1,22 PHILlPS TECHNICAL REVIEW VOL. 15, No. 4,

THE DETERMINATION OF DROPLET SIZE IN ARC WELDING BYHIGH-SPEED CINEMATOGRAPHY

by P. C. van del' WILLIGEN and L. F:DEFIZE.

In arc welding, material is transferred from the we{ding electrode to the work in the formof droplets. To a,certain extent the size of these dropiets tietermines the welding characteristicsof the electrode. In order to provide data regarding the droplet size, films have been made ofthe transfer of the weld metal, with a camera taking up to 3000 frames per second. With thesame camera, colourfilms have also been made which present 'a ueryclear picture of the weldingprocess as a whole.

621.791.75: 778.534.83

Introduetion,

Investigation into the manner in which the metalis transferred from the welding electrode to the workin are welding, has revealed that this usually takesplace iÎl the form of free droplets of molten metal.Sometimes the droplets are so large that theyconstitute a short circuit while flowing from the,electrode into the pool - in this case the dropletscan no longer be considered to' be wholly free.

To i great extent the droplet size determines thewelding characteristics of the electrode; it-is knownthat small droplets result in a smooth bead, andthat the smaller the droplet, the more stable theare, This stability relationship is especially notice-able in A.t. welding, where the are is establishedtwice per cycle. Electrodes producing small drop-lets, moreover, melt more quickly than those whichgive the larger drops, '.because a large. drop usuallyclings to the electrode for a relatively longer time,thus impeding the melting of fresh weld metal.On the other hand the welder will generally preferto employ an electrode giving a large drop when itis' required to span the gap between two plates.

The number . of different kinds of electrodemanufactur~d has gradually increased in the courseof time, and the need has arisen for a quantitativedetermination of the droplet size occurring withthe warleus electrodes. In order to meet this need,systematic measurements have been carried out in. the Philips laboratories at Eindhoven, and adescription of the method employed is now pre-sented. To illustrate the results obtained, theoxide-silicate and gas-shielded electrodes will becompared with the "contact" electrode which wasderived from these types 1). The results obtainedwith a "basic" or low-hydrogen electrode ar~ also'mentioned.

1) For a review of the characteristics and composition ofwelding electredes see for example J. D. Fast, The functionof the coating of welding rods, Philips tech. Rev: 10, 114,1948.

.Before low-hydrogen and contact electrodes were introduced(the first of these before, and the second just after the lastworld war), mild steel and low-alloy steels were welded almostexclusively with o~ide-silicate or gas-shielded electrodes,of which the 'Philips 50 and Philips 48 are examples.Basic electrodes are so named because the mainconstituents

of the coating are basic oxides. The outstanding feature of thiskind of electrode is that the coating gives off very little watervapour which otherwise produces atomic hydrogen in the arc;this is liable to be absorbed by the molten metal, with detri-mental effects on the weld 2). These low-hydrogen electrodesexcel by reason of their good mechanical properties and lowsensitivity to impurities (especially sulphur) in the metal tohe welded. However, they have to be used with a very shortare if porous welds are to he avoided, and this imposes ratherheavy demands on the skill of the welder.Contact electrodes are noted for their easy welding, and more

particularly, for their high welding speed 3). They do notrepresent a new 'type of electrode in the sense that the coatingyields a new kind of slag, i.e. one not met in other kinds ofelectrode. Their essential feature lies rather in a modificationof existing types, to produce an electrode suitable for contactwelding, i.e. the are is struck automatically when the coatingis brought into contact with the work; the electrode with itscoating is then rested on the work while the weld is beingmade. To make this possible, a semi-conductive coating isused, which is also thicker than usual, owing to the fact thata large part of the core metal is included in the coating in theform of powder.

. Method of measurement ind apparatus used

In order to determine the 'size of the droplets,films have been made of the are, a method oftenemployed to demonstrate the transfer of the weldmetal from the, electrode to the work. It wasalready known that the processes taking placewithin v the arc are so rapid that very high filmspeeds (> 1500 frames per sec) would be requiredto permit these processes to be followed 4). Use

2) J. D. Fast, Low-hydrogen welding rods, Philips tech. Rev.14, 96, 1952 (Nos. 3-4).

3) P. C. van der Willigen, Contact arc-welding, Philips tech.Rev. 8, 161, 1946 and 8, 304, 1946.

4) A. Hilpert, Werkstoffübergang im Schweisslichtbogen,Z. Ver. dtsch. Ing. 73, 798-799, or Welding J. 8, 21, 1929and 12, 4-8, 1933.

OCTOBER 1953 DROPLET SIZE IN ARC WELDING 123

was made of an Eastman high-speed camera typeIll,which is capable of taking a maximum of 3000frames per sec. on 16 mm film; exhibited at thenormal speed of 24 frames per sec. the picturesare seen slowed down by a factor of up to 125. Thecamera does not of course operate at full speedfrom the moment of starting; this speed is developedin accordance with the curve shown in Jig. 1.frames per sec.

7561ü

3000!.---+--V-:f- --

V

I ---

I/

i2000

7000

75 20 25-+ metres of fil;~

Fig. 1. To determine the diameter of the droplets transferredfrom welding electrode to work, a cine camera taking up to 3000frames per sec was used. The curve in the diagram showshow the speed develops after the camera is started.

5

To obtain an indication of the speed relating toevery part of the film, 50-cycle A.C. was used forthe welding are; the current thus passes throughzero 100 times per second, so that the are extin-guishes that number of times per second. Anotherargument in favour of filming an A.C. are is theincreasing use of the A.C. welding technique, whichinvolves much less difficulty due to magnetic areblowand requires less expenditure on equipment.

The fumes, as well as the light emitted by theare, tend to make filming difficult. The fumes canbe blown from the field of view by means of acurrent of air directed across the are. To minimizethe effect of too much light from the are itself, apowerful light source was placed behind it (fig.2);originally a 60 A are lamp from a cinema projector

75611

Fig. 2. Diagram of set-up for the filming in silhouette of theweld metal droplets in the welding arc. The light from acarbon are lamp A is concentrated in the welding arc B by aconcave mirror, so that the droplets can be seen against abright background. C camera.

3()

was used, the light being concentrated around thewelding are by a concave mirror, to produce abackground of high and constant brightness 5).The welding are thus became almost invisibleagainst the background, and the droplets ofmolten weld metal could be filmed in silhouette.However, this light source was cumbersome and

proved rather unmanageable in use, and it wassubsequently replaced by a water-cooled mercuryvapour lamp, this being much more convenient tohandle. To concentrate the light, a lens was mountedbetween the lamp and the are, Another advantageof this type of light source is that it introduces nointerfering objects in the beam; the carbon arcand mirror had to be very carefully watched, toensure that no shadow from the carbon-holder fellacross the welding are (see fig .. 5).

Fig. 4 depicts the actual apparatus used, with themercury vapour lamp. The welding electrode wasmounted in such a way that the arc remained in thesame place while the electrede was being consumed,with the electrode-holder lying in the direction ofthe electrode itself and supported by four guiderollers. The work was pulled along below the arcat the required speed by an electric motor. It wasnot found necessary to use automatic feed for theelectrode-holder to maintain a constant are length.

At,I

75261

a bFig. 3. Increase in the field of view obtained by replacing theare and mirror by a high-pressure mercury vapour lamp inthe apparatus depicted in fig. 2.a) Field with carbon arc and mirror. The shadow of one of the

carbon-holders is visible on the left.b) With the mercury vapour lamp the shadow is eliminated.

The silhouette of the small bead already deposited canbe clearly seen.

The arc length was checked by means of an opticalsystem with calibrated eye-piece, and the camerawas started as soon as the are attained the desiredlength. The operator hat no difficulty in keepingthe are length constant for the few seconds neededto take the shot (15 m of film).

Another very attractive method of photographing thewelding process is by means of X-rays 6), the advantage of

5) A similar arrangement was first employed by L. Bull;see M. Lebrun. La soudure électrique à l'arc et ses appli-cations, Bibl. off. centr. acétylène et soud. autog., Paris,1931, pp. 39-44.

6) J. Sack, A new method for the investigation of the transferof material through the welding are, Iron and Steel Inst.Symp, on the welding or iron and steel, part 2, 553-559,London, 1935. J. Sack, "leIding and welding rods, Philipstech. Rev. 2, 129, 1937.

124 PHILlPS TECHNICAL REVIEW VOL. 15, No.

75429

Fig. 4. Actual apparatus employed for filming the transfer ofweld metal in the welding are, A water-cooled mercury vapourlamp, the light from which is concentrated on the welding areby a condenser lens B. C welding electrode (whitened withchalk to distinguish it from the holder). D insulated rollersbetween which the electrode-holder passes. E motor for draw-ing the work slowly under the are, F jet (16 X 5 mm) throughwhieh 20 litres of air per minute is blown, to disperse the smokelaterally. G gauge indicating air velocity. H cine camera (upto 3000 frames/sec). J vertically mounted glass plates forprotecting camera and condenser from spattering metal.

this being that it is thus possible to distinguish between thecoating or slag and the metal. Further, no difficulties are thenencountered owing to the light from the welding are, or withfumes, so that it is not necessary to take steps to counteractthese. As far as we are aware, however, no equipment has so farbeen designed that will give cinematographic X-ray picturesof sufficient frame speed. At the same time, efforts are beingmade to increase frame speed in X-ray filming techniques,for instance, Slack and eo-workers ') describe a camera thatwill take 150 frames per second.

In order to obtain comparative data of the droplet size, itwould be sufficient to have available a large number of in-stantaneous pictures, not necessarily in the form of film; theslow-motion film, however, does provide an opportunity forstudying the whole process of droplet formation, the transferof the weld metal and the form and motion of the are,

Determination of characteristic droplet size

The individual pictures of the film are projectedone by one on the screen, from which the dia-meters of the droplets as well as the diameter of theelectrode are measured. As the actual electrode sizeis known, the real diameter of the droplets is easilyascertainable.It is first necessary, however, to define the size

of the droplet to be regarded as characteristic of anelectrode. The transfer of metal is determined bythe volume of the droplets, so it is desirable to workill terms of droplet volumes as obtained from the

') C. M. Slack, L. F. Ehrke, D. C. Dickson, and C. T. Zavales,High speed cine-radiography, Non-destr. Test., Spring] 949, pp. 7-11.

observed diameters. Let ni be the number of drop-lets found of which the volume is Vi. It will be clearthat the arithmetic mean droplet volume Enitï/ Eniwill not be suitable for representing the volume ofthe characteristic droplet. As against the verylarge number of small droplets, the influence of thelarger drops, which, after all, are those that con-tribute most towards the transfer of metal, wouldthus remain obscured 8). This objection is removedby counting each drop a number of times propor-tional to its volume, i.e. to the contribution of thedroplet to the transfer of weld metal. The charac-teristic droplet volume is then defined as:

Enivi2Vc =

Now each droplet is encased ill a protectivecoating of slag 9), and the volumetrie percentageof slag will certainly vary between one dropletand another. However, in order to take the averagevolume of slag into account in the calculation, wereduce the characteristic droplet volume by the(known) volumetrie percentage of slag in the moltenbead, taking the result to be the characteristicvolume of the iron core of the droplets. From thiswe then obtain the characteristic diameter dc of thedroplet core.As an example, jig. 5 shows a histogram of the

number of observed droplets as a function of their

J5 75612

JOf-JLI

-- --.JhI~

LLn-, Irt-r h .-

15

10

5

oo 2,5 3 J,5

-- d0,5 1,5 de 2 4mm

Fig. 5. Diameter of droplets transferred through the arc ofcontact electrode C 18-4, used with free are, Vertical axis:number of droplets whose diameters lie between the limitsdefined by the intervals (1/7 mm in width) on the horizontalaxis. (Here the characteristic diameter de of the core of thedroplet is 1.8 mm). For the measurement, the film was project-ed with an enlargement of 7 x. Droplets of smaller diameterthan 3/14 mm (1.5 mm on the screen) could not with anycertainty be distinguished from blemishes in the film, andare therefore not included.

8) Some of the metal evaporates. Strictly speaking, all themetal atoms in the vapour should also be taken into ac-count as separate droplets, but this number of atoms isso large that the average droplet volume would be almostequal to the volume of a single a tom.

9) See articles referred to in footnote 6).

OCTOBER 1953·

\

DROPLET SIZE IN ARC WELDING 125

diameter, for the contact electrode 10), C 18-4(with ~hich 38% of the total volume of the bead'of molten material is slag). Another complicationis encountered with contact electrodes in that thesize of the droplet cannot be determined from thesilhouette if the weld is made in accordance withusual practice. Normally, the cup of the electroderests on the work, and the space between thecoating and the work. is then so small that theformation of the dropléts cannot be observed. Toenable the droplet size to be determined, the weldingwith the contact electrode was therefore done using, an open are, It remains a moot point whether thesize o~ the characteristic droplet in the normalcontact welding procedure would be the same.The values obtained for contact electrodes C18-4

and C20:4, as well as'for the oxide-silicate electrodePhilips 50-5 (from which contact electrode C20

"was derived), are shown in the table. A few of theresults of measurement on the basic (low hydrogen)electrode Philips 56-5 are also included.

designed to give a large 'droplet, a~d the film mfact showed that metal-was transferred from thiselectrode only when the are was short-circuited(fig. 6a.and b). It is therefore somewhat surprisingthat with the Cl8 electrode, which is' the contactversion of the Philips 48, the material is transferredonly in the form of small, free, droplets (fig. 6c). Theconversion to contact' elect~ode has thus resultedin an appreciable decrease in the size of the droplet.Just the opposite has taken place, however, in theoxide-silicate electrode Philips 50 and its contactcounterpart, the. C20; the table shows that theratio dc/D' of the C20 is almost twice the rdtiodc/D of the Philips 50. No explanation has as 'yetbeen found for these facts.

The ratios dc/D' of the Cl8 and C20 electredes. 'are practically the same." It is remarkable that amuch larger percentage of very small droplets wasfound for the Cl8 than for any other of the elec-'trodes meásured. In the calculation of dc, however,thése smaller droplets are of minor importance:

Table. Diameter of the iron core of the characteristic droplet for several types, of weldingelectrodes. ,. ,. .

'The molten metal at the extremity of the Philips50 electrode, as well as that in the metal pool, is

. very mobile .. 'Ihis fact, coupled-with the smallnessof the droplet (the smallest measured) is relatedto the comparatively high FeO content of themolten metal (approx. 0.04%), which greatlyreduces the viscosity and surface _tension of themolten metal. The droplets transferred are irregularin shape. .

The Philips 56, low-hydrogen electrode gives thelargest droplets (figs. 6d and eh the-spherical formof the droplet is characteristic of this electrode.It is know;;: that the FeO content of the weld metalof this basic electrode is very low (0.001 %); theviscosity and surface tension are accordinglyhigh.

- Current Arc Arc Dropsdc .dclD rlc/n' (for

Electrode voltage length per I contact(amp.) (volt) (mrp.) sec. (mm) electrodes)

C 18-4 ;180 45 4.8 185 1.9 0.47 .ti 0.41,; 180 4·7 6.6 133 2.0 0.50 0.44- ,

6.1 145 0.44,.

" 165 42 1.8 • 0.38

I I I I.,

C 20-4 190 35 6.1 91 2.3 0.58 0.44

Philips 50-5 I 200 35 I 4.8 I 162 I 1.3 0.26

Philips 56-5 230 37. 8.0 29 2.7 0.54

" 230 37 6.2 32 2.8 0.56," 230 38 5.3 42 3.3 0.66

" 200 38 5.5 24 2.6 0.53 "

In general, larger droplets are to be expected toresult from larger core diameters D, so that, tocharacterise the type of welding rod, the value ofdc obiained should be divided by the value of D.For the' contact electrode, moreover, the ratiodc/D' is also given, where D' is the core diameterwhich the contact electrode would have if all theiron in the' coating were included in the core. (Inthe C18, 24% of the iron is in the coating as powder;in the C 20, 41%.)Discussion of results

The gas-shielded electrode Philips 48 was specially

10) The second figure in the designation of an electrode in .every case indicates the core diameter in mm.

126 PHILlPS TECHNICAL REVIEW

Sack 11) has also measured the droplet size of the Philips50-5 and 4B-5 electrodes, in this case by welding rapidly overa copper plate, so that the droplets were caught separately.These droplets, from which the slag was removed, were sortedaccording to weight, and the characteristic diameters of thedroplets computed along much the same lines as those de-scribed above. For the Philips 50-5 electrode, Sack obtaineda value of dc/D = 0.7, which is roughly twice the value found

1 'fj• •2 ta• •8 ·taca •15 ··tA• •24 'tjDI •34 '_j11 •36 ·ti

a

1 ti• •2 ·ti• •3 ·ti- -5 fl .• -8 , ,0• -11 'Q.. ..13 -(I

b

VOL. 15, No. 4

The "optical efficiency"In order to gain some impression whether the

transfer of weld metal was actually recorded quan-titatively on the film, the volume of material (ironand slag) transferred per second was first deter-mined from the measured droplet diameters. This

1 ta• •2 ta• •4 ti• •5 ~

liJ •6 "ti• •7 'ti• -9 tA

by us. It is very probable, however, that by this method thedroplets obtained are not droplets in the true sense, but largermasses formed by the merging of several droplets beforesolidification. For the Philips4B-5, Sack obtained droplets forwhich dciD = 0,9, but the method employed by him doesnot really lend itself to revealing whether the transfer ofmaterial is accompanied by shorting of the arc.

Fig, 6. Silhouette pictures of droplet transfer in are welding. Some of the intermediateframes have been omitted from each strip, as will be seen from the nnmbering. Thesereproductions do not do justice to the film as exhibited; for example, it is barely possibleto see from the above whether the arc is burning or not, whereas in the actual film the arecan be clearly seen to extinguish and re-ignite.a.) Metal transfer from Philips 4B-5 gas-shielded electrode. The short-circuiting of the are

is clearly visible, starting at the moment the alternating current passes through zeroand lasting almost one whole cycle (1/60 sec.). Film speed 2000 frames/sec.

b) Further pictures of thc Philips 4B-5 electrode. In frames 1, 2 and 3 the are is extinguished(current e- zero ). In 2 there is a momentary short circuit, but no material is transferred.From 5, B, 11 and 13 it will be seen that the arc exerts a force on the droplet, thusdistorting it (see H. von Conrady, Del' Werkstoffübergang im Schweisslichtbogen,Elektroschweissung 11, 109-114, 1943.)

c) With the contact electrode CIB-4, which is derived from Philips 4B electrode, themetaltransfer is in the form of small droplets. Weld made with free are,

d) The low-hydrogen (basic) electrode, Philips 56-5. Two free droplets are seen, the'upper one of which has the dimensions of the characteristic droplet (dc/D = 0,57). Theare is extinguished only in frame 3.

e) The Philips 56-5 again. Here a large droplet is seen in process of being transferred. Theare is extinguished in 1 and 23. The 23 frames represent one half-cycle of the A.C.(1/100 sec).

11) J. Sack, Overhead welding, Philips tech. Rev. 4, 9,1939.

c d e

result was then divided hy the volume of iron andslag per second ascertained from the actual weightof the quantities of material deposited and theknown specific gravities. For the C18-4 this "op-tical efficiency" was found to be 100%, which ofcourse it should he. The basic electrodes gavevalues of 150 to 200%, and the Philips 50 only

Fig. 7. Extracts (not sequential) from three colour films of are welding with A.C. Thedifferences in colour between the films are due to the use of different filters. (a) and (b)were filmed with the set-up shown in fig. 8; (e) with that of fig. 9.a) Philips 50-5. In the top and hottom frames the arc is just re-igniting after the current

has passed through zero. The air-blast used for dispersing the smoke also tends toblow away the are, In the last frame but one the are is extinguished (current passesthough zero). Note the spatter on the right-hand side.

b) Contact electrode C20-4 developed from the Philips 50. The electrode is now touchingthe work. The are is almost entirely shrouded by the cup formed by the coating andtherefore burns quietly. No spattering is seen, and the are is not affected by the air-blast used for dispersing the smoke. The arc is able to withstand a much greater airvelocity than Philips 50 before it begins to "stutter". The current passes through zeroin the second frame from the top.

e) Contact electrode drawn across the work at an angle that affords a view into the cup.The arc is extinguished in the top picture and is almost out in the last but one. It isclearly seen that the are originates in the core of the electrode and that, irrespectiveof the direction of the current, it fans out towards the work.

OCTOBER 1953

a

DROPLET SIZE IN ARC WELDING

b

127

c

128 PHILIPS TECHNICAL REVIEW VOL. IS, No. 4

50%. By moulding the droplets from the basicelectrodes in a synthetic resin and then cuttingthem through, it was shown that many of thesedropiets contained cavities. Hence the dropletvolumes observed from the film were greaterthan their true volumes. (In normal welding, thedroplets merge in the liquid pool and the cavitiesentirely disappear.)

In the case of the Philips 50 electrode, a numberof droplets were apparently not observed; thesewere presumably quantities of very small dropletswhich had been rendered invisible by the smoke,or by the fact that their silhouettes were indistin-guishable from blemishes in the film. Clearly, inthis case, the value of dc obtained for this electrode,which is already very low, is nevertheless too high.

Colour films

In addition to the black and white silhouettefilms, some films were also made which were in-tended to give an overall picture of the weldingprocess. On the advice of Prof. J. Brillié of Paris,

Fig. B. Path of the rays in filming with reflected light usingcarbon are lamp A as light source. This 1S the set-up used fortaking the colour films. B welding arc. C camera.

these films were taken in colour (fig. 7). The lightsource was mounted so as to illuminate the work atslightly more than grazing incidence (figs,8 and 9).The camera was similarly mounted, so as to photo-graph the work from an angle. In order to get asmuch reflected light as possible into the camera, thesurface of the work was polished before each shot;in spite of this, so much light was lost that the arewas much brighter than the background.

The advantage of filming from an angle is thatthe pool of metal under the are, with the layer ofliquid slag on its surface, is then clearly visible.This gives a good perspective view of the weldingprocesses. In addition to the advantage of moreor less natural colouring, the use of colour film -owing to its laminated structure - gives picturesalmost entirely free from the halation that other-

1

ftFig. 9. Apparatus used for filming the processes in the crater ofa contact electrode. The strip shown in fig. 9c was filmed inthis marmer. For meanings of letters A to J see fig.4. K filter(Scott GG 3). The welding electrode has again been whitened.The weld is made in the plane defined by the light source A(here, a high pressure mercury are) and the optical axis ofthe camera (instead of as in fig. B, where the weld is shownalong a direction perpendicular to this plane).

wise occurs when the camera is directed towardsthe light source. With black and white films, halationis found to be a drawback at the beginningof the shot, when the camera has not yet attainedits propel' speed and the exposure time is accordinglytoo long.The series of pictures reproduced in fig. 7 are

taken from three of these colour films. One seriesshows welding with an electrode with free arc (a),and the other two show contact welding withcontact electrodes.These films demonstrate the steadier are and

smaller spatter losses when the contact electrodeis used; they also show the shape of the are. Thefilming with the contact electrode at an obliqueangle, giving a view into the cup (see figs 9 and 7c),confirms that the end of the core metal is alwayscovered with a layer of slag during welding.

Summary. Systematic measurements have been carried out toascertain the size of the droplets of weld metal transferredfrom welding electrodes to the work. For this purpose, films ofthe welding are were made at a picture frequency 0 fup to3000 per second. For the background a light of such brightnesswas used that the droplets were filmed in silhouette, notwith-standing the brilliant light emitted by the are itself. Some ofthe measurements are discussed.

Colour films have also been made, at the same frequency, togive a more general impression of the processes taking placeduring welding. Certain features observed in the film arebriefly reviewed.