Shrinkage of concrete is caused by loss of water by evaporation or by hydration of cement, and also by carbonation. The resulting reduction in volume as a fraction of the original volume, i.e. volumetric strain, is equal to three times the linear strain, so in practice shrinkage is simply measured as a linear strain. The units are mm per mm, usually expressed as microstrain (10ÿ6). When freshly laid and before setting, concrete can undergo plastic shrinkage due to the loss of water from the surface or by suction by dry concrete below, a situation that can lead to plastic cracking (see Section 3.3.2). Plastic shrinkage is greater the larger the cement content of the mix and it can be minimised by complete prevention of evaporation immediately after casting. Even when no moisture movement to or from the set concrete is possible autogenous shrinkage occurs, which is caused by loss of water used by the hydrating cement. Except in large volume pours (mass concrete), autogenous shrinkage is not distinguished from shrinkage of hardened concrete due to loss of water to a dry surrounding environment. In normal strength concrete, auto- genous shrinkage is small (50 to 100 10^-6),but can be large in high performance concrete, i.e. high durability and high strength concrete. Such concrete usually contains a high cementitious materials content consisting of cement and a mineral admixture, such as silica fume or fly ash. In addition, the mix has a low water/cementitious materials ratio so that a superplasticising chemical admixture is required to make the mix workable. Such composition yields a finer pore structure than normal strength concrete, which causes high early autogenous shrinkage when measured from initial set but, when measured from the age of 28 days, it is small. If concrete is stored continuously in water during hydration, the concrete expands due to absorption of water by the cement paste, a process known as
swelling. In concrete made with normal weight aggregate, swelling is 5±10% of shrinkage of hardened concrete. On the other hand, swelling of lightweight aggregate concrete can be much greater, viz. 25±80% of shrinkage after 30 years (Brooks, 2005).
The loss of water from hardened concrete stored in dry air causes drying shrinkage. To some extent the process is reversible, i.e. re-absorption of water will cause expansion of the concrete, but not back to its original volume (see Fig. 3.3(a)). In normal concretes, reversible shrinkage is between 40 and 70% of the drying shrinkage depending upon the age of the concrete when first drying occurs. If the concrete is cured so that it is fully hydrated at the time of exposure, more of the drying shrinkage is reversible but, if drying is accompanied by hydration and carbonation, the porosity of the cement paste will decrease, thus preventing some ingress of water (Neville and Brooks, 2002).
Another type of shrinkage that occurs in concrete is carbonation shrinkage, which normally accompanies drying shrinkage, although it is different in nature. Carbonation, which is known to be a potential cause of corrosion of steel in reinforced concrete, describes the reaction of calcium hydroxide with the carbon dioxide of the atmosphere in the presence of moisture. The carbon dioxide first dissolves in moisture and then reacts with calcium hydroxide to form calcium carbonate, the process resulting in a volume contraction known as carbonation shrinkage. The rate of carbonation depends upon the permeability of the concrete, the moisture content and the relative humidity of the environment, the severest conditions for high carbonation shrinkage being 55% relative humidity and high water/cement ratio (Neville, 1995). The effect of carbonation shrinkage in concrete is normally small as it is restricted to the outer layers, but can cause warping of thin panels. A practical benefit is in the manufacture of porous concrete blocks where curing in an atmosphere of carbon dioxide results in the calcium carbonate products being deposited in the pores. This restricts moisture movement (reversible shrinkage) and enhances strength. The pattern of concrete subjected to cycles of drying and wetting, simulating daily weather changes in practice, is illustrated in Fig. 3.3(b). The magnitude of the cyclic change depends upon the duration of periods, the ambient humidity and the composition of the concrete, but it is important to note that drying is slower than wetting. Consequently, shrinkage resulting from a prolonged dry weather can be reversed by a short period of rain. Generally, shrinkage of lightweight aggregate concrete is more reversible than that of normal weight concrete. Three mechanisms are thought to be responsible for reversible drying shrink- age: capillary tension, disjoining pressure and changes in surface energy (Mindess and Young, 1981). Removal of water from larger capillaries of the cement paste to the drier outside air causes little shrinkage, but this disturbs the internal equilibrium so that water is transferred from smaller capillaries to larger ones. When capillaries empty, a meniscus forms and a surface tension is developed. This induces a balancing compressive stress in the calcium silicate hydrate (C-S-H) and results in a volume contraction or shrinkage. Stresses are higher in smaller capillaries and when the humidity is low due to the increasing curvature of menisci, but at very low humidity capillary stresses do not exist because the menisci are no longer stable. With the disjoining pressure theory, the adsorption of water on the C-S-H particles affects the Van der Waals surface forces of attraction between adjacent particles in areas of hindered adsorption. The adsorbed water creates a disjoining pressure, which increases with the thickness of the adsorbed water. When the disjoining pressure exceeds the Van der Waals forces the particles are forced apart and swelling occurs. Conversely, as the pressure decreases due to a reduction in relative humidity, the particles are drawn together and drying shrinkage occurs. The change in surface energy is thought to be responsible for drying shrinkage occurring at very low humidity (below 40%) when capillary stress and disjoining pressure are no longer present. Solid particles are subjected to a pressure due to surface energy and the pressure is decreased by water adsorbed on the surface. Loss of water will allow the surface energy pressure to increase, resulting in further shrinkage. A significant part of the initial drying shrinkage is irreversible and this is explained by the changes that take place in the C-S-H. When adsorbed water is removed on first drying, additional physical and chemical bonds are formed as the particles become more closely packed. Moreover, additional bonds can occur due to hydration and carbonation (see later). Consequently, the porosity and connectivity of the pore system of the C-S-H change with drying, which reduces ingress of water on re-wetting. Drying shrinkage of concrete is affected by several factors, the main ones as recognised by Codes of Practice being: water/cement ratio, aggregate, relative humidity, size of member and time. Figure 3.4 demonstrates that, for a constant volume of aggregate, drying shrinkage increases as the water/cement ratio increases and, for a constant water/cement ratio, drying shrinkage increases as the volume of the aggregate decreases. The influence of the aggregate is to restrain the shrinkage of the cement paste. Aggregates of low stiffness (low modulus of elasticity) provide less restraint and result in more drying shrinkage than non- shrinking good quality aggregates. Thus, lightweight aggregate concrete has a higher drying shrinkage than normal weight aggregate concrete as shown in Fig. 3.5. The size of aggregate hardly affects shrinkage but, at a constant water/ cement ratio, larger aggregate allows the use of a leaner mix (more aggregate by volume) to achieve the same workability, which results in less shrinkage. Most aggregates are dimensionally stable; however, there are exceptions and the use of aggregates having high drying shrinkage should be avoided. As already mentioned, the relative humidity of the air surrounding the concrete is a main factor and the lower the humidity the greater the loss of water and drying shrinkage. Similarly, an important factor is the size of member. Drying shrinkage of a large member is less than that of a small member because it is more difficult for water to escape from the former, which has a longer drying path The effect of size is expressed as the volume/surface area ratio (V=S) or effective thickness (ˆ 2V=S), the surface area being that exposed to drying (see Fig. 3.6). There is a secondary influence of shape of member on drying shrinkage that is normally neglected. Drying shrinkage occurs over a long period of time with a high initial rate after exposure to drying, which then gradually decreases to a very low rate after several years. It is believed that shrinkage does not have an ultimate value, although for design purposes, a final value after a time of 50 years is often assumed. Typically, a shrinkage-time characteristic would be 20% of 20-year shrinkage occurring in two weeks, 60% occurring in three months and 75% occurring in one year (Neville and Brooks, 2002).
Accelerators, retarders and other chemical admixtures that affect the rate of strength development of concrete will also affect the rate of drying shrinkage; however, the long-term shrinkage is not affected to a great extent. With water- reducing admixtures (plasticisers and superplasticisers) there is a general increase in drying shrinkage of 20% due to the presence of the admixture itself, e.g. as in the case of flowing concrete (Brooks, 1999) but, when used as a water reducer, shrinkage is affected by the change in mix proportions as well as the admixture. Nowadays, shrinkage-reducing chemical admixtures are available, which appear to reduce shrinkage by suppressing hydration of cement. Any reduction in strength can be offset by decreasing the water/cement ratio, thus reducing shrinkage even more. When the mix proportions are unchanged, the use of certain mineral admix- tures (fly ash, blast-furnace slag) as partial replacement of cement do not affect long-term drying shrinkage appreciably (Brooks, 1999). On the other hand, for a constant water/cementitious materials ratio, increasing the level of cement replacement by silica fume and metakaolin reduces both autogenous and drying shrinkage of high performance concrete. For example, drying shrinkage is reduced by approximately 25% for a 10% replacement of cement (Brooks and Megat Johari, 2001). Drying shrinkage causes loss of prestress in prestressed concrete, increases deflections of asymmetrically reinforced concrete and, together with differential temperature, contributes to the warping of thin slabs. As discussed in Section 3.3.6, restraint of shrinkage often leads to cracking. Several methods are available for estimating drying shrinkage and swelling for design purposes and these are considered together with creep in Section 3.2.4.
