Moulding in sand

Iron that comes out of a furnace can only be poured, so almost everything made of cast iron began as a hole in packed sand. How to run the casting stations in the game is in the molten metal guide, and the pig bed, the simplest sand mould of all, has its own page. This page follows the moulder through one mould, then says what the game keeps of it.

Every period claim below comes from one working moulder's book, Bolland's The Iron-Founder of 1892, American in its examples. The work he describes is older than his book, but this page cannot say by how much.

Ramming a mould by hand in a foundry shop in Kaduna, Nigeria, today casting aluminium pots. The tool is the short wooden rammer Bolland gives to the bench moulder, and the mould is two boxes, one on top of the other.
Ramming a mould by hand in a foundry shop in Kaduna, Nigeria, today casting aluminium pots. The tool is the short wooden rammer Bolland gives to the bench moulder, and the mould is two boxes, one on top of the other.
The finished mould from the same shop, closed and resting on its joint: the upper box is full of rammed sand and the only way in is the hole cut in its top.
The finished mould from the same shop, closed and resting on its joint: the upper box is full of rammed sand and the only way in is the hole cut in its top.
The game's floor flask in its drawing, one of four flasks that differ in what they can hold: the small flask, the floor flask for tall and flat work, the long flask with its hinged cope, and the flask sunk in a pit., iso
The game's floor flask in its drawing, one of four flasks that differ in what they can hold: the small flask, the floor flask for tall and flat work, the long flask with its hinged cope, and the flask sunk in a pit.

The work in order

The pattern. A mould starts as a wooden copy of the casting.

Patterns kept on a foundry wall in north Wales: gear wheels, part-wheels and a grid, each cut in wood and kept for the next order.
Patterns kept on a foundry wall in north Wales: gear wheels, part-wheels and a grid, each cut in wood and kept for the next order.

It has to come out of the sand again without tearing it. Bolland's complaint about wheel patterns is the whole difficulty in one job: a pattern with loose arms makes "extra labor in the machine-shop", and one lifted with a jar breaks the teeth of the mould, which then get pushed back and patched (Bolland p. 314). His remedy is to bed core plates against the teeth and lift them clear of the pattern, so that the mould comes away "exact" (Bolland p. 315). For a long run of one article a wooden pattern was not enough, and halves of a finished casting were "finished up and pinned together and used for the pattern" (Bolland p. 326); an iron pattern could be cast from a wooden model (Bolland pp. 344-349).

The flask. The box that holds the sand costs labour with every shovel, so it was made no larger than it had to be: just big enough for the rammer to work 2 to 3 inches off the pattern (Bolland p. 37).

A 14-inch iron flask for small work (Bolland p. 37, Fig. 38), on its bottom board. Two halves, lifted by the handles, located by pins through the lugs and held for casting by the two clamps. Iron could be smaller than wood: a wooden box needs a thicker body of sand between the casting and itself, or it burns.
A 14-inch iron flask for small work (Bolland p. 37, Fig. 38), on its bottom board. Two halves, lifted by the handles, located by pins through the lugs and held for casting by the two clamps. Iron could be smaller than wood: a wooden box needs a thicker body of sand between the casting and itself, or it burns.
A snap flask (Bolland p. 38, Fig. 39), hinged at one corner and latched at the opposite one. Once the mould was closed it was opened and taken away for the next, and the bare block of sand stood under a weight with a hole left in it for the runner. Bolland's point is the number of flasks it saves, and the speed.
A snap flask (Bolland p. 38, Fig. 39), hinged at one corner and latched at the opposite one. Once the mould was closed it was opened and taken away for the next, and the bare block of sand stood under a weight with a hole left in it for the runner. Bolland's point is the number of flasks it saves, and the speed.

Large flasks were built up of loose sides and bars cast in open sand (Bolland p. 42). The lower part is the drag and the upper the cope, with cheeks between where a job needed more depth (Bolland p. 38).

Facing and backing. Sand came in two grades. A fresh facing mixture went against the pattern, with sea-coal in it for green work (Bolland p. 310). Old floor sand made up the bulk behind it and was used over again (Bolland pp. 133, 302).

Ramming. "No amount of finishing will rectify faulty ramming" (Bolland p. 29).

How to ram beside a pattern (Bolland p. 29, Fig. 15). Rammed too far off, the casting swells; rammed too close, it scabs. Where one layer meets the next, at A, a seam is left on the casting unless the joint is packed first with a small rammer, at B. Then the ordinary rammer goes in at C, reaching the bottom of the layer each stroke and never closer than two inches to the pattern.
How to ram beside a pattern (Bolland p. 29, Fig. 15). Rammed too far off, the casting swells; rammed too close, it scabs. Where one layer meets the next, at A, a seam is left on the casting unless the joint is packed first with a small rammer, at B. Then the ordinary rammer goes in at C, reaching the bottom of the layer each stroke and never closer than two inches to the pattern.

Bench moulders used short wooden rammers, often one in each hand, and floor moulders worked standing with long iron ones (Bolland pp. 27-28). Sand hanging under the bars of a cope was carried on iron hooks called gaggers (Bolland p. 27).

Venting. Damp sand makes steam the moment iron reaches it. Too much water leaves holes in the upper faces of the casting from steam that "no ordinary amount of venting will carry away" (Bolland p. 26). The vent was a hole pushed through the rammed sand with a wire (Bolland p. 26).

The joint and the lift. On a kettle mould parting sand is thrown on before the cope ring goes on (Bolland p. 181), so that the two will come apart. The cope was lifted, the pattern drawn, the mould mended, and the cope closed down again on an empty space (Bolland pp. 314-315).

A closed mould (Bolland p. 98, Fig. 97), labelled for this page. The iron runs down the gate and up into the mould. It then bears on every part of the mould as a column of liquid to the top of the basin, so the cope is pushed up as if iron were standing on it. Bolland's worked case is the sum at the right (Bolland p. 99). A cope that is not weighted or bound rises, and the iron runs out at the joint.
A closed mould (Bolland p. 98, Fig. 97), labelled for this page. The iron runs down the gate and up into the mould. It then bears on every part of the mould as a column of liquid to the top of the basin, so the cope is pushed up as if iron were standing on it. Bolland's worked case is the sum at the right (Bolland p. 99). A cope that is not weighted or bound rises, and the iron runs out at the joint.

The joint also decides how good each face will be. In a floor mould Bolland finds "smooth upper surfaces, and equally rough and unsightly lower surfaces" (Bolland p. 46). Work that mattered went into dry sand or loam, where every face could be reached and finished (Bolland pp. 46-52).

Cores. A hollow in a casting is a separate piece of sand set into the mould.

Cores for the steam jacket of a cylinder (Bolland p. 263, Figs. 227 and 228). Fig. 227 is one core cut across: the sand at A has an iron bar through it and vents along it, because gas from a core surrounded by iron has to get out by its ends. Fig. 228 is the iron as it lies in the core box. Cores like these were rammed, dried, and the set built up in the mould one at a time.
Cores for the steam jacket of a cylinder (Bolland p. 263, Figs. 227 and 228). Fig. 227 is one core cut across: the sand at A has an iron bar through it and vents along it, because gas from a core surrounded by iron has to get out by its ends. Fig. 228 is the iron as it lies in the core box. Cores like these were rammed, dried, and the set built up in the mould one at a time.

Bolland's best mixture for cylinders is 8 parts fire-sand, 2 of moulding sand and 1 of flour, in clay-water (Bolland p. 124), dried in an oven (Bolland pp. 52-62). Each core needs its own vent (Bolland p. 262).

Pouring and feeding. The iron went in hot and quickly, because metal crossing a mould loses heat to the sand (Bolland pp. 79-80). Green-sand work was poured as cool as would still fill it (Bolland pp. 69-70).

A plain cylinder 3 feet across and 6 feet long, cast in loam (Bolland p. 148, Fig. 128). The length above the flange B, up to C, is a head 6 inches deep, cast on "to receive the sullage, or dirt, which gathers as it is being cast". The dirt rises into the head instead of staying in the cylinder.
A plain cylinder 3 feet across and 6 feet long, cast in loam (Bolland p. 148, Fig. 128). The length above the flange B, up to C, is a head 6 inches deep, cast on "to receive the sullage, or dirt, which gathers as it is being cast". The dirt rises into the head instead of staying in the cylinder.

Iron shrinks as it freezes, and freezes from the outside, so the middle of a thick section is the last to set and the first to go hollow. A "riser or feeding head" keeps hot iron against that place, and counteracts the shrinkage "in some measure at least" (Bolland p. 75). A cylinder has a head on its top end (Bolland p. 243), and a 16-foot gun carries a 3-foot sinking head (Bolland p. 264).

Shaking out. The casting was broken out, and the sand went back on the heap, though not for ever. A heap "very rotten on account of a preponderance of burnt sand" needs new sand mixed in (Bolland p. 303).

Period basis, and what is the game's own

  • Period: three classes of mould: green sand, cast damp; dry sand, baked first; and loam, built and struck to shape (Bolland pp. 147-148, 233). The game's flasks work in green sand; loam and dry-sand moulds are not modelled, and a dry-sand core is planned.
  • Period: a wooden pattern, drawn from the sand and used again. The game's pattern is a tool cut from planks that wears a little with each impression and is cut again from its diagram when it is spent.
  • Period: a flask is the smallest box that lets the rammer work (Bolland p. 37). The game's small flask has a brick-walled box today; a plank box with iron straps is planned, so that it can be built before any iron has been cast.
  • Period: ram, turn, ram the cope, lift, draw, close. The game's ordinary flask keeps this as one rammed body and a single press of the pattern. The long flask has a real hinged cope, to be closed and wedged for work that needs it; ramming that cope stands for ram, lift, draw and close together. The long flask's cramps must be wedged first: poured unwedged, the cope lifts and the iron runs out at the joint, by Bolland's rule (Bolland pp. 97-99).
  • Period: a mould is poured hot and filled fast (Bolland pp. 79-80). The game keeps one check, a misrun: iron that arrives below the pattern's minimum pour temperature gives scrap for the whole cavity. A short pour keeps its full pieces and scraps the rest.
  • Period: a head is added to take the shrinkage and the dirt (Bolland pp. 75, 148). The game plans a head for bored work only, its metal coming back as scrap. Today no pattern carries one.
  • Period: the core is a separate article, made in a box and dried (Bolland pp. 124, 262-263). The game plans a core rammed in a wooden box, drying by itself for a day as a bow stave does, and spent at shake-out.
  • Period: facing mixed fresh and old sand reconditioned (Bolland pp. 303, 310). The game's own: sand is never used up. A flask takes sand once and shake-out leaves it full, so a casting costs labour and pattern life and no sand.
  • The game's own: gates, vents, gaggers and the joint's rough face are left out, and a gate is carved only in the sand bed. See vanilla against reality: the base game has no sand casting at all, only clay moulds.

The numbers

Casting amounts follow the density rule; the rest are the game's own, as the scaling page explains.

In the gamePeriod figureHow it was scaledGame value
Sand to commission a flaskNone givenThe game's own, sand reused for ever1 green sand, once
Impressions from a wooden patternNone givenThe game's own240, then re-cut from its diagram
Feeding head on bored work6 in on a cylinder of 6 ft (Bolland p. 148), about 8% of the lengthShare of the impressionPlanned about 8%, returned as scrap
Pour gate for every patternNo temperatures givenThe game's own: 100 C over the iron's melt, under the coolest cupola tap of 1339.6 C1300 C, against a melt of 1200 C
Hold-down on a cope450 lb on a 12 in cube under a 12 in head (Bolland p. 99)Bolland's rule, as a pass or failThe long flask's two wedged cramps hold it; unwedged, the pour runs out
Core dryingOven dried, no time read (Bolland pp. 52-62)The game's own, as bow stavesPlanned 24 in-game hours

Open, and what you can tell us

Bolland writes in 1892, for American shops, and this page draws only on him. If you know a source for how an English or Scottish moulder vented, weighted and mended a mould in the first half of the nineteenth century, and how it differed, we would like to read it. Bolland gives no temperature for a pour; if you know of one measured before pyrometers were common, the misrun gate could use it.

Picture credits