mitigation of root defect in laser and hybrid laserarc ... research march 2015 / welding journal...

WELDING RESEARCH

MARCH 2015 / WELDING JOURNAL 73-s

Introduction

High-power laser and hybrid laser-arc welding (Refs. 1– 5) offer fasterwelding speeds, lower heat inputs, anddeeper penetration over traditional arcwelding processes in a range of differ-ent construction and fabrication in-dustries (Ref. 6). However, theseprocesses are also susceptible tounique defects associated with theirhigh aspect ratio and deep penetra-tion. Two of the most common ofthese defects include porosity fromkeyhole instability in partial-penetra-tion welds (Refs. 7–9) and complete-joint-penetration welding root defects.The latter defect has also been called

chain of pearls (Ref. 10), dropping(Ref. 11), and root humping (Ref. 12).and is characterized by the formationof weld metal spheroids at the bottomsurface of a complete-joint-penetra-tion weld. An example of root defectsin a DH36 steel hybrid laser-arc weld isshown in Fig. 1A. As higher laser pow-ers become available and deeper pene-trations are obtained, these defectswill become more problematic, and adeeper understanding of the mecha-nisms driving these defects will benecessary to take advantage of thesehigh laser powers. Root defects have been character-ized at both high (Refs. 13–16) andlow (Refs. 11, 13, 17, 18) heat inputs.

Consequently, changes in laser poweror welding speed resulted in the ap-pearance or disappearance of root de-fects. For example, in 304 stainlesssteel, Zhang et al. (Ref. 14) and Kaplanand Wiklund (Ref. 16) found that rootdefects occur at lower welding speeds(i.e., higher heat input) during laserwelding of 12- and 16-mm-thickplates, respectively. In other cases, in-creasing the heat input leads to accept-able welds. Havrilla et al. (Ref. 11) in-creased the laser power by 1 kW from7.75 to 8.75 kW at a constant weldingspeed to eliminate root defects in 12-mm-thick steel. Ilar et al. (Ref. 12) em-ployed high-speed imaging to studythe formation of root defects in realtime during the laser welding of 8-mm-thick 304 stainless steel plate.The high-speed videos showed the ini-tiation of bulges immediately behindthe keyhole, and these bulges wouldoccasionally build up and solidify asroot defects. Ilar et al. (Ref. 12) con-cluded that gravity, surface tension,and melt availability play a role in theformation of root defects. One method for avoiding root de-fects is supporting the weld pool fromthe bottom through the use of electro-magnetic forces from an oscillatingmagnetic field (Refs. 15, 19, 20). Bach-mann et al. (Refs. 15, 20) comple-mented physical experiments by utiliz-ing a 3D numerical heat transfer andfluid flow model to calculate the EMforces necessary to balance the hydro-static pressure, which promoted rootdefects in 10- and 20-mm-thick steeland aluminum plates, respectively.

Mitigation of Root Defect in Laser and Hybrid LaserArc Welding

A model is developed that predicts the range of processing conditions that produce defectfree, completejointpenetration welds

BY J. J. BLECHER, T. A. PALMER, AND T. DEBROY

ABSTRACT Even though laser and hybrid laserarc welding processes can produce singlepass,completejointpenetration welds in excess of 12 mm, root defects, such as root humping, have been observed at these greater plate thicknesses. The competition betweenthe surface tension and the weight of the liquid metal in the weld pool is expected togovern rootdefect formation. A series of laser and hybrid lasergas metal arc welds hasbeen completed in which each force is independently varied. The internal morphologiesof the resulting root defects are characterized by Xray computed tomography andfound to vary significantly when welding with either the laser or hybrid laserarcprocess. In order to compute the surface tension and liquid metal weight, a modelbased on the approximate geometry of the weld pool is developed and successfullypredicts the range of processing conditions where root defects form. Process maps arethen constructed for lowcarbon steel and 304 stainless steel alloy systems. These mapscan then be used to select welding parameters that produce defectfree completejointpenetration welds over a wide range of plate thicknesses.

KEYWORDS • Laser Welding • Root Defect • Hybrid Welding • Carbon and LowAlloy Steels

J. J. BLECHER ([email protected]) and T. DEBROY are with the Department of Materials Science and Engineering, and T. A. PALMER is with the Applied ResearchLaboratory, The Pennsylvania State University, University Park, Pa.

Blecher Supplement_Layout 1 2/12/15 1:09 PM Page 73

WELDING RESEARCH

WELDING JOURNAL / MARCH 2015, VOL. 9474-s

However, the predicted EM force val-ues were slightly less than the experi-mental values necessary to hold theliquid in place since only the weight ofthe liquid metal column above the bot-tom pool surface was taken into ac-count. While EM support can be usedto weld thick sections, utilizing theprocess in a production environmentmay not be practical, and the applica-tion of EM forces can change the fluidflow patterns during welding (Ref. 20).Therefore, it is necessary to develop adeeper understanding of the condi-tions that promote root defect forma-tion in order to intelligently selectwelding parameters that suppress it. In this paper, the formation of rootdefects is investigated for laser and hy-brid laser-arc welding under a variety ofwelding conditions. For the first time,the 3D internal structure of the root de-fect nuggets in a structural steel plateweld was characterized by X-ray-com-puted tomography (CT) and was foundto depend on the welding process em-ployed. The melt volume and the sur-face tension of the molten DH36 steelwere independently varied to determine

the effect of each on root defect forma-tion. Melt volume was varied by chang-ing the heat input of the welds, and sur-face tension was altered by removingthe oxide scale on the bottom surface ofthe plate prior to welding. Increasingmelt volume or decreasing surface ten-sion led to root defects being formed. Inorder to quantify each effect, a force bal-ance considering the weight of the liq-uid steel and the surface tension at theweld root is developed for an idealizedweld pool and used to determine theconditions for the formation of root de-fects. Process maps for defect-free, com-plete-joint-penetration laser welds, forwhich substantial experimental resultshave been reported, and selected hybridlaser-arc welds were developed for low-carbon steel and 304 stainless steel.

Experimental Methods

Bead-on-plate laser and hybridlaser-arc welds were performed on 4.8-and 9.8-mm-thick DH36 steel plate.An IPG Photonics® YLR-12000-L yt-terbium fiber laser with a Precitec®YW50 welding head was used for laser

A

B

C

Fig. 1 — A— The typical weld root defects formed during hybrid lasergas metal arcwelding with a laser power, welding speed, and welding wire feed rate of 5 kW, 30mm/s, and 229 mm/s, respectively; B — the roots of hybrid welds made under identicalprocess conditions (welds 7, 8) without bottom surface oxide scale; C — with bottomoxide scale. On the plate with scale, root defects formed due to the low surface tension.

Fig. 2 — A — A comparison of thetransverse hybrid weld (welds 7, 8)cross sections with identical weldingconditions with the exception of bottom surface oxide scale, which was notpresent; B — the bottom surface oxidescale was present. The sizes of thewelds are similar, suggesting that theweight of the liquid metal is similar andthat the reduction in surface tension isdue to the oxide scale presence, whichled to the root defects.

Table 1— Welding Conditions for LBW and HLAW Welds

Weld Number Oxide Removed Welding Process Plate Thickness (mm) Root Defects Weld Speed (mm/s) WFS (mm/s)

1 no LBW 4.8 no 30 —2 no LBW 4.8 no 40 —3 no LBW 9.5 yes 15 —4 no HLAW 4.8 yes 30 1275 no HLAW 4.8 yes 30 1526 no HLAW 4.8 yes 30 2297 yes HLAW 4.8 no 40 1278 no HLAW 4.8 yes 40 127

Note: Oxide removed refers to the bottom surface oxide scale being removed.

A

B

Blecher Supplement_Layout 1 2/12/15 11:28 AM Page 74

WELDING RESEARCH


welding. The optics system utilizes col-limating and focusing lenses with 200-and 500-mm focal lengths, respective-ly. The 1-m laser wavelength is trans-ported to the welding head through a200-m-diameter process fiber. Thefocused spot size and full divergenceangle were measured with a Primes®Focus Monitor and are 0.52 mm and64 mrad, respectively. A Lincoln Elec-tric® Power Wave 455 M/STT powersource with a Binzel® WH 455D water-cooled welding gun was used with ER

70S-6 welding wire for the hybridwelding experiments. In both the laser and hybrid laser-arc welding processes performed here,laser power, defocus, laser-arc separa-tion distance, and arc voltage werekept constant at 5 kW, 8 mm, 3 mm,and 31 V, respectively, while wire feedspeed, arc current and travel speedwere varied when welding on 4.8-mm-thick plate. A positive 8-mm defocusindicates that the position of focus isabove the plate. When welding on a9.8-mm-thick plate, 7-kW laser powerand zero defocus were selected. Addi-tionally, to test the effect of the oxidepresence on the bottom surface, twotypes of plate were used, one with onlythe top surface of the plate sand blast-ed to remove the oxide. The other

plate had the oxide removed on bothsides. A summary of the welding pa-rameters is given in Table 1. Oxide re-moved refers to whether the oxidescale was removed on both sides priorto welding. The welding processeswere laser beam welding (LBW) andhybrid laser-arc welding (HLAW).Standard metallographic techniqueswere used to inspect the transversecross-sections of the welds. X-ray CT images were captured witha General Electric® v|tome|x CT sys-tem. The accelerating voltage and cur-rent for each scan were 280 kV and 180A, respectively. The voxel (i.e., 3D pix-el) size with a magnification of 20×was50 m. DatosX® software handled thereconstruction of the individual X-rayimages to produce the 3D image. Thedefect detection module in the VolumeGraphics® VGStudio Max software wasused to highlight the internal porestructure of the weld defects.

Results and Discussion

Root Defect Formation and Characterization

Based on previous research (Refs.12, 15, 19, 20), surface tension andweld pool volume are thought to play apart in the formation of root defects.The surface tension at the bottom ofthe plate will restrain the liquid metalin the weld pool and discourage rootdefect formation. On the other hand,the weight of the molten metal in theweld pool will act to promote the for-mation of root defects. The competi-tion between these two forces will de-termine whether defects will form. In order to investigate the relation-ship between the surface tension andweld pool volume, both were inde-pendently varied during welding ex-periments. The presence of oxygen inmolten iron has a significant effect onthe surface tension. At high oxygencontents between 0.06 and 0.1 wt-%,the surface tension of liquid iron islowered by 50% or more (Ref. 21) ofits oxygen-free value, 1.91 N/m (Ref.22). Oxygen content can be controlledindirectly by the removal of the oxidescale, which is present on both sur-faces of the plate. In these experi-ments, welds with a high surface ten-sion were ensured by first removingthe oxide scale from both surfaces of

Fig. 3 — The transverse weld cross sections for: A — A laser weld (weld 1); B— a hybrid weld (weld 4) are shown.The laser conditions are the same, butthe hybrid weld has increased heatinput, larger amount of melted volume,and greater weight that must be supported by the surface tension force.

Fig. 4 — The internal structure of rootdefects in: A — Hybrid lasergas metalarc weld in 4.8mmthick plate (weld6). The direction of welding is indicatedby the arrow. The strands of porositythat stretch from the bottom of theplate and connect in the center of thedefect were common in all the hybridwelds where root defects formed; B —laser weld in 9.5mmthick plate (weld3) as determined by Xray computedtomography. Only individual pores areobserved. Evidence of gouging due tomaterial loss to root defects can be observed at the top of the plate.

A

B

B

A


WELDING RESEARCH


the steel plate. Low surface tensionweld pools were acquired by only re-moving the oxide scale from the topsurface and not grinding the bottomsurface. Oxygen pickup is a contribu-tor but not the sole source of root de-fects. In this case, the oxide scale actsas a source of oxygen that leads to alower surface tension of the moltenmetal at the root of the weld but is nota necessary condition for root defectformation. The volume of the weldpool was varied by increasing the heatinput by transitioning from laser tohybrid laser-arc welding or by increas-ing plate thickness. Higher heat inputsand greater plate thicknesses producemore molten metal during completejoint penetration welding. In examining the effect of lowersurface tension, hybrid welds (welds 7,

8) were made under thesame conditions, with theexception of the oxidescale presence on the bot-tom surface. The weldwith scale (weld 8)formed root defects mostlikely because the re-

straining surface tension force wasmuch lower due to the increased oxy-gen content. On the other hand, theweld without scale (weld 7) formed noroot defects. A comparison of the bot-tom surfaces and transverse weld crosssections are shown in Fig. 1B, C andFig. 2, respectively. In Fig. 1, irregular-ly spaced root defects can be observedin the weld with bottom surface oxidescale. While the weld without scalecontains some reenforcement on thebottom surface, there are no observ-able defects. As can be observed in Fig.2, root defects have a profound effecton the weld cross section. The weldwithout scale has a clearly identifiablearc zone at the top, which is muchwider than the rest of the weld. In theweld with scale, the weld width is

much narrower, and the complexshape of the porosity in the root de-fect nugget is observable. Eleven hy-brid welds were made with and eighthybrid welds made without bottomsurface oxide scale. In all cases, thewelds without bottom surface scaledid not form root defects, while thosewith scale did form defects. The effect of heat input and liquidmetal volume on the root defect for-mation was investigated next for a se-ries of welds made at a travel speed of30 mm/s. As shown in Table 1, thelaser weld (weld 1) has a heat input of167 J/mm, while the hybrid welds(welds 4–6) had estimated heat inputsbetween 426 and 695 J/mm. Thehigher heat inputs should increase thesize of the weld pool, which will in-crease the overall weight that must bebalanced against the surface tensionforce at the bottom of the pool. At thelowest heat input, there were no rootdefects, but defects formed at the highheat inputs. The effect of heat inputon the size of the weld and amount ofmaterial melted can be observed in

Fig. 6 — Process maps for: A — Lowcarbon steels; B — 304 stainless steel— were constructed from available reports of laser welding and cuttingexperiments. The maps indicate with what processing conditions acceptable complete penetration welds can be made with respect to other interaction modes.

Fig. 5 — The comparison of surface tension force andweight force for the experimental welds considered. Thenumbers next to each weld indicate the processing conditions as shown in Table 1. Compared to experimentalresults, all the welds are found to be on the correct sideof the Fg = Fs line.

Table 2 — Average and Standard Deviations of Three Measurements for the Top Surface and Bottom Surface Weld Width and Weld Length and the Estimated Arc Force

Weld Number D (mm) d (mm) l (mm) Arc Force (mN)

1 2.2 ± 0.1 1.6 ±0.2 3.7 ± 0.8 0.02 1.9 ± 0.2 1.5 ± 0.1 3.4 ± 0.7 0.03 4.2 ± 0.2 2.5 ± 0.1 6.9 ± 0.6 0.04 4.9 ± 0.2 2.5 ± 0.1 13.8 ± 0.7 9.05 6.4 ± 0.9 2.6 ± 0.3 17.0 ± 2.3 39.46 7.8 ± 0.4 3.3 ± 0.4 28.6 ± 1.8 74.37 6.1 ± 0.3 2.2 ± 0.3 15.8 ± 2.2 14.08 3.9 ± 0.5 2.0 ± 0.2 8.4 ± 2.2 14.0

A B

Blecher Supplement_Layout 1 2/13/15 2:06 PM Page 76

WELDING RESEARCH


Fig. 3, which shows the transverseweld cross sections of the laser weld(weld 1) and the lowest heat input hy-brid weld (weld 4). The hybrid weld ismuch larger in terms of cross-sectionalarea compared to the laser weld, andthe high pool volume combined withthe suppression of surface tensionfrom the bottom surface scale promot-ed the formation of root defects. Postweld observation showed thatsome of the defect nuggets were miss-ing portions of the outer wall, whichcan be seen in one of the nuggets ofFig. 1C and indicates that porositymay be present. In Figs. 2 and 3, mi-crographs show complex pore shapes.The internal structure of the root de-fect nuggets was characterized with X-ray CT, which nondestructively evalu-ates internal defects by differentiatingregions of different density. Within ametal structure, pores appear brightersince fewer X-rays are absorbed. Typical internal structures of rootdefects formed during hybrid laser-arcwelding and laser welding are shownin Fig. 4. The porosity is representedby the yellow colored shapes withinthe defects. The large pore in the hy-brid weld is fully interconnected. Eightarms start at the edges of the weld andextend down to the bottom of the rootdefect nugget where they connect to alarge central pore. This interconnectedporosity contrasts with pores shown inFigure 4B for a laser weld fabricated inthicker plate. In this defect, there isonly a dispersion of smaller sphericalpores, which are only present in thetop half of the defect. Large gouges inthe top surface resulting from the lossof material to the defect are also visi-ble in Fig. 4B. Clearly, the laser and hybrid weld-ing processes produced different poreshapes and sizes within the weld de-fects. Typically, in laser weldingprocesses, where large spherical poresare present, keyhole collapse (Ref. 7)results in pores centered on the laserbeam axis. Based on video evidence ofroot defects forming and growingalong the length of the pool away fromthe keyhole (Ref. 12) and X-ray CT im-ages of complex pore networks anddifferent characteristics for differentwelding processes, keyhole dynamicsprobably cannot be used to explain theporosity structures within the weld de-fect nuggets. The most likely explana-

tion is that the additional forces in arcwelding, such as the arc pressure anddroplet impact force, led to the net-work of pores observed in Fig. 4A.

Mechanism of Root Defect Formation

The formation of weld root defectscan be viewed as a force balance be-tween the weight of the liquid metal inthe weld pool and the surface tensionforce. A model has been developed tocalculate the magnitude of each forceand utilizes an approximate weld poolshape and measured weld bead dimen-sions. The details of the model are givenin Appendix A. The results of the forcebalance between weight and surfacetension are shown in Fig. 5. The num-bers next to each point indicate whichweld is plotted. Error bars represent thespread in values calculated with the un-certainty for each measured dimensiongiven in Table 2. The model is physicallyconsistent, since as heat input increas-es, the calculated values for weight dueto gravity and surface tension force alsoincrease. The Fg = Fs line defines theboundary between regions where rootdefects will (to the right of the line) andwill not (to the left of the line) occur. The laser welds (welds 1–3) havethe lowest weight forces (i.e., pointsfarthest to the left in Fig. 5) due to thelow heat input of the laser and the re-sulting small pool volumes comparedto hybrid welds. Laser welds fabricatedon 4.8-mm plate (welds 1, 2) did notform a defect since the surface tensionforce of 4 mN easily restrained theweight force of 1 mN. However, whencomplete joint penetration laser weldswere made on thicker plate of 9.5 mm(weld 3), the weight force increased to7.5 mN and exceeded the surface ten-sion of less than 7 mN, and defectsformed. The force balance captures thedifference in plate thicknesses andpredicts the observed outcome for thelaser welds. For the laser and hybrid laser-arcwelds with the same 30 mm/s weldingspeed (welds 1, 4–6), the model predictsincreasing weight due to gravity from 1to 23 mN due to increasing heat input,but the surface tension only increasedfrom 4 to 9 mN. The model predictionsfor these welds agree with the experi-mental observations. The force balancealso captures the differences in welds onplates with and without bottom surface

oxide and identical process conditions(welds 7, 8) with similar weight forces of8 and 10 mN, respectively. With thepresence of the oxide scale, though,weld 8 possesses only 40% of the sur-face tension force of weld 7, which hadno oxide scale. As a whole, agreement isgood between the observed and predict-ed formation of root defects with all ofthe green symbols, indicating no ob-served root defects, falling above the Fg= Fs line and all the red symbols, indicat-ing observed defects, below the sameline. These results indicate that repre-sentation of the root defect formationphenomenon as a quantitative force bal-ance between weight and surface ten-sion allows for a qualitative predictionof an important welding defect.

Process Maps for Complete Joint Penetration Laser Welding

The results described above indi-cate that both heat input and platethickness affect the formation of rootdefects in DH36 steel and are expectedto play a role in other alloys. For givenlaser welding parameters, materialproperties, and plate thicknesses,however, the developed model cannotprovide a broader predictive capabilityof root defect formation. Additionally,the process parameters that producedefect-free complete-joint-penetrationwelds are bounded by other laser-material interaction modes, such ascutting and partial-penetration laserwelding. In order to address these ad-ditional complexities, process mapsthat tie together laser welding parame-ters, material properties, heat input,plate thickness, and laser-material in-teraction modes are constructed fordifferent alloy systems. Comparison ofprocess maps for different alloysshould show where similarities existand what conclusions apply across ma-terial types. Process maps have been construct-ed for laser processing of low-carbonsteels and 304 stainless steel. Thenondimensional heat input per unitlength, which is similar to that used byDe and DebRoy (Ref. 23) includes laserwelding parameters and material prop-erties and is defined as

(1)2=

η

ρ + Δ πH*

PU

(h H ) rv b


WELDING RESEARCH


where is absorptivity, P is laser pow-er, rb is the laser beam radius at focus,U is the welding speed, is the liquidmetal density, h is the enthalpy of theliquid at the boiling point, and Hv isthe heat of vaporization. Both the nu-merator and denominator have unitsof energy per unit length, J/m. Equa-tion 1 compares the amount of energyabsorbed from the laser per unitlength in the numerator to theamount of energy per unit length nec-essary to heat the alloy from roomtemperature to the boiling point (de-nominator). For the experiments con-sidered here, H* is typically greaterthan one, indicating that absorbedlaser energy is a multiple of the energyto heat the alloy to the boiling point.The material property values that wereused to calculate H* for each alloy aregiven in Table 3. Estimated resistivi-ties (Ref. 24) were used to calculate ab-sorptivity (Ref. 25), and density, en-thalpy, and heat of vaporization wereobtained from available references(Refs. 24, 26). When H* is plotted as a function ofplate thickness, four regions definingcutting, complete-joint-penetrationweld without defects, complete-joint-penetration weld with defects, andpartial-penetration weld, can be iden-tified. In the case of cutting low-car-bon steel or stainless steel in the pres-ence of oxygen gas, laser energy ac-counts for approximately half of thetotal energy input into the systemwith oxidation of liquid iron account-ing for the other half (Refs. 27, 28).Other processes, such as hybrid laser-arc welding, are not considered inthese maps, but as shown previously,the addition of another heat sourcecan increase the chances of root defectformation. The process maps for the laserwelding of low-carbon steel (Refs. 18,27, 29–53) and 304 stainless steel(Refs. 14–16, 39, 40, 54–70) areshown in Fig. 6. Experimental H* datafor cutting, complete-joint-penetra-

tion welds without root defects, andpartial penetration welds are deter-mined by macrographs of the welds orexplicit statements in the text andplotted for every material type. Rootdefects reported in the literature areassumed to form because of the com-petition between the surface tensionand liquid metal weight forces and notother phenomena, such as keyhole in-stability, which leads to macroporosity.The lines defining each region are fitto the complete-joint-penetrationlaser welding data. All the conditionsshown are for a laser only, except asindicated in Fig. 6A, where a set of hy-brid laser-GMAW conditions (Ref. 18)is used to show another case of rootdefects in low-carbon steel. The points indicating root defectformation for low-carbon steel in 12-mm-thick plate (Ref. 31) are situatedclose to the cutting-complete jointpenetration transition line. On theother hand, because of the availabilityof data, a line indicating the transitionfrom defect-free welds to welds withroot defects can be drawn for 304stainless steel and fully encloses thecomplete-joint-penetration laser weld-ing space for the alloy. According tothe process map, the laser welding pa-rameters for complete-joint-penetra-tion welds in excess of 16 mm are verylimited, so only careful selection ofprocessing parameters can produce adefect-free weld. For the same space inlow-carbon steel, relatively thickplates of 25 mm can be welded atnondimensional heat inputs of lessthan 40. Comparison of the process mapsshows relatively similar behaviors atlower heat inputs and plate thickness-es. This similarity across process mapsis highlighted by the thick solid blackline, which encloses an identicalprocess space in the defect-free zone,and can be used to produce defect-freecomplete-joint-penetration welds foreach alloy. For example, process pa-rameters yielding an H* of 12 will pro-

duce a defect-free weld in 8-mm-thickplate of low-carbon steel and 304stainless steel. Identical H* values forplate thicknesses between 3.5 and 10mm can be used across alloy systems.The parameters shown in Fig. 6 canspeed process parameter development,especially in thick plates, wheregreater heat inputs lead to longer cool-ing times between trial welds. For ex-ample, if laser optics are fixed (i.e.,minimum laser beam radius cannotchange), then welding engineers canquickly select nondimensional heat in-put from Fig. 6 for a given plate thick-ness and calculate via Eq. 1 the laserpowers and welding speeds that willproduce defect-free, complete-joint-penetration welds. At higher heat inputs and platethicknesses, the behaviors of each al-loy diverge. A very small defect-freecomplete-joint-penetration processspace is observed in 304 stainlesssteel. The opposite is observed in low-carbon steels. The process maps indi-cate that two conditions are necessarybefore root defects can form duringlaser welding. First, the plate thicknessmust be 10 mm or greater. Second, thenondimensional heat input must begreater than 16, which is also a mini-mum value associated with root de-fects. Once these two conditions aremet, the formation of root defects be-comes possible. However, as shown inthe process maps, satisfying the condi-tions does not guarantee root defectformation. For example, the mean av-erage H* value for root defect forma-tion in low-carbon steel and 304 stain-less steel laser welds is 33, so thechances of forming root defects in-creases as the heat input increases. From the process maps, 304 stain-less steel is more susceptible to root de-fect formation than low-carbon steel.The reason is the difference in surfacetension, which is 1.91 N/m (Ref. 22)and 1.17 N/m (Ref. 79) for steel andstainless steel, respectively. The densityof each liquid alloy is 7030 kg/m3 for

Table 3 — Values Used to Construct the Process Maps

Alloy – 1m - 10.6 m (kg/m3) h (kJ/kg) Hv (kJ/kg)

Steel 0.34 0.12 7030 2390 6260304SS 0.34 0.12 7070 2290 6330

Note: Absorptivity, , depends on the wavelength of the laser beam.


WELDING RESEARCH


steel and 7070 kg/m3 for 304 stainlesssteel (Ref. 24). Both surface tension anddensity should affect the formation ofroot defects as shown in Eq. A5 and A7,and only surface tension is very differ-ent between the two alloys. Close inspection of Fig. 6A, B indi-cates that root defects are associatedwith relatively higher heat inputs. Forlow-carbon steel, the root defects inboth laser and hybrid welding condi-tions are close to the cutting transi-tion line, indicating high heat input. Inthe case of 304 stainless steel, all theheat inputs associated with root de-fects are greater than the defect-freeheat inputs for the same plate thick-nesses, 10, 12, and 16 mm. Some researchers have concludedthat the root defects are associatedwith low heat input (Refs. 11, 18, 31).However, the data compiled in Fig. 6A,B suggest that the opposite is true. Incases where lower heat inputs certain-ly led to root defect formation, it ap-pears that the researchers were alreadyoperating at a high heat input. For ex-ample, for 12-mm-thick low-carbonsteel plate (Ref. 31) and nondimen-sional heat inputs between 35 and 40,defects formed, but at lower heat in-puts of 29 and 33 and higher heat in-puts of 45 and 50 root defects did notform. Additionally, all of these heat in-puts are relatively high because com-plete-joint-penetration defect-freewelds were made at nondimensionalheat inputs as low as 5 and 7 (Ref. 31).By considering the whole range ofprocess parameters captured in thenondimensional heat input parameter,the fact that root defects are a highheat input phenomenon becomes clearand unambiguous.

Summary and Conclusions

The root defect in complete-joint-penetration laser and hybrid laser-arcwelding has been experimentally andtheoretically investigated. Welding pa-rameters, plate preparation, and platesize were varied to produce welds withand without root defects. Optical mi-croscopy and X-ray CT characterizedthe internal structure of the defectnuggets for different welding process-es. A force balance between the weightof the liquid metal and the surface ten-sion was developed to describe thecompeting forces driving the onset of

defect formation. Process maps fortwo alloy systems have been con-structed based on the experimentalwelding and cutting parameters re-ported in the literature. The conclu-sions of this work are listed below. 1) The qualitative effect of surfacetension and weight of liquid metal onthe formation of root defects was deter-mined by varying the welding parame-ters. A decrease in surface tension dueto the presence of oxide scale on thebottom plate surface led to the defectformation while no defects formed forthe same conditions on a plate with theoxide scale removed. Larger weld poolswere formed either by increasing heatinput with the addition of an arc orlaser welding on 9.5-mm-thick plate.The larger pools led to root defects,while the laser welds on 4.8-mm-thickplate formed smaller pools and did notresult in root defects. 2) With the use of X-ray CT, the in-ternal structure of defect nuggetsformed during hybrid laser-arc and laserwelding were found to be different. Inhybrid welding, the structure consistedof a network of large pore strands thatstretched from the edge of the bottomweld bead to the center of the defectnugget. On the other hand, defectnuggets resulting from a laser weldshowed a dispersion of small sphericalpores. The additional arc pressure anddroplet impact forces in hybrid weldingare the likely factors for the differencein porosity structure. 3) Based on the observations ofsurface tension and weight of liquidmetal, a force balance between the twowas developed for an idealized weldpool and applied to the experimentalconditions used in the study. The forcebalance calculations matched the ex-perimental observations in terms ofroot defect formation for all of the cas-es considered. The results showed thatthe force balance has utility for pre-dicting the defect formation, assumingthe pool geometry is known or can becalculated. 4) The process maps for low-carbonsteel and 304 stainless steel revealedthat identical H* values between 5 and15 can be used to fabricate defect-freewelds in plate thicknesses between 3.5and 10 mm for the two alloys consid-ered. 5) The compiled data show that twoconditions, plate thicknesses greater

than 10 mm and H* values greaterthan 15, must be met before root de-fects can form. Consideration of theheat inputs necessary to form root de-fects in low-carbon steel and stainlesssteel demonstrated that root defectsare a high heat input phenomenon, soin most cases, reducing heat input willeliminate defect formation.

The authors would like to thankMr. Jay Tressler for performing thewelding experiments and Mr. Ed Goodfor preparing the metallographic andX-ray specimens. This research wasperformed using funding receivedfrom the DOE Office of Nuclear Ener-gy’s Nuclear Energy University Pro-grams under Grant Number 120327.

1. Zhang, J., Shan, J. G., Ren, J. L, andWen, P. 2013. Reducing the porosity in die-cast magnesium alloys during laser weld-ing. Welding Journal 92(4): 101-s to 109-s. 2. Roepke, C., Liu, S., Kelly, S., and Mar-tukanitz, R. 2010. Hybrid laser arc weldingprocess evaluation on DH36 and EH36steel. Welding Journal 89(7): 140-s to 150-s. 3. Sachez-Amaya, J. M., Boukha, Z.,Amaya-Vazquez, M. R., and Botana, F. J.2012. Weldability of aluminum alloys withhigh-power diode laser. Welding Journal91(5): 155-s to 161-s. 4. Victor, B., Farson, D. F., Ream, S., andWalters, C. T. 2011. Custom beam shapingfor high-power fiber laser welding. WeldingJournal 90(6): 113-s to 120-s. 5. Wu, S. C., Yu, X., Zuo, R. Z., Zhang,W. H., Xie, H. L., and Jiang, J. Z. 2013.Porosity, element loss, and strength modelon softening behavior of hybrid laser arcwelded Al-Zn-Mg-Cu alloy with synchro-tron radiation analysis. Welding Journal92(3): 64-s to 71-s. 6. Ribic, B., Palmer, T. A., and DebRoy,T. 2009. Problems and issues in laser-archybrid welding. Int. Mater. Rev. 54(4):223–244. 7. Matsunawa, A., Kim, J.-D., Seto, N.,Mizutani, M., and Katayama, S. 1998. Dy-namics of keyhole and molten pool in laserwelding. J. Laser Appl. 10(6): 247–254. 8. Tucker, J. D., Nolan, T. K., Martin, A.J., and Young, G. A. 2012. Effect of travelspeed and beam focus on porosity in alloy690 laser welds. JOM 64(12): 1409–1417.

Acknowledgments

References

Blecher Supplement_Layout 1 2/12/15 11:28 AM 9

WELDING RESEARCH


9. Madison, J. D., and Aagesen, L. K.2012. Quantitative characterization ofporosity in laser welds of stainless steel.Scripta Mater. 67(9): 783–786. 10. Salminen, A., Piili, H., and Purto-nen, T. 2010. The characteristics of highpower fibre laser welding. Proc. IMechE,Part C J. Mech. Eng. Sci. 224(C5):1019–1029. 11. Havrilla, D., Rominger, V., Holzer,M., Harrer, T., and Andreev, A. 2013. Ad-vanced welding techniques with optimizedaccessories for high brightness 1 mlasers. Proc. SPIE 8603, High-Power LaserMater. Process.: Lasers, Beam Deliv., Diagn.,and Appl. II. Ed. F. Dorsch, 86030N. SPIE. 12. Ilar, T., Eriksson, I., Powell, J., andKaplan, A. 2012. Root humping in laserwelding — an investigation based on highspeed imaging. 7th Conf. Laser Assist. NetShape Eng./Int. Conf. Photonic Eng. Eds. M.Schmidt, F. Vollertsen, and M. Geiger, pp.27–32. Elsevier Sci. 13. Punkari, A., Weckman, D. C., andKerr, H. W. 2003. Effects of magnesiumcontent on dual beam Nd:YAG laser weld-ing of Al-Mg alloys. Sci. Technol. Weld. Join-ing 8(4): 269–281. 14. Zhang, M., Chen, G., Zhou, Y., andLiao, S. 2014. Optimization of deep pene-tration laser welding of thick stainless steelwith a 10 kW fiber laser. Mater. Des. 53:568–576. 15. Bachmann, M., Avilov, V., Gu-menyuk, A., and Rethmeier, M. 2014. Ex-perimental and numerical investigation ofan electromagnetic weld pool support sys-tem for high power laser beam welding ofaustenitic stainless steel. J. Mater. Process.Technol. 214(3): 578–591. 16. Kaplan, A. F. H., and Wiklund, G.2011. Advanced welding analysis methodsapplied to heavy section welding with a 15kW fibre laser. Weld. World 53: 295–300. 17. Ohnishi, T., Kawahito, Y., Mizutani,M., and Katayama, S. 2013. Butt weldingof thick, high strength steel plate with ahigh power laser and hot wire to improvetolerance to gap variance and control weldmetal oxygen content. Sci. Technol. Weld.Joining 18(4): 314–322. 18. Piili, H., Salminen, A., Harkko, P.,and Lehtinen, J. 2008. Study of phenome-non of fibre-laser-mig/mag-hybrid-weld-ing. ICALEO 2008 – 27th Int. Congr. Appl.Lasers Electro-Opt. pp. 506–515. LaserInst. of America. 19. Avilov, V. V., Gumenyuk, A., Lam-mers, M., and Rethmeier, M. 2012. PA po-sition full penetration high power laserbeam welding of up to 30-mm-thick AlMg3plates using electromagnetic weld poolsupport. Sci. Technol. Weld. Joining 17(2):128–133. 20. Bachmann, M., Avilov, V., Gu-menyuk, A., and Rethmeier, M. 2012. Nu-merical simulation of full penetration laser

beam welding of thick aluminium plateswith inductive support. J. Phys. D: Appl.Phys. 45(3): 035201. 21. Keene, B. J. 1988. Review of datafor the surface tension of iron and its bina-ry alloys. Int. Mater. Rev. 33(1): 1–37. 22. Keene, B. J. 1993. Review of datafor the surface tension of pure metals. Int.Mater. Rev. 38(4): 157–192. 23. De, A., and DebRoy, T. 2004. Asmart model to estimate effective thermalconductivity and viscosity in the weld pool.J. Appl. Phys. 95(9): 5230–5240. 24. Gale, W. F. 2004. Smithells MetalsReference Book, 8th Edition. Burlington,Mass., Elsevier Butterworth-Heinemann. 25. Dowden, J., and Phiroze, K. 1995. Amathematical investigation of the penetra-tion depth in the keyhole welding withcontinuous CO2 lasers. J. Phys. D: Appl.Phys. 28(11): 2252–2261. 26. Pakratz, L. B. 1982. Thermodynam-ic properties of elements and oxides. U.S.Dept. of the Interior, Bureau of Mines, Dis-trict of Columbia. 27. Hsu M. J., and Molian, P. A. 1994.Thermochemical modeling in CO2 lasercutting of carbon steel. J. Mater. Sci.29(21): 5607–5611. 28. Ivarson, A., Powell, J., and Magnus-son, C. 1991. The role of oxidation in lasercutting stainless and mild steel. J. LaserAppl. 3(3): 41–45. 29. Krasnoperov, M. Y., Pieters, R. R. G.M., and Richardson, I. M. 2004. Weld poolgeometry during keyhole laser welding ofthin steel sheets. Sci. Technol. Weld. Joining9(6): 501–506. 30. Sokolov, M., Salminen, A.,Kuznetsov, M., and Tsibulskiy, I. 2011.Laser welding and weld hardness analysisof thick section S355 structural steel.Mater. Des. 32(10): 5127 –5131. 31. Rominger, V., Haug, P., Speker, N.,and Holzer, M. 2013. High-power full pen-etration welding behavior. Laser Tech. J.10(3): 36–40. 32. Assuncao, E., Ganguly, S., Yapp, D.,Williams, S., and Paradowska, A. 2011.Characterization of residual stress state inlaser welded low carbon mild steel platesproduced in keyhole and conduction mode.Sci. Technol. Weld. Joining 16(3): 239–243. 33. Cui, C., Hu, J., Gao, K., Pang, S.,Yang, Y., Wang, H., and Guo, Z. 2008. Ef-fects of process parameters on weld metalkeyhole characteristics with CO2 laser buttwelding. Lasers Eng. 18(5-6): 319 –327. 34. Hamatani, H., Miyazaki, Y., Otani,T., and Ohkita, S. 2006. Minimization ofheat-affected zone size in welded ultra-fined grained steel under cooling by liquidnitrogen during laser welding. Mater. Sci.Eng. A 426(1–2): 21–30. 35. Ion, J. C., Salminen, A. S., and Sun,Z. 1996. Process diagrams for laser beamwelding of carbon manganese steels. Weld-

ing Journal 75(7): 225-s to 232-s. 36. Oefele, F., Musiol, J., Zaeh, M. F.2008. Influence of remote-laser-weldingparameters for an 8 kW fibre laser on theseam quality of steels. ICALEO 2008 – 27thInt. Congr. Appl. Lasers Electro-Opt. pp.399–405. Laser Inst. of America. 37. Mori, K., and Miyamoto, I. 1997.In-process monitoring of laser welding bythe analysis of ripples in the plasma emis-sion. J. Laser Appl. 9(3): 155–159. 38. Vollertsen, F., Grunenwald, S., Reth-meier, M., Gumenyuk, A., Reisgen, U.,Olschok, S. 2010. Welding thick steelplates with fibre lasers and GMAW. Weld.World 54(3–4): R62–R70. 39. Brown, R. T. 2008. Keyhole weldingstudies with a moderate-power, high-brightness fiber laser. J. Laser Appl. 20(4):201–208. 40. Nakamura, S., Sakurai, M.,Kamimuki, K., Inoue, T., and Ito, Y. 2000.Detection technique for transition be-tween deep penetration mode and shallowpenetration mode and shallow penetrationmode in CO2 laser welding of metals. J.Phys. D: Appl. Phys. 33(22): 2941–2948. 41. Kumar, N., Kataria, S., Shanmu-garajan, B., Dash, S., Tyagi, A. K., Pad-mananbham, G., and Raj, B. 2010. Contactmechanical studies on continuous waveCO2 laser beam weld of mild steel with am-bient and under water medium. Mater. Des.31(8): 3610–3617. 42. Miranda, R., Quintino, L., Willams,S., and Yapp, D. 2010. Welding with highpower fiber laser API5L-X100 pipelinesteel. 5th Int. Mater. Symp./14th Conf. Soc.-Port.-Mater. Eds. L. G. Rosa and F. Margari-do, pp. 592–596. Trans Tech Publ. 43. Quintino, L., Costa, A., Miranda, R.,Yapp, D., Kumar, V., and Kong, C. J. 2007.Welding with high power fiber lasers — apreliminary study. Mater. Des. 28(4):1231–1237. 44. Suder, W. J., and Williams, S. 2014.Power factor model for selection of weldingparameters in CW laser welding. Opt.Laser. Technol. 56: 223–229. 45. Chen, S.-L. 1998. The effect of gascomposition on the CO2 laser cutting ofmild steel. J. Mater. Process. Technol.73(1–3): 147–159. 46. Lo, K. H. 2012. A comparative studyon Nd:YAG laser cutting of steel and stain-less steel using continuous, square, andsine waveforms. J. Mater. Eng. Perform.21(6): 907–914. 47. Malikov, A. G., Orishich, A. M., andShulyat’ev, V. B. 2009. Experimental opti-mization of the gas-assisted laser cuttingof thick steel sheets. Quantum Electron.39(6): 547–551. 48. O’Neill, W., and Steen, W. M. 1995.A three-dimensional analysis of gas en-trainment operating during the laser-cut-ting process. J. Phys. D: Appl. Phys. 28(12):

Blecher Supplement_Layout 1 2/12/15 11:28 AM 0

WELDING RESEARCH


12–18. 49. Powell, J., Al-Mashikhi, S. O., Ka-plan, A. F. H., and Voisey, K. T. 2011. Fibrelaser cutting of thin section mild steel: anexplanation of the ‘striation free effect.Opt. Lasers Eng. 49(8): 1069–1075. 50. Shariff, S. M., Sundararajan, G., andJoshi, S. V. 1999. Parametric influence oncut quality attributes and generation ofprocessing maps for laser cutting. J. LaserAppl. 11(2): 54–63. 51. Sobih, M., Crouse, P. L., and Li, L.2007. Elimination of striation in laser cut-ting of mild steel. J. Phys. D: Appl. Phys.40(22): 6908–6916. 52. Wandera, C., Kujanpaa, V., and Salmi-nen, A. 2010. Laser power requirement forcutting thick-section steel and effects of pro-cessing parameters on mild steel cut quality.Proc. IMechE Part B: J. Eng. Manuf. 225(B5):651–661. 53. Zaied, M., Miraoui, I., Boujelbene, M.,and Bayraktar, E. 2013. Analysis of heat af-fected zone obtained by CO2 laser cuttingof low carbon steel (S235). 3rd Int. Congr.Adv. Appl. Phys. Mater. Sci. Eds. A. Y. Oral,Z. B. Bahsi, and A. Sonmez, pp. 323–326.Amer. Inst. Phys. 54. Nath, A. K., Sridhar, R., Ganesh, P.,and Kaul, R. 2002. Laser power couplingefficiency in conduction and keyhole weld-ing of austenitic stainless steel. Sadhana27: 383–292. 55. Bhargava, P., Paul, C. P., Mundra, G.,Premsingh, C. H., Mishra, S. K., Nagpure,D., Kumar, A., and Kukreja, L. M. 2014.Study on weld bead surface profile and an-gular distortion in 6-mm-thick butt weldjoints on SS304 using fiber laser. Opt.Laser Eng. 53: 152–157. 56. Nakabayashi, T., Wani, F.,Hayakawa, A., Suzuki, S., and Yasuda, K.2003. Thick plate welding with Nd:YAGlaser and COIL. 1st Int. Symp. High-PowerLaser Macroprocess. Eds. I. Miyamoto, K. F.Kobayashi, K. Sugioka, R. Poprawe, H. Hel-vajian, pp. 416–421. SPIE. 57. Panstar, H., Salminen, A., Jansson,A., and Kujanpaa, V. 2004. Quality andcosts analysis of laser welded all steel sand-wich panels. J. Laser Appl. 16(2): 66–72. 58. Kawahito, Y., Mizutani, M., andKatayama, S. 2007. Elucidation of high-power fibre laser welding phenomena ofstainless steel and effect of factors on weldgeometry. J. Phys. D: Appl. Phys. 40(19):5854–5859. 59. Ghany, K. A., and Newishy, M.2005. Cutting of 1.2-mm-thick austeniticstainless steel sheet using pulsed and CWNd:YAG laser. J. Mater. Process. Technol.168(3): 438–447. 60. Hsu, M. J., and Molian, P. A. 1995.Off-axial, gas-jet-assisted, laser cutting of6.35-mm-thick stainless steel. Trans. ASME117(2): 272–276. 61. Ilavarasan, P. M., and Molian, P. A.

1995. Laser cutting of thick sectionedsteels using gas flow impingement on theerosion front. J. Laser Appl. 7(4): 199–209. 62. Purtonen, T., and Salminen, A.2014. A study on the effect of cutting posi-tion on performance of fiber laser cuttingof stainless steel tubes. Weld. World 58(2):193–204. 63. Sheng, P. S., and Joshi, V. S. 1995.Analysis of heat-affected zone formationfor laser cutting of stainless steel. J. Mater.Process. Technol. 53(3–4): 879–892. 64. Sparkes, M., Gross, M., Celotto, S.,Zhang, T., and O’Neill, W. 2008. Practicaland theoretical investigations into inertgas cutting of 304 stainless steel using ahigh brightness fiber laser. J. Laser Appl.20(1): 59–67. 65. Stelzer, S., Mahrle, A., Wetzig, A.,and Beyer, E. 2013. Experimental investi-gations on fusion cutting stainless steelwith fiber and CO2 laser beams. 7th Int.WLT Conf. Lasers Manuf. Eds. C. Emmel-mann, M. F. Zaeh, T. Graf, and M. Schmidt,pp. 392–397. Elsevier Sci. BV. 66. Wandera, C., Salminen, A., and Ku-janpaa, V. 2009. Inert gas cutting of thick-section stainless steel and medium-sectionaluminum using a high power fiber laser. J.Laser Appl. 21(3): 154–161. 67. Kuo, T. Y., and Jeng, S. L. 2005.Porosity reduction in Nd-YAG laser weld-ing of stainless steel and inconel alloy byusing a pulsed wave. J. Phys. D: Appl. Phys.38(5): 722–728. 68. Rizzi, D., Sibillano, T., Calabrese, P.P., Ancona, A., and Lugara, P. M. 2011.Spectroscopic, energetic, and metallo-graphic investigations of the laser lapwelding of AISI 304 using the responsesurface methodology. Opt. Lasers Eng.49(7): 892–898. 69. Hartwig, L., Ebert, R., Kloetzer, S.,Weinhold, S., Dreschel, J., Peuckert, F.,Schille, J., and Exner, H. 2010. Materialprocessing with a 3 kW single mode fibrelaser. J. Laser Micro/Nanoeng. 5(2):128–133. 70. Kawahito, Y., Mizutani, M., Kataya-ma, S. 2009. High quality welding of stain-less steel with 10 kW high power fibrelaser. Sci. Technol. Weld. Joining 14(4):288–294. 71. Mills, K. C. 2002. Recommended Val-ues of Thermophysical Properties for SelectedCommercial Alloys. Cambridge, Woodhead. 72. Blecher, J. J., Palmer, T. A., and De-bRoy, T. 2014. Solidification map of a nick-el base alloy. Metall. Mater. Trans. A 45A(4):2142–2151. 73. Brohnstein, I. N. 2007. Handbook ofMathematics, 5th Edition. New York, N.Y.,Springer. 74. Halliday, D., Resnick, R., and Walk-er, J. 2005. Fundamentals of Physics, 7th Ed.Hoboken, N.J., John Wiley & Sons, Inc. 75. Lin, L., and Eagar, T. W. 1986. Pres-

sures produced by gas tungsten arcs. Met-all. Trans. B. 17B(3): 601–607. 76. Rai, R., and DebRoy, I. 2008. Nu-merical simulation of heat transfer andfluid flow in GTA/ laser hybrid welding.Sci. Technol. weld. Joining 13(8): 683–693. 77. Zhang, W., Kim, C.-H., and DebRoy,T. 2004. Heat and fluid flow in complexjoints during gas metal arc welding — PartII: Application to fillet welding of mildsteel. J. Appl. Phys. 95(9): 5220–5229. 78. Basu, S., and DebRoy, T. 1992. Liq-uid metal expulsion during laser irradia-tion. J. Appl. Phys. 72(8): 3317–3322. 79. He, X., Norris, J. T., Fuerschbach, P.W., and DebRoy, T. 2006. Liquid metal ex-pulsion during laser spot welding of 304stainless steel. J. Phys. D: Appl. Phys. 39(3):525–534. 80. Garandet, J. P., Vinet, B., and Gros,P. 1994. Considerations on the pendantdrop method: a new look at Tate’s law andHarkin’s correction factor. J. Colloid InterfaceSci. 165(2): 351–354.

Appendix A

Calculation of Surface Tensionand Liquid Metal Weight

To calculate the force balance be-tween surface tension and liquid metalweight, the approximated weld pool issplit into two general parts as illustrat-ed in Fig. A1. The leading segment ofthe weld pool, where the laser is inter-acting with the liquid metal, is round-ed and semicircular in shape and ex-tends through the thickness of theplate. The trailing section of the weld,especially at higher welding speeds,can take a triangular appearance at thetop surface but does not extendthrough the entire plate thickness.There is some boundary between thepool and solidified material that ex-tends from the end of the pool on thetop surface to the edge of the pool onthe bottom surface. Similarly, the vol-ume of the idealized weld pool, shownin Fig. A1, is split into two parts, acylinder surrounded by half of a trun-cated cone, which are numbered 1 and2, respectively, near the heat sourcesand an overlapping trapezoid basepyramid, which is divided between arectangular base pyramid (3) and twotriangle base pyramids (4 and 5). Thisidealized volume is comparable to nu-merical modeling results of partial-penetration laser welds made undersimilar conditions with depths of pen-etration close to the plate thickness


WELDING RESEARCH


used here (Ref. 72). The various shapes are numberedas 1) cylinder, 2) truncated cone, 3)rectangle base pyramid, and 4) and 5)triangle base pyramid. The volumesare calculated as (Ref. 73)

V1 = 1⁄4πd2t (A1)

V2 = 1⁄24(πt)(D2 + d2 + Dd)− 1⁄2V1 (A2)

V3 = 1⁄3dtl − 1⁄2V1 (A3)

V4 = V5 = 1⁄12(D − d)tl (A4)

where d is the bottom surface weldwidth, t is the plate thickness, D is thetop surface weld width, and l is thedistance from the position of maxi-mum weld width to the trailing edge ofthe weld pool. Plate thickness isknown, and the top and bottom sur-face widths can be measured from theweld bead. The distance, l, can be esti-mated from the top surface weld bead.Three measurements were made forthe values D, d, and l in each weld andare given in Table 2 along with thestandard deviations. Most of the liquid steel in the weldis supported partially by underlyingsolid material, which either solidifiedor never melted. So, the entire weightof the liquid metal is not supported bythe surface tension force at the bot-tom of the pool. This situation is ac-counted for by assuming each columnof liquid is supported by a surface act-ing as an inclined plane (Ref. 74). Thevolume of liquid unsupported by anysurface is the cylinder lying in the cen-ter of the truncated cone. The force ofthe liquid metal in the idealized weldpool that must be supported by thesurface tension force is

Fg = g[V1 + (V2 + V4 + V5 + Vs)sin21

+ V3sin22] (A5)

where is the density of liquid iron,7200 kg/m3 (Ref. 24), g is accelerationdue to gravity, 1 is the angle that thecone and the triangle base pyramidsmake with the bottom of the plate,and 2 is the angle that the rectangle

base pyramid makes with the platebottom. Vs is a small volume that isnot considered in the five volumes andis determined by disc integration oftriangle ABC in Fig. 7B. Some of the welds considered arehybrid laser-arc welds, where fillermetal is added to the molten pool. Inthis case, the filler metal was assumedto be spread evenly over the area ofthe idealized weld with each of the fivevolumes increasing based on top sur-face area fractions of each shape. Thevolume of filler metal added over thelength of the weld pool is

where vw is the wire feed speed, dw isthe diameter of the filler metal wire,and U is the welding speed. The firstfraction describes the volume of fillermetal added to the weld per unit time.The time necessary for the arc to tra-verse the length of the weld pool is cal-culated in the second fraction. Addi-tionally, arc pressure from the plasmawill act on the weld pool during hybridlaser-gas metal arc welding. The totalarc force has been measured by Linand Eagar (Ref. 75) in gas tungsten arcwelds for various torch angles andwelding currents. The estimated forces(Ref. 75) are listed in Table 2, andsince the arc is located over the rectan-gle base pyramid, the forces have beenadded to the rectangle base pyramid. In the above derivations, recoilforce and droplet impact force havenot been taken into account. The re-coil force results from evaporation atthe molten pool surface, and thedroplet impact force is due to the addi-tion of liquid metal to the pool fromthe consumable electrode. The recoilpressure is a function of the equilibri-um vapor pressure, which itself de-pends on the temperature of the liquidmetal. Using the vapor pressure of liq-uid iron (Ref. 24), the recoil force (Ref.76) with a 46 mm2 weld pool surfacearea and 2500 K surface temperatureis 0.4 mN. Due to the relatively lowvalue of recoil force at a fairly hightemperature, the effects of recoil pres-sure were neglected. Similarly, when

the droplet mass and acceleration areestimated (Ref. 77) for the weldingconditions studied, the maximumforce is less than 1 mN, which is negli-gible compared to the other forces. The surface tension force is the re-straining force that holds the liquidmetal in the pool and prevents the for-mation of the root defect. This force iscalculated as (Refs. 78,79)

Fs=d (A7)

where is the surface tension of liquidsteel at the melting point. In thisgeometry, the root of the weld poolacts as a pendant drop prior to detach-ment (Ref. 80). According to this for-mulation, the maximum surface ten-sion force is normal to the plate. Thedetermination of the surface tensionfor the pendant drop is calculated thesame as Eq. A7 when the force is setequal to the weight of the droplet. Forthe plate that was ground on the bot-tom, the surface tension was taken as1.91 N/m (Ref. 22), the value for pureiron. For those welds made on platescontaining oxide scale on the bottomsurface, the surface tension was takenas 0.88 N/m (Ref. 21), a value consis-tent with oxygen impurities in iron.

VV d D l

UMw w

4

12 (A6)

2= π + Fig. A1 — The idealized weld pool,

which is used to estimate the volumeand weight of the liquid metal, isshown as — A — A topdown view outlining the five volumes used to calculatetotal volume; B — a 3D view with thevarious variables, dimensions, and individual volumes.

A

B


mitigation of root defect in laser and hybrid laserarc ... research march 2015 / welding journal...

Documents