The structural performance of hollow reinforced concrete elements has gained increasing attention due to their material efficiency, sustainability benefits, and structural adaptability. This study presents a comprehensive data-driven assessment and numerical investigation of vertical bearing resistance in quadrilateral void-filled cementitious support elements, with a focus on square hollow reinforced concrete columns under axial compression. The research integrates empirical modelling, statistical analysis, and theoretical mechanics to evaluate the load-bearing capacity and failure mechanisms of such structural members.

A hybrid methodological framework is adopted, combining experimental findings from existing literature with numerical simulations and data-driven predictive modelling. Key influencing parameters—including wall thickness, reinforcement ratio, material properties, and geometric configuration—are systematically analyzed. The study further evaluates the applicability of design codes such as ACI 318-19 and Eurocode 2 in predicting axial capacity. Machine learning-based approaches are also examined to enhance prediction accuracy and capture nonlinear structural behavior.

The findings indicate that vertical load capacity is highly sensitive to geometric proportions and material heterogeneity. Data-driven models outperform conventional empirical equations in capturing complex interactions among variables. Numerical simulations reveal that failure mechanisms are governed by combined effects of local buckling, material cracking, and instability. While design codes provide conservative estimates, they often lack precision in accounting for void geometry and confinement effects.

The study contributes to the advancement of structural engineering by proposing an integrated predictive framework that combines statistical modelling, computational analysis, and theoretical principles. The results have significant implications for sustainable construction practices, enabling optimized design of hollow concrete columns with improved structural efficiency. Limitations related to data availability and model generalization are acknowledged, highlighting the need for further experimental validation and multi-scale modelling approaches.