The International Journal of Advanced Manufacturing Technology

https://doi.org/10.1007/s00170-018-2927-4

ORIGINAL ARTICLE

Effect of in-process active cooling on forming quality and efficiency

of tandem GMAW–based additive manufacturing
Junbiao Shi 1,2 & Fang Li 1,2 & Shujun Chen 1,2 & Yun Zhao 1,2 & Hongyu Tian 1,2

Received: 10 May 2018 / Accepted: 23 October 2018

# Springer-Verlag London Ltd., part of Springer Nature 2018

Abstract
Wire arc additive manufacturing (WAAM), utilizing welding arc to melt metal wire into shaped parts, has become a promising
manufacturing technology recently. Tandem GMAW–based WAAM (TG-WAAM), in which two wires are fed into the molten
pool simultaneously, has the potential to double the efficiency of traditional WAAM. However, the high wire-feed speed is
accompanied with high heat input that is likely to cause molten pool overflowing, especially at upper layers because of decreased
heat dissipation and increased heat accumulation. An in-process active cooling technology based on thermoelectric cooling is
introduced into TG-WAAM in this research. Its effect on forming quality and efficiency of TG-WAAM is investigated exper-
imentally. The results show that the additional cooling well compensates for the excessive heat input into the molten pool, which
enables not only increased maximum wire-feed speed (9–15%) but also reduced inter-layer dwell time (42–54%), while main-
taining the desired forming quality. The overall efficiency is improved by more than 0.97 times in the case study. This research
provides a feasible scheme to solve the conflict between forming quality and efficiency during WAAM.

Keywords Wire arc additive manufacturing . Tandem GMAW . Active cooling . Forming quality . Efficiency

1 Introduction their high cost and low efficiency (typically 0.12 kg/h for Ti–
6Al–4V) greatly restrict their application in fabricating large
Additive manufacturing (AM), which fabricates components metal components [4, 5].
in a layer-by-layer pattern, has become a promising In contrast, wire arc additive manufacturing (WAAM), uti-
manufacturing technology with huge growth potential in the lizing welding arc to melt metal wire into shaped parts, well
past 30 years [1]. It offers much higher geometrical flexibility overcomes the obstacle above [6]. The maximum deposition
and greater potential of time and cost savings compared with rate can reach up to 2.6 kg/h for Ti–6Al–4V (much higher than
conventional manufacturing technologies. Recently, AM for SLM and EBM) and the cost of metal wire is about 1/10 of
functional metal components, instead of plastic prototypes, metal powder (much lower than SLM and EBM) [4, 7]. These
has become a hot topic in order to meet the demanding re- characteristics make WAAM a good option to fabricate large
quirements of the aerospace industries. For any metallic AM metal components among various AM technologies [8]. For
technology, the basic idea is to utilize a high energy source to example, Cranfield University applied WAAM to fabricate a
melt metal material into shaped parts [2]. Laser and electron 1.2-m-long spar part for an airplane wing and Beijing
beam–based AM processes are commonly used [3]. However, University of Technology applied it to fabricate an integrally
stiffened panel [9–11]. Different WAAM processes have been
developed to meet different geometrical requirements, includ-
* Fang Li ing GMAW-based, GTAW-based, and PAW-based processes.
[email protected] For example, CMT, a modified GMAW, is suitable to produce
thin-walled structures, while PAW is suitable to produce wall
1
College of Mechanical Engineering and Applied Electronics structures of large width [12].
Technology, Beijing University of Technology, Beijing 100124, Efficiency is crucial for WAAM when large metal compo-
China
nents are concerned. In order to obtain higher efficiency, the
2
Engineering Research Center of Advanced Manufacturing wire-feed speed (WFS) should be increased to deposit more
Technology for Automotive Components-Ministry of Education,
Beijing University of Technology, Beijing 100124, China
material per unit time and so is the welding current to increase
 Int J Adv Manuf Technol

Fig. 1 a IGM Robot system for

TG-WAAM. b, c Thermoelectric
cooler

the melting energy. However, this will inevitably bring extra single special torch simultaneously along with the direction of
heat into the molten pool that is directly coupled with the heat travel, whereas for twin-wire GTAW two wires are fed from
for melting the metal wire [13]. Along with high heat input are two separate wire-feed systems [17]. They provide the possi-
reduced viscosity of the material, large arc force, and strong bility to double the efficiency of traditional single-wire
droplet impingement, which are likely to lead to molten pool WAAM. However, multiple wire processes require much
overflowing, making it difficult to obtain the desired forming higher heat input than single wire processes to increase the
quality. As the deposition height increases, the heat dissipation melting energy, which means that the molten pool shape and
condition will become worse and the heat accumulation will the forming quality are more difficult to control. Previous
become more serious. This is because in this case the heat is studies mainly focus on the application of multiple wire pro-
dissipated not only to the substrate through conduction but cesses to the production of intermetallic and functionally grad-
also to the air through convection and radiation. However, ed materials, but few focus on how to realize the full potential
the latter is less effective than the former [14]. These reasons of their high WFS [18].
add more difficulty to control the shape of the molten pool and During WAAM, the molten pool shape is affected by not
ultimately the forming quality. Therefore, the forming quality only the heat input but also the heat dissipation. Though the
and efficiency conflict each other during WAAM. It is chal- heat input is high, it is still possible to well control the molten
lenging to achieve a high efficiency while maintaining the pool shape by enhancing the heat dissipation via additional
desired forming quality [15, 16]. cooling to compensate for the excessive heat input. This pro-
An alternative way to increase the efficiency is to employ vides the possible to solve the conflict between forming quality
multiple wire processes, such as tandem GMAW and twin- and efficiency during WAAM [19]. Fayolle applied cryogenic
wire GTAW. For tandem GMAW, two wires are fed from a liquid argon cooling to prevent heat accumulation during

Table 1 Experiment design

Group no. Heat dissipation condition WFS (m/min) TS (cm/min)

I At the lower layer without cooling 2, 3, 4, 5, 6, 7 20, 30, 40, 50, 60, 70, 80, 90, 100
II At the upper layer without cooling 2, 3, 4, 5, 6, 7 20, 30, 40, 50, 60, 70, 80, 90, 100
III At the upper layer with cooling 2, 3, 4, 5, 6, 7 20, 30, 40, 50, 60, 70, 80, 90, 100
Int J Adv Manuf Technol

Fig. 2 a Deposition on a flat

substrate. b Deposition on an
existing wall without cooling. c
Deposition on an existing wall
with cooling

WAAM and the efficiency was proved to be tripled. Henckell 2 Methodology

presented that gas cooling improved the buildup condition that
enabled a continuous manufacturing process without heat ac- 2.1 Experimental setup
cumulation in the top layers [20, 21]. However, the air cooling
setup should not be too close to the molten pool because the Figure 1a shows the experimental setup to implement TG-
stability of the welding arc is likely to be disrupted. Therefore, it WAAM. An IGM industrial robot is used to control the
affects the molten pool mainly during the period after it is movement of the torch. Two separate wire-feed units
solidified that contributes to reducing the inter-layer dwell time, along with two Fronius power supplies are used to control
but has limited effect on the shape of the molten pool before it is the two wires, which are fed through the same torch and
solidified. To address this issue, this research introduces in- are placed in-line with each other along the direction of
process active cooling into TG-WAAM based on the thermo- the travel. The power supplies are operated under the
electric technology, which can be quite close to the molten pool. pulse transfer mode. The substrate is high-strength alumi-
With the help of the strong heat conduction of the thermoelec- num alloy 2219 (6.3% Cu), which has wide application in
tric cooling technology, the low heat dissipation rate via con- the aerospace industry [22]. The wire electrode is alumi-
vection and radiation is enhanced significantly. Therefore, it has num alloy 2325 with similar chemical composition as
significant effect on the molten pool both before and after it is 2219 and a 1.2-mm diameter.
solidified, thereby leading to both increased WFS and reduced Two thermoelectric coolers are employed for in-process
inter-layer dwell time. The effect of in-process active cooling on active cooling, which are partly placed symmetrically to-
the forming quality and efficiency of TG-WAAM is investigat- wards the two sides of the wall structure, as shown in
ed in the rest of this paper. Fig. 1b. The basic function of a thermoelectric cooler is

Fig. 3 a Process parameter area of various forming appearances: b forming appearances at the lower layer (group I); c forming appearances at the upper
layer (group II)
 Int J Adv Manuf Technol

to transfer heat from the hot side to the cold side, con-
suming electrical energy. Then the excessive heat input
into the molten pool can be transferred to the ambient
environment quickly. Compared with gas or liquid
cooling methods, the primary advantages of thermoelec-
tric coolers include no leakage, flexible shape, long life,
and controllable cooling rate [23]. For the thermoelectric
cooler used in this study, its cooling power can be easily
controlled through adjusting the DC voltage [24].
Preliminary experiments show that if the thermoelectric
cooler does not contact well with the side of the wall,
given that the wall surface is not flat due to the stair-
stepping effect, the cooling effect is not obvious. To ad-
dress this issue, a silicone rubber with high thermal con-
ductivity is employed to guarantee good contact between
the thermoelectric cooler and the wall. Its thickness is
2 mm, as shown in Fig. 1c. This cooling setup can also
move in the Z direction as the deposition height increases,
driven by a step motor.

2.2 Experimental design

Several experiments were conducted to explore the effect

of different heat dissipation conditions (group I–group III)
on forming quality and efficiency of TG-WAAM under
various process parameters, as given in Table 1. Travel
speed (TS) and wire-feed speed (WFS) were selected as
the main influencing factors. WFS was varied from 2 to
7 m/min with an interval of 1 m/min and TS was varied
from 20 to 100 cm/min with an interval of 10 cm/min.
Different combination of WFS and TS generated different
forming quality and efficiency.
The experiments were conducted on a flat substrate
and an existing wall in group I and group II/III respec-
tively to reflect the heat dissipation condition at different
heights. The flat substrate in group I has dimensions of
150 mm (width) × 150 mm (length) × 6 mm (height) (see
Fig. 2a) and the existing wall in group II and III has the
same dimensions as the flat substrate (see Fig. 2b, c).
From these experiments, we can analyze (1) the effect
of deposition height on forming quality and efficiency
of TG-WAAM by comparing group I and group II and
(2) the effect of the additional cooling on forming quality
and efficiency of TG-WAAM by comparing groups II
and III.
In these experiments, the molten pool was detected through
a high-speed camera MotionPro Y4S1. The bead geometry
after the solidification of the molten pool, characterized by
bead width and bead height, was measured through a laser
Fig. 4 Bead width contour a at the lower layer without cooling (group I), displacement scanner. The thermal cycling was recorded
b at the upper layer without cooling (group II), and c at the upper layer through a thermocouple of type K, which was mounted at a
with cooling (group III)
depth of 5 mm from the upper surface.
Int J Adv Manuf Technol

Fig. 5 Forming appearance and

molten pool shape: a at the upper
layer without cooling (group II)
and b at the upper layer with
cooling (group III)

3 Results and discussions However, the molten pool contains excessive heat that easily
results in overflowing and wrinkles. In region III with rela-
3.1 Comparison of forming appearance in the three tively large WFS and large TS, the bead width and height is
cases likely to change irregularly, even producing scallops and
humps. This can be explained by the strong droplet impinge-
Based on the experimental results (according to Table 1), the ment and the large arc force. In region IV, the forming appear-
relation between the forming appearance with the process pa- ance is quite excellent with no obvious defects and spatters
rameters under different heat dissipation conditions was and therefore is suitable for WAAM.
established, as shown in Fig. 3. The contour plot was also By comparing the process windows of group I and group
generated to represent graphically the relation between the II, it can be concluded that the process window obtained at the
bead width and the process parameters, as shown in Fig. 4. lower layer is larger than that obtained at the upper layer, as
The process window, which reflects the feasible parameter shown in Fig. 4a, b. At the high heat input region, the forming
ranges that allows good forming appearance, is confined to appearance is excellent at the lower layer as shown in the
the area where WFS/TS is between 5 and 20 basically. For any group I, but is poor at the upper layer as shown in the group
of the three cases, the process parameter area can be divided II. This result indicates that the heat dissipation condition of
into four regions, which differ in their heat input. As is known, the molten pool becomes worse with the increase of the depo-
the heat input per unit length increases with the increase of the sition height. Besides, the constraint of the base for the molten
welding current and the welding voltage, and the decrease of pool is relatively weak at the upper layer. Therefore, the mol-
the TS. Because the control system is operated under the uni- ten pool overflowing is more likely to occur when the heat
fied parameter regulation control mode, the welding current is input is high. In other regions, though both their forming ap-
an increasing function of the WFS. In region I with relatively pearances are satisfactory, their bead width contours differ a
small WFS and large TS, the heat input is quite small. The lot. The contour line at the upper layer is shifted towards to the
welding arc is not stable and the weld toe is incompletely lower-right direction compared with that at the lower layer,
melted. In region II with relatively large WFS and small TS, which implies that the bead tends to be wider under the pro-
the heat input is large enough to spread the molten pool. cess parameters. This is because at lower layers with relatively

Fig. 6 Microstructures a with

active cooling and b without
active cooling
 Int J Adv Manuf Technol

Table 2 Maximum wire-feed speed (deposition rate) versus bead width

Bead width

8 mm 10 mm 12 mm 14 mm

DR (kg/h) WFS (m/min) DR (kg/h) WFS (m/min) DR (kg/h) WFS (m/min) DR (kg/h) WFS (m/min)

Without cooling 1.83 5 1.97 5.5 2.12 5.8 2.03 6

With cooling 2.0 5.5 2.19 6.2 2.34 6.4 2.38 6.5
Improvement 10% 10% 11% 11% 9% 9% 15% 15%

fast heat dissipation, the molten pool solidifies quickly with- exhibits a cone shape. This proves that with active cooling it is
out enough time to form a wide bead and therefore the bead possible to eliminate the impact of excessive heat input on the
width tends to be smaller. At upper layers with relatively slow molten pool shape and ultimately the forming quality.
heat dissipation, the viscosity of the molten pool material is Further, how active cooling affects the microstructures of
decreased and therefore the liquid metal tends to spread more additively manufactured parts was investigated. The micro-
easily, which leads to a larger bead width. The average in- structures in the middle of the beads from Fig. 5 were charac-
crease of the bead width is 2 mm. terized by dendrite grains, as shown in Fig. 6. The average
In group III with in-process active cooling, the process grain size was 8.85 μm when active cooling was applied,
window is enlarged and the contour line is shifted towards compared to 16.53 μm obtained without active cooling. This
to the upper-left direction compared with the situation in is because active cooling generates a much larger heat dissi-
group II, as shown in Fig. 4c. The bead width decreases ob- pation rate that helps increase the number of particles and
viously (about 2 mm) with the same process parameters. This therefore refine the grain size.
proves that the active cooling method is more effective than
traditional convection and radiation, which leads to quicker
solidification of the liquid molten material and therefore 3.2 Comparison of deposition rate in the three cases
smaller bead width. It is interesting to observe that the process
window and the contour line in group III are more close to This contour plot above gives an instruction about how to
those in group I, which implies that the heat dissipation of the select appropriate WFS and TS to obtain the desired bead
upper layer might reach the same level as that of the lower width. For a certain bead width, we can obtain different com-
layer with the aid of active cooling. Figure 5 compares the bination of WFS and TS. From the viewpoint of high efficien-
molten pool shapes and the corresponding forming appear- cy, the WFS should be as high as possible because the depo-
ances obtained in groups II and III under the same parameters sition rate (DR) is determined by Eq. (1) as follows:
of WFS = 5.5 m/min and TS = 37 cm/min (at the high heat
input region). Without cooling, the molten pool flows ran- ρπd 2
domly with interactions of various forces and the edge has a DR ¼  WFS ð1Þ
4
tendency to collapse. The remelting of the substrate (previous
layer) is also obvious as shown in Fig. 5a. This phenomenon The maximum allowable WFS as well as the maximum DR
disappears when active cooling is applied and the molten pool without and with active cooling is obtained from Fig. 4, which

Fig. 7 Comparison of thermal

cycling curves with and without
cooling
Int J Adv Manuf Technol

Table 3 Inter-layer dwell time versus inter-layer temperature

Inter-layer temperature 50 °C 75 °C 100 °C

Cooling time without cooling 180 s 131 s 70 s
Cooling time with cooling 91 s 60 s 40 s
Reduction − 50% − 54% − 42%

is a function of the bead width as summarized in Table 2.

Active cooling allows enlarged process window and therefore
more process parameters can be chosen for a certain bead
width. The maximum deposition is improved by 9–15% for
the bead width of 8–14 mm.

3.3 Comparison of inter-layer dwell time in the three

cases

Heat accumulation is another factor that significantly affects

the molten pool shape and the forming quality [25]. To avoid
this, the inter-layer temperature should be well controlled
through setting the appropriate inter-layer dwell time.
Figure 7 compares the thermal cycling curves without and
with cooling measured by thermocouples, from which we
can obtain the peak temperature and the inter-layer dwell time
as a function of the inter-layer temperature. With active
cooling, the peak temperature is decreased obviously from
415 to 353 °C and the cooling rate is increased significantly.
As a consequence, the inter-layer dwell time is decreased by Fig. 8 Wall structures when: a WFS = 3 m/min and TS = 30 cm/min
42–54% for the inter-layer temperature of 50–100 °C, as given without cooling, b WFS = 5 m/min and TS = 50 cm/min with cooling, c
in Table 3. WFS = 5 m/min and TS = 50 cm/min without cooling
From the above analysis, we can conclude that active
cooling has significant effect on the molten pool both before
WFS and TS were 3 m/min and 30 cm/min respectively without
and after it is solidified. Therefore, it allows for increased
cooling (case a), which took 3.3 min in deposition time and
efficiency through two ways. One is to increase the maximum
54 min in cooling time. If active cooling was applied, WFS
deposition rate (9–15%) and the other is to decrease the inter-
and TS could be increased to 5 m/min and 50 cm/min respec-
layer dwell time (42–54%).The overall improvement of the
tively while maintaining the same bead width (case b), which
efficiency depends on the actual case.
took 2 min in deposition time and 27 min in cooling time. The
decrease of the cooling time is more significant than that of the
3.4 Case study deposition time. The overall efficiency is improved by 0.97
times. The third case employed the same process parameters
To further verify the benefits of active cooling, three walls were as the second case but the active cooling was not applied. The
fabricated with different process parameters, which were obtained bead width was about 12 mm and the required total
100 mm long and were made of 10 layers, as given in time was 56 min, which means lower material utilization and
Table 4. The inter-layer temperature was retained at room tem- lower efficiency compared with the case with cooling. The
perature. In order to build a wall of 10-mm width, the required advantage of active cooling is clearly demonstrated (Fig. 8).

Table 4 Parameter settings and results in the case study

Case no. WFS (m/min) TS (cm/min) Width (mm) Height (mm) Deposition time (min) Cooling time (min) Total time (min) Cooling condition

a 3 30 10 22 3.3 54 57.3 Without

b 5 50 10 22 2 27 29 With
c 5 50 12 18 2 54 56 Without
 Int J Adv Manuf Technol

4 Conclusions 7. Ding D, Pan Z, Cuiuri D, Li H (2015) Wire-feed additive

manufacturing of metal components: technologies developments
and future interests. Int J Adv Manuf Technol 81(1–4):465–481
This paper proposes an in-process active cooling technology 8. Ding D, Pan Z, Cuiuri D, Li H (2015) A practical path planning
to compensate for the excessive heat input associated with the methodology for wire and arc additive manufacturing of thin-
high WFS during TG-WAAM. Its effect on forming quality walled structures. Robot CIM-INT Manuf 34(C):8–19
9. Zhan Q, Liang Y, Ding J, Williams S (2016) A wire deflection
and efficiency is investigated. The conclusions are as follows:
detection method based on image processing in wire + arc additive
manufacturing. Int J Adv Manuf Technol 89(1–4):1–9
(1) The heat dissipation condition becomes worse with the 10. Li F, Chen S, Shi J, Zhao Y (2018) In-process control of distortion
increase of the deposition height, which confines the in wire and arc additive manufacturing based on a flexible multi-
process window and is likely to cause molten pool point support fixture. Sci Technol Weld Joi. https://doi.org/10.1080/
13621718.2018.1476083
overflowing. 11. Li F, Chen S, Shi J, Tian H, Zhao Y (2017) Evaluation and optimi-
(2) Thermoelectric technology–based active cooling con- zation of a hybrid manufacturing process combining wire arc addi-
tributes to enhancing the heat dissipation of upper layers tive manufacturing with milling for the fabrication of stiffened
that leads to larger process window and better control of panels. Appl Sci 7(12):1233
12. Xiong J, Lei Y, Chen H, Zhang G (2017) Fabrication of inclined
the molten pool shape. The bead tends to be narrower thin-walled parts in multi-layer single-pass GMAW-based additive
and the average decrease is about 2 mm. manufacturing with flat position deposition. J Mater Process
(3) With the assist of active cooling, the maximum WFS can Technol 240:397–403
be increased by 9–15% and the inter-layer dwell time can 13. Xiong J, Zhang G, Zhang W (2015) Forming appearance analysis
in multi-layer single-pass GMAW-based additive manufacturing.
decreased by 42–54% while maintaining the desired Int J Adv Manuf Technol 80(9–12):1767–1776
forming quality. 14. Geng H, Li J, Xiong J, Lin X (2016) Optimisation of interpass
(4) The case study shows that the overall efficiency is im- temperature and heat input for wire and arc additive manufacturing
proved by more than 0.97 times when active cooling is 5A06 aluminium alloy. Sci Technol Weld Joi 22(6):472–483
15. Xiong J, Zhang G, Hu J, Wu L (2014) Bead geometry prediction for
applied attributed to both increased maximum WFS and
robotic gmaw-based rapid manufacturing through a neural network
decreased inter-layer dwell time. and a second-order regression analysis. J Intell Manuf 25(1):157–
163
Funding information This work was supported by the National Natural 16. Xiong J, Yin Z, Zhang W (2016) Closed-loop control of variable
Science Foundation of China (no. 51805013) and Foundation Research layer width for thin-walled parts in wire and arc additive
Fund of Beijing University of Technology (no. 001000546318526). manufacturing. J Mater Process Technol 233:100–106
17. Ye D, Hua X, Wu Y (2013, 2013) Arc interference behavior during
Publisher’s Note Springer Nature remains neutral with regard to jurisdic- twin wire gas metal arc welding process. Adv Mater Sci Eng:
tional claims in published maps and institutional affiliations. 937094
18. Sproesser G, Chang YJ, Pittner A, Finkbeiner M, Rethmeier M
(2017) Environmental energy efficiency of single wire and tandem
gas metal arc welding. Weld World 61(4):733–743
References 19. Li F, Chen S, Shi J, Zhao Y, Tian H (2018) Thermoelectric cooling-
aided bead geometry regulation in wire and arc-based additive
1. Thompson MK, Moroni G, Vaneker T, Fadel G, Campbell RI, manufacturing of thin-walled structures. Appl Sci 8(2):207
Gibson I, Bernard A, Schulz J, Graf P, Ahuja B, Martina F (2016) 20. Ding J (2012) Thermo-mechanical analysis of wire and arc additive
Design for additive manufacturing: trends, opportunities, consider- manufacturing process. Dissertation, Cranfield University
ations, and constraints. CIRP Ann Manuf Technol 65(2):737–760 21. Henckell P, Günther K, Ali Y, Bergmann JP, Scholz J, Forêt P
2. Li F, Chen S, Wu Z, Yan Z (2018) Adaptive process control of wire (2017) The influence of gas cooling in context of wire arc additive
and arc additive manufacturing for fabricating complex-shaped manufacturing—a novel strategy of affecting grain structure and
components. Int J Adv Manuf Technol 96(1–4):871–879 size. TMS 2017 146th Annual Meeting & Exhibition
3. ZhongY RLE, Wikman S, Koptyug A, Liu L, Cui D, Shen ZJ Supplemental Proceedings. Springer International Publishing
(2017) Additive manufacturing of ITER first wall panel parts by 22. Cong B, Qi Z, Qi B, Sun H, Zhao G, Ding J (2017) A comparative
two approaches: selective laser melting and electron beam melting. study of additively manufactured thin wall and block structure with
Fusion Eng Des 116:24–33 al-6.3%cu alloy using cold metal transfer process. Appl Sci 7(3):
4. Martina F (2014) Investigation of methods to manipulate geometry, 275
microstructure and mechanical properties in titanium large scale 23. Meng JH, Wang XD, Zhang XX (2013) Transient modeling and
wire+arc additive manufacturing. Dissertation, Cranfield dynamic characteristics of thermoelectric cooler. Appl Energy
University 108(5):340–348
5. Baufeld B, Biest OVD, Gault R (2010) Additive manufacturing of 24. Zhao D, Tan G (2014) A review of thermoelectric cooling: mate-
Ti–6Al–4V components by shaped metal deposition: microstruc- rials modeling and applications. Appl Therm Eng 66(1–2):15–24
ture and mechanical properties. Mater Des 31(1):S106–S111 25. Wu B, Ding D, Pan Z, Cuiuri D, Li H, Han J, Fei ZY (2017) Effects
6. Xu X, Ding J, Ganguly S, Diao C, Williams S (2017) Oxide accu- of heat accumulation on the arc characteristics and metal transfer
mulation effects on wire + arc layer-by-layer additive manufacture behavior in wire arc additive manufacturing of Ti6Al4V. J Mater
process. J Mater Process Technol 252:739–750 Process Technol 250:304–312