FROST HEAVE AND WATER UPTAKE RELATIONS IN VARIABLY SATURATED AGGREGATE BASE MATERIALS PAPER NO

FROST HEAVE AND WATER UPTAKE RELATIONS IN VARIABLY SATURATED AGGREGATE BASE MATERIALS PAPER NO. 03-4391 By W. Spencer Guthrie, E.I.T. Texas Transportation Institute Texas A&M University College Station, TX 77843-3135 (979) 845-9919 Fax (979) 845-1701 s-guthrie@ttimail.tamu.edu and Åke Hermansson Swedish National Road and Transport Research Institute, VTI SE-581 95 Linköping, Sweden +4613204072 Fax +4613141436 ake.hermansson@vti.se Transportation Research Board 82 nd Annual Meeting January 12-16, 2003 Washington, D.C. EQUIVALENT WORD COUNT: 6,209

Guthrie and Hermansson 1 ABSTRACT The occurrence of frost heave in soils and aggregates can be attributed to the redistribution of water in the soil profile. Frost heave testing performed in this study on 71 variably saturated specimens of aggregate base material indicates that while the uptake of new water from outside the soil body is a primary source of moisture in the formation of segregation ice, internal water residing within the soil or aggregate structure can also serve as an important supply of water to the freezing front. This paper provides a brief overview of frost heave concepts relating to unsaturated soil conditions, describes the laboratory methodology employed in this study, and discusses relationships between the physical properties of the specimens and their frost heave behavior. Degrees of saturation ranging from 45 to 84 percent were evaluated, and heave-uptake ratios as high as 2.24 were calculated. Ratios less than 1.09 suggest that sufficient porosity exists in the sample matrix to allow the formation of ice without causing frost heave, and higher ratios designate samples which are nearly saturated and which experience substantial upward redistributions of existing water during the initial freezing process that give rise to measurable heave even before additional water is imbibed by the sample. This paper suggests that the entry of air into freezing soils and aggregates plays an important role in their frost heave behavior. Key words: Frost Heave, Aggregate Base Materials, Soil Freezing, Water Uptake

Guthrie and Hermansson 2 INTRODUCTION In cold climates, the use of frost-susceptible base and subgrade materials in pavement construction gives rise to the potential for frost heave damage in roadways. When free moisture is available in subsurface pavement layers, its upward movement toward the freezing zone can lead to the formation of ice lenses and the occurrence of differential frost heave in the pavement section. During thawing the accumulation of meltwater in the upper roadbed can cause a markedly reduced bearing capacity of the pavement that often requires specification of load restrictions during spring seasons to avoid excessive pavement damage. Because of these negative effects on pavement performance, the flow of moisture in freezing soils and aggregates remains an important topic in cold regions transportation research. In explaining the frost heave behavior of freezing soils, numerous authors have partitioned total heave into two types (1, 2, 3, 4). The first is in-situ freezing, or primary frost heave, where existing water is frozen in place at its original location within the soil matrix. Primary heave contributes only a small amount to the total heave in highly frost-susceptible soils (1, 4). The second is segregation freezing, or secondary frost heave, where water flows toward the freezing front under thermally induced suction gradients to create ice lenses. For a constantly saturated soil column with a supply of free water at its lower boundary, the occurrence of segregation freezing is directly related to the uptake of new water from outside the soil body (3, 5). That is, in laboratory test analyses, secondary frost heave of saturated specimens is completely attributed to the freezing of incoming water. Consequently, considering the nine percent expansion of water upon freezing, the volumetric heave that occurs in the secondary mode is presumed to be larger than the volumetric water uptake by a factor of 1.09. However, in natural environments where soils and aggregates are variably saturated, the relationship between frost heave and water uptake during freezing cannot be so readily defined. For example, the redistribution of existing water and the entry of air within the soil matrix can give rise to measurable frost heave without the addition of any new water at all. Conversely, the expulsion of air during freezing of a static water structure may lead to relatively reduced amounts of frost heave (1). In order to quantify the relationship between frost heave and water ingress occurring in variably saturated freezing specimens, this work defines a heave-uptake ratio and investigates the range in heave-uptake ratios that can occur in aggregate base materials tested at natural water

Guthrie and Hermansson 3 contents. Frost heave tests were conducted at the Texas Transportation Institute (TTI) on 71 variably saturated specimens representing aggregate base materials from Indiana, Minnesota, Pennsylvania, Texas, and Virginia. This paper provides a brief overview of frost heave concepts relating to unsaturated soil conditions, describes the laboratory methodology employed in the experimentation, and discusses relationships between the physical properties of the specimens and their frost heave behavior. FROST HEAVE BEHAVIOR With the onset of winter in cold climates, soil water begins to freeze as sufficient heat flows across the pavement surface from underlying layers into the air. At the freezing front, not all pore water changes into ice at the same temperature, however (6, 7). In order to be frozen, water molecules must have sufficient mobility to be physically rearranged into the crystal structure of ice (8). Mechanisms that prevent freezing include adsorption to mineral surfaces, pore water salinity, and compression forces. Water molecules that remain unfrozen as a result of these mechanisms exist as water films in the soil matrix and strongly influence the hydraulic transport properties of the affected layers (9). Because of the intermingling of individual effects, a clear picture illustrating the separate influence of each mechanism on the structure of unfrozen water networks in a freezing system may not be easily established. Nonetheless, research has demonstrated that in many cases a sufficient quantity of water remains unfrozen in a freezing soil or aggregate for water to flow in the pavement in response to suction gradients (10). While water molecules tightly bound to the mineral surface may not be free to move, water molecules in unfrozen saline films can migrate in freezing soil systems to create ice lenses (8). Limited by the hydraulic properties of the medium and the rate of heat removal, the flow of water and the formation of ice can begin to cause frost heave of the soil matrix when the pore space at a given depth within the frozen zone reaches saturation (11). As the ice crystals then grow to form ice lenses, usually parallel to the pavement surface, the aggregate particles may be forced apart, generating a heaving pressure on the overlying pavement layers. As the particle-toparticle contact is increasingly disrupted by ice inclusions, the effective stress in the soil matrix is reduced. The weight of the overburden is subsequently transmitted from the soil matrix to the ice-water structure, leading to a pressurization of the unfrozen water films along the soil-ice

Guthrie and Hermansson 4 interfaces (11). Additional water can enter the film only when the matric suction at the interface is sufficient to overcome the counteracting overburden stresses and the matric suction of the underlying soil from which the water would come (6, 11). Thus, depending on the suction that develops in the freezing zone relative to the suction of the underlying soil matrix, water can be redistributed from lower layers upwards toward the freezing front. This implies that frost heave could occur even in closed systems given a sufficiently high level of saturation and the possibility of air intrusion or expansion. In such a case, assuming a stationary frost front, moisture in lower layers would be transported to upper layers until the difference in suction potentials between them was ultimately balanced by the drying of lower layers. The redistribution of water under thermally induced suction gradients is the premise for the discussion of laboratory results given in this paper. TEST PROCEDURES The laboratory test program included 71 specimens representing aggregate base materials from Indiana, Minnesota, Pennsylvania, Texas, and Virginia. Samples were tested in standard plastic cylinders of 152.4-mm inside diameter and 304.8-mm height. At approximately 6 mm above the outside bottom of the mold, 1.6-mm-diameter holes were drilled around the circumference of the mold at a horizontal spacing of 12.7 mm. One hole was also drilled in each quadrant of the bottom of the mold about 50 mm from the center. Aggregate samples were scalped on the 25-mm sieve and compacted at their respective optimum moisture contents in four lifts of 50 blows each with a 4.5-kg hammer dropped from a height of 457 mm to a finished height of about 200 mm inside the pre-drilled mold. After drying at 40 C for four to six days, the specimens were placed in a 12.7-mm-deep bath of distilled water at room temperature for a 10-day soak period as shown in Figure 1. The moisture profiles developed at the end of this conditioning were considered natural water contents representative of those likely to exist in the field given the availability of moisture. Specimen weights and surface dielectric values were monitored during soaking to assure equilibration of the moisture profile by the end of the conditioning period. Dielectric values are most sensitive to the amount of unbound water in a soil matrix (12). Afterwards, the specimens were moved with the water bath to an environmental chamber maintained at 17 ºC for continued soaking and frost heave measurements through another

Guthrie and Hermansson 5 period ranging from 9 to 62 days. As shown in Figure 2, styrofoam insulation was added around the samples to control the direction of frost penetration. Thermostat-controlled heat tape was utilized in the water bath to keep the water from freezing during the test. A rigid plastic cover with a shallow, 2-mm-diameter hole drilled in the center was placed on the surface of each aggregate sample to facilitate frost heave measurements. Increases in the heights of the tested specimens were determined by measuring the distance from the top of a rigid metal bar placed in turn across the top of each sample container to the bottom of the center hole in each cover. Measurements were made with a digital caliper to a resolution of 0.0025 mm. The air temperature in the environmental room and the water temperature in the bath were checked regularly, with the latter being consistently around 5 C for the duration of the testing. With this testing arrangement, surface temperatures just under the sample covers usually did not descend below 4 C, likely due to constant heat conduction from the bath upwards through the samples. Thus, with an estimated 9 C temperature difference over a sample height of 200 mm, the maximum temperature gradient in the specimens was about 0.045 C/mm. The bath water level was also monitored throughout the testing and replenished as necessary. At the conclusion of frost heave testing, specimens were weighed to determine the amount of water uptake that occurred in each, and then the samples were oven-dried to facilitate computation of moisture contents before and after freezing. Washed sieve analyses and specific gravity testing were also performed on many of the specimens. TEST RESULTS The results of testing and data analysis are given in Figures 3 through 11, which address relationships between optimum moisture contents, dielectric values, porosities, water contents before and after freezing, frost heave measurements, degrees of saturation, fines contents, and amounts of water uptake during freezing. In each figure, the data is divided according to the duration of the frost heave testing to which the specimens were ultimately subjected. Figures 5 and 8 are based on data from 56 specimens whose specific gravities were determined, and Figure 9 is based on 62 specimens for which washed sieve analyses were performed. All other figures contain data for all 71 specimens. Figure 3 depicts the relationship between the optimum moisture contents and the gravimetric water contents attained by the specimens during the 10-day conditioning period just

Guthrie and Hermansson 6 prior to frost heave testing. In general, the soaked gravimetric water contents approximated the optimum moisture contents at which the specimens were compacted. Figure 4 shows the relationship between surface dielectric values and gravimetric water contents measured at the end of the equilibration period. The depth of investigation of the dielectric probe used in this study was approximately 25 mm. As noted earlier, dielectric values are most sensitive to the amount of free water in a soil or aggregate matrix so that higher water contents yield higher dielectric values. Figure 5 presents the relationship between sample porosities and gravimetric water contents attained by specimens during the conditioning period. This figure suggests that greater void space within the pore structure increases the water storage capacity, which, depending on the size of the pores and the availability of water, could lead to increased natural water contents of aggregate base materials in the field. Figure 6 displays the correlation between the gravimetric moisture contents after conditioning and the gravimetric moisture contents after freezing for each of the freezing times employed in the frost heave testing. In this study, the water content after freezing was between 3 and 70 percent greater than the water content before freezing, with an average increase of 29 percent. As can be observed from the figure, the scatter among data points for the 9-day through 27-day freezing tests makes these trends practically indistinguishable from one another, while the water contents after freezing are somewhat higher for the 62-day tests. The strength of the overall correlation in Figure 6 lies in the fact that the water content before freezing was not an arbitrary moisture content, but a natural condition resulting from the unique suction and hydraulic characteristics of each material that also govern moisture flow during freezing. While only 23 of the 71 specimens imbibed more than their optimum moisture content during the soaking period, with an average excess of 1.3 percent moisture, the optimum moisture contents of 54 specimens were exceeded during the freezing process, with an average excess of 2.5 percent moisture. In the most extreme case, a limestone specimen with a high fines content reached a moisture level of 7.8 percent above its optimum moisture content after freezing for 11 days. Upon thawing in the field, this additional water would lead to extreme conditions of supersaturation in the roadbed. Figure 7 shows the relationship between the gravimetric water contents after freezing and the amounts of frost heave exhibited by the tested specimens. In general, higher water contents after freezing were associated with greater amounts of frost heave, an observation also illustrated

Guthrie and Hermansson 7 in Figure 8. However, some specimens with relatively high water contents did not exhibit substantial heave, presumably because of the presence of salts in the pore water that depressed its freezing point (13). In several such cases, limited electrical conductivity measurements obtained during the conditioning period confirmed this hypothesis. Variation in the amounts of frost heave experienced by different specimens at similar water contents suggests that the freezing characteristics of pore water and the extent of its redistribution within the soil matrix during frost heave testing varied substantially among the specimens. Figure 9 gives the relationship between frost heave measurements and the fines contents of tested specimens, where greater amounts of frost heave were generally associated with greater fines contents. Figure 10 addresses the frost heave behavior of specimens tested in this work relative to the heave-uptake ratio of 1.09 discussed earlier. In the case of unsaturated soils, the air space that exists within the soil matrix at the onset of freezing must become filled with water and ice at a given depth within the freezing zone before substantial frost heave can occur in that region. During this period of imbibition, the majority of the water entering that locality would not be contributing toward frost heave, but toward saturation, and the expected ratio of volumetric heave to volumetric water uptake would be less than 1.09. This is the case for all the data points located beneath the 1.09 slope line in Figure 10. Only after a soil element is saturated does additional incoming water contribute completely toward frost heave as it freezes, bringing the heave-uptake ratio to 1.09 for that element. If the new water arriving at the saturated zone was originally held by lower layers within the soil column at a moisture content higher than can be sustained during freezing, so that air has displaced some of the original water in the lower layers, then the overall heave-uptake ratio for the soil body may be greater than 1.09, which is the case for all the data points located above the 1.09 slope line in Figure 10. Because water from outside the soil body as well as water present within the matrix at the time of freezing can be transmitted from lower layers upwards to the freezing front, specimens with higher water contents can experience volumetric frost heaves much greater than the expansion that would be predicted from the combined effects of primary and secondary frost heave when the latter is based only on the new water entering the soil body. That is, existing water is not limited to participating in only the primary mode of frost heaving. If air entry into subsurface pavement layers is possible, which is a reasonable expectation in the field, then the observed magnitude of frost heave for an unsaturated material can far exceed the amount

Guthrie and Hermansson 8 predicted by using a heave-uptake ratio of 1.09. In fact, as illustrated in Figure 11, the most frost susceptible specimens in this study exhibited heave-uptake ratios as high as 2.24, indicating an especially efficient use of existing water in the creation of ice lenses within the soil structure. These results are consistent with the recent findings of at least one other similar study (14). In a project in which frost heave tests were performed on samples of unsaturated, undisturbed medium silt, a heave-uptake ratio of 1.7 was measured, and the researcher concluded that for initially saturated fine-grained soils, the volumetric heave can be as much as three times greater than the volumetric water uptake during the first few days of freezing. Subsequent tests using samples at natural moisture contents demonstrated the occurrence of substantial heave without the addition of any new water at all (14). Thus, the source of water in such cases does not necessarily need to be an external supply of free water. Internal water held within a soil or aggregate matrix by capillary forces can also lead to the formation of ice lenses and subsequent frost heave of the soil column. CONCLUSION The frost susceptibility of pavement base layers is a critical aspect of materials selection for roadway construction in cold climates. The occurrence of frost heave in soils and aggregates can be attributed to the redistribution of water in the soil column. Frost heave testing of variably saturated specimens of aggregate base material performed in this study indicates that while the uptake of new water from outside the soil body is a primary source of moisture in the formation of segregation ice, internal water residing within the soil or aggregate structure at the time of freezing can also serve as an important supply of water to the freezing front. In conditions of constant saturation, the expected ratio of volumetric heave to volumetric water uptake is 1.09, but when air entry and expulsion are permitted, this ratio can vary significantly from soil to soil depending on the unique suction, hydraulic, and freezing characteristics of the medium. Materials that experience substantial frost heave will have higher heave-uptake ratios than those that are less frost susceptible. Heave-uptake ratios as high as 2.24 were calculated for specimens tested in this study, for example, demonstrating an especially efficient use of existing water in the creation of ice lenses. Lower ratios suggest that sufficient porosity exists in the sample matrix to allow the formation of ice without causing frost heave.

Guthrie and Hermansson 9 Because of the potential for roadway damage due to frost heave and the subsequent reduction in bearing capacity due to the accumulation of meltwater during thawing, identification of frost susceptible materials remains an important task for pavement engineers. While acknowledging that further research is needed to gain additional understanding of the relationship between frost heave and water uptake in variably saturated specimens during the freezing process, this paper suggests that the entry of air into freezing soils and aggregates plays an important role in their frost heave behavior. ACKNOWLEDGMENTS The authors wish to express appreciation to Stephen Sebesta of TTI for his assistance in preparing many of the samples tested in this study. Lee Gustavus of TTI was instrumental in facilitating laboratory needs. REFERENCES 1. Taber, S. Freezing and Thawing of Soils as Factors in the Destruction of Road Pavements. Public Roads, Vol. 11, No. 8, 1930, pp. 113-132. 2. Miller, R. D. Freezing and Heaving of Saturated and Unsaturated Soils. In Highway Research Record 393, TRB, National Research Council, Washington, D.C., 1972, pp. 1-11. 3. Loch, J. P. G. Influence of the Heat Extraction Rate on the Ice Segregation Rate of Soils. Frost i Jord, No. 20, 1979, pp. 19-30. 4. Konrad, J.-M. and N. R. Morgenstern. A Mechanistic Theory of Ice Lens Formation in Fine- Grained Soils. Canadian Geotechnical Journal, Vol. 17, No. 4, 1980, pp. 473-486. 5. Konrad, J.-M. Soil Freezing Characteristics Versus Heat Extraction Rate. National Research Council of Canada, DBR Paper No. 1257, 1984. 6. Penner, E. Ground Freezing and Frost Heaving. Canadian Building Digest, CBD-26, National Research Council, Canada, 1962. http://www.cisti.nrc.ca/irc/cbd/cbd026e.html. Accessed Aug. 1, 2001. 7. O Neill, K. and R. D. Miller. Exploration of a Rigid Ice Model of Frost Heave. Water Resources Research, Vol. 21, No. 3, 1985, pp. 281-296. 8. Anderson, D. M. The Interface between Ice and Silicate Surfaces. Journal of Colloid and Interface Science, Vol. 25, No. 2, 1967, pp. 174-191.

Guthrie and Hermansson 10 9. Kaplar, C. W. Phenomenon and Mechanism of Frost Heaving. In Highway Research Record 304, TRB, National Research Council, Washington, D.C., 1970, pp. 1-13. 10. Anderson, D. M. and N. R. Morgenstern. Physics, Chemistry, and Mechanics of Frozen Ground: A Review. In Permafrost: North American Contribution to the Second International Conference, NAS, Washington, D.C., 1973, pp. 257-288. 11. Cary, J. W. A New Method for Calculating Frost Heave Including Solute Effects. Water Resources Research, Vol. 23, No. 8, 1987, pp. 1620-1624. 12. Saarenketo, T. Electrical Properties of Water in Clay and Silty Soils. Journal of Applied Geophysics, Vol. 40, 1998, pp. 73-88. 13. Guthrie, W. S., Hermansson, Å., and T. Scullion. Determining Aggregate Frost Susceptibility with the Tube Suction Test. In Cold Regions Engineering: Cold Regions Impacts on Transportation and Infrastructure, American Society of Civil Engineers, Reston, VA, 2002, pp. 663-674. 14. Hermansson, Å. Frost Modelling and Pavement Temperatures. Luleå University of Technology, Licentiate Thesis, 2000.

Guthrie and Hermansson 11 LIST OF FIGURES 1. Specimen Conditioning 2. Insulation Required for Frost Heave Testing 3. Relationship Between Optimum Moisture Content and Water Content Before Freezing 4. Relationship Between Dielectric Value and Water Content Before Freezing 5. Relationship Between Porosity and Water Content Before Freezing 6. Comparison of Water Contents Before and After Freezing 7. Relationship Between Water Content After Freezing and Frost Heave 8. Relationship Between Degree of Saturation Before Freezing and Frost Heave 9. Relationship Between Fines Content and Frost Heave 10. Variation in Frost Heave with Water Uptake 11. Variation in Heave-Uptake Ratio with Frost Heave

Guthrie and Hermansson 12 FIGURE 1 Specimen conditioning.

Guthrie and Hermansson 13 FIGURE 2 Insulation required for frost heave testing.

Guthrie and Hermansson 14 Water Content Before Freezing (%) 14 12 10 8 6 4 2 0 4 5 6 7 8 9 10 11 Optimum Moisture Content (%) 9 10 11 12 17 27 62 Freezing Days FIGURE 3 Relationship between optimum moisture content and water content before freezing.

Guthrie and Hermansson 15 Water Content Before Freezing (%) 14 12 10 8 6 4 2 0 4 9 14 19 24 29 34 Dielectric Value Before Freezing 9 10 11 12 17 27 62 Freezing Days FIGURE 4 Relationship between dielectric value and water content before freezing.

Guthrie and Hermansson 16 Water Content Before Freezing (%) 12 10 8 6 4 2 0 14 16 18 20 22 24 26 28 30 Porosity Before Freezing (%) 9 10 11 12 17 27 62 Freezing Days FIGURE 5 Relationship between porosity and water content before freezing.

Guthrie and Hermansson 17 Water Content After Freezing (%) 18 16 14 12 10 8 6 4 2 0 2 4 6 8 10 12 14 Water Content Before Freezing (%) 9 10 11 12 17 27 62 Freezing Days FIGURE 6 Comparison of water contents before and after freezing.

Guthrie and Hermansson 18 Frost Heave (mm) 40 35 30 25 20 15 10 5 0 0 5 10 15 20 Water Content After Freezing (%) 9 10 11 12 17 27 62 Freezing Days FIGURE 7 Relationship between water content after freezing and frost heave.

Guthrie and Hermansson 19 Frost Heave (mm) 40 35 30 25 20 15 10 5 0 40 50 60 70 80 90 Degree of Saturation Before Freezing (%) 9 10 11 12 17 62 Freezing Days FIGURE 8 Relationship between degree of saturation before freezing and frost heave.

Guthrie and Hermansson 20 Frost Heave (mm) 40 35 30 25 20 15 10 5 0 0 5 10 15 20 25 30 35 40 Percent Passing 0.075 mm Sieve (%) 9 10 11 12 17 27 62 Freezing Days FIGURE 9 Relationship between fines content and frost heave.

Guthrie and Hermansson 21 700 600 Frost Heave (cm 3 ) 500 400 300 200 100 0 0 100 200 300 400 500 600 Water Uptake (cm 3 ) 9 10 11 12 17 27 62 Freezing Days 1.09 Slope FIGURE 10 Variation in frost heave with water uptake.

Guthrie and Hermansson 22 2.5 Heave-Uptake Ratio 2.0 1.5 1.0 0.5 0.0 0 5 10 15 20 25 30 35 40 Frost Heave (mm) 9 10 11 12 17 27 62 Freezing Days FIGURE 11 Variation in heave-uptake ratio with frost heave.