How Sand Casting Works: Eight Steps from Drawing to Finished Part

Jun 27th, 2025

How Sand Casting Works: Eight Steps from Drawing to Finished Part

Sand casting is eight steps. Each one has a single number that decides whether the casting is sound: shrink allowance on the pattern, moisture in the sand, ramming density in the mould, magnesium residual in the iron, pouring temperature, shakeout temperature, blast standard, and the inspection sample that signs it off.

Get those eight numbers right and the process is boring. Get one wrong and you spend the afternoon grinding.

This is the sequence as it runs in a plant that casts iron every day, not as it is described in a textbook. Where a step has a failure mode, it is named.

Step 1: Pattern, shrink and draft

The pattern is the part plus the metal's contraction plus the machining stock, minus the draft. Nothing else happens until those three allowances are fixed.

Shrink allowance for iron runs 0.8 to 1.0 percent for gray iron and 0.7 to 1.0 percent for ductile iron, applied as a scale rule rather than a per-dimension calculation. Ductile iron is not a single number because graphite expansion partly offsets liquid contraction, and a constrained section shrinks less than a free one. On a 300 mm dimension that spread is 0.6 to 0.9 mm. It is why the tolerance class has to be agreed before the pattern is cut, not after the first sample.

Machining stock is 2 to 3 mm on faces that will be cut. A pattern maker who adds 5 mm "to be safe" is adding cost to every piece for the life of the tool.

Draft is 1 to 1.5 degrees on a wooden pattern and 0.5 to 1 degree on aluminium or cast iron, and insufficient draft does not show up as a bad casting at all. It shows up as a torn mould wall on the drag side and a scab on the casting, and the moulder will blame the sand.

What fails here: wrong shrink direction on a dimension toleranced from a machined face. No draft on a deep pocket. A parting line placed where no core print can be supported.

Step 2: Gating and feeding

The running system decides how fast the metal fills and where the hot spots end up. It is designed before the first mould, and changed only when a defect proves it wrong.

Three numbers matter. The pouring time for a 5 to 20 kg iron part is typically 5 to 15 seconds. Shorter than that and the metal jets and erodes sand. Longer, and the thin sections freeze before the mould fills. The gate area follows from the pouring time and the metallostatic head. The riser has to stay liquid longer than the section it feeds, which in ductile iron means a modulus about 1.2 times what the feeding distance requires. Not a riser the size of the boss.

Ductile iron adds a fourth item. The riser must not feed through a thin section, because the graphite expansion in a sound mould partly compensates, and an oversized riser on a thin-walled part causes shrinkage elsewhere by pulling metal back.

What fails here: gas entrapment from a badly placed ingate. Sand erosion from a gate aimed at a mould wall. Shrinkage porosity under a riser that froze first.

Step 3: Sand, mixed and tested

Green sand is not a material you buy. It is a loop you maintain: silica sand, bentonite, water, and the fines and dead clay that accumulate. The loop is the process.

Sand propertyWorking range for ironTest frequencyConsequence if out
Moisture3.2 to 4.0 percentEvery shiftWet sand: blowholes and scabs. Dry sand: crumble, mould wall collapse
Compactability35 to 45 percentEvery shiftLow: weak mould. High: sand sticks, poor fill
Green compressive strength0.12 to 0.18 MPaEvery shiftLow: mould erosion. High: poor shakeout, hot tears
Permeability80 to 140 (AFS)DailyLow: gas defects at the mould wall
Loss on ignition3 to 6 percentDailyHigh: excessive gas, brittle mould
Compacted densityTracked, not usually specifiedContinuous on automatic linesLow: the single largest source of dimensional scatter

The last row is where automatic moulding lines earn their money and where they fail quietly. A DISA line squeezes sand to a density a hand rammer cannot approach, and consistency of that density across the whole mould face is what actually holds CT 9 to CT 11 on a production part. When density drifts, dimensions drift with it and no amount of pattern measurement explains the trend.

What fails here: blowholes and pinholes from high moisture and low permeability, sand inclusion from low strength, rat tails from over-wet sand against a hot pattern.

Step 4: Moulding, coring and closing

Moulding makes the cavity. Coring makes the internal shape. Closing puts them together, and closing is where tolerances are quietly lost.

Cores for iron are normally shell (resin coated sand) or cold box. A shell core holds its shape for months and gives a clean internal passage at Ra 6 to 12 um. A green sand core is cheaper and rougher, and has to be used the same shift. Core print clearance has to be tight enough that metal does not enter the gap and loose enough that the core does not crack on closing. That window is a few tenths of a millimetre, and it belongs on the core box drawing, not in the moulder's judgement.

Closing alignment on a vertically parted line is set by the machine. On a horizontal flask it is set by pins, and if the pins are worn the casting is offset and the wall thickness on one side drops. Offsets under 0.5 mm pass a visual check. They fail a wall-thickness check.

What fails here: core shift producing a wall that is thin on one side. Sand drop from an unsupported core. Mould wall damage from poor closing.

Step 5: Melting, treatment and pouring

This is the step with the most measurement and the least tolerance for improvisation.

Base iron is melted and its chemistry checked by spectrometer before treatment. For ductile iron, magnesium is added by sandwich or tundish method to a residual of 0.035 to 0.055 percent. A sample pour is checked for nodularity before the production moulds are poured. Nodularity of 85 percent is the usual floor. Pressure and structural parts often specify 90 percent.

Inoculation happens in the stream during pouring, not in the ladle. The effect decays in minutes. A part poured ten minutes after inoculation with no late inoculant has a different microstructure at the last mould than at the first.

Pouring temperature for gray iron is 1,320 to 1,400 C and for ductile iron 1,350 to 1,450 C, taken high in the range for thin sections and low for heavy ones. Temperature is measured with a disposable thermocouple at the furnace and again at the ladle, because the second reading is the one that actually fills the mould.

Pouring rate is controlled by the operator's eye on the sprue cup, which is a weak control. On critical parts we pour under a timed sequence and record it. A fast pour on a cored part washes the core. A slow one cold-shuts a thin wall.

What fails here: blowholes from wet sand meeting hot metal. Cold shuts from low temperature. Misruns in thin sections. Dross and slag from a poorly skimmed ladle. Degenerate graphite from magnesium fade.

Step 6: Solidification and shakeout

The casting stays in the mould until it is solid enough to move without cracking, and shakeout temperature above roughly 400 C is where distortion and cracking start, particularly on thin, long parts.

Ductile iron is the sensitive case. It is still transforming in the mould, and a casting shaken out hot can crack at a section change that looked fine on the bench. We hold heavy ductile parts in the mould longer than the metallurgical minimum for that reason. It costs flask turnover, which is a real production constraint when the line is full.

What fails here: hot tears at section changes. Distortion on long thin parts. Cracking found only after machining.

Step 7: Cleaning and finishing

Shot blast to Sa 2.5 for parts that will be coated. Gate and riser removal by cut-off wheel or press. Grinding of fins and parting line flash, then a 100 percent visual for surface defects.

Cleaning is the step where cosmetic repairs happen: a small sand inclusion is dug out and welded, and a porous patch on a non-critical face is filled with a compound that will not show under paint. Whether that is acceptable depends on the drawing, and a supplier who repairs without telling you is making a decision that belongs to your engineering department.

What fails here: cracks opened by aggressive gate removal, distortion from improper fixturing during machining setup, coatings applied over a badly blasted surface.

Step 8: Inspection

Inspection is not one gate at the end. On a production iron part it is four. Chemistry at the furnace. Nodularity and microstructure on a sample. Mechanical properties on a separately cast test bar. Dimensional and visual on the part.

Ultrasonic and magnetic particle testing are added for pressure and safety parts, while CMM is used for the first article and for the dimensions the drawing marks as critical, not for every part, because the cost would not survive the quotation.

What fails here: a part that passes dimensional and fails pressure, because porosity was never looked for on the section that matters.

The map from step to defect

Defects are not random. Each one points at a step, and knowing which step saves a day of arguing.

DefectMost likely stepFirst thing to check
Blowhole, pinholeSand or pouringMoisture, permeability, pouring temperature
Sand inclusionSand or closingGreen strength, mould wash, core support
Shrinkage porosityGating and feedingRiser modulus, hot spot location, pouring temperature
Cold shut, misrunPouringTemperature at the ladle, pouring time, section thickness
Core shiftCoring and closingPrint clearance, core print wear, closing pins
Flash and finningClosingParting line flatness, flask clamping force
Crack after shakeoutShakeoutShakeout temperature, section change design
Wrong nodularityTreatmentMagnesium residual, inoculation timing, cooling rate

Prototype and production are two different routes

A single casting for a prototype does not need the process a 20,000-piece order needs. Pretending otherwise wastes money in both directions.

For one to twenty pieces, a wooden pattern on a hand-moulded flask is the cheapest route. Tolerance is CT 12 to CT 13, machining stock is generous, and the part is good for fit and function checks, not for production approval.

For production, the pattern becomes aluminium or cast iron, the moulding moves to an automatic line, and the gating is designed rather than improvised. Same casting, different economics. The decision point is usually somewhere between 50 and 200 pieces, driven by pattern life rather than by unit price.

The mistake buyers make is ordering a production pattern for a design that is still moving, then paying for the pattern changes twice and losing a month on each one. The second mistake is entering production with a prototype pattern and discovering the dimensional scatter is too wide for the machining fixture.

Where the calendar actually goes

Tooling is one to three weeks for a simple pattern and first article is days after that, so the long items are usually not the foundry at all. A coating or plating subcontract, a heat treatment cycle for ductile iron, and the customer's own approval loop on the sample are what fill the calendar.

If a schedule matters, tell us the date the parts must be at your dock, not the date you want the quotation. That changes the sequence. It may be worth casting a sample from a temporary pattern to start machining trials while the production tool is being made.

FAQ

How long does sand casting take?

A single part can be poured within a week of pattern completion. A production run depends on the moulding schedule and the finishing steps; a coated or heat-treated part adds days after casting, not hours.

What metals can be sand cast?

Iron, steel, aluminium, copper alloys and most others with a melting point the sand can survive. Our own lines pour ductile and gray iron only.

What tolerance can sand casting hold?

CT 9 to CT 13 for green sand to ISO 8062-3, and CT 7 to CT 9 for shell moulding. Critical dimensions are normally machined rather than held as-cast.

How much machining stock should be left?

2 to 3 mm on a green sand casting and 1.5 to 2 mm on a shell casting for a machined face. More stock does not improve the result, it increases the weight you pay for and the time on the machine.

Why does my casting have a rough surface on one side only?

Usually a mould wall problem, not a metal problem. Over-wet sand against the pattern. A pattern surface that is worn or polished through. Mould wash applied unevenly. The side that is rough tells you which half of the mould to look at.

Can sand casting produce a pressure-tight part?

Yes, with a gating design that puts the feed where the section is thick, and with ultrasonic or pressure testing to prove it. Ductile iron valve and pump bodies are sand cast and pressure tested as routine.

What to send

A drawing with the material grade, the tolerance class marked per dimension, the machined faces identified, the annual quantity, and the date the parts are needed. Five items, and every one of them changes the quotation.

If you have a sample of the part you are replacing, send it as well. A machined sample tells us more about the wall sections than a 2D drawing usually does.

The green sand line page covers what the DISA line holds. Shell moulding covers the tighter end of the tolerance range, and the ductile iron overview lists the grades we pour. If your part is aluminium and the volume is high, read the die casting comparison before you go further, because sand may not be the right answer.

Product Reviews

0

nico@zyqfoundry.com

82b6bea3fa8b013700d5b8cb6263e9a
82b6bea3fa8b013700d5b8cb6263e9a

86-18561567088

86-18561567088