Seismic Risk Assessment and the Needs for Site Response Analysis: Advances, Challenges, and Future Directions
- Dr. Youssef M A Hashash, at University of Illinois Urbana-Champaign
Having just returned from the 13th National Conference on Earthquake Engineering (13th NCEE) in Portland, Oregon (USA), I am impressed by the amazing progress our community of researchers and practitioners has made in addressing the various pieces of the scientific and technical puzzle needed to evaluate, manage and to the extent feasible reduce the seismic risk to communities and infrastructure worldwide. Seismic risk assessment has evolved toward higher-fidelity, region- and site-specific estimates. Risk assessments increasingly consider multiple hazards and cascading hazards rather than evaluating a single natural hazard in isolation.
Stakeholders, including scientists, engineers, social scientists, insurers, and government agencies, are engaged in “all-hands-on-deck” efforts to understand these risks and develop necessary tools, approaches, and frameworks needed to navigate the complexities associated with characterizing and reducing this risk. Yet, significant knowledge gaps remain; data, large and small, better and robust simulations, opportunities for machine learning and other AI tools.
From Prescriptive Code Compliance to Performance Evaluation
At the core of risk assessment are advanced engineering evaluations of the seismic hazard, site effects, soil-structure interactions and structural responses. Seismic analysis workflows are evolving to support a shift from conventional code compliance, where the objective is to meet minimum design requirements, toward performance-based assessment and functional recovery. These approaches allow infrastructure owners to define target performance levels for individual components or entire systems and evaluate whether those targets can be met under specified hazards. Those targets can also incorporate functional recovery objectives, such as the acceptable time required to restore critical services or resume normal operations following an event.
Available seismic analysis tools assess expected damage, serviceability, and recovery requirements across a range of earthquake scenarios. By connecting component-level behaviour to system-level performance, these tools can help stakeholders identify vulnerabilities and evaluate design or retrofit options based on specific resilience objectives.
Capturing Uncertainty in Engineering Performance Evaluations
Representing and communicating uncertainty consistently across all elements of seismic risk evaluation is a challenge the earthquake community continues to grapple with. Uncertainty in seismic hazard dominates the overall uncertainty in risk evaluation, while evaluating uncertainty in engineering systems response (site effects, soil-structure interaction, structure response) is daunting because of limited data, geometric complexity and high computational costs. Two types of uncertainty are typically considered:
- Epistemic uncertainty, which results from limited knowledge, data, or model accuracy. Epistemic uncertainty can be reduced by improving the underlying models.
- Aleatory uncertainty, which reflects inherent variability in seismic events, site conditions, and system response. Aleatory uncertainty can be better represented by accounting for variability in input parameters.
Quantifying uncertainty requires computationally efficient analysis tools that are accessible to practicing engineers and researchers at a reasonable cost.
Rising Expectations from Engineers and Project Stakeholders
Over the past decade, expectations for seismic analysis software have shifted toward higher-fidelity results combined with greater speed, accessibility, and ease of use. As projects increase in complexity, engineers are expected to produce higher-resolution analyses and evaluate many realizations efficiently. Software must provide more refined results that can reduce unnecessary conservatism and its associated costs. Results must also be well documented and clearly communicated to support transparency and confidence among stakeholders.
Engineering practice is moving toward consistent, end-to-end estimates that connect the different components of risk assessment. Faster analysis workflows are increasingly important because they allow engineers to evaluate multiple design iterations, compare alternatives, and optimize designs while managing design resources and overall project costs.
Evaluating the role of local ground conditions remains a cornerstone in all seismic risk assessments. Understanding site effects is not limited to conventional engineering evaluations. At the 13th NCEE I saw broad applications of RSSeismic 1-D site response including use by insurance companies. While significant advances have been made in evaluating site effects, many challenges and opportunities remain.
Challenges we continue to face in site response analysis and opportunities
Evaluating local site effects due to seismic shaking remains a challenging task. We need to estimate the ground shaking, characterize often complex three-dimensional geology, stratigraphy and cyclic soil response including liquefaction. End-to-end numerical assessments remain challenging because of their computational requirements and limits on the frequency ranges that current methods can resolve in addition to the limited available 3-D site characterization.
Centella (2026) demonstrated the feasibility of large scale 3-D nonlinear site response analyses for specialized research applications. I expect that in the near future we will see 3-D site response evaluation in complex geologic setting for high value infrastructure. Nevertheless, 1-D site response analysis remains the workhorse for large swaths of engineering applications.
Why RSSeismic v3 Matters
RSSeismic v1 & v2, built on the 25-year legacy of DEEPSOIL, introduced efficient workflows for equivalent linear and nonlinear 1-D site response including access to a large ground motion database and representation of uncertainty within a numerically efficient simulation framework that is first in class. RSSeismic v3 introduces major new capabilities that greatly enhance site response workflows and address uncertainty quantification:
- Introduces a finite element (FE) based 1-D site response capability that reduces epistemic uncertainty through the introduction of:
- Horizontal bi-directional (and in the future tri-directional) shaking whereby both horizontal shaking components are considered. This is a capability that is absent in conventional equivalent-linear or nonlinear lumped-mass 1-D analyses as well as 2-D FE and finite difference (FD) analysis platforms. Currently the capability for horizontal bi-directional shaking requires the use of 3-D FE and FD versions of commonly used analysis platforms.
- A practical 3-D soil-constitutive model that can represent small strain nonlinearity, non-Masing hysteretic damping, shear-induced volumetric response, excess pore-water pressure generation, and phase transformation during liquefaction. This also provides a bridge to 2-D and 3-D site response and seismic soil-structure analyses.
- A highly efficient computational platform that is on par with the efficiency of the lumped mass nonlinear 1-D analyses. This enables the evaluation of aleatory uncertainty through a computationally efficient platform capable of supporting large numbers of analyses on laptops or workstations whereby HPC systems are only needed for very large studies.
- Integrates enhanced ground motion selection and development frameworks to further simplify site response analysis workflow.
- Incorporates workflow automation capabilities through Python scripting including batch analysis and processing of large-scale RSSeismic analyses
Details on these new capabilities are provided in separate articles.
Meeting Evolving Real-World Needs
RSSeismic v3 is designed to meet the evolving needs of the engineers’ analyses. The bi-directional shaking makes site response analysis compatible with modern ground motion models that use effective amplitude spectra (EAS). EAS based models are currently under development as part of NGA-West 3, and EAS based site terms have been developed for these models (Ilhan et al., 2026).
The upcoming NEHRP Recommended Seismic Provisions for New Buildings and Other Structures (BSSC, 2026) will require that site response analysis for site with potential liquefaction be able to capture the phase transformation (sometimes referred to as dilation spikes) in the simulations. The introduced FE & advanced soil modelling capabilities will be able to represent this behaviour and make them available to researchers and engineers in an accessible and computationally efficient framework.
Looking Ahead
Seismic risk assessment still relies on individual tools that are time-consuming to operate and create bottlenecks in engineering practice and research. These tools are also fragmented, with limited integration and substantial manual data transfer between different stages of an assessment. Improvements in computational performance would make it possible to run larger number of simulations needed to evaluate uncertainty across seismic hazards, site conditions, and system response. Greater automation can connect tools, standardize data exchange, coordinate simulations, and track assumptions and results across the workflow.
Agentic AI can further support this process by setting up analyses, executing multi-step workflows, checking inputs and outputs, and identifying where additional simulations are needed. There are many exciting opportunities for continued development, and much work lies ahead!
References
Centella, D. (2026). Modeling 3D Wave Propagation Effects in Seismic Site Response Analysis. (Doctoral dissertation, University of Illinois at Urbana-Champaign).
Ilhan, O., Centella, D., Sulaimain, A., Sung, CH., Hashash, Y. M., Abrahamson, N. (2026). Simulated Nonlinear Site Amplification Models for Western United States. Earthquake Spectra.
NEHRP Recommended Seismic Provisions for New Buildings and Other Structures. National Institute of Building Sciences, Washington DC, final publication pending, September 2026.
