Statistical Variation in Creep of Concrete—An Experimental Investigation

Abstract

Design of creep-sensitive concrete structures requires not only prediction of mean values of creep strains but also bounds to account for deviations. Hence, there is a need to understand how mean and deviations of the creep strains from its mean values evolve in time, to develop an appropriate stochastic model. Such studies are scant since it requires high-speed data acquisition. Results of experimental investigations carried out at CSIR-SERC, on creep and shrinkage of OPC concrete for the loading duration of 505 days after loading at 28 days, under both sealed and unsealed conditions, are presented in this paper. The rates of variation in basic and total creep suggest that basic creep rates are higher at ages beyond ten days and also noted that percentage contribution of basic creep towards total creep is higher at later ages due to moisture presence in sealed specimen; hence, consideration of basic creep is imperative in the design of closed web structures (viz. box girders) and mass concrete structures (viz. dams). From a statistical analysis of creep strains, it is noted that the histograms of square of instantaneous creep strain deviations from its mean value \(\upvarepsilon_{\text{dev}}^{2}\) are unimodal up to 0.93 days after loading, bi-modal from 0.93 to 40.56 days after loading and tri-modal after 40.56 days after loading. From the variations of mean and mean-square deviations of creep strains with respect to time, it is noted the creep evolution is a non-stationary stochastic process. Moisture presence and local load variations in specimen in the initial stages after loading seem to set the initial conditions for variation in the evolution of creep strains with time. More experimental investigations need to be carried out in this direction.