beads
introduced by WAAM could lead to unrealistically large imperfection
amplitude measurements. To filter these unwanted features, while
retaining the underlying imperfection profiles, a 10 mm moving average
was applied to smooth the obtained imperfection distributions. Typical
local imperfection distributions measured before and after smoothing
are presented in Fig. 12. The maximum deviation of each smoothed
curve from the fitted reference line was then taken as the local imper­
fection amplitude for the corresponding face.

6
C. Huang et al.
Fig. 12. Local geometric imperfection distributions of Specimen S70×70×3.5-340-2: (a) before smoothing; (b) after smoothing.
Empirical CDF
Fig. 13. Histogram and CDF of local geometric imperfection amplitudes.
The local imperfection amplitude of the compression flange of each
beam specimen w0 is reported in Table 3, while a histogram of 56
imperfection amplitude measurements (one measurement per section
face for 14 specimens) is presented in Fig. 13, together with the cumu­
lative distribution function (CDF) curve. No clear link between the local
imperfection amplitudes and the cross-sectional dimensions of the
WAAM specimens was found in the limited dataset. The measured
imperfection amplitudes generally lie between 0.5 mm and 2.5 mm,
which are higher than those typically seen in conventionally produced
sections [37,38] and similar to the dimensional accuracy of about ±1.0
7
Engineering Structures 272 (2022) 114922
Fig. 14. Comparison of an SHS specimen with its corresponding part of the
MX3D bridge [20].
to ±2.0 mm typically quoted for WAAM elements [39–41].
5. Four-point bending tests
A total of 14 WAAM beam specimens, comprising eight SHS, two
RHS and four irregular hollow sections, with a wide range of local
slendernesses, were tested to investigate their cross-sectional behaviour
under uniform bending. A subset of the WAAM beam specimens were
dimensioned based on the cross-sectional profiles and width-to-
thickness ratios of the key components of the MX3D bridge, in order
to assess the flexural behaviour of the bridge elements. A comparison of
C. Huang et al. Engineering Structures 272 (2022) 114922

(a) DIC cameras Load actuator

Special bearing
Spreader beam

Fixed roller

Inclinometers WAAM
beam Free roller
Rubber sheet High strength
steel SHS

Inclinometers Roller support

String
DIC cameras potentiometers

(b) Load
Rubber sheet High strength steel SHS
WAAM beam

String Inclinometer
potentiometer Wooden block Support

140 500 60 L 700

Fig. 15. Four-point bending test set-up: (a) photograph; (b) schematic diagram (dimensions in mm).

a typical WAAM SHS specimen with its corresponding part on the MX3D
bridge is presented in Fig. 14.
5.1. Test setup
Prior to testing, each WAAM specimen was welded to two steel end
plates and then bolted to a 700 mm long high-strength steel SHS at each
end to extend the member length. All extended specimens were tested in
the four-point bending configuration shown in Fig. 15, with the central
WAAM tube located within the constant moment region. The specimens
were simply supported on a pair of steel rollers, and symmetrically
loaded through two further steel rollers, with one fixed in position to
restrain the longitudinal movement of the specimens. The shear span
length was 500 mm, and the overhanging length at each end was 140
mm, as dimensioned in Fig. 15(b). The vertical load was applied using an
Instron 2000 kN hydraulic actuator, and transferred to the specimens
through a special bearing, a spreader beam and the top two rollers; the
special bearing allowed rotations about the axis of bending and
eliminated possible gaps between the loading points and specimen,
thereby enabling symmetric vertical loading. Rubber sheets were uti­
lised to spread the loads onto the high-strength steel SHS, inside which
wooden blocks were also inserted to prevent web crippling at the
loading points and supports. A range of instrumentation was employed
during the tests, including four load cells (two for each roller) to mea­
sure the applied vertical loads, four inclinometers to measure the rota­
tions at the loading points and supports, and three string potentiometers
to measure the vertical displacements at the mid-span and both ends of
the WAAM specimens.
A four-camera LaVision Digital Image Correlation (DIC) system was
also employed to monitor the evolution of deformations and strains of
the WAAM specimens throughout testing. The DIC speckle patterns were
produced on the top surface (i.e. the compression flange) and a side
surface of each beam specimen; the surfaces were sprayed with white
paint, over which random patterns of approximately uniformly sized
black speckles were subsequently applied. Matt white and black paint
was adopted to prevent data loss from reflection during image capture.

8
C. Huang et al. Engineering Structures 272 (2022) 114922

10 mm 10 mm 10 mm
(a) (b) (c)
Fig. 16. DIC speckle patterns for Specimens (a) S70×70×3.5-340-1; (b) R75×110×8.0-550 and (c) S150×150×3.5-750.

Table 4
Summary of key results for WAAM tubular beams.
Specimen c/tε Mu (kNm) Mel (kNm) Mpl (kNm) κu (×10-4 mm− 1) κpl (×10-4 mm− 1) κpl,u (×10-4 mm− 1) Mu/Mel Mu/Mpl R
a
S150×150×3.5-750 51.5 28.0 28.7 33.4 0.84 0.45 – 0.97 0.84 0.00
S120×120×3.5-600 41.3 21.1 17.4 20.3 1.61 0.56 2.01 1.21 1.04 2.56
S110×110×3.5-550 38.9 18.3 13.9 16.3 1.74 0.62 3.13 1.32 1.13 4.07
S105×105×3.5-525 35.4 15.6 13.0 15.3 1.92 0.65 2.25 1.20 1.02 2.47
S100×100×3.5-500-1 33.4 15.3 11.8 13.9 2.87 0.68 5.06 1.30 1.10 6.41
S100×100×3.5-500-2 33.7 14.0 11.6 13.6 1.94 0.68 3.24 1.21 1.03 3.73
S70×70×3.5-340-1b 21.5 8.4 5.4 6.5 17.87 0.99 21.07 1.56 1.30 >20.25
S70×70×3.5-340-2b 21.6 7.6 5.3 6.4 7.19 0.99 16.95 1.42 1.18 >16.07
R80×120×3.5-590 40.6 11.9 10.4 12.0 2.04 0.83 2.04 1.15 1.00 0.00
R75×110×8.0-550 23.8 16.9 13.7 16.3 4.43 0.61 6.88 1.23 1.04 10.36
IR55×95×3.5-500b 17.1 12.4 7.8 10.1 5.16 0.72 7.86 1.59 1.22 >9.97
IR130×100×3.5-500a 37.5 17.1 13.4 17.3 1.09 0.53 – 1.28 0.99 0.00
IR45×75×3.5-500 25.6 4.2 3.3 3.9 3.95 1.47 6.80 1.25 1.07 3.63
IR90×100×3.5-500 35.5 16.0 11.8 13.6 1.95 0.69 4.37 1.36 1.18 5.32
a
Test moment failed to reach Mpl (Mu < Mpl).
b
Test terminated before moment fell below Mpl on descending curve.

inclinometers, were all recorded at 0.5 s intervals using the data

acquisition system DATASCAN. The DIC images were captured at a
frequency of 1 Hz, as adopted in [16,20,21], to achieve both accuracy
and computational efficiency, and were processed using Davis 10 [43].

5.2. Test results

Fig. 17. Definition of rotation capacity R and parameters for calculation of R.
Special attention was paid to the determination of the DIC speckle size,
given the fact that larger speckles cause poor spatial resolution while
speckles that are too small can cause aliasing. The average speckle size
for each specimen was calculated to correspond to 4 pixels in the DIC
images, following the procedure recommended in [22,42]. Typical
speckle patterns for the WAAM specimens are presented in Fig. 16,
where the average size of the speckles is shown to increase with the
specimen length. The bending tests were performed under load control;
the loading rate was determined to achieve a strain rate of approxi­
mately 0.1% min− 1 at the extreme fibre of each WAAM cross-section.
Test outputs, including the applied loads from the load cells, the ma­
chine displacement and the readings from the string potentiometers and
The key results from the four-point bending tests are summarised in
Table 4, including the measured ultimate bending moment Mu, the
elastic and plastic moment capacities Mel and Mpl, taken as the product
of the 0.2% proof stress σ0.2 and the elastic and plastic section modulus
respectively, the normalised ultimate moment (Mu/Mel and Mu/Mpl), the
curvature at the ultimate moment κu, the elastic curvature at the plastic
moment κpl, the curvature when the applied moment falls back to the
plastic moment κpl,u, and the cross-sectional rotation capacity R, as
defined in Fig. 17. The local slenderness of the compression flange,
which is the dominant element in determining the local stability of the
studied cross-sections, c/tε, where c is the compressed flat width, t is the
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
plate thickness and ε = (235 / σ 0.2 )/(E / 210000), is also reported in
Table 4 for all beam specimens. Note that Mel, Mpl and ε were calculated
based on the mechanical properties of the as-built material, as given in
Table 1.
The normalised moment–curvature curves M/Mpl–κ/κpl for all tests
are shown in Fig. 18. The moment in the flexural span M was determined
on the basis of the shear span length and the applied loads at the two
loading points, with the latter varying within 5% of each other
throughout testing, according to the readings from the load cells. It
should be noted that, as the specimens deformed during testing, M
consisted of the bending moments induced by not only the dominant
vertical but also the small horizontal components of the applied loads;

9
C. Huang et al. Engineering Structures 272 (2022) 114922

Fig. 18. Normalised moment–curvature curves of tested beams: (a, b) SHS; (c) RHS; (d) irregular hollow sections.

this was considered in the calculation of M, following the recommen­
dations of Meng and Gardner [44]. The average curvature of each
specimen κ was calculated using both Eqs (1) and (2), based on the
measured vertical displacements and the end rotations respectively:
8(dL − dM )
κ=
4(dL − dM )2 + L2
where dL is the average vertical displacement measured from the two
string potentiometers at the ends of the WAAM beams, dM is the vertical
displacement at mid-span and L is the length of the WAAM beams.
θ1 + θ2
κ=
L vertical deflections, and hence, the corresponding rotation capacities R
where θ1 and θ2 are the rotations measured from the inclinometers at the
two ends of the WAAM beams. A close match between the curvatures
determined from the two methods was achieved up to the onset of local
buckling, while upon the development of local instability, Eq. (2) was
shown to yield more stable and reliable results; this calculated curvature
was therefore used for the moment–curvature curves plotted in Fig. 18.
It can be seen from Fig. 18 that most of the test specimens were
(1) capable of reaching their plastic moment capacities Mpl, except for
Specimens S150×150×3.5-750 and IR130×100×3.5-500, due mainly to
their relatively high local slendernesses and imperfection amplitudes;
the rotation capacities R of these two specimens are therefore equal to
zero, as reported in Table 4. Three bending tests (i.e. S70×70×3.5-340-
1, S70×70×3.5-340-2 and IR55×95×3.5-500) were terminated before
their moment–curvature curves fell back below Mpl due to excessive
(2)
were determined based on the maximum recorded curvatures, as high­
lighted in Table 4. Fig. 18(b) shows the moment–curvature curves of two
pairs of repeat tests, where more variable capacities between the repeat

Fig. 19. (a) Correlation of normalised moment capacities with local slenderness; (b) variation in relative response of specimens with geometric imperfec­
tion amplitude.

10
C. Huang et al.
Fig. 20. Failure modes of tested WAAM beams.
WAAM specimens are observed than those generally exhibited by
conventionally formed stainless steel tubular sections [45,46].
The normalised moment capacities M/Mpl of the test specimens are
plotted against the local slenderness parameter c/tε in Fig. 19(a), where
a linear regression trend line, denoted by ρlinear, is also presented,
revealing the general trend of decreasing moment capacity with
increasing local slenderness. The normalised moment capacity M/Mpl of
each specimen is further normalised by ρlinear and plotted against the
11
Engineering Structures 272 (2022) 114922
local imperfection amplitude w0 in Fig. 19(b). It can be seen that the
relative flexural performance of the WAAM beams degrades with
increasing values of w0, and that geometric imperfections are the key
reason for variable structural responses between the repeat specimens.
Fig. 20 shows the tested WAAM beam specimens, all of which dis­
played local buckling of the compression flange and the upper portion of
the webs. The location of failure (i.e. local buckling) in the specimens
related to the geometric imperfection distributions and thickness
C. Huang et al.
Fig. 21. Correlation between geometric variability and failure locations, shown for Specimens (a) S70×70×3.5-340-2 and (b) R75×110×8.0-550.
variations. This can be seen in Fig. 21, where the distributions of the
local geometric imperfections w and average wall thickness of the
compression flange tc along the member length for two typical speci­
mens (covering two nominal thicknesses) are shown, along with the
longitudinal strain distribution over the side face of each specimen from
the DIC results to highlight the failure location. It can be observed that
local buckling is triggered in regions of both high geometric imperfec­
tions and low wall thicknesses. The longitudinal strain and out-of-plane
displacement fields derived from the DIC data for Specimen
IR90×100×3.5-500 are plotted alongside the corresponding
moment–curvature curve in Fig. 22, where the evolution of longitudinal
strains and the development of local buckling can be clearly seen.
6. Comparisons with existing design methods
In this section, the test results are used to assess the applicability of
existing cross-section design methods to WAAM stainless steel tubular
beams. The suitability of the local slenderness limits specified in the
current European [25] and American [26] standards is first evaluated
and discussed. The test results are subsequently compared against the
cross-section resistance predictions from the design functions of EN
1993-1-4 [25], AISC 370 [26] and the continuous strength method
(CSM) to assess their applicability to the design of WAAM sections in
bending. Note that all comparisons have been made based on unfactored
strength predictions, i.e. with all partial factors set equal to unity.
6.1. Assessment of local slenderness limits
The concept of cross-section classification is employed in EN 1993-
1–4 [25] and AISC 370 [26] for the treatment of local buckling [47,48].
Cross-sections are classified into Class 1–4 according to Eurocode 3
(EC3), or into compact, noncompact and slender categories in AISC 370
12
Engineering Structures 272 (2022) 114922
[26]. The slenderness limits for internal plate elements in compression
are assessed using the test data on the WAAM stainless steel cross-
sections generated herein.
Assessment of the Class 1 slenderness limit was carried out on the
basis of a rotation capacity requirement of R = 3, which is widely
adopted for the plastic design of steel structures [49]. The rotation ca­
pacities R of the tested WAAM sections are plotted against the local
slenderness parameter c/tε (calculated based on the as-built material
properties) in Fig. 23, where test data on conventionally produced
stainless steel sections collected from the literature [37,46,50–52] and
the AISC compact slenderness limit are also depicted for comparison
purposes. The collected rotation capacity data points are rather scat­
tered, due mainly to the varying stainless steel grades, levels of geo­
metric imperfections and moment gradients, as well as to the interaction
of constituent plate elements [53]. However, the WAAM beam data can
be seen to follow a similar trend to the conventional beam data and the
current EC3 Class 1 slenderness limit of c/tε = 33 is on the safe side.
The experimental ultimate moment capacities Mu obtained from this
study are normalised by the plastic moment capacities Mpl and plotted
against the c/tε ratio of the compression flange in Fig. 24, where the
collected test data on conventionally manufactured stainless steel beams
[37,46,50,52,54,55], as well as the EC3 Class 2 slenderness limit and
AISC compact slenderness limit, are also depicted. Both the as-built and
machined material properties, as given in Table 1, are used for the
normalization and analysis of the test data on the WAAM sections. It is
shown that the data points associated with the machined material
properties consistently lie below the collated test dataset of conven­
tionally produced sections, in line with the findings of Kyvelou et al.
[21]. In contrast, use of the as-built material properties, which incor­
porate the effect of the geometric undulations, results in the WAAM data
points generally lying within the range of the test data on the conven­
tionally formed sections. With the adoption of the as-built material
C. Huang et al.
Fig. 22. Typical DIC results for Specimen IR90×100×3.5-500: (a) top face and (b) side face.
properties, the EC3 Class 2 and AISC compact slenderness limit of 35
appears to be suitable for WAAM plate elements in compression.
To appraise the suitability of the EC3 Class 3 limit and AISC non­
compact slenderness limit for internal compression elements, the
bending test results from this study, together with the compression test
results on WAAM sections from a previous study [21] and the collected
test data on conventionally produced stainless steel beams, are pre­
sented in Fig. 25, where the ultimate moment capacity normalised by
the elastic moment capacity Mu/Mel (for beams) and the ultimate load
capacity normalised by the yield load Nu/Aσ0.2 (for stub columns) are
13
Engineering Structures 272 (2022) 114922
plotted against the local slenderness parameter c/tε. From Fig. 25, it can
be observed that the EC3 Class 3 and AISC noncompact slenderness limit
of 37, and the CSM yield slenderness limit of 38.6 [56], appear to be
generally applicable to WAAM sections in compression and bending.
Overall, Figs. 23-25 demonstrate a clear trend between an increasing
local slenderness and a decreasing rotation capacity and normalised
moment capacity. The current codified slenderness limits are shown to
be generally applicable to the examined WAAM tubular beams, provided
that the influence of the surface undulations associated with as-built
WAAM elements are accounted for through the use of material
C. Huang et al. Engineering Structures 272 (2022) 114922

properties derived from coupons tested in the as-built condition. Further

test data are nonetheless considered necessary to confirm the findings
and provide a stronger basis for a reliability assessment.

6.2. Assessment of resistance predictions

In this subsection, the cross-section resistances of the tested WAAM

beams are compared against the cross-section resistances predicted
using the EC3, AISC and CSM design functions. Based on the afore­
mentioned cross-sectional classifications, the unfactored design re­
sistances for bending about one principal axis of a cross-section (MEC3
and MAISC) are given by Eqs (3)-(5) in EN 1993-1-4 [25]:
MEC3 = Mpl = Wpl σ0.2 for Class 1 and 2 cross − sections (3)

MEC3 = Mel = Wel σ0.2 for Class 3 cross − sections (4)

MEC3 = Weff σ 0.2 for Class 4 cross − sections (5)

Fig. 23. Assessment of EC3 Class 1 and AISC compact slenderness limits for
internal compression elements. where Wel, Wpl and Weff are the elastic, plastic and effective section
moduli, respectively, and Eqs (6)-(8) in AISC 370 [26], where the no­
tation has been harmonised between the specifications for simplicity.
MAISC = Mpl = Wpl σ 0.2 for compact cross − sections (6)
( )
λ − λpf
MAISC = Mpl − (Mpl − Mel ) for noncompact cross − sections
λrf − λpf
(7)

MAISC = Weff σ 0.2 for slender cross − sections (8)

where λ = c/tε is the local slenderness, and λpl and λrl are the slenderness
limits for compact and noncompact flanges, equal to 35 and 37,
respectively. Note that Weff was calculated using the effective width
method given in EN 1993-1-4 [25] and AISC 370 [26], which provide
slightly different formulae for the local buckling reduction factors, ρEC3
and ρAISC, as given, for internal elements, by Eqs (9) and (10),
respectively:
0.772 0.079
ρEC3 = − 2
⩽1 (9)
λP λP
Fig. 24. Assessment of EC3 Class 2 and AISC compact slenderness limit for
internal compression elements. ( )
1 0.10
ρAISC = 0.772 − 2
⩽1 (10)
λp λp

where λp is the plate slenderness of the compression flange, calculated

using Eq. (11) and reported for all specimens in Table 5. Note that the
beam webs were fully effective in all cases.

b/t
λp = √̅̅̅̅̅ (11)
28.4ε kσ

Fig. 25. Assessment of EC3 Class 3, AISC noncompact and CSM yield slen­
derness limits for internal compression elements.
where b is the flat width of the compression flange, taken as B-3t for the
EN 1993-1-5 calculations [57] and taken as B-2R for the AISC 370 cal­
culations [26], and kσ is the plate buckling factor, taken as 4 for flanges
in uniform compression across their widths according to EN 1993-1-5
[57].
The continuous strength method (CSM) [56,58–60] has also been
employed to predict the bending moment resistances of the WAAM
tubular beams. The CSM is a deformation-based design approach
featuring two key components: (1) a base curve to provide a non-
dimensional measure of cross-section deformation (strain) capacity
and (2) a material model, which allows for strain hardening and, in
conjunction with the strain measure, can be used to determine the cross-
section bending resistance. The base curve adopted herein for SHS/RHS
is given by Eq. (12), comprising two parts: for non-slender (λp,cs ≤ 0.68)
and slender (0.68<λp,cs ≤ 1.6) cross-sections respectively:

14
C. Huang et al. Engineering Structures 272 (2022) 114922

Table 5
Comparisons of test results with predicted moment resistances for tubular beams.
Specimen EC3 Class AISC Class c/tε λ p,EC3 λ p,AISC λ p,cs Mu/MEC3 Mu/MAISC Mu/Mcsm

S150×150×3.5-750 4 Slender 51.5 0.94 0.91 0.84 1.17 1.16 1.10

S120×120×3.5-600 4 Slender 41.3 0.76 0.73 0.68 1.35 1.33 1.22
S110×110×3.5-550 4 Slender 38.9 0.72 0.68 0.65 1.37 1.34 1.12
S105×105×3.5-525 3 Noncompact 35.4 0.66 0.62 0.59 1.20 1.05 1.06
S100×100×3.5-500-1 2 Compact 33.4 0.62 0.59 0.56 1.10 1.10 1.12
S100×100×3.5-500-2 2 Compact 33.7 0.63 0.59 0.57 1.03 1.03 1.05
S70×70×3.5-340-1 1 Compact 21.5 0.41 0.38 0.39 1.30 1.30 1.11
S70×70×3.5-340-2 1 Compact 21.6 0.41 0.38 0.39 1.18 1.18 1.01
R80×120×3.5-590 4 Slender 40.6 0.76 0.71 0.66 1.28 1.25 1.12
R75×110×8.0-550 1 Compact 23.8 0.45 0.42 0.43 1.04 1.04 0.95
IR55×95×3.5-500 1 Compact 17.1 0.34 0.30 0.35 1.18 1.18 0.98
IR130×100×3.5-500 4 Slender 37.5 0.70 0.66 0.60 1.33 1.28 1.08
IR45×75×3.5-500 1 Compact 25.6 0.49 0.45 0.44 1.07 1.07 0.98
IR90×100×3.5-500 3 Noncompact 35.5 0.66 0.63 0.61 1.36 1.22 1.24
Mean 1.21 1.18 1.08
COV 0.10 0.09 0.08

⎧ ( )
⎪
⎪ 0.25 0.1εu
⎪ 3.6 ⩽min 15,
⎪ for λp,cs ⩽0.68
⎪
⎨ λp,cs εy
εcsm
= ( ) (12)
εy ⎪
⎪ 0.222 1
⎪
⎪ 1 − for 0.68 < λp,cs ⩽1.6
⎪
⎩ 1.05 1.05
λp,cs λp,cs

where εcsm is the limiting strain (i.e. the deformation capacity) of the
cross-section, εy and εu are the material yield and ultimate strains,
respectively, and λp,cs is the slenderness of the full cross-section [53],
defined by Eq. (13):
√̅̅̅̅̅̅̅
Mel
λp,cs = (13)
Mcr

where Mcr is the elastic local buckling moment, obtained using numer­
ical software ABAQUS [61]. The material model for austenitic stainless
steel sections is given by Eqs (14) and (15):
Esh = (σu − σ 0.2 )/(0.16εu − εy ) (14)

εu = (1 − σ 0.2 /σu ) (15) Fig. 26. Comparison of test results with EN 1993-1-4, AISC 370 and CSM
strength predictions.
where Esh is the strain hardening modulus. Combining the CSM base
curve and material model, the cross-section moment resistance Mcsm is
determined from Eq. (16):

⎧ [ ( ) ( ) ( )2 ]
⎪
⎪ Esh Wel εcsm Wel / εcsm
⎪
⎨ W pl σ 0.2 1 + − 1 − 1 − for λp,cs ⩽0.68
E Wpl εy Wpl εy
Mcsm = (16)
⎪
⎪ εcsm
⎪
⎩ Wel σ0.2 for 0.68 < λp,cs ⩽1.6
εy

into behavioural classes, the current design provisions of EN 1993-1-4

The test ultimate moment capacities Mu are compared with the EN
1993-1-4, AISC 370 and CSM resistance predictions Mu,pred calculated
based on the as-built material properties; the comparisons are depicted
in Fig. 26 and reported in Table 5. The graphical and numerical com­
parisons show that the EC3, AISC and CSM design approaches generally
yield safe-sided strength predictions for the WAAM beams, with the
ratios of the test ultimate moments to the predicted resistances, Mu/
MEC3, Mu/MAISC and Mu/Mcsm, equal to 1.21, 1.18 and 1.08 on average,
respectively, and reaching up to 1.37, 1.34 and 1.24, respectively; the
corresponding coefficients of variation (COV) are equal to 0.10, 0.09
and 0.08, respectively. Based on the concept of placing cross-sections
[25] and AISC 370 [26] yield comparable moment resistance pre­
dictions; the only major difference lies in the design of Class 3 (AISC
noncompact) cross-sections, where the AISC design method offers
improved design accuracy and efficiency over the elastic EC3 method by
means of utilisation of partial plastification in bending. Similar elasto-
plastic design rules for Class 3 steel cross-sections have recently been
incorporated into prEN 1993-1-1 [49,62]. The CSM allows for a rational
exploitation of the spread of plasticity and material strain hardening,
and thus, as expected, provides more accurate and consistent strength
predictions than those from the EC3 and AISC design approaches.

15
C. Huang et al.
7. Conclusions
An experimental programme has been undertaken to investigate the
cross-sectional behaviour of WAAM stainless steel tubular beams, as
well as to complement the safety verification of the MX3D bridge [20].
Four-point bending tests on a total of 14 square, rectangular and irreg­
ular tubular cross-sections, covering a wide range of local slenderness
values, have been conducted. A set of advanced measuring methods,
including 3D laser scanning, silicone casting and Archimedes’ mea­
surements, was employed to determine the as-built geometric properties
of the test specimens, while digital image correlation (DIC) was used to
monitor the deformation responses of the specimens during testing.
Similar to equivalent, conventionally formed sections, the tested
WAAM beams revealed the anticipated trend of decreasing normalised
moment capacity and rotation capacity with increasing local slender­
ness, reflecting increasing susceptibility to local buckling. However,
owing to the inherently greater variation in geometry, the WAAM beams
exhibited more variable flexural capacities between repeat specimens
than generally displayed by conventionally produced stainless steel
sections. The test results were used to assess the applicability of the
cross-section design provisions in EN 1993-1-4 and AISC 370 as well as
the continuous strength method (CSM) to WAAM stainless steel tubular
beams. On the basis of the as-built material properties, the current
codified local slenderness limits were shown to be generally applicable
to WAAM sections in bending; the EC3, AISC and CSM design ap­
proaches generally yielded safe-sided strength predictions for WAAM
tubular beams, with the CSM providing the most accurate and consistent
results. Further test data are nonetheless considered necessary to
confirm the findings and assess the reliability. The presented results lay
the foundations for the development of numerical models alongside
more suitable structural design rules for WAAM elements in future
research.
CRediT authorship contribution statement
Cheng Huang: Data curation, Formal analysis, Investigation,
Methodology, Software, Writing – original draft, Writing – review &
editing, Visualization. Xin Meng: Formal analysis, Investigation,
Methodology, Writing – review & editing. Leroy Gardner: Conceptu­
alization, Funding acquisition, Supervision, Writing – review & editing.
Declaration of Competing Interest
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influence
the work reported in this paper.
Data availability
Data will be made available on request.
Acknowledgements
This experimental programme was possible thanks to the funding
and support from the Data Centric Engineering programme at the Alan
Turing Institute (ATI) and the China Scholarship Council (CSC). This
research also benefitted from the Engineering and Physical Sciences
Research Council (EPSRC) funding under Grant No. EP/R010161/1 and
from the UK Collaboratorium for Research on Infrastructure and Cities
(UKCRIC) Coordination Node EPSRC grant under EP/R017727/1. The
authors gratefully acknowledge MX3D for the provision of the test
specimens, and would also like to thank Les Clark, Gordon Herbert, Bob
Hewitt, Wing Wang and Pinelopi Kyvelou for their assistance during the
experiments.
16
Engineering Structures 272 (2022) 114922
References
[1] Lim S, Buswell RA, Le TT, Austin SA, Gibb AGF, Thorpe T. Developments in
construction-scale additive manufacturing processes. Autom Constr 2012;21:
262–8.
[2] Campbell I, Bourell D, Gibson I. Additive manufacturing: rapid prototyping comes
of age. Rapid Prototyp J 2012;18:255–8.
[3] Bourell D, Kruth JP, Leu M, Levy G, Rosen D, Beese AM, et al. Materials for additive
manufacturing. CIRP Ann - Manuf Technol 2017;66(2):659–81.
[4] Ngo TD, Kashani A, Imbalzano G, Nguyen KTQ, Hui D. Additive manufacturing (3D
printing): a review of materials, methods, applications and challenges. Compos
Part B 2018;143:172–96.
[5] Buchanan C, Gardner L. Metal 3D printing in construction: A review of methods,
research, applications, opportunities and challenges. Eng Struct 2019;180:332–48.
[6] Ye J, Kyvelou P, Gilardi F, Lu H, Gilbert M, Gardner L. An end-to-end framework
for the additive manufacture of optimized tubular structures. IEEE Access 2021;9:
165476–89.
[7] Kanyilmaz A, Demir AG, Chierici M, Berto F, Gardner L, Kandukuri SY, et al. Role
of metal 3D printing to increase quality and resource-efficiency in the construction
sector. Addit Manuf 2022;50:102541.
[8] ASTM International. Additive manufacturing—general principles—terminology.
ISO/ASTM 52900, West Conshohocken, PA. 2017.
[9] Sasahara H, Tsutsumi M, Chino M. Development of a layered manufacturing system
using sheet metal-polymer lamination for mechanical parts. Int J Adv Manuf
Technol 2005;27(3-4):268–73.
[10] Buchanan C, Matilainen VP, Salminen A, Gardner L. Structural performance of
additive manufactured metallic material and cross-sections. J Constr Steel Res
2017;136:35–48.
[11] Rodrigues TA, Duarte V, Miranda RM, Santos TG, Oliveira JP. Current status and
perspectives on wire and arc additive manufacturing (WAAM). Materials (Basel)
2019;12:1121.
[12] Ding D, Pan Z, Cuiuri D, Li H. Wire-feed additive manufacturing of metal
components: technologies, developments and future interests. Int J Adv Manuf
Technol 2015;81(1-4):465–81.
[13] Haden CV, Zeng G, Carter FM, Ruhl C, Krick BA, Harlow DG. Wire and arc additive
manufactured steel: Tensile and wear properties. Addit Manuf 2017;16:115–23.
[14] Hönnige JR, Colegrove PA, Ahmad B, Fitzpatrick ME, Ganguly S, Lee TL, et al.
Residual stress and texture control in Ti-6Al-4V wire + arc additively
manufactured intersections by stress relief and rolling. Mater Des 2018;150:
193–205.
[15] Dinovitzer M, Chen X, Laliberte J, Huang X, Frei H. Effect of wire and arc additive
manufacturing (WAAM) process parameters on bead geometry and microstructure.
Addit Manuf 2019;26:138–46.
[16] Huang C, Kyvelou P, Zhang R, Britton TB, Gardner L. Mechanical testing and
microstructural analysis of wire arc additively manufactured steels. Mater Des
2022;216:110544.
[17] Gu J, Ding J, Williams SW, Gu H, Bai J, Zhai Y, et al. The strengthening effect of
inter-layer cold working and post-deposition heat treatment on the additively
manufactured Al-6.3Cu alloy. Mater Sci Eng A 2016;651:18–26.
[18] Buchanan C, Wan W, Gardner L. Testing of wire and arc additively manufactured
stainless steel material and cross-sections. China: Hong Kong; 2018.
[19] Laghi V, Palermo M, Gasparini G, Trombetti T. Computational design and
manufacturing of a half-scaled 3D-printed stainless steel diagrid column. Addit
Manuf 2020;36:101505.
[20] Gardner L, Kyvelou P, Herbert G, Buchanan C. Testing and initial verification of the
world’s first metal 3D printed bridge. J Constr Steel Res 2020;172:106233.
[21] Kyvelou P, Huang C, Gardner L, Buchanan C. Structural testing and design of wire
arc additively manufactured square hollow sections. J Struct Eng 2021;147:1–19.
[22] Huang C, Meng X, Buchanan C, Gardner L. Flexural buckling of wire arc additively
manufactured tubular columns. J Struct Eng 2022;148:04022139.
[23] Kyvelou P, Slack H, Daskalaki Mountanou D, Wadee MA, Britton TB, Buchanan C,
et al. Mechanical and microstructural testing of wire and arc additively
manufactured sheet material. Mater Des 2020;192:108675.
[24] Kyvelou P, Buchanan C, Gardner L. Numerical simulation and evaluation of the
world’s first metal additively manufactured bridge. Structures 2022;42:405–16.
[25] CEN (European Committee for Standardization). Eurocode 3—Design of Steel
Structures—Part 1-4: General Rules—Supplementary Rules for Stainless Steel. EN
1993-1-4:2006+A1:2020, Brussels, 2020.
[26] AISC (American Institute of Steel Construction). Specification for structural
stainless steel buildings. ANSI/AISC 370, Chicago, Illinois, 2020.
[27] Gardner L, Yun X, Walport F. The Continuous Strength Method – Review and
outlook. Eng Struct 2022. https://doi.org/10.1016/j.engstruct.2022.114924.
[28] Ramberg W, Osgood W R. Description of stress-strain curves by three parameters.
Technical Note No. 902, National Advisory Committee for Aeronautics.
Washington, D.C., USA, 1943.
[29] Mirambell E, Real E. On the calculation of deflections in structural stainless steel
beams : an experimental and numerical investigation 2000;54:109–33.
[30] Rasmussen KJR. Full-range stress-strain curves for stainless steel alloys. J Constr
Steel Res 2003;59:1–44.
[31] Arrayago I, Real E, Gardner L. Description of stress-strain curves for stainless steel
alloys. Mater Des 2015;87:540–52.
[32] Yun X, Wang Z, Gardner L. Full-range stress–strain curves for aluminum alloys.
J Struct Eng 2021;147:04021060.
[33] About MX3D. Available from: https://mx3d.com/company/about/ [Accessed 25
June 2021].
[34] Ziemian RD. Guide to stability design criteria for metal structures. 6th ed. 2010.
C. Huang et al.
[35] Geomagic Wrap 2017 (Version 2017.0.2:64) [Software] 2017 3D Systems,
Incorporated and its licensors, 2017.
[36] Rhino 3D (Version 5 SR14 64-bit) [Software] Robert McNeel & Associates, 2017.
[37] Zhao O, Rossi B, Gardner L, Young B. Behaviour of structural stainless steel cross-
sections under combined loading - Part I: experimental study. Eng Struct 2015;89:
236–46.
[38] Meng X, Gardner L. Testing of hot-finished high strength steel SHS and RHS under
combined compression and bending. Thin-Walled Struct 2020;148:106262.
[39] Laghi V, Palermo M, Gasparini G, Girelli VA, Trombetti T. Experimental results for
structural design of Wire-and-Arc Additive Manufactured stainless steel members.
J Constr Steel Res 2020;167:105858.
[40] Jafari D, Vaneker THJ, Gibson I. Wire and arc additive manufacturing:
opportunities and challenges to control the quality and accuracy of manufactured
parts. Mater Des 2021;202:109471.
[41] Li Y, Li X, Zhang G, Horváth I, Han Q. Interlayer closed-loop control of forming
geometries for wire and arc additive manufacturing based on fuzzy-logic inference.
J Manuf Process 2021;63:35–47.
[42] Lionello G, Cristofolini L. A practical approach to optimizing the preparation of
speckle patterns for digital-image correlation. Meas Sci Technol 2014;25.
[43] Davis 10 (Version 10.1.1) [Software] La Vision GmbH, 2017.
[44] Meng X, Gardner L. Cross-sectional behaviour of cold-formed high strength steel
circular hollow sections. Thin-Walled Struct 2020;156:106822.
[45] Theofanous M, Chan TM, Gardner L. Flexural behaviour of stainless steel oval
hollow sections. Thin-Walled Struct 2009;47:776–87.
[46] Theofanous M, Gardner L. Experimental and numerical studies of lean duplex
stainless steel beams. J Constr Steel Res 2010;66:816–25.
[47] Gardner L. Stability and design of stainless steel structures – Review and outlook.
Thin-Walled Struct 2019;141:208–16.
[48] Bu Y, Gardner L. Local stability of laser-welded stainless steel I-sections in bending.
J Constr Steel Res 2018;148:49–64.
17
Engineering Structures 272 (2022) 114922
[49] CEN (European Committee for Standardization). Eurocode 3—Design of Steel
Structures—Part 1-1: General Rules and Rules for Buildings. prEN 1993-1-1:2005+
A1:2020, Brussels, 2020.
[50] Afshan S, Gardner L. Experimental study of cold-formed ferritic stainless steel
hollow sections. J Struct Eng 2013;139:717–28.
[51] Theofanous M, Saliba N, Zhao O, Gardner L. Ultimate response of stainless steel
continuous beams. Thin-Walled Struct 2014;83:115–27.
[52] Bock M, Arrayago I, Real E. Experiments on cold-formed ferritic stainless steel
slender sections. J Constr Steel Res 2015;109:13–23.
[53] Gardner L, Fieber A, Macorini L. Formulae for calculating elastic local buckling
stresses of full structural cross-sections. Structures 2019;17:2–20.
[54] Gardner L, Nethercot DA. Experiments on stainless steel hollow sections-Part 2:
Member behaviour of columns and beams. J Constr Steel Res 2004;60:1319–32.
[55] Zhou F, Young B. Tests of cold-formed stainless steel tubular flexural members.
Thin-Walled Struct 2005;43:1325–37.
[56] Afshan S, Gardner L. The continuous strength method for structural stainless steel
design. Thin-Walled Struct 2013;68:42–9.
[57] CEN (European Committee for Standardization) Eurocode 3—Design of Steel
Structures—Part 1-5: Plated structural elements. EN 1993-1-5:2006+A2:2019,
Brussels, 2019.
[58] Zhao O, Afshan S, Gardner L. Structural response and continuous strength method
design of slender stainless steel cross-sections. Eng Struct 2017;140:14–25.
[59] Gardner L, Wang F, Liew A. Influence of strain hardening on the behavior and
design of steel structures. Int J Struct Stab Dyn 2011;11:855–75.
[60] Gardner L. The Continuous Strength Method. Proceedings of the Institution of Civil
Engineers – Structures and Buildings 2008;161(3):127–133.
[61] ABAQUS (Version 6.13-1) [Software] Hibbitt and Karlsson and Sorense,
Pawtucket, RI, 2013.
[62] Meng X, Gardner L, Sadowski AJ, Rotter JM. Elasto-plastic behaviour and design of
semi-compact circular hollow sections. Thin-Walled Struct 2020;148:106486.