Accidental eccentricities, frame shear forces and ductility demands of buildings with uncertainties of stiffness and live load

•An assessment of typical code recommendations on accidental building torsion is carried out based on an exceedance-probability analysis of accidental eccentricities and frame shear forces.•Increments of frame shear forces, caused by variations of building mass and stiffness, ranged from 10% to 40%.•Based on an exceedance probability of 2%, uncertainty on both live-load mass and stiffness lead to frame shear forces 16% larger than those computed for the corresponding symmetrical case.•Nonlinear analysis of a building model considering the uncertainties in live load (magnitude and position) and stiffness of structural elements were conducted using different earthquake records.•The dispersion of the computed ductility demands due to accidental eccentricity is small. Ductility demands are not significatnly affected from one earthquake record to another.

Monte Carlo simulations of 3D structural models with 4, 6 and 15 stories, subjected to bidirectional seismic excitation, were generated to study the accidental torsional response of frame buildings. Variations of accidental eccentricities, story shear forces and story-drift ductility demands were studied. The probabilities of exceeding typical accidental-eccentricity recommendations of building codes were analyzed. Simulations assumed the following variables as random: (1) live-load magnitude, (2) live-load spatial distribution, and (3) flexural stiffness of columns and beams. The other structural variables, as well as the excitation, were considered as deterministic. Results show that, although the plan aspect ratio seems important for the selection of a design accidental eccentricity, this ratio does not have a significant influence on frame shear forces. Results indicate that maximum accidental eccentricities decrease as the building height decreases and that a normalized accidental eccentricity ea/b = 0.05 seems acceptable for the tallest models studied, whereas a normalized eccentricity of ea/b = 0.10 seems acceptable for shorter buildings. When an exceedance probability of 2% of frame shear forces is considered, results indicate that, for the unsymmetrical case, the frame shear force results roughly 20% larger than the computed force of the corresponding symmetrical case. Results also suggest that the design of regular frame buildings can be simplified by providing formulas to increase frame shear forces directly, instead of estimating and distributing building torsional moments among frames. The variations of both live loads and stiffness used in this study led to frame shear-force increments that varied between 10% and 40%, as compared with the nominally symmetrical case. The dispersion of the computed ductility demands caused by accidental eccentricity resulted small.