Portland cement paste shrinks primarily in response to the removal of free water from its capillary and gel pores. Immediately after mixing, while the paste is a murky fluid, its volume will be reduced as gravity forces the solid cement particles to settle. In this initial stage, the lost paste volume equals the volume of water bled to the surface. Once the paste’s solid structure begins to form, its reduction in volume changes to a decreasing fraction of the water lost. Logically, the mechanisms that give rise to the lion’s share of the paste’s shrinkage throughout the hardening process must involve the replacement of the original liquid water with a gaseous mixture of air and steam (i.e. water vapor).
The shrinkage associated with self-desiccation is termed autogenous shrinkage. It is strictly an internal process that begins as soon as the water is introduced to the mix and persists as long as the cement continues to hydrate. The shrinkage associated with the loss of water to the surroundings is termed drying shrinkage. It is strictly an external process that persists as long as water is able to escape from the paste. Of all factors, evaporation is by far the most significant contributor to the paste’s long-term shrinkage.
Here is the overall picture: All practical portland cement concretes contain large amounts of surplus water, and all start out as grossly aerated, solid-in-liquid suspensions. At water-cement ratios, all the mix water fully occupies two different kinds of spaces within the paste: the capillary pores and the very much smaller gel pores. From the moment of batching, and continuing essentially to eternity, both internal and external processes—self-desiccation, base wicking, bleeding, and evaporation—act to replace the liquid water originally permeating the paste with air (sucked in from the nearest entrapped bubbles) and vaporized water. It is the persistent substitution of humid air for liquid water within the paste’s capillary and gel pores that leads inexorably to the paste’s, and thus the concrete’s, drying shrinkage.
There are two ways that this gas-for-liquid replacement is believed to force the paste’s skeleton to contract: loss of disjoining pressure between adjacent C-S-H crystals and compression of the capillaries by surface tension. For complicated reasons, in an emulsoid (remember that hydrated cement paste is just really hard gelatin), the pressure in the liquid filling the space between two closely adjacent particles is higher than the pressure in the liquid surrounding the particles. Because this higher pressure tends to push the particles apart, when it is relieved by the liquid’s replacement with a gas, the particles tend to move toward one another and thereby shrink the paste’s solid structure.
At equilibrium, six factors determine the height and shape of the boundary—called the meniscus—that forms between two immiscible fluids in an open vertical capillary tube. Four of the factors—the particular surface tensions characteristic of the three interfaces that exist between the various materials present (i.e. the water, steam, and C-S-H), and the radius of the capillary tube itself—are conceivably modifiable.
Given the colossal value that would attend the development of a practical method for eliminating drying shrinkage, this last observation immediately suggests the existence of at least four sure-fire get-rich schemes. Indeed, all you have to do is find an inexpensive way to do just one of the following:
1. Eliminate the surface tension between water and steam.
2. Equalize the surface tensions between water and C-S-H, and air and C-S-H.
3. Equalize the surface tensions between water and C-S-H, and water and steam.
4. Make the capillary and gel pore radii large enough to flatten out the menisci, but still small enough to allow the paste to gain strength.
Although schemes 1 and 4 have already been tried using shrinkage reducing admixtures and autoclaving, schemes 2 and 3 have yet to attract commercial interest.