Genetic Algorithm-Based Cost Optimization of Piled Raft Foundations on Soft Clay: A Performance-Based Geotechnical Design Approach

Abstract

Foundation engineering on soft cohesive deposits remains one of the most challenging tasks in geotechnical practice, where conventional pile group designs often produce excessively conservative and economically suboptimal solutions. This paper presents a comprehensive Genetic Algorithm (GA)-based optimization framework for piled raft foundations on soft clay, integrating nonlinear finite element analysis, probabilistic reliability assessment, and life-cycle cost considerations into a unified design methodology. A 15 m × 12 m piled raft system supporting a 5,000 kN column load on a 15 m thick soft Holocene clay deposit (Su = 20–40 kPa) overlying medium-dense sand is investigated. The optimization framework employs six design variables pile diameter, pile length, pile spacing, number of piles, raft thickness, and pile arrangement pattern within ten geotechnical and structural constraints derived from IS 2911 and IS 1904. Three-dimensional finite element analyses using PLAXIS 3D with the Soft Soil Creep constitutive model validate GA solutions against settlement, bearing capacity, and load-sharing criteria. The GA-optimized configuration achieves a 31% reduction in total foundation cost (₹14.8 lakh saving on the test case) compared to the conventional 4×4 pile group design, while maintaining FOS = 2.6 (above the 2.5 minimum), maximum settlement of 23 mm (below 25 mm limit), and reliability index β = 2.98 (target = 3.0). The raft carries 42% of total load, falling within the optimal 30–60% range reported for piled raft systems. Sensitivity analysis identifies pile length as the most influential design variable, followed by pile diameter and spacing ratio. The proposed framework is generalizable to other soft soil profiles and provides immediate practical value for routine geotechnical design practice in Indian urban construction.