Properties and Durability of Aggregate

Surface strain experienced by mortar in wetting - drying cycles and in deicer salt application.

by Peter P. Hudec and Martin Ondrasik

Synopsis: The scaling of concrete and mortar involves sub-parallel de-laminations of material from the surface. To produce this phenomenon, differential stresses parallel to the surface and resulting in differential strain must be active. This research measured the differential strain developed along the surface of specially shaped mortar bars upon their wetting, drying, and osmosis due to application of deicer salts.

Mortar bars were made at a w-c ratio of 0.4 and 0.6 with shaley sand and high quality dolomite as aggregate. The sand is known to cause surface scaling. The bars were cast in a 'half circle' shape. The normally cured samples were dried, and all but the outer surface of the 'half circle' were sealed.. This allowed the ingress of water and solutions from one direction only, such as would occur in an 'infinite' concrete surface. Steel pins were secured to the ends of the half circle to facilitate measurement of the strain.

The strain of the specimens was measured during the following states: 1. dry samples, 2. saturating in water, 3. drying, 4. saturated, placed in saline solutions, 5. then placed in pure water.

The results show that as the water entered or left the surface, stresses developed which were sufficient to deform the ends of the half circle up to 0.6% of the diameter distance. Largest deformations took place upon wetting, followed by those on drying, and the least deformation resulted from osmotic forces. When the samples had equilibrated, i.e., became either fully saturated, dried, or the pore fluid composition equaled that of the saturating medium, the strain was relaxed. Water-cement ratio influenced the time of maximum strain development and aggregate and cement type determined the magnitude of the strain.

Hudec, P.P., and Ondrasik, M., 1997, Surface strain experienced by mortar in wetting-drying cycles and in deicer salt application, Fourth CANMET/ACI International Conference on Durability of Concrete, Sydney, Australia, ACI Special Publication 170-44, pp. 853-878.

Peter P. Hudec, PhD, is the Professor of Engineering Geology at the University of Windsor, Windsor, Ontario, Canada. He has been working in the area of aggregate and concrete technology for the last 30 years. In particular, he has been studying the effect of pore size and pore size distribution of rock aggregate and mortar on their durability properties.

Martin Ondrasik, Msc, was a graduate student at the University of Windsor, and is now an Engineering Geologist with the Slovak Geologic Survey.

INTRODUCTION

Slab concrete or mortar is wetted, dried, heated or cooled from the surface down. The exposed surface undergoes rapid changes under conditions which are only gradually transmitted to the sub-surface. The surface responds to these changes primarily by changing its volume. The surface of the concrete is thus first to show distress when the volume of the surface changes and the resultant stresses exceed its tolerance limits.

The theories offered for deterioration are many and varied. The most often quoted theories deal concern critical saturation, freezing, osmotic pressures, and associated phenomena of thermal shock on cooling freezing and thawing.

Powers (1) suggested hydraulic pressure theory - destructive flow of a high-pressure water in the pores. The driving force is freezing water. Ice with its 9% expansion of volume acts as piston on water which, due to a small size of pore, or due to dissolved chemical content, is not frozen.

Helmuth (2) worked out osmotic pressure theory. Ice forms first in larger pores. This increases concentration of the unfrozen solution, decreases a freezing point of this solution, and induces osmotic migration of water from solution of lower concentration into those with higher concentration. This results in osmotic pressure on pore walls. Fagerlund (3) combined these two theories.

Rosli and Harnick (4) presented as a possible deterioration mechanism a thermal shock idea. Deicing salt applied on a surface of frozen concrete initiate melting of surface ice. Melting ice absorbs heat from a sample. This creates temperature difference of up to 9 C between sample surface and 5 cm deep from the surface, introducing potentially detrimental tension.

Another deterioration mechanism was elaborated by Adkins (5) and Moukwa and Adkins (6). It is based on a sun shine - freezing scaling. A layer of frozen water is trapped between the surface heated by sun and the body of concrete that is heated by inner accumulated heat. Such a phenomenon induces high pressure below the surface and causes scaling.

Scaling due to different deicing salt concentration profiles was suggested by Harnick et al, (7). Salt on the surface and near the surface is often washed away. This results in maximum concentration approximately 10 mm below the surface. Scaling is initiated due to the differential stress. The stress is caused by the freezing of the layers of solution at different temperatures.

Concrete and mortar are relatively impermeable, and the moisture and salt ingress take days to penetrate few millimeters into the surface. As the moisture penetrates, the pores become progressively filled.  Initially, the water is adsorbed on the internal surfaces.  Then, capillary water begins to fill the pores.  The water in capillaries is under tension that is proportional to the capillary radius.  When the capillaries fill and their menisci flatten, the tension is reduced.  With time, the adsorbed water begins to act as an osmotic fluid, and some of the pores come under positive pressure.  Thus, as the moisture conditions in the successive layers change, so do their dimensions. The volume change causes stress; however, the value of the stress is different for each layer. There is no effective way of measuring the differential stress developed in the various layers or mortar or concrete. One aim of this research was to devise a methodology for observing and measuring differential stresses and resultant strains that develop upon uni-directional wetting, drying and salt application to a mortar surface. Mortar rather than concrete was chosen because it allowed working with smaller specimens while still approximating the conditions found in concrete.

BACKGROUND DISCUSSION

Water saturates porous, permeable materials such as soils, rock, and mortar by capillary action. Water is a polar, wetting fluid, and is attracted to mortar and aggregate surfaces. If the surface is in the form of a pore (capillary), it is pulled into it. In a partially saturated solid, the capillary is occupied by both water and air, with a curved meniscus interface between them. The surface tension of water strives to flatten the meniscus, and pulls the water deeper into the solid, working against gravity, resistance to flow, or pressure of entrapped air (or all three). The meniscus remains curved and the water in tension until the capillary is filled. The curvature of the meniscus and the tension developed is a function of the capillary diameter, and can be expressed by the equation:

Equation 1

Where:

H_c = Capillary rise, cm      D_w =density of water, g/cm³

F = surface tension,           g/s² r = capillary radius, cm

G = gravitational constant, cm/sec²        2 = interfacial angle between water and solid

The capillary rise is equal to the capillary tension existing within the water.

Kelvin's equation relates the relative vapour pressure existing over the capillary meniscus of a given radius by the following equation:

Equation. 2

Where:

p, p₀ = vapour pressure over flat and concave (meniscus) surfaces

r = capillary radius, cm

M = molecular weight of liquid, g

R = the Gas constant

d = density of liquid, g/cm³

(In variations of the equation above, molar volume V can be substituted for M)

As the wetting front advances into the dry solid, the relative vapour pressure at the front boundary is low and a function of the median pore size. Only the smallest capillaries are filled. The capillary tension P_c in this zone is correspondingly high. There is thus a differential stress applied across the wetting front. Ahead of the front, in the dry region, the material is relaxed. Behind the wetting front, the entire wetted zone is in tension, since the capillary tension is hydrostatic, and extends across the full length of the capillary system. After the solid is fully saturated, the meniscus flattens, and tension is released.

The theoretical magnitude of this stress shown in table below, calculated from equations 1 and 2. As can be noted, although the calculated magnitude of the stress is relatively small in terms of pascals (N/m²), it becomes very large when considered as a point stress; in terms of N/mm², an approximation of the small area of the capillary, the magnitude increases by a factor of 10⁶.

The pore system discussed above is similar to that expected in mortars and aggregates. Setzer (8) gives the range of capillaries to be found in mortar, and his micro- and meso- capillaries would generate considerable stress upon wetting. Hudec (9) classifies rock pores in aggregates, and his 'force' pores also would result in appreciable differential stresses upon wetting. Thus, theoretically, the differential stresses possible due to both wetting and drying are demonstrated. It remains to measure the actual stresses present.

EXPERIMENTAL DESIGN AND METHODOLOGY

Measurement of a differential stress created in rock or mortar by an advancing wetting front is difficult. The wetting front advance must be uni-directional, i.e., from one surface only. The strains produced must be measured over time at different levels in the specimen as the front advances. Since the strain levels are relatively low, either a very sophisticated equipment is required, or the deformation must be amplified. Considering these requirements, the normal bar or cube test specimens will not yield the desired results.

A novel specimen shape employed in this research allowed the control of all the above-mentioned parameters. A half-circle or a horse-shoe shaped specimen, when wetted from the outside perimeter, transfers the expansion/contraction of the outside layer to the arms or ends of the specimen, in much the same way as metals of two dissimilar thermal expansion coefficients used in thermostatic switches.

The basics of the design are shown in Fig. 1. The specimen is essentially a half circle with a diameter of 13.5 cm, and 1.8 cm x 1.8 cm in cross-section. The side and inner surfaces are sealed, leaving only the lower outside surface exposed.

Stainless steel pins 3 mm in diameter and 9 cm in length are embedded in the ends of the specimens to facilitate the expansion measurement, and to amplify the expansion.

The distance between the pins changes as the relative stress and the corresponding strain in concentric layers of the specimen change. If, for instance, segment 'A' is in tension, the arms (pins) will move outward; converse will happen if the segment 'A' is in compression. The movement is linearly amplified along the pins as a function of the distance from the base of the pins. In most cases, it is the relative contraction or expansion that is more informative than the absolute values.

The measuring points on the pins are established as grooves into which fit the ends of gage callipers. The gauge, sensitive to 0.001 mm was equipped with conductivity meter to indicate proper contact between the steel pins and the gauge Three measuring positions along the pins are established, one near the bottom, one in the middle, and one near the top of the pins.

The specimens are placed in different environments (water, air, salt solution, etc), and briefly removed for measurement. Thus, any desired number of specimens with different properties can be measured at the same time.

SPECIMEN PREPARATION

The fine aggregate used in the preparation of mortar represents two extremes in durability: a natural shaley sand with relatively poor service record, and a good quality manufactured dolomite sand. The two were chosen to see what effect the aggregate has on the differential expansion of the mortar. The cement used was CSA Type 10 (ASTM Type 1).

The sample gradation used was that for normal mortar according to ASTM C-227. The aggregate-cement ratio of 2.25, and water-cement ratio 0.4 and 0.6 were used in specimen preparation. Additionally, a neat cement specimen with w-c ratio of 0.4 was also cast for comparison purposes. The mixing procedure followed ASTM C-305, using a Hobart-type mixer.

The forms were prepared from 4" (10.16 cm) and 6" (15.24 cm) diameter PVC pipes. The pipes, 34 cm in length, were cut lengthwise into two halves. Plexiglass was used at bottom and sides to mount the semi-circular pipes, one inside the other such that the distance between the inner surface of the larger and the outer surface of the smaller pipe was 2 cm. Removable screws were used to hold the plexiglass and the PVC half-pipes together, and as molds for holes into which steel pins were eventually inserted. The finished form is shown in Fig. 2.

The mixtures were placed into the forms from the top in layers and tamped down. After two days, the forms were taken apart, leaving a semi-circular, trough-shaped mortar specimen which was then cut into 1.8cm wide horseshoe shaped pieces. One casting yielded 15 'horseshoes'. The 'horseshoes' were placed in lime water for three days, and then hot cured in lime water at 60⁰ C for 24 hrs. Two 9 cm steel pins were epoxied into the holes at the ends of the 'U'. The inner and side surfaces were sealed initially with Thompson's Water Seal to allow moisture to enter the specimen from the outside surface only. The water seal did not fill the larger pores, so candle paraffin wax was used to additionally seal the specimens. A finished specimen is shown in Figure 3.

EXPERIMENTAL PROCEDURE

The various combinations of water-cement ratios and aggregates produced the following set of specimens:

Most of the specimens were then subjected to the following sequential exposure conditions:

Because of the low permeability of the mortar specimens, the full saturation/dryness of the specimen took well over a month per condition change. The length changes were measured periodically (every two or three days) by means of an mechanical gauge with a sensitivity of +/- 0.001 mm.

Measurements were taken at the three measuring locations on the pins: at the base (d₁), in the middle (d₂), and near the tips (d₃). A typical graph of circumferential length change (strain) on wetting vs. time is given in Figure 4. As shown, the distance between the pins at the three measuring points is contracting (negative strain) proportionately to the points' distance from the specimen.

The actual strain experienced by the specimen can be calculated from the relationship of similar angles as shown in Figure 5. In this case, the outer layer is assumed to be contracting relative to the inner layer. The pins would move out relative to the center of the specimen 'O', and strain would be positive as measured between the pins. The outside surface can be considered as having moved from point (a) to point (a') along the arc. The angle subtended by the arc is equal to the angle that the tangent to the circle at a' (the pin) makes with a line passing through the origin and the original position (a). The angle is related to the arc length by the following expression:

   Equation 3

The angle can be closely approximated by the relationship:

Equation 4

Thus, combining equations 1 and 2, the measured strain is equated to the arc length change, i.e., the deformation of the outer layer, by:

Equation  5

Equation 3 can be applied to a typical measurement as follows: The radius o - a' in the specimens is 65 mm. The length to the furthest measuring points on the arm varies from 5 mm, 40 mm and 90 mm. For given set of measurements, R and arm length are constant. Taking the maximum strains shown in Figure 4 at the three measuring points, the actual radial expansion is calculated as 0.16 mm, or 0.02% of the outside circumference. This value is close to the measured value at the 5 mm pin position. Therefore, rather than converting the results to the radial strain, the measured strain at 5 mm point is used in all the results that are discussed in the following section.

RESULTS AND DISCUSSION

The Wetting Cycle:

The results of the wetting experiments are discussed first. The specimens were dried at room temperature conditions prior to saturation. The response of the specimen to saturation is shown in Figures 6, 7, and 8. The curves represent averages of three mortars for each condition of water-cement ratio and aggregate type.

Figure 6 shows the strain as function of time. Negative strain indicates the expansion of the outer layer relative to the inner layer. The behaviour of the mortars is largely the function of the water-cement ratio. Although the sand contains shale that expands on wetting, its presence does not affect expansion, since mortars containing non-expansive dolomite behave in the same way. The fine aggregate content, however, does affect expansion on wetting, since the neat cement mortar exhibits a behaviour which is quite different from those containing aggregate. The higher w/c ratio specimens show slightly lower expansion and show it much earlier. Changing the cement source (but not the type) as shown by the sample labeled 'cement2' does not affect the results.

The difference in response based on the w-c ratio can be ascribed to the differences in permeability of the mortar. The mortars with higher w/c ratios are more permeable, and allow the water to penetrate half-way into the specimen faster. They are also shown to saturate fully earlier (Figure 7). The 50% degree of saturation corresponds to the maximum expansion of the outer layer at about 1.5 hours for w/c ratio of 0.6, and 20 hours for w/c ratio of 0.4. The neat cement specimen at w/c ratio of 0.4 has a 50% degree of saturation and very much greater expansion at 2.2 hours. The expansion is ascribed to the greater number of micro-capillaries per unit volume of the cement paste as opposed to aggregate occupying the same space in the other samples.

The 50% degree of saturation and maximum expansion times coincide, as expected, and can be used to calculate the rate of water ingress or hydraulic conductivity of the mortar. Thus, for the 0.6 and 0.4 w/c ratio mortars, the hydraulic conductivities are approximately 56 x 10^-5 and 1.4 x 10^-5 cm s^-1 respectively.

The total water content (effective porosity) of the specimens as shown in Figure 8 is also largely the function of the w-c ratios, and is not influenced by the aggregate content. The mortars with w-c ratio of 0.4 have between 3 and 4 percent water at saturation, whereas those at w-c of 0.6 have almost 8% saturation. Neat cement paste has the highest water content at about 13%.

The near-saturation of the specimens is shown by the return of the strain to near 0 in Figures 6 and 7, and, and by the flattening of both the degree of saturation and water content curves in Figure 8. It should be noted that the time required to fully saturate the sample is about 10 times longer than required for 50% saturation for all samples.

The Drying Cycle:

As may be expected, the behaviour of the 'horseshoes' in the drying cycle is reversed - as the outer surface dries, it contracts, and the arms of the sample diverge, i.e., show positive strain. The results are shown in Figures 9, 10, and 11. There are, however, some major difference:. Because the drying is much slower than wetting, and the moisture content in the sample more diffuse, the strains are more muted. There is also less of the influence of w/c ratio on the strain. The shaley sand mortars assume a permanent deformation after drying rather than returning to the original position. Fig. 10 shows that the strain levels are small but constant over a wide range of moisture contents. Only when most of the water has evaporated does the strain decrease and then reverse to indicate permanent deformation of the sample.

The mortar samples show similar behaviour based on the w/c ratio; however, the cement paste sample shows a much greater strain, and a much slower water loss than the mortars.

The maximum strain occurs at about 30 to 40% of total saturation, and is a function of the w/c ratio. Although Fig. 11 shows that the drying is gradual, log-normal in nature, and very much dependent on the pore structure (w/c ratio), the differential strain of drying is relatively constant over wide moisture range of mortars.

Osmotic Effects:

The next sequence of experiments was designed to test the differential strain due to osmotic pressures that are thought to be generated by deicing salts. The mortars were saturated in fresh water, and then exposed to relatively high concentrations of NaCl brine (24% and 31% by weight), and their differential expansion monitored.

The next set of figures illustrate the results. Figure 12 shows that as a fresh-water mortar is placed into the saline environment, the outer surface expands. The expansion is most likely due to the fresh water streaming from inside the specimen towards the saline front on the outer surface. This causes capillary contraction of the inner surface, and expansion of the outer surface. It is interesting to note that the osmotic expansion/contraction is approximately one magnitude less than that experienced upon wetting or drying. The w/c ratio has a significant effect on osmotic expansion/contraction. The more permeable mortar with w/c of 0.6 tends to equilibrate more quickly than the 0.4 w-c mortar. The latter did not equilibriate even after 100 days, but continued to expand at the conclusion of the experiment.

After exposure to salt water of 24% and 31%, the respective specimens of w/c 0.4 and 0.6 were placed in fresh water. Although an contraction of the outer layer was expected, the specimens showed a small expansion which remained constant over the 40-day experimental period. The only explanation offered is that there was insufficient osmotic gradient to cause volume change in this environment.

Figure 13 illustrates the relative degree of saturation of the specimens as function of their strain. The effect of w/c ratio is particularly evident: When water saturated samples are placed in brine solutions, the outer surfaces of the 0.4 w/c ratio mortars contract while loosing water. The 0.6 w/c ratio mortars contract and then expand while gaining moisture. The brine concentration does not have much effect on these processes. When the brine-saturated mortars are placed in fresh water, they show slight contraction, but a significant gain in water. It would seem that the water gained does not exert much force on the pore walls.

CONCLUSIONS:

1. The U-shaped mortar configuration, sealed on all but the outside surface, was quite effective in demonstrating the differential stresses that exist in a mortar and concrete surfaces upon wetting, drying and influence of de-icing salts.

2. The water-cement ratio has a major influence on the strain, more so than the aggregate type. The smaller the pores (lower w-c ration), the greater the effect. Hydrated cement specimen showed the greatest effect. The mortars with w-c ratios of 0.4 and 0.6 had comparable strains, but the w-c 0.4 mortars showed had a more rapid strain development and equalization than the w-c 0.6 mortars. This is explained by the higher permeability rates of the 0.6 w-c mortars.

3. The greatest strains averaging 0.2% were observed upon wetting. About ½ that amount was observed upon drying. Osmotic strains averaged 0.04%, or 1/5 of those developed on wetting.

4. Aggregate type had minor influence on strain development under all conditions. Although only two types of aggregates were used, they represented the two extremes of quality - very high quality, durable dolomite, and a poor quality shaley sand.

5. The brine concentration had little effect on osmotic behaviour, probably because both concentrations of 24% and 31% were relatively high (initial trials with low concentration brines showed minimal response).

6. Fresh-water saturated 0.4 w-c mortars lost moisture when placed in brine; the 0.6 w/c mortars showed a minor gain. Both types of mortars after saturation in brine showed gain in moisture when placed in fresh water. Strains experienced under all osmotic conditions were small and indicated relative expansion of the outer surface.

ACKNOWLEDGMENTS

The above research was supported by a grant from Natural Science and Engineering Council of Canada (NSERC).

REFERENCES

(1) Powers, T.C. 1945, A working hypothesis for further studies of frost resistence of concrete, Journal of the American Concrete Institute, Vol 16, N 4, pp. 245-272.

(2) Helmuth, R.A 1953, Theory of volume changes in hardened cement paste during freezing, Proceedings of the Highway research board, Vol 32, pp. 285-297.

(3) Fagerlund, G. 1975: Studies of the destruction mechanisms at freezing of porous materials, Proceedings of the sixth international congress on the problems raised by frost action, Le Havre, France, Fondation Francaise d'Etudes Nordiques. pp. 166 - 196.

(4) Rosli, A., and Harnick, A.B., 1980: Improving the durability of concrete to freezing and deicing salts, Durability of building materials and components, ASTM STP691, pp. 464-473.

(5) Adkins, D.F. 1986: Laboratory duplication of surface scaling. Concrete International, vol.2, February, pp. 35-39.

(6) Moukwa, M. and Adkins, D.F. 1988: New approach to a concrete scaling test based on field conditions, Cement, Concrete and Aggregates, Vol.10, No.2, pp. 102-108.

(7) Harnick, A.B., Meier, U., Rosli, A. 1978: Combined influence of freezing and deicing salt on concrete Physical aspects, ASTM special technical publication STP691, pp. 474-484.

(8) Setzer, M.J., 1990: Interaction of water with hardened cement paste, in Advances in Cementatious Materials, Amer. Ceramic Soc. Trans., vol 16, pp. 415-439.

(9) Hudec, P., 1987, Deterioration of rocks ans function of grain size, pore size, and rate of capillary absorption of water. Journal of Materials in Civil Engineering, vol. 1, pp. 3-9.

FIGURES

'Horseshoe' mortar & Measuring setup.