Soil structure effect on soil erosion potential

Soil erosion poses a significant threat to water-related infrastructure such as bridges, dams, quays, and levees by detaching and transporting soil grains downstream, thereby compromising the structural support of these installations. While erosion damage is acknowledged in current design practices, understanding soil erosion parameters requires scrutiny. However, existing soil erosion databases mainly rely on reconstituted soil samples, which may differ substantially from in situ erosion due to alterations in soil structure. This study scrutinizes and contrasts the erodibilities of in situ and reconstituted soils. In situ soil samples were obtained using thin-walled Shelby tubes from Victoria, Canada, while reconstituted specimens were prepared in a slurry state and consolidated to match the overburden pressure on-site. A custom rotational erosion testing apparatus facilitated erosion testing on both Shelby tube and reconstituted specimens. The findings shed light on the influence of soil fabric on soil erosion potential, an aspect currently lacking in comprehensive understanding.


Introduction
Soil surface erosion represents the natural process of scouring exterior surfaces of earthen embankments or streambeds, where flowing fluids detach and transport soil particles.Recognized as a major contributor to infrastructure failures in water bodies, such as bridges and levees, surface erosion's significance is underscored by research.Shirole and Holt (1991) found that over 60% of bridge structural failures in the US were erosion-related [1], while more than 30% of dam failures resulted from overtopping due to surface erosion, triggering severe downstream flood events [2].
Recent laboratory tests aimed at assessing earthen material erodibility under varying hydraulic conditions have yielded valuable insights.Kandiah and Arulanladan (1974) explored cohesive soil erosion using a straight open flume [3], highlighting the impact of soil moisture and eroding fluid salt content on erosion behavior.Hanson and Hunt (2007) investigated the effects of compaction and moisture on soil erodibility using a submerged jetting device [4], while Briaud et al. (2001) proposed an erosion-function apparatus (EFA) for surface erosion testing, leading to an erosion chart classifying earth materials into six erodibility categories [5,6].
Many laboratory tests have relied on artificially created specimens, which may not accurately represent field conditions.The soil structure, including particle arrangement and inter-particle bonding, developed in situ during depositional and post-depositional processes, can significantly differ from that of laboratory-prepared soils [7].Young et al. (1975) conducted a comparison between laboratory and in situ flume erosion testing results, revealing that critical shear stress obtained in the laboratory could be overestimated by a factor of two [8].However, limited research has been undertaken to investigate the effects of soil structure on soil erosion.Consequently, there exists a gap in understanding in-situ soil erosion behavior based on the existing database, primarily established using laboratory-prepared specimens.
This study seeks to address this gap by comparing in situ and laboratory-reconstituted soils.In situ specimens were collected using thin-walled Shelby tubes from Victoria, Canada, while reconstituted specimens were prepared and consolidated to match on-site conditions.Complementary tests evaluating compressibility, strength, and surface erosion were conducted using a purpose-built rotational erosion testing apparatus, facilitating comprehensive comparison and analysis.

Materials
In this study, both in situ and reconstituted soil specimens were examined, and their basic properties are detailed in table 1.In situ soil samples were collected via thin-walled Shelby tubes from a depth of 20-22 feet below the ground surface at sites in Victoria, Canada.Prior to experimentation, the Shelby tube specimens were extruded using a hydraulic jack.The soil was identified as high-plasticity clay (CH) with a liquid limit of 51.9% and a plastic limit of 26.6%, determined by the percussion cup method and thread-rolling test (ASTM D4318) [9].Reconstituted soil specimens were prepared following the method recommended by Burland (1990) to achieve intrinsic behavior [10].The soil was thoroughly mixed above the liquid limit to disrupt its structure.The resulting soil slurry was then consolidated at the same overburden pressure (90 kPa) onsite until primary consolidation was complete, utilizing a consolidation frame, as depicted in figure 1.

Standard oedometer test.
A standard oedometer test was performed using Terraload S-450 to evaluate the one-dimensional consolidation properties of the soil, following ASTM D2435/D2435M-11 [11].The discrepancy in compressibility between the in situ soil and the reconstituted soil indicates a soil structural difference.

Direct shear test.
A direct shear test was performed under consolidated drained conditions according to ASTM (D3080) [12].The shearing process was established at a low shear rate to avoid excess pore water pressure build-up.Moreover, direct shear tests were performed at low effective normal pressures to mimic the debonding mechanism of the erosion process [13].

Erosion test.
Surface erosion tests were conducted using a purpose-built Rotational Surface Erosion Testing Apparatus (RSEA), illustrated in figure 2. This apparatus consisted primarily of a motor, power supply, rotating acrylic chamber, torque measurement assembly, and data acquisition system.Soil samples were positioned coaxially within the outer chamber and secured using two specimen caps, which remained stationary during erosion tests.The apparatus generated rotational flow, eroding the side surface of a cylindrical soil specimen.As the chamber rotated, fluid in the annulus was dragged by the moving wall, imposing hydraulic shear stress on the specimen surface.Chamber rotation speed (RPM) was controlled and measured using a power supply and tachometer, respectively.The erosive force, i.e., hydraulic shear stress, was balanced by the spring force in the torque measurement assembly, which included a rotary sensor, pulley, and tension spring, as depicted in figure 2. Using the known spring constant and measured spring deformation, the corresponding hydraulic shear stress was computed for erosion analysis [14].Each erosion test lasted two minutes to assess soil erosion behavior under different hydraulic conditions (RPM).Following each test, eroded soil particles were collected with the eroding fluid using a vacuum pump and dried to a constant weight.Erosion rate was calculated by dividing the mass of eroded soil by the erosion duration and surface area of contact.Further details on device design, operation, calibration, and calculations were provided by Lin et al. ( 2023) [14].curve of in-situ Shelby tube specimens exceeded that of reconstituted soil.Natural soil structure restrained deformation under low effective vertical pressures, as evidenced by the initial portion of the consolidation curve in figure 3.However, beyond the consolidation yield stress (approximately 13 kPa), natural soil became more sensitive to changes in vertical effective stress [15].Based on the linear portion of the consolidation curve in figure 3, the compression index of natural soil was 34% higher than that of reconstituted soil, as listed in table 1.This suggests that the difference in void ratio between natural and reconstituted soils decreased with increasing effective vertical stress after the consolidation yield stress.

Figure 3. Void ratios at different consolidation pressures
Figure 4 displays odometer test results using normalized void ratios following Burland (1990), calculated using Eqs.(1) [10].Intrinsic compression (ICL) was determined based on reconstituted soil consolidation test results, while a sediment compression line (SCL) was employed empirically based on a consolidation database of multiple natural soils [10].Over the 10-1000 kPa range, ICL and SCL were approximately parallel.At a given void index over this range, effective overburden pressure carried by SCL was about five times that carried by ICL [10].Odometer test results for natural soil are plotted in figure 4. Natural soil was less compressible before yield consolidation stress due to deformation constraint from internal soil structure.Difference in void index (I_v) between natural soil and ICL at the same stress level increased with rising vertical effective stress up to yield stress.This trend reversed after a brief overlap with SCL and tended to converge with ICL due to restructuring from increased overburden.
where   is the void index,  is the void ratio of the point of interest,  100 * and  1000 * are the intrinsic void ratios corresponding to normal effective stresses of 100 kPa and 1000 kPa, respectively.The e1000*intercept of  1000 * was used in this study.

Structural effect on soil strength
Soil surface erosion manifests at the soil-water interface, where normal effective stress is negligible.Hydraulic shear stress initiates erosion by breaking inter-particle bonds, akin to direct shearing procedures studied previously [13].In this research, direct shear tests were conducted at low normal effective stress levels for both natural and reconstituted soils to explore their impact on soil strength, reflecting inter-particle bonding and erosion resistance.Figure 5 illustrates stress-strain relations of natural and reconstituted soils under different normal effective stresses.Reconstituted soil stress-strain curves resembled those of natural soil, albeit with notable distinctions.The artificial remolding process during reconstitution disrupts inter-particle bonding and damages the primary load-carrying structure, leading to strength reduction.Notably, a peak shear strength was evident for natural soil, absent in reconstituted soil.This disparity may arise from natural soil initially bearing load through its structure during shearing, resulting in peak stress once threshold resistance is surpassed.Conversely, reconstituted soil lacks a distinct structure, leading to uniformly distributed shear stress and "harden type" stress-strain curves devoid of peaks.

Structural effect on soil erodibility
Soil surface erosion tests were conducted using RSEA under varied hydraulic conditions for both natural and reconstituted soils.Erosion tests were controlled by adjusting chamber rotation speed and voltage via a power supply.Fig. 6 depicts soil erosion rates under each hydraulic condition, indicating a 38% lower erosion rate for natural soil.The erosion mechanism parallels that of direct shear tests but is driven by hydraulic shear force.Due to structural development during deposition and post-deposition, natural soil exhibits greater resistance against shearing, as evidenced by direct shear tests.Consequently, natural soil also displays heightened resistance against erosion-induced shearing, explaining the reduced erosion rate.Additionally, erosion curves based on the linear excess shear stress equation  =   ( −   ). are depicted in Fig. 6.The erosion curve for natural soil closely mirrors that of reconstituted soil, suggesting similar erodibility coefficients (Kd) for both soil types.However, the natural soil's critical shear stress (τc), indicating erosion initiation, was not captured in RSEA testing, likely due to soil particle disintegration in water and subsequent erosion susceptibility, a phenomenon reported in other erosion experiments [16].This research underscores the dominance of structural effects over void ratio effects in soil erosion.Prior studies consistently highlight an inverse correlation between void ratio and soil erodibility [17][18][19].Soil with lower void ratios typically exhibits higher erosion resistance due to increased interparticle friction.However, natural soil in this study, despite having a larger void ratio (figure 3) than reconstituted soil, displayed higher erosion resistance.This disparity likely arises from increased erosion energy required to remove fabric-induced clogging and disrupt inter-particle bonding in natural clay [7].

Conclusions
To summarize, this study delves into the impact of soil structure on soil surface erosion.Naturally structured soil collected via Shelby tubes was compared with reconstituted soil prepared at the liquid limit and reconsolidated to an equivalent stress level.Standard oedometer and direct shear tests were conducted to assess void differences and shear resistance, respectively.Erosion tests utilizing RSEA examined soil erodibility under varying structures.The findings suggest: (1) Natural soil exhibits larger void ratios but converges with intrinsic compression after initial structure disruption; (2) Natural soil demonstrates superior shear resistance due to natural fabric's bonding and clogging effect; (3) Natural soil exhibits lower erodibility than reconstituted soil, with a 38% lower erosion rate under equivalent hydraulic conditions.

Figure 2 . 3 . Results and discussion 3 . 1 .
Figure 2. Rotating Surface Erosion Testing Apparatus (RSEA)3.Results and discussion3.1.Structural effect on compressibilityFigure3illustrates soil void ratios at varying consolidation stresses based on one-dimensional odometer tests.It's evident that compression curves varied slightly.The compression curve of natural soil was convex, while that of reconstituted soil was slightly concave upward, consistent with Burland's findings (1990)[10].Odometer test results for reconstituted soil reflected intrinsic compression behavior.Differences in natural soil compressibility mirrored soil structure developed during deposition and aging, including fabric (particle arrangement) and interparticle bonding[7,10].In general, the compression

Figure 4 .
Figure 4. Void index at different consolidation pressures