Nonstationary, storm-type-resolved wind hazard for design

Hazard models that separate thunderstorm downbursts, extratropical and tropical cyclones, represent their directional structure, and let parameters evolve with climate covariates and projections. Methods combine XIMIS, peaks-over-threshold and superstation analysis with Bayesian nonstationary models, synthetic tropical cyclone datasets and downscaled climate output.

How will shifts between convective and synoptic events change design wind speeds and directionality factors? How should lifetime exceedance probabilities be defined under a nonstationary hazard?

Reliability-consistent design factors for performance-based wind engineering

Reliability-based calibration of load factors, directionality factors and acceptance criteria for both strength and serviceability limit states, across building classes rather than a single prototype. Time-dependent reliability over a building’s service life will let design targets account for future hazard, and studies of the dynamic behavior of nominally rigid low-rise and long-span light-frame buildings will test where current code criteria fall short.

Goal: evidence that informs the next generation of ASCE 7 provisions and performance-based wind design.

Experimental validation of computational predictions

Applying wind-induced structural demands, derived from aerodynamic databases and computational models, as equivalent loads to laboratory specimens, and comparing measured response with predictions to improve understanding of dynamic behavior. Longer term: developing specialized wind-loading capabilities alongside structural testing facilities, and building light-frame wood specimens to extend the work beyond steel low-rise systems.

Builds on hands-on experience in RPI’s hydraulics, structural and wind tunnel laboratories.

Field-calibrated directional reliability

Calibrating directionality and reliability parameters, currently derived largely from climatology, against observed building performance. Emerging methods for estimating near-surface wind speed and direction from post-disaster damage surveys and imagery provide directional wind information, and observer-level uncertainty in field damage assessments will be treated explicitly.

Draws on my post-earthquake damage assessments in Nepal after the 2015 Gorkha earthquake.

Compound extremes and regional resilience with physics-informed surrogates

Hurricanes combine wind, rain intrusion and surge; convective storms bring wind, hail and flash flooding. I will develop machine learning surrogates constrained by aerodynamics and structural mechanics, trained on wind tunnel, CFD and field data, that plug into Monte Carlo and multi-hazard risk frameworks at regional scale. Applications include community loss and recovery under compound events, prioritizing retrofit and adaptation investments, extending the framework to multiple structural systems and toward combined wind and seismic risk, and transferring hazard and fragility knowledge to data-scarce regions such as Nepal and the wider Global South through superstation methods and transfer learning.