This study presents a robust methodology for calibrating the Hardening Soil– Brick (HS-Brick) constitutive model using both laboratory and field tests. Model parameters are derived and verified against a benchmark problem involving a spread footing founded on overconsolidated sandy soil in Texas, based on large-scale experiments by Briaud and Gibbens. The results show that wide variations in confining stress complicate direct parameter estimation from triaxial tests, necessitating the use of global optimization methods. To reduce
computational cost, a metamodeling approach based on Latin hypercube sampling is employed, enabling efficient surrogate predictions that serve as the computational engine for evolutionary algorithms. The proposed framework provides accurate and computationally efficient settlement predictions and is readily adaptable for reliability-based geotechnical design applications.
