Results and Discussion

In the first experiment, cement pastes of w/c=0.3 and 0.45 were prepared and immediately placed in their block molds. While these blocks were exposed to a variety of drying conditions [8], here we shall focus on the set of blocks that was immediately exposed to the chamber environment of 50 % RH and 23 ºC. For these specimens, X-ray measurements were performed over a period of 7 d.

After the initial settling (setting) of the cement paste, surprisingly, the drying profiles are observed to change in a relatively uniform manner throughout the thickness (depth) of the specimen, as shown in Fig. 3. This is in contrast to the inward progression of a relatively sharp drying front often observed in porous materials. While these cement paste specimens are only 4 mm to 5 mm thick, similar results have recently been obtained for cement-based systems 50 mm thick using magnetic resonance imaging [14]. For hardened concrete, Selih et al. [15] have observed relatively uniform drying for a few days, followed by the development of a significant moisture gradient originating at the drying surface. In the fresh cement paste, the combination of a relatively wide pore size distribution and a high permeability allows a rapid rearrangement of internal water due to capillary forces, so that the largest pores throughout the specimen thickness empty first. This results in a fairly uniform drying throughout the specimen thickness. The scale over which this mechanism operates in fresh concrete is yet to be determined, but the results of Coussot [14] suggest that it is at least 50 mm, similar to the depth of the steel reinforcement in exposed concrete members. In Fig. 3, it can also be noted that while there is a 1 mm diameter pinhole (collimator) in front of the X-ray detector, the spatial resolution of the system in practice is significantly higher than this as the sharp interfaces present at the tops and bottoms of the specimens seem to be resolvable to the 0.2 mm spacing used in sampling.

  

Figure 3: Normalized density profiles for open block cement paste specimens. Bottom of cement paste specimen is at position 28.5 mm (in system coordinates) and top exposed surface is at position 33 mm to 34 mm.

  

Figure 4: Differential density profiles (relative to 3 h) for layered cement paste (0.3 over 0.45) open to the chamber environment. Exposed top surface is at position 35.5 mm.

  

Figure 5: Relative water content vs. time for exposed layered cement paste specimens.

Even more interesting are the results for layered composite specimens. For example, Fig. 4 shows the differential density profiles for a layered composite consisting of a layer of w/c=0.3 cement paste over a layer of w/c=0.45 cement paste. Even though it is the 0.3 cement paste that is exposed to the drying environment, the 0.45 paste ``dries out'' first (between 3 h and 7 h). Only at later times, after the 0.45 paste has nearly reached complete drying, does the 0.3 paste experience significant water loss. The capillary forces present in the fresh cement paste are easily able to redistribute the water from the lower 0.45 paste layer to the 0.3 paste layer resting on top of it. The results in Fig. 4 are consistent with the mass loss measurements shown in Fig. 5, with a large amount of water loss occurring during the first 10 h to 20 h of exposure, and very little thereafter.

This rearrangement of capillary water is even observed in a capped specimen which loses very little capillary water (only about 10 % over the course of 80 h -- see Fig. 5), as shown in Fig. 6. Here, the more dense paste at the bottom of the specimen is seen to further increase in density (negative values on the differential density plot) at the expense of the less dense paste in the top half of the specimen. In this case, a substantial movement of water is observed to occur between 34 h and 46 h as the finer pore structure of the w/c=0.3 paste imbibes water from the w/c=0.45 paste to replace that "lost" due to chemical shrinkage/hydration.

  

Figure 6: Differential density profiles (relative to 3 h) for layered cement paste (0.45 over 0.3) sealed with a cap.

Differences in the initial pore size distribution of the fresh cement paste can also be produced at a constant w/c by using cements ground to different finenesses (PSDs). In general, coarser cement particles imply coarser "pores" between them [10]. Figures 7 and 8 show the results obtained for both configurations of w/c=0.45 layered composites of the "5 µm" and "25 µm" cements. Regardless of which paste is exposed to the chamber environment, the 25 µm cement paste is always seen to dry out first (3 h to 7 h), while the differential counts for the 5 µm cement paste remain near zero (indicating no loss of water). Once again, the capillary forces draw the water from the coarser pores to the finer ones. Within each layer, however, the drying is still observed to occur uniformly, with no indication of a sharp drying front. Similar experimental observations using magnetic resonance imaging have been made by Coussot et al. [16] for composites consisting of layers of packed glass beads of different sizes, once again with equivalent porosities in each layer.

  

Figure 7: Differential density profiles (relative to 3 h) for layered w/c=0.45 cement paste (25 µm over 5 µm) open to the chamber environment. In this case, the 25 µm cement paste is directly exposed to the environment, with its top surface at position 41 mm.

  

Figure 8: Differential density profiles (relative to 3 h) for layered w/c=0.45 cement paste (5 µm over 25 µm) open to the chamber environment. In this case, the 5 µm cement paste is directly exposed to the environment, with its top surface at position 41 mm.

Composite specimens consisting of a layer of cement paste over a layer of mortar and vice versa (at a constant w/c) were also investigated. In these cases, the drying was relatively uniform throughout both layers simultaneously with no clear indication of differential water movement between the layers, as shown in Fig. 9 for a w/c=0.45 composite. In this case, the coarser pores which most likely exist in the interfacial transition zones (ITZs) surrounding the sand particles in the mortar are balanced by the finer pore structure of the bulk cement paste in the mortar [17]. If the w/c is increased in the ITZs in the mortar, it must also be decreased in the bulk paste relative to the nominal w/c [17]. Thus, most likely, water moves locally from the ITZ to the bulk paste regions within the mortar (which would not be detectable using the X-ray absorption measurements), with little detectable bulk movement from the cement paste to the mortar or vice versa. It should be noted, however, that the observed X-ray absorption results would also be consistent with the mortar being a simple two-phase (paste and sand) composite with no distinguishable ITZ regions. In this case, the cement paste would be nominally identical in the paste and mortar specimens and no differential water movement due to capillary forces would be expected.

  

Figure 9: Differential density profiles (relative to 3 h) for layered w/c=0.45 mortar over cement paste open to the chamber environment. In this case, the mortar is directly exposed to the environment, with its top surface at position 38.5 mm.

Several of the systems studied experimentally were also simulated using the new version of the CEMHYD3D model. For these simulations, the drying rates were chosen to approximately match those observed experimentally (Fig. 5). After each time of hydration, the results were spatially averaged over each consecutive 200-pixel thick layer to obtain data with a spatial resolution the same as that of the experimental data, namely 0.2 mm. Figures 10 and 11 show simulation results for porosity within a 4 mm thick composite specimen consisting of a 2 mm thick layer of w/c=0.45 cement paste covered by a 2 mm thick layer of w/c=0.3 cement paste. The similarity between the results in Fig. 10 and their experimental counterpart shown in Fig. 4 can be clearly observed. Because the fresh cement paste has a high permeability and the capillary forces are relatively large (leading to a rapid rearrangement of the water), we can successfully model the drying process by simply emptying the largest pores in the microstructure first without direct consideration of the kinetics of the water movement.

For the simulated system, little drying is observed to occur within the w/c=0.3 cement paste layer for the first 7 h of exposure, as instead, water is drawn from the w/c=0.45 cement paste layer underneath it. After 70 h, as shown in Fig. 11, while much more water has dried out from the w/c=0.45 paste layer, the empty porosity due to self-desiccation is much less and quite similar within the two layers. With the CEMHYD3D model, information on the degree of hydration of the cement paste is also readily available. Fig. 12 shows the degree of hydration obtained after 72 h of hydration as a function of depth. Because it retains water-filled porosity for a longer time period, the w/c=0.3 cement paste layer is observed to achieve a substantially higher degree of hydration than the w/c=0.45 cement paste. Results for earlier hydration times such as 3 h (not shown) indicate a much more uniform degree of hydration with depth, as the initial drying and emptying of the largest pores has minimal effects on the initial hydration rate [12,13].

  

Figure 10: Simulation results for "dried" porosity fraction vs. depth for layered composite specimen. Positions 0 to 2000 correspond to the w/c=0.45 cement paste layer while positions 2000 to 4000 are the w/c=0.3 paste layer that is directly exposed to the drying environment.

  

Figure 11: Simulation results for porosity fractions vs. depth for layered composite specimen after 70 h of curing. Positions 0 to 2000 correspond to the w/c=0.45 cement paste layer while positions 2000 to 4000 are the w/c=0.3 paste layer that is directly exposed to the drying environment. "Initial water-filled" indicates the initial distribution of capillary porosity in the fresh (0 h) cement paste. "Water-filled" indicates the remaining water-filled porosity (after 70 h of curing). "Dried" indicates empty porosity created due to drying, while "Self-dessicated" indicates empty porosity created due to self-dessication and chemical shrinkage during hydration.

  

Figure 12: Simulation results for degree of hydration vs. depth for layered composite specimen after 70 h of exposure. Positions 0 to 2000 correspond to the w/c=0.45 cement paste layer while positions 2000 to 4000 are the w/c=0.3 paste layer that is directly exposed to the drying environment.

The computer simulations were also applied to simulating the drying/hydration behavior of the systems with different PSDs. Results for drying profiles are provided in Figures 13 and 14, while those for simulated degree of hydration are given in Fig. 15. The simulated drying profiles closely mimic their experimental counterpart shown in Fig. 8. Once again, water is seen to be removed preferentially during drying from the coarser pore size distribution microstructure. In this case, the differences in degree of hydration between the two layers (Fig. 15) are even more striking as the 5 µm cement achieves a much higher degree of hydration than the 25 µm cement, due to both its ability to retain water and to its inherently higher hydration rate (because of its smaller size cement particles and higher contents of C₃S and C₃A).

  

Figure 13: Simulation results for "dried" porosity fraction vs. depth for layered composite specimen. Positions 0 to 2000 correspond to the coarser 25 µm cement paste layer while positions 2000 to 4000 are the finer 5 µm paste layer that is directly exposed to the drying environment.

  

Figure 14: Simulation results for porosity fractions vs. depth for layered composite specimen after 60.2 h of curing. Positions 0 to 2000 correspond to the coarser 25 µm cement paste layer while positions 2000 to 4000 are the finer 5 µm paste layer that is directly exposed to the drying environment. Symbol labels are the sames as those defined in the caption of Fig. 11.

  

Figure 15: Simulation results for degree of hydration vs. depth for layered composite specimen after 60.2 h of exposure. Positions 0 to 2000 correspond to the coarser 25 µm cement paste layer while positions 2000 to 4000 are the finer 5 µm paste layer that is directly exposed to the drying environment.

While the model and experimental results compare favorably, two phenomena that are not included in the model results shown above are the settling of the cement particles (bleeding) prior to the setting of the paste and carbonation reactions between the calcium hydroxide produced during cement hydration and CO₂ present in the environment. These reactions proceed rapidly at intermediate relative humidities [18]. Preliminary efforts to include these two effects in the model drying codes appear promising. For example, simulations of a 1.2 mm thick cement paste (initial w/c=0.6) undergoing evaporation/settling/carbonation have been conducted to compare to available experimental data [3,4]. It has been assumed that after settling/evaporation, the final thickness of the paste is 1.0 mm. Drying and carbonation rates were taken from available experimental data [3,4], and it was assumed that all carbonation results in the formation of the calcite form of calcium carbonate. Since the molar volume of calcite (36.93 cm³/mol) is greater than that of calcium hydroxide (33.08 cm³/mol) and the reaction occurs on a 1:1 molar basis, extra calcium carbonate pixels are (probabilistically, p=0.1164) generated at the reaction sites to maintain the appropriate volume stoichiometry for the carbonation reaction. Also, the water released from the calcium hydroxide during carbonation is made available for further hydration of the cement.

A comparison of the experimental scanning electron microscopy/image analysis (SEM/IA) results for capillary porosity and the ratio of anhydrous cement to solid material (anhydrous cement + hydration products) [4] to those based on the simulations of varying complexity are provided in Table 1. It is observed that the best quantitative agreement is observed when both settling and carbonation are included in the model hydration/drying codes. Thus, the development and evaluation of these enhanced codes will be the subject of future research.

 

Table 1: Hydration Models of Varying Complexity

	Experimental	Hydration +	Hydration + Settling	Hydration + Settling/Carbonation
% Anhydrous/Solids	40.	38.	48.	43.
% Porosity	40.	50.	44.	46.

Finally, Figures 16 and 17 provide three-dimensional microstructures from the top surface and an interior 100 pixel x 100 pixel x 100 pixel section of a simulated 1 mm thick w/c=0.6 microstructure (no settling or carbonation). In these figures, empty porosity is white, water-filled porosity is blue, and the unhydrated cement particles are red. In comparing the two images, one can definitely observe the preferential drying at the immediate top surface (due to the larger pores present there) and a more "uniform" drying within the interior of the specimen, as the larger pores at all thicknesses are the first to empty during the evaporation process. The emptying of these pores at the top surface is not observable using the current X-ray equipment due to limitations in spatial resolution, 0.1 mm experimentally vs. about 0.015 mm in Fig. 16.

  

Figure 16: Three-dimensional image (100 pixels x 100 pixels x 100 pixels) of top surface of cement paste (w/c=0.60) microstructure after 2.9 h of hydration/drying. Dried (empty) porosity is white, water-filled porosity is blue, and unhydrated cement particles are red.

  

Figure 17: Three-dimensional image (100 pixels x 100 pixels x 100 pixels) of middle section of cement paste (w/c=0.60) microstructure after 2.9 h of hydration/drying. Dried (empty) porosity is white, water-filled porosity is blue, and unhydrated cement particles are light red.

The practical implications of these experimental and computer modeling studies are potentially quite significant. Two areas where a correct understanding of water movement are critical are the curing of high performance concrete and the application of repair materials [8]. Care must be taken when applying a curing membrane (or compound) to concrete to ensure that the pores in the curing layer are coarser than those in the fresh concrete. If this is not the case, the so-called curing layer will remain saturated to the casual observer while it is continuously drawing water from the concrete underneath it. Obviously, this will result in a concrete that does not achieve its potential hydration, strength development, or durability even though "the surface appears to remain saturated." These same principles of differential water movement due to capillary forces also apply to the use of saturated lightweight fine aggregates to supply internal curing water for high performance concretes [19,20]. In the area of repair materials, the differential water movement between the substrate and the repair material are critical to both the proper hydration of the repair material and the development of an adequate bond between the two layers.