High Deposition Wire Arc Additive Manufacturing of Mild Steel - Strategies and Heat Input Effect On Microstructure and Mechanical Properties

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

Journal of Manufacturing Processes 58 (2020) 615–626

Contents lists available at ScienceDirect

Journal of Manufacturing Processes


journal homepage: www.elsevier.com/locate/manpro

High deposition wire arc additive manufacturing of mild steel: Strategies


and heat input effect on microstructure and mechanical properties
E. Aldalur a, *, F. Veiga a, A. Suárez a, J. Bilbao b, A. Lamikiz c
a
TECNALIA, Basque Research and Technology Alliance (BRTA), Advanced Manufacturing Department, Paseo Mikeletegi 7, 20009 San Sebastián, Spain
b
ADDILAN Fabricación Aditiva S.L., Eguzkitza 1, 48200 Durango, Spain
c
Department of Mechanical Engineering, University of the Basque Country UPV/EHU, Faculty of Engineering of Bilbao, Plaza Torres Quevedo 1, 48013 Bilbao, Spain

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 layer-by-layer production concept that is proposed as a promising
Additive manufacturing alternative to traditional subtractive techniques due to its potential to manufacture large metallic components of
WAAM medium geometrical complexity. In recent years, this technology has created interest due to the advantages it
High deposition rate
proposes, such as a low cost and high deposition rates in comparison with other Additive Manufacturing (AM)
ER70S-6 steel
Oscillatory strategy
techniques. In this study, the Gas Metal Arc Welding (GMAW)-based WAAM system was developed to deposit
mild steel material. To that end, a comprehensive methodology to understand the main process variables and
their influence in the final component properties is shown. The aim of the present study was to compare the
influence of oscillatory and overlapping deposition strategies in terms of productivity, growth per layer, obtained
microstructure and mechanical properties using the control volume concept to compare volume-dependent pa­
rameters. Although some characteristics of these strategies have been documented in literature, the heat input
influence has not yet been comprehensively demonstrated and understood. For example, heat input, analyzed
deeply in the present study, is a critical factor for the stability of the WAAM process, influencing the mechanical
properties and microstructural evolution of as-fabricated parts. The results indicate that the oscillated strategy
increases productivity.

1. Introduction is a process that has evolved into an encouraging alternative to


manufacturing metallic large-scale parts with relatively low and me­
Steel refers to ferrous alloys that are most commonly used as engi­ dium structural complexity. Automatic welding is the fundamental
neering materials. Depending on their composition, different properties concept behind WAAM [5]. This process is mainly composed of an en­
can be achieved but, in general, steels are high tensile strength and low ergy source, a wire feeding system, a kinematic platform (an articulated
cost materials. Among the different types of steels, mild steel has a low robot or a Gantry machine) and some accessories, such as the shielding
carbon content and, depending on its alloying element proportion and gas system [6]. In this process, the power source melts the wire and the
the deposition parameters, its microstructure is defined. Steels are substrate with an electric arc to deposit material on the top of the sub­
widely used in the construction, naval, gas & oil, wind energy and strate, layer by layer, obtaining the desired preform. In addition, during
automotive industries [1–4]. In these sectors, mild steel, in combination the process, the melt pool is protected with shielding gas to prevent
with additive manufacturing technologies, can be utilized to manufac­ contamination from atmospheric gases. Rodrigues et al. [4] review
ture large parts with a low global cost. different studies that analyze the effect of the type of shielding gases and
Within AM technologies, Wire Arc Additive Manufacturing (WAAM) the flow rate, concluding that these parameters affect the mechanical

Abbreviations: OVL, overlapping; OSC, oscillating; DR, deposition rate; WFR, wire feed rate; TS, travel speed; Δx, control volume dimension in the x-axis; Δy,
control volume dimension in the y-axis; Δz, control volume dimension in the z-axis; l, traveled distance; ρ, density; T, period of the current signal; Vdeposited, deposited
material volume; A, wire transversal area; ΔV, volumetric layer growth; ΔVm, average volumetric layer growth; p(x)i, transversal profiles of the walls; P, power; I,
intensity; V, voltage; HI3D, in-volume heat input; HAZ, heat affected zone; UTS, ultimate tensile strength; YS, yield strength; Elong., elongation.
* Corresponding author.
E-mail addresses: eider.aldalur@tecnalia.com (E. Aldalur), fernando.veiga@tecnalia.com (F. Veiga), alfredo.suarez@tecnalia.com (A. Suárez), jbilbao@addilan.
com (J. Bilbao), aitzol.lamikiz@ehu.eus (A. Lamikiz).

https://doi.org/10.1016/j.jmapro.2020.08.060
Received 19 June 2020; Received in revised form 24 August 2020; Accepted 28 August 2020
Available online 2 September 2020
1526-6125/© 2020 The Society of Manufacturing Engineers. Published by Elsevier Ltd. All rights reserved.
E. Aldalur et al. Journal of Manufacturing Processes 58 (2020) 615–626

properties and wall appearance. rate WAAM processes, without taking into account the influence of some
Depending on the type of the energy source, three types of WAAM parameters, such as heat input. In this paper, all the parameters that can
processes are distinguished: (1) WAAM based on Gas Metal Arc Welding influence the obtained properties or the productivity have been
(GMAW), (2) WAAM based on Gas Tungsten Arc Welding (GTAW) and considered.
(3) WAAM based on Plasma Arc Welding (PAW). Artaza et al. [7]
compared the mechanical properties of parts manufactured by 2. Experimental procedures
GMAW-based and PAW-based WAAM processes on Mn4Ni2CrMo steel.
Their study showed that, with both technologies, higher mechanical 2.1. Material and set-up
values than analogous forging standards are achieved. In the present
study, among the different WAAM technologies, the GMAW-based sys­ The WAAM technique, based on GMAW, was used to manufacture
tem was chosen. In this process, the wire itself is the consumable elec­ walls with different strategies in ER70S-6 mild steel (EN ISO 14341-A - G
trode that is introduced in the process coaxially [8]. 42 3 M21 3Si1). Commercially produced ER70S-6 wire with a diameter
Considering the nature of the WAAM process and, taking into ac­ of 1.2 mm was employed as the deposited material (the elemental
count that it should be economically competitive, the most important composition of the wire, as provided by the supplier, is given in Table 1).
advantage is the high deposition rate obtained compared to other ad­ Furthermore, 8 mm S235JR steel (DIN ST37-2) plates were used as a
ditive manufacturing technologies. In general, depending on the mate­ substrate.
rial and the process parameters, the GMAW technology deposition rates To carry out the experiments, an automatic welding system was
are between 1–10 kg/h [5], which is 2–3 times higher than those ob­ utilized. Fig. 1 shows the used 3 + 2-axis Addilan WAAM machine [18]
tained with the WAAM processes, based on GTAW and PAW. In the with a GMAW welding system, composed of a Titan XQ 400 AC puls
GMAW process, liquid droplets are transferred from the electrode (the generator and an M drive 4 Rob5 XR RE wire feeder from the EWM
wire) to the melt pool. Depending on the electrode current signal welding manufacturer. This machine has high versatility due to its
characteristics, different metal transfer modes are embraced. This cur­ particular architecture and the open CNC software. In addition, it offers
rent signal is defined by the welding machine manufacturers creating the opportunity to manufacture parts utilizing additional AM technol­
work modes, such as CMT and Pulse [9]. For instance, CMT or the Cold ogies, such as WAAM-based Gas Tungsten Arc Welding (GTAW) or
Arc mode obtains walls with thin thicknesses and low deposition rates, WAAM-based Plasma Arc Welding (PAW). This machine is also equip­
owing to its low heat input and its small melt pool [10,11]. ped with a close chamber which allows for the manufacturing of parts in
Although diverse studies that have dealt with different aspects of reactive materials.
GMAW-based WAAM technology have been carried out, there are still In the GMAW process, the electric arc melts the wire to deposit it on
topics that have not been studied in detail. Amongst them, to improve the top of the substrate obtaining the desired geometry. Furthermore, to
productivity and geometrical accuracy, the oscillation deposition strat­ control the process, a geometrical laser scanner (Laser Scanner Q4 Se­
egy is proposed. Ma et al. [12] realized a comprehensive study about the ries) and a pyrometer (Optris CT) were placed in the welding torch
oscillatory (weaving) strategy for producing large and thin-walled (Fig. 2). For the measurement of the surface temperature, an emissivity
structures in aluminum alloys. The aim of this study was to use the of 0.5 was considered for mild steel following the recommendations
oscillatory strategy to improve the top surface flatness, concluding that provided by the supplier. The laser scanned the geometry of the walls
is an effective system. Caballero et al. [13] performed experiments to every two layers, obtaining a cloud of points that represented the
determine the influence of different process parameters, including transversal geometrical shapes of the deposited layers and, thanks to this
different deposition paths (single, oscillated and parallel pass), on the cloud of points, the operator can decide the Z position for the following
microstructure and mechanical properties in 17− 4 pH stainless steel layer. For more detail, this laser measures 244 points per profile and the
parts. These parts were manufactured by CMT welding equipment and, profiles are given every 2 mm in the longitudinal direction. On the other
depending on the utilized strategy, different properties were obtained. hand, the pyrometer was used to control the temperature of the walls
On the other hand, the comparison of the effect of the parallel pass, and determine the inter-layer cooling time.
triangular oscillation and square oscillation on bead geometry was
carried out by Xu et al. [14] on maraging steel, concluding that a 2.2. Deposition strategy and parameters
triangular oscillation strategy is the most recommended. Lastly, Dirasu
et al. [3] compared the strength and fracture toughness properties ob­ In this research, the synergic MAG-PULS mode of the welding ma­
tained on steel parts manufactured by oscillatory and unique-pass stra­ chine was used. The test samples were two 220 mm long straight walls
tegies. All this research was performed with low deposition rate GMAW with the same width (20 mm) and height (70 mm) created using two
work modes, such as CMT, but the influence of the oscillated strategy on different deposition strategies: Wall 1 overlapping and Wall 2 oscil­
high deposition rate processes still has not been studied. lating, as seen in Fig. 3. In both cases, alternating layers were deposited
Another aspect that has not received the attention that it deserves, in opposite directions to maintain geometrical accuracy and to prevent
which is covered in this article, is heat input. During the WAAM process, the effect reported by Zhang et al. [19] in the rectangular part—namely,
the multiple cycles of fusion and solidification create non-uniform heat the deposition was made with a bi-directional movement of the torch. In
propagation in the as-built components. This nonlinear heat propagation Wall 1, three beads were overlapped in each layer due to the impossi­
and accumulation determine the obtained part formation, microstruc­ bility of obtaining a 20 mm wide wall with a single linear bead. In this
ture and mechanical properties [15]. Similarly, Li et al. [16] found that case, the overlap ratio was 65 % of the bead width (6.9 mm). On the
the heat input influences the microstructure and mechanical properties other hand, in Wall 2, to obtain the same width as Wall 1, the amplitude
in Al-7Si-0.6 Mg components. Moreover, the heat input also affects in­ of the oscillation was 10 mm and the frequency was 1.5 Hz.
ternal defect formation, residual stress, distortion and phase trans­ The deposition was performed with the torch in the vertical position
formations [17]. Therefore, the need for in-process temperature control and the stick-out length (the distance that the wire comes out from the
and monitoring systems is evident. welding contact tip) was set to 15 mm. Following the recommendation
The motivation behind the present study is to compare the over­ of the feedstock supplier, a shielding gas, argon blend (20 % CO2 and 80
lapping deposition strategy of linear weld beads and oscillating depo­ % Ar) was supplied at a flow rate of 17 L/min. Since the main objective
sition strategies for high deposition rate GMAW-based WAAM processes. of this work was to achieve a stable process with good penetration, in
The characteristics and suitability of each case are evaluated. The Table 2 the necessary set of welding parameters in both walls is shown.
analysis centers on mild steel parts since this is the most widespread case These optimized parameters were decided by the preliminary study for
in the industry. While most of the studies to date focus on low deposition parameter selection, carried out by the authors in [20].

616
E. Aldalur et al. Journal of Manufacturing Processes 58 (2020) 615–626

Table 1
Chemical composition of ER70S-6 wire in weight %.
Mn Si C Cr Cu Ni S P Mo Ti Zr Fe

1.64 0.94 0.06 0.02 0.02 0.02 0.016 0.013 0.005 0.004 0.002 bal.

2.3. Synergic GMAW

Synergic pulsed GMAW welding is a variant of common pulsed


GMAW welding and, nowadays, it is the most used method in welding
equipment. The synergic GMAW welding set has a digitally controlled
power source that creates current pulses. In each of these pulses, an
identical drop of molten material is released from the wire to the sub­
strate [21]. These pulses, combined with a particular relationship be­
tween process parameters, ensure stable wire burn off.
Depending on the material, shielding gas and wire diameter, the
recommended pulse parameters (peak/background current, peak dura­
tion, base duration, pulse frequency, etc.) are unique and they are pro­
grammed by the welding equipment producer. Once these pulse
characteristics are pre-selected, the wire feed rate governs this control,
automatically amending the values of these pulse parameters.
Fig. 1. 3 + 2-axis Addilan machine [18]. The automation of the choice of these parameters depending on the
wire feed rate is made through the design of an algorithm that con­
templates calculation routines and a database with the involved material
information. This synergic control of parameters guarantees uniform
weld bead profile and penetration. In comparison, the principal disad­
vantage of non-synergic sets is that they require the individual setting of
the pulse parameters for each given wire feed rate.
Fig. 4 shows the theoretical waveform of the current signal of both
manufactured walls. It is obtained by interpolating from the database
designed by EWM for steel material, 80 % Ar–20 % CO2 shielding gas,
1.2 mm wire diameter and 8 m/min wire feed rate. The drop formation
frequency in this case is 138.2 Hz. This figure also shows the process of
drop formation and its correlation with the current signal.

2.4. The mechanical and microstructural characterization method

For metallographic analysis, samples were transversely cross-


sectioned (XZ), longitudinally cross-sectioned (YZ) and sectioned be­
tween layers (XY) (axes referenced in Fig. 12) with a metallic saw,
mechanically polished and etched with 2% Nital solution (Nitric and
Ethanol acid), successively. Once the samples were metallographically
prepared, microstructures and macrostructures were observed with the
Eclipse MA200 (Nikon) microscope. For macro-examination and Vickers
hardness testing, samples were taken, transversally, in the central part of
the walls at three heights: the top, the middle and in the bottom. Vickers
Fig. 2. The welding torch. hardness testing was performed at room temperature, utilizing the
Duramin A-300 (Struers) hardness testing machine, according to stan­
dard ISO 6507-1 [22].
On the other hand, as shown in Fig. 13, twelve tensile specimens,
according to specific dimensions—neck diameter 4 mm and length
22 mm—were cut from the overlapped and oscillated walls. Half of the
tensile specimens were cut in the vertical direction and half in the
horizontal direction. Tensile specimens were prepared following ISO
6892-1 [23]. The uni-axial tensile test was carried out at room tem­
perature using an Instron 5585H electronic machine. The tensile testing
was performed with a contact extensometer Instron EX2620-602 with a
load cell of 100 kN.

3. Results and discussion

3.1. Material deposition and growth per layer

To study the material deposition and growth per layer, a 3D control


volume (Δx, Δy, Δz) was defined in both walls, as shown in Fig. 5. In
Fig. 3. Paths of overlapping and oscillating strategies.
general, the control volume is the reference volume necessary to have

617
E. Aldalur et al. Journal of Manufacturing Processes 58 (2020) 615–626

Table 2
Overlapped and oscillated wall manufacturing parameters.
Wall WFR (m/min) Current (A) Voltage (V) Thickness of single bead (mm) TS (cm/min) DR (kg/h) Pyrometer limit (ºC)

Overlapped 8 229 27.1 8 65 4.26 400


Oscillated 8 244 26.6 8 20 4.26 400

Fig. 6. Torch trajectory and drop deposition during oscillated and overlapped
walls manufacturing in the control volume.
Fig. 4. Theoretical current signal and the process of the drop formation.

Fig. 7. Calculation of the volumetric layer growth of layers 25 and 26 in both


walls. Note that the curves represent the subtraction between the 26th layer
transversal profile and the 24th transversal profile in both walls. The areas
under the curves are the volumetric layer growths for layers 25th and 26th in
both walls.

comparable results of some volume-dependent parameters, such as


material deposition or heat input. In this and future studies, the control
volume concept should be used to enable the comparison between the
obtained results. In this case, taking into account that the main objective
is to manufacture walls as accurately as possible to the defined geometry
(wall width 20 mm and wall height 70 mm), in both walls the control
volume embraces all the deposited material in a 20 mm wide (Δx) and
20 mm long (Δy) area. The control volume height (Δz) is given every
two consecutive layers because the laser scans every two layers.
Regarding the material deposition inside the control volume, both
deposition strategies are presented in Fig. 6. As previously mentioned, in
each unit current pulse, the synergic pulsed GMAW welding process
detaches identical molten droplets, as represented in Fig. 6. As the wire
Fig. 5. Control volume definition. feed rate is the same on the two walls compared in this case, the amount
of material in every drop is also equal.
However, inside the control volume the number of deposited drops is
higher during oscillated wall manufacturing than in overlapped wall

618
E. Aldalur et al. Journal of Manufacturing Processes 58 (2020) 615–626

Eq. 2, where DR is the deposition rate, WFR is the wire feed rate and A is
the wire transversal area. As seen previously, the number of drops within
the control volume is greater in the oscillated wall than in the over­
lapped wall; thus, the deposited volume is also greater in the oscillated
wall, since all the drops have the same volume, thanks to the synergic
GMAW process. Table 3 shows the Vdeposited values in the control volume
(two layers) in both walls:
DR ∗ l
Vdeposited = (2)
TS ∗ ρ

where DR = WFR ∗ ρ ∗ A.
Moreover, another important parameter in the WAAM process is the
volumetric layer growth (ΔV). To calculate the volumetric layer growth
inside the control volume, clouds of points that represent the transversal
profiles of the walls (p(x)i) are used. These data were obtained by
scanning the walls every two layers with the geometrical laser during
the manufacturing process. For example, Fig. 7 shows the volumetric
Fig. 8. Average volumetric layer growth. Note that error bars indicate the layer growth of layers 25 and 26 in both walls. To determine this growth,
maximum and minimum values. the subtraction between two consecutive transversal profiles is calcu­
lated and represented, as in Fig. 7. The area under the curve is calculated
manufacturing—829 and 765 drops, respectively. The deposited num­ and the volumetric layer growth inside the control volume is obtained.
ber of drops can be calculated as Eq. 1, where l is the traveled distance, This process can be defined with Eq. 3, where p(x)i are the transversal
TS is the travel speed and T is the period of the current signal. In the profiles scanned with the geometrical laser, Δx is the length taken in x-
oscillated wall, more drops are deposited in the corners than in the axis direction—in this case, the control volume width is 20 mm—and Δy
central part due to its velocity field. In this wall, the travel speed is not is the length taken in y-axis direction—in this case, the control volume
constant along the path, becoming null in the corner and maximum in length is 20 mm.
the center. However, in the overlapped wall, the deposition is uniform ∫Δx
because the welding torch has a constant travel speed along the entire ( )
ΔV = p(x)i+2 − p(x)i ∗ dx ∗ Δy (3)
trajectory.
0
l
Drop Number = (1) Although more drops of material were deposited in the oscillated
TS ∗ T
wall inside the control volume, the average volumetric layer growth
The deposited material volume (Vdeposited) can be calculated using (ΔVm) in the oscillation strategy is lower than in the overlapped

Fig. 9. (a) Temperature evolution during the intra-layer cooling time in the eighth layer of the overlapped wall; (b) Inter-layer cooling time evolution along the
overlapped wall; (c) Temperature evolution during the inter-layer cooling time in the first eight layers in the oscillated wall; (d) Inter-layer cooling time evolution
along the oscillated wall.

619
E. Aldalur et al. Journal of Manufacturing Processes 58 (2020) 615–626

Fig. 10. (left) Macrostructures of the top and bottom parts of the overlapped wall in transversal direction (plane XZ); (right) Microstructures of the overlapped wall.
Note the microstructures are taken in the areas marked with the boxes in the macrographs.

strategy—1162 mm3 and 1414 mm3, respectively—as shown in the layer growth:
Fig. 8. This is due to the different heat input that supports the walls,
Vdeposited − ΔV
creating a different dilution rate. As a consequence, the number of layers Dilution = ∗ 100 (4)
Vdeposited
necessary to obtain the desired height is smaller in the overlapped wall
than in the oscillated wall—40 layers and 48 layers, respectively. Finally, the deposited material volume in the control volume, the
Furthermore, the dilution can be achieved through Eq. 4, where average volumetric layer growth and the dilution rate of each wall are
Vdeposited is the deposited material volume and ΔV is the volumetric represented in Table 3. As can be seen, the oscillated wall presents

620
E. Aldalur et al. Journal of Manufacturing Processes 58 (2020) 615–626

Fig. 11. (left) Macrostructure of the top and bottom parts of the oscillated wall in transversal direction (plane XZ); (right) Microstructure of the oscillated wall. Note
the microstructures are taken in the areas marked with the boxes in the macrographs.

higher dilution due to the high heat input and heat accumulation on it. even if the stick-out length is the same.
Aspects related to the heat input are discussed in the next section. Furthermore, Table 4 shows the heat input parameters that are
calculated for the control volume. It is worth noting that this value is the
heat input of two-layers because within the control volume there are two
3.2. Heat input layers. Lastly, it is possible to calculate the heat input per cubic milli­
meter grown, which is higher in the oscillated wall.
While the heat input concept in welding is well defined, in additive As can be seen in Table 4, in the overlapped wall the introduced
manufacturing it is not such a trivial concept. In this case, to calculate energy is less, and it is added in three parts, having intra-layer cooling
the in-volume heat input is necessary to determine a control volume (Δx, times between each deposited bead. On the other hand, the introduced
Δy, Δz) as the one defined in the 3.1 section that embraces all the energy in the oscillated wall is greater and it is inserted, continuously, in
deposited material. It is also important to quantify the heat input one go. As a result of torch oscillation being maintained for a long time
introduced in this control volume, taking into account the multiple in the same area, the wall retains more heat; namely, the heat accu­
passes. Notice that the introduced energy can be determined as mulation is higher. Therefore, the dilution increases in the oscillated
described by Eq. 5 and the in-volume heat input can be expressed ac­ wall and, consequently, the average growth per layer is lower and the
cording to Eq. 6, where I is the intensity, V is the voltage and TS is the number of layers needed to obtain the desired wall is greater, as was
travel speed. verified in Section 3.1. Moreover, the energy input also affects the
I ∗ V ∗ Δy ∗ Number of beads microstructure and mechanical properties of the parts, as found by
Introduced energy = (5) Rodrigues et al. [24]. This research expresses that high heat input results
TS
in a lower surface undulation and large grain size microstructure. This
HI3D =
Introduced energy Introduced energy
= (6) effect is also reflected in the present study since, in the two manufac­
Volume Δx ∗ Δy ∗ Δz tured walls, variations in microstructure and mechanical properties are
To calculate and compare the heat input in both walls, the intensity shown. Dirisu et al. [3] reported the same effect, comparing the oscil­
and voltage values were monitored during the process. The average lated strategy with the single-pass strategy in high strength low alloy
values of all the layers are shown in Table 4. Theoretically, having the structural steels.
same wire feed rate in both cases (8 m/min), the intensity and voltage
values should be equal since synergic equipment is used, as explained
above. However, in practice, the exact arc length affects these values,

621
E. Aldalur et al. Journal of Manufacturing Processes 58 (2020) 615–626

Fig. 12. (left) Microstructure of the center parts of the oscillated wall in YZ and XY planes; (right) Microstructure of the overlapped wall in YZ and XY planes.

3.3. Cooling time analysis and fabrication time


Table 3
Parameter values related to the growth per layer in both walls.
Depending on the material and the parameters of the WAAM process,
Vdeposited (mm3) ΔVm (mm3) Dilution (%) it is necessary to have a cooling time between subsequent passes to
Overlapped wall 1669.51 1414 15.30 ensure the growth of the part and in order not to have excessive heat
Oscillated wall 1808.64 1164 35.75 accumulation. In this way, the heat input determines the necessary
interpass temperature. The interpass temperature, by definition, is the
temperature of the preceding layer just before the deposition of the
Table 4 following one. In addition, the cooling time is dependent on this inter­
Parameters related to the heat input in both walls. pass temperature and it affects the total manufacturing time and, ulti­
I (A) V (V) Introduced energy (J) HI3D (J/mm3) mately, productivity. As Rodrigues et al. [24] reviewed, if this
temperature is too low, the cooling time can extend for too long,
Overlapped 229.34 27.09 68818.94 48.67
Oscillated 244.45 26.61 78057.77 67.17 rendering the total production time impracticable. However, on the
other hand, if the interpass temperature is too high, the deposition can
be erratic and may induce the collapse of the part. Therefore, to analyze
the cooling time in both manufactured walls, after the deposition of each
bead, the temperature is measured at the starting point of the next bead

622
E. Aldalur et al. Journal of Manufacturing Processes 58 (2020) 615–626

using the pyrometer. In both cases, waiting time is required until at this created because of the different deposition sequences and paths, it is
point, the temperature is less than 400 ◦ C (interpass temperature) and expected that the microstructure in the oscillated wall and the over­
the temperature evolution is monitored. lapped wall will be different.
On the one hand, in the overlapped wall within the same layer (in In previous studies, Choi and Hill [26] analyzed the microstructure
this case, layer 8), the intra-layer cooling time increases considerably on ER70S-6 welds, which is helpful to analyze the microstructure of our
from the first bead to the third because of heat accumulation in the part, walls. In Fig. 10a,b the macrographs of the upper and lower part of the
as can be seen in Fig. 9a. In Fig. 9b, inter-layer cooling time is graphed. overlapped wall are observed in the transversal direction of the wall
To calculate inter-layer cooling time in this wall, the sum of the three (plane XZ) (axes referenced in Fig. 12). In these images, each contrib­
intra-layer cooling times of the three beads is considered. As can be seen uted bead and HAZ are clearly visible. In this heterogeneous micro­
in this graph, these inter-layer cooling times do not change much structure, areas with polygonal ferrite and acicular ferrite can be
throughout the different layers—that is, they remain constant. Even so, differentiated. The creation method of acicular ferrite in weld beads was
it is true that, in the first layers, the inter-layer cooling time is shorter reviewed by Babu [28].
because the substrate is at an atmospheric temperature from the In the upper part of the wall (Fig. 10 a.1) a mixture of acicular ferrite,
beginning. The average value of the inter-layer cooling time in this case allotriomorphic ferrite and bainite is seen since this part has not expe­
is 115.50 s. rienced the thermal effect of the following layer. A similar microstruc­
On the other hand, in the oscillated wall, as can be seen in Fig. 9d, the ture was obtained by Shassere et al. [29] in a single-pass strategy in the
inter-layer cooling time is variable along the wall. In the first layers, the upper part of the manufactured wall. In Fig. 10 a.1, the darker areas
inter-layer cooling time is shorter since the substrate is still cold, but as correspond to acicular ferrite and bainite microstructures, and the
the number of deposited layers grows, the cooling time increases until it lighter areas to allotriomorphic ferrite microstructures. In the remainder
stabilizes around layer 27. In this case, the only cooling time that is of the wall, polygonal ferrite areas (Fig. 10 a.3, b.2) and acicular ferri­
recorded is the inter-layer since in each layer only one bead is deposited. te/bainite areas (b.1) are intercalated. This non-uniformity in the
Fig. 9c shows the evolution of the inter-layer cooling time in the first microstructure is due to the non-uniform energy input caused by the
eight layers in detail—note the increase in the cooling time for the first characteristics of the deposition strategy. It is evident that the acicular
layer until the layer eighth. In this case, the average value of the inter- ferrite parts have not received the thermal effect of the surrounding
layer cooling time for each layer after stabilization is 49.38 s. passes; however, the polygonal part has. In Fig. 10 a.3 and b.2, polyg­
Although the energy introduced in the process in each layer is greater onal ferrite with some perlite in the grain boundaries (black features)
in the oscillated wall, as seen in the previous section, the average inter- can be seen. Dirasu et al. [3] and Haden et al. [30] obtained the same
layer cooling time is shorter in the oscillated wall because this energy microstructure in the single-pass manufactured wall central part in
enters in a different way. As previously explained, in the oscillated wall, ER70S-6 steel. For more detail, in Fig. 10 a.2 the transition for the
unlike the overlapped wall, all the energy is introduced in one go. On the acicular ferrite microstructure to the polygonal ferrite microstructure
other hand, in the overlapped wall, the energy is introduced in three can be seen.
parts separated by cooling times. Therefore, the sum of these three Furthermore, the ferritic grain size variation can also be observed
cooling waiting times in the overlapped wall is longer than the unique along the wall. There are areas with fine grain size, ASTM 12 and areas
cooling waiting time in the oscillated wall in all of the layers. with bigger grain size ASTM 10. In Fig. 10 b.2, this variation can be seen,
Regarding fabrication time, it takes more time to manufacture the with the grain size being bigger in the top part compared to the lower
wall by overlapping than oscillating, even if more layers are deposited in part. This variation is also a consequence of the different thermal history
the oscillated method. This fabrication time includes the adding time that the areas have undergone. For example, the grain coarsening in the
and the cooling time in both walls, as shown in Table 5. The added time HAZ zone was reported widely in the literature [26,28,30].
is longer for the oscillated wall because the average speed in the entire On the other hand, Fig. 11c,d present the macrographs of the upper
path is lower and the distance traveled is similar. In contrast, as and lower part of the oscillated wall in the transversal direction of the
demonstrated before, the total cooling time is longer for the overlapped wall (plane XZ) (axes referenced in Fig. 12). In this case, the micro­
wall. It can be concluded that the oscillating method is more productive structure is totally different. The heat input layer by layer, reported in
than the overlapped method using these data. the previous section, produces the formation of layer bands. These bands
are easily differentiated in Fig. 11c,d and the distances between the
3.4. Macrostructure, microstructure, and hardness bands are equal to the layer height. In the upper part of the wall, as in the
case of the overlapped wall, a mixture of acicular ferrite, allo­
As it is known, different process parameters affect both geometric triomorphic ferrite and bainite microstructures (Fig. 11 c.1) is observed
accuracy and metallurgical properties, such as macrostructure, micro­ because this part has not undergone the thermal effect of the successive
structure and hardness [27]. Furthermore, since WAAM technology deposited beads. In the remainder of the wall, a homogeneous micro­
parts are manufactured layer by layer, the heat input generated in the structure of polygonal ferrite with some perlite can be seen. Since only
manufacturing process causes a partial re-melt in the previous layers. As one oscillatory bead is deposited in each layer, the thermal input is
a result, the microstructure of the melted material and heat-affected uniform throughout all the layers and, therefore, this homogeneity in
zone (HAZ) is subjected to successive thermal cycles. Depending on the microstructure is observed. Even so, small grain size variations are
the heat input of each weld bead and the deposition strategy, areas with noticed at different heights of the wall but, in general, the grain size is 10
a different thermal history may evolve. Therefore, it is evident that ASTM, larger than the overlapped one. This larger grain size is due to the
different regions of the part will have different microstructures. Taking heat accumulation effect and the consequent low cooling rate. The same
into account this characteristic of WAAM technology, Cunningham et al. behavior was obtained by Oliveira et al. [17] and Dirasu et al. [3] in
[25] studied the improvements that can introduce, in this aspect, the their research.
different ancillary processes. In our case, due to different thermal cycles Fig. 12 presents the axes referenced, taken for the microstructural
analysis, and the microstructures of the center parts of the oscillated and
Table 5
overlapped walls in planes YZ and XY. In these planes, the polygonal
Fabrication times. ferrite microstructure can also be observed and, as in the XZ plane,
slightly larger grain sizes can be observed in the oscillated wall than in
Adding time Cooling time Fabrication time
the overlapped wall.
Overlapped 46 min 46 s 75 min 38 s 2h 2 min 24 s With regard to the hardness analysis, the hardness and the micro­
Oscillated 52 min 36 s 32 min 29 s 1h 25 min 5 s
structure are closely related, and hardness mapping can illustrate the

623
E. Aldalur et al. Journal of Manufacturing Processes 58 (2020) 615–626

level of homogeneity of the microstructure. In the manufactured walls, Table 7


Vickers hardness testing was done at the top, middle and bottom part of Transversal hardness evolution test of the produced walls.
both walls. The average values obtained in each part are presented in Hardness (HV)
Table 6. In both walls, the central and bottom part hardness measures
1 2 3 4 5 6
are between the limits of the wrought ASTM A36, which is similar in
composition to the ER70S-6. Furthermore, Haden et al. [30] obtained Overlapped wall 143 141 155 149 141 141
Oscillated wall 130 129 129 128 128 131
similar values of hardness for WAAM single-pass manufactured ER70S-6
walls.
As can be seen in Table 6, in both walls there are higher hardness
values in the upper part of the wall due to the presence of the acicular
ferrite microstructure. This acicular ferrite has greater hardness because
of its needle shape structure. In addition, at the three heights the over­
lapped wall has higher values of hardness than the oscillated wall due to
its non-uniformity in the microstructure. This non-uniform microstruc­
ture intercalates acicular ferrite and polygonal ferrite areas, obtaining
higher hardness values if the indentation is in the acicular ferrite area.
To check this phenomenon, the transversal hardness evolution is
analyzed inside both walls. Therefore, in the middle part of both walls,
six indents were made, and the results are shown in Table 7. As can be
seen, the hardness values along the overlapped wall are less uniform
than in the oscillated wall because of its heterogeneity in the
microstructure. Fig. 13. Arrangement of specimens for the mechanical properties analyses in
both walls.

3.5. Evaluation of mechanical properties

To determine the mechanical properties obtained in both strategies a


tensile test is executed and in Fig. 13 the sample distribution of both
walls is presented.
The mechanical behavior of the parts manufactured by WAAM, for
the most part, depends on the fabrication process, heat flow and thermal
gradient, resulting grain size, microstructure and the composition of the
wire. Fig. 14 shows the stress–strain curves of the horizontal and vertical
samples of both walls, obtaining similar curves to the previous literature
[3]. Making the comparison, in the vertical direction, UTS and YS values
are higher in the overlapped wall than in the oscillated one, but the
Elong. values are smaller in the overlapped wall. Nevertheless, in the
horizontal direction, this difference is not so evident. The higher UTS
and YS values in the vertical direction are a consequence of the smaller
grain size achieved in the overlapped wall. It is well known that the
small grain size increases the strength of the material by means of
fine-grain strengthening mechanisms. Dirasu et al. [3] also reported a
bigger grain size in an oscillated wall than in the single-pass wall
manufactured by the GMAW-based WAAM process. In addition, another
effect of the grain size on mechanical properties is the shortening of the
creep zone. In the zooms of Fig. 14, it can be seen that the creep zones
are larger in oscillated wall samples in both directions. Larger grain size
on the oscillation strategy results in an increase in the creep zone due to
the blocking dislocation motion on tensile testing at room temperature.
Furthermore, Table 8 highlights the average ultimate tensile strength
(UTS), yield strength (YS) and elongation percentage (Elong.) values
and its standard deviations obtained from the tensile test for both walls.
The oscillated wall values are similar to the ASTM A36 wrought values
and Ghaffari et al. [31] and Shassere et al. [29] achieved similar results Fig. 14. Stress–strain curves for (a) horizontal and (b) vertical samples.
in their thin walls manufactured by GMAW-based WAAM in ER70S-6.
4. Conclusions

The effects of the different deposition strategies were investigated in


low-carbon steel (ER70S-6) walls manufactured by GMAW-based
Table 6 WAAM. The two strategies studied were the three linear beads over­
The hardness values of the produced walls. lapping strategy and the oscillatory strategy. Different process aspects
Hardness (HV)
and parameters were compared in order to clarify the advantages and
disadvantages of each. The main aspects studied were material deposi­
Overlapped wall Oscillated wall
tion, growth per layer, heat input, fabrication time, and the micro­
Top 163 154 structure and mechanical properties. The results obtained from these
Centre 146 136
studies lead to the following conclusions:
Bottom 144 135

624
E. Aldalur et al. Journal of Manufacturing Processes 58 (2020) 615–626

Table 8
Tensile test average results and its standard deviations.
Overlapped wall Oscillated wall

UTS (MPa) YS 0.2 % (MPa) Elong. (%) UTS (MPa) YS 0.2 % (MPa) Elong. (%)

Horizontal 498 ± 8.8 368 ± 12.2 36 ± 3.5 478 ± 6.41 354 ± 13.54 38 ± 2.6
Vertical 501 ± 2.87 368 ± 4.15 32 ± 1.1 474 ± 0.94 338 ± 4.37 36 ± 2.2

• The GMAW-based WAAM process is suitable for manufacturing parts interests or personal relationships that could have appeared to influence
without defects in ER70S-6 steel employing both strategies studied. the work reported in this paper. All authors have read and agreed to the
The viability of this technology to produce large metallic parts is published version of the manuscript.
proven.
• Although more material is deposited in the oscillated wall inside the CRediT authorship contribution statement
control volume the average volumetric layer growth is lower.
Therefore, to manufacture the desired walls, more layers are needed E. Aldalur: Methodology, Software, Formal analysis, Investigation,
in the overlapped wall. Data curation, Writing - original draft, Visualization. F. Veiga:
• The average cooling time is lower in the oscillated wall because, Conceptualization, Software, Validation, Formal analysis, Resources,
although the heat input is higher in this wall, this heat is introduced Data curation, Writing - review & editing, Visualization. A. Suárez:
at once. As a result, productivity is higher in the oscillated strategy, Conceptualization, Methodology, Validation, Resources, Writing - re­
even if more layers are deposited. view & editing, Supervision, Project administration, Funding acquisi­
• Due to different thermal cycles created in both walls because of the tion. J. Bilbao: Validation, Writing - review & editing. A. Lamikiz:
different deposition sequence and path, the microstructure in the Validation, Writing - review & editing.
oscillated wall and the overlapped wall is totally different. In the
overlapped wall, each deposited bead is clearly observed, and a References
heterogeneous microstructure is obtained with intercalate areas of
polygonal ferrite and acicular ferrite. This acicular ferrite appears in [1] Queguineur A, Rückert G, Cortial F, Hascoët JY. Evaluation of wire arc additive
manufacturing for large-sized components in naval applications. Weld World 2018;
areas that the heat input of the next layer does not affect, such as the 62:259–66. https://doi.org/10.1007/s40194-017-0536-8.
deposited last layer, at the top of the wall. [2] Hejripour F, Binesh F, Hebel M, Aidun DK. Thermal modeling and characterization
• On the other hand, in the oscillated wall, the acicular ferrite also of wire arc additive manufactured duplex stainless steel. J Mater Process Technol
2019;272:58–71. https://doi.org/10.1016/j.jmatprotec.2019.05.003.
appears in the last layer for the same reason. In the remainder of the [3] Dirisu P, Ganguly S, Mehmanparast A, Martina F, Williams S. Analysis of fracture
wall, layer bands are observed, and a homogeneous polygonal ferrite toughness properties of wire + arc additive manufactured high strength low alloy
microstructure is obtained. In this case, the grain size is larger than in structural steel components. Mater Sci Eng A 2019;765:138285. https://doi.org/
10.1016/j.msea.2019.138285.
the overlapped wall due to the heat accumulation suffered. [4] Rodrigues TA, Duarte V, Miranda RM, Santos TG, Oliveira JP. Current status and
• The hardness and the microstructure are closely related properties. perspectives on wire and arc additive manufacturing (WAAM). Materials 2019;12:
As a result, higher values of hardness are obtained in the top part of 1121. https://doi.org/10.3390/ma12071121.
[5] Williams SW, Martina F, Addison AC, Ding J, Pardal G, Colegrove P. Wire + arc
both walls where the acicular ferrite microstructure is present. In
additive manufacturing. Mater Sci Technol 2016;32:641–7. https://doi.org/
comparison, the overlapped wall hardness values are greater than 10.1179/1743284715Y.0000000073.
the oscillated one because of the heterogeneity in the microstructure. [6] Pan Z, Ding D, Wu B, Cuiuri D, Li H, Norrish J. Arc welding processes for additive
• Isotropic mechanical properties are achieved in both manufactured manufacturing: a review. Trans Intell Weld Manuf 2018;1:3–24. https://doi.org/
10.1007/978-981-10-5355-9_1.
walls; namely, in the vertical and horizontal directions, the me­ [7] Artaza T, Suárez A, Murua M, Garcia J, Tabernero I, Lamikiz A. Wire arc additive
chanical properties are similar. Furthermore, in overlapped wall, manufacturing of Mn4Ni2CrMo conference steel : comparison of mechanical and
ultimate tensile strength and yield strength values are greater than metallographic properties of PAW and GMAW. Procedia Manuf 2019;41:1071–8.
https://doi.org/10.1016/j.promfg.2019.10.035.
the oscillated values because of the smaller grain size obtained. This [8] Wu B, Pan Z, Ding D, Cuiuri D, Li H, Xu J, et al. A review of the wire arc additive
is a good example of a fine-grain strengthening mechanism. manufacturing of metals: properties, defects and quality improvement. J Manuf
Process 2018;35:127–39. https://doi.org/10.1016/j.jmapro.2018.08.001.
[9] Wang L, Xue J, Wang Q. Correlation between arc mode, microstructure, and
Further studies are expected to test these strategies on large steel mechanical properties during wire arc additive manufacturing of 316L stainless
pieces to see the influence of all the parameters analyzed on more steel. Mater Sci Eng A 2019;751:183–90. https://doi.org/10.1016/j.
complex structures. In the present study, it was demonstrated that the msea.2019.02.078.
[10] Tabernero I, Paskual A, Álvarez P, Suárez A. Study on arc welding processes for
heat input is a critical parameter in WAAM due to its influence on some high deposition rate additive manufacturing. Procedia CIRP 2018;68:358–62.
aspects of the process, such as microstructure. Thus, it will be important https://doi.org/10.1016/j.procir.2017.12.095.
to analyze methods to monitor and control this parameter. Furthermore, [11] Selvi S, Vishvaksenan A, Rajasekar E. Cold metal transfer (CMT) technology—an
overview. Def Technol 2018;14:28–44. https://doi.org/10.1016/j.dt.2017.08.002.
different strategies could be detailed to redefine the obtained micro­
[12] Ma G, Zhao G, Li Z, Yang M, Xiao W. Optimization strategies for robotic additive
structure and to enhance the mechanical properties. and subtractive manufacturing of large and high thin-walled aluminum structures.
Int J Adv Manuf Technol 2019;101:1275–92. https://doi.org/10.1007/s00170-
Funding 018-3009-3.
[13] Caballero A, Ding J, Ganguly S, Williams S. Wire + arc additive manufacture of 17-
4 PH stainless steel: effect of different processing conditions on microstructure,
The authors acknowledge the Basque Government for financing the hardness, and tensile strength. J Mater Process Technol 2019;268:54–62. https://
PROCODA project, ELKARTEK 2019 program (KK-2019/00004), HAR­ doi.org/10.1016/j.jmatprotec.2019.01.007.
[14] Xu X, Ding J, Ganguly S, Diao C, Williams S. Preliminary investigation of building
ITIVE, HAZITEK 2017 program (ZE-2017/00038), HARIPLUS, HAZITEK strategies of maraging steel bulk material using wire + arc additive manufacture.
2019 program (ZL-2019/00352) and QUALYFAM project [kk-2020/ J Mater Eng Perform 2018;28:594–600. https://doi.org/10.1007/s11665-018-
00042]. 3521-5.
[15] Wu B, Pan Z, Ding D, Cuiuri D, Li H. Effects of heat accumulation on microstructure
and mechanical properties of Ti6Al4V alloy deposited by wire arc additive
Declaration of competing interest manufacturing. Addit Manuf 2018;250:304–12. https://doi.org/10.1016/j.
addma.2018.08.004.
The authors declare that they have no known competing financial

625
E. Aldalur et al. Journal of Manufacturing Processes 58 (2020) 615–626

[16] Li C, Gu H, Wang W, Wang S, Ren L, Wang Z. Effect of heat input on formability, [25] Cunningham CR, Flynn JM, Shokrani A, Dhokia V, Newman ST. Invited review
microstructure, and properties of Al–7Si–0.6Mg alloys deposited by CMT-WAAM article: strategies and processes for high quality wire arc additive manufacturing.
process. Appl Sci 2020;10:70. https://doi.org/10.3390/app10010070. Addit Manuf 2018;22:672–86. https://doi.org/10.1016/j.addma.2018.06.020.
[17] Oliveira JP, Santos TG, Miranda RM. Revisiting fundamental welding concepts to [26] Choi CL, Hill DC. A study of microstructural progression in As-Deposited weld
improve additive manufacturing: from theory to practice. Prog Mater Sci 2020; metal. Weld J 1978;57:232.
107:100590. https://doi.org/10.1016/j.pmatsci.2019.100590. [27] Zhang Y, Chen Y, Li P, Male AT. Weld deposition-based rapid prototyping: a
[18] ADDILAN WAAM Technology Machines. https://www.addilan.com/. preliminary study. J Mater Process Technol 2003;135:347–57. https://doi.org/
[19] Zhang Y, Chen Y, Li P, Male AT. Automated system for welding-based rapid 10.1016/S0924-0136(02)00867-1.
prototyping. Mechatronics 2002;12(1):37–53. [28] Babu SS. The mechanism of acicular ferrite in weld deposits. Curr Opin Solid State
[20] Aldalur E, Veiga F, Suárez A, Bilbao J, Lamikiz A. Analysis of the wall geometry Mater Sci 2004;8:267–78. https://doi.org/10.1016/j.cossms.2004.10.001.
with different strategies for high deposition wire arc additive manufacturing of [29] Shassere B, Nycz A, Noakes MW, Masuo C, Sridharan N. Correlation of
mild steel. Metals 2020;10:892. https://doi.org/10.3390/met10070892. microstructure and mechanical properties of metal big area additive
[21] Kah P, Suoranta R, Martikainen J. Advanced gas metal arc welding processes. Int J manufacturing. Appl Sci (Switzerland) 2019;9:787. https://doi.org/10.3390/
Adv Manuf Technol 2013;67:655–74. https://doi.org/10.1007/s00170-012-4513- app9040787.
5. [30] Haden CV, Zeng G, Carter FM, Ruhl C, Krick BA, Harlow DG. Wire and arc additive
[22] Normalización Española, Materiales Metálicos. Ensayo de dureza Vickers. Parte 1: manufactured steel: tensile and wear properties. Addit Manuf 2017;16:115–23.
Método de ensayo. UNE- EN ISO 6507-1. October 2018. https://doi.org/10.1016/j.addma.2017.05.010.
[23] Normalización Española, Materiales Metálicos. Ensayo de tracción. Parte 1: [31] Ghaffari M, Vahedi Nemani A, Rafieazad M, Nasiri A. Effect of solidification defects
Método de ensayo a temperatura ambiente. UNE-EN ISO 6892-1. February 2017. and HAZ softening on the anisotropic mechanical properties of a wire arc additive-
[24] Rodrigues TA, Duarte V, Avila JA, Santos TG, Miranda RM, Oliveira JP. Wire and manufactured low-carbon low-alloy steel part. JOM Miner Met Mater Soc 2019;71:
arc additive manufacturing of HSLA steel: effect of thermal cycles on 4215–24. https://doi.org/10.1007/s11837-019-03773-5.
microstructure and mechanical properties. Addit Manuf 2019;27:440–50. https://
doi.org/10.1016/j.addma.2019.03.029.

626

You might also like