2006 | OriginalPaper | Chapter
Flat-jack tests and parameter identification for diagnostic analysis of dams
Authors : R. Fedele, G. Maier, L. Marazza
Published in: III European Conference on Computational Mechanics
Publisher: Springer Netherlands
Activate our intelligent search to find suitable subject content or patents.
Select sections of text to find matching patents with Artificial Intelligence. powered by
Select sections of text to find additional relevant content using AI-assisted search. powered by
In many existing concrete dams chemical-physical processes, such as the alkali-aggregate reaction (AAR) [
1
] and/or past extreme loadings, have given rise to reductions of local stiffness and strength during decades of service life. Moreover, concrete expansion due to AAR and possible geological motions in the foundation may have generated self-equilibrated stress states, additional to stresses caused by external loadings.
In this communication an experimental-computational method centred on flat-jacks is outlined for the assessment of local stress states and possibly deteriorated properties of concrete in existing dams. Herein the synergistic combination of a novel experimental pattern, computer simulation of the tests (by conventional finite elements) and inverse analysis techniques allows to exploit the experimental data more effectively than in the current practice of flat-jack tests [
2
], without recourse to traditional semi-empirical formulae relating measurable quantities to material parameters. At suitably chosen locations on the free surface of the monitored dam, the proposed technique identifies all the components of local (plane) stress state, the Young moduli in vertical and horizontal direction (often different due to the compacting process), tensile strength and fracture energy of the dam concrete.
The inverse problem in point is formulated as a sequence of parameter estimations: the constrained minimization of least-square objective functions is performed by means of a gradientbased, interior-point Trust-Region algorithm (see e.g. [
3
]). As an alternative to conventional identification techniques, the application of artificial neural networks to the present inverse analysis problem is investigated, in view of their routine, cost-effective use in situ as a “black-box”, by a procedure similar to the one recently proposed in [
4
] for dilatometric in-depth tests.