Exothermically Assisted Shielded Arc

Adding exothermic flux to SMAW enhances heat generation and increases melting rate

BY J. W. ALLEN, D. L. OLSON AND R. H. FROST

ABSTRACT. Exothermic flux additions to such reactions in SMAW flux coatings where I is current, S is travel speed, V is SMAW electrodes can assist in the gen- could assist heat generation during arc arc voltage, AV is arc instability, q is flux eration of heat and increase the rate of welding. Theoretical calculations indi- viscosity, y is interfacial tensions and K' melting. The exothermic addi- cate that with high concentrations of and C are constants for a given flux sys- tions consisted of various combinations exothermic additions it may be possible tem (Ref. 3). and concentrations of aluminum and to formulate a welding flux that can sup- Heat input, H, is the quantity of en- powders in a mixture with ply all or most of the heat required for ergy introduced from the arc per unit hematite. Aluminum reactive additions welding. length of weld. It is expressed in J/mm were significantly more effective than An investigation of alternative heat and is calculated as the ratio of the total magnesium-rich additions at melting the sources must consider the total quantity power input (W = Vl) to the travel speed electrode, melting the weld bead and of metal, both electrode and base metal, of the welding arc in mm/s multiplied by penetrating the base plate. Magnesium that is melted to form the weld deposit. a dimensionless efficiency factor, ~. was less effective at supplying heat be- Two weld deposit features that relate di- VI cause its reaction rate was too fast with rectly to the weld parameters, and thus H = ~-- (2) respect to the electrode melting rate. This indirectly to heat input, are the weld pen- S magnesium "exothernic meltback" pre- etration depth and the weld cross-sec- The efficiency term, ~, compensates for vented effective transfer of the chemi- tional area. Jackson and Shrubsall (Ref. 1) losses during energy transfer from the cally generated heat to the base plate. have studied the effect of current, voltage power supply to the workpiece. A direct, The rate at which the exothermic reaction and travel speed on penetration for gas linear correlation between the weld bead occurs with respect to the melting rate of metal arc welds. Olson and Schwemmer cross-sectional area and the heat input the electrode determines the efficiency (Ref. 2) suggested that there are other fac- has been reported (Ref. 4). Theoretically, with which the exothermic heat is used tors that also influence penetration, such one joule of heat will melt 0.0854 mm 3 for welding. Exothermic reactions can as arc stability, viscosity and interfa- of , assuming the temperature of the produce as much as 25-30% of the total cial tension forces. A combined form of molten steel during welding is 1750°C. heat necessary to produce a shielded the Jackson-Shrubsall and Schwemmer Because the weld bead is formed from metal arc weld. equations that takes into account both both the melted base plate and the welding process parameters and slag be- melted electrode, the effects of welding Introduction havior has been suggested, and is given variables on electrode melting character- as istics have been investigated (Refs. 5-11 ). The welding arc normally supplies the It has been shown that the electrode energy required to fuse the base plate +C (11 melting rate increases with current. In material and the metal transferred from fact, a number of variables have been the consumable electrode. The resulting shown to affect electrode melting rate, heat flow determines many physical and including current, polarity and electrode chemical changes that occur during arc diameter. Hummitzsch (Ref. 12)investi- welding and is responsible for the forma- gated the effect of the ionization poten- tion of the weld deposit. This research KEY WORDS tial of the SMAW flux coating on the elec- was to determine the heating potential of trode melting rate. He related the heats of exothermic additions to the flux coating SMAW formation, vaporization and ionization of of shielded metal (SMAW) Exothermic the compounds in the coating to the elec- consumables. Oxidation reactions of el- Electrodes trode melting rate. The lowest melting ements such as magnesium, aluminum or Aluminum Powders rates were reported for cerium and bar- titanium are known to produce signifi- Magnesium Powders ium compounds, while higher melting cant amounts of heat. It is possible that Hematite rates were observed for magnesium, tita- Melting Rate nium and silicon compounds. J. W. ALLEN, D. L. OLSON and R. H. FROST Calorimeter The addition of different compounds are with the Center for Welding, Joining and to a welding arc can change both its tem- Coatings Research, Colorado School of Mines, perature and its ability to transfer heat to Golden, Col.

WELDING RESEARCH SUPPLEMENT r 277oS 1.4o ,.

A

A E E

.~ 1.2o- 1.10-

c: "11" -1- | 1.t0- '5 1.00-

• AI (excess hematite) • | • 125A,25V | I • 96A, 32V 1.00 0.g0 o.~o Io'.oo 2o1oo :,o'.oo 4o1o0 o.~o Io'.oo 2oi® 301oo 4o'.0o Weight Percent React~ Addition(%) Weight PeroentAluminum + Hematite(%)

Fig. 1 -- Measured heat input as a function of reactive addition. Boxed Fig. 2 -- The effect of welding parameters at the same power on mea- points indicate approximately equal measured heat. The effect of ex- sured heat input of the aluminum-reactive electrodes. cess hematite on measured heat input for the aluminum-reactive elec- trodes can be seen by comparing line A with line B. the weld deposit. Flux additions that are hematite and magnesium with hematite in heating due to exothermic reactions. capable of producing heat through are calculated to be 566 kJ/mol 02 and He reported increases from 3 to 26% in exothermic interactions may or may not 1060 kJ/mol 02, respectively. The most heat generation and increases of 8 to change the arc efficiency, but are poten- common reaction is 29% in melting efficiency depending tial arc heat generators. This investigation upon the amount of exothermic addi- considers the potential usefulness of such 2AI + Fe203 = tions in the flux coating. exothermic flux additions and quantifies A/203 + 2Fe + 566 kJ / Mol 02. (3) These results suggest that exothermic their heating and melting abilities additions could enable large SMAW Glushchenko (Ref. 15) has investi- through calorimetry. electrodes to be used with small, gated the effects of exothermic additions portable, low-power electrical sources. in fluxes on de- Exothermic additions may also allow Exothermically Assisted position rate and melting efficiency. He welding consumables to be used with Welding Consumables has derived equations that were reported lower welding currents, which could to predict these behaviors for a given conceivably alleviate overheating and Exothermic welding systems are not a exothermic flux mixture. Zarechenskii arc blow complications. new concept to the welding industry. (Ref. 16) has used the exothermic reac- Thermite welding was developed about a tion between a magnesium/aluminum century ago. Goldschmidt discovered alloy and hematite to increase the melt- Preliminary Investigation that if finely divided metallic oxides are ing efficiency of the flux cored arc weld- mixed with aluminum powder they will, ing process. This study indicates that the A preliminary investigation was per- if ignited, fuse and produce temperatures additional heat produced by the reactive formed that identified the effects of po- of 2500°C in less than 30 s (Ref. 13). The addition assists in melting the flux core, tential reactive additions on the welding which is thought to be the limiting factor process. The preliminary results led to the amount of heat generated during the ox- to melting efficiency for flux cored arc selection of the appropriate flux systems idation of can be calculated using welding. used in the primary investigation. tabulated thermodynamic data (Ref. 14). Karpenko (Ref. 17) investigated the ef- These calculations make it possible to fects of exothermic (aluminum + evaluate the potential productivity of Table 1 -- Beginning Flux Base oxide) additions on the melting charac- Composition Selected for Exothermic specific chemical heat sources. Metals teristics of SMAW electrodes. The results Modification such as calcium, magnesium and alu- of this study indicated that reactive addi- minum are among the most reactive and tions increase the weld deposition rate Ingredient Weight-Percent have very large negative values for their and that there exists a particular set of Alumina 3.0 free energies of formation (AGf) and en- welding parameters and oxide-to-reac- Feldspar 7.5 thalpies of formation (AHf). The more tive element ratio that optimizes produc- Calcium Carbonate 13.63 negative the enthalpy of formation, the tivity, loffe, et aL (Ref. 18), studied the ef- Iron-Manganese 2.46 more heat that is generated during the re- fects of a titanothermite mixture on Potassium Titanate 12.07 Rutile 51.96 action. The quantities of heat produced melting behavior and derived theoretical Silica 9.39 by the reactions between aluminum with equations to predict the percent increase

278-s I JULY 1998 The selected flux system to be investi- addition of calcium fluoride was added gated was initially developed by Flem- to the base flux, giving the new compo- Table 2 -- Flux Base Composition Modified with Calcium Fluoride Addition ming, et al. (Ref. 19), and has the ease of sition listed in Table 2. Calcium fluoride use that is characteristic of acidic SMAW is commonly employed to increase the Ingredient Weight-Percent fluxes, but still maintains a low weld fluidity of for steel making. It was metal oxygen content. The composition found that the flux with the additional Alumina 2.7 for this formulation is listed in Table 1. calcium fluoride produced a more uni- Feldspar 6.7 Calcium Carbonate 12.2 This flux was mixed with an aluminum + form bead with a larger cross section and Iron-Manganese 2.2 hematite (one-sixteenth of the stoichio- deeper penetration. This new base com- Potassium ~tanate 10.8 metric ratio) mixture to form O, 3 and 10 position was then tested with a 40 wt-% Rutile 37 wt-% mixtures of reactive addition in the addition of the exothermic reagent and Silica 8.4 base flux. The fluxes were then bound was found to perform acceptably. The Calcium Fluoride 20.0 with a potassium silicate binder and ex- weld bead was uniform throughout its truded onto a low- steel core wire. length and possessed consistent proper- mitted control of the acquisition rate and The electrodes were baked at 325°C for ties. The new base composition (Table 2) range and allowed real-time monitoring 2 h and then used to make bead-on-plate was selected to be a suitable flux for the of the acquisition process. welds on ASTM A36 steel. primary investigation. Initial verification and calibration of The weld deposits were cut, ground, the calorimeter was accomplished by polished, photographed and measured Primary Investigation measuring the heat content of a with the image analysis system. The de- test block. The test block was heated in a posited slags and weld beads were visu- A liquid nitrogen calorimeter was muffle furnace and its heat content was ally examined for uniformity and poros- used to measure the amount of heat that measured with the calorimeter at tem- ity. Up to the 3 wt-% level, the weld bead was transferred to the base plate when peratu res ra ngi ng from room temperatu re and slag were uniform and no porosity welding with exothermically assisted to 1173°C. Theoretical calculations of was present in the weld deposit. The consumables. This technique relates the the heat content were used to verify the weld bead area and penetration depth amount of liquid nitrogen vaporized by a measurements made with the calorime- were found to actually decrease with in- weld specimen to its heat content (Ref. ter. The calculations were based upon creasing exothermic reagent. At the 10- 20). The calorimeter consists of a Denver tabulated thermodynamic data (Ref. 14). wt-% level the weld bead and slag were Instrument DI-8K, 8000-g-capacity bal- Some manipulation of the measured heat nonuniform and there was a drastic drop ance equipped with a level-compatible data was required to obtain the actual in the nugget size and the penetration. It RS-232C interface and a 6-L stainless heat contents of the copper test blocks. is thought that increasing exothermic ad- steel dewar flask with lid. The dewar flask Consideration of the amount of liquid ni- ditions promoted the formation of alu- acts as a reservoir for liquid nitrogen and trogen that would boil off due to the am- mina through the exothermic reaction sets on the balance. The RS-232C inter- bient temperature in the room, the buoy- and raised the viscosity of the molten face allows weight information to be ant force exerted by suspending the block slag. This behavior caused the irregular transmitted from the balance to a Fortis in the liquid nitrogen and the sensible bead due to the lack of fluidity in the 386-SX personal computer. The data ac- heat of the block at room temperature molten flux. quisition software for this experiment was necessary. Based on these findings, a 20 wt-% was written in Visual Basic, which per- The ambient boil-off rate (ABR) of the

30.(]0 n. I. 1 I I

25.00- 1.30" f

• 2O.OO- E g t.20- 15.00-

| e- 1@00- 'k 1,10"

5.00- • M0 • bp~ .n uo~ (5o15o) Im ~ (5og5o) 0.00 1.00 0.00 10'.(]0 20:00 30:00 40'.00 o~ loi® 2oi® ~i® 4oi® Welght PercentRuctive ,~ldit~ (%) Weight PercentReact~e Add~on (%)

Fig. 3 -- Boxed points indicate weld bead area produced with the same Fig. 4 -- Measured heat input as a function of reactive addition. Boxed measured heat input. points indicate approximately equal measured heat.

WELDING RESEARCH SUPPLEMENT I 279-s ~o.oo

25.00 25.00 A ~ ~.00 v ~EE3°'°° t ' , , , , 15.00

15"°°1 •

O.W 10.001 0.00 I0.00 20.00 30,00 40.00 WeightPercent Re~-tive ,a, dditiofl 06)

Fig. 5 -- Boxed points indicate weld bead area produced with the same Fig. 6 -- The effect of excess hematite on weld bead area. measured heat input. liquid nitrogen was determined at two bulent boiling of the liquid nitrogen on approximately 20 min. After mixing, the different temperatures, 23 and 26°C. A the sides and lid of the dewar. ingredients were emptied into a stainless best fit regression line was used to deter- steel mixing bowl and thoroughly com- mine the amount of liquid nitrogen lost Materials and Experimental Consumables bined with a potassium silicate binder. during the time of the actual welding The amount of binder necessary to obtain tests. Fourteen electrodes were formulated the correct consistency for extrusion var- The buoyant force produced by both with three types of exothermic reagents. ied depending upon the composition of the copper calibration block and three Four fluxes were made with aluminum the mixture. This dependence was pri- different masses of steel was measured thermite, three fluxes with magnesium marily due to the difference in particle using the calorimetric apparatus and cal- thermite, four fluxes with a 50/50 wt-% sizes, the smaller particles requiring culated using available material physical aluminum/magnesium alloy thermite, more binder. In general, 1400 g of flux re- parameters. Because of the difficulty in one flux with the pure base flux compo- quired between 300 and 500 g of potas- test specimens of precisely sition and two fluxes with excess her- sium silicate binder. the same mass (and therefore the same natite, which is the primary oxygen The flux was extruded onto AISl-SAE buoyant force and sensible heat), it was source for the reaction. The electrodes 1008 steel cored wire, which had a di- necessary to construct calibration curves were formulated with increasing con- ameter of 3.175 mm. The coating was ex- that related sample mass to sensible heat centrations of exothermic reagents. The truded with a 5.26-mm-diameter die, giv- and buoyant force. After determining the exothermic formulations were made by ing a coating-to-wire ratio of 1.66, which ABR, the buoyant force and the sensible mixing a 10-kg batch of the base compo- is within the recommended ratios re- heat of the copper test block, the sition and then adding the exothermic ported in the l iteratu re (Ref. 21 ). The elec- calorimeter was calibrated to measure reagents in the various weight percent- trodes were allowed to air dry for 24 h heat contents within the range of interest ages indicated in Table 3. Initially, the and were then baked at 325°C for 2 h. (20-30 kJ/mol). ratio of hematite-to-exothermic reagent Sixty 19.1-mm-thick ASTM A36 steel A consistent overmeasurement of the was stoichiometric, but poor extrudabil- base plate specimens were cut 54 mm heat content of the test block was ob- ity later required a ratio of 16% of stoi- wide and 127 mm long. Approximately 2 served. This overmeasurement was at- chiometry. ft ( 0.6 m) of wire was GMA tributed to an additional absorption of The ingredients were dry mixed in a welded to one edge of each plate. The heat by the calorimeter caused by the tur- V-blender with ceramic grinding balls for wire would later suspend the as-welded

Table 3 -- Exothermic Additions to the Base Flux for the Final Test Matrix of Exothermically Assisted Welding Electrodes

Electrode Designation and Composition (Weight-Percent) Ingredient $1 A5 A10 A25 A40 M5 M25 M40 MA5 MA10 MA25 MA40 A10X A40X Hematite X 1.7 3.3 8.3 13.3 0.83 1.66 6.7 1.1 2.23 5.55 8.9 6.6 26.6 Aluminum X 3.3 6.7 16.7 26.7 X X X X X X X 6.7 26.6 Magnesium X X X X X 4.16 8.33 33.3 X X X X X X Mg/AI Alloy X X X X X X X X 3.4 7.77 19.45 31.1 X X (50/5O)

280-s I JULY 1998 2.50 n u u i | 7.{~, ] ~. i i i i

2.00 6.00 o.001 • E g 5.00 "5. 1.00 ~ t- | ,.®4 .o

i~ 1.00, 4.00 •

0.00. 3.00 •

• ,

0:00 Io'.® 2oi® 3o'.oo ,o100 3o0 o.~o " ,o'.00 " 20.o0 " 00:00 " ~'.oo WeightPercent Reactive Addition (%) WeightPercent Reactive Addition (%)

Fig. 7 -- Penetration depth as a function of type and amount of reac- Fig. 8 -- The effect of type and amount of reactive addition on elec- tive addition. trode melting rate.

plate in the calorimeter and prevent it alyzed and the necessary information ex- types of exothermic reagents. The largest from touching the dewar walls. The tracted. The welded plates were weighed sources of error in the data are probably plates were sandblasted, cleaned with again to determine the buoyant force and electrode inhomogeneity and variable acetone and weighed to the nearest 0.1 g sensible heat of the as-welded plate. power delivered by the welding ma- immediately prior to welding. The welded plates were sectioned in chine. With all three types of exothermic the transverse direction, ground, pol- addition (Mg, AI, Mg/AI), there is a defi- Welding Arrangementand Procedure ished and etched. The weld bead cross nite trend toward increasing measured sections were photographed and mea- heat with increasing concentration of re- A fixture was designed to thermally sured using the image analyzer. Six 1-g active addition. insulate the base plate to be welded from samples were sectioned from the weld It is apparent from Fig. 1 that, in all the work bench. A base plate of prede- metal of each plate. These samples were three cases, a maximum measured heat termined size could be held firmly in quantitatively analyzed for oxygen, sul- exists in the range of 25-30 wt-% reac- place on pins in the fixture by a quick re- fur, nitrogen and carbon content with a tive addition. Beyond this maximum, fur- lease latch. The latch enabled the speci- LECO interstitial analyzer. ther addition of reagent does not increase men to be removed easily and quickly the measured heat. This plateau in mea- from the fixture and placed into the liq- Results and Discussion sured heat indicates that there may be a uid nitrogen calorimeter. limit to the amount of reagents that can A Hobart 300-series constant-current Calorimetric measurement of the as- participate in the exothermic reaction. transformer-rectifier machine was used welded plates indicated that exothermic One limiting factor could be an exhaus- to make the welds. The head assembly heating does occur when welding with tion of available oxygen in the region of consisted of an ARCON voltage con- these electrodes. Visual examination of the reaction. The conditions of welding troller connected to a semiautomatic the arc length indicated that, in all cases, indicate that the exothermic reaction oc- travel carriage. The voltage controller the arc length at the same potential in- curs in a molten metal environment; maintained a set arc length, which creased as the concentration of exother- thus, the solubility limit of iron oxide in helped maintain a constant electrical mic reagent increased. Increasing the arc iron can limit the amount of oxide avail- heat input during welding. Three bead- length tends to broaden the distribution able to react. on-plate welds were made with each of arc heating and decrease the testing in- It is expected that the exothermic re- electrode formulation. The welding para- tensity of the arc. Longer arcs also radiate actions occur primarily at the tip of the meters were direct current, reverse po- more energy and convect more heat to electrode, significantly affecting the elec- larity, 25 V, 125 A and a travel speed of the atmosphere. Therefore, increases in trode melting rate. Given this assump- 2.54 mm/s. One 90-mm weld was de- the arc length with increasing exother- tion, the peak in the measured heat input posited on each of the prepared plates mic additions will tend to mask or de- curve can also potentially be explained and immediately submerged in the liquid crease the actual heating effect. The true by any physical limitations of heat trans- nitrogen calorimeter after the arc was ex- exothermic heating potential must be fer to the molten electrode tip. Limita- tinguished. The amount of time that greater than that which is measurable tions can result from chemical heating elapsed between extinguishing the arc with the equipment used in this research. occurring at locations above the molten and submerging the welded specimen Figure 1 shows how calorimetric mea- tip and by convection losses to the at- was always less than 5 s. The computer surements of heat input change with in- mosphere. For exothermic additions in recorded the cooling data, which was an- creasing concentrations of the three excess of 25-30 wt-%, these modes of

WELDING RESEARCH SUPPLEMENT [ 281-s 0.00 | I I AI 28.00 • Mg ~ 5.~. • Mg/AI(50/50) .E 22.00 .,.,.•5.00. 20.00 c "1 /. I 4.50- W ::I / -- ~ 4.00- <°° I / . ~., 3.50- 'o-t_/• =Ew 8.~ "• ~[.,~_.~._~.i 3.00 o.'oo 10100 2o:® 3o1oo 4o:oo Electrode Melting Rate (g/mini Weight Percent Reaclive Addition (%)

Fig. 9 -- Melting rate as a function of weld bead area for the three types Fig. 10 -- Melting rate normalized with respect to measured heat input of exothermic additions. for the three types of exothermic additions. heat loss may occur at rates exceeding than lower currents (Ref. 1). How weld- the detachability of these slags deterio- the rate of heat transfer into the molten ing parameters affect exothermic heating rated with increasing concentration of re- tip, resulting in the optimum in the mea- cannot be determined from this data. active addition, no porosity was trapped sured heat input curves. The heat that is However, Glushchenko (Ref. 15) has in the weld metal and the beads were rel- not lost from the molten electrode tip is shown that the melting efficiency of his atively uniform and symmetric along transferred to the weldment by the exothermic electrodes is optimized at their lengths. molten droplets. currents in the range of 300-325 A. The A series of electrodes (electrode- Figure 1 also illustrates how the mea- covering compositions with twice the sured heat changes with increases in alu- Characteristics of the Slags ratio of hematite to aluminum) produced minum-reactive addition with excess slags that trapped excessive porosity in hematite in the mixture. The curve la- Slag viscosity and interfacial tensions the weld metal. Recalling the measured beled A was produced from consumables have been shown to affect weld bead heat data for these welds, which indi- with twice the amount of hematite as the morphology and penetration (Ref. 2). For cated a plateau in the chemical heat pro- rods that produced curve B. These curves this reason, it is necessary to review the duction, it seems reasonable that indicate that additional hematite, which performance of the slags in at least a hematite in excess of its solubility in iron is thought to be the primary source of qualitative sense before discussing the re- would release oxygen in the solidifying oxygen for the exothermic reaction, does sults of the bead area and penetration weld pool, producing a porous weld increase the measured heat input. This study. bead. result confirms that the availability of In general, the detachability and pore- oxygen influences the heat generated in removing ability of the slags deteriorated Exothermic Meltback the exothermic reaction. as the concentration of exothermic addi- To determine the effect of current and tion increased. It was found that magne- Once ignited, a purely exothermic re- voltage on measured heat input, welds sium/aluminum alloy reactive additions action can propagate without further ad- were made with the aluminum-reactive up to 10 wt-% produce slags that readily dition of energy. Some of the electrodes series of electrodes at 33 V and 96 A, detach during and after cooling and form with 25 and 40 wt-% reactive addition which gives approximately the same a relatively symmetric and uniform layer would spontaneously ignite ahead of the electrical heat input as the previous of slag protection. At 25 and 40 wt-% re- arc during welding. With the exothermic welds received (1.2 kJ/mm). The mea- active addition, the slag deposit adheres reaction outrunning the process of melt- sured heat inputs at the two welding pa- tenaciously to the weld, traps pores in the ing the rod, a separation of the energy rameters for the aluminum-reactive elec- weld metal and is nonuniform along its source from the melting rod tip devel- trodes as a function of increasing reactive length. The shape of the slag deposits oped. After the arc was extinguished, the addition are shown in Fig. 2. At a current were observed to become narrower and reaction would propagate up the length of 96 A and a voltage of 33 V, the mea- taller with increasing amounts of of the electrode without melting the core sured heat contents are significantly less exothermic reagents. Narrow, tall slag wire, very much like a sparkler. These re- than those welds produced at 125 A and deposits tend to be associated with actions usually were not visible until after 25 V. This behavior is seen because higher viscosity slags. 3.5 in. (90 ram) of weld bead was de- higher currents are more efficient at The slags formed from the aluminum posited during the calorimetric tests. transferring the arc heat and produce series of reactive consumables demon- These premature reactions can seriously larger weld bead and greater penetration strated the best performance. Although limit the applicability of exothermically

282-s I JULY 1998 assisted welding electrodes. addition. Thus, the type of reactive addi- One solution to the premature reac- tion influences how the heat input melts tion problem could be to fill a hollow and produces the weld bead. tube with the reactive flux, as in flux Figure 6 shows how the weld bead cored arc welding. A flux cored rod area changes with increasing weight per- would better contain the heat generated cent of aluminum reactant with and with- by the exothermic reaction and may have out excess hematite in the mixture. Re- better electrode melting characteristics. call that Fig. 1 indicated that excess The extra heat imparted to the electrode hematite, and thus excess oxygen, signif- would then be transferred to the base icantly increased the heat inputs mea- plate. sured with the calorimeter. However, Fig. 6 shows that the weld beads formed with Penetration and Weld Nugget Size the excess hematite are considerably smaller than those weld beads formed The effect of the type and amount of without excess hematite. Once again, it reactive addition on weld bead area is il- is apparent that heat input is not the pri- lustrated in Fig. 3. For the aluminum and mary variable determining weld bead magnesium additions, the weld bead area. This indicates that oxygen may play area increases with increasing concen- a role in determining how the weld bead tration of reactive component. For the is formed. magnesium/aluminum alloy reactive ad- Figure 7 illustrates weld penetration dition, the weld bead area actually de- depth as a function of the type and creases and then increases slightly. amount of reactive addition. The general The general trend is toward increasing trend is toward increasing penetration Fig. 11 -- Mi(rt~/4mphs near the tusi

WELDING RESEARCH SUPPLEMENT I 283-S sion temperature pool of sulfur more effectively than the coefficient is nega- aluminum-reactive additions. This clean- 1~.® ' ' ' " ' ' ' " ' tive, producing a ing would explain the differences in size l~ I • AI wide pool with shal- and shape of the weld beads formed • Mg low penetration. under the action of the three types of 1~.®- A l • Mg/AI(50/50) The dominant fluxes. ~~ I • Al (excesshematite) pool stirring mode At currents less than approximately has been shown to 150 A, weld pool stirring is controlled by

v eoo.oo- be related to the the surface tension gradient, which is in ®c magnitude of the turn influenced by weld metal sulfur and O~ welding current. oxygen (Ref. 22). Chemical heat genera- ,~ ~.®. ~ Below approxi- tion during exothermic welding is con- mately 150 A, pool trolled by the reaction of a metal with ~_ stirring is dominated oxygen. Thus, a measure of the weld 4~.®~ by the surface ten- metal oxygen should, in some fashion, sion gradient. Above reflect the measured heat input, weld 150 A, pool stirring bead area and penetration depth. Figure 2~.®- is dominated by the 12 shows weld metal oxygen content as . Lorentz force. a function of the type and amount of re- o.~ Io'.~ 2oi® 3o1o0 40100 Metallographic active addition. For each type of reactive observation of the addition, weld metal oxygen content de- Weight PercentReactive Addition (%) weld beads revealed creases with increasing weight-percent evidence of stirring reactive component. Recalling the mea- Fig. 12 -- The effect of type and amount of reactive addition on weld patterns. Figure 1 1 sured heat input curves in Fig. 1, it is ap- metal oxygen, shows micrographs parent that they are reciprocal of the data taken near the weld shown in Fig. 12. This correspondence interface of the indicates that the plateau in heat input Weld Pool Stirring welds from two different types of may be due to the exhaustion of available exothermic-assisted electrodes. Two dis- oxygen to react. Additionally, excess It has been shown that minor element tinct patterns are evident in the welds. hematite (the reaction oxygen source) additions can influence the stirring pat- The weld made with the A40 electrode produced the greatest measured heat terns in the weld pool and thus affect the (aluminum addition) shows patterns con- input and a stable weld pool oxygen con- weld bead morphology (Ref. 22). Al- sistent with negative convective flow, i.e., tent. This result suggests a saturation though these studies used the GTAW metal moving downward from the top point at which the size of the molten process, the results presented here sug- centerline toward the root of the weld. droplet controls the amount of reactive gest that weld pool stirring may play a The welds made with the MA40 (magne- addition (metal plus oxide)that can react. part in the SMAW process as well. sium-aluminum combination) electrode The level of weld metal oxygen can be- Two primary factors contribute to show evidence of positive convective come very low and may approach the weld pool stirring. Current flowing into flow, i.e., metal moving radially outward transition value for negative to positive the weld pool exerts a force (Lorentz from the centerline. stirring. force) on the liquid metal, which will cir- Two features present in the weld culate the molten metal downward at the beads indicate the type of stirring: the Summary center and radially outward at the bottom overall shape of the weld bead and the of the pool. This pattern of metal flow en- shape and orientation of the undercut Calorimetric measurements of heat hances the flow of heat and tends to pro- base plate near the weld interface. Neg- input indicate that exothermic mixtures duce a relatively deep weld pool (Ref. ative string tends to form deep, narrow of magnesium, aluminum and magne- 22). The second force originates from sur- weld beads, like that of the A40 weld sium/aluminum alloy plus hematite can face tension gradients (Marangoni effect) bead. Positive stirring forms shallow, assist in heat generation during shielded produced by temperature gradients at the wide weld pools, like the M40 and MA40 metal arc welding. Increases in heat weld pool surface. Near the center of the weld beads. Negative stirring, directed input up to approximately 25 to 30% of weld pool, the arc heating is greater than downward toward the root of the weld the heat required to weld were attributed the weld pool edge, which is solidifying. bead and then up along the edges of the to chemical heat generation. There ap- Depending upon the weld pool compo- pool, tends to undercut the base plate pears to be slight differences in the quan- sition, this temperature gradient can pro- near the weld interface (mushy zone), tity of heat generated by different reactive duce a positive or negative surface ten- the wave-like structures present components. The supply of oxygen to the sion gradient, which influences surface in the A40 weld bead. Positive flow tends reaction limits heat generation. metal flow (Ref. 20). Low concentrations to undercut in the opposite fashion, pro- The melting rate data clearly indicate (less than 150 ppm) of surface-active el- ducing an uplifted base plate such as is that the exothermic reactions occur at the ements (S, O and Se) have been found to present in the M40 weld bead. molten electrode tip. The heat from these substantially alter the shape of GTA weld Magnesium additions of 3 and 6 wt- reactions increases the melting rate of the deposits. In the presence of surface-ac- % to the flux coating of SMAW elec- electrode and is transferred to the plate in tive elements, the surface tension tem- trodes has been shown to produce very the molten droplets. The aluminum perature coefficient is positive, produc- low weld metal oxygen and sulfur con- exothermic reactions occur at a rate that ing a relatively deep and narrow bead tents. Magnesium and magnesium/alu- allows the resulting heat to be incorpo- geometry. With low concentrations of minum alloy additions clean the weld rated into the electrode droplet, giving surface-active elements, the surface ten-

284-s I JULY 1998 higher electrode melting rates and more occur predominantly at the tip of the ing rate and metal transfer in gas-shielded base-plate heating than the magnesium electrode, with the chemically generated metal arc welding. Welding Journal 37(9): exothermic reactions. The magnesium heat being transferred to the weld pool by 343-s to 353-s. 11. Wilson, J. L., Claussen, G. E., and Jack- exothermic reactions occur before the the transfer of molten metal droplets. son, C. E. 1956. The effect of 12R heating on arc melts the droplet, and so the excess electrode melting rate. Welding Journal, 35(1 ): heat is not incorporated into the droplet, Acknowledgment 1-s to 8-s. but is instead lost to the atmosphere or 12. Hummitzch, W. 1948. New facts on electric-arc reactions of coated welding elec- used to accelerate the reaction of the The authors acknowledge and appre- exothermic components higher up in the trodes. Schweisstechnik, 2, 50-54, May; ciate the support of the U.S. Army Re- 67-70, June; 84-89, July. electrode coating. search Office. 13. Underwater cutting process surfaces Increasing concentrations of exother- for new applications. 1989. Welding Journal, mic mixtures in the coating of SMAW References 68(7): 59-60. consumables increase measured heat 14. Chase, M. W., Davies, C. A., Downey, input, weld nugget area, weld penetra- 1. Jackson, C. E., and Shrubsall, A. E. 1953. J. R., Frurip, D. J., McDonald, R. A., and Control of penetration and melting ratio with Syverud, A. N. 1985. JANAF thermodynamic tion depth and electrode melting rate and tables. Journal of Physical and Chemical Ref- decrease weld metal oxygen content. welding technique. Welding Journal, 32(4): 172-s to 178-s. erence Data, 3rd Ed., Vol. 14. Weld bead area and penetration mea- 2. Schwemmer, D. D., OIson, D. L., and 15. Glushchenko, A. S. 1980. Determina- surements correlate very well with the Williamson, D. L. 1979. The relationship of tion of the productivity of the thermit arc weld- melting rate data. The higher melting rate weld penetration to the welding flux. Welding ing process under an exothermic flux. Svar. obtained with the aluminum additions Journal, 58(5): 153-s to 160-s. Proiz., 37-39, September. 3. Natalie, C. A., Olson, D. L., and Blander, 16. Zarechenskii, A. V. 1982. Increasing produced larger weld bead areas and the efficiency of flux cored strip melting. Svar. greater penetration depth. Weld metal M. 1986. Physical and chemical behavior of welding fluxes. Material Science Annual Re- Proiz, 38-39, July. sulfur and oxygen content, through their view, Vol. 16, pp. 389-413. 17. Karpenko, V. M. 1980. The melting effect on weld pool stirring, may play a 4. Welding Handbook, 7th Ed., Vol. 1. parameters of welding electrodes with an role in determining the melting charac- 1981. American Welding Society, Miami, Fla. exothermic coating mixture. Svar. Proiz, teristics of the weld bead. p. 37. 33-37, September. 5. Jackson, C. E. 1959. The science of arc 18. Ioffe, O. M. 1980. The effect of the ti- welding. Welding Journal, 38(5): 177-s to tanothermite mixture in the electrode coating Conclusions 119-s. on the increase in the productivity of welding. 6. Chandel, R. S. 1990. Electrode and plate Svar. Proiz, 29-32, March. 1) Exothermic reactions can assist in melting efficiencies of submerged arc welding 19. Flemming, D. A., Bracarense, A., Lui, the production of heat during shielded and . Materials Science S., and Olson, D. L. 1996. Toward developing and Technology, 6(9): 772-777. a SMA welding electrode for HSLA-100 grade metal arc welding. steel. Welding Journal, 75(6): 171-s to 183-s. 2) It is essential to control the oxygen 7. Chandel, R. S. 1987. Mathematical modeling of melting rates for submerged arc 20. Watkins, A. 1989. Heat Transfer Effi- content to achieve both optimized chem- welding. Welding Journal, 66(5): 135-s to 140- ciency in Gas Metal Arc Welding. Master's ical and physical responses when devel- s. thesis, University of Idaho, Moscow, Idaho. oping exothermic welding consumables. 8. Robinson, M. 1961. Observations on 21. Hoffmann, R. L. 1979. Mild steel elec- 3) Weld deposition is enhanced more electrode melting rates during submerged arc trode coverings -- How they work. Welding Journal, 57(5): 408-410. with aluminum-reactive additions, pro- welding. Welding Journal, 40(10): 503-s to 515-s. 22. Burghardt, P., and Campbell, R. 1992. ducing a greater electrode melting rate 9. Niles, R. W., and Jackson, C. E. 1975. Chemistry effects on stainless steel weld pen- and a larger weld bead, than with mag- Weld thermal efficiency of the GTAW process. etration. Key Engineering Materials, 69 and 70, nesium-rich additions. Welding Journal 54(1 ): 25-s to 32-s. p. 379, Transactions of Technical Publications, 4) The exothermic reactions need to 10. Lesnewich, A. 1958. Control of melt- Switzerland.

WELDING RESEARCH SUPPLEMENT I 285-s