Comparison of Masonry Homogenization Methods – Macromodelling and Micromodeling of Walls Behaviour Made of Autoclaved Aerated Concrete Masonry Units

The adopted method of empirical homogenization strictly determines the degree of faithful reproduction of the masonry structure's work in terms of the analysis of cracking forces, destructive forces, and the mechanism of structure destruction. The high level of detail of the numerical model may make it impossible to perform calculations and predict internal forces for larger structures or entire buildings. The study aims to compare two different masonry homogenization techniques and determine the advantages and disadvantages of the adopted methods. The concept of a micromodel, in which the contact of two materials - a masonry unit and a mortar, was simulated using contact elements in the interface planes and a macromodel in which the wall was modelled as a homogeneous, isotropic material, omitting contact surfaces. The analysis subjects were standard wall models made of autoclaved aerated concrete (AAC) masonry units in axial and diagonal compression tests. In the numerical calculations, the elasto-plastic model with degradation implemented. The Menetrey William boundary surface describes the compression phase, and the Rankine criterion determines the tensile phase. In the axially compressed walls, the relations of forces and vertical and horizontal deformations compared, and in the shear walls, the forces and values of strain angles analyzed. In both models, the mechanisms of wall destruction and scratching were considered. The initial parameters of the elasto-plastic model derived from the results of wall tests using various model validation techniques. The calibration coefficient was used in the micromodel, determined as the quotient of the wall's compressive strength and masonry unit's compressive strength. The fracture energy value was also corrected. In the macromodel, the masonry's modulus of elasticity and the tensile strength value calibrated. Calculations based on the micromodel were consistent with the test results at the relative error level of 2%. The observed damage and scratches to the walls after the tests were consistent with the numerical projection. The macromodel calculations showed the convergence of the results in scratch morphology, scratching and destructive forces. The most significant differences occurred in shear deformations. The macromodelling approach allowed for capturing the wall's global tendency to deteriorate without opening the contact surfaces locally (cohesive cracks), as is the case during the tests.