NATURAL HAZARDS, RELIABILITY, AND INSURANCE
Instructor: Mircea Grigoriu, mdg12@cornell.edu
Generalities:
– Why probabilistic models for natural hazards:
– Wind speeds, seismic ground accelerations, tsunami waves, and other natural
hazards exhibit significant variability.
– Available records are insufficient to estimate long-term system performance, e.g.,
wind speeds likely to be exceeded in 500 years.
– Steps of analysis:
Records =) Probabilistic models =) System response =) Performance metrics
– Performance metrics for reliability and insurance.
Primer of probability/statistics:
– Probability space: Framework for constructing probabilistic models of natural hazards
and estimating system performance under natural hazards.
– Random variables/functions as models of natural hazards. Moments, correlations,
distributions, and other properties of random elements are defined, illustrated by numerous
examples, and estimated from records of natural hazards.
– Monte Carlo simulation for Gaussian and non-Gaussian random variables, vectors, and
functions of time and/or space with applications to natural hazards.
Probabilistic models for natural hazards:
– Model constraints: Computational, i.e., models must be sufficiently simple for calibration/
analysis, and physical, i.e., models must be consistent with physics.
– Model selection and calibration: Examples include models for seismic ground acceleration,
wind pressure field, hurricanes, floods, and other natural hazards.
Performance metrics for reliability and insurance:
– Similarities: Probabilistic models of natural hazards are inputs to both reliability and
insurance metrics.
– Differences: Reliability focuses on safety of individual structures and insurance concerns
with repair costs for large portfolios of structures.
– Methods for estimating reliability/insurance metrics: Monte Carlo simulation and approximate
methods for estimating properties of quantities of interest.
Grades: Individual and group assignments, a midterm, and a final project will be used to
assess performance. The final projects for CEE 6770 will be more demanding than those for
CEE 4770.
