Materials and Design 66 (2015) 169–175

Contents lists available at ScienceDirect

Materials and Design

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

Effect of pulse current on mechanical properties and dendriticmorphology of modified medium manganese steel welds metal

http://dx.doi.org/10.1016/j.matdes.2014.10.0500261-3069/� 2014 Elsevier Ltd. All rights reserved.

⇑ Corresponding author at: School of Materials Science and Engineering, TianjinUniversity, No. 92 Weijin Road, Tianjin 300072, China. Tel.: +86 22 27405889; fax:+86 22 27405889.

E-mail address: dixinjie@tju.edu.cn (X. Di).

Xinjie Di a,b,⇑, Shengjie Deng a,b, Baosen Wang c

a Tianjin Key Laboratory of Advanced Joining Technology, Tianjin University, Tianjin 300072, Chinab School of Materials Science and Engineering Tianjin University, Tianjin 300072, Chinac Baosteel Research Institute, Baoshan Iron & Steel Co., Ltd., Shanghai 200431, China

a r t i c l e i n f o

Article history:Received 18 July 2014Accepted 17 October 2014Available online 27 October 2014

Keywords:Modified medium manganese steelPulse-gas metal arc weldingSide bend testDendritic arm spacingMicrosegregation

a b s t r a c t

Modified medium manganese steel (MMMS) samples were joined using gas metal arc welding (GMAW)and pulse-GMAW (P-GMAW) techniques. The joints were examined using optical microscope, scanningelectron microscope, hardness tests, tensile tests and side bend tests. The use of P-GMAW was foundto be superior to the GMAW process, resulting in a noteworthy enhancement of the plastic deformationcapacity of the weld joint while maintaining comparable tensile properties. Microstructural study andmeasurement of primary and secondary dendrite arm spacing were also performed to better understandthe important aspects of weld metal solidification, i.e., pulse current. The treatment with pulse currentrestrained the dendrite growth in the welds, resulting in finer dendritic grains, which improved the duc-tility of the weld joint. The refinement of the microstructure can be attributed to the application of thepulse current, which intensified the effective vibration of the molten pool, facilitated the diffusion ofthe alloy and reduced the microsegregation.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction

High manganese austenitic steel and its excellent work-harden-ing properties during high-impact use have been a major area ofstudy in the field of wear resistance over the past century eversince the English metallurgist Sir Robert Hadfield inventedHadfield’s steel in 1882 [1], and studies on welding this type ofsteel have been progressively emerging since that time [2–5]. In1963, to meet the requirements of low-impact applications, theAmerican Metal Climax company introduced modified mediummanganese wear-resistant steel (MMMS) [6], which also had anaustenitic structure but with lower stability, and this propertyleads to better wear resistance performance under low stress abra-sive wear conditions via the formation of numerous strain-inducedmartensites and twins [7].

The emergence of new types of steel is often accompaniedwith corresponding weldability problems. Usually, certain issues,such as hot cracking and coarse columnar grains, which appearin the welds of austenitic steel, will also arise in the welds of

MMMS because of the similar austenitic structure [8]. To obtaina weld deposit with better crack resistance, the electrodes orwires used for welding austenitic manganese steel should beaustenitic with high nickel or molybdenum contents when nec-essary [2,5]. In addition, typical issues, such as intergranularbrittleness due to the carbide precipitation in heat-affected zone(HAZ) during welding of austenitic manganese steel, will alsooccur [2].

Unfortunately, there are few works that have reported thewelding of MMMS despite the fact that this type of steel is widelyused; jointing MMMS is extremely urgent. In general, gas metal arcwelding (GMAW) can obtain a low heat input if a large current, finewire, and high-speed welding can be used, and thus, a smaller HAZwidth and carbide precipitation tendency could be achieved. Mostimportantly, this technique improves productivity. To obtain grainrefinement of the solidification structure of a weld and a betterHAZ performance, different dynamic grain refining techniques,including arc oscillation and weaving, ultrasonic vibration andweld stirring using magnetic force, have been applied to fusionwelding [9–11].

Pulse welding can effectively control the heat input, leading toless distortion and improved quality, which is especially impor-tant for welding highly heat-sensitive alloy materials [12–15].However, the pulse applied to the welds of MMMS is provided

Table 1Chemical composition of the base metal and weld metal (wt.%).

Element Base metal Weld metal

C 0.98 0.18Mn 6.5 6.3Cr 1.2 16.8

170 X. Di et al. / Materials and Design 66 (2015) 169–175

less. Two different GMAW processes were presented to study therole of current pulsation on the weld solidification structure ofMMMS and its corresponding mechanical properties. This paperreports and discusses the results of this experimental investiga-tion, and sound welds and improvements are expected with pulsewelding.

Ni – 7.2Si 0.19 0.58Mo 2.9 0.11V 0.2 –Nb 0.006 –P 0.003 0.006S 0.002 0.002

Fig. 1. Microstructure of the base metal.

2. Experimental details

The chemical compositions of the MMMS and weld metal arelisted in Table 1. The base metal and filler metal have different ele-ments and elemental contents, and the P and S contents are verylow. The microstructure of MMMS, which is shown in Fig. 1, isaustenitic at room temperature. The test plates used in this inves-tigation are divided into two groups, with dimensions of 500 mmin length, 220 mm in width and 30 mm in thickness. The detailsof the weld joint are given in Fig. 2. Both are welded using G18–8 Mn wire, which is 1.2 mm in diameter, according to BS EN12072–2000 [16].

A shielding gas of 80% Ar + 20% CO2 was used, and the gas flowrate was 15–20 L�min�1. Weld deposition was performed at a60 cm�min�1 travel speed and without any preheating or post-weldheat treatment. The feed speed was 5 m�min�1. The group withoutadding pulse was designated GMAW, while the other group, whichhad applied pulse current, was labeled P-GMAW. After earlyattempts, an appropriate set of square wave pulse parameters wasused to obtain good bead appearance. During the welding process,the welding current and voltage were collected using a data acquisi-tion card PCI8622. The average voltage and current in the GMAWprocess, which were calculated using the collected data, were 25 Vand 224 A, respectively. Fig. 3 is a pulse current waveform collectedfrom the actual welding process in 100 ms. The peak current wasapproximately 300 A, and the background current was approxi-mately 70 A. The main parameters of the rectangular current pulseare given in Fig. 4. The joint distortion was kept to a minimum byalternating the successive weld passes between the two sides ofthe double-V-groove weld. The number of GMAW weld passes was19, and for P-GMAW, this number was 26.

Metallographic samples were truncated from the two welds formicrostructure observation. Standard metallographic techniqueswere applied to prepare the samples. The welds were subjectedto microscopic examination using an optical microscope and ascanning electron microscope (SEM) equipped with an energy dis-persive X-ray detector (EDX) spectrometer. This analysis was per-formed at 20 kV. Then, the alloy contents were measured to studythe intergranular and intragranular microsegregation. Themechanical tests for the welding of specimens were performedaccording to the ISO standards [17–19]. The microhardness testswere performed under a 10 kg load.

Fig. 2. Details of the double-V-groove butt joint.

3. Results and analysis

3.1. Assessment of parameters related to solidification

The HAZ of austenitic manganese steel will become brittlewhen subjected to extensive heating due to the temperaturedependent carbide precipitation and perlite formation in theaustenitic structure [20]. Therefore, the suggested maximum heatinput per unit length of the weld, q, should be less than19.2–23.0 kJ cm�1, considering the arc efficiency of gas metal arcwelding, and the heat input can be calculated using the followingequation [21]:

q ¼ gIUm¼ g

m�

Xn

i¼1

IiUi

nð1Þ

where q is the heat input in kJ�cm�1, I is the welding current in A; Uis the welding voltage in V; v is the welding speed in cm�s�1; g is thearc efficiency, for gas metal arc welding, g = 0.75–0.9 [10], n is the

total counts, andPni¼1

IiUi is the total power during the counting

period.In this investigation, the heat input values calculated using the

resulting voltage and current data corresponding to GMAW andP-GMAW were 4.19–5.03 kJ�cm�1 and 4.38–5.27 kJ�cm�1, respec-tively, which are far less than 19.2–23.0 kJ�cm�1, and acceptable.

3.2. Optical microscopy

Welds with good bead appearance and macroscopic morpholo-gies were obtained, and no solidification cracking was foundbecause the amount of sulfur and phosphorus was very low andthe manganese content was adequate; thus, the segregation of alow melting eutectic phase could be suppressed [8]. The austenitic

Fig. 3. An actual pulse current waveform over 100 ms.

Fig. 4. Main parameters of the pulse signal.

Fig. 6. Microhardness of the base metal, HAZ and welds.

X. Di et al. / Materials and Design 66 (2015) 169–175 171

structure, which formed during solidification, was kept at roomtemperature in that there was no solid transformation. The ratesof heating and cooling were very high during welding, which leadto oriented heat dissipation, so the coarse columnar grains werevery remarkable. Metallographic examination shows no slag,porosity or other defects in the weld zone. The macroscopic mor-phology is shown in Fig. 5. The depth of the P-GMAW weld isgreater than that of the GMAW weld, and macrosegregation existsin both welds.

3.3. Mechanical testing

Hardness test results from the welds and base metal are shownin Fig. 6, indicating that the hardnesses of the welds made usingdifferent processes are similar: both them are 20–25 HV10, whichis higher than the base metal.

Tables 2 and 3 show the results of the tensile and side bendtests, respectively. Tensile test specimens, taken transverse to theweld, exhibited ultimate tensile strengths of 570–600 MPa. Anobvious result is that these two groups of welds all broke at thebase metal during the tensile tests, suggesting that the strength

Fig. 5. Macroscopic morphology of the two welds.

of the weld joints meet the requirements. The two processes resultin welds with good tensile properties, and the P-GMAW processcan guarantee the joint strength.

The plastic deformation of a specimen can be measured usingthe bend test. The bend properties are also a measurement of thematerial plasticity index. In the side bend test of GMAW welds,all four specimens failed to bend to 180�. In contrast, none of theP-GMAW weld samples broke during the test. Comparing theresults of the bending experiments, shown in Table 3 and Fig. 7,a different conclusion from the tensile test can be drawn: P-GMAWsignificantly improves the bending properties of the weld joint.

According to the results obtained by Gowrisankar [22], as thenumber of weld passes increases during welding, the hardnessand tensile strength properties of the austenitic steel weldsincrease systematically, while their ductility and toughnessdecrease progressively. However, in this study, the ductility andtoughness of the P-GMAW welds have been significantly improvedcompared with the GMAW welds despite the fact that the numberof passes during P-GMAW welding is greater than the othermethod, which slightly conflicts with the conclusions of Gowrisan-kar, while the difference may be attributed to the influence of thepulse current on the microstructure of the weld metal.

3.4. Morphology of dendrites

Depending on the growth conditions, the dendrite will developarms of various orders. A dendritic form is usually characterized interms of the primary (dendrite trunk) spacing, k1, and the second-ary (dendrite arm) spacing, k2. Generally, the spacing is measuredas the perpendicular distances between branches. The value of k1

measured in the solidified microstructure is the same as that

Table 2Results of the tensile tests.

GMAW P-GMAW

Specimen Tensile strength Rm (MPa) Location of failure Specimen Tensile strength Rm (MPa) Location of failure

1 592 Base metal 1 598 Base metal2 571 Base metal 2 598 Base metal3 595 Base metal 3 596 Base metal4 589 Base metal 4 592 Base metal

Table 3Results of the side bend tests.

GMAW P-GMAW

Specimen Angle (�) Result Specimen Angle (�) Result

1 180 Unqualified 1 180 Qualified2 180 Unqualified 2 180 Qualified3 180 Unqualified 3 180 Qualified4 180 Unqualified 4 180 Qualified

Fig. 7. Results of the bend tests.

Fig. 8. Microstructure of the MMMS welds made with and without pulse, etched with aqk1 and k2.

172 X. Di et al. / Materials and Design 66 (2015) 169–175

during growth, whereas the secondary spacing is often accompa-nied by a ripening process because of the long contact time inthe melt [23]. Fig. 8 shows the solidification microstructure ofthe welds gained using the GMAW and P-GMAW processes and aschematic of the measurement of k1 and k2 [24].

The primary dendrite arm spacing k1 and secondary dendritearm spacing k2 were measured from these structures and areshown in Fig. 9.

Fig. 8 is a comparison of the dendrite arm spacing between thetwo types of weld. At the same magnification, the structureobtained using the GMAW process has coarser columnar grain, asshown in Fig. 8(a) and (b), while the dendrite arm spacing of theP-GMAW weld is smaller and the weld metal has a more refinedstructure. Based on these metallographic properties, the dendritearm spacing was subjected to measurement statistics. The resultsare shown in Fig. 9. As Fig. 7 shows, the average value of k1 ofthe P-GMAW weld is approximately 18.48 lm, whereas the spac-ing in the GMAW weld is approximately 31.18 lm, which is almostdouble the value of the former weld. The pulse current can signif-icantly inhibit the columnar grain growth during the solidificationof the weld. Despite the similar welding parameters and more weldpasses, the results of the mechanical properties tests show that theP-GMAW process still resulted in a greater improvement in themechanical properties of the joint, and this improvement in bendtoughness can be attributed to the refinement of the substructureand decrease in dendrite arm spacing.

ua regia; (a) P-GMAW weld, (b) GMAW weld; (c) a schematic of the measurement of

Fig. 9. Measured dendrite arm spacing.

Table 4Microsegregation ratio between intergranular and intragranular regions.

Element Cr Ni Mn Fe Si

P-GMAW 1.06 1.04 1.12 0.97 1.23GMAW 1.12 1.11 1.21 0.93 1.36

X. Di et al. / Materials and Design 66 (2015) 169–175 173

As mentioned above, the two different processes have similarheat inputs. Considering the relationship between the cooling rateand the heat input, the cooling rate can be expressed as [10]

@T@t

� �x

¼ @T@x

� �t

@x@t

� �T

¼ �2pkmðT � T0Þ2

Q¼ �2pk

T � T0ð Þ2

qð2Þ

where k is the thermal conductivity, T0 is the preheating tempera-ture and Q = g � U � I is the net heat. At the same welding speed,the cooling rate decreases with increasing heat input. However,

Fig. 10. SEM images and EDS analysis. (a) SEM images of the weld metal microstructu

there is little difference between the two cooling rates due to thesame welding speed and similar heat input. The variation in thecooling rate has a significant impact on the degree of grain refine-ment; with increasing cooling rate, the dendrite arm spacing ofthe weld becomes finer [25,26]. However, the influence of the cool-ing rate on the dendritic arm spacing can be omitted due to the sim-ilar cooling rates of the GMAW and P-GMAW processes. The factorsrelated to dendrite arm spacing have little to do with externalparameters and are dependent only on the addition of the pulse sig-nal. The solidification process is often associated with the variationin the alloying constituents and their microsegregation, especiallyin high-alloy steels [18]. Variations in the alloying elements resultin a variety of dendrites and dendrite arm spacing in the weldmetal.

Fig. 10 shows SEM images at high magnification of the two weldmetals and EDS analysis of the intergranular and intragranularalloy contents. Data were collected every ten points to calculatethe average composition of the intergranular and intragranularregions. Then, the ratio of the elemental contents of the intergran-ular region to the intragranular region was obtained to studymicrosegregation in the alloy. Table 4 shows the results of themicrosegregation ratio between the intergranular and intragranu-

re. (b) EDS analysis of the intergranular and intragranular regions indicated in a.

Fig. 11. Growth of the protruding points at the interface.

174 X. Di et al. / Materials and Design 66 (2015) 169–175

lar regions. The microsegregation in the GMAW weld metal is moreserious than in the P-GMAW weld metal.

4. Discussion

4.1. Primary dendrite arm spacing

Atsumi [27] has some basic views on solidification. Solidifica-tion in the liquid usually begins at the location that has the maxi-mum undercooling. The heat flow promotes grain growth, andsolute segregation at the interface inhibits it. The crystals growpreferentially at the interface where the segregation is smaller.

The formation of dendritic grains begins at breaks at unstableplanar interfaces [27], as shown in Fig. 11, with the cooling of mol-ten metal; these protrusions preferentially grow and form trunks.If there is no movement at the interface, the branches will formequidistantly, and thus, the primary dendrite arm spacing in differ-ent places is equal. However, there are temperature and composi-tion fluctuations in the actual molten metal, so the primary armsare uneven.

Electrical pulses, as an energy intervention, are essentiallystructural variation behavior of the local melt under the effect ofan external field, which can affect the liquid metal with the helpof convection and conduction of the fluid [28]. When the weldingcurrent changes following a certain rule, the convection of the mol-ten metal changes accordingly. A sudden change of the pulse signalwould produce a certain electromagnetic stress, which will pro-duce some type of vibration within the molten pour during thesolidification of the weld. The pulse current and its resulting stir-ring of the molten pool may also cause some temperature fluctua-tions [26,29]. During the solidification of the weld metal withoutpulse, the initial nucleation regions are discretely distributed,and the distribution is uneven. Some locations favor nucleationand growth, while other locations restrict these processes due tosolute segregation [30]. However, the weld metal made with pulsehas more homogeneous nucleation sites in the entire molten metalbecause of the vibrational perturbation and temperature fluctua-tions caused by the pulse signal, which will increase the nucleationrate, and the nucleation can occur over a wide range. The columnar

Fig. 12. Schematic of the reduction

grains in the molten pool will stop growing when they contacteach other, and thus, the growth of the crystals is suppressed.Accordingly, the higher the nucleation rate, the greater the numberof nuclei. Additionally, the columnar crystals are suppressed moreeasily and more rapidly, and the primary arm spacing is corre-spondingly smaller.

4.2. Secondary dendrite arm spacing

The small bumps in the trunk are amplified under perturbationuntil significant peaks and valleys emerge at the perturbed inter-face. The tip of the trunk grows faster than the valley because itcan discharge solute more easily. If the dendrite growth conditionsare met along the surface of the trunk, a secondary arm will form.When solidified, the interface of the solid becomes enrichment viathe discharge of solutes or becomes dilute via the absorption of sol-utes that the segregation produced. Then, constitutional underco-oling reduction will be caused by this segregation, as shown inFig. 12, and the solidification of this area will be suppressed, whichcauses the places with greater undercooling near the solidificationinterface to grow preferentially [27].

The addition of pulse signal changes the activation energy of theatoms, increases the diffusion coefficient of the alloy elements and,subsequently, reduces the constitution segregation. In contrast, thestirring of the molten pool via pulse current makes it more uni-form, resulting in less segregation, which will allow secondarydendrite arms to grow more easily over the entire pool. The den-drite growth is stopped when they contact each other. The second-ary arms are smaller because of the decreased segregation.Nevertheless, in the solidification of the weld metal without pulse,the growth in locations that accumulated the solutes dischargedduring solidification is suppressed, see Fig. 13; even the branchesproduced at the roots of the trunks are restricted. In addition, oth-ers branches become coarser because they suffer less impediment.

The pulse signal can also enhance the convection of the moltenpool, which allows the enriched solutes near the solidificationinterface to easily dissipate, or in the case of solute depletion, thesolute can easily be replenished from the raffinate, which reducesthe segregation at the interface. In general, secondary dendriteswill undergo a ripening process after their formation; one possiblecoarsening mechanism is that small secondary dendrites melt andthe diameters of the coarser branches increase [10]. The stirringand convection effect of the pulse current on the molten poolreduces the overall temperature, weakens the ripening process,and, most importantly, prevents the annihilation of small dendritearms; therefore, the dendrites become uniform and fine.

In view of the above discussion, the improved mechanical per-formance can be attributed to the attainment of fine primary andsecondary dendrites, and the use of pulse current reduces themicrosegregation of the dendrites and enhances equidistributionof the alloy.

of constitutional supercooling.

Fig. 13. The growth of dendrite branches. (a–c) Schematic diagrams of the dendritegrowth, showing the suppressed branches due to the enrichment of the solute; (d)is the real dendrite morphology.

X. Di et al. / Materials and Design 66 (2015) 169–175 175

5. Conclusions

The MMMS samples in 30 mm thickness were welded byGMAW and P-GMAW techniques for the purpose of improvingthe microstructure and mechanical properties of the joints. The fol-lowing conclusions can be drawn according to the experimentalresults and theoretical analysis:

(1) Modified medium manganese steel can be effectivelywelded using GMAW or P-GMAW processes with low heatinputs. The joints made using the pulse current welding pro-cess can ensure good strength, and the use of P-GMAW pro-vides comparatively better bend toughness.

(2) In comparison to GMAW, the microstructure of P-GMAWis finer with less developed columnar subgrains. Theaverage k1 value of GMAW weld is greater than doublethe value of P-GMAW weld, and the secondary dendritespacing of the P-GMAW weld is smaller than the GMAWweld.

(3) The improvement in the bend toughness of the P-GMAWweld can be attributed to the refinement of the substructureand the decrease in the dendrite arm spacing.

(4) The use of pulse signals causes effective vibration in themolten pool, changes in the solidification state of the weld,improves the segregation conditions, and thus, producessmaller dendrites, which are beneficial to the ductility ofthe weld joint.

Acknowledgements

This research is financially supported by Tianjin Natural ScienceFoundation, No. 11JCYBJC06000, and Key Project of Tianjin Munici-pal Science and Technology Support Program, No. 11ZCGYSF00100.

References

[1] Hadfield RA. Hadfield’s manganese steel. Science 1888;12:284–6.[2] Mendez J, Ghoreshy M, Mackay WBF, Smith TJN, Smith RW. Weldability of

austenitic manganese steel. J Mater Process Technol 2004;33(4):596–602.[3] Zhang FC, Lv B, Hu B, Li YG. Flash butt welding of high manganese steel

crossing and carbon steel rail. Mater Sci Eng, A 2007;454:288–92.[4] Quiroz MSV, Gumenyuk IA. Laser beam weldability of high-manganese

austenitic and duplex stainless steel sheets. Weld World 2012;56(1–2):9–20.[5] Zorc B, Nagode A, Kosec B, Kosec L. Analysis of weld cracking in shotblasting

chambers made of Hadfield steel. Eng Fail Anal 2013;33:48–54.[6] Norman E. US 3113861: austenitic steel alloy. US-Patent, US; 1963.[7] Jing TF, Zhang FC. The work-hardening behavior of medium manganese steel

under impact abrasive wear condition. Mater Lett 1997;31:275–9.[8] Erich F. Welding Metallurgy of Stainless Steels. Germany: Springer-Verlag

Wein New York; 1988.[9] Davies GJ, Garland JG. Solidification structures and properties of fusion welds.

Int Metall Rev 1975;20(1):83–108.[10] Sindo K. Welding metallurgy. 2nd ed. America: A Wiley-Interscience

Publication; 2002.[11] García RMA, López MVH, García HR, Curiel LF, Lemus-Ruíz J. Effect on the

microstructure and mechanical properties of the electromagnetic stirringduring GMA welding of 2205 DSS plates. Mater Sci Forum 2013;755:61–8.

[12] Pal K, Pal SK. Effect of pulse parameters on weld quality in pulsed gas metal arcwelding: a review. J Mater Eng Perform 2011;20(6):918–31.

[13] Razal RA, Manisekar K, Balasubramanian V, Rajakumar S. Prediction andoptimization of pulsed current tungsten inert gas welding parameters to attainmaximum tensile strength in AZ61A magnesium alloy. Mater Des2012;37:334–48.

[14] Mirzaei M, Arabi JR, Yazdipour A, Zangeneh-Madar K. Study of weldingvelocity and pulse frequency on microstructure and mechanical properties ofpulsed gas metal arc welded high strength low alloy steel. Mater Des2013;51:709–13.

[15] Hadadzadeh A, Ghaznavi MM, Kokabi AH. The effect of gas tungsten arcwelding and pulsed-gas tungsten arc welding processes’ parameters on theheat affected zone-softening behavior of strain-hardened Al–6.7 Mg alloy.Mater Des 2014;55:335–42.

[16] BS EN ISO 14343: 2009. Welding consumables – Wire electrodes, wires androds for arc welding of stainless and heat-resisting steels – Classification.British Standards Institution; 2010.

[17] ISO 4136. Destructive tests on welds in metallic materials – transverse tensiletest. International Standardization Organization; 2001.

[18] ISO 5173. Destructive tests on welds in metallic materials – bend tests.International Standardization Organization; 2001.

[19] ISO 9015-1. Destructive tests on welds in metallic materials – hardness testing– Parting 1: Hardness test on arc welded joints. International StandardizationOrganization; 2001.

[20] Gruzin PL, Grigorkin VI, Mural’ VV, Moskaleva LN. Transformations inaustenitic manganese steel. Metal Sci Heat Treat 1969;11(1):5–8.

[21] Vaselin JM. US 1978052: Welding of Austenitic Manganese Steel. US-Patent,US; 1978.

[22] Gowrisankar I, Bhaduri AK, Seetharaman V, Verma DDN, Achar DRG. Effect ofthe number of passes on the structure and properties of submerged arc weldsof AISI type 316L stainless steel. Weld J 1987;66:147s–54s.

[23] Kurz W, Fisher DJ. Fundamentals of solidification. 4th ed. Switzerland: TransTech Publications Ltd; 2010.

[24] Grugel RN. Secondary and tertiary dendrite arm spacing relationships indirectionally solidified Al–Si alloys. J Mater Sci 1993;28(3):677–83.

[25] Cicutti C, Boeri R. On the relationship between primary and secondarydendrite arm spacing in continuous casting products. Scr Mater2001;45(12):1455–60.

[26] Flemings MC. Behavior of metal alloys in the semisolid state. Metall Trans B1991;22(3):269–93.

[27] Atsumi O. The solidification of metals. Japan: Chijin Shokan Co. Ltd; 1975.[28] Zhang S. Pulse electroplating and pulse welding power. Beijing: China machine

Press; 1988 [in Chinese].[29] Ghosh PK, Goyal VK, Dhiman HK, Kumar M. Thermal and metal transfer

behaviours in pulsed current gas metal arc weld deposition of Al–Mg alloy. SciTechnol Weld Join 2006;11(2):232–42.

[30] Wang JZ, Qi JG. Electrical pulses processing theory and application of metalmelt. Beijing: Science press; 2011 [in Chinese].