Stochastic Emulators for Performance-Based Optimization: Enhancing Structural Resilience under Extreme Hazards

• Grant number: 25/13211-7
• Support Opportunities: Scholarships in Brazil - Post-Doctoral
• Start date: January 01, 2026
• End date: December 31, 2028
• Field of knowledge: Engineering - Civil Engineering - Structural Engineering
• Principal Investigator: André Teófilo Beck
• Grantee: Isabela Durci Rodrigues
• Host Institution: Escola de Engenharia de São Carlos (EESC). Universidade de São Paulo (USP). São Carlos , SP, Brazil
• Associated research grant: 24/10207-6 - Performance-based wind engineering: economy, safety, and reduction of carbon emissions in the built environment, AP.R

Abstract

Performance-Based Engineering (PBE) has become a leading methodology for assessing the structural response of systems subjected to natural hazards, as it enables the incorporation of multiple sources of uncertainty and stakeholder-oriented performance metrics such as life safety, repair cost, and downtime. When combined with optimization techniques, this framework evolves into Performance-Based Risk Optimization (PBRO), which seeks to identify structural designs that balance safety, cost, and performance under probabilistic constraints. However, despite the available solutions in the literature, the implementation of PBRO remains computationally intensive due to the need for repeated nonlinear dynamic analyses across a wide input space, often involving record-to-record variability and parameter uncertainty. Stochastic emulators have recently emerged as an efficient alternative to represent the output of such computational models while preserving their intrinsic randomness. Unlike traditional surrogate models that provide point estimates or confidence intervals around a mean prediction, stochastic emulators model the full distribution of possible responses for a given input, offering richer insights into structural behavior under stochastic loading. Despite their potential, no studies to date have integrated stochastic emulators into structural optimization frameworks based on the PBRO methodology. This research addresses that gap by proposing a computational framework that combines stochastic emulation with PBRO strategies to optimize structural performance while significantly reducing computational cost. The methodology includes the selection, calibration, and validation of stochastic emulators and their integration into risk-based and multi-objective optimization algorithms. Applications will focus on reinforced concrete and steel structures subjected to earthquakes and winds, and will also consider carbon emissions as a design variable to explore trade-offs between structural robustness and environmental impact. Expected outcomes include: (i) theoretical advancements in stochastic surrogate modeling, (ii) significant reductions in computational cost for high-fidelity simulations, (iii) enhanced robustness of the resulting structural designs under probabilistic performance objectives, and (iv) contributions to the integration of advanced optimization tools into engineering workflows. Ultimately, this research aims to support more resilient and efficient infrastructure design, aligning with current priorities in structural safety, environmental impact, and computational engineering. (AU)