Cast iron: graphite, silicon and repair

Kautny saw the mending of broken iron castings as one of the largest uses of the welding flame, and a century later it still is. Machine beds, pump casings, gearbox housings, engine blocks and valve bodies are cast in iron because it is cheap, damps vibration and casts into complex shapes. When they crack, replacement can take months. Welding them is possible, but cast iron punishes the habits of steel welding more harshly than any other common metal: it is brittle, it hardens beside the weld, it carries residual stresses from casting, and it can lose the very element that keeps it grey.

This chapter explains why, using Kautny’s insight about silicon and TWI’s modern guidance, and sets out the repair plan for the pump casing’s cracked foot.

The family of cast irons

Cast irons are iron alloys with more than about 2% carbon. How that carbon is held, as graphite flakes, graphite nodules or iron carbide, decides almost everything about the metal, including how it welds. TWI’s job knowledge article on cast irons summarises the main types.

Figure 1Types of cast iron and how they weld

	Structure	Weldability
Grey	2.0 to 4.5% C, 1 to 3% Si; graphite flakes	weldable with care; flakes make it brittle
Nodular (ductile, SG)	3.2 to 4.5% C; graphite spheroids	more ductile; welds more easily
Malleable	heat-treated white iron	weldable; depends on type
White	carbon held as iron carbide	normally considered unweldable

Grey iron, the commonest, is also the most difficult of the weldable types. Its graphite flakes act as built-in cracks, so it has little ductility, and it cannot stretch to accommodate the shrinkage of a weld. Nodular iron, whose graphite forms round nodules, is much more ductile and, TWI notes, forms martensite less readily, so it welds more easily.

Why cast iron cracks

TWI lists the reasons, and each connects to an earlier chapter.

• Brittleness. The graphite flakes in grey iron cannot withstand the cooling stresses of a weld, which Chapter 3 showed are large.
• Hard zones. With so much carbon available, the heat-affected zone and any fused parent metal readily form iron carbides and martensite as they cool, hard and brittle structures that crack, as Chapter 4 described for high-carbon steel.
• Uneven heating and thermal shock. Uneven preheating causes unequal expansion, and fast cooling causes thermal shock, either of which can crack the casting.
• Contamination of nickel deposits. Nickel weld metal, used in many repairs, can pick up sulphur and phosphorus from the casting and crack as it solidifies, a problem Chapter 8 explores.

Castings also carry their own residual stresses from uneven cooling in the mould, which Kautny noted are highest where heavy sections join light ones, such as the rim and spokes of a wheel. A weld adds its stress to those already present.

Kautny’s silicon insight

Kautny explained the hardening of cast iron welds in terms of silicon and manganese, and the substance of his explanation holds. Silicon promotes the formation of graphite as cast iron solidifies; manganese works against it, favouring iron carbide. Grey iron stays grey because it has enough silicon. If silicon is lost under the welding flame, the balance tips, and the weld zone solidifies as white iron, hard and brittle, which explained to him why welds in grey iron so often became hard.

His remedy was a filler of grey cast iron richer in carbon and especially in silicon than the casting, to replace what was lost, together with a flux to protect the molten metal from the air and to deal with oxides. That is still the basis of gas fusion welding of cast iron with matching cast iron rods.

One detail needs correcting. Kautny attributed the silicon loss to evaporation, describing silicon as having a low boiling point. Silicon in fact boils at a very high temperature. Its loss in the flame is by oxidation: silicon combines readily with oxygen, which is one reason a neutral flame and a flux matter so much. Fast cooling compounds the problem, because rapid solidification itself favours white iron. Both effects are reduced by a high preheat and slow cooling, which is why gas fusion welding of cast iron uses the highest preheats of any method.

Hot and cold repair methods

Modern cast iron repair falls into two broad approaches, with braze welding as a third option.

Hot repair preheats the whole casting to a high temperature and fuses it with a matching cast iron filler, usually by oxy-acetylene, so that the weld solidifies slowly and grey. TWI’s table gives a preheat of about 600 °C for gas fusion welding of the common grades. The result can match the casting in colour, structure and machinability, but the casting must be heated and cooled uniformly, often in a furnace.

Arc repair with nickel-based fillers uses lower preheats, around 300 °C for grey irons by manual metal arc or MIG in TWI’s table, and fillers of nickel, nickel-iron or nickel-copper. Nickel weld metal does not form hard carbides; TWI explains that it absorbs the carbon picked up from the casting and gives a ductile, machinable deposit. The technique uses small runs, peening and slow cooling. For large castings TWI also describes a low-heat-input technique of small stringer beads, each peened while hot, with the weld area kept from building up heat.

Braze welding, with copper-based fillers that melt below the casting’s melting point, avoids melting the cast iron at all and so avoids most of the cracking. TWI notes that the surfaces must be cleaned thoroughly, for example in a salt bath, because mechanical cleaning smears graphite over the surface and prevents wetting.

Figure 2Preheat for grey cast irons by method

	Manual metal arc	Gas fusion
Ferritic flake grey iron	300	600
Pearlitic flake grey iron	300 to 330	600
Pearlitic nodular iron	200 to 330	600

Preparing the casting

Preparation decides much of the outcome. The full extent of the crack is found first, usually by dye penetrant testing, because cracks in cast iron often run further than they appear. Material containing the crack is removed by grinding or with a carbide burr, and TWI advises lightly grinding the prepared surface to remove any hardened material. Castings that have run in oil, such as machine bases and gear cases, absorb it into their porous structure, and it boils out under the arc as porosity, so oil-soaked castings are degreased and often heated to drive the oil out before welding. Where the repair will be loaded, a qualified engineer should confirm that a welded repair is acceptable at all.

The casing foot repair plan

This is an illustration. The pump casing’s drawing identifies it as a pearlitic grey iron. The crack is 120 mm long through a 25 mm foot. The workshop chose arc repair with nickel-based fillers, which avoids the 600 °C heating of a gas fusion repair, and followed the sequence TWI describes.

1. Find and remove the crack. Dye penetrant confirmed the crack’s full length and showed it stopped short of the radius into the body. It was ground out into a 90° single-V groove through the 25 mm foot. The groove’s cross-section is about half of 50 mm wide by 25 mm deep, 625 mm², so the groove holds about 625 × 120 ≈ 75 cm³ of weld metal.
2. Preheat the whole casting. The casting went into the furnace and was brought to about 300 °C, the lower end of TWI’s range for pearlitic flake iron, checked at the foot, the radius and the far side of the body. Chapter 2 estimated the heat needed at about 10.9 MJ.
3. Butter the groove. A first layer was laid on each groove face with a 2.4 or 3 mm nickel electrode, in short runs, each peened while hot and cleaned of slag, to form a soft, ductile layer over the cast iron.
4. Fill with nickel-iron. The groove was filled with 3 or 4 mm nickel-iron electrodes, in short runs working from the body end towards the free edge, as Chapter 3 recommended, each peened and cleaned, with the casting returned to the furnace whenever its temperature fell towards the lower limit.
5. Cool slowly. The casting went back into the furnace and cooled with it.
6. Inspect. After cooling, the repair was dye penetrant tested and the foot machined flat.

The repair is as much a thermal cycle as a welding operation, and the workshop planned it that way.

Figure 3Thermal cycle for the casing foot repair

1. Stage 1Heat in furnacewhole casting to about 300 °C, checked at three points
2. Stage 2Buttershort nickel runs, each peened while hot
3. Stage 3Fillnickel-iron runs from body end to free edge; reheat as needed
4. Stage 4Slow coolback into the furnace, cooled with it
5. Stage 5Inspectdye penetrant, then machine the foot flat

When not to weld

Welding is not the only way to repair a cracked casting, and sometimes it is the wrong one. Metal stitching, a cold mechanical repair in which the crack is closed by a series of drilled and fitted locking pieces and threaded pins, introduces no heat at all and is widely used on large castings that cannot be heated uniformly, such as machine beds and engine blocks. Bolted straps, clamps and fitted inserts can carry load across a crack. A replacement casting, or a fabricated steel replacement for a broken bracket, may be cheaper and more reliable than a repair.

For castings that contain pressure, carry safety-critical loads or are covered by a design code, the repair method is not a workshop choice at all. It must be approved by a competent engineer and, where the equipment is regulated, by whoever the regulator or code requires. The pump casing in the case is a low-pressure water casing and the crack is in a mounting foot, not the pressure envelope; had it run into the volute, the workshop would have recommended a replacement casing.

A failure: the long hot runs

This is an illustration. On a trial repair of a scrap casting of the same type, done to qualify the procedure, a welder used 4 mm electrodes at the top of their current range with a wide weave, to fill the groove quickly. The weld looked smooth. Dye penetrant testing showed a crack running along the fusion line on one side, and a file skated over a narrow band of the casting beside the weld, a sign of a hard white or martensitic zone.

The cause was heat concentration. Long, wide, hot runs melted more cast iron into the weld, raising the carbon and sulphur picked up by the nickel deposit, and created a larger fused and heat-affected zone that cooled as one large mass with high shrinkage. The procedure was corrected to limit run length, electrode size and current for the buttering layer, to forbid weaving beyond a stated width and to require peening of every run. The qualification test was repeated and passed. Trial repairs on scrap castings are a cheap way to find this kind of fault before it reaches the customer’s part. They also give the welder who will do the real repair practice on the same metal, section and position, under the same thermal cycle, which no amount of steel welding experience replaces.

What the repair record holds

A cast iron repair leaves few visible clues about how it was done, so the record matters more than usual. For the casing foot, the workshop recorded the casting’s identification and grade, the dye penetrant result before and after, the groove dimensions, the furnace temperatures and the points where they were measured, the electrode types, sizes, batch numbers and drying, the welder, the peening, the reheating events and the cooling. If the foot cracks again in service, that record is the difference between a guess and a diagnosis.

Machinability and appearance

Customers often care whether a repair can be machined and whether it matches the casting. Nickel deposits machine readily, which is a main reason for choosing them; the hard zone beside a poorly controlled weld does not, and a tool that skids on it is an early sign that the procedure needs attention. Nickel and nickel-iron welds are a different colour from cast iron, which matters on visible surfaces. Gas fusion repairs with matching filler can match the casting in colour and structure but require the high preheat. Kautny claimed that a properly executed grey iron weld would have a better structure than the original casting; whether a repair matches, exceeds or merely joins the casting depends on the method chosen and how carefully it is controlled.

Exercise: preheating a nodular casting

A 40 kg pearlitic nodular cast iron bracket is to be repaired. Using TWI’s table, give the preheat range for manual metal arc welding and the preheat for gas fusion. Then calculate the heat needed to raise the bracket from 20 °C to 330 °C and to 600 °C, using 0.46 kJ/kg·K.

TWI’s table gives 200 to 330 °C for manual metal arc and 600 °C for gas fusion. Heating to 330 °C needs about 40 × 0.46 × 310 ≈ 5,704 kJ; heating to 600 °C needs about 40 × 0.46 × 580 ≈ 10,672 kJ, nearly twice as much, and a longer, more carefully controlled cooling afterwards. Nodular iron’s better ductility makes the lower-preheat arc repair attractive, provided the filler and procedure are qualified and the casting is cooled slowly.