Grain Refinement in Electroslag Weldments by Metal Powder Addition

The electroslag process when operated with a modified strip electrode shows promise of becoming a viable alternative to high-speed welding

BY S. LIU AND C. T. SU

ABSTRACT. In this study, the perfor­ weldments made using this process also joining thick sections, usually 19.1 mm mance of a metal powder cored strip showed significant improvement over (3/i in.) and greater. The materials welded electrode in electroslag welding was the conventional electroslag weldments. are generally limited to low-carbon steels, studied. The process demonstrated sta­ Additionally, chemical analysis of the low-alloy steels, some stainless steels, and ble operation and produced welds with welds indicated that it is possible to some nickel alloys. If the welding must be full penetration. Weld deposition rate control the chemistry of the weld pool interrupted before completion, restarting was as high as 38 kg (84 Ib) per hour. and final weld properties by incorporat­ is always a problem because of the diffi­ Using a combination of experimental ing different alloying elements in the core culty of completely removing the solidi­ measurements and theoretical heat flow of the electrode. In summary, the metal fied slag bath. Finally, electroslag welding calculations, a constitutive equation, powder cored strip electrode presented can only weld in the vertical or near- which correlates welding parameters very promising results and can be consid­ vertical position. such as current, voltage, wire feed rate ered as an alternative technique for elec­ In addition to the operational charac­ and weld travel speed, was derived. troslag welding. teristics described above, it is also known Specific heat input of the process was that electroslag welds are significantly found to be directly proportional to the Introduction different from the welds made by con­ fourth power of voltage and inversely ventional open arc processes. The cool­ proportional to the square of current. Electroslag welding is mainly used to ing conditions produced by the electro­ The new consumable aiso minimized the achieve high deposition rate in thick- slag welding process are extremely slow, deleterious effect of the long thermal section joining applications. Plates from resulting in excessively coarse grains in cycle of slow heating and cooling rate in 52 to 300 mm (2 to 12 in.) in thickness can the as-solidified weld metals, constituted electroslag welding, with grain refine­ be welded at a rate up to 0.6 to 1.2 m (2 mainly of grain boundary ferrite and ment in the weld metal and heat-affected to 4 ft) per hour, with one or more some ferrite sideplates (Ref. 1). Further­ zone (HAZ). By increasing the welding electrodes. Despite the recent decrease more, the extremely large columnar current from 1000 to 1500 A at 40 V, the of interest of fabricators in electroslag grains are often segregated with impuri­ weldment structure was modified with welding and the economic conditions for ties causing grain boundary separation to the total HAZ width reduced from the heavy manufacturing industries, occur. The effect of cooling rate can also approximately 11 to less than 8 mm (0.43 major applications of this process can still be noticed in the base metal next to the to 0.31 in.). Grain refinement in the be found in shipbuilding, pressure vessel, fusion line. The extended period of heat­ coarse grain heat-affected zone (CGHAZ) structural and basic machinery fabrica­ ing at high temperatures coarsens the also occurred. The average prior austen­ tion. The advantages of electroslag weld­ inclusions or second phase particles, ite grain size in the CGHAZ of the higher ing, over the other arc welding pro­ allowing the austenite grains to grow. current welds was less than 300 fim. cesses, include simple joint preparation, Subsequent slow cooling will allow the Temperature distribution along the fusion high deposition rate, minimal amount of formation of large volume fractions of line also demonstrated the influence of distortion, vertical and single-pass opera­ allotriomorphic structure with carbide the metal powder cored strip electrode tion. However, there are also many dis­ precipitation and ferrite sideplates, reduc­ on the heat flow conditions. In spite of advantages and limitations associated ing the toughness of the CGHAZ of the similar peak temperatures, higher current with the process. It can only be used for weld (Ref. 2). Consequently, electroslag welds showed significantly shorter ther­ welds often present poor mechanical mal cycles with higher heating and cool­ properties, low strength and toughness. ing rates. Mechanical properties of the Postweld heat treatments are often spec­ ified to restore the impact toughness of KEY WORDS the weld metal and the HAZ such that Electroslag Weldment they satisfy the design requirements. This 5. LIU is with the Center for Welding and Grain Refinement is sometimes uneconomical, and other Joining Research, Department of Metallurgical Metal Powder Additive times, impractical. and Materials Engineering, Colorado School of Cored Strip Electrode Mines, Golden, Colo. C. 75 SU is with the Metal Powder Electrode University of Texas, Department of Industrial High Productivity Engineering, Arlington, Tex. Constitutive Equation Heat/Cooling Effects Electroslag Processes This research work was partly conducted at Hardness Measurements More recent efforts in arc and electro­ the Pennsylvania State University, and the Charpy Impacts slag welding focus on process develop­ original paper was presented at the 68th Alloy Additions Annual AWS Meeting, held March 22-27, ments which improve productivity as 1987, in Chicago, III. well as mechanical properties (Refs. 3-

132-s|APRIL 1989 BASE METAL

SOLID FLUX'

ELECTRODj ? -CONSUMABLE GUIDE IK COPPER |-T I - WATER COOLING SHDSHOEF —' I 2- ELECTRODES 3- INERT GAS METAL POWDER DELIVERY 4-MAGNETIC SCREENING

Fig. 1 — Narrow groove electroslag welding with single and double strip Fig. 2 — Equipment setup for square-grooved high-speed welding with electrode and solid flux blocks to prevent arcing (Ref. 13) inert gas metal powder delivery system. Left-Plate thickness 10-40 mm; Right-Plate thickness 40-100 mm (Ref. 18)

20). One of the two most used input was 1 kj/mm2 (645 kj/in.2). Despite ing adopted in the 1960's at the same approaches is to decrease the overall the improved weld mechanical proper­ time that powdered filler metals were heat input of the process by decreasing ties reported, the production rate is still being applied in arc welding. These two the volume of the weld deposit and the not high enough, and the specific heat concepts were combined in two ESR dimension of the root opening. This leads needs to be further lowered. facilities, one in the U.S. to produce to an increase in the cooling rate to refine Venkataraman, ef al., and Atteridge, ef stainless steels (Refs. 8, 9) and one in the weld metal and HAZ microstructure. al. (Refs. 5, 6), investigated the effects of Europe to produce high-quality forging The other method is to increase the root opening and reported that narrow steels (Refs. 10, 11). In both facilities, one deposition rate and welding speed such groove welds resulted in higher base or more low-carbon steel strips carried that the process heat is distributed metal dilution than standard groove the electric current, and alloy powders throughout a long section of the weld, welds. Dilution was also observed to were conveyed to the strips to which it lowering the effective heat input. increase linearly with decreased joint adhered by the magnetic field generated spacing. The specific heat input of these by the currents carried by the strip elec­ Narrow Groove Approach narrow groove welds was comparable to trodes. This led to the use of powdered Avramenko, et al. (Ref. 12), suggested that reported by the Japanese research­ filler metals in electroslag welding. that the most expedient and simple way ers (Ref. 13). Smirnov and Efimenko (Ref. 17) report­ of increasing the productivity of the elec­ ed that powdered filler metal in electro­ Metal Powder Addition troslag welding process is the reduction slag welding refined the weld metal and of groove width with a simultaneous In arc welding, the addition of pow­ HAZ microstructure. Low-temperature increase of electrode wire feed rate. This dered metal, which can be easily melted impact strength of the weldments in the procedure decreases mainly the amount by the excess process heat of the arc, as-welded condition increased by a fac­ of weld metal deposited in the weld joint was introduced in the early 1960's to tor of two. Deposition rate was approxi­ and the time needed to complete the increase the deposition rate in what mately 13 to 24 kg (30 to 50 Ib) per hour weld. However, the extent to which the became known as the "Arcmetal" (Ref. with slight reduction in the specific heat groove opening can be reduced is limited 14) and the "Bulkweld" process (Ref. 15). input. by the stability of the process and the size Reynolds and Kachelmeier reported Eichhorn, ef al. (Refs. 18-20). added of the electrode and guide tube inserted increased deposition rate, decreased powder metal filler and increased the into the opening. If the groove is too heat input, reduced HAZ, and lower deposition rate while reducing the over­ narrow, arcing may occur between the consumption of fluxes in hardfacing and heating of the weld pool. The ferromag­ electrode and the base metal. cladding with powdered metal filler. The netic metal powder was supplied to the Using solid flux blocks to avoid arcing powder-to-wire weight ratio was ap­ weld by a flow of argon gas —Fig. 2. The between a solid strip electrode in a thin proximately 1:1. Campbell and Johnson powder-to-electrode weight ratio was channel consumable guide and the base (Ref. 14) used powdered filler metals to 1.2 : 1. He used single and dual electrode metal, Watanabe, ef al. (Ref. 13), was reduce the heat input of heavy butt and configurations and different powder able to weld with groove openings as fillet welds in both submerged arc and delivery locations to increase the welding small as 12 mm (0.5 in.). Figure 1 shows gas shielded arc welding processes. speed to approximately 230 mm (9 in.) the electrode setup for the narrow Thomas (Ref. 16) reported on the use of per minute. A simultaneous reduction of groove process. Sound welds with good highly alloyed powders to produce energy input was also observed. Notch penetration were made at approximately cobalt-based hardfacing deposits using impact testing results measured in sam­ 800 A and 34 V, with 100-mm (4-in.) thick unalloyed cobalt strip. ples taken at distances of 1 and 2 mm plate and an opening of 15 mm (0.6 in.), The development . of electroslag from the fusion line showed very good attaining a welding speed of about 15 remelting (ESR) as a process for refining Charpy impact toughness. mm (0.6 in.) per minute. The specific heat alloy steels and other alloys was becom­ Reynolds (Ref. 21) reviewed the vari-

. WELDING RESEARCH SUPPLEMENT 1133-s Table 1—Chemical Composition of the Electrode and Base Metal Used in the Welding Experiments (wt-%)

C Mn Si P S Cr Ni Mo Al Ti Co V Cu Nb B Electrode 0.022 1.05 0.34 0.028 0.018 0.08 0.04 0.01 0.007 0.002 0.013 0.002 0.065 0.005 0.0005 Base metal 0.26 1.05 0.21 0.022 0.008 0.04 0.03 0.01 0.032 0.001 0.004 0.0002 0.012 0.002 0.0003

ous methods of improving productivity in cessful welds is also more restricted. For present approach involves the design of both electroslag and electrogas welding. example, a 12- to 15-mm (0.5- to 0.6-in.) a special cored strip electrode which The experiments carried out included narrow groove may require machined adds metal powder to the weld pool in a varying the weld root opening and the plate edges in thicknesses 100 mm (4 in.) more efficient way, eliminating the need metal powder addition. Figure 3 shows and greater. Furthermore, this raises the of a special inert gas metal powder deliv­ the window for acceptable welds as a question as to whether the average fabri­ ery system. It incorporates the ideas of function of metal powder ratio and heat cator of heavy plates is equipped to reduced root opening, electrode geome­ input for root openings of 51 mm (2.0 control the parameters within the narrow try, chemical composition modification, in.). ranges required for such improvements and high electrode melt rate. to be obtained. It is the intent of this paper to report Furthermore, the distortion of the thin Filler Wire Addition and discuss the performance and feasibil­ strip guides, even though restrained by ity of the metal powder cored strip Rather than metal powder additions to fusible flux blocks and the thin strip elec­ electrode as an electroslag welding con­ absorb the excess thermal energy in the trode, raises questions as to the feasibility sumable and how it affects the metallur­ electroslag bath, Yakushin, ef al. (Ref. 22), of the process in long joints under pro­ gical, chemical and mechanical properties provided a separate wire fed into the duction conditions. The more conven­ of the weldments produced. slag. This resembles the feeding of a cold tional groove width of the metal powder approach would seem to be more feasi­ wire into plasma arc or gas tungsten arc Experimental Procedure welding processes. The authors demon­ ble. strated that not only was the productivity Processes which feed loose metal ASTM A588 grade structural steel was of the process increased, but like the powders into a welding zone have pre­ chosen for the consumable guide weld­ powder fed systems, the HAZ width was sented production control problems, ing experiments. Plates of 305 X reduced by increasing the welding speed. often militating against their use in pro­ 305 X 38 mm (12 X 12 X V/i in.) were The authors suggested that filler wires duction. But improvements in powder used. A hollow low-carbon manganese could also be used to sense the depth of flow systems offer hope that the steel "strip" electrode, which allowed for the metal pool, thus assisting in the con­ approach of Eichhorn, etal. (Refs. 18-20), chemical composition change by adjust­ trol of the weld pool geometry and may be found useful for some heavy ing the alloy additions to the core of the reducing the hot cracking tendency. plate fabrications. Control of the unifor­ electrode, was devised. In this work, the Little, if any, production experience has mity of the alloy, particularly in high-alloy cavity of the electrode was filled with been obtained on the developments systems, presents the same problems ferro-silicon and ferro-manganese pow­ reported in the preceding paragraphs. faced by the producers of metal cored der at 50% compaction and a The results reported by the various wires for arc welding and by the users of 3 : 1 metal powder-to-electrode weight authors point to a potential for increased loose unalloyed powders in thermal ratio. The chemical composition of the productivity and improved mechanical spraying applications. The requirement base metal and electrode used in this properties in the as-welded condition. that the powders be ferromagnetic is a work are given in Table 1. Both the narrow groove approach (Refs. further drawback for highly alloyed pow­ With cross-sectional dimensions of 12,13) and the addition of metal powders der additions. 35.4 X 4.5 mm (1.4 X 0.18 in.), the metal to the slag bath appear to yield a consid­ The objective of this research is to powder cored strip electrode was made erably lower specific heat input with develop a high deposition rate welding to carry high current and achieve high improved mechanical properties. On the technique that will also refine the HAZ melt rate. And as shown in Figure 4, the other hand, the operating range for suc­ and weld metal grain structure. The surface of the electrode is corrugated for

I- ACCEPTABLE ACCEPTABLE < < or WELDS CC WELDS or"? u*i £O UJ o o a * o UNACCEPTABLE O O o UNACCEPTABLE a. a. WELDS WELDS

134-s [ APRIL 1989 ELECTRODE

.GUIDE

<

Fig. 4—Photograph of metal powder cored strip electrode illustrating Fig. 5 — Electrode-to-joint configuration for the welding experiments corrugated surface for feeding and retention of powder with the metal powder cored strip electrodes the feeding mechanism and also the of the weldments. Within the 800 to 1500-A range, the retention of powder in case the end of Finally, some preliminary work on grain feed rate of the metal powder cored strip the electrode is cut open. The width of refinement by electromagnetic stirring of electrode varied from 5 to 20 mm/s (0.2 the electrode covers almost the entire electroslag weld metal was also per­ to 0.8 in./s), which is much slower than thickness of the base plate promoting formed. A magnetic field was coupled to that of conventional solid electrodes, typ­ more uniform heating of the weld pool. the weld to promote magnetic stirring of ically 70 to 200 mm/s (2.8 to 7.9 in./s) In addition, a pair of low-carbon-steel the weld pool. The two weldments (Ref. 4). Of course, the cross-sectional consumable guide plates were also used. obtained were examined using metallo­ area of the strip electrode is much larger The root opening was held constant at graphic techniques. than that of the conventional solid elec­ 38.1 mm (Vh in.) for all welds, with the trode. For the same length of both elec­ opening at top 3.2 mm {Vs in.) greater trodes, strip electrodes can carry much than the opening at the bottom. Figure 5 Results and Discussion more current than the conventional solid shows the electrode-to-joint configura­ ones. Several investigators (Refs. 2, 23- tion in this process. Moreover, because Process Variables and Welding Behavior 26) reported a linear relation between of the specific electrode-to-joint geome­ It was found that voltage has strong current and electrode feed rate, while try, no lateral oscillation of the electrode effects on both process stability and base others (Refs. 27, 28) found a nonlinear was needed. metal penetration. Voltage should be behavior. From the data obtained in this For the determination of temperature maintained between 36 to 46 V to study, the relation between current (I) distribution and heat extraction condi­ produce stable operation of the process. and electrode feed rate (W) was: tions in the HAZ, thermocouples were A voltage below 36 caused short circuit­ I = A W035 (1) inserted into drilled holes in the plates ing between electrode and molten weld where A is a proportionality constant. and monitored throughout the heating metal with frequent arcing. Excessively Figure 6 is a graphical representation of and cooling cycles. Temperature profiles high voltage also resulted in severe spat­ Equation 1 with the experimental data. of welds were obtained and compared ter from flux bath and arcing. However, The coefficient of correlation, r2, is 0.942 to data available in the literature. stable operation within 36- to 46-V range and indicates a good fit. The welds were produced at a con­ did not guarantee fully penetrated welds. The average weld deposition rate was stant potential of 40 V with the welding A minimum of 40 V was required to approximately two to three times higher current varying from 800 to 1500 A. The achieve full penetration. than the solid wire process, reaching details of welding parameters selection have been presented previously (Refs. 3, 4, 7) and will not be discussed in this Table 2--Typical Chemical Composition of a Running Flux for Electroslag Welding (wt-%) paper. Commercial starting and running fluxes were used to carry out the weld­ CaO MgO MnO CaF Si0 AI2O3 Ti0 K20 Na 0 FeO Bl ing, and the flux bath was maintained at a 2 2 2 2 depth of 35-51 mm (T/2-2 in.). The 12.2 2.3 22.5 8 b 33.0 8.3 8.0 0.9 0.6 1.8 0.9 chemical composition and basicity index (Bl) of the running flux is shown in Ta­ ble 2. Extensive metallographic analyses Table 3—Chemical Analyses of Some Representative Weldments were performed to study the macro and microstructure of the weldments. Chemi­ Weld C Mn Si P S Cr Ni Mo Al Ti Co V Cu Nb B cal analyses of the welds were carried 10CX) A 0.15 0.89 0.19 0.016 0.012 0.05 0.04 0.01 0.011 0.001 0.004 0.001 0.041 0.002 0.0003 out using an emission spectrometer, and 1200 A 0.11 0.87 0.17 0.014 0.015 0.07 0.04 0.01 0.022 0.002 0.006 0.001 0.051 0.003 0.0003 the composition given in Table 3. Stan­ 1300 A 0.10 0.79 0.16 0.010 0.014 0.07 0.04 0.01 0.015 0.001 0.005 0.0005 0.048 0.003 0.0003 dard ASTM procedures were followed to 1400 A 0.11 0.90 0.18 0.012 0.014 0.07 0.03 0.01 0.012 0.001 0.005 0.001 0.054 0.002 0.0003 1500 A 0.10 0.88 0.18 0.011 0.013 0.06 0.04 0.01 0.014 0.001 0.005 0.001 0.048 0.002 0.0003 determine the Charpy impact toughness

WELDING RESEARCH SUPPLEMENT 1135-s also that this equation took into consider­ ation the electrode to joint geometry, which is related to the electrical and thermal transport in the process. Using the same approach, relationships between heat input, current, potential, wire feed rate, the geometry of the weld and consumable configuration are estab­ lished for electroslag welding with metal powder cored strip electrodes. The con­ stitutive equation has the following form: Fig. 6 — Welding I = B' W035V103 (4) current as a function 4 6 8 10 12 14 of electrode feed ? B' is the proportionality constant. The 2 rate ELECTRODE FEED RATE (cm/s) coefficient of correlation, r , is 0.8813, indicating a fairly good statistical fit for the power function. Figure 7 plots the above 38 kg (84 Ib) per hour. Specific mal transport at the electrode/slag inter­ current as a function of wire feed and heat input of the welds produced was face. Using an oscillating consumable voltage. found to be lower with respect to the guide in electroslag welding of 102-mm The specific energy input of the metal 2 other electroslag welds, 0.8 kj/mm (520 (4-in.) 2'/4Cr-1Mo steel plate, Frost, powder cored strip electrode process 2 k)/in. ). Edwards and Rheinlander (Ref. 29) was also found to present a different derived functional relations between the relationship than that shown by Equation various process variables for the selection 2, and that E, as shown in Equation 5, was Constitutive Equations for Welding Current, of a voltage and current range within found to be proportional to V4/l2 rather Voltage and Travel Rate which the energy input is just sufficient to than WVI as in Equation 3. achieve penetration of the parent plate One of the major objectives for all 4 but without overheating the HAZ. A con­ (L g ~ 2 wgtg) V welding process studies is to control the E = (5) stitutive equation of the following form 2 power input per unit length of weld since [L + (g - 2 tg)] (We te) I was determined: this is the primary variable to influence That is, the process with metal powder the degree of melting of the base materi­ /2 ,/3 I = B W' V (2) cored strip electrode was more sensitive al, the amount of dilution of the weld to current than the conventional circular metal, and the thermal history of the where V is the welding voltage and B is a wire process, and energy input can be HAZ. In electroslag welding, current and proportionality constant. The total reduced significantly by increasing the travel speed are not independent vari­ energy input per unit area of weld sur­ welding current. In this equation, wg and ables as in other arc welding processes. face (E) was aiso determined: tg are the width and thickness of the Holding constant the welding potential, consumable guide plates used, and we Lg V the current and power are both functions E = B- - (3) and t are the width and thickness of the • H (L + g) I e of the resistance of the slag pool. Hence, strip electrode, respectively. The differ­ they are functions of the electrode feed In this equation, L is the thickness of the ent degrees of specific energy input rate, the mechanics of electrode melting, base plate; g is the root opening and r is dependency on the welding current of and the nature of the electrical and ther- the radius of the welding electrode. Note the two processes are due to the differ­ ent electrode geometry, current path, current density and thermal transport Fig. 7— Welding conditions. In the case of metal powder current as a function cored strip electrode welding, the area of electrode feed „ 2 ratio, A /A , in Equation 3 (Lg/27rr ) is rate and voltage g w < substantially greater than that shown in •/ Equation 5 (wgtg/wete), hence the consid­ 1500 — erable difference in the energy depen­ h- dence on the voltage/current ratio not­ 2 UJ ed. Figure 8 is a plot of specific heat input • as a function of welding current. rr '• rr Unlike arc welding processes where an Z 1000 increase in current results in greater energy input and deeper penetration, the reverse occurs in electroslag welding. An (D increase in current requires an increase in Z electrode burnoff rate. This increases the Q welding speed and decreases the energy u 500 input per unit area of weld. An excessive $ increase in the current at constant volt­ age can shift the process from penetra­ tion to nonpenetration of the base plate. Based on experimental welding results 0 / I I I I I I I I and theoretical heat flow calculations, the 20 40 60 80 process control boundaries for successful electroslag welding can be developed p0.35 .03 V (Ref. 29). It took into consideration the

136-s | APRIL 1989 o 80 > - t :- D

RO - \ 2 l \ c^ >>~— ; c O \ a. 40 A-4 J ^ . \ METAL ~ A^>r-^ NA POWDER a _WIRSOLIDE / *r -- CORED j (REF. 29) STRIP LJ ELECTRODE 3 20

0 1 | 1 900 JIOO 1300 1500 500 1000 1500 CURRENT (A) WELDING CURRENT (A) Fig. 8-Variation of specific heat input as a function of welding Fig. 9 — Operating parameter windows established for electroslag weld­ current ing with conventional solid wire and metal powder cored strip electrode to obtain fully penetrated welds power contributions required to heat and cific heat input of the process. However, Solidification Structure and Refinement melt the electrode, guide plates, and base the much narrower operating space metal surfaces, and the various heat observed indicates that this process may In a more recent publication, Yu, ef al. losses, such as flux pool radiation, weld require more critical process and param­ (Ref. 30), characterized the solidification puddle superheating and copper shoe eters control. structure of low-carbon structural steel heating. Figure 9 illustrates the boundaries From the heat balance calculations, the electroslag weldments and related them determined for electroslag welding with role played by the various factors in the with the welding current and weld pool metal powder cored strip electrodes. The consumption and transport of energy form factor (pool width/pool depth). boundaries for solid wire electroslag was also determined. Table 4 shows the They observed both cellular and colum­ welding (Ref. 29) are also included for effect of each of the heat terms for nar dendrites, named previously as comparison. Boundaries A through D several welds prepared at different cur­ coarse and fine columnar grains by Paton enclosed an operating space exhibiting rent levels. As the welding current (Ref. 24). The coarse columnar grains good process stability and base metal increased, the percent of energy dissipat­ protrude from the fusion line toward the fusion. Outside the indicated boundary, ed in the form of heat loss to base metal center of the welds. At some distance the process often appeared to operate decreased, indicating the possibility of from the fusion line, fine columnar grains well but with inadequate base metal lowering the specific heat input and could be observed. The location of cellu­ penetration. Boundary A represents the improving the HAZ properties of the lar to columnar dendrite transition and voltage threshold for parent plate fusion welds. the angle of inclination that the grains at low power inputs. Boundary B repre­ sents the constitutive equation for ade­ quate penetration at high power levels. Table 4—Energy Balance of Electroslag Welding with Metal Powder Cored Strip Electrodes Boundary C represents the maximum power output of the welding power Energy Dissipated (%) supply and Boundary D represents the Major Power limit of electrode feed rate at which the Contributions 1300 A 1100 A 900 A 700 A wire electrode melts by ohmic heating. Heat input to the base metal 32.9 35.7 38.5 41.2 Compared to the conventional electro­ Resistive heat loss to guide plates 27.8 21.7 15.6 10.1 slag process, a higher current level is Heat input to melt electrode 22.1 23.2 25.8 27.6 needed in order to obtain fully pene­ Heat loss to copper shoes 14.0 15.2 16.4 17.5 trated welds. This is not of concern since Heat loss by flux pool radiation I.9 2.1 2.3 2.4 higher current will also decrease the spe- Heat loss by weld pool superheating I.I I.I 1.2 1.3

WELDING RESEARCH SUPPLEMENT 1137-s 100 - • COARSE COLUMNAR ZONE o FINE COLUMNAR ZONE A EQUIAXED GRAINS ZONE 80

TYPE I TYPE III TYPE IV TYPE II U K 60- w

40 -

INCREASING WELDING CURRENT, SPEED 20 - INCREASING WELDING POTENTIAL, HEAT INPUT

INCREASING FORM FACTOR

_L _L 1000 1100 1200 1300 1400 1500 I Fig. 10 — Variation of solidification mode in carbon steel electroslag 600 weldments as a function of welding current, voltage and travel speed WELDING CURRENT (A) (Ref. 30) Fig. 11— Variation of solidification structure with welding current meet at the weld center were some of The coarse as-welded structure of lays, Nakano, ef al. (Ref. 32), used mag­ the criteria used in the classification. Fig­ electroslag weldments is a major concern netic field to promote fluid flow and drive ure 10 shows the variation of solidifica­ to the welding engineers and several the molten slag layer from the hotter tion mode in carbon steel electroslag attempts were made to refine the as- region to the colder region to eliminate weld metal. Low-current welds show solidified structure of these high heat undercutting. In this work, some prelimi­ shallow molten metal pools with a high input welds. Venkataraman, et al., and nary tests were done to investigate the form factor. High-current and fast travel Atteridge, et al. (Refs. 5, 6), investigated effect of electromagnetic refinement of speed welds will generally exhibit deep the effects of mechanical vibration on electroslag weld metal. The weldments pools and low form factors. Figure 11 electroslag weld pools. The observed with magnetic field coupling showed evi­ shows that the present findings are similar movement of the slag surface as a result dence of grain refinement. The length of to that reported in the literature and of plate vibration was insufficient to the columnar grains decreased signifi­ discussed above. As the welding current cause turbulence at the slag-metal pool cantly, while the equiaxed dendritic zone was increased, the volume fraction of interface for grain refinement to occur. increased. This is shown in Fig. 12. The coarse columnar grains decreased and However, refinement of the as-deposited results seem to indicate that the induced the volume fraction of fine columnar grains resulted from the use of an exter­ convective fluid motion in the weld pool grains increased with current. However, nally shrouded electrode guide tube as an disrupted the solidification process and the increase became more gradual with extended stirrer. The form factor of the growth of columnar dendrites. current levels above 1300 A, with the fine weld pool decreased from 3 to 1.5. columnar grains replaced by the thicken­ Multiple orientation and finer columnar ing of the equiaxed grained zone. A grains were observed throughout the HAZ Microstructures minimum of volume fraction of equiaxed entire cross-section of the welds. The The CGHAZ and the fine-grained HAZ grains was observed in the 1200- and authors concluded that because of the (FCHAZ) sizes were measured and com­ 1300-A welds. quartz shielding, a sharp thermal gradient pared with data obtained in the literature. Yu, ef al. (Ref. 30), also observed crack was established in the slag pool generat­ Under similar operating conditions, the development along the centerline of the ing an intense stirring action. The turbu­ present technique with metal powder welds or along the interfaces of the fine lent convection which oscillates the hot cored strip electrodes produced welds dendrites when the welding current was zone in the slag and metal pool led to with reduced HAZ. The thickness of the increased to the 1300/1500-A range. periodic solidification and remelting of heat-affected zone also varied with the However, no cracking of that nature was some dendrite arms. Subsequent renu- current level. At 1500 A, the CGHAZ and observed in this work, even though the cleation resulted in the multiple orienta­ FGHAZ were approximately 5 and 2 mm, welds were made at similar current and tion growth of the columnar grains. East- respectively. This represented a reduc­ voltage levels. In addition, the welds did erling (Ref. 31) also mentioned several tion of almost 50% from the 1000-A not exhibit Type III nor IV solidification techniques for refining welds which weld. Figure 13 shows the HAZ reduction structure. Even at the higher current lev­ include magnetic stirring, ultrasonic cavi­ as a function of welding current. Further­ els and high travel speeds, the solidifica­ tation and the use of chemical inocula- more, grain size reduction in the HAZ tion pattern resembled closely the solidifi­ tors. With ultrasonic vibration at 20 kHz, was also observed. Figures 14 A and B cation Modes I and II. This seems to the grain structure along the centerline of show representative HAZ micrographs. indicate the importance of electrode- an electroslag weld was refined signifi­ At 1000 A, the average CGHAZ grain size to-joint geometry in the electroslag weld­ cantly. was approximately 550 nm. At 1500 A, ing operation, affecting particularly the Even though magnetic field application the average CGHAZ grain size was molten pool formation, heating and cool­ is quite common in the processes of arc reduced to approximately 250 /zm. Figure ing conditions of the weldment. This welding and electroslag refining, the 15 illustrates graphically the relationship effect will be further discussed in a later technique is not generally applied to between CGHAZ grain size and the weld­ section. electroslag welding. For electroslag over­ ing current. When compared to conven-

138-s | APRIL 1989 s **<• ^^S^BEIFBL' :|3 ... ,,. £ HAZ oQ. •mp~... I 0 - > ui Q X 8 - o CGHAZ tr < Ul cn ui

- ¥ » -•" * . *•• •* IM •FGHAZ-A. < a o —I ui \* >

Fig. 12 — Macrophotographs of two electroslag weldments to illustrate X _l_ _1_ the effect of magnetic stirring. A - With superimposed magnetic field; 0 o 900 1000 I [00 I200 I300 I400 1500 B— without superimposed magnetic field. Notice that the length of the

tional electroslag welds, which generally ing is a positive indication of the ability of delineated by the grain boundary ferrite exhibit grains of 600 to 800 nm, the HAZ grain refinement by controlling ther­ network. On the other hand, the lower- S present technique produced welds with mal history of the welding process. current weld showed microstructure Q. much refined grain size in the CGHAZ. comparable to conventional welds with o Grain size variation was observed to be extremely large prior austenite grains. Weld Metal Microstructures more significant at current levels around Figure 17 shows the variation of prior 1200 and 1300 A. austenite grain size with the welding cur­ Figure 16 shows representative weld X It is recognized that an austenite grain metal micrographs of low- and high- rent. Again, a sharp decrease in austenite o grain size was observed in the 1200- to ce size of 250 jum is still quite coarse, partic­ current welds taken at equivalent posi­ < ularly when compared with the newer tions. As expected, the high-current weld 1300-A range welds. Both Paton (Ref. 24) Ul cn fine-grained steels, yet the present find­ exhibited finer prior austenite grain size, and Yu, ef al. (Ref. 30), indicated that Ul tr

600 - Oa. -j >ui 500 Ul a ox 400 tr < ui < CO - " \"' cr Ul IS ui Ul a x a I00 cc Ul J_ _L J_ cn 1000 1100 1200 1300 1400 1500 1600

Fig. 14 —Representative HAZ microstructure of the: A— 1000-A; B- WELDING CURRENT (A) 1500-A, 40 V welds (200 nm) Fig. 15 — Variation of CGHAZ grain size with welding current

WELDING RESEARCH SUPPLEMENT 1139-s A ij&\ *.'

• > - I700 K UJ £ 1500 rOHU 5 - • WJ^PA < — ^KJ*L~S Q I300 TfwVv-^^ |— S*', • \ weld meta < II00 e> • UJ 900 \ H \ z \ w 700 \ tf>l- 3 V "* 500 - \ CE O \ 5 300 Q_ — inn I I I I- I I000 IIOO I200 I300 I400 1500 I600 WELDING CURRENT (A) fig. 76 — Representative weld metal microstructure of the: A — 1000-A; Fig. 17 — Prior austenite grain size in the weld metal as a function of the B 1500-A, 40 V welds welding current coarse columnar grain zones usually con­ transform into fine columnar austenite Heating and Cooling Rate Effects tain more acicular ferrite, while fine grains with grain boundary and sideplate columnar grains zone exhibit higher vol­ ferrite as the predominant structure. This The temperature profiles of the lowest ume fraction of grain boundary ferrite. is also verified in this work. Increasing and highest current welds, shown in Fig. During cooling, cellular dendrites trans­ current led to an increase in volume 18, were chosen to verify the effects of form into coarse columnar austenite fraction of fine columnar grains. Subse­ heating and cooling rate on the HAZ grain grains and result in an acicular ferrite quent transformation of finer austenite size control. The readings represented dominant room-temperature microstruc­ grains resulted in high volume fractions of real-time measurements taken at 1.5 mm ture. Columnar dendrites will generally grain boundary ferrite. (0.06 in.) from the fusion line, after

I400 A I500 A • I000A 200 (THERMOCOUPLE POSITION* 1.5 mm FROM F.L.)

I000

UJ rr 800 - < tr UJ o.

UJ 600 - I-

400

200

I00 200 300 400 500 600 700 800 900 TIME (S) Fig. 18— Temperature profiles of high- and low-current welds indicating the differences in heating and cooling rates of the two welds

140-sl APRIL 1989 steady-state welding regime was 600 reached. The 1500-A weld temperature profile showed a steeper temperature increase with time during the heating cycle with a peak temperature of approx­ 500 imately 1400°C (2552°F). The lower cur­ rent weld experienced a more sluggish temperature increase, and the peak tem­ perature was slightly below 1400°C. Dur­ 400 ing cooling, the higher current welds showed faster cooling rate than the low­ < er current samples. For welds of similar chemical composition, the different ther­ N 300 mal history experienced by the welds will promote the formation of distinct o microstructures. Excessive grain growth in the CGHAZ 200 occurs mainly as a result of the coarsen­ ing and dissolution of nonmetallic parti­ cles such as carbides and nitrides (Ref. Fig. 19 — Variation of 33). The presence of these particles is I 00 CGHAZ grain size as essential to the stability of the structure, 0.5 1.5 2.0 a function of specific since they can interact and pin the grain SPECIFIC HEAT INPUT (kj/mm2) heat input boundaries. In the case of an aluminum killed fine grain steel, the grain coarsening temperature is related to the aluminum rates which limited particle coarsening or to 1300-A range resulted in a significantly nitride dissolution temperature. The alu­ dissolution. This also explains the resulting larger amount of electrode being depos­ minum content of the base metal used in finer CGHAZ grain size observed in the ited into the weld joint and accelerating the present experiment is 0.032 wt-%, higher current welds. However, Ats/s in the welding process (Ref. 34). Conse­ and the grain coarsening temperature the low-current weld was not observed quently, heat input into the molten metal was estimated to be approximately to be much longer (approximately 150 s, pool is spread out to a longer section of 1000°C (1832°F) (Ref. 33). However, as compared with the higher current the weld decreasing the specific heat with heating and cooling rates greater weld reading of 120 s). This is confirmed input. This also leads to a faster heating than those corresponding to the equilibri­ by the observation that the CGHAZ and cooling cycle resulting in effective um conditions, the dissolution of these structure of both welds were composed control of HAZ grain growth. particles in the HAZ will depend on the of proeutectoid ferrite networks along peak temperature, the concentration of the prior austenite grain boundaries, aluminum and nitrogen in the alloy, and Hardness Measurements and Chemical intragranular ferrite and pearlite matrix the time duration that the base metal is Composition Variation with carbide and bainite. subjected to the severe thermal cycles Hardness measurement of the weld during welding. In terms of peak temper­ specimens are shown in Fig. 20. When ature, the low- and high-current welds Specific Heat Input Effect considering the highest and lowest cur­ had experienced quite similar tempera­ Grain refinement in weldments is a rent welds prepared, only a slight differ­ ture readings, 1340° and 1400°C (2444° complex matter and generally cannot be ence in the HAZ hardness readings was and 2552°F), respectively. However, the achieved by changing one single parame­ observed. This is consistent with the pre­ two welds showed distinct heating and ter, even though the faster heating and vious observation that even though the cooling characteristics. The lower current cooling rate of the higher current electro­ HAZ size and the grains were refined in weld had a heating rate of approximately slag weldments were shown to have a the higher current welds, the constituents 10°C/s (18°F/S) and its cooling rate was strong influence on the HAZ grain growth found in the microstructures were gener­ 6°C/s (11°F/s). On the other hand, the restriction. In fact, the temperature pro­ ally the same. Therefore, the increase in high current weld showed approximately file itself is due to a combination of other HAZ hardness with increasing current 15°C/s and 30°C/s (27°F/s and 54°F/s) factors such as welding current, welding was minimum. However, the average in the heating and cooling cycle, respec­ speed, geometry of the strip electrode, hardness of the weld metal with the tively. This observation is particularly thermal properties of the materials used, refined solidification structure was ap­ important in the interpretation of the and specific heat input. Specific heat proximately 10% lower than that of the grain refinement event in higher current input of the higher current welds pro­ coarse structure weld. Venkataraman, ef welds. The higher heating rate does not duced in this work was found to be al. (Ref. 5), reported similar findings that allow for the equilibrium heating and approximately 0.8 kj/mm2 (520 kj/in.2) the hardness of the quartz shroud refined dissolution of the aluminum nitride parti­ and lower than that of an average con­ weld metal was lower than that of the cles, even up to a very high temperature. ventional electroslag weld. The effect of standard electroslag weld. The decrease During the entire thermal cycle, particle specific heat input and welding speed on in hardness is related to the microstruc­ pinning of the austenite grain boundaries CGHAZ grain size is illustrated in Fig. 19. ture discussed previously. Since higher was never ceased. In addition, the time The sharp decrease in grain size at specif­ current welds showed smaller austenite that the CGHAZ experienced tempera­ ic heat input of approximately 1.3 kj/ grains with large amounts of grain bound­ tures above 1000°C for the lower and mm2 follows the trend described previ­ ary ferrite, the average hardness was higher current welds were 90 and 50 s, ously, for it corresponds to the 1200- to observed to be lower. A weld with respectively. Therefore, one can con­ 1300-A range. Due to the nonlinear microstructure of coarser austenite grains clude that the higher current weld expe­ behavior of electrode feed rate with and higher volume fraction of acicular rienced a shorter thermal cycle above increasing current, a faster increase in ferrite results in higher hardness. 1000°C with faster heating and cooling deposition rate with current at the 1200- An interesting feature observed, how-

WELDING RESEARCH SUPPLEMENT 1141-s • 1500 A 95 O 1300 A A 1400 A A 1000 A

90

85

33 o o

75

70-

_L _L 0.5 1.0 1.5 2.0 DISTANCE ( INCH) Fig. 20 - Hardness profiles of electroslag weldments prepared with the Fig. 21 — Variation of relative chemical element loss in the weld pool metal powder cored strip electrode with welding current ever, is the variation of weld metal hard­ larger loss of manganese, silicon, and ing chromium, manganese, silicon, ness with the welding current. A mini­ phosphorus occurred. A lower carbon molybdenum, oxygen, and aluminum mum in hardness was observed in the pickup was also observed within the increased with increasing current density. specimens at welding current around same weld. Lower chemical hardenability The concentration changes observed 1200 to 1300 A. This shows that the contributed to the lower weld metal were directly proportional to the welding decrease in weld metal hardness is not hardness. The loss observed at this partic­ current and to the current efficiency for due to the grain size variation alone. ular current range is very interesting since the reactions, and were inversely propor­ Process characteristics of the metal pow­ it follows the same trend as observed in tional to the melt rate. As compared to der cored strip electrode technique may CGHAZ grain size and weld metal prior to conventional electroslag welding prac­ have affected the slag/metal interaction, austenite grain size decrease with weld­ tice, this difference arises from the signif­ particularly within the 1200- to 1300-A ing current. This indicates that the icant difference in the ratio of the elec­ current range. Therefore, relative chemi­ observed behavior is originated from the trode cross-sectional area to that of the cal element loss was determined for each thermal history of the process, which weld groove and the welding current. of the major alloying elements. includes the heating and cooling cycle, The area ratio, Ag/Aw, in Equation 3 2 Relative chemical element loss of an temperature distribution and molten met­ (Lg/2irr ) is substantially greater than that element (AX) is expressed as the differ­ al pool chemistry. Element recovery in shown in Equation 5 (wgtg/wete), hence, ence between the calculated (Xth) and the weld pool will also depend on current the considerable difference in the energy measured composition (Xmeas) of the ele­ efficiency. Frost, Olson and Edwards (Ref. dependence on the voltage/current ratio ment X in the weld. It was observed in 35) concluded that the current efficien­ noted in an earlier section. The balance Fig. 21 that at 1200- to 1300-A range, a cies for electrochemical reactions involv- between increasing current and melt rate determines the alloying element recovery in this process. Fig. 22 —Charpy impact toughness as Mechanical Properties: Charpy Impact a function of testing Toughness temperature of the metal powder cored A set of 5 welds covering the 1000- to strip electrode welds 1500-A range was chosen for the Charpy impact toughness testing and the results are summarized in Fig. 22. The effects of grain refinement in the CGHAZ is clearly shown. The upper shelf energy of the 1500-A weld is significantly higher than that shown by the 1000-A one. The lower shelf energy of the two welds, however, are quite similar. This seems to indicate that at high testing temperatures, the fine precipitate population and distri­ bution of the higher current welds were -40 -20 0 20 40 60 80 I00 not affected very much by the faster TESTING TEMPERATURE (°C) heating and cooling rate. More energy

142-s | APRIL 1989 was required for a crack to propagate O - 0*C (32°F) through the specimen, moving from pre­ • --20*C (0°F) cipitate to precipitate. On the other A--30*C (-20°F) hand, the coarsening of the precipitates and the decrease in number of precipi­ tates in the lower current weld, because of longer thermal cycles, will lower the energy for crack propagation. The duc­ tile-to-brittle transition temperature (DBTT) of the welds was also observ­ ed to decrease with increasing welding •x Fig. 23 — Comparison current. The 1500-A weld exhibited m of the Charpy the lowest DBTT of approximately impact toughness of — the metal powder -40°C (-40°F) with an energy of 100 ) — cored strip electrode (70 ft-lb). MIN. SPECIFICATION FOR CRITICAL APPLICATIONS-'' welds with the Finally, when the Charpy V-notch minimum toughness energy absorbed at 0°C (32°F), -20°C specified for (-4°F), and -30°C (-22°F) of the metal I000 I200 I400 IS00 I800 engineering powder cored strip electrodes welds are WELDING CURRENT (A) applications compared with the generally accepted specification of 25 ft-lb at 0°C for weld resulting in distinct carbon, manganese, Gatlinburg, Tenn., pp. 401-411. acceptance, as shown in Fig. 23, it can be silicon, and phosphorus transfer behavior 8. Malim, T. H. 1967. Electroslag "instant" seen that the welds generated in this within the range of 1200 to 1300 A. steelmaking. Iron Age, August. work performed satisfactorily and would 9. Dorschu, K. E. 1973. Production of spe­ 9) Charpy impact toughness data be qualified for many of the so called cial-steel billets by continuous electroslag pow­ showed that it is possible to obtain welds critical applications. der melting. Conf. Proc. Electroslag Refining, with acceptable properties by this pro­ Iron and Steel Institute, Sheffield, England. cess. 10. Descamps, J., and Etienne, M. Produc­ Conclusions tion of large ingots by continuous electroslag powder melting, ibid. A ckno wledgments 1) Electroslag welding with the metal 11. Levaux, j., and Sunnen, ). 1972. Contin­ powder cored strip electrode and con­ The authors gratefully acknowledge uous electroslag powder melting for the pro­ sumable guide plates demonstrated sta­ the research support of the National duction of large ingots. Iron and Steel Engineer, ble operation with increased travel speed Science Foundation under Grant DMC March. and decreased specific heat input, indi­ 8505204, and the equipment and materi­ 12. Avramenko, V. I., Lebedev, B. F., and cating the strong potential of the process Bozhko, V. I. 1973. Some ways of increasing al support of The Hobart Brothers Co. to become an alternative for high-speed the productivity of electroslag welding. Svar. and Alloy Rods, Corp. In addition, the welding. Proiz., No. 10, p. 16. discussion and suggestions of Mr. S. Fer­ 13. Watanabe, K., Sejima, I., Kokura, S., 2) Expressions were developed to ree of Alloy Rods and Mr. j. Hannahs of Taki, G., and Miyake, H. 1975. Problems and relate welding current, voltage, electrode Tri-Mark are much appreciated. improvement of large heat input electroslag feed rate, travel speed, weld and con­ welding. Journal of Japan Welding Society, sumable geometry, and specific heat p. 519. input of the electroslag welding process References 14. Campbell, H. C, and Johnson, W. C. using a metal powder cored strip elec­ 1967. Granular metal filler metals for arc weld­ trode. 1. Campbell, H. 1970. Electroslag, electro­ ing. Welding lournal 46(3):200-206. 3) Process boundaries for obtaining gas, and related welding processes. WRC 15. Reynolds, G. H., and Kachelmeier, E. J. sound welds with successful penetration Bulletin 154. 1978. Adding powdered metal filler speeds 2. Liby, A. L, and Olson, D. L. 1974. Metal­ deposition rate. Metal Construction, p. 426. for the metal powder cored electrode lurgical aspects of electroslag welding —a 16. Thomas, R. D., Jr. 1966. Corrosion resis­ was determined. review. The Quarterly of the Colorado School tant weld overlays by the dual strip process. 4) Even at high current levels and of Mines Vol. 69, No. 1. British Welding Journal, May. welding speed, Types I and II solidification 3. Liu, S., and Su, C. T. 1986. Innovations in 17. Smirnov, S. A., and Efimenko, L. A. structures containing essentially cellular electroslag welding. NSF Conf. Proc. on Manu­ 1973. Special structural features and mechani­ and columnar dendrites were observed, facturing Processes, Systems and Machines, cal properties of electroslag welded joints with no signs of hot or cold cracking. pp. 29-35. made using powdered filler metal. Svt Svarka, 5) With the metal powder cored strip 4. Su, C. T. 1987. High Speed Electroslag No. 9, p. 46. electrode, both CGHAZ and FGHAZ size Welding. M.S. Thesis, Penn State University, 18. Eichhorn, F., and Remmel, J. 1983. Situ­ State College, Pa. ation of research in electroslag welding —a were reduced with increasing welding 5. Venkataraman, S., Devletian, j. H., tendency of further development. Indian current. Wood, W. E., and Atteridge, D. G. 1982. Grain Welding Journal, No. 4, p. 37. 6) Substantial grain refinement in the refinement dependence on solidification and 19. Eichhorn, F., Remmel, ]., and Wubbels, CGHAZ was achieved with the metal solid-state reactions in electroslag welds. Grain B. 1984. High-speed electroslag welding. powder cored strip electrode. Average Refinement in Castings and Welds. AIME, St. Welding Journal 62(1):37. grain size was reduced from above 600 Louis, Mo., pp. 275-288, 20. Eichhorn, F., and Remmel, j. 1985. Effi­ to approximately 250 /jm. 6. Atteridge, D., Venkataraman, S., and cient fillet welding in the vertical welding 7) The faster heating and cooling rates Wood, W. E. 1982. Improving the Reliability position with electrogas and electroslag weld­ of the metal powder cored strip elec­ and Integrity of Consumable Guide Electroslag ing methods. IIW Doc. Xll-908-85. Weldments in Bridge Structures. Research 21. Reynolds, G. H. 1980. Property and trode process provided adequate condi­ report, Oregon Graduate Center, Beaverton, Productivity Improvements in Electroslag and tions for grain growth control in the Or. Electrogas Welding. Final report, Material Sci­ CGHAZ. 7. Liu, S., and Su, C. T. 1986. Performance ences Northwest, Inc. January. 8) The metal powder cored strip elec­ evaluation of a metal powder cored strip 22. Yakushin, B. F., Basheve, L. F., Tikhonov, trode affected the heat distribution in the electrode in high-speed electroslag welding. V. P., and Zaletov. 1982. Improving the capac­ molten weld pool as well as the chemistry Advances in Welding Science and Technology, ity of electroslag welded joints for resisting hot

WELDING RESEARCH SUPPLEMENT 1143-s cracking. Translation from Russian by Auto­ Electroslag welding of pressure vessel steels. with ESW. Kawasaki Steel Technical Report, matic Welding, October Welding Research Proc. Electroslag Welding for Marine Applica­ No. 2, pp. 31-42. Abroad, June/July 1984. tions, U.S. Naval Academy, Paper 2, March. 33. Gladman, T. 1963. The effect of alumi­ 23. Electroslag and electrogas welding. 29. Frost, R. H., Edwards, G. R., and Rhein- num nitride on the grain coarsening behavior AWS Welding Handbook. Vol. 2, Chapter 7, Iander, M. D. 1981. A constitutive equation for of austenite. Iron and Steel Institute Special pp. 226-260, 1978. the critical energy input during electroslag Report 81: Metallurgical Developments in Car­ 24. Paton, B. 1962. Electroslag Welding. welding. Welding Journal 60(1):12-62. bon Steels, pp. 68-70. American Welding Society, Miami, Fl. 30. Yu, D., Ann, H., Devletian, )., and 34. Liu, S., and Su, C. T. 1987. Grain refine­ 25. Thomas, R. D., jr. 1960. Electroslag Wood, W. 1985. Solidification study of nar­ ment in the HAZ of high-speed electroslag welding. Welding Journal. 39(2): 111. row-gap electroslag welding. Welding Re­ welding. Welding Metallurgy of Structural 26. Solari, M., and Biloni, H. 1977. The search: The State of the Art, ed. E. Nippes and Steels, AIME/TMS, Denver, Colorado, pp. effect of wire feed speed on the structure in D. Ball, ASM, Toronto, Canada, pp. 21-32. 439-453. electroslag welding of low carbon steel. Weld­ 31. Easterling, K. 1983. Introduction to the 35. Frost, R. H., Olson, D. L, and Edwards, ing Journal 56(9):274-s. Physical Metallurgy of Welding. Butterworths. G.R. 1983. The influence on electrochemical 27. Cary, H. 1970. Porta-slag Welding. 32. Nakano, S., Nishiyama, N., Hiro, T., and reactions on the chemistry of the electroslag Hobart Brothers Co. Tsuboi, J. 1981. MAGLAY process - electro­ welding process. Modeling of Casting and 28. Frost, R., Jones, )., and Olson, D. 1985. magnetic controlled overlay welding process Welding Processes ll, AIME.

WRC Bulletin 338 November 1988 Interpretive Report on Electroslag, Electrogas and Related Welding Processes By R. D. Thomas, Jr., and S. Liu

These processes are characterized with emphasis on fundamentals of heat flow conditions, metal transfer, weld pool morphology and the chemical and electrochemical aspects of the slag and weld pool reactions. A total of 146 references are included in this report. Publication of this report was sponsored by the Interpretive Reports Committee of the Welding Research Council. The price of WRC Bulletin 338 is $20.00 per copy, plus $5.00 for postage and handling. Orders should be sent with payment to the Welding Research Council, 345 E. 47th St., Suite 1301, New York, NY 10017.

WRC Bulletin 339 December 1988 Development of Tightness Test Procedures for Gaskets in Elevated Temperature Service By A. Bazergui and L. Marchand

In this report, different elevated temperature gasket tightness test procedures are compared. A two-tier test approach, involving aging of the preloaded gasket in a kiln followed by a short duration tightness test was evaluated. The procedures were evaluated using spiral-wound gaskets with two different fillers: a mica-graphite filler and an asbestos filler. Publication of this report was sponsored by the Subcommittee on Bolted Flanged Connections of the Pressure Vessel Research Committee of the Welding Research Council. The price of WRC Bulletin 339 is $16.00 per copy, plus $5.00 for postage and handling. Orders should be sent with payment to the Welding Research Council, 345 E. 47th St., Suite 1301, New York, NY 10017.

WRC Bulletin 340 January 1989 Interpretive Report on the Mechanical Properties of Brazed Joints By M. M. Schwartz

This report summarizes the mechanical properties, brazing procedures and testing of materials, such as aluminum, beryllium, copper, steel, nickel, superalloys, reactive metals, ceramics and graphite. Over 120 references are included in this interpretive report. Publication of this report was sponsored by the Interpretive Reports Committee of the Welding Research Council. The price of WRC Bulletin 340 is $24.00 per copy, plus $5.00 for postage and handling. Orders should be sent with payment to the Welding Research Council, 345 E. 47th St., Suite 1301, New York, NY 10017.

144-s | APRIL 1989