Research

We specialize in seismic analysis and design of underground structures, focusing on the inelastic response of reinforced concrete and nonlinear soil behavior. Using high-fidelity models with erosion and fiber elements in Abaqus, LS-DYNA, DIANA FEA, OpenSees, FLAC, and PLAXIS, we collaborate across structural and geotechnical disciplines.

• High-fidelity inelastic simulation of RC underground structures using fiber and erosion elements
• Damage evolution of cut-and-cover and bored tunnels
• Development of fragility curves for underground structures
• Machine learning-based seismic performance prediction

A central focus for over two decades, our site response research addresses the propagation of seismic waves through soil deposits and bedrock formations. Funded by NRF and KHNP, with global collaborations including UIUC and UCLA, we develop both equivalent-linear and fully nonlinear ground response analysis frameworks tailored for the Korean geological environment.

• High-fidelity equivalent-linear and nonlinear site response analysis algorithms
• Probabilistic site response analysis incorporating parameter uncertainty
• Shallow bedrock amplification models for the Korean Peninsula
• H/V spectral ratio-constrained site amplification models
• Machine learning-based seismic site amplification prediction

Artificial intelligence and machine learning techniques are applied using data obtained from field measurements and rigorously conducted site response and numerical analyses. By integrating physics-based understanding with data-driven approaches, we enhance the interpretation and prediction of seismic behavior of soils and underground structures.

• Physics-informed neural networks for seismic site response prediction
• Data-driven surrogate models trained on high-fidelity numerical simulations
• Machine learning-based classification and regression for liquefaction triggering
• AI-enhanced interpretation of field measurement and monitoring data
• Hybrid physics–data frameworks for efficient seismic assessment of geotechnical systems

We develop probabilistic seismic hazard analysis (PSHA) procedures tailored for the Korean Peninsula, addressing epistemic and aleatory uncertainties in seismicity catalogs and ground motion models. Through iterative catalog-based methods and logic-tree frameworks, our research provides the foundation for national seismic hazard maps and site-specific design spectra for critical infrastructure.

• Regional ground motion models (GMMs) accounting for unique site amplification effects
• Machine learning algorithms for GMM development using Korean and Japanese datasets
• Generation of regional uniform hazard spectra (UHS) for design applications
• Iterative seismicity parameter calibration for Korean national hazard maps
• Logic-tree uncertainty quantification for source and attenuation models

We develop and validate liquefaction assessment methods through both physical testing and advanced numerical simulation. Using cyclic simple shear and centrifuge test data, in collaboration with the University of Illinois at Urbana-Champaign and the University of Naples, we establish robust procedures for evaluating liquefaction triggering and post-liquefaction behavior in diverse soil conditions.

• Accumulated stress-based excess pore water pressure model development
• Development of normalized liquefaction resistance curves from laboratory data
• Effective stress analyses for site-specific liquefaction potential evaluation
• Centrifuge model testing for validation of numerical liquefaction predictions
• Magnitude scaling factor development tailored for Korean seismicity

We evaluate the seismic performance of natural and engineered slopes using dynamic nonlinear finite element and finite difference models. This research is essential for transportation infrastructure resilience, encompassing highway embankments, railway cuts, and dam abutments subjected to earthquake-induced ground motions across a wide range of intensities.

• Calibration of 2D nonlinear slope models against field and centrifuge data
• Characterization of topographic amplification and critical failure surfaces
• Development of seismic fragility curves for natural and engineered slopes
• Newmark-type sliding block analysis with advanced constitutive models
• Machine learning-based seismic slope displacement prediction

We employ high-fidelity numerical models in LS-DYNA and Abaqus to simulate blast-induced wave propagation and evaluate the potential for blast-induced damage to underground infrastructure. In parallel, we develop practical attenuation prediction methods and sophisticated rock fracture simulations that capture near-field characteristics unique to drilling and blasting operations.

• Blast-induced wave propagation simulation in layered soil and rock media
• Novel attenuation curves for free-field and within-profile conditions
• High-fidelity rock fracture modeling using coupled hydro-mechanical approaches
• Structural damage assessment of RC structures under blast-induced pulses
• Near-field attenuation characterization for construction blasting operations

We develop numerical tools for real-time damage assessment of tall buildings and underground spaces under severe earthquakes. Using 3D full-scale soil–structure interaction simulations on high-performance computing workstations, we demonstrate that superstructure–substructure interaction critically influences the overall system response and cannot be neglected in practical design.

• Characterization of unique superstructure–basement seismic interaction mechanisms
• Quantification of seismic earth pressure on deep basement walls
• 3D full-scale SSI simulation of high-rise buildings with multi-level basements
• Machine learning-based seismic performance prediction of basement structures
• Development of practical seismic design guidelines accounting for SSI effects

We conduct extensive model tests and numerical simulations to evaluate the performance of various offshore structures, including bucket, spudcan, and pile foundations. Using advanced methods such as the Coupled Eulerian–Lagrangian (CEL) approach in Abaqus, we address large-deformation geotechnical problems critical to the design and installation of offshore energy infrastructure.

• High-strain penetration analysis of spudcan and bucket foundations in layered soils
• Development of simulation-based bearing capacity equations for bucket foundations
• Global response and damage mechanisms of jack-up barges under punch-through failure
• Coupled Eulerian–Lagrangian (CEL) modeling for large-deformation offshore problems
• Physical model testing and validation of offshore foundation performance