13 th World Conference on Earthquake Engineering Vancouver, B.C., Canada August 1-6, 2004 Paper No. 933 AN IMPROVED SPREAD PLASTICITY MODEL FOR INELASTIC ANALYSIS OF R/C. FRAMES SUBJECTED TO SEISMIC LOADING Michael KYAKULA 1 and Sean WILKINSON 2 SUMMARY This paper proposes an improved spread plasticity model that correctly identifies the initiation of yielding anywhere in the beam, takes into account the gradual spread of plasticity, the shift of the points of contra- flexure, the variable location and actual length of the yield zones. The model assumes that columns and beam column joints remain elastic. Beams are made up of elastic and spread plasticity sub elements connected in series. When a beam yields, its stiffness reduces and flexibility increases. Before yielding spread plasticity sub element has a null matrix, as the beam yields, the magnitude of its coefficients increase. At each time step, the model updates the flexibility matrices of the spread plasticity sub elements. Unlike existing spread and concentrated plasticity models, moments within the span are monitored and the effect of their yielding or “unyielding” taken into account. A number of examples are presented that demonstrate the limitations of the existing spread plasticity models. The paper concludes that spread plasticity models that only consider plasticity at the beam column connections are only accurate for lower stories and structures where the applied/design gravity load < 0.8. The examples also show that compared to existing spread plasticity models, the proposed model improves the accuracy in calculation of global displacements, joint rotations and inter story drift ratios by up to 25%, 69% and 55% depending on the ratios of applied/design gravity load and bottom/top reinforcement. INTRODUCTION A major problem in seismic analysis and design of reinforced concrete structures is the modelling of non- linear behaviour. All existing methods of seismic analysis and design suffer from limitations caused by modelling assumptions and therefore can benefit from improvements. One of the ways in which improvements can be effected is through experimental work and field observation of damaged or surviving structures during and after an earthquake. The other is to perform parametric studies of multi degree of freedom structures using more elaborate numerical models and analysis methods. The most sophisticated modelling procedure is the finite element method. However the complexity of the behaviour and size of the problem limits its application to validation of simpler models applied to analysis of simple structures and beams rather than non-linear analysis of multi degree of freedom building structures. Riva [1]. On the other hand, the most accurate method of dynamic analysis is the time history analysis. This too is computationally expensive and its application is generally limited to models less sophisticated than finite element.