Effect of air voids on salt scaling and internal freezing

By combining calorimetric measurements with dilatometry, it has been possible to calculate the contributions of thermal expansion, pore pressure, and crystallization pressure of ice to the strain observed in a mortar during freezing/thawing cycles. Air-entrained mortars contract upon freezing, while non-air-entrained mortars expand. The expansion of the latter is attributed primarily to hydraulic pressure, owing to the rapid growth of ice, which nucleates at low temperatures in laboratory samples. Poromechanical calculations account quantitatively for the contraction of samples with air entrainment, assuming that ice crystals form in the air voids. As originally proposed by Powers and Helmuth, those crystals create suction in the pore liquid that offsets the crystallization pressure of ice in the mesopores of the paste, resulting in a net contraction. Ice in the matrix also contributes significantly to the increase in the thermal expansion coefficient of the mortar. The magnitude of the contraction in air-entrained mortar is shown to account for a reduction of salt scaling damage. According to the glue-spall theory, the damage results from cracking of the ice on the surface of concrete, when the thermal expansion mismatch stress exceeds the strength of the ice. The contraction of the mortar caused by air entrainment offsets the thermal expansion mismatch sufficiently to prevent cracking. Based on observations of the nucleation temperature of ice in laboratory samples of various sizes, it is estimated that there is one site capable of nucleating ice at -1 degrees C in a cube of mortar roughly 34 cm on an edge (or, one per square meter in a stab 3 cm thick). This suggests that ice nucleates in the field at high temperatures, compared to what is typically seen in the laboratory, and propagates slowly through the pores as the temperature drops. This mode of growth may lead to fatigue damage over many cycles, owing to local stresses from crystallization pressure, where the contribution of hydraulic pressure is insignificant. (C) 2009 Elsevier Ltd. All rights reserved.