X-ray inspection for castings: how it works and when to use it
Radiography shows internal voids, gas porosity and inclusions without cutting the part. This page covers the mechanism, the source and detector choices, the practical limits of wall thickness and geometry, and the cases where an X-ray callout costs more than it finds. Written for design engineers and buyers who have to sign off a drawing note.

In this article
- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
What x-ray inspection for castings actually measures
A cast part sits between a source and a detector. X-rays pass through the metal and lose intensity along the path. The detector records what is left, so the image is a map of how much material the beam crossed. A shrinkage cavity or a gas pore removes metal from the path. The beam arrives brighter there and the void prints dark on film or as a high grey value on a digital panel.
This is a volumetric method. It sees inside a wall that no visual or dye penetrant check can reach. It does not care whether the void is open to the surface. That is the main reason drawings for pressure-tight or safety-critical castings carry a radiography note at all.
Dense inclusions behave the other way. Sand, slag or a trapped chunk of chill or tramp metal absorbs more than the surrounding alloy, so it prints light. Reading a radiograph is therefore a two-sided skill: dark for missing metal, light for extra metal. Both are defects, and both need a severity grade before anyone can accept or reject the part.
The output is a 2D projection of a 3D object. Every feature along the beam path stacks into one image. A pore near the source side and a pore near the detector side land in the same place. That single fact drives most of the practical limits described below.
Which internal defects radiography can find
The classic targets are shrinkage porosity, gas porosity, sand and slag inclusions, hot tears, cold shuts and core shift. Each has a signature. Shrinkage tends to be irregular and clustered near the last metal to freeze, often at a heavy section feeding a thin one. Gas porosity is rounder and more evenly spread, and it often appears in the same place on every part from a given die or mold.
Hot tears look like jagged dark lines, usually at a hot spot where a section change restrained contraction. Cold shuts appear as linear dark indications where two metal fronts met without fusing. Core shift is not a void at all: it shows up as a wall thickness change, so you need a reference feature in the image to measure against.
Detectable size depends on section thickness and technique. Under good conditions a void of roughly 1 to 2 percent of the penetrated thickness is a realistic reference. A 25 mm steel section supports detection in the 0.25–0.5 mm range; a 6 mm aluminum wall does far better in absolute terms.
That number is a reference, not a guarantee. Contrast falls as thickness rises, and scatter from a thick part blurs small indications. A drawing that calls for 0.5 mm sensitivity in a 60 mm steel wall is asking for a technique the physics will not deliver without a linear accelerator and careful collimation.
How the inspection process runs, step by step
Start with the drawing callout. It should name the standard, the severity level and the coverage: which zones get 100 percent, which get sampled, which get none. Without that, the inspection house sets its own criteria and the argument happens after the parts are made.
Source selection follows thickness. X-ray tubes in the 100–450 kV range cover light alloys and thin steel sections. Above roughly 50 mm of steel, gamma isotopes such as iridium-192 or cobalt-60, or linear accelerators in the 1–9 MeV range, become necessary. Lower energy gives better contrast in thin sections but needs longer exposure.
Focal spot size governs geometric sharpness. A small spot gives a sharper image, but the source is weaker and the exposure runs longer. For a fine-porosity callout on a thin wall, a microfocus tube is worth the cycle time. For a heavy steel casting, the spot size is a secondary concern next to penetration.
Then come the views. A single shot covers one projection. A complex part needs multiple angles so that no defect hides behind a rib or a boss. Each added view adds setup time, and that is where piece price comes from. Finally, the image is graded against reference radiographs from the named standard.
- 1Drawing firstStandard, severity level and coverage zones before any exposure.
- 2Source by wallTube up to about 50 mm steel; isotope or linac above that.
- 3Views by geometryOne projection per direction that can hide a defect.
- 4Grade, don't guessCompare against reference radiographs, not opinion.
Acceptance criteria come from a standard, not a handshake
ASTM E505 covers radiography of aluminum and magnesium castings. E446 covers steel castings up to 2 inches. E186 covers heavy-wall steel castings from 2 to 4.5 inches, and E280 covers heavy-wall steel above 4.5 inches. These standards publish reference radiographs graded by severity level, so a drawing callout translates directly into an accept or reject decision.
The practical value is repeatability. Two inspectors looking at the same image against the same reference plate should reach the same grade. Without a named standard, acceptance drifts between shifts and between suppliers, and a perfectly good casting gets scrapped on one inspector's opinion.
A callout that says only "no porosity" is not a specification. It is a wish. State the standard, the severity level, and the zones it applies to. If a machined surface will open a pore later, say so on the drawing so the foundry can account for stock removal in the section it pours.
Film, digital detectors and computed tomography
Film radiography remains the reference method. Fine grain, high contrast and a permanent analog record are hard to beat. The drawbacks are consumables, chemical processing and physical storage. A film of the required density must be filed and retrieved, and that archive cost is real on a long-life part.
Digital radiography with flat-panel detectors or computed radiography plates cuts cycle time to seconds per exposure. Images can be stacked, filtered and measured on screen. The trade-off is capital cost and a periodic detector calibration. For high-volume or high-mix work, the per-shot saving usually wins.
Computed tomography reconstructs a full 3D volume from hundreds of projections. That allows slice-by-slice porosity analysis, wall thickness mapping, internal dimension checks and reverse engineering of as-cast geometry. Voxel sizes commonly land in the 50–200 μm range on small-to-medium parts.
The CT trade-off is part size and cycle time. Dense or large castings may not fit the beam, and a full scan can take far longer than a handful of film shots. CT earns its cost when the question is dimensional or when the porosity has to be located in 3D rather than graded in 2D.
Practical limits: thickness, geometry and access
Geometry limits what any technique can deliver. Complex internal cores, ribs and intersecting walls superimpose structure onto the image. A defect can hide behind a thicker feature, and a thin wall next to a heavy boss is hard to expose in one shot. Multiple views or compensated techniques are the answer, and both add cost.
Sections that vary widely in thickness are the classic problem. An exposure set for the heavy section will wash out the thin one, and an exposure set for the thin one will saturation-black the heavy one. A step-wedge comparison or a multi-exposure stack handles it, but the technician has to plan for it before the part goes on the table.
Access matters too. The source and detector must sit on opposite sides of the region of interest. A closed internal cavity with no line of sight cannot be radiographed from that direction, and no amount of source power fixes it. Review the part geometry against the beam path early, before the drawing note is fixed.
Material density sets the penetration ceiling. Aluminum and magnesium are easy. Steel and nickel alloys need more energy, and thick sections of either push you toward isotopes or a linear accelerator. Titanium sits in between. The rule of thumb: if the wall is heavy enough that the exposure needs an accelerator, the inspection cost has stopped being a line item and become a design decision.
When radiography is the wrong choice
Radiography is the wrong call when the defect you fear is a surface defect. Cracks that break the surface are cheaper to find with dye penetrant or magnetic particle testing. A full radiograph of every part to catch a surface crack is a slow and expensive way to solve a problem that a ten-second penetrant check handles.
It is also the wrong call when geometry blocks the beam. If the critical zone has no source-to-detector line of sight, the exposure will show a busy image and prove nothing. For those parts, sectioning a first article or a CT scan on a sample is a better spend than 100 percent film on every unit.
And it is the wrong call as a substitute for process control. If porosity keeps appearing in the same location, the fix is in the gating, the die temperature or the fill time, not in more inspection. Inspecting a process that is drifting only tells you it drifted. Catch it at the source and the radiography bill drops with the scrap rate.
- 1Surface cracksPenetrant or magnetic particle testing is faster and cheaper.
- 2No beam pathSection a first article or CT a sample instead.
- 3Repeating porosityFix gating, die temperature or fill time first.
Process control alternatives to full radiography
Real-time X-ray on the production line is one option. A digital panel at the casting cell lets an operator catch a drift within a shift instead of after a batch. It costs capital, but it turns inspection into a control loop rather than a post-mortem.
First-article radiography plus periodic sampling is the cheaper pattern. Scan the first part of a run in full, confirm the process is set, then sample at a defined frequency. If the foundry holds its parameters, the sample rate can drop. If it does not, the sample rate was never the problem.
For dimensional questions, coordinate measuring machines and structured-light scanning often beat radiography on cost and speed. Use X-ray for what is inside the metal, and use metrology for what the drawing dimensions describe. Mixing the two tools wastes money on both.
Film vs digital panel vs CT for castings
Pick the detector by the question you are asking, not by the newest equipment on the floor.
| Criterion | Film | Digital panel | Computed tomography |
|---|---|---|---|
| Best for | Reference-grade 2D images | Fast 2D production screening | 3D porosity and wall mapping |
| Cycle time | Minutes per exposure | Seconds per exposure | Hours per part |
| Image record | Permanent analog film | Digital file, easy to share | Full 3D volume dataset |
| Part size limit | Large, flexible | Limited by panel area | Limited by beam and cabinet |
| Cost driver | Consumables and storage | Capital and calibration | Scan time and compute |
| Typical use | Code acceptance and archive | High-volume foundry line | First-article and R&D |
The verdict
If the risk is an internal void in a pressure-tight or safety-critical casting, specify x-ray inspection for castings with a named ASTM standard and severity level. If the risk is a surface crack or a drifting process, spend the money on penetrant testing and process control instead.
Frequently asked questions
What size of defect can X-ray inspection for castings realistically find?
Under good conditions, a void around 1 to 2 percent of the penetrated thickness is a realistic reference. That means roughly 0.25–0.5 mm in a 25 mm steel section.
Thicker walls and scattered radiation reduce that sensitivity. Ask the inspection house what it can hold on your actual section before you write a tighter number on the drawing.
Do I need film, or is a digital panel enough?
A digital panel is enough for production screening and most acceptance work, and it is far faster. Film still wins when a customer or a code requires a permanent analog record.
For first articles and dimensional questions, computed tomography gives more information than either.
Can X-ray find a crack in a casting?
Only if the crack has enough opening and orientation to change the beam path. A tight crack parallel to the beam is easy to miss.
Surface-breaking cracks are better found with dye penetrant or magnetic particle testing. Use radiography for internal volumetric defects.
How much does inspection add to piece price?
Cost is driven by part size, required coverage and view count, not by the method alone. One view on a small casting is a small adder. Full coverage of a large, complex part can dominate the piece price.
The cheapest saving is a tighter drawing callout that limits radiography to the zones that actually matter.
Does CT replace radiography for castings?
No. CT answers 3D questions: where the porosity sits, how the wall varies, whether an internal feature is in position. It is slow and size-limited.
Radiography remains the faster and cheaper tool for grading 2D projections against a reference standard.
Can GreatLight help with the drawing callout before casting?
Yes. We review the part geometry, the section changes and the machined surfaces that will open a pore later, then recommend a standard, a severity level and a coverage zone.
Quotation and DFM analysis come back within 12 hours, and an NDA is available on request.
Send the drawing, get a casting inspection plan
We review your casting geometry and tell you which zones need radiography, which need penetrant testing, and which need nothing at all. Quotation and DFM analysis within 12 hours.
12-hour quote100% inspection before shipmentNDA on request