OpenSees Earthquake Engineering with Machine LearningNew

Background

Seismic risk studies must simulate thousands of building models under many ground motions. OpenSees provides the nonlinear finite-element core (fiber-section beams, beam-column joints, contact, soil-structure interaction) trusted by the earthquake engineering community.

Classical workflow

1. Build nonlinear structural models with material definitions
2. Run response-history analyses under suites of ground motions
3. Extract engineering demand parameters (drift, acceleration, residual)

Where ML accelerates

• Surrogate models trained on OpenSees output predict damage states across building portfolios
• Learned ground-motion-to-response mappings enable real-time regional estimates
• Model-order reduction keeps accuracy where nonlinearity matters

Use cases

• Regional seismic loss estimation for insurance and planning
• Retrofit prioritization: which buildings gain most from strengthening
• Performance-based design checks against code limits

Resources

• OpenSees documentation and wiki examples
• PEER reports and journal papers on OpenSees validation benchmarks