BOND BEHAVIOUR BETWEEN RECYCLED AGGREGATE CONCRETE AND DEFORMED STEEL BARS

Abstract

Bond of deformed steel bars, having diameters in the range of 8 mm – 25 mm, embedded in recycled aggregate concrete made with coarse recycled concrete aggregate has been investigated with the help of pullout and a selected number of splice beam tests. Three grades of recycled aggregate concrete corresponding to normal-strength, medium-strength and high-strength were studied in the pullout tests whereas in the splice beam tests only the behaviour of normal- and the high-strength concrete has been compared. Natural coarse aggregates in the control concrete mixtures were substituted with an equal weight of the coarse recycled concrete aggregates (in the saturated-surface-dry moisture condition) at replacement levels of 25 %, 50 %, 75 % and 100 % and the effect on the mechanical properties and bond behaviour was monitored. The 56-day compressive strength of the three concrete grades decreased by approximately 33 %, 30 % and 27 % respectively upon 100 % substitution of the natural coarse aggregates (NCA) with the recycled concrete aggregates (RCA). For the aforementioned condition, the 56-day splitting tensile strength however increased by about 18 %,13 % and 19 % for the normal-, the medium- and the high-strength concrete respectively. A bond strength predictive model for short rebar embedded lengths and which is valid for both NCA as well as RCA concretes (with cylinder crushing strengths of up to 70 MPa) has been proposed and validated. The predictive efficacy of the proposed model was better compared to that of the model in the fib Model Code 2010, which is widely referred to for short embedded lengths. In the proposed model, the effect of concrete properties on bond strength has been represented using the square root of the compressive strength. Across all the three concrete grades, the pullout specimens embedded with the 8 mm, the 10 mm and in some cases with the 12 mm bars, showed pullout failure and the normalised bond strengths in the case of these specimens increased with an increase in the amount of RCA in concrete. This trend in the normalised bond strength has been explained in terms of fracture toughness of concrete estimated using brittleness index, an analogous parameter from rock mechanics. The trend of increasing normalised bond strength with increasing RCA replacement levels was not clearly evident in the case of the pullout specimens embedded with the 16 mm, the 20 mm, the 25 mm and to some extent with the 12 mm bars, the observed failure mode in all these cases being pullout failure induced by through splitting. In all the three concrete grades, the measured bond stress-slip relationships for all the bar sizes under investigation were similar in the NCA and the RCA concretes and no significant difference was noted in the interfaces of the two concrete types. For the specimens failing in pullout, five distinct stages of bond behaviour could be identified whereas for the specimens failing in pullout induced by through splitting four stages of ii bond behaviour were noted. The predictions of the proposed empirical model for the bond stress-slip relationship associated with two failure modes noted in this investigation were in good agreement with the measured data. Failure modes of the NCA and the RCA concrete specimens in the splice beam tests were similar and a bond strength predictive model for long embedded lengths typical in splice beam testing (and in actual construction) has been proposed for both NCA and RCA concrete in terms of parameters which are widely accepted to influence bond behaviour. This model, which is valid for cylindrical compressive strengths of up to 70 MPa, accounts for the effect of   ' 1 4 c f . It was noted that the bond strength model in the ACI 408R-03, originally developed for NCA concrete, gave reasonably accurate predictions for the bond strengths measured in the splice beam tests.