Non-destructive predictive modeling of traditional building materials for historic conservation and digital heritage applications

Abstract The conservation of historic buildings requires diagnostic tools that assess material performance without destructive testing. This study introduces a predictive modeling framework that integrates mechanical, thermal, and moisture-related properties across seven traditional construction materials: adobe brick, lime mortar, limestone, sandstone, marble, volcanic stone, and wood plank. Using data synthesized from twelve peer-reviewed studies (2015–2024), we applied Pearson correlation, regression, Principal Component Analysis (PCA), and hierarchical clustering to identify key relationships and material groupings. Results confirm that porosity strongly predicts compressive strength (R² ≈ 62%, p = 0.035), while density correlates with thermal conductivity (R² ≈ 85.5%, p = 0.003). PCA and clustering distinguished lightweight, porous, moisture-sensitive materials from dense, durable ones, offering a comparative classification tool for conservation planning. Unlike earlier works that examined materials or properties in isolation, this study systematically integrates multiple parameters into a single predictive framework with direct applications in heritage diagnostics, HBIM, and energy-efficient retrofitting. Future validation through field monitoring and HBIM integration will further enhance predictive accuracy.