RESEARCH

SUPPLEMENT TO THE WELDING JOURNAL, MARCH 1978 Sponsored by the American Welding Society and the Welding Research Council li II j

Slag/Metal Interaction, Oxygen and Toughness in Submerged

Fused fluxes containing 45% CaF2/ 35% Al203 and 20% CaO yield less oxygen in the weld deposit than do agglomerated fluxes with the same constituents,

and Al additions to a 45% CaF2, 35% Al203/ 20% CaO lower the oxygen contents of weld deposits to below that of the while the addition of up to 1.2% Al to the flux improves weld notch toughness

BY T. H. NORTH, H. B. BELL, A. NOWICKI AND I. CRAIC

Part I—Flux Formulation and Oxygen Content During Moreover, the final weld deposit composition is the end-point of a complex interplay of physical and chemical factors. Submerged arc fluxes ABSTRACT. The influence of flux was not feasible. for welding steels are composed of a formulation on the oxygen content of Aluminum additions to (45 wt-% complex mixture of oxides, halides, submerged arc deposits has been CaF,, 35 wt-% ALO,, 20 wt-% CaO) flux carbonates, silicates and ferro-alloys. investigated. When using low oxygen lowered the oxygen content of depos­ Consequently, the formulation of potential flux formulations composed its to below that of the electrode. Such fluxes has a major influence on the of mixtures of CaF.,, ALO., and CaO, low oxygen levels were obtained at the final deposit oxygen content, and on deposit oxygen contents were marked­ expense of increased contents of its notch toughness. ly lower using fused fluxes than in aluminum in solution. This paper examines the factors using agglomerated fluxes. The main determining the transfer of oxygen to source of oxygen when welding with Introduction the weld metal during submerged arc fluxes composed of CaF,,, ALO:, and welding. Considerable accent has The submerged arc welding process CaO was that due to decomposition of been placed on the concept of basicity is superficially simple in operation. flux constituents. of submerged arc fluxes as a means of Oxygen pick-up during welding producing low oxygen content weld with CaO-ALOrCaF,, fluxes was main­ deposits. In terms of our knowledge of ly determined by gas/metal reactions Paper to be presented at the AWS 59th slags, such use of basicity ratios is occurring at the electrode tip region. Annual Meeting in New Orleans, Louisiana, incorrect, and it is important to differ­ There was no correlation between the during April 3-7, 1978. entiate between the basicity of the FeO content of slags and the oxygen Dr. T. H. NORTH is a Lecturer in , flux employed and the oxygen poten­ content of deposits. Consequently H. B. BELL is Professor of Metallurgy, and A. tial during welding. slag/metal equilibrium was not at­ NOWICKI and I. CRAIG are Graduate Basic oxides such as FeO and MnO tained, and derivation of an effective Students, University of Strathclyde, Glas­ have relatively high oxygen potentials, reaction temperature during welding gow, Scotland. e.g.,

WELDING RESEARCH SUPPLEMENT I 63-s 2Fe + O, ^2FeO • log P„„ (at 1600 C, basicity considerations. The oxygen was decided that a flux formulation 2912 F) = -8.2 potential of flux constituents can be composed of highly-stable oxides assessed from the Ellingham diagram, and/or halides should be employed. while an acidic oxide such as Al203 has e.g., oxides such as Al203 are very As a result, laboratory-prepared fluxes a very low oxygen potential, i.e., stable while oxides such as FeO and based on the CaO-CaF,-AI203 system CuzO are not so stable. Since calcium are investigated. A number of lines of 4AI + 30, ss 2ALO;1 • log P„, (at 1600 C, metal and magnesium metal boil at investigation were followed: 2912 F) = -19.9 1500 and 1100 C (2732 and 2012 F) 1. Evaluation of flux formulations in The oxygen potential represents the respectively, CaO and MgO decom­ the system CaO-CaF2-AI203 depositing driving force for oxygen transfer from pose to oxygen and calcium and the lowest oxygen contents when a the oxide to the metal and depends on magnesium vapors during the high single electrode composition was em­ the reaction: temperature cycle of welding Simi­ ployed; factors such as loss of alloy larly, Na,0 and K,0 decompose, elements to the slag during welding Oj3S2 [0],;AC° = -57,000 -1.38 T producing sodium and potassium va­ and the form of flux manufacture cal pors and, at temperatures above 1100 (agglomerated and fused) on weld When basicity is considered, the C (2012 F) they have higher oxygen metal composition were also investi­ important factor is the oxygen ion potentials than FeO. Oxides such as gated. Under ideal conditions, welding activity in the slag. While single ion Si02 and TiO, have oxygen potentials tests would have been carried out activities cannot be determined, the intermediate between FeO and using pure iron wire, but this was not following charge reactions may be Al203. available and the research program considered: The quantity of oxygen transferred employed commercial con­ to the metal is markedly temperature taining various manganese contents. 0*-(. ) + '/2S,U^S'-'( , ) + Vi lag s a8 dependent, since the solubility of 2. Examination of the means of o8y n) oxygen in iron increases with tempera­ oxidation and deoxidation during sub­ ture increase, i.e., from 0.19% at 1550 C merged arc welding with particular and, (2822 F) to 0.79% at 2000 C (3632 F). As emphasis on the role of deoxidant 5 Vi p.,(B) + /4 02y + vi a result, for a given oxygen potential additions in the flux in determining O'-(„„) ss POJ-U.) (2) the higher the temperature the more the electrode tip and final deposit oxygen will be transferred. In the case of reaction (1), a high oxygen contents. oxygen ion concentration favors reac­ The foregoing discussion has as­ 3. Estimation of the attainment of tion to the right, while a high oxygen sumed the attainment of thermody- equilibrium during welding with labo­ potential favors reaction to the left, namic equilibrium during submerged ratory-prepared fluxes. i.e., basicity and oxygen potential arc welding. Elucidation of whether 4. Examination of the effect of counteract each other. In the case of thermodynamic equilibrium actually oxygen and aluminum on the notch the reaction involving phosphorous occurs is an important facet of this toughness of submerged arc deposits; (i.e., reaction (2) ), oxygen potential investigation. Certainly the time of this work is presented in Part II of this and oxygen ion activity act together, contact of metal and slag is limited; paper. and both should be high for dephos- temperatures range from 2400 C (4532 phorization to occur. F) (the droplet formation stage) to 1520 C (2768 F) (the fusion line region), Experimental Procedure The oxygen potential during steel and it is readily apparent that a simple making is generally considered as the Materials slag-metal interaction at a constant partition of oxygen between metal and temperature is not being considered. The chemical composition of the slag, and is related to the ferrous oxide mild steel plate used in welding tests is content of the slag, i.e., given in Table 2. Proprietary electrode Aims of Research Program Fe+ [0]%ss FeO(slag) wires were employed throughout and Since submerged arc welding is a encompass a range of manganese and the oxygen partition (L) is given complex process, simply-formulated, levels—Table 2. as: laboratory-prepared fluxes provide an The experimental fluxes employed L = activity of oxygen in the metal essential tool for examination of such in this program were based on the (a0)/activity of ferrous oxide in the slag a system. The advantage of employing CaO-AI,03-CaF2 ternary system. Meth­ simple fluxes of known formulation is (aPe0) ods of flux preparation were as that the thermophysical properties and follows: During welding the activity of oxy­ solution thermodynamics are avail­ Agglomerated Fluxes (Table 3). gen in the metal (a„) is given by the able. Consequently, meaningful de­ 1. Dry mixing of 4 kg of pure wt-% oxygen content. If equilibrium is ductions concerning slag/metal inter­ components to give a desired flux attained during the welding operation, action, oxidation and reduction, and formulation for 30 min. the oxygen content would be propor­ alloy transfer during welding can be 2. Wet mixing after addition of tional to the ferrous oxide activity in made. alkali silicate binders (corresponding the slag. The activity coefficient of Preliminary tests were carried out to 2.5 wt-% Na,0, Vi wt-% K..O and 1 ferrous oxide is higher in basic slags using fluxes based on the systems wt-% SiO, in the flux). than in acid slags; this means that for a CaO-MgO-SiO, and CaO-MgO-TiO,. 3. After agglomeration the flux was given ferrous oxide content in the slag Table 1 shows the oxygen content of baked on trays at 450 C (842 F) for 3 the oxygen content of the metal would bead-on-plate deposits when using h. be higher with basic slags than with different formulations. The high oxy­ Fused Fluxes (Table 3). acid slags (if equilibrium conditions gen contents found were due to the 1. Dry mixing of 10 kg of flux in a are attained during welding). presence of Si02 in the CaO-MgO- mixer; the CaO employed was ob­ It is readily apparent that flux formu­ Si02 formulations and to Ti02 in the tained from calcination of CaCO;, at lations yielding the lowest weld metal CaO-MgO-Ti02 formulations. These 1000 C (1832 F) for 2 h. results confirm data of other investiga­ 2. Melting of flux in a graphite oxygen contents must be derived from 1 4 a consideration of the oxygen poten­ tors. - crucible using a high-frequency melt­ tial of flux constituents and not from On the basis of preliminary tests, it ing furnace.

64-s I MARCH 1978 Table 1—Flux Formulations and Deposit Oxygen Contents During Bead-on-Plate Table 3-Chemical Compositions of Welding-S3 Electrode Used Throughout, % Laboratory-Prepared Fluxes ("A"—Agglomerated Fluxes; "B"—Fused Oxygen Fluxes), Wt-% deposits

(1650 C, Designation"' CaF2 Al203 CaO Designation CaO MgO TiO SiO, lagO KsO 3002 F) 1B 100 — _ CMS 1 20 14 - 63 2.5 0.5 0.085 2B 100 - - CMS 2 30 14 - 53 2.5 0.5 0.086 4A 5 55.0 40.0 CMS 3 40 14 - 43 2.5 0.5 0.054 5A 5 58.0 37.0 CMS 4 50 14 - 32 2.5 0.5 0.042 6A 10 34.0 56.0 CMT 1 20 7.5 70 2.5 0.5 0.063 7A and 7B 10 45.0 45.0 CMT 2 30 7.5 60 2.5 0.5 0.068 8A 10 52.5 37.5 CMT 3 40 7.5 50 2.5 0.5 0.056 9A 10 57.5 32.5 10A 15 50.0 35.0 11A 15 47.0 38.0 12A 20 35.0 45.0 Table 2—Chemical Compositions of Electrodes and Plate Materials, % 13A 20 45.0 35.0 14A 20 50.0 30.0 Al Al 15A 20 56.0 24.0 (O) (O) acid acid 16A 25 45.0 30.0 (1650 C, (1800 C, Al sol­ insol­ 17A and 17B 25 50.0 25.0 Designation C Mn Si S P 3002 F) 3272 F) (total) uble uble 18A 25 55.0 20.0 19A 30 40.0 30.0 0.065 0.84 0.15 - - — — Base metal B1 0.01 0.016 0.0053 20A 30 50.0 20.0 Base metal B2 0.15 0.45 0.23 0.009 0.011 0.006 - - - - 21A 30 53.0 17.0 0.012 - • Electrode S1 0.08 0.47 0.18 0.02 0.031 - - - 22A 35 40.0 25.0 Electrode S2 0.10 0.98 0.17 0.008 0.02 0.013 - - - • - 23A 35 44.0 21.0 Electrode S3 0.12 1.59 0.29 0.012 0.012 0.008 - - - - 24A 35 52.0 13.0 Electrode S4 0.11 2.10 0.23 0.03 0.03 0.017 0.016 0.0045 0.0025 0.002 25A 40 35.0 25.0 26A 40 45.0 15.0 27A 40 50.0 10.0 3. Crushing of flux to the desired initiation and to keep flowing during 28A and 28B 45 35.0 20.0 particle size. the welding operation. 29A 45 45.0 10.0 Electrode tips were collected by 30A and 30B 45 48.0 7.0 In order to maintain consistency retracting the electrode into the argon with agglomerated fluxes the initial stream when stable welding condi­ "'All fluxes except 1B contain 2.5 wt-% Na-O, 0.5 wt- K O and 1 wt-% SiO, respectively. fused fluxes contained 2.5 wt-% Na,0, tions had been achieved during bead- z 0.5 wt-% K,0 and 1 wt-% SiO,. In tests on-plate welding. This means of elec­ where deoxidant additions were made trode tip collection providing material mean oxygen content of the electrode during welding, it was convenient to from a range of possible situations, i.e., tip region at any welding condition. add aluminum via the flux since main­ droplet transfer may have just oc­ Since it can be difficult to assess where tenance of constant current and curred or droplets growing at the elec­ the collected droplet begins and the voltage conditions throughout pro­ trode tip having various dimensions unfused electrode ends, only the last vided a reproducible means of achiev­ may be collected. ing any chosen aluminum level.5 millimeter length of collected elec­ In order to overcome this problem When aluminum additions were re­ trode tip samples were analyzed for 10 electrode tips were collected and quired, these were made by mixing 200 oxygen. analyzed for oxygen content at any mesh aluminum powder with crushed Vacuum fusion analysis was em­ particular welding condition. In e'ffect, fused flux for 8 h in a roller drum. In ployed throughout. Two test tempera- the oxygen content found was the control tests where aluminum was tures-1650 C (3002 F) and 1850 C (3272 added to the 45 wt-% CaF,, 35 wt-% F)—were examined since previous in- ? Al,03, 20 wt-% CaO formulation, no vestigators" noted differences in the Na,0, K,0 or SiO, additions were determined oxygen contents of alumi­ present. num-killed samples when tested at different temperatures. At any temper­ WATER COOLING ature the vacuum fusion analysis gave Welding Conditions a reproducibility of ±0.002%. Reverse polarity was employed throughout. Ten-high pads were used Chemical Analysis to overcome problems caused by dilu­ Aluminum analysis was carried out tion during welding operations. Ten- using the British Standard (B.S.19) high pads were deposited as follows: photometric method of analysis." The four weld beads, followed by three, eriochrome-cyanine method gave val­ then by two and finally by the test ues for the acid-soluble and acid- weld. The welding conditions insoluble aluminum contents when throughout were 600 A, 32 V, 0.25 m/ COPPER CONTACT the residues of the initial acid-solution min (9.8 ipm). TUBE treatments were separately analyzed. Slag samples were analyzed by Electrode Tip Collection during Welding means of absorption spectrophotome- Electrode tip collection was carried try. Prior to slag analysis, samples were out using a modified torch design—see crushed and iron particles were Fig. 1. This allowed argon gas to pass Fig. 7.—Modified torch for electrode tip removed by means of two procedures, through the welding head prior to arc collection tests i.e., using a strong magnet, and by

WELDING RESEARCH SUPPLEMENT I 65-s means of a liquid separation technique of iron particle removal from slag by previous investigators.5-910 using methylene iodide. Both methods samples had been found satisfactory Nitrogen analysis of weld samples was carried out using the Kjeldahl method and had an accuracy of CaO ±0.002%.

Results and Discussion

The initial tests were carried out with agglomerated fluxes made from CaC03, and CaF2. CaC03 was used to avoid problems of moisture absorp­ tion. Figure 2 shows the range of flux formulations examined—from 5 to 45 wt-% CaF2, 34 to 58 wt-% Al,03 and 7 to 56 wt-% CaO. All flux formulations were chosen to have melting points below 1500 C (2732 F), and Table 4 gives the final deposit analyses (of the tenth run of ten-high pads) when employing an electrode having 0.12%C, 0.29%Si, 0.012%S, 0.012%P, 1.59%Mn, 0.008% [O]. The welds deposited using agglom­ erated fluxes had relatively high oxy­ gen contents, and there was a correla­ tion between the CaC03 content of fluxes and the oxygen content of weld deposits. Increasing calcium carbonate content in fluxes increased the oxygen content of weld deposits (Fig. 3) and Fig. 2—Agglomerated flux formulations and deposit oxygen contents when contradicted the results of a previous welding with S3 electrode containing 0.008% oxygen study." The relationship between the oxygen content of weld deposits and the calcium carbonate content in the flux formulations (given as the equiva­ Table 4—Chemical Analyses of Weld Deposits Made Using Agglomerated Fluxes ("A" lent CaO content) was as follows: Series) and Fused Fluxes ("B" Series)—S3 Electrode Employed Throughout, wt-% % [OJaeposit = 0.001 (%CaO) + 0.042

Oxygen The loss of manganese from the (1650 C electrode during welding also in­ Designation C s, S P Mn Mo Cu 3002 F) creased as the calcium carbonate content of the flux increased—Fig. 4. 0.77 0.02 0.0725 4A 0.090 0.050 0.016 0.013 0.11 The relationship between the loss of 5A 0.087 0.060 0.015 0.014 0.63 0.02 0.12 0.1270 manganese during welding and the 6A 0.090 0.020 0.016 0.013 0.68 0.02 0.11 0.0750 7A 0.075 0.050 0.016 0.014 0.77 0.03 0.12 0.0941 calcium carbonate content of flux 8A 0.085 0.050 0.017 0.015 0.81 0.02 0.12 0.0640 formulations (given as the equivalent 9A 0.104 0.076 0.016 0.015 0.71 0.02 0.11 0.0930 CaO content) was: 10A 0.088 0.064 0.015 0.015 0.77 0.02 0.12 0.0761 11A 0.089 0.074 0.015 0.014 0.73 0.02 0.11 0.0860 (% Mn loss) = 0.362 (% CaO) + 39.81 12A 0.073 0.098 0.015 0.013 0.68 0.02 0.11 0.0852 MnHpotrode 13A 0.086 0.062 0.015 0.015 0.73 0.02 0.11 0.0962 where % / KA \ .—„, 14A 0.080 0.050 0.017 0.015 0.81 0.03 0.12 0.0671 Mn loss = ( ~Mn-'" ) 100% 15A 0.100 0.060 0.020 0.013 0.81 0.03 0.13 0.0780 Mlpliar* 16A 0.064 0.076 0.018 0.014 0.79 0.03 0.11 0.0666 While deposits made using agglom­ 17A 0.075 0.060 0.017 0.014 0.82 0.03 0.12 0.0640 erated fluxes had relatively high oxy­ 18A 0.097 0.082 0.016 0.015 0.76 0.03 0.13 0.0746 gen contents, they did supply useful 19A 0.075 0.050 0.017 0.014 0.80 0.03 0.11 0.0743 information: 20A 0.075 0.070 0.019 0.015 0.88 0.03 0.11 0.0740 21A 0.080 0.060 0.019 0.015 0.81 0.02 0.10 0.0645 1 The main source of oxygen dur­ 22A 0.081 0.070 0.018 0.015 0.79 0.02 0.11 0.0731 ing welding was C02 evolved from 23A 0.069 0.100 0.017 0.014 0.82 0.02 0.11 0.0680 calcination of the CaCO,, in agglomer­ 24A 0.075 0.075 0.018 0.015 0.93 0.03 0.11 0.0460 ated fluxes. 25A 0.074 0.096 0.017 0.015 0.87 0.03 0.11 0.0700 2. The slag composition formed 26A 0.074 0.092 0.017 0.015 0.83 0.03 0.11 0.0486 27A 0.065 0.08 0.017 0.013 0.93 0.03 0.11 0.0470 during welding was similar to that in 28A 0.075 0.06 0.018 0.015 0.77 0.03 0.11 0.0471 later tests using carbonate-free fluxes; 29A 0.064 0.10 0.017 0.014 0.97 0.03 0.11 0.0473 this was of particular value when con­ 30A 0.100 0.09 0.019 0.016 0.94 0.02 0.14 0.0345 sidering the possible formation of 1B 0.105 0.21 0.005 0.010 1.35 0.05 0.05 0.007 some pseudo-equilibrium reaction 1 2B 0.11 0.18 0.0 0.019 1.35 0.02 0.19 0.064 temperature during welding. 7B 0.095 0.19 o.ot 0.017 1.07 0.02 0.14 0.0264 3. The FeO contents of slag samples 0.15 0.0215 17B 0.105 0.18 0.0". 0.015 1.04 0.03 were similar to those of carbonate-free 28B 0.100 0.12 P.014 0.016 1.12 0.01 0.15 0.026 fluxes—Table 5. 30B 0.09 0.14 0.014 0.017 1.07 0.02 0.15 0.0175 In order to remove the effects of

66-s I MARCH 1978 CaC03 in flux formulations, a series of fused fluxes were made based on the CaO-AI203-CaF, system. In order to 0-12 maintain consistency with agglomer­ ated flux formulations, fused fluxes 0-10 contained 2.5 wt-% Na20, 0.5 wt-% K,0 %I0] and 1 wt-% Si02. The oxygen content of deposits made with fused fluxes DEPOSIT 0.08 were lower than those made with agglomerated fluxes—Fig. 3 and Ta­ ble 4. 0-05 Employment of fused CaF, as a flux for submerged arc welding produced 004 weld deposits with oxygen contents equal to that of the electrode—Table 4. 002 Apart from slight voltage fluctuations (±2 V), about any chosen value CaF, flux performed satisfactorily during deposition of ten-high pads. However, in industrial welding situations the low %[CaO] viscosity of this flux would lead to Fig. 3—Relation between the deposit oxygen content and the calcium carbonate inadequate bead profiles. content (given as the equivalent CaO content) of agglomerated fluxes. • are The addition of 2.5 wt-% Na,0 and agglomerated fluxes and A are fused fluxes) 0.5 wt-% K,0 to fused CaF2 flux produced a marked increase in the deposit oxygen content, i.e., from pared. further suggested that oxygen pick-up 0.007% oxygen (that of the electrode 2. The (FeO) content of slags do not at the electrode tip is determined by a employed) to 0.064% oxygen. In effect, differ markedly even when the differ­ gas/metal reaction. The main source the deposition of low oxygen content ence in deposit oxygen content are of oxygen is due to carbonate and weld metal is promoted by the large. oxide decomposition during welding, employment of fluxes containing the 3. Small additions of Na,0 and K,0 i.e., CaCO, decomposition (in agglom­ minimum content of alkali oxides raise the oxygen content of weld erated fluxes), CaO, Na20 and K,0 consistent with adequate arc stability deposits and oxygen-free CaF flux decomposition (in fused fluxes), and during welding. produces weld metal having the same Na,0 and K20 decomposition in the oxygen content as the electrode wire case of fused CaF2 flux. The reactions occurring are: Commentary used. All the above effects can be CaCO, ^CaO + C02(,,) The important factors apparent from explained by assuming that oxygen CaOssCa(g) + 72 02(s) a consideration of the tests using pick-up occurs at the electrode tip Na20^2 Na(s) + Vt 02Q agglomerated and fused fluxes are as region during welding, i.e., the impor­ follows: tance of C02 evolution in agglomer­ FeO is produced during welding and 1. Almost identical slag composi­ ated fluxes, of Na20 and K20 addi­ comes from oxidation of the metal. tions produce deposits with markedly tions, and the lack of correlation Some of the iron oxide could be due to different oxygen contents when ag­ between FeO content in slags and desorption of oxygen from the weld glomerated and fused fluxes are com- oxygen content in weld deposits. It is metal during cooling and solidifica-

Table 5—Flux Formulations Where Fe,0:l and MnO Additions Are Made to CaF2 • Al203 • CaO Fl uxes

Fe203 Al additn. additn. MnO additn. O weld FeO MnO Electrode Flux designation to flux, % to flux, % to flux, % metal, % (slag), % (slag), %

S1 1B — _ 0.032 1.44 0.32 S1 1B 0.20 - - 0.015 1.36 0.22 S1 7B — - 0.043 2.04 0.65 S1 7B 0.20 - - 0.034 1.49 0.19 S1 7B 0.80 - - 0.030 1.43 0.71 S1 28B - - 0.032 0.79 0.19 SI 28B 0.40 — - 0.030 0.69 0.13 S2 7B - - 0.029 1.73 0.19 S2 7B 0.40 - - 0.024 1.58 0.13 S2 28B - - 0.031 0.69 0.26 S2 28B 0.40 - - 0.024 0.63 0.19 S2 28B 2.0 - 0.037 1.49 0.38 S2 28B 3.0 - 0.047 2.48 0.64 S2 28B - 1.0 0.034 1.22 0.77 S2 28B - 2.0 0.040 1.07 1.22 S3 30A - - 0.0345 1.29 0.29 S3 7A - - 0.084 1.87 0.33 S3 7B 0.025 1.40 0.25 S3 - - 7B 0.20 0.016 1.35 0.31 S3 - - 7B 0.40 - - 0.012 1.68 0.33

WELDING RESEARCH SUPPLEMENT I 67-s 1.2 ' %Mn %[0] WELD METAL ELECTRODE TIP

11

to

•9 ^S. • •

•8 • •

• • •

-7 • ^ DATUM LEVEL • •

• -6

%AI in FLUX

•5 Fig. 5—Relation between the aluminum content in the flux and oxygen content of electrode tips (the datum level is that when electrode tips are collected using fused CaF., flux). S4 electrode and (45 wt-% CaFz, 35 wt-% Al,0„ 20 wt-% CaO) flux .4

c 10 20 30 40 50 60 70 CONDUCTIVITY (mhos) %CaO in FLUX ohm cm Fig. 4—Relation between the manganese content of deposits and the calcium carbonate content (given as the equivalent CaO content) of agglomerated fluxes. S3 electrode employed through­ out

%[0) WELD METAL

°/o AI in FLUX

Fig. 6—Effect of aluminum additions in the flux on deposit oxygen contents. S4 electrode and (45 wt-% CaF,, 35 wt-% Al.,0„ 20 wt-% CaO) flux

Fig. 7—Specific electrical conductivity vs. temperature data lor laboratory-prepared and proprietary fluxes, where A is Lincoln No. tion. However, the amount of FeO 7 flux; B is BX7 electroslag flux; C is BOC-MUREX Muraflux A; D is present in slags is too great to have Oerlikon OP41TT flux; E is BOC-MUREX 1009 flux; F is (45 wt-% come wholly from this source. Other CaFi, 35 wt-% AI..O* 20-wt% CaO) flux possible sources could be from the dissolution (by the slag) of an oxide surrounding droplets. Certainly there content of weld deposits. film formed on droplets or from the is no relation between the FeO Additional experiments were carried oxidation or iron vapor in the gas film content of the slag and the oxygen out to confirm that oxygen pick-up

68-s I MARCH 1978 during welding was dependent on Table 6-Weld Deposit Analyses When Using 45 wt-% CaF , 35 wt-% ALO „ 20 wt-% CaO gas/metal reactions at the electrode 2 Flux Containing Aluminum Additions, Wt-% tip region:

1. Addition of Fe,03 and MnO to W1 W3 W5 W7 W9 W11 Fluxes. Two series of welds were de­ (AW)"' (AW)"' (AW)"" (AW)"" (AW)"" (AW)"" W2 (SR) W4 (SR) W6 (SR) W8 (SR) W10 (SR) W12 (SR) posited using (45 wt-% CaF2/ 35 wt-% Al203, 20 wt-% CaO) flux containing C 0.11 0.13 0.11 0.13 0.13 0.12 2% and 3% Fe,0.„ and 1% and 2% MnO. Mn 1.49 1.52 1.52 1.56 1.60 1.59 The results of these tests are given in Si 0.11 0.14 0.16 0.18 0.20 0.21 Table 5. S 0.013 0.008 0.010 0.010 0.009 0.009 Although there was a slight increase P 0.020 0.020 0.019 0.019 0.019 0.019 in oxygen content of welds due to O at 1650 C 0.031"" 0.0265"" 0.0205"" 0.0145"" 0.0105"" 0.0095"" O at 1800 C 0.034"" 0.026"" 0.0204"" 0.0157"" 0.0145"" 0.011"" dissociation of Fe,03, it was notable Al (flux) 0.40 0.80 1.20 1.60 2.00 that the fractional increase in oxygen - Al (total) 0.006 0.0095 0.0125 0.0195 0.031 0.045 content of deposits was very much less Al (acid soluble) 0.0015 0.003 0.0055 0.0125 0.022 0.034 than the increase in the FeO content Al (acid insolubl ;)0.0045 0.0065 0.007 0.007 0.009 0.011 of the slags. These results again Nitrogen 0.011 0.009 0.008 0.010 0.013 0.012 confirmed that there was no correla­ (%AI)2 X (%0)' 1.4 X 10" 1.2 x 10 • 1.3 x 10"' 1.5 X 10-" 2.9 x 10" 2.7 X 10-" tion between the FeO content of the ""AW—as-welded; SR—stress-relieved. slag and the weld deposit oxygen ""Four tests carried out. content. 2. Examination of Electrode Tip Ox­ 0.30 poise at 1600 C (2912 F), surface 1 ygen Contents. Electrode tips were tension of 300 erg/cm- in the Widgery "' indicated that the vac­ collected when welding with 45 wt-% 1500-1700 C (2732-3092 F) temperature uum fusion analysis method gave range, and a liquidus temperature of much lower oxygen values than the CaF,, 35 wt-% Al203, 20 wt-% CaO flux-Fig 5. It is readily apparent that 1370 C (2948 F)." This particular flux neutron activation analysis technique the mean oxygen content of the formulation also had a specific elec­ on similar samples, and that inclusion trical conductivity similar to that of content evaluations based on metallo­ electrode tip region was slightly higher 17 than that of the weld deposit when other proprietary flux brands —Fig. 7. graphic counting could not be related using the same welding condi­ Increasing aluminum contents in the to the oxygen values found by the vacuum fusion method. Also when tions-Fig. 6. This confirmed that oxy­ flux during deposition of ten-high deoxidizing steel melts with aluminum gen transfer took place at the elec­ pads produced a continuous decrease additions, Franklin" found that the trode tip and little change occurred in in the oxygen content of the final vacuum fusion method at a test the weld pool. runs—Fig. 6. As the aluminum content of the flux increased, there was a temperature of 1650 C (3002 F) gave The droplet formation stage is char­ continuous increase in the aluminum values 60 ppm lower than when acterized by approximately 2400 C testing was carried out at 1800 C (3272 (4352 F) high temperatures,12 vigorous content of weld deposits. Figure 8 shows the relationship of aluminum F). In the case of silicon-killed steel motion of droplet material and emana­ melts Franklin found similar oxygen tion of the arc from localized anode content in the flux, total aluminum content of the weld metal, and the values at both 1650 and 1800 C (3002 spots (during reverse polarity weld­ and 3272 F) testing temperatures. In ing). At such temperatures, all avail­ acid-soluble and acid-insoluble alumi­ num contributions to the total deposit Table 2 it is apparent that some weld able oxygen will be dissolved in liquid deposits gave higher oxygen contents aluminum content. It is readily appar­ iron. The only other source of oxygen when tested at the higher tempera­ ent that the acid-insoluble aluminum in solid weld deposits is trapped flux. ture, i.e., by as much as 40 ppm. Since the solubility of oxygen in solid level was uninfluenced by increasing iron is extremely low (0.003% ±0.003% aluminum contents in the flux—Table The relationship between the alumi- at 1295 to 1380 C (2363 to 2516 F))," oxygen is precipitated as iron, manga­ 3 nese and silicon oxides (or combina­ %AL10 tions of these) during weld metal 70 in DEPOSIT solidification. Since the rate of solidifi­ cation is high, it is unlikely that these 60 oxides will have an equilibrium com­ position. 50 Employment of Deoxidants During Welding 40 yS / Deoxidation by means of silicon and aluminum additions is an inherent 30 total >^ / - feature of steelmaking practice. Con­ sequently, aluminum additions were 20 made to the fused flux formulation (45 ^*^ ^ ' acid soluble wt-% CaF2, 35 wt-% Al203, 20 wt-% CaO) in order to lower the final weld •—————- 10 '"T-" ~~ deposit oxygen content. This particular - i— — T^^"^ "* acid insoluble

flux formulation provided excellent ,— . — • ' running characteristics and good slag 0 04 OS 12 16 20 removal properties, and had known %Al in FLUX thermo-physical properties—viscosity Fig. 8-Relation between aluminum content in the flux and total, acid-soluble of 0.10 poise at 1500 C (2732 F), and and acid-insoluble aluminum content in submerged arc deposits

WELDING RESEARCH SUPPLEMENT I 69-s num content in the flux and the Derivation of an effective reaction program it is likely that the reaction is oxygen content at the electrode tip temperature during submerged arc determined by precipitation of Al,03 during welding is shown in Fig. 5. welding assumes that equilibrium ex­ from the weld metal during cooling, Since the oxygen content of ten-high ists during welding and that mean­ i.e., the solubility product of alumina pads deposited using pure calcium ingful extrapolations can be made [%AI]2 [%OJ3 should be considered. fluoride flux gave oxygen levels equal from equilibrium data applying to Solubility product values in sub­ to that of the electrode wire (see Table lower temperatures and to much merged arc deposits range from 4), the datum level for tests where simpler melts. If slag/metal equilib­ 1.29,10-' to 2.9.10"' and are typical of electrode tips were collected was that rium exists during submerged arc values occurring during conventional when using this particular flux, i.e., welding using laboratory prepared steelmaking practice.1" In effect, the 0.016% oxygen (the mean of ten fluxes, there should be a distinct rela­ conditions attained when aluminum tests). tionship between the FeO content of additions are made during submerged It is readily apparent from Fig. 5 that the slag and the oxygen content of arc welding are similar to those occur­ the presence of aluminum in the flux deposits. ring when aluminum additions are modified the oxygen content of metal The activity coefficient y°FeO is made to molten steel containing oxy­ at the electrode tip. In effect the high in CaF,-rich slags. However, gen in the ladle or mold during steel presence of 1.2 wt-% aluminum in the measurements of the activity of FeO in manufacture. It also follows that flux successfully counteracted the ox­ CaF2-AI,03 and CaF2-CaO slags sug­ employment of a solubility product idizing tendency of the 45 wt-% CaF2, gests that Y°FeO in the slags used value of 10" during submerged arc 35 wt-% Al203, 20 wt-% CaO flux during submerged arc welding is close welding will permit calculation of the formulation. In this connection the to a value of one. This is particularly total aluminum content when the addition of 1.2 wt-% aluminum to the true if temperatures in the range of deposit oxygen content is evaluated flux decreased the oxygen content of 2000 C (3632 F) are considered. It has using vacuum fusion analysis. the final run of ten-high pads to that of been shown that there is no distinct the electrode employed—Fig. 6. In­ correlation between the FeO content creasing the aluminum content of the of slags and the oxygen content of flux beyond 1.2 wt-% permitted depo­ weld deposits when welding with a Conclusions sition of ten-high pad weld deposits number of electrodes of varying man­ 1. The final oxygen content of with oxygen contents less than that of ganese content, and with various flux submerged arc weld deposits depends the electrode employed. However, formulations. It follows that slag/ on the oxygen content of the elec­ production of such low oxygen values metal equilibrium is not attained trode and on the chemical composi­ was counteracted by a marked in­ during submerged arc welding. Since tion of the flux employed. When using crease in the content of aluminum in the derivation of effective reaction an oxide-free flux such as fused CaF, solution in deposits—Fig. 8. temperature depends on this as­ the weld metal oxygen content sumption, the concept of an effective At the inception of this program it equaled that of the electrode. When reaction temperature during welding was thought that aluminum would act using low oxygen potential flux formu­ is untenable. mainly in the weld pool, i.e., by deox­ lations composed of mixtures of CaF2, idizing the molten weld metal. How­ Weld deposits were analyzed for Al,03 and CaO, deposit oxygen con­ ever, the effect of aluminum additions oxygen and for the total, soluble and tents were markedly lower in fused on the oxygen content of electrode insoluble aluminum contents when fluxes than in agglomerated fluxes. tips suggested that it had an influence deoxidant additions were made during 2. The main source of oxygen when on the oxygen content of the elec­ welding. The aluminum content of welding with fluxes composed of CaF2, trode tip region, i.e., that it "mopped liquid weld metal comprises dissolved Al,03 and CaO was that due to up" oxygen in the arc cavity and aluminum and precipitated Al,0.,. On decomposition of flux constituents reduced the oxygen potential sur­ solidification, the dissolved oxygen during welding, i.e., CaC03 decompo­ rounding growing droplets. content in liquid weld metal precip­ sition in agglomerated fluxes, CaO Using aluminum additions it was itates as further Al,03. Consequently, decomposition in fused fluxes, and possible to reduce the oxygen content the insoluble aluminum content Na20 and K20 decomposition when of weld deposits below that obtain­ which was analyzed in deposits came alkali oxide additions were made to able with aluminum-free fluxes; in­ from two distinct sources. In addition fluxes. deed, with aluminum additions in rapid cooling during weld solidifica­ 3. Oxygen pick-up during welding fused CaF2 flux it was possible to lower tion could allow oxygen precipitation with CaF,-AI,03-CaF2 fluxes was deter­ the deposit oxygen content to below as FeO or hercynite (FeALO,). mined by gas/metal reactions occur­ that of the electrode—Table 5. Table 5 Bearing this in mind, examination of ring at the electrode tip region. No also indicates that the low oxygen aluminum deoxidation during sub­ correlation existed between the FeO contents of welds made using alumi­ merged arc welding is best ac­ content of slags and the oxygen num additions were not reflected in complished using the total analyzed content of weld deposits. Conse­ significantly lower FeO contents in aluminum content values—Table 6. quently, slag/metal equilibrium was slags and confirmed the previous data Deoxidation using aluminum de­ not attained, and derivation of an on aluminum-free fluxes. pends on the reaction: effective reaction temperature during welding was not feasible. 2 [Al] + 3 [0]^AL03 Chemical Equilibrium and 4. When welding with (45 wt-% Effective Reaction Temperature CaF2, 35 wt-% Al,03, 20 wt-% CaO) and, the equilibrium quotient (K) is flux, aluminum additions lowered the During Welding given as: final weld metal oxygen content to below that of the electrode. Such low Many investigators have deduced "AI..O, oxygen levels were obtained at the effective reaction temperatures for K 2 3 slag/metal interaction during welding [a„] [a„] expense of increased contents of and values range from 1520 to 2500 C The equilibrium quotient (K) has a aluminum in solution. Aluminum ad- (2768 to 4532 F), depending on the value of 1014 at 1600 C (2912 F)."; ditons in the flux modified the oxygen element considered and the process Although the activity of alumina is less content of the electrode tip region, employed. than one for the slags employed in this and emphasized the importance of

70-s I MARCH 1978 electrode tip reactions in determining lournal ol Iron & Steel Inst., 1972 March, Experimental Procedure the final deposit composition during pp. 153-162. submerged arc welding. 8. "British Standard Method for Alumi­ The materials employed (plate, elec­ nium Analysis in Iron & Steel," 8.5. Hand­ trode and laboratory-prepared fluxes) References book, No. 19, Photometric Method. were similar to those noted in Part I. 9. Christensen, N., and Chipman, )., The welding conditions were also 1. Lewis, W. )., Faulkner, C. E., and Riep- "Slag-Metal Interaction in Arc Welding," identical. pel, P. )., "Flux and Filler Wire Devel­ WRC Bulletin No. 15, 1953. Tests examining the effect of oxygen opments for Submerged Arc Welding of 10. Belton, C. R., Moore, J., and Tankins, HY80 Steel," Welding lournal, 40 (8), Aug. S., "Slag-Metal Reactions in Arc Welding," content on the notch toughness of 1961, Research Suppl., pp 337-s to 345-s. Welding lournal, 42 (7), July 1963, Research submerged arc welds employed fluxes 2. Palm, J. H., "How Fluxes Determine Suppl., pp. 289-s to 297-s. comprising varying contents of CaF2, the Metallurgical Properties of Submerged 11. Tuliani, S. S., Boniszewski, T., and AljOj and CaO, and electrode with the Arc Welds," Welding journal 51 (8), Aug. Eaton, N. F., "Carbonate Fluxes for following composition: 0.12% C, 0.29% 1972, Research Suppl., 358-s to 360-s. Submerged Arc Welding of Mild Steel," Si, 0.012% S, 0.012% P, 1.59% Mn, 3. Potapov, N. N., and Lyubavskii, K. V., Welding & , July 1972, 0.008% (O). "Oxygen Content of Weld Metal Deposited pp. 247-259. The effect of aluminum on notch by Automatic Submerged Arc Welding," 12. Erohin, A. A., "Heat Balance of Elec­ toughness was investigated using the Welding Production, 1971, No. 1, pp. 16-20 trode Melting Process in Arc Welding," (Welding Institute translation from Rus­ Physics of the Arc Symposium, 1962, control flux formulation of 45 wt-% sian). London, pp. 164-170. CaF,, 35 wt-% Al203, 20 wt-% CaO and 4. Potapov, N. N., and Lyubavskii, K. V., 13. Sims, C. E., "The Nonmetallic Con­ an electrode with 0.11% C, 0.24% Si, "Interaction between the Metal and Slag in stituents of Steel," Trans. Met. Soc, 0.03% S, 0.03% P, 2.10% Mn, 0.017% the Reaction Zone During Submerged Arc A.I.M.E., |une 1959, V. 215, pp. 364-393. (O). Welding," Welding Production, 1971, No. 7, 14. Nafziger, R. H., "Oxide-Fluoride Flux Charpy specimens were cut from the pp. 14-18 (Welding Institute translation). Systems for Electroslag Melting," /. of final run of ten-high pads and, there­ 5. North, T. H., The Distribution of Metals, V. 25, Nov. 1973, pp. 55-61. fore, notch toughness results apply to Manganese between Slag and Metal during 15. Widgery, D., "De-Oxidation Practice single-run weld metal. Two forms of Submerged Arc Welding," Welding Re­ for Mild Steel Weld Metal," Welding lour­ search Abroad, Jan. 1977, Vol. XXIII, pp. nal, 55 (3), March 1976, Research Suppl., Charpy specimens were employed in 2-40. pp. 57-s to 68-s. this program—2.07 mm (0.08 in.) diam­ 6. Franklin, A. O, Rule, O, and Widdow- 16. Fitterer, C. R., "Oxidation and Reoxi- eter specimens and 10 X 10 mm son, R., "Effect of De-Oxidation Technique dation in Modern Steel Refining," Chemical (0.39 X 0.39 in.) specimens. When on the Oxygen Content of Inclusion Type Metallurgy ol Iron & Steel, 1971, University 10 X 10 mm (0.39 X 0.39 in.) Charpy in 0.2% Carbon Steel," journal of Iron & of Sheffield, published by The Iron & Steel specimens were employed, great care Steel Inst, 1969 Sept., pp. 1208-1218. Institute, pp. 184-191. was taken to cut samples only from the 7. Bruch, )., "Determination of Cases in 17. Noor, M., North, T. H., and Bell, H. B. final run to ten-high pads. In all tests Steel and Application of the Results," to be published. longitudinal Charpy specimens were employed, and notches were cut at right angles to the direction of weld­ ing. Part II—Oxygen Content, Aluminum and Notch Toughness The content of pro-eutectoid ferrite present in weld deposits was evalu­ of Submerged Arc Deposits ated using lineal analysis techniques. A Vicker's microscope with an automatic counting facility was employed ABSTRACT. The influence of alumi­ toughness properties of submerged arc throughout. The percentage of pro- num additions on the notch toughness welds, i.e., Lewis et al'-" noted that eutectoid present at any given alumi­ of submerged arc welds has been 0.027% aluminum in the deposit num level in the deposit was evalu­ investigated. Aluminum concentra­ improved notch toughness properties, ated from an examination of sixty tions up to a critical level of 0.0195% in and greater contents of aluminum had regions, i.e., ten traverses on each of 6 deposits (corresponding to 1.2 wt-% Al a markedly detrimental effect on Charpy specimens. in the flux) improved notch toughness notch toughness when welding HY80 properties. At higher aluminum con­ steel. Also, Kushnarev21 noted that centrations, the toughness properties aluminum additions up to 0.015% had Results and Discussion were worsened. a negligible effect on notch toughness Oxygen Content and Notch Toughness Aluminum additions had the dual and that an aluminum content of advantage in that they lowered the 0.020% was extremely detrimental to Different flux formulations in the the notch toughness of submerged arc final deposit oxygen content (raising CaF2 • Al203 • CaO system produced the upper Charpy shelf values) and welds. different oxygen potentials during modified the weld metal microstruc­ In this connection, contradictory welding (mainly due to differing ture (so that the transition tempera­ results have also been obtained when CaCO., contents in agglomerated ture was lowered). aluminum additions were made during fluxes). For this reason, weld deposit gas-metal-arc welding, i.e., Ul'yanov et analyses and weld metal microstruc­ a/22 noted that aluminum additions up Introduction tures varied. to 0.60% during GMA-CO, welding Changes in deposit microstructure High oxygen content in weld depos­ were always detrimental to notch were partly due to changes in weld 23 its decreases the notch toughness of toughness properties. Widgery noted metal composition and possibly due to weld deposits.1819 It has been shown in that 0.12% aluminum had no detect­ different cooling rates during the Part I that aluminum additions during able effect. welding operation, e.g., increasing car­ submerged arc welding produce very In this paper a systematic study of bonate contents in submerged arc low oxygen contents in deposits. the effect of aluminum additions on fluxes can modify weld cooling rates Contradictory results have been the oxygen content, microstructure due to the endothermic decomposi­ indicated concerning the effect of and mechanical properties of sub­ tion of limestone during welding.22 aluminum additions on the notch merged arc deposits is carried out. Weld samples were consequently

WELDING RESEARCH SUPPLEMENT I 71-s and corroborates previous data.1819 CHARPY ENERGY . Effect of Aluminum Content on Mechanical Properties joules Tensile Tests. Table 7 shows the 60 influence of aluminum on the tensile properties of submerged arc welds in the as-welded and stress-relieved con­ 50 ditions. The stress relief treatment was carried out at 600 C (1112 F) for one hour. L0 With up to 1.6 wt-% aluminum in the flux (corresponding to a deposit aluminum content of 0.031%), no 30 observable effect on tensile properties 002 0-04 005 008 was noted. The only significant change %[0] DEPOSIT occurred in the weld deposited using flux containing 2% aluminum, i.e., the Fig. 9-Relationship between the 20 C Charpy values and the oxygen content of 0.2% proof stress was raised. It is heat-treated deposits (lor sub-standard 2.07 mm diameter meter Charpy spec­ apparent from Table 3 in Part I that the imens) nitrogen content of deposits made with different aluminum additions in fluxes did not vary markedly; also, the Table 7- Tensile Testing Results presence of aluminum nitride precipi­ tation was not detected in the weld Ultimate deposited using flux containing 2% Al 0.2% proof tensile Reduction aluminum. (flux) stress, strength, Elongation, of area, The relationship between aluminum 2 2 % Condition MN/m MN/m % content and tensile properties in Table 0.4 As-welded 514.3 646.9 25 56 7 applies to single-run weld metal and 0.8 As-welded 492.8 597.4 29 57 is similar to that existing when up to 1.2 As-welded 517.8 666.1 28 55 1% aluminum is added to wrought 1.6 As-welded 500.1 647.1 27 57 plate material.24 In multipass welding 2.0 As-welded 542.4 637.9 30 55 situations aluminum additions are 0.4 Stress-relieved 413.3 545.5 31 63 associated with increased tensile 0.8 Stress-relieved 430.4 562.6 31 60 strength, e.g., in submerged arc weld­ 1.2 Stress-relieved 428.2 566.7 29 61 ing,2" in GMA (argon) and GMA (CO,) 1.6 401.0 539.3 31 60 Stress-relieved welding,2225 and in manual-metal-arc 2.00 Stress-relieved 471.5 536.2 29 62 welding.26 Notch Toughness. Figures 10-13 show the marked effect of aluminum given a high temperature furnace noted that manganese changes from additions on the notch toughness of treatment-2 h at 925 C (1697 F), 0.6 to 1.6% caused no observable submerged arc welds, i.e., aluminum followed by 2 h at 650 C (1202 F), and change in the notch toughness proper­ additions up to a critical level improve then air-cooled—in order to eradicate ties of single-run submerged arc de­ the notch toughness properties. microstructural effects. posits, it is possible to relate the Figure 14 shows the relationship Weld composition variations were impact values at 20 C (68 F) with the between the aluminum content of most apparent when considering the weld deposit oxygen content values. deposits and the Upper Charpy shelf manganese content (see Table 4, Part Figure 9 shows that high oxygen values. Upper Charpy shelf values are I). Since Dorschu and Stout" have contents decrease notch toughness raised when aluminum is added up to

TEMPERATURE'C. TEMPERATURE C. Fig. 10— Effect of aluminum additions of the as-welded notch Fig. 11—Effect of aluminum additions on the as-welded notch toughness properties of weld deposits (for 10 x 10 mm Charpy toughness properties of weld deposits (for 10 x 10 mm Charpy specimens). Flux formulation (45 wt-% CaF^ 35 wt-% A/2Oj, 20 wt-% specimens). Flux formulation (45 wt-% CaF,, 35 wt-% AlzO^ 20 wt-% CaO) with S4 electrode CaO) with 54 electrode

72-s I MARCH 1978 TEMPERATURE °C. TEMPERATURE C Fig. 12—Effect of aluminum additions on the stress-relieved notch Fig. 13—Effect of aluminum additions on the stress-relieved notch toughness properties of weld deposits (for 10 X 10 mm Charpy toughness properties ol weld deposits (for 10 x 10 mm Charpy specimens). Flux formulation (45 wt-% CaF.,, 35 wt-XA^O-,, 20 wt-% specimens). Flux formulation (45 wt-% CaF2 35 wt-% AltO„ 20 wt-% CaO) with S4 electrode CaO) with S4 electrode

0.0195% (in the deposit) and are ues at any temperature than the alumi­ toughness properties in different ways, lowered when further aluminum is num-free flux. Lewis et a/20 noted that i.e., decreasing aluminum contents added. the as-welded toughness properties of improved room temperature impact The effect of aluminum on the submerged arc welds in HY80 steel values and lowered sub-zero impact upper shelf values is analogous to that were higher when 0.027% aluminum values. existing when other deoxidant addi­ was added and that further increase in The as-welded toughness results of tions are made during welding, e.g., aluminum content had a markedly Kushnarev21 showed no observable silicon additions in manual-metal-arc detrimental effect on notch tough­ improvement in notch toughness 26 welding. The initial improvement in ness. These workers also indicated that when aluminum was added up to upper shelf values is associated with aluminum additions modified room- 0.015% (in the deposit) and showed a decreased oxygen content of deposits temperature and sub-zero notch markedly deleterious effect of alumi­ (see Fig. 6, Part I) and with decreased num when added up to 0.020%. It is inclusion content. important to emphasize that the Typical inclusions occurring in weld UPPER SHELF ENERGY results in Figs. 10 and 11 apply to deposits made with fluxes containing JOULES single-run welds and the results of 2 21 aluminum additions include alumina, Lewis " and Kushnarev apply to alumino-silicates, inclusions rich in multipass welding situations. calcium and aluminum, and manga­ Figure 16 shows that aluminum nese sulphide. Figure 15 shows inclu­ levels up to 0.0195% produced a sions present in weld deposits, i.e., a marked improvement in the transition 1.5 micro alumino-silicate inclusion, a temperature of stress-relieved welds. 2.5 micron inclusion rich in calcium stress-relieved Further increases in aluminum content and aluminum (which is possibly a slag had a detrimental effect on the tran­ particle caught during weld solidifica­ sition temperature of stress-relieved tion), and a 3 micron alumina inclu­ welds. The transition temperature val­ sion. ue employed was the temperature When the aluminum content of corresponding to an energy absorption deposits exceeds 0.0195%, the content of 70 joules; it was chosen since a of aluminum in solution rises markedly direct relation between the 70 joule (see Figure 8, Part I), and the fall in transition temperature and the 0.4 mm upper shelf values is associated with c.o.d. transition temperature has been increasing contents of aluminum in noted previously in submerged arc solution. Erasmus27 noted a similar welds.28 effect in wrought steels, i.e., aluminum additions in excess of that required to Table 6 in Part I shows that remove nitrogen from solution de­ increasing aluminum contents in de­ creased notch toughness since the posits were associated with increasing soluble aluminum content was in­ manganese and silicon contents in creased. deposits. The manganese content in­ creased from 1.49 to 1.6% while the silicon content increased from 0.11 to Transition Temperature 0.21%. A 0.1% manganese increase It is apparent in Figs. 10 and 11 that 0 10 20 produces an improvement in tran­ the as-welded properties of deposits sition temperature of approximately 4 Fig. 14—Effect of aluminum content in the 29 made using fluxes containing alumi­ flux on the upper Charpy shelf values of C (7 F) in wrought CMn steels, a 5 C num produced higher toughness val­ weld deposits (9 F) improvement in multipass GMA

WELDING RESEARCH SUPPLEMENT I 73-s Fig. 15—Typical inclusions in weld deposits where aluminum additions were made to the flux; left to right A—1.5 micron alumino-silicate inclusion (X10.000); B—2.5 micron inclusion rich in calcium and in aluminum (X.10,000); C—3 micron alumina inclusion (X10.000). A, B, and C reduced 50% on reproduction

(argon) welds,"' a 7 C (13 F) improve­ readily explained by changes in silicon this particular structure promoted im­ ment in single pass GMA (CO,) and manganese content of weld de­ proved notch toughness. The in­ welds,22 and no observable effect on posits. fluence of aluminum in promoting the the toughness of single pass sub­ Change in the aluminum content of formation of acicular ferrite is similar merged arc welds when the manga­ fluxes produced deposits with distinct to that found when vanadium is added nese content was varied from 0.6 to microstructure variations. Figure 17 to welds in MnMoNb pipe steels32-33 1.6%.31 shows the relation between the alumi­ and when titanium is added to CMn An increase in silicon content pro­ num content of the flux and the weld deposits." The deleterious effect duces no observable effect on the content of pro-eutectoid ferrite in the of high aluminum contents in excess notch toughness of single pass sub­ deposit microstructure. Decreasing of 0.0195% in deposits was due to the merged arc deposits when added up to contents of pro-eutectoid ferrite (as rapid increase in the aluminum con­ 0.5%," and promotes decreased tough­ the aluminum content increases) favor tent in solution (see Fig. 8, Part I). ness in GMA (argon) welds when improved toughness in a manner added in the range 0.35 to 0.80%.30 similar to that occurring in GMA (CO,) 23 Conclusions Since the deposits examined in this welds. program were single run deposits, the As the aluminum content was 1. The aluminum content in the flux marked improvement in notch tough­ increased, the content of acicular which reduced the oxygen content of ness of stress-relieved welds was not ferrite increased and the presence of the deposit to that of the electrode

40 TRANSITION 5C % PRO-EUTECTOID TEMPERATURE FERRITE. 30 °C

20 40

30

20

10

20 0 -20 -40 % Al in FLUX . TRANSITION TEMPERATURE °C Fig. 16—Effect of aluminum content in the flux on the transition Fig. 17-Relation between the pro-eutectoid ferrite content and the temperature (corresponding to 70 joules) of stress relieved welds transition temperature of stress relieved welds

74-s (MARCH 1978 employed gave best notch toughness, 1961, Research Suppl., pp 337-s to 345-s Additions of Forgeability, Austenite Coars­ i.e., highest upper Charpy shelf values 21. Kushnarev, D. M., and Svetsinckii, ening Temperature, and Impact Properties and lowest transition temperature. In V. C, "Certain Special Features of the Struc­ of Steel," lournal of Iron & Steel Inst., (an. the submerged arc tests examined, the tures of Submerged Arc Welds Made with 1964, pp. 32-41. High-Basicity Fluxes," Automatic Welding, 28. Farrar, R. A., Juliani, S. S., and critical aluminum level was 1.2 wt-% Al 1972, No. 12, pp. 17-21 (Welding Institute Norman, S. R., "Relationship between Frac­ in the flux giving 0.0195% Al in the translation from Russian). ture Toughness and Microstructure of Mild deposit. 22. Ul'yanov, V. I., Parfessa, C. I., and Steel Submerged Arc Weld Metal," Weld­ 2. Aluminum additions had a dual Shevchuk, R. N., "Effects of the Aluminium ing and Metal Fabrication, Feb. 1974, pp. 68- advantage in that they lowered the in Electrode Wire on the Strength of C02 73. final deposit oxygen content (raising Weld Metal in St. 3 Steel," Automatic Weld­ 29. Van der Ween, "Development of the upper Charpy shelf values) and ing, 1974, No. 2, pp. 15-19 (Welding Insti­ Steels for Off-shore Structure," Rosenhain modified the weld metal microstruc­ tute translation from Russian). Conference, London 1976, pp. 223-232. ture (so that the transition tempera­ 23. Widgery, D., "Deoxidation Practice 30. Moll, R. A., and Stout, R. D., "Compo­ for Mild Steel Weld Metal," Welding lour­ sition Effects in Iron-Base Weld Metal," ture was lowered). nal, March 1976, 55 (3), Research Suppl., Welding lournal, 46 (12), Dec. 1967, pp. 57-s to 68-s Research Suppl., pp. 551-s to 561-s. References 24. Sage, A. M., and Copley, F. E. )., "The 31. Dorschu, K. E., and Stout, R. D., Notch Ductility and Tensile Properties of "Some Factors Affecting the Notch Tough­ 18. Masubuchi, K., Monroe, R. E., and Some Synthetic Mild Steels," lournal of Iron ness of Steel Weld Metal," Welding lournal, Martin, D. C, "Interpretive Report on Weld & Steel Inst., August 1960, pp. 422-438. 40 (3), March 1961, Research Suppl., pp. 97-s Metal Toughness," WRC Bulletin No. 19, 25. Sibley, C. R., "The Effect of Alumi­ to 105-s. pp. 1-38. num Additions to Mild Steel Weld Metal," 32. Sawhill, ). M., and Wada, T., "Proper­ 19. Kubli, R. A., and Sharav, W. W„ "Ad­ Welding lournal, 35 (8), Aug. 1956, Research ties of Welds of Low Carbon Mn-Mo-Cb vancement in Submerged Arc Welding of Suppl., pp. 361-s to 368-s. Line Pipe Steels," Welding lournal, 54 (1) High Impact Steel," Welding lournal, 40 26. Sakaki, H., "Effect of Alloying Ele­ Jan. 1975, Research Suppl., pp 1-s to 11-s. (11), Nov. 1961, Research Suppl., pp. 497-s ments on the Notch Toughness of Basic 33. Dolby, R. E., "Factors Controlling to 502-s. Weld Metals," lournal Japan Welding Soc, Weld Toughness—The Present Position," 20. Lewis, W. )., Faulkner, C. E., and Report I. Effect of Silicon, 1959, V. 28, pp. Part 2-Weld Metals, Welding Institute Rieppel, P. |., "Flux and Filler Wire Devel­ 858-863 (Translation by Welding Inst. from Report 14/1976/7 1976, pp. 1-29. opments for Submerged Arc Welding of lapanese). 34. Ito, Y., and Nakanishi, M., Interna­ HY80 Steel," Welding lournal, 40 (8), Aug. 27. Erasmus, L. A., "Effect of Aluminium tional Institute of Welding Document, I.I.W. Doc, XIIA-133-75.

WRC Bulletin 229 August 1977

(1) Dynamic Fracture-Resistance Testing and Methods for Structural Analysis by E. A. Lange

The potential for the initiation of fast fracture can be predicted by the recently developed technology of linear elastic fracture mechanics (LEFM) which has produced the basis for an analytical approach to fracture resistance and structural integrity. To make fracture mechanics a viable engineering design tool, empirical correlations between practical dynamic test results and the basic parameters are needed. In this paper the attributes and limitations of the Charpy, Drop Weight-Nil Ductility Transition Temperature, Drop Weight Tear, and Dynamic Tear tests are discussed with respect to providing information useful in structural integrity analyses.

(2) Junction Stresses for a Conical Section Joining Cylinders of Different Diameter Subject to Internal Pressure by W. J. Graff

The conical transition section of a cylindrical pressure vessel was instrumented inside and outside with electric resistance strain gages, and from the longitudinal and circumferential strains measured experimentally, the corresponding stresses were determined. The results were compared with calculated stresses from the theory of shells. The experimentally determined stresses exceeded code membrane stresses in the immediate vicinity of the junctions. Publication of this paper was sponsored by the Welding Research Council. The price of WRC Bulletin 229 is $7.50 per copy. Orders should be sent with payment to the Welding Research Council, United Engineering Center, 345 East 47th Street, New York, NY 10017.

WELDING RESEARCH SUPPLEMENT I 75-s