Computational and experimental investigation of microcapsule-based self-healing polymers: Macro-mechanical finite element modeling and artificial neural network predictions

Abstract

This study presents a comprehensive computational and experimental analysis of self-healing polymers containing multi-core microcapsules. A macro-mechanical finite element model (FEM) was developed to evaluate healing efficiency, incorporating homogenized and impacted regions simulated in Abaqus-Explicit with VUSDFLD subroutines to capture stress–strain responses. An artificial neural network (ANN) was employed to predict the effects of microcapsule volume fraction (VF) (5–10 %) and nanoparticle reinforcement (multi-walled carbon nanotubes (MWCNT), nanoclay, nanosilica) on tensile strength and healing performance. The FEM demonstrated high accuracy (<5% error) compared to experimental data, outperforming multi-scale models despite higher computational costs. The ANN revealed that increasing microcapsule content enhances healing efficiency, while nanoparticle reinforcement reduces it due to restricted healing agent release from mechanically stronger capsules. SEM analysis identified three healing mechanisms: localized healing near ruptured microcapsules, crack-path filling, and wide-crack bridging. Validation against experimental and multi-scale data confirmed the model’s reliability for optimizing self-healing material design. Key findings indicate that 10 vol% microcapsules maximize healing efficiency, though nanoparticle reinforcement trades mechanical strength for reduced self-repair capability. This work advances computational tools for autonomous healing materials, providing a framework for balancing mechanical performance and self-healing functionality in structural applications. © 2025 The Author(s).