New Ti-6Al-4V DED Study Shows One Residual-Stress Test Cannot Release a Repair

A study published online on 2026-08-19 in the Journal of Failure Analysis and Prevention tested how substrate thickness changes residual stress in Ti-6Al-4V deposited by laser-based directed energy deposition with powder feedstock (DED-LB/p). Across four substrate thicknesses, the researchers found tensile stress in the deposit of roughly 200–500 MPa and compressive stress in the substrate of about −100 to −200 MPa (paper).

The buyer-relevant result is not simply that a repair contains residual stress. It is that the stress exists at different length scales, and each verification method sees a different part of it. A surface X-ray result, a shallow hole-drilling profile and a through-section contour map cannot be treated as interchangeable certificates.

Thickness Changes Both Heat Flow and What the Instrument Sees

The team deposited plasma-atomized AMS7017 Ti-6Al-4V powder with a 45–75 μm particle range onto AMS4911 Ti-6Al-4V substrates. The tested substrate thicknesses were 3.2, 5, 13.7 and 17 mm. The samples were made with the same deposition settings so thickness could be isolated, although the detailed industrial process parameters remained confidential.

Thicker substrates increased thermal gradients and mechanical constraint. The deposit and interface carried tensile stress, while the substrate balanced it with compression and, in thicker samples, visible macroscopic bending. Thin substrates showed a larger microscale contribution associated with cooling rate and phase transformation; thick substrates shifted more of the response toward macroscopic stress redistribution.

This matters because “same alloy, same DED recipe” does not create the same stress state on every repair. A thin blade edge, a thick boss and a locally constrained pocket remove heat and accommodate contraction differently. Substrate geometry is therefore part of the process variable set, not merely the piece onto which material is added.

Three Methods Answer Three Different Questions

The researchers combined three techniques rather than choosing a single winner.

• Laboratory X-ray diffraction (XRD) was primarily sensitive to Type II and Type III microscale stresses near the measured surface. It captured microstructural differences relevant to local fatigue behaviour, but did not map the complete internal force balance.
• Incremental hole drilling (IHD) responded to Type I and Type II stress and resolved gradients below the surface. It is semi-destructive and useful near the repair zone, but its depth and local sampling volume remain limited.
• The contour method (CM) mapped Type I macroscopic stress across a cut section. It exposed internal redistribution and bending, but it is destructive and cannot be used as a routine final-part inspection on the exact component being delivered.

All three methods agreed on the broad pattern—tension in deposited material and compression in the substrate—but reported different magnitudes because they sampled different scales and volumes. That is not measurement failure. It is the physical reason a one-line “residual stress passed” statement is incomplete.

A Stress-Scale Release Map for Titanium Repair

For suppliers of fabricated titanium equipment and titanium sheet and plate, this map prevents a common category error. Nondestructive XRD may be the right production screen, but it cannot replace the destructive bulk method used during qualification. Conversely, a contour map from one development coupon does not prove that every production repair has the same surface microstress.

The Evidence Boundary Is as Important as the Result

The work used controlled samples, not repaired turbine blades or blisks in service. It did not establish a universal acceptable residual-stress limit, fatigue life, crack-growth rate or component-release criterion. The authors explicitly call for validation on representative industrial geometries and for repeatability studies of both the deposition process and the measurement methods.

The starting substrates also came from two material conditions: thinner samples were sheet, while thicker samples were machined from wrought stock. The paper discusses this microstructural difference, so thickness should not be read as the only possible cause behind every numerical difference.

The defensible conclusion is narrower and more useful. DED-LB/p repair creates a multiscale stress field whose balance changes with substrate thickness. A buyer should choose the verification method from the failure mode, required depth and destructive allowance, then preserve the correlation between development coupons and the released component. One instrument can screen one risk; it cannot certify the whole stress state.