Predictive Modeling for Mechanical Strength in 3D-Printed ABS Polymer Specimens Using Machine Learning Algorithms

Abstract

The advancement of machine learning (ML) offers significant potential for optimizing mechanical properties in 3D-printed polymer materials, which are widely used in industries requiring durable and precise components. This study focuses on developing predictive models for estimating tensile strength in 3D-printed acrylonitrile butadiene styrene (ABS) specimens, fabricated through Fused Filament Fabrication (FFF) technology. Optimizing FFF parameters such as nozzle type, printing speed, filament color, and tensile coupon’s distance is crucial for enhancing strength. Using two Ender 3 Pro 3D printers, we produced 162 ABS dogbone specimens. One printer utilized a standard 0.5mm nozzle, while the other incorporated a novel extruder system capable of annealing each layer during printing. All specimens were printed at room temperature within an enclosed box to maintain consistent conditions. Predictor variables included nozzle type (enhanced vs. standard), printing speed (categorical), filament color (black vs. white), and the distance of the tensile coupon. We implemented multiple ML algorithms, including Decision Trees, Random Forests, Gradient Boosting Machine, Support Vector Regression (SVR), and Neural Networks, to develop the prediction models. The performance of each model was evaluated based on RMSE and R² metrics, allowing us to determine the most effective model for accurately predicting tensile strength. Ultimately, our study seeks to identify optimal FFF conditions for improved material performance, providing valuable insights that could contribute to advancing additive manufacturing practices and encouraging data-driven decisions in 3D printing processes.