Types of Casting Defects and How to Catch Them Before They Ship

Nov 21st, 2025

Types of Casting Defects and How to Catch Them Before They Ship

Most defect lists are organised by metallurgy. That is not much help on a shop floor. A buyer standing next to a pallet does not need to know that a pinhole is a form of gas porosity. They need to know whether the part in front of them can be saved, and who pays for it if it cannot.

So this article sorts defects by the moment they appear: at the pouring line, after shakeout, after machining, and after pressure test. Where a defect surfaces tells you what caused it. More usefully, it tells you whether rework is even possible. Nine families cover most of what gets scrapped on our four lines, and the split between them is not academic. Roughly speaking, a defect caught at pouring costs a melt. The same defect caught after machining costs the casting plus the machining hours plus the delivery date.

Caught atTypical defectsReworkWho absorbs it
Pouring lineCold shut, misrun, sand washNo, remeltFoundry
After shakeoutSand inclusion, scab, rat tail, expansionSometimes, if outside tolerance zoneFoundry, before machining
After machiningSubsurface porosity, shrinkage, hard spots, wall breakthroughRarelyDisputed, contract-dependent
After pressure testLeak path, micro-porosity through wallNoDepends on agreed test standard

That split is the whole argument. Everything below fills in the judging rules.

At the pouring line: the cheap defects

These are visible before the sand is even broken out, and they are the only defects that cost almost nothing.

Cold shut. Two streams of liquid metal meet but do not fuse. You see a seam with a rounded edge, and under magnification there is an oxide film trapped in it. Ductile iron poured below roughly 1350 °C on a long thin section is where this shows up. Wall sections under 4 mm with a long flow path are the usual suspects. Thickness matters more than total part size. A 30 kg housing with uniform 8 mm walls pours more easily than a 2 kg bracket with a 3 mm rib.

Misrun. The metal simply stops before filling the cavity. Low pouring temperature, slow pouring, or a vent that cannot keep up will all do it. On our automatic vertical line the tell is a smooth, rounded, often oxidised leading edge rather than a jagged break.

Sand wash. Metal scours the mould surface and carries sand away, leaving a rough depression usually right downstream of the ingate. It means the sand surface strength was too low for the pouring speed, or the ingate was aimed at a weak spot.

All three are judged by eye, on the spot, and the part goes back to the furnace. On a 6000–8000 tonne annual shop that is a few minutes of labour and a few kilowatt-hours. Painful, not expensive.

After shakeout: where inclusion gets misdiagnosed

Once the sand is off, the surface defects appear. This is the stage where most arguments between foundry and buyer start, because two defects that look similar have completely different causes, and only one of them is the foundry's fault.

Sand inclusion versus slag inclusion. This one distinction saves a lot of email.

FeatureSand inclusionSlag inclusion
Cavity wallRough, granular, sand grains visibleSmooth, glassy, sometimes dark green or black
PositionBottom of cavity, dead corners, core shadowNear ingate, top surface, last-to-freeze zone
ShapeIrregular, follows sand pocketElongated, sometimes stringy
CauseMould or core strength, ramming, core shiftPoor slag removal, turbulence, low pouring temperature
Fix ownerFoundry processFoundry melt practice

If you can see sand grains under a 10× loupe, it is sand. If it looks like a piece of glass was pulled out, it is slag. Buyers who learn this one check stop paying for the wrong corrective action.

Sand inclusion is strongly process-dependent. Our shell mould line, where the mould is a cured resin-bonded shell with very low moisture, runs a visibly lower inclusion rate than the green sand line on the same part. That is not a sales claim, it is the reason shell moulding exists. The trade is cost: resin-bonded shell moulds carry a higher per-mould price, so they pay off on thin walls, fine surface, and parts where a 2 mm inclusion is a scrap event. On heavy, thick-section parts where a small inclusion sits outside the machined envelope, green sand is the sensible choice. Details of what each line suits are on the clay sand casting and shell mould casting pages.

Scab and rat tail. Both come from mould expansion. The sand surface lifts, metal flows behind it, and you get a shallow crust (scab) or a tail-shaped ridge (rat tail). High pouring temperature makes it worse. So does sand with too much fine material and too little refractoriness. These are cosmetic on non-machined faces and fatal on sealing faces.

Expansion defect at the mould wall. Usually visible as a local bulge. On a machined face it becomes an out-of-tolerance condition, which moves it into the next, more expensive category.

After machining: the expensive ones

This is where defect cost multiplies. By the time a defect appears here, the part has already consumed casting, fettling, and machining hours.

Subsurface porosity. You machine 0.5–2 mm off and a cluster of small holes opens up. Pinholes are typically under 2 mm; blowholes are larger and usually sit deeper. The mechanism is gas that could not escape through the sand: too much moisture in the mould, low permeability, a damp ladle, or insufficient degassing. The reason it hides is simple. It is not on the surface, so no visual inspection will find it, and the wall may still be thick enough at that depth to pass a dimensional check.

Shrinkage porosity. Spongy, irregular voids in the last region to freeze — the thermal centre. Ductile iron is worse here than grey iron, and the numbers explain why. Grey iron expands during graphitisation, which partly offsets liquid contraction; ductile iron's nodular graphite does the same thing less effectively because the nodules grow in a more isolated way. Practical linear contraction figures used for pattern allowance sit around 0.8–1.2% for ductile iron and 0.7–1.0% for grey iron. Get the feeder or the riser neck wrong, and you get a sponge in the heaviest section.

Unlike gas porosity, shrinkage is detectable before machining, if anyone looks. Ultrasonic testing will pick up the density change in a section thick enough to couple. Radiography shows it clearly but only in the direction it is shot. Which brings up a limit worth stating plainly: no single NDT method sees everything. RT misses tight planar defects unless the beam is aligned; UT struggles on thin, rough, as-cast surfaces; MPI only finds surface-breaking indications in magnetic material.

Hard spots and chill. A local hardness spike that destroys a cutter. In grey iron, a thin section that cools too fast can freeze as white iron instead of grey. In ductile iron, edge chill and carbides form the same way. A normal HT250 grey iron casting sits around 180–220 HB; a chilled edge can run past 300 HB in a small patch. The machinist finds it as a broken insert, not as a defect.

Wall breakthrough. The cutter breaks into a core cavity or through the wall. Either the wall thickness drifted (core shift, mould wall movement) or the machining stock was calculated on nominal dimensions that the process cannot hold. On parts with a 4 mm wall and a ±0.5 mm casting tolerance, there is not much room left.

After test: the ones that fail in front of the customer

Pressure-retaining parts get tested, and test is unforgiving.

Leak paths through porosity. A pump housing or valve body that passes dimensional and visual inspection can still weep under pressure. Interconnected porosity that never broke the surface becomes a leak path once the wall is thin enough. This is why pressure-retaining castings are usually specified with a test pressure and a hold time rather than a visual acceptance criterion. The specific test pressure belongs in the drawing or the applicable product standard, not in a blog post.

Dimensional drift after machining. Not a defect in the casting sense, but it fails the same way. It usually traces back to pattern wear, moulding sand variation, or a fixture that was set to nominal rather than to the actual casting.

How we actually judge a suspect part

Judging is a sequence, not a single check.

  1. Confirm the defect type by appearance and, where ambiguous, by a loupe or a section cut. Sand or slag is decided here.
  2. Locate it against the machined envelope and the drawing's acceptance zone. A 2 mm inclusion in a non-functional boss is not the same as a 2 mm inclusion on a sealing face.
  3. Choose the method that can actually see it. Surface-breaking on a magnetic part: MPI. Internal in a thick section: UT or RT. Leak path: pressure test. Each has a blind spot, and picking the wrong one produces false confidence.
  4. Decide: accept with concession, rework, or scrap. Weld repair is possible on many iron castings but it is not free and not always allowed — structural and pressure parts often prohibit it in the specification, and the heat-affected zone becomes its own quality question.

Where the defect type and the process correction are linked, the detail lives in the companion notes on preventing sand holes in iron castings and on what actually drives iron casting quality.

What we will not take on

Being clear about this saves both sides a month.

  • Ductile iron walls below about 2.5 mm on the green sand line. The process cannot hold it with any repeatability, and quoting it would be dishonest. Thin-wall work belongs in investment casting or die casting.
  • 100% radiography on high-volume parts with complex cores. We can do RT, but the inspection cost per piece often exceeds the casting price, and the schedule becomes an inspection schedule rather than a production schedule.
  • Castings above 200 kg on the automatic line. It can be poured, but the economics and the handling change, and it needs to be quoted as a separate job.
  • Anything specified only as "no defects." Acceptance needs a zone, a size, and a method, or the first shipment becomes a negotiation.

FAQ

What is the most common casting defect?

Sand inclusion on green sand lines, subsurface porosity on parts with machined walls. Which one dominates depends on the process, not on the material.

Can porosity be repaired?

Sometimes. Surface porosity on a non-critical face can be sealed or welded under an approved procedure. Interconnected porosity in a pressure wall usually cannot, because the repair does not restore the wall's integrity.

How do I tell sand inclusion from slag inclusion quickly?

Look at the cavity wall with a 10× loupe. Sand grains mean sand. A smooth glassy lining means slag. Position helps: sand sits at the bottom and in dead corners, slag sits near the ingate and on top.

Are black spots after machining always porosity?

No. They can be graphite nodules pulled out, small inclusions, or burnt-on sand. A spot that is perfectly round and 0.1–0.3 mm across on a ductile iron surface is often a pulled nodule, which is cosmetic unless the surface is a seal.

Why does the same drawing produce more defects at a different foundry?

Because the drawing specifies geometry, not process. Moulding method, sand system, gating, and pouring temperature are the foundry's choices. A part designed for shell moulding will run at a higher defect rate on green sand, and no amount of inspection fixes a process mismatch.

Can X-ray find every internal defect?

No. Radiography is directional. A planar crack parallel to the beam is nearly invisible. UT is better for planar defects in thick sections but needs a sound surface. The honest answer is that inspection coverage has to be designed around the defect you are worried about.

Before you send a drawing

If the concern is a specific defect you have been fighting, send the drawing with the failed area marked, the acceptance standard you are working to, and the quantities. We will tell you which of our lines can hold it and which cannot. That answer is more useful than a catalogue. For pressure-retaining work, the pump body and pipe fitting pages show the test and material specifications we normally quote against.

Product Reviews

0

nico@zyqfoundry.com

82b6bea3fa8b013700d5b8cb6263e9a
82b6bea3fa8b013700d5b8cb6263e9a

86-18561567088

86-18561567088