PREDICTION OF PEELING FAILURE OF REINFORCED-CONCRETE BEAMS WITH EXTERNALLY BONDED STEEL PLATES

A review of available literature on design against premature failure ( due to plate peeling) of reinforced concrete beams strengthened by the gluing of mild steel plates to their soffits suggested strongly that there existed a number of unresolved problems in the field, The previo us approaches had been either purely experimental or (at their best) o f a semiempirical nature, resorting to calibration factors based on te st data from laboratory specimens which (although sometimes quite exte nsive in number) were not believed to have covered the full range of f irst order (i.e. controlling) design parameters as regards the plate p eeling phenomenon. There appeared to be a number of disagreements abou t the nature of the basic models) of failure and the choice of other p rimary material and/or geometrical parameters to be used in developing appropriate design recommendations. The present paper reports details of a simple theoretical model which (in a systematic fashion) provide s an insight into the mechanism of the appropriate mode of premature f ailure and the outcome of which is supported by an extensive set of te st results from other sources. The final formulations, which do not us e any calibration factors, provide upper and lower bounds to the magni tude of the critical axial tensile stress in the plate at the instance of peeling failure which is found to be the primary factor controllin g the ultimate load. It is shown that the magnitude of this critical s tress depends (among other factors) on the spacings of the flexural cr acks in the concrete cover, and, owing to large variations (by a facto r of, say, 2) in spacings of flexural cracks in practice, wide scatter is to be expected in the test data from even closely controlled exper iments. It is therefore concluded that a lower bound approach would be the appropriate design technique. The paper discusses the possible pr actical limitations of the proposed theory and, for example, suggests that, in certain cases, relatively small-scale test specimens may not correctly simulate conditions, such as the so-called shear lag phenome non, which is likely to be significant in beams under service conditio ns with sufficiently long portions of steel plates which have been ter minated within the critical shear span. Nevertheless, for most practic al situations, the present model has identified the relative significa nce of the first order design parameters and, if these are controlled, it should be possible to avoid premature failures in practice.