1 INTRODUCTION In the last decades a large research effort has been spent on the study of response of RC structures sub- jected to seismic excitation. These share with struc- tures of steel, masonry and timber the problems tied to the procedures of analysis; however, for the pecu- liar nature of the material they are composed of, the prediction of their response is a complicated task, due to computational problems in reproducing the non-linear behaviour of concrete, steel and the bond between the two. Based on the results provided by the research on the materials behaviour, different levels of refinements can be adopted in the structural discretization, namely the micro-scale of the finite element method, with its subset of the meso-scale of the fibre models for 1D elements, and the macro- scale of the beam elements. The micro and meso- scale approaches are able to follow at local level the material constitutive relations in terms of stress- strain, and differ mainly for the richness of the kin- ematic field, quite general for the micro scale and tied to geometric hypotheses on the cross-section behaviour for the meso-scale (Bernoulli or Timo- shenko beam models). The macro scale approach tracks the material behaviour in terms of sectional properties, as moment-rotation or moment-curvature relations, and often relies on a flexibility approach, with a less rich kinematic field. The reduction in ac- curacy is paired by the smaller computational effort required, in terms of CPU time, amount and simplic- ity of input data required and output data provided, that are usually in terms of quantities familiar to the design engineers. Aim of this paper is to explore the capabilities and the limits of a macro-scale model in the seismic analysis of RC walls. The adoption of macro-scale models is appealing not only for the above- mentioned reasons, but also for the fact that these represent a synthesis of research studies at more re- fined scales that can be offered outside the scientific community. In fact, the introduction of simplified non linear procedures into seismic codes creates the neces- sity for non linear elements, capable to reproduce the most important non linear phenomena of RC struc- tures under earthquake loading, while being simple enough to be used by practitioners. This work focuses on a spread plasticity model (Coronelli & Mulas 2001); this accounts for the non linear behaviour in bending which is typically distributed along the length of the element, starting from the end cross- sections. In addition, in the study of moment- resisting frames, a non-linear rotational spring can be inserted at the two ends of each element, to ac- count for the typical pinching behaviour tied to bond deterioration effects inside the beam-column joints. This model – not including the end springs - has been adopted in this work to reproduce the experi- mental results of a prototype lightly reinforced shear wall, named CAMUS I. The wall was designed ac- Macro-scale modelling for the seismic analysis: a case study P. Martinelli & M. G. Mulas Department of Structural Engineering, Politecnico di Milano, Milano, Italy ABSTRACT: The performance and the possible improvements of an existing beam spread-plasticity model have been investigated in this work, focusing on the case study of a lightly reinforced shear wall. The experi- mental response of the CAMUS I wall, tested on a shaking table under a sequence of five accelerograms, has been assumed as a benchmark. The wall response is strongly influenced by two non linear phenomena, namely the strength reduction due to M-N interaction and the stiffness degradation produced by cyclic shear, here quite pronounced due to the small amount of transverse reinforcement. The reproduction of these phenomena have been tackled first of all through an accurate set-up of the numerical model. Secondly, without modifying the model formulation, the hysteretic relationship governing the behaviour of the plastic hinge regions has been upgraded to include both degraded unloading and pinching branches. The numerical results match satis- factorily the experimental data, confirming the model capability in the non linear dynamic analyses of the wall at study.