- Author
- Pnelli, J. P. | Simiu, E. | Gurley, K. | Subramanian, C. | Zhang, L. | Cope, A. | Filliben, J. J. | Hamid, S.
- Title
- Hurricane Damage Prediction Model for Residential Structures.
- Coporate
- Florida Institute of Technology, Melbourne, FL National Institute of Standards and Technology, Gaithersburg, MD Florida Univ., Gainesville Florida International Univ., Miami, FL
- Journal
- Journal of Structural Engineering, Vol. 130, No. 11, 1685-1691, November 2004
- Keywords
- hurricane | damage | weather effects | structures | disasters | risk analysis | simulation | residential buildings | Monte Carlo method
- Identifiers
- Venn diagram for basic damage modes of masonry home; Venn diagrams of combined damage states; calculation of damage matrices; probabilities of occurrence of subdamage modes conditional on wind speed intervals; intermediate output to be used for validation with observed data; sample of simulated probabilities of combined damage states, conditional on wind speed intervals; to be used for damage calculations
- Abstract
- The paper reports progress in the development of a practical probabilistic model for the estimation of expected annual damage induced by hurricane winds in residential structures. The estimation of the damage is accomplished in several steps. First, basic damage modes for components of specific building types are defined. Second, the damage modes are combined in possible damage states, whose probabilities of occurrence are calculated as functions of wind speeds from Monte Carlo simulations conducted on engineering numerical models of typical houses. The paper describes the conceptual framework for the proposed model, and illustrates its application for a specific building type with hypothetical probabilistic input. Actual probabilistic input must be based on laboratory studies, postdamage surveys, insurance claims data, engineering analyses and judgment, and Monte Carlo simulation methods. The proposed component-based model is flexible and transparent. It is therefore capable of being readily scrutinized. The model can be used in conjunction with historical loss data, to which it can readily be calibrated.