WO1981000531A1

WO1981000531A1 - Improvements in and relating to electroslag welding

Info

Publication number: WO1981000531A1
Application number: PCT/AU1980/000048
Authority: WO
Inventors: I French; G Kimpton
Original assignee: Commw Scient Ind Res Org; I French; G Kimpton
Priority date: 1979-08-24
Filing date: 1980-08-22
Publication date: 1981-03-05
Also published as: EP0034598A1

Abstract

A process for electroslag welding permitting utilization of a gap width of from 3 to 8 mm between members (16, 18) to be welded, in which there is used a sheet electrode (28) having a width of the order of the width of the members (16, 18) at the gap; the electrode (28) having a central layer (33) of metal and, on each of the major surfaces of the metal layer (33), a layer (34) of slag forming flux.

Description

IMPROVEMENTS IN AND RELATING TO ELECTROSLAG WELDING

This invention relates to improvements in or for narrow gap electroslag welding.

The normal electroslag welding process is used for welding relatively thick sections of metal separated by a gap with the gap being unwardly disposed during welding. The process is generally applicable for joining sections, of metals such as steel, cast iron, aluminium alloys and titanium alloys, having a thickness of greater than about 25mm. The gap or joint opening between the metal sections to be welded is convention- O ally about 25mm to 35mm. The metal sections are positioned with their edges in spaced substantially parallel relationship so as to form the gap or joint opening. A consumable metal rod or wire electrode is connected to an electrical power source, and is supported in and guided by a guide tube and fed continuously into a molten pool of slag in the joint opening until the joint is filled. It is a well-accepted practice to separately introduce flux suitable for forming the slag into the joint opening as the welding process proceeds . The conduction of the current through the molten slag causes resistance heating which melts the metal electrodes . Whilst there is no arc involved during the actual welding of the sections to be joined, it is conventional to initiate the welding process by striking an arc between the bottom end of the metal electrode and a metal contact such as steel wool, placed at the bottom of the gap.

The electroslag welding process is considered to have a number of advantages over conventional metal arc welding processes for joining thick metal sections or workpieces, these being as follows: 1) a high weld^' metal deposition rate;

2) the extent of joint preparation is not as critical as with arc welding processes, since a parallel sided gap with flame cut edges will suffice;

3) the heating and cooling cycles are much more gradual than for arc processes and this gives rise to fewer weld defect problems such as weld cracks and porosity;

4) it is a single-pass process and thus electroslag welds are less prone to presence of inclusions than

< » •li^'O

^■ " with welds made by multi-pass arc processes wnere slag inclusions may form between passes. Electroslag welding is recognised as a very effective process for joining thick sections of metals. However the relatively large gap size between the sections of metal to be welded results in certain drawbacks with this process. The gap size is conventionally approximately 25mm to 35mm, and hence large quantities of electrode material are consumed. Also, because this large quantity of metal is melted, the heat input per unit height of weld is high and this leads to degradation of mechanical properties of the weld metal and surrounding zone.

The Applicant has recognised that the electroslag process could be made significantly more efficient if the gap or joint opening between the sections to be joined could be substantially reduced in width. It is considered that a narrower gap or joint opening would give rise to less metal electrode consumption and lower electrical energy consumption for a given joint, with a reduced operating time. Gaps narrower than about 20 to 25mm, to the

Applicant's knowledge, have rarely been used previously for electroslag welding for two basic reasons; firstly, the most commonly used electrodes in known electroslag welding processes comprise wires having a diameter of between 3 and 6 mm. Such wire electrodes are passed through tubular guides which may be either consumable or non-consumable and which, on account of their size, cannot be inserted into a gap narrower than about 20mm. Secondly, it is usual procedure to add the electroslag flux in granular or powdered form after the initial arc has been struck to initiate the electroslag welding mode and, to be able to do this, one must have a space of at least 6mm between the electrode (or the outer surface of the guide tube) and the edge defining one side of the gap or joint opening. Whilst reference has been made to wire electrodes, the Applicant is aware of electroslag welds having been made with metal plate rather than wire electrodes . The main advan¬ tages of using plate electrodes are (i) mechanical simplicity compared to the equipment required to support and guide wire electrodes, and (ii) more uniform heating of the molten slag pool. To the Applicant's knowledge however, plate electrodes have been used mainly for joining exceptionally thick metal sections, the plate electrodes having a thickness of around

5 8 to 12mm, and with even larger than usual gap or joint opening sizes.

More recently, the Applicant is aware of a number of attempts which have been made to develop an electroslag welding process which will operate with narrower gap sizes. The

"L_Q narrowest gap so far reported at which successful electroslag welds have been made is about 12mm. To the Applicant's knowledge, this has been achieved using a strip or band electrode having cross-sectional dimensions of 0.8mm (thickness) to 20mm (width) and a consumable strip electrode 5 guide.

It is the main object of the invention to provide an improved electroslag welding process permitting use of a gap, between the workpiece sections to be joined, which is narrower than the gap sizes used heretofore for electroslag welding.

2o According to the invention, there is provided a process for welding of metal members in which the members are positioned so that respective surfaces at which they are to be welded are in opposed, spaced relation to define an upwardly extending gap between the surfaces, the gap having a width

25 between those surfaces that is substantially less than both the width and vertical extent of the surfaces, inserting endwise, downwardly into the gap a consumable sheet electrode having a thickness less than the gap width so that the electrode has its width dimension disposed substantially horizontally and

30 substantially parallel to said surfaces; the electrode having a central layer of metal and, on each of opposed major surfaces of that layer, a layer of slag forming flux; supplying electrical power to the electrode to form a slag pool into which the electrode extends and to initiate an electroslag welding 5 operation; and continuing that operation, while lowering the electrode, to form a weld between the surfaces of the members.

The process normally is applicable to the provision of a weld between surfaces of the members which are substantially planar and which preferably are substantially vertical and. for simplicity of further description, such conditions are assumed. However, it is to be appreciated that the surfaces may be correspondingly or similarly curved, that they may be somewhat irregular in surface finish and tnat the spacing between the surfaces and, hence, the gap width, can vary to a degree. Also, the surfaces can be inclined to the vertical to an extent appreciated in the art.

The width dimension of the electrode preferably is not substantially less than the width of the pieces to be joined. Hoxever, in some instances, it is found that this can result in a region of unsound weld metal at vertical edges of the weld. In one form of the invention which overcomes such unsound metal, the width of the electrode is substantially equal to or greater than that of the pieces to be joined. In such form, at least where the width of the electrode is greater than the width of the pieces to be joined, end pieces used to close the gap at the sides of the pieces during the weld operation are grooved or shaped to accommodate the edges of the electrode. In general, the electrode width need be no more than 20mm greater than that of the pieces to be joined, such that the electrode can readily be accommodated by otherwise conventional end pieces.

The process of the invention can be operated with metallic end pieces or shoes closing the gap along vertical edges of the surfaces to be welded, as in conventional electroslag processes. However, as with conventional processes, such end pieces often are difficult to remove after completion of welding as they fuse to the weldment. Applicant has found that significant practical advantages can be achieved by utilization of aion-metallic end pieces.

In one form of the invention, the gap between surface to be welded is closed along or adjacent vertical edges of the surfaces by end pieces of refractory material. Highly suitable for this purpose are end pieces formed of the oxide of aluminium, silicon or magnesium. However, other materials also can be used and preferred examples of these are flux materials, including conventional electroslag flux, and limestone.

The non-metallic end pieces can be formed in any convenient manner. However, where flux material is used, this can be mixed with an adhesive or bonding agent and either preformed to the required shape and dried or applied across the end of the weld gap. Particularly in the latter case, the end piece most conveniently is backed by a support, such as of metal foil or thin metal sheeting, which holds the flux material in position. Where the flux material is applied as a paste, the support may be deformable to a required configuration corresponding to that of sides of the members being welded. In the case of the above oxides, limestone and also flux, these materials in granular or particulate form can be cast into a required shape. In one suitable method the material can be mixed with sodium silicate, and the mixture cast in a block or mould and hardened by heating in an atmosphere of carbon dioxide. The end piece can be backed by a support, as described above.

For most applications, end pieces of the above- mentioned oxides are preferred. These have a melting point significantly above the temperature reached by the molten slag pool and thus retain their integrity during welding. After completion of welding, they can be removed by tapping as they then simply crumble away. The use of such end pieces also results in a sound weld along the sides of the welded sections and, while a slight bead of weld metal may protrude from the weld, this can be controlled and normally is acceptable.

The flux end pieces tend to melt away during welding and allow relatively large beads of weld metal to protrude from the edges of the joint. However, this in fact can be an advantage in some instances, such as in joining irregular sections. End pieces of limestone behave similarly, but to a lesser extent.

With oxide and limestone end pieces, it is preferred to provide an insert of flux over the portion of the surface of the end piece extending across the weld gap. For this purpose a groove or channel is provided in that surface and is filled with flux. Such provision of flux can be used to produce additional slag to offset loss of slag derived from the flux of the electrode. End pieces as described have the advantage of being easily removed compared with metal end pieces. They also enable a sound weld to be obtained across the full width of surfaces of welded sections without the need for use of an electrode of greater width. Additionally, compared with metal end pieces, the non-metallic end pieces do not signficantly contribute to heat loss during a welding operation.

The width of the gap may be less than 8mm in the process of the invention, for example from 6 to as low as 3mm. For a gap width of 6 to 8mm, an overall electrode thickness of 4 to 6mm is suitable. In such electrode, the metal layer may be from 1 to 4mm thick. However, for narrower gaps, the metal layer may be as thin as 0.3mm.

Satisfactory welds have been produced between surfaces of up to 100mm x 100mm. Larger sections can be welded, although problems can be experienced simply in handling electrodes for providing a weld between surfaces of relatively large width or, in particular, height. However, electrodes may be of a height of up to 500mm or more and corresponding width with, particularly when operating at larger gap widths of 6 to 8mm. The principal difficulty to be overcome with larger electrodes _% is in avoiding buckling and in having too long an electrical stickout.

The overall thickness of the electrodes, i.e. the thickness of the metal layer and flux should be a balance betwee the need to have sufficient flux for slag formation on the one hand and managability of the electrode and avoidance of arcing on the other hand. However, these factors can be substantially offset, particularly when providing a weld between relatively smooth surfaces, by providing an initial quantity of flux in the gap and by a vertically slidable assembly to which the electrode is attached for controlled movement within the gap. Thus, the principal requirement in this regard is for a slight clearance for the electrode sufficient only for its movement into the gap.

It is found that most favourable results are obtained with the invention by use of a relatively low voltage power source operating at a relatively high current. Thus, a voltage of 20 to 30 volts is preferred, compared with 40 to 50 volts

OMPI typically used in prior electroslag welding processes ., Higher voltages tend to cause difficulty in establishing the electroslag mode and unwanted arcing during welding, while lower voltages make the process difficult to start. With the present invention, as applied to welding pieces of less than 70mm in width, currents of 800 to 1000 amps are found to provide optimum results, compared with a current generally of from 400 to 500 A for prior processes. Currents less than 600 A may not reliably provide continuous welds while those between 600 and 800 A can tend simply to fill the gap with molten metal and thus result in welds of minimal penetration. Currents in excess of 1000 A tend increasingly to result in a very large melted zone, particularly near the centre of the gap length. In general, the optimum current required for a weld will be directly proportional to the width of sections to be welded, although it is found that a current density of from 5 to 15, preferably 5 to 10, A per sq. mm of electrode cross-sectional area provides satisfactory welds with a D.C. supply. However, as will be appreciated, the process can be carried out using an A.C. power supply with minor modifications; and current densities similar to those for a D.C. supply are suitable.

For the purpose of further describing the invention, reference now is made to the accompanying Figures, in which: Figures 1 and 2.. show a schematic representation of an assembly operable in accordance with the invention; Figure 3 illustrates voltage and current traces taken during a welding process according to the invention; Figure 4 illustrates a sectioned weld produced by the invention; and

Figure 5 illustrates half-sectional representations of (a) a conventional electroslag weld and (b) a weld produced by the invention.

In Figures 1 and 2, the assembly 10 is shown at an intermediate stage in formation of a vertical weld between opposed surfaces 12,14 of 25mm thick mild steel plates 16,18; the surfaces being 70mm wide and 80mm in height. The assembly arrangement is however applicable to providing a weld between surfaces of a wide range of sections, provided the gap between the surfaces can be oriented so as to be upright, preferably vertical. Surfaces 12,14 define a vertical gap 20 therebetween of 6mm optimum width. Prior to commencement of welding, the length and height of gap 20 corresponded to the width and heigh of the surfaces. The ends of gap 20 are closed during the welding operation by a respective end piece 22, of which only one is shown for simplicity of illustration. Across the bottom of gap 20, a base plate 24 is provided; while run-off tabs 26 extend across plates 16,18 adjacent the top of gap 20. Plate 24 is shown as having a channel 27 extending below gap 20 such that plate 24 provides a function similar to run-on tabs. Assembly 10 also includes a sheet electrode 28. For the welding operation, electrode 28 is attached to a rigid vertical slide (not shown) to prevent lateral movement and, as consumed, electrode 28 is lowered into gap 20 by the slide, either manually or by means of an electric motor drive. Each end piece 22 has a body 29 of non-metallic material, consisting of refractory material such as limestone or an oxide of aluminium, silicon or magnesium. The body is grooved along the portion of its surface to extend across an end of gap 20, with the groove being filled with flux 30. Body 29 is cast into a metal support 31 and, after flux 30 is added, the body and flux is hardened such as in the manner described above. Alternatively, body 29 and flux 30 may be hardened by curing of a suitable resin, such as a thermosetting phenolic or melamine resin, premixed with the material, of body 29 and with flux 30.

Electrode 28 initially had a height of 240mm, and has a width of 70mm and thickness of 4 to 5mm. It has a centra layer 33 of 1.5mm thickness electrode metal, with a layer 34 of flux over each of its major surfaces opposed to a respective one of surfaces 12,14. Flux 34 may be of any suitable composition, such as MF38 (Kobe Steel Company) . Flux 34 may be coated onto layer 33 in any suitable manner. One convenient way is to prepare a thick slurry of granular flux material and a dilute solution of suitable adhesive, spread this to the required thickness on layer 33 and finally dry the slurry τ.υ provide an adherent flux coating.

Above electrode 28 there is a continuation 33a_ of layer 33 which has only a thin film of flux and, at its upper end is connected to a power source via conductor 36. A second conductor 38 is connected to plate 24 for completion of the electrical circuit. The film of flux at 33a is provided to prevent arcing but can be substantially thinner than the flux coating at 34 as the latter can provide the necessary amount of slag for operation in the electroslag mode.

With the components of assembly 10 positioned, power was supplied using, for example, a Lincoln Idealarc source having a capacity of 1500 A direct current operating in its constant voltage mode. A chart recorder (not shown) provided traces of current and voltage during welding operations, and one such record is shown in Figure 3. The welding process was initiated by lowering electrode 28 to contact some steel wool or iron powder placed in the bottom of channel 27 , to produce arcing until enough of flux 34 was melted to form slag 40 for the electroslag welding mode to become established. With reference to Figure 3, arcing began at point X, and electroslag conditions provided molten metal pool 42 and form¬ ation of weld 44 were established at point Y. The most satisfactory operating voltage was found to be about 25V, since higher voltages caused unwanted arcing during welding and lower voltages made the process difficult to start.

The current used could be controlled by varying the speed with which electrode 28 was lowered. It could be increased by lowering the electrode more quickly and decreased by lowering more slowly. The magnitude of the current was found to have a marked effect on the thickness of the melted zone, with the optimum current range for the geometry used being 800 to 1000 A.

A polished and etched section through one of the joints welded using such assembly with a current of about 900 A and a voltage of 25V is shown in Figure 4. The region of non- continuous weld metal at the bottom of the joint is where arcing occurred prior to the electroslag mode, due to run-on tabs not having been used. However, this defect can be eliminated by use

OMPI _ Λ. WIPO „ >/ of such tabs. The wide melted zone present at the top. of Figur 4 would not normally be part of the joint if run-off tabs are used. Also in Figure 4 it will be noted that the weld is thicker at a central zone; although this can be avoided, by control of electrode feed rate, to achieve a weld of substantially uniform thickness.

The heat input for this welding technique was calculated. The rate at which the joint was filled under the above-mentioned optimum conditions of gap width, current and voltage was about 1 nuns . Heat input was therefore

20 to 25 KJ per mm of height for an electrode width of 70mm or approximately 0.3 KJ per mm of height per mm of width. As shown in Figure 5 the gap width in the weld according to the invention is significantly less than with a conventional electroslag weld. This leads to a difference in volume of weld metal by a similar factor. The Figure also shows the resultant difference in the melted (and heat affected) zone in the welded members.

A chemical analysis of the parent plate, electrode and weld metal for one of the trial welds was made and the results are shown in Table I, based on use of a mild steel sheet for the electrode layer 32. That sheet had considerably lower C, Mn and Si content than either the parent plate had or a commercial electrode would be expected to have. The weld metal had evidently picked-up som Mn and Si from the flux since it contained more of these elements than either the parent plate or the electrode.

OMPI TABLE I Chemical compositions - weight percent

Two particular features of such welds produced by the invention offer considerable advantages over conventional electroslag welds. Firstly the gap between components to be joined is smaller by a factor of 4 to 6 than the gaps normally associated with electroslag welds. This difference in gap width is illustrated by Figure 4 which shows half cross-sections through (a) a conventional electroslag weld and (b) a narrow gap weld according to the invention. As shown, the difference in gap width leads to a difference in voluireof weld metal by a similar factor while the extent of -the heat affected zone also is signficantly reduced in the section resulting from the invention. The second major difference is between the welding speeds and therefore the heat inputs of the two techniques. In conventional electroslag welds produced over a range of conditions considered typical of industrial practice, the welding speeds range from 0.19 to 0.38 mms with an average value of 0.28 mms^" . The process of the invention permits welding speeds about 3.5 times higher than this average value. The heat inputs per mm of height per mm of width conventionally range from 1.1 to 2.0 KJ with an average of 1.5 KJ. The heat input for the process of the invention, found to be 0.3 KJ per mm of height per mm of width, was therefore lower by a factor of 5 than the average for conventional welds.

To further illustrate the process of the invention, description now is directed to further specific examples.

"gU ^EAlT OMPI

Λ . WIPO < ^>y Exa ple 1

In this Example the narrow gap electroslag welding process was used to butt-weld two components (in the form of plates) of mild steel each having cross-sectional dimensions of 100mm x 100mm. In this, the pair of mild steel plates were clamped onto a heat sink prior to welding in order to prevent burn through. The plates were arranged in spaced relationship so as to form a gap or joint opening having a width of 6mm and, at adjacent vertical edges of the plates a respective closure member was secured by suitable means so as to enclose the ends of the gap. A bottom closure member also was placed in position to close off the base of the gap. With the gap directly beneath a manually operable vertical slide assembly to which a suitable electrode was attached, the electrode then was vertically lowered into the gap. In this Example, the consumable electrode consisted of a mild steel sheet of thick¬ ness 1mm and width approximately 100mm. The lower end portion of the sheet was coated with a layer of starting flux which, in this instance, was a titania based flux taken from stick electrodes. The strip or sheet electrode also was provided with a coating of an electroslag flux, this being Kobe Steel flux MF38, sufficient to provide molten slag which forms a pool upon commencement of the welding operation. The amount of electroslag flux coated onto the sheet electrode can be less at the upper end thereof as its principal purpose is to act as an insulator to prevent arcing as the strip or sheet electrode is fed into the gap. It was subsequently found that a starting flux layer at the lower end portion of the electrode was not necessary as the electroslag flux itself was sufficient. When the strip or sheet electrode contacted a piece of steel wool placed in the bottom of the gap, arcing was initiated. The electrode was fed into the gap to maintain arcing, so that flux melted and formed a molten slag pool. As is well known by those skilled in tne art, the molten flux extinguishes the arc to establish electroslag welding conditions, and the electrode is consumed by melting at its lower edge. The sheet or strip electrode is progressively fed into the gap at a rate necessary to maintain the current in the

-BUREAT

OMPI ^l ^j? _AT\ correct range.

The Applicant has established that currents of approximately 800 amps or above give satisfactory penetration of the weld metal into the base metal. However, very high currents lead to excessive penetration, and can cause burn through, of the plates or workpieces being welded. Lower currents in the order of 600 amps tended to give inadequate penetration and simply filled the gap with weld metal. It was found that if the current was decreased below around 300 amps, the slag tended to solidify leaving large inclusions in the weld. For the conditions of the present Example, it was found that the optimum current range was between 800 and 1,000 amps.

The optimum on-load voltage was around 25 volts and at this voltage, the welding process was easily controlled and the weld free of large inclusions. Higher on-load voltages namely above 30 volts, have been tried but the welding process proved difficult to control. Example 2 This was carried out in a very similar manner to that described in Example 1 above except that an electrical/ mechanical drive system was used to continuously feed the electrode into the gap or joint opening between the plate sections to be joined. The sheet or strip mild steel electrode, in this Example, had a thickness of .75 mm and thin edge strips of flux were coated on the edges of the electrode to improve insulation and so prevent possible shorting of the electrode to the surfaces being welded. In order to initiate the welding process, steel wool again was placed in the bottom of the gap or joint opening and the electrode lowered, with the electrical welding power supply turned off, until the sheet or strip electrode was about 5 mm from the steel wool. The welding power then was connected to energise the electrode which once again was slowly lowered until arcing starts . The arcing was maintained by lowering the electrode until enough flux had melted for electroslag conditions to prevail. At this point, the rate of lowering of the electrode was adjusted to give a constant current between 800 and 1,000 amps. It was found that if the rate of lowering of the electrode is too high the

IJUREATΓ

OMPI

.s_A WIPO .« current will increase while, if the rate is too low, the curren will decrease.

In other trials conducted by the Applicant, it has been shown that if the electrode is lowered into the gap so tha it is not equidistant from the sides thereof, the amount of weld metal penetration is greater on the side to which the electrode is closer. This feature can be used to advantage in instances where parts of unequal thickness are to be welded.

As with conventional electroslag processes, a region of unsound weld metal is unavoidable at the start of an electroslag weld in the invention because the process starts in an arcing mode. The problem is usually overcome by using a run-on tab, which may be subsequently removed, to start the weld. For conventional electroslag welds these run-on tabs are 50 to 75 mm in height. However, with the narrow gap permitted by the invention, a height of 25mm is adequate and the tabs may be only from 5 to 10mm in height. Similarly a region of unsound weld metal can exist at the finish of electroslag welds due to shrinkage cavities and the fact that the slag pool must be maintained. This is normally overcome by using a run-off tab and, again, such a run-off -tab is necessary with the narrow gap welds but its height can be small compared with those used for conventional welds (20mm compared with 50 to 75mm) and may be only from 10 to 15mm. Also in the process of the invention, the sheet electrodes can become unmanagably long if joints of much greater height than those in the present trials were to be welded. For the electrode thickness and gap used, the electrod preferably is about 3 times the height of the joint. With electrodes up to 500mm in length handling problems can be overcome but longer electrodes could give rise to difficulties with buckling and electrical stickout unless care is taken in selecting electrode metal layer thickness and with feeding the electrode. The electrodes described above had a metal layer of commercial mild steel because of the more ready availability of that material in a suitable form. It is to be expected that the mechanical properties of the weld metal can be improved by appropriate choice of electrode metal for specific metals to be welded. Also, the effect of the mass of the sections .to be " welded requires consideration, since the method of the invention is well suited to welding of relatively long lengths. The significant practical advantages provided by the invention will be apparent from a consideration of the foregoing. It is considered that these make possible more efficient welding operations, such as in structural welding and also butt-welding of members such as railway-line sections. Finally, it is to be understood that various alterations, modifications and/or additions may be introduced into the constructions and arrangements of parts previously described without departing from the spirit or ambit of the invention.

Claims

1. A process for welding of metal members in which the members are positioned so that respective surfaces at which they are to be welded are in opposed, spaced relation to define an upwardly extending gap between the surfaces, the gap having a width between those surfaces that is substantially less than both the width and vertical extent of the surfaces; inserting endwise, downwardly into the gap a consumable sheet electrode having a thickness less than the gap width so that the electrod has its width dimension disposed substantially horizontally and substantially parallel to said surfaces; the electrode having a central layer of metal and, on each of opposed major surfaces of that layer, a layer of slag forming flux; supplying electri power to the electrode to form a slag pool into which the electrode extends and to initiate an electroslag welding operation; and continuing that operation, while lowering the electrode, to form a weld between the surfaces of the members.

2. A process according to claim 1, wherein power is supplied to the electrode at a current density of at least 2 5 amps/mm of electrode metal cross-section.

3. A process according to claim 2, wherein the current

2 density is from 5 to 12 amps/mm of electrode metal cross- section.

4. A process according to claim 2 or claim 3, wherein the power is supplied at a voltage level of from 20 to 30 volts

5. A process according to any one of claims 1 to 4, wherein the width of the gap between the surfaces of the metal members is not greater than 8mm.

6. A process according to claim 5, wherein the width of the gap is from 3 to 6mm.

7. A process according to any one of claims 1 to 6, wherein said gap is closed across respective sides of said members by non-metallic end pieces.

8. A process according to claim 7, wherein said end pieces are of a refractory material.

9. A process according to claim 8, wherein said material is selected from limestone and oxides of aluminium, silicon and magnesium.

10. A process according to any one of claims 7 to 9,

( O ^{' ■} wherein each said end piece is provided with flux over.the portion of its surface extending across said gap.

11. A process according to claim 7, wherein said end pieces are of flux material.

12. A process according to any one of claims 7 to 11, wherein each said end piece includes a support member.

13. A process according to any one of claims 1 to 12, wherein the width of the electrode is not substantially less than the width of the surfaces of said members.

14. A process according to claim 13, wherein the width of the electrode is substantially equal to or greater than the width of the surfaces of said members and, at least where greater, end pieces for the gap are grooved or shaped to accommodate the electrode.

15. A process according to any one of claims 1 to 14, wherein the metal layer of the electrode has a thickness not greater than 4mm.

16. A process according to claim 15, wherein the metal layer has a thickness of from 0.3 to 1mm.

17. A process according to any one of claims 1 to 16, wherein the electrode is connected to a vertically movable slide assembly so as to be movable into the gap by manual • or motorised movement of the assembly.

18. A process according to any one of claims 1 to 17, wherein a run-on tab is provided at the bottom of the gap and/or a run-off tab is provided at the top of the gap, the or each such tab having a width substantially equal to that of the gap and extending substantially the full length of the gap.

19. A process according to claim 18, wherein the or each such tab has a height up to about 25mm.

O...PI