This paper proposes a critical approach on some procedures for constructing plate models to describe the out of plane mechanical behaviour of “masonry” panels with a regular texture. The word “masonry” is here used to describe a system of blocks which interact with elastic interfaces (i.e. mortar joints). Identification procedures between the 3D discrete model and the 2D plate continuum model, have been here reported based on standard homogenisation methods and based on direct identification method. i.e. by equating the internal work in the 3D discrete model and in the 2D plate model for a class of regular motions - equivalent compatible model -. A crucial problem, with the choice of an identification procedure, is how the kinematic, the dynamic and the constitutive prescriptions of the discrete system are transferred to the continuous one. Hence, constitutive functions of the plate 2D models may be different. A Love-Kirchhoff plate model based on standard homogenisation for linear elastic periodic brickwork has been already proposed by Cecchi and Sab [
]. This model has been also developed in the case both of infinitely rigid blocks and of elastic blocks connected by elastic interfaces taken into account shear effects leading to the identification of a new Reissner-Mindlin homogenised plate model [
]. In this case the identification between the 3D block discrete model and the 2D plate model is based on an identification at the order 1 in the displacement and at the order 0 in the rotation. A Mindlin-Reissner model when block are rigid blocks based on a compatible identification at the order 1 both in the displacement and in the rotation has been performed by Cecchi and Rizzi [
]. Here this model has been implemented also in the case of elastic blocks. All these models permit to obtain the Mindlin-Reissner 2D plate constitutive functions in an explicit symbolic form. The idea is to compare the different constitutive function trough a critical analysis of the identification procedure and to evaluate the accuracy of these identification models by comparison with a 3D F.E. model for a meaningful case.