Flat-jack tests have been employed for decades for the assessment of stresses and
Young moduli in possibly deteriorated concrete dams and masonry structures. We
propose a procedure for such tests that includes several innovations: identification of
Young moduli and shear modulus in the presence of orthotropy, of pre-existing
normal and shear stresses, and of tensile and compressive strength and fracture
energy; use of full-field displacement measurements by digital image correlation
(instead of extensometers); computer simulations performed once-and-for-all
and productive of results which are subsequently processed out by proper
orthogonal decomposition and its truncation; and identification of parameters
in situ, soon after the tests, by portable computer with software able to
perform inverse analyses by mathematical tools newly introduced into this
context. The proposed procedure is validated by means of pseudoexperimental
numerical exercises, by employing comparatively, as central computational tools,
artificial neural networks and a trust region algorithm implying only first-order
derivatives (with respect to the sought parameters) of the discrepancy function to
minimize.
Dedicated to the memory of
Marie-Louise Steele and to Charles Steele.