Research On Stochastic Damage Constitutive Law Of Masonry

The constitutive law of masonry is an important subject of the basic mechanical properties, especially the compressive stress-strain curve. It has a great value in theo-retical research and engineering applications, but an widely used constitutive model has not been established until now. Although experienced constitutive model agrees with experimental results well, but they are lack of theoretical foundations. The nonlinear properties of masonry can be reflected, but the discreteness can be not.This paper mainly studies the stochastic damage constitutive law of masonry, focus on the following:Based on the established constitutive relationships and the failure phenomena of masonry, the failure mechanism is analyzed from mesoscopic level. A mesoscopic physical model of masonry under monotonic axial loading, which can explain the nonlinear performance of stress-strain curve, has been developed.By selecting the damage variable with physical meaning and using of the equilibrium of force, a damage constitutive model of masonry subjected to monotonic axial loading is established. The failure strain is assumed as a stochastic variable, and considering the stochastic properties of elastic modulus, the mean value and variance constitutive relationship can be obtained through theoretical derivation of damage mechanics. If the correlation of stochastic variables is not considered, the stochastic damage constitutive relationship turns into a deterministic one.With fly-ash brick masonry as the research object, the experiment of stress-strain curve of nine prism specimens is carried out. The mechanical properties and damage characteristics of macroscopic are investigated. The complete stress-strain curves, which can provide reliable data to verify the constitutive model, are obtained through the experiment.With the help of stochastic modeling principle and Powell optimization algorithm, the random field parameters and material parameters of the established constitutive model can be recognized. The model results have a good agreement with experimental data. The development of macroscopic cracks can also be predicted from the damage level.