In the frame of Cosserat continuum theory, an upscaling procedure for the assessment of the in-plane strength domain of discrete media is developed. The procedure is the extension to the Cosserat continuum of a procedure initially formulated for the Cauchy continuum, based on the kinematic approach of limit analysis and the classical homogenisation theory. The extension to the Cosserat continuum is made in order to take into account the effect of particles' rotation on the strength of the discrete medium. The procedure is illustrated with regard to periodic assemblies of blocks in contact and is then generalised to the whole class of discrete periodic media with particles of the same type. The case of masonry is considered as an application. Strength criteria of columns and walls are formulated in terms of non-symmetric stresses and in-plane couples. The procedure allows to show how the in-plane strength of the medium is reduced as a result of particles' rotation.

The paper deals with a probabilistic modeling of the thermo-mechanical behavior of cardboard-plaster-cardboard (CPC) multilayer plates submitted to re load. The proposed model takes into account data and model uncertainties. This work is justied by the fact that re resistance tests of plasterboard-lined partitions are made impossible when their dimensions exceed those of furnaces. A fundamental key to solve such a problem is the development and the experimental validation of a deterministic and probabilistic model of CPC multilayer submitted to re load. The rst step of this work concerns the constitution of an experimental thermo-mechanical data base for a CPC multilayer and for its components. These experimental tests are carried out by the use of a bench test specially designed for this work. The second step is the development of an homogenization thermo-mechanical mean model for the CPC multilayer. The third step is the development of a probabilistic model of uncertainties based on the

SUMMARYThe paper deals with probabilistic modeling of heat transfer throughout plasterboard plates when exposed to an equivalent ISO thermal load. The proposed model takes into account data and model uncertainties. This research addresses a general need to perform robust modeling of plasterboard-lined partition submitted to fire load. The first step of this work concerns the development of an experimental thermo physical identification data base for plasterboard. These experimental tests are carried out by the use of a bench test specially designed within the framework of this research. A computational heat transfer model is constructed using data from the literature and also the identified plasterboard thermophysical properties. The developed mean model constitutes the basis of the computational stochastic heat transfer model that has been constructed employing the nonparametric probabilistic approach. Numerical results are compared to the experimental ones.