Forensic Analyses and Rehabilitation of a Failed Highway Embankment Slope in Texas

Abstract

A comprehensive investigation was designed and conducted to identify the potential causes of failure of a highway embankment slope in Texas and evaluate the effectiveness of lime treatment to rehabilitate the failed slope. Highway slopes built with high plasticity clays often experience shallow slope failures after exposure to repeated wet–dry weathering cycles. Lime stabilization generally reduces the swell–shrink potential, enhances the engineering properties of problematic clayey soils, and can potentially prevent surficial slope failures. However, exposure to wet–dry cycles can negate some of the benefits of lime treatment and therefore a study was conducted to address the use of this lime treatment to stabilize embankment slopes. Extensive laboratory tests were conducted to study the effect of weathering cycles on the degradation of hydro-mechanical properties of untreated and lime-treated soils. Rainfall-induced slope stability analyses were performed to investigate the probable causes of slope failure and evaluate the stability of lime-treated surficial slope. The optimum stabilizer dosage and treated layer thickness required for the slope rehabilitation were determined based on laboratory tests and numerical analysis results. The stability analysis results indicate that the degradation of surficial soil’s hydro-mechanical properties and the development of a perched water table during prolonged rainfall possibly caused the slope failure. The post-treatment increase in shear strength properties, reduction in moisture fluctuations recorded by embedded moisture sensors, and the presence of newly installed underlying drains are expected to prevent recurrence of surficial slope failures. Salient results from this study are covered in this paper.

High plasticity clayey soils are prevalent in many parts of the world and are often used to build highway embankments. These soil slopes are mostly present in unsaturated conditions and undergo significant volume changes because of seasonal fluctuations and wet–dry cycles ( 1 ). In dry periods, the decrease in moisture content increases the surface tension between soil particles, and consequently the soil tends to shrink. Tensile stresses are induced when the shrinkage potential is restrained by internal or external boundary conditions, including stress concentration or moisture gradient ( 2 , 3 ). The surficial soil experiences desiccation cracking when the tensile stress exceeds the tensile strength ( 4 ). Even though the following wet period tends to seal the cracks, some of the cracks might stay open, resulting in permanent deterioration in the soil’s hydro-mechanical properties ( 5 ).

The consequence of desiccation cracks on the hydraulic properties of soil has been investigated by many researchers ( 6 – 8 ). Li et al. ( 9 ) reported that the hydraulic properties of clayey soil are primarily controlled by the crack network geometry in the soil mass. Daniel ( 10 ) found that the difference in permeability values between field conditions and laboratory-prepared specimens can be up to four orders of magnitude because of the formation of desiccation cracks. Albrecht and Benson ( 11 ) observed a 100 times increase in permeability value after the first drying cycle and up to 500 times increase within three cycles. The desiccation cracks also reduce the shear strength properties of the soil to fully softened strength (FSS), which can potentially affect the stability of the slope. The effects of cracks on degradation of hydro-mechanical properties are prominent in the surficial layers of the embankment and are negligible beyond a depth of about 7 ft (2.1 m) from the surface ( 12 , 13 ). These slope stability issues could be mitigated by suppressing the swell–shrink potential and reducing desiccation cracking of plastic clays using chemical stabilization techniques.

Lime stabilization is one of the most preferred chemical stabilization techniques for treating soils with a high plasticity index (PI). Lime treatment reduces the swell–shrink behavior of the soil and enhances the shear strength properties ( 14 , 15 ). Adding lime to clayey soil generally increases the plastic limit (PL), reduces the liquid limit (LL), and consequently decreases its PI and volume change characteristics ( 16 ). For these advantages, many slope failures have been repaired by treating the soil with 2% to 8% lime ( 17 , 18 ). The required lime dosage is often determined based on recommendations of design manuals developed by different regulatory agencies. Soil-lime pH tests, reduction in plasticity characteristics, decrease in swell–shrink potentials, and improvements in strength properties are generally used as measures for assessing the extent of improvement in engineering properties in the mix-design phase ( 19 , 20 ). The laboratory mix-design process often portrays short-term improvements in soil properties and does not precisely account for the long-term detrimental influence of climatic factors that the treated soil experiences in the field ( 21 ).

Even though lime stabilization has generally performed well in most projects, recurring failures have been observed in some rehabilitation projects. At the Alexandria-Ashland Highway in Kentucky, the top 6 in. of subgrade was stabilized using 6% lime considering the increase in bearing capacity of the soil. Though the post-treatment unconfined compressive strength (UCS) values increased about fivefold in six years, longitudinal and transverse cracks were still observed on the highway pavement, probably because of insufficient depth of treated layer and high volumetric changes of the underlying untreated soil ( 22 ). Abrams and Wright ( 17 ) reported three failed lime stabilized slopes in Texas, where the slopes failed within three years of construction. Cracks were observed on the treated surface of slopes before the failures. The failures were attributed to the possible restraint of internal drainage because of lime stabilization, which caused excess hydrostatic water pressures in the soil and thereby reduced the soil strength and stability.

Although the exact causes of these post-treatment slope failures are still debatable, the probable reasons for the failure include lack of comprehensive laboratory testing in the design phase, inadequate depth of treatment and dosages, underestimation of wet–dry cycle effects on the properties of treated soil, and inadequate drainage systems. Furthermore, most of these studies relied only on basic laboratory testing, and extensive numerical analyses were not conducted to predict the long-term performance of the treated slopes. The main objective of this study is to address these issues by conducting comprehensive experimental and numerical studies for rehabilitation of a failed embankment slope in Texas. The strength properties and durability of the untreated and treated soils were assessed and compared by conducting a series of laboratory tests. The possible causes of slope failure and expected performance after lime treatment were evaluated by performing rainfall-induced slope stability analyses for various rainfall events. The moisture variations recorded by sensors embedded in the slopes at the time of rehabilitation construction works were analyzed and used to gauge the effectiveness of lime treatment. The following sections present the site details, experimental program, numerical analyses, and field construction procedure adopted to rehabilitate the failed slope.

Site Details

The failed slope section is located in the Paris district, along the southbound U.S. 75 Frontage Road between Randell Lake Road and State Highway 91 in Denison, Texas (Figure 1a). The embankment slope is 11.6 m (38 ft) in height, with a slope ratio of three horizontals to one vertical (3H:1V). Desiccation cracks were first noticed in 2014 close to the toe of the slope, and the slope experienced surficial failures in December 2015. In 2016, the slope failures started damaging the pavement structure, and consequently, the road was closed to service by the Texas Department of Transportation (TxDOT).

Figure 1.

(a) Location of the failed slope in Denison, Texas (source: TxDOT) and embankment slope condition in (b) December 2017; and (c) June 2019.

During the first preliminary site investigation in December 2017, severe cracks and surficial slope failures were observed. The slope failure affected approximately 18.3 m (60 ft) of the pavement along the southbound lane, and the depth of failure was about 1.8 m (6 ft) (Figure 1b). The extent of the failure zone progressed with time, and the length of the damaged slope segment increased to 134.7 m (442 ft) in June 2019, damaging both lanes of the pavement (Figure 1c). Soil specimens were collected from the failure scarp, and the hydro-mechanical properties of the soil were studied in the laboratory to investigate the causes of failure. The collected soil was also used for the lime stabilization mix design, and the optimum dosage required for the rehabilitation work was determined based on the results of extensive laboratory testing and rainfall-induced slope stability analyses.

Experimental Program

Basic Soil Characterization Tests

The basic physical, chemical, and engineering properties of the soil were evaluated as per American Society for Testing and Materials (ASTM) and TxDOT standard procedures, and these test results are summarized in Table 1. The soil exhibited swelling and shrinkage characteristics, and this high swell–shrink potential is consistent with the high PI and clay content of the soil. Lime treatment was selected as the most suitable chemical stabilization option because of the high plasticity properties, swell–shrink potential, and low soluble sulfate content of the soil.

Table 1.

Basic Soil Characterization Test Results

Property	Value	Standard
Liquid limit (LL)	58%	ASTM D4318
Plastic limit (PL)	21%	ASTM D4318
Plasticity index (PI)	37%	ASTM D4318
Clay fraction	55%	ASTM D7928
Activity value	0.67	na
USCS classification	CH	ASTM D2487
Optimum moisture content (OMC)	19.5%	ASTM D698
Maximum dry unit weight (MDUW)	16.8 kN/m³ (107 pcf)	ASTM D698
One-dimensional (1D) free swell strain	7.2%	ASTM D4546
1D linear shrinkage strain	17.7%	TEX 107-E
Soluble sulfate content	480 ppm	TEX 145-E

Note: USCS = Unified Soil Classification System; ASTM = American Society for Testing and Materials; pcf = pounds per cubic foot; na = not applicable.

The minimum lime dosage was determined as per the TEX 121-E pH test. The target was a minimum pH of 12.4 to sustain long-term pozzolanic reactions after lime treatment. The pH test results indicated that 5% lime dosage (by weight of dry soil) satisfies the target pH of 12.4. As the pH test does not provide any insight into the permanency of the stabilized material, a higher percentage of the lime additive is often required to ensure durability. Therefore, shear strength and durability tests were performed on the untreated control soil and the 5% and 8% lime-treated soil specimens. The optimum lime dosage was then selected based on these test results.

Estimation of Unsaturated Soil Properties

The properties of unsaturated soil dictate the moisture variation during wet–dry cycles. In unsaturated conditions, soil suction contributes to soil strength and enhances the stability of an embankment slope, whereas the matric suction and associated shear strength of the surficial soil decreases after a rainfall event. A slope stability analysis, incorporating the effect of a rainfall event, requires comprehensive characterization of both saturated and unsaturated soil hydraulic properties. The soil permeability function, which depicts the variation in hydraulic conductivity properties with soil suction, was estimated using the soil water retention curve (SWRC) and the saturated permeability value determined from laboratory tests.

The SWRC tests were conducted using untreated and treated soil specimens of 25.4 mm (1 in.) height and 63.5 mm (2.5 in.) diameter. The untreated specimens were prepared by statically compacting the soil at optimum moisture content (OMC) of 19% and dry unit weight of 16.0 kN/m³ (101.7 pounds per cubic foot [pcf]) (95% of maximum dry unit weight [MDUW]). The 5% and 8% lime-treated specimens were statically compacted at water contents and dry unit weights of 21.5% and 15.4 kN/m³ (98.2 pcf), and 23.5% and 13.5 kN/m³ (85.8 pcf), respectively. These compaction conditions for the treated soil specimens correspond to water contents 2% wet of the respective OMC (OMC+2%) and 95% of the respective MDUW. The compacted treated soil specimens were cured for 7 days in sealed zip-lock bags (≈ 100% relative humidity) at room temperature.

The SWRC was determined using a Tempe cell for measuring suction values ranging between 0 and 500 kPa (0–10.4 kilo pounds per square foot [ksf]) and dew point potentiometer device for measuring the suction values higher than 500 kPa (10.4 ksf). The SWRC best fitting curve was developed using the Fredlund and Xing function ( 23 ), and model parameters were determined. The saturated permeability tests were conducted on triplicate specimens for untreated and 7-day cured treated specimens using a triaxial test setup following the procedure outlined in Bhaskar et al. ( 24 ). After determining the SWRC and saturated permeability values, the unsaturated hydraulic conductivity functions were determined as per Krahn ( 25 ). These unsaturated soil properties were later used in the transient seepage analyses.

Estimation of Shear Strength Parameters

The peak and FSS strength parameters were determined for both untreated and treated soils to study the changes in strength properties before and after desiccation cracking. The peak shear strength parameters, which represent the newly-compacted condition of the soil, were measured by conducting drained direct shear (DS) tests as per ASTM D3080. DS tests were conducted on treated soil specimens prepared and cured similar to the SWRC test specimens. The FSS parameters, which characterize the long-term condition of soil after exposure to wet–dry cycles, were determined using a torsional ring shear (TRS) device, according to ASTM D7608-18. The tests were conducted at slow shearing rates of 4.4 × 10⁻³ mm/min (1.7 × 10⁻⁴ in./min) and 0.018 mm/min (7.1 × 10⁻⁴ in./min) for DS and TRS tests, respectively. The obtained shear strength parameters were utilized for the slope stability analysis studies, which are detailed in later sections.

Durability and Strength Retention Studies

Wet–dry cycles affect the engineering properties of clayey soils exposed to semi-arid environmental conditions. The post-treatment permanency, after exposure to repeated wet–dry cycles, was investigated using durability studies. These test results provided insights into the impact of climatic variation on the long-term performance of lime-treated soil. Durability tests for untreated, and 5% and 8% lime-treated, specimens were conducted on triplicate specimens, as per ASTM D559 method. Specimens of 101.6 mm (4 in.) diameter and 116.4 mm (4.58 in.) height were compacted statically in three layers at the same moisture content and dry unit weight as those used for preparing the SWRC and DS test specimens. The untreated and 7 day cured treated soil specimens were submerged in water for 5 h and then oven-dried at 70°C for 42 h. After completing each wetting and drying cycle, the volumetric strain changes in swell and shrinkage conditions were measured, and the durability test was continued until completion of 14 cycles (or failure of the specimens). The strength retention characteristic of specimens was studied by conducting UCS tests after completion of 0, 3, 7, and 14 cycles. The UCS tests were performed as per ASTM D2166 at a strain rate of 1 mm/min (3.9 × 10⁻² in./min). The durability and strength retention test results were analyzed to select the optimum lime dosage required to stabilize the soil. The next section elucidates the procedure adopted to utilize the abovementioned laboratory test results and perform the rainfall-induced slope stability analyses.

Numerical Analyses Studies

The potential causes of failure of the U.S. 75 frontage road embankment slope and the expected long-term performance of the lime-treated rehabilitated slope were evaluated by conducting rainfall-induced slope stability analyses. The slope model was developed based on the geometric configuration details obtained by TxDOT, and the failed slope characteristics observed during site visits (Figure 2). The numerical simulations were performed in three stages: assignment of initial pore water pressure (PWP) profile to the embankment slope model, transient seepage analyses, and slope stability analyses.

Figure 2.

Numerical analysis model of the slope section.

The initial PWP profile, which depicts the distribution of pore water pressure in the embankment slope before a rainfall event, was assigned to the numerical model based on the location of the groundwater table. The boring log showed the groundwater table at a depth of 15.7 m (51.5 ft) from the pavement surface. Therefore, the groundwater table was defined at the bottom of the numeric analysis model at an elevation of 180.0 m (590.6 ft). Typically, a linear increase in negative PWP is assumed to approximate the variation of matric suction with height above the groundwater table. However, this led to significant matric suction levels on the surficial unsaturated soil layers, similar to that reported by Lee et al. ( 26 ). Therefore, the matric suction was assumed to increase linearly from 0 kPa at the groundwater table to 20 kPa at an elevation of 182.0 m (597.1 ft). The matric suction was kept constant at 20 kPa (418 pounds per square foot [psf]) above the elevation of 182.0 m (597.1 ft). This maximum suction level was determined from the SWRC based on the average water content of the soil specimens collected from the site.

Transient seepage analyses were conducted using a commercially available finite element software program. The impact of rainfall infiltration on the slope soil moisture variation was studied considering different rainfall intensities and durations obtained from the National Oceanic and Atmospheric Administration. The compiled rainfall data set was generated based on a 10-year mean recurrence interval. For the nearest rainfall station, a total amount of 15.8 cm (6.2 in.), 20.8 cm (8.2 in.), and 25.7 cm (10.1 in.) rainfall were predicted for one, four, and 10 days of continuous rainfall, respectively. However, these predictions did not provide an hourly variation of precipitation. Therefore, simulations were performed with constant rainfall intensities of 1.8 × 10⁻⁴ cm/s (7.1 × 10⁻⁵ in./s), 6.0 × 10⁻⁵ cm/s (2.4 × 10⁻⁵ in./s) and 3.0 × 10⁻⁵ cm/s (1.2 × 10⁻⁵ in./s), for durations of one, four, and 10 days, respectively. After conducting the transient seepage analysis, the obtained PWP distributions were transferred to the slope stability analysis software program, and the factor of safety (FOS) of the critical slip surface was computed using the Morgenstern-Price method of limit equilibrium analysis. A half-sine interslice force function was used, and the entry and exit points of probable slip surfaces were specified for the slope stability analyses.

The rainfall-induced slope stability analyses were conducted for two different scenarios—initial and long-term conditions; these conditions represent the state of the surficial soil before and after desiccation cracking, respectively. The stability analyses representing the initial condition were conducted by assigning peak shear strength parameters and hydraulic conductivity value obtained for newly-compacted soil specimens. The long-term stability of the slope was assessed using degraded soil strength properties and increased hydraulic conductivity because of desiccation cracking.

During the site visit, the depth of failure was measured as 2.1 m (7 ft) from the original slope surface. Therefore, the numerical model of the slope section was divided into two layers: surface layer and deep layer. The top 2.1 m (7 ft) surface layer was modeled and assigned the hydro-mechanical properties of weathered soil. The FSS shear strength parameters were used, and the hydraulic conductivity obtained for a newly-compacted specimen was increased by four orders of magnitude to simulate the effect of desiccation cracks, as reported by Daniel ( 10 ). Since the properties of deeper layers are not affected by desiccation cracking, the peak soil strength properties and the permeability value for a newly-compacted specimen were assigned to the deep layer. Based on the unsaturated soil properties, the contribution of soil matric suction toward shear strength was incorporated by using φ^b of 15°, in accordance with Zhang and Fredlund ( 27 ).

The numerical simulation process mentioned above was repeated with untreated and lime-treated soil properties. The numerical analysis results with untreated soil properties facilitated understanding of the possible reasons for the slope failure. Three stabilized layer thicknesses of 30.5 cm, 61.0 cm, 91.4 cm (1 ft, 2 ft, and 3 ft) were considered, and the anticipated long-term FOS values were compared to select the design thickness of the lime stabilized layer. The need for the drainage system underneath the lime-treated layer was then investigated by conducting similar numerical analyses, with and without the provision of drainage. The following section presents the results obtained from the experimental program and numerical analyses.

Analysis and Discussion of Results

Unsaturated Soil Properties

The SWRC presented in Figure 3 illustrates the impact of lime treatment on the hydraulic properties of the unsaturated soils. The Fredlund and Xing ( 23 ) fitting parameter a, which is related to air entry value (AEV), increased from 940 for untreated soil to 1,640 and 1,780 for 5% and 8% lime-treated soils, respectively. The higher a value in 7 day cured treated soils could be attributed to the formation of calcium silicate hydrate (C-S-H) gels. C-S-H gels have porous structures, high specific surface area, and negative surface charge, and these characteristics possibly increased the AEV after treatment. The fitting parameter n, which is related to the pore size distribution of soil, increased moderately from 0.73 for untreated soil to 0.90 and 1.02 for 5% and 8% lime-treated soil specimens, respectively. The post-treatment flocculation-agglomeration and soil modification resulted in a more homogeneous pore size distribution in treated soil, which might be responsible for the observed increase in n value. The m parameter, which is related to the adsorption characteristics of the soil, decreased marginally from 1.71 for untreated soil to 1.65 and 1.62, for 5% and 8% lime-treated soils, respectively. This suggests that the residual water-holding characteristic of the soil was not affected appreciably after lime treatment.

Figure 3.

Soil water retention curve of untreated and lime-treated soils.

The hydraulic conductivity functions estimated using the SWRC and saturated permeability values are represented in Figure 4. The saturated untreated soil had a low permeability value of 3.0 × 10⁻¹⁰ m/s (9.9 × 10⁻¹⁰ ft/s), which is characteristic of clayey soils. The saturated permeability values of 5% and 8% lime-treated soil specimens increased to 3.9 × 10⁻⁹ m/s (1.3 × 10⁻⁸ ft/s) and 1.5 × 10⁻⁸ m/s (5.0 × 10⁻⁸ ft/s), respectively. Typically, the formation of pozzolanic reaction products partially blocks the interconnected pores and reduces the permeability of the treated soil matrix. The observed increase in saturated permeability after lime treatment is possibly because of the combined effect of flocculation and agglomeration that made the soil friable, and the limited precipitation of cementitious phases after a short curing period of 7 days.

Figure 4.

Hydraulic conductivity functions of untreated and lime-treated soils.

Shear Strength Parameters

The shear strength parameters obtained from DS and TRS tests are presented in Table 2. The test results revealed that the untreated soil underwent significant strength loss after desiccation cracking because of softening behavior. Although the soil friction angle did not change considerably, the cohesive properties decreased significantly from 13.2 kPa (275.7 psf) to 1.3 kPa (26.8 psf). This reduction in soil cohesion could be detrimental to the stability of surficial slopes since the low effective overburden pressure on soil layers near the surface is inadequate to derive strength from the angle of internal friction and maintain stability, especially during rainfall events.

Table 2.

Shear Strength Parameters of Untreated and Lime-Treated Soils

Soil	Peak (direct shear)		FSS (TRS)
Soil	c′ (kPa/psf)	φ′ (deg)	c′ (kPa/psf)	φ′ (deg)
Untreated	13.2/275.7	23.7	1.3/26.8	23.4
5% lime-treated	43.5/908.5	41.2	5.8/120.3	31.0
8% lime-treated	37.0/772.8	39.1	6.6/137.6	30.6

Note: FSS = fully softened shear strength; TRS = torsional ring shear; psf = pounds per square foot.

Lime treatment caused significant improvement in the peak shear strength properties. The initial soil modification, flocculation-agglomeration, and some pozzolanic reaction during the 7 days of curing increased both cohesion and friction angle of the treated soils. The extent of improvement in 8% lime-treated specimens was less than 5% lime-treated soil, possibly because of higher OMC and lower MDUW used for specimen preparation. The excess unreacted lime in 7 days cured 8% lime-treated specimens might have been partially responsible for lower strength properties as lime itself does not have appreciable cohesion and friction angle. Nevertheless, the beneficial influence of lime treatment is apparent from the shear strength parameters presented in Table 2.

The FSS test results of lime-treated soils suggest that exposure to wet–dry cycles negates some of the benefits of treatment. However, the treated soil specimens had considerably high FSS strength parameters, compared with untreated specimens, because of soil modification reactions and residual inter-particle cementitious bonds. This cohesive property could help prevent surficial slope failure, even after exposure of the lime-treated soil layer to weathering cycles.

Durability and Strength Retention Studies

The permanency of the stabilized soil was studied using wet–dry durability and strength retention tests. The untreated soil specimens exhibited high volumetric swelling of 13% and disintegrated during the first wetting cycle. Unlike the untreated soil specimens, the 5% and 8% lime-treated specimens were able to withstand 7 and 14 wet–dry durability cycles, respectively. Lime treatment was also beneficial in reducing the volumetric changes during the wetting and drying cycles, and these results are summarized in Figure 5. The lime-treated soil specimens did not experience considerable volumetric changes until the end of the third weathering cycle. However, after the fourth cycle, 5% lime-treated soil specimens started exhibiting a gradual increase in swell–shrink potential. The volumetric swell–shrink strains reached approximately 5% at the seventh cycle, indicating deterioration of the stabilized material. The 5% lime-treated specimens failed during the eighth cycle because of the breakage of cementitious bonds, volumetric changes, and overall disintegration of the treated soil. The 8% lime-treated specimens did not exhibit a considerable increase in volumetric strain until the 11th cycle. In the 14th cycle, these specimens experienced volumetric strain of approximately 2.5%. Overall, 8% lime treatment outperformed 5% lime treatment throughout the wet–dry durability test and successfully endured all 14 wet–dry cycles (Figure 6).

Figure 5.

Volumetric changes during wetting and drying cycles.

Figure 6.

(a) Typical lime-treated specimen before durability test; (b) 5% lime-treated specimen after seven durability cycles; and (c) 8% lime-treated specimen after 14 durability cycles.

The strength retention characteristics of soil specimens exposed to wet–dry durability tests were also studied, and these results are presented in Figure 7. In newly-compacted condition, treated soil UCS value increased to more than double that of untreated soil specimens, from 319.2 kPa (46.3 pounds per square inch [psi]) to 797.0 kPa (115.6 psi) and 700.0 kPa (101.5 psi) for 5% and 8% lime treatment, respectively. During the first three cycles, the UCS values of both 5% and 8% lime-treated specimens increased significantly. The high-temperature conditions during drying cycles increased lime–soil reactions and resulted in this strength improvement. After the third cycle, both 5% and 8% of lime-treated specimens started to lose the retained strength properties because of the detrimental effect of wet–dry cycles. In the case of 5% lime-treated soil, UCS value decreased to 476.5 kPa (69.1 psi) at the seventh cycle and the specimen failed at the eighth cycle. The 8% lime-treated soil did not experience a significant strength loss at the seventh cycle and had a retained UCS of 1055 kPa (153 psi). Even after the 14th weathering cycle, 8% lime-treated specimens showed considerably higher retained strength properties as compared with the initial strength of untreated specimens.

Figure 7.

Strength retention of soil specimens after select durability test cycles.

The test results highlight the importance of incorporating durability and strength retention studies in the laboratory mix design phase. Even though 5% treatment satisfied the initial soil-lime pH value of 12.4 and exhibited improvement in strength properties in the DS and TRS tests, it did not ensure durability against inclement weather conditions and experienced failure at the eighth wet–dry cycle. A higher lime dosage of 8% was required for the soil to endure all 14 wet–dry durability cycles. The extra 3% lime dosage facilitated sustaining the pozzolanic reaction during the weathering cycles and imparted additional strength properties required to resist the deleterious wet–dry cycles. Based on the shear strength and durability study results, the optimum lime treatment dosage of 8% was selected for numerical analyses and field construction works.

Numerical Analyses Results

The rainfall-induced slope stability analyses were performed with the shear strength properties of untreated soil to identify the potential causes of failure. The stability analysis was first conducted with hydro-mechanical properties of the untreated soil before exposure to the wet–dry cycles. For all the different rainfall scenarios considered in this study, the FOS of the critical slip surface was greater than the minimum FOS of 1.5 recommended by different regulatory agencies. However, the FOS of the critical slip surface dropped to 0.92 when the hydro-mechanical properties of the desiccated untreated soil were considered for the surficial layer and the slope was exposed to a total of 20.8 cm (8.2 in.) rainfall in 4 days.

Figure 8a presents the PWP distribution after the rainfall event along with the FOS of the critical slip surface. A localized perched water table was formed in the desiccated surficial layer because of the increased hydraulic conductivity properties that allowed rainfall infiltration. Consequently, the contribution of suction-induced strength decreased in the surficial layer. The deeper layers, which did not experience desiccation cracking, had low hydraulic conductivity values, and were not affected by the infiltrating rainwater. The decrease in shear strength parameters from peak to FSS, development of perched water table, increase in PWP, and associated reduction in suction-induced shear strength properties resulted in the surficial slope failure. Wet patches on the slope surface (Figure 8b) and a perched water table (Figure 8c) were also observed during the field visits. These observations corroborate the findings of the numerical analyses and confirm that the desiccation-cracking-induced degradation in hydro-mechanical properties and the formation of perched water table after rainfall were the primary causes of the slope failure.

Figure 8.

(a) Pore water pressure variation and factor of safety (FOS) of the untreated slope after exposure to 20.8 cm (8.2 in.) rainfall; (b) wet patches on slope surface; and (c) perched water table observed during a site visit.

The FOS values of the critical slip surface obtained considering hydro-mechanical properties of weathered untreated and 8% lime-treated surficial soil layers are presented in Table 3. The variation in FOS values with different stabilized layer thicknesses, for various rainfall events, with and without the presence of underlying drainage systems, are also presented in Table 3. While the untreated slope section experienced failure after 20.8 cm (8.2 in.) rainfall, the expected FOS values after lime treatment of the surficial slope were greater than one, regardless of the rainfall event and depth of treatment. The increase in post-treatment shear strength parameters, particularly effective cohesion values, enhanced the stability of the slope (Figure 9). Furthermore, the presence of the underlying drainage systems increased the FOS values for all the stabilized layer thicknesses. The underlying drains prevented rainwater accumulation and enhanced the shear strength and stability of the treated surficial layer.

Table 3.

Factor of Safety Values of Untreated and Lime-Treated Slope Section

Drainage system	Treatment depth	FOS after different rainfall events
Drainage system	Treatment depth	15.8 cm (6.2 in.)	20.8 cm (8.2 in.)	25.7 cm (10.1 in.)
	Untreated (na)	1.76	0.92	0.67
No	30.5 cm (1 ft)	1.99	1.81	1.36
	61.0 cm (2 ft)	2.03	1.89	1.61
	91.4 cm (3 ft)	2.14	2.02	1.64
Yes	30.5 cm (1 ft)	1.99	1.87	1.45
	61.0 cm (2 ft)	2.08	1.95	1.67
	91.4 cm (3 ft)	2.24	2.09	1.76

Note: FOS = factor of safety; na = Not applicable.

Figure 9.

Pore water pressure variation and factor of safety (FOS) of the 8% lime-treated slope after exposure to 20.8 cm (8.2 in) rainfall.

Based on the results presented in Table 3, 8% lime treatment of the top 61 cm (2 ft) of the slope surface is required to attain a minimum required FOS of 1.5, considering the detrimental effects of wet–dry cycles on the engineering properties of lime-treated soil. The drainage system below the treated layer was also beneficial in enhancing the stability of the slope. These results were used to develop the construction plan for rehabilitation of the failed slope, and a schematic representation of the rehabilitated slope is presented in Figure 10.

Figure 10.

Schematic representation of the rehabilitated slope (figure is not to scale).

Rehabilitation of Failed Slope

The rehabilitation construction work commenced in July 2019 and was completed in May 2020. The vegetation remaining on the surface of the slope was removed, and the surficial soil was excavated. The excavated soils were transferred and stockpiled at a nearby area of flat ground. After removing the surficial soil along the embankment slope, the lime treatment started from the toe of the slope. The existing slope was benched using a motor grader to ensure proper interlocking between the treated layer and underlying untreated soil and avoid any potential sliding failure. The previously excavated soil was transferred from the stockpile area in dump trucks and placed along the slope in lifts 0.6 m (2 ft) thick and 3.7 m (12 ft) wide (Figure 11a). A distributor truck was then used to spread the hydrated lime slurry to attain the target lime dosage of 8% (Figure 11b). The soil and lime were thoroughly mixed with a rotary mixer (Figure 11c) and then compacted with a sheep foot roller (Figure 11d). Perforated PVC drainage pipes of 152.4 mm (6 in.) diameter were enclosed in geotextile to avoid clogging and placed along the length of the embankment slope at a spacing of 5.5 m (18 ft) (Figure 11e). The process was repeated to repair the entire slope and a grader was used to finish the slope surface.

Figure 11.

Rehabilitation construction work: (a) soil placement; (b) lime application; (c) soil-lime mixing; (d) compaction; (e) installation of drainage pipes; and (f) placement of a moisture sensor.

During the construction, a total of three moisture sensors were embedded in the middle of the slope (Figure 11f). The sensors were placed at depths of 46 cm (1.5 ft), 76.2 cm (2.5 ft), and 167.6 cm (5.5 ft), measured from the finished surface of the rehabilitated slope. The moisture sensors at depths of 46 cm, 76.2 cm, and 167.6 cm provided information about moisture variations in the middle of the treated layer, at the bottom of the treated layer and in the untreated soil, respectively. All the moisture sensors are capable of automatically recording the volumetric moisture content data every 15 min. These data were recorded and stored in a data logger since December 2019 and were retrieved once every month. The diurnal precipitation data from the closest rainfall station were also collected from December 2019 to October 2020. The obtained results are illustrated in Figure 12.

Figure 12.

Rainfall data and moisture content recorded by embedded sensors.

Both the moisture sensors placed in the treated soil layer recorded similar trends of moisture fluctuations over the seven-month recording period. The moisture sensor placed at a depth of 76.2 cm (2.5 ft) from the surface recorded maximum moisture variation of 10.5% (between 34.3% and 44.8%), which is considered a moderate fluctuation. In contrast, only 7.7% change in the volumetric water content was recorded by the moisture sensor placed in the untreated soil over 10 months. These results suggest that the moisture fluctuations were primarily limited to the surficial layer because of the absence of desiccation cracks after lime treatment and underlying untreated layer did not experience high saturation rate. The 61 cm (2 ft) lime-treated surficial layer has been successful till now in preventing the underlying untreated soil from experiencing drastic moisture variations, associated volume changes, and detrimental degradation of hydro-mechanical properties.

Summary and Conclusions

The potential causes of failure of a highway embankment slope in Texas were investigated in this study. The expected long-term performance of the rehabilitated slope after lime treatment was evaluated, accounting for the detrimental changes in the treated soil properties because of wet–dry cycles. Comprehensive laboratory tests were performed on untreated and treated soils, and the optimum lime dosage was selected based on the improvements in shear strength and durability properties. Extensive numerical analyses were performed, and the thickness of the lime-treated surficial layer required for the slope rehabilitation was determined. The failed slope was repaired, and the volumetric water content data recorded by moisture sensors embedded during the construction were analyzed to study the rehabilitated slope’s performance. The salient findings of this study are summarized as follows:

The untreated soil lost its cohesive properties after experiencing desiccation cracking and exposure to wet–dry cycles. Lime treatment enhanced both peak and FSS parameters because of soil modification and formation of cementitious pozzolanic reaction products. Even after experiencing wet–dry cycles, lime-treated soil specimens exhibited considerably high cohesion values as compared with untreated soil.

Soil-lime pH test results suggested that a lime dosage of 5% by weight of dry soil is sufficient to sustain the pozzolanic reaction. Wet–dry durability and strength retention tests indicated that a higher lime dosage of 8% is required to withstand the wet–dry cycles. Therefore, a higher lime dosage than that determined from soil-lime pH test may be required to account for the detrimental effect of weathering cycles.

The rainfall-induced slope stability analysis was performed with the degraded hydro-mechanical properties of untreated soil, and this facilitated identification of the probable causes of the embankment slope failure. An increase in soil permeability after desiccation cracking, reduction in shear strength properties to fully softened shear strength, and development of a perched water table in the surficial layer of the slope possibly resulted in the slope failure.

The numerical analysis results indicated that 8% lime treatment of the top 2 ft (61 cm) of the surficial layer with an underlying drainage system could prevent surficial slope failures in the future. The volumetric moisture content data recorded by the moisture sensors embedded during the rehabilitation construction work indicate that the lime-treated surficial layer is shielding the underlying untreated soil from detrimental wet–dry cycles and associated moisture fluctuations.

Footnotes

Acknowledgements

The authors acknowledge the support of Mr Noel Paramanantham, P.E. and Mr. Dan Perry P.E. of Paris District, TxDOT for their support of this study. U.S. DOT’s Transportation Consortium for South-Central States (Tran-SET) research grant 18GTLSU06, and the software support provided by GeoStudio for the slope stability studies are acknowledged. The authors would also like to acknowledge the NSF Industry-University Cooperative Research Center (I/UCRC) program funded Center for Integration of Composites into Infrastructure (CICI) site at Texas A&M University, College Station, Award # 1464489 (Phase I) and Award # 2017796 (Phase III), Program Directors: Dr Gregory Reed, Dr Prakash Balan, and Dr Andre Marshall.

Author Contributions

The authors confirm contribution to the paper as follows: study conception and design: A. J. Puppala and B. Boluk; data collection: B. Boluk and P. Bhaskar; analysis and interpretation of results: B. Boluk and S. Chakraborty; draft manuscript preparation: B. Boluk, S. Chakraborty, P. Bhaskar, and A. J. Puppala. All authors reviewed the results and approved the final version of the manuscript.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This interagency contract was funded by Texas Department of Transportation (TxDOT)—Paris District, Transportation Consortium for South-Central States (Tran-SET) research grant 18GTLSU06. The authors would also like to acknowledge the NSF Industry-University Cooperative Research Center (I/UCRC) program funded Center for Integration of Composites into Infrastructure (CICI) site at Texas A&M University, College Station, Award # 1464489 (Phase I) and Award #2017796 (Phase III).

References

Puppala

A. J.

Manosuthikij

Chittoori

B. C. S.

Swell and Shrinkage Characterizations of Unsaturated Expansive Clays from Texas. Engineering Geology, Vol. 164, 2013, pp. 187–194.

George

K. P.

Cracking in Pavements Influenced by Visoelastic Properties of Soil-Cement. University of Missisisppi, 1968.

Acharya

Enhanced Shrinkage Characterization of Clayey Soils. Doctoral dissertation. University of Texas, Arlington, TX, 2015.

T. D.

Pouya

Hemmati

Tang

A. M.

Numerical Modelling of Desiccation Cracking of Clayey Soil. Proc., E3S Web of Conferences, EDP Sciences, Vol. 9, 2016, p. 08009. https://doi.org/10.1051/e3sconf/20160908009.

Fickies

R. H.

Fakundiny

R. H.

Mosley

E. T.

Geotechnical Analysis of Soil Samples from Test Trench at Western New York Nuclear Service Center, West Valley, New York, New York State Geological Survey, Vol. 21, NUREG/Cft-0644. International Atomic Energy Agency (IAEA), Vienna, Austria, 1979.

Lau

J. T. K.

Desiccation Cracking of Soils. Doctoral dissertation. University of Saskatchewan, Canada, 1987.

Kodikara

J. K.

Choi

A Simplified Analytical Model for Desiccation Cracking of Clay Layers in Laboratory Tests. Proc., 4th International Conference on Unsaturated Soils, Carefree, AZ, 2006, pp. 2558–2569.

Peron

Laloui

L.-B.

Hueckel

Formation of Drying Crack Patterns in Soils: A Deterministic Approach. Acta Geotechnica, Vol. 8, No. 2, 2013, pp. 215–221. https://doi.org/10.1007/s11440-012-0184-5.

J. L.

Guo

L. B.

Zhang

Experimental Study on Soil-Water Characteristic Curve for Silty Clay with Desiccation Cracks. Engineering Geology, Vol. 218, 2017, pp. 70–76. https://doi.org/10.1016/j.enggeo.2017.01.004.

10.

Daniel

D. E.

Predicting Hydraulic Conductivity of Clay Liners. Journal of Geotechnical Engineering, Vol. 110, No. 2, 1984, pp. 285–300. https://doi.org/10.1061/(ASCE)0733-9410(1984)110:2(285).

11.

Albrecht

B. A.

Benson

C. H.

Effect of Desiccation on Compacted Natural Clays. Journal of Geotechnical and Geoenvironmental Engineering, Vol. 127, No. 1, 2001, pp. 67–75. https://doi.org/10.1061/(ASCE)1090-0241(2001)127:1(67).

12.

Konrad

Ayad

Desiccation of a Sensitive Clay: Field Experimental Observations. Canadian Geotechnical Journal, Vol. 34, 2011, pp. 929–942. https://doi.org/10.1139/t97-063.

13.

Anderson

D. C.

Brown

K. W.

Thomas

J. C.

Conductivity of Compacted Clay Soils to Water and Organic Liquids. Waste Management & Research, Vol. 3, No. 4, 1985, pp. 339–349.

14.

Herrier

Puiatti

Chevalier

Froumentin

Üonelli

Fry

J. J

. Lime Treatment: New Perspectives for the Use of Silty and Clayey Soils in Earthen Hydraulic Structures. 9th ICOLD European Club Symposium, April 2013, Venice, Italy, pp. 1–10.

15.

Knodel

P. C.

Lime in Canal and Dam Stabilization. U.S. Department of the Interior, Bureau of Reclamation, Denver, CO, 1987.

16.

Daneshmand

Huat

B. B.

Asadi

Farshchi

Slope Stability Analysis with Applicability of Lime in Capillary Barrier Effects. Electronic Journal of Geotechnical Engineering, Vol. 16, 2011, pp. 1287–1299.

17.

Abrams

T. G.

Wright

S. G.

A Survey of Earth Slope Failures and Remedial Measures in Texas. The University of Texas, Austin, TX, 1972.

18.

Dronamraju

V. S.

Studies on Field Stabilization Methods to Prevent Surficial Slope Failures of Earthfill Dams. The University of Texas, Arlington, TX, 2008.

19.

Basma

A. A.

Tuncer

E. R.

Effect of Lime on Volume Change and Compressibility of Expansive Clays. Transportation Research Record: Journal of the Transportation Research Board, 1991. 1295: 52–61.

20.

Charles

Herrier

Chevalier

Durand

An Experimental Full-Scale Hydraulic Earthen Structure in Lime Treated Soil. Proc., 6th International Conference on Scour and Erosion ICSE, Paris, France, 2012, pp. 1223–1230.

21.

Chittoori

B. S.

Puppala

A. J.

Saride

Nazarian

Hoyos

L. R.

Durability Studies of Lime Stabilized Clayey Soils. Proc., 17th International Conference on Soil Mechanics and Geotechnical Engineering: The Academia and Practice of Geotechnical Engineering, Alexandria, Egypt, IOS Press, 2009, pp. 2208–2211.

22.

Meade

B. W.

Allen

D. L.

Evaluation of Lime Stabilized Subgrade (AA-19). Report No. KTC-93-251993. Kentucky Transportation Center, Lexington, KY, 1993.

23.

Fredlund

D. G.

Xing

Equations for the Soil-Water Characteristic Curve. Canadian Geotechnical Journal, Vol. 31, No. 4, 1994, pp. 521–532.

24.

Bhaskar

Boluk

Banerjee

Shafikhani

Puppala

Effect of Lime Stabilization on the Unsaturated Hydraulic Conductivity of Clayey Soil in Texas. In Geo-Congress 2019: Geotechnical Materials, Modeling, and Testing ( Meehan

C. L.

Kumar

Pando

M. A.

Coe

J. T.

eds.), American Society of Civil Engineers, Reston, VA, 2019, pp. 773–783.

25.

Krahn

Seepage Modeling with SEEP/W: An Engineering Methodology. GEO-SLOPE International Ltd., Calgary, Alberta, Canada, 2004.

26.

Lee

L. M.

Gofar

Rahardjo

A Simple Model for Preliminary Evaluation of Rainfall-Induced Slope Instability. Engineering Geology, Vol. 108, No. 3–4, 2009, pp. 272–285. https://doi.org/10.1016/j.enggeo.2009.06.011.

27.

Zhang

Fredlund

D. G.

Examination of the Estimation of Relative Permeability for Unsaturated Soils. Canadian Geotechnical Journal, Vol. 52, No. 12, 2015, pp. 2077–2087.