Screw pull-out force predictions in porcine radii using efficient nonlinear µFE models including contact and pre-damage

Abstract

Nonlinear micro finite element (µFE) models have become the gold-standard for accurate numerical modeling of bone-screw systems. However, the detailed representation of bone microstructure, along with the inclusion of nonlinear material and contact, and pre-damage due to pre-drilling and screw-insertion, constitute significant computational demands and restrict model sizes. The goal of this study was to evaluate the agreement of screw pull-out predictions of computationally efficient, materially nonlinear µFE models with experimental measurements, taking both contact interface and pre-damage into account in a simplified way. Screw pull-out force was experimentally measured in ten porcine radius biopsies, and specimen-specific, voxel-based µFE models were created mimicking the experimental setup. µFE models with three levels of modeling details were compared: Fully bonded interface without pre-damage (FB), simplified contact interface without pre-damage (TED-M), and simplified contact interface with pre-damage (TED-M + P). In the TED-M + P models, the influence of pre-damage parameters (damage zone radial thickness and amount of damage) was assessed and optimal parameters were identified. The results revealed that pre-damage parameters highly impact the pull-out force predictions, and that the optimal parameters are ambiguous and dependent on the chosen bone material properties. Although all µFE models demonstrated high correlations with experimental data (R 2 > 0.85), they differed in their 1:1 correspondence. The FB and TED-M models overestimated maximum force predictions (mean absolute percentage error (MAPE) > 52%), while the TED-M + P model with optimized pre-damage parameters improved the predictions (MAPE <17%). In conclusion, screw pull-out forces predicted with computationally efficient, materially nonlinear µFE models showed strong correlations with experimental measurements. To achieve quantitatively accurate results, precise coordination of contact modeling, pre-damage representation, and material properties is essential.