000013236 001__ 13236
000013236 005__ 20161114160333.0
000013236 04107 $$aeng
000013236 046__ $$k2009-06-22
000013236 100__ $$aBertero R., D.
000013236 24500 $$aReliability decomposition of dynamic wind load on structures

000013236 24630 $$n2.$$pComputational Methods in Structural Dynamics and Earhquake Engineering
000013236 260__ $$bNational Technical University of Athens, 2009
000013236 506__ $$arestricted
000013236 520__ $$2eng$$aMulticorrelated stationary random process, such as wind pressure fluctuations on structures can be transformed into a set of subprocesses by diagonalizing their covariance matrix trough eigenvector decomposition. This eigenvector decomposition is known as the proper orthogonal decomposition (POD). The numerical advantage of the POD technique, akin to the modal analysis in structural dynamics, relies on the reduced-order representation through truncation of the higher eigenmodes associated with small eigenvalues. However, truncation of higher modes may not work effectively in the case of local response, which may lead to an underestimation of the local wind loads and their effects. In this paper a new method, called Reliability Decomposition, is proposed to overcome the weaknesses of the POD technique. In the proposed method the subprocess is selected based on the most probable load distribution at failure. This load distribution is obtained using first order reliability analysis. Since the Reliability Decomposition depends on the local effect under analysis, only one subprocess is needed in each case to obtain accurately results. In this paper, the methodology to obtain the load distribution to compute any internal force or displacement is presented. An example is developed to show the comparison among Reliability Decomposition and POD computations and results.

000013236 540__ $$aText je chráněný podle autorského zákona č. 121/2000 Sb.
000013236 653__ $$aReliability Decomposition, Proper Orthogonal Decomposition, multicorrelated random process, wind load distribution. Abstract. Multicorrelated stationary random process, such as wind pressure fluctuations on structures can be transformed into a set of subprocesses by diagonalizing their covariance matrix trough eigenvector decomposition. This eigenvector decomposition is known as the proper orthogonal decomposition (POD). The numerical advantage of the POD technique, akin to the modal analysis in structural dynamics, relies on the reduced-order representation through truncation of the higher eigenmodes associated with small eigenvalues. However, truncation of higher modes may not work effectively in the case of local response, which may lead to an underestimation of the local wind loads and their effects. In this paper a new method, called Reliability Decomposition, is proposed to overcome the weaknesses of the POD technique. In the proposed method the subprocess is selected based on the most probable load distribution at failure. This load distribution is obtained using first order reliability analysis. Since the Reliability Decomposition depends on the local effect under analysis, only one subprocess is needed in each case to obtain accurately results. In this paper, the methodology to obtain the load distribution to compute any internal force or displacement is presented. An example is developed to show the comparison among Reliability Decomposition and POD computations and results.

000013236 7112_ $$aCOMPDYN 2009 - 2nd International Thematic Conference$$cIsland of Rhodes (GR)$$d2009-06-22 / 2009-06-24$$gCOMPDYN2009
000013236 720__ $$aBertero R., D.
000013236 8560_ $$ffischerc@itam.cas.cz
000013236 8564_ $$s148728$$uhttps://invenio.itam.cas.cz/record/13236/files/CD343.pdf$$yOriginal version of the author's contribution as presented on CD, section: Statistical and probabilistic methods in computational mechanics to treat aleatory and epistemic uncertainties in structural and/or geotechnical systems and their loading environment - i (MS).
000013236 962__ $$r13074
000013236 980__ $$aPAPER