Casting defects sort into a small number of families, and the family tells you what physically went wrong: gas the freezing metal could not hold, contraction that feeding never replaced, metal that arrived too cold or too broken up to knit together, a mold that moved or gave way, and foreign material that ended up inside the metal. Every defect named below belongs to one of those.
Those families are not just vocabulary. They decide who owns the fix, and they decide what you can negotiate. ASTM E446, the reference radiograph standard for steel castings up to 2 in. (50.8 mm) in section, grades gas porosity, sand and slag inclusions, and shrinkage on five severity levels each; cracks and hot tears appear in it as single ungraded illustrations. Volumetric discontinuities are a matter of degree. Planar ones are not, and no acceptance level makes a crack acceptable.
1. Gas Defects
Gas defects are voids left by gas the metal could not hold as it froze. Solubility is much higher in the liquid than in the solid, so gas is rejected ahead of the advancing solid until the remaining liquid is supersaturated and bubbles form. In cast iron the two gases that do this are hydrogen and nitrogen.
The gas comes from the mold and the melt shop, not from the alloy: moisture, damp refractory, the mold atmosphere, the breakdown of clay, limestone and iron oxides, and the degradation of organic binders during pouring. Pouring is itself a source: faster filling means more turbulent contact between the stream and the mold atmosphere.
The margin between normal and defective is narrower than it looks. Usual nitrogen content in cast iron runs 20 to 80 ppm, and castings with nitrogen defects are reported at 110 to 150 ppm.
Porosity
Porosity is gas present as many small rounded voids dispersed through the regions that froze last. The walls are smooth and the shape is close to spherical, because a bubble in liquid takes the shape surface tension gives it.
In steel, dissolved gas content is normally very low, which makes porosity a reliable diagnostic rather than an ambiguous one: if a steel casting shows porosity, the working assumption is that liquid feeding was cut off, not that the melt was gassy. In aluminium the same evidence proves much less, because hydrogen porosity is a major factor in its own right and confounds the reading.
Blowholes
A blowhole is a single large cavity left by one trapped gas bubble, smooth-walled and usually rounded or pear-shaped. Where porosity is a population, a blowhole is an event.
They sit where gas collected and could not vent — under the cope skin, against a core, at the top of an enclosed section. The gas is mold gas rather than dissolved gas: wet or over-bonded sand, a core with no vent path, binder burning off faster than the mold can breathe it out.
Pinholes
Pinholes are clusters of small holes at or just under the casting skin, usually invisible until machining or shot blasting opens them up. They are the reason a casting can pass visual inspection and fail at first cut.
The mechanism is local rather than sectional: hydrogen and nitrogen come out of solution at the metal–mold interface itself, fed by moisture in the mold face and by oxides and hydroxides breaking down there. That is why pinholes track mold condition and coating practice rather than section thickness.
2. Shrinkage Defects
Shrinkage defects are the volume the metal loses on freezing, in the places feeding failed to replace it. Carbon steel to A216 WCB contracts 4.49 percent as it solidifies, and no gating change reduces that number — feeding decides only whether the missing volume ends up in the riser or in the casting.
Foundries locate the risk before pouring rather than after. Every casting simulation package outputs the Niyama criterion, a local measure of temperature gradient against cooling rate near the end of freezing, and it is used qualitatively — to mark the regions likely to contain shrinkage porosity, not to predict a percentage. Worth knowing when you review a supplier’s simulation report: the threshold below which porosity forms is established for steel and largely unknown for other alloys.
Shrinkage Cavity
A shrinkage cavity is one large void at the last point in a section to freeze, with jagged, dendritic walls. The rough interior is the diagnostic that separates it from a blowhole of similar size — gas leaves a smooth wall, starved feeding leaves the shape of the dendrite network that was there when the liquid ran out.
Cavities form where a heavy section is cut off from its feed path: an isolated boss, a junction that is thicker than everything around it, a heavy section sitting behind a thin one that froze first and closed the road.
Shrinkage Porosity
Shrinkage porosity is dispersed micro-voids formed through the mushy zone, where interdendritic flow could no longer keep up with contraction. The driver is a shallow temperature gradient: the pasty region is long, the pressure drop needed to feed through it is high, and at some point flow stops.
Chills are the counterintuitive case, and it is worth having straight. Modelling work on shrinkage prediction shows that shrinkage porosity rises with cooling rate, because both feeding velocity and the pressure drop across the mushy zone rise with it. A chill helps in spite of that, through the steep gradient it creates — the faster cooling it also causes works against you.
3. Filling and Solidification Defects
These are the defects of metal that arrived too cold, too slow or too broken up to knit into one body. They occur in every casting material and in every molding and casting process — the mechanism is premature solidification of the melt during the pour, not a property of one mold type.
The causes: pouring or mold temperature too low, pouring rate or metallostatic head too low, poor gas permeability in the mold holding back the front, and an alloy whose fluidity is being asked for more than it has.
Cold Shut
A cold shut is a seam where two flow fronts met and did not fuse. It shows as a rounded lap line on the surface, often with the edges visibly turned, and it runs into the section as an unbonded plane — a crack, mechanically speaking, that happens to be smooth.
It is a reject in most acceptance regimes because it is planar and because it is usually visible to the unaided eye. Anywhere two streams converge — around a core print, where two ingates meet, at the far side of a thin flange — is where to look for it.
Misrun
A misrun is a casting the metal never finished filling. The part comes out incomplete, with thin sections and the extremities furthest from the ingates missing, and the edges of what did fill rounded rather than sharp.
Misrun and cold shut are the same failure at two severities: the front lost the heat and the head it needed to keep moving. A misrun means it stopped; a cold shut means it stopped and another front reached it.
Cold Shot
Cold shots are separate globules of metal embedded in or near the casting surface, visible as beads that never merged with the body. They come from the stream breaking up during filling, not from the front freezing.
Once metal is entering above the velocity at which its own surface stays intact, droplets detach ahead of the main front. Any that skin over before the front catches them cannot fuse, and they are trapped where they landed. This makes cold shot a gating and pouring-practice defect: the cure is a filling system that keeps the metal below that velocity, not a hotter pour.
4. Cracking Defects
Casting cracks split into two families by when they form: while the mushy zone still holds liquid, and after the section is fully solid. That is not a fine distinction — they have different causes and different fixes.
Hot Tears
A hot tear forms where the mushy zone has been cut off from liquid feeding and is pulled in tension at the same time. Both conditions are required; either alone produces something other than a tear.
The tension comes from restrained contraction — cores inside a cylindrical casting, geometry that cannot move, a gating layout that ties the casting down while it shrinks. The initiation site is porosity that forms late in solidification along grain boundaries, which is why in steel castings the appearance of porosity and the appearance of tearing point at the same root cause.
They appear as cracks on the surface or internally, sometimes large enough to see and sometimes only findable by magnetic particle inspection. Once a steel casting has hot tears the options are weld repair or scrap, which is why the money is spent on design instead: experimental work on T-shaped steel castings by the Steel Founders’ Society of America concluded that filleted corners, smaller section transitions and an unrestrained casting and gating layout avoid them, by improving feeding and lowering the tensile strain at once.
Cold Cracks
Cold cracks form after the section is fully solid, when contraction stress exceeds the strength of metal that no longer has any liquid in it to accommodate movement. Everything the hot-tear mechanism depends on is over by then.
The practical consequences of the difference: a hot tear follows the interdendritic path and is usually ragged and oxidised, while a cold crack is a cleaner, sharper break that runs across grains and can appear after knockout, during handling, or during heat treatment. A hot tear is answered by fillets, feeding and a freer gating layout; a cold crack is answered by shakeout timing, handling and stress relief.
5. Mold and Sand Defects
Mold and sand defects come from the mold failing under the metal rather than from the metal itself — bonded aggregate erodes, expands, breaks down and can be wetted by the alloy. Every defect in this section is a sand casting defect in the strict sense; a steel die and a ceramic shell fail in their own ways, covered in sections 10 through 12.
Sand Inclusion
A sand inclusion is mold material that broke loose and was carried into the casting, ending as an irregular subsurface void lined or filled with sand. The void is the giveaway: it is angular rather than rounded, and it sits wherever the metal front deposited what it was carrying.
Radiographically the industry does not separate them from slag — ASTM E446 grades sand and slag inclusions together as one category on five severity levels. The causes of sand inclusions sit in mold strength, mold handling and the gating that decides how hard the stream hits the mold face.
Scab
A scab is a rough plate of metal raised above the casting surface with a layer of sand trapped underneath, formed where the mold face buckled and metal ran in behind the lifted sand. The mechanism is an increase in surface sand volume that has nowhere to go, so surfaces overlap.
The temperature everyone quotes here is usually quoted backwards. Bonded silica sand expands as it heats until the alpha-to-beta quartz inversion at about 1,063 °F (573 °C); the American Foundry Society’s own testing on sand expansion describes that point as where the sand stops expanding and starts losing volume, regardless of binder type. What produces the defect is the differential between a mold face that has passed the inversion and the sand directly beneath it, which has not.
Cuts and Washes
Cuts and washes are erosion scars: the stream tore sand out of the mold, so the casting carries a rough excess-metal ridge where the sand used to be. They are directional, and they point back along the flow path to the gate that caused them.
Two conditions produce them — stream velocity high enough to abrade the mold face, and sand not strong enough at that point to resist it. Whatever sand left also has to go somewhere, so cuts and washes usually come with sand inclusions further downstream in the same casting.
Metal Penetration
Metal penetration is liquid metal entering the pores between sand grains, leaving an adhesive sand crust on the casting that cannot simply be brushed off. The working criterion is geometric: penetration is present once the roughness depth reaches or exceeds the average grain radius of the molding aggregate.
It is a pressure balance at the metal–mold boundary. Metal enters the surface pores under metallostatic pressure until interfacial tension, wetting behaviour and capillary forces balance it, so anything that raises the pressure or lowers the resistance drives it: coarse grain with a wide size distribution, insufficient binder, high mold compressibility, poor gas permeability, tall heads and high pouring temperatures.
Burn-On
Burn-on is a thin layer of sand fused to the casting skin, sitting one step short of penetration on the same scale — the metal has bonded to the grains rather than filled the spaces between them. It survives shot blasting and has to be ground off.
It concentrates where thermal load is highest and holds longest: heavy sections, inside corners, the drag face under a tall head. For the buyer the practical cost is not the grinding but the inspection — a burnt-on layer hides the surface it covers until it is removed.
6. Dimensional and Mold-Alignment Defects
These defects leave the casting sound and complete but put the metal somewhere the drawing does not allow — halves offset, core moved, part distorted. Radiography and magnetic particle inspection will not find any of them; layout and CMM will.
Mismatch or Mold Shift
Mismatch is a lateral offset between cope and drag, showing as a step at the parting line and as wall thickness that is heavy on one side and light on the other. It matters most where the parting line crosses a machined feature or a wall with little stock to spare.
The question that actually gets asked is where normal variation ends and a defect begins, and it depends on the molding process. The Steel Founders’ Society of America publishes a suggested mold parting allowance of +.010 in. (.25 mm) for shell molded castings, and its tolerance grades come with an explicit caveat — they are a suggested starting place for agreement between manufacturer and customer, not a specification. Across the common molding methods, shell holds the tightest tolerance, no-bake next, green sand last.
Core Shift
Core shift is a core that moved from its intended position, producing walls thick on one side and thin on the other around a cored feature. Buoyancy in the liquid metal and clearance in the core prints are the two things it comes from, and the thin side is where it turns into a leak or a wall that machines through.
Core construction sets what is achievable before anything moves. A one-piece core holds tighter tolerances than split cores glued together, which may be slightly out of round; SFSA’s tolerance work adds an extra .03 in. allowance for split cores to cover the parting line between the glued halves.
Warpage
Warpage is distortion of the casting as a whole — a flat face that is no longer flat, a bore no longer parallel — caused by non-uniform contraction against restraint after the metal is solid. Unlike mismatch and core shift, nothing was in the wrong place when the metal froze; the part moved afterwards.
It is normally compensated rather than eliminated. The tendency to warp is one of the factors that sets machining allowance in the first place, alongside casting size, shape and surface condition, so a part known to move gets stock to move into.
7. Surface Defects
Surface defects leave a casting that is dimensionally correct and internally sound but whose skin is not acceptable. Whether it is acceptable gets settled by eye: ASTM A802, the surface acceptance standard for steel castings, grades nine classes of surface condition by visual comparison against graded reference comparators rather than by any measured roughness value — worth knowing before an Ra number goes on a casting drawing.
Rough Surface
Rough surface is roughness on the scale of the sand grain itself, and grain size is the direct cause: coarser sand fractions give rougher castings. It concentrates at the sections and spots that carried the highest thermal load.
Rough surface and metal penetration are the same physics at two severities. Below one threshold the metal only begins to enter the surface pores of the mold wall, which reads as roughness; above a higher one it enters the gaps between grains, which reads as penetration. The lever for both is the same — finer, better-bonded sand and a coating that keeps the metal out.
Rat Tails and Buckles
Rat tails and buckles are sand expansion defects: the mold face expands, fails and folds, and the casting records it as a fold line or a broad depression. ASTM A802 classes them under exactly that name, sand expansion discontinuities, which is a useful thing to be able to point at in a dispute.
Drop
A drop is sand that fell from the cope into the cavity before or during pouring. It leaves two marks on one casting: a raised, irregular patch where the metal filled the void left in the cope, and a hard inclusion wherever the fallen lump ended up.
It is a mold-strength and mold-handling defect, not a metal defect — weak cope sand, an unsupported overhanging section, rough closing, vibration during transport to the pouring line.
8. Inclusion Defects
Inclusions are non-metallic material held inside the solidified metal: sand, slag, refractory from ladle and runner, and oxide generated during the pour. They are the one defect family that gets into the casting from outside rather than forming inside it.
Both acceptance routes treat them as a graded quantity, and this is where the negotiating room is. ASTM A802 covers nonmetallic inclusions as one of its nine visual levels, and E446 grades sand and slag inclusions on five radiographic severity levels — so an acceptance level for inclusions is a normal thing to specify, unlike an acceptance level for a crack.
What makes inclusions worth more attention than a cleanliness statistic is that they seed other defects. Shrinkage pores nucleate on pre-existing bubbles associated with oxides and other inclusions, and gas bubbles nucleate preferentially on oxide inclusions because the energy barrier there is far lower than in clean liquid. Two independent lines of work converge on the same point: cleaner metal is not only cleaner, it is less porous.
9. Oxide and Film Defects
Oxide films are a separate family from inclusions because they are not carried in from outside — they are manufactured during the pour, by the metal’s own surface, and they are planar rather than lumpy. That geometry is the whole problem: a lump is a stress raiser, a film is a crack.
Oxide Films
An oxide film defect, or bifilm, is a doubled-over oxide film with a gas-filled crevice between its two faces, folded into the metal during filling. Because the two faces are unbonded oxide, the result behaves as a pre-existing crack of near-zero thickness inside a casting that never cracked.
It has one entry condition, and it is a velocity. Above roughly 0.5 m/s — a figure published work on bifilm behaviour gives as applying to most liquid metals, not only aluminium — the surface film folds over onto itself and is submerged with air trapped inside. Below it the surface stays intact and no bifilm is generated.
The damage shows up in the mechanical data before it shows up in an inspection report. Bifilms lower tensile and fatigue properties and, more tellingly, increase their scatter, which a buyer reading test certificates experiences as lots that are inconsistent for no visible reason. They also initiate hydrogen porosity and nucleate iron-rich intermetallics.
This is why filling velocity is the master variable in a running system. The same turbulence that folds films into the metal also raises dissolved gas pickup during filling — one root cause, two defect families, and one design decision that addresses both.
10. Die Casting Defects
Die casting’s own defects are gas porosity, soldering, die erosion and flow lines, and they follow from two conditions: the die is steel and permanent, and the metal is injected far above the velocity at which a liquid metal surface stays intact. Neither condition exists in a sand mold, so neither do these defects.
Gas porosity is the structural consequence of the second condition. A 2024 study of injection velocity in die-cast aluminium measured average porosity rising when the fast injection rate was raised from 1.0 to 1.5 m/s, because a faster flow front breaks up more finely and folds in more air. High pressure die casting accepts that trade knowingly — entrained air and oxides in exchange for filling thin sections before they freeze — which is why entrapped gas that expands during a solution treatment can raise blisters, and why heat treatment of die castings is treated as a problem rather than a routine step.
Soldering is the consequence of the first condition, and it is not adhesion. Molten aluminium reacts metallurgically with the steel die, growing intermetallic layers at the interface that hold the casting to the die face and tear the surface on ejection.
The rest of the permanent-die set follows from the die being a wearing tool. Die erosion is progressive loss of die surface where the incoming metal impinges, and it prints as rising roughness and drifting dimensions across a campaign rather than as a defect on one part. Flow lines are surface streaks marking where partially chilled fronts met without fully merging — cosmetic on a bracket, disqualifying on a visible housing.
11. Investment Casting Defects
The dominant defect source in investment casting is the ceramic shell itself, and the moment of highest risk is de-wax rather than pouring. Research on shell cracking reports that cracking affects roughly 20 percent of the moulds built, with estimates of how many of those cracks originate in the de-wax autoclave running from 80 to 90 percent.
The mechanism is a pressure vessel working in reverse. When autoclave steam heats the assembly, the wax expands faster than the ceramic around it and pressurises the shell from the inside.
Geometry decides which patterns crack. Overhangs where wax pools during de-waxing, long thin sections with ceramic cores that restrict the wax’s escape route, and sharp edges that concentrate stress are the three that recur — which is why a design review before tooling is worth more here than inspection afterwards.
For the buyer the consequence is rarely a cracked shell, because that is caught in the foundry. It is what a crack produces downstream: rework, metal run-out, extra machining, distortion, and inclusions in the casting as shell material enters the metal. The mitigation carries the same cost — venting the shell to relieve de-wax pressure gives the wax more ways out, but it also gives the shell more ways in, so it raises inclusion risk in the cast.
Wax residue belongs to the same chain. Wax that penetrates the shell wall or remains in the cavity has to be burned out before pouring; anything left behind is carbon and gas at the metal–shell interface, which is why the burnout cycle, not just the de-wax cycle, is part of the defect discussion.
12. Gravity Die Casting Defects
Gravity die casting inherits its defects from both neighbours: it fills under gravity like a sand casting, and it fills into a permanent steel die like a die casting. Nothing on this list is unique to it, and that is the point — it gets the filling defects of one process and the die defects of the other.
From the gravity side come the filling and film defects. Metal entering above the roughly 0.5 m/s threshold at which a liquid metal’s surface folds in on itself entrains oxide films exactly as it does in a sand mold, and because the steel die chills the metal much faster than sand does, cold shuts and misruns stay live risks in thin sections that pressure-driven die casting would fill through. Non-metallic contamination is the cause specifically named for cold shuts in this process.
From the die side come the interface defects. The die is steel and the alloy is usually aluminium, so the soldering reaction and progressive die erosion apply here as they do in high pressure die casting, without the pressure that masks some of their symptoms.
Shrinkage behaves differently than in sand, and it is worth flagging where the two get confused. The thresholds used to predict shrinkage porosity are sensitive to casting conditions and specifically to sand mold versus steel mold, so feeding rules carried over unchanged from a sand casting will not transfer — the die’s much steeper gradients change both where the last liquid sits and how long the feed path stays open.