Adaptive Non-Intrusive Substructure Modeling Framework for Multiscale Nonlinear Damage Analysis of Large-Scale RC Structures

Damage in large-scale reinforced concrete (RC) structures evolves across multiple scales, from material microcracking to RC component damage and global structural failure. However, multiscale damage analysis of realistic large-scale RC structures remains challenging. To overcome this difficulty, this study proposes an adaptive hierarchical non-intrusive substructure modeling (NSM) approach that selectively builds separate refined substructure models for local critical regions that are automatically identified using predefined criteria, without altering the upper-level main structure model (i.e., non-intrusively). A correction-force-based coupling method and a gradual model-transition strategy are developed to ensure consistent main-substructure interaction and stable model replacement. To address the path-dependent nature of damage evolution in nonlinear regions, the substructures are initialized through state reconstruction using the response history of the main structure up to the current time. This non-intrusive strategy enables efficient parallel computing through limited data exchange, infrequent main-substructure interaction, and low synchronization overhead. Furthermore, to better simulate fracture behavior, peridynamic (PD) models are incorporated in this paper within the finite-element (FE) framework to leverage PD’s strength in simulating microcrack evolution. The proposed method is implemented in the open-source platform OpenSees and validated using a realistic 10-story RC hospital building subjected to seismic loading, demonstrating the feasibility of multiscale damage analysis for realistic large-scale RC structures. These results show that the proposed approach can simulate the evolution of material-level damage and its connection to macro-scale failure modes, thereby supporting early damage detection and assessment of potential structural collapse.