Introduction

Moisture transport in porous media plays an important role in a wide variety of processes of environmental and technological concern, such as the degradation of building materials (e.g. mortar and concrete), the spread of hazardous wastes in the ground, oil recovery, and the containment of nuclear wastes [2],[3],[4],[5]. The presence of water in building materials can lead to cracks which result from freeze/thaw cycles or, in combination with very low permeabilities, to the spalling [6] of high performance concrete exposed to fires. In addition, the invasion of water in building materials provides a mechanism and path for the penetration of deleterious materials like chloride and sulfate ions. While the primary transport mechanisms by which chloride and sulfate ions ingress concrete are diffusion and capillary action, diffusion alone can be a very slow process, hence it may be that capillary transport, especially near an unsaturated concrete surface, is the dominant invasion mechanism. Clearly, an understanding of moisture transport in concrete and mortar is important to estimate their service life as a building material and to improve their quality.

Presently, the only standard test for the ingress of chlorides into concrete which, at least in part, depends on capillary transport, is the so called "Ponding Test [7]." In this test, a specified solution of a chloride salt is placed on top of a diked section of a concrete slab. The penetration of chloride ions is then monitored by taking cores and measuring the amount of acid-soluble chloride as a function of depth over a period of about 90 days. Since the concrete slabs are typically dried for two weeks prior to initiation of the ponding test, it is likely that capillary transport is the main driving mechanism for the chloride ion transport, at least at early times as the surface pores rewet. It is interesting that there have been attempts to correlate another standard test, the rapid chloride test [8], with the ponding test although the rapid chloride test is more a measure of the conductivity (or diffusivity) of chloride ions. Indeed, recent studies [9] have shown that such correlations can be weak. Clearly, an improved understanding of moisture transport is needed to separate the contributions of diffusion and capillary action to mass transport in order to better predict the penetration of chloride or other deleterious ions in concrete.

Tests which directly measure the rate of capillary sorption, such as the Covercrete Absorption Test (CAT) [10] and the Initial Surface Absorption Test (ISAT) [11] typically fit the total water uptake to the following equation

W / A = St^1/2 + S_o (1)

where W is the volume of water absorbed, A is the sample surface area exposed to water, S is the sorptivity coefficient, t is time and S_o is a correction term added to account for surface effects at the time the specimen is placed in contact with the water (see the appendix for a more detailed description of this equation). Such tests are made over a period of less than one hour. Clearly, measurements over such short periods of time will only probe surface effects and cannot provide sufficient information for the modelling of capillary transport over longer periods, which is needed for service life prediction.

Further, Equation 1 is based on parallel tube models of porous media (see appendix) and hence cannot accurately model capillary suction in a random porous material like concrete. The pore surface topology is far more complex so that, as the air/water interface moves through the porous medium, there are many orientations of the local interface which may be stable despite the smallness of the pore size. As an example, Figure 1 shows a meniscus, subject to a small applied pressure from below, which is unstable in a narrow pore yet is stable in a neighboring pore. In addition, the above theory of capillary suction applies to the case where the porous medium is initially dry. Clearly, the rate of capillary sorption will depend on the degree of saturation of the porous medium [5]. Finally, analysis of the flow through concrete is further complicated by the fact that water can react with the solid matrix, possibly causing a change of the pore structure with time or changing the pore solution composition.

Previous experiments measuring the capillary suction of water in concrete have exhibited conflicting results concerning the time dependence of the total water uptake. Instead of the standard t^1/2 behavior of simple capillary sorption theories other t$t^{\alpha}$ behavior (also called anomalous scaling) [12], [13] where 0.25 < < .05 has been observed. It has been suggested that this anomalous scaling in concrete is the result of modification of the pore structure due to leaching [15] or further hydration as the water is absorbed. For instance, while hydration will reduce the typical pore size in the cement paste matrix, slowing the sorption of water, leaching opens up pores, making them larger and more connected such that capillary sorption could be enhanced. However, it has not been quantitatively demonstrated how such alterations of the pore structure affect a material's sorptivity. In addition, the dissolution of salts may reduce the rate of capillary sorption [14]. Finally, as mentioned above, moisture transport in concrete must depend on factors such as the degree of saturation and environmental conditions.

FIG. 1. Stable and unstable menisci in a pore.

In this paper we present results of a study concerning capillary transport, over periods of about one year, in mortars and concrete. The validity of Equation 1 is tested. Variables considered were the water-to-cement ratio (W/C), sand size distribution, and amount of curing. We discuss several issues concerning sample preparation and boundary conditions. The existence of two different regimes associated with capillary sorption in mortar and concrete was found. An empirical function which describes capillary sorption over much longer periods is introduced. We also discuss the utility of sorptivity measurements for service life prediction.