Relative Density (Per Cent of Theoretical)

Divide the density of the part you built by the density the same alloy has when it is fully solid, and quote the answer as a percentage. It is the number every parameter-development campaign is trying to move, the number that appears in the abstract of every process paper, and the number most likely to be believed further than it deserves.

The arithmetic is trivial. The interesting part is entirely in what the number cannot say.

The arrangement of the porosity matters far more than its amount. Two parts both reported at 99.5 per cent can be quite different components. Round gas pores — from keyhole collapse, or from argon dissolved in the powder during atomisation — are relatively benign under static load and are largely closable by hot isostatic pressing. Flat, elongated lack-of-fusion voids, left where adjacent tracks or successive layers failed to overlap, are already crack-shaped, lie in planes, and behave like cracks in fatigue. A part at 99.7 per cent riddled with lack-of-fusion defects can have a shorter fatigue life than one at 99.2 per cent with spherical pores. A single density figure cannot distinguish them. A polished cross-section or an X-ray CT scan can, and when the number matters those are the arbiters.

Archimedes measurement has a specific blind spot and it runs in the flattering direction. The method finds the part's bulk volume by displacement, so any porosity that is open to the surface fills with water and gets counted as though it were metal. A part with an interconnected pore network can therefore report a density the method has no way of questioning. Meanwhile the as-built surface roughness traps air bubbles, which pushes the answer the other way, and the water temperature has to be corrected because its density moves about 0.02 per cent per degree. This is why a density claim without its method is not a claim, and why serious specifications name the technique.

The reference density is not one number either. The theoretical density of an alloy moves with composition inside its specification range, with which phases are present, and between as-built and heat-treated conditions — a martensitic as-built structure and an annealed one are not the same density. A relative density computed against a handbook figure for wrought bar is only as good as that assumption, and quoting five figures against a two-figure reference is a familiar way of appearing more precise than the data allows.

Finally, the sample. Density is usually measured on a small coupon, and a coupon does not necessarily represent a large part with thick and thin sections, downskin surfaces, contour regions and overhangs, all of which have their own local density. The bulk of a part is scanned with the hatch parameters; its skin is scanned with contour parameters; and the two are not equally dense. Density on a 10 mm cube is a parameter-development number, and it should not be quoted as though it were a statement about the component.