SCREW PULL-OUT FORCE PREDICTION IN PORCINE RADII USING COMPUTATIONALLY EFFICIENT NONLINEAR µFE MODELS Pia Stefanek (1), Juan Diego Silva-Henao (2), Andreas Reisinger (2), Dieter H. Pahr (1,2), Alexander Synek (1) 1.Institute of Lightweight Design and Structural Biomechanics, TU Wien, Austria; 2.Division Biomechanics, Karl Landsteiner University of Health Sciences, Austria Introduction In mechanical simulations of bone-implant systems, modelling the screw-bone contact interface and the pre- damage due to screw-insertion are major challenges. To maintain computational efficiency, the contact interface can be simplified using a fully-bonded interface or by applying a method developed by Steiner et al. [1], which is based on the deletion of interface elements under tension. Pre-damage is either not considered [2] or included in a simplified way by reducing the elastic modulus in proximity to the screw [1]. In this study, we test if a highly efficient materially nonlinear µFE solver (ParOSol-NL [3]) can predict the maximum screw pull- out force in porcine radii. Four different µFE simulation types were performed to test if accurate predictions were required including the simplified screw-bone contact interface and/or the pre-damage. Methods Ten cylindrical specimens (diameter and height ~20mm), were derived from porcine distal radii. A central pilot hole was drilled, and a 2.5mm locking screw was implanted. Using a custom-designed testing apparatus, cyclic tensile loading with increasing amplitude was applied until failure. Segmented µCT images were acquired from the unloaded, pre-drilled specimens. Images were cut in cuboids and resampled to a resolution of 36µm. Virtual screw insertion was done to obtain voxel-based µFE models (Fig. 1a). Linear-elastic material properties were assigned to the screw and a nonlinear damaged-based material model was used for the bone [3]. Material parameters were adapted for porcine bone according to [4]. Different simulations were performed using ParOSol-NL and compared (Fig. 1b): A fully-bonded bone-screw interface (FB), interfacial tensionally-strained element deletion (TED, see Steiner et al. [1]), TED with inclusion of a pre-damaged region around the screw (TED-P) and FB using adapted material parameters (FB-A). For TED-P a small cylindrical region around the screw was selected where the pre-damage was initialized with 0.9. The material parameters for FB-A were found by calculating the mean scaling factor for the FB results to match the experiments. Experimental and simulated maximum pull-out forces were compared. Results Simulations and experiments showed high correlations (R2 ≥0.84, Fig. 2). The root mean squared error was highest for FB (239N), lower with TED (143N), and lowest for TED-P (43N) and TED-A (44N). 1:1 agreements was reached for TED-P and FB-A models. Conclusion Experimental pull-out forces could be predicted with high correlations using ParOSol-NL. The FB model led to an overestimation of the maximum forces. Using a simplified contact model alone (TED) slightly improved the predictions, whereas adding pre-damage finally led to a good 1:1 agreement. FB-A revealed that a similar “tuned” prediction of maximum force could also be attained through the adaptation of the elastic modulus alone. However, this study is based on only one loading scenario and a rather small sample size. Further research is needed to fully determine the relevance of contact and pre-damage implementations for accurate pull-out force predictions in bone-screw simulations. References 1. Steiner et al, J Orthop Res 35:2415-2424, 2017. 2. Ovesy et al, J Mech Behav Biomed Mater 126: 105002, 2022. 3. Stipsitz et al, Biomech Model Mechano 19:861-874, 2019. 4. Chen et al, J Mech Behav Biomed Mater 65:644-651, 2017.