FluxCoating Development for SMAW Consumable …files.aws.org/wj/supplement/WJ_2014_08_s271.pdf ·...

Introduction

With the development of a weldcomposition that demonstrates abili-ties to mitigate ductility dip cracking(DDC) and hot cracking in high-nickelalloys while maintaining their hightemperature and corrosion resistance,an existing GTAW wire was used as anSMAW core rod with a flux coating de-signed specifically to meet the new

predefined target weld metal composi-tion. This work presents a designmethodology for developing a fluxcoating formulation to convert aGTAW wire into a SMAW consumableelectrode for high-nickel alloys.

The primary functions of the fluxcoating of an SMAW consumable elec-trode are to provide arc stability, pro-tect the molten metal from theatmosphere, and refine the weld pool(Refs. 1, 2). Other functions of a flux

coating include providing a slag layerfor a smoother weld contour and beadshape, and surface cleanliness of theweld metal. Additionally, the flux coat-ing should be formulated with ingredi-ents that produce a low-density slag,provide proper viscosity for out-of-position welding, promote slag detach-ability, and reduce spatter and fume.

Fluxes can be separated into acidic,basic, and neutral depending on whichoxides are dominant as flux ingredi-ents. The combination of these oxidesdetermines the slag properties such asmelting temperature, thermal expan-sion, density, electrical conductivity,and viscosity. Acidic oxides such as sili-cates are primarily slag or network for-mers. Basic oxides (CaO, MgO, Na2Oand K2O) decrease viscosity since theybreak down silica networks. Fluoridesand chlorides are also added to se-lected flux compositions as aggressivechemicals that assist in removingstrong oxides from the weld joints. Ad-ditions of fluorides decrease the vis-cosity of the slag, depending on thetype of fluoride. Slipping agents andbinding agents are added to improvethe flow of the wet flux mixture dur-ing extrusion and bind the flux ingre-dients to each other and to the corerod to form a solid coating. Sodium sil-icate, potassium silicate, or mixturesof the two are common binders usedto manufacture electrodes.

Welding arc stability is importantto the weldability of the electrode andthe integrity of the weld properties.Some ingredients used to enhance arcstability are potassium silicate, sodiumsilicate, rutile (TiO2), and potassium

WELDING RESEARCH

AUGUST 2014 / WELDING JOURNAL 271-s

FluxCoating Development for SMAW ConsumableElectrode of HighNickel Alloys

Extrudability, recovery analysis, and welding characteristics were incorporated into ashielded metal arc welding electrode for welding highnickel alloys

BY K. SHAM AND S. LIU

ABSTRACTThere is interest in improved highnickel alloy electrodes for shielded metal arc

welding (SMAW) to replace gas tungsten arc welding (GTAW) with cold wire feed forpressure vessel fabrication/repair in the powergeneration industry. Without alteringthe GTAW wire composition, a proper flux coating was designed to produce a SMAWelectrode to give an enhanced weld composition. SMAW electrodes were designed bydeveloping proper flux coatings for extrusion, welding characteristics, and weldintegrity. The flux formulation design began with an equal distribution of three primaryingredients: cryolite, rutile, and calcium carbonate to clean the weld, create a slag, andprovide shielding gas, respectively. The system was then optimized to enhance extrudability, slag detachability, electrode weldability, alloying recovery, arc stability, and anumber of weld properties. Detailed mass balance calculations were performed to determine alloying element transfer across the arc and to understand weld metal recovery from the flux coating, filler metal wire, and base metal. For chromium andmanganese, an average of 95 and 60% weld recovery was confirmed and determined,respectively. The elemental recovery data determined were important for calculatingthe proper additions of these elements into the flux formulation. Comparisons revealedcomparable welding characteristics for the inhouse manufactured and commercialSMAW electrodes. Arc voltage analysis was also performed to determine arc stability ofthese electrodes. Results indicated differences in metal transfer modes between thedifferent flux coating iterations, and the final electrodes exhibited more consistent voltagetime traces. This work demonstrates that even though it is common practice to usea GTAW wire to build a SMAW electrode, the flux coating can be designed tobeneficially alter the weld composition.

KEYWORDS•Covered Electrode • SMAW Consumable • Nickel Alloys • Flux Coating •Extrusion • Electrode Manufacture • Composition

K. SHAM and S. LIU are with Colorado School of Mines Golden, Colo.

SUPPLEMENT TO THE WELDING JOURNAL, AUGUST 2014Sponsored by the American Welding Society and the Welding Research Council

Sham 8-14_Layout 1 7/15/14 1:22 PM Page 271

titanate (Ref. 3). Even though potas-sium silicate acts predominately as thebinder in the flux coating, it plays acritical role in the stabilization of thearc in AC applications because potas-sium ionizes at a lower energy, 4.3eV/atom. Sodium ionizes at 5.1eV/atom and is therefore more readilyused as the binder for arc stabilizer ofDC applications where the ability toreestablish the welding arc is not asvital (Ref. 4). The addition of fluoridestypically produces a more erratic arc,despite its ability to remove surfaceoxides from the joint region.

Slag detachability is of great con-cern because the presence of slag onthe weld surface will influence the in-tegrity of multipass weldments and affect process economics. Rutile-con-taining fluxes generally produce slagsthat exhibit good detachability. On theother hand, basic flux systems typi-cally exhibit less than desirable wet-ting. Consequently, complete slagremoval may be more difficult in thesesystems. Slag removal has also beendocumented as more difficult when

the flux coating con-tains fluorides. Slagsthat contain thespinel (AO.B2O3)where A and B repre-sent cations such asMg and Al andcordierite(Mg2Al4Si5O18) glassstructure have alsobeen deemed diffi-cult to detach fromthe weld metal. An increase in aluminacontent in the flux composition hasshown improved slag detachability.Another important factor to considerfor slag detachability is the differencein coefficient of thermal expansion(CTE) between the slag and the weldmetal deposit (Refs. 5–7). It is desir-able to have a slag with a CTE differentfrom that of the weld metal. Too earlyslag detachment, however, may pres-ent undesired heavy temper colors(Ref. 1). To further explain the ther-mal expansion mismatch, Equation 1describes the strain that is produceddue to a temperature change at thebonding interface between the slagand the weld metal.

S = Δα(ΔT) (1)

S is defined as the strain, the Δα is thedifference between the thermal expan-sion of the slag and the weld metal, andΔT, known as the differential contrac-tion, is defined as the temperature dif-ference at the bonding interface duringthe welding thermal cycle (Ref. 8).

Another important aspect of theflux coating of the SMAW consumableelectrode is that it can be utilized as aconduit for alloying the weld metalcomposition. The flux coating ratio de-fined as the diameter of the core roddivided by the diameter of the SMAWelectrode affects the amount of alloy-ing elements recovered in the weldmetal (Refs. 9–11). There are threemain components that contribute tothe end weld composition: the corerod, the base metal, and the flux coat-ing. Thus, it is important to obtainchemical compositions for all threecomponents in order to calculate re-covery directly from the flux coating.

Delta quantity is described as theamount of a specific element that isgained or lost during the weldingprocess. The delta quantity is repre-sented by the following mathematicalequation:

(ΔQuantity)i = (Analytical Content)i – (Estimated Content)i

(2)The analytical content of element i isdetermined by chemical analysis of the

WELDING RESEARCH

WELDING JOURNAL / AUGUST 2014, VOL. 93272-s

Fig. 1 — Experimental GTAW wire in coil form.

Fig. 2 — Experimental GTAW wire decoiled, straightened, andcut into 356mm (14in.) lengths.

Fig. 3 — Schematic illustration of the extruder, top view (Ref. 14).

Table 1 — Commercial HighNickel Core RodCompositions

Element E1 (wt%) E2 (wt%)

Carbon 0.006 0.028Silicon 0.05 0.16

Chromium 16.8 27.3Manganese 2.5 0.37Niobium 0.06 0.02

Iron 8.75 10Nickel 71.3 60.4

Sham 8-14_Layout 1 7/14/14 4:39 PM 72

weld metal, whereas the estimatedcontent is determined by adding thecontributions from the core rod andbase metal. A negative delta quantitygives a loss of the specific element ei-ther to the slag or in a gaseous form. Apositive delta quantity implies fluxcontribution of a specific element,having been transferred to the weldmetal to achieve a certain target weldcomposition (Refs. 12, 13). The deltaquantity concept plays an importantpart in understanding the role of theflux ingredients in their contributionto the final weld composition.

Experimental ProceduresFlux coating chemistries for com-

mercial consumables were determinedusing a variety of techniques. The fol-

lowing techniques were all used to-gether: X-ray diffraction (XRD), energydispersive X-ray spectroscopy (EDS),wavelength-dispersive X-ray spec-troscopy (WDS), atomic absorption(AA), and inductively coupled plasmaoptimal emission spectroscopy (ICP-OES). The combination of chemicalanalysis techniques are used together todetermine both elemental compositionand crystalline phases to determinepossible mineral ingredients used to for-mulate the flux coatings.

Experimental wire of 3.2-mm (1⁄8-in.) diameter was procured in the formof a large coil (Fig. 1) that had to bedecoiled, straightened, and cut into356-mm (14-in.) lengths — Fig. 2.Table 1 gives the core rod composi-tions of the commercial SMAW elec-trodes ENiCrFe-3 (E-1) and ENiCrFe-7

(E-2) used for comparison in this re-search. Due to the sensitive nature ofthis work, the exact composition ofthe experimental wire is not revealed.

The experimental flux matrix pre-sented in Table 2 displays all the majorheats and changes to reach a final fluxcomposition titled, “CSM Final.” Aninitial formulation of a 1:1:1 massratio of cryolite (Na2AlF6), rutile, andcalcium carbonate (CaCO3) was takenas a starting point in the formulationprocess. The descriptors “decrease”and “increase” represent decreasing orincreasing quantities of an ingredientfrom the previous heat to promote no-ticeable variations, in particular, slagproperties or weld behavior. The ma-trix was designed to address the fol-lowing properties: extrudability, fluxcoating finish, slag viscosity, slag de-

WELDING RESEARCH


Fig. 4 — Pictorial depiction of diameter of filler metal rod and diameter of electrode.

Fig. 5 — Schematic of SMAW. Fig. 6 — A relationship between the correctamount of binder added and the percentageof relative humidity in the working area.

Table 2 — Experimental Flux Matrix for CSM HighNickel SMAW Electrode Development

Flux Heat 1 Heat 1A Heat 2 Heat 3 Heat 4 Heat 5 Heat 6 CSM FinalIngredient (wt%)

Cryolite 21.6 Decrease Increase Decrease Increase Increase – –Rutile 21.6 Decrease Increase – – – – Increase

Calcium Carbonate 21.6 Increase Decrease – – – – –Silicon Dioxide n/a – – – – Increase – –

Talc n/a – Increase – Decrease Decrease – –Kaolin Clay n/a Increase – Increase – Dercrease – –Bentonite Clay n/a – – – – – Increase DecreaseCellulose n/a – – – – – – IncreaseAlginate n/a – – – – – – Increase

Ferroniobium n/a – – – – – Increase –Manganese n/a – – – – – Increase –Chromium n/a – – – – – Increase –

Binder Bal. Bal. Bal. Bal. Bal. Bal. Bal. Bal.


tachability, effect of cryolite on weldcleanliness, welding characteristics,arc stability, alloying additions, andSMAW electrode mass production.

Manufacturing SMAW Consumable Electrodes

For the SMAW consumable elec-trode extruder, a total of 2 to 3 kg (ap-proximately 4 to 6 lb) of the wet mix isneeded to make a successful extrusionrun. The extruder used for the experi-mental small batch production for thiswork is capable of pressures rangingfrom 500 to 1000 lb/in.2 (3.5 to 6.9MPa) as compared to a commercial ex-truder averaging around 5000 to 6000lb/in.2 (35 to 40 MPa). Figure 3 gives adetailed top-view schematic of howthe extruder functions during extru-sion. A successful extrusion is depend-ent on many factors such asexperience with the extruder, success-ful adjustment of the eccentricity ofthe electrodes, proper preparation ofthe core rods, adding the right amountof binder and slipping agent, adequateflux coating ratio, and having a good

particle size distribution for the drymix. One major manufacturing aspectof concern is electrode eccentricity,which affects the weldability of theSMAW consumable electrode. Elec-trode eccentricity is measured by therelative position of the core rod withrespect to the layer of flux coating. Aneccentric electrode with uneven thick-ness coating surrounding the core rodwill produce an erratic arc with unevenmelting of the core rod and preferen-tial flaking of the flux coating.

Another important aspect of extru-sion is determining the appropriatethickness of the flux coating, which isadjustable during manufacturing. In-appropriate flux coating thickness canaffect the integrity of the weld and thecomposition of the weld metal. Thethickness of the flux coating is meas-ured as a flux coating ratio defined asthe ratio between the electrode diame-ter and the diameter of the core rod —Fig. 4. Commercial electrodes showedan average flux coating ratio of 0.4.Thus, a die size comparable to 0.4 wasused for the manufacturing of the in-house manufactured electrodes giving

an average flux coating ratio of 0.42.Calipers and image analysis were usedto verify results. After drying in air for24 h following extrusion, the elec-trodes were then baked in an oven at175°C (350°F) for one h, 260°C(500°F) for three h, and then 175°C(350°F) for one h. The drying sequencewas consistent with industrial practiceand selected after input from commer-cial electrode manufacturers (Ref. 11).

Welding Process

Using ASTM A36 base plates andInconel® 690 base plates for the finalwelds, bead-on-plate welds were de-posited using a Hobart TIG-300 con-stant current power supply. Usingdirect current electrode positive(DCEP) configuration, an average lin-ear heat input of 12 kJ/mm (0.5kJ/in.) was obtained. Based on themanufacturer-recommended weldingranges for E-1 (90–110 A) and E-2(75–100 A), the respective averages of100 A and 88 A were adopted for weld-ing for the 3.2-mm- (1⁄8-in.-) diametercommercial electrodes. The optimalwelding current for the in-house man-ufactured electrode was subsequentlydetermined to be 85 to 95 A. Table 3gives the welding parameters for E-1,E-2, and the “CSM Final.” Arc signalswere acquired during the weldingprocess for the bead-on-plate samplesfor both the commercial and in-housemanufactured electrodes. These volt-age signals were obtained using a data

WELDING RESEARCH


Fig. 7 — Particle size distribution for coarse, fine, and50% coarse and 50% fine combined for rutile.

Fig. 8 — Beadonplate weld produced with Heat 2 flux composition. Part ofthe slag was retained for illustration.

Fig. 9 — Beadonplate weld produced with Heat 3 flux formulation withdecreased cryolite content.

Table 3 — Welding Parameters for E1, E2, and CSM Final

Electrode E1 E2 CSM Final

Electrode Feed Rate (mm/s) 4.9 4.9 5.1Travel Speed (mm/s) 2.1 2.1 1.7Deposition Rate (g/s) 0.57 0.62 0.45Welding Time (s) 36 43 42


acquisition system with a samplingrate of 1000 Hz and LabView® 7.1.These digital voltage signals wererecorded for the duration of everyweld, which ranged from 35 to 50 s.

(3)

Chemical Compositions andRecovery Calculations

The chemical compositions of thebase metal, all the filler metal wires,and welds were analyzed using OES.Filler metal wires were also meltedinto buttons to simulate the chemicalrecovery of arc welding in an inert at-mosphere. Alloying element recoverycalculations were done with the meas-ured compositions. Dilution calcula-tions were performed on allbead-on-plate welds in order to calcu-late alloying element recovery in theweld metal. Dilution percentage of abead-on-plate weld was calculated as37.1%. With these known contribu-tions from the base metal and thefiller metal, the weld composition canbe determined numerically. Weldmetal recovery was determined foreach of the major alloying elements:niobium, manganese, and chromium.

A mass balance calculation wasconducted to determine the actualamounts of the electrode consumed(both weight and length) to form theweld. The amount of weld deposit(area of deposit and area of penetra-tion) was also determined using Im-ageJ® for the mass balancecalculations. ImageJ is a public do-main image processing software de-veloped at NIH (National Institute ofHealth). With all the data necessary, aseries of calculations using the equa-tions below was performed in se-

quence. Figure 5displays all as-pects of SMAWto assist in theunderstandingof the recoverycalculationsthat follow.

(4)

(5)

(6)

(7)

(8)

(9)

After completing calculations usingEquations 3–9 for each element of in-terest, the following Equation 10 isused to determine the actual overallweld recovery:

(10)

where WMR is weld metal recovery, Zrepresents the concentration of the el-ement (i) in discussion, w is the weldmetal, f is the flux coating, r is thefiller metal core rod, and bm is thebase metal.

Results and Discussion

The flux coatings of the two com-mercial nickel-alloy electrodes, E-1 andE-2, were examined as references forcomparison with the flux formulationeffort in this investigation. Table 4 pres-ents the compositions of the E-1 and E-2 flux coatings. As can be seen, thethree principal ingredients in the fluxcoating of commercial high-nickelSMAW electrodes are cryolite, rutile,and calcite. E-1 contained a fourthmajor ingredient, feldspar, which is gen-erally treated as filler material, addedmainly for slag volume. A combinationof strontium and calcium carbonate wasused in the commercial flux coatings be-cause strontium carbonate decomposesat a higher temperature, 1100°C(2012°F), than calcium carbonate andprovides better coverage to the weldpool. An additional advantage of stron-tium carbonate is that it is not as hygro-scopic as calcium carbonate. Picking upless moisture, the electrode flux coatingis expected to produce a more stable arc.The E-2 flux coating contained a smallamount of silica and some other alloy-

AFCC g

RFC in

IFC inAFC g

( ) =

− ( )( )

⎛⎝⎜

⎞⎠⎟

⎛

⎝⎜⎞

⎠⎟× ( )1

( ) ( )( )

=

×−⎛

⎝⎜⎞⎠⎟

FM g AFMC g

OES FM wt%

100

AFMC GRE in

IE inAFM g( ) = − ( )

( )⎛⎝⎜

⎞⎠⎟

⎛

⎝⎜⎞

⎠⎟× ( )1

( ) ( )( )=

×⎛⎝⎜

⎞⎠⎟

FC g AFCC g

FC wt%

100

)()

)

(

(

=

× ×

BMP g AP cm

LW cmg

cc

7.8

2

( ) ( )( )

=

×−⎛

⎝⎜⎞⎠⎟

BM g BMP g

OES BM wt%

100

( )( ) ( ) ( )( )

= +

×−⎛

⎝⎜⎞⎠⎟

WM g BMP g WMD g

OES WM wt%

100

WMRZ

Z Z Ziw

f r bm i

=+ +

⎛

⎝⎜

⎞

⎠⎟ × 100%

WELDING RESEARCH


Fig. 11 — Ternary phase diagram of CaOTiO2SiO2 (Ref. 15) withcompositions marked.

Fig. 10 — Beadonplate produced with CSM Final flux formulation with cryolite adjusted to 24 wt%.


ing elements for weld composition ad-justment such as manganese, nickel,and chromium. Due to the inert natureof nickel, the value and importance of achemically aggressive flux ingredientcannot be overemphasized. Togetherwith fluorides that are chemically ag-gressive for cleaning the weld joint, cry-olite prevents joint surfaces fromoxidation and improves the cleanlinessof the weld metal surface. Rutile prima-rily functions as a slag former and arcstabilizer. Calcium carbonate decom-poses around 890°C (1634°F) into CaOand CO2. CaO serves as a slag former byreacting with TiO2 and SiO2. CO2 is theshielding gas that displaces the air fromthe weld joint and blankets the entireweld region. The performance of eachflux formulation determined theamounts of addition or subtraction ofeach of the ingredients for the subse-quent formulation.

Evaluation of Electrode Extrudability

Extrusion of SMAW electrodes is nota process that can be easily quantified.It depends on a number of variables in-cluding the type and source of the min-eral, the appropriate amount of water,the particle size distribution, amongothers. The extrusion variables studied

were the amount of binder, the mixingtechnique, the slipping agents, and theparticle size distribution of the dry mix.The extrusion environment is also criti-cal. For example, relative humidity andtemperature affect significantly theflow characteristics of the wet mix dur-ing extrusion. An extrudable flux canquickly lose moisture and become diffi-cult to extrude during the time of man-ufacturing one batch of electrodes.Therefore, this process is subjective andbased on experience in observing theconsistency of the wet mix, the fluidityof the wet mix, and the adhesion be-tween the core rod and the flux coating.

Extrusion was performed on eachof the experimental flux formulations,including the base formula, to verifythe extrudability of the flux mixtures.As expected, the first flux formula-tion, Heat 1, was not extrudable be-cause an excessive amount of thebinder (at ~35 wt-%) was utilized.Since the binder is typically 50% sili-cate and 50% water, some data werecollected to determine if the amountof binder added is dependent on thehumidity of the manufacturing envi-ronment. The correct amount ofbinder is dependent on the relative hu-midity of the working area at the timeof mixing and extrusion. The relativehumidity in the laboratory fluctuated

between 34 and 55% with an averagerelative humidity at 42%. The averageamount of binder added to the dry mixto form the wet mix was 30 wt-% witha standard deviation of less than orequal to 2.5 wt-% depending on thehumidity. When the relative humiditywas above the average relative humid-ity of 42%, the amount of binder wasless than 30 wt-% but within the devi-ation of 2.5 wt-%. Figure 6 gives agraphical representation of the rela-tionship between binder amount andrelative humidity in the working area.The point with the highest relative hu-midity of 95% was data recorded dur-ing the mass production of theelectrodes at a commercial electrodemanufacturer in Houston, Tex. It canbe seen that the amount of binderneeded in making the wet mix de-creases as the relative humidity of thework area increases with 23.5 wt-% ofbinder being the lowest amount ofbinder needed.

The poor extrudability of Heat 1 ledto a decrease in binder and the addi-tion of a slipping agent, kaolin[Al2Si2O5(OH)4], to the flux paste toimprove electrode extrusion. Heat 1Ahad improved extrudability with toomuch slipping agent added; the fluxpaste flowed faster than the rod inter-mittently even though the velocity ofthe compressor was decreased to aminimum. For Heat 2, a combinationof kaolin and talc, Mg3Si4O10(OH)2,was used in a decreased amount fromHeat 1A. These two slipping agentsgreatly improved extrudability of thebase flux coating. For the latter extru-sion runs of the nickel SMAW consum-able electrodes, including the largerbatches, bentonite clay,[Al2O3(4SiO2H2O)] was used instead

WELDING RESEARCH


Fig. 12 — Self releasing slag systems, E1 (top) and CSM Final (bottom).

Fig. 13 — Heat 2 beadonplate exhibiting poor slag detachability.

Table 4 — E1 and E2 Flux Coating Composition

Mineral E1 (wt%) E2 (wt%)

Cryolite (Na3AIF6) 21 25Rutile (TiO2) 14 17

Strontium Carbonate(SrCO3) and Calcium 24 28Carbonate (CaCO3)

Silicon Dioxide (SiO2) n/a 7Feldspar

(K2O.Al2O3.6SiO2) 23 n/aManganese 13 12

Nickel 1 2Iron n/a 3

Chromium n/a 3


of kaolin clay due to what was avail-able. Other than the difference inchemical composition, both bentoniteclay and kaolin clay, and talc functionas slipping agents to ease extrudabil-ity. Availability, proper particle sizedistribution, and desired weld compo-sition may be the deciding factors forselecting one and not the others. Fi-nally, a small quantity of cellulose wasadded to enhance homogeneous mix-ing of the wet mix during the commer-cial mass production. To maintainextrudability and minimize water ab-sorption, a small amount of alginatewas added as a slipping agent.

In Heat 5, the amount of slippingagent was further reduced and a betterparticle size distribution for the dry mixwas used to maintain extrudability. Inaddition to the extrusion aspect, parti-cle size distribution of the dry mix alsoaffects the flux coating finish of aSMAW electrode. Using the combina-tion of slipping agents, proper mixing

technique, the correctamount of binder, and an im-proved particle size distribu-tion, “CSM Final” gave a

much improved extrusion experience.The dry mix of the CSM Final was opti-mized by using 50 wt-% fine grades and50 wt-% coarse grades for the calciumcarbonate and rutile. Figure 7 depicts anexample of how the combination of thefine and coarse grades of rutile affectedthe overall particle distribution. Ascompared to Heat 1, which consisted oflarger quantities of 325 and 400 meshsize, “CSM Final” had lesser quantitiesof 325 and 400 mesh size and displayeda more normal distribution around 200mesh, which is the practice of industryformulators. The CSM Final electrodeshave good extrudability and their fluxcoatings have a smooth appearance.

Weld Cleanliness

The first experimental Heat 1 of theflux formulation used an even distribu-tion of cryolite, rutile, and calcium car-bonate as the starting point. Heat 1Aand Heat 2 contained 18–19 wt-% of

cryolite, which produced weldmentswith good surface cleanliness. Surfacecleanliness was inspected visually with ashiny, silvery weld having good surfacecleanliness. Figure 8 depicts a bead-on-plate weld produced with Heat 2 fromthe experimental flux coating designmatrix displaying good surface cleanli-ness. To test the effectiveness and opti-mal amount of cryolite as a chemicallyaggressive flux ingredient to enhancethe cleanliness of the weld metal, theamount of cryolite was decreased from19 wt-% in Heat 2 to 13 wt-% in Heat 3.As a result, the bead-on-plate weld from

WELDING RESEARCH


Fig. 14 — Voltage data for E1.

Fig. 16 — Voltage data for Heat 2.

Fig. 15 — Voltage data for E2.

Fig. 17 — Voltage data for CSM Final.

Table 5 — Flux Ingredients Broken Down intothe Slag Form

Flux Ingredients Primary Oxidesin Slags

Na3AIF6 Na20, Al2O3TiO2 TiO2CaCO3 CaOAl2Si2O5(OH)4 Al2O3, SiO2Mg3Si4O10(OH)2 MgO, SiO2SiO2 SiO2K2SiO3.nH2O K2O, SiO2

⇒


Heat 3 produced a dull, oxidized weldmetal surface finish — Fig. 9. In Heat 4,cryolite was increased back to the 19wt-% used in Heat 2 to restore weldcleanliness. In Heat 5, cryolite was fur-ther increased to optimize weld cleanli-ness. The “CSM Final” flux formulationoptimized cryolite to 24 wt-% and therespective bead-on-plate weld can beseen in Fig. 10. The “CSM Final” bead-on-plate weld displays the lustrous, sil-very metallic appearance of nickel thatHeat 3 (Fig. 9) does not with the de-creased amount of cryolite. Cryoliteplays a critical role in the cleanliness ofthe weld metal surface. The amountmust be optimized for weld cleanlinesswithout affecting the viscosity of theslag flow since fluorides act as networkbreakers and decrease the viscosity ofthe slag. Part of the design methodol-ogy for flux coatings in a SMAW con-sumable electrode is to balance themultiple functions of each flux ingredi-ent to optimize extrudability, weldabil-ity, and weld integrity.

Slag Detachability

Slag detachability is defined as theease with which the solidified protec-tive slag layer can be removed. Poorslag detachability can negatively im-

pact productivity and weld integrity.The differences in the CTE betweenthe slag and the weld metal play avital role in the ability for the slag todetach easily.

Using a phase diagram of the CaO-TiO2-SiO2 ternary system, shown inFig. 11, mass balance equations wereused to estimate the glass phases pro-duced in the slag. To perform thesemass balance equations, the flux in-gredients were first broken down totheir respective slag form (expressedin the form of primary oxides)—Table5). Similar oxides were then groupedtogether such that the ternary systemdiagram of CaO-TiO2-SiO2 can beused. Na2O, MgO, and K2O weregrouped together with CaO, and Al2O3was grouped with TiO2. This approxi-mation does not give an exact compo-sition of the slag, but themethodology allows for the estimationof the predominant glass phase in theslag. The determination of the domi-nant glass phase allows the estimationof the differences in the CTE valuesbetween the slag and weld metal,which helps predict slag detachability.Values of linear CTEs were collectedfrom literature for the weld metal andthe various crystalline phases thatcould be formed based upon the com-

position of the flux coating. These val-ues are shown in Table 6. With thecompositions of the E-1 and E-2 fluxcoatings, calculations were made to de-termine the respective crystallinephases that could form in the resultingslags. As can be seen in Fig. 10, thetwo commercial slags were determinedto produce the perovskite phase,which has a larger CTE, 15.8 ×10–6/°C, than that of the weld metal,10.4 × 10–6/°C (Refs. 15, 18). The CTEof Inconel® 690 alloy is used for theweld metal since it is the most similarto the composition of the experimen-tal weld wire used.

Figure 11 shows the ternary systemwith the respective phases that formfor Heats 2 through “CSM Final,” andthe commercial electrodes. Examina-tion of this figure shows that the com-mercial electrode compositions fellinto the perovskite phase. Heat 2 pro-duced a dominant sphere phase in theslag with a CTE of 6 × 10–6/°C, whichis smaller than the weld metal, 10.4 ×10–6/°C, and displayed poor slag de-tachability. Heat 3 also produced asphene phase with poor slag detacha-bility (Ref. 15). Furthermore, the slagdeveloped a porous layer that adheredtenaciously to the weld metal, which isan indication that a thin spinel phaseis likely to be present on the surface ofthe weld bead. The occurrence ofspinel on top of a weld bead is usuallyin the form of a thin layer on the orderof several atomic layers thick. Thethermal strain associated with con-traction and expansion of this layer istypically small. As such, the spinellayer is expected to have great adhe-sion with the weld bead surface.

Heat 4 electrodes produced a domi-nant perovskite phase in the slag,which exhibited much better slag de-tachability due to the higher CTE val-ues. With a CTE value that is higherthan that of the weld metal ofInconel® 690 alloy, the perovskite

WELDING RESEARCH


Table 8 — Known Alloy Recovery from SMAW Flux Coating (wt%) (Ref. 5)

Alloy Element Form of Material in Electrode Covering Approximate Recovery of Element, wt%

Aluminum Ferroaluminum 20Boron Ferroboron 2

Carbon Graphite 75Chromium Ferrochromium 95Niobium Ferroniobium 70Copper Copper Metal 100

Manganese Ferromanganese 75Molybdenum Ferromolybdenum 97

Nickel Electrolytic nickel 100Silicon Ferrosilicon 45

Titanium Ferrotitanium 5Vanadium Ferrovanadium 80

Table 7 — Weld Recovery Results (wt%)

Elements Heat 2 (%) Alloy 1 Heat 2 (%) Alloy 2 E2 (%) E1 (%)

Chromium 100 100 100 88Manganese 54 65 47 51

Niobium 89 100 n/a* n/a*Iron 100x 100 94 95

Nickel 83 99 100 100

*Note — Niobium composition was not determined.

Table 6 — Room Temperature CTE Values ofTwo Glass Phases and Inconel 600 and 690Alloy (Refs. 15–18)

Materials CTE at RT (10–6/°C)

Sphene (CaTiSiO5) 6

Perovskite (CaTiO3) 15.8

Inconel 690 alloy 10.4


phase will contract much more thanthe weld metal upon solidification.The greater tensile stress induced inthe slag phase will promote better slagdetachability. In order to maintain slagdetachability for CSM Final flux for-mulation, rutile was increased to enhance slag formation. The observa-tions described above help to explainthe slag detachability behavior ob-served in the experiments. CTE differ-ences between the dominant slagphase and weld metal can be used as atool to help design flux coatings withenhanced slag detachability.

Weld Metal Recovery

Following the series of calculationsdiscussed in the experimental proce-dures section, Table 7 gives the weldrecovery results of importance to theintegrity of the weld metal. Chromiumgives an experimentally determinedaverage of 97% weld recovery with re-spect to the three sources of contribu-tion, the core rod, the flux coating,and the base metal. Manganese dis-played an average weld recovery ofabout 54%. This value was slightlylower than that typically used (around70 to 80%) in the industry. However,being the major deoxidizer in the weldsystem, a low recovery rate is not en-tirely surprising. Niobium was onlycalculated for the in-house electrodesat an average recovery value of about94%. The commercial electrodes werenot taken into consideration becausethe niobium content in those fluxeswas not determined. The ASM MetalsHandbook Vol. 6 provided niobium re-covery from the flux ingredient in theform of ferroniobium as 70% alongwith additional known alloy recoveryfrom SMAW flux coating (Table 8)(Ref. 5). Another element of impor-tance is the iron content, which mustbe carefully monitored because it too,like chromium, forms the undesirableM23C6-type carbide. Average meas-

ured values in this research gave ironrecovery at 97%. Lastly, average recov-ery value of nickel was also calculated tobe about 95%. These recovery values areimportant for the calculation of neces-sary additions in the flux coating forfinal weld composition.

Another series of recovery calcula-tions was performed on the buttonmelts to understand the percentage ofrecovery directly from the core rodswithout other influences. These re-sults are shown in Table 9. Chromiumand iron contents were almost 100%in recovery in an ideal welding atmos-phere. The recovery of manganesefrom the commercial core rods (Alloy1) averaged about 95% whereas the re-covery of manganese from the in-house manufactured Heat 2 (Alloy 2)averaged about 89%. It is clear withthis information that the replenishingof manganese must be made via flux.The ASM Metals Handbook reportedrecovery from ferromanganese flux asabout 75%. Alloying through the fluxmust be done with the appropriate in-gredients to minimize iron increase.Niobium recovery from the experi-mental core rods was relatively high atan average of 94%. Iron recovery fromthe core rods was reported as an aver-age of 76%. The different recoveryrates from the core rod and from theflux coating reported in this work areinvaluable for future efforts in flux de-sign and formulation.

The findings showed that it was pos-sible to obtain reliable recovery valuesfrom all four of the most important al-loying elements. These values were usedto incorporate alloying additions in theflux coating design and optimization

once the base flux formula was com-pleted. The recovery rate used forchromium additions in the flux was95%. A value of 60% was used for themanganese recovery in the flux coating.Since there were no calculated values ofrecovery percentages for niobium, theASM Metals Handbook Vol. 6 value of70% was taken as the recovery rate foralloy additions in the flux coating.

Using the recovery rates determinedpreviously, mass balance calculationswere performed to determine theamounts for replenishing the alloyingelements. Two empirical values wereused to predict the percentage of the al-loying elements additions to achieve thetarget weld composition. Each electrodeweighed an average of 20.5 g and the av-erage weight of the flux coating was 10g. These data originated from averagecalculations of weighed core rods andelectrodes manufactured in-house atCSM. For chromium, a calculated 5%addition into the flux coating is re-quired. For manganese and niobium,the calculated values were 3 and 4.1%,respectively. Using these calculated val-ues, the target end weld compositionwas reached. These three main alloyingadditions were reached in each of thelarge batch extrusions demonstratingthe robustness of the compositionrange. Note also that the last three ex-truded flux formulations possessed thesame cryolite, calcium carbonate, rutile,ferroniobium, chromium, and man-ganese additions with small adjust-ments to other minor flux ingredientsto obtain the correct amount of siliconand iron in the target weld composition.

The final base flux coating alsoserved as the starting point for alloy-

WELDING RESEARCH


Table 9 — Recovery Results from the ButtonMelts with Respect to the Core Rods (wt%)

Elements Alloy 1 (wt%) Alloy 2 (wt%)

Chromium 99 100Manganese 95 89

Niobium 100 89Iron 62 60

Nickel 100 100

Table 10 — Comparison of Core Rod and NoDilution Weld Composition Compared to a PredefinedTarget Weld Composition (wt%)

Elements Target Weld Experimental Heat 5 Heat 6Composition Core Rod

(HD 52)

Cr 27 26.8 24.7 27.2Fe 2.5 2.6 2.91 4.9C 0.035 0.012 0.009 0.02Ti 0.3 0.15 0.03 0.05

Nb 2.5 1.82 1.11 2.58Mn 3 3.02 2.12 2.97Ni Bal Bal Bal BalS <0.0015 0.001 0.001 <0.001P <0.005 0.001 0.001 0.005Si 0.05 0.08 0.58 0.60

Mo 0.01 0.12 0.11 0.10Cu 0.1 <0.1 0.0015 0.05


ing. Table 10 gives the chemical com-position of the experimental high-nickel core rod (Alloy HD 52) used formanufacturing the SMAW consumableelectrodes. Even though most of theelements are very close to the targetweld composition (also defined inTable 10), alloying elements were lostthrough the SMAW process, thus re-quiring replenishment of chromium,niobium, and manganese as done inHeat 6. Using 95% recovery rate forchromium, 70% recovery rate for nio-bium, and 60% recovery rate for man-ganese, the compositions of thesethree elements were restored. Table 10shows that the chromium, niobium,and manganese values in the no dilu-tion weld composition paralleled thoseof the target weld composition.

The overall weight-percentage ofrutile was increased approximately 5wt-% to increase the oxygen content,which subsequently lowered the ironrecovery without compromising nio-bium recovery. Silicon dioxide was re-moved to decrease the amount ofsilicon in the weld, but increasing therutile content compensated for the re-moval of silicon dioxide to minimizesilicon recovery since rutile is also anetwork former. Talc was also re-moved and one percentage of alginatewas added instead as a better slippingagent without introducing more sili-con into the flux system. The amountof bentonite clay was decreased a fewpct. to lower the aluminum and siliconin the flux coating. For CSM Final for-mulation, the iron decreased from 4.9to 3.6 wt-% and the silicon contentwas decreased from 0.6 to 0.1 wt-%.To achieve the predefined target com-position of the no dilution weld metal,talc was completely removed and ben-tonite clay was decreased significantly.

Comparison of Qualitative andQuantitative WeldingCharacteristics

General qualitative and quantitativewelding observations were documentedfor the characterization of the overallwelding characteristics that each elec-trode displayed. Table 11 gives the qual-itative characteristics for thecommercial and in-house manufacturedelectrodes according to input from agroup of professional welders. Compar-ing four experimental electrodes andtwo commercial electrodes, the bestperforming electrode was the CSM Finalwith a soft and stable arc, self-releaseslag, and thick, homogenous slag cover-ing. Professional welders use the terms“harsh” and “soft” to describe their ob-servation of the welding arc. A harsh arcgenerally implies a stronger arc force,but is more erratic in nature, and moredifficult for a welder to control. A softarc, on the other hand, is more uniformand easier for a welder to control. Feed-back from several industrial weldersconfirmed these observations. Figure12 shows the slags collected from E-1and CSM Final to demonstrate goodslag detachability. In contrast, slag de-tachability for Heat 2 was very poor asshown in Fig. 13, which shows the slagadhered tenaciously to the weld metal.All slag-related measurements weretaken at a constant electrode melt rate.Slag thickness measurements weretaken at various locations along the slagand consisted of the center and the sidewalls along the slag. The average thick-nesses of E-1, E-2, and the CSM Finalslag systems are given in Table 12. Thecommercial consumable designated E-1had the thinnest average slag thicknessat 1.52 mm. E-2 had an average slag

thickness of 1.91 mm. The highest aver-age slag thickness was the CSM Final at2.03 mm.

Arc Quality

The arc characteristics were de-scribed previously in a qualitative man-ner (Table 11). Arc voltage data werecollected to characterize the arc stabilityand relate with the metal transfermodes during welding. During welding,E-1 and CSM Final both exhibited softand stable arcs. E-2, however, displayeda harsh, bright, but stable arc. As to theintermediate flux systems, Heat 2 gavea less stable arc whereas Heats 3 (Fig. 9)and 4 gave unstable arcs due to elec-trode eccentricity issues. When an elec-trode is eccentric, the filler metal corerod is not centered in the flux coatingcausing the thinner side of the fluxcoating to burn away faster and the arcto blow out on that side. This causes thearc to widen and wander producing anunacceptable bead shape. Voltage signalanalysis was not performed on Heat 3and 4 because of the difficulties in-volved in welding with eccentric elec-trodes. The voltage data should assist inquantitatively substantiating the de-scriptive welding lingo of a harsh andsoft arc.

Comparison of Welding Arc Signal Processing

The commercial electrodes, E-1 andE-2, exhibited lower average voltages of24 and 23 V with a standard deviationof ±1.5 and ±2 V, respectively, and ex-hibited consistent and regular short-cir-cuiting events — Figs. 14, 15. Thein-house manufactured electrodes usingHeat 2 displayed a higher average volt-age of 29 V with a standard deviation of±1.9 V. Heat 2 displayed mainly short-circuiting metal transfer mode withsome globular events occurring in-be-

WELDING RESEARCH


Table 11 — Welding Observations at Optimal Welding Parameters

Electrode E1 E2 Heat 2 Heat 3 Heat 4 CSM Final

Slag Self Good Not Good Not Good Good SelfDetachability Release to Poor to Poor Release

Slag Thinner, Thicker Thicker Thicker Thicker ThickerThickness Porous

Slag Flow Less Viscous Viscous Viscous Viscous ViscousViscous

Arc Soft, Harsh, Harsh, Unstable Unstable Soft,Stable Stable Less Arc Arc Stable

Stable

Table 12 — Quantitative Values Related to theSlag

Electrode Electrode AverageMelt Rate Thickness(mm*s1) (mm)

E1 4.9 1.52E2 4.9 1.91

CSM Final 5.1 2.03


tween. Figure 16 shows a portion of thevoltage data from Heat 2 displaying thechange from short-circuiting to globularmetal transfer mode. Changes in metaltransfer modes throughout the weldingof an electrode show problems with arcstability. Considering an electrode feedrate of 5.5 mm/s for Heat 2, thesedroplets were quite sizeable. Large glob-ular transfer is generally an indicationof transfer instability and spatter. CSMFinal average voltage was 22.5 V with astandard deviation of ±1.6 V — Fig. 17.The dominant transfer mode for E-1, E-2, and CSM Final is short-circuiting oc-curring at similar frequencies. Thefrequencies of the short-circuitingmetal transfer mode for E-1, E-2, andCSM Final are 3, 4, and 6 Hz, respec-tively. Calculating the actual amounts oftransfer per short-circuiting event, E-1on average transferred 0.19 grams (0.57g/s, Table 3, divided by three short-cir-cuiting events per second) while theCSM Final transferred 0.075 g/s (0.45g/s, Table 3, divided by six short-circuit-ing events per second). The smallermass transferred is expected to result insmoother transfer and less erratic arcbehavior. In summary, the CSM Finaldisplayed the lowest overall averagevoltage and also the highest and mostconsistent short-circuiting metal trans-fer mode throughout the entire weld.

ConclusionsThe major conclusions of this re-

search follow:• A design methodology for develop-

ing SMAW consumable electrode fluxcoatings for high-nickel alloys is pre-sented focusing on enhancing extrud-ability, weld cleanliness, slagdetachability, alloying recovery, weldingcharacteristics, and arc quality.

• A GTAW wire can be converted intoa SMAW consumable electrode, therebyproving that a unified welding wire is vi-able for both flux coated and bare wirewelding processes.

• In enhancing extrudability, an ap-propriate amount of binder with respectto relative humidity is important, ade-quate slipping agent, and a dry mix witha normal particle size distribution at200 mesh is desirable.

• Adequate amount of cryolite is nec-essary for enhanced weld cleanliness forhigh-nickel SMAW consumable elec-trodes and maintaining slag viscosity.

The amount must be optimized for weldcleanliness without affecting the viscos-ity of the slag flow since fluorides act asnetwork breakers and decrease the vis-cosity of the slag.

• The CTE values for the slag and theweld metal is important for good slagdetachability. Perovskite, with a CTEvalue of 15.8 × 10–6/°C, had a larger CTEvalue than the weld metal (CTE value of10.4 × 10–6/°C) which exhibited themost desirable slag detachability. With aCTE value of 6 × 10–6/°C, Sphene, adominant slag phase, displayed poorslag detachability.

• A series of mass balance equationsis presented to calculate accurate recov-ery rates for alloying in the flux coatingdeveloped in this research.

• Flux alloying recovery rate forchromium in this high-nickel alloy weldsystem was determined as 95% same asfor low-carbon steel SMAW consumableelectrodes.

• Manganese was calculated as 60%flux alloying recovery for this high-nickel alloy weld system which is muchlower than the 75% reported in theASM Handbook.

• Electrode formulation “CSM Final”exhibited the best arc stability with thelowest average voltage at 22.5 V and thelowest mass per transfer event at 0.075g per droplet. Flux formulation is a com-plex, interwoven, balancing process toensure good extrudability, welding char-acteristics, and weld integrity.

The authors acknowledge the helpand discussions of Joe Scott and KeithMoline of Devasco International, Inc.,and Bob Hamilton of Johns Manvilleduring the course of this research. Theauthors also acknowledge the supportof Dr. George Young for the useful discussions.

1. Jackson, C. E. 1973. Fluxes and Slags inWelding. Weld Res. Bull. 1973, No. 190,Welding Research Council, New York, N.Y.

2. Linnert, G. E. 1965. Welding Metal-lurgy. Vol. 1, Ch. 8, American Welding Soci-ety, Miami, Fla., pp. 367–396.

3. Natalie, C. A., and Olson, D. L. 1985.Physical and Chemical Behavior of WeldingFluxes. Center for Welding Research, Col-

orado School of Mines, Golden, Colo. 4. Liu, S., Frederickson, G. L., Johnson,

M. Q., and Edwards, G. R. 1994. ShieldedMetal Arc Welding Consumables for AdvancedHigh Strength Steels (HSLA-130 Steel WeldingConsumables). Center for Welding and Join-ing Research, Colorado School of Mines.

5. ASM Handbook. Welding, Brazing, andSoldering. 1993, Vol. 6, ASM International,Materials Park, Ohio.

6. Vornovitskii, I. N., Medvedev, A. Z.,and Cherkasskii, A. L. 1973. Influence of co-efficient of thermal expansion of slag on itsdetachability from weld metal. Svarochn.Proizvodstvo 3: 35–37.

7. Vornovitskii, I. N., Malashonok, V. A.,and Cherkasskii, A. I. 1975. Procedure forquantitative evaluation of slag detachability.Svarochn. Proizvodstvo 2: 47–48.

8. Olson, D. L., Liu, S., and Fleming, D. A.1993. Welding Flux: Nature and Behavior. Cen-ter for Welding and Joining Research, Col-orado School of Mines.

9. Donchenko, E. A., Panasenko, L. K.,and Larochkin, V. K. 1978. Some propertiesof welding slag. Svar. Proiz. 6: 38–40.

10. Claussen, G. E. 1949. The metallurgyof covered electrode weld metal. WeldingJournal 28: 12–24.

11. Scott, J., and Moline, K. 2008.Devasco International, Inc., private communication.

12. Liu, S., Dallam, C. B., and Olson, D. L.1982. Performance of the CaF2-CaO-SiO2system as a submerged arc welding flux for aniobium based HSLA steel. Proc. Intl. Conf.on Welding Technology for Energy Applications,Gatlinburg, Tenn., AWS-ORNL, ASM,p. 445.

13. Indacochea, J. E., and Olson, D. L.1983. Relationship of weld metal mi-crostructure and penetration to weld metaloxygen content. Journal of Materials of En-ergy System 5: 139.

14. Perez Guerrero, F. 2003. Effect ofnickel additions of rutile electrodes for un-derwater welding. MS thesis, ColoradoSchool of Mines.

15. Devries, R. C., Roy, R., and Osborn, E.F. 1955. Phase Equilibria in the System CaO-TiO2-SiO2. American Ceramic Society 38(5):161.

16. Wu, C., et al. 2009. Plasma-sprayedCaTiSiO5 ceramic coating on Ti-6Al-4V withexcellent bonding strength, stability, and cel-lular bioactivity. J. R. Soc. Interface 6(31):159–168.

17. Redfern, S. A. 1996. High-tempera-ture structural phase transitions in per-ovskite (CaTiO3). J. Phys.: Condensed Matter8: 8267–8275.

18. Special Metals. 2008. Inconel® 600Alloy: Technical Bulletin.

19. Natalie, C. A., and Olson, D. L. 1985.Physical and Chemical Behavior of WeldingFluxes. Center for Welding Research, Col-orado School of Mines, Golden, Colo.

WELDING RESEARCH


Acknowledgments

References


FluxCoating Development for SMAW Consumable …files.aws.org/wj/supplement/WJ_2014_08_s271.pdf ·...

Documents

Transcript of FluxCoating Development for SMAW Consumable …files.aws.org/wj/supplement/WJ_2014_08_s271.pdf ·...