A Bayesian–Hybrid Surrogate Framework for Viscoplasticity in Rocks and Heavy Construction Materials

Rocks and heavy construction materials may undergo time-dependent deformation that is partly recoverable and partly permanent. Under sustained or cyclic loading, this response can involve creep, rate-dependent yielding, hardening or softening, microcrack growth, pore-pressure effects, and eventual localization or failure. A useful constitutive model must therefore link stress, deformation, time, damage, and environmental variables without violating thermodynamic constraints.This paper presents a synthesis and a proposed modelling framework for viscoplastic rocks and heavy construction materials. It combines continuum mechanics,pressure-sensitive yielding, Perzyna-type overstress flow, internal variables for hardening and damage, and poromechanical effective stress. Bayesian inference isproposed to quantify parameter uncertainty and update predictions as laboratoryor field measurements become available. A hybrid surrogate strategy is outlined inwhich a physics-based constitutive model supplies the governing structure, whilea constrained surrogate approximates computationally expensive response components. The framework is intended for creep assessment, underground excavations,deep foundations, rock structures, and long-term infrastructure analysis. It is aresearch formulation requiring material-specific calibration and validation; it is nota universal calibrated law.