PSI - Issue 42

Theodoros Marinopoulos et al. / Procedia Structural Integrity 42 (2022) 903–910 T. Marinopoulos et al./ Structural Integrity Procedia 00 (2019) 000 – 000

906

4

2.3. Numerical framework A dedicated FEA model was developed in ABAQUS software and used to simulate the experimental loading conditions. The original socket geometry was imported and meshed using first-order tetrahedral elements. The residual-limb manikin ’s geometry was imported and meshed with the same elements. The load-application levers were introduced as square and cylindrical beam elements reproducing the experimental setup. All materials were assumed to be homogeneous and linear elastic. The Young’s modul i of 0.7 GPa and 2 GPa were used for PLA and plaster, respectively. The respective parameter for steel and aluminum was assigned as 210 GPa and 69 GPa. The components were assembled to match the testing configuration seen in Fig. 2a. The bottom lever was fixed with regard to all degrees of freedom at point A and a load of 8000 N was applied at point B on the top lever. The points of the hollow cylindrical surface of the manikin were kinematically coupled to the end of the top lever to recreate the load transfer between the mandrel and plaster. 2.4. Redesign using FEA results The stresses on the outer surface of the original socket were calculated. The values of von Mises stress were normalized for the nodes on the outer surface with the respective maximum value. The obtained dimensionless magnitudes were multiplied for each node with the maximum additional thickness of 3 mm, and the nodes were moved accordingly in the horizontal direction. This changed the constant 3 mm thickness of the socket to a variable thickness of 3 – 6 mm, with the maximum thickness being in the area of the highest stress concentration. This process was repeated for maximum levels of thickness of 4.5 mm and 9 mm, creating 3 new designs in total (Fig. 1c). From these designs, the one with 6 mm thickness was selected to be tested experimentally. Five samples were printed using the same AM parameters as for the original design. All designs were used in numerical simulations to replicate static load tests. 3. Results 3.1. Experimental results Mixed results were found for the tested 3D printed PLA sockets. The original sockets printed in aPLA demonstrated a mean maximum load-bearing capacity of 2050 N with all samples failing to meet the standard-based requirements of 3890 N. On the other hand, the original sockets printed in bPLA exhibited a mean load-bearing capacity of 7500 N, with all the samples successfully passing the requirements. A significant variation was found in the mechanical performance of bPLA AM sockets, with a minimum of 4950 N and a maximum in excess of 10000 N. In the latter instance the socket withstood the load without visible failure. A significant variance was also observed for redesigned sockets printed in aPLA. More specifically, a mean load bearing capacity of 3750 N was found. Although the mean was lower than the instructed requirements, some redesigned samples performed successfully, with a maximum of 5800 N. Closer performance was observed fror redesigned bPLA sockets. All the samples successfully passed the requirements, with a mean load-bearing capacity of 6300 N, in a range between 6000 N and 7000 N. The results for the ultimate strength tests are presented in Fig. 3. Comparing the average weight of the printed sockets, a 8.84% difference was found for the original sockets, with bPLA being heavier. The bPLA redesigned sockets were also heavier, by 4.94%. They performed better than the aPLA ones, demonstrating a threefold improvement in load-to-weight ratio for the original design and a 50% improvement for the redesigned sockets. The mean weights and load-to-weight ratios are given in Table 1. Table 1. Weight and load-to-weight ratios for original and redesigned (with 4.5 mm) sockets printed in aPLA and bPLA. aPLA bPLA Weight (g) 202.39 253.29 220.29 265.81 Load/weight (N/g) 9.93 15.99 33.05 24.09