Fraunhofer IWM makes peening and deep rolling of AM metal parts calculable, up to 40% more fatigue strength

On 4 August 2026 the Fraunhofer Institute for Mechanics of Materials IWM published a model-based strategy for designing mechanical surface post-treatments of additively manufactured (AM) metal parts. The method predicts how shot peening, deep rolling or burnishing change the surface layer. It then feeds that into a fatigue assessment based on the FKM guideline, which many German design offices and small firms already use.

The institute says the chain was validated on AlSi10Mg, 316L and Ti6Al4V made by laser powder bed fusion (LPBF) and Cold Metal Fusion (CMF). In model tests it reports up to 40% higher fatigue strength and a life extension of “mehrere Dekaden” (several decades, which on a logarithmic life axis we read as orders of magnitude in cycles). The calculations were checked by fatigue tests on specimens and components.

What the method does

The press release describes three linked steps:

Step	What is predicted or chosen
Component assessment	Which treatment suits the geometry and the location of the most highly stressed zone
Process simulation	Residual stress, roughness and work hardening as a function of parameters such as blasting pressure, contact pressure and track overlap
FKM-based life assessment	Fatigue life gain, with the FKM procedure adapted to the printed and treated materials

Fraunhofer IWM names the root causes it is attacking: surface roughness, tensile residual stress and defects such as gas pores and lack of fusion, which act as crack starters. The treatments densify the surface layer, reduce defect density near the surface and introduce compressive residual stress. The institute also notes that printed parts mostly fail from the surface, so a tailored treatment can partly compensate for scatter in the print process. It adds that effects can be estimated for new alloys from yield and tensile strength.

Why it matters for design engineers

The as-built surface is a notch. Roughness valleys on an unmachined AM surface behave like micro-notches, and near-surface porosity adds to them. In a hand calculation that shows up as a surface factor (Marin $k_a$, or the FKM roughness factor) and effectively as extra $K_t$ on top of the geometric one. At a fillet or bore where geometric $K_t$ is already 2 to 3, the combined knockdown can dominate the design. Start with the geometric term in our stress concentration factor calculator, then apply a separate surface or defect factor that you can justify with test data.

A 40% gain is not an allowable. The figure is a best case from model tests, not a statistically based design value. Compressive residual stress can relax under high mean stress, overloads or elevated temperature, and peening only helps where the media or roller actually reaches. Internal features and lattice struts often cannot be treated at all, so internal pores and unreached surfaces can become the new critical sites.

A calculable treatment gives you something to specify. If the benefit is tied to process parameters, the drawing or specification can call up those parameters and their verification.

If you	Check
Take credit for peening or rolling on an AM part	Which surfaces are treated, coverage or overlap, and a residual stress measurement (e.g. X-ray diffraction depth profile) on a witness part
Use as-built surfaces in a fatigue zone	Measured roughness (Ra, Rz or Sa) on the actual build orientation, and fatigue data taken on that surface condition, not on machined bars
Accept a supplier’s fatigue curve	Specimen surface state, HIP status, build orientation, porosity level (CT) and number of lots behind the data
Rely on treated surface in service	Operating temperature and mean stress versus residual stress relaxation, plus repair rules if the treated layer is damaged or machined
Have treated fillets or bores	Recalculate $K_t$ with the finished geometry, as rolling and burnishing change local radii and dimensions; check fits with our ISO 286 chart

An AM part’s fatigue life is set at its surface; a surface process you can model and verify beats a generic knockdown you hope is conservative.