94 PHILlPS TECHNICAL REVIEW VOLUME 20

ENCLOSED WELDING OF RAIL SECTIONS

/

One very well-Known practical consequence ofthermal expansion is that enough space has to beleft between the rails of the permanent way toallow them to expand to the anticipated maximumwithout any danger of buckling. Now the railwayauthorities would be glad enough to do away withthese gaps between rails, for their elimination wouldresult in an immense saving in maintenance on thetrack and rolling stock. It has been estimated thatno less than 80% of maintenance costs are attri-butable to jolts set up as trains pass over the gaps.The reduction of jolting and noise would also meanmore comfort for the passenger. However, thedanger of buckling has always made the gaps anecessary evil.

In recent years there has becn a change for thebetter in this respect, thanks to improvements thathave been made in most countries in the manner oflaying the rails. Rails are now fixed more firmlyto the sleepers, the distance between sleepers hasbeen reduced, and the gravel metalling has beenimproved. These improvements have reduced therisk ofbuckling to such an extent that it has becomeunnecessary to leave gaps between sections of rail.

For fullest advantage to be taken of this fact,the ends of the sections should be welded together.Until a few years ago, however, no satisfactorymethod existed that could be employed for doingso in situ. Non-electric methods, such as thermitewelding, do not in general give very good results.In arc-welding, one comes up against the difficultythat the greater part of the weld has to be madebetween vertical faces, namely the cross-sectionsof the web and head of the rail. This means thatonly a moderate welding current can be used, other-wise the :fillermetal will run away before it has timeto solidify. Making the weld is therefore a slow job.Worse still, on account of the low current value toolittle heat is communicated to the parent steel oneither side of the weld, which therefore cools quickly,becoming very hard and brittle in places owing toits high carbon content. The only electrical methodof satisfactorily welding rails was resistance welding.That method cannot of course be employed on thetrack. It is in fact employed sometimes in railwayworkshops, but the overall length of the weldedsections is restricted by the need to transport themon a string of flats wagons. A total length of 60yards - 200 yards in exceptional cases - is notusually exceeded; and the problem of joining thewelded sections together still has ~o be solved.

It became clear that the whole question was of

621.791.75:625.143

more than local interest when, in quick succession,inquiries reached Philips Welding Department fromfour railway companies in differént countries. Webelieve we have found the answer to their problem inthe shape of an are-welding method that was inprocess of development when the inquiries weremade. For reasons that will become clear in amoment, we refer to the method as "enclosed"welding.The investigations that eventually led to the

enclosed welding method were undertaken in thefollowing circumstances.Metallurgical research had established that hy-

drogen was frequently the cause of unsatisfactoryresults in are-welding by the normal method. Fairlylarge quantities of that element are present in thedeposits from normal welding electrodes, such asmineral-coated and rutile electrodes. Basic types ofelectrode coatings ("low-hydrogen electrodes") havebrought about a big improvement in this respect 1).A characteristic difference between basic (or

low-hydrogen) and non-basic electrodes becomesapparent when molten steel and liquid slag arepresent in equilibrium. Where non-basic electrodeshave been used, the steel may contain slag inclusionsand the slag itself may be free from metal (fig. la);

a b Q3Qb3

Fig. 1.Diagrams to showequilibrium states of molten steel andliquid slag when (a) non-basic electrodes have been used,(b) when basic electrodes have been used. In case (a) it ispossible for slag to be present in the steel, but not for steelto be present in the slag; precisely the opposite applies incase (b).

when basic electrodes are employed, however, theweld steel is slag-free, although drops of metal maybe present in the slag (fig. lb). (It may be observedthat the question as to which of the two states willresult is not merely a matter of specific gravities.)It was clear that this particular property of basic

1) J. D. Fast, Philips tech. Rev. 14, 96, 1952/53.

1958/59, No. 4 ENCLOSED WELDING 95

electrodes, which plays no part in ordinary welding,might assume a practical importance: even whenworking with a large weld-pool, it would he possibleto obtain a deposit essentially free from the macro-scopic slag inclusions that do a great deal to weakena weld.

Our investigations, referred to above, were directedtoward finding a practical way of exploiting this fact.After a series of tests, we arrived at the arrange-ment shown in fig. 2formaking a vertical butt weld.

-------_ ..6

~ ---.t.2mm

L 7 I It

{3 2+---r2mm

54-

--94096

Fig. 2. Arrangement for enclosedwelding of two steel bars 1and 2. The enclosure consists of a base-plate 4 and blocks5 and 6, all of copper. Welding is carried out inside thecompartment' 3. The clearances through which the slag runsout are visible in the plan view.

Two steel bars 1 and 2 to be welded together, arearranged vertically with a gap 3 of about 15 mmbetween their edges. The bottom of the gap isclosed by a copper base-plate 4, and its sides areblocked by copper blocks 5 and 6. A basic weldingelectrode is introduced into the compartment 3 thusformed, and the are is struck against the lower partof one of the bar ends. A thick electrode (5 to 8mm) and a corrcspondingly heavy welding current(250 A to 450 A alternating current) is preferablyemployed. A large pool of molten steel, with liquidslag on top of it, is now formed in the compart-ment. Since narrow clearances (2 mm) have beenleft between the workpieces and the copper blocks

and base-plate - see view from above in fig. 2 -the thin liquid slag runs out through these openings,but the steel remains. The function of base-plate 4and blocks 5 and 6 is rather similar to that of themoulds ("shuttering" ) used for concrete structures-they form a temporary enclosure for the concrete,vhile it is setting.These tests confirmed that, when basic electrodes

are used, the deposited steel will he free of macro-scopic inclusions if only it has remained molten fora long enough time. That condition appears to besatisfied by the arrangement shown in fig. 2, pro-viding halts for electrode changing are of veryshort duration (when large cross-sections are beingwelded, dozens of electrodes are necessary for theone job). There is a big difference between enclosedwelding and normal welding in layers: in the formerthe whole weld actually constitutes a single layer;in the latter case, thin layers up to the requirednumber are deposited one on top of the other. As aresult of this, and as a consequence, too, of theheavier welding current, the amount of heat appliedper unit length of an enclosure-welded joint is aboutseven times as much as that applied per unit lengthand per layer in normal welding. In enclosed weldingthe parent steel facing the weld cools so slowly thatno appreciable hardening takes place in the transi-tion (heat-affected) zone. A second consequence ofthe rapid heating rate is that the work proceedsquickly. The components of the enclosure can beused over and over again, for hundreds of jobs. Allthat is necessary is now and again to regrind thosefaces that have roughened with use.Although the process looked promising, it was

found that even low-hydrogen electrodes still con-tained too much hydrogen for enclosed welding.The difficulty is that the high column of moltensteel prevents the escape of the hydrogen 2) whichmust be removed as the temperature drops. Theenclosed welding of vertical and almost verticaljoints only became a complete success once a special("extra low-hydrogen") electrode.had been develop-ed. To prevent these new electrodes (Philips 56 R)absorbing moisture in transit or during storagethey are supplied only in airtight tins.The manner of welding rail sections end-to-end

can be divined from the simple arrangement in fig. 2.However, the shape of the rail cross-section callsfor enclosure components that are more elaboratethen those for flat plates. Figs. 3a and b show thecomponents required and their positioning. Weldingof rail sections begins at the foot of the rail. Here a

2) P. C. van der Willigen, De metallurgie van het lassen vanstaal, Lassymposium Utrecht 1957, page 25 (in Dutch),

96 PHILIPS TECHNICAL REVIEW VOLUME 20

difficulty arises that always gives trouble whenmassive objects are to be welded: the bulk of therail remains cold, and. hence initially the parentsteel i~ the vicinity of the weld cools quickly, there-by becoming brittle. The difficulty may be overcomeby delaying the cooling process. A simple way of

E'E

Fig. 3. Enclosure for welding rail sections. a) With the base-plate A and small blocksBand B' in position, the foot is welded.Blocks C .and C' are then added and the lower part of the webis welded. b) Blocks C and C' are replaced by large blocks Dand D', and the weldingof the web is completed. Adding blocksE and E' allows the head to be welded.Base-plate A and blocksB ... E' are of copper. F is a steel plate on which rest theclamps for blocks pand D'.

doing so -is to heat the foot with an oxyacetyleneflame prior to welding, raising its temperature toabout 4~0 °C. Fig. 4a shows the enclosure in posi-tion: foot and web of the rails have already beenwelded, and the welding of the head can now begin.The completed weld may be seen in fig. 4b.

Important advantages of enclosed welding arethat no preparative machining and but little finish-ing are required and that, as already stated; it is aquick method. (Providing the gap in the rails is ofthe right size, the weld can be made and finishedoff within half an hour.) Further practical detailsare given in an article 3) that has appeared elsewhere.

Suitable enclosures can equally well be designed forother cross-sections; among the enclosures designed,there is one for rods of circular section such as usedfor the reinforcement of concrete.

A new welding method cannot be recommendedunless both laboratory investigation and practicaltests have proved its soundness. As regards the for-mer, the Philips Laboratory at Eindhoven has madephotographs of etched sections and radiographs ofnumerous specimens of enclosed welds. These testifyto the absence of macroscopie inclusions (fig. 5).In addition, the Vickers-Lips hardness meter, whichis ideallysuited to the purpose 4), has been used todetermine local variations in hardness in the welditself, in the transition zon~s and at places beyondthat part of the rail affected by the welding opera-tion. Hardness curves obtained in this way areshown in fig. 6, from which it may be seen that thevalue of 300 VPN 10 5) - which is still quiteacceptable - is nowhere exceeded.

A great variety of methods are used by railwayengineers in various countries for testing themechanical endurance of rails. All such tests havebeen carried out by outside organisations, sincePhilips does not possess the appropriate equipment.Pulsation tests and tup tests are the most importantkinds.

Pulsation tests, the purpose of which is to deter-mine fatigue strength, have been conducted by theNetherlands Railways and British Railways, byT.N.O.6) Delft (on behalf of Kloos and Sons'Workshops at Kinderdijk, Holland) and also at theMunich Technische Hochschule (on behalf of theBamag Works at Butzbach, Germany). In thesetests the operative quantity is P, the tensile stressin the outermost fibre of the foot of the rail. A stressP is caused to oscillate between a low minimumvalue Pmin (e.g. one or two kgjmm2) and a graduallyincreasing maximum value Pmax. The fatiguestrength is taken to be the value of Pmax at whichfracture just fails to occur after say 1000000 pulsa-tions have been passed (however, the number ofpulsations to which the test piece is subjected maybe as high as 3 X 106, 5X 106 or even 10X 106).As always, finish-machining and polishing the weldhas a good effect on the test results, since this re-moves surface irregularities at which fracture is

3) G. Zoethout, Het bekist lassen van spoorstaven, Lastech-niek 23, 274-277, November 1957 (in Dutch).

4) H. T. Schaap, Hardheidsmeters en hun bruikbaarheid voorhet onderzoeken van de overgangszones van lasnaden, Las-techniek 18, 27-33, 1952 (in Dutch).

ó) VPN = Vickers' pyramid number. The figure 10 signifiesthat the measurement was made with a force of 10 kg.

6) Dutch National Council for Industrial Research.

1958/59, No. 4 ENCLOSED WELDING 97

aFig.4. Enclosed welding of rail sections. a) Foot and web have already been welded, andthe welding of the head can now begin. b) The completed weld.

b

liable to nucleate. In tests by the Netherlands Rail-ways, and in tests by British Railways, in whichonly the underside of the foot had been so finish-machined, the welds were found to have fatiguestrengths approximately half that of the unweldedrail. Fatigue strengths but little inferior to that ofthe rail itself were found for the finished and polish-ed specimens used for experiments at T.N.O. andMunich. For practical purposes it is recommendedthat at least the foot, the tread and the inside edgeof the head should be finished and polished.The Netherlands Railways have carried out so-

called four-point bend tests, as illustrated in fig. 7.To prevent the rough surface ofthe weld prejudicingthe reproducibility of the result, the load wasapplied a few centimetres on either side of the joint.Typical results appear in Table Ij a mean fatiguestrength of 19 kgjmm2 was found for these speci-mens.

Fig. 5. Photograph of the etched cross-section of a weldmade by the enclosed method between two rail sections(0.8 X actual size). No macroscopie slag inclusions are present.The dark regions on either side of the weld are the transition(heat-affected) zones.

98 PHILlPS TECHNICAL REVIEW VOLUME 20

300VPN10

200

.-0.

.-;:: _..o--'~ \0

/ .»>__.,:r <,

~n ---- ~/0 »-: 0 '0

I0

~/0 I a-

q

280

260

240

220

lBO

260VPN10

160

/V

0 -«~o

1\----0"" ~~ .....,_ 0

0 --k --__----..:; 6

0 r---- -

240

220

200

lBO

14093972

Fig. 6. The photograph shows a detail of the head of the weld appearing in fig. 5 (3.2 Xactual size). A row of indentations made by the Vickers-Lips hardness meter may be seenbetween A and A'. The hardness values determined from the indentations and expressedin Vickers units are plotted in curve a. Curve b corresponds to the indentations B-B'.All values occurring in the curves are well below 300 VPN 105).

1958/59, No. 4 ENCLOSED WELDING 99

F

cr

fd -?.\.4+d= 50mm

I : I =1000mm

8A

IIIOk-------------~r.-+--------------~_X

b +M

Fig. 7. Four-point bend test on enclosure-welded railway sec-tions, as conducted by the Netherlands Railways. a) Thespecimen of welded rail C rests on supports A and B (distanceapart I = 1 m) with the weld in the middle. A pulsating forcetF is applied at two points a distance of -Itd = 50 mm oneither side of the weld; the weld has not been finished orpolished, but at the points where the forces are applied thesteel has its original smooth surface. b) Bending moment M asa function of length x along the specimen. Between the pointsof application of the load the bending moment has a uniformmaximum value of Mruux = tF(I-d).

Tests on behalf of British Railways, carried out inthe Derby laboratory of the British TransportCommission, gave comparable results.

Table 11 contains the results of tests on twofinished and polished weld samples. The tests werecarried out at the Institute for Rail and RoadConstruction of Munich Technische Hochschule,under the direction of Prof. Meier. The results agreefairly well with those of corresponding tests doneby T_N_O_,in which small rods cut out of the footof the weld sample constituted the test objects.

Tup tests have been conducted by the railwayauthorities III England (British Railways) andFrance (S.N.C.F.).The British Railways tests, carried out in their

laboratory at Redbridge, consisted in allowing atup of It tons (1270 kg) to fall on the weld froma height that was progressively increased by 6

Table I. Results of pulsation tests carried out by the Nether-lands Railways on six welds between rail sections, only thefoot of the weld having been finish-machined. The rails weigh-ed 46 kg per metre. Pmin = 2 kg/mm2.

Pmax, Number ofkg/mm2 pulsations

17 270000018 276000019 3 160 00020 217000020 243000020 3200000

Result

No fractnreNo fractnreNo fractureFractureFractureFracture

93965

Table IT. Pulsation tests carried out by Munich TechnischeHochschule on two samples of finished and polished weldsbetween rail sections weighing 49 kg/m. Pm;o = 1 kg/mm''.

Sam- Pmax, Number of Resultpie kg/mmê pulsations

I

---- ---

I22 2000000 No fracture24 2000000 No fracture26 2000000 No fracture

1

I

28 2000000 No fracture30 2000000 No fracture31.5 I 2000000 No fracture

I32.5 983000 Fracture

Total 12 983 000--- --

~I 25 5000000 No fracture

2 28 5000000 No fracture( 31 3900000 No fracture

ITotal13 900 000

inches until fracture occurred. The distance betweenthe two supports on which the test object restedwas 1.20 m, and the rails weighed about 54 kg permetre. Eight out of ten welds fractured at 5.33ton. metres, the remaining two at 7.00 ton. metres.

In their laboratory at Moulin-Neuf, the S.N.C.F.carried out six tup tests whereby a tup of 1000 kg

Fig. 8. Switchpoints, welded at hoth ends to normal rail sec-tions. The joints are indicated by arrows. The section betweenthe welds is milled out of a single steel block ("monoblock"construction).

100 PHILlPS TECHNICAL REVIEW VOLUME 20

was dropped on the weld from a height of 6 m andthe work done in causing fracture was recordedautomatically. Here the distance between the sup-ports was 0.70 m and the rails were of the U33type. The results obtained were 4.37, 3.21, 3.165,4.83, 1.51 and 5.23 ton.metres. (They cannot becompared with the English results on account ofthe quite different conditions under which the testswere performed.)

We may conclude with some notes on the actualemployment of enclosed welding on the railways.The workshops of K100s en Zonen at Kinderdijk,to which reference has already been made, were thefirst to put the enclosed welding of rail sections into

practice: some 90 switchpoints of the so-called"monoblock" type were' welded to normal lengthsof rail (fig. 8). These are now in use in the mar-shalling yard of the IJmuiden Blast Furnaces.Since then, welding by the new method has beendone on a large scale in a number of other shuntingyards in the Netherlands (Rotterdam and Eind-hoven) and Belgium (Beringen). Having had goodresults with enclosed welding in a shunting yardnear Toten, British Railways are now using theprocèss on both branch and main lines. A numberof gangs are proceeding with this work daily, andhave already made thousands of welds:

G. ZOETHOUT.