Mechanical Characterization of an Artificial Clay

Transcription

1 Mechanical Characterization of an Artificial Clay A. Turan 1 ; S. D. Hinchberger 2 ; and M. H. El Naggar 3 Abstract: Glyben is an artificial soil comprising bentonite mixed with glycerin that has been used recently in scaled model tests to study seismic soil structure interaction. In spite of recent interest in glyben, factors affecting the dynamic properties of this material have not been well established. This paper presents the results of vane shear tests, cyclic triaxial tests, resonant column tests and bender element tests undertaken to characterize the dynamic properties of glyben. The results show that the modulus ratio of glyben decreases with increasing shear strain amplitude similar to that observed for natural clays. However, there are significant thixotropic changes in the properties of glyben after mixing bentonite with glycerin. In addition, glyben exhibits time-dependent volumetric compression after the application of isotropic consolidation pressure, the damping ratio of glyben is higher than that of natural clays and the dynamic properties of glyben are strongly influenced by temperature. These factors should be considered when interpreting the results of scaled physical model tests using glyben. DOI: / ASCE :2 280 CE Database subject headings: Geotechnical models; Soil dynamics; Clays; Damping; Shear; Thermal factors; Cyclic tests. Introduction Scaled physical modeling is an economic and effective approach to study soil structure interaction during earthquakes. One of the challenges of such studies is to obtain a model soil that can be scaled to adequately simulate the seismic response of prototype soils. Several natural and synthetic soil mixtures have been proposed for scaled model tests involving structures founded on or in clay deposits e.g., Seed and Clough 1963; Seah 1990; Meymand Although, most model soils have proven to be useful, their properties can be strongly influenced by stress history, thixotropy, and consolidation during spin-up in centrifuge tests. In addition, model soils can be difficult to prepare and place in laminar shear boxes and they can generally be used only once due to drying and consequent desiccation during tests. These characteristics can limit the usefulness of model soils, especially for centrifuge tests. Thus, there is a need for an artificial soil that has fewer limitations compared to conventional model soils. This paper investigates the effect of time or thixotropy, temperature, confining pressure, and cyclic stress history on the dynamic properties of glyben. Other characteristics such as Atterberg limits, vane shear strength, and compaction behavior are also presented. Glyben see Mayfield 1963; Kenny and Andrawes 1997; Rayhani and El Naggar 2006 is a synthetic clay 1 Research Assistant, Dept. of Civil and Environmental Engineering, the Univ. of Western Ontario, London Ont., Canada N6A 5B9. aturan@uwo.ca 2 Assistant Professor, Dept. of Civil and Environmental Engineering, the Univ. of Western Ontario, London Ont., Canada N6A 5B9 corresponding author. shinchberger@eng.uwo.ca 3 Professor, Dept. of Civil and Environmental Engineering, the Univ. of Western Ontario, London Ont., Canada N6A 5B9. helnaggar@ eng.uwo.ca Note. Discussion open until July 1, Separate discussions must be submitted for individual papers. The manuscript for this paper was submitted for review and possible publication on September 14, 2006; approved on May 2, This paper is part of the Journal of Geotechnical and Geoenvironmental Engineering, Vol. 135, No. 2, February 1, ASCE, ISSN /2009/ /$ mixture comprising bentonite and glycerin. The stiffness, shear strength and damping of glyben can be varied by altering the percentage of glycerin by mass g/c in the mixture. Glyben behaves as a cohesive material, it consolidates but at a very slow rate compared to natural clays, and as such, it possesses properties that are favorable for scaled model tests requiring cohesive soil behavior e.g., where particle size scaling is not required. Although glyben has been used recently for model tests involving cohesive soil response e.g., Rayhani and El Naggar 2006, factors affecting the dynamic properties of this material are not well established. Consequently, a series of vane shear, cyclic triaxial, resonant column, and bender element tests were conducted to characterize the dynamic behavior of glyben. The results of this study are considered to be of interest to researchers designing and conducting scaled physical model tests at 1 g and in a centrifuge. Background Several modeling soils have been developed for use in scaled physical modeling applications. In some cases, researchers have used reconstituted soils for scaled model tests see Nunez and Randolph 1984; Burr et al. 1997; Moss et al When reconstituted soils are used, the soil is normally slurried and placed in a test container. Then it is consolidated to obtain the desired strength and stiffness. This process can be performed either prior to or during spin-up in centrifuge tests, but it can be impractical for 1-G tests, due to the large quantity of soils required and the length of time needed for consolidation. In addition, reconstituted soils generally cannot satisfy all similitude criteria such as those for the undrained shear strength and dynamic shear modulus. As an alternative to reconstituted soils, a wide variety of artificial and natural soil mixtures have been developed including: supersil, plastellina, aerosil, veegum, silicon gum, plasticine, polyacrylimide, and modified sands, clays, and clayey silts. Tavenas et al developed an artificial model soil using kaolinite, Portland cement, and bentonite to simulate brittle Champlain clay. Overconsolidated soils have been modeled by Ko et al. 1984, using a kaolinite water mixture, and by Blaney and Mallow 280 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING ASCE / FEBRUARY 2009

2 Table 1. Summary of Dynamic Tests Specimen Test g/c % Parameter investigated Tx-1, Tx-2, Tx-3 Cyclic triaxial a 47.5 Thixotropic effects Fig. 4 St-1, St-2, St-3, St-4 Cyclic triaxial a 47.5, 45, 42.5 and 40 G and Figs. 5 and 6 Src-1 Resonant column a 45 G and Figs. 5 and 6 Ct-1 Cyclic triaxial b 47.5 G and versus confining pressure Fig. 7 Cr-1 Resonant column b 45 Effect of confining pressure Fig. 8 Nc-1 Cyclic triaxial c 42.5 Effect of loading cycles Fig. 9 T-1, T-2 Cyclic triaxial a and bender elements 47.5 Temperature effects Figs. 10 and 11 R-1, R-2, R-3, R-4 Cyclic triaxial a 47.5 Repeatability Fig. 13 V-1, V-2, V3 Cyclic triaxial c 42.5 Verification tests for G and Figs. 5 and 6 Mv-1 bentonite and water w=45% Cyclic triaxial c 45 Coefficient of consolidation Fig. 14 Mv-2 45 a Procedure 2: cell pressure held constant, shear-strain amplitude ramped. b Procedure 1: cell pressure ramped and shear-strain amplitude ramped at each cell pressure. c Procedure 3: cell pressure held constant, shear-strain amplitude constant using fumed silica mixed with bentonite and water. Biscontin and Pestana 2001 studied the influence of torque rate on the vane shear strength of a lightly cemented bentonite kaolinite mixture. Iskander et al modeled an artificial transparent clay using amorphous silica. The most extensive studies involving model soils for seismic applications have been conducted at University of California, Berkley UCB. In 1963, Seed and Clough 1963 developed a kaolinite, bentonite, and water mixture w=200% for 1-G shake table studies. The seismic characteristics of the UCB soil can be found in Kovacs Sultan and Seed 1967, Arango-Greiffenstein 1971, Bray 1990, Lazarte 1996, and Meymand 1998 subsequently used UCB soil with some modifications to study the seismic performance of structures, such as clay core dams, and piles and pile groups in clay. Glyben is a mixture of sodium bentonite and glycerin that seems to overcome some of the shortcomings of traditional synthetic modeling soils. Mayfield 1963 conducted a series of triaxial tests on glyben and concluded that it behaved as a cohesive u =0 material. Sutherland 1988 studied the uplift capacity of piles in cohesive soils using glyben and noted that there was negligible desiccation of glyben at room temperature. Later, Kenny and Andrawes 1997 conducted undrained triaxial tests and vane shear tests and noted that glyben behaved as a purely cohesive material, u =0, during quick loading. In addition, Kenny and Andrawes 1997 observed that compacted glyben gave excellent repeatability during testing. Recently, Rayhani and El Naggar 2006 investigated the seismic performance of glyben using resonant column, T-bar, and in-flight hammer tests during centrifuge tests. They concluded that glyben exhibited trends of modulus reduction and increased damping that were similar to soft and medium clays. However, Rayhani and El Naggar 2006 found that the damping ratio of glyben at small strain amplitudes was higher than that measured for natural clays. In general, these studies point out a general interest in glyben and a need for further characterization of this material. Testing and Methodology In this study, a series of compaction tests, vane shear tests, Atterberg limit test, cyclic triaxial, bender element, and resonant column tests were performed to investigate the mechanical properties of glyben. Details of these tests and the procedures followed are given in the following sections. Table 1 provides a detailed list of each specimen, the type of test performed, and its glycerin or water contents. Material Preparation Glyben was prepared by mixing bentonite with glycerin in a kneading type mixer Blakeslee, model B-20, Chicago, IL for at least 30 min. Both bentonite and glycerin were blended very slowly in the mixer over a period of about min to ensure as uniform a mixture as possible. After mixing, specimens were then prepared by compacting the glyben into a split mold in four lifts using a drop hammer. The number of lifts and blows with the drop hammer were selected so that the bulk density of each specimen was 95% of the maximum bulk density determined from standard compaction tests. Specimens were subsequently removed from the split mold after compaction, wrapped in plastic and stored at room temperature 22 1 C until testing. Unless otherwise stated, specimens were tested at least 5 days after preparation to avoid thixotropic effects. Compaction, Vane Shear, and Atterberg Limit Tests A series of standard geotechnical tests were performed to provide preliminary characterization of glyben. Compaction tests were conducted in accordance with ASTM D-698 on glyben mixtures with glycerin contents g/c of 35, 37.5, 40, 42.5, 47.5, and 50%. In conjunction with the compaction tests, shear vane tests were performed according to ASTM D-2573 to measure the shear strength of glyben. For the vane shear tests, glyben was compacted into a 20 cm deep stiff metal container 30 cm 30 cm to 95% of the maximum bulk density. Then vane measurements Pilcon, 19-01, Hampshire, England were taken in the container over a period of time as the glyben cured. In addition, glyben mixtures with different g/c ratios were prepared and the liquid limit LL and plastic limit PL were determined in accordance with ASTM D Cyclic Triaxial Testing A Wykeham Farrance cyclic triaxial apparatus was used for this study. The triaxial apparatus is a digitally controlled, servopneumatic, closed-loop system, which controls three parameters: axial stress, confining pressure, and backpressure. Axial load is applied JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING ASCE / FEBRUARY 2009 / 281

3 by a double acting digitally controlled 5 kn pneumatic actuator and a coaxially mounted displacement transducer provides a feedback signal to the control system for precise displacement control and data acquisition. Although the actuator can generate frequencies up to 70 Hz, the testing frequency is dependent on the type of sample tested. In this study, strain controlled testing was used in accordance with ASTM D-3999 Method B. Tests were performed on glyben specimens with a g/ c, percent glycerin by mass, of 40, 42.5, 45, and 47.5%. During each test, a specimen was placed on the triaxial pedestal with a top cap. A latex membrane was placed over the specimen using o-rings to seal the membrane against the top cap and base pedestal. Finally, the triaxial cell was filled with water, the top cap was connected to the actuator and the cell water was pressurized to the desired cell pressure, c. Backpressure was not applied during the tests i.e., the backpressure valve was open to air. This was considered to be more representative of conditions during scaled physical model tests, where glyben is compacted in a laminar shear box and tested either at 1 G or in a centrifuge without application of backpressure. The cyclic triaxial tests were carried out at cell pressures, c, of 50, 100, 200, and 300 kpa using a sinusoidal peak-to-peak strain controlled loading applied to obtain axial strain amplitudes of 0.1, 0.2, 0.4, 1, 2, 5, and 10%. Table 1 provides a summary of the tests. In total, 21 glyben specimens and 1 bentonite water specimen were tested and the dynamic properties were measured using three different procedures. For Procedure 1, individual glyben specimens were placed in the triaxial cell and the cell pressure was subsequently ramped from 50 to 300 kpa. At each cell pressure, the dynamic properties were measured at shear strain amplitudes of 0.1, 0.2, 0.4, 1, 2, 5, and 10% in accordance with ASTM D3999 Method B. This procedure, which is referred to as ramped cell pressure and ramped shear strain amplitude in Table 1, was used during testing of Specimen Ct-1 to study the effect of confining pressure on the dynamic properties of glyben. For Procedure 2, specimens were placed in the cyclic triaxial cell and the cell pressure was held constant, whereas the dynamic properties were measured at shear strain amplitudes of 0.1, 0.2, 0.4, 1, 2, 5, and 10% e.g., cell pressure held constant and shear strain amplitude ramped. Procedure 2 was used to study the influence of thixotropy on the dynamic response of Specimens Tx-1, Tx-2, and Tx-3, the effect of shear strain amplitude on the response of Specimens St-1, St-2, St-3, and St-4, temperature effects on Specimens T-1, and T-2 and to investigate measurement repeatability using Specimens R-1, R-2, R-3, and R-4, respectively. Finally, select specimens were tested individually at constant cell pressure using only one shear strain amplitude. This procedure, which is labeled Procedure 3 in Table 1, was performed on Specimens V-1, V-2, and V3 to verify the results obtained using Procedures 1 and 2 described earlier. In addition, Procedure 3 was used during tests conducted on Specimens Mv-1 and Mv-2 to study the effects of consolidation on the dynamic properties of glyben see Table 1. For each shear strain level and consequent data point, 20 load cycles were used to measure the shear modulus and damping ratio. Fig. 1 shows a typical hysteresis loop obtained from the testing. The dynamic shear modulus, G, and damping ratio,, were obtained from each hysteresis loop as illustrated in Fig. 1. Full details of the data interpretation for these tests can be found in Turan et al Fig. 1. Typical elastic modulus and damping ratio from hysteresis loop Bender Extender Element Tests Bender elements see, Viggiani and Atkinson 1995; Jovicic et al. 1996, Lee and Santamarina 2005 were used to measure the small-strain dynamic properties of glyben. In this study, however, a bender extender element system was used manufactured by IPC Global, which is capable of generating square, sinusoidal, and user defined compression waves and shear waves P waves and S waves. In this test program, sinusoidal p waves and s waves were generated at frequencies of 0.33, 0.5, 1, 2, and 5 khz and the corresponding wavelengths were measured. A final measurement was then made using a frequency that produced a wavelength equal to about half the sample thickness. This approach has been reposted to minimize near-field effects in bender element tests see Viggiani and Atkinson For each test, 100 mm diameter and 100 mm thick cylindrical specimens were prepared by compacting glyben into a split mold to achieve 95% of the maximum bulk density and tests were performed on both unconfined specimens and confined specimens in a triaxial cell. For each bender element test, a glyben specimen was placed on a triaxial pedestal that had a p-wave and s-wave transmitter embedded in it. A top cap equipped with an embedded receiver was placed on the top of each specimen. The tips of both the transmitter and receiver penetrated 3 mm into the specimen. For confined tests in the triaxial cell, a latex membrane was placed over the specimen with o-rings on both the top cap and bottom pedestal to seal the specimen and to permit application of cell pressure. Unconfined specimens were tested in the same manner, but without applying a membrane or cell pressure. The primary measurement from bender element tests is the peak to peak arrival times of s waves, T S, and p waves, T P, and the distance, d, between the transmitter and receiver tips e.g., Viggiani and Atkinson From d, T S and T P, the s-wave velocity, V S, and p-wave velocity, V P, can be calculated from which the small-strain dynamic shear modulus, G max, and Poisson s ratio, may be deduced. Turan et al gives details of typical p-wave and s-wave traces and their interpretation. Resonant Column Tests The resonant column is widely used to measure the small-strain dynamic properties of soil. A description of the apparatus and 282 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING ASCE / FEBRUARY 2009

4 Table 2. Variation of G max /C u and Bulk Density with g/c Ratio Glycerin content g/c G max /c u Bulk density kg/m Fig. 2. Variation of compacted bulk density and cohesive strength versus glycerin content applicable procedures can be found in Drnevich et al and Morris and Delphia In this study, resonant column tests were performed on glyben specimens with g/c=45% in accordance with ASTM D-4015 using torsional loading. These tests were performed according to Procedures 1 ramped cell pressure and ramped shear strain amplitude and 2 constant cell pressure and ramped shear strain amplitude described previously for cyclic triaxial tests. Specimen preparation for resonant column tests was identical to that described earlier for the triaxial tests. First, 70 mm diameter and 145 mm high specimens were prepared by compacting glyben into a split mold to achieve 95% of the maximum bulk density. Then, the specimen was assembled in the resonant column and a latex membrane was placed over the specimen and sealed to the bottom pedestal and top loading cap using o-rings. Procedure 1 was used to Test Specimen Cr-1 see Table 1 using ramped cell pressures, c, of 100, 200, and 300 kpa and ramped shear strain amplitudes ranging from to 0.12%. The tests performed on Cr-1 were used to study the effect of confining pressure on the dynamic properties of glyben. Specimen Src-1 see Table 1 was tested according to Procedure 2 using a cell pressure of 100 kpa and shear strain amplitudes that were ramped from to 0.12%. The results from tests on Specimen Src-1 were used to study the effect of shear strain amplitude on the shear modulus and damping ratio. From the resonant column test, the fundamental angular frequency, n, is measured and the shear wave velocity, V S, is calculated from n see Turan et al for additional details. respectively see Fig. 2. Thus, the glycerin content by mass g/c may be varied to achieve a wide range of shear strength. Table 2 summarizes the small strain shear modulus of glyben versus bulk density. The G max values were measured using bender elements. As shown Table 2, the G max /C u ratio of Glyben was found to vary from 224 at a bulk density of 1,645.4 kg/m 3 to 306 at a bulk density of 1,676 kg/m 3. The Atterberg limits of glyben were also determined. The LL, PL, and plasticity index PI were found to be 59, 34, and 25, respectively. Thixotropic Effects The thixotropic behavior of glyben was investigated using a combination of vane shear and cyclic triaxial tests. Fig. 3 compares the vane shear strength of glyben, at g/c=47.5%, with the vane shear strength of bentonite water prepared with a water content of 65%. The different pore fluid contents were required to obtain equal vane shear strength e.g., about 18 kpa after 10 days, which is often a scaled parameter in scaled model tests. As shown in Fig. 3, glyben reached a steady-state vane shear strength of 18 kpa in 4 days whereas it took 9 days for the vane shear strength of the bentonite water mixture to equilibrate. The shear strength of glyben increased by 12.5% after mixing, whereas the strength of the bentonite water mixture increased by about 10% Fig. 3. Consequently, there are significant thixotropic increases in the vane shear strength for both glyben and the bentonite-water mixture. The time frame over which the thixotropic changes occur is longer for bentonite water than for glyben. Fig. 4 shows the results of cyclic triaxial tests performed, 1, 5, and 20 days after mixing bentonite and glycerin. The shear modulus and damping ratio are plotted in Fig. 4 versus shear strain amplitude for glyben with g/c=47.5%. The results indicate that Results Compaction, Vane Shear Strength and G max /C u The results of standard Proctor compaction and vane shear test are summarized in Fig. 2. The results summarized in Fig. 2 correspond to the maximum bulk densities from each mixture, not the maximum dry density. The maximum bulk density of glyben was 1,765 kg/m 3 for mixtures with g/c=40%. In addition, for the investigated range of g/c, there is a linear increase in the vane shear strength versus decreasing glycerin content. Natural soils exhibit a similar increase in strength with decreasing water content. The vane shear strength of glyben was found to range from 12 to 64 kpa for glycerin contents g/c ranging from 50 to 35%, Fig. 3. Variation of the cohesive strength of glyben and bentonitewater mixture versus time JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING ASCE / FEBRUARY 2009 / 283

5 Fig. 4. Variation of shear modulus and damping ratio versus time there is a time dependent increase in shear modulus of about 32% over the first 5 days after mixing for a shear strain amplitude,, of 0.2%. The damping ratio decreased by about 21% at the same level of shear strain amplitude over the same time period 5 days. However, from 5 to 20 days, the results in Fig. 4 show that there are very minor changes in the shear modulus and damping ratio of glyben. These changes, which may potentially be due to some diagenetic process, would introduce some small but tolerable experimental error in scaled model tests. Effect of Shear Strain Amplitude Fig. 5 shows variation of the modulus ratio, G/G max, versus shear strain amplitude for the various glyben mixtures studied. Modulus ratio measurements were obtained at both low and high shear strain amplitudes. For comparison purposes, Fig. 5 also shows typical curves of G/G max versus shear strain amplitude for natural soils from Vucetic and Dobry 1991 and Ishibashi and Zhang 1993 corresponding to a PI of 25%. As discussed earlier, the tests undertaken in this study were performed at about 95% of the maximum bulk density of each mixture see Fig. 2 as determined by standard Proctor compaction tests. Fig. 5 shows four significant trends with respect to the dynamic shear modulus of glyben for tests undertaken at a confining Fig. 6. Damping ratio versus shear strain amplitude stress, c, of 100 kpa. First, at high shear strain amplitudes cyclic triaxial tests, the measured modulus ratio, G/G max, is close to that of natural clays with PI=25%. Second, the effect of g/c on G/G max is not that significant for glycerin contents between 42.5 and 47.5%. Third, at low shear strain amplitudes resonant column tests, the modulus ratio, G/G max, plots closest to the curve for natural clay obtained from Ishibashi and Zhang Fourth, there is a separation or gap in the data from the resonant column and cyclic triaxial measurements. This can be attributed to the difference in mode of loading in resonant column compared with the cyclic triaxial apparatus and the different excitation frequencies Kim et al.1991; Stokoe et al. 1995; D Onofrio et al. 1999; Matešić and Vucetic Fig. 6 summarizes the damping ratio,, versus shear strain amplitude for tests conducted at a confining pressure, c, of 100 kpa at both low and high shear strain amplitude. In addition, curves from Vucetic and Dobry 1991 and Ishibashi and Zhang 1993 corresponding to PI=25% are also presented for comparison purposes. Referring to Fig. 6, it can be seen that the damping ratio of glyben is significantly higher than that expected for natural clays. At a shear strain amplitude of 0.02%, the damping ratio is about 0.15 whereas natural soils lie within the range of 0.02 and These results, in terms of the high damping ratio, are similar to the findings of Rayhani and El Naggar In addition, results for bentonite and water are also plotted in Fig. 6. It can be seen that the damping ratio of bentonite and water mixtures is comparable to that of natural clays at low shear strain amplitudes. Thus, glyben has a damping ratio that is significantly higher than that of natural soils due to the viscosity of glycerin as discussed in the upcoming section entitled Temperature Effects. Fig. 5. Normalized shear modulus versus shear strain amplitude Effect of Confining Stress Cyclic triaxial tests were also conducted on glyben specimens with g/c of 47.5% to investigate the effect of confining stress on the dynamic properties. Fig. 7 shows the dynamic shear modulus and damping ratio measured at confining stresses of 50, 100, 200, 300, and 500 kpa and shear strain amplitudes of 0.2, 0.4, and 1%. It should be noted that these tests were undertaken immediately after application of isotropic confining stress without allowing time for consolidation of the glyben. The consequence of this will be discussed later in this paper. From Fig. 7, it can be seen that 284 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING ASCE / FEBRUARY 2009

6 Fig. 7. Variation of shear modulus and damping ratio with confining pressure at high strain levels Fig. 9. Variation of shear modulus and damping ratio with number of cycles there is a clear increase in the dynamic shear modulus as the confining pressure increases. Conversely, the damping ratio decreases as the confining pressure increases. This type of behavior is commonly encountered in natural soils e.g., Teachavorasinskun et al. 2002; Cai and Liang Similar behavior is evident from the results shown in Fig. 8 for resonant column tests. The resonant column results show a similar effect of confining pressure on the dynamic stiffness and damping properties of glyben for g/c=45% as compared to g/c=47.5% in the cyclic triaxial tests. Influence of the Number of Cycles Fig. 9 summarizes the effect of the number of loading cycles on the dynamic shear modulus and damping ratio of glyben with a g/c=42.5%. The triaxial tests were performed at a confining pressure of 100 kpa, a frequency of 1 Hz and shear strain amplitude of 1%. A relatively high shear strain amplitude was used because it was considered to be a severe test of the impact of cycles on the dynamic properties of glyben. From Fig. 9, it can be seen that the number of cycles has a relatively small impact on both the dynamic shear modulus and damping ratios for the conditions considered. The dynamic shear modulus was found to vary from 2,320 to 2,370 kpa and the damping ratio varied from 0.20 to These variations are relatively small and in the order of 1 and 5% for the shear modulus and damping ratio, respectively. Consequently, the dynamic response of glyben does not appear to be strongly affected by the number of loading cycles for at least up to 500 cycles. Such behavior suggests that glyben is not structured like some natural soils e.g., Leroueil and Vaughan Temperature Effects It has been shown that glyben has a damping ratio that is considerably higher than that of natural soils. This could be attributed to the different physical properties of glycerin compared to water the normal pore fluid. Table 3 summarizes the bulk modulus, viscosity, dielectric constant, and unit weight of both glycerin and water. From Table 3, it can be seen that the bulk modulus of water and glycerin are comparable. However, the viscosity of glycerin is several orders of magnitude greater than that of water Dorsey In addition, the viscosity of glycerin changes significantly between 20.3 and 37.8 C and as a result, temperature could have a significant effect on the engineering properties of glyben. Cyclic triaxial tests c =100 kpa and unconfined bender element tests were performed to investigate the effect of temperature on the dynamic properties of glyben at low and high strain amplitudes. The results of these tests are summarized in Figs. 10 and 11. For the bender element tests, two different procedures were adopted to measure the influence of temperature on the smallstrain dynamic modulus of glyben. The first test procedure involved heating a glyben specimen with g/c=47.5% to 40 C in an Table 3. Temperature-Dependent Variation of the Viscosity and Dielectric Constant of Glycerin and Water Material Viscosity, MPa s Bulk modulus N/m 2 23 C Dielectric constant Density kg/m 3 Fig. 8. Variation of shear modulus and damping ratio with confining pressure at low strain levels Water C C 1, C C Diluted glycerin a C C 1, C C Glycerin 1, C C 1, C C Note: From Dorsey 1940 and Huck et al a 50% glycerin and 50% water solution. JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING ASCE / FEBRUARY 2009 / 285

7 Fig. 10. Effect of temperature on the shear wave velocity of glyben g/c=47.5% Fig. 12. Variation of damping ratio with pore fluid viscosity oven for 24 h. The specimen was subsequently removed from the oven and bender element tests were performed on the specimen as it cooled in air to 22 C. A thermocouple was inserted into the core of the sample to measure the internal temperature. Based on a finite element analysis of the test Turan et al. 2006, the thermocouple provides an approximate measurement of the internal temperature of the specimen due to the temperature gradients that develop during sample cooling. In addition to the cooling test, four specimens were heated in an oven to 22, 25, 30, and 37 C, respectively for 24 h and tested immediately upon removal from the oven. The results obtained from both of these test procedures are plotted in Fig. 10. As shown in Fig. 10, temperature has a significant effect on the small strain dynamic shear modulus of glyben. At 22 C, the small strain dynamic shear modulus, G max, is 5,159 kpa whereas it decreases significantly to 2,632 kpa at 37 C. The decrease in G max is greater than 45% over the temperature range from 22 to 37 C. Fig. 11 summarizes the results of cyclic triaxial tests undertaken at temperatures of 23 and 28 C. For this series of tests, the triaxial specimen and cell water were heated to the desired temperature prior to measuring the cyclic properties. Fig. 11 shows similar trends in behavior as shown in Fig. 10. For all shear strain Fig. 11. Effect of temperature on the dynamic shear modulus and damping ratio of glyben g/c=47% amplitudes, there is at least a 10% reduction in the dynamic shear modulus of glyben when the temperature is increased from 23 to 28 C. Temperature has a similar effect on the damping ratio where the damping ratio is typically 7 10% higher at 23 C compared with that measured at 28 C. The results of cyclic triaxial and bender element tests illustrate the significant influence of temperature on the dynamic properties of glyben. As only the viscosity of glycerin is strongly influenced by temperature see Table 3, it appears that variations in the damping ratio and shear modulus of glyben versus temperature can be attributed to variation of the viscosity of glycerin in the pore space and in the viscous double layer. glycerin is a polar molecule. Fig. 12, which shows the damping ratio of glycerin versus pore fluid viscosity, further suggests a correlation between pore fluid viscosity and variations in the damping ratio of glyben. Repeatability of Dynamic Properties A series of unconfined bender element and confined cyclic triaxial c =100 kpa tests were conducted on glyben specimens with g/c=47.5% to assess the reproducibility of the laboratory results. For each type of test, glyben specimens were prepared from three different batches of bentonite and glycerin. Each test specimen was compacted to 95% of the maximum bulk density see Table 2 and then tested using either bender elements or a cyclic triaxial apparatus. Table 4 summarizes the bender element results and Fig. 13 summarizes the results of cyclic triaxial tests conducted at shear strain amplitudes between 0.1 and 10%. From Table 4, it can be seen that measurements of the small strain dynamic properties of glyben were generally reproducible for the number of tests performed. The maximum variation was about 3% for tests conducted at a frequency of 0.33 khz. From Fig. 13, it can be seen that measurements of the large strain dynamic modulus and damping ratio were also fairly reproducible; although less reproducible than the bender element results. Using the cyclic triaxial apparatus, the measured dynamic modulus tended to vary by not more than 50 kpa at all levels of shear strain. In addition, the percent variation in the measured parameters tended to increase from about 3% at a shear strain amplitude of 0.2% to about 15% at a shear strain amplitude of almost 10%. Such variations are typical of soils. 286 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING ASCE / FEBRUARY 2009

8 Table 4. Small Strain Level Reproducibility of Dynamic Glyben Properties Bender Element Test Frequency khz Batch number s-wave velocity m/s p-wave velocity m/s Standard deviation S wave Fig. 13. Large strain level reproducibility of dynamic glyben properties Fig. 14. Degree of consolidation versus dimensionless time for glyben and bentonite water mixture Consolidation Behavior of Glyben To conclude, the consolidation response of glyben is summarized in this section. Fig. 14 compares the consolidation response of a typical bentonite water specimen and glyben specimen during isotropic consolidation. Both specimens were prepared with the same void ratio see Fig. 14, placed in a cyclic triaxial apparatus without lateral strip drains and isotropically consolidated for 10 days at a cell pressure of 200 kpa. The axial and radial strains were measured during compression to obtain volumetric strain. Again, backpressure was not applied during the consolidation phase to simulate conditions similar to those in scaled model tests. The large strain shear modulus =0.1% and damping ratio were measured periodically during consolidation and the shear wave velocity was measured before and after consolidation. Shear modulus measurements are also presented in Fig. 14 as described in the following. Referring to Fig. 14, the volumetric response of bentonite water after application of a hydrostatic confining pressure of 200 kpa is fairly typical and can be interpreted using conventional consolidation theory. From the measured response, the coefficient of consolidation of the bentonite water specimen is cm 2 /s. For the glyben specimen, there is also time-dependent volumetric strain after application of isotropic confining stress, which can likewise be interpreted with conventional consolidation theory. As glycerin is a polar molecule with a dielectric constant comparable to that of water see Table 3, it is expected that glycerin should form a double layer with bentonite and that the resultant glyben mixture would consolidate with time after the application of confining stress. Thus, Fig. 14 shows the interpreted degree of consolidation of glyben versus dimensionless time, T v =c v t/d 2. To derive the percent consolidation of glyben, it was assumed that the drained bulk moduli of the bentonite water and glyben specimens are equal at equal void ratio note: 100% consolidation of the glyben specimens could not be achieved in a practical time. From this interpretation, it can be seen that the degree of consolidation for glyben after 10 days is approximately 21% see the symbols in Fig. 14. In addition, the coefficient of consolidation, c v, for the glyben specimen is approximately cm 2 /s, which is about 1/1,382 times the c v of bentonite water. The ratio of c v for bentonite water versus c v for glyben is comparable to the ratio of the viscosity of water to glycerin 1/1,244 at 23 C. Overall, the consolidation process of glyben appears to be much slower than that of bentonite water. Further, the slow rate of consolidation may be attributed primarily to the high viscosity of glycerin noting that there are probably other factors affecting the response such as variations of the glycerin bentonite double layer thickness, the characteristics of the double layer and differences in the free pore space available for pore fluid flow. Last, changes in the dynamic properties of glyben during isotropic consolidation in a triaxial cell are presented in Fig. 14. Large strain shear modulus, G, values obtained from cyclic triaxial tests at the start of the consolidation t=0, and t=4 and 9 days after the start of consolidation are denoted by the drop down arrows this figure. In addition, shear wave velocity measurements made before and after consolidation using bender elements are presented in the lower left corner of Fig. 14. Table 5 shows additional measurements of G during isotropic consolidation from resonant column tests. From the bender element results in Fig. 14, it can be seen that the shear wave velocity of glyben changed by about 1.9% due to the time-dependent volume change JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING ASCE / FEBRUARY 2009 / 287

9 Table 5. Changes in Dynamic Properties due to Consolidation Resonant Column Test Time a Confining stress kpa Strain level % that occurs during 9 days of consolidation. Similarly, there is a corresponding 8% increase in the large strain =0.1% dynamic modulus, G, of glyben over the same time frame. Thus, the time-dependent compression of glyben that occurs after application of confining stresses is expected to introduce some experimental error in scaled model tests. For 1-g model tests, confining stresses are negligible and similarly the volumetric compression of glyben should be negligible as well. Thus, it would be sufficient to account for thixotropy and temperature effects alone in 1-g model tests. For n g centrifuge tests, however, confining stresses are significant. Fortunately, the time frame for most centrifuge tests is less than about 1 h. In addition, accounting for typical drainage paths in n g model tests, the degree of consolidation and consequent change in the dynamic properties of glyben would be much less than 8% with proper experimental design. For conventional soils, consolidation effects are more significant as shown in Fig. 14, which would require full consolidation of the material before testing to reduce experimental error adding considerable time for both 1-g and centrifuge tests. Summary and Conclusions S-wave velocity m/s Damping ratio 5 min days a Time after application of confining pressure in the resonant column. Glyben is an artificial soil comprising glycerin mixed with water and it is suitable for modeling the undrained response of cohesive soils, only, during scaled model tests conducted at 1 g or ngin a centrifuge. This paper has presented the results of an experimental investigation into the effect of time or thixotropy, temperature, strain amplitude, and cyclic stress history on the mechanical properties of glyben. The testing program described earlier comprised vane shear tests, compaction tests, cyclic triaxial tests, bender element tests, resonant column tests, and isotropic consolidation tests and the results should be sufficient to assist with experimental design of scaled model tests at 1 g and ngin a centrifuge. The following is a summary of the results and conclusions arising from this study. 1. The vane shear strength of glyben decreases as the glycerin content g/c increases. The vane shear strength of natural clays exhibit a similar decrease with increasing moisture content. 2. The maximum density of glyben in standard Proctor compaction tests was 1,765 kg/m 3 at a glycerin content g/c of 40%. 3. There were significant thixotropic changes in the vane shear strength, dynamic shear modulus, and damping ratio of glyben during the first 5 days after mixing bentonite and glycerin. There were small and tolerable changes in these properties beyond five days after mixing. Thus, it is concluded that scaled model tests should ideally be undertaken after allowing sufficient time for thixotropic changes to take place. 4. The modulus ratio, G/G max, of glyben is generally within the normal range for natural clays for shear strain amplitudes between 0.2 and 10%. For strain amplitudes between 0.01 and 0.2%, G/G max is slightly higher than that expected for natural clays. Such behavior is satisfactory for use in scaled model tests simulating cohesive soil response. 5. There is a step or gap in the modulus ratio, G/G max, of glyben at 0.2% shear strain amplitude see Fig. 5, which can be attributed to the different modes and frequencies of loading used in resonant column tests torsion at about 20 Hz versus cyclic triaxial tests compression at 1 Hz. 6. The damping ratio,, of glyben is higher than that of natural soils for all levels of shear strain amplitude investigated. This has to be accounted for when interpreting the results of dynamic tests at small shear strain amplitudes using glyben. 7 The damping ratio and shear modulus of glyben are dependent on the confining stress. As the confining stress increased the damping ratio decreased and the dynamic shear modulus increased. Changes in the damping ratio and dynamic modulus were found to be significant for confining stresses up to 200 and 300 kpa, respectively. 8. The dynamic shear modulus, G, and damping ratio,, of glyben were not significantly affected by the number of loading cycles from 1 to 500 cycles for the high level of shear strain amplitude investigated. This behavior would permit multiple dynamic tests from a single scaled model improving the economics and efficiency of such tests. This is one advantage of glyben over other model soils. 9. Temperature has a significant effect on the dynamic properties of glyben. From Figs. 10 and 11, G max, G, and varied by as much as 20% over the temperature range 23 to 28 C. From this behavior, it is concluded that the temperature effects should be accounted for and/or the temperature controlled during the scaled model tests. 10 Glyben exhibits volume change versus time after application of confining stresses, which can be interpreted with conventional consolidation theory. The coefficient of consolidation of glyben is approximately cm 2 /s, which is about 1/ 1,382 times that of coefficient of consolidation of bentonite water. In addition, the ratio of c v is comparable to the ratio of the viscosity of water to the viscosity of glycerin 1/1,244 at 23 C. 11 There is 8% change in the small strain dynamic properties of glyben with time during the first 9 days after application of confining stresses. These changes are due to the timedependent consolidation of glyben. Although 8% variation of the dynamic properties appears to be significant, it is possible to design and conduct centrifuge tests, which control such changes to tolerable levels by limiting the test duration and controlling the drainage path of the model. 12. Finally, glyben has the following advantages over other model soils. First, it can be compacted into place instead of resedimented in a slurry form. This improves the ease of handling. Second, it consolidates at a very slow rate after the application of confining stress, which can be used during experimental design to avoid a prolonged consolidation phase during spin up in a centrifuge. Last, glyben does not desiccate significantly with time and it can be used multiple times during tests as its dynamic properties do not undergo permanent changes at large strain levels and load cycles. This last advantage permits multiple tests to be performed on a single scaled model, without significant alteration of the dynamic response of the model soil. 288 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING ASCE / FEBRUARY 2009

10 Acknowledgments The research reported in this paper has been funded by grants from the National Research Council of Canada and the University of Western Ontario Academic Development Fund. Notation The following symbols are used in this paper: c cohesive strength; d distance between transmitter and receiver tips; E eq dynamic elastic modulus; f frequency at which the testing was performed; G dynamic shear modulus; G/G max shear modulus reduction factor; g/c percentage of glycerin in glyben mixture by mass ; T P arrival time for P waves; T S arrival time for S waves; V p compression wave velocity; V s shear wave velocity; w water content; shear strain; a axial strain; Poisson s ratio; damping ratio; axial stress; a c isotropic confining pressure; friction angle; and n fundamental angular frequency. References Arango-Greiffenstein, I Seismic stability of slopes in saturated clay. Ph.D. dissertation, Univ. of California, Berkeley, Calif. Biscontin, G., and Pestana, J. M Influence of peripheral velocity on vane shear strength of an artificial clay. Geotech. Test. J., 24 4, Blaney, G., and Mallow, W Synthetic clay soil for dynamic model pile tests. Proc., Dynamic Response of Pile Foundations Experiment, Analysis, and Observation, Geotech. Spec. Pub. 11, T. Nogami, ed., ASCE, Reston, Va., Bray, J The effects of tectonic movements on stresses and deformations in earth embankments. Ph.D. dissertation, Univ. of California, Berkeley, Calif. Burr, J., Pender, M., and Larkin, T Dynamic response of laterally excited pile groups. J. Geotech. Geoenviron. Eng., 123 1, 1 8. Cai, Y. Q., and Liang, X Dynamic properties of composite cemented clay. J. Zhejiang Univ., Sci., 5 3, D Onofrio, A., Silvestri, F., and Vinale, F Strain rate dependent behaviour of a natural stiff clay. Soils Found., 39 2, Dorsey, N. E Properties of ordinary water-substance, Reinhold, New York, 184. Drnevich, V. P., Hardin, B. O., and Shippy, D. J Modulus and damping of soils by resonant-column method. Special Technical Publication No. 654, Symp. on Dynamic Geotechnical Testing, ASTM, West Conshohocken, Pa., Huck, J. R., Noyel, G. A., and Jorat, L. J Complex permittivity and relaxation time of supercooled aqueous dielectrics. Proc., 5th Int. Conf. on Dielectric Materials, Measurements and Applications, Canterbury, U.K., IEEE, Ishibashi, I., and Zhang, X Unified dynamic shear moduli and damping ratios of sand and clay. Soils Found., 33 1, Iskander, M. G., Liu, J., and Sadek, S Transparent amorphous silica to model clay. J. Geotech. Geoenviron. Eng., 128 3, Jovicic, V., Coop, R., and Simic, M Objective criteria for determining Gmax from bender element tests. Geotechnique, 46 2, Kenny, M. J., and Andrawes, K. Z The bearing capacity of footings on a sand layer overlying soft clay. Geotechnique, 47 2, Kim, D. S., Stokoe, K. H., and Hudson, W. R Deformational characteristics of soils at small to intermediate strains from cyclic tests.. Research Rep. No , Center for Transportation Research, Bureau of Engineering Research, Univ. of Texas at Austin, Austin, Tex. Ko, H., Atkinson, R., Goble, G., and Ealy, C Centrifugal modeling of pile foundations. Proc., Analysis and Design of Pile Foundations, J. R. Meyer, ed., ASCE, New York, Kovacs, W. D An experimental study of the response of clay embankments to base excitation. Ph.D. dissertation, Univ. of California, Berkeley, Calif. Lazarte, C The response of earth structures to surface fault rupture. Ph.D. dissertation, Univ. of California, Berkeley, Calif. Lee, J. S., and Santamarina, J. C Bender elements: Performance and signal interpretation. J. Geotech. Geoenviron. Eng., 131 9, Leroueil, S., and Vaughan, P. R The general and congruent effects of structure in natural soils and weak rocks. Geotechnique, 40 3, Matešić, L., and Vucetic, M Strain-rate effect on soil secant shear modulus at small cyclic strains. J. Geotech. Geoenviron. Eng., 129 6, Mayfield, B The performance of a rigid wheel moving in a circular path through clay. Ph.D. dissertation, University of Nottingham, Nottingham, U.K. Meymand, P. J Shaking table scale model tests of nonlinear soil-pile-superstructure interaction in soft clay. Ph.D. dissertation, Univ. of California, Berkeley, Calif. Morris, D. V., and Delphia, J. C Resonant column testing of dynamic rock properties. Proc., 9th Conf. on Engineering Mechanics, ASCE, Reston, Va., Moss, R. E. S., Rawlings, M. A., Caliendo, J. A., and Anderson, L. R Cyclic lateral loading of model pile groups in clay soils. Proc., 3rd Conf. Geotechnical Earthquake Engineering and Soil Dynamics, ASCE, Reston, Va., Nunez, I., and Randolph, M Tension pile behavior in clay Centrifuge modelling technique. Proc., Symp. on the Application of Centrifuge Modelling to Geotechnical Design, Manchester, Balkema, Rayhani, M. H. T., and El Naggar, M. H Characterization of Glyben for seismic applications. Rep. No. GEOT-3-06, The University of Western Ontario, London, Ont., Canada. Seah, T. H Anisotropy of resedimented Boston blue clay. Ph.D. thesis, Massachusetts Institute of Technology, Cambridge, Mass. Seed, H. B., and Clough, R Earthquake resistance of sloping core dams. J. Soil Mech. and Found. Div., 89 1, Stokoe, K. H., Hwang, S. H., Lee, J. N.-K., and Andrus, R. D Effects of various parameters on the stiffness and damping of soils at small to medium strains. Proc., 1st Int. Conf. on Pre-Failure Deformation Characteristics of Geomaterials: Pre-Failure Deformation of Geomaterials, Balkema, Rotterdam, The Netherlands, Sultan, H. A., and Seed, H. B Stability of sloping core earth dams. J. Soil Mech. and Found. Div., 93 4, Sutherland, H. B Uplift resistance of soils. Geotechnique, 38 4, Tavenas, F., Roy, M., and La Rochelle, P An artificial material JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING ASCE / FEBRUARY 2009 / 289