PSI - Issue 83

C. Mallor et al. / Procedia Structural Integrity 83 (2026) 130–137

132

2.1. Experimental Setup The experimental prints were carried out on an Aconity platform, chosen because its open architecture makes it suitable for tightly coupled experimentation and model development. In the present study, the machine acts as both a manufacturing system and a benchmark test bench: it generates the physical prints that define the validation targets, and it provides the process context that the numerical framework must reproduce. The Aconity Mini 3D printer used in the present research for the metal AM is shown in Fig. 1.

Fig. 1. Aconity Mini compact LPBF 3D printer.

The selected benchmark is a stainless-steel twin-cantilever geometry derived from the type of calibration specimen often used in commercial AM simulation practice. The benchmark is simple enough to be measured and modelled with confidence, yet sufficiently sensitive to residual-stress release to reveal clear distortion trends. This leads to deviations between the intended and built geometries as shown in Fig. 2.

Fig. 2. Desired shape versus bult shape for a twin-cantilever build geometry.

The geometry for the use case involves a metal twin-cantilever structure as presented in Fig. 3. The part consists of two equal cantilevers joined at the centre and supported by multiple legs. The nominal geometry is 80 mm long, 7.2 mm wide, and 8 mm high, with legs 5.75 mm tall. A 3 mm support structure is printed beneath the specimen, while the build plate is modelled as a 20 mm thick substrate. The material for the twin cantilever part, the supports, and the build plate is stainless steel 316L. Commercially gas-atomized 316L powder from Carpenter [25], with a powder size distribution of 15-45 µm, is used. Printing is conducted under an argon atmosphere of 99.999% purity, with the oxygen content controlled below 500 ppm. This geometry is well suited to macro-scale distortion assessment because the deflection after partial release is easy to visualize, measure, and compare with model predictions.

Fig. 3. Geometry model general dimensions.