Stress concentration decoded: FEA-ML synergy overrides trial-and-error design in lightweight Mg-Al syntactic foams

Achieving ultra-lightweight magnesium-aluminum syntactic foams (MASFs) with retained compressive strength and energy absorption is hindered by severe property degradation at high hollow microspheres (HMs) volume fractions, primarily driven by microstructural instabilities. We present a synergistic multiscale framework integrating experiments, high-resolution finite element analysis (FEA) decoding the micromechanics of deformation and failure, and machine learning (ML) models trained on combined experimental and simulated data to establish predictive performance maps. FEA simulations quantify how increasing Ni-HMs fraction (30 → 60 vol%) shifts failure from uniform matrix yielding to localized stress concentration around pores and at microsphere contact points, leading to progressive collapse and explaining the observed 48.3 % strength drop. Nickel coating enhances load transfer initially but geometric constraints at high fractions dominate failure, limiting gains. ML models (SVR R² = 0.94/0.91) leverage FEA-validated microstructure-performance relationships to achieve high-fidelity prediction. Feature importance analysis confirms volume fraction and sintering temperature as key levers for microstructural control. This framework provides a physics-aware pathway to navigate the lightweighting-strength-absorption trade-off, enabling rapid identification of optimal processing conditions.