Reliability-Based Design Optimization of Mass Timber Self-Centering Rocking Walls

Seismic structural optimization aims to achieve high-performance design by balancing seismic objectives with resource efficiency. Yet optimization of deterministic designs often produces solutions on the edge of feasibility, which may not be reliable under the uncertainties involved. This study addresses uncertainties in the design of self-centering mass timber rocking wall systems by proposing a system-level multiobjective reliability-based design optimization framework that adaptively updates hazard-consistent loading as structural properties evolve, enabling multilimit-state reliability assessment. To ensure performance objectives are met, various seismic design targets are considered, while a parametric finite element model is employed to simulate structural behavior for reliability assessment. A hybrid computational strategy uses a low-fidelity metamodel for deterministic constraints, while high-fidelity finite element model and adaptive seismic loading evaluate probabilistic limit states within the reliability assessment loop, improving accuracy and efficiency beyond existing single-fidelity approaches. The framework also includes an initial global sensitivity analysis to reduce dimensions and computation as well as Gaussian process regression to detect trends and quantify uncertainty. The framework is demonstrated on self-centering mass timber rocking wall systems, illustrating trade-offs between its objectives. The wall length is shown to be the most important input. Drift ratio and moment limit states at the design earthquake hazard level are identified as the most critical ones for near-optimal designs.