Welding; theory, practice, apparatus and tests, electric, thermit and hot-flame processes

Preface

In spite of the numerous data on the theory, practice, apparatus, and tests of welding contained in the trade journals and metallurgical books, no previous attempt has been made to present this data in sequence under one cover. But in the last fifteen years the subject has begun to be of interest and importance. The electric, thermit, and hot-flame processes are welding all of the metals and are doing repeat and repair work that has never before been attempted. New brazing methods have also been successfully tried out and the range of good solders greatly increased.

I have given separate chapters to the commercial metals. Few of the metallurgies give much space to the working properties of the metals, especially the welding property, which is often merely mentioned.

Test and cost data must be taken with a grain of salt. The test data have been compiled from various sources. Those tests given for iron are standard and cannot be questioned. Those tests given for the special processes are more recent and in most cases have been made by interested parties. They are no doubt accurate, but at present the special processes cannot be so well represented by test data, as by the actual work they turn out. The same may be said of cost data. The prospective purchaser of welding machinery must figure the cost of his apparatus plus the cost of labor and the depreciation. But above all, he must satisfy himself that the apparatus he chooses is the best for his kind of welding.

I wish to express my thanks for assistance received from the different welding companies mentioned herein; also to James H. DeLong, for special analyses; Dr. Edward Hart, Dr. Joseph W. Richards, Prof. Oliver P. Watts, Otis Allen Kenyon, E. A. Colby, of Baker & Co., and to many others.

R. N. Hart.

Los Angeles, Cal., October, 1910.

V

Contents

PAGE

Preface . . , v

Definitions and introduction xi

Part I— The Metals IRON

Pure or nearly pure iron is readily weldable at a white heat. Malleable, or nearly pure wrought iron, is known as weld iron. The range of temperature in which it can be welded is very wide: it runs from the imperfect weld at cherry- red, to dazzling white, when the welding property is lost just before the iron melts. According to Pouillet, the different colors of heated metals are represented approximately by the temperatures given:

Deg. C

Incipient red 525

Dark red : 700

Incipient cherry 800

Clear cherry-red 1000

White 1300

Dazzling white 1500

Melting 155°

Wrightson ' states that he found iron to increase in volume as much as 7 per cent, when passing into the plastic stage.

Malleable Iron. — Weld iron is a name occasionally used to describe malleable iron, indicating that it is pure enough to be welded. Malleable iron is produced by the puddling process. It is made by melting up pig iron and scrap in an open-hearth furnace and burning out the greater part of the silicon, manganese, and carbon in the order named. As the burning continues the melting point of the iron rises, and it becomes a pasty mass permeated with slag from the stirring. This nearly pure iron is gathered into "puddle balls," and taken to the rolls at a white heat. It is rolled or hammered out into long strips, the strips are cut, reheated to white heat, and welded together between the rolls or beneath the hammer. To produce the finest wrought

' Journal of the Iron and Steel Institute, 1895, ^^o'- I- P- 4^3-

I iron, the process of rolling and welding is repeated. This is the puddling process in brief; its object is to squeeze and work the slag out of the iron and to give the iron a fibrous structure like rolled copper.

The principal chemical difference between wrought iron and steel is the carbon content. In fact, except for its high carbon, basic open-hearth steel is purer than most malleable irons.

All weldable iron is not malleable, but it is weldable in proportion to its malleability.

As is seen in the foregoing process, the iron is welded

several times.

Fig. I. — ^\'iews of scarfed bars for tt j.^ iir^u t*^».

lap welding. HoW tO Wcld IrOIl.—

The mechanics of a "smithed" weld is in the main as follows: Suppose the smith wishes to join two short bars of malleable iron, of cross-section I by 2 inches. He first hammers or cuts a convex bevel or "scarf" on the ends of each bar, as shown in Fig. i.

The heating must be done in a coke or coal forge. Coke is the best fuel for the reason that it gives a good reducing flame. The fire is blown with the bellows until it is at a high heat. The ends of the bars are thrust well into the midst of the ignited coke and more coke is piled over them. The blowing is continued until the bars are red-hot on the ends, when the smith takes them out with pliers and dips them into pure sand or borax near by. This is the flux, and it so acts on the surface of the iron as to clean it of rust and y, ^

forms a glass that protects ^ t, • . . . , , ,.

, ^ Fig. 2. — Bars in position for lap welding.

the fresh iron from further

rust. The smith puts his irons back into the forge again and

heats them until white.

When white, or nearly white-hot, the smith takes one bar in his pliers, his helper takes the other, and they place them together on the anvil, the one lapping the other (Fig. 2). Both the smith and his helper give the pieces quick, light blows with their hammers until the plastic iron is well-joined. They turn the irons while hammering to get an even weld on all sides.

IRON 3

The smith will try to weld at a white, but not a dazzling white, heat, and he will try to complete it without putting it back into the fire for a reheating.

When the weld is finished, he may give it a special shape by placing it between swage blocks and hammering.

This simple operation of welding requires much skill, as will appear. The smith may find that with all his skill his joint is bad or that his iron will not join at all.

Points in Practice. — In practice there are the usual number of details requiring skill that the smith must observe.

The contact faces should always be shaped so that their middle points touch first and so that hammering causes union first at the middle. This prevents slag becoming enclosed.

The heating should be done rather slowly, so that the pieces will be of uniform heat at the weld. Both irons should be of the same temperature. To effect this the smith must not blow up his fire too hot and must watch the color of both irons. A deep bed of coals will give a flame the least oxidizing, because most of the oxygen will burn to carbon monoxid or dioxid.

The welding heat is quite different for different irons. The smith must use his judgment, taking good care he does not overheat his irons. Overheated iron, miscalled "burnt iron," will be brittle on cooling. Steel is much more sensitive to overheating than iron. It is seldom safe to heat steel above redness.

The iron should be well fluxed directly on the contact surfaces, and not merely around them, as is too often the case. The common fluxes for iron are pure silica (river sand) and borax. Impure malleable irons are self-fluxing to a certain degree, but this cannot be relied on. Other fluxes are calcined borax and sal ammoniac. One writer^ recommends powdered marble, which is limestone, the flux of the blast furnace. Borax is recommended for steel on steel or steel on iron, as silica is inadequate.^ Besides fluxing the bars before raising them to welding heat, the smith may sprinkle calcined borax on the irons when they are ready to weld, to replace it on surfaces accidentally rubbed bare in the forge.

' "The Blacksmith's Guide," J. S. Sallows.

^ "Testing of Materials of Construction," W. C. Unwin, 1899, p. 292.

The hammering, or "working," is very important, and must be done rapidly. The smith manipulates his bars so that welding begins at a middle point and works outward, driving the slag away from it. The first few blows of the hammer should not be heavy. If the pieces are large or the smith slow, his heat will fall before the weld is finished and he must put his bars back in the fire. Second heating must be avoided if possible; slag is apt to

Fig. 3. — Lap weld. No upset. Fig. 4. — Lap weld. Upset.

get in the weld or the pieces may become " burnt," to say nothing of the time lost.

The different kinds of welds known to the smith — ^butt, lap, scarf, jump, cleft — are variations of simple welds. Illustrations of them and of how the stock should be shaped are given in figures 3 to 7. The method of hammering will suggest itself in each case. Jump and butt welds will have sufficient upset to work on, while scarf welds will not unless the pieces are jumped or over-lapped considerably.

Welding Fires, etc. — The ordinary fire used by the smith for

ZD

3(1

Fig. 5. — Jump weld. Fig. 6. — Butt weld.

his welding is a deep bed of coke. But the fire may be fed with hard or soft coal; it may also be a gas or oil flame.

For blacksmithing, the coke fire is much the best. In this fire nearly pure carbon is burned, and the resultant gases are carbon monoxid and dioxid. The carbon monoxid and the coke are reducing in their action. They will prevent the metal heated from oxidizing and will even clean the metal of scale.

IRON ^

Hard coal can be used, but in a small forge it is impossible to get a hot enough fire. Soft coal is a poor coal for forge work. It has, as a rule, high sulphur, which the hot iron readily absorbs to its detriment. While the thick carbon smoke of the flame is apt to collect on the iron in an oily soot unless the flame can be kept hot enough.

Gas and oil flames are often used for chain welding and similar operations. The gas or oil should be fairly free from sulphur.

The gas flame is made by in-

jecting the gas through a large «

Bunsen burner. Air is intro-

1 , 1 , , . , Fig. 7.— Cleft weld.

duced through holes m the

burner a short distance from the burner end. Or if the gas is under sufficient pressure to make a roaring flame, it will take in enough air at the end of the burner. The proper mixture gives a nearly colorless flame. The quickest and best way to tell if a flame is right for welding is to heat in it a small piece of steel or soft iron. If the flame has too much air, the metal piece will rust and scale off in the flame. If too little air, the soot will collect on the metal and it will not heat quickly. Gas flame is the most convenient for most purposes. It is easy to regulate and can be turned on the iron while welding. In this way iron can be welded without flux. When chain is' made the Hnks are hung on a bar above a long narrow furnace of fire bricks. The flame is played beneath.

The oil flame is a cheaper flame than gas. It can be manipulated in the same way as a gas flame. But the oil must be free from sulphur compounds to make a safe flame.

It has lately been tried to burn gases that have been preheated. If coal gas, starting to burn at 35 deg. C., will give a combustion point of about 2000 deg., then by preheating it to about 800 deg. C. and burning, there will be a gain of 400 or 500 deg. in the temperature of the flame. In other words, the welder has boosted his flame up to about 2500 deg., a very respectable temperature.

The preheating is done by passing the gases through copper coils which are nested above the welding-flame flue, and which receive the heat from the welding flue.

O WELDING

Causes of Poor "Welds. — Suppose we have an iron or steel that will weld. There are still many good reasons for discrediting the safety of the joint until it has been tested. They are:

I. Imperfect Contact.

a. The two surfaces may give the appearance of perfect union, but a considerable percentage of the central portion may be faulty. For if the metal in course of puddling is not relieved of its enclosed slag, this will be finally pressed out into thin laminae. These clea\ing planes will be parallel with the bar until it is welded, when they will be upset in all directions at the joint. A network of such planes now running across the bar instead of parallel to it will greatly diminish the tensile strength.

b. Then, again, the smith may carelessly allow some of his flux to stay within the weld. All the hammering in the world

Fig. 8. — Correct shapes for jump Fig. 9. — Incorrect shapes for jump

and lap welds. and lap welds.

will not remedy it. In most cases one of the pieces can be scarfed or pointed, so that the first contact of the hot pieces will be at a central point. As the hammering proceeds, the slag will be forced out of the joint as fast as the pieces weld (see Fig. 2). This caution applies equally to. the smithed weld or the Lafitte joint (see page 158).

c. Or the smith may not flux his bars correctly, thereby allowing unreduced and uncombined oxid of iron to get in the weld. It is not a difficult thing to get this rust incorporated in the weld, because it is soluble in iron. Dissolved rust makes a "burnt" or brittle joint. Clean silica sand is the common flux. The smith dips his hot bars into a box of the sand kept near by, and often sprinkles a handful over the pieces when hammering if the pieces are large. The reaction between the sand and the rust is rapid. Iron silicates, easily fusible, are formed. They

IRON cover the fresh iron surface with a thin glass, which prevents further oxidation.

2. Insufficient hammering or ''working''' may account for the poor weld. The iron may be perfectly joined, yet the structure is weak and uncertain. Wrought iron has a fibrous structure similar to wood. The smith must do his best to continue this structure through the weld. With a lap weld he can start with a heavy upset which can be hammered down. Hammering serves the double purpose of consolidating the metal and of laying the cleavage planes perpendicular to the blows. Much hammering is impossible with jump and butt welds. The lapwelding of wrought-iron pipe and the scarf-welding of chain linksmust depend on pressure for their perfect finishing. Welds of steel cannot often be much benefited by working. Steel is an alloy, without structure. To be strong it must be solid and homogeneous. Campbell's advice is to keep the critical temperature high and keep down the silicon and phosphorus when choosing your steel.

3. Too high a heat is responsible for some bad welds. Very great heat will unsettle the structure of the iron in the stock a considerable distance from the joint. Because the bar breaks near, but not on, the joint, is no proof of the soundness of the weld. The structure of the iron is impaired wherever the heat is beyond the critical point of crystallization of any important impurity. By this is meant that if there be an appreciable amount of carbon, for example, it will tend to mineralize the iron when the latter is cooled down past the crystallization point of the first carbon-iron mineral. A number of other impurities, silicon, aluminum, arsenic, etc., are presumed to act in like manner. So the smith must work his weld and also all of the surrounding metal that has been raised above this critical temperature. Just what is the critical temperature is unknown to him, because it varies with every iron, according to the amount and proportion of its impurities.

Oftentimes welds are made which cannot be worked to advantage. Such welds are apt to be weak. The only present remedy is to select for such purpose iron as free as possible from objectionable impurities. Machine welds, such as small chain links and small-bore pipe, where the entire piece must be heated, are especially subject to this disadvantage.

4. Large welds are unsafe.

Just what is the maximum cross-section of practicable welding cannot be safely laid down. Welding is not advisable where it is difficult to get a good welding heat, to flux properly, to joint the pieces well and accurately, or to work the weld when made. To weld a ship's sternpost or skeg or rudder-post was a matter of weeks of expensive and often hopeless labor before the advent of thermit (see page 147). To weld a large driving- rod, an embossing die, crushing roll, or propellor shaft was impossible. The largest iron pipe used, however, could be welded; also chain links a foot or more long, and 2- or 3-inch stock could be safely welded.

Campbell says,^ "Welds of large rods of forging steel are entirely unreliable. Electric methods do not offer a solution of the problem, for during the process the metal is heated far beyond the critical temperature." Thermit, since discovered, does offer a partial solution, because a thermit weld can be reinforced with as much metal as needed, and the metal itself can be varied to meet the requirements. Its cost is to be considered. The oxy-acetylene melt-weld should recommend itself for like reasons. More than one torch can be used for large work (see page 75).

Effect of Impurities, etc. — In order to make a good weld of iron, the metal must possess plasticity, which is directly affected by the contained impurities. The welding temperature seems to be the point at which the iron becomes semiplastic, but is still not sufficiently hot to burn or oxidize rapidly. For the reason that this temperature varies considerably for the different alloys of iron, it is impracticable to make any definite rules. A number of impurities or alloy formers affect the welding property to a marked degree even when present in fractional parts of i per cent. Substances which cause "red shortness," such as sulphur, or which oxidize to form an impervious skin, such as aluminum, must be avoided in iron.

In the case of cast iron, which is iron high in carbon and silicon, the metal passes suddenly from a crystalline to a liquid condition.

» "Metallurgy of Iron and Steel " p. 588.

IRON 9

It cannot, therefore, be welded over the smith's forge, and can be welded only by one of the recent special processes.

There are an infinite number of kinds of iron, due to the proportions of the combined impurities and the method of treatment during the milling. Each of these so-called alloys will act differently at the welding temperature, while many of them cannot be welded by the smith at any temperature. There are a few general rules which will help the worker in the ^election of his material. But there is so much room for failure in welding the best irons that what cannot be blamed on the smith and his material must be laid to the devil.

Carbon in the combined state should be kept below 0.30 per cent, because it is a hardener and makes brittle in the combined state. Carbon has been found to amalgamate with iron to form a number of alloys, which are unworkable in proportion to their carbon content. W. C. Unwin^ gives 0.90 carbon as the limit for welding iron with silica flux; i.io carbon as the limit with any flux. These alloys,^ pearile, martensite, cementite, etc., mineralize in the iron and form a granitic structure on cooling. It is likely that all of the mineral impurities of iron behave in this manner.

In the uncombined form, as graphite, carbon does not affect the welding property, except that it becomes an enemy to the homogeneous structure of the metal, just as does slag.

Slow cooling, drawing the temper, keeps the combined carbon at a minimum. Silicon in a quantity approximately over i per cent and aluminum even more potently, also reduce combined carbon to graphite. However, the objection to silicon is that it makes a brittle alloy, and aluminum oxidizes to the disadvantage of the metal. Slow cooHng is valueless for the reason that welding is done at a high heat, when iron is capable of absorbing carbon in greater quantity.

Hence high carbon in the stock cannot be toned down and should be avoided. It should never be above 0.50, and ought to be below 0.25 for the best results.

Silicon causes brittleness, which does not yield to welding heat. It must be kept below 0.20 per cent. Silicon is the

' "Testing of Materials of Construction," W. C. Unwin.

2 Wm. Campbell, Electrochemical and Metallurgical Industry, Feb., 1904.

lO element which differentiates cast iron from mild irons and steel. An iron with gray-colored fracture contains too much silicon to use in welding.

Welding Property of Silicon Steels^

The weld

Si c p s

Mn

Unsatisfactory

0.21-5.08

0.14-0.26

0.08

0.05

0.14—0.29

Perfect

0.01-0.504

0.028-0.094

0.455-0.622

It will be noted that the second series of steels gave perfect welds with a content of 0.504 silicon, by above analysis; probably accounted for by the high manganese, which will prevent crystallization.

Phosphorus makes the "rotten iron" for thin and sharp castings when present in quantity approximately over o. 10 per cent. Malleable iron for welds should contain less than 0.03 per cent of it, not so much to assist welding, as to insure a strong joint on cooling.

Sulphur is one of the deadly enemies of the smith. Its effect on iron is especially disastrous at welding temperature, even when not otherwise apparent. In quantity approximating over o.io, it causes " red shortness," brittleness. It should be kept as low as possible. Manganese is the remedy for sulphur, used in the production of iron, but the smith cannot correct it with manganese himself. Very low sulphur content is the secret of the success of Swedish iron in the arts.

Manganese may be present up to about i . 50 per cent, in iron or steel to be welded. Up to this limit its effect is generally beneficial. Campbell states that between 2 and 6 per cent it forms a brittle unworkable alloy. Hadfeld's steel, with about 7 per cent, manganese, again becomes malleable.

Manganese is the cheapest tonic of the iron producer. It reduces both sulphur and oxygen in the iron and rises to the top as slag. The remainder, if any, forms a tough workable alloy with the iron.

'Campbell, "Metallurgy of Iron and Steel."

IRON II

Nickel steel welds readily at all compositions. Nickel is a valuable addition to iron, because small percentages of it greatly increase the tensile strength without impairing the elasticity. It is also valuable because it prevents rust in its iron alloys to a marked degree. Nickel steels of from 2.05 to 4.95 per cent nickel are specifically mentioned as weldable if the carbon is kept down.^

Chrome steeP can be welded. The first hammering must be very gentle so that the metal will not fly to pieces.

Aluminum. — There is some uncertainty about the effect of small quantities in iron. In amounts above 3 per cent it forms a valuable alloy with steel, which is highly fluid on melting. 0.50 per cent increases the tensile strength and elastic limits from 3000 to 8000 pounds and lessens the ductility.^ Odelstjerna^ says that only 0.002 aluminum gives inferior steel castings, the fracture being coarsely crystalline.

A serious count against aluminum, if it be true, is that it oxidizes in its alloys and coats them with a skin. This would seriously affect the welding, because this skin is a very refractory, unmanagable substance. Reliable data are wanting.

Copper is generally believed to be harmful in malleable iron. However CampbelP in his welding experiments used bars containing 0.35 per cent, with excellent results. He says: "The critical temperature at which the steel ceases to be malleable and weldable varies with every steel. It is lower with each associated increment of copper; it is higher with each unit of manganese, and it is lower in steel that has been cast too hot."

Arsenic steel, with less than 0.20 per cent arsenic will weld as usual. Between o. 20 and i . 20 per cent a flux of borax and sal ammoniac is needed. 2.75 per cent arsenic prevents welding altogether, and the iron behaves hke pig iron.** Campbell claims that so small an amount as 0.093 impairs the welding property.

^Iron Age, July 25, 1905.

^ R. Brown, Journal of the Iron and Steel Institute, i8q6, Vol. I, p. 472.

^"Metallurgy of Iron and Steel," Wm. Campbell, p. 477.

* Transactions American Institute Mining Engineers, Vol. XXIV, p. 312.

^ "Metallurgy of Iron and Steel," p. 467.

® Iron Age, April 13, iSqq.

^ "Metallurgy of Iron and Steel," p. 478.

He says that o. 20 per cent, of arsenic increases the strength and reduces toughness.

Nitrogen content is not a subject for the smith to bother about. Nitrogen has been blamed recently for otherwise unaccountable failures of chemically good iron. E. J. Sjostedt^ claims that infinitesimal quantities of it cause red and yellow shortness. He claims that a furnace producing trisilicate slag gave iron with 0.003 P^r cent nitrogen; bisilicate slag, 0.016 per cent nitrogen; monosilicate slag, .024 per cent nitrogen. 0.006 per cent, he says, is the limit for good steel and, presumably, for good welding steel.

However, there is at present no easy way of recognizing its presence, and it cannot be guarded against.

Tests of Smith Welds. — Campbell made a series of tests of smithed welds of all of the different kinds of steel and of wrought iron, the results of which are given in tabulated form in his "Metallurgy of Iron and Steel." Four smiths of ability and experience welded the metal in flats and rounds of size and shape most convenient for handling. And though the men knew their bars would be tested, their welds were often far from satisfactory. "Picking out the worst individual weld of each workman, blacksmith 'A' obtained only 70 per cent, of the value of the original bar, 'B' 54 per cent., 'C 58 per cent., and 'D' only 44 per cent. The forging steel showed one weld with only 48 per cent., the common soft steel 44 per cent., while even the pure basic steel gave one test as low as 59 per cent."

But the tensile strengths of the bars are fairly uniform when compared with the elongation. "In some cases where the break took place away from the weld, the elongation was nearly up to the standard." The elongation test of a basic open-hearth steel of low carbon gave greater elongation in welded pieces than in the natural bar; "but in the other pieces the stretch was low and the fracture so silvery that it was plain the structure of the bar had been ruined. In most cases where the test bar broke in the weld, the pieces parted at the surfaces of contact, showing that no true union had taken place; one or two fractures were homogeneous, but they showed the coarse crystallization that follows overheating." ^

* Iron Age, May 5, 1904. ^ "Metallurgy of Iron and Steel," Wm. Campbell.

IRON

Welding Tests of the Royal Prussian Testing Institute

Kind of metal

Ultimate strength lb. per sq. in.

Per cent, elongation in 20C mm.

= 7.87''

Per cent, of reduction of area

Average of 6 tests natural

Average of 9 tests welded

.Average of 6 tests natural

Average of 9 tests welded

Average of 6 tests natural

Average of 9 tests welded

Medium O. H. steel .

Soft 0. H. steel

Puddled iron

72110 64570 57890

41820

45800 47080

20.8

25-1 22.2

3-2

5-1

7-7

34-9 44-7 39-5

4-5 10.5 14.0

These results agree substantially with those of Campbell. The lowest result for soft steel was ^^ per cent., the average 71.

The lowest result for medium steel was 23 per cent., the average 58.

The lowest result for puddled iron was 62 per cent., the average 81.

These results confirm the general impression that puddled iron is the best iron for welding. Contrary to one authority who says that iron before puddling welds more easily because of the presence of the slag in the iron.

The welding tesf- is occasionally specified in this country on account of the common use of welds in structural work. Two iron bars of the metal under test, of about i inch section, are scarfed, heated to white heat, and joined without flux. The joint is worked with an 8- or lo-pound hammer, and brought down to unit section. It is cooled without chiUing. This bar is then tested for tensile and elastic strength; and a similar weld is half cut open, bent until fractured, and examined for structure. Results of Tests by Prof. Bauschinger

Kind

Section, sq. in.

Ratio of strength at weld to strength of bar, per cent.