Grout injection is used to improve loose contractive sands that are prone to liquefaction. Injected chemical grouts bind soil grains, increase the undrained shear strength, and thus reduce the liquefaction potential of soils (Gallagher and Mitchell 2002; Xanthakos 1994; Karol 2003). Typical materials include cement, gypsum, sodium silicate, acrylates, acrylamides, and polyurethanes. Various efforts have characterized and modeled the mechanical behavior of soils cemented with brittle chemical grout materials (e.g., Clough et al. 1981; Airey and Fahey 1991; Lade and Overton 1989; Fernandez and Santamarina 2001; Wang and Leung 2008). However, the injection of chemical grouts is often problematic due to water pollution and environmental regulations (DeJong et al. 2010). With societal demands for environmentally friendly construction materials and techniques, the use of gel-like biopolymers has recently garnered significant interest as an alternative to chemical grout materials for ground improvement because of their nontoxic and biodegradable characteristics (Ivanov and Chu 2008; DeJong et al. 2010; Cabalar et al. 2017; Im et al. 2017). Furthermore, biopolymers are known to improve erosion resistance, increase water retention, and promote vegetation (Kwon et al. 2017, 2019; Cho and Chang 2018; Ham et al. 2018; Tran et al. 2019; Kim et al. 2019). In contrast to brittle cemented soils, the impact of soft viscoelastic inclusions, such as gel-like biopolymers, on the mechanical response of treated soils remains poorly understood.
The sudden contractive failure of saturated sands results from positive excess pore pressure generated during undrained loading (Terzaghi et al. 1967; Castro 1969; Casagrande 1976; Yamamuro and Lade 1997; Santamarina et al. 2019). This excess pore pressure reduces the effective stress and hence the undrained shear strength of sands, and in some extreme cases may lead to flow liquefaction (Robertson and Wride 1998; Yoshimine et al. 1999). The inclusion of viscous biopolymers, such as guar gum, gellan gum, beta-glucan, and xanthan gum, has been proven to effectively increase soil strength, although the extent of improvement differs with the host soil’s baseline strength, curing time, water content, and biopolymer characteristics (e.g., Chang and Cho 2012; Chen et al. 2013; Latifi et al. 2016; Muguda et al. 2017; Soldo et al. 2020). Earlier studies used uniaxial compression (UC), unconsolidated-undrained loading (UU), vane shear, fall cone, and direct shear tests, and had limited control on the confining stress or fluid saturation. In fact, the undrained load–deformation response of biopolymer-treated sands under triaxial stress conditions is scarcely examined.
This study explores the load–deformation behavior of loose contractive sands treated with soft viscoelastic biopolymers when subjected to undrained loading. The analysis emphasizes the extent to which soft viscoelastic pore fillers alter the contractive behavior and improve the undrained shear strength. Gelatin was chosen as a model biopolymer to represent the soft viscoelastic medium. The confinement and gelation history may affect the mechanical behavior of the gelatin-treated sands in a way similar to that of cemented sands (Clough et al. 1981; Consoli et al. 2010; Khan et al. 2006; Yun and Santamarina 2005; Dai and Santamarina 2017). For example, the failure of cementing bonds governs the overall strength of a cemented sand when the cementing process precedes the confinement. On the other hand, when the confinement precedes the cementing process, both the baseline strength of the granular frame and the strength of the cementing agent contribute to the cemented sand strength.
These contrasting results call for the examination of two extreme confinement-gelation sequences: consolidation before gelation (CbG), and confinement after gelation (CaG). We use triaxial compression tests to identify the effect of gelatin treatment and formation sequence on the stress–strain responses of loose and contractive sands during undrained loading while simultaneously monitoring changes in shear wave velocity. Results provide a unique experimental data set and new insights into the feasibility of soft viscoelastic biopolymers to improve the undrained strength of loose contractive sands that are prone to liquefaction.
Materials and Methods
Gelatin was chosen as a model biopolymer to represent soft viscoelastic inclusions. It is a translucent, colorless, nearly tasteless solid substance derived from the collagen in animal skins and bones. Gelatin is an irreversibly hydrolyzed form of collagen. It consists of various amino acids, predominantly composed of glycine, proline, hydroxyproline, glutamic acid, arginine, and alanine, and its chemical formula is typically expressed as
Gelatin is a viscoelastic material and shows a wide range of physical properties, which vary with the gelatin-water mixing ratio and curing conditions, such as temperature, humidity, and curing time (Imeson 2010; Gómez-Estaca et al. 2011). Gelatin has thermoreversible characteristics due to hydrogen and van der Waals bonds. The sol-gel transition is reversed by heating and cooling, and can be repeated several times without loss of gel characteristics (Imeson 2010; Kavanagh et al. 2013).
Curing and gelation took place in this study over 24 h at an ambient temperature of 20°C. We mixed 8.7, 13.64, 19.05, and 25 g gelatin with 100 g deionized water to prepare four gelatin concentrations:
, 12%, 16%, and 20% (Table S1). The mixing temperature was 60°C to ensure the complete dissolution of gelatin. Gelation binds gelatin and water molecules (Bohidar and Jena 1993); all available water was bound to gelatin in this study, and hence there was no free water left on the cured gelatin surface.
The mechanical properties of pure gelatin samples were measured for the different gelatin concentrations and included uniaxial compressive strength and both small- and large-strain elastic moduli (Figs. S1–S3). The measured uniaxial compressive strength ranged from
to 15.17 kPa and the Young’s modulus ranged from
to 32.4 kPa, as a function of gelatin concentration (Fig. S1). Rod and shear wave velocity values ranged from
Fig. 1 presents the instrumented triaxial cell used in this study. The axial compression load was imposed at a constant displacement rate. The pressure panel controlled both the cell and fluid pressures. The instrumentation included a load cell to measure the vertical load and a linear variable differential transformer to monitor the vertical displacement. A pair of bender elements was installed on the top and bottom caps to capture shear wave signals during deviator loading; the input signal was a square wave with a 20-Hz repetition rate and an amplitude of 10 V (function generator: Keysight, Model 33210A, Santa Rosa, California). An oscilloscope (Keysight, Model DSOX3024A) stored the signals received by the top bender element after preconditioning by a filter amplifier (Krohn-Hite, Model 3384, Brockton, Massachusetts, bandpass filter from 500 to 200 kHz).
The experiments used clean coarse silica sand (KAUST 20/30, BMS, Jeddah, Saudi Arabia) with the following index properties: specific gravity
, mean grain size
, maximum void ratio
, and minimum void ratio
The reference tests used gelatin-free sand specimens. The target relative density was controlled to be less than
to promote a contractive response and greater than
to ensure dilative behavior. Dry sand was first air-pluviated into a membrane stretched over a split mold. All specimens were prepared by air pluviation and had a diameter of 50 mm and a target height of 100 mm. After applying a vacuum of
to the dry sand specimen, the split mold was removed, the pressure cell was assembled, and the confining stress
was applied. Deaired and deionized water was introduced into the specimen for water saturation with a pressure difference of 3 kPa, followed by a 200- to 300-kPa backpressure at a constant effective stress of 30 kPa. At the end of saturation, the B-value was greater than 0.94 in all tests. Thereafter, sand specimens were subjected to the effective confining stress of
by simultaneously controlling the cell and back pressures. Specimens were kept for more than 1 h until volume change ceased. The final relative density of the biopolymer-free specimens ranged from
to 61% after isotropic effective confinement.
Preparation of the biopolymer-treated sand specimens started by pouring a warm gelatin solution of 60°C into the stretched membrane. The dry sand was then wet-pluviated into this solution to ensure homogeneous mixing and full saturation of the specimens. The final relative density of these biopolymer-treated specimens ranged from
to 48% after isotropic confinement.
The loosely packed sand specimens saturated with the warm gelatin solution were subjected to a cell pressure of
while hot water at
circulated through the triaxial cell to completely melt the gelatin contained within the specimens. We also heated an external flow line to avoid gelation (Fig. 1). The confining stress was then elevated to the target initial confining effective stresses of either
under drained conditions for 30 min. The drained gelatin volume allowed calculations of specimen volume changes during consolidation. Following the consolidation phase, the gelatin-treated sand specimens were cured for 24 h at
while maintaining a constant cell pressure. Through additional batch experiments, we confirmed that the gelatin volume did not measurably change during the cooling from 75°C to 20°C.
The sand specimens saturated with warm gelatin solutions were cured for 24 h at
without confining stress. After complete gelation, the cell pressure was elevated to the initial confining stress of either
. The pore pressure valve remained open for a consolidation time of approximately 30 min; however, there was no measurable volume change in the specimen.
The pore pressure valve was closed to enforce undrained deviator loading for both CbG and CaG specimens. The vertical deformation rate was kept constant at
, equivalent to a vertical strain rate of
. Shearing continued to a vertical strain greater than 20%. Gelatin plugged pores and flow lines once the gelatin hardened; therefore, pore fluid pressure measurements were not possible during the undrained deviator loading of the biopolymer-treated sands. Shear wave signatures were acquired every minute during deviator loading (see Figs. S4–S6 for shear wave signatures).
The undrained shear strength
is the deviator stress
at large strains herein determined at a vertical strain of
[ASTM D4767 (ASTM 2011); Thevanayagam 1998]. Fig. 7 shows the measured undrained shear strength plotted versus gelatin concentration for the various initial confining stresses and loading histories. The undrained shear strength
increases with gelatin concentration in both the CbG and CaG cases. These results suggest that biopolymers with higher stiffness and strength provide greater support to the granular frame and prevent the buckling of chains [Figs. 7(a) and S1]. Previous studies using soils treated with beta-glucan and xanthan biopolymers show similar trends [unconfined compression: Chang and Cho (2012), Chen et al. (2013), Latifi et al. (2016), and Soldo et al. (2020); vane shear: Cho and Chang (2018)].
Fig. 7(b) depicts the undrained shear strength normalized by the initial confining stress. Results demonstrate that the impact of biopolymers on undrained shear strength is more pronounced at lower confining stress levels and for higher biopolymer concentrations. These observations are analogous to cementation treatments (Dupas and Pecker 1979; Acar and El-Tahir 1986; Dass et al. 1994; Fernandez and Santamarina 2001).
Loose coarse-grained soils have a low coordination number and grains form granular columns that are prone to buckling due to the limited lateral support during loading, as depicted in Fig. 8(a) (Santamarina et al. 2001; Hasan et al. 2008; Kim et al. 2013; Kuei et al. 2020). Buckling collapse in loosely packed sands generates excess pore pressure under undrained conditions (Vaid and Chern 1985; Ishihara 1993).
Biopolymers such as gelatin fill the pore space, surround the particle chains, and contribute the viscoelastic resistance that prevents buckling. The spring-dashpot system in Fig. 8(b) is analogous to the pore-filling biopolymers (the elastic moduli and damping coefficients obtained for gelatin under longitudinal and shear vibrations are shown in Figs. S2 and S3). We anticipate that the grain support provided by the biopolymers differs with the loading rate due to their viscoelastic nature. Furthermore, the elastic modulus and viscous damping coefficient of gelatin are concentration dependent (Fig. S3); therefore, the increased gelatin concentration leads to greater lateral support to the granular frame and explains the increased undrained shear strength of the specimens that were consolidated before gelation [Fig. 7(a)].
Effective stress and volumetric paths plotted in the
space allow us to identify failure states and infer either contractive or dilative soil behavior. The hardened gelatin plugs pores and prevents pore pressure measurements. Yet, we can gain insight into the evolution of the mean effective stress
from the measured shear wave velocity
= shear wave velocity at 1 kPa; and
= shear wave velocity sensitivity to changes in effective stress. Experimental results show
for the sand used in this study (Fig. S7).
Figs. 9 and 10 show the effective stress paths inferred from changes in the shear wave velocity for CbG and CaG specimens. These figures also include the paths measured for the biopolymer-free specimens during undrained loading. The baseline cases show the consequences of the excess pore pressure generation during undrained shearing in both contractive specimens (positive excess pressure and postpeak softening behavior) as well as dilative specimens (negative excess pore pressure). The biopolymer-treated loose specimens show increases in effective stress
due to the presence of biopolymers together with a strain-hardening behavior that is compatible with a dilative tendency (Fig. 9). Terminal states appear to be affected by gelatin concentration.
The estimated initial mean effective stress
is significantly lower than the applied confining stress
in CaG specimens (Fig. 10). Still, the presence of gelatin hinders bucking and particle rearrangement, which results in the high peak deviator stress
of treated sands. Furthermore, postpeak softening decreases with higher gelatin concentrations, suggesting that gelatin tears at large strains.
CbG cases simulate the condition where a contractive soil layer prone to liquefaction is treated by injecting a viscoelastic biopolymer grout, whereas CaG resembles cases where the viscoelastic grout treatment is applied at every lift during backfilling.
Results in Figs. 9 and 10 show that the confinement-gelation sequence has a pronounced effect on the undrained shear strength. The marked improvement in undrained shear strength in CbG specimens compared to the CaG specimens has important implications on field implementation of biopolymer grout injections. In particular, this finding highlights potential benefits of treating the soil postconstruction rather than before or during construction.
Still, the field implementation of biopolymer treatments in geotechnical practice faces a number of uncertainties related to the durability and life span associated with thermal and moisture cycles, indigenous microbial activity, pore water chemistry, and groundwater flow–driven dissolution.