Exploring How Cracks Grow in PLA With Dual Keyhole Notches: Blending Experiments, Simulations, and Smart Predictions

Abstract

Polylactic acid (PLA), a biodegradable polymer, is gaining attention as a sustainable alternative to steel in civil engineering, yet its fracture behavior under complex loading remains underexplored. This study examines crack propagation in PLA samples with dual keyhole notches under tensile loading, integrating experimental tests, finite element simulations, and machine learning predictions. Six PLA specimens (200 × 50 × 10 mm, crack length 10 mm, angles 60° and 70°, notch radii 0.5–2 mm) were tested experimentally, while 400 samples (angles 1°–80°, radii 0.5–4 mm) were simulated in ABAQUS. Artificial Neural Networks (ANN) and Long Short-Term Memory (LSTM) models, implemented in MATLAB, analyzed the results. Experimental peak stress reached 33.3 N/mm² (0.5 mm notch, 60°), while simulations predicted up to 65.225 N/mm² (0.5 mm, 1°). ANN with Bayesian Regularization outperformed other models, offering precise predictions of crack behavior. These findings provide a fracture criterion for PLA, advancing its potential in sustainable structural applications.