A New Approach to Structural Reliability in Fatigue Failure
Abstract
The project objective is to develop formalism for assessing fatigue fracture of welded parts of ship structures based on a newly developed automatic Learning Expert System (LES), using extensive knowledge of material characteristics, experimental results, and results obtained through computational simulations. Any study of the welds on a ship structure must take into account the uncertainties and random characteristics of the loading, as well as the fatigue lifetime of the welds. Due to the stochastic nature of the loads, the available deterministic approaches are not sufficient to give a reliable evaluation of the structural safety. Although there has been some effort to fill this gap by probabilistic approaches, these are yet of limited usefulness because of the limited available databases. The research project will ultimately demonstrate the applicability and effectiveness of LES to this class of engineering mechanics problems. A special procedure to produce the intelligent descriptors has been developed. These descriptors are then used to represent the important scalar fields at the critical points of the structure. Two commercial Finite Element codes, NISA and ALGOR, are used in this procedure. After the FEM model is generated, fixed size windows are set up to cover the hot points in the specimen. Inside each window, the maximal, minimal and average values of the important scalar fields, such as I1 and J2, are calculated. To reduce the influence of mesh density, five intervals between the minimum and maximum values of each scalar field have been set up and the percentage of the area under the scalar value within each of the five intervals is calculated to express the distribution of the scalar field in the window.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 01, 1998
- Accession Number
- ADA345639
Entities
People
- Joseph Zarka
- Sia Nemat-nasser
Organizations
- University of California, San Diego